Converging Grammars: Constructions in Singapore English 9781614514091, 9781614515715

Citation preview

Debra Ziegeler Converging Grammars

Language Contact and Bilingualism

Editor Yaron Matras

Volume 11

Debra Ziegeler

Converging Grammars Constructions in Singapore English

ISBN 978-1-61451-571-5 e-ISBN (PDF) 978-1-61451-409-1 e-ISBN (EPUB) 978-1-5015-0063-3 ISSN 2190-698X Library of Congress Cataloging-in-Publication data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2015 Walter de Gruyter GmbH, Berlin/Boston Typesetting: Compuscript Ltd., Shannon, Ireland Printing and binding: CPI Books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Table of contents Preface and acknowledgements Abbreviations  xi List of figures and tables  xiii List of tables  xiv

 ix

Chapter 1 Introduction 1 1.1 Main terms and definitions 1.2 Overview 9

5

Chapter 2 Singapore English 15 2.1 Introduction: Historical background 15 2.2 The sociolinguistics of Singapore English 20 2.2.1 Earlier studies 20 2.2.2 The situation today 24 2.3 Grammatical morphology and discourse features 2.3.1 Tense, aspect and modality 28 2.3.2 Other grammatical features 34 2.4 Overview 39

28

Chapter 3 Construction grammars and the paradox of ‘mixed’ construction types 3.1 Introduction: constructions in variational contexts 41 3.2 Questions of construction descriptions 43 3.2.1 Identification and terminology 43 3.2.2 Compositionality 48 3.2.3 Meaning 53 3.2.4 Form-meaning alignment in other accounts 55 59 3.3 Construction(al)isation 3.4 Construction development and coercion 67 3.4.1 Cyclical interaction 70 3.5 Summarising the current position 72 Chapter 4 Transitivity and causativity 77 4.1 Introduction 77 4.2 What is a conventionalised scenario? 80 4.2.1 Earlier reference to the conventionalised scenario

80

41

vi 

 Table of contents

4.2.2 Adversative conventionalized scenarios 82 4.2.3 Constraints on the use of CSs 85 4.2.4 The causative‑­resultative alternate 88 4.2.5 Adversative resultatives 90 4.3 Conventionalised scenario constructions in Singaporean English 93 4.3.1 Substrate influence 94 4.3.2 The quantitative survey 99 4.3.3 Results 101 4.3.4 Comparative overview 105 4.4 Discussion 107 4.4.1 Pragmatic mechanisms of causativity reduction 4.4.2 The subject role 111 4.5 Summary 114

108

Chapter 5 Experiential aspect 117 5.1 Introduction 117 5.2 The ever construction 118 5.2.1 Negative polarity ever 118 5.2.2 SCE ever 119 5.3 Contact and substrate languages 123 5.4 Contact grammaticalisation as a possible explanation 5.5 Historical functions of English ever 132 5.5.1 Universal quantifier uses of ever in SCE 135 5.6 Logical explanations of meaning changes 136 5.7 Discussion 140 5.8 Summary 142 Chapter 6 The past tense construction 143 6.1 Introduction 143 6.2 Tense marking in habitual aspect in SCE 145 6.3 Tense marking in habituals in other languages 6.3.1 Slavic 150 6.3.2 Chinese dialects 152 6.4 Preliminary survey data 154 6.4.1 Search items 155 6.4.2 Examples of the use of pasts-for-presents (PFP constructions) 156

149

127



Table of contents 

 vii

Distributional frequency 6.4.3 168 6.4.4 Interim summary 170 6.5 Discussion 170 6.5.1 Present-perfectives and the realis-irrealis interface in English 173 6.6 Summary 178 Chapter 7 Bare noun constructions 181 7.1 Introduction 181 7.2 Number marking in Singapore Colloquial English count nouns 182 7.3 Specific and non‑­specific nouns 184 7.4 Bare Noun Constructions in creole systems 187 7.5 Further examples of the Bare Noun Construction in SCE 7.5.1 Zero‑­plural BNCs 191 7.5.2 More recent data 194 7.5.3 Specific markers in SCE 197 7.6 Number marking and the Chinese substrate 200 7.7 The Bare Noun Construction and construction coercion 7.8 Applying the coercion hypothesis to the contact data 7.8.1 A grammatical metaphor 211 7.9 Summary 213

191

204 207

Chapter 8 The Merger Construction: a model of construction convergence 215 8.1 Introduction 215 8.2 Mechanisms of contact construction development 216 8.2.1 Convergence 216 8.2.2 Material and pattern copying 219 8.2.3 Grammaticalisation 222 8.2.4 Equivalence, and other constraints 225 8.2.5 Relexification and systemic transfer 228 8.3 The case studies in the present volume 232 8.3.1 Transitivity and the conventionalised scenario construction 8.3.2 The experiential aspect construction 236 8.3.3 The past tense construction 238 8.3.4 The bare noun construction 240 8.4 Previous studies of contact constructions 242 8.5 The Merger Construction Model 247

234

viii 

 Table of contents

8.5.1 The Transitive Merger-Construction 248 8.5.2 The Experiential ever Merger-Construction 8.5.3 The Past Tense Merger-Construction 253 8.5.4 The Bare Noun Merger-Construction 255 8.6 Summary 258 Chapter 9 Concluding remarks References Index

289

271

261

251

Preface and acknowledgements The main inspiration for this collection of studies developed out of one of the studies, summarised in Chapter 7, in which a case for construction coercion was held up for challenge against the data obtainable from Singapore English. In particular, it was inspired by a question posed by one of the referees of Constructions and Frames, who asked whether Singapore English should be understood (in construction terms) as actually a dialect of English or a different language. For the purposes of the present work, the question extends far beyond the puzzling data produced as part of that study, and certainly could be asked of any mixed language situation, creole or otherwise. In response to that initial prompting, I felt it was necessary to approach the problem of constructions in contact situations using data from the more difficult areas of contact such as those in which the transparency of substrate modelling is not so clearly in evidence, and furthermore, to attempt to create a theoretical platform based on the use of ambiguous construction frameworks. The confusion of contact constructions has been tackled by few researchers so far, and it is hoped that the present volume of studies can contribute towards resolving many of the problems that arise from applying a strict construction-based descriptive approach to cases of languages in which the syntax and the lexicon derive from different languages. Amongst the other papers that have contributed to this volume are included one that was written as the output of a one-year UK ESRC grant held at the U ­ niversity of Manchester (2003) (R000223787), with the help of Sarah Lee. Chapter 4 appears as an extended version of the original study, and includes more detail than was permitted to appear in its original published form due to space limitations. I am grateful to all the colleagues and students at Manchester who might have helped in providing scholarly comment or assisting in any way to shape this study. Chapters 5, 6 and 7 were for the most part both written during an enriching, two-year Visiting Fellowship at the National University of Singapore Department of English Language and Literature, 2007-2008, while working in a “dream” office overlooking a romantic tropical harbour, dotted with shining white cargo ships and lined with coconut palms. I am grateful to colleagues and students at the Chinese Department of the National University of Singapore for their comments on the study in Chapter 5, and for inviting me to present an earlier version of the study as a colloquium in their department. Chapter 6 was the output of a small grant provided by the NUS Faculty Staff Support Scheme (2008-9), and I thank in particular my student assistant, Amelyn Thompson, for her ideas and assistance in data gathering for that study. To the many colleagues from the Department of English Language and Literature at NUS, including Bao Zhi Ming, Vincent

x 

 Preface and acknowledgements

Ooi, Peter Tan and Lionel Wee, who might have at one time or another provided ­comments or suggestions on these studies, I am also very grateful. But most of all I must acknowledge the enthusiasm and interest of the undergraduate students of the Department of English Language and Literature at NUS, whose creative passion and infinite respect for learning and scholarship provided me with enormous inspiration for research during my two-year visit to Singapore. Other ideas in the studies were presented in France at the AFLiCo conference, University of Lyon II, May 2011, the AFLiCo conference, University of Paris 10, Nanterre, May 2009, and at colloquia and presentations at the Sorbonne-Nouvelle University Paris 3, 2012 and 2013. I am grateful to colleagues including Hilary Chappell, Salikoko Mufwene, and to audience members at those presentations for their assistance and useful comments. The editors and anonymous referees of English World-Wide, Language Sciences, and Constructions and Frames must also be thanked for their suggestions and ideas in contribution to the original papers that formed the basis of Chapters 4, 6, and 7. I am also especially indebted to Yaron Matras, and the editorial team of the Language Contact and Bilingualism Series, for their patience and wise advice on the structure of this book. Others who should be thanked for their comments and suggestions on the work presented in the studies include Werner Abraham, Markku Filppula, Marc Fryd, Larry Horn, Elisabeth Leiss, Diana Lewis, Lucia Tovena, Elizabeth Traugott, and colleagues and students at the Sesylia seminars at Université Paris III, where sections of the book were presented. Last but not least, I must thank as always the patient support and encouragement of Tuck, and courageous little Kuai Leui, whose constant and loyal companionship on my computer desk never ceases to provide me with the right philosophy on life. Debra Ziegeler Paris, November 2014

Abbreviations 1 1st person 2 2nd person 3 3rd person ACC accusative agt agent ART article ASP aspect marker BA pre-transitive marker BNC bare noun construction CAUS causative CL classifier CR causative-resultative (construction) CRS currently relevant state (le) CS conventionalised scenario DEMPART demonstrative particle EE experiential ever EVD evidential F feminine FUT future INDF indefinite M masculine N(P) noun (phrase) NEG negative marker NPE negative-polarity ever Nom. nominative OBJ object OBL oblique P(P) preposition (phrase) PART participle or converb PASS passive PFP past-for-present PFV perfective PL plural PST/ past PAST pat patient QP question particle

xii 

 Abbreviations

rec recipient REL relator RP relative pronoun SCE Singapore(an) Colloquial English S/SG singular SUBJ subject SSE Singapore(an) Standard English/Standard Singapore(an) English V(P) verb (phrase)

List of figures Fig. 1: Three principle construction types showing form-meaning alignment differences Fig. 1a: Langacker’s Cognitive Grammar Fig. 1b: Croft’s (2001) Radical Construction Grammar Fig. 1c: Goldberg’s construction grammar: composite fused structure caused motion + put (Goldberg 1995: 52) Fig. 1d: Goldberg’s (to appear) caused-motion construction Fig. 2: Cyclical interaction between generalisation and lexical constraints in constructionisation Fig. 3: The transitive construction (Goldberg 1995: 117) The ACTION FOR RESULT metonymy: metonymic links between Fig. 4:  the ­causative-resultative construction (X gets X’s hair cut) and the ­conventionalised scenario (X cuts X’s hair). Fig. 5: Advertising flyer illustrating the variable feature of zero-plural BNCs in Singaporean (commercial) English. Fig. 6: Singapore traffic warning sign (April, 2008) Fig. 7:  Form-function poles of the Bare Noun Construction in Singapore ­Colloquial English Fig. 8: The Transitive Merger-Construction Fig. 9: The Experiential ever Merger-Construction Fig. 10: The Past Tense Merger-Construction Fig. 11: The Bare Noun Merger-Construction

List of tables Tab. 1: Responses to the elicitation survey, showing frequency of CSs relative to causative resultatives (CRs). Tab. 2: Average of totals for causative-resultatives and conventionalised scenarios, in percentages. Tab. 3: Responses using habitual aspect in answer to the lead posting: How long do you take for shower? What’s the first thing that you will mostly do in the bathroom? (ex. 171) Tab. 4: Probability of the occurrence of the specific determiner one relative to indefinite article use in a restricted portion of the Flowerpod Corpus (the Pen Club thread). Tab. 5: The components and constraints of nominal type-shifting. Tab. 6: The components and constraints of count-to-mass type-shifting, using ambiguous examples in (192–212).

Chapter 1  Introduction The vernacular dialect of Singapore English, commonly known as ‘Singlish’, has provided the source for an overwhelming number of research efforts in the past 50 years since the island of Singapore first became an independent nation. In this regard, many or even most of the linguistic studies undertaken on Singlish have focused more particularly on sociolinguistic or socio‑­historical factors under‑ lying its development. Such studies have referred, in the past, to proficiency or educational levels (e.g., Platt 1978), diglossia (e.g., Gupta 1991, 1994), and more recently, the diglossic account has been extended to refer to a clear, functional distinction between Singlish and the standard variety of English spoken on the island as well (Alsagoff 2010), which differs little from other standards except perhaps in pronunciation (Gupta 1994). As such, Singapore English in general has not suffered from a dearth of descriptive accounts, reflecting, above all, the way in which the dialect and its sub‑­varieties have changed and evolved during the brief period of time in which there has been a consciousness of an autono‑ mous variety of English. In addition, recent studies on contact languages have extended their horizons to the study of Singlish and its fluctuating conditions of contact, ­initially having a multilingual substrate made up of at least 10 langua‑ ges spoken in ­southern China as well as southern Indian languages and Malay, and now in parallel contact with Mandarin Chinese, as well as the standard sub‑­ variety itself. The approaches of these former studies, with the exception of a few (e.g., Bao 2010) have remained descriptive throughout. The present volume does not deny the need for additional research into descriptive approaches to the study of Singapore English, but it will also aim to support the application of a more theoretical approach than has been the focus of many earlier studies. A selection of grammatical characteristics, the object of previous research, will be reviewed from within the descriptive framework of unrelated languages and non‑­contact varieties of English, as well as from the contrastive point of view of Standard Singapore English with which the dialect is in constant contact. It is intended in this way to raise awareness of the study of contact varieties of English as a sub‑­typological domain of enquiry open to the various influences of universal cognitive and functional tendencies, rather than a pervasive object of continual (and often prescriptivist) sociolinguistic debate. For this reason, it is considered an ideal target of investigation for the applica‑ tion of new cognitive linguistic theories such as construction‑­based approaches. Such approaches have rarely been applied to contact data in previous accounts (Pietsch 2010 and Höder 2014 being two exceptions, as discussed in Chapter 8), but have not been exploited in the same way as in the present account, which aims

2 

 Chapter 1 Introduction

for a unified construction grammar ranged across the total contact situation. It also aims in this way to bring to light a number of problems surrounding the ­theories of constructions themselves, such as the notion originally proposed by Goldberg (1995) of a construction semantics determined by the form or syntax of the construction alone. This notion is a perennial problem for construction theory, as recent definitions still attest to such claims, e.g., Boas (2011: 1273): (1) Constructions are form‑­meaning pairings that exist independently of the particular words that instantiate them. Such definitions require investigation from all perspectives, and Boas’ defini‑ tion, along with other similar ones, will be part of the main focus of analysis in Chapter 3. The underlying research questions which motivated the compilation of the present volume revolve around the problem of describing constructions in mixed‑­ language situations. Although Singapore Colloquial English has not been classed technically as a mixed language, i.e., generally‑­speaking, one that is seen to contain the grammar of one language and the lexicon of another (see, e.g., Bakker and Mous (eds., 1994), Matras and Bakker (eds., 2003)), it contains a number of features that could feasibly contribute to the application of such a definition (such as the entire aspectual system, as pointed out by Bao (2005) or the nominaliser, one (Bao 2009); see Chapter 8). Ansaldo (2009: 52), also describes Singlish as an Asian variety of English with a ‘‘... Sinitic‑­type grammar’’. The features descri‑ bed by Bao are grammaticalised and argued to reflect Chinese ­features ‘‘... filte‑ red through the morphosyntax of English’’ (2005: 237), which raises a number of questions for the challenge of cognitive approaches. One is the integration of constructions in contact dialects: if it has been demonstrated in the more funda‑ mentalist construction literature that constructions follow precise, form‑­function compositional correspondences, with the result that it is the form of the con‑ struction type that may ultimately determine its meaning, at least at a certain level as suggested in (1), then how can the mixed language situations described above cope with a form‑­dictated construction semantics? It would mean that the ­original lexifier, English, not only provides the form for the constructions in Singlish, but also the meanings associated with those forms. However, we know that that is not always the case, as the Singlish aspectual system functions have been transferred from the Chinese substrate according to Bao (2005), and English aspectual categories are quite different. Thus, one of the major purposes of the present volume is to attempt to derive a contact construction semantics in which the form‑­meaning correspondences of the original lexifier of Singlish, English, are plausibly accommodated. The forms of English constructions have often been selected by speakers of Singlish to articulate functions of the substrate languages,

Introduction 

 3

and it is reasonable to expect that the selections must be based on judgements of specific analogies made across contact. Such judgements will become the topic of discussion in Chapter 8. Other researchers such as Croft (2001) and Traugott (2014), for example, ­postulate a much broader definition of constructions than Boas’ (2011) d ­ efinition, one which includes ‘atomic’ form‑­meaning correspondences such as single‑­word/ phoneme items that could not possibly be covered by the definition in (1) (e.g., for Trousdale 2012, a construction is simply a conventionalised form‑­meaning correspondence and nothing more). Boas (2003, 2011) and others, such as Iwata (2002) have proposed the existence of more general schematic semantic levels that underlie a construction‑­based semantics: ‘mini‑­constructions’ in the case of Boas, and Croft (2003) has also discussed ‘verb‑­specific’ constructions, while Goldberg (2009a) admits to the possibility of prototype constructions (­Ziegeler 2010 describes a similar phenomenon as a Lexical Prototype Construction –­see Chapters 3 and 8). All these descriptions will be reviewed in Chapter 3, as they indicate the potential for syntactic analogy to rely on lexical analogy much more significantly than earlier construction theorists had originally predicted. Given the possibility of a lexicon‑­driven construction inventory (or constructicon), it will be argued throughout the cases studies presented that the description of a contact construction is thus much easier to derive. However, what also needs to be taken into consideration are the capacities for constructions to express not just form‑­meaning correspondences, but also form‑­function ones. Little attention has been devoted to such areas of research in previous literature on construction grammars. For the main objectives of the present volume, the definition in (1) is therefore certainly important, given the most searching question in the face of construction theory is how to describe the constructions of a language in the case of contact varieties of that language which exhibit the syntax of one language (usually the substratum, or model language) and the lexicon of another. Do they possess only the constructions of the lexifier, or the constructions of the language which sup‑ plies the (morpho)syntax? Furthermore, how is the form‑­meaning relationship described in contact situations? It is not always clear what comprises ‘form’: phonological form or syntactic form, and it is made clear in Chapter 3 that this may differ from one theory to another. In some cases of contact constructions, such as those which are ‘borrowed’ from a substrate language (material‑­copied, according to Matras and Sakel 2007), both the form and the meaning are transfer‑ red constructions. However, in other cases, if the semantics of a construction are derived autonomously from the syntax, then is it the syntax of the substrate that supplies the semantics of the construction, albeit infused with the lexicon of a dif‑ ferent language? (This question could be asked in the case of the ‘false’ transitive

4 

 Chapter 1 Introduction

constructions discussed in Chapter 4.) If it is the case, then the ­development of more schematic construction layers linked to semantic relations with items of lexical input may be problematic for a theory of constructions. Although it has been claimed (Traugott 2007: 525) that the syntax and semantics of constructions are specific to time and variation, it does not mean they are independent of time and variation, since varieties are determined ultimately by their own variation (see, e.g., Hudson 1986: 24), and all languages are susceptible to variation at any level –­there can be no variation‑­free language. The principle aim of the present volume will be, therefore, to come to terms with the problems of describing the replication of constructions in a typical contact situation, using the contact dialect of English represented in Singa‑ pore. Contact varieties and contact languages are seen to be characterised by a mixed ­construction inventory, one that has been associated with ‘convergence’, or pattern replication, by Matras and Sakel (2007). Convergence is essentially a result of form‑­function mapping (but see further discussion in Chapters 3 and 8), and Matras and Sakel (2007: 841) have referred to similar situations formerly described as exhibiting substrate influence in second‑­language acquisition, or relexification of form‑­function patterns found in the substrate. With regard to the replication of construction patterns in contact, the question arises how to ­accommodate the fact that in many cases, the construction meaning is derived from the nature of its most prototypical lexical associations (e.g., caused‑­motion verbs in Sam sneezed the napkin off the table), and if the model ­construction‑­types are replicated in contact, then the model construction semantics may be replica‑ ted as well. Often, of course, this will not cause problems. It is naïve, though, to assume that a typical Singlish example, such as the following (observed perso‑ nally), might represent the same form‑­meaning pairing (and thus the same cons‑ truction) as it does in other varieties of English: (2) Salt‑­water can’t breed. (Ziegeler 2008a: 412). (i.e., ‘[In] saltwater, [mosquitoes] can’t breed’), and yet, we might ask, is this still a variety of the same language, demonstrating as it does the topic‑­­prominent, dropping features familiar to Chinese, with which it is in constant subject‑­ contact? The lexicon is English, but construction grammarians call for a unifi‑ cation of the syntax with the lexicon. So which lexicon? The problem emerges of having to determine the construction semantics from the syntax of a language, in the top‑­down approach to constructions usually advocated by proponents of the Goldbergian revolution, without over‑­proliferating the construction inventory to account for every conceivable variant found in every English dialect. There is also the problem of determining the semantics of forms that do not belong to a periphrastic syntactic structure, in constructions consisting of a single



1.1 Main terms and definitions 

 5

word, as in the Ever Construction discussed in Chapter 5. A related question is how to determine the meaning of zero morphosyntactic representation, as in the Bare Noun Construction discussed in Chapter 7, where only a lexical input item carries the meaning, or so it seems. The replication of constructions in contact situations is seen to be problematic, from the point of view of traditional const‑ ruction grammar, and it is one objective of the present study to develop the theory of constructions in such a way that it readily accounts for situations of variation, contact or otherwise. In this way, the general purposes of the volume are two‑­ fold: to bring to light a fundamental problem of construction‑­based approaches not so far researched at any depth, and to use the findings of such research to explain certain grammatical phenomena of a contact dialect which have been left relatively under‑­researched for far too long.

1.1 Main terms and definitions Singapore Colloquial English (Singlish) has been variously described in past literature not only as a contact language, but also as a second‑­language variety (L2), as well as a ‘new’ English (Platt et al. 1984). The latter two terms are less frequently preferred in recent studies as they refer more to the situation of contact of around the time of the nation’s independence (1965) and up to about 20 years after. Chapter 2 will cover in more detail the sociolinguistic adequacy of such terms, but it is felt that the term L2 variety is probably only relevant historically nowadays, as is ‘new’ English (though Schneider (2003) still adheres to this term, using it to describe Australian English as well). In other ways, it could be descri‑ bed as a former L2 variety with L1 speakers, though such terms still tend to carry connotations of hierarchical ranking of language sub‑­varieties. The linguistic term contact variety is easily the most socio‑­politically neutral descriptive term to use. As a ‘new’ English, it may only be understood in comparison to ‘old’ Englishes, though Australian English is roughly the same age and is not normally considered ‘new’, due to the fact that it was transported and transmitted without generational interruption. All such terminology is replete with unavoidable political nuances. For the purposes of the present volume, the sub‑­variety of Singapore Collo‑ quial English (SCE, or Singlish) is contrasted with Standard Singapore English (SSE according to Gupta’s 1991 diglossic categories); from the point of view of present‑­day usage, the two sub‑­varieties are in constant contact, and thus it is inevitable that they may also influence one another. The concept of a lexifier is no longer relevant and therefore not the same as in many other contact situ‑ ations, in which the language being replicated in contact is associated with a much more recent time period. The original English lexifier, which according to

6 

 Chapter 1 Introduction

Gupta (1998) could have come from a variety of international sources, including ­non‑­native ones, is no longer the language from which norms are accepted, to adopt Kachru’s (1992) terms. Thus, if there is any concept of a lexifier still relevant to the ­Singapore situation of contact today, in the 21st century, it is the norms of a local standard variety with which Singlish remains in contact, hence, Stan‑ dard Singapore English. Thus, for the studies to be investigated in the present volume, the benchmark against which features of variation will be analysed is the local standard English itself, rather than the abstract, notional, standard English of international usage, regardless of how little the local sub‑­variety varies from other international varieties. From the point of view of comparison, then, the two varieties of English which will be involved in the cross‑­contact, merged const‑ ruction to be discussed in Chapter 8 will be the two sub‑­varieties of Singapore English: Singapore Colloquial English and Standard Singapore English. A further point concerns the use of the terminology to describe the languages of contact. The term substrate has been used above for reasons of clarity, but it has often been associated with creole or pidgin situations, and it has been reiterated in many previous accounts that the situation of English in Singapore was never that of a typical creole; in fact, it has been labelled a creoloid by Gupta (1991), for example. Thus there is no real substrate‑­superstrate divide as is traditionally dis‑ cussed in contact studies on creoles. This also accords with the fact that the lexi‑ fier is not a term that can be readily applied to the Singapore situation either. For this reason it is appropriate to use the terms used in Heine and Kuteva (2005) of model language or M‑­language, and replica language being the language created out of contact. More recently, Heine (2012) has distinguished M, a Model language as one providing the model for replication in contact (regardless of whether or not it is a substrate), R, the Replica language being the language of replication of the model, and another contributor, the *R, or pre‑­contact replica language, marked with an asterisk to indicate that in many cases, it needs to be reconstructed histo‑ rically. In their (2005) study, Heine and Kuteva made no distinction between the language often called the lexifier, and the replicating language of contact; in the present study, the term pre‑­contact replica language will be used to indicate the language usually called the lexifier in creole situations, in the Singapore case, a hypothetical variety of colonial English. Thus, it is because the lexifier is an ephemeral entity in most creole situa‑ tions, and certainly in the history of Singapore English, that it is preferable to benchmark SCE with present‑­day Standard Singapore English (SSE), rather than a historical reconstruction (but this may be referred to more generally throughout as simply standard English, or the standard language/dialect). The replica lan‑ guage, in turn, is the term used to refer to Singapore Colloquial English, and the model language may vary amongst any of the contact languages once spoken, or



1.1 Main terms and definitions 

 7

now spoken, in Singapore, such as the southern Chinese dialects or Mandarin. The more general term of a contact language or a contact dialect will be intended to refer to the English replica language of contact, Singlish, unless specifically indicated, rather than to any of the languages with which Singlish may come into contact. This is in line with Ansaldo’s (2009: 1) use of the term in reference to new languages of contact, rather than the pre‑­existing languages which contri‑ bute to the formation of the new languages (the pre‑­existing languages may be referred to simply as the languages of contact). The term (model) substrate may also be used, since these languages are technically and historically classifiable as substrate languages of contact in Singapore. On the other hand, as indicated in Chapter 5 and elsewhere, the potential for Standard English (Singaporean or otherwise) to provide the model, is not out of the question, as it has also been discussed in Ziegeler (2014a) in a study of contact grammaticalisation. Standard ­Singapore English, as well as Mandarin Chinese and the other official languages of ­Singapore (Malay and Tamil) may be considered potentially adstrate languages. More terms will be explained in the succeeding chapters. There are also a number of underlying assumptions which are understood as associated with the theoretical approach to be taken throughout the book. One of these is the presence of semantic continuity as a guiding principle of change across time and variation (see also Ziegeler 2006, Chapter 1). It has been often stated in studies on grammaticalisation (e.g., Heine 1997) that the main purpose for the use of lan‑ guage is to convey meaning. Thus, notwithstanding the efforts of the meticulous descriptive architectures of current and past theories on construction grammars (and similar approaches, as discussed in Chapter 3) to capture language struc‑ ture, a construction serves only to embody meaning and nothing more. Whether this is derived primarily from the lexical input of a construction or any semantic memory associations with a specific syntactic frame (as would be apparent from Goldberg’s early work) is less of a problem than the need to establish semantic associations between forms of the standard dialect and meanings of the newly‑­ created replica language itself. Thus, there is a plausibility factor in the selection of form to embody meaning across contact, and it is upheld by the fact that spea‑ kers of the replica language, at the best of times, attempt to maintain semantic continuity between the contact languages they are replicating in various ways, explaining form in function. In a contact situation, the replica language speaker’s efforts to maintain semantic continuity may be part of what is involved in the process of convergence, according to Matras and Sakel (2007: 835), and the com‑ promise between structural retention and structural accommodation. Form is rarely selected arbitrarily to convey meaning or function, especially in contact. In the present volume, it will be seen that semantic continuity across contact is maintained by the transparency of polysemies in the standard contact language

8 

 Chapter 1 Introduction

identified by speakers in attempting to replicate not necessarily only the f­ unctions of a ­substrate model language, but also those of the standard contact language itself. These strategies will be discussed at length in Chapter 8. A more general, underlying assumption of the present volume is therefore that of an essentially pragmatically‑­based, functional approach to the theoreti‑ cal development of the model, rather than a syntactically‑­based one. In terms of the theory of merger constructions to be developed, and the foregoing prin‑ ciple of semantic continuity, this is hardly surprising. The underlying postulate that speakers use language to convey meaning is at the heart of most modern, cognitive‑­functional theory, including constructional approaches to describing language. The understanding of a coalescence of form and meaning in the unifi‑ cation grammars such as construction grammars is not new; it was presupposed in most grammaticalisation studies of the 1980s and 1990s also. However, it is treated in various ways depending on the underlying theoretical persuasions of the protagonists presenting it, and the presuppositions of a functional‑­pragmatic approach to the treatment of construction grammars should not be taken for granted. Many construction‑­based approaches attribute a priority to syntactic and morphosyntactic representations (e.g., Fried 2008); others, e.g., Trousdale and Traugott (2013) prioritise pragmatic‑­functional explanations in adapting con‑ struction theory to grammaticalisation theory. However, for the moment, suffice it to suggest that constructions are equally the product of change and variation which has been affected by pragmatic forces; for example, the metonymic accom‑ modations and metaphorical analogies in the argument‑­structure constructions of Goldberg (1995: Chapter 3), which were seen as connected by devices known as Inheritance Links, as in: (3) a. Sam kicked the ball over the fence. and b. John gave Mary an apple. in which (3a) represents a caused‑­motion construction, and (3b) a ditransitive construction. The Inheritance Links between the constructions involve a meta‑ phorical pragmatic device, in which a meaning of transfer links the two const‑ ruction types to one another, and to other construction types in the same family, such as resultatives, where the transfer goal of a place (in caused‑­motion const‑ ructions) is replaced or substituted by that of a person (in ditransitives) or a state (in resultatives). In the same way, metonymic forces allow for the infusion of new arguments into the same construction type, e.g.: (4) Sam sneezed the napkin off the table. is metonymically aligned with the caused‑­motion schema in (3a), in which the meanings of causative transfer of an object to a place have been extended in (3a)

1.2 Overview 

 9

to include verb meanings of unintentional, forceful propulsion in (4). Kövec‑ ses and Radden (1998: 39) provide their own definition of metonymy as “… a ­cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same domain, or ICM.” Thus in (3a), the basic conceptual sense of a caused‑­motion transference, which is schematic to the lexical meaning of kicking an object to another position, is used to provide access to other lexical meanings of caused transfer of objects, unintentional or otherwise, as in (4). In the postulation of a merger construction in Chapter 8, the same metonymic and metaphoric forces are often perceived to associate one construction form with a different construction function in another dialect of the same language. In this way, it will be seen that the possibility of two different dialects of contact (rather than one and the same dialect sharing the same construction inventory) requires a closer review, and the notion of an arbit‑ rary mixing of construction types across contact is highly questionable.

1.2 Overview The present volume is a collection of earlier studies selected for the fact that they present difficulties of analysis from the point of view of traditional contact ­theories, as outlined in Chapter 8, in that they represent neither borrowed or replicated forms from the point of view of phonological and syntactic contact transfer. Each of the studies has been published previously, but the ideas are built upon in Chapter 8, and expanded beyond the limited analyses of the previous studies. In the present volume, they have been analysed from the point of view of a construction‑­based approach, by means of which the construction inventory of a contact dialect situation is seen to span over two or more genetically distant sub‑­dialects of the same language. In this way, the studies are pertinent also to the development of construction theories, in that they allow for the inclusion of features of contact which are not necessarily transferred to become part of the construction inventory of the contact dialect. Thus the volume represents, in effect, an extended anthology of case studies. Chapter 2 introduces the situation of Singapore English today, comparing a number of approaches to the sociolinguistic situation, including the former contact continuum approach advocated by Platt (1977) and Ho and Platt (1993). This approach has been superseded by the diglossic approaches of more recent times, such as Gupta (1994) and Alsagoff (2010). The main questions raised are whether the original substrate languages still have as much influence on the linguistic situation in Singapore as they once did, given the powerful politically‑­dominated motivations for streamlining the number of languages in current usage on the

10 

 Chapter 1 Introduction

island today. The emergence of Singlish as a contact language is thus subjected to more stigmatisation than in its colonial past, in the promoting of standard Manda‑ rin as the preferred, and only Chinese language to be used alongside English, and the local language policy which determines an obligatory lingua franca depending on one’s ethnic background and the promotion of standard English. This is bound to affect the range of distinctive grammatical features in Singlish which provide the base for a constructicon or construction inventory, but the representation of a typical creole context in this respect is not an accurate description in the thriving multilingual and multicultural community of Singapore. Chapter 3 raises some of the most searching questions regarding the theoreti‑ cal approaches to constructions and constructions grammars, among them, what is the standard nomenclature for a construction, and how to identify them? In many descriptions, a construction type is just a form‑­meaning pairing, but such a genera‑ lised definition is not without its problems, when the definition of meaning is also somewhat obscure in many accounts. The labelling of a construction is sometimes made on account of a syntactic feature, sometimes a functional one, and sometimes a relational or even a lexical one. The need to distinguish what is not a ­construction becomes increasingly more evident also, and it is concluded that a construction coincides with a constituent ‘chunk’ in most accounts, since n ­ on‑­compositional meaning cannot be explained by any other means. Other problems for ­construction grammarians are discussed in Chapter 3, such as the meaning of construction meaning itself, and why this is sometimes confused with construction function, and whether or not it is the lexical components of a construction which are ultimately the source of a construction meaning. The question of mutability of constructions is also approached, and the constructionalisation accounts of more recent research (such as Traugott and Trousdale 2013) are also reviewed. The first case study in Chapter 4 (published with Sarah Lee (2006), as ‘­Causativity reduction in Singaporean English’, English World Wide 27/3: 265–294), deals with the relationship between a simple transitive construction and what is known as a conventionalised scenario in Goldberg (1995). As with all the studies, the form of a single construction is shared across what appear to be two distinct construc‑ tion types. Examples such as I’ll pull my tooth (i.e., at the dentist’s) were found to be in frequent use in Ziegeler and Lee (2006), with no interference from (albeit implausible) ambiguities; i.e., the surface structure is represented as a transitive construction, but the semantics are not, being described in terms of Goldberg’s (1995) conventionalized scenarios, causative constructions which take the form of a transitive construction but with the intervening cause ignored. The chapter studies the frequency of the conventionalised scenario in SCE, comparing the frequency with that of British English speakers, and it is found that SCE spea‑ kers use conventionalised scenarios up to four times more frequently than British

1.2 Overview 

 11

English speakers. The reasons suggested for such frequency are the tendency for topic‑­prominence in all syntactic structures, a factor which is not independent of substrate influence, but which has an effect on the diachronic development of subject‑­predicate relations and the concomitent relations of agentivity between the subject and the main verb in SCE. Such relations are most likely overlooked in a dialect in which the development of the subject role is relatively under‑­ grammaticalised (see, e.g., Ziegeler 2008a). The reduction of implicit causativity usually associated with conventionalised scenarios (as in You cut your hair, when in fact the hairdresser was the intervening causee) is taken way beyond its limited potential in standard varieties of English to enable the conventionalised scenario construction to extend to almost any causative act. The chapter also investigates the constraints on the use of such constructions in standard varieties, revealing the construction to be a form of reflexive causative with a more restricted genera‑ lisation than is found in SCE. Chapter 5 discusses a quite different type of construction which has been labelled the ever construction, for the purposes of the present study. The study was published as Ziegeler (2011) in Groninger Arbeiten zur germanistischen Lin‑ guistik 53.2, as a shorter and less detailed version. The use of the adverb ever to express experiential aspect in SCE is well‑­known, and is believed to have been derived from the adverb never, interpreted as ‘not once’, as a back‑­formation, in a number of previous studies (e.g., Ho and Wong 2001). The study in Chapter 5 reviews the previous studies and questions the legitimacy of such an a ­ nalysis given the fact that negatively‑­polarised ever is found alongside experiential ever in the same dialect, as well as the fact that the adverb once can be found in ­non‑­redundant co‑­occurrence with it. The findings reveal then, that the use of ever may accommodate both minimiser meanings of ‘(at least) one possible time’ as well as experiential aspectual meanings of ‘(at least) one necessary time’ simultaneously. The case for substrate or model language influence is ruled out in the selection of this adverb to serve the functions of experiential aspect, though the motivation for expressing experiential aspect in Singlish is influenced by the substrate languages, particularly the Chinese dialects. As well as constructions expressing aspect, the volume discusses the some‑ what unusual use of the past tense construction in SCE in Chapter 6, to apply to situations which in standard varieties would normally use a form of the simple present tense, as in habitual aspect, performatives, and present tense anterior subordinate clauses, e.g., (155 in Chapter 6): (5) I am the kind who can tell myself I will do well even if I didn’t study … The possibility of substrate influence is considered to be of little relevance, since studies such as Ho and Platt (1993) have discussed the use of past tense in SCE as

12 

 Chapter 1 Introduction

restricted to applying to punctual and perfective verb types. The frequent use of the past tense in present habitual situations seems to contradict the hypothesis that grammatically imperfective aspect is usually left unmarked for past, espe‑ cially since the past habitual is usually expressed using present tense forms in SCE, as discussed in Ho and Platt (1993) also. However, a survey of the expression of past habituals in other languages reveals that there is frequent ambivalence regarding tense marking in this aspectual category, since it is a hybrid category expressing imperfective grammatical aspect but using verbs which are lexically perfective aspectually (as was previously pointed out by Givón 1994). Thus, there is also the possibility that the use of past tense for present habitual aspect could be reflecting a need to mark the verb as lexically perfective or punctual, regard‑ less of the grammatical aspect. However, such an explanation does not account for all of the data, and there are also examples in which past tense marks an anterior clause, regardless of its time reference (it may express an event which has not even taken place yet, as in (5)). Other examples include the use of past tense to mark a performative expres‑ sion, e.g., (178 in Chapter 6): (6) Ya.. I agreed. And recently it isn’t easy to get a job The study in Chapter 6 was first published as Ziegeler (2012a) as “On the interaction of past tense and potentiality in Singaporean Colloquial English’’, in Language Sci‑ ences 34: 229–251, in a special issue discussing the realis‑­irrealis ­interface, using data from an empirical study funded by the National University of ­Singapore Staff Support Fund (2008). Many of the data were derived from a new corpus which was created for this project using materials from online social networks in Sin‑ gapore (the Flowerpod Corpus). The results of the empirical study revealed that the different, ­non‑­standard functions of the past tense construction all had one thing in common, and that was that they tended to resemble the functions of the Old English and Germanic preterite‑­present verb category, in expressing the completion of a past event with an implicit result‑­state. The function is described in Chapter 6 as expressing precedence, or the perfectivity of an assumed anterior event which has implications for a potential future or present situation. Chapter 7 is based on a study which was first published as Ziegeler (2010) in the journal Constructions and Frames, initially with the aim of using data from contact dialects such as SCE to question the hypothesis of Michaelis (2003, 2004) of coercion between count and mass noun categories (some ideas from this study also appear in Ch. 3). In dialects such as SCE, there are no count or mass noun distinctions, and yet the same Bare Noun Constructions (BNCs) occur as those which have been type‑­shifted from count to mass status in standard varieties. In other words, the rule of maximality of reference or mass noun status applied to

1.2 Overview 

 13

arguments immediately adjacent to the verb phrase does not necessarily apply in the case of non‑­standard dialects of English. The BNC in Singlish is not a mass noun but a non‑­specific noun, and as such is unmarked for countability. The marking of non‑­specific nouns as bare noun constructions is not unusual in creole situations, but only partially reflects the system of the substrate languages, especially the Chinese dialects. Because of this, it was felt that the hypothesis of coercion between categories needed to be revised with respect to a more uni‑ versal approach to the cross‑­linguistic marking of specific vs. non‑­specific noun categories instead. The chapter deals with the problem of coercion by positing a framework for the integration of mass noun semantics within the perspective of the SCE system of using the bare nouns to mark non‑­specific reference. The data were taken from a range of sources, including commercial advertising flyers coll‑ ected over a period of about 10 years, as well as the Flowerpod Corpus. The Bare Noun Construction is discussed in Chapter 8 as using the form of a mass noun to express the function of a non‑­specific noun, selected (as with the other cases) under analogical identification of semantic resemblances. Chapter 8 reviews some of the most important literature on theories of contact to date, and arrives at the conclusion that few elements of such theories are useful in explaining the problems of contact presented in the preceding case studies. The chapter also discusses two recent theories of contact constructions, (Pietsch 2010 and Höder 2014) which aim to propose a merger construction of a different kind, most typically representing a merger of the functions of a model language with the contact language functions; in other words, a means of per‑ petuating the substrate model language functions in the form of a new contact language. The present theory of merger constructions takes the previous research one step further in accounting for the contact speakers’ identification of replica language form as an exponent of possible functions restructured from the sub‑ strate model language, with the proviso that in many cases (with the exception perhaps of the experiential perfect discussed in Chapter 5), it is not always clear which functions are being replicated. The possibility of innovations in contact is also discussed, as is the arbritary selection of replica material in the establish‑ ment of equivalence across contact (Heine 2012). Thus it is not always syntactic or functional replication which is the target of replica language creation. The resulting hypothesis of a Merger Construction is therefore one which is based on perceived analogical comparisons and matchings created by speakers of the replica language, and reanalyses of the meanings of the contact standard language material to account for the specific needs of the replica language. The Merger Construction thus contains not just a form‑­meaning pairing, but a three‑­ part, form‑­function‑­meaning coalescence, in which only the functions to which the constructions are applied differentiates them from those of the local standard

14 

 Chapter 1 Introduction

language. The merger of the construction types across contact is achieved by the search for polysemic meanings in the meaning component of a ­construction pairing, which may be considered to be shared across both the replica language and the contact standard language.1 Such polysemies may be found at any sche‑ matic level, so that even the more generalised construction types such as the transitive construction are able to provide the semantics upon which a merger can be created (in this case, the schematic semantics of causativity). The level of generalisation of the semantic component of the construction is thus dependent on the type of construction that is involved in the merger. The volume aims to present a new perspective on contact conditions and the capacity for contact linguistics to be aligned with modern, cognitive theory. The data are taken from a number of studies which, as noted above, were left partly unresolved in previous publications, since they could not be compared with most studies of contact which dealt with the representation of syntactic or grammatical restructuring using the form of an imposed lexifier language and the a ­ ttribution of all syntactic structure to the substrate alone (the matter of lexical borrowing need not be involved in the present research). The four case studies are all selec‑ ted for the fact that they can only be described with reference to a merger of con‑ structions types across contact; and in this way, they aim to contribute to the development of construction‑­based theory as much as construction‑­based theory may contribute to the explanations required for the case studies. The result is a merger construction hypothesis, a compromise construction accounting for the form of the contact standard variety, the meanings of the local vernacular, Sin‑ glish, and the functions of other local languages of contact or substrates. It will be seen in the subsequent chapters how such a hypothesis operates to serve the aims of the present volume.

1 The search for polysemies across contact was also noted by Nau (1995) as mentioned in Matras and Sakel (2007), and discussed in Ziegeler (2014a) as partly explaining the recaptulative proces‑ ses of some types of contact grammaticalisation.

Chapter 2  Singapore English 2.1 Introduction: Historical background The introduction of English into Singapore, according to Gupta (1998), did not begin with the foundation of the colony, as English traders had visited the island long before the establishment of the British colony in 1819 by Sir Stamford Raffles. A full historical account of the demographic development of Singapore is summa‑ rised in the introductory chapters of Leimgruber (2009), amongst others, though earlier studies such as Gupta (1998) had focused specifically on the development of the linguistic profile of Singapore. In the 21st century, we find an independent republic, populated largely by Chinese descended originally from the southern provinces of mainland China (referred to locally now as PRC – Peoples’ Republic of China) and speaking a range of languages known either as Chinese dialects (e.g., Ho and Platt 1993; Bao 2001) or Sinitic languages (Chappell (ed.) 2001, as noted in Matthews 2010) depending on the author’s perspective on their typolo‑ gical identity. According to Ho and Platt (1993), around 10 Chinese dialects were originally spoken in the region, the most significant being Hokkien (Southern Min), Cantonese (Wu), Teochew (or Chaozhou, also a variant of Southern Min and mutually intelligible with it according to Lim (2007: 454)), with others such as Hakka (Kejia) playing an important role as well. The language of the original inhabitants of the island was Malay, which still remains the national language of the republic today. Later, languages from southern India and Sri Lanka, such as Tamil, were also introduced with the immigration of south Asians, creating a linguistic melting pot in a vigorous trading community, the ideal situation for the development of lingua francas and creoles and a vibrant context for the emer‑ gence of contact languages. Gupta (1998) refers to the explosion of the population in the 19th century; Leimgruber (2009: 3) notes that it rose from 11,000 in 1824 to 81,000 in 1860. In the beginnings of colonisation, the majority had been Malays, but by 1867 the Chinese represented 65%. Today, the population of Chinese is closer to 76.9% estimated by Ho and Platt (1993), with around 14.6% Malays, and 6.4% Indians.2 In spite of this, the number of British at the time, according to Leimgruber, was as low as 300 (mainly civil servants in administrative roles).

2 It is likely that the total Chinese population has been increased in recent years by the government’s policy to encourage migrants from mainland China, though the numbers of ­original Singaporean Chinese may have remained stable. Leimgruber (2009) estimates the ­balance of ethnicities in today’s population to be approximately the same as Ho and Platt’s (1993) figures.

16 

 Chapter 2 Singapore English

The present‑­day situation of Singapore offers equality to all ethnic groups in that it accommodates four official languages in a population of no more than four million inhabitants.3 Media broadcasts and public announcements may be heard in either English, Mandarin Chinese, Malay, and Tamil, with all four being used consecutively in public domains such as in the metropolitan rail transport system. Speakers are encouraged to use English and what is known locally as the ‘mother tongue’, or the official language of one’s ethnic group, i.e., Mandarin Chinese for the Chinese, Malay for the Malays, and Tamil for the Indian popula‑ tion. In this way, a policy of active bilingualism has long been promoted by the ruling government party which has remained in power ever since independence in 1965. The perpetuity of a dominant, originally visionary form of government has enabled, over more than 40 years, a certain amount of linguistic, as well as administrative and economic stability in the region. According to Leimgruber (2009: 9), all legislation and the majority of administrative processes use English, which is classed as a lingua franca in the region and a language which is neutral to ethnic divisions. The standard English used administratively, though, varies greatly from the colloquial sub‑­variety known informally as Singlish, and, accor‑ ding to Gupta (1992a, 1994), it varies little from any other standard variety else‑ where in the world, as noted in Ch. 1. In spite of the 19th century situation of intense contact amongst a large number of languages, and the existence of a lively trading community, it has been argued by Gupta (1998) that there was no previous evidence of a pidgin English in the early Straits Settlements which also included the island of Penang and the town of Malacca (Malaysia). The 19th century lingua franca was Bazaar Malay, a Malay‑­based pidgin, and Bao (2001) cites Denny (1878a) as offering a valuable description of this language. However, Bao (2001: 285) also claims that a pidginised form of English (Chinese pidgin English) was likely to have preceded the use of English in Singapore today, citing a number of travellers’ sources such as Thompson (1864) and Dennys (1878b) as suggesting that it replaced Bazaar Malay, having spread to Malaya and Singapore from the south China coast after the Taiping Rebellion of 1850 in China. Lim (2007: 456) claims that Bazaar Malay was not replaced by English as the lingua franca until the late 1970s and early 1980s; however, she is referring only to the use of an official standard form. Bao (2001) describes the colloquial sub‑­variety of English used in Singapore as an ­endogenised pidgin‑­creole, citing a term used by Chaudenson (1977) to ­distinguish creoles emerging out of commerce or trading situations from those typically developing from isolated slavery and agricultural settings. However

3 3,844.8 million residents, according to latest figures (www.singstat.gov.sg, accessed 23/11/2013).



2.1 Introduction: Historical background 

 17

obvious it must be that a lingua franca was in existence at the time in Singapore, it is likely that the main local language spoken on the island before colonisa‑ tion was Malay, and therefore Malay would have formed the base of any future pidgin‑­creole. Thus, it is unlikely that a language such as China Pidgin English would have contributed to the main structural development of Singlish today, though it is possible that, given the likely similarities in substrate and adstrate languages, it could have served a contact role.4 Platt (1989: 397), who termed Sin‑ gapore English a ‘creoloid’ rather than a creole, emphasised that the difference between an indigenized English and an English‑­based pidgin or creole was in the fact that an indigenised English was used originally as a medium of education. At the same time, he did find some features of the local variety which resembled the features of creoles discussed by Bickerton (1981) (see below). However, it was not the medium of education that created the indigenised variety familiarly known as Singlish today, but the need for inter‑­ethnic communication amongst school‑­ children outside the classroom setting that promoted the colloquial sub‑­variety. This point is also stressed by Gupta (1991) in her study of the emergence of a diglossic situation in Singapore English (see below). It might be argued that the researchers who stress the importance of a dis‑ tinctive difference of (educational) setting for the emergence of the more collo‑ quial sub‑­variety are also those who over the years may have been instrumental in attempting to promote the teaching of a standard variety, within the classroom, and therefore wish to associate the emergence of a non‑­standard sub‑­variety with factors beyond classroom control. Bao’s (2001: 285) claims for a pre‑­Singlish form of pidgin are at least partially convincing as well, given that as he claims, mass education in English medium did not formally become institutionalised until after independence in 1965, and before that time, only a small minority of rich mer‑ chants could afford to send their children to English‑­based schools. This leaves a very short period of time for the establishment and indigenisation of an English variety with the unique grammatical profile that Singlish has, as pointed out by Ansaldo (2010), combining features of Chinese and Malay into a single contact language. In view of the powerful motivation of trade and commerce before that time in Singapore, some sort of English‑­based pidgin must have been inevita‑ ble; in fact Bao (2001: 285) cites historical records dating back to 1897 in which the non‑­standard use of English in the Straits Settlements was deplored as much as it is today. It is possible then, that the use of Singlish in the English‑­medium schools was already incipient by the time of the introduction of mass education;

4 A comprehensive history of China Pidgin English can be found in studies such as Bolton (2002). He does not, however, mention its spread from south China to Malaya.

18 

 Chapter 2 Singapore English

perhaps it is only then that it became more noticed as a distinct sub‑­variety in its own right, and worthy of description. Gupta (1998) devotes a large amount of description to the history of English‑­ medium education in Singapore, also stressing that the teachers of the variety of English which was transmitted through the educational system prior to inde‑ pendence were not always speakers of standard varieties of English, for a start, and furthermore, were often non‑­native speakers themselves, some having little knowledge of the language and few credentials for teaching it apart from being seen as European (not necessarily British). She notes (citing Ho 1964) that histo‑ rical records of 1884 list, amongst the names of 31 staff in government schools, only eight European names. Other names appear to be Indian or Chinese, with 18 described as ‘‘probably Eurasian’’, of which seven have Portuguese names. There were also Catholic missionary teachers from the United States, as well as others from the UK and Australia, Ireland, France and Belgium. (Davydova (2013) also noted the influence of Irish Catholic missionaries in the 19th century colonial period, which she claimed to have had an effect on the transmission of the English perfect aspectual system in Singapore Colloquial English.) Gupta (1998) outlines two phases of educational transmission of English in Singapore: a stable period in the 19th century when the majority of children educated in the English medium were from English‑­or Malay‑­speaking homes, and a later period at the turn of the 20th century when Chinese‑­speaking children started entering English‑­medium schools. She believes it was in the first two decades of the 20th century, when children from mixed language backgrounds started to enter English medium edu‑ cation that the first signs of a colloquial sub‑­variety started to appear, the situa‑ tion being enhanced in particular by the education of girls who would be likely to give rise to a new generation of native speakers in later years (the assumption that females are likely to perpetuate the language, as mothers, of course, need not be always the case; in many homes, the father’s native language may have become the household lingua franca as well). Newbrook (1987: 10) noted that in 1987 that Tamil‑­medium education and Malay‑­medium education was in the stages of being phased out in favour of uni‑ versal English‑­medium education, with Chinese‑­medium education also likely to disappear, and this has been the situation ever since, (though Chinese‑­medium education in selected schools has been making a comeback since as early as 1991 (Gupta 1991: 13), probably due to the focus on economic values underlying lan‑ guage policy gradually shifting to mainland China). Newbrook estimated in 1987 that nearly 100% of Singaporean teenagers knew English, though proficiency in skills such as writing fell short of what he described as native standards. He also noted the appearance of nativised and regular patterns of usage replacing what were originally held to be errors, and that distinctions of ethnicity were no longer



2.1 Introduction: Historical background 

 19

apparent in spoken English usage (1987: 11). Gupta (1998) also referred to what had been previously described as errors as simply dialectal features. At the same time as the English language was reaching maturity in Singapore, western values associated with English were generally considered to be incompatible with Asian values apart from ‘‘ ... superficial areas such as pop consumer culture’’, according to Newbrook (1987: 12). The emerging situation he was observing in 1987 was the development of a ‘new’ English, according to Platt, Weber and Ho (1984), one which by the 1980s would see the then current school intake of children already having acquired English in the home as about 20–30% (Gupta 1992a: 324). Today, this figure is likely to have increased significantly; in fact Lim (2007: 456) and Lim and Foley (2004: 6) claim for around 70% of children entering school for the first time in 1990 with English as a dominant language. Today, the figures are likely to be much higher.5 Lim (2007) proposed a series of overlapping stages of linguistic ­development in Singapore labelled Ages, similar to the stages of development of i­ nternational varie‑ ties of English on a global scale hypothesised by Schneider (2003). Age 1 referred to the development of substrates from pre‑­colonial times to ­post‑­independence, Age 2 referrred to the development of official languages from the mid‑­1970s to the present day, and Age 3 proposed an age of global media languages from the 1980s to the present day. Age 4 predicted the return of emphasis on local languages in order for enhanced regional communication (1990s onwards). In this way, Lim presents the life histories of the languages which have been considered as domi‑ nant in Singapore at various times in the past, some perceived as replacing one another along a time line, others becoming obsolete (e.g., Bazaar Malay). She notes in particular (2007: 453–4) the role of the Babas in the history of English in Singapore, a race of people sometimes known as Straits‑­born Chinese, who were descendants of the 19th century Chinese immigrants having ­inter‑­married with local Malay women. The Babas created a new and exotic culture and a mixed language with elements of both Hokkien Chinese and Malay integrated, and this language in turn served as an influential substrate in the region as well. The role of the Babas was significant in the developing society at the time; they were a powerful economic and social elite group, also considered to be the best educated and holding a high regard for English‑­medium education and the use

5 Leimgruber (2012: 3) shows statistics of frequency of use of English in the home to be only 32% (with 36% for Mandarin Chinese, 14% ‘other Chinese’, 12% Malay, and 3% for Tamil). The statis‑ tics are from Wong (ed.)’s (2011) census data, and thus reflective of self‑­reporting. Nevertheless, they reveal an almost equal use of English and Mandarin, something which has long been the objective of government bilingual policies for the majority Chinese population.

20 

 Chapter 2 Singapore English

of English in the home. Their command of English led to their ability to occupy ­prominent official posts in administration, and to their often being referred to as the King’s Chinese (see also Gupta 1998). Gupta (1998) refers to the Straits Chinese as the first Chinese group to educate their daughters in English‑­medium schools, and names them, along with the Eurasians, the Singapore Jews, and the local Armenians, as being instrumental in promoting the use of English in trading situations in the mid‑­19th century. The historical scene, then, as described by numerous accounts, was pre‑ sented as one of a rich intermixing of cultures and languages. Many of these have survived into the 21st century, though some are rapidly disappearing, due mainly to the careful engineering of local government language planning policy. Thus, while English was instituted as a dominant lingua franca, the public use of various Singapore regional Chinese dialects, such as Hokkien and Cantonese, has become suppressed, especially with the aid of the 1979 Speak Mandarin ­Campaign. However, it cannot be denied that, in place of the lost cultures and languages, there has arisen a new, more united cultural profile and national iden‑ tity, epitomised in the growth of the internationally‑­famous sub‑­variety, Singlish, or Singapore Colloquial English. The sociolinguistic classification of English in Singapore cannot avoid reference to its presence, and this will be discussed in the following section.

2.2 The sociolinguistics of Singapore English 2.2.1 Earlier studies As noted above, amongst the earliest research on Singapore English were some accounts which attempted to describe it with the use of standard varieties as a contrastive benchmark, highlighting the notion of errors (e.g., Crewe 1984; Tongue 1979). At the same time, however, researchers such as Platt (1977) were noticing that the typical speaker of Singaporean English at the time, although deviating from a prescriptive standard found in other varieties, had “a native‑­like fluency in his speech variety” (Platt 1977: 83), and recognising that an ­educated sub‑­variety was also beginning to emerge (1977: 84). This pattern represented the beginnings of nativisation as we see it in Singapore English today, where any inconsistency amongst speakers in the structural features of Singapore English is no longer immediately obvious. At the time, though, the changes observed by Platt (1977) needed a taxonomic description, as it was clear that with the emer‑ ging lingua franca, a rule‑­based grammatical and phonological system was also starting to become apparent. Platt and his colleagues had resorted to the analogy



2.2 The sociolinguistics of Singapore English 

 21

with a post‑­creole situation proposed by DeCamp (1971) for Jamaican Creole, or Bickerton (1975) for Guyanese Creole, in which the decreolising situation could be described in terms of movement between the basilect, or most informal sub‑­ variety, the mesolect, or middle‑­range sub‑­variety, and the acrolect, being the most educated sub‑­variety and one that is closest to the standard (Platt 1977: 84). Platt did not intend to class Singapore English technically as a creole, though, as noted above, since it was introduced mainly through the education system, a situation which is not typical for creole development. Instead he labelled it (as noted earlier) as a creoloid.6 However, his classifications were definitive of three separate sub‑­varieties, and were accompanied by distinguishing characteristics at each level; for example: the acrolect contained only phonological features dis‑ tinguishing it from standard English varieties, while the mesolect contained not only ­phonological distinctions, but also grammatical ones such as absence of verbal and nominal inflections, and variable realisation of the copula verb. The basilect, at the lowest level, contained more of the features of the mesolect rea‑ lised with higher frequency, while other characteristics such as the replacement of a tense system by an aspectual system were definitive of the closer contact of this sub‑­variety with the Chinese substrate languages, as indicated by Bao (2001) much later. Platt (1977) supported his predictions with statistical data from typical spea‑ kers of each sub‑­variety, e.g., a mechanical engineer for the acrolect, a secretary for the mesolect, and a waitress for the basilect, each differentiated also by their highest level of education attained. The study was somewhat sketchily descri‑ bed, but it yielded the hypothesis that Singapore English was represented as a continuum, marked not only by sociolinguistic divisions, but also by levels of formality in the situation; i.e., Platt introduced the possibility of style‑­shifting into the continuum‑­based approach. The style‑­shifting entailed that the speaker switched between the three sub‑­varieties on the basis of stylistic needs as well as proficiency, so that the sociolects represented degrees of formality alongside the educational level of the speaker. At the highest level, the acrolectal level, the speaker would have the widest range of styles available from which to choose,

6 Ho and Platt (1993: 1) defined ‘creoloid’ on the basis of Platt (1978: 55): “A speech variety which has developed through the educational system such that a non‑­native or introduced prestige variety is taught to speakers of another speech variety (or other speech varieties) in a situa‑ tion where the introduced variety comes to be used in everyday situations, to be acquired by some children before they commence school and to become a virtual ‘native’ speech variety for some or all speakers.” The sociolectal distinctions in this definition are not totally transparent. ­Ansaldo (2009: 113), however, cites Platt’s (1975) definition of a creoloid as a creole without a pidgin ancestor.

22 

 Chapter 2 Singapore English

e.g., a business executive whose highest sociolect was the acrolect would pro‑ bably use this for a formal occasion such as an interview, but may switch to a mesolect variant while having lunch. A more relaxed, informal occasion (such as being caught in a rain‑­shower) would invoke the use of the basilect by the same speaker, or some basilectal features (Platt 1977: 92), indicating also that it was not only the situation, but also the subject matter of the speech event that determined which sociolect was to be used. The speaker whose highest range was the meso‑ lect would only have recourse to this and also the basilect in shifting between formal and informal use, while the speaker whose highest range was the basilect would use it for both formal as well as informal situations. Leimgruber (2009) summarises the sociolectal continuum model suggesting that although it is still quite representative of the Singaporean speech situation, there are problems in that many English‑­educated speakers can no longer fully comprehend the basilect (2009: 31), and he cites a study by Hussain (2006) to testify to this. It is possible that 40 years after Platt’s original studies, the quasi‑­ post‑­creole, sociolectal continuum has moved away from the more basilectal, creoloid end which was the object of Platt’s earlier observations. There are other problems with a model such as Platt’s, amongst them the difficulty in separating the different lects on the basis of a limited inventory of identifying features, which is rarely stable and constantly open to new members in any case, as pointed out by Alsagoff (2010: 338) discussing Gupta’s (1991, 1992a, 1994) classifications (see below). It must also be remembered that in a contact situation, it is not only contact with the substrate that is relevant, but also contact between the various sub‑­varieties of the contact language itself. The problems of accurately isolating the different sociolects at any one time must indicate a model too constrained to be operable. In the 1990s, the research accounts tended towards reducing the number of sociolects available to the speakers to remain at two, and the creation of a diglos‑ sic hypothesis for the classification of Singapore English. The diglossic propo‑ sal was at first initiated by Gupta (1991), who adopted Platt’s earlier (1977: 91) mention of the basilect as a colloquial sub‑­variety, and used the term Singapore Colloquial English (SCE) to refer politely (and more technically) to what had gene‑ rally become known as Singlish (even by linguists researching in it, as Leimgru‑ ber (2012) observes). It was this sub‑­variety which became the L(ow) sub‑­variety of the diglossic situation (modelled on Ferguson 1959), with Standard Singapore English (or just Standard English, as Gupta (1991: 9) referred to it, as the H(igh) sub‑­variety. The two diglossic poles were discussed as functioning with the same stylistic determinism of Platt’s continuum, with factors such as the forma‑ lity of the occasion sometimes being influenced by the educational level of the ­interlocutor as well as the speech setting.



2.2 The sociolinguistics of Singapore English 

 23

Gupta (1991: 10) saw SCE as a form which differed little from a creole, though, as with Platt (1989) and others before her, she preferred to avoid the use of the term creole. Instead she proposed that it was a contact variety with a disconti‑ nuity in its transmission, one that is nevertheless spoken natively. In this way she was preferring to adopt the label of a mixed language, after Thomason and Kaufman (1988: 3), i.e., a language which has not followed a common genetic line in its evolution. It is understood that her use of the term mixed language is probably much more general than, for example, Bakker (2000), who describes such mixing as intertwining, where the two languages concerned consistently combine the lexicon of one language and the syntax of another (2000: 596), e.g., Media Lengua. However, for many features of SCE, it has been claimed that they bear the lexicon of English and the syntax of Chinese. Platt (1989: 397) questioned the following SCE expression as being English, suggesting that the words (apart from the discourse particle, lah) were English but the syntax was Chinese: (7) Passenger depend lah, good one also got bad one also got which could be roughly translated as ‘‘it depends on the passengers, sometimes there are good ones, sometimes there are bad ones too”. He noted that the speaker was an English‑­medium educated Singaporean (taxi‑­driver) who had failed his primary‑­school leaving examination. The use of the term mixed language may also be at variance with a description of a variety in which there are not only dis‑ course particle borrowings, as shown in (7), but also lexical borrowings, as in the following example, taken from the Flowerpod Corpus: (8) eh... actually i very kiasu de. 1st i use my ettusais acne clear makeoff, then use a cotton pad to wipe off. then when i go bathe, i use the acne clear makeoff again, use water to rinse off. then i use my normal cleanser. Posted by: >ichigo< Oct 23 2005, 12:55 PM ‘Actually I’m (a) very careful (person). First I use my [product name] acne clear makeoff, then use a cotton pad to wipe it off. Then when I go to bathe, I use the acne clear makeoff again, and use water to rinse it off. Then I use my normal cleanser.’ In (8) the example illustrates the frequent use of the adjective kiasu which, according to Butler (1999: 197), means ‘afraid of losing out to someone else: anxious not to be disadvantaged’, and is borrowed from Hokkien meaning ‘afraid to lose’. In the context in (8) the meaning has been extended as it often is in present‑­day SCE, to mean simply ‘cautious’ or ‘careful’. Note also the use of the nominaliser de in (8), a borrowing from Mandarin; this is often substitu‑ ted with an English calque, one as discussed, for example, by Bao (2009). (In (8) the absence of the copula can also be noted, a feature frequently discussed in

24 

 Chapter 2 Singapore English

Singaporean English grammars; e.g., Alsagoff and Ho (1998), Ansaldo (2009).) Thus, it can be seen that in SCE there is no necessarily strict mixing of the lexicon of one language with the syntax of another; lexical items may derive from any of the substrates in borrowings or calques; thus there is material copying as well as pattern replication. However, it is often also the case, as will be seen in further chapters, that the substrate sources for some syntactic pattern replications are not always that transparent.

2.2.2 The situation today Later studies such as Alsagoff (2007, 2010) were to capture the diglossic model within a different framework, perhaps one more adapted to the vastness of the range of uses to which English is assigned in Singapore and internationally by ­Singaporean speakers. Alsagoff believed that neither proficiency (as in Platt’s (1975, 1977) model) nor register (as defined by the diglossic approach of Gupta (1991, 1992a, 1994) could account for the type of style shifting observed in ­Singaporean English (2010: 336).7 Alsagoff (2010: 337) also pointed out, quite ­justifiably, that Platt’s model was replaced by the diglossic model of Gupta (1991, 1992a, 1994) as it did not explain sufficiently why speakers whose normal lect was the acrolect would choose to also use the basilectal, non‑­standard sub‑­variety. The diglossic model, she notes, was used to enhance communicative goals by enabling the speaker to express solidarity and informality through the Low variety, rather than simply providing for less competent speakers (2010: 337). However, she also points to the problems of Gupta’s model (1994) in that, like the continuum approach advocated by Platt (1975, 1977), it cannot be used to isolate the boundaries of the structural features that define the two diglossic poles of Singlish and standard English. Numerous features appear in other studies which are not necessarily included in the original diglossic structural classifica‑ tions, in which a restricted sample of four features marks the colloquial pole (the Low variety): the occurrence of pragmatic particles, the presence of verb groups without subjects, the use of ‑­ing forms as finite or verbless complements, and the use of conditional constructions without subordinating conjunctions (Alsagoff

7 The term Singaporean English was used by Ho and Platt (1993) to distinguish the English spoken generally in Singapore from Singlish, which they referred to as Singapore English.



2.2 The sociolinguistics of Singapore English 

 25

2010: 338). For example, she notes that Gupta classifies the following sentence as SCE only because of the pragmatic particle, lah: (9) She’s given him the correct file, lah whereas the following: (10) She give him what? would fall outside the SCE category as it does not include any of the four crite‑ ria mentioned above, though the word order is reflective of the contact langu‑ ages as is the use of the uninflected verb form. Further problems with the use of such a model are that either it does not account for the incidence of lexical borrowing, which could reasonably mark an expression as Singlish or SCE, nor can it account for degrees of phonological approximation to a standard form of pronunciation. Another, more significant problem that was often addressed by researchers such as Newbrook (1987) was that it was not always known exactly what standard English really was. Gupta (1991) maintained that the standard Sin‑ gapore sub‑­variety varies little from any other standard. However, in fact, there may be numerous characteristics that mark it as distinctively local, though not non‑­standard, as seen in the following examples: (11) We noted that Singlish is a spoken language and the best way to study it is to examine its usage in speech, lest tone influences the usage of these terms (Singapore university students’ essay, 2007) (12) No stopping at all times (ACE. The Basic Driving Theory Test (2007, ­MultiNine Corporation Pte. Ltd.) The use of archaic lest in (11) (not followed by the subjunctive form of the verb) rather than unless may be quite possibly an adoption from current (popular) e­ cclesiastical usage (based on biblical sources); it has not been observed, to present knowledge, in any previous studies. The use of a universal quantifier rather than an existential one in the scope of the negative as in (12) is also found in Singapore English, and yet it might not be taken as non‑­standard, nor even as an endonormative standard, as it is not unknown in other dialects (see Newbrook 1997: 240). Other characteristics of local usage will be discussed below. At the same time, universal features of non‑­ standard urban (mainstream) dialects, such as like used in place of a conjunction or the redundant relative pronoun what in comparative clauses, may be starting to make their appearance amongst younger speakers: (13) She felt like her heart was on fire (14) ... are not always as fickle or temporary or unstable as what Hopper claims.

26 

 Chapter 2 Singapore English

both of which were observed personally, December 2007, in university ­students’ writing (another observation was the use of like as a discourse hedge, and object pronouns for subject pronouns in conjoined subject phrases, e.g., Me and X were V‑­ing ... etc. in spoken usage). The appearance of such urban non‑­standardisms are an indication of the maturity of the dialect, as they are not in any way related to the local contact substrate languages, but are familiar to other international (L1) English dialects as well.8 Alsagoff (2010) proposes instead of Gupta’s diglossic model a model of English usage in Singapore known as the Cultural Orientation Model (COM), which accom‑ modates English as both a global, international language and a local, inter‑­ethnic lingua franca. In other words, it has become ‘glocalised’, in that it e­ ncompasses both global and local functions at the same time. Singlish, according to this model, is not condemned as basilectal and non‑­standard but is recognised as serving an important cultural role in its contribution as a unifying lingua franca amongst multiple ethnic groups. Singlish characterises the soul of Singapore, and it is seen as a symbol of national identity to many Singaporeans (Alsagoff 2010: 342). According to Alsagoff (2010: 343): “The globalist orientation looks outwards at the world, and the localist looks within.” The 21st century has seen the growth of mass education in Singapore and the facility with which educated ­Singaporeans are able to communicate internationally has nothing to do with the retention of Singlish as a local, informal dialect. Rather than a diglossic local approach, Alsagoff has raised the possibility of a diglossic international approach, defined by speaker preferences and discriminations, not speaker competence. The speaker may also engage the use of standard English to express authority on a local level, while at the same time incorporating elements of Singlish to indicate adherence to the community (2010: 344). However, the COM model suggests that it is hardly likely that an educated speaker would consciously attempt to impose the use of the local sub‑­variety at an international level of communication, and necessarily expect to be understood. Leimgruber (2009, 2012) succinctly summarises some of the inadequacies of the former approaches to the classification of English in Singapore, pointing out first, that Platt’s (1975, 1977) model assumed proficiency as a guide to sociolec‑ tal usage, a hypothesis based on educational attainment rather than speaker’s choice. In rejecting the model, he also mentions (like Alsagoff 2010) its inability

8 In discussing the ‘maturity’ of a new variety of English, it is not intended here to enter into the debate outlined in Anchimbe (2009) which tends to equate maturity with a goal‑­oriented notion similar to decreolisation. It is considered, on the other hand, that the variety has simply reached a level of conventionalisation that may be different from that of other contact varieties.



2.2 The sociolinguistics of Singapore English 

 27

to accurately define at which point of the sociolectal continuum the utterance is located, when it may bear some characteristics of one lect and some of another; for instance, increase in the use of the copula verb marks a speaker of a higher educational background, but it is not easy to pinpoint exactly where the speaker’s most usual sociolect belongs (2012: 5). Thus, Platt’s model is problematic due mainly to questions of identification of the sociolects. He also criticises Gupta’s diglossic model on the basis of its lack of homogeneity, something which Gupta herself had recognised as ‘leaky diglossia’ at a later stage (2006). In many ways, a diglossic situation is incompatible with such a contact situation, as contact does not rule out contact between the two diglossic poles, and the ultimate melt‑­down of an infinite range of structural features characteristic of the many languages spoken in the region. Leimgruber commends the use of the COM of Alsagoff (2007, 2010), since it enables speakers to exercise not only shifting between styles, but also to assert cultural membership, and a certain economic status. Insofar as the latter factor is concerned, the development of such linguistic flexibility is also inadvertently a sign of a developing class structure in the local community. However, Leimgruber goes further than Alsagoff in building on her model to discuss instead an indexical approach to the use of English in Singapore. In using such an approach, he bases his studies on the hypotheses of Eckert (2008) in which certain sociolinguistic variables, such as the phonological variable of an (un) released /t/ in American English, were intended to convey indicators of social stance, emotion, status, or any other socially‑­prescribed meaning in the discourse. Leimgru‑ ber adapts the indexical approach to account for the situation of code‑­mixing of Sin‑ glish, Standard English, Mandarin, Malay, Hokkien, and any other local language in the same discourse, introducing a phenomenon which has been largely overlooked by past researchers focussing solely on Singapore English. Code‑­switching is rife in Singapore, and amongst the Chinese, the strict bilingual, Mandarin‑­English‑­mother‑­ tongue policy objectives of equivalent competence in both languages have brought with them the inevitable consequences of speakers frequently employing words or phrases, seemingly haphazardly, from either language in a single utterance. The fol‑ lowing example from Leimgruber (2012: 4) illustrates this well: (15) Yong tea wash it off lah. In (15) yong is Mandarin for ‘use’; the speaker has been advising the interlocutor to use the tea in front of her to wash the oil off the spoon. Leimgruber questions whether the speaker has also switched from standard English to Singlish in the same utterance, with the addition of lah as a discourse particle (see below, regar‑ ding discourse particles). However, the question is whether the remainder of the utterance was standard English or Singlish anyway. The serial verb word order is typical of Mandarin (or other dialects) at least, and there is no to‑­infinitive.

28 

 Chapter 2 Singapore English

The approach taken by Leimgruber is similar to Alsagoff’s COM in that it employs a theory of variable expressions of stance and assertion of social values; however, it was not clear in Alsagoff’s studies that she accounted for switching at the utte‑ rance level in this way. The more recent situation in Singapore demands a far more detailed account of the use of the contact languages and its c­ onsequences for the future of English in the region generally. In addition, there is a need for greater attention to the theory of code‑­switching in a situation of contact such as ­Singapore presents. It is unlikely that Singapore’s standard variety of English would be affec‑ ted by the presence of such changes. However, Singlish itself, without a doubt, appears to be changing rapidly under the influence of the bilingual policy, and examples like (15) are typical of a trend pitched ultimately towards a blend of two languages, Mandarin and Singlish. Since Mandarin was never a true substrate lan‑ guage in Singapore, it is thus questionable whether today’s Singlish is the same language that was observed in the works of Platt and his colleagues 30 years ago. Apart from a changing Singlish, in 21st century Singapore, we see a situation of continued diglossia, but a diglossia with non‑­mutually‑­exclusive functions in the society, and one that is flexible enough to allow the individual access to both codes in order to assert or define his or her status or affiliations within a given speech situation. The emergence of a local standard is already established, but SSE, though naturally intelligible on an international basis, is nevertheless cha‑ racterised by local features which mark it as unmistakenly Singaporean, quite apart from a distinctive phonological system, as demonstrated in Bao (2003). It is unlikely that any variety of English will have a neutral, standard sub‑­variety totally unmarked by localisms, grammatically or lexically. There are shared, general exonormative features in all standards (e.g., word order), but conventi‑ onalisation takes place at a local level as well as at a global level (consider, for example, spelling and lexical conventions in US English). However, this need not imply unintelligibility, simply variation. At the non‑­standard level, the ­sub‑­variety (SCE), now defined universally as Singlish, is accepted locally as a means of expressing identity, empathy and solidarity, and it would be ­unthinkable for use in international exchange. However, as a contact language, it provides a wealth of insights into language development and change.

2.3 Grammatical morphology and discourse features 2.3.1 Tense, aspect and modality In spite of the large number of studies in recent years attesting to the ­systematic grammatical autonomy of Singlish or SCE, Bao (2003) points out that the



2.3 Grammatical morphology and discourse features 

 29

vernacular variety spoken in Singapore cannot be analysed without reference to L1 varieties of English. He uses phonological data to illustrate that social stig‑ matisation actually forms a barrier to the conventionalisation of non‑­native ele‑ ments. In other studies, such as Alsagoff and Ho (1998), it has been claimed that levels of grammaticality in Singlish or SCE are independent of the levels of gram‑ maticality found in other varieties, and must inevitably be considered as such. Perhaps there is a difference between phonological and grammatical stabilisation in a new variety of English where this is concerned, and the local tolerance for non‑­standard accents may be more reduced than for non‑­standard grammatical usage, some of which is not always readily identified as being different from exo‑ normative uses in any case (e.g., Newbrook 1997 – as noted above). There is also a difficulty in classification, even in academic accounts, of what may be termed SCE and what may be classed simply as evidence of a local standard usage. Other problems arise in the case of features which are not always typically classed as standard, but appear in educated usage to varying extents in varieties spoken outside of Singapore, as is the case for the conventionalised scenarios descri‑ bed in Chapter 4. In such cases, the only differences from external usage are in the frequencies of occurrence of such items. Furthermore, as noted above, some non‑­standardisms may be shared across varieties, whether historically L1 or L2 in classification, as seen above in (13–14). In the chapters to follow, the studies used will instead refer to the classifications of sub‑­varieties in accordance with the function of use, and will use stylistic factors such as of degrees of formality and mode of communication (speech, e‑­text, or writing) to act as diagnostics of their sociolectal categorisation. In Singlish or SCE, the influence of the contact and substrate languages is of course most prominent, and this factor has also been emphasised on many occa‑ sions, though not necessarily with any verifiable means of testing it. Many of the features of the Chinese substrate are remarkably similar to those of many creoles, and yet Chinese is not a creole language, as pointed out long ago by Mufwene (1990: 10), so it is often difficult to discern whether a feature is derived from sub‑ strate influence or is simply a typical contact feature found universally. Alsagoff and Ho (1998: 137) cite Platt and Weber’s earlier (1980) studies, for example, in which tense is not marked on many of the verbs in SCE, but time reference relies mainly on temporal adverbs to express what is marked morphologically on the verb in standard varieties, e.g., (Platt and Weber 1980: 61): (16) A. My Mum she come from China many years ago! B. Oh, I see him last week. Time reference is in fact grammaticalised in Chinese through the means of aspect as well as time adverbs, as in many other languages, as discussed, for

30 

 Chapter 2 Singapore English

example, by Bao (1995, 2005). Bao’s (2005) study convincingly illustrates a heavy substratum influence in the use of aspectual markers such as already, which is nevertheless described as filtered through the grammatical constraints of the lexifier, in the present case, Standard Singapore English. Bao illustrates that the aspect marker already substitutes readily for the aspect marker ‑­le in Chinese, in marking the completive aspect where English would use either the perfect or the past tense, e.g., (2005: 252): (17) I see the movie already. ‘I saw the movie’. By use of the term completive, Bao does not rule out uses in which the ­completion of the event takes place in the future, as in the case of anteriors in subordinate clauses (2005: 241): (18) After it rain already, we can go out ‘After it has rained, we can go out.’ There are other uses for already which are unrelated to tense in English (such as inchoative uses), but what is important is the filtering influence of the lexifier discussed by Bao, indicating that the transfer from the substrate is not uncon‑ strained, and must accord with certain restrictions. Lexifier filtering suggests a distinct case for convergence in two grammars in the use of already, as illustrated in the fact that already cannot be used in SCE with events which are unintentio‑ nal or accidental (Chen 2008), e.g., I broke my arm already, a factor most likely due to semantic constraints of expectedness of the event it modifies (Dahl 2006). Such constraints may be associated with the lexical origins of the lexifier adverb, which were derived from a phrase ‘all ready’, as indicated in the Oxford English Dictionary Online, and dating back to 1275, indicating that the nominal that it modified referred to a state of preparedness, and this meaning has not been lost altogether from the present‑­day univerbation. The substrate argument is not necessarily weakened by the fact that source forms with the equivalent meaning already or finish are represented by Dahl (2006) (what he labels lamitives) as a regional, or areal feature of the grammaticalisation of aspect in Southeast Asia as well as in languages spoken in West Africa. Thus, to attribute the presence of already solely to the influence of Mandarin Chinese, as Bao (2005) seems to do, may be a little restrictive (even though, as Bao claims after Chao (1968), the principle features of the grammar of Chinese languages are reasonably uniform across all of the dialects including the southern ones). Ho and Platt (1993) is a comprehensive study of tense in SCE, and one that bases its hypotheses on the classification of sociolect in accordance with educa‑ tion or proficiency level. It was probably an appropriate means of classification



2.3 Grammatical morphology and discourse features 

 31

for its time (Ho’s original study was completed in 1986), though as noted above, the diglossic stratification of English in Singapore is now widely accepted. What is important in Ho and Platt (1993), though, is that it reveals that past tense is not always absent in Singlish, but is variably absent according to the lexical aspect of the verb and the grammatical aspect of the expression: punctual verbs are marked with past more frequently than non‑­punctual, and completed, perfective actions in the past are marked with past tense more frequently than imperfec‑ tive (past progressive or past habitual). The conclusion was that past tense is not being used to mark past time reference, but perfectivity, either lexical or gramma‑ tical, according to the speaker’s needs, and that the patterns follow those of the substrate languages, notably the Chinese dialects. Examples of such contrasts include the following (personally‑­observed) example: He [Edgar Hoover] goes after his political opponents, he bugs them and (19)  keeps files on them ... assassinated a few. In (19), the verbs expressing habitual aspect in the past are in the present tense, as they refer to repeated activities, and are thus grammatically imperfective, while assassinated is in the past tense, as it refers to a punctual event, bounded by the quantification of the object referent. Another frequently used distinctive feature of SCE is the alternation between the modal forms will and would, in examples like This office would be closed next Saturday. This was first observed in accounts such as Tongue (1979) and Crewe (1977), as noted in Alsagoff and Ho (1998: 142), who believed that would was often substituted for will as a means of expressing a more polite register, as in requests. This has been translated as a need to express tentativeness in some studies (e.g., Deterding 2003), though the alternation seems to resist all other explanation, and is also one of the few residual features of SCE that has extended into SSE, as often seen in students’ academic writing. As such it may be considered a marker of Singaporean English, regardless of the sub‑­variety being used. However, will is often used to express present tense habitual aspect in Singaporean English (see Ziegeler 2014a), and would is substitued even in these environments, in which it would not be used in standard varieties, e.g., (Deterding 2003: 37): (20) I usually would ... study in school until the evenings Deterding notes that the speaker is talking about what she usually does in the eve‑ nings, at the time of speaking. There is no possibility of an underlying m ­ otivation to express something with more tentativeness or politeness. In Teow (1988) this type of alternation is described as free variation, and the same free variation is claimed to exist between the modals can and could in SCE. While it may be descri‑ bed in terms of frequency as free variation, there are other, more subtle reasons

32 

 Chapter 2 Singapore English

for such alternation: the diachronic grammaticalisation of past tense crosslin‑ guistically is hypothesised by Bybee et al. (1994: 92) to involve an earlier stage of perfective marking in which only non‑­stative verbs are marked with past tense (these being the most prototypical contexts for past as they can refer to events which are terminated), leaving stative verbs unmarked for past. It is only at later stages that perfectivity may be applied to stative verbs and imperfective situations in which no meanings of event completion are possible, and it is at this stage that it may be identifiable as a true tense, since time reference is only required where it is not inferrable from the lexical aspectual semantics of the verb itself. This is also the case with modal verbs as modals are inherently stative in their lexical aspect, referring to meanings of desire, obligation etc.9 However, the consequences of the generalisation of past tense marking to statives are that past tense statives leave open the possibility of the completion of the states they refer to (Bybee 1995), thus creating a sense of irrealis or hypotheticality with regard to their actuality in the present. For Singaporean speakers, in which past tense is less frequently marked on stative verbs anyway (as Ho and Platt’s 1993 study illustrates), the irrealis meanings may be less obvious, and the use of would becomes overgene‑ ralised. There is thus free variation between will/can and would/could because for speakers working within an aspect‑­based rather then a tense‑­based system there is less likely to be a retention of irrealis meanings from their evolutionary development; i.e., for such speakers, past tense marking carries less salience on stative verbs. Much of this argument has been summarised in Ziegeler (1995: 326). Other features involving the expression of modality in SCE include the ubiqui‑ tous use of the bare modal can as an expression of consent, agreement, permis‑ sion, possibility and other expressions of positive judgement on a situation. This item has not been researched to a great extent in the literature, though it derives its origins almost undoubtedly from the substrate (keyi in Mandarin serves almost identical functions, bu keyi for the negative). The following examples appeared in the Flowerpod Corpus: (21) a. i ts wheatgrass! with lemon.. taste very nice.. even babies also can drink.... just like one cup a day can liao. Posted by: bell. Oct 11 2005, 12:52 AM

9 The term lexical aspect follows roughly what has been discussed in previous accounts as ­Aktionsart, or ‘kinds of action’ (Comrie 1976: 6–7); Comrie prefers to refer to it as inherent aspect, or semantic aspect, Sasse (2002) refers to it as Aspect2, while Smith (1991) and Bybee et al (1994) have referred to the same category as situation aspect. It is used in opposition to grammatical aspect, or Aspect1 (Sasse 2002) (viewpoint aspect of Smith 1991 and Bybee et al (1994)).



2.3 Grammatical morphology and discourse features 

 33

b.  must we put it in the fridge? in air con room can or not? wan to buy a bottle to put in office Posted by: rainisnow Aug 3 2005, 04:00 PM Note the accompanying particle liao in (21a), borrowed from Hokkien, which has a meaning similar to le in Mandarin; i.e., it acts as a perfective marker, (see, e.g., Bao 1995, 2005). In (21a), can has the meaning of ‘is ok’, while in (21b), the speaker is asking the forum if it is possible to keep the item in an air conditioned room instead of the fridge – can is used more in the sense of possibility here. The negative, cannot, is also used as a bare modal meaning often, ‘not allowed’, or ‘not possible’, as in the following: (22)  I gave my mom $200 but she will nag and say why so little, i explained that i’m tight and she will say “next month cannot ah” this went on for every other months.. haha.. Posted by: kerinong Nov 3 2006, 01:42 PM The speaker is contributing to a discussion on the amount of money remitted to the household by younger members of the family who are working and still living with their parents. In (22), the meaning is basically ‘it won’t do’ (ah is a discourse particle). Apart from this, other studies point to the more specific aspects of modal categories in SCE. Ziegeler (2000a) discusses the use of modals in the predicates of hypothetical wish‑­matrices, as well as counterfactual conditionals. Bao (2010b) discusses the grammaticalisation of must in the International Corpus of English (Singapore), and arrives at the conclusion that compared with other varieties of English, it is less advanced in its stages of development to an epistemic modal, because the motivation is not present from the Chinese substrate form, which in Mandarin is bixu, a modal auxiliary with only deontic functions. He also suggests an infrequency in the use of epistemic environments such as the modal perfect (must + have V‑­en). Bao’s analysis suggests that any development in the contact language must rely on a similar stage of development in the structure of a suitable substrate lan‑ guage in order to be justified; if it is not there, it will not take place independently. Previous studies such as Ziegeler (2000a) illustrated that grammaticalisation in contact may develop at a slower rate than in the lexifier when there are inhibi‑ ting constraints from substrate languages (e.g., the past tense in SCE, constrained mainly to non‑­stative verbs). At the same time, grammaticalisation may proceed at a more accelerated rate than in the lexifier (hypergrammaticalisation) when there are fewer traces of lexical retention associated with the grammaticalising form for the SCE variety. This is particularly noticeable with hypothetical modal

34 

 Chapter 2 Singapore English

verbs in the dialect, such as would. However, taken to the extreme, Bao’s (2010b) analysis presupposes that not only the presence of a feature in the substrate ensures its replication in the contact language, but also that the absence of a function for that feature in the substrate entails its subsequent absence in the contact language as well. If that were the case, then there would be an unlimited number of functional absences to explain in a contact situation. In some cases, it is worth considering a broader explanation in which the different functions may be served by other forms interacting elsewhere in the system.

2.3.2 Other grammatical features There are many other distinctive features of the colloquial sub‑­variety that have been described in detail in recent studies. In (16) for example, it is noted that there is often no distinction in SCE between masculine and feminine 3rd person singular pronouns, following a pattern familiar to spoken Chinese, in which male and female 3rd person singular pronouns are homophonous, though their written form shows differences (both are pronounced tā). Alsagoff and Ho (1998: 147) also note a blend of both English and Chinese in the expression of relative clauses; while in English, the head precedes the relative clause, in Chinese languages, it is always the relative clause which precedes the head, e.g., as in the Mandarin example (Alsagoff and Ho 1998: 145–7): (23) a.

Mandarin Nie jie-jie de  neige haizi hen huaidan Pinch  sister RP  that  child very  naughty ‘The child who pinched my sister is very naughty’.

b. SCE That boy pinch my sister one very naughty. In (23a) it is seen that the relative clause precedes the head, neige haizi, while in (23b), the head (That boy) precedes the relative clause, with the ­relative pronoun one occuring in clause‑­final position, exactly as de occurs in (23a). Thus the resul‑ ting construction is a clear merger of both the Chinese and the English relative clauses, or as Bao (2009) claims, it reflects the grammar of Chinese de ­filtered through the morphosyntactic constraints of the English specific pronoun, one. Other syntactic features frequently referred to include the absence of subor‑ dinate conjunctions, formerly mentioned in Platt, Weber and Ho (1984: 125) in reference to conditionals: (24) You go by metre, you have to pay.



2.3 Grammatical morphology and discourse features 

 35

This reflects a pattern of expressing conditionals in Chinese in which the subor‑ dinate and main clauses are simply juxtaposed, as noted by Comrie (1986: 82), for Mandarin: he jiu, wo (25) Zhangsan   Zhangsan drink wine 1SG  ‘If Zhangsan drinks wine, I scold him’.

ma scold

ta him

Bao and Lye (2005) discuss similar examples as a feature of the transfer of topic‑­ marking in Singapore English. Similar examples are listed by Alsagoff and Ho (1998: 148): (26) You make like that can work or not? ‘If you make it like that, can it work?’ (27) You turn 21 can have a big party. ‘When you turn 21, you can have a big party’. (28) She never say hello. Walk in, sit down, talk to my sister. Always do like that. ‘She never says hello. She walks in, sits down, and then talks to my sister. She always does that.’ In (26) the polar question tag or not is used; this can take other forms such as can or not?, reflecting the grammatical influence of the Chinese languages in which, e.g., bu shi in Mandarin (‘not is’) may also be found in sentence‑­final position. A universal question tag, is(n’t) it, replaces all other forms of question tagging, as well as serving to substitute for back‑­channelling expressions of confirmation or surprise, e.g., ‘really?’ This invariable question tag also reflects a similar form in the Chinese substrate languages (in Mandarin, shi bu shi, lit. ‘is‑­not‑­ is’ may appear in similar functions, sentence‑­finally). The use of is(n’t) it may be classed as a calque, but it is not possible to determine this for certain, as it is found in other dialects of English which have no Chinese substrates, such as Indian English, Welsh English, and also Irish English (see, e.g., Trudgill and Hannah 1986[2002]). The following example is extracted from the Flowerpod Corpus: (29) She sounds like a bad mannered b**ch! so everyone treat her like queen, is it? Posted by: Happy_Daze Aug 25 2008, 06:42 PM (29) is also an interesting example of the use of a bare nominal, queen, (‘so ­everyone treats her like a queen’). The absence of determiners for non‑­referential nouns will be discussed in detail in Chapter 7, so will not be dealt with further in the present chapter, nor will other features such as ­topicalisation and subject and object deletion, known as pro‑­drop in Alsagoff and Ho (1998).

36 

 Chapter 2 Singapore English

More examples illustrating the systematicity of SCE may be found in recent studies such as Bao and Wee (1999), in which the marking of the adversative passive in SCE is detailed. According to Bao and Wee, there are two ways of expressing adversative passives in Singaporean English: one using a borrowed form from Malay, kena, and one using give. Only the second is a contact gram‑ maticalisation of the Chinese model form, (e.g., bei in Mandarin), but both are roughly equivalent in function to the get‑­passive of standard English, which also has migrated into the domain of malefactives and adversatives. Thus, the fol‑ lowing ­comparisons are possible (Bao and Wee 1999: 2): (30) The thief kena caught by the police and for give: (31) The dog give the boy kick (Bao and Wee 1999: 5).10 They maintain that the subject of a give‑­passive must be considered as contributing in some way to its own misfortune, and therefore, must be animate. In (31), it is the Chinese dialects, in particular, Hokkien, which are claimed to be the contact source for the use of give as a passive‑­marking verb in SCE, e.g., (1999: 7): (32) eeay kow hor ee that dog give 3SG   ‘That dog got beaten by him/her’.

pha beat

The only difference in linguistic distribution between the contact source lan‑ guages such as Hokkien and Mandarin and the SCE lexifier, English, is that the passive marker in Chinese does not necessarily require its subject to be animate, while in the use of English give there are constraints imposed by the semantics of the lexifier and give requires an animate subject (Bao and Wee 1999: 8). Thus, when a form is modelled from the languages of contact, it also appears to retain some traces of the semantics from the lexico‑­grammatical interface of the lexifier. Bao and Wee (1999: 8) use this evidence to suggest that substrate influence is structural while superstrate influence is lexical, as a general rule, citing accounts such as Muysken and Smith (eds., 1986) and Mufwene (ed., 1993, 1996). However, in the example of kena we have lexical and structural substrate influence, i.e., both material copying and pattern copying, so the general rule may not always apply. On the other hand, in the case of give the lexical influence of the lexifier

10 Bao (2010: 805) notes that although the kena passive is frequently heard, the give‑­passive is now rarely used in spontaneous conversation.



2.3 Grammatical morphology and discourse features 

 37

language interferes with distributional patterns, so it is not easy to tease these two factors apart. Other studies which seem to fly in the face of such assumptions include the many studies of discourse particles in Singaporean English, such as Lim (2007) and Wong (2004), to cite more recent accounts. In such cases, the entire particle is transported wholesale into the grammar of SCE, without, in most cases, lexical modification from the superstrate at all, a clear case of mate‑ rial replication. It is questionable whether such particles are an element of the grammar or not (see recent studies such as Kaltenböck et al. (2011). However, they are part and parcel of an autonomous system, and also contribute to a determi‑ ned series of functions, according to the studies discussing them (Lim (2007) also mentions Gupta (1992b) as claiming for 11 such particles, Wee (2004) as listing 8, and Ler (2006) as listing 10). It would appear as a general assumption that the dis‑ course particles have as one function that of lexifying prosodic intonation effects which cannot be so easily articulated in the substrate languages because of tone distinctions. The introduction of the class of particles into Singaporean English does not necessarily suggest, though, that SCE is a dialect without the intonation that SSE has. All it is suggesting is that the emotive contrasts expressed by such particles probably lend a more colourful perspective to the landscape of Singlish than if English intonation alone were used. Lim (2007: 448) names lah [la] as the most well‑­known particle, so well‑­ known that it has entered the Oxford English Dictionary. It has been described as having a range of functions, according to various analyses dating as far back as Tongue (1974). Lim’s study is certainly comprehensive in that it does not simply discuss the discourse function for the use of the particle, but also the tones used, which are carried over into SCE from the relevant Chinese dialects.11 Lah is found, according to Lim (2007), (i) in Singaporean English, where it is said to indicate solidarity, familiarity and informality in appealing for accommodation, (ii) in Bazaar Malay, where as well as marking solidarity, it expresses emphasis, (iii) in Hokkien, where it indicates persuasiveness, or finality, (iv) in Cantonese, where it indicates a lack of forcefulness, and has a softening effect, and (v) in Mandarin, where it simply provides emphasis (2007: 460). The origin of the particle in SCE is therefore still undetermined, since there are so many substrate possibilities, though Lim (2007: 465) believes that it is a hybrid feature of SCE, having emerged from the typological convergence of the particle in Hokkien and Bazaar Malay.

11 Platt and Ho (1989) also mention tone as carried over into Singaporean English particles, ­according to Wong (2004: 761), and Lim (2007: 448) notes that Platt (1987), Kwan‑­Terry (1992), and Lim (2004) all note two distinct tone differences in the pronunciation of la.

38 

 Chapter 2 Singapore English

Wong (2004: 763) does not attempt to determine the source of the Singa‑ porean English particles, emphasising that they have been transferred into the common Singaporean English dialect, which is spoken by speakers of all lan‑ guage backgrounds who can employ them all with the same facility. He describes three different functions for la, all differentiated by tone: (i) imposition of the speaker’s attitudes or opinions, e.g., (2004: 764): (33) A. Go to gym at 9 am [tomorrow]? B. 9.30 la3. Don’t be so unearthly la3. (ii) propositional la, in which the speaker uses it to propose an idea or some advice to the hearer, e.g., (2004: 768): (34) A. You don’t know this, right? B. (in mock defiance) Ya la4. and (iii) persuasive uses, e.g., (2004: 771): (35) You tell him la2; he’ll listen to you la2 . Wong uses the Mandarin tone system to describe the differences, though he does not suggest that the particles were transferred from Mandarin. If the particle came from Malay, it is not likely to have carried tone in any case, as Malay is not a tone language. However, Lim (2007: 463) suggests that tonal features are being lost from Singaporean English particles, and that this loss is already quite advanced with la[h] (a form of bleaching, since tone carries lexical significance). It may appear that all three of Wong’s uses are derived from one another, since the meanings are quite similar,though Lim’s (2007) proposal of a typological conver‑ gence where la is concerned may place doubts on such a suggestion. Other discourse particles which have various functions according to their possible origins include ah, marking a question, or acting as a back‑­channeller, lor, which generally expresses obviousness, hor, expressing agreement or con‑ firmation, leh, marking comparison or a question, meh marking a question or doubt, and ma indicating obviousness (Lim 2007: 461). Another particle, what, appears to be derived from English, as its borrowed form is undetermined, if it had one at all. Lim rejects the possibility of its derivation from archaic discourse particles in English, suggested in Brown (1999), considering instead a near‑­ homophone in Cantonese wo21. However, she claims that the vowel in the Can‑ tonese form was a rounded back vowel, while in the Singaporean English form it is a central unrounded vowel. She prefers instead to consider the analysis of Kwan‑­Terry (1978: 34) that it was a calque of ma from either Hokkien, Cantonese or Mandarin, suggesting emphasis as well as a certain annoyance as part of its meaning. However, the question remains open why what was selected for this calque, amongst all the possible English forms which could have been selected in

2.4 Overview 

 39

its place, and why ma was not just borrowed into Singlish. Wong (2004) spells the form as wut, and claims that it expresses a degree of denial; he offers examples such as the following (from Kwan‑­Terry 1978: 25): (36) A. Why didn’t you come in? B. You told me to wait here wut. In standard English, a sentence‑­final adverb like though might serve the same function, but the use of appropriate intonation might rule out the need for a par‑ ticle altogether. It is not certain whether the form could have been a retentionist feature from archaic (military) colonial usage, as suggested, e.g., by Kuteva (2013). Whatever the case, the search for its origins still goes on.

2.4 Overview In the situation of Singapore, we see one account of an established dialect of contact English which has developed rapidly from its earlier stages possibly as a creoloid used by children of different language backgrounds in the playg‑ rounds of the first, post‑­independence schools, when mass education had been encouraged across all ethnicities. Other studies such as Roberts (2000) cited in Hopper and Traugott (2003) illustrate a similar setting for the development for Hawaiian Creole. On the other hand, as Bao (2001) indicates, it would not have been impossible for some form of a basilectal, endogenous pidgin‑­like variety to have emerged out of the needs of a trading situation in the rapidly developing colony of the 19th century. The tendency to accord early language development to the manipulation of children has always been popular (since Bickerton 1981, at least), and Platt (1989) is adamant in his claims for a second‑­language variety as having emerged through the educational system. However, it must be remembe‑ red that adults have an important role to play in the development of a language also, as illustrated in studies such as Romaine (1999) on Tok Pisin. The current situation of English in Singapore, according to recent accounts, remains a diglossic one, having passed through at least three descriptive stages of classification: the sociolectal continuum stage of Platt et al. (1984), the diglos‑ sic stage of SCE and SSE advocated by Gupta (1991, 1994), and the more recent diglossic stage of Alsagoff (2010). Clearly, accounts have varied as to the nature and function of the two diglossic poles. Recent research reveals a distinct prefe‑ rence for educated speakers to adopt Singlish as a trademark of their Singapo‑ rean identity, switching to standard English as the need arises. Such a situation extends considerably the stylistic shifting illustrated by Platt and his colleagues some 30 years ago, in which speakers were restricted by their education level and

40 

 Chapter 2 Singapore English

consequently by their proficiency. In the present‑­day Singapore, proficiency is not an issue, and English is spoken not so frequently in competition with the wide range of Chinese dialects which were once used transactionally or domestically for one function or another. As a result, the trademark of the traditional Chinese dialect clan‑­culture is more frequently exchanged for a sense of coherent nation‑ hood in the use of a language which is neutral to any form of cultural sovereignty or chauvinism in the region. English maintains, therefore, the advantageous role of eclipsing ethnic divisions in favour of its function as a lingua franca. In the above examples are illustrated a range of distinctive features of Singlish some of which are traceable to substrate or adstrate origins, such as the bare modal can, or is it used as a universal question tag, though there are also a number of alter‑ native explanations which rely more closely on universals of language development to account for the presence of such features. The above account has not mentioned, for example, the progressive aspect in Singlish, which extends to stative verbs (e.g., I am having a cold). Such generalisation was found in Old and early Middle English up to 1000 years ago (see, e.g., Ziegeler 2006), and curiously enough, it cannot be accounted for by the Chinese substrate, as Bao (2005) points out. The usage is also found in other new English dialects, such as Indian English (Sharma 2009), and in East African English as well (Ziegeler, to appear). In such cases, one can only resort to universalist explanations to account for its distribution. In other cases, there is a clear instance of replica grammaticalisation occurring (see Heine and Kuteva 2003, 2005), in which a form which is modelled on a similar function in the substrate lan‑ guage will undergo either the same processes of grammaticalisation as the model in the substrate, or else grammaticalise via universal processes of grammaticali‑ sation. In the case of the progressive, it is possible that another process may be in operation – the grammaticalisation paths follow instead those of the the lexifier (see Ziegeler (to appear) for a more detailed account). Whatever the case, it cannot be denied that first, there is a distinctive variety of English used in Singapore which is at the disposal of all speakers, regardless of edu‑ cational level, and which is recognised internationally as belonging to Singapore, functioning to mark a form of cultural cohesion amongst the various ethnicities living on the island. Second, it is clear that such a variety is characterised by both material and pattern replication of the numerous Chinese and other local langua‑ ges that are, or have been, spoken alongside it in the small, but densely populated region. Questions of the origins of certain features which characterise it will always be raised, and debated; however, what is important is not so much what theoreti‑ cal explanation can be given to determine their origins, but what the presence of such features can offer to the development a theory of contact that offers predictive appeal to more general frameworks of linguistic research beyond its application to Singlish alone. It is to such objectives that the present study is mainly directed.

Chapter 3  Construction grammars and the paradox  of ‘mixed’ construction types 3.1 Introduction: constructions in variational contexts It has been more than twenty years since Fillmore and his Berkeley colleagues (e.g., Kay 1984; Lambrecht 1986; Fillmore Kay and O’Connor 1988) first invoked the notion of a unified grammatical object of analysis, bringing to light a number of studies initially devoted to the discussion of idiomatic phrases and expressions that seemed to avoid conventional analysis by the structuralist trends of syntacticians of the time. At the same time, another extremely vigorous area of linguistic research also led by principles of unification was taking the stage as well, perhaps more so in the arena of typological studies, and that was the field of grammaticalisation research (see, e.g., Bybee 1985; Traugott 1988). While these two fields of research were not perhaps intended to be heading in the same direction, it is interesting to look back and compare their objectives, since both espoused the central hypothesis of a holistic description of language, in which form and meaning could not be separated, and in fact, were dependent on one another. However, it was not until the Berkeley awakening that alternative ways of considering language data began to emerge, bearing with them new models and new sets of constraints. In many ways, grammaticalisation had followed the pathway of the constructionists, in the priority it gave to pre-grammatical, lexical forms from which more abstract schematic meaning may develop. However, while grammaticalisation regards the synthesis of meaning and form as belonging to a diachronic, evolutionary process, construction grammars generally describe the composition of meaning and form in synchronic terms, in that the meaning of a construction is held to lie in the syntactic arrangement of its constituents, with or without reference to any lexicalised, diachronic precedence. A more recent, alternative viewpoint, though, in which constructions are considered to undergo grammaticalisation themselves, is provided by Traugott (2007, 2008) and Traugott and Trousdale (2013) as discussed below. In this way, they have shown that the two theoretical domains share common insights. It is undeniable that there are always questions arising in any new theoretical direction, particularly as the popularity for the theory increases and is applied to an ever-expanding range of linguistic situations. It was noted earlier that the present collection has two principle theoretical objectives: first, to shed light on the problems of contact and generally to explain the phenomenon of

42 

 Chapter 3 Construction grammars and the paradox

interlingual identification in a more theory-specific manner, and second, to enable the problems of contact to reveal what has been largely overlooked in the field of constructions and construction grammars to date, that is, how to account for the nature of instability in constructions. Variation and change may never accommodate the notion of a seamless, immutable constructicon, i.e., the language-specific inventory of constructions (Traugott, to appear), or the set of constructions specifying the particular lexical characteristics and combinatory possibilities required to build constructions (Sag, Boas and Kay 2012: 18). With regard to instability, it is shown, for example, in Colleman and De Clerck (2011) that the range of lexical input available to the double-object construction in English has become significantly restricted since the 17th century, an effect similar to specialisation in grammaticalisation (Hopper 1991); for example, the verb banish can no longer be used in a double-object (ditransitive) construction today (Colleman and De Clerck 2011: 192): (37) And a man that could in so little a space, first love me, then hate, then banish me his house13 but was amongst a range of verbs used in Late Modern English that were available for use in ditransitive constructions of the 18th century. Their data would suggest that the English construction is vulnerable to change over time, particularly to the changing nature of the lexicon, a point that is also made by Traugott (to appear) in stressing the overlap between grammaticalisation and construction-building. Such changes will be discussed further below. The present chapter will examine some of the descriptive problems associated with the principle approaches to constructions discussed in the past 20 years, in order to offer a more elegant definition suitable for use in variational and contact language situations. Section 3.2 will consider some questions on the identification and labelling of a construction, including what is understood by the terms form and meaning, as well as the central question of compositionality. Section 3.3 will look at the relevance of grammaticalisation to constructions and review some of the recent studies in that field. Section 3.4 offers an argument against the notion of coercion in constructions, much of which was first published in Ziegeler (2007, 2010) in which data from Singaporean English revealed that

13 This example is taken from the Corpus of Late Modern English Texts (compiled by Hendrik de Smet, University of Leuven); the text was Pamela, or Virtue Rewarded (Samuel Richardson). The meaning of banish according to Colleman and De Clerck is ‘to officially order someone to leave somewhere’. Thus, the present-day meaning would involve a caused-motion construction: ‘banish me from his house’.



3.2 Questions of construction descriptions 

 43

coercion could not be upheld as an independent construction-related process. Section 3.5 consolidates the present position regarding form and meaning, and summarises the chapter.

3.2 Questions of construction descriptions 3.2.1 Identification and terminology It may well be emphasised that a construction is a form-meaning relationship and nothing more (this is the usual definition found in various manifestations in recent studies such as Boas (2013), Sag, Boas and Kay (2012), Croft (2001, 2013), Goldberg (2013, see below), Traugott (to appear) or Trousdale (2010, 2012) or a form-meaning relationship which is indivisable and built up on clusters of conventionalised semantic, pragmatic or textual features (Fried, to appear), or further, a “special, bipolar schematic structure that captures generalizations over form-meaning pairs” (Bergen and Chang 2005: 157). However, in defining a construction, it is often difficult to fully reconcile the notion of indivisability with the role played by conventionalised, mainly semantic components, and the question of just how far this conflicting situation may be taken is one of the key issues in current construction grammar theory. Originally, the definition offered by ­Goldberg (1995: 4) emphasised the non-compositionality of a construction, suggesting also the reducibility of a construction type: (38) C is a construction [if and only if] C is a form-meaning pair such that some aspect of Fi or some aspect of Si is not strictly predictable from C’s component parts or from other previously established constructions. This definition was applied very successfully to the types of constructions that Goldberg (1995) first studied: argument structure constructions, in which the syntactic positioning of the arguments revealed the schematic semantic generalisations of transfer meanings as belonging to the syntax, rather than to the verb types that were used in them. In more recent years, with the increasing awareness of the factor of the lexical input in building constructions, this has become a contentious issue and Goldberg (2006: 94) has modified her original definition of constructions to accord a stronger role to the nature of usage frequency in defining a construction: (39) Any linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable from its component parts or from other constructions known to exist. In addition, patterns are

44 

 Chapter 3 Construction grammars and the paradox stored as constructions even if they are fully predictable as long as they occur with sufficient frequency …

The constraints on semantic decomposability have thus been somewhat relaxed in order to allow for the more recent accounts in which construction syntax cannot be seen to be entirely responsible for the semantic interpretation of a construction (see, e.g., Boas 2003; Croft 2003; and others to be discussed below). Interestingly enough, though, the predictability of the semantic interpretation of a construction does not in any way detract from the original glamour of the theory, nor does it reduce in any way the usefulness of what Goldberg (2013) calls “constructionist approaches”, accounting for the variety of sub-fields into which the original Berkeley theory has propagated today. Even Goldberg herself has now generalised the definition of a (grammatical) construction to accommodate the variation in such approaches (Goldberg 2013: 17): (40) Constructions are defined to be conventional, learned form-function ­pairings at varying levels of complexity and abstraction … leaving out completely the aspect of semantic predictability in their definition, and suggesting, as did Croft (2001) that they can be reduced to their most atomic form, which in such revisions may consist of only one word or morpheme. In the end, with the redundancy of the predictability element in the definition, one may well ask, if all constructions can be so readily decomposed semantically, have we still got constructions after all? And further, could it be said that the theoretical premise is now reducible to nothing but a basic methodology for s­ tudying form and meaning together in a language? The original blend of syntactic ­complexity and semantic parsimony by which constructions were identified as ­non-decomposable expressions or idioms, such as the way-construction, or the let-alone construction (lexicalisations constructed out of multi-word phrases or expressions), has become partially eclipsed by a prolific upsurge of new approaches, in which almost any unitary linguistic element can participate as a candidate for construction-hood. Goldberg’s (2006: 5) examples of constructions types are listed by Boas (2013), and include the following (English) categories: (41) single morphemes (e.g., -ing, pre-); lexical words such as avocado, ­including complex or compound forms such as daredevil, conjunctions such as and, complex words such as [N-s], expressing plurality in English, idioms, including partially filled ones (jog X’s memory), co-variational conditionals (the X-er, the Y-er), ditransitives, and passives. One of the obvious consequences of such proliferation is that it is sometimes difficult now to tell what is not a construction, or where its constraints lie.



3.2 Questions of construction descriptions 

 45

For example, if one were to take (39) as a definition of a construction, then in the paradigm of the verb be in English, one would not deny that each of its members are constructions in themselves, since their forms or functions are not predictable from their component parts (which reduce to just a few phonemes in some cases, e.g., /ɪz/). However, the existence of each member is fully predictable from other constructions if the other members of the paradigm are also constructions. The definition in (40) is insufficient to refer to a construction, since it is often difficult to determine the limits of conventionalisation without adequate empirical verification. Reference to form-function pairing is also problematic, as will be shown in the present study, since functions may vary both within a language and across varieties of a language, not necessarily involving a change of form, e.g., the use of the negative modal mustn’t has an epistemic function in north-western (British) English which is not found in southern British varieties of English: (42) a. He mustn’t be in – his car is gone (northwestern British English – epistemic) b.  He mustn’t go out today (northwestern British English and southern British English – deontic) (see, e.g., Trudgill and Hannah 1986: 47). In such cases, given the universal application of the term construction to account for any form-meaning relationship, does the negative epistemic modal mustn’t constitute a different construction in the northern British dialects from its identity in southern dialects? If so, it may be questioned how far such differences may extend before we can say we have a different language. On the other hand, many approaches have referrred to ­form-meaning pairings, while others do not distinguish between meaning and function, e.g., Hoffman and Trousdale (2011: 2), who note the evidence of a lexicon-syntax continuum discussed by construction grammarians which should entail the existence of synchronic and diachronic variation. Others refer to meaning when in fact they are discussing function, or parenthesising function under the general label of meaning (e.g., Traugott, to appear, though in her diachronic approach this is hardly surprising since meaning and function are seen to collapse together over time). Similar mnemonic confusion is reflected in the vast array of terminological variation used in describing different constructions; for example, Croft (2013: 214) refers to the Active Voice construction, a phenomenon which in English must surely be defined more by what it is not than what it is, since the formal identification of such a construction is not usually evident without the comparison of a passive-marked counterpart construction. Croft’s answer to the question of identification and labelling of constructions is that they can be identified by their semantics, and labelled with whatever terminology is desired (2001: 50). However, such a methodology is less than ideal, given

46 

 Chapter 3 Construction grammars and the paradox

that sometimes the same construction is identified by its lexical ­components (e.g., the ­kick-construction), sometimes by its lexical category level (e.g., the ­transitive verb construction), and sometimes by its clause-level label (e.g., the Clause construction) (2001: 57), depending on the level of schematicity required for analysis. This taxonomic hierarchy (p. 56) illustrates a similar inconsistency: ­sentence-functional labels being situated at higher levels than item-functional labels, with lexical inserts at the bottom level, resembling inescapably a pattern reminiscent of hierarchical trees of traditional structuralism in its conflation of lexical, syntactic and functional information at various levels in the model. The problem of labelling a construction is not restricted to Radical Construction Grammar, of course, and it would appear that terminological labels are selected at the convenience of the researcher, depending on what aspect of a construction’s complexity is relevant to the purposes of investigation.14 Nevertheless there are implications involved in construction identification and it is thus questionable whether it is possible to determine the limits of a constructicon itself. Croft approaches this problem resorting to a prototype analysis (cf. Rosch 1978) in which the principle determinants of a construction type are found in the number of features it possesses which are not shared by other constructions, suggesting that constructions form taxonomies (2001: 52–53). Apart from such allusions, he has little more to suggest except to add that the identification of constructions involves psychological research into the creation of taxonomies and categorisation processes (the same reference to categories and taxonomies as a means of discriminating constructions is mentioned in Croft (2013)). However, such a problem is crucial for the present study, since cross-variational data, and especially data influenced by contact, is especially requiring of a more rigorous definition to distinguish constructions, given that, as discussed below, if constructions are language-specific, it will only be formal distinctions rather than universal functional distinctions that distinguish one construction from another. Croft’s (2001) Radical Construction Grammar was probably the first attempt to generate constructions from minimal features of a language, being radical in the sense that it unleashed the original Berkeley theories from their restricted applications to idioms, to account for a broader, typological perspective, but also because it undermined the traditional generative syntactician’s efforts to isolate a Universal Grammar comprised of similar syntactic categories across all languages. Croft repeatedly emphasised in his 2001 study that constructions

14 This carries the implication also that certain construction types may be either too generalised to be useful, such as the English Verb Construction, or may be so specific that they are rarely referred to in any empirical studies.



3.2 Questions of construction descriptions 

 47

are language-specific and non-universal (2001: 29, 50, 59), though he goes on to maintain (p.  60–1) that they are not language-specific in their functions (thus implying that ­grammatical functions are shared across languages, which is functionalist domains anyway). As emphasised above, accepted doctrine in ­ then, if a construction is a form-function pairing, then the only reason for their language-specific nature is the nature of their forms. The fact that such formmeaning/function pairs may be considered language-specific is one of the major issues to be approached in the present volume, which seeks to answer the question posed in the Introduction of how constructions can be described in a contact replica language in which the syntax derives principally from one language and the lexicon from another. The solution to this question may require a review of the entire ontology of constructions, and, in particular, to consider again the question of construction identification. However, it is not the task of the present volume to attempt to overturn existing theoretical postulates in this blossoming new field of linguistic research. The aims, as stated in the Introduction, are first to use contact situational data to resolve some of the problems of constructions and construction grammars, and second, to use construction-based approaches in order to explain some of the questions of contact. The role of problems such as construction identification will thus become clearer as the work progresses, but it is rather the investigation of constructions in dialectal variation and in contact that forms the principle motivation for compiling the present study. Boas’ work also contains no recourse to a more improved method of construction identification, the problem that seems to pervade most major theoretical positions. Goldberg’s array of construction types listed in (41) leads to her later revisions of the definitions suggesting that a construction is just about anything that has a form and a corresponding unitary meaning attached to it. However, as noted earlier, a cursory glance at the list creates the impression that a c­ onstruction may be labelled by any salient feature of its form, its semantics, or its ­grammatical function. In (41) it appears that the more complex the constructions becomes, the more likely the terms labelling it are to refer to its functional properties. Elsewhere, we find terms such as the VP construction and the WH construction (referring to formal criteria), the Subject-Auxiliary Inversion construction (referring to a s­ yntactic process), (Goldberg (2013: 28), the way-construction (labelled by a salient lexical feature – Goldberg 1995), or the Subject-Predicate construction (labelled by a grammatical function of syntactic constituents) (Boas 2008: 131). In Fried (to appear) we even have the Positive Suggestion ­construction, which takes the form Why don’t you (p)?, a pragmatic or speech act function being ­highlighted, and the Group Identity Noun Phrase construction, with examples of constructs such as the poor, the elderly etc. in which the common feature is an adjective functioning as a noun expressing human generic reference to a selected (social)

48 

 Chapter 3 Construction grammars and the paradox

grouping.15 The absence of any uniformity in the labelling of c­ onstructions may, of course, be useful in demonstrating their holistic character, but it does little to contribute to the essential validity of each research domain. Perhaps it could be suggested that they can be identified by any common minimal feature which is not found in other constructions, though this definition risks circularity, and would be misleading in cases such as the Positive Suggestion constructions, for example, if the same syntax were associated with a different meaning, e.g., Why don’t you go and take a running jump! in which case the suggestion is negative, or simply an enquiry into the regular habits of the interlocutor: Why don’t you believe in God? Why don’t you love me? All such constructions share the same form, and yet they refer to different functions. Thus, the matter of determining terminological distinctions for constructions carries a vast number of problems.

3.2.2 Compositionality Although applauded for its efficiency, Croft’s (2001) study was also criticised by Boas (2010) in particular for its lack of adequate data sampling to illustrate crosslinguistic construction similarities, for example, in the domain of voice and transitivity. Croft (2013: 231), however, argues for a balanced sample of languages for the purposes of comparing constructions across languages; this need not imply a minimum size. The criteria for examining and comparing crosslinguistic examples of constructions are based on similarities of function, since functional criteria are the only criteria which may be shared (“universals of ­grammatical constructions” (2013: 231)), according to Croft’s accounts. In suggesting that there are such universals, Croft is not rejecting the fact that the symbolic ­relationship between form and meaning associated with constructions is not a crosslinguistically universal one. However, ultimately, it is such symbolic relationships which create the universal distribution of functions across languages, and this is the consequence of language-specific pairings. Croft also indicates the need to use other methodologies for analysing possible universal properties in constructions (p. 232).

15 The term construct (meaning an actual expression) derives originally from Goldberg (2002: 348), who also claims that constructs may be composed of numerous different constructions; e.g., the construct What did Mina buy Mel: Ditransitive, Q[uestion]-construction, ­Subject-­Auxiliary inversion, VP constructions, NP construction, Indefinite Determiner construction, and the ­ ­atomic constructions: Mina, buy, Mel, what, and do. In such listings, we see again that constructions are labelled by an indeterminate mix of both formal criteria as well as functional and semantic features.



3.2 Questions of construction descriptions 

 49

One such methodology, frame semantics, has been studied in work of Boas (2003, 2005, 2008), and in his work on mini-constructions. Boas (ed., 2010) had also attempted to search for similarities in constructions across languages. However, such studies are useful only in the evidence they present of certain means of coding generalisations across languages, or in examining properties of constructions in order to seek out, for example, translational equivalents of ­constructions crosslinguistically. Thus, although studies of this type are ­important in motivating further research, they still do not dismiss the fact that constructions must be language-specific, if they are to retain their original definitional qualities of form-meaning symbolic relations (since it is necessarily the case that most ­languages are distinguished from one another principally by formal means). Comparative results similar to constructional comparisons may just as easily be obtained by looking at grammaticalisation studies, in which universal grammatical functions are perceived to develop from universal lexical source items – such an approach also involves, in many cases, complex constructions (e.g., the be going to future in English; see below). The advantages of Boas’ mini-constructions, however, are seen more perspicuously in their application to language-specific examples. Mini-constructions are defined as individual verb-senses, with their own frame-semantic, pragmatic and syntactic specifications to be referred to in the presence of over-generation of abstract senses of a construction (Boas, 2003, 2013: 7). Boas frequently refers to a lexical-constructional model, and a mid-point (of hierarchical networks) found in constructions,16 which earlier accounts (e.g., Goldberg’s original model) had not taken into consideration, especially in the description of argument structure constructions. In earlier studies, he had described mini-constructions in terms of event-frames, since they were originally conceived in order to provide an alternative viewpoint on the argument structure constructions used by Goldberg (1995), for example, the now classic: (43) Sam sneezed the napkin off the table. (Goldberg 1995: 29) which has been subject to considerable analysis by various accounts ever since for the fact that it demonstrates the way in which the construction syntax derives ‘fake’ transitivity out of a verb that is normally intransitive. In the absence of a mini-construction analysis, the original Goldbergian theory had proposed that the semantics are derived from the construction itself, and that it was the syntactic positioning of the arguments which led to the interpretations

16 Similar descriptive accounts known as lexical construction models appear in Iwata (2002) and Ruiz de Mendoza Ibáñez and Usón (2008).

50 

 Chapter 3 Construction grammars and the paradox

of dynamic transfer possible with novel examples such as (43). Thus, any new verb could be inserted into the verb slot and the same senses of dynamic transfer could be recreated each time. Boas’ mini-constructions proposed instead that it was the verb semantics that led to the association of a particular construction meaning with a particular syntactic configuration. Goldberg went on to discuss the usefulness of construction polysemy in accounting for related constructions of the same family, such as the resultative (1995: 81): (44) Pat hammered the metal flat. and the ditransitive (1995: 89): (45) John gave Mary an apple. In all of the related constructions, the basic sense of dynamic transfer is still present. It is not possible to substitute verbs which do not contain this sense element (without some kind of special interpretation); for example: (46) ??Sam blinked the napkin off the table thus suggesting that the original semantic architecture, X CAUSES Y TO MOVE (TO) Z, was what was determining which lexical verbs could be used in which constructions. It was also seen in later accounts that there were limits on the productivity of prepositional phrases used in such constructions, for example, not just any syntactic configuration of NP V NP NP series would result in a ditransitive interpretation, as can be seen in possible examples such as the following: (47) William married Kate, a rich heiress from Milton Keynes. Many later accounts which had attempted to build on Goldberg’s original ingenuity had also overlooked the fact that some of the proposals they were making had been alluded to by Goldberg herself in the first place, albeit without being highlighted as intrinsic to the model. The central senses of the three argument structures: X CAUSES Y TO MOVE (TO) Z (caused-motion), X CAUSES Y TO BECOME Z (resultative), and X CAUSES Y TO RECEIVE Z (ditransitive) had provided the ­foundation to what are nowadays described as schematic semantic structures from which the productivity of the construction-type can be determined, though they differ, naturally, in their explication. Amongst such later accounts Boas’ mini-constructions postulated that verb senses were contained in event-frames, using the ideas originally attributed to Fillmore (1982). Such frames subcategorised for participant slots, and semantic and pragmatic information such as the specific lexical or phrasal associations of resultative phrases, either an adjective

3.2 Questions of construction descriptions 



 51

phrase or a noun phrase, e.g., in Jack painted the barn red a nominal complement may be substituted for the adjectival one: a light shade of blue (Boas 2005: 452). Such elements are introduced as part of the semantic frames associated with a lexical item and are shown to react on the grammatical structure in which the item is found. Boas (2010: 8) cites Petruck (1996: 2) as defining a frame as “a cognitive structuring device, parts of which are indexed by words associated with it in the service of understanding”. These collocational restrictions, it would appear, may be little different from conventionalised conversational inferences in determining what may or may not co-occur alongside a particular lexical item (as also implied in Fillmore 1988: 46), and are listed in Fillmore’s (1988) frame models as appearing in square brackets. For example, in the frame of the verb contribute are found its valence elements, as follows: (48) GF: SR: MS:

subject agent N

(object) patient N

[complement] recipient P [to]

suggesting that although it represents a conventionally transitive verb, the grammatical function (GF) of object and complement was optional, though present in the valence quota of the verb, entailing also the optionality of the semantic role (SR) of patient and recipient, and of the morphosyntactic (MS) items encoding them. Interestingly, Fillmore interprets the subject and object categories as grammatical functions, while Goldberg (1995) had re-categorised them as belonging to the syntax. Fillmore (1988) also refers to an “indefinite interpretation” to explain the optionality of the object and conversational “givenness” to explain the optionality of the complement. The justification for postulating a conversational implicature analysis for the composition of an event-frame is the element of world-knowledge. Boas (2005: 457) notes that such encyclopedic knowledge is essential for the interpretation of a construction, but it must be combined with crucial contextual information as well, and he maintains that the meaning of a verb is situated on a continuum with encyclopedic knowledge at one end and lexical knowledge at the other. Goldberg (1995: 29) did not overlook the element of world knowledge as essential to the productivity of examples such as (43); in fact, she qualifies its acceptability by adding that it is essential to know that sneezing involves a forceful expulsion of air, thus justifying the use of the verb sneeze in a caused-motion construction. Boas, however, has taken this criterion further, to describe it as significant “off-stage” information. Also implicit in the mini-constructions account is the use of analogy to determine the productivity of constructions. Goldberg (1995) had discussed polysemy and inheritance links to demonstrate the relationships between each of the argument structure construction types; such inheritance links were stipulated to be

52 

 Chapter 3 Construction grammars and the paradox

metaphorically governed.17 Boas, however, accentuates the inferential process of analogical reasoning in gauging the productivity of construction-types, obviating the need for a fusion of verb with construction semantics. In the end, it is only the verb senses that derive the construction senses (2005: 454), and novel verbs like sneeze are plausibly analogised to construction senses by their associations with verbs already conventionalised in such constructions (such as blow in (43)). However, even analogy fails on certain occasions, as Boas (2003: 113–116, 2008: 122–3) has shown in his examples of transitive verbs which, like those in (48), may appear with the implicit object omitted, i.e., what Goldberg (2000) had labelled the Deprofiled Object Construction, for example (cf. also Levin 1993: 213–217): (49) a. Pat ate is possible, but b. *Pat devoured is not. Boas resigns to the fact that it is not always easy to determine which verbs may appear in an independent construction, and which may not, and in the above case, there appear to be rules determining which verbs may allow loss of implicit arguments. However, in such cases, the implicitness of the object argument must surely be a factor of frequency of usage, since eat is likely to be more frequently used than devour. Perhaps also the idiosyncrasies of certain construction types may be due to historical factors, as devour is of Latinate or French origin, and likely to have been introduced into the language much later than eat, of Germanic origin. This need not suggest that verbs of Anglo-Saxon origin are to be treated differently from those entering the English lexicon at more recent stages of history, but accordingly, more recent synonyms may not be so conventionalised in their usage as older forms. This is one of the difficulties with dealing with English-only data, since English is historically a hybrid language. However, the factor of usagefrequency in creating new constructions by “chunking” (Bybee 2013) might also have a part to play in the creation of construction families, in which it is natural for frequently-used verb types to be among the first lexical domains into which constructions may propagate (see below). Interestingly, for all its value in providing a stronger definition of construction description, the proposal of mini-constructions (and its associated theories such as Croft’s (2003) verb-specific constructions, or Iwata’s (2002)

17 It is possible that some inheritance links may have diachronic significance, if (1) can be taken as an example of a shift for the verb banish from a ditransitive frame to a caused-motion one.



3.2 Questions of construction descriptions 

 53

distinctions of phrase-meaning from lexical-meaning) cannot avoid challenging somewhat the original foundations of Goldberg’s theories (what is now labelled Cognitive Construction Grammar (e.g., by Boas 2013)) in its requirements for a bottom-up approach (Boas 2008: 127) to the problems of composition. If constructions are usually considered to be unpredictable combinations of form and meaning, then the separation of form from meaning familiar to more analytical, ­lexical-constructional approaches needs further amplification. Boas (2008: 125) indeed goes so far to propose that there is a split between the syntax and the lexicon, and maps the lexical entries separately from the constructions in which they appear, quite a revolutionary move already at such an early stage of theoretical development. However, he does not discount the fact that there is interaction between the lexical entries and the constructions; one would go further to suggest, not just interaction, but dependency is needed in order to retain constructional integrity. Boas’ accounts are limited, though, in their data scope, mainly covering the area of English argument structure constructions (probably due to the fact that the mini-constructions were intended for such data in the first place), and mostly standard English examples.

3.2.3 Meaning A further problem is in defining meaning, which both Fried and Boas do face. For example, Boas (2013: 2) states that the term meaning “is understood to include all of the conventionalised aspects associated with a construction’s function … ”, offering as an example the fact that certain verbs can tolerate argument omission, as seen in (49a). Given such an approach in which lexical meaning appears to be regarded as separate from construction meaning, then this leaves only the grammatical function meaning as the meaning end of the form-meaning relationship. However, earlier, we have found that verb meaning in mini-constructions, for example, is on a continuum with lexical meaning at one end and encyclopedic knowledge at the other. Since a verb holds the property of expressing lexical meaning as well as fulfilling a grammatical function in a construction, ­ uestioned then, which aspect of meaning really is the dependent aspect in it is q the form-meaning relationship of a construction? It is this element of the construction definition which is crucial also to the labelling process, and to the consistency of the theory as a whole. Fried (to appear) reveals more inconsistencies in the diversification of constructional approaches, and attempts to resolve the problem of meaning headon, by suggesting (p. 9) that it is not a specific semantics which is defined in the composition of a construction, but their grammatical functional meaning.

54 

 Chapter 3 Construction grammars and the paradox

For example, the type of Modification construction she discusses [Adj N], consisting syntactically of a head noun and its (adjectival) modification, is claimed to have a meaning representation: “restrict reference of the noun by the property expressed by the modifier”. However, the meaning could just as well be represented as: “augment reference of the noun by the property expressed by the modifier”, depending on how the modification of nouns is regarded (the former description presupposes that by adding feature X, you are subtracting the possibility of alternatives). In addition, the explication of a grammatical function is something that Croft (2001), as discussed above, named as universal to constructions, suggesting that they are only universal in their grammatical functions and nothing more (hence the omission of a grammatical functional stratum in Croft’s form-meaning symbolic structures). If the meaning component were to derive from their grammatical function, then all constructions should be universal at this level, but according to Croft (2001) and Traugott (to appear: 3), they are language-specific entities. Naturally, it could be argued that it is only the grammatical function of, e.g., modification, that is universal, and that each language has different formal means of expressing it, thus ensuring that it is always the correlation between form and function that is language-specific. But that is not always the case, either, as the form [Adj N] is found in more than one language (if the notion of form could be represented by the terms adjective + noun). Such correlations leave only the phonological element of the formal pole as the languagespecific element (see below for further discussion). However, Fried (to appear: 10) uses this correlation as evidence for non-compositionality, demonstrating in another example, that the formal constituents (noun + finite verb) need not always serve the function of subject and predicate, and it is their combination in a Subject-Predicate construction which gives them the grammatical roles of subject and predicate. Plausibly, then, a subject cannot exist outside of a specific construction type, and this internal functional dependency is at the basis of Goldberg’s cognitive construction grammar: it is the syntactic relations which create the non-compositionality of a construction rather than their lexical input. However, Fried also states (p. 11) that non-compositionality in the narrow, propositional sense need not comprise a condition for construction candidacy. The functional-grammatical approach to meaning rejects the role of the lexicon in construction composition, as noted above, and it is only this aspect of construction composition that qualifies the construction for language-specificity. In the latter types, we are looking at constructions influenced by the lexical input of their arguments, and although Fried (to appear) did not discuss further argument structure constructions since she felt they represented only one type of construction, it is questionable why one type of construction should have a different interpretation of the meaning pole from those of others. Furthermore,



3.2 Questions of construction descriptions 

 55

in the grammatical functional approach to meaning, there is little likelihood of ­delimiting the productivity of a construction, since there is no recourse to lexical constraints to account for the compatibility of co-occurring elements. For example, while the construction a blue moon reveals incompatibility in the lexical relationship between modifier and noun, it is simply listed as an idiom, having a unitary lexical status, as Fried points out. However, in other cases, for example, (50) My trousers are broken or (51) That was [the sound of] my duvet hitting the wall18 the lexical input to the syntactic frame requires a closer association with similar, more frequently-occurring input items that usually occur in such constructions, that is, the existence of mini-constructions, and such objects may serve more functions than just argument-structure functions in accounting for analogical comparisons between conventionalised and non-conventionalised input items. Nevertheless, Fried goes on to suggest (p. 11) that there are several interpretations of the term meaning that are open to constructions grammarians, e.g., ­idiomatic lexical meaning, such as the rich, blue moon, etc., grammatical function or dependency such as modification, or speech act function such as politeness. Thus, we are back to where we started from in our quest for a more fine-grained interpretation of the meaning pole of the form-meaning relationship. Perhaps what has not been appreciated in most construction accounts so far is that meaning is a by-product of a developmental process in constructionalisation, or the grammaticalisation of constructions, which, as in most grammaticalisation accounts, explains that what started out as a lexical meaning may gradually acquire grammaticalfunctional meaning as it proceeds along the path towards becoming a grammatical item (see below). It is not clear how formal approaches to constructions may account for such metamorphosis in construction meaning, but it is clear that this is a shady area which has been left under-researched for far too long.

3.2.4 Form-meaning alignment in other accounts In a recent study, Langacker (2005a) also paid homage to Goldberg’s early innovations, while at the same time emphasising that Cognitive Grammar (Langacker

18 (50) and (51) are actual examples (observed personally as uttered by a Singaporean speaker), (51) referring to the buttons on a duvet.

56 

 Chapter 3 Construction grammars and the paradox

1987; 1991) had always proposed a construction-type approach of one kind or another, in which all language could be expressed as symbolic form-meaning relationships too. In the Langackerian approach to constructions, the presence of a grammatical layer of form is decidely absent: the only poles of the formmeaning relationship are found in phonology and meaning, not grammatical form and meaning. Such an account is strengthened by the fact that grammatical structure, as shown in grammaticalisation accounts, is always seen to be moving or unstable (or emerging, as Hopper (1987) would have it), and that the margins between what is a grammatical construction and a lexical one are often blurred, as noted above. Thus, Langacker’s symbolic structures are also pairings, composed of a semantic structure and a phonological one, and are relevant to the changing status of form-meaning relationships across time. He notes (2005a: 105–6) that grammar is not distinct from semantics but is incorporated in the semantic pole. In all other construction accounts, he observes, it was usually the syntax or the morphosyntax that was paired with semantic structure and was the primary object of linguistic investigation under a construction-based analysis; in Croft’s Radical Construction Grammar, phonological structure is aligned with grammatical structure as just one component of the form end of the form-meaning polarity, while in Goldberg’s constructions, phonology has no particular role to play at all. However, for Langacker, there is no other form apart from phonological form (and its written expression, if applicable). He particularly emphasises his departure from all other accounts of constructions and construction grammars, by assigning grammatical structure to be schematic over phonological form and semantic structure (2005a: 106). He goes on to raise the question to what extent a noun categorisation is a matter of form, for example, when speakers do not carry in their speech production any such category labels or have any idea of grammatical classes or membership?19 In his (2005b) study, Langacker continues further to suggest (p. 160) that the use of such terms makes sense only in the case of grammatical form, which is considered then as existing independently of semantic and phonological structure – presupposing an autonymous syntax, something that most constructionists have tried to dispense with. Speakers know only the objectives of expressing meaning, and the phonological means by which it can be done, both phonological form and semantic structures being linked to form symbolic structures in Langacker’s terms (2005a: 112). Semantic structure, on the other hand, is represented in Langacker’s account as everything that is not formal in the sense of phonological.

19 In Langacker’s (1987) study, nouns and verbs are shown to share similar crucial semantic features, and in fact are considered in parallel in many ways.

3.2 Questions of construction descriptions 



 57

The form-meaning relationships in Langacker’s cognitive grammar are thus compared with Croft’s Radical Construction symbolism, and Goldberg’s constructions, in the diagrammatic models, adapted from Langacker (2005a: 5) in Figure 1. Symbolic structure

Symbolic structure

Semantic structure

Semantic structure Grammatical form

Phonological structure

Phonological structure

Fig. 1a: Langacker’s Cognitive Grammar.

Sem CAUSE-MOVE




Fig. 1c: Goldbergʼs Construction Grammar: composite fused structure Caused Motion + put (Goldberg 1995: 52). Fig. 1: Three principle construction types showing form-meaning alignment differences.

It is seen that while Langacker’s Cognitive Grammar constructions only pair phonological structure with semantic structure, Croft’s (2001) model includes a diversification of the form pole of the symbolic structure, pairing both phonology and grammatical structure with semantic structure. It might also be noted that Radical Construction Grammar includes in grammatical form both morphological properties as well as syntactic properties, and that the semantic pole is restricted to semantic properties, pragmatic properties and discourse-functional properties. Traugott (2014: 89), after Croft (2001), also extends the form pole of a construction to refer to syntax, morphology or phonology, and the meaning pole to refer to semantics, pragmatics or discourse function. In neither of these two models is there a level for grammatical functional properties, though Fried (to appear), as noted above, has included this at the semantic pole. Problematic to such descriptions is the fact that grammatical function is neither a semantic

58 

 Chapter 3 Construction grammars and the paradox

characteristic nor a formal one, as discussed above. Goldberg’s typical models, as exemplified in the composite fused structure in Figure 1c, show only an opposition between the syntax as the formal component of a construction, and the semantics, revealing that the original function of argument structure constructions was to demonstrate the initial claims for syntax-generated semantics. In her earlier studies, the lexical semantics are not accounted for at all, and the semantic pole indicates only the syntactic relational roles of the lexical items in the construction concerned. However, more recently, Goldberg (to appear) has simplified the diagrammatic model of many of the earlier constructions to account for form-function oppositions, rather than form-meaning ones. The Caused Motion Construction is now represented as in Figure 1d: CAUSED-MOTION CONSTRUCTION Form: Function:

V

{NP

cause-move (causee,

PP} path)

Fig. 1d: Goldbergʼs (to appear) Caused Motion construction.

Goldberg does not elucidate on the differences between the two models, except to say that category labels are used (NP) rather than relationship labels such as OBJ, in order to maintain consistency with current discussion on the verb-particle construction, the topic of the study. There is no explanation as to the use of terms such as function introduced into the study, nor why the term Sem(antics) is no longer used, nor Syn(tax). However, this is a recent descriptive change which, to present knowledge, has not been used elsewhere, nor is it used consistently by Goldberg throughout the same study either. Langacker’s Construction Grammar may only be criticised only for the fact that it does not entertain constructions which are not entrenched or highly conventionalised (e.g., 2005b: 159). No construction theory should account for only well-formed expressions in a language – a construction theory ought to explain anything that is uttered, as long as it is meaningful. In terms of contact varieties of languages, it is clear that a specific definition of a construction is needed, especially in the face of situations in which the phonological form derives from the lexifier language and the syntactic form, frequently, derives from a model language or substrate. The problems of definition of constructions are noted above, and we shall see the significance of such problems as the present study progresses. However, for the moment, there are further questions with regard to construction-types according to the Langackerian model: if the lexifier (or source language) provides the model for the phonological form, is this really a lexifier

3.3 Construction(al)isation 

 59

after all where constructions are concerned? Phonological form expresses both lexical meaning and grammatical meaning, so what is meant by the capacity to lexify a replica language? Given what is known about the inconsistencies in formmeaning relationships between contact varieties of languages and non-contact ones, the better term to use in the case of the source language is the phonologiser language. On the other hand, if syntax is not a formal structure, as Langacker suggests, and is part of the semantic structure, then a contact replica language will have identical phonological forms with (sometimes) unidentical semantic structures with regard to their source language counterparts. The question then arises which model is most suitable to adopt in the case of contact languages, and it may appear that replica languages share some constructions with their lexifiers but not others. It is considered that Langacker’s approach to constructions as phonological-semantic symbolic structures is the one most suitable to adopt in the case of contact constructions, since a formmeaning relationship in which form equals syntax, or grammatical categories, is one which will inevitably lead to positing multiple constructicons for every variety of a language possible, even when the language shares the same phonological form. If so, the replica language would not be a variety of the source language, but a new language in its own right. Such an analysis would lead to hopeless proliferation of languages, since if according to Hudson (1986: 24) a variety is determined by its variation alone, that need only be represented by a single feature such as the second vowel in the pronunciation of the word tomato, or the epistemic vs. deontic use of negative must as in (42). If form-meaning correspondences determine constructions, and constructions are language-specific, then every time a feature of variation (of form or meaning) emerged in a language, there would be a new language. In order to deal with this problem, it will be necessary to describe constructions in a different way, accounting for variation in such a way that subtle changes of variation do not lead to the establishment of a new language based on the form-meaning definition of constructions. The description cannot be in the region of form, but in meaning alone. In later chapters, it will be seen how best to handle this situation. In the meantime, it is worth also considering a new field of construction theory has recently begun to take shape, and that is the field of diachronic construction grammar, or construction(al)isation.

3.3 Construction(al)isation One of the first accounts to apply attention to the problems of the difference between lexical constructional meaning and grammatical constructional meaning was that of Traugott (2007) and later, Trousdale (2010). Traugott has

60 

 Chapter 3 Construction grammars and the paradox

since revised many of the ideas first appearing in her (2007) paper, though the essential claims remain intact, and they are principally, that construction ­development and grammaticalisation have much in common, and in fact, complement each other (Traugott 2008: 240). Traugott (to appear) also specifically distinguishes lexical constructions from grammatical constructions, which could appear problematic for some cases of grammaticalisation, when a developing form is considered intermediate between having lexical status and grammatical status. In her earlier work, she discusses what can be labelled, for present purposes, phrasal grammaticalisation (as opposed to morphemic grammaticalisation), in which a particular grammatical function comes to be expressed by a complex, phrasal item, such as [NP of NP] in a shred of X, a lot of X etc. in the development of degree modifiers in English. Such cases provided ideal examples to illustrate what could be addressed simultaneously as construction development as well as grammatical development, because of their ultimate loss of compositional analyticity of the [NP of NP] syntagm in acquiring quantifier functions. Traugott (2007: 525) divided the construction inventory into three typical chronological stages of development, ranked by levels of generality: (52) a. Macro-constructions (high-level construction schemas, e.g., ­ditransitives, degree-modifier constructions), which may contain several sub-types; b. Meso-constructions (similarly-behaving types expressing ­grammatically-distinct functions, e.g., a bit of, distinct from sort of ) c. Micro-constructions (individual examples of the previous level, such as a lot of); d.  Constructs (actual tokens of micro-constructions). The last three levels can be summarised as items that converge to the general patterns appearing at the macro-level, and the definition of constructs is no different from that represented in most other accounts since then. The objective in her study was to illustrate that constructions develop diachronically by continuously assimilating with other constructions at greater and greater levels of semantic schematicity, e.g., at the micro-level, constructs such as a bit came to co-occur with unbounded nouns, while a bit of came to co-occur with both bounded and unbounded nouns. As the quantifier became more generalised, it became ­analogised at the meso-level to assimilate with degree adverbs such as very, modifiying adjectives as well (Traugott 2007: 541). Later developments included analogisation of other degree modifiers to patterns exhibited by the new constructs, and expansion of the entire group encompassed by the macro-degree modifier ­construction (2007: 541).

3.3 Construction(al)isation 

 61

The levels of construction-types (e.g., macro-, meso- and ­micro-constructions) listed originally by Traugott (2007; 2008) and now adopted by Trousdale (e.g., 2010, 2012), following her, appear to be equally visible as clusterings to which different construction types may gravitate at different times of development as stages along one continuous construction-development. At each stage, there are changes in the accompanying nuances of both form and meaning which could classify one construction-type as identifiably different from another at a different historical period; for example, in the meaning changes of the degree modifier, sort of in English (adapted from Traugott 2008: 237), the time periods shown as illustrated: (53)   Stage 1 (1400) Grammatical PreFunction partitive Meaning

2 (1600) Partitive

3 (1700) Degree modifier

3a (1800) 4 (1900) Degree Degree adverb adjunct

class of member of downtoner/ hedge container contained approximator

tag, response

The examples of degree modifiers in English illustrated by Traugott (2007) and (2008) clearly demonstrated that the foundations for a combined approach to grammaticalisation and constructionalisation were in evidence. Traugott later revised some of the terminology used, replacing the terms macro- and mesoconstructions with simply schemas and sub-schemas, whilst retaining the term micro-constructions, e.g., the Cause-not-receive subschema of the ditransitive construction schema may be realised with a micro-construction such as deny somebody something (to appear: 4). Important to her development of the theoretical premise is the fact that (a) she was able to illustrate clearly the overlap between constructionalisation and grammaticalisation (something that had been alluded to many times previously (e.g., by Bybee, Perkins and Pagliuca 1994: 11; Lehmann 1992: 406; Himmelmann 2004: 31), as noted by Traugott (2008: 221), but never adopted as a theoretical model, and (b) she distinguished between grammatical constructions and lexical constructions, the former having mainly relational semantics and the latter mainly referential semantics.20 Nevertheless, in the constructicon, she claimed for a gradient between the two construction types (Traugott, to appear: 4).

20 Traugott (2012b) also refers to Icxns – intermediate constructions such as the adverb luckily and the way-construction which are mid-way between being lexical constructions and grammatical constructions.

62 

 Chapter 3 Construction grammars and the paradox

Traugott’s construction model was built on Croft’s (2001) Radical C ­ onstruction Grammar, with form-meaning structures split between on the one hand, syntax, morphology and phonology (for form), and on the other, semantics, pragmatics, and discourse function (for meaning). Like Croft, she does not distinguish ­grammatical function meaning from lexical meaning, perhaps because of her ­diachronic approach in which lexical items are seen to develop gradually into grammatical constructions; it is the lexical meaning which becomes eventually the source for grammatical function meaning.21 For a diachronic approach, the blurring of distinctions in such a way is not unusual. However, in a synchronic construction grammar, it creates problems, as discussed above. It will be seen later in the present volume what consequences this has for discussing constructions in mixed or contact languages. In her (2008) study, Traugott illustrates (p. 237) meticulously the stages attained in the constructionalisation of the degree modifiers a sort of, a lot of, and a shred of in English, using the framework proposed by Croft (2001), and demonstrating which aspects of form and/or meaning were involved in the reanalysis [NP1 [of NP2]] to [[NP1 of] NP2] at which time period in history, and which elements of the symbolic pairings illustrated by Croft (2001) are exemplified at which stage. The presentation articulates lucidly Croft’s abstract model, but one question arises from her (2008) study and other similar historical accounts: while conceptual structure and pragmatic levels of a construction are held to be universal, it was also claimed that syntax and semantics are languageand period-specific (Traugott 2007: 525). With this in mind, it may be questioned whether each diachronic stage of development is a sub-stage of the same construction, or a different one at each stage. In consideration of the developmental pathways shown in many of Traugott’s studies, it would seem that we are looking at an emerging, single construction-type. However, if the syntax and semantics of a construction are period-specific (and in this respect, Traugott accords with Croft in distinguishing semantics from pragmatics and discourse functional meaning), then the construction type is different at every historical stage of its development. In Traugott (2012a; to appear), the present-day be going to construction was discussed, and did not appear to be identifiable as such until the early eighteenth century, when inanimate subjects incapable of motion were beginning to be associated with the auxiliary (to appear, p. 23). Clearly at such stages, there is a distinct shift in the semantics along with a syntactic reanalysis, substantiating the claims that the syntax and semantics are period-specific. Traugott (2012; to appear), in fact, questions whether be going to originated in a construction, and

21 Traugott (to appear: 11) stresses, however, that it is not the source, but the outcome, which is of importance in considering constructionalisation, and the result may be lexical or ­grammatical.

3.3 Construction(al)isation 

 63

provides evidence to suggest that it began in the form of the construct go (as a lexical source), unifying with a pool of other construction types, such as the preprogressive, purpose markers, and the passive. It was in the contextual environment of these three construction types that the present-day future auxiliary first began to be identifiable as a distinct construction of its own. Thus, also necessary for the development of a construction were extensions to environments which were considered to be incompatible with the original lexical meaning of the construct; these are typical of any grammaticalisation process, and have been described in Heine (2002) as switch-contexts (as noted by Traugott), but mistakenly related to features of mis-match by constructionists working in non-diachronic approaches (e.g., Michaelis 2004), as pointed out in Ziegeler (2007). Thus, the answer to the question of the history of a present-day construction is that it is not the same construction as in earlier periods, but reveals a life history in which it interacts with other constructions to eventually develop into the form-meaning pairing associated with it today. With regard to terminology and identification of a construction, the diachronic approach is clear in its illustration of the contexts in which a construction-type may emerge. In grammaticalisation terms, a construction emergence corresponds with a grammaticalisation: the construction is there when the function it is seen to serve is first observed historically, and non-compositionality is a fact of construction-creation just as bleaching is of grammaticalisation accounts. Trousdale (2008a, b, 2010 and elsewhere) also discusses the overlap between the development of constructions and grammaticalisation, emphasising as part of his definition the characteristic of frequency of use in determining what is and what is not a construction (e.g., 2010: 53). The element of usage frequency is, however, only secondary to the fact that the identifying characteristics of constructions are the common form-meaning behavioural patterns they share. What is common to grammaticalisation and constructions in general, as he points out, is that the results of grammaticalisation and the identification of a construction are always schematic objects (ranged over a selection of lexical structures). Trousdale (2010: 54) thus rejects as peripheral to the identification of a construction the issue of compositionality or non-compositionality. In doing so, he is acknowledging, as does Traugott (to appear), that it is the outcome of constructionalisation that is most important. In many ways, then, the compositionality of a construction (such as the way- construction, in (54) I knitted my way across the Atlantic (Goldberg 1995: 213) is trivial, if not sometimes obscure. However, in other ways, the compositionality is crucial to the productivity of the constructions, since, as also acknowledged by Trousdale (2010), and by Traugott (to appear), ­construction

64 

 Chapter 3 Construction grammars and the paradox

productivity calls upon the process of analogy in order to diffuse through the range of lexical input required to substantiate the syntactic meaning. In order for (54) to be a felicitous form-meaning correspondence, the verb expresses a meaning of manner analogising to other manner verbs in the same construction.22 ­ rovided Trousdale’s earlier work was significant in that, for the first time, it p an investigation into argument structure constructions from a diachronic point of view, and arrived at a succinct account of the grammaticalisation of a construction type over time. Trousdale (2008a, b and 2010) adopt the diachronic ­construction level types listed by Traugott (2007), portraying the transitive construction (a construction type discussed little by Goldberg 1995), as a ­macro-construction, or superordinate kind, which embraces what could still be described nowadays as non-prototypical transitives, such as (55) a.  The tent sleeps six and b.  The hotel forbids dogs (Trousdale 2008b: 313), and illustrating with such examples that they are not translatable as transitives in languages such as German, a language still marked with a complex morphological case system. The grammaticalisation of transitive constructions over time since Old English is seen to first affect individual constructs formerly with dative subjects, e.g., (56) hu him se sige gelicade (‘how the victory pleased him’, lit. ‘how to-him the victory pleased’) (Trousdale 2010: 56), moving to micro-constructions of the same type (i.e., others using the same case-marking morphology and specific verb types), then to mesoconstructions; for example, the category of Impersonal transitives versus Personal transitive constructions). In general, however, there is a gradual increase in the productivity of the transitive construction from the time of the loss of morphological case, when the transitive construction arose mainly through the effects of word order regulation as a generalisation over other construction types. Alongside the impersonal constructions, Trousdale also observes the appearance of a construction type in Old English which was marked by a nominative subject and an accusative (direct) object, e.g., (2008b: 307): (57) he 3M-Nom.SG

acwealde kill-1/3S.PST

þone the-ACC

dracan dragon-ACC

‘he killed the dragon’ (Ælfric Homilies (Supp.), XXL, 455) 22 Goldberg notes that in this particular case the action expressed by the verb (in (54) is, unlike other verbs in the same construction, not a means of self-propulsion.

3.3 Construction(al)isation 

 65

He notes such types as mainly marked by telicity, volitionality, and expressing actions, and thereby considers the case-marking feature as a means of organising grammatical material to accord with symbolic conceptual content (2008b: 307). This construction type which was associated with the Personal construction type as opposed to the Impersonal type became more productive in the history of transitivity, thus providing the model for the development of transitive constructions as we know them today. Although Trousdale is discussing morphological case in this particular instance, the patterns were already in place for the assimilation of argument structure word order to specific semantic role information, even as early as Old English times. Thus it is no mystery that synchronic observations, such as Goldberg (1995), reveal the persistence of such conceptual symbolism in the semantics of the arguments contained in them. It may well be asked, though, how the syntactic arrangement of arguments in a construction can give rise to the observation that it is the syntax which is governing the semantics of the construction. But such an effect is grounded in history, as the regularisation of word order in English, a process taking place gradually over four hundred years from Old English times until 1500 (Hopper and Traugott 2003: 67), was not unmotivated: it was in fact taking over the functions of the former case morphology in expressing roles and relations. The same process has been noted in the history of English by Croft (2000: 124) in the encoding of passive subjects in the nominative case, a case typically reserved for subjects, when they had formerly been encoded in the dative case. The conceptual residue of such sentence relations once grammaticalised in affixes on verb arguments then continues to adhere to the argument positions alone at later stages, regardless of what lexical items occupy those positions; as noted by Trousdale (2010: 67), “… argument structure is implicated in the change from lexical to structural case assignment”. The recognition of a schematic semantics for such grammatical constructions enables the generalisation of role relations to apply to whatever lexical constructions may interact with them. The work of Traugott and Trousdale thus marked a milestone in the fact that it articulated clearly a number of earlier allusions to the possibility of grammaticalisation in or through construction-types. However, it was not without its critics, as shown, for example, by Noël, who caustically complained (2007: 178) that the advent of constructional approaches marked the “… human need to find strength in numbers by banding together behind a standard”, and that “… construction grammar … [might] indeed turn into a paradigm within which most ­non-generativists will be able to congregate”. In such criticisms, he was implying that the functional-cognitivist field had, until the construction revolution, lacked a solid theoretical base of its own, thus presumably suggesting that the new theoretical status of construction grammar should not necessarily replace the earlier, more established foundations of traditional grammaticalisation

66 

 Chapter 3 Construction grammars and the paradox

(“… what is there to stop grammaticalization theorists from joining the growing army of ­ construction grammarians?” – 2007: 178). Behind such an attack, though, Noël raises a number of questions, amongst them: how can constructions, which are already grammatical units, become more grammatical? Also, if grammaticalisation involves a semantic change to more grammatical meaning, which meanings can be counted as grammatical; in other words, at what point does the construction emerge as part of the grammar? (2007: 189). Similar questions have been raised above, with respect to the identification of a construction, though not within a diachronic context. Both questions, however, show evidence of a little misapprehension. In the first place, construction-development is not limited to the development of grammatical constructions, as pointed out by Trousdale (2012: 193) in response (while maintaining that similar processes are also involved in lexicalisation), and in the second place, even in grammaticalisation, there is not always a readily perceptible cut-off point between the lexical status and the grammatical status of a grammaticalising item, as emphasised continuously in the classical sources, such as Heine (1992, 1993), Heine, Claudi & Hünnemeyer (1991), Hopper (1991), and Hopper & Traugott (1993[2003]). So why should construction grammaticalisation be any different? Trousdale also maintains (2012: 193) that grammaticalisation of schematic constructions is an observable fact, and requires a more explicit account than grammaticalisation theory alone can give it. Noël (2007: 193–194), citing Schøsler’s (2005) study of the development of the French dative over 1800 years since Latin, makes a case for the rejection of developmental accounts of argument structure grammaticalisation, claiming that the end-result of such constructions is a morphosyntactic patterning essentially void of meaning, leading to a grammaticalisation interpretation rather than a construction account which would entail the presence of a meaning aspect. He protests that the attribution of a constructionalisation account to such historical developments is an exaggeration on the part of the authors who wish to apply semantic significance to syntactic configuration (2007: 195), and that what is needed is a closer investigation of, for example, the semantic roles of the arguments in such constructions. However, in English, such roles are not marked morphosyntactically on lexical arguments, so it is a moot point whether the source for the semantics of argument structure constructions is a sum of the word-order rules, the semantic roles played by the participants, or both. In any case, grammaticalisation does not presuppose a complete emptying of meaning, as Traugott (1988) established, but a replacement of semantic meaning often with pragmatic inferential development. From the contrastive data from Singapore Colloquial English, as we will see later in the volume, it is also clear that argument structure constructions in some varieties



3.4 Construction development and coercion 

 67

are not entirely empty of all meaning, as in examples such as the following, taken from Singaporean advertising material (Low and Brown 2003: 66): (58) Polish your car while you shop. an action which would be impossible to comprehend in standard varieties. The transitive construction in Singapore English is seen to conflict with an o ­ ver-extension of the paradigm of what Goldberg (1995) labelled conventionalised scenarios, an indirect causative construction in which the intervening causee is omitted from the construction. These constructions will be discussed further in Chapter 4, but it is clear that the infelicity of many such examples points to an underlying transitive semantics that is still apparent and associated with the syntactic configurations of argument structure constructions as a form of prototype retention.23 This type of prototype retention is essential to the understanding of another theoretical issue in construction semantics: that of mismatch and coercion.

3.4 Construction development and coercion The development of constructions has been sometimes associated with the notion of coercion, which (as indicated in Ziegeler 2007) amounts to little more than a form of metonymy operating at the syntax-semantics interface. One of the main reasons for scrutinising the notion of coercion so closely is for what it reveals about the nature of the prototype semantics of constructions, as well as the view of constructions as supplying semantics from their surface syntactic structure. The latter view is challengeable, as noted earlier, in the face of evidence related to contact dialects such as Singaporean English, which do not require the same coercion accounts to describe their construction data (see Chapter 7). Coercion does not appear to require any particular status independent of other, ordinary pragmatic processes of language change, and may even be considered illustrative of some of the processes of grammaticalisation in construction-development. Coercion is described as a mechanism of contextual reinterpretation “triggered by the need to resolve [semantic] conflicts” (Michaelis (2004: 2), citing De Swart (1998: 360)), or an enriched representation

23 The role of retention in grammaticalisation has been discussed in Ziegeler (2007) and will be mentioned again below. In the case of prototype retention, there is no single lexeme to which the retained semantics may be traced historically, but often a prototype construction, as with the transitive, or a semantic nuance associated with roles or relations in earlier constructions (such as agentive associations with the subject argument).

68 

 Chapter 3 Construction grammars and the paradox

produced by the means of reconciling implicit type-shifting, an inferential procedure used to bridge semantic indiscrepancies in the morphosyntax (Michaelis 2005: 47). It is obvious from such studies that this particular type of “coercion”, i.e., the resolution of semantic-syntactic mismatch, is intrinsic to constructions, as earlier studies, e.g., Michaelis (2004; 2003), Francis and Michaelis (2003) have demonstrated, though Lauwers (2008: 166) points to a number of studies in which the same types of mismatch are found between items of morphology, semantics and syntax, or between category and function, the latter being most important to his study of French nominalised adjectives. It was seen in Ziegeler (2007 and 2010) that the phenomenon of coercion could be interpreted in a number of different ways, depending on how the term was used; for example, the case of aspectual coercion. The English progressive (aspect) construction was one example which was discussed at length in that study following a review of Michaelis’ (2004) treatment of this as an aspectual type-shifting process. The main problems that seemed to arise from Michaelis’ account were that the progressive construction represented two conflicting descriptions of coercion that seemed to be irreconcilable. In the first instance, Michaelis (2004: 7) had defined aspectual coercion as taking place when “… the interpreter must reconcile the meaning of a morphosyntactic construction with the meaning of a lexical filler” (2004:7).24 Examples of type shifting in the progressive included the following (Michaelis 2004: 36): (59) a.  I’m liking your explanation. b.  Right now she’s believing there’s going to be a reconciliation. (59) was used to illustrate a type of coercion resulting from the distributional constraints of the progressive which require a non-stative main verb, and believe and like are stative. In the case of the progressive, the coercion apparently perceived can be regarded as nothing more than the reflection of historical changes which had taken place in the diachronic development of the progressive aspect: it began as an aspectual marker of imperfectivity on durative verbs and in time-stable situations (including generic situations), and it was not until after 1600 that a notable rise in the number of transitive, non-durative predicates was observed (see ­Ziegeler 1999). Accomplishments and Achievement verb classes are also claimed by Michaelis (2004) to be in a mismatch relation with the stative form of the head, the stative auxiliary. The picture becomes very confusing when the historical facts of the development of the progressive are not taken into consideration: only Activity verb classes do not type-shift and Activity types, being prototypically durative, were amongst the verb classes with which the progressive originally combined (which did, incidentally, allow for stative verbs too). Thus, if any coercion can be claimed to have

24 This definition does not necessarily presuppose mismatch as a trigger for coercion, though it is understood as such elsewhere.



3.4 Construction development and coercion 

 69

taken place, it must have happened as a historical change, when lexical verbs other than durative, imperfective lexical classes began to be admitted into the paradigm. The frequent correlation of ­diachronic change with mismatch has been discussed in earlier works (e.g., ­Faarlund (2004) and also Francis and Yuasa (2008)). A number of other cases of coercion have been described in the literature, all, or nearly all of them pertaining more closely to grammatical or cross-categorial metonymies (see Ziegeler 2007 for more examples). Even the prototypical argument structure constructions of Goldberg (1995) have been associated with the notion of coercion, and noted as becoming productive through the introduction of new arguments, a process Goldberg (1995: 157–9) prefers to term accommodation, e.g., the conversion of a locative prepositional complement (as in inside the jar) into a directional complement when combined with verbs of motion, such as Sam squeezed the rubber ball inside the jar. These examples can be linked to an explanation of metonymical associations as the grounds for productivity, in which a preposition expressing location is extended to stand for not only location but direction towards the location as well. It is clear from such examples, particularly those involving transfer and ditransitive arguments, that in spite of the apparent influence of the construction syntax in contributing to the extent of the construction productivity, there are anomalous combinations, e.g., in (46), repeated below as (60), (60) ??Sam blinked the napkin off the table which, although it bears the same syntactic template as a caused-motion construction, cannot under any plausible analysis be made semantically acceptable as, unlike all the other verbs that may occur in such constructions, the predicate does not share with them nor does it have the potential to share with them the semantic sub-feature of dynamic transfer. In such examples, there is no possibility of coercing a meaning at all, and the construction syntax cannot make for a plausible accommodation of the arguments. Clearly, the identification of a construction syntax providing the semantics for the generation and extension of the construction is impossible without resorting to what may be referred to as a Lexical Prototype Construction (Ziegeler 2010), which carries a lexical blue-print that may be generalized to enable future productivity. The Lexical Prototype Construction explains how the semantics of ostensibly incompatible lexical entries may be so effortlessly subordinated to the syntactic semantics on the point of unification, and why other lexical entries fail to be coerced, or accommodated, as in (60). From a construction(al)isation perspective, though, the semantics of a construction are readily accounted for by the lexical input of the source constructions from which the present-day construction types have emerged over time (as in the case of the progressive aspect construction, once only compatible with durative, imperfective, Activity verb types). Thus, from a synchronic point of view, there is no call for an independent construction semantics, the semantics are in effect, the diachronic legacy of the construction source elements.

70 

 Chapter 3 Construction grammars and the paradox

3.4.1 Cyclical interaction In order to accommodate such extension and generalisation within a grammaticalisation account of constructions, it is necessary to consider a two-way model of development, which involves the interaction of both metonymy and metaphor in propelling the construction onward and into new lexical domains. It was proposed in Ziegeler (2010) that constructions may expand their range of uses through diffusion into new lexical domains.25 The situation seems to entail that constructions have semantic associations with the Lexical Prototype, or are extensions of the Lexical Prototype (characteristically in the form of family resemblances – see, e.g., Rosch (1978) and will proceed to take on all lexical combinatory possibilities in their path, as they advance to greater and greater extents of generalisation. Anomalies, when not correlated to alternative construction types, will be rejected on the basis of their inability to share a relationship with the Lexical Prototype, e.g., (60). The cyclical pattern of extension and generalisation will continue in the same way, as the construction, in turn, gradually absorbs its semantics from the (novel) lexical meanings of the source items with which it is most often correlated, and then, goes on to determine the nature of the lexical items with which it combines, placing restrictions on its own productivity. The propagation of the construction through the lexicon may be represented, then, as a cycle, in which distributional semantic effects feed back into the prototype, as in Figure 2 (fr. Ziegeler 2010): Prototype determination

Lexically-constrained constructionisation

Lexically-constrained constructionisation

Distributional generalisation Fig. 2: Cyclical interaction between generalisation and lexical constraints in constructionisation.

25 The term diffusion is used in the spirit of the term lexical diffusion, which describes the way phonological change infiltrates through the lexicon, as noted by Bybee (1998). Bybee (1998) has shown also that faster diffusion is often the result of more frequently-used lexical items.



3.4 Construction development and coercion 

 71

It is thus construction generalisation that gives rise to distributional semantic effects over the lexical fields, which when combined with the generalising construction, will contribute to the main schematic elements of grammatical meaning (e.g., transfer in ditransitives and other argument structure constructions). Such meanings then control unification in further distribution.26 This model of cyclical construction generalisation crucially involves the speaker’s access to the semantics of the Lexical Prototype Construction as a mechanism of retrieval based on identification of the semantics of the extended construction type by its later distributional possibilities through the lexicon. ­Goldberg (2009a: 101) also notes the presence of a semantic prototype in the development of constructions, as do Ruiz de Mendoza and Usón (2008). Croft (2003) discusses Verb-Specific Constructions, and Boas (2003) also refers to the presence of mini-constructions, as noted earlier. Nevertheless, the disagreement between the verb-specific approaches and the syntax-governed approaches to the analysis of argument structure meaning goes on, and Noël (2007) also takes the side of the proponents of verb-meaning contributions, relying solely on data illustrating the analogical correspondences of verbs occurring in the argument structures of the historical examples of French datives. He believes that the syntactic patterning accompanying such structures took its meaning from a verb frequently occurring in it, and the pattern was later to extend to verbs with a similar meaning (2007: 193). The theoretical premises of Boas’ mini-constructions and Croft’s ­verb-specific constructions predict the same process of development. However, in the end, the problem is perceivable as a chicken-and-egg situation: the structural frame must have been there in the first place, associated with such verbs, for it to grammaticalise to increasingly wider lexical fields. It is for this reason that it has been proposed instead that argument structure construction types not only derive their semantics from the lexical items with which they most frequently combine, but also diffuse through the lexicon, so determining in a cyclical manner the nature of the lexical items that may combine with them. The process of the development of construction semantics is therefore a two-way process, built on the interchange between the lexical nature of the combining verbs, and the syntactic frame which takes its semantics from the verb senses combining with it, and then, inexorably grammaticalises to include a broader lexical range. The process is similar to

26 It should be noted at this stage that not all constructions generalize in the way demonstrated in Figure 2, and the model need not be treated as a strictly diachronic one. For example, the quantifying expression, a deal (of) + NP is relatively restricted in distribution compared with a great deal of + NP, and even more so compared with a lot of + NP (see, e.g., Traugott 2007). Until researched further, then, the developments shown on Figure 2 must therefore be limited initially to a somewhat conservative application.

72 

 Chapter 3 Construction grammars and the paradox

that of hyperanalysis (see Croft 2000: 121, 131), in which the syntax may absorb meaning from the semantic context with which it interacts, and in some cases, then jettison the item from which it took its semantics (e.g., present-day French ne V pas (negative marking) > V + pas). The fact remains that constructions do not propagate through the lexicon haphazardly, but seek out contiguities with the lexical predicate types which they subsume. Thus, there is ample justification to admit that construction semantics are simply an abstract, epiphenomenal variation of the nuclear lexical components from which the construction was originally composed. This argument will be seen to form the basis of the theoretical position taken in Chapter 8, since it is the lexical composition of constructions which is most readily accessible and identifiable in situations of language contact.

3.5 Summarising the current position It has been shown above that the solution to the problem of ­construction-definition is still a little obscure, especially if diachronic accounts are to be taken into consideration. On the one hand, the classic definition of a construction is a formmeaning relationship, and as we have seen, there are differences of opinion on what constitutes form and meaning. It was questioned from the studies on diachronic development of constructions whether we are looking at the same construction through time, or a series of different ones. Similarly, it will be questioned in subsequent chapters whether we are looking at one construction across different dialects or at several variations of the same construction. Given that Traugott (2007: 525) had claimed for the period-specific characterisation of syntax and semantics, the possibility of maintaining a consistent construction-type through diachronic periods is therefore difficult to uphold. However, in the equally transitory diffusion of a Lexical Prototype Construction through the lexicon, as seen in the previous section, it is not diachronic development we are looking at, but a more immediate generalisation of a construction type. The two positions are difficult to reconcile: the Traugott and Trousdale construction(al)isation, which explains grammaticalisation of phrasal chunks as construction-types, and the construction(al)isation-as-generalisation account described above, in which construction spread is based not so much on the creation of grammatical construction material, but the increased range of distribution of an already-established prototype construction. What they do have in common, however, is an increase in schematicity, in that the greater the generalisation, the more abstract their semantics become, exactly as in classical grammaticalisation studies. It could be argued that there are both diachronic and



3.5 Summarising the current position 

 73

synchronic dimensions to construction(al)isation: diachronic as the formation of a construction, and synchronic as its generalisation through the lexicon, which in turn, reinforces construction-based semantics. From such viewpoints, the distributional cycle in Figure 2 could be taken as a later development of construction grammaticalisation itself. As noted earlier, the nature of meaning in Traugott’s account, as well as in Croft (2001) (on which it was first modelled), was intended to encompass all kinds of meaning fields including pragmatics, and discourse function. The form pole includes grammatical functional categories under a Langackerian account, such as noun, adverb etc., but also shifts in the pragmatic weighting of the categories themselves (reanalysis of head nouns where [NP1 of [NP2]] becomes reinterpreted as [[NP1 of] NP2]. Traugott (to appear) describes constructionalisation as the emergence of a new construction, therefore it is unlikely that whatever form-meaning pairs had co-existed that gave rise to the new construction would ever be considered to function as the same construction as the one which develops historically. And yet, they form the network of earlier constructions that contributed to the pairing of form and meaning identifying the construction in current usage. What is of essence in the entire approach is that grammaticalisation can be viewed as the development of a syntagmatic entity rather than an isolated morpheme. The identification of a construction in current usage, if the examples in (41) are to be taken as typical construction-types, nevertheless appears vague, and lacks rigorous classificatory delimitations. All language form carries meaning – if only as a result of its interaction with other form. Construction types as those listed in (41) may be conceived of ad hoc according to linguistic purpose and endeavour, and yet there is a clear, constituent-based framework underlying the classification of constructions which is difficult to ignore; constituents are in fact, sense packages. One is unlikely to find, for example, a construction-type known as the [if the] construction, simply because there is no conventionalised formmeaning pairing involved in which the two items if and the together comprise a new meaning, however much they can be recognised as construction-types in their own right. It would seem that construction-types are identifiable by the purposes for which they are intended, and form prototype clusterings on the basis of distinctive, eliminable features shared by other constructions; i.e., they are identified by assimilation to their families. In the open market of generalised construction types, the construction-based enterprise, for all its value and insight, may be little more than an elaborate, taxonomic methodology carrying principles that were obvious to functionalists all along but never really made prominent until the advent of early research into idiomaticity and lexical meaning beyond the lexeme (e.g., Fillmore, Kay and O’Connor 1988).

74 

 Chapter 3 Construction grammars and the paradox

This is another reason why the case of argument structure constructions is crucial to the theoretical foundations of construction-based approaches. The original appeal of argument structure constructions was in their ability to contribute their own semantics to affect the meaning of the items found in them, as in the classic example (43). However, Goldberg (2002: 335) acknowledges along with many others at the time, that there was a place for the consideration of verb semantics in the construction, as well as the semantic roles of the arguments most frequently found in the argument slots, points that had previously been made by Boas (2003), Croft (2003), Iwata (2002), and Tomasello (2000) in different frameworks. Langacker (2005a), furthermore, had vouched for the need to examine the conventionalisation of examples like (43), repeated below for convenience as (61): (61) Sam sneezed the napkin off the table suggesting that until such examples become established and conventionalised (perhaps along with the establishing and conventionalisation of the activities they refer to!) they will not be understood as contestants for a caused-motion sense. Langacker adopts a dictionary-method of analysing such examples, suggesting that verbs such as send and kick already have caused-motion possibilities built into their lexical meaning, while sneeze does not (2005a: 152). As noted above, he refers instead to the need for entrenchment, psychologically, and in the speech community. These are difficult points to uphold, since the entrenchment of a new usage in the speech community has to start somewhere, and whether or not the napkin-sneezing examples will perpetuate and propagate through the speech community is often a question of the popular success of the neologism, a sociolinguistic matter, not a linguistic question (see Haspelmath 1999). All change starts out by diffusion from a single instance or set of instances (see, e.g., Croft 2000: 180–1). In their function across contact situations, constructions then face another set of problems: the very issue of unification itself is in jeopardy, and this is what coercion accounts seem to overlook. Given that one of the objectives of the present study was to attempt to provide a stronger basis for the curious phenomenon labelled interlingual identification, we may well ask how speakers in a contact situation in which substratum constructions differ considerably and genetically from those of the lexifier or source language may begin to develop a new construction inventory. The paradox of the contact situation is thus how to accommodate the syntax of one language with the semantics of another, and, if the development of a new constructicon involves using such mixed sources as the materials of construction-building, to determine what the factors are which guide the selection of materials from either language for pattern replication of



3.5 Summarising the current position 

 75

constructions across contact. The principle problems of construction theory have thus been outlined in the above chapter to involve: (a) identification of a construction (b) the question of decomposability (c) the meaning of meaning in a construction, and (d) how constructions become generalised. It has also been suggested that the definition of a construction as aligning with a constituent-like sense package may be sufficient for many purposes, whatever the schematic level of the package itself. It is with these problems in mind that the present study aims to tease apart the anatomy of a construction composed of mixed language sources and created from first principles in a new language situation such as Singapore presents.

Chapter 4  Transitivity and causativity 4.1 Introduction One of the most under‑­researched constructions to appear in Goldberg’s (1995) monograph was the construction known as the transitive construction, an argu‑ ment structure construction about which she had little to say, perhaps since its semantics seemed apparently, self‑­explanatory. Curiously enough, even in later studies on constructions based around Goldberg’s (1995) theories, as well as those of Croft (2001), few researchers have given a great deal of attention to this construction‑­type, in spite of the fact that transitivity has long been researched extensively in typological studies. In her (1995) treatment of the construction, Goldberg admits (p. 117) that although its composition is based on Dowty’s (1991) Proto‑­roles of Agent and Patient as Subject and Object, respectively, there are many cases which defy the initial compositional nature of the construction, clearly illustrating a construction‑­type that has, in English at least, extended its inheritance boundaries way beyond the limits of its original form‑­meaning pairing frame, in effect, a construction gone feral. Bearing in mind, however, that Subject and Object are functional roles rather than formal ones, it could be argued that any two‑­place argument structure will fit the criteria of a transitive construction, one of Goldberg’s earliest dilemmas, which she explained could not be permitted under the Principle of Maximized Motivation (1995: 67). Under such a principle, constructions needed to be be linked to one another semantically, if related at all, and the transitive construction has grammaticalised to such an extent that its protoypical source semantics, in which the subject and the object are linked to agent and patient roles respectively, are no longer recoverable in many cases. Goldberg’s solution was to posit a hierarchy of multiple senses (1995: 118), thus sacrificing the elegance and parsimony of her account and raising the question once again of a how much unification is actually involved in the postulation of the construction as the atomic linguistic unit. One form, according to such an account, cannot be responsible for the generation of so many different meanings. Goldberg’s (1995) appeal to prototypical acquisition routes and crosslinguis‑ tic discrepancies was nevertheless too early for the later work that was to follow, involving diachronic grammaticalisation of construction types. Grammaticali‑ sation accounts of constructions permitted a breakthrough in the fact that they allowed for such distributionally‑­diverse construction types to be explained by a functional motivation, in which the extension of a construction through an incre‑ asing range of lexical environments is accompanied by a certain loss of its proto‑ type semantics over time. More recent studies such as Trousdale (2008b) enabled

78 

 Chapter 4 Transitivity and causativity

the transitive construction to be considered as one of the most highly schematic construction‑­types, while it could still be related, if only distantly, to the proto‑ type from which it evolved historically (see Chapter 2). Thus, it is not necessarily the case of multiple meanings for one form, but of an extension of a form into a range of increasingly schematic meanings; in a sense, reflecting the tendencies of grammaticalisation routes towards ultimate semantic bleaching. Such an evoluti‑ onary perspective on constructionisation raises questions regarding the possibi‑ lity of such increased lexical diffusion affecting the semantics of the construction itself, as alluded to in Chapter 3, and also in Boas (2008), in which he maintains that novel constructional diffusion is made possible not only from the syntactic positioning of arguments but by analogy to existing constructions, known as mini‑­constructions. In many cases, a transitive construction appears to become less affected, not more affected, by the lexical input with which it assimilates, as transitivity is as much an abstract syntactic notion as a semantic one. The study in the present chapter illustrates one example where the notion of transitivity is shown to be somewhat opaque. The original transitive construction was thus simply portrayed as in the fol‑ lowing diagram (Goldberg 1995: 117):

Sem

Syn

proto-agent

proto-patient

↓

↓

SUBJ

OBJ

Fig. 3: The transitive construction (Goldberg 1995: 117).

In such models, the syntax often refers to the argument structure functions of the respective NPs that fill the appropriate slots in the syntax, and it is clear from the model that the semantic roles of agent and patient are seen to play a significant part in the combinatory possibilities of the syntax and semantics (the syntax does not even include the verb in Figure 3). If this is the only description we can refer to in the present chapter, it will be questioned whether the same con‑ struction description can be applied to a form with a similar surface syntax and semantics closely resembling those of a transitive construction, which is not actu‑ ally a transitive construction at all, and this is the case of the conventionalised scenario (Goldberg 1995); this is a syntactically transitive construction which permits the implicit presence of an intermediate causee. Associated with the

4.1 Introduction 

 79

study of such constructions is the additional problem of classifying subjects: if a transitive ­construction is a unification of subject‑­agent (+ verb) + object‑­patient, how can the surface syntax account for the semantics of transitive constructions that align the subject with an indirect form of agency rather than a direct one? In the present chapter, then, the notion of transitivity will first be approached as a contact problem, in the republication of the studies on which the present chapter is based (Ziegeler and Lee 2006, 2009), illustrating the frequency of the conventionalised scenario. It was found that Singapore English exhibits a relatively higher incidence of such constructions, when compared with British English, and that in some cases, the presence of conventionalised scenarios is found in Singapore English where they would be considered unacceptable or treated as ordinary transitive constructions in British varieties. The implausibi‑ lity of interpreting many of the examples as ordinary transitives, though, raises the question of how they are to be described in terms of construction types, and whether the term conventionalised scenario can continue to be used to describe them. It also raises the question of the selection of the transitive construction in the contact situation to express superficially implausible causative events and its generalisation across contact to express indirect as well as direct causativity. The use of a metonymic analysis helps to explain the relation of conventionalised scenarios to existing causative constructions in British varieties of English, with which they may be seen as competing. The present chapter initially studies the construction as a reduced causative construction, at the same time illustrating the limitations on its diffusion in other varieties of English. The chapter will investigate the distribution of these forms in Singaporean English, first reviewing any previous discussion on the use of conventionalised scenarios or their causative alternates (section 4.2), and in section 4.3, subst‑ ratum and contact factors are discussed. Also in section 4.3, the methodology of the studies and results of investigations into their frequency are explained in detail. Section 4.4 offers a number of possible hypotheses to account for the apparent frequency of conventionalised scenarios over causative‑­resultatives in Singaporean English, bringing to light, among other things, the consequences of language contact, in that the reduced salience of agentivity in topic‑­prominent languages may be the main factor enabling (though not directly causing) the greater tendency towards causativity‑­reduction. It is hypothesized, furthermore, that the relationship between resultative constructions and conventionalised scenarios is one in which a resultative construction is simply replaced by a single‑­ clause transitive construction, through the mechanism of a particular metony‑ mic device operating when the causee is not expressed in the complement. The consequences of such a replacement for a construction account of transitives in contact will also be discussed.

80 

 Chapter 4 Transitivity and causativity

4.2 What is a conventionalised scenario? 4.2.1 Earlier reference to the conventionalised scenario The term ‘conventionalized scenario’ was first introduced by Goldberg (1995: 168) to describe a causative construction in which there is no actualization of the cau‑ sativity at all, let alone formal expression of an intermediate causee. It is represen‑ ted as a syntactically‑­reduced expression of indirect causativity, but one which is expressed as a transitive construction in terms of its surface syntax. The causative senses are thus intuitively derived from knowledge about the world and the way in which causative acts are accomplished. Goldberg gives the following examples: (62) a. The invalid owner ran his favorite horse (in the race). b. Chris cut her hair at the salon on University Avenue. c. She painted her house (when in fact the painters did the painting). d. Farmer Joe grew those grape vines. She goes on to explain that these examples are conventionalized in the sense that one normally expects a causee to carry out the act described, and that the internal causative structure is being ignored (1995: 169). However, not all of these examples are the same functionally. To begin with, it is important in examining such examples not to confuse the process taking place with simple cases of transitivization of intransitive verbs. Shibatani (2002: 2) gives an example of a morphologically derived causative verb in Turkish, kos‑­tur ‘cause to run’, which could well translate the use of the verb ran in (1a). In such instances, it would appear that the causativization process taking place is the same as a transitivization of conventionally intransi‑ tive verbs, what may be perhaps a fairly productive process in English. This is not the same as in (1b–c), in which the verb is originally transitive in the first place. In (1d), the process taking place occurs with a verb which may be either transitive or intransitive, i.e., a labile verb (Haspelmath 1987; cited in Li & Thompson 1994) or one which has been described as a lexicalized causative verb, having a cau‑ sative/inchoative function (Croft 1990; Shibatani 2002: 3). Labile verbs usually have a patient causee, and are thus prime candidates for clause union (as was also shown in Ziegeler (2004), in the loss of causative meaning in do after Middle English). What is more important for the present analysis, though, is what (1a–­d) are competing with in functional terms. It is hypothesised that (1b), for example, is represented in most standard uses as a causative‑­resultative (Chris had her hair cut), as will be discussed below. But this is not shown in the other examples The invalid owner ran his favorite horse no more relates to The invalid owner had his favorite horse (to be) run than does The dog‑­owner walked his dog relate to The



4.2 What is a conventionalised scenario? 

 81

dog‑­owner had his dog walked, and Farmer Joe grew those grape vines does not derive from Farmer Joe had those grape vines grown. In other words, there is a disparity of function and not all of the examples require the reconstruction of an unexpressed causee. As such, they may not all, therefore, fall into the description of a conventionalized scenario if that entails, as Goldberg claims (1995: 169), that one normally expects an intermediate causee to carry out the act described. Talmy (2000: 540–541) provides an account of a similar phenomenon known as Inducee‑­conflation, which perhaps serves as a clearer description. Examples include: (63) She took all her furniture with her when she moved to New York (that is, where professional movers did the actual transporting) (64) I cleaned my suit (at the cleaner’s). indicating that the construction is probably used quite frequently in certain con‑ texts describing commercial services and trades. In none of these examples can there be any likelihood that the process is one of simple transitivization of an intransitive verb, and they cannot be interpreted as the causative alternates of causative/inchoative verb types either. (Example (63) is doubtful in terms of cau‑ sativity since the act of taking something with one cannot be easily attributa‑ ble to a third participant.) What is happening is that the construction dispenses with the intermediate causee, either for reasons of lack of salience or for the fact that the causee NP referent is indefinite or non‑­specific, or simply understood as an integral part of the entire caused event and therefore not in need of overt expression. It is Talmy’s suggestion that such examples are derived from confla‑ ted caused‑­agency constructions, e.g., (as adapted from Talmy 2000: 541): (65) HAD‑­ENTITIES‑­do > did However, in the present study, is it hypothesized that it is the causative‑­resultative alternate: (66) NP‑­CAUSE‑­NP V‑­ed > did and not the caused‑­agency alternate as shown in (65), that competes with conven‑ tionalized scenarios to produce Inducee‑­conflation, as the ­causative‑­resultative construction shares with it the demotion of the unexpressed causee; in (65) the causee (ENTITIES) is still overtly expressed. In the conventionalized scena‑ rio described here, the causee is demoted beyond recoverability. It is felt, therefore, that the conventionalized scenario is a cognitive derivative of the ­causative‑­resultative in (66), and as will be seen later in the paper, native speaker intuitions confirm that this is indeed the case.

82 

 Chapter 4 Transitivity and causativity

Conventionalized scenarios may instead be examples of grammatical ­metonymy. Panther and Thornburg (2000) describe causative expressions in English which fall within the classification of a RESULT FOR ACTION metonymy, e.g., Have your documents ready. A possibly similar device may be illustrated ­ ONTROLLER FOR CONTROLLED metonymy, illustrated by Kövecses & in the C Radden (1998). Such metonymies include the once‑­topical Schwarzkopf defeated Iraq, or others such as: (67) Bush bombed Baghdad. From our knowledge about the world, we are able to reconstruct the possible intermediate agents, such as the U.S. Air Force, but as with the conventionalized scenarios described by Goldberg (1995), the internal structure of any interve‑ ning causative event is being ignored. In interpreting the metonymy, Kövecses and Radden (1998) refer to the presence of an Idealized Cognitive Model, usually abbreviated as ICM (Lakoff & Johnson 1980; Lakoff 1987; Croft 1990), i.e., a cogni‑ tive domain in which specific entities are contiguously located in proximity rela‑ tionships; in this case it is the Control ICM. Such relationships can be called upon in reconstructing the target or targets of the metonymy, i.e., in the present situa‑ tion, the Control ICM may assist in retrieving the missing participants of the inter‑ nal causative structure of the event. Thus, for Bush bombed Baghdad, we could suggest that Bush stood for Bush’s Air Force, or in Chris cut her hair yesterday, we could suggest that Chris stands metonymically for (Chris and) Chris’s hairdresser. However, it is not clear from this metonymic shift whether indirect causativity was involved at all or whether the metonym is indeed a conventionalized scena‑ rio, alternatively expressible as a causative‑­resultative construction such as Bush had/ordered Baghdad [to be]bombed. Given world knowledge, though, there is a more likely possibility that (67) will not be ambiguous between a causative and a non‑­causative reading, as a conventionalized scenario might be. Whatever the case, examples such as (67) may still fall into the constructional categorisation of a typical transitive construction, on the basis of the syntax alone. However, this is problematic if the subject is not a direct agent of the action described in the predicate.

4.2.2 Adversative conventionalized scenarios Even more problematic are conventionalised scenario constructions that may be considered not to be causative at all, which at the same time exhibit the syntax of ordinary transitive constructions, and such constructions express situations of adversity involving their subjects; i.e., the subject role could be thought of



4.2 What is a conventionalised scenario? 

 83

as a non‑­causative recipient, not an agent. Talmy, for example, illustrates the ­following examples (2000: 517–518): (68) a. I developed a wart on my ear. b. I broke my arm (when I fell). Similar examples are discussed in Davidse (1992: 120) under the classification of SUPERVENTIVE relations; i.e., those not engendered by the participant. In neither of these examples can there be said to be a causative agent as subject, regard‑ less of the formal similarity with the typical conventionalized scenario described above. It could be said, however, that the construction from which they are derived could be the same as an adversative resultative construction, e.g., I had/got my arm broken. In such cases, the intermediate causee (which may be an event or a human participant) is implied in the resultative construction, though demoted to the point of deletion, as in the non‑­adversative type. The difference between the adversative type and the non‑­adversative type is that in the latter, the subject may be perceived as the agent of the verb, while in the former, the subject may be perceived as only an Author, to use Talmy’s (2000) categorization; i.e., an unin‑ tentional causative participant. The analysis of such constructions as transitives is therefore difficult to render if the parameters of Hopper and Thompson (1980) are taken into consideration (e.g., including volitionality, action and agency), and yet, the syntax bears all the characteristics of a typical transitive structure, and must, of necessity, imply classifiying them as such. The fact that such construc‑ tions are in common currency may be due to the weakening of causativity in the causer subject, which is not so far advanced in other conventionalized scenarios; that is, the causativity is lost first when there is an environment least susceptible to ambiguity: one does not intentionally cause one’s own arm to be broken, and the construction is accepted on those terms (for this reason, they will be labelled adversative conventionalised scenarios). It will be seen later in the chapter that the responses of Singaporean informants bear out similar predictions on acceptability. Talmy describes adversative examples as also involving an Undergoer subject,26 rather than a recipient subject (not relating to the general classification

26 The use of the term Undergoer may appear to be inconsistent with the use of the term Author, as previously discussed, and Talmy is not very clear on the differences; however, there is a subtle distinction: in the first instance, the adversative event appears to be initiated by an agentive act but with no knowledge of its adversative consequences, e.g., The careless kid (accidentally) broke his arm in hitting it playfully with a hammer (Talmy 2000: 517). However, in the case of the Undergoer, the initiating event is not agentive: The hapless fellow (by misfortune) broke his arm when he fell (2000: 517). Since the grammatical form of the conventionalized scenario is the same in both cases, it may be most accurate to consider the ambivalent subject role as a type of ‘involved recipient’.

84 

 Chapter 4 Transitivity and causativity

of macro‑­roles introduced by Foley and Van Valin 1984), i.e., one that is a sentient being to which something is happening. It also includes the subjects of such non‑­ agentive acts as losing a possession. It may be questioned at this stage, therefore, if there are any lexical constraints on the development of CSs, and careful exa‑ mination of the prototypical cases suggests that there are. The examples in (62) suggest that, although the surface form of a CS is structurally identical to a simple transitive construction (as termed by Goldberg 1995), it is not the case that just any surface transitive construction can be interpreted as a CS. The only example which is unambiguously a CS is (62b), and the lack of ambiguity arises partly from what we know about the world, or common ground. Thus, there is no formal means of distinguishing a CS from a simple transitive construction in which there is no implied causee (even (1c) requires a parenthetical note to eliminate the pos‑ sibility that the subject was the immediate agent of the act). However, there is a semantic difference between (62b) and (62c); in (62b) the subject is, partially, an undergoer of the act of hair‑­cutting, since hair is an inalienable possession of the subject, and the subject, Chris, cannot avoid being affected in some way by the caused event. Since there is no (plausible) ambiguity likely in (62b), we may hypothesise that amongst the most readily available situations for the develop‑ ment of CSs are those in which the subject of the caused event is not only a causer but more importantly an affectee, recipient or an undergoer of the caused event. Such a hypothesis leads us to explain why adversatives leave us in no doubt as to the level of indirect causativity involved: the subject of an adversative CS is the undergoer, as illustrated in (68), and is at least partially affected by the caused event, and the body‑­part involved in the caused event is an inalienable possession of the subject. We could not expect any ambiguity over the identity of the participa‑ ting agentive cause of the arm‑­breaking or the wart‑­developing in such examples, and if anything, such examples are conventionalised because there is little pos‑ sibility that the subject would engage the assistance of an intervening causee to undertake such acts. The acceptability of such constructions leads us to hypothe‑ sise that another situation in which CSs are most readily available is that in which the subject is not only an undergoer, recipient or affectee, of the caused event, but in which the likelihood of an intervening cause is virtually absent altogether; i.e., those in which there is some adversity directed toward the subject. In such cases again, the subject bears both agent (or author, as Talmy suggests) and undergoer role at the same time, but the undergoer role is more in focus. Although, as Talmy (2000: 517) claims, such constructions with undergoer subjects are not really cau‑ satives but only mistakable for causatives, there are some features which are shared in common with causatives, and it is for this reason that it would not be necessary to consider the adversative type differently from the non‑­adversative type.

4.2 What is a conventionalised scenario? 



 85

4.2.3 Constraints on the use of CSs (62c), however, does not involve a subject that is partially affected by the caused event at all; in fact, the subject is not an undergoer or affectee in such construc‑ tions. If anything, the subject is merely a recipient of an action performed by an intervening cause. The recipient semantics of the subject role, however, must be reinforced by additional information, as shown; otherwise, it is quite possible to interpret (62c) out of context with the subject, She, as agent of the act of painting. The same may be said for (64), which also contains parenthetical information to disambiguate the identity of the agent role. Such examples may be conside‑ red, then, as distributional extensions of the more prototypical examples (such as adversatives) in which the subject is simultaneously an author and affected undergoer of the act, and recipients may be considered as unaffected undergoers. Thus, different types of conventionalised scenarios may be graded as more or less prototypical exemplars of the construction, depending on whether their surface representations can or cannot be alternatively interpretable as simple transitive constructions. This grading can be shown in the following examples: (69)  Example   I broke my arm   Chris cut her hair   I cleaned my suit

Nature of Subject NP Nature of Object NP Author/Affected Undergoer Agent/Affected Undergoer Agent/Recipient

Inalienable possessee of subject NP Inalienable possessee of subject NP Alienable possessee of subject NP

Such characteristics reveal that the use of the CS may reflect in distributional range from a prototypical core use (adversative) and less prototypical uses in which the meaning may still be ambiguous between a CS and a simple transi‑ tive structure. The direction in which the extension is leading is indicated by the ambiguity of examples such as (64) which, out of context, still require some clarification over the identity of the immediate agent; this is not the case with the prototypical examples in which the agent is also affected by the causative act. The adversative examples are also less prototypical transitives in Hopper and Thompson’s (1980) terms, for the fact that the object of the verb is not so readily individuable from the subject, being an inalienable possession. However, the adversatives are more prototypical CSs, having a subject least likely to be the agent of the action described. Thus the CS construction could be considered as expanding into the territory of the transitive construction, or of subsuming its syntax, with the obvious outcome of ambiguities in cases in which the agent of the action is possibly indeterminate from the syntax.

86 

 Chapter 4 Transitivity and causativity

Other constraints are obvious. The CS is not likely to arise in cases in which the ambiguity with a simple transitive construction would be too great to warrant using one. Thus, while (62c): She painted her house is understood (albeit with the aid of explanatory notes) to refer to an act carried out by an intervening participant, (70) a. She painted his house might be mistaken for an act of love and devotion on the part of the subject, or b. She painted the school‑­house might be considered to be a contribution of benevolence toward her local com‑ munity (even if she can refer to the new headmistress, an authority figure, it still leaves in doubt the identity of the actual agent of the event). There is no reason to invoke the possibility of an intervening causee in such circumstances, and yet, on the surface, (70a–b) have a form identical to that of the conventionalised scena‑ rio. There could only be semantic reasons: in (62c) the object NP is a possessee of the subject NP, while in (70a–b) it is not. As we have seen above, for the more prototypical examples that do not require disambiguation, the object NP is in an inalienable possessor‑­possessee relationship with the subject, while in (62c) the object NP is, though not inalienable, still a possessee. In many ways, then, one could hypothesise that the CS offers a distinct way of marking a benefactive cons‑ truction in English, which, if alternatively expressed using a causative‑­resultative construction could in some instances be ambiguously interpreted as an adver‑ sative (e.g., She got her hair cut (unintentionally)). It should also be noted that examples such as She cut her hair cannot supply the adversative reading that is obtained from examples such as She broke her arm, leading one to believe that the construction has grammaticalised further from an originally restricted use associated with mainly adversative situations to include benefactive situations as well. Thus, for this reason also, the types of situations represented in the deve‑ lopment of CSs from (62c) and (64) are considered to be distributional extensions of the more prototypical ones in which the object NP is an inalienable possessee of the subject: the restrictions on alienability being weakened to leave only the constraint of a general relationship of possession between subject and object NP. Finally, however, there cannot be a conventionalised scenario if it cannot be interpreted alternatively as a simple transitive construction. It will be seen later, in the examples cited from Singaporean English, that such a constraint is not observed so frequently in such dialects (for example, Do you deliver your[own] ­newspaper? No, I don’t deliver, heard as an informal observation on a Singapore radio station, Jan. 2003 (it is not possible to deliver something to yourself)). However, it is also not possible to use a CS when the action referred to by the verb is one of movement away from the subject source NP, or accompanying



4.2 What is a conventionalised scenario? 

 87

movement, as shown in (63) above: one cannot literally take something with one, if an intervening cause is involved. Similarly, examples such as: (71) She took her suit to the cleaners (?when in fact someone else actually took it for her) cannot be said if an intermediate participant is involved; in such cases, a lexical causative, send, is used to convey the same information instead, and (72) She led her children into the garden (?when in fact someone else actually led them) leaves no ambiguity as to who is actually doing the leading. In other words, there are selection restrictions or co‑­occurrence constraints on the type of verb which may appear in a CS construction. This means that only verbs in which the action refer‑ red to does not require the physical accompaniment of the agent or does not refer to a directional path away from the subject source can be used in CSs, and of those verb types, one would not expect to find many verbs that did not express a causa‑ tive change of state in their semantics. This is why (62a) is marginal, why intransitive verbs in general cannot be found in CSs, even when used transitively; (e.g., I walked the dog is not an example of a CS, as the dog is not an intervening causee) and simi‑ larly, why Goldberg’s (1995: 169) example, The company flew her to Chicago for an interview which is the transitive means of expressing The company caused her to fly, are not bona fide examples either. (In both the dog‑­walking case and the flying case, it is the raised transitive object which performs the action of walking or flying – albeit with the aid of a machine in the latter example. Such a situation is not parallel to the prototypical hair‑­cutting CS, as hair cannot cut itself). It is for the same reasons that Goldberg’s (1995: 169) caused‑­motion examples do not have corresponding CSs (e.g.,??Farmer Joe grew those vines onto the roof is the example she provides, but Chris cut her hair onto the floor would better illustrate the point); such examples have a location away from the source as goal, and a restricted definition of a CS as discussed in the present study would permit only the causer source to be the goal (or recipient) of the causative action. The constraints can be thus summarised as follows: (73) a. The subject of a CS must be either an affected agent + undergoer, or an unaffected agent + recipient of the action specified in the verb;27 i.e., both a source and a goal of the causative event;

27 The use of semantic role terms such as agent in causative expressions may invite some con‑ fusion, since it is not clear whether the primary agent (i.e., causer) or the secondary agent (i.e., causee) should be labelled as such; for this reason it will be preferred to distinguish such roles by terms such as causer and causee instead.

88 



 Chapter 4 Transitivity and causativity b. The subject NP of the CS is usually in a possessor/possessee ­relationship with the object NP; c. Verb types cannot include accompaniment verbs or verbs expressing direction away from the causative source and must refer to a causative change of state; d. The CS must have the form of a simple transitive construction, and be alternatively interpreted as one, but not be plausibly ambiguous with one.

It is for such reasons that the causative‑­resultative construction has been linked to the conventionalised scenario –­it has a passive complement, and therefore the subject is ambiguously both agent (or author) and affected undergoer (or recipi‑ ent), the object NP is in a possessee‑­possessor relationship with the subject NP, since the lexical source of the causativity is a verb of possession, the direction of the causative action is never away from or accompanying the agent‑­source but directed back towards it, and the resultative nature of the construction means that it expresses a change of state. Furthermore, the dual role of the agent/recipi‑ ent subject, involved in the causing and receiving of entire causative acts, is the result of the lexical source meanings of causative verbs of agentive acquisition such as have and get having meanings of possession, thus further testifying to the association of the CS with such constructions. Thus, I broke my arm is linked to I had/got my arm broken, I cut my hair to I had/got my hair cut, and I cleaned my shirt to I had/got my shirt cleaned. The causativity in causative‑­resultatives is thus reflexive, which in other languages (such as Thai, Mandarin, and Cantonese) provides an environment in which causative expressions progress to become pas‑ sives (see, for example, Yap and Iwasaki 2003). In English, however, the reflexive causativity appears to be the source for a construction in which the causativity is not expressed at all, but only implied, making it ambiguous in many accounts with a simple transitive construction in which the subject is the direct agent of the transitive act. What we are seeing instead, then, in the conventionalised scenario, is also a reflexive causative situation in which a resultative construction is linked by a process of voice‑­conversion, not to a passive construction but to an active one. Such a process could only take place under specific pragmatic conditions. 4.2.4 The causative‑­resultative alternate The causative‑­resultative construction in the present chapter is typically defined as of the following surface form: NP CAUSE NP V‑­ed, the complement form being a past participle.28 This type of construction does not appear to have been isolated

28 Bencini and Goldberg (2000: 650) use get‑­causative resultatives as examples of resultative constructions according to Goldberg’s (1995) use of the term, e.g., Dana got the mattress inflated. Goldberg’s original resultatives, however, did not include participle complements.



4.2 What is a conventionalised scenario? 

 89

as such in the literature; it may be compared with what Talmy (2000) has labelled ‘the caused‑­agency’ construction, a more general term encompassing many dif‑ ferent types of periphrastic causatives, all definable by the fact that the causer is an agentive entity and capable of imposing his or her will on the resulting caused events, and marked by the presence of an infinitive (with or without to) in the complement of the medial NP. In the latter semantic category it is possible to classify the active‑­voice‑­complement counterparts to the passive‑­complement construction under investigation; i.e., those of the surface form: NP CAUSE NP (to) V (NP). In the present study, the verb form represented as the first verb of the resultative constructions so described (CAUSE) most frequently appears as a possessive‑­causative verb, either have or get, e.g., (74) I had my tonsils removed (International Corpus of English‑­Great Britain: S1A‑­051)29 Causative‑­resultatives have also been classified under the label ‘causeeless cau‑ satives’ (Kemmer & Verhagen 1994: 139), or ‘Inducee‑­nominal deletion’ (Talmy 2000: 538), referring to those constructions in which the causee has been demoted to the extent that it is absent from the construction altogether. Causeeless causa‑ tives are seen by Kemmer and Verhagen to represent constructions in which the most peripheral attention is given to the causee participant, as the causative act is described without the need for an overtly marked causee. In many languages, the causee is not deleted, but Talmy (2000: 538) describes the Inducee in the same way that most accounts describe the causee, and gives the example from English: (75) I had a shirt made. as well as Yiddish (76) Ikh hob gelozn makhm a hemd. It is interesting to note that in Yiddish, as well as in Dutch and German, the ­complement verb is not a participle but an infinitive, e.g., in Dutch (Kemmer & Verhagen 1994: 138): (77) Zij laten een 3PL let INDF.ART  ‘They are having a house built’30

huis house

bouwen build

29 Such constructions are to be considered distinct from what is known as the get‑­passive, as discussed in Givón and Yang (1994), in which the medial NP is considered an ellipted reflexive of the subject. The coding for these examples refers to the text category and hierarchy within the International Corpus of English‑­Great Britain; for (S1A‑­051) –­face‑­to‑­face conversation. 30 Interlinear gloss supplied by DZ.

90 

 Chapter 4 Transitivity and causativity

Similar examples are also found in German (Panther & Thornburg 2000: 218): sich nicht zur (78) Lassen Sie Let you yourself not to‑­the ‘Don’t let yourself be forced into signing.’

Unterschrift signature

zwingen force

Although Panther and Thornburg’s examples are restricted to imperative uses of the main verb, it is clear that the passive or adjectival complement is not possible in German when the main verb is stative. The facility with which this occurs in the English translations is explained by them as due to the RESULT FOR ACTION metonymy (noted above), which they believe is far more grammaticalised in English than in German, i.e., the expression of the result (the holding of a com‑ pleted state) may be used to stand for the prior action presupposed by the result (2000: 215). The resulting construction, as shown in (78), translates similarly to the causative‑­resultative in English (NPi CAUSE NPi V‑­ed). 4.2.5 Adversative resultatives Also mentioned in the literature are a number of cases in which (as for the ­adversative CSs above), the structure NP HAVE/GET NP V‑­ed cannot be held to be causative at all, though it bears the same surface structure features as the causative‑­resultative. These are adversative resultatives, which, like adversative CSs, are so‑­named because their complements refer to some kind of adversative result affecting the subject of the main verb. Such examples have been referred to as early as Lakoff (1971) as one of the emerging senses of the three‑­way ambi‑ guous sentence: (79) John got/had his dishes washed. Lakoff is puzzled by the ambiguities: (i) the adversative, unintentional reading, (ii) the achievement reading, and (iii) the causative reading, though she admits that the adversative sense is one in which the recipient meanings of get are focal. The addition of an adverbial expression, e.g., on purpose is added to similar con‑ structions to determine whether the responsibility for the action resides in the subject; this is not the case with be‑­passives. If the causative‑­resultative can be considered an incipient passive itself, then it is still affected by the retention of meanings more closely related to its source meanings, i.e., the subject is still only a recipient, and not yet a prototypical patient. It is interesting that such examp‑ les are expressed using the same formal means as causative‑­resultatives, as they represent a stage of grammaticalization of the have or get causative‑­resultative which is bleaching of all causative meaning. An unambiguous adversative



4.2 What is a conventionalised scenario? 

 91

example from the International Corpus of English‑­Great Britain, using have as the main verb, shows no residual sense of causativity whatsoever: (80) Students failing to disclose this fact are liable to have their registration ­cancelled (ICE‑­GB W2D‑­007 015)31 As is shown in Ziegeler (2000b) for the case of the bǎ‑­construction in Chinese, the similar rise of an adversative sense in an erstwhile causative construction is an indication that the meanings of causativity are being lost, and the construction is at a later stage of grammaticalization. The subject role is no longer an agent, shif‑ ting to become a recipient subject as the meanings of possession gradually take over the earlier meanings of causativity. There are examples in the Oxford English Dictionary Online dating back to Early Modern English times, e.g.: (81) Another had one of his hands burnt

1719 Defoe. Robinson Crusoe II, x.

The rarity of earlier examples attests to the later stage of grammaticalization; one does not deliberately cause one’s hands to be burnt, and such uses could be assumed to be on the way towards a passive interpretation. It is hypothesised that the development of adversative causative‑­resultatives indicated the bleaching of causative senses in the subject of such constructions at later stages. This in turn might have opened the gates for the single‑­clause, conventionalised scenario constructions in which the (indirect) causativity is only implicit in the transitive‑­ like surface structure. In linking the two constructions pragmatically, it is necessary to seek out a means of semantic continuity by which an association between the two construc‑ tions may be held; it is not possible for just any construction to replace another with no logical connection between the two, and there is clearly no syntactic con‑ tinuity between them. The term used to link the constructions has been described above as a kind of voice‑­conversion, whereby a passive‑­like construction becomes replaced with an active counterpart, preserving at the same time the semantics of its passive participle in the active verb form replacement. The replacement by an active verb form as main predicate of the converted form suggests that the causer‑­recipient of the countervailing resultative construction is now the sole agent of the caused event; however, whether or not we can maintain the semantic role of causer‑­recipient in the converted form is not important, since the causee (or secondary agent) of the caused event in the parallel resultative construction

31 The text codes listed in the International Corpus of English‑­Great Britain refer to this example (W2D‑­007 015) as from administrative, or regulatory writing.

92 

 Chapter 4 Transitivity and causativity

is present only as a conversational implicature in any case. The reason that the causee must be interpreted as a conversational implicature is due to the reflexi‑ vity of the caused event; it is less possible to perform actions on oneself without the intervention of another participant (the conversational implicature may be revealed in the cancellation of I got my hair cut with (but nobody else did it)). In such cases where it is possible, the use of the causative‑­resultative with a causa‑ tive verb of possession indicates successful achievement of a difficult task (e.g., of cutting one’s own hair), and the focal sense changes from causativity to accom‑ plishment. The agency of a causee participant is thus implied in the fact that the result would not be so easily obtained if there were not an intervening partici‑ pant. The neutral meaning of the reflexive causative‑­resultative is thus one of suc‑ cessful achievement of a caused event, regardless of who performed the action. Thus the three‑­ways ambiguous example provided by Lakoff (1971) in (79) easily allows for semantic continuity in a transitive conversion: John washed his dishes, where no intervening causee need be present at all, and such a conversion may be considered possible through semantic continuity of the ‘achievement’ reading in John got his dishes washed. The achievement sense is evoked by the subject’s involvement in performing and completing an action for his/her own benefit (hence an additional benefactive sense is associated with such types), and may be reinterpreted as performance in the corresponding transitive, John washed his dishes. The CS is not necessarily an achievement expression, but may be derived by analogy with possible cases such as John washed his dishes, in which the causee is totally absent and not even an inference in the meaning. The possibility for such an analogy is due to the fact that the causee is only an impli‑ cature, e.g., in I got my hair cut, the implicature brought about mainly through world knowledge about the way certain events take place. It is thus not part of the logical meaning of the construction; i.e., it would not result in a contradiction to derive such a conversion to I cut my hair, with the subject as the true agent. The conversion may thus be expressed in the following semantic pathway: (82) CAUSED‑­RESULT < ACHIEVEMENT < PERFORMANCE/TRANSITIVITY in which the possession of caused‑­result is reinterpreted as an achievement of the subject, which then implies successful prior performance on the part of the causer‑­recipient (just as for have‑­perfects, which is possibly why they are inter‑ preted as expressing prior performance readings in many Romance languages). As long as the causee is either absent or has only an implied presence, analogical extension to other, non‑­achievement uses would be possible. On the other side of the coin, in adversatives the likelihood of a causee is even more suppressed but for different reasons; e.g., He got his arm broken in which the cancellation (but nobody else did it) is somewhat redundant – nobody causes another person



4.3 Conventionalised scenario constructions in Singaporean English 

 93

to do injury to them. The two semantic pathways of malefactive and benefactive CRs appear to lend themselves most easily to conversion, due to the fact that the implicature of an intervening causee participant is the least likely in such cases; in the former case because of the adversative effect of the event on the subject, and in the latter case because of the prominence of the subject‑­causer in achieving the action. The adversative CS, then, is a ‘fake’ transitive, because it is least likely to be interpreted as one; the achievement CS is a true transitive which can be related to the CR of achievement over adversity. This analysis may help to clarify the relation between causative‑­resultative constructions and CSs as revealed in the empirical study described below.

4.3 Conventionalised scenario constructions in Singaporean English The original hypothesis for the present study was based on personal observations of what was considered to be an over‑­extended use of conventionalised scenarios in Singaporean English, in expressions such as you cut your hair (when somebody else did it). Although such forms are noted as in current usage in other dialects, it was felt, as noted above, that they were used a little more frequently in Singaporean English than elsewhere. The only previous observations found of such a feature are in Brown (1999) and Low and Brown (2003: 66), who provide examples of impera‑ tive uses such as: (83) Spray paint your car at our workshop seen on Singaporean advertising material, and (84) Polish your car while you shop observed in a Singapore shopping centre car park (discussed also as (58) in Ch.3.4). They note that in the first instance, ambiguity is marginally possible, though certainly not in the latter example (84), in which the construction is so highly grammaticalised that the understood, main clause causer subject which is not the agent of the verb may be co‑­referenced with the expressed subject of a sub‑ ordinate clause which is actually the agent of the following verb. Other examples include: Laminate your driving license here; I must repair my watch; I need to wash my car; and I want to take my passport photo (Low and Brown 2003: 67). Low and Brown’s ­publication in which the forms were reported was intended partially as a coursebook for English teacher trainees at the National Institute of Educa‑ tion, Singapore, and therefore alternative suggestions for what are represented

94 

 Chapter 4 Transitivity and causativity

as somewhat marginally standard uses are also provided. The alternative sugges‑ tions which appear alongside these examples were have‑­causative‑­resultatives, which are considered to be the construction type most likely to be in competition with the conventionalised scenario, as noted above. However, evidence from such token examples does not necessarily imply that the construction is used frequently or is associated generally with the the local vernacular: for this reason it was considered necessary to undertake a quantita‑ tive survey of usage amongst native speakers of Singaporean English, in order to determine whether it is an entrenched feature of common usage in the dialect. It was also considered that the substrate and local contact languages spoken along‑ side English in Singapore might be providing the model construction on which the English CS is based. The following sections will therefore review translations of the conventionalised scenario and causative‑­resultative constructions in some of the main substrate and contact languages spoken in Singapore.

4.3.1 Substrate influence As noted in Chapter 2, the main substratum languages, if any, likely to have influ‑ enced Singaporean English in the past are the southern Chinese dialects such as Hokkien and Teochew (or Chaozhou), both Southern Min dialects, Cantonese, and to some extent, Malay and Bazaar or Baba Malay, the latter two being more likely to have had an influence on the early development of English in Singapore than on its continued development. Tamil is the official language of the Indian community, but as only less than 10% of the population are Indian and speak a range of Indian languages and dialects, in numbers too small to be of contact sig‑ nificance (Bao Zhi Ming, p.c.), it is not a dominant substratum language.32 It was also noted in Chapter 2 that in more recent years, Mandarin Chinese may be added as an influencing contact language especially in Singapore, as the government’s Speak Mandarin Campaign has been in force since 1979 to be used as a means of eliminating regional Chinese dialects among Singaporeans of Chinese ethnicity. However, it is well known that some features of the original substratum languages have become conventionalized into the local colloquial variety and are passed on from generation to generation, e.g., discourse particles such as lah (from Hokkien or Malay). Thus, it is not always easy to distinguish what is carried over as part and parcel of the dialect from the beginnings of Singaporean English in contact, 32 It should be noted that some aspects of Indian English grammar are shared with Singaporean English grammar; e.g., the use of the progressive with stative verbs, and the use of plural mar‑ king on collective or partitive nouns (Trudgill & Hannah 1986).

4.3 Conventionalised scenario constructions in Singaporean English  



 95

and what is associated with its present‑­day contact in daily competition with the prescribed Chinese dialect, Mandarin. It may be questioned whether the construction type known as the conventi‑ onalised scenario may be understood to have been transferred directly from the substrate languages as a pattern copying or whether it may have evolved inde‑ pendently in the new dialect as a reduced form of the causative constructions with which it may originally have been in contact. In Malaysian Hokkien, the most preferred expressions used to express causative‑­resultative situations are those in which no indirect causativity is marked, a construction identical to the conventionalized scenario, as shown below. The following examples illustrate the ways in which indirect causativity is expressed in the main local languages of Singapore, using translations of the predicted English responses to the survey questions shown in section 4.3.2. Hokkien33 (khi) ka (85) a. wa 1SG (go) cut ‘I get my hair cut’

tau‑­mor‑­i hair‑­3SG

There is no passivity allowing for an implied causee participant in the Hokkien translation, though in other examples there is some variation:   (bueq) kio lokun b. wa 1SG (would) call/ask doctor ‘I would get it removed by a surgeon’

kuat(‑­khi) cut(‑­off )

hor chiã‑­ui siu‑­li c.  wa 1SG let/allow car‑­place fix ‘I get my car fixed (at the garage)’.34

i 3SG

chiã car

The latter two examples allow the introduction of an active causee p ­ articipant, though an alternative means of translating such expressions need not use a medial causee at all, e.g.:   i‑­chiã (khi) siu‑­li d. wa 1SG (go) fix 3SG‑­car ‘I get my car fixed’.

33 Hokkien examples were provided by Sarah Lee, Mandarin examples by Bao Zhi Ming. Canto‑ nese examples are by Richard Wong, and the Malay examples were provided by Ismail Talib and Foong Ha Yap. Shoniah Supramani provided the Tamil. 34 The tilde marks nasalization.

96 

 Chapter 4 Transitivity and causativity

In Mandarin, similar syntax is used to translate the examples. The following examples are provided by Bao Zhi Ming (p.c.): Mandarin (86) a. Wǒ bǎ tóu‑­fa 1SG BA hair jiǎn le b. Wǒ 1sg cut ASP ‘I get my hair cut’.

jiǎn cut

le ASP

tóu‑­fa hair

Sentence (86b) is the preferred means of expressing such constructions in Man‑ darin, though it should be noted that in (86a), the pre‑­transitive marker bǎ, while represented here as a grammatical function word, is grammaticalized from a possession‑­based lexical source meaning ‘take’ or ‘get’, and that English get‑­ resultatives have been translated by native speakers of Taiwanese Mandarin into Mandarin bǎ‑­constructions in exercises of direct elicitation (see Ziegeler 2000b). It might be suggested then, that there does exist a possessive‑­causative in Manda‑ rin, though it is less frequently used than a manipulative, single‑­clause transitive causative expression, and the causative verb is not so closely related to its origi‑ nal lexical sources as get might be in English. The other two examples yielded similar results: (87) a. Wǒ 1SG

bǎ BA

b. Wǒ xiū‑­le 1SG repair‑­ASP  ‘I get my car fixed’.

chē car

xiū le repair‑­ASP

chē car

c. Wǒ jiào yīshēng bǎ zì 1SG call doctor BA mole  ‘I would have the mole removed by a doctor’

qiè‑­le cut‑­ASP

Note that there is no alternative, mono‑­clausal construction for (87c). It seems that, as in Hokkien, the use of indirect causatives and the introduction of a causee participant in Mandarin correlates with the amount of control attribu‑ table to the medial causee, and surgical operations are certainly not within the control of the causer, who acts as an instigator of the causative act and nothing more from then on.

4.3 Conventionalised scenario constructions in Singaporean English  



 97

Other dominant substratum languages, such as Cantonese, show similar ­translations, involving conventionalized scenarios in the complement clauses, e.g.: Cantonese heoi (88) a. ngóh 1SG go ‘I get my hair cut’ ló b. ngóh 1SG take ‘I get my car fixed’

zín cut

taùh faat hair

gaa CL

chē car

heoi go

seúng heoi tuet c. ngóh 1SG want go remove ‘I would have a mole removed’

saūleí repair

mák mole

Like the Hokkien examples, but unlike the Mandarin examples using bǎ, there is no resultivity explicitly marked in the complement, and the corresponding struc‑ tures contain more serial verbs than the other two dialects. The Malay examples are interesting in that they do seem to contain comple‑ ments expressing result or passivity, though there are preferences for quite diffe‑ rent types of constructions: Malay dapat (89) a. saya 1SG get ‘I get my hair cut’

rambut hair

saya my

di‑­gunting PASS‑­cut

b. Kalau saya saya akan minta (supaya) tahi lalat yang ada if 1sg 1sg FUT ask (so‑­that) mole REL exist di‑­hilang‑­kan PASS‑­disappear‑­CAUS    ‘If it were me, I would ask (that/so that) any existing mole would be removed.’ For (89a), a preferred means of expressing the same thing would be: saya gunting rambut – ‘I‑­cut‑­hair’; similarly, for (89b), an alternative means would be tahi‑­lalat saya akan potong – mole‑­I‑­want‑­remove. The use of minta (‘ask’) in such cases, would not be necessary. In the case of (89b), the sense of hypotheticality in the main clause must be conveyed by an accompanying, overtly expressed conditional

98 

 Chapter 4 Transitivity and causativity

clause (kalau saya – ‘if it were me’), unlike in English where the c­ ondition can be merely implied by the use of tense inflections (Foong Ha Yap, p.c.). It is interesting that when the causee is not mentioned, the causative‑­passive morphology must be expressed, and that a possessive verb (dapat) is used in Malay to translate the English causative‑­resultative. It remains to be seen whether this is likely to be an factor influencing the frequency of the competing constructions in Singaporean English. The standard Tamil causative is constructed morphologically by using the main verb and a suffix – v/bi, which converts a non‑­causative verb into a causa‑ tive one. However, it is being renewed in its function in present‑­day usage by an auxiliary form, chei (‘do’), as shown below. Again the distinction between causa‑ tive result and caused agency is not made, though the main verb is expressed as non‑­finite when co‑­occurring with the auxiliary form:35 Tamil en mudi‑­yai (90)  a. Naan I my hair‑­ACC  ‘I will get my hair cut.’ en mudi‑­yai b. Naan I my hair‑­ACC  ‘I will get my hair cut.’

vettu‑­vi‑­pp‑­ en cut‑­CAUS‑­FUT‑­1SG vettu‑­maaru cut‑­PART

chei‑­v‑­en do‑­FUT‑­1SG

From all of the data above, it is possible to predict that the effects of substra‑ tum tendencies in the marking of causativity may influence the selection of a single‑­clause, conventionalized scenario structure over a ­causative-­resultative in ­Singaporean English. It is clear that the means by which ­causative‑­resultatives are expressed in many of the substratum and contact languages either do not express indirect causativity at all, or if they do, do not distinguish ­caused‑­result meanings from other caused‑­agency expressions. The ­consequences of the pre‑ sence of such constructions in the contact environment were thus examined in the form of a brief survey investigating the means by which indirect causativity was expressed by educated native speakers of ­Singaporean English.

35 Greg Anderson is gratefully acknowledged for his assistance in the morphological analysis of the Tamil example.



4.3 Conventionalised scenario constructions in Singaporean English  

 99

4.3.2 The quantitative survey The forms under investigation in the first direct elicitation survey were three con‑ ventionalized scenario constructions, or their variants, and the production rather than the comprehension of the feature was taken into consideration. The reasons for this were that it would be pointless to produce a conventionalized scenario and attempt to elicit comprehension: in most cases the meaning of the construc‑ tion was easily retrievable by pragmatic means if it was obvious that there was an intermediate cause, and if not, as in I cut my hair, the ambiguity, out of context, would do little to demonstrate what conditioning factors, if any, held for the dis‑ tribution of a CS across dialects. Any likelihood of distributional constraints may be assumed from the informants’ readiness to produce such a form or not given an appropriate context. The three questions used to supply the context were the following (extracted from longer surveys): (91) Question 1. What do you do when your hair gets too long? Question 2. What do you do if your car has a broken tail light? Question 3. What would you do if you had a mole on your body that is a bit painful and the doctor says it is cancerous? The questions were designed to elicit such causative‑­resultative forms as get it removed, have it cut, get it fixed, and variations of these responses, thus also confir‑ ming whether the conventionalized scenario is in competition with the causative‑­ resultative in the dialects under investigation. In order to ascertain that responses contained reference to actions with an implied causee, a further s­ upplementary question was added to the original questions (Questions 1 and 2), i.e., for (Q.1): Who is your favorite hairdresser or barber? and for (Q.2): At which garage or repair shop is your car serviced? It was not felt necessary to add a supplementary ques‑ tion to (Q.3), as no ambiguity in the causer/causee relationship was likely. The Singaporean groups (a minimum of 22 and a maximum of 61 for any single question) were almost all taken from undergraduate students, who were approached out of classroom hours in canteens, recreational areas, and in halls of residence at the National University of Singapore. A small number (around 10) were given surveys to take home at the end of a lecture (students from the Depart‑ ment of English Language and Literature); it is not known how long they took to complete the tasks, but they were returned a few days later. The Singaporean stu‑ dents approached on campus were from various disciplines, and were also given approximately 20 minutes to complete the tasks. They were mainly under 25 years

100 

 Chapter 4 Transitivity and causativity

of age. All responses, including the Control data, were provided in writing, since the written form is the most formal of styles, and it is generally noted that once a variant has become acceptable in written usage, it is considered an entrenched item in the system (see e.g., Backus, Dogruoz and Heine (2011), while spoken usage is subject to much more variation and idiomaticity. Questions 4–5 were also surveyed on the campus of the National University of Singapore, but at a later stage, under the same conditions as the first three questi‑ ons, using spontaneously‑­elicited responses to particular written questions desi‑ gned to produce a causative response. The questions are as follows: (92) Question 4. What would you do if you had a bad tooth and your dentist says it can’t be saved? Question 5. You need ID photos and you don’t like the old ones you kept. What would you do? All the data were obtained informally by approaching groups of students or individuals directly. Question 4 elicited 22 causative responses and Question 5 elicited 25. The anticipated responses to Question 4 were expected to produce structures like: have it pulled out, have it extracted, and for Question 5: have/ get new ones taken. It was found that the causative responses to these last two questions were overwhelmingly biased towards the use of the CS construction, as seen in Table 1. In addition, it was felt necessary to survey a group of British speakers of English, since the contructions were originally associated with non‑­contact vari‑ eties of English, as assumed in Goldberg (1995). The British control group for the first three questions was a group of first year English undergraduates from the University of Manchester, surveyed at the end of a lecture. There was a maximum of 66 and a minimum of 47 respondents for any one question. They were also given approximately 20 minutes to answer the total number of tasks supplied on each questionnaire. They were mainly under 25 in age, there being only 10 over that age in total. Thus, although the conditions for sampling were not controlled in as far as situation or setting was concerned, there was a certain degree of uni‑ formity with regard to the timing of the exercise, and also to the informants’ edu‑ cation level and socio‑­economic status, most of the participants being university students or qualified for university‑­level education. An additional British control group was used for the later task (Questions 4–5), consisting of members of two additional groups of undergraduate students also surveyed at the end of a lecture at the University of Manchester. As with all the other data, responses which did not contain a causative form of one kind or another were discarded. The results are provided below.



4.3 Conventionalised scenario constructions in Singaporean English  

 101

4.3.3 Results Tab. 1: Responses to the elicitation survey, showing frequency of CSs relative to causative resultatives (CRs). Caused agency types and examples using other causative verbs are listed under other causatives. Numbers vary for each question, depending on the selection of a ­causative expression in response. Figures are in percentages.

Question 1 SgE (N = 61) BrE. (N = 51) Question 2 SgE (N = 48) BrE (N= 41) Question 3 SgE (N = 50) BrE (N = 41) Question 4 SgE (N = 22) BrE (N = 36) Question 5 SgE (N = 25) BrE (N = 34)

have CRs

get CRs

other causatives

1.6 15.6

13.1 64.7

0 0

85.2 19.6

4.2 4.8

47.9 78

6.2 9.7

41.6 7.3

30 53.6

10 46.3

0 0

60 0

13.6 72

0 7

13.5 18.6

73 1.3

4 26.4

12 59

0 0

84 14.7

CSs

Note: SgE: Singaporean English; BrE: British English (control group)

As can be seen in Table 1, the majority preference is for CS constructions for Sin‑ gaporean speakers, in answer to all questions. The following examples are typical of some of the responses received from the Singaporean group. (93) Have CRs a. I would go repair it or have it repaired. b. Actually I would bring it to a garage to have it serviced. c. Have it remove [sic.] d. You have it surgically removed. e. I would have it removed. f. Have it removed by an operation. g. Have it extracted. h. Have it extracted, of course! i. Go to the nearest photo booth to have my picture taken. (94) Get CRs a. I go to the hairdresser to get it cut. b. Get it repaired as soon as possible.

102  c. d. e. f. g. h. i. j.

 Chapter 4 Transitivity and causativity I’ll get it changed. Get it fixed. I will go to the mechanic to get it repaired. I’ll ask what I can do to get it removed. I will get it removed, go for an operation. Get it removed. Go to the instant photo shop and get one done. Go to the nearest photo booth and get myself ‘shot’.

(95) CSs a. Cut it. b. Tie it up, then cut it when I have the time. c. Cut. d. I cut it. e. Cut my hair. f. Cut it or style it with gel/mouse [sic.] h. Repair it. i. Replace it. j. Repair. k. Fix it. l. Take it to repair. m. Operate it. n. Remove it. o. Operate on it and remove it.36 p. I will remove it by surgery. q. Take it of [sic.] r. Remove it then.

36 Note that in response to Question 3, the range of conventionalised scenarios included the verb operate, which is often used in informal Singaporean English with the patient (literally‑­speaking) as the subject of the verb, rather than the surgeon, e.g., he operated yesterday can be uttered in full knowledge that the subject of the verb was not a qualified surgeon, but the person being ope‑ rated on. There is, therefore, no relationship of agentivity at all between the verb and the subject in such uses. Such expressions are not, strictly speaking, conventionalised scenarios, according to the constraints listed in (73), but have been classified as such for reasons of their functional similarity: they all express an action in which an implicit, indirect cause is involved (see Chapter 8 for more discussion). Such examples are perhaps directly related to the syntactic structure of many contact languages, which would not use a passive means of expressing such meanings. The causative counterpart to (95m) and (95o) is have/get it removed or have/get it operated on, the latter of which includes a preposition in the result‑­state complement. These examples have been included though the use of the verb operate was not found in the British control data.



4.3 Conventionalised scenario constructions in Singaporean English  

 103

s. I might remove it. t. Pluck it out.37 u. Extract the tooth. v. Extract and replant another tooth. w. If it is a back tooth I’ll just pull it out. x. Take another new one. y. Go take one at some photo shop. z. I will go to a photo studio/shop to take some more ID photos. (96) Other causatives a. I ask a guy to fix it for me.38 b. Get him to extract it. c. Let the dentist remove it. No additional questions were asked accompanying Question 4 to ensure that the subject does not extract his or her own teeth, as it was considered unnecessary for such a question (notwithstanding the horror that responses such as (95v) or (95w) evoke!), as the dentist was already a participant in the question eliciting the response. The remainder of answers to Question 4 revealed that have CRs were a possible alternative (11.6%) but not competing in the Singaporean dialect for the same function. As for all the other questions, there were a number of responses which did not use a causative alternative at all; these were discarded. It should be noted also for Question 5, that it was not felt necessary to add questions to determine the causee, as is clearly shown in (95y–z) that the causa‑ tive sense is understood and there will be an intermediate causee. For some of the responses to Question 1, which were intended to link to a supplementary ques‑ tion, it was clear that the respondents had made the connection, for example, for one who had answered with what looked on the surface like a CS, the sup‑ plementary question had evoked the response: I am my favourite hairdresser. For Question 2, also, the connection to the follow‑­up question seemed also to be understood, as one respondent replied: I do my car repairs at the motor shop near my place, clearly illustrating again with another CS that an implied causee participant was actually involved.

37 Verbs such as pluck for ‘pick’ or ‘pull’ are examples of what Ho & Platt (1993) describe as ‘lear‑ ned vocabulary’ alternatives frequently heard in Singaporean English. 38 The verb ask may not be considered a causative verb in some accounts, but it was found in some of the accompanying tasks that this verb was used in Singaporean English in place of get (+ NP + V) apparently as a more ‘polite’ substitute.

104 

 Chapter 4 Transitivity and causativity

Examples of data from the British control group included the following res‑ ponses (to all questions): (97) Have CRs a. I have it cut b. Have it removed c. Arrange to have it removed, as soon as possible d. Have it extracted e. Have it taken out and replaced f. Have it replaced, preferably on the NHS g. Have a false tooth fitted h. I would have a new photo taken i. Throw the old ones in the bin and go to the shopping centre to have a new set of photos taken in the photo booth. (98) Get CRs a. Get it cut b. I go to the hairdressers to get it cut c. Get it fixed d. Take it to a garage to get it fixed e. Get it removed f. Get a new photograph taken g. Get another taken h. Get new ones done i. I would go to a photo booth and get a new photograph taken (99) CSs a. I cut it (“My favourite hairdresser is Toni and Guy …”) b. I cut my hair when it gets too long (“My favourite hairdresser is called Tina”) c. Take another! d. Depending on what it was for – but I am most likely to take some new ones (100) Other causatives a. Let the dentist carry out whatever action was appropriate. b. Let him pull it out. It can be seen by the variation in the control group responses that the option of using a causative construction is not restricted to any particular lexical verb, no matter which verb appears in the participle. The absence of constraints sugges‑ ting the need for a human intermediate causee with the have‑­causative, as noted by Kemmer and Verhagen (1994), are not observed, though, in (97i) where the process is usually performed by a machine. The speaker who produced (97i) was



4.3 Conventionalised scenario constructions in Singaporean English  

 105

relaxing the constraint to refer not just to humans but to instruments also (as did the speaker who produced (93i)). However, the tendency for have to be used when there is an obvious intervention by a human causee, as in Q4, is still very strong. It is not clear from many of the get responses to Q5 whether the taking of the photo would be done by the causer or another participant, but it is possible for either. It could be hypothesised that such environments in which there is ambi‑ guity over the presence or absence of an intermediate causee, or those in which the causee is merely an instrument such as a photo‑­machine, lend themselves most readily to replacement by CS constructions; they are also environments of declining causativity, as are adversatives (such as Chris broke his leg), but unlike adversatives, can be reinterpreted more easily as a typical transitive construction.

4.3.4 Comparative overview Table 2 illustrates the average scores for the highest scoring causative construc‑ tions across both groups and both surveys. Tab. 2: Average of totals for causative‑­resultatives and conventionalised scenarios, in percentages.

Singaporean British Av. Totals

Have CRs

Get CRs

9.9 37 23.4

14.7 47.5 31.1

CSs 71 8.75 39.8

In order to explain the more frequent presence of the CS in Singaporean English, it is necessary to first understand why the construction is not as productive in the British data, as shown in Table 2. In the five lexical environments selected for ana‑ lysis, there are sufficient data (in Table 1) to be able to assess that the use of the CS in British English is confined largely to two out of the three environments selected, while in Singaporean English the distribution can be extended to a wider range of lexical environments overall. It can be hypothesised that there may be a combina‑ tion of factors controlling the use of CSs across the two dialects surveyed, on the basis of the present data, which may be operating independently of one another, one related to the restricted range of lexical environments in which the CS can be used in L1 English, as discussed above, and the other related to the reasons for extended generalisation of its use in the L2 or contact dialect, relative to its use in an L1 dialect. With regard to the first factor, the CS response to Q1 (main

106 

 Chapter 4 Transitivity and causativity

survey) represents a prototypical situation as the constraints discussed in (73) are in operation: I cut it/my hair is a situation in which the causer subject represents the dual roles of causer + undergoer, being partially affected by the causative act, the patient‑­object is an inalienable possessee of the subject, and the causative act represents a reflexive change of state. The CS may be less likely to be preferred in answer to Q2 (main survey), though, as the subject cannot be an affected under‑ goer, only a causer/recipient, the relationship of possession between the subject and the object NP is only very loose (one possesses a broken tail‑­light only by virtue of it being part of the entire car, which is the real possession, though still an alienable one). Q.3 is also less likely to motivate the use of a CS, according to (73), as although the mole is actually a possessee of the subject (inalienable or not), and the undergoer subject is certain to be affected by its removal, the con‑ struction is less prototypical in that the causative act is not reflexive and refers to direction away from the source (remove = ‘take away’, which is why examples such as She removed the litter from in front of her house do not render themselves unambiguously to interpretation as CSs either). Question 4 refers to a similar situ‑ ation in which direction away from the source is also indicated, though the tooth may or may not be an inalienable possession of the subject, and the undergoer subject is nevertheless affected by the action.39 Question 5 refers to a situation in which the possessor‑­possessee relationship is weaker, and the photo becomes an alienable possession of the subject. The subject is nevertheless both source and goal of the causative event, and the direction of the causative force is not away from the subject. Thus, it is observed that of the five lexical contexts for the use of CSs by both Singaporean and British speakers, the two most frequent, Q1 (85.2% SgE group to 19.6% BrE group) and Q5 (84% SgE group, to 14.7% BrE), correspond to (i) situations in which the direction of causative force is not away from the speaker, (ii) frequently‑­occurring situations in which any ambiguity with a simple tran‑ sitive would not usually be plausible, (iii) a possessor‑­possessee relationship between the subject and the object, and (iv) situations in which the subject is both the source and the goal of the entire causative event. With regard to (i), unlike Q3 and Q4, the responses do not contain reference to situations of removal or disposal indicating direction away from the subject; and with regard to (ii),

39 Note that the CS responses to the Questions 3 and 4 do not necessarily contain explicit syn‑ tactic reference to possession, as in, for example, I would remove my tooth, but the possessor‑­ possessee relationship is understood from the context and in the questions which elicited the CS responses, suggesting that the semantic constraints on the use of such constructions are not necessarily expressed syntactically but must be present in the discourse.

4.4 Discussion 

 107

the reflexive meaning of the CS construction does not allow ambiguity with a simple ­transitive construction, the subject being both source and goal or agent and undergoer of the caused event, unlike in a typical transitive construction in which the subject is only an agent, and the direction of causative force is usually away from the subject. There are indications in the present data, though, that frequency of use in certain contexts will eventually lead to the grammaticalisa‑ tion of CSs in English as a construction bearing the same surface structure as a transitive construction but different semantics from one. The second factor to understand is the reasons why such constraints may not be so frequently operable in Singaporean English. There is a high number of CS responses for all questions for the Singaporean group, the highest score being 85.2% for Q1 (Q2 – 41.6%, Q3 – 60%, Q4 – 73%, and Q5 – 84%). The substra‑ tum languages which do not use a passive form to mark recipient causativity may well be motivating the increased distributional range for most of the questions. Thus it could be hypothesised that both factors are in interaction in determining whether or not a CS will be used in a particular context: the overgeneralisation of the distribution to environments which may not be the preferred environments for British speakers (such as Q3 and Q4) could be the result of substratum and contact influences in Singaporean English, in which conventionalised scena‑ rios are a reflection of a similar means of expressing indirect causativity through single‑­clause constructions, or through particular serial‑­verb constructions. But this cannot explain all the tendencies.

4.4 Discussion It seems evident from the data that CSs are in partial competition with other cau‑ sative constructions, especially resultatives, when a causative construction is selected to answer the survey questions, and that the competition applies to both Singaporean and British dialects to varying degrees; i.e., it is not entirely attribu‑ table to contact. It is clear from the above data that the use of the CS construction is more prevalent in Singaporean English, though it is present in British English in certain lexical contexts, with scores of 19.6% of CSs for Q1 and 14.7% for Q5 for the British speakers. It is therefore obvious that substratum languages, in which resultative‑­causativity is not always morphologically encoded, cannot explain the entire story, and the British CS certainly does not reflect the influence of the contact constructions associated with Singaporean English. It is possible, then, that there may be reasons for the use of CSs in either dialect that are merely accelerated by substratum and contact features in the L2 dialect. Such reasons may include prag‑ matic devices, and the rate of grammaticalisation of certain constructions.

108 

 Chapter 4 Transitivity and causativity

4.4.1 Pragmatic mechanisms of causativity reduction The apparent shift between the constructions might be describable as a ­replacement; however, if so, it is not a lexical replacement which is taking place, but a syntactic one, in which an entire construction replaces the form that was previously used to serve the same function. Li & Thompson (1976a) discuss the syntactic replacement of serial verb causatives in the history of Mandarin Chinese with compound causatives, with which the serial verbs were in competition dia‑ chronically for a prolonged time period ranging from the 7th century BCE to the 9th century CE in total. The co‑­existence of two different syntactic devices for marking the same function made it appear as though the later development was derived from the earlier one; however, this was not the case, and the change that took place was a straight replacement of one form by another for the same func‑ tion. Similar replacements in grammaticalisation theory are termed renewals, or renovations (see, e.g., Hopper (1991); Lehmann (1995)), and one would anticipate that the same processes of renewal and renovation are equally applicable to the grammaticalisation of entire constructions. In the study of the relationship of conventionalised scenarios to causative constructions, a similar approach might be upheld, the common function to both constructions being that of expressing causee‑­less causativity. However, it cannot be revealed without a diachronic study that there is a chronological ordering of the two constructions; it is only known that the conventionalised scenario appears to have constraints related to situations of reflexive causativity, and is not just another (unconstrained) means of expressing the same functions as the causative‑­resultative; i.e., it is used only in situations in which it could not be mistaken for a simple transitive construction counterpart. This seems to indicate that it is simply a means of reducing excess morphological baggage in environments where such morphology is least needed. What needs to be defined, now, is whether there is a type of pragmatic device that results in such a derivation. The explication of the implicit action of the causative‑­resultative allows a conceptual link to the next construction, which in the present study, is considered to be in the form of a metonymy. In Panther & Thornburg (2000), as noted earlier, the possibility was discussed of a RESULT FOR ACTION metonymy to explain the way in which stative verbs could be used in the imperative with a result‑­state complement, e.g., have your passport ready. In the present study, it would seem as if the reverse were in operation: an ACTION FOR RESULT metonymy is the conceptual link which forms the bridge between the two constructions: the conventionalised scenario and the causative resulta‑ tive. Kövecses & Radden (1998) and Radden & Kövecses (1999) maintain that met‑ onymies, unlike metaphors, are often reversible, because of their location within the domain of ICMs. The ACTION FOR RESULT metonymy is also mentioned

4.4 Discussion 

 109

by Radden & Kövecses (1999), but as a referential metonymy only; e.g., in verb‑­derived nominals such as production, standing for ‘action of producing’. The process of metonymy is one which is exemplified by contiguous links of meaning; in the ACTION FOR RESULT metonymy, as shown, the links are a continuation of the former RESULT FOR ACTION metonymy, in which the action passively implied as part of the meaning of the participle is linked to the action expressed in the finite verb of a single clause construction, i.e., part of the meaning of the source form is transferred to the target through the contiguity of the lexical , Causative-resultative construction: X gets X s hair cut.

Sem

CAUSE-BECOME | means




GET




↓

↓

↓

↓

↓

V

SUBJ

OBJ

PP

Ø

Syn

IMY: Action for Result

Sem

Syn

CAUSE-BECOME | means




↓

↓

↓

↓

V

SUBJ

OBJ

Ø

, Conventionalised scenario: X cuts X s hair. Fig. 4: The ACTION FOR RESULT metonymy: metonymic links between the causative‑­resultative construction ( X gets X’s hair cut) and the conventionalised scenario (X cuts X’s hair). (Note: IMY = metonymic inheritance links.)

110 

 Chapter 4 Transitivity and causativity

constraints illustrated in (73). In this way, entire constructions may serve as ­metonymic ­substitutions for other constructions, as noted by Stefanowitsch (2003) for the ABILITY‑­FOR‑­ACTION metonymies involved in the interpretation of indirect speech acts (such as can you pass the salt?), and resulting in the production of grammatical metonymies in which an entirely new construction stands for an old one. The mapping of such constructional metonymies is made possible through the same models as are used, e.g., in Goldberg (1995), to illustrate metaphoric mappings between constructions, using inheritance links, as shown in Figure 4.40 The causative‑­resultative construction resembles the Resultative construc‑ tion of Goldberg (1995) with the exception that the subject is represented as the semantic causer‑­recipient of the action which is in fact attributable to an agent (causee) having no realisation in the syntax; i.e., it is implicit, and thus with the causer is defined in Figure 4 by the use of parentheses and non‑­bold font. The same pattern is continued in the conventionalised scenario construction, with the difference that the result‑­goal expressed in the syntax of the causative‑­resultative construction is no longer present and the passive action implied in the participle form is now reactivated in the main verb of the syntax in which the causativity and result of the semantics are fused into one predicate. Thus, what is normally conceived as a causative action verb, cut, now stands for both the caused result and action that caused it. The subject remains the causer‑­recipient of the action, and the implicit agent/causee that was not expressed in the syntax in the alterna‑ tive CR construction also remains unrealised in the CS. It could be argued that the subject in the syntax should correspond to the agent in the CS construction, but such an interpretation would make the CS con‑ struction identical to a transitive construction. It is clear that although the syntax is identical to that of a transitive construction, the semantics still accounts for the fact that the indirect causativity of the source construction is still implicit in the CS construction, although not expressed. Thus, like many other constructions in English which are transitive in surface syntax, there is no automatic correlation with a prototypical semantic representation of transitivity; i.e., containing a pro‑ totypical agent and a prototypical patient – they can be classified, as it were, as ‘false transitives’. The agent of the verb is an absent participant, and the subject remains logically the recipient of the action. In describing the metonymy as an ACTION FOR RESULT metonymy, the passive CR‑­construction is convertible to an

40 Goldberg & Jackendoff (2004) note that get‑­resultatives are classifiable as intrinsically resul‑ tative, or verbal resultatives. Inheritance links are given little specific definition in Goldberg’s (1995) introduction, but they usually are held to express the relationship of the inheritance of properties between derived constructions.

4.4 Discussion 

 111

active CS‑­construction, and an active verb form stands for both the former passive result form and the implied prior action together.

4.4.2 The subject role What remains to be explained is why examples such as (95o, p, s, and t) are s­ uperfically implausible as transitives and yet remain perfectly acceptable in Sin‑ gaporean English. The relative frequency of use of the CS construction across the two dialects surveyed may be associated with developing features of the contact situation. Examples from the Singaporean data, such as I will remove it by surgery appear to be implausible due to the unimaginable circumstances in which the subject is also the agent of the transitive clause.41 The development of the syn‑ tactic category of subject was discussed by Shibatani (1991) as a viable instance of grammaticalisation, which exhibits cross‑­linguistic variation, being further advanced in English than in languages such as Japanese. Shibatani considered the role of subject as the result of the merging of topic and agent nominal and the generalisation of agent over other semantic roles (1991: 103). The idea that sub‑ jects are grammaticalised topics is not new; Lehmann (1976) first introduced the possibility that subject‑­predicate languages in Indo‑­European were developed out of topic‑­comment languages. In English, the generalisation of agentivity in the subject has proceeded to include referents which are not necessarily agentive, such as inanimate objects; other languages do not always allow such referents to appear case‑­marked as subject; e.g., Spanish me gusta la cerveza (‘I like beer’), in which me is case‑­marked for a non‑­agentive, experiencer role in the dative, and German, in which inanimates cannot be translated as they appear in English middle‑­voice constructions: *Dieses Zelt schlaft funf ‘this tent sleeps five’ (1991: 102). Shibatani suggests that a clear distinction is made between a topic sentence and a topicless sentence, the topic sentence exhibiting less of a tightly integrated unit with the discourse comment than the subject does with its predicate, and that some languages, e.g., Tagalog and Cebuano, demonstrate that the diachro‑ nic change of a topic into a subject is currently still in progress (1991: 105).

41 They also appear to be unlikely candidates for CS constructions, if the constraints listed in (73) require that the action described cannot be in a direction away from the source. However, they share with other CSs the fact that the subject‑­causer is nevertheless, affected by the action, and in all but the adversative cases, a benefactive recipient. Such examples will be discussed further in Chapter 8.

112 

 Chapter 4 Transitivity and causativity

While the sentence topic in English is relatively invisibly merged into the subject role, there are instances in which it may be still seen to exert a presence in the selection of subject, for example, Shibatani’s example A boy is in the garden is considered to be intuitively less acceptable than There is a boy in the garden, since the subject, a boy, is indefinite and therefore not mentioned previously in the dis‑ course as given information, a requirement of all topics. What may be acceptable, then, as subject, needs also to be acceptable as topic, indicating the prevailing influence of earlier stages of grammaticalisation, a tendency summarised in dia‑ chronic grammaticalisation as retention (Bybee and Pagliuca 1987) or persistence (Hopper 1991), as mentioned above. In English, however, the senses of agentivity which have been generalised over the subject nominal emerge as what were once referred to as selection restrictions (Lehmann (1976); Li and Thompson (1976b)). Little understanding was given at the time to formally defining what selection restrictions may repre‑ sent in semantic terms. It is also stated by Li and Thompson (1976b) that topics do not in general exhibit evidence of selection restrictions with the verb. Langa‑ cker (1987: 282) describes selection restrictions as merely the conflict in semantic specifications between two entities placed in correspondence, and demonstra‑ tes that they can apply to noun‑­adjective constituents as well as to subject‑­verb constituents. What this may suggest for the purposes of the present study is that for native speakers of substratum or contact languages with primarily topic‑­ prominent sentence structure, there will be a much wider distribution of verb arguments than for those with subject‑­predicate structure. In the case of Sin‑ gaporean English, the majority of the influencing contact languages are all of the former type, and it has been claimed that for Singapore Colloquial English, topic‑­comment structure is preferred to subject‑­predicate structure (Bao 2001), allowing all major phrasal categories to serve as topics. This does not mean that CS constructions in Singaporean English do not typically express subject‑­ predicate structure; however, the influence of the contact languages’ sentence structure on the local English dialect means that a wider range of subjects may be permitted than is usual in other dialects due to the possible overgeneralisa‑ tion of subject‑­predicate relations. Thus, for some Singaporean speakers, con‑ ventionalised scenarios, such as I cut my hair, and I extract it (a tooth), may tend to be more acceptable since the agentive associations manifested as selection restrictions are not so salient in a language in which subject‑­predicate informa‑ tion structure may be less than fully grammaticalised. On the other hand, the British speakers will tend to reject the alternative use of such forms if the rela‑ tionship between the verb and the subject is understood as more agentive, and more tightly integrated as in a subject‑­predicate sentence structure. For British speakers, then, the retention of agentive senses associated with the prototypical

4.4 Discussion 

 113

subject‑­predicate relationship may inhibit the development of conventionalised scenarios, especially when they conflict superficially with identical transitive constructions in which no intermediate causee is implied at all. For the Singa‑ poreans, greater freedom of discourse structure patterned on substratum and contact language models permits a looser, more context‑­based interpretation to be given to such expressions; hence the greater frequency of use and wider dis‑ tribution of contexts shown. Although Shibatani (1991) and Li and Thompson (1976) pre‑­date much of the more current work on the grammaticalisation of constructions in general, the discussion of the subject category as a target for grammaticalisation ­presumes the accompaniment of the appropriate construction types with which it is normally associated; i.e., subject‑­predicate constructions, which may be ­considered as more schematic instances of transitive constructions. The more frequent use of non‑­ adversative conventionalised scenarios by Singaporean speakers means that for such speakers, the transitive construction has been ­hyper‑­grammaticalised (Ziegeler 2000a: 12). Hyper‑­grammaticalisation may be described as ‘­over‑­generalisation’ of a grammaticalising form or construction. It may be c­ ontrasted with ‘hypo‑­grammaticalisation’, or under‑­generalisation of a grammaticalising construction, which is often related to constraining contact features in the languages in which it occurs. Hyper‑­grammaticalisation refers to grammaticalisation patterns extending beyond its standard distributive ­restrictions, due to the less frequent influence of semantic retention as a cons‑ traint on grammatical distribution; in the present case, the less frequent asso‑ ciations of causativity and agency in the semantic relationship between subject and verb. In other words, the subject‑­predicate structure of the CS in Singaporean English is an overlay on the topic‑­prominent information structure of the substra‑ tum and contact languages, and is applied relatively unconstrainedly. The presence of the more conventionalised adversatives (e.g., He cut his finger), in which there is not likely to be any ambiguity with an agentive tran‑ sitive construction, is evidence that there can be more rapid reduction in the expression of indirect causativity when the CS is not likely to be ambiguous with an existing transitive construction. However, it may be questioned what the difference is between these adversative examples of apparent self‑­ mutilation and the examples provided by the Singaporean speakers referring to such actions as extracting one’s own teeth and performing surgery on oneself. The difference is again, plausibility, since many of the actions described by such bizarre examples cannot be performed on oneself, which means that an ordinary adversative is more acceptable due to the possible absence of cause altogether; i.e., the action it describes can be attributable to accident. However, getting one’s hair cut can scarcely qualify as an adversative, and yet Question 1

114 

 Chapter 4 Transitivity and causativity

seemed to yield the highest statistical frequency from both groups. Whatever the case, the underlying presence of topic‑­comment information structure still permits a higher incidence of CSs amongst the Singaporean speakers, and this is thus an enabling factor, rather than a causal factor, for the Singaporean spea‑ kers to use the construction with more frequency. Whether this will eventually result in the replacement of one construction with another is a matter for future research.

4.5 Summary It has been demonstrated in this study that there is variation shown in the expression of causative result across contact, and the relationship between the causative‑­resultative construction and the conventionalised scenario with which it is in (partial) competition has been hypothesised to be a metonymic one, in which the action implicit in the result‑­state in the form of the passive participle of a CR construction is expressed as a active predicate in the single‑­clause CS construction. The metonymy has been identified as the ACTION FOR RESULT metonymy, i.e., a stage further than the metonymy used by Panther & T ­ hornburg (2000) to account for constructions in which a typically stative verb used in the imperative is associated with producing a final result‑­state. In the ACTION FOR RESULT metonymy the implicit action/performance of the passive participle becomes explicated in a single‑­clause structure, resulting in an equivalent cons‑ truction meaning alternate of a CR‑­construction. The relationship of the surface subject to its semantic role, though, remains unchanged in the CS construction, since the causative agency of an intermediate participant is still left with zero‑­ expression in the CS. The metonymic extension between the two constructions is therefore only related to the nature of the predicate, in which a non‑­focal element in the meaning of the CR, the implicit action, is manifest in the CS construction as an active predicate. The metonymic link between the two constructions, however, explains only the conceptual relationship between the two competing causative constructions, not the main factors motivating greater frequency of use of CSs in Singaporean English. The greater frequency is more probably due to the structure of the sub‑ stratum and contact languages, coupled with the weaker selection restrictions between subject and verb due to the prevailing topic structure of those langu‑ ages with which the variety of English spoken is still in daily contact. This does not mean that the subject always is understood as a topic in such dialects, but it does mean that the role of subject imposed as an overlay feature on the under‑ lying topic structure will not be so semantically constrained with regard to the

4.5 Summary 

 115

retention of agentive characteristics of the NP to which it refers. Much more work still needs to be done in this area; thus the final picture is very complex indeed, and there are four factors hypothesised to be influencing the variation: (i) the semantic retention from earlier grammaticalisation stages of the prototype tran‑ sitive construction that constrains the distribution of the CS in non‑­contact dia‑ lects; (ii) the substrate/contact model patterns permitting wider distribution of the CS in Singaporean English, (iii) the relatively unrestricted distribution of CSs in Singaporean English due to the hypergrammaticalisation of subject‑­predicate relations, and (iv) the grammaticalisation of the transitive construction diachro‑ nically to extend to ever more schematic environments in which it is syntactically, though not semantically, ambiguous with its prototype illustrated in Figure 3. These four factors will be taken into consideration in the proposal of a transitive merger construction‑­type in Chapter 8.

Chapter 5  Experiential aspect 5.1 Introduction The present chapter first appeared in a paper titled: ‘Experiential aspect in ­Singapore English: the depolarisation of ever’, for the Groninger Arbeiten zur ­germanistischen Linguistik 53.2, and leaves open still many unanswered questions. The adverb ever, used affirmatively, which in SCE has the meaning ‘at least once’, is one of the most puzzling items of the contact grammar of Singapore Colloquial English. Its function is that of serving to lexify the implicit experiential aspect of the present perfect of indefinite past in Standard Singapore English, which is unmarked as a distinctive category in the latter sub-variety. In terms of the analysis represented in the present volume, it functions as a construction type which, for present purposes, may be labelled the experiential construction, since it replicates a function typical of the Chinese substrate of Singaporean English, an experiential perfect marker in many dialects of Chinese (e.g., the Mandarin verb guò – ‘pass’), as observed by Bao (2005) and Ho & Wong (2001). As a construction type, then, it carries the form of the lexifier, (standard) English, but the function of the languages of contact, and thus is worthy of a merger construction analysis, as a form-meaning relation consisting of a single unitary item. The present chapter will examine a number of possible explanations for the reason that ever was recruited in the Singaporean situation to express the substrate functions of experiential aspect, and endeavour to account for its use more accurately by the postulation of a common semantics across contact. At the same time, it will be suggested also that the selection of the source for the grammaticalisation of experiential aspect in SCE cannot avoid reference to universal strategies of reanalysis often found associated with similar developments in non-contact situations; i.e., it is necessary to explore the possible diachronic pathways of development which led to the adverb becoming a negative-polarity item in the first place. The problems to be analysed are thus, as in earlier chapters, semantic: first, how a speaker selects a form from the lexical source language for representing structures of the substrate (or model languages) and second, how this random selection can be explained in terms of semantic continuity. Semantic continuity is essential for the construction-convergence model as introduced in the present volume, as well as for many processes of pragmatically-motivated grammatical change, as outlined in many previous studies, e.g., Traugott & Dasher 2002, Ziegeler (2006). In the next section, the present-day use of ever in Singapore English will be described, as recently outlined in the previous literature; section 5.3 will discuss the role of substrate structures in Chinese and other contact languages, and 5.4 will

118 

 Chapter 5 Experiential aspect

raise questions of transfer via either ordinary contact-induced grammaticalisation or replica grammaticalisation (Heine and Kuteva 2003, 2005). In section 5.5, the diachronic development of ever in the history of English will be briefly reviewed. In section 5.6 the logical relations between negatively-polarised contexts and the use of ever as an experiential marker will be discussed, and the basis for the selection of the adverb as an experiential marker will be established, enabling a more parsimonious explanation for the functional role it plays in SCE today than previous studies have provided. Sections 5.7–8 will summarise the findings

5.2 The ever construction 5.2.1 Negative polarity ever Israel (1998) establishes four basic functions for present-day ever: Existential, Universal, Emphatic, and Derivational, supplying the following examples:42 (101) a. Existential: Have you ever robbed a liquor store? b. Universal: COLAs … are positive feedback loops that drive inflation ever upwards. c. Emphatic: IBM is like a battleship ever so slowly turning in the wind. d. Derivational: You can leave whenever you want. Israel describes ever as a polysemous, negative polarity item (NPI), and concludes that polarity sensitivity is related to its capacity to encode emphasis and informativity. Informativity is related to a scalar notion, in which items with a high quantitative value are maximally emphatic in positive polarity contexts, while those with a low quantitative value are maximally emphatic and thus also informative in a negative polarity context. Universal ever is a positive polarity use and Existential ever is a negative polarity type. In terms of emphasis, though, it can be further observed that in (101a and 101d) ever marks emphatic possibility, and in (101b and 101c) it marks emphatic necessity; the underlying meanings of the adverb, are therefore, vaguely modal in character. Israel (1998) also compares ever with any as a free choice item, eliminating the possibility of a free choice ever on a number of grounds, significant among them being that while any applies to individuals as a nominal determiner, ever applies to propositions as an adverbial qualifier, and that the differences amount to just that. From the functional

42 Most appear in his corpus of examples from the Wall Street Journal, though Israel does not supply the date span.

5.2 The ever construction 

 119

­ ifferences between the two items in standard English, it may seem at first glance d that there is no particular reason to align experiential ever (henceforth EE) in SCE with free-choice any. However, it will be seen that from the functions to which it has been extended in SCE, it may be the case that there is good reason to make comparisons with free-choice any. The typical contexts of negative polarity are listed by Israel (2004) as including: the scope of negation, including negation expressed by adverbs such as hardly, and rarely, as well as not, negative quantifiers such as nobody, nothing, or never, the complements of predicates such as doubt, be surprised that (to which one could also add regret that, deny that, and forget that, wonder whether, amongst others), the antecedent clause of a conditional, the restriction of a universal or generic quantifier, the nuclear scope of only, and the focus of a polar interrogative, as in (101a). Israel (2004) adds to this list rhetorical information questions, comparative and equative constructions, and subordinate clauses introduced by before and long after. Israel (2006) also notes the frequent presence of what are known as minimisers in negative polarity contexts, such as a wink in I didn’t sleep a wink. Such minimisers often express a minimal degree or measure, which contributes, according to Israel, to the emphatic nature of an expression; they are attracted to negative polarity because this is where they are most informative, and Israel (2006) includes ever amongst these items. It will be seen below that there is adequate evidence of the use of negatively-polarised ever (NPE) cooccurring alongside EE in SCE, and that its experiential uses are not the only ones used by Singaporean speakers. However, it remains to determine how speakers of SCE could de-polarise the use of the adverb when its negative-polarity use is found alongside it.

5.2.2 SCE ever It was noted above in Chapter 2 that Mandarin Chinese is rapidly becoming a vigorous language of contact in Singapore, being spoken alongside English by the majority of the population. The use of ever in marking experiential aspect in SCE reflects a functional category present in Mandarin Chinese but not ­distinctively marked in SSE, and SCE does not recruit the counterpart lexical conceptual source from Mandarin (a verb meaning ‘pass’ or ‘cross’) in order to grammaticalise the category. Ever has been claimed as having its origins instead in Hokkien and Malay (Ho & Wong 2001) but in the present study, this will be open to question also, since there are no conceptual counterparts identifiable in those languages either. The need to search for a more universal strategy in the contact grammaticalisation of the aspectual function is therefore compelling.

120 

 Chapter 5 Experiential aspect

Chapter 2 also discussed the many earlier studies in which the aspectual system of Singapore English mirrors very closely that of Mandarin, and studies have referred frequently to the adverb already as a marker of perfectivity (e.g., Bao 1995, 2005), with its functions parallel to those of le in Mandarin Chinese. It was also noted that Ho and Platt (1993) discussed the past tense system in Singaporean English as reflecting the aspectual functions of the substrate rather than expressing tense per se; they also discuss the use of the progressive aspect with stative verbs, as does Bao (2005). The use of experiential ever (henceforth EE) as a marker of experiential aspect, however, has been less frequently discussed in the literature, though some earlier studies provide an introduction, such as Ng (1999), Ho & Wong (2001) and Bao (2005). Ho & Platt (1993), as well as Low & Brown (2003), also mention its usage in Singaporean Colloquial English appearing in affirmative statements that are not found in standard varieties today, for example (Ho & Platt 1993: 76): (102) I ever go/went/gone dere which can be roughly glossed in Standard Singapore English as ‘I have been there’, with an additional implicit sense conveyed of ‘at least once’, in which the present perfect alone conveys the indefiniteness in specification of the anterior event in the time period leading to the moment of speaking. Such specification requires explicit marking in SCE, since, as Ho and Platt (1993) and Ziegeler (1995) noted, the functions of the present perfect in general are often merged with those of the past tense, and the past tense conveys perfectivity (of a unique event). The result is that the range of functions associated with standard varieties such as the experiential sense expressed in (102) may be relatively restricted, and the present perfect in SCE may not express precisely the same meanings as in SSE, thus indicating a need for an additional marker of indefinite anteriority. Ho & Platt (1993) also mention the use of never as a general form of negation which is not restricted to co-occurrence with perfect aspect, suggesting that it may be influenced by non-standard negation from standard varieties, but they pursue the possibility no further. Thus, little research has been undertaken, to date, on the curious depolarisation expressed by ever in SCE, though some preliminary work can be found in students’ academic exercises, for example, Ng (1999) provides an introductory description to the usage of ever as an experiential marker in SCE, attributing the function, not surprisingly, to a corresponding pattern in Mandarin Chinese. She notes a number of correlations with the Mandarin Chinese verb, guò, for example, (i) it expresses the perfective viewpoint, (ii) it is restricted to situations that show a departure from one stage to another, and (iii) it co-occurs with stative verbs or with non-statives, specific or indefinite situations, expressing

5.2 The ever construction 

 121

new situations in co-occurrence with statives, and acquired experiences in cooccurrence with non-statives. Most importantly, it is restricted to events that can recur, for example in: (103) John ever eat/at/eaten this apple before. the intended meaning can only be that John has eaten this type of apple before, not a particular apple, since an apple can never be eaten more than once.43 Ever is also seen to occur in habitual situations of extended duration in the past (1999: 27): (104) John ever play/played truant everyday, but doesn’t do so anymore. Like other Singaporean accounts, Ng concludes her study by attributing the use of ever to strong substrate influence, suggesting it is a calque of guò in Mandarin Chinese, an experiential perfect marker with the function of marking events that took place at least once in the past. The only difference in usage, it would appear from Ng’s study, is syntactic positioning: in Mandarin, the experiential marker follows the verb, while in SCE, it always takes a position preceding the verb, in accordance with the constraints of the standard contact variety, SSE. Such a tendency is therefore related to the respective word order patterns in both languages, due to a factor that has been described by Bao (2010, for example), as the lexifier filter in contact (as will be discussed in Chapter 8). The argument for a calque could also be affected by the fact that the two forms are not parallel in grammatical categorization: ever is an adverb, and guò is a former lexical verb (meaning ‘pass’, ‘cross’, having grammaticalised further to function as a directional complement, an anterior aspect marker, and later an evidential (Chappell 2001; (1992: 83), though it still retains its verbal source functions). As such it is most likely to occupy a post-verbal position. Ho & Wong’s (2001) study is one of the few published accounts of experiential ever in SCE and it uses data collected from a spoken corpus of 300 transcripts and interviews, as well as written samples of students’ work at secondary and undergraduate levels. Amongst the functions illustrated by Ho & Wong are included spoken examples in which the adverb may appear as a response to polar interrogatives (2001: 80): (105) a. A. Your husband ever bring fish home to eat or not? B. Ever.

43 Wu (2008) provides some counterexamples to this constraint; they will not be discussed in the present paper. However, it is observable that most of the examples Wu offers have multipleparticipant subjects, or at least the presupposition of other participants in the discourse, thus suggesting repeatability across participants, if not events.

122 

 Chapter 5 Experiential aspect

b. A. You ever work in other jobs ah? B. Yes, ever work in other jobs also. In (105) the use of EE co-occurs with its standard use in negatively-polarised counterpart (NPE), giving the impression that EE was simply formed on analogy with NPE. Ho & Wong note that EE may appear in exchanges in which it has not been used in the preceding question: (106) A. Do you go to Change Alley? B. Oh! Change Alley, ever. In (106) the time reference of the response is not the same as that of the question, referring to the indefinite past rather than the indefinite present. Other, noninterrogative contexts include: (107) This share ever hit forty dollars! in which it seems to convey the same meaning as ‘once’. EE was also observed by Ho & Wong as co-occurring with the adverb before, e.g., I ever seen you before. Ho & Wong define the main functions of ever as expressing an action that has taken place at least once before in the past (2001: 82). More examples of EE are available from a local Singaporean internet site: (108) ya, pretty interesting to me, was one of my favourite modules in poly[technic university]..[I] ever thought of joining this industry, but heard lots of people saying tt [‘that’] u play the bad guy in the co[mpany] and stuff. http://flowerpod.com.sg/forums/Career-Talk-f28.html&st=40 (Posted by: diamonds Aug 21 2007) However, examples also appear on the same site by the same speaker with NPE occurring alongside its experiential use: (109) oo.. thats good. i think if i ever work in an office, i’ll choose to work in hr[‘human resources’] dept. http://flowerpod.com.sg/forums/Career-Talk-f28.html&st=40 (Posted by: diamonds Aug 21 2007) The ratio of EE to NPE uses is not significant, standing at 1: 14 in the topicselected corpus of 85,909 words (including posting information) from which the above examples were selected.44 Thus, it can be seen, at least in this small sampling, that there is adequate evidence that EE has not ousted the negative polarity

44 Also appearing was one example of a universal usage (see (126)).



5.3 Contact and substrate languages 

 123

functions of ever, which co-occurs quite happily alongside it. It is hardly likely, furthermore, that speakers of SCE are insensitive to its negative polarity usage either. The presence of EE in Singaporean English requires explanation from the point of view of contact transfer, but many established theoretical approaches to contact (e.g., Lefebvre 1998, 2001) do not provide sufficient an explanation to account for the selection of one conceptual source item in the lexifier over another, relying almost exclusively on substrate explanations). The following section will therefore review possible substrate reasons for the selection of this adverb in SCE.

5.3 Contact and substrate languages Chappell (2001) argues that all the experiential markers in Chinese are more accurately described as expressing evidentiality, and lists a number of features associated with such markers in a variety of Chinese languages, including Taiwanese Southern Min, Fuzhouese (Northeastern Min), Hong Kong Cantonese (Yue), Changsha (New Xiang), Hakka, Gan (Nanchang), Shanghainese (Wu), and standard Mandarin. They include: (i) immediate or inferential evidence and first person split (ii) discontinuity, repeatability, and (incompatibility with) verbs of destruction (iii) contrary-to-expectation meaning (iv) phase complement marking for completion (v) partitive meaning (‘partial effect on an object’) (vi) future reference clauses and imperatives (‘do VERB again’), (vii) irrealis and conditional uses (viii) scope of evidential markers. Of all these features and functions associated with cognate forms in the Chinese languages, it may be argued that it is only (ii) restriction to discontinuity, repeatability, and incompatibility with verbs of destruction that are also associable with the functions of ever in SCE. Thus only a restricted domain of functions of guo and its cognates can be said to be carried over into the experiential aspect marker in SCE, and it is for this reason that the present study will continue to analyse ever in SCE as an experiential marker rather than an evidential one – it does not grammaticalise the full range of functions demonstrated by guo and its cognates. With respect to the languages with which SCE is in contact, Ho & Wong find counterparts not only in Mandarin Chinese guò, but also in Hokkien koe/khi and Cantonese gwoh/kwo (2001: 84), all possible cognate forms of guò in ­Mandarin.45 Such items may be considered as merely sharing the same grammatical function as ever, as noted above, though naturally not transferring the conceptual source

45 Note that there is no strict standard for the romanisation of non-Mandarin Chinese dialects.

124 

 Chapter 5 Experiential aspect

of grammaticalisation to the SCE experiential aspect marker (see d ­ iscussion below). However, in addition to the ‘pass’ verbal sources for the aspect marker, they also note another form in Hokkien, which is rendered as bat or pat, suggesting a meaning of ‘ever’ and that it appears pre-verbally, as does ever in SCE, implying a closer word order correspondence between the two functional matches (2001: 85): (110) Goa bat khi I EVER go ‘I’ve been to Japan (before)’

Jit-pun Japan

It is also found in response to questions, just as with ever: (111) Q. Li bat khi You EVER go ‘Have you ever been to Japan?’

Jitpun Japan

bo? not

A. Bat. EVER ‘I have/Once’46 Ho & Wong also note the occurrence of pernah in Malay, which they suggest is lexically translatable as ‘ever’ and also precedes the verb and can be used as an affirmative response to a question, in the same way as bat in (110–111). For example: (112) Q. Awak pernah-kah pergi You EVER-QP go ‘Have you ever been to Japan?’

ke to

Jepun? Japan

A. Pernah. EVER ‘I have/once’ Again, the word order matches that of SCE ever,which is used pre-verbally, unlike Mandarin guò and its cognates. Because of such parallels, Ho & Wong conclude their brief study by suggesting that (rather than Mandarin), it is the influence of Hokkien and Malay that has motivated the use of affirmative ever in SCE. They also allude to the derivation of ever from antonymous never, with the interpretation ‘at any time’ being a possible route of introduction, but go no further with

46 The glosses in (110) and (111) are my own.



5.3 Contact and substrate languages 

 125

this ­possibility; nor do they explain the loss of negative polarity entailed in its SCE transformation, or the fact that EE and NPE co-occur in the same dialect used by the same speakers, as well as in SSE which is also in contact with SCE. As such, their explanations leave room for further exploration into the reasons for the selection of ever in SCE. It could be questioned whether the Malay form, pernah translates as ‘once’, since its experiential aspect gloss in (112A) is not possible in standard usage, only in SCE. In Hassan and Muhammed (1994), the English adverb once glosses in Malay as dulu, sekali, segera, bekas, and dahulu, though there is no indication of the different contexts in which these forms may be used. Pernah does not appear as the translation of ‘once’, but instead, of ‘ever’, alongside sentiasa, selalu, and selamanyu, so in this way, it accords with the claims of Ho & Wong for a direct calquing, but also indicating that Malay must have a form with the meaning of positive polarity ‘ever’. However, if so, it makes it translatable in standard English only by ‘at least once’ as its nearest equivalent in meaning, a translation they do not use possibly because of the many alternative glosses for ‘once’ in Malay. Since once is not represented by pernah in Malay either, we are left wondering what the true source meaning of this adverb really was, and how it could be so readily translated in affirmative contexts by an adverb that is normally restricted only to negative polarity contexts in Standard Singapore English.47 Native speakers of Malay offer different suggestions, pernah being considered to have the status of an auxiliary (Foong Ha Yap, p.c.), as it can co-occur with a question particle (QP, as shown in (112)). The lexical source, however, is not transparent, though Foong considers it may be cognate with a similar form, parna, in Ilokano, a Philippine language, as shown in the following: (113) Kalman parna yesterday have ‘Yesterday there was rain.’

tudu rain

In such uses, it appears to express existentiality (via the H-possessive schema – see Heine & Kuteva (2002)). If this is the source in Malay also, then there is a viable case for a polysemous link between pernah and ever in a shared sense of existentiality of an event. No similar source meanings appear in the Sinitic

47 It is also difficult to assume that ever is used simply as an adverb meaning ‘once’ in SCE, as it can co-occur non-redundantly alongside once. This is shown in the following (personal observation) of a speaker of Singaporean English: I ever had a grant once that was funded by Melbourne (24/2/14).

126 

 Chapter 5 Experiential aspect

languages studied by Chappell (2001); the possibility of pernah as the source for ever therefore is still open to question. The translation of ‘never’ in English is tidak pernah ‘not ever’, but it is clear that pernah is not restricted to negative contexts. In the same way, bat in Hokkien may also be negated (Chappell 2001): (114) i m bat siū koè goá ê khì 3SG NEG EVD receive EVD 1SG L anger ‘She has never before borne the brunt of my anger.’48

koè EVD

(114) illustrates the co-occurrence of both koè and bat in the same utterance, both grammaticalising evidential functions, indicating a functional overlap or layering (Hopper 1991), as Chappell indicates: it is possible that there is also a replacement strategy in progress here. The experiential markers are, furthermore, glossed as evidential markers in order to illustrate the equivalences proposed by Chappell. The appearance with negatives in both Malay and Hokkien leads one to assume that there are close parallels with the development of the SCE form, although it is not possible to determine for certain that the negative form might have preceded the introduction of the positive use in these substrate languages. However, this still does not explain, of course, why the form ever, with no functional parallel in affirmative contexts in present-day standard English, was selected to cover the needs of affirmative experiential markers in the model languages. Bodman (1955) does not supply an affirmative form of bat, though there are verbs appearing in the vocabulary lists that are romanised as bat with the meaning ‘know’, ‘be able.’ The lexical source of this form was indeed a verb with the meaning of ‘to know, by experience’ (Hilary Chappell 2001; Chappell 1992: 83) which lends itself easily to a meaning suitable for the grammaticalisation of an experiential marker referring to actions experienced at least once. Furthermore, Chappell (2001) illustrates the use of similar lexical sources from verbs meaning ‘know’ as experiential/evidentials in other Min languages. Bao (2005) also discusses ever in SCE, and relates this marker to the more immediate and conspicuous influence of the adstratum language, Mandarin Chinese, rather than the older, less-used contact dialects such as Hokkien and Cantonese. He does not mention Malay as an influence, but also notes that as for Mandarin guò, and as discussed for (103), the aspect marker may not be used if the event it marks is not repeatable (Bao 2005: 245): (115) ?He ever old ‘He was once old (and no longer is)’

48 Chappell provides the Hokkien Chinese characters in the original example. (EVD = evidential marker; L is not explained in the original.)



5.4 Contact grammaticalisation as a possible explanation 

 127

Thus, ever is lower-bounded rather than upper-bounded in scope in that it must occur with events and states that occur at least once, and have the possibility of recurring (the same restrictions may apply to NPE, e.g., ??Was he ever old? – though see (126) below) . Bao also discusses the use of never, which in SCE is seen to share closely the functional specification of meǐ in Mandarin Chinese, rather than bù (used with stative verbs), as a perfective negator. Bao’s principle hypothesis is that the entire aspectual system of Chinese has been transferred wholesale to the situation in Singapore Colloquial English, where it becomes relexified. This type of transferral may well accord with a closely-related theoretical analysis of contact grammaticalisation (Heine & Kuteva 2003, 2005), which to present knowledge has not been taken into account in a great deal of depth in the Singaporean situation. However, it is obvious there are different possibilities with regard to substrate claims: Hokkien, Malay and Mandarin all having a role to play, according to previous research, and the claims for calquing sources are many and varied. It is for such reasons that other analyses of contact situations should be examined, in order to determine if they have any motivation in the selection of ever as a means of grammaticalising experiential aspect in SCE.

5.4 Contact grammaticalisation as a possible explanation Grammaticalisation theory has recently come to the attention of researchers working with contact languages, as is evident, for example, in the work of Heine and Kuteva (2003, 2005), Mufwene (2008), and Bruyn (2009); see also volumes such as Wiemer et al. (eds. 2012). A frequent observation associated with grammaticalisation in contact situations is its relatively accelerated rate of ­development: what normally takes up to one thousand years to accomplish in terms of ordinary grammaticalisation situations will often happen very suddenly, perhaps over only two or three generations in a contact situation such as in the development of pidgins and creoles (Heine & Reh 1984: 89–90). The reason for this is that a contact situation in linguistic terms is a situation of communicative urgency (Communicative Pressure – Hagège 1993) with greater need to move to more advanced levels of automation of the language system in as short a time period as possible.49 An alternative argument for the speed at which contact ­languages grammaticalise is offered by Ansaldo (2009: 111), who believes that it is the restricted pool of

49 Givón (1989: 251–261) discusses grammaticalisation across a number of different dimensions as an example of the autom(is)ation of the code; i.e., a process associated increasingly with repeated, rehearsed tasks and routinised, conventionalised activities, illustrating a bottom-up shift from attended processing to automated processing. The accelerated automation of the system found in contact situations is thus representative of accelerated grammaticalisation.

128 

 Chapter 5 Experiential aspect

speakers in a contact situation which contributes to faster development of grammatical morphology. The contact situations most often described in the literature include those of trade and commerce, or situations in which a large number of people of different language backgrounds are brought together for some functional reason, and require as efficiently as possible to establish a conventionalized lingua franca in order to carry out day-to-day transactions. Early Singapore was one typical example, as noted in the preceding chapters, though, as also noted in the literature, English may not have been the first lingua franca to appear on the trading scene in Southeast Asia. Perhaps it is Hagège’s (1993) hypothesis of Communicative Pressure which is the driving force behind the use of certain forms in an over-extended sense: they become used preemptively in environments which in older varieties of the language are not yet part of its distributional range. Although it is anticipated that the forms may eventually extend their range of uses to an increased number of environments in the lexifier, if they do, the time taken will be much longer, generally. This may not mean, in the present case, that the standard variety is also on the verge of developing a experiential function for ever, but in the case of Singapore, the functional needs of the substrate languages are not accounted for in the lexifier as transparently as they might be: thus, the communicative pressure is still present in the urgency to articulate such needs with whatever lexifier material may be suitably at hand. Heine & Kuteva (2003; 2005) discuss (ordinary) contact-induced grammaticalisation as a strategy for transferring some grammatical concept from the model language (M) to the replica language (R) (2003: 533). This strategy involves the following stages: (116) a. Speakers of language R notice that in language M there is a grammatical category Mx. b. They develop an equivalent category Rx, using material available in their own language (R). c. To this end, they draw on universal strategies of grammaticalisation, using construction Ry in order to develop Rx. d. They grammaticalise construction Ry to Rx. They also note that the process is a gradual one and may take several centuries to complete, especially at the last stage (d). Quite often, the R language is a pidgin or creole, and the M language is a substrate (though Heine & Kuteva do not employ such terms, as the theories are expected to have a universal application to any situation of contact, not only those of pidgins and creoles). The process outlined above does not presuppose that the selection of the lexifier item to grammaticalise Mx will be determinate; in general, what is selected is often a crosslinguistically



5.4 Contact grammaticalisation as a possible explanation 

 129

universal lexical source, such as the form already in SCE, the Ry used to express the perfective aspect Mx in the Chinese substrate languages, as pointed out in Matthews and Yip (2009). It was noted by Dahl (2006) that forms with the meaning of ‘already’ or ‘finish’ (what he labels lamitives) are an areal feature of the ­Southeast Asian region. This justifies reference to universal strategies of grammaticalisation in re-creating the function of perfective aspect in SCE as ordinary contact-induced grammaticalisation (as previously suggested by Matthews and Yip (2009). The processes may therefore be mapped in the following way: (117) a. Speakers of language R notice that in language M [= SUBSTRATE CHINESE DIALECTS] there is a grammatical category Mx [= PERFECTIVE ASPECT]. b. They develop an equivalent category Rx [= SCE PERFECTIVE ASPECT], using material available in their own language (R) [= ALREADY] c. To this end, they draw on universal strategies of grammaticalisation, using construction Ry [=ALREADY] in order to develop Rx [= PERFECTIVE ASPECT]. d. They grammaticalise construction Ry to Rx. It may be questioned, then, whether EE could possibly have been an example of ordinary contact-induced grammaticalisation, like already. Heine and Kuteva (2002) do not list ever as a grammaticalisation source, nor do they list the experiential aspect as an individual category. Dryer and Haspelmath (2011) in the World Atlas of Linguistic Structures mention the experiential perfect, but do not list any source grams or maps. Given the conceptual distance between the Chinese lexical source and the English adverb, there is little reason to suspect a crosslinguistically-universal strategy is involved in grammaticalising experiential aspect from ever in SCE (though see also (129–131) below). But the question remains of the selection of ever from the material available in language R. Heine & Kuteva (2003; 2005) also discuss the notion of replica grammaticalisation, which differs from ordinary contact-induced grammaticalisation in that it is not simply a grammatical concept that is transferred from the M-language to the R-language, instead an entire grammaticalisation process found in the M language is replicated in the R language, as explained below (Heine & Kuteva 2003: 539): (118) Replica grammaticalisation: a. Speakers of language R notice that in language M there is a grammatical category Mx. b. They develop an equivalent category Rx, using material available in their own language (R).

130 

 Chapter 5 Experiential aspect

c. To this end, they replicate a grammaticalisation process they assume to have taken place in language M, using an analogical formula of the kind [My> Mx] = [Ry > Rx]. d. They grammaticalise category Ry to Rx. The essential difference between these two processes means that Ry in ordinary contact-induced grammaticalisation may involve the use of any available ­material – it need not have any conceptual relation to the parallel function in My, and, moreover, that the diachronic grammaticalisation of Ry need not be seen to match that of the model function it is emulating. As Gast & Van der Auwera (2012) also note, the latter type involves the M language as its motivation, while the former relies on universally common grammaticalisation patterns. In replica grammaticalisation, however, Ry and My may often share similar lexical source concepts (e.g., the Ry give in SCE which is used to replicate the function of grammaticalising passive-marker geǐ in Chinese, and undergoes the same evolutionary processes, albeit within a relatively limited time frame, as illustrated by Matthews & Yip (2009): lexical ‘give’ > permissive > passive. The stages are seen as represented in the bilingual acquisition routes of Hong Kong children learning English and Cantonese, but similar chains of development are held to be relevant to SCE. However, although these models portray quite lucidly the mechanisms of transfer, it is still necessary for us to investigate whether various attributions of the conceptual sources for ever discussed in the previous section could be explained by the models. It is clear that replica grammaticalisation is not applicable, as the forms used to express experiential aspect in the model languages have been shown to be completely unlike the adverb ever both lexically and functionally. The question thus remains whether EE in SCE is an example of ordinary contactinduced grammaticalisation, and if so, how the Ry feature is selected to perform the functions of the My. This is discussed at length by Gast & Van der Auwera (2012) and also Matras & Sakel (2007) who refer to polysemy, cognitive strategies in the extrapolation of concrete senses from more abstract, grammaticalised replica forms, and the phenomenon of ‘pivot-matching’, in which the process of replication involves the identification of a structure in the replica language which matches that of the model language, and from which grammaticalisation may proceed (2007: 830). Important also to the present argument is the factor of ‘respect’ in the grammaticalising replicated form for the constraints of the replica language itself (2007: 830). Such matters will be discussed further below. It must be recalled in cases of ordinary contact-induced grammaticalisation that there is no necessary conceptual relation between the form in the contact language and the form serving an equivalent function in the model language. The selection of material used to grammaticalise the function is left open in ordinary



5.4 Contact grammaticalisation as a possible explanation 

 131

contact-induced grammaticalisation, where only the functional requirements of the model language are replicated, that is, the category Mx. Previous accounts (such as Ho and Wong 2001) in which it is assumed that function X in SCE ‘comes from’ a substrate/adstratum language form Y need further support, as they do not explain in any principled or theoretical manner exactly what is derived from the model language, nor how it is. Quite apart from this, it is also possible that the process we are observing in the case of ever may not refer to contact-induced grammaticalisation at all, nor any type of grammaticalisation. Thus, in ordinary contact-induced grammaticalisation, as noted earlier, what is not always accounted for is the speaker’s motivation for the selection of Ry. Since the grammatical function of experiential aspect is only implicit in standard varieties of English often in the form of the present perfect, it is most likely considered to be under-specified by speakers of SCE, for whom it is an obligatory category in the M-languages. One means of explaining selection, though, is provided in the phenomenon of convergence (Matras 1998) as the means by which speakers of a group-internal language attempt to adapt material from that language to correspond with the functional requirements of an external counterpart structure, in an effort to match the mental processing operations of both languages (Matras and Sakel 2007: 834–5). This often requires the interlingual identification of polysemies, according to Matras & Sakel (2007), and the requirement in the case of ever to have some form of semantic identification with the model language verb forms discussed above. Grammaticalisation may be seen as a process taking place subsequent to the process of convergence, as the creation of more abstract functions from the selected material (Matras and Sakel 2007: 858), which contributes to pivot-matching. Such operations will be considered below, and in Chapter 8. On the other hand, replica grammaticalisation may be considered in cases in which the lexifier or source language is providing the model (see Ziegeler 2014a), in the manner of grammaticalising experiential aspect as a diachronically and typologically universal strategy that is followed through whenever the same source feature recurs across languages, or whenever experiential aspect requires encoding. This would mean that the positive polarity effects shown by ever in SCE may be reflecting the precedence of some stage in the history of ever in standard English varieties, due to the fact that contact languages grammaticalise the same stages of their non-contact counterpart varieties at a more accelerated pace of development, as noted above. It does not necessarily entail that every stage of grammaticalisation will be replicated chronologically; in retentionist theoretical approaches (e.g., Pietsch 2009), present-day functions of grammatical items are traced to the time of contact historically, but subsequent developments are not necessarily followed through (e.g., the Irish medial perfect). In order to determine

132 

 Chapter 5 Experiential aspect

if either situation holds in the case of SCE ever it is necesary to investigate the diachronic development of ever for possible semantic changes over the historical context which may be relevant to the situation holding in present-day SCE.

5.5 Historical functions of English ever The historical contexts in which ever appeared in the English language can be found in the Oxford English Dictionary Online (under the entry for ever), and it is seen that the adverb had an early appearance as a universal quantifier with the meaning of ‘always, continually’, ‘at all times.’ At no time in the historical records provided in the Oxford English Dictionary is it defined as meaning ‘once’, though it is noted as assuming a meaning of ‘at any time’. This meaning is listed as going back as far as the year 1000CE, but even as early as this it appears in the scope of a negative, and it could be argued that its universal quantifying uses co-occurred with its negative-polarity functions right from the start. One such example (from early Middle English) is the following: (119) 1382 WYCLIFJohn i. 18 No man euere syʒ God, no but the oon bigetun sone. ‘No man ever sees God, none but the one begotten son.’ The meaning of the universal temporal quantifier is therefore presumed to be intimately linked with that of its existential counterpart meaning (‘at least one time’) from quite an early historical period. Leuschner (1996) notes that the Old English source of ever (æfre) meant both ‘always’ and ‘ever’ (its universal and existential quantifier meanings both present in the one adverb). The entries listed for ever with the existential meaning ‘at any/one time’, according to the Oxford English Dictionary, are restricted (primarily) to occurrence in negative and interrogative sentences, and in hypothetical and subordinate clauses; in other words, the examples given all appear in negatively-polarised contexts. It is likely that the existential meanings would be found with no greater frequency than the universal meanings are for the same time period, as they do not represent the majority of senses supplied for the affirmative form in the Oxford English Dictionary. Early uses of the form meaning ‘always’ are equally apparent: (120) c1175 Lamb. Hom. 57 þet is and wes and efre scal beon iblecced ofer al. ‘that is and was and ever shall be blessed over all’ Similar uses remain today in formal, ecclesiastical contexts. However, as noted by Israel (1998), such uses are now considered archaic,50 since the adverb has

50 To quote Israel (1998: 4): “The formerly widespread Universal ever has by now been reduced to a few islands of idiomaticity, a small archipelago of constructions in a sea of obsolescence.”



5.5 Historical functions of English ever 

 133

become restricted to distribution in negative contexts; i.e., it has become chiefly a negative polarity item (NPI) and as a consequence, expresses mainly the existential meaning.51 How this occurred is not precisely explained by either Israel (1998) or Leuschner (1996), though the use of negation with a universal quantifier will always give rise to an existential meaning (via logical equivalences between external and internal negation, in which the meaning always-not gives way to one of not at any time (never) – cf., e.g., Jespersen 1917; Horn 1989: 216, see below). The resulting loss of the universal meaning was accompanied by a suppletion of the universal component of meaning in all-forms in Dutch and German, as well as English, where the earliest all-form ((e)-alne weg) occurred in Old English (Leuschner 1996), thus around the same time as the existential uses of ever were appearing in negative polarity contexts. Examples of such uses include (119) above, as well as others from the OXFORD ENGLISH ­DICTIONARY Online: (121) 1662 STILLINGFL., Orig. Sacr. III. ii. §17 We deny that ever his Atoms with all their occasions would ever produce those things which are in the Universe. In the Helsinki Corpus, universal affirmative uses start to become rare in Early Modern English. (122) is one of only 8 out of 55 (14.5%) of such tokens found for the period 1500–1570:52 (122)  ther Screvener ever wrytyng ower namys man by man As we entyred in the presens of the seyd lordis, ‘their Scribe always writing our names man by man as we entered into the presence of the said lords’ 1517. Richard Torkington, Ye Oldest Diarie of Englysshe Travell. Ed. Loftie. p. 22 Negative uses (never) for this period also include a meaning of ‘no longer’, or ‘not still’ which may be considered obsolete in today’s English: This Jaff was Sumtyme a grett Citee, as it appereth by the Ruyne of the (123)  same, but nowe ther standeth never an howse but oonly ij towers,

51 There may be the exception of environments such as comparative clauses, which Zepter (2003) claims restrict it to the universal meaning. However, examples such as her Today is hotter than it ever was before, while acceptable under a universal intepretation (all the days before), are equally acceptable when read as an existential (any day before). The coincidence of both universal and existential meanings in such environments indicates a possible context of shift, though it need not. 52 These uses represent 36 out of 157 (22.9%) of such tokens in the period from 1570–1640. However, more than one third appear as end salutations of personal letters; the proportion may be biased by such contexts.

134 

 Chapter 5 Experiential aspect

‘This Jaff was at one time a great city, as it appears by the ruin of the same, but now there stands no longer a house but only nine towers’ 1517. Richard Torkington, Ye Oldest Diarie of Englysshe Travell. Ed. Loftie. p. 24. The meaning of never in (123) refers to the discontinuation of a preceding state, not the meaning that we normally associate with never today, i.e., ‘always-not’. The examples provided by the Oxford English Dictionary Online and the EME (I–III) portion of the Helsinki Corpus reveal that the tendency for existential, negatively-polarised contexts is overwhelming. However, one example of a EE was actually found in the EME section of the Helsinki Corpus: My Lords, I take it, he that has been examined, has ever been asked at the (124)  time of his Examination, if it be according to his meaning … 1570–1640. The Trial of Sir Walter Raleigh, ed. Hargrave, PI, 210.C1. A similar example is the following: (125)  Suche a sorte of herytykes ho ever sawe, that wyll nother reverence the croose of Chryste 1500–1570. The Autobiography of Thomas Mowntayne, ‘Such a kind of heretic he had seen, that will neither reverence the cross of Christ’ ed. Nichols, p. 209. The meaning in (124), out of context, could refer to a single event in the past, possibly equivalent to the function in SCE of marking affirmative experiential aspect; however, it is possible that it is being used as a split free relative pronoun (= ‘who …. ever’), according to the context,53 while (125) could carry the universal meaning, ‘always’ as well. The existence of such examples nevertheless reveals that EE may have possibly appeared in the history of English, though was certainly rare at the time, and its frequency would be difficult to estimate on the basis of so few examples. It is certainly not the same function, though, as that of the experiential aspect in SCE. Even more unusual, though, is the following example: (126) ...that yf ever he died before her, he wold never give her anythinge. (1552–1602), The Autobiography and Personal Diary of Dr. Simon Forman, ed. Halliwell, p. 10.

53 I am grateful to Marc Fryd for this suggestion.



5.5 Historical functions of English ever 

 135

In this example, the use of ever appears ungrammatical in today’s English as it violates the constraints of repeatability normally associated with the meaning, both for EE as well as NPE. However, as noted in [2], ever is an adverb with a function of emphasising the possibility or the necessity of the event it applies to, so given such examples, there is no necessary reason to suggest that event repeatability is actually part of its semantics; it may be simply a conversational inference derived from the meanings of minimal event reference. It is possible that this usage could reflect a non-standardism of the time, but to interpret the meaning as ‘if at any (future) time he died before her …’ would seem to focus on the multiple possibilities of a unique event in the future, rather than the repeatability of the event itself, and may not be so ungrammatical. This cannot be confirmed without more evidence; however, it is interesting that the same semantic constraint is common to SCE and present-day English uses of ever (see (115)), and that this provides evidence of its links with aspectual situations which are restricted to describing (repeatable) generic or habitual actions.

5.5.1 Universal quantifier uses of ever in SCE From the few data surveyed so far, the appearance of ever, then, as an affirmative experiential aspect marker in Singlish, can be shown to have a questionable precedence in the history of English. From the examples in SCE that we have seen so far, there is little chance that the same early distribution of universal functions of ever could be repeated as a replicated archaism in the SCE experiential marker. There are apparently no examples of ever in the diachronic survey which reveal an experiential function at any time in its history. However, one example of an apparently universal usage of ever appears in a Singapore internet site: (127)  And she? She senang senang taking care of only staff training and staff leave and yet drawing her high pay and she still hv [‘have’] a HR mgr [manager] who is ever helpful to do things for her. http://flowerpod.com.sg/forums/Career-Talk-f28.html&st=40 (23 August 2008) More, similar examples appear in the International Corpus of English-Singapore spoken corpus, but they are not prolific. In (127), the meaning is ‘always’,54 indicating that the universal meanings associated with past uses of ever may also appear

54 Senang senang is an adverbial reduplicative expression, borrowed from Malay, meaning ‘having an easy life’ (T.C. Choy, personal communication).

136 

 Chapter 5 Experiential aspect

in SCE, though examples of frequently-used universal contexts such as ever since are equally common. It is also possible that the universal usages and the existential (NPE) usages occasionally co-occur in such dialects because they are in fact bi-directional reanalyses of each other (see below). From the historical picture, it is conclusive that ever shed many of its functions as a universal temporal quantifier, having been replaced by always at the same time as its existential functions began to increase through frequent use in negatively-polarised contexts (the two functions not being mututally exclusive at any one time). This pattern of relations holding between universal and existential quantifiers is not uncommon. However, the present-day frequency of negative existentials as well as negatively-polarised ones raise the question of which negatively-polarised environment could have provided the source for the derivation of the experiential marker in SCE: the existential (ever) or the negative (never) (or both). It should be borne in mind also, that the experiential reading in SCE presupposes the existence of the event it qualifies, albeit leaving open the possibility of its referentiality in the discourse, whereas the various uses of NPE do not always presuppose the existence of the event they qualify.55 The semantic relations between NPE and EE therefore remain to be accurately defined.

5.6 Logical explanations of meaning changes Ho & Wong’s (2001) brief allusion to the possibility that ever possibly arose as an antonym of the negative form never raises the principle question why this antonym can co-exist alongside NPE in the same dialect without being considered ungrammatical or pleonastic. Haspelmath (1997, and Leuschner (1996), citing Haspelmath 1993)) claim that universal quantifiers, not existential ones, emerge out of negative polarity contexts, through the intermediate stage of free-choice quantification, though it is questioned whether this category concerns the development of ever. Perhaps the emergence of negative polarity was attributable to the reverse situation in which existential quantifiers (nominal or propositional) could arise from universal contexts. The emergence of existential determiners with the meaning of ‘any’ from universal ones meaning ‘every’ has been noted by ­Haspelmath (1997: 156n) for Turkish and Hebrew, but he adds that though possible, such cases are

55 In most cases of NPE, this is likely. There are uses, though, discussed by Horn (2000a) in which ever is suffixed to wh- indefinite pronouns, such as who-, how-, where-, what-, which he describes as ‘indiscriminatives.’

5.6 Logical explanations of meaning changes  



 137

relatively rare.56 If such cases are possible in other l­ anguages, the claims for contact-induced grammaticalisation must be thus weighted against the likelihood of universal strategies of reanalysis as mechanisms of change in contact as well. For such strategies to provide the strongest explanations for language change, they must have a theoretical application outside the data sets to which they are originally applied, and beyond the contact situations in which they are first observed. Given the former meanings of ever to mean ‘always’, the flexibility with which meanings of existential and universal ever interacted historically can therefore be attributed to the logical means by which universal quantifiers may be reinterpreted as existential quantifiers, and vice versa. These tripartite equivalence relations between quantifier values such as in the following sets were first isolated by Jespersen (1917) and are cited later in Horn (1989: 216), A and C rows representing absolute quantifiers, the B rows intermediate quantifiers: (128) A B C

all some/a none/no

everything something nothing

everybody somebody nobody

always sometimes never

everywhere somewhere nowhere

Jespersen had added modal values to the sets, and considered the members to be interdefinable using negation in such a way that for each column listed, ~A = B (e.g., not always = sometimes), ~C = B (e.g., not never = sometimes), A … ~ = C (always not = never) and C … ~ = A (never not = always). The principle behind such interchanges is simply that if the negation is placed before the absolute quantifier (externally), it negates only the quantifier, allowing for intermediate scales, while if it is placed after the absolute quantifier (internally) it negates the propositional scope of the quantifier. These equivalences can be interpreted linguistically as bi-directional reanalyses, since they may result in instantaneous changes in meaning that can affect either the positive, universal pole of quantification, or the negative, existential one.57 The perceived co-existence of the A and B values for always and ever can therefore be related to diachronic changes in the typical

56 French is one language, though, in which quantifiers meaning ‘all’ can be used to mean ‘any’ in the absence of a determiner, e.g., En tout cas ‘at any rate/in any case’, and Tout dossier incomplet sera refusé (‘Any incomplete dossier will be refused’), personal observation, 27/09/10. The quantifier tout, typically meaning ‘all’ co-occurs with a singular noun and singular verb agreement in such examples, suggesting an existential, rather than a universal reading. 57 Note that Horn (2001: 399) considers pairs like sometimes/ever to be suppletive positive/negative polarity opposites; a further analysis might consider them as event-reference adverbs contrasting in the identifiability of the event they qualify. We shall treat ever, there fore, as a B-value on the table in (128).

138 

 Chapter 5 Experiential aspect

Aristotelian square of oppositions, as discussed in Horn (1989, Chapter 4), with the meaning of ever in Old English as ‘always’. The following shifts of meaning then appear to have taken place: (i) ever became negated externally, to create never (næfre, according to Horn, in press); (ii) næfre came to mean ‘always not’ (negated internally) since its external negation meaning (‘not always’) could be lexicalised via a sub-contrary, sometimes; (iii) the meaning ‘not always’ was thus never lexicalised; (iv) æfre (‘ever’) was eventually supplanted by ‘always’ in many of its universal functions. The suppletion by always (which dates to 1250, according to the Oxford English Dictionary Online) was the natural consequence of the increasing attraction of ever to negative-polarity contexts. Thus in the case of NP ever, which, currently as an existential quantifier, is a B-item, a meaning probably emerged diachronically via the external negation of its former universal senses of , ‘always’, ambiguous between a contradictory meaning of always (‘always not’), and its contrary meaning (‘not always’ > ‘sometimes’), by which a further external negation (‘not sometimes’) implicates ‘always-not’, in any case. Thus the negated form næfre is inevitably destined to finish up in the E – corner (of universal negation) of the Aristotelian square, whatever happens. The attraction to negative polarity of ever is merely a restricted distribution, not a change in the meaning of ever. The data from the Helsinki Corpus bear witness to the increasing restriction of such uses to negatively-polarised contexts over time. For this reason it is hypothesised that the meaning of NP ever in standard usage is polysemous and its development in SCE as an experiential marker is affected by modal nuances of meaning. It could be argued that speakers of SCE have reinterpreted the meaning of never as the emphatic minimiser ‘not once’ rather than ‘not at any possible time’, and thus derived the experiential meaning of ‘once’ by backformation from never. However, if this were the case, it would not take into account the lower-bounded sense conveyed by experiential aspect that whatever occurred did so at least once, if not more frequently. It is therefore more likely that the backformation was derived from NPE, where ever carries the meanings of minimal possibility of the event, reanalysed as minimal (negative) necessity in negative contexts (~ ◊ → □ ~ → □). Thus, it is the modal sense of necessity which survives in a metonymic backshift where experiential ever = [□ (at least) one time], giving the appearance of a morphological back-shift, though in fact it is a semantic one. The evidence suggests that ever in Singapore English may simply be expanding its range of functions due to contact needs. However, independent evidence for the latter possibility, in which there are no contact motivations, can be found in other languages, such as Modern Dutch (Hoeksema 1998) in which the adverb

5.6 Logical explanations of meaning changes  



 139

ooit ‘ever’ is polysemous between having a meaning ‘at (possibly) one time’ in negative polarity contexts, and one of ‘at (necessarily) one time’ in others; e.g., (129) Niemand heeft nobody has ‘Nobody ever knew it.’

het it

stond ooit (130) Hier here stood once ‘A mill stood here, once.’

ooit ever

een a

geweten known 58

molen mill

Again, the difference between ooit in (129) and (130) is simply that in (130) the modal meanings of possibility of the minimal existence of P associated with (129); ‘nobody at at least one possible time ever knew it’ have given way to meanings of experientiality, and of the necessity of the minimal existence of P in (130): ‘a mill stood here, (necessarily) at least one time’. Any time is thus the ­negatively-polarised counterpart of once. Curiously, another example is supplied by Cornillie (2009: 55) from Modern Dutch, in which it occurs within the scope of an evidential statement: (131) Naar to

verluidt appears

moet must

minister minister

Stevaert Stevaert

ooit ever

tijdens while

een losse babbel de volgende gevleugelde uitspraak hebben gedaan a loose chat the following concise declaration have done ‘According to reports, minister Stevaert is said to have made the following judgment in an informal chat’. The use of ooit meaning ‘ever’ is noted by Cornillie to contribute to the evidential reading of (131), along with the adjective volgende. Again, it appears that a shift in meaning is taking place in such examples, wherein the function of ever in past time reference of emphasising possible minimal event existence in negative contexts is being reanalysed in positive ones as one of expressing necessary minimal event existence, and further, experiential meaning as a validation for evidential expressions. One is reminded of the association by Chappell (2001) of evidentiality with experiential aspect in certain cases; thus, (131) provides evidence outside of the Sinitic languages for an evidential marker that carries the same meaning as the experiential marker in SCE. There could be a case, then, for establishing a relationship between experiential aspect and evidentiality in SCE; this requires further research and observation.

58 (Author’s gloss (DZ)).

140 

 Chapter 5 Experiential aspect

5.7 Discussion The nature of the reanalysis described above is hypothesised to be a semantic (not a morphological) back-formation, a process previously identified by Queller (2003), but seldom discussed in relation to grammatical change. It is not a morphological back-formation because ever existed as an autonomous word prior to its reanalysis in SCE, with a universal meaning, and not necessarily restricted to negative contexts, as the historical examples above clearly demonstrate. As a morphological back-formation, it would require reanalysis as a new lexeme created out of an earlier morpheme that formerly had no perceptible lexemic autonomy (e.g., televise < television). However, we are looking only at a back-formed meaning for a preexisting form, not the emergence of a new morpheme. While it may be argued that EE may have “emerged” out of never in SCE (which would make it a morphological backshift), the co-existence of NPE in the same dialect, the historical evidence, and the semantic equivalences just discussed makes it just as likely that NPE could have served as the source for EE in SCE, as it clearly is in the Dutch examples above. However, both NPE and the negation of ever are linked in that they both share the same meaning of possible minimality of existence, albeit negative possibility in never. The link between the two ever constructions is, pace Israel’s (1998) opinions, quite possibly one in which ever may be compared to free-choice any in its experiential function. One could therefore suggest that what EE has in common with its NP counterpart is then the evoking of the minimal existence of P (as a past event). Also important is a certain equivalence of occurrence probability.59 The two functions are interrelated insofar as invoking only the minimal existence of a single event, while historically ever once served the function of marking the necessity of the event at all times, not just on one minimal occasion. It is thus such backgrounded associated senses which have been taken up and exploited by the speakers of SCE at the point of contact in order to fulfill a function in their substrate languages which has no parallel in the lexical source language. The meanings conveyed by ever in its negative-polarity function of possible minimal event-existence are identifiable to the speakers of SCE as pivotal points of interlingual identification by which the same minimal event-existence expressed by the experiential aspect is analogised as a necessary, not a possible minimal event occurrence. The only difference is that in negative-polarity

59 I am grateful to Lucia Tovena for her comments regarding such possibilities.

5.7 Discussion 

 141

context, ever invokes the non-existence of a minimal event, implicating the nonexistence of additional, similar events, while EE invokes its existence. Thus it is questionable why speakers should use a grammatical marker to express existence, when the mere mention of a referent in discourse implies its existence, either in some imaginery world or in the real world. The obvious reason for using a marker of minimal existence, either of an entity or of an event, is that there is some presupposition of non-existence to contradict. This is a presupposition shared by both NPE and EE and it is in this respect that it is also from that point on that grammaticalisation of experiential aspect is also possible (also in Dutch, as shown in (129–131). Horn (2000b: 182) refuted at one time the possibility of linking free-choice any with ever, due to ever’s inability to ‘sponsor’ generic reference. Generic reference in both cases is implicit and is encapsulated in the repeatability factor: what is repeatable in terms of spatial entities may equally be repeatable in terms of temporal events, given the constraints discussed above (cf.*if ever he died before her...., in example (126)), and clearly articulated in the studies of ever in SCE (e.g., Bao 2005). In example (126) it is clear that the minimiser meaning of ever has not yet begun to create the inferences of repeatability by the early 17th century: while (126) can be paraphrased as ‘if at any time (at one possible time) he died’, referring to a future event, the semantics of repeatability are likely to have developed later as conversational inferences ruling out the plausibility of such examples: ‘!if at (at least) one time, he died …’ . It appears the lower boundedness of the minimiser in the semantics of ever at the time was measured only by the possibility of the event and not its realisation. The criterion of repeatability also described by Chappell (2001) as one of the semantic ingredients of experiential markers in Chinese also provides the essential semantic substructure required for the creation of the pivot construction, a merger construction from the point of view of the present theory. The pivot thus revealed is not a concrete sense that can be extrapolated from the meaning, but a more abstract, implicit pragmatic inference associated with existential quantifiers in general, and that is the identification of ever’s lower-boundedness on a scalar dimension, as noted by Israel (1998). The factor of repeatability is more likely to be a conversational implicature derived from the minimiser meaning associated with the negative polarity uses of ever in SSE. Moreover, we find that the constraints associated with the use of the item in the lexical source language (or SSE) are not there in the use of experiential ever, which is grammaticalised in SCE way beyond the restricted negatively-polarised contexts of standard English.

142 

 Chapter 5 Experiential aspect

5.8 Summary It was questioned in the introduction to the present study how an adverb such as ever, largely restricted to negative polarity contexts in most dialects of English, could become the source construction for the grammaticalisation of experiential aspect in a contact language such as SCE, despite the constraints of its grammaticality in Standard Singapore English and other standard varieties. The likelihood of replica grammaticalisation was ruled out by the lack of evidence of historical data on the use of ever as an experiential marker, though the continued but rare use of the adverb as a universal quantifier in SCE appears to justify the classical reanalysis of universal quantifiers as existential ones in SCE as well as standard English. Ordinary contact grammaticalisation, though a likely explanation for the development of more abstract, aspectual functions, was deemed insufficient to explain the selection of the item as a pivot for interlingual identification. The role of the substrate, or model languages, in the present case, mainly Sinitic languages, is seen to be more salient insofar as the semantics of experientiality as described by Chappell (2001) are concerned. Experiential aspect in such languages is comprised of a number of semantic and functional ingredients, only one of which has been exploited by speakers of Singapore English as providing the pivot match necessary to seek a similar construction type in the source language: the marking of minimal existence. This is also the semantic feature shared by free choice any, suggesting that EE may qualify for free-choice status if only on the basis of its capacity to implicate the repeatability of the events it qualifies. If so, could be hypothesised that ever in SCE may eventually expand its functional range as a universal quantifier of events, according to Haspelmath’s (1993) observations that universal quantifiers emerge out of NP items through the intermediate stage of free choice quantification. However, such hypotheses are speculative for the moment, without sufficient evidence to support them. In terms of a possible grammaticalisation path, it can be seen also from Dutch that an adverb of emphatic possibility in negative contexts, already reanalysed from an adverb of emphatic necessity, can be grammaticalised in some languages as an experiential aspect marker, where negative emphatic possibility is seen as semantically contiguous with the expression of minimal (positive) necessity (i.e., upper-bounded ‘not at at least one time’ > lower-bounded ‘at least one time’). This is an interesting development that to present knowledge has not been the subject of previous research efforts, and provides ample scope for further investigation in the future.

Chapter 6  The past tense construction 6.1 Introduction Another curious feature of Singapore Colloquial English has been the v ­ ariation in the marking of habitual aspect. The present chapter first appeared in Language Sciences (as Ziegeler 2012a) titled, ‘On the interaction of past tense and potentiality in Singaporean Colloquial English’, and was earlier presented as part of a workshop on irrealis organised by Caterina Mauri and Andrea Sansò at the 2008 annual conference of the Linguistic Society of Europe, Forli, Italy. The original ideas were pursued in a research project funded by the Singapore Faculty Staff Research Support Scheme at the National University of Singapore, 2008–09, and with the assistance of Amelyn Thompson, who also compiled the corpora used as the principle data source, the Flowerpod Corpus. The problems and questions raised at the time were studied from the point of view of the aspect-modality interface, an area which is critical to the understanding of the expression of habituals and generics in general. The ideas have not been significantly revised since then, as there have been few observable changes in the expression of present habitual aspect using past tense forms from which to work on; the study has not, however, been given a construction account, and it is with this aim that is it presented in summarised form below. Although the present chapter deals with the expression of habitual aspect in SCE, it is the variation in the use of tense to express such aspectual distinctions which is the main object of investigation. As noted earlier in Chapter 2, past tense is used less often in SCE than in Standard Singapore English (SSE), due to the influence of marking aspectual priority, and that it was more likely to represent perfective aspectual tendencies than tense marking (Ho & Platt 1993). Less ­frequently observed though, are cases in which a past tense form has been seen to appear in a present tense function; for example, in the context of referring to habitual activities of present time reference. Such past-for-present (PFP) constructions have also been observed in Ho & Platt (1993), but were left largely unexplained. The same use of past tense for what is usually encoded in the simple present is found occasionally in the antecedent clauses of ‘open’ conditionals (Quirk et al.’s 1985 term), otherwise known as indicative or predictive conditionals (­Kaufmann 2005), or in other subordinate-clause counterparts of such conditionals, such as those defined by Brisard (1997: 278) including temporal when-clauses and concessive subordinate clauses. Such contexts are marginally irrealis in that they

144 

 Chapter 6 The past tense construction

express non-assertion, but in many languages there is ambiguity as to whether they are marked as realis or irrealis (see, e.g., Givón 1994).60 The present chapter examines such uses, first discussing comparative and crosslinguistic data in which one of the categories, present habitual aspect, is marked by past tense, and later ­considering the more general situation of marking present time reference with perfectivity. What is particularly intriguing about the SCE situation, though, is the fact that at the same time as present habituals are sometimes found marked for past, past habituals are often left unmarked for past tense, reflecting the precedence of grammatical aspect over tense, though the reasons remain unknown. The present study will endeavour to explain such incongruities, using a construction-based approach. The chapter will first discuss the interaction of habitual aspect and tense in Singaporean Colloquial English, (Section 6.2). A review of some crosslinguistic evidence for the marking of perfectivity in present habituals will be provided in section 6.3, and section 6.4 will present evidence of the types of environments in which past tense marking may appear in SCE in data derived from local internet blog-spots. 6.5 will discuss the marking of present habitual aspect and real conditionals with past tense in SCE, and it will be questioned why past tense is sometimes used to mark present habituals in SCE, while present tense may mark past habituals. The chapter also takes into consideration the fact that habitual and generic expressions lie at the margins of aspect and modality. A few background assumptions are then in order. The category of perfective aspect is taken in the present volume to refer crosslinguistically to bounded events that must be viewed in their entirety (see, e.g., Comrie 1976; Smith 1991; Bybee, Perkins and Pagliuca 1994). The categorisations of perfective vs. imperfective aspect will cut across the two aspectual types making a clear-cut distinction between what Sasse (2002) called Aspect2 or what is known in the classical literature as ‘Aktionsart’ (i.e., ‘kinds of action’—Comrie 1976: 6–7), and Aspect1 or grammatical aspect, being the binary aspectual distinction between perfective and imperfective found in Slavic, for example, or Chinese. Other terms for the former have included ‘objective’ (Verkuyl 1993), situation (Bybee et al. 1994; Smith 1991), and lexical, semantic, or inherent aspect (Comrie 1976, who rejected the term Aktionsart). Terms for the latter have included viewpoint aspect (also

60 The terms realis and irrealis have been used extensively in the crosslinguistic literature to refer to modal oppositions, particularly in non-Indo-European languages. Givón (1994) used the terms to signify degree of assertion, in that a realis proposition is strongly asserted to be true, while an irrealis one is weakly asserted as either possible, likely or uncertain (1994: 268).



6.2 Tense marking in habitual aspect in SCE 

 145

used by Bybee et al., after Smith 1991). In the present volume, the terms lexical and grammatical aspect will be used in preference to any alternative nomenclature that may have appeared in the literature from time to time. It will also be understood that in the majority of cases, grammatical perfectivity is attracted to lexically perfective environments, by means of a Principle of Relevance (Bybee 1985). However, in the case of the habitual aspect (as well as the progressive), grammatical imperfectivity is attracted to lexically perfective environments, thus increasing the potential for modal interpretations, as discussed in Ziegeler (2006, Chapter 1), a result of aspectual conflict. At the same time, the grammaticalisation of past tense has been hypothesised to be an end-result of the grammaticalisation of perfective grammatical aspect, in cases in which perfectives are extended to the domain of stative verbs (see Comrie 1976; Bybee et al. (1994: 92)). The problems at hand are thus extremely complex: the appearance of past tense in SCE in imperfective environments such as habituals and generics, and also in the protases of ‘open’ conditionals: is this a question of marking for perfective aspect, or marking for tense? The need to use past tense to mark present tense habitual aspect may have more to do with the expression of factuality than of time reference, since if past tense is a means of marking perfective aspect in SCE, it also carries with it the assumption of an event completed. But it is necessary first to consider some examples.

6.2 Tense marking in habitual aspect in SCE In Ho & Platt (1993), a punctual aspectual form refers to one which is not only lexically punctual (i.e., referring to actions occurring at a particular point in time), but also covers a category in which the entire situation in which it occurs is punctual, thus including in their description features of telicity as well as ­completion (1993: 82–3). Such a description is slightly muddling, though, as the category of ­non-punctuals refers to the use of inherently punctual aspectual types ­co-occurring with imperfective grammatical aspect; e.g., that of habituals, which typically involve the use of lexically punctual verb types, as in the following, ­extracted from Ho & Platt (1993: 154): (132) TS.18. He makes it a point to to-to come down to de[their] level (speaker talking about his former principal) Other examples come from personal observation: (133) He [Edgar Hoover] goes after his political opponents, he bugs them and keeps files on them … assassinated a few … .

146 

 Chapter 6 The past tense construction

In (132) the lexical aspect (Aktionsart) is punctual or perfective (make it a point) but it is not marked for past as it is occurring in a grammatically imperfective environment, the (past) habitual. (133) illustrates the contrast between non-past marking on the grammatically imperfective verb forms, and past marking on the only form which can be understood as perfective (since it is bounded by a quantified NP object). Ho & Platt (1993) believe that the absence of past marking in such environments is due to the non-occurrence of the perfective marker le in past habituals in Mandarin Chinese, e.g., (1993: 150): (134) tā zuótiān wèi le 3SG yesterday feed-PFV ‘Yesterday s/he fed the dog.’

goǔ dog

(135) *qùniān tā měitiān wèi le last year 3SG every day feed-PFV ‘Last year s/he fed the dog every day.’

goǔ dog

(135) illustrates the use of a lexically punctual verb type in a grammatically nonpunctual environment, in which it is not marked for grammatical perfectivity. The perfective marker le in Chinese cannot mark lexically completive functions in a grammatically imperfective (non-completive) environment, as it has not progressed to the anticipated stage at which it may parallel a tense marker. The other important point that was raised in Ho and Platt’s study was that there are a number of unanswered questions with regard to past tense marking in SCE, which cannot be explained by the hypotheses presented in either Ho & Platt (1993) or Bao (1995; 2005). Amongst such questions are those that have to do not with the absence of past marking where it would be found in SSE, but with the objects of investigation of the present study, i.e., the occasional use of past tense in environments in which a present tense form is used in SSE, e.g., in present habituals, and in the protases of indicative conditionals. This feature does not appear to be categorical, but a number of previously observed examples are given below, for example: (136) For six days a week, I have to work as a person doing sales, I have s­ omewhat a “flexi” time to complete my official duties. Usually I finished work at around six o’clock. Currently I am attending night classes about thrice a week. (136) was a written example. More examples of diary observations include the following, sent on a mobile phone message (May 2008): (137) Have forwarded your message to Lilyn who will be contacting you as I no longer worked in the clinic. Have a blessed day.



6.2 Tense marking in habitual aspect in SCE 

 147

The speaker was female, aged around 30, and educated to at least ­school-leaving. Another example illustrates the parallel usage appearing in the antecedent clauses of conditionals or conditional equivalents: (138) ... Until the market picked up, speculators are likely to stay away from the market ... uttered by a female property agent on MediaCorp (television) News service, 12/6/08. Ho & Platt (1993: 148) found similar occurrences, for example, in the dialect of a secondary-school educated speaker: (139) He can be a very good learner – you can force a person to learn but reach to certain age, what happen, he landed Woodbridge! [ = ‘he ends up at Woodbridge’] The use of the past tense in the last sentence substitutes for the present – ‘but what happens, he lands in Woodbridge’ (a mental hospital). Ho & Platt (1993: 148) consider such verbs used in past form to be part of a ‘learned’ vocabulary of items which are often found in participle form such as allowed (as in ‘not allowed’) admitted, and excited, which has an adjectival function in the present in any case, and thus is generalised to any context. However, such an explanation does not account for all the instances of past tense found in previous observations. Ho & Platt (1993: 147) discuss past-for-present habitual marking as problematic, and unresolved, stating that the informants are using realis marking to describe the events of a general situation which occurred in the past; e.g., (1993: 147): (140) Every Sunday dey sure want to go one. (One is an emphatic marker. The children insist on attending Sunday school every Sunday.) Heavy rain also, dey wanted to go. (Even if it rains, they’d still want to go.) Another example from a secondary-level educated subject occurs in the following (1993: 147): (141) (Talking about Bugis Street, a famous night spot for tourists.) Aah, ­European a lots ‘know. Every touris(t) dey will send by cart, buses, some dey go by demself; a lot of Navy (seamen) wen(t) dere. I see dem, some of dem wearing caps, Navy, New Zealand. Some are Navy … Even Japanese also went dere. Mmm. The speaker goes on: (142) Oh! I always went dere y’see. (She still does.) At nights I not(h)ing to do – I went dere. … Aah majority (frequently) because ah, always I go from work – aroun(d) t(h)ree – one week two times.

148 

 Chapter 6 The past tense construction

The examples in (142) using irregular past tense forms are an indication that the variation may be due to semantic or functional reasons and not simply to a phonological over-extension, which would be the case if only regular past forms were found. Ho & Platt note that the alternation between went and go in the above examples might be attributed to whether or not the verb is expressing anterior actions (1993: 147). However, they do not link this possibility to any of the other instances, nor do they explore it further. The phenomenon is not, to present knowledge, reported in any other accounts, except for Loh (1998: 41), who provides the following (constructed) examples from a possible exchange in SCE: (143) A. What should we do for Joe’s dinner tonight? KFC? B. Cannot lah (‘No we shouldn’t’)

a. B. He don’t eat fried chicken b. B. He didn’t eat fried chicken.

(144) A. What should we buy for Joe? Jeans? B. Cannot lah. a. B. He don’t look good in jeans, b. B. He didn’t look good in jeans. Loh explains the (alternative) (b) responses as indicating that the state expressed refers to a definite period of time in the past, from t1 to t2. Thus, Joe ate KFC on one occasion in the past and did not like it, and this leaves open the possibility of questioning when the event took place. The (b) responses are then taken to express, according to Loh, more specific instances of the (a) responses. However, they seem to suggest that the past tense form is acting to refer to past time reference in (143–144), when in actuality, it is interchangeable with a present tense form in the (a) examples. This raises doubts as to the form-function correspondences intended, if past tense is to be considered a construction type, as it would be under the perspective of a universal constructional approach to morphology. The question then is to find out what functions may be served by the past-present tense contrast in SCE, and how they differ from standard correspondences. Given that past tense has been claimed by Ho & Platt (1993) to express mainly perfective aspect in SCE, it may be hypothesised, then, that perfective aspect is prioritised over imperfective in SCE habituals also. It must be remembered again that habitual aspect constructions form a hybrid category with perfective lexical aspect and imperfective grammatical aspect. If this is the case, then it is the perfective lexical aspect that is given marking priority over the imperfective



6.3 Tense marking in habituals in other languages 

 149

grammatical aspect, in an otherwise ambiguous aspectual context. In the context of such a hypothesis, the examples from SCE will be compared against the evidence of perfective marking in habituals found in other languages.

6.3 Tense marking in habituals in other languages The interaction of habituals with tense marking apart, there are other idiosyncrasies in relation to the marking of habituals as modal categories. Recent discussions of the marking for habitual aspect crosslinguistically have noted the fact that it may be marked in some languages with the same marking as irrealis modality, and in others as equivalent to a realis modal category, and for that reason has been labelled a ‘swing’ category in relation to modality (e.g., Givón 1994). However, the modal nature of the category is only secondary to its aspectual function. Its somewhat ambivalent modal character is a reflection of its dual aspectual function referring to both individual events which are inherently bounded in time (perfective) and multiple events spanned over an unbounded time period (imperfective). As noted earlier, perfectivity has close associations with realis and imperfective with irrealis, since perfectivity expresses the completion of an event which gives rise to inferences of actuality or factuality, and imperfectivity refers to incomplete events thus implicating non-actualisation and thus irrealis notions as well. Because of such modal ambiguity, though, habitual aspect receives quite variable marking crosslinguistically (as also noted in Cristofaro 2004), and is frequently found to be unmarked, or zero-marked, in languages in which durative imperfectives (progressives and continuous aspects) are marked overtly in opposition (Bybee, Perkins & Pagliuca 1994: 151). The marking of habitual aspect therefore appears to be often dependent on the presence or absence of a grammaticalised marking for a complementary imperfective aspectual category in a given language, and varies according to the ways in which that language may carve up its tense-aspect system. However, it is clear from crosslinguistic studies that specific marking for the present habitual aspect is a rarity, and in most languages marking for present habitual aspect is shared with the progressive aspect or a more general present tense (Bybee, Perkins & Pagliuca 1994: 153). Much crosslinguistic evidence for the grammaticalisation of habitual aspect also reveals a tendency for such functions to be encoded by modal verbs, rather than tense markers. Such modals are those which eventually may grammaticalise to function as modals of possibility; sources include those listed by Bybee et al. (1994: 154) related to meanings such as ‘know’, but verbs meaning ‘live’, and in one or two cases, e.g., Yagara, verbs meaning ‘see’ may also become the sources for habituals. In the latter case, such verbs are metaphorically related to ‘know’,

150 

 Chapter 6 The past tense construction

not only in Yagara, but also in the Latin cognate sources for Germanic verbs of cognizance (Hewson & Bubenik 1997: 225). Such verbs develop meanings expressing mental ability (‘know-how’), then physical ability, and eventually grammaticalising into modals of possibility (Bybee 2003). Many of the cited sources for grammaticalised habitual markers occur in creoles; e.g., Singler (1990) refers to Kru (a West African, English-based creole), the forms ken and we (derived from English can and will) which have both irrealis and habitual functions. From such evidence, it may appear, then, that the use of modals to mark habituals focuses on the irrealis notions of the hybrid aspectual form, not the realis facet of the category, in the conceptualisation of habitual action as potential action in the subject. However, in Cristofaro (2004) it is shown that it is the past tense habitual in many languages which attracts irrealis marking, while the present tense habituals are often marked by forms used to encode actualised events, e.g., in languages such as Dahalo, Etsako, Gurr-goni, Hausa, Malayalam, and Tamil (the last two are minority substrate languages in Singapore, and Tamil is also an official language). In only 11 of the 200 languages surveyed, she finds that the present habitual shares a grammatical marker with a form marking unactualised events (e.g., Azari, Karimojong, and Tibetan) – this is a general habitual marker whatever the tense. Thus, while there is a tendency (according to Bybee et al. (1994) and Singler (1990)) across many languages generally to mark habitual aspect with a stative lexical verb more closely associated with imperfective aspect than perfective aspect, Christofaro (2004) shows that when present time reference is used, a form expressing actuality is more frequently used instead. Thus according to such accounts, in SCE the situation of the past habitual being unmarked for past, and the present habitual being marked for past, would not be atypical if past tense, as a marker of perfectivity, were considered as a means of expressing actualised events, and present tense forms were more closely associated with unactualised events. It is in such environments, then, that the functional convergence between perfective aspect, expressed as past tense in SCE, and realis modality is most easily visible.

6.3.1 Slavic Examples of languages or dialects in which there may be a link between habitual aspect and actualisation of the events include Slavic languages discussed by Dahl (1996) and Comrie (1976). Dahl (1996: 419–20) includes such languages as Czech, Slovak, Serbian, Serbo-Croatian, and Slovene, and Russian as cases in which the perfective may appear in the imperfective territory usually associated



6.3 Tense marking in habituals in other languages 

 151

with the habitual. This occurs when the sentence refers to a bounded event, as quoted from Mønnesland (1984: 62) for Serbo-Croatian: (145) Svako jutro popijem čašu rakjie ‘Every morning I drink a glass of brandy’

(Dahl 1996: 420)

The perfective aspect popijem is used in such cases rather than the imperfective pijem because of the inherent lexical aspect of the situation which is telic, and it is at this level, the level of the event, that the aspectual choice is determined, despite the present time reference. Thus, any imperfectivity that would normally be associated with a habitual event is determined at the sentence level by frequency adverbs or other quantificational means of marking the timelessness of an event; however, in languages which appear to mark perfectivity in habitual sentences, it is the lexical level, rather than the grammatical level, which ­supplies the justification for the choice of marking. In the Slavic languages, therefore, the lexical aspectual semantics appears to be more salient and to take priority over the grammatical level of aspect marking. Dahl (1996: 420) considers that aspectual distinctions are usually neutralised in habitual/generic sentences, but such examples, and the data discussed above, suggest otherwise. Comrie (1976: 31–2) discusses a similar phenomenon to that of Serbo-Croatian in Bulgarian, which has two combined aspectual types: Perfective/Imperfective and Aorist/Imperfective (the latter occurring in past tense only). The Perfective/ Imperfective can be used in an iterative function (in 146a) of which the individual occurrences would each be described using the Perfective Aorist (in 146b); e.g.: (146) a. štom pukneše, (Pfv. Imperfect) zorata, izkarvax ovcite navәn ‘as soon as dawn broke, I used to drive the sheep out’ b. štom pukna zorata … ‘as soon as dawn broke’

(Comrie 1976: 31–2)61

The Perfective Imperfective, then, in this example reflecting past time r­ eference, is described by Comrie as taking “ … a situation which would in itself be d ­ escribed by a perfective form (Perfective), and then superimposes upon this imperfectivity, or rather one of the possible subtypes of imperfectivity, namely habituality, ie. this is the imperfective (or more specifically, the habitual) of a perfective, or the

61 Comrie (1976) and Dahl (1996) did not supply an interlinear gloss for these examples in the original citation.

152 

 Chapter 6 The past tense construction

Imperfective of a Perfective.” The combined form, then, is an aspectual distinction which subsumes both the lexical aspect of a verb or event, together with the grammatical aspect of the entire situation in one portmanteau formal category. As such, it may well reflect the ambivalence given to the marking of the habitual aspect in other languages. Thus, the examples shown above from Serbo-Croatian and Bulgarian also illustrate the means by which perfectivity may be found to reside in what are normally imperfective environments; these may also be examples of what appears to be a ranking of aspect over tense in such languages. 6.3.2 Chinese dialects Another possible case where markers for perfectivity or anteriority share functions with the present habitual is that of the verb u in Taiwanese Southern Min, a local L1 dialect spoken alongside Mandarin in Taiwan, and one of the Min family of Chinese dialects spoken in the south-eastern provinces of China (known collectively as Hokkien). The form is genetically related to the Mandarin verb, yǒu, a possessive verb which had the original meaning of ‘to hold meat in the hand’ or ‘to hold the moon in the hand’ (M-S Bao 1971). Hence the original meaning ‘to hold X in the hand’ becomes later generalised to ‘to hold’ as the lexical source meaning begins to weaken. In the negative and in polar (yes/no) interrogatives, yǒu remains in standard Mandarin. However, in the Hokkien dialects including Taiwanese, the counterpart of yǒu, u, has additional functions and is still used in the affirmative as a perfect and perfective aspect marker, e.g., (147) i chahng u chia. 3SG yesterday U eat ‘He had a meal yesterday.’ ‘He did have a meal yesterday.’

png rice

(Lu 1991: 36)

Lu’s translation of u reflects also another function of u that is not parallelled in the Mandarin form – that of marking emphasis. Tsao (1995) does not discuss the present-day use of u noted by Cheng (1979 and 1997) as marking a habitual function. Cheng (1997: 35) notes that u is obligatory in expressing habitual action, unless there is a time adverbial. In the following examples, it appears with unemphatic usage: u chiah   hun (148) a. Li 2SG U inhale    smoke ‘Do you smoke?’

bo? NEG.

6.3 Tense marking in habituals in other languages 



 153

b. Goa bo chiah   hun. 1SG NEG inhale   smoke ‘I do not smoke.’ c. U, goa U 1SG ‘Yes, I do.’

u U (Cheng 1979: 184)

(It should be noted that the negative form bo in (148b) means both ‘not’ and ‘not have’ and has generalised from being a negator of possession/existence to a general negative marker occurring with any verb.) In standard Mandarin the counterpart form yǒu would be ungrammatical, as the habitual aspect is usually expressed by zero. (148c) would then be translated simply as Wǒ chōu yān (‘1sg inhale smoke’) with no auxiliary. The Mandarin counterpart form, yǒu, though, sometimes surfaces in Taiwanese Mandarin, as evidence of contact or substratum influences. Similar constructions are found in Cantonese, in which a cognate form yauh, serves to optionally mark the present habitual; e.g., (149) Bouji yaht yaht dōu yáuh góng gúsíh newspaper day-day all have talk stock-market ‘The newspaper talks about the stock market every day’ (Matthews and Yip 1994: 281) and in Hakka, using yu (also in Chaozhou dialects (Li and Thompson 1981 [1989]: 431)). Because of this, we find that the perfect construction using u is often ambiguous with the expression of habitual action (Chinfa Lien, p.c.) in Taiwanese, which might suggest that the two functions could be at adjacent stages of grammaticalisation. The allusions made by Tsao and Cheng, though, are to meanings of existence (of an action) rather than of possession (which might be more closely related to the source meanings of the verb form) and which indicate that there are inherent meanings of factivity surrounding the use of both yǒu and u to express habituality in the Chinese dialects. Such meanings are possibly derived from the extension of the ‘existence’ meanings of the same verb from referring to the existence of an entity to the existence of an action or event (in other words, factuality). It is less likely that the appearance of verbs expressing both anterior aspectual distinctions as well as present habitual in the substrate Southern Min and Cantonese dialects may be linked to the use of past tense in SCE by contact: have-perfects are not used to express habitual aspect in SCE (and not often for anteriors either) and yet the have-possessive source for perfects is available in

154 

 Chapter 6 The past tense construction

SSE. Furthermore, it will be seen in the data below that there are other environments besides the present habituals in which past tense is used for present time reference in SCE, and these are not found in the substrate dialects at all. But in the first instance, it was considered necssary to investigate the range and extent of the use of past tense constructions in present time reference contexts in SCE, and for such a task, a survey of typical uses was constructed.

6.4 Preliminary survey data In order to examine the frequency and extent of the use of past-for-present in SCE, a search was conducted during a period of eight weeks (from May–June 2008) for tokens appearing on teenagers’ internet blog-spots in Singapore. Many of these forums discuss everyday habits and activities of their subscribers, with little reference to narrative style or the expression of past time, unless they are clearly describing past experiences. Much of the content is thus fairly superficial and based on giving advice, and discussing fashions, cosmetics, shopping, hobbies, superstitions, and daily activities, which provide ideal contexts for the use of present tense habitual forms, or open conditionals. The initial problems were in devising a search mechanism, and also in confirming the reliability of the mechanism as well as of the informants’ backgrounds. Internet data, it must be admitted, are deficient in such areas, and should not be taken as necessarily representative, by any means. However, the content of the postings is often quite revealing as to the nature of the sociolinguistic factors that would otherwise need to be elicited from interviews and surveys (see Ooi, Tan & Chiang 2008), and is sufficient to provide a reasonable perspective on the frequency of use of certain forms under investigation. The style and register of such forums is inevitably informal, often intimate and often reduced to the code of telephone text-messaging: according to Claridge (2007), internet message boards provide a medium of communication which is intermediate between spoken and written forms, and should be considered a ‘new’ medium. In all cases, the advantages of such data are that the anonymity of the participants as well as any ethical protection are easily guaranteed. Open-class forms are notoriously difficult to search for in any data-base; so in the first instance, a range of strong past forms was selected (based partly on Ho & Platt’s (1993) selection of strong verbs) which was narrowed down according to frequency of occurrence of individual items. Strong verbs rather than regular pasts were selected at first (SCE regular past tense forms are often marked in speech by the absence of –ed suffixes, particularly where a consonant cluster occurs, and such elision is often carried over into the written form, making it



6.4 Preliminary survey data 

 155

often difficult to determine whether past or present forms were intended.) Stative strong verbs were also avoided, as the marking for past tense on such forms is predictably variable, since statives are not generally marked for past anyway in SCE.62 Once a pool of frequently-repeated instances was found, this was established as a limitation on the data, leaving out less-frequently-occurring items. The narrowing of the lexical pool in this way could be criticised as biasing the data, but in other ways, it could be seen as an advantage, in that it provided a means of determining if there were any common lexical factors influencing the data as well. It must also be remembered that any new items found in the process of searching for the old items were added to the pool and re-searched in new attempts. A score was kept of the total number of past forms found for each item within approximately 10 hours of searching per week. The method of haphazard searching was maintained through the first 8 weeks, and forms the preliminary survey for the study. The items are listed as follows.

6.4.1 Search items The original lexical pool of items comprised: came, contained, did(n’t), felt, found, forgot, kept, landed, spent, studied, thought, took, wanted, went, these being items taken from previous observations as well as examples used in previous studies. Additions, after 5 weeks, included: bought, brought, chose, gave, had, left, lost, met, paid, saw, slept, sold, told, wore, wrote. Control examples, added after the 7th week, included the following: blew, ran, sat, spoke, talked, and walked. The control items (all Activity types and hence most often used imperfectively) were selected on the basis of a preliminary hypothesis linking the use of past tense with a lexically perfective verb type. Others, taken from personal observation and Ho & Platt (1993) included: came, ended up, finished, graduated, picked up, rose, worked (all having occurred once each, except for ended up (2 altogether)). The frequencies at the final count (after 8 weeks of searching) were the following: took (32), found (12), kept (6) ate (5), saw (5), did(n’t) (3), made, spent, thought, (2 each), and bought, graduated, lost, recovered, tried, used, went, wore, all occurring once each. From such input, over the 8 weeks of searching the blog spots as well as from evidence of previous studies (e.g., Ho & Platt 1993) and personal observations, a limited pool of items was accounted for by which the marking of past tense in

62 For this reason, stative auxiliaries and modals were also not included in the totals.

156 

 Chapter 6 The past tense construction

present habitual contexts could be defined. Some examples of their usage will be examined below.

6.4.2 Examples of the use of pasts-for-presents (PFP constructions) Examples of each of the selected lexical pool will be presented below. A rough gloss is also provided below each item, where the language used may become vague due to the overuse of abbreviated forms, or ‘creative re-spellings’ (Ooi, Tan and Chiang (2008)). Some examples of personal observation may also be included in the examples below.

6.4.2.1 Ate The occurrence of ate [medicine] below is a possible loan translation from Mandarin Chinese chī yào where reference to taking medicine is usually described as ‘eating’ medicine. (150) Original Post: Sorethroat, Any remedy to relieve/cure it? A: yest 23th June... sore throat severe again and slight fever... ate Chinese Med now.. cos western med cant help me... ‘Yesterday, 23rd June, [I had a] sore throat again and a slight temperature. [I] take Chinese medicine now, because Western medicine can’t help me’. http://www.flowerpod.com.sg/forums/index.php? showtopic=24344&st=40 (24/06/08) The use of the past tense form in (150) is not explained by any reference to past time, as it is accompanied by the adverb now, indicating that the time reference is in the present. However, the adverb now is often translatable in SCE by already, as noted above; it refers to entering into a new state brought about by an implied anterior change of state: in this case, from taking Western medicine to taking Chinese medicine. It is quite likely, then, that ate in this example is grammatically perfective, as would be an inchoative perfective form, and is competing with eat + already. (Why the speaker has chosen not to use already, though, is not known.) However, this explanation does not always apply, as the following example seems to indicate: (151) i drink 2 litre of plain water and ate alot [sic.] of fruits. http://www.flowerpod.com.sg/forums/index.php?showtopic=1387&st=400 (240608)



6.4 Preliminary survey data 

 157

The use of ate in (151) refers to an ongoing situation, already signified in the ­discourse by co-ordination with the present habitual form drink preceding it. There is no apparent reason for the sudden shift to past tense, except for the possibility that the incompatibility between perfective lexical aspect and imperfective grammatical aspect is variable or unresolved for this speaker. Another speaker uses ate without a corresponding present habitual: (152) Prunes good enough la. I just ate it as a snack and i lao sai :lol:63 ‘Prunes are good enough. I just eat them as a snack and I get diarrhoea.’ http://www.flowerpod.com.sg/forums/index.php?showtopic=1387 (240608) There appears to be no clear explanation for the use of past tense in such examples.

6.4.2.2 Bought The only example which turns up is in the 7th week of searching, and it is used intransitively. (153)  Original Post: Hey guys, just curious, where do you usually buy your food from – supermarket, wet market or take-out? For me, it’s usually the supermarket! I also don cook....bcos my auntie will cook.... she bought from wet market mostly..... somtime also go supermarket.... ‘I don’t cook either … because my auntie cooks … she buys from the wet market mostly … sometimes she also goes to the supermarket’. httpt://www.sgclub.com/singapore/ whats_routine_food_74864.html (02/07/08) The use of the frequency adverb mostly indicates that the habitual aspect was intended, and the topic question sets the time reference as present, or at least non-past. The use of will + infinitive, an alternative means of marking present habituals in SCE, is also illustrated in this example (will cook). A non-past habitual form, go, is also found co-occurring with bought.

63 The form lol is an (international) abbreviation derived from either ‘laugh out loud’ or ‘lots of laughs’, and transferred to the email text from SMS phone messaging jargons. Lao sai (‘diarrhoea’) is another borrowing from Hokkien. Note also that the copula is frequently omitted in SCE and the singular 3rd person pronoun it serves for both singular and plural inanimates.

158 

 Chapter 6 The past tense construction

6.4.2.3 Came This was another relatively infrequent verb, only occurring once in the ­preliminary data taken from Yahoo Classifieds. (154) Toys & Games Tow Cobra sz60 heli Jurong West, sg Description:  Nitro sz 60 scale tow cobra for sale. fast deal sms 97839574 jj. item came complete with 6 servo (JR/Hitec), gyro(futaba), os60 heli engine, and necessary wires and cable. http://sg.classifieds.yahoo.com/?intl=sg – Classifieds 06/03/06. It is hardly likely that the seller of this item will be advertising the fact that the item previously came complete with all the accessories listed, implying that they are no longer available, suggesting that a non-past time reference is clearly intended. However, there were no similar examples in the forums searched. It is quite possible that came in (154) illustrates a perfective sense (completion) which might not be so easily rendered by the non-past form and is primed by the presence of the adverb complete in the context. The aspectual marking in (154) belongs to the entire proposition, not to the verbal semantics alone.

6.4.2.4 Did (n’t) The first example of the PFP use of this verb came from a students’ blog-spot: (155)  I am the kind who can tell myself I will do well even if I didn’t study … But the results proves otherwise la … Just that I am wondering whether its is worth getting lousy grades if I get a free meal. wakaranai.winter-muse.net/august2004.html (29/08/04) It is not clear from the text whether the speaker intended an anterior use in (155) or a habitual, as even if I haven’t studied could just as easily be substituted, with an iterative aspectual sense. The only other example comes from a later forum: (156) For those who took flaxseed oil, did you all have any side effect? http://www.flowerpod.com.sg/forums/index.php?showtopic= 425&st=600 (24/06/08) The usage in (156), decontextualised, might appear to be in reference to a past time situation, but given the nature of the forums and the fact that speakers are not addressing known participants with any specific reference to a particular place, time or event, it would be hard to imagine that a particular instance of taking



6.4 Preliminary survey data 

 159

flaxseed oil is being referred to. The example also contains a very frequently used item in the list (took) – see below for discussion. It is almost certainly the case that the time reference is indefinite here, or referring to an indefinite anterior time period.

6.4.2.5 Ended up This verb is semantically similar to landed, discussed by Ho & Platt (1993, p. 148) for ex. (139). One example observed in casual conversation with a Singaporean speaker (male, 50s, June 2008) is: (157) a. A lot of students ended up in condominiums. Another example was found in the Flowerpod forum (11/06/08): b. erm. . . i always got such problem of taking a long while and ended up late for most of my time, e.g.: work, meet up with friends etc.. http://www.flowerpod.com.sg/forums/index.php?showtopic=66699&st=200 In both cases, the speaker was referring to an indefinite, on-going situation, not a past event (got is used with present time reference in SCE to mean ‘have’ or ‘exist’). The topic in (157b) was the time taken to have a shower (see (171)). It should also be noted that in SSE, once the past tense is used, the referential status of the subject changes from indefinite to definite: A lot of students end up in condominiums usually would refer to an indeterminate number of students, since the construction is generic, while with a past tense verb, there is an implication that the number of students is limited and identifiable. However, the consequencs for the truth conditions of (157a) are not necessarily relevant if it is understood as a generic construction.

6.4.2.6 Found 10 examples of found used as a PFP construction were found in the data (compared with only one example of find as a present habitual); there was a number from previous searches in Yahoo Classifieds as well. The previous examples included the following: Lost & Found Lost Of Malaysian International Passport; Choa Chu Kang, sg (158)  Whoever found it pls contacts Mr. [Y] at [xxxxxxx] and will be REWARDED WITH MONEY! http://sg.classifieds.yahoo.com/?intl=sg – Classifieds (08/04/06)

160 

 Chapter 6 The past tense construction

The use of found in (158) seems at first glance to presuppose optimistically that the passport has already been found. In actuality, the speaker is using a past form to anticipate anteriority in the future, as the form occurs in a subordinate clause and indicates indefinite time reference, a typical irrealis ‘threshold’. More examples include the following: (159) (Original question: Things about SG buses … Anything comments about them?) I often take SBS and found nothing comment. [‘nothing worth commenting about’] http://www.sgclub.com/singapore/things_about_sg_ 35277.html (09/06/08) The speaker in (159) may be substituting a past form for a present perfect in an anterior usage, but it is more likely that the present non-past could be used, since the preceding verb, take, is in the simple present and expresses a present habitual action. In this case, it is not a case of the anterior use of the past tense.

6.4.2.7 Graduated This verb does not appear in the forums, but one instance was observed in casual conversation of a female (undergraduate) speaker, aged 21 (9/6/08): (160) ... they graduated next year [talking about some of her friends] It is an unusual case, since it does not express habitual aspect, but future perfectivity, as it would if co-occurring with the aspect marker already (next year they graduate already). As this was the only main-verb future function of the past tense found in the data, it may be considered simply idiosyncratic.

6.4.2.8 Kept Kept is found in the semi-auxiliary use only (the lexical meaning of keep in SCE is literally both ‘put away’ as well as ‘hold’ (as a result of putting (something) away), but no examples of these uses appear in the data, instead we find the semi-auxiliary use appearing in the past tense. It is possible that the meanings associated with its lexical source (see, e.g., Heine 1993) may be still adhering to the semi-auxiliary (keep V-ing) usage in the past tense, even though in most standard varieties, the lexical semantics of the form used in such functions is



6.4 Preliminary survey data 

 161

somewhat opaque. One example occurs in a discussion forum on sharing residence with one’s parents: (161)  for me, i want to stay with my parents becos i am used to it le....although we kept quarrel over small small things but i think its fun with them ard...heehee. http://www.flowerpod.com.sg/forums/index.php?showtopic = 63157 (8/04/08)64 Another occurs in a discussion about cosmetics: (162) ic... gotta finish using my current pressed powder first... i kept on buying cosmetics when my current ones haven’t even finish half! http://www.flowerpod.com.sg/forums/index.php?showtopic = 16733&st=160 (20/05/08) In both (161) and (162) the time reference is present, or indefinite – in (162) this is indicated by the form of the present perfect in the subsequent clause. Although the verb is not aspectually completive in its lexical semantics, the action of repetitiveness expressed in keep V-ing seems to justify for the speakers a need to resort to past tense forms. Perhaps the continued iterativity is felt to be best expressed by expressions referring to repeated actualisation of the event in the past. This possibility will be further discussed below.

6.4.2.9 Landed Ho & Platt (1993: 148) discuss one example (139), though no new examples appeared in the data.

6.4.2.10 Lost One example of lost as a PFP construction was found in a discussion about the time spent getting to work in the morning: (163)  Den if i choose to come SUPEr early, and i mean super early like 1 hr b4 the time we suppose to meet... den i will surely be bored and den i go play

64 Le or leh in (161) is a discourse particle expressing comparison. See Lim (2007) for more detail on its usage and derivation.

162 

 Chapter 6 The past tense construction arcade ... After that, lost track of time and den in the end play arcade overshot.. End up late again LOL http://www.sgclub.com/singapore/how_long_take_48606_2.html (30/06/08)

It should be noted that this example is found alongside simple present forms marking the habitual, as well as the overtly marked use of the modal will expressing habituality.65 It is uncertain why lost is specifically selected to be expressed in the past tense, when the other forms remain unmarked for past, including end up, which seems to attract past marking elsewhere. However, it does express an event which is punctual (an Achievement), unlike choose, go, and play. 6.4.2.11 Made This example occurred three times, twice in a subordinate clause suggesting that it was functioning in an anterior role. In (164), the speaker uses a PFP form twice (used as well as made) in the same posting, which is a disclaimer on his or her postings: (164) Disclaimer: I am interested in an intelligent discussion about things. Sometime I used sarcastic remarks in my arguments. If you feel offended, don’t be! Please don’t be an emo. Go back to your room if you do. I point my remarks to your argument, not you. If you don’t like my post, just “skip it”. If you made any complaint, that just show that the “matured” forum is needed afterall. http://www.sgclub.com/singapore/things_about_sg_35277_2.html (090608) (Matured is commonly used in SCE where SSE would use mature.) Both examples seem to point to an anterior aspect function for the PFP construction in which the past form could quite readily be interpreted as evidence for expressing a habitual present, and the same anterior use seems to be evident in another example, in the following: (165) friends just come and go.. unless u really made an effort in contacting them or wat lo... http://www.sgclub.com/singapore/friends_come_go_ 42672_2.html (10/06/08) In both (164) and (165) the PFP made is found in a subordinate clause, which provides the context for an anterior use of the past form. The presence of

65 It is unusual to find will appearing with so many 1st person subjects, as these contexts do not allow for a predictive inference to arise from the observation of a habitual or generic situation.



6.4 Preliminary survey data 

 163

corresponding uses of PFP constructions in anterior sub-clauses in this way could indicate that the forms are expressing a dual function, that of precedence in discourse, and at the same time, potentiality for recurrence in time of the event referred to; i.e., precedence in discourse is understood as precedence in time. Bybee et al. (1994: 239) dismiss any possibility of past habitual uses of modal verbs being related to their hypothetical uses, and yet the same constructions are at issue: conditional constructions in which anteriority can be reinterpreted as a premise for open predictions (potentiality). In (164–165), the SCE use of the PFP is to signal anteriority at the expense of potentiality; in SSE, the simple present form has implicit anteriority, but leaves open the possibility of future occurrence as well.

6.4.2.12 Recovered Only one example of this form appears in the data. The full context is provided below, to illustrate the anterior usage again, this time referring to an event preceding another event in the future: (166) [A: But i’m sick and on medication. I stopped taking EPO now. Does this matter ah? Can i cont eat or stop first?  B: Guess if u don’t take EPO for a while, its ok becos u are supposed to take it consistently over a period of time b4 u can see gd results.]  A: OK. When i recovered from my sickness then take EPO again. Thanks for the advise (sic.) http://www.flowerpod.com.sg/forums/index.php?showtopic= 425&st=80 (230608) The substitution here is most likely for what could be either a present perfect or a simple present form of the verb, expressing future perfectivity rather than habitual aspect (note also this possibility with the use of stopped in the earlier quotation above it). Such uses appear to be distinct from the habitual aspect uses; they will be discussed below. They were not included in the original list, as they are weak pasts.

6.4.2.13 Saw A number of examples (3) appeared in this form for PFPs. Two clearly habitual uses occur in a discussion about women’s fashions: (167) (a) Saw lotsa people wear boots nowadays, damn stylo http://www.flowerpod.com.sg/forums/index.php?showtopic =22714&st=160 (01/07/08)

164 

 Chapter 6 The past tense construction (b) whenever i saw girls wearing boots, i can’t help but take a 2nd look. It’s so lovely lo! Stylish. http://www.flowerpod.com.sg/forums/index.php?showtopic = 22714&st=280 (01/07/08)

Both of these examples are accompanied by evidence of present or indefinite time reference, and there is no indication that the anterior is what is being marked here, though they may be substituted with the present perfect used as an anterior (have seen). The past marking on saw in (167b) may be related to its lexical aspect (as it means ‘perceive’); this will be discussed below.

6.4.2.14 Spent This form occurs once in the earlier data, in a classified advertisement: (168) Classic sprint150. Fully restored. need no dressup. ... Do not want to sell but wife jealous that i spent too much time with scooter! http://sg.classifieds.yahoo.com/?intl=sg – Classifieds 08/07/07 The context refers to spending time, and is habitual and indefinite. There is one more example appearing in the recent data, again referring to the spending of time: (169)  sometimes if nt rush maybe spent 30mins... but if in a rush, 10mins is enough le66 http://www.sgclub.com/singapore/how_long_take_ 48606_4.html (30/06/08) Spent in both examples does not occur in a temporal subordinate clause, and it is unlikely that the past tense is expressing anteriority.67

6.4.2.15 Thought This example appears twice in the recent data, and an interesting instance occurs in the older (pilot) sample (the topic was about men wearing female perfumes and vice versa): (170)  the Issey Miyake for men is not bad. thanks 4 the suggestions. i actually thought of buying a female frangrance [sic.] to use as it is more unique if

66 Le is a discourse particle expressing uncertainty. 67 The use of spent for spend could be due to phonological reasons (Bao Zhi Ming, p.c.).



6.4 Preliminary survey data 

 165

someone ask me. any recommendations? hope not those frangrance not to more for female use one. or else pple tought i gay or ah-kua thanks 18/5/06 http://www.flowerpod.com.sg/forums/index.php? showtopic=3531 ‘The Issey Miyake for men is not bad. Thanks for the suggestions. I ­actually thought of buying a female fragrance to use as it is more unique, if anyone asks me [what it is]. Any recommendations? Hope not to use those fragrances that are too feminine, or else people will think I am gay or ah kua [Hokkien word for ‘gay’]. Thanks’. The speaker’s use of tought reflects the pronunciation-guided spelling quite often found in the Singapore blog-spots, and in this particular instance the past form is used in place of a habitual or future form ‘will think’.

6.4.2.16 Took Took has by far the largest number of tokens overall, a total of 32 across the 8 weeks of searching. The reasons for this are probably that the contexts in which it is necessary to use the item often were biasing the lexical selection; e.g., taking medicine, and taking a shower are two topic-based contexts which yielded a large number of repeated instances. Examples include: (171) [Original lead] a) Q: How long do you take for shower? What’s the first thing that you will mostly do in the bathroom? http://www.flowerpod.com.sg/forums/index.php? showtopic=66699 (020608) The thread continues: b) A: I took about 20 mins minimum..... http://www.flowerpod.com.sg/forums/index.php? showtopic=66699 (020608) c) A: i took around 15 minutes http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=40 (020608) d) A: I took abt 20 mins, shampoo->conditioner/treatment, shower ­foam->body scrub, face cleanser http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=80 (020608)

166 

 Chapter 6 The past tense construction

e) A: for me is at least 30 mins. max is 1 hour.... i tried to bathe fast but in vain... the hair already took up 15 mins to wash liao.... and why hurry when its the time to relax? Hehee http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=80 (020608) f) A: I took my shower for an estimate timing of 30 minutes. But when I am abroad, my shower time reduced to half or even 75%. http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=120 (020608) g) A: i took about 20 mins....izzit consider average or long?? http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=120 (020608) h) ... normally i took 20–30 mins. http://www.flowerpod.com.sg/forums/index.php? showtopic=66699&st=160 (020608) The repetition of the past form may be being primed by each successive use of it, though it cannot be ascertained how many of the speakers using took for take had read the previous postings before posting their own contribution (the examples did not occur in sequence). The aspectual environment is undeniably habitual or iterative (reduced in (171f) is most likely interpretable as a participle without a copula in a passive structure). Another context is that of taking medicine: (172)  Lately i took FANCL’s Kale juice. It helps! NO MORE constipation!: good: Just one sachet with hot water every morning. http://www.flowerpod.com.sg/forums/index.php? showtopic=1387&st=240 (240608) (172) seems to be substitutable with the present perfect progressive (have been taking), though this could not be said of all the examples: How many litres of water you drink everyday?? How many capsules you (173)  took everyday?? http://www.flowerpod.com.sg/forums/index.php? showtopic=1387&st=400 (240608) (173) refers to an indefinite time period.



6.4 Preliminary survey data 

 167

6.4.2.17 Tried Again this occurs in the context of discussing daily habits and activities, ­co-occurring here with another example of PFP took in (171e) above. (Liao in (171e) is another discourse particle, borrowed from Hokkien.) Only one example of tried appears. It is possible that in this example, the speaker is substituting the past tense form for a perfect or anterior, referring to only one occasion in the past (if so, then took is equally likely to refer to the same past occasion). However, try is equally substitutable. The verb is not lexically perfective.

6.4.2.18 Used (See 6.4.2.11 for the single example.)

6.4.2.19 Went Only one example of went appears in the data, though this verb is discussed more extensively in Ho & Platt (1993). The example is the following, in the discussion of time taken to get to work/school in the morning: (174) hmm.. usually my sec schl starts at 7.15... i wake up 6am .. bath till 6.15.. den dress all till 6.30... and went down to the bustop near my house.. 6.35-45am the bus arrive and strght to school.. 7-7.10 arrive... http://www.sgclub.com/singapore/how_long_take_ 48606_3.html (30/06/08) Again, there is clearly a habitual context for the use of the verb. Ho & Platt (1993: 107 and 147) mention this verb form at least twice, and indicate that the present tense form go may be used when it is expressing duration (past or present), but the past tense form is used when there is a terminal boundary or destination referred to, e.g., went dere (‘there’). The explanation alluded to (p. 147) that it is the anterior that is marked on the present habitual went there does not account for the same form used in the context of a narrative about a past event, where it is contrasted with go also referring to the past (1993: 107): (175) ... we go by ourselves – after that we wen(t) dere – we reach Taipei ah – so just go ourse(lves) lah. For this example at least for some speakers, the verb form must be determined by the same aspectual conditions of the context as in (174), not the time reference,

168 

 Chapter 6 The past tense construction

which, although past in (175) and present/indefinite in (174), does not seem to determine the tense marking at all. However, in both (174) and (175) the use is accompanied by a telic complement, in contrast to the atelic uses of the other verbs in the context.

6.4.2.20 Wore This verb was mentioned above as another example of a state-for-process type, in which the verb does double duty in referring to the state that results from a previous past action, as well as the action itself, and thus can also mean ‘put on’. (Another verb of the same type is carry (also meaning ‘pick up’), though no examples of this were observed.) One example of wore appears in the data: (176) actually i stare @ girls who wore boots is cos they r so nice. http://www.flowerpod.com.sg/forums/index.php? showtopic=22714&st=280 (01/07/08) It is more idiomatic to suggest that the meaning here, though, is ‘wear’, rather than ‘put on’, though the dual meaning may be influencing the use of the verb in the past, as the implied sense is perfective and resultative at the same time (this could also be a subordinate – clause anterior use).

6.4.3 Distributional frequency The examples shown above may appear at first to be somewhat haphazard and unsystematic, but it must be borne in mind that Singaporean Colloquial English is not a learner variety, nor is it in a stage of acquisition (as most of its speakers, particularly of the younger, computer-literate generation) are L1 speakers and it is thus not affected by interlanguage ‘errors’ such as are typically described in the early days (1970s and 1980s) of research into Singapore English. Studies such as Alsagoff and Ho (1998) vouch for the inherent, internal systematicity of SCE. However, even an established contact variety may exhibit deviations from its own standards of conventionalisation, showing the continued effects of contact, and change vs. stability, over the short time period that it has been in existence as an L1 variety. A more accurate survey of the data above might therefore use direct elicitation methods amongst a representative population sample in order to obtain some idea of the frequency of occurrence of the anomalies. However, with the

6.4 Preliminary survey data 



 169

original data having been taken from internet sources, it was considered essential to maintain the same genre in any frequency counts. For this reason, a ‘topicbased’ approach was used; i.e., the original archived lead question was followed as the most natural elicitation method for obtaining the appropriate functiontype in responses. For example, a question on daily habits became the means to elicit forms expressing the present habitual (e.g., 171a). In order to obtain an approximate estimate of the frequency of PFP ­constructions relative to other forms which could have been used but were not, a total count was taken of the entire posting of the Flowerpod Forum which elicited the large number of took PFPs in (171). The following counts were taken: Tab. 3: Responses using habitual aspect in answer to the lead posting: How long do you take for shower? What’s the first thing that you will mostly do in the bathroom? (ex. 171) Present tense forms

Will + V

PFPs

Total

84.5 (65.5%)

32 (24.8%)

12.5 (9.7%)

129

The count included PFP constructions other than took, though all but one (ended up) of the PFPs found was the verb took. Half values were given to respondents whose replies included two different forms of expressing habituality. Odd examples using can + V and would + V have not been included (though examples in which a real conditional, with a simple present in the protasis and a will + V in the apodosis, have been included, as the simple present is a possible alternative to the will + V in the apodosis). There were 276 posted replies in total but the figures in Table 3, a total of 129, represent the number of replies that used a finite verb form expressing habitual aspect. This small sample provides some indication of the relative proportion of the past tense form to other forms likely to be selected in marking the habitual aspect. At roughly 9.7% of the total given, it does not seem to represent a significant tendency at all. It must be remembered though, that habitual aspect is not the only environment in which PFP constructions are found, and a higher total percentage might be found in a broader context in which habitual aspect is not so predictable in the responses. With such figures, then, only time will tell if the feature could be described as either an ‘emerging’ feature of grammaticalisation (after the spirit of Hopper 1987), or one that is recessive, and progressive diachronic evidence would be needed to support the latter case. In addition, the possibility of weak verbs occurring in the PFP construction cannot be discounted altogether. This will be discussed further below.

170 

 Chapter 6 The past tense construction

6.4.4 Interim summary In the data discussed so far, it is clear that there is evidence of a tendency across a wide range of lexical contexts for the past tense to appear in contexts in SCE in which the present tense would be most natural in SSE. The data do not show a strong tendency for such uses to be found in any particular lexical verb, nor with any aspectual characteristics of the verbs concerned; however, at the same time because of the nature of the searching it is difficult to quantify the results. In many ways, though, the data reveal that the marking of past tense in present tense contexts is not restricted lexically – it appears to turn up almost anywhere and quite unpredictably. A much more extensive survey across a corpus with a greater word-count might produce interesting findings; however, the present data are useful in that they provide ideal contexts for the production of present habitual aspect. On the other hand, the marking of simple presents as past in SCE is not restricted to habitual aspect alone, as will be discussed below.

6.5 Discussion The most frequently-appearing forms in the earlier data so far were the following: took (32 tokens), found (12), kept (6), ate (5), saw (5), did(n’t) (3), thought, made, spent (all 2 each), and recovered, lost, went, wore, used and bought all appearing once. However, the frequency of each individual item need not be taken as explanatory for the present hypotheses, as there is no way of telling as yet whether the frequency is due to the nature of the subject matter used (e.g., for the topic of taking a shower, as in (171)), or because the frequency is there whatever the subject matter. Interestingly, though, in the eighth week of searching, the same items recurred: did (not) – once, kept (twice, once with a lexical meaning), found (2), saw (2), and said (in a subordinate clause, suggesting anteriority). The possibilities that there were lexical factors restricting the types of verbs that can appear as PFP constructions were considered originally, that habituals are marked as anteriors in SCE (e.g., Ho & Platt 1993: 147), and anterior is expressed using the past tense, or that past tense marks both anteriority as well as lexical aspect (while already marks only grammatical perfectivity). However, it is also well-known that grammatical perfectivity is constrained by the lexical perfectivity of the verbs it co-occurs with anyway (Bybee et al. 1994: 92), and already appears in anterior functions as well. Furthermore, the appearance of stative items such as tried or kept in the past form, would seem to belie a hypothesis based on the priority of lexical aspect alone. There are in some cases correspondences between the SCE form and a translation in Mandarin which would necessitate

6.5 Discussion 

 171

a resultative compound predicate (i.e., one comprised of two juxtaposed verbs, the first verb expressing the action and the second verb the result – see Li and Thompson (1981[1989]: 54)), but there are also exceptions and the entire sample of PFP constructions found does not constitute a complete defined set of such cases. In order to further rule out a possible lexical bias then, a search of all the regular pasts found in PFP constructions in the ‘Career talk’ thread sub-component of the Flowerpod corpus site was undertaken,68 and a number of new lexical items appeared, also PFPs: asked, blamed, created (2), decided, enjoyed, faced, lacked, managed, pampered, realised, refused, required (2), resulted, seemed, started, stretched, wanted (2), wished, worked, all occurring once each, except where marked. Moreover, two new verbs were added, both normally stative but in the use in which they appeared, expressing performative speech acts: I agreed was used six times out of a total of 40 (which included 2 × 0.5 token counts for a single speaker who used it twice); i.e., 15% of the uses of I agree were in the past tense, and I believed appeared two times out of a total of 10 (8 × I believe). In the latter case, there would not be sufficient data to predict whether the relative proportion of past to present usage for the performative function would continue with more data, but in the former case, it looked slightly more likely. Altogether, a total of 63 tokens of PFP constructions were found in the entire corpus sample of 86,005 words. Examples of these include the following: (177) (a) Posted by: clarissavera Jul 22 2008, 01:18 AM Realised that our young working adults still kind of getting allowances from their parents, some even refused to work after graduation, knowing that their parents will still provide for them. Do share your views on this current trend? (b) Posted by: hondavvvtec Jul 22 2008, 04:14 AM well, i dont blame these youngsters. i blamed the parents instead. (c) Posted by: chemsium Nov 6 2007, 12:07 AM hmm can. maybe go study some Hr courses at SHRI. then go try apply for an HR job. do it while u are young. dun wait until next time then decided u want to switch career then too late liao. ‘Hmm, possibly. Maybe go and study some Human resources courses at [name of college]. Then go and try to apply for an HR job. Do it while you are young. Don’t wait until later then decide you want to switch careers, as then it is too late [discourse particle]’.

68 The ‘Career talk’ sub-component corpus examples spanned a period from 17/05/08 to 5/09/08, and the word-count included posting details and dates.

172 

 Chapter 6 The past tense construction

(d) Posted by: XiaoPinky May 14 2008, 11:41 PM ya u are right, those ppl in IT field mostly wished to get out of IT lol. ‘Yes you are right, people in the field of IT mostly want to get out of it [lol].’ All of the above examples come from the ‘Career talk’ forum of the Flowerpod website (http://www.flowerpod.com.sg/forums/index.php?s=914f361d54e8ce5 775841cc04264fae3&showforum=28) and occur in contexts that may express or imply potentiality in SSE but are marked using prototypical realis forms in SCE by the speakers who posted them. Note also in (177c) that the speaker has matched the definiteness of the past tense verb use with a definite demonstrative determiner those for the subject people which would not normally be required if the construction were considered to be a generic one, as can be seen by the gloss. It appears from this that the construction is not substitutable for a generic construction at all in the speaker’s dialect, and (see also (157) above) that the use of the past tense is clearly motivated by a functional need and is not simply an unanalysed hypercorrection. Moreover, the occurrence of performative uses of the past enables a new stance to be taken on the entire data and offers reason to reject the earlier hypothesis based on lexical priority, for example: (178) Ya.. I agreed. And recently it isn’t easy to get a job so I’m happy with what I’m offered.. Hopefully the working environment there is good. (Posted by: evelyn83 Jun 19 2008, 03:12 AM) Evidence from examples such as (177–178) indicate that a broader explanation is required. In the use of already as a grammatical perfective marker, the categories of expressing perfectivity in Mandarin Chinese are replicated very closely, as shown by Bao (2005) in his hypothesis of systemic transfer of the entire paradigm. However, with regard to English, which is not the model language (Heine and Kuteva 2005) on which the aspectual system is transferred, there are paradigm gaps that may not seem apparent on first appearances. The simple present tense has been described by some accounts (e.g., Comrie 1976: 68; Dahl 1985; Langacker 1982, 2001) as a means of marking perfectivity in present time reference. Although the functions in which present perfectives may occur are somewhat restricted in English, nevertheless they exist, and are not overtly marked at all, as the simple present is seen as the default category for present tense, the progressive occupying the function of the marked category (Bybee 1994 prefers to label categories like the simple present as grammaticalised ‘zero’ counterparts of their unmarked oppositions). Thus, in the use of a simple present form, the speaker is often expressing an event, which merely by virtue of contrast with the progressive, is seen in its entirety but c­ o-occurring with the moment of speaking, as in stage directions, sports commentaries and ­play-by-play accounts. Performative functions also fall within this category.

6.5 Discussion 

 173

Perfectivity expressed as concurrent with the moment of speaking is not marked with already in SCE as it is not marked with le in Mandarin Chinese either. The situation of contact therefore leaves a class of perfectives that cannot always be complemented by the perfectives of the model language; there is thus no cause for replication of the class using the grammaticalisation of already. It is not known at this stage what factors motivate selection of one form or the other, but there is, though, clearly a need to establish the existence of such a competition between modelling the model (substrate) languages, and modelling the source (lexifier) language in the contact situation, something that has been overlooked in previous accounts of contact grammaticalisation, to some extent.

6.5.1 Present-perfectives and the realis-irrealis interface in English What this means is that many of the functions normally served by a simple present form in SSE are served by a past tense form, the prototypical marker of perfectivity, for some speakers of SCE.69 These may include, as shown by the majority of examples in the data, (i) the antecedent clauses of temporal or conditional constructions; i.e., ‘projected’ anteriors (e.g., 155), (ii) the habitual present (e.g., 153), (iii) the use of performatives (e.g., 178). Although not marked by an irrealis marker in standard English, such environments are considered to be fairly prototypically irrealis in that they exhibit a degree of possibility or potentiality, particularly the environment of future plans. (According to Langacker (1997: 221), such environments are amongst those typically operating in a mental space distinct from current reality, since they refer to events that are not constrained by the observation of the event as it actually occurs.) However, they are encoded for some speakers in SCE by the marking usually attributed to a realis form: past tense (notwithstanding the arguments for metaphorical association between past tense and irrealis expressed in studies such as Fleischman (1989)). This indicates a highly volatile interactive domain which in one dialect (or language) may be regarded as irrealis, while in another the same domain seems to be considered as realis. The reasons for the variation between realis and irrealis depend very much on the particular function involved. For (i), according to Langacker (2001), the use of the simple form of the present tense expresses present perfective and is justified by the fact that the events described are virtual, and awaiting actualisation (2001: 269),

69 Given that the progressive in Singlish is closer to a true imperfective, appearing in lexically imperfective situations (e.g., stative verbs), it is not surprising that the simple form of the ­non-past verb is marked in accordance with the perfective functions it conveys.

174 

 Chapter 6 The past tense construction

and are thus extensions of the basic sense in which actualisation is concurrent with the moment of speaking. Hewson and Bubenik (1997: 210), however, have pointed to a category labelled ‘Immanent aspect’ in Indo-European languages, which is simply represented by the non-progressive non-past. This nomenclature embraces both an aspectual as well as a modal function; in fact, they also note that in Old English and in early Germanic the simple non-past tense represents both present and future (p. 219), i.e., it served to express future irrealis meanings as well as those relating to present, with indefinite time reference, as noted earlier. Thus, although Langacker (2001) may have regarded the future use of the simple present today in terms of virtual actuality, it is clear that if historical patterns are taken into consideration, it is the imperfectivity of the present time reference inherent in the use of the simple present that seems to be called upon to express an extended, future meaning, not the metaphorical transposition of the perfectivity it seems to express at the same time. Such implicit, grammatical perfective meanings may have simply arisen gradually in default opposition to the emerging presence of a general imperfective form that had contributed to the source of the present-day progressive aspect in Old English (see, e.g., Ziegeler (2006)).70 A different problem arises with habituals (function (ii)). We have seen the way in which in other languages the lexical aspect of the predicate is p ­ erceived to take priority over the more abstract imperfective grammatical aspect of the category, resulting in perfective marking. Habituals, although a type of generic aspect, refer to situations that have no anchorage in time or space, never ­co-occur with the moment of speaking, and are thus quasi-modal in c­ haracter (as mentioned earlier in reference to Givón (1994)). Any aspectual domain where perfective lexical aspect and imperfective grammatical aspect coincide, or vice versa, it has been noted, presents a potential semantic conflict from which the outcome is modal inferences, as noted in Ziegeler (2006).71 This is because

70 According to Quirk and Wrenn (1957: 77), Old English only had two tense inflections, present and preterite, and any time reference apart from past or present was usually contextually determined, often using adverbs. 71 The combinatory hypothesis states that: “Modality arises principally as an inferential byproduct of particular lexico-grammatical co-occurrence conditions” (Ziegeler 2006: 21). In such cases, it is found that the clash of perfective lexical aspect with imperfective grammatical aspect, and vice versa, often produces (irrealis) modal inferences. In the cases of the perfective presents illustrated, this hypothesis could only apply to habituals, and projected anteriors, which carry potential meanings in context (i.e., perfective lexical aspect in an imperfective grammatical aspectual environment). For performatives, an unambiguously grammatically perfective environment, the lexical aspect of the performative verb may determine whether there is aspectual conflict (producing irrealis inferences) or aspectual concord (producing realis meanings).

6.5 Discussion 

 175

aspectual ­indeterminacy gives rise to truth-conditional indeterminacy (if the difference between aspectual meanings such as completion vs. non-completion may be interpreted as the difference between fact and non-fact). It is not surprising, then, that there is a need to mark present habituals, an area of irrealis modal derivation in some languages, with a form most often associated with the expression of realis in others. For function (iii), the environment is one of a de facto use of perfectivity: predicates in speech acts such as I agree, I believe, I hope, and so on, are stative verbs and as such should be incompatible with any form of the perfective. However, communicative speech act verbs such as I announce/pronounce/name and similar may be considered to be non-stative in the context of utterance of a performative: they effect an action which is a speech act, enabling the actualisation of the event referred to in their predicates. Verbs of perception and cognition such as agreeing, hoping and believing, are construed as performative speech act verbs in the same way, since the very communication of the speaker’s thoughts in this way constitutes an act of (metaphorical or actual) change. As such, they can acquire the perfective aspectual encoding by analogy with their speech act counterparts, and are perceived as perfective in nature by the speakers who use past tense forms to express them.72 For the realis-irrealis interface, though, they may be only marginally realis, as they are inherently stative and cannot be viewed as bounded. The de facto nature of their perfective encoding is also made evident by the fact that they cannot co-occur with the perfective adverb already in SCE (Bao Zhiming, p.c.). In other words, speakers who do encode them in the past tense are acknowledging the perfectivity of the grammatical context which cannot be expressed systematically in SCE by any other means. Thus, there may be more than one reason for marking such items as perfective using the past tense in SCE: the use of past tense as simply an alternative to already with anterior uses (i), and also, the speaker’s inability to use already in marking perfectivity in environments such as habitual and the performatives ((ii) and (iii) above) where it would be ungrammatical Singlish. The reason could be due to the possible inferences of potentiality that seem to arise in such contexts, which would be incompatible with an aspect marker of completion, marking an event in its entirety. What is curious is the fact that in standard usage, the functions listed above as (i) and (ii) are considered to be slightly modal in

72 Comrie (1976: 20) discusses the similar way in which perfective aspect may be coerced into the meaning of stative verbs such as see, sit, and know, by inducing an inchoative phase into their meaning.

176 

 Chapter 6 The past tense construction

character, though in the data from SCE they are marked using a form usually assigned to non-modal functions, since past tense marks what has occurred, not what is potentially likely to occur. At the same time, it may be questioned just how much realisness is attributable to the SCE PFP constructions. It has been suggested73 that there may be a relation between the use of past tense in habitual environments and a need to express evidentiality (i.e., the past is being used as evidence for expressing habituality, which is, after all, a predictive aspect). This suggestion has its drawbacks, as there exist in the dialect additional means of expressing such evidentiality in terms of experiential aspect,74 and that is in the use of the adverb ever, as we have seen earlier. Evidentiality is also a highly complex category which involves notions related to hearsay, speaker commitment, direct and indirect observation, and many other factors that do not seem to be immediately relevant to the problems presented (for example, there does not appear to be a need to ascribe an evidential function to the use of past tense in performatives, or habituals with first-person subjects). However, the suggestion accords well with accounts such as Langacker (1997) in which habituals are seen as referring to actual instances of an event type, whereas certain generics, such as This door opens to the inside (p. 198) need not be observed in order for their truth value to be ascertained. Similar distinctions are outlined by Lenci and Bertinetto (2000), and Krifka et al. (1995), who consider even unmodalised generics as ‘modal’, in the fact that they refer to open-ended classes of entities (e.g., A lion has a mane). It is felt, along with Ziegeler (2008b), that such borderline examples render themselves to a modal interpretation simply because, as Langacker (1997) illustrates, they are remote from observed reality, not remote from reality. It could be argued, therefore, that that if any evidential function for the use of the past tense with habituals is proposed, it is because of a need to distinguish observed reality from the not-necessarily observed truth value of a non-habitual, generic statement. Thus, speakers of SCE are using the past tense form to express not just what is known to have occurred, but what is known to have occurred and may well occur again, in other words, habituality is an implicit sense of prediction contained in such constructions. But habituals are not the only contexts with which PFP constructions are associated. There is also a marginal functional analogy with the Old English ­preterite-present verb types, as noted earlier. According to Leiss (2008), the

73 By Caterina Mauri and Andrea Sansó (p.c), in editing the previous publication of the study. 74 See Chappell (2001) on the relation between experiential aspect and evidentiality (in Chinese languages), as discussed in the previous chapter.

6.5 Discussion 

 177

encoding of binary grammatical aspectual distinctions in the older stages of the ­Germanic languages functioned to express modality in the absence of modal verbs in the same way as the event semantics of a preterite-present verb. Perfective verbs in the present tense, in the older stages of Germanic, as well as in Slavic, are understood as futures (Leiss 2008: 30), and the loss of aspectual divisions in Germanic resulted in the need for functional renovation using an analytic system of modal verb auxiliaries + infinitives to express the modal meanings formerly conveyed by aspect alone. The preterite-present forms were the sources for many modal verbs in Germanic, since they contained a portmanteau, aspect-modal semantics consisting of a result-state, which was a consequence of a prior perfective event implicit in the use of a preterite stative verb (see, e.g., Bybee et al. (1994: 92). Kotin (2008: 374) gives an example of the preterite-present verb class from German: ich kann, meaning, ‘something is in the state of having been/being recognized/understood by me’, i.e., ‘I have understood/recognized X’, therefore implying: ‘I now know X’). Clearly, the functional parallels with the examples seen above point to such an analogy, in that there are inferences of potentiality that inevitably arise from the completive use of a perfective or past form, expressing prior actualization that has a bearing on a current and future situation arising from it. If such parallels can be (speculatively) postulated at all, it seems plausible that speakers of SCE might prefer to encode the potentiality of a future situation as pragmatically implied in the expression of one that is now complete. This may involve the continued state of habitual physical events, such as in (171), i.e., I took around 15 minutes, [before, therefore continue to do so in the future], or the extended state resulting from a single, performed event of cognition, as in (178): I agreed [now, and therefore continue to hold the agreement]. The corresponding use of the past tense in anteriors, discussed in 6.4.2.11, in which precedence is prioritized over potentiality, is a more transparent example of the functions shown, as precedence of events is reflected in the iconicity of discourse sequencing. However, the fact that the extension of a state resulting from a past event is only pragmatically implied invites the typical grounds for the creation of irrealis inferences: only the perfective event is expressed as actualized, not the inferred situations resulting from it. The function of the PFP construction in such contexts can therefore be isolated as grammaticalising PRECEDENCE, i.e., a past tense gram-type that expresses perfectivity but also carries pragmatic inferences of potentiality as part of its extended meaning. In this way, then, it mirrors the function of the perfective (simple) present construction in SSE and other standard uses, which carries a hybrid semantics of both completion and potentiality for recurrence in its aspectual composition. In such contexts, the distinctions between realis and irrealis also become highly elusive.

178 

 Chapter 6 The past tense construction

6.6 Summary The present study illustrates that the marking for past tense in present tense environments in SCE, a contact variety of English, is related to the need to express perfectivity at the expense of imperfectivity, an ambivalence found commonly associated with habituals crosslinguistically with both present and past time reference. Given that the implications of using a perfective marker are those of expressing realis as well, the findings of the study also illustrate the possibility of a realis-irrealis distinction in SCE outside the limitations of other modality markers, in which past tense marks events that are actualised or in evidence at the moment of speaking, while present tense marks those that are as yet unactualised, or no longer in evidence at the moment of speaking (as in counterfactual readings of past habituals). However, the primary function of the Singaporean English PFP appears to grammaticalise precedence, a function expressing perfectivity but implying reference to potential recurrence or continuation in the future at the same time. This applies to all three functions discussed: anteriors, habituals, and performatives. In addition, it should also be noted that there is apparent competition between past tense marking and already in at least one of the environments, (i), projected anteriors, which may be indicative of the possibility that the picture represented by the data is one of a merging of two complementary systems: the PFP construction as a means of marking perfectives, with its inferences of potentiality through precedence, and the grammatical aspectual means of marking simply perfectivity through the use of already, with no nuances of irrealis whatsoever. Such findings aside, the study also highlights a need to reconsider what contact varieties may reveal regarding the category distinctions in non-contact or mainstream varieties. That there is a realis-irrealis distinction possible in such contact varieties, and one that is subject to so much variation, throws light on the areas in the standard language system in which ambiguity or vagueness as to category boundaries may be found. The ambiguity may readily be explained by the presence of implicit senses of irrealis in what are often inferred as realis contexts in English (using the simple present perfective), but are not generally considered as such because English has no specific form of grammaticalising the differences between realis and irrealis as distinct binary modal domains anyway. The past tense is selected for this function by the speakers of a contact variety in which it represents the transferred realisness of perfective aspect into the domain of potential reality, and the implicit sense that an event that was completed in the past is considered as actualised, and relevant to future recurrence. In the case of present tense marking, it is clear that it is used with past habituals in SCE to indicate that the past habitual state no longer holds. For the same reason, it would

6.6 Summary 

 179

be natural to encode present habituality with past tense, indicating that the state expressed continues to hold, or is an observable reality. Thus, for speakers of SCE, the selection of tense marking correlates quite harmoniously with realis/ irrealis distinctions at certain levels. The study leaves open the question of the identification of the past tense as a schematic construction-type in SCE, and why it was selected to fulfil the functions of expressing precedence when such functions are not obligatorily encoded in Standard Singapore English, or, if at all, require the experiential and indefinite sub-function of the present perfect. The relations with an experiential marker such as ever, are quite clear also, though whereas in the case of ever, the experiential construction marks an event only as happening at least once in the past, with the past tense construction in SCE, the marking is of events that have happened in the past and furthermore, carry implications of recurrence in the future. Ever does not hold the function of prediction in the same way. It is clear from the data that habitual aspect is a category that requires some type of explicit marking in SCE, and cannot be left unmarked as a default category in opposition to the progressive as in SSE (note also the tendency to over-use the modal verb will for habituals in Table 3). This may be partly due to the irrealis potential of the habitual aspect, which is implicit in the use of PFP constructions in SCE. The need to reconcile the past tense construction in Standard Singapore English with that used in SCE will be defined more clearly in Chapter 8.

Chapter 7  Bare noun constructions 7.1 Introduction One of the most frequently observed features of Singapore Colloquial English is the presence of bare nominals, unmarked by any form of determiner or modification for number, which gives them the appearance of mass nouns. The first studies I undertook on bare nominals in Singlish were published in 2003 in a paper titled ‘On the zero‑­plural in Commercial Singaporean English’, which appeared as a chapter in a volume on the grammar of Singapore English, edited by David Deterding, Ee Ling Low and Adam Brown: Singapore English: Research on Grammar (­Singapore: McGraw Hil). Many of the earlier data examples in the present chapter first appeared in that study, as well as some of the discussion in section 7.4. Much of the inspiration for the entire volume is derived, however, from a paper presented at a workshop on coercion (‘On the concept of coercion’) which took place at the AFLiCo Conference, University of Paris 10, Nanterre, in May 2009, convened by Peter Lauwers and Dominique Willems. The paper was then published in Constructions and Frames 2 (2010), 33–73, under the title ‘Count‑­mass coercion, and the perspective of time and variation’, and some ideas from the paper have also been discussed in the ­Handbook of World Englishes (Ziegeler 2014b) in a chapter titled: ‘Are c­ onstructions dialect‑­proof? The challenge of English variational data for construction‑­grammar research’. Further sections of the present chapter appeared in another paper titled ‘Towards a composite definition of nominal modality’. The latter was published in a collected volume compiled following the 2011 Linguistic Society of Europe Con‑ ference, held at Logroño, Spain, titled Covert Patterns of Modality, and edited by Werner Abraham and Elisabeth Leiss and published by Cambridge Scholars Publis‑ hers (Ziegeler 2012b). In that volume, the ideas originally focusing on the use of bare nouns in SCE to express generic reference were extended further to embrace a type of nominal modality. The modality of nouns will not be the primary concern of the present chapter, though, since, as for all the case studies so far, we are dealing with the question of describing certain constructions of SCE which overlap in form with their standard variety counterparts, but not in meaning. Thus, the present chapter reviews again the discussions in those previous papers, and in particular, raises the question how surface syntax alone can provide the sole source for construction meaning, viewed from a cross‑­dialectal perspective. The topic of the main discussion is the bare noun phrase construction (BNC), a construction‑­type with ambiguous semantics seen in the light of contact dia‑ lects of English such as Singapore Colloquial English, and represented as either the absence of the indefinite article on singular count nouns or the absence of

182 

 Chapter 7 Bare noun constructions

plural marking (zero‑­plural) on plural count nouns. In standard English, bare nouns, unmarked for number and occurring without determiners, can only be represented as mass nouns, unless they undergo type‑­shifting to mass noun status, as discussed below. In SCE, it is possible to find unmarked count nouns as well, indicating that the category of number marking overlaps semantically with the category of referentiality, since all such bare noun constructions are non‑­referring, or non‑­specific, in terms of their distribution in discourse. As such, they present a viable contradiction to the hypothesis of construction coercion and count‑­to‑­mass type‑­shifting accounts, as discussed later. One of the aims of the present chapter, therefore, is to test the coercion and type‑­shifting hypothesis against such data. The present chapter also notes the significance of pragmatic mechanisms such as metonymy and metaphor in dismissing a type‑­shifting ana‑ lysis of such bare noun constructions: coercion explanations, for the most part, are exhausted by the application of such mechanisms in standard varieties. Chapter 7 will therefore address the following issues: in the next section it will introduce briefly the BNC in SCE and the interpretations given to it so far; in the third section, it will review the properties of specificity and countability in nouns. The fourth and fifth sections will discuss more data from SCE, ­comparing the SCE examples of BNCs with those of creoles and other contact languages. The sixth section will discuss the use of the BNC in Chinese, comparing this with present‑­day SCE. Section 7.7 will present the argument for a coercion analysis of bare noun constructions, which has been used to explain the presence of bare count nouns type‑­shifted to mass noun status and of mass nouns marked with indefinite articles and type‑­shifted to count noun status in standard varieties of English. Section 7.8 will argue against the use of the coercion analysis with SCE data, and Section 7.9 will summarise and review the chapter.

7.2 Number marking in Singapore Colloquial English count nouns The Bare Noun Construction in SCE refers to a construction which generally refers to a count noun but has the form of a mass noun; i.e., it is unmarked for number, singular or plural, neither with the indefinite article nor with the plural ‑­(e)s. To date, there appears to have been little research of any depth on the feature of plural marking in Singaporean English, and it has not been correlated to any great extent with parallel research into the absence of indefinite article on count nouns. It was observed by Platt (1991) that plural marking occurs more frequently with more ‘proficient’ speakers, though such a factor need not be uniquely asso‑ ciated with the feature of noun plural marking. What is of interest in the present



7.2 Number marking in Singapore Colloquial English count nouns  

 183

study is not the presence or absence of the plural –(e)s morpheme, but the ­semantic characteristics which govern its absence in particular environments, such as in reference to an indeterminate quantity. In fact, Ho and Platt (1993: 22) found that the use of plural marking was highest in environments in which a quantifier or determiner preceded a count noun (e.g., both, many) – 82% of the speakers surveyed, followed by environments in which a quantifier or determiner preceded a count or non‑­count noun (e.g., any, all) – 82% of speakers, and least used when an indefinite plural noun has no determiner preceding it (e.g., We get customer like that). They assumed no direct link with the Chinese substratum. Ho and Platt (1993) also cited a study of Hong Kong English (Budge 1986) in which the marking of plural occurs more frequently with nouns preceded by a quantifier, and Platt (1991) noted the same results for a study of plural marking by Indonesian Chinese students acquiring English (Sudipa 1986). Platt’s conclu‑ sions were that there was some semantic influence, but he takes the investigation no further. Ho and Platt (1993) made an allusion to Bickerton’s (1981) Bioprogram of typical features of creoles, suggesting that it was related to a specific‑­non‑­ specific distinction rather than a definite‑­indefinite opposition, and zero‑­plurals marked non‑­specific reference. It is clear that definiteness is more grammatica‑ lised than specificity, but it is not clear whether the manifestation of this in the manipulation of standard English morphology to express such distinctions is due to a universal characteristic of contact languages, or to the cognitive constraints of the Chinese substrate, as the similarities between (at least) Mandarin Chinese and a typical creole have been noted before (Mufwene 1990: 10). As mentioned in Chapter 2, Mandarin Chinese is not a substratum language in Singapore, but it is has an active role as a Chinese lingua franca in present‑­day Singaporean society. Other studies, such as Alsagoff and Ho (1998: 143) proposed the notion that the zero‑­plural of the noun phrase in Singaporean Colloquial English was an example of a noun phrase which might be considered either count or non‑­count; e.g., ticket in the following example: (179) a. She queue up very long to buy ticket for us i.e., it has a use in (179) similar to that of a mass noun. What they actually mean though, is that there is no countability requirement in SCE for unquantified NPs. Similarly, they suggest that SCE often treats singular count nouns as non‑­count by the omission of an article (1998: 144): b. She got car or not? ‘Does she have a car’ though they do not venture any further to explain the reason why this may be so. The absence of plurality and indefinite articles in such cases suggests

184 

 Chapter 7 Bare noun constructions

the (possibly variable) absence of a countability system in SCE, which may be ­rendered ­redundant in the variety closest in form to the languages of the sub‑ strate. However, the frequency with which number marking may be applied to nouns in SCE is difficult to estimate on the basis of the data presented so far in earlier studies. A corpus search would be equally difficult to undertake, given that the search items are open‑­class items, and searching for the indefinite article or the suffix ‑­s would yield an overwhelming amount of redundant data. Ho and Platt’s (1993) statistics may be somewhat out‑­dated by present‑­day comparisons, but the absence of number marking on count nouns is still pervasive, as seen in the data selected in section 7.5. It could be considered that count nouns with no referentiality or specificity are being conceptually marked as mass nouns by speakers of SCE. In non‑­contact dialects, such a phenomenon has been described as a construction coercion; however, it is open to question first whether coer‑ cion exists as an independent mechanism to explain such conceptual shifts, and second, whether if it does exist, it applies to the data at hand. In order to explain the feature using a construction‑­based analysis, there are other possibilities to consider, in the first instance, including comparisons with similar features in creoles, as will be seen below. In addition, the distinctions of specific and non‑­ specific nouns must be defined and outlined.

7.3 Specific and non‑­specific nouns The category of specific reference has been a highlight of work underaken in the context of creoles; e.g., Bickerton (1981), amongst others, and also a topic of consideration in the grammar of African languages (Givón 1973, 1978, 1981). Givón preferred the terms referential/non‑­referential distinctions, rather than specific/non‑­specific, but used them with almost the same definition: that the nominal expression should have non‑­empty references; i.e., “to ‘exist’ within a particular universe of discourse” (1978: 293). In the same study, he alludes to the fact that ambiguity between referential and non‑­referential reference is asso‑ ciated with the scope of modality, though where the noun must be interpreted ­referentially, the ambiguity does not exist (1978: 294). Givón (1973) provides a comprehensive description of all the possible grammatical environments in ­ which ­non‑­referentiality may be interpreted in nominals. More recently, the distinction of specific/non‑­specific reference leaves many more mysteries in its wake. Geurts (in press) summarises much of the previous research on the formal semantic definition of specificity as outlined in Kasher and Gabby (1976), Fodor and Sag (1982), Manga (1996), Kratzer (1995), Yeom (1998), and van Rooy (1999) in which it is claimed as referring to the knowledge



7.3 Specific and non‑­specific nouns 

 185

of the speaker of a particular NP referent, distinguished from a definite NP by the fact that the hearer does not know the identity of the referent. He adds that there are no linguistic devices dedicated to the function of expressing the speaker’s knowledge alone. In this respect, he refers to Haspelmath’s (1997) claims of the number of languages available in which the same indefinite NP can signal the knowledge of the referent by the speaker as the absence of speaker‑­knowledge of the referent; e.g., German irgendein, which can be used specifically or non‑­ specifically (Geurts, in press: 4): (180) Wilma hat vor, irgendeinen Schweden zu heiraten Wilma intends some‑­or‑­other Swede to marry. The English translation: ‘Wilma intends to marry a Swede’, is thus equally ambi‑ guous between specific and non‑­specific, illustrating also that the English inde‑ finite article is also not marked for specificity. However, in classic cases such as (180) (in English as well as in German), it is arguable what form of specificity is actually under consideration: the absence of a social identity (i.e., the name of the Swede) or the absence of the speaker’s personal identification of the referent (which may involve any kind of characteristic). In either case, the category of spe‑ cificity renders the interpretation of nominal reference to involve a high degree of subjectivity. Geurts’ main argument is that specificity should not be considered a distinct feature of many languages, as it is encoded in the same way as either an indefinite or a definite expression, e.g., in Lillooet Salish (citing Matthewson 1999) where the indefinite article can only appear in the scope of what are commonly referred to as negative polarity items. The presentative function of the article across lan‑ guages though, be it definite or indefinite, is usually specific; this is due to the nature of the information structure of the utterance, as Geurts points out. Thus, there is therefore no hard‑­and‑­fast crosslinguistic rule determining whether spe‑ cificity should be encoded in the same way as definites or in the same way as indefinites, specifics seem to be at an intermediate point between the two dis‑ course categories, as is shown in many languages (see also Givón 1978). Underlying the specificity/non‑­specificity distinction is the fact that s­ pecificity implies existence of a referent, whereas non‑­specificity may imply non‑­existence, but need not. Accordingly, Von Heusinger (2002) characterises specificity as ­existentially presupposed (2002: 246), and disambiguated (in English) by the use of a certain. Von Heusinger (2002) attempts to seek out a more rigorous semantic definition of specifics, suggesting that the speaker’s knowledge of the referent is inadequate as a definition (after all, if this is not overtly encoded, how does one know what the speaker has in mind?). It could be argued that a certain rein‑ forces the lost sense of specificity that the indefinite article once had historically.

186 

 Chapter 7 Bare noun constructions

Von Heusinger isolates the specificity distinction in (English) NPs as dating back to Baker (1966) and Fillmore (1967), also citing Givón’s (1978) use of the term referentiality to mean the same thing. However, his own explanation of the term specific is upheld by discourse representation theories (as is Geurts’, to a certain extent), and he suggests that the use of a non‑­specific nominal is an indication that it cannot be linked to a previously established discourse referent (2002: 252). He approaches the problem of definition by maintaining that specificity is just a matter of linking the referent to another already introduced referent in the dis‑ course, or to the speaker. Thus, it appears that the category of specific vs. non‑­specific nominals is something of an overlay on other, marked distinctions such as definiteness, and that some languages align specificity with definite markers while others align it with indefinite markers. Von Heusinger (2002: 254) claims that there are no articles dedicated to specific distinctions alone in any Indo‑­European languages. We have seen that there are at least two accounts which indicate a ­discourse‑­pragmatic explanation for the use of specificity distinctions; such accounts cater for the dis‑ tributional occurrences of specific vs. non‑­specific markers across languages, and the same observational approaches were put forward by Givón (1981). However, with regard to the many uses of specific indefinite articles such as in English, it could be questioned why an indefinite article is even actually used, when it only renders ambiguous the referentiality of the NP in question to the hearer, as in the English translation of (180): (181) Wilma intends to marry a Swede. The specificity often needs to be disambiguated by supplementary information, e.g.: (182) I was introduced to him at Wimbledon last year. One might then ask, if the speaker needs always to clarify the identity of the inde‑ finite referent, why bother using the indefinite article in the first place? Under a Gricean pragmatic account, the speaker would be accused of being under‑­ informative. However, if the speaker is adhering to the Quantity 1 maxim (Grice 1975), and supplying as much information as known, then the information in (182) should not be required. But there is a reason for using an indefinite specific, and that is if the speaker may not intend to disclose the identity of the referent, in other words, it serves as a means of withholding information, and flouting the Quantity 2 maxim for whatever reason (e.g., in (181) the Swede may be someone unmentionable, or simply have an unpronounceable name, or the speaker may wish to focus on his nationality rather than his identity). It could be argued, then, that the use of specific indefinites may be somewhat restricted in today’s English,



7.4 Bare Noun Constructions in creole systems 

 187

a fossil of a historically older system that must have once held more salience in discourse than it does in the present day. In such cases, then, the noun reference system of SCE is unlike any European language in the fact that it actually distin‑ guishes specific from non‑­specific nouns, the latter being represented in the bare form as noted above, singular or plural.

7.4 Bare Noun Constructions in creole systems In discussing specificity distinctions in nominals, it should also be noted that only count nouns are eligible for classification as specific indefinites (at least in English); that bare mass nouns are inherently non‑­specific, e.g., in (183) There is water on the floor water is non‑­specific. However, this is not because it is not known to the speaker, but because of its non‑­individuable characteristics. In order to make it specific, a determiner such as some might be used, though even then, the specificity is usually only present when there is contrastive stress placed on some. This raises the question of individuation as a necessary ingredient for determining specifi‑ city, and for this reason it is also useful to study the evidence from pidgins and creoles, which often have a rigorous system for marking individuation in nomi‑ nals, without necessarily an accompanying indefinite article system. Such pat‑ terns are illustrated in Mufwene (1981) and more recently, Stewart (2011), in the case of Jamaican Creole, where non‑­individuation may apply to either count or mass nouns, as shown in the following examples (Mufwene 1981: 229–230): (184) Buk de aal uova di tiebl ina im afis ‘There are books all over the table in his office.’ In (184) the absence of plural marking on buk indicates that non‑­specific plural count nouns in Jamaican Creole are considered as non‑­individuable, in the same way as mass nouns: (185) Wata de aal ouva di tiebl ina im afis ‘There is water all over the table in his office.’ In order to express singular individuation, either the determiner di or wan may be used, depending on whether the speaker wishes to express definite or indefinite individuation:

188 

 Chapter 7 Bare noun constructions

(186) Di/wan buk (de) pan di tiebl ina im afis  ‘The book is on the table in his office.’  ‘There is a book on the table in his office.’ Definite plurals may also be individuated using the plural marker dem: (187) Di buk dem de aal uova di tiebl ina im afis   ‘The books are all over the table in his office’ These examples do not necessarily suggest that there is no article system in Jamaican Creole, only that there is an implicit system for marking specific vs. non‑­specific distinctions in indefinite reference that aligns the formal expression of non‑­specific plural forms with mass nouns, due to their absence of potentiality for individuation. Mufwene, however, did not mention the likelihood of specificity distinctions, though these seem to emerge as a natural consequence of the need to distinguish individuation in the language, suggesting, further, that specificity overlaps with individuation in nouns. Patrick (2004) notes that the specificity dis‑ tinction in Jamaican Creole is linked with tense and time reference, so that past temporal expressions all have specific referents (Leiss 2008 has observed similar correlations between perfective aspect and definiteness in Gothic and Old High German). Patrick further states that bare NPs are always generics though some bare NPs in the scope of past time reference can be specific. Similar evidence is supplied in Stewart (2011), who supplies one example of a singular non‑­specific (which she refers to as a singular generic), left unmarked for number, just as for non‑­specific plurals (2011: 373): (188) Waya a paizn wire BE poison ‘Wire is a poisonous thing’.

ting thing

The single example is taken from Bailey (1966: 27).75 Additional examples are found in Hawaiian Creole, as Janson (1984) discus‑ ses, suggesting that the numeral one (wan) was introduced in the functional role of a new indefinite article, but did not replace the indefinite article with non‑­ specific nouns because of the inappropriateness of using a numeral in cases where no specific referent can be counted (1984: 297). Again, we see an interes‑ ting overlap between specificity and individuation: specific indefinites being

75 Patrick (2004) notes that in equatives like (188), the copula is a, a reduced form of the older form da, and that this varies with is/was in the mesolect. He also notes that the zero copula does occur but that Rickford (1996) shows that the copula has a more than 80% frequency of ­occurrence. However, he cites Bailey (1966) as mentioning that it is omitted in predicative NPs.



7.4 Bare Noun Constructions in creole systems 

 189

marked by wan and non‑­specifics aligning with bare mass nouns in having zero‑­ marking, even if they are count nouns. Janson, however, finds varying patterns occurring in other creoles, and uses the data from Seychelles French Creole and Papiamentu to argue against the monogenesis hypothesis of Bickerton (1981) that a specific‑­non‑­specific distinction is universal across creoles. Instead, he proposed that there cannot be a single, universal explanation to the presence or absence of articles in creoles, as indeed there are many non‑­creole languages which do not have an article system (such as Russian). In one example, he pro‑ vides evidence from Cameroon English Pidgin (from Todd 1979) to suggest that the language does not have any marker for indefinite nouns, whether specific or non‑­specific. However, he notes that, sometimes a determiner sohm is preposed in front of a noun (Janson 1984: 312): (189) trohki si sohm bush dia ‘Tortoise saw a wild deer’ (Todd 1979: 44) This item is not explained further, though it does appear as an optional nominal determiner under his listing of indefinite singular nouns (1984: 313). Janson conti‑ nues to insist that the specific/non‑­specific distinction is not encoded; however, the paradigm is almost identical to Mufwene’s (1981) noun classifications in Jamaican Creole: indefinites (singular or plural) are left unmarked except for an optional marker in the singular, in this case, sohm, and definites are always marked with a definite article di. Although Janson’s work pre‑­empts grammaticalisation studies to some extent, it does not ignore historical universals, and he is aware of the development of inde‑ finite articles in Indo‑­European languages from numeral sources meaning ‘one’, and definite articles from demonstrative sources meaning ‘this’ or ‘that’. It is only later that we see that Bickerton’s early (1981) monogenesis hypothesis might have survived better had it been more closely structured in terms of a compa‑ rative, historical appeal to grammaticalisation universals. For the same reason, Janson might have provided a more convincing argument if he had investigated the sources of the indefinite article in Old English, and found that a form meaning ‘one’ had co‑­existed alongside a form sum (= ‘a certain’ – see Quirk and Wrenn 1957) both of which had a function of marking specificity in the discourse (non‑­ specific nouns were left bare). Thus, the use of the optional determiner sohm in Cameroons English Pidgin in (189) could represent a specific indefinite that is reminiscent of one used in the early historical development of the English lan‑ guage itself. If these were the sources of the now grammaticalised determiner forms in present‑­day English, it would not be surprising to see the same pat‑ terns of development emerging in a new language like a creole. Universals do have a strategy, and all other things being equal, in the redevelopment of a new

190 

 Chapter 7 Bare noun constructions

language, it would not be unexpected that the same grammaticalisation sources should be found, neutral to time differences, since grammaticalisation strate‑ gies are universal in diachrony as well as across languages (see, e.g., Heine and Kuteva 2002; Ziegeler 2014a). Similar patterns of the salience of specific/non‑­specific reference to plural marking are summarised in Tagliamonte, Poplack & Eze (1997) for Nigerian Pidgin English, in which quantified NPs may occur with plural marking, e.g., (1997: 110): (190) A dey sɪks mɔns wɪt am fɔ Englan. (01/13) ‘I was with them for six months in England.’ Such examples may be contrasted with the bare generic plural in the following (1997: 112): (191) Na de wey got de slip. (09/955) ‘That’s where goats sleep.’ Tagliamonte et al. note a general tendency for the zero‑­plural to be conditioned by two factors: animacy and nominal reference, with animacy cross‑­cutting referentiality, so that inanimate nouns which are bare are interpreted as generic and non‑­specific, but bare human nouns are construed as singular unless marked overtly for plural. Nevertheless, the factor of genericity scores higher on the probability scale for bare NPs (0.57) than that of inanimacy (0.54). Such differences are marginal and it should be considered that both factors interact in some ways. Singler (1989) and (1991) also notes the same constraints found in Liberian Settler English, generic NPs scoring lower for plural marking (0. 41) than non‑­generic (0.59). Singler establishes the factor as a general creole characteristic, something which Tagliamonte et al. (1997) disagree with, suggesting African substratum influ‑ ences from languages such as Igbo to be of more importance. Poplack, Tagliamonte & Eze (2000: 95), furthermore, suggest a syntactic motivation, governed by the pre‑ sence or absence of pre‑­modification. For Nigerian Pidgin, on the other hand, Tag‑ liamonte et al. (1997: 114) suggest a ‘counter‑­functional’ motivation, i.e., the plural marker is used in environments in which it would be noticed if absent, i.e., it is a marked feature, used for referring plural NPs which have more salience in the discourse (i.e., it may be used redundantly with pre‑­modifying quantifiers) What appears to be operating in Singapore Colloquial English, according to Ho and Platt (1993), is the same ‘counter‑­functional’ tendency for marked referring NPs to be assigned plural marking, while unmarked, non‑­referring NPs are more frequently observed to appear bare. Thus, in the use of the BNC, a distinct semantic feature is also grammaticalised, and that is the referential status of the NP.



7.5 Further examples of the Bare Noun Construction in SCE 

 191

7.5 Further examples of the Bare Noun Construction in SCE 7.5.1 Zero‑­plural BNCs The examples used in Ziegeler (2003) were taken from a number of informal samples of advertising literature, haphazardly collected at intervals as diary notes over the previous ten years. In this way, the data used by no means reflects a formal, quantitative sociolinguistic approach. However, the pervasiveness of the feature under observation through the sampling period in particular non‑­ referring environments was obvious, and was consistent with the tendencies dis‑ cussed in previous studies by Ho and Platt (1993) and Platt (1991). One of the most interesting examples came from a flyer for a bargain sale in the Singapore Grand Hyatt Hotel, December, 2001:

Fig. 5: Advertising flyer illustrating the variable feature of zero‑­plural BNCs in Singaporean (commercial) English.

It can be seen that the clothing items mentioned are variably marked for plural, as follows: (192) jeans T‑­shirt shirt

192 

 Chapter 7 Bare noun constructions jacket belt wallet bags and many more item.

Interestingly, under the list was written: (193) Must buy. All items $9.00 onward. In (193), the word item is marked for plural, where it is not in (192). This is in accord with findings so far by Ho and Platt (1993) and for the creole situa‑ tions mentioned above that the use of certain determiners (here, All) shows a higher incidence of plural marking than for a bare noun phrase. The difference between All items and many more item is in the fact that All in this case refers to a conceptually‑­delimited set of items, while more is quantitatively indefinite. Why bags receives plural marking, but not the other items (apart from jeans which is inherently plural), remains unresolved; perhaps for the writer of the flyer, the use of the zero-plural BNC was simply not categorical in such environments. Another sample came from a business card collected during the same period from a hairdressing and beauty salon near Bugis Street, in which the following examples of the zero‑­plural BNC included: (194) Post‑­Natal Stretch Mark; Eyebrow and Eyeliner Tattoo; Hot and Cold Mask; Pimple; Acne Scar; Freckle; Black Head. However, amongst those NPs marked for plural included the following items: (195) Personal Make‑­up Courses; and Personal Skin Treatment Courses. The reason for (195) to receive plural marking may be again that the courses available represent a delimited set, while tattoos and freckles are non‑­specific and unquantifable. Other items have been observed in advertising signs displayed inside stores and outside shop fronts, but more recent advertising literature, collected over the period May – August, 2002, includes the following (emphasis cited): (196) Applicant must be Singapore Citizens or Permanent Residents and be between 21 and 65 years old to qualify. (Robinsons Loyalty Card Application Form) and from the Jiamei Roast Duck and Catering Service: (197) Rental of wine glass and porcelain plate also available. (198) Fried Egg Noodle; Hong Kong Style Fried Noodle; Butter Prawn;

7.5 Further examples of the Bare Noun Construction in SCE 



 193

However, (199) Nonya Otak in Banana Leaves and (200) Mini Custard Puffs appeared on the same menu. It would appear that the variation might be statis‑ tically governed, and not an obligatory or categorical rule in every case and for every speaker, or perhaps that noodle is less determinate a quantity than a fixed specific number of custard puffs. The same variability is found in the following, from a public notice on the introduction of a local transport card: (201) Only applicable to Singapore Citizens, Permanent Residents, and holders of Employment Pass, Work Permit, and Student Pass in which the NPs referring to humans are preferred to have plural marking over those referring to inanimate objects. The contrast between plural marking for human referents and zero‑­marking for inanimates accords well with the findings noted above for Nigerian Pidgin English, suggesting that there might be a similar tendency in some cases. Other items include: (202) 10% off all hair‑­do only (Hair Studio Promotions card) in which hair‑­do is unmarked for plural, possibly on analogy with Cut, Wash & Blow, both appearing on the same card. Incidentally, Student (with Pass) $8/–­ also appears, next to Senior Citizens $8/ –­, in which the plural is marked. Another item was found in a Singtel advertising voucher: (203) Offer valid for new female customer only. and (204) Chan Brothers Travel – China Tour in which Tour is unmarked for the generic plural. Other examples included the following, from an internet web‑­site for job‑­seekers: (205) Job Opportunities Part Time Staffs a. Computer Knowledge b. Willing to work on Sunday and Public Holiday c. Short Term Basis

194 

 Chapter 7 Bare noun constructions

In each example, the occurrence of a bare noun construction is always ­associated with a plural generic or non‑­specific NP (Staffs is an overgeneralisation of plural marking to collective NPs). Although this tendency is not categorically represented, it does signify the existence of a counter‑­functional motivation in that when plural marking is found, it is consistently found with nouns which are unquantified. 7.5.2 More recent data The data previously investigated in Ziegeler (2003) found that the use of the bare plural in SCE was most frequent with count nouns that were non‑­referring or generic in nature. The use of the bare count noun is clearly not categorical, and appears to be more frequent in the absence of any nominal pre‑­modifiers (in (196) it clearly functions to mark as non‑­specific the noun, applicant, which would be more likely to carry referentiality in any subject position). Many examples appear on public notices, for example, a traffic sign, observed close to a shopping centre in Bukit Batok, read: (206) BEWARE OF PEDESTRIAN as shown in Figure. 6:

Fig. 6: Singapore traffic warning sign (April, 2008)

In (206), in fact, the opposite interpretation may be produced with regard to referentiality (in other dialects) – the pedestrian becomes understood as a single, specific one, rather than an indefinite quantity of non‑­specific ones.76

76 Similar interpretations can be associated with traffic signs such as Beware of Vehicle observed next to a car park exit in Coleman Street, Singapore (observed January 2009, and earlier).



7.5 Further examples of the Bare Noun Construction in SCE 

 195

Another traffic sign, observed in Chinatown, Singapore, during March 2009, read: (207) PEDESTRIAN THIS WAY The reference in both cases is not to a single (dangerous) pedestrian but to an unquantified number of (benign!) pedestrians who may from time to time need to cross the street. Other warnings included the following, observed on a wall in Tan Tock Seng Hospital, April 2009: (208) DANGER! LOOK OUT FOR ONCOMING TROLLEY! Again, the sign was intended to refer to the danger of an unquantifiable number of oncoming trolleys, not to a single instance of one, un‑­named but specific runaway trolley, the interpretation usually associated with bare nouns in sign‑­ post genres. However, the unquantified nature of the noun phrase renders the noun conceptually equivalent to a mass noun, in the local dialect, i.e., it takes no marking for plurality, and no determiners marking it as singular; according to Michaelis (2003), the absence of any marking for referentiality implies a mass noun status. Another interesting example from an email received (26 March 2009) read: (209) Electronic dictionary is not allowed [in exams] In (209) it is seen that the agreement of the verb is also expressed as singular, indicating that the absence of the plural marker carries with it the conceptual significance of the indefinite quantification of a mass noun, and that it was not simply a morphological omission or ‘error’. However, it differs from a mass noun in that the reference of an unquantified number of dictionaries is understood from the context, while with a mass noun interpretation, the reference is only to an unquantified substance, not to an unquantified number. At the same time, it could not be argued that electronic dictionary in (209) is referring (at least in non‑­ Singaporean dialects) to an unquantified substance of dictionary, and in (207), the form is motivated for use as part of the grammatical system. It will be ques‑ tioned below, though, whether such bare nouns in other dialects may be consi‑ dered as coercing mass noun status on count noun categories, and it is for this reason that the SCE data has been introduced as a point of comparison. Gil (2003) noted the tendency in SCE for count and mass nouns to be indis‑ tinguishable, at least morphologically; similar reference was made in Alsagoff and Ho (1998: 143), and Ho and Platt (1993), the latter account attributing the requirements of the language to mark specificity rather than definiteness on the noun phrase. Such earlier studies did not attempt to provide an in‑­depth analysis of the functional constraints on bare plurals, until Ziegeler (2003) observed their

196 

 Chapter 7 Bare noun constructions

frequency in commercial Singaporean English as noted above, where the NPs appear in isolation and decontextualised, on nominal expressions referring to a generic, rather than a specific type, and not always as verb arguments. At the same time, however, non‑­specific or non‑­referential singular count nouns may also be left unmarked by the indefinite article appearing as verb arguments, as in the following examples, taken from various Singaporean internet blog spots in the Flowerpod Corpus, the first from a discussion on the health benefits of honey: (210) a. I use plastic spoon to scoop my honey & put my honey jar in cool dry place. b. Yesterday i went for foot and back massage @ Suntec n [and] bought one set of this honey. http://www.flowerpod.com.sg/forums/index.php?showtopic=8627, 4/8/05 Another (amusing) contribution came from a posting on the subject of superstitions: (211) And my bf [boyfriend] was thinking of booking hotel room to boink boink http://www.flowerpod.com.sg/forums/index.php?act=ST&f=2&t= 261931/8/08 The following example illustrates a typical BNC used in SCE, where the means of transport is not marked for referentiality: (212) hen, I’d ride very fast with my bike to school (usually take bus if I’m not so late). http://www.sgclub.com/singapore/things_about_sg_35277.html The BNC (212) could be a non‑­specific definite or indefinite, it is not clear from the context. What is clear though is that the absence of reference in SCE has func‑ tional salience in the constructicon, and is frequently marked by the absence of a determiner on singular nouns or the absence of plural marking on plural count nouns. In all cases in SCE, then, the form of a bare noun carries the function of non‑­specific reference, as in Figure 7:

FORM

bare NP ↕

FUNCTION

↕

non-specific noun

Fig. 7: Form‑­function poles of the Bare Noun Construction in Singapore Colloquial English



7.5 Further examples of the Bare Noun Construction in SCE 

 197

It should be noted that such a pairing of form and function is not exclusive to SCE nominals and similar pairings may be associated with mass nouns in any dialect, as discussed in Chapter 8. Note also that the meaning of the construction type has not been included in the diagram; the reason for this is that as discussed in Chapter 3, meaning is a highly illusory component of construction composi‑ tion, and should not be confused with function in the first instance. The need for a more schematic approach to meaning is raised again in Chapter 8. 7.5.3 Specific markers in SCE It is clear that in SCE the bare singular count noun is not categorical by any means, and is found with varying frequency across varying numbers of speakers (needless to say, it is associated frequently with colloquial, spoken usage, though the nature of the online internet forums presents a medium which is semi‑­informal, interme‑ diate between speech and writing (Claridge 2007)). The items outlined in all of the data above are understood by the speakers of SCE to be referentially non‑­specific, but it may be questioned, then, how SCE marks nominals which are specific. It was observed as long ago as Platt, Weber and Ho (1984) that Singapore English marks specificity with the use of the numeral one, exactly as in Hawaiian Creole, and Jamaican Creole discussed above; (210b) clearly illustrates one example of the usage, in which the speaker is referring to a specific set of honey products. The use of one for this purpose, however, is becoming relatively infrequent, as a recent study shows (Ziegeler, to appear), occurring not much more frequently than once per 10,000 words in both the International Corpus of English‑­Singapore and the International Corpus of English‑­India, though it is often difficult to discern from written texts whether the usage is simply contrastive (enumerating a single item rather than several) or specific (in which case spoken usage would not carry any contrastive stress). The example in (210) does not appear to reflect numerical cont‑ rast; other examples can be found in both the Singapore and the India sub‑­sections of International Corpus of English (International Corpus of English‑­SIN and Interna‑ tional Corpus of English‑­IND, respectively): (213) But Kallang got one Fun world or what lah ‘But there is a Fun World at Kallang!’77

77 What and lah are discourse particles used in SCE; the origin of what is indeterminate, though lah is a frequently‑­used discourse particle expressing emphasis or certainty, borrowed from either Malay or Hokkien (see, e.g., Lim 2007 for more details).

198 

 Chapter 7 Bare noun constructions

(214) do this blur you know that day one one lecturer came and ask uh uh ask me to help him lah so I went then78 (215) And I know one professor uh ... in Selam who was our professor also (216) Like uh District Industries Corporation ... industries ... there is uh one corpo‑ ration called uh Karnataka State Financial Corporation In examples (213) and (216), the use of the specific determiner is constrained to presentative functions, got being the verb often used in SCE to mean ‘there is/are’, influenced by the Chinese contact languages, in which a possessive verb is used to express existence, e.g., yoǔ ‘have’ in Mandarin; (see also Heine (1997) for more crosslinguistic examples). The presentative use of one in Indian English, another new variety of contact English, is also believed to be related to substrate factors (Sharma 2005).79 The variability with which the distinctions of specific‑­ non‑­specific reference are marked in SCE is evident also, as seen in the following example, in which the indefinite article is used with specific reference as in SSE: (217) I got them from a shop in Bras Basah Complex. (Flowerpod Corpus, Hardwarezone file. Posted by Lehnsherr 04‑­10‑­2008) It is obviously not possible to determine the frequency of use of specific one for any particular speaker, based on the the corpus examples alone. Heine and Kuteva (2005: 134) referred to the marking of NPs in this way as transnumeral marking, replicating the patterns from Chinese contact languages. However, as seen above, similar patterns appear across other varieties of English. In fact in Indian English data discussed by Sharma (2005) even in the case of spe‑ cific reference the NP is unmarked with the indefinite article (she does not show data for plurals), although in all cases except for one speaker surveyed, the fre‑ quency for marking plural on specific NPs as against non‑­specific was lower. The other noticeable aspect of Sharma’s study was that she quantified the non‑­use of articles according to position in the sentence and sentence function (theme, rheme, etc.). This method was not felt necessary with the present data, since many of the bare plural NPs were found in commercial usage in Ziegeler (2003) in

78 Blur = ‘stupid’ in SCE. 79 Sharma (2005) also notes the absence of the non‑­specific indefinite article in Indian English, a factor that she attributes to substratum features, e.g., Then he thought, what about getting Ø girl to marry from India? (2005: 545).



7.5 Further examples of the Bare Noun Construction in SCE 

 199

listings of isolated generic items, and were not found to be significant to Sharma’s findings. The use of one for the indefinite article with specific indefinite use was not frequent in the Indian English sample, but nevertheless this is a shared feature with Singaporean English. In terms of universal grammaticalisation strategies, it is not unusual to find in many of the world’s languages indefinite determiners derived from forms expressing the numeral ‘one’, and so it is difficult to deter‑ mine whether the specific indefinite marker one in SCE and in Indian English is the result of substrate transfer or universal grammaticalisation strategies. Heine and Kuteva (2002: 220–1) list Albanian, Turkish, German, French, Ewe, Moré, Hun‑ garian, Lezgian, Tamil, and Easter Island as languages in which this grammatica‑ lisation has taken place. However, in their (2002) data, the numeral marks both non‑­specific indefinites as well as specific, just as the English indefinite article does today. In Heine (1997: 72–4), a five stage‑­sequence of changes is discussed as involved in the grammaticalisation of the indefinite article from the numeral ‘one’ crosslinguistically. The stages are presented thus (1997: 72–4): Stage 1: The numeral, in which specific indefinite reference may be left unmarked. Stage 2: The presentative marker, introducing a new participant into the discourse. Stage 3: The specific marker, when the form develops a discourse function in marking any nominal participant that is known to the speaker but presu‑ mably not to the addressee (this means that in many cases non‑­specific reference is marked in opposition by bare nominals. Stage 4: The non‑­specific marker, no longer restricted to marking specific refe‑ rence, but used whenever an indefinite singular nominal is referred to. (At this stage, it may be said that the indefinite article marks not only actual reference, but potential reference.) Stage 5: The generalised article, not restricted to determining singular count nouns but possible with plural and mass nouns also e.g., Spanish: unos hombres (one Masculine‑­Plural men: ‘some men’). It may be estimated that the SCE specifc marker one is roughly at Stage 3 of this universal sequence, while the standard sub‑­variety spoken alongside SCE, SSE, will be at Stage 4. The co‑­existence of the two sub‑­varieties may result in some variation where such marking is concerned, but it was noted in Ziegeler (to appear), that the frequency of one versus the indefinite article was not very high, as seen in the following table:80

80 Since this was only a restricted portion of the corpus, the restricted numbers of contributors to the particular discussion from which it was taken also has to be taken into account (32 contri‑ butors, many of whom have repeat postings).

200 

 Chapter 7 Bare noun constructions

Tab. 4: Probability of the occurrence of the specific determiner one relative to indefinite article use in a restricted portion of the Flowerpod Corpus (the Pen Club thread). Specific one

Specific indefinite article

Total

19 = 11.17%

151 = 88.82%

170

The presence of bare non‑­specific count nouns is clearly an option in such contact dialects, which is governed to a certain extent by the presence or absence of determiners or quantification. However, it may appear to carry a coercive effect to speakers of dialects for whom the bare noun form is immediately construable as a mass noun (see section 7.7). Thus, even though pedestrian in (206–207) refers to a potentially countable number of referents, it may be construed by speakers of other dialects as a mass noun (or potentially uncountable), if the absence of any determiners or plural inflection carries the salience of a distinct nominal category assignment. For speakers of SCE, such nominals share with mass nouns the common characteristic of under‑­specification of quantity and thus non‑­referentiality; however, speakers of other dialects may find that, along with the meanings of non‑­referentiality, the additional semantic inference of non‑­ individuation is also implicated, which is in conflict with the truth specifications of the count noun semantics; e.g., the non‑­referential noun pedestrians is only unquantified, it does not necessarily indicate non‑­individuated reference.

7.6 Number marking and the Chinese substrate From the data above from Singapore Colloquial English, it could, of course, be argued that SCE reflects the number and countability marking strategies of its sub‑ strate languages, the majority of which represent a dialect of southern Chinese. Gil (2003) discusses correlations between number and definiteness, and citing Nichols (1992), notes that articles and number marking are a predominant lingu‑ istic feature of European languages, but are generally absent in the languages of East Asia and Africa; hence, the contact situation of English in Singapore is one of extreme contrast of Indo‑­European categories with those of Southeast Asian lan‑ guages as well. As regards Singapore Colloquial English, which he labels by the familiar term ‘Singlish’, Gil discusses an animacy‑­definiteness hierarchy which governs the patterns of number marking in many other languages: (218) 1st proform; 2nd proform; 3rd proform; human; animate; all count



7.6 Number marking and the Chinese substrate 

 201

In such a hierarchy, number is more likely to be marked the higher up the hier‑ archy one goes, Singlish reflecting the same patterns of Mandarin Chinese and Singapore Malay, where number marking is obligatory for all human personal pronouns and for inanimates as long as they are determined. It is for this reason also that plural inanimate nouns in Singlish are often anaphorically pronomina‑ lised in the singular, rather than the plural, as in the following example from a Singapore internet site: (219) Re: what is on your mind rite now ... my work as wondering when the stupid auditor will courier me the documents as need it urgently http://www.sgclub.com/singapore/mind_rite_now_20325.html [Ican 12‑­07‑­2007 10:02 AM] The speaker was referring anaphorically to the documents as there was no other noun in the discourse that could have provided the co‑­referential source. Crossing this hierarchy is the definiteness hierarchy: determined; all count, which suggests that the higher one goes up the scale of animacy, the more likely a noun is to be marked for definiteness. Number and definiteness are therefore seen as going hand‑­in‑­hand in such languages, the likelihood of number marking increasing the more definite the noun phrase is, and this increases, in turn, with the animacy hierarchy, a first‑­person nominal being perceived as more requiring of number marking than a third person nominal, since it is definite by default. Gil sees the notion of individuation as being the underlying conceptual justification for the intersection of the hierarchies. According to Gil, the absence of plural marking on nominals in the Chinese languages that formed the substratum in Singapore is due to the singular interpre‑ tations of the classifiers with which the nouns must co‑­occur. However, the subst‑ ratum effects need not be the reason that some nouns in Singlish are unmarked for plurality. In pidgin and creole languages, it has been observed that there are com‑ peting factors underlying the marking or non‑­marking of nouns for number, as we have seen in section 7.4, e.g., in Nigerian Pidgin English (Tagliamonte, Poplack and Eze (1997)), where nouns marked for quantity receive plural marking, while in others, quantified nouns do not receive plural marking, e.g., Jamaican English (Mufwene 1981). As noted above, in the case of Singlish, it is more likely to be found on nouns that co‑­occur with some form of pre‑­modifying quantifier, as in (193): All items, though it cannot be said that this is entirely rule‑­governed. What is interesting, though, is that Gil’s interpretation of an intersecting hierarchy of animacy and definiteness may also be applicable in the case of (193) and (194), All items being higher in definiteness than the generic reference to the list of items appearing in the same context. It could be claimed that the grammaticalisation

202 

 Chapter 7 Bare noun constructions

of plural marking on nouns in Singlish is vaguely emerging, then, as a form of agreement (with co‑­occurring nominal pre‑­modifiers). It was clear from Ziegeler (2003), though, that the absence of plural marking in generic lists of items appea‑ ring within a commercial genre in Singlish was the setting in which it was found to be most predominant. In Mandarin Chinese, one of the principle contact languages in Singapore today, we find that definiteness/indefiniteness distinctions are less significant, or less grammaticalised, than what Li and Thompson (1981[1989]: 126) term refe‑ rential distinctions. Givón (1981) had used the term referential (as noted above) in the same way as specific is used, perhaps more frequently, to refer to the speaker’s knowledge of the existence of a referent. However, Li and Thompson’s definition of referentiality does not necessarily encompass the criterion of specificity or identification, since it is used to refer to the reference of an entity, whether real or imagined, singular or plural. They do add, though, that referential expressi‑ ons are used to name some entity, and examples of non‑­referential noun phrases seem to coincide readily with the usual interpretations of non‑­specificity else‑ where, e.g., (Li and Thompson 1981[1989]: 127); i.e., they are left bare, as in the example below: (220) Xìnměi shi gōngchéngshī Xinmei be engineer ‘Xinmei is an engineer’ They go on to list a number of other categories of nouns that may be classed as non‑­referential, which under most accounts would classify as generic (1989: 128–9): (221) wǒmen zhòng we grow ‘We grow peanuts’

huāshēng peanut

(222) māo xǐhuān cat like ‘Cats like to drink milk’

hē drink

niú nǎi cow milk

or non‑­specific, occurring in the scope of a negative (1989: 129): (223) wǒ méi yǒu I not exist ‘I don’t have a pencil’

qiānbǐ pencil



7.6 Number marking and the Chinese substrate 

 203

Li and Thompson suggest that referential nominals can be marked for ­definiteness or indefiniteness, a distinction which aligns loosely with what elsewhere is described as specificity distinctions: a definite noun referring to an entity which is known to the speaker, and an indefinite one referring to an entity which the speaker believes is unknown to the hearer. Thus noun phrases determined by demonstratives are always definite, but those preceded by numeral quantifiers are necessarily indefinite. However, definiteness is a distinction which is not fully grammaticalised in Chinese, as it is often the context which determines whether a noun phrase is definite or indefinite (Li and Thompson 1981[1989]: 131): (224) wǒ mǎi le shuǐguǒ I buy PFV fruit ‘I have bought the fruit/some fruit’

le CRS

The definite interpretation is applied only if the speaker and hearer are aware of the reference of the fruit. They also note that the demonstrative is beginning to be used in some cases as a definite article, and the numeral yí ‘one’ is starting to function in many cases as an indefinite article (1989: 132). However, the latter development being perhaps more pronounced in Cantonese (see Matthews and Yip 1994) and perhaps other southern dialects of China, where it is the classifier alone used without the unitary numeral that is gradually being reanalysed as an indefinite article (possibly a hyperanalysis in Croft’s 2000 terms). Li and Thompson’s analysis appears to reflect a description based on a com‑ parison with languages that have obligatory categories such as definite and inde‑ finite articles, as such it presents a particularly Eurocentric perspective on the grammar of Chinese. From such examples, though, it is clear that there exists at least a weak correlation between number and the marking of specificity distinc‑ tions in Chinese, and that while nouns preceded by a numeral quantifier may be indefinite/non‑­specific, at the same time they are unmarked for plurality them‑ selves, since the language is transnumeral. The bare NP in Chinese, then, may be specific or non‑­specific, though clearly when it does not occur with any form of modification at all it has a greater chance of being interpreted non‑­referentially. The up‑­shot of this is that plural marking as a form of hypothetical quantification over noun phrases with no reference does not exist in languages such as Singapore Colloquial English, a language in constant contact with Mandarin Chinese today.

204 

 Chapter 7 Bare noun constructions

7.7 The Bare Noun Construction and construction coercion In present‑­day SCE, the Bare Noun Construction appears unmarked for number, reflecting only partially the semantic requirements of the major substrate or adstrate languages, since clearly, the BNC in Mandarin Chinese and other d ­ ialects of Chinese is not restricted to non‑­referential nouns as it is in SCE. Similar phe‑ nomena in standard varieties have been explained by construction coercion and type‑­shifting (between count and mass nouns), as described in Chapter 3. Apart from the BNC, the other two construction types that are represented in the present case are the Indefinite Article Construction, and the Plural Construction. These two construction types are served by no more than a single morpheme (with its combined lexical input) in most cases of the English examples represented in those studies. It is maintained in Michaelis (2003: 268), that even grammatical formatives can function as constructions; e.g., a mono‑­phonemic suffix such as ‑­s (together with its allomorph ‑­es) in English can act as a head of a partially lexically‑­ filled construction‑­type. The grounds for attributing such complexities to these residual elements of historically‑­eroded (but still actively‑­functioning) morpho‑ logy are simply their ability to bear meaning and determine what co‑­occurs with them. However, in the examples below, it shall be seen that the meaning con‑ veyed in such items is less significant than a construction‑­based approach would otherwise predict. The hypothesis of coercion was discussed at length in Chapter 3 as the resolu‑ tion of syntactic‑­semantic mismatch in a construction, and numerous examples were discussed. The same definitions apply to nominal as to other types of coer‑ cion, so that if we are discussing, for example, the Plural Construction, the mor‑ pheme ‑­(e)s represents the syntax with which a lexical input filler, i.e., a count noun in English, must combine. In the case of mismatch, the lexical input would be representative of a noun category other than a count noun (or perhaps one with an irregular plural, such as sheep). This is one type of coercion; another typical example of coercion in nominals is one cited by Michaelis (2003: 261): (225) She had a beer in which the mass noun, beer, is the lexical input fused (ostensibly incompatibly) into the Indefinite Article Construction, the construction component which deter‑ mines what can co‑­occur with it. Or at least that is what it seems, at first glance. Such an explanation is seen from the point of view that the semantic‑­syntactic incongruity is the result of the misplacing of lexical items, rather than that the indefinite article is over‑­extending its function, as it grammaticalises to include more and more lexical combinations in its scope. The syntax is thus perceived as basic and prior, and may be seen under such an analysis to attract new lexical



7.7 The Bare Noun Construction and construction coercion  

 205

items with which it may be slightly incompatible (e.g., Michaelis, 2004: 22). However, it may be questioned why the lexical items are attracted to occur in inappropriate syntactic configurations, when they are incompatible with the syn‑ tactic environment in the first place. Similar examples to (225) have in fact been discussed in the literature simply as metonymies (e.g., the CONTAINED FOR CON‑ TAINER metonymy of Kövesces and Radden (1998: 58). However, the type‑­shifting from mass to count nouns is of less relevance to the present data as that from count to mass noun status. The incompatibility of the lexical with the syntactic environment is discussed by Michaelis (2003, 2004) as resolved by a principle in which the meaning of the syntax wins out over the lexical meaning of the item to be coerced into the envi‑ ronment. The principle was designed to account for a perspective on coercion as related to type‑­shifting in construction‑­based grammars; e.g., the mass noun ‘type’ of the lexical item, beer, in (225), has been shifted to allow co‑­occurrence with an indefinite article (the Indefinite Article Construction), the marker of a count noun type. This is known as Implicit type‑­shifting and is always linked to coercion effects; the shift is accounted for by the principle known as the Override Principle, in which a lexical item which is semantically incompatible with its syntactic or grammatical environment will always take on the semantics of the environment at the expense of its own type semantics (Michaelis 2003: 268). Thus, beer acquires count‑­noun status. However, when the construction is conventionally associa‑ ted with the lexical item, a process known as Explicit type‑­shifting is involved; e.g., the use of the Partitive Construction in a piece of bread, explicitly shifts the mass noun bread to be construed as a bounded entity because it is unified with a construction referring to a bounded entity. This type of mismatch situation is not typically associated with coercion as the association between the lexical item and the construction type is conventionalised. Traugott (2007) revealed that such conventionalisation can be shown to reflect diachronic developments of const‑ ruction types. In standard varieties of English, the process works either way, in the case of count versus mass nominals. Michaelis (2003) goes on to discuss count nouns which are coerced into mass noun syntax, i.e., they occur as Bare Noun Construc‑ tions, e.g., apple in the following (2003: 268): (226) You have apple on your shirt. According to the syntactic conditions described by Michaelis (2003: 269), the Verb Phrase Construction can be combined with a bare lexical noun object only if the noun is a mass noun. Therefore, the noun apple in (226), which is a count noun, must shift its type in order to combine directly with the Verb Phrase construction, so that it receives a mass interpretation. The assignment of the interpretation,

206 

 Chapter 7 Bare noun constructions

then, is held to be due to the combinatory potential of the construction in which it is located. Such instances may therefore omit the article indicating individuation and be interpreted in the same way as mass nouns. It appears that the default nominal structure for count nouns must always contain an article or marking for plural, and therefore, when a count noun appears without such marking, it con‑ stitutes a special interpretation in which the count noun is perceived as a mass noun. However, how this interpretation is derived is a matter for consideration. For Michaelis (2003, 2004), the default form comes within a construction, and it is the construction meaning that determines, to a greater or lesser degree, what lexical items can slot into it. That is, the construction has a semantics of its own, built out of the entrenchment of regularly and frequently‑­occurring lexical items, which have, in turn, lent their own semantics to the unified form. The semantic traces of the most frequently‑­occurring lexical items found over the life‑­history of the construction leave an indelible impression which governs what may predicta‑ bly be found in such slots in any future construction of the same syntactic pattern (in much the same way as persistence effects in grammaticalisation (Hopper 1991), mentioned in Chapter 3). This is why, for such accounts, it is the new lexical item that must be accommodated to the syntax of the construction, not the other way round. Thus, for proponents of such top‑­down theories of syntax, the notion of forcing an intrusive lexical item into an incompatible syntactic arrangement makes sense; the syntax is already pre‑­determined and immutable, and the cons‑ truction semantics are affected, recursively, by the lexical environments in which they most frequently are found, as was noted in Chapter 3. However, as seen from the examples provided so far it is obvious that coer‑ cion does not result in ungrammaticality, or in any overt violations of selection restrictions. If the examples represent anything infelicitous or unacceptable at all, it is not always immediately apparent. A beer is perfectly understood in its context of a measure of the liquid to which it refers, while bare apple is equally understood to refer simply to the substance of apple. Because of this, the selec‑ tion of examples would therefore seem to suggest that some kind of conventio‑ nalisation is taking place. Michaelis (2003, 2004) has stated that a Verb Phrase Construction cannot normally co‑­occur with a bare noun phrase object argument because of the requirement that all ‘sisters’ of the head verb must be maximal categories, i.e., in the case of a noun, it must have reference (be quantified or at least anaphorically linked to a referring noun) (2003: 269). If it does not have maximality, it will automatically denote a mass noun type. Thus, the unquanti‑ fied amount of apple on the hearer’s shirt in (226) enables the default sense of a mass noun, and the same may be said of the noun apple when it occurs in a Subject‑­Predicate construction (2003: 268):



7.8 Applying the coercion hypothesis to the contact data 

 207

(227) Apple dries well. It need not be argued, even at this early stage, that the use of a bare noun any‑ where, regardless of its construction frame, may be considered to be a mass noun, as for example in the noun CHEESE seen displayed in signs above the dairy shelves of any local supermarket, denuded of any co‑­occurring constructional alliance, in spite of the fact that a range of different kinds of cheese could be on sale. The examples from SCE of lists of commercial items also convey this impres‑ sion. However, it is still incontrovertible that the use of apple in (226–227) requires some sort of mental adjustment from its usual count noun status to that of a noun referring to a substance rather than a single entity, and that is at the basis of count‑­to‑­mass noun coercion, as Michaelis would like to describe it: it exhibits an example of implicit type‑­shifting, in which its categorical status has been altered (via the Override Principle so that the construction determines the interpretation of the lexical items contained within it) to implicitly correspond to the require‑ ments of the syntactic constructions in which it appears. If the coercion hypothe‑ sis is indeed a viable explanation for the appearance of unlicensed Bare Noun Constructions, it may be considered in relation to the present data from SCE, in which, as we have seen, bare NPs appear more frequently than in standard or non‑­contact varieties of English.

7.8 Applying the coercion hypothesis to the contact data What is problematic to the coercion hypothesis is that the appearance of bare nouns in dialects such as SCE may not necessarily be explained by a type‑­shifting account, even though they are count nouns which are unmarked for number and do appear as verb arguments. Such evidence appears to belie the need for a coer‑ cion account to explain the similarities. In the case of count nouns appearing in a bare form, unmodified by determiners or by plural marking, the unmarked form is used to mark the function of non‑­referentiality, or non‑­specificity of quantity or reference in the discourse. Unlike the examples provided in (226–227) above, users of Singaporean English do not ‘coerce’ a mass noun type from a count noun; the dis‑ tinctions of numeration remaining only secondary to the more salient distinctions of referentiality. In addition, as noted above, non‑­referential count nouns were always left unmarked for number by the indefinite article in the history of English (see, e.g., Quirk and Wrenn 1957), and it was only when the indefinite article began to grammaticalise to include non‑­referential nouns within its scope that such ­determiners began to modify nouns. This is not to deny that in the case of (226–227), the type‑­shifting combines the functional objective of marking non‑­referentiality

208 

 Chapter 7 Bare noun constructions

in the discourse, just as in the history of English, and the same absence of refe‑ rentiality is salient to the use in the Singaporean contact variety of the language. What is being questioned, then, is the fact that Michaelis’ analysis of such const‑ ructions restricts non‑­referentiality to mass nouns only, since in standard English varieties, a bare argument of a verb can only be a mass noun. However, as we have seen, the lack of referentiality in the subject of examples like (227) is due to the elliptical nature of the noun apple, standing for an unquantified mass of apple substance, while in examples such as (209), the lack of referentiality is due to the unquantified number of dictionaries the noun phrase stands for in the uni‑ verse of dictionaries. Lack of quantification is not the same as mass reference: whereas apple in (227) is conceived as containing reference to a plausible mass entity, most of the examples in (192–209) are not. Thus, the BNC in (206), for example, pedestrian, rather than apple in (227), should more likely be represented as a true count‑­to‑­mass conversion (if coercion resolves mismatch), while apple in (227) is merely a metonymical shift of reference transfer. Pedestrian in (206) is not a coercion in the dialect in which it is used, as it expresses a systematic feature of noun‑­marking or non‑­marking, as the case may be. What is important, though, is that the SCE data illustrate that the same surface representations may yield non‑­coercive interpretations. The construc‑ tions presented in the data exemplify typical Subject‑­Predicate constructions (e.g., 196, 209) or Verb Phrase constructions (e.g., 211, 212), and yet the criterion laid down by Michaelis (2003) that the arguments in such constructions must have reference or be mass nouns or coerced into mass nouns does not apply. It may be argued that since the dialect is in contact with languages which do not mark these distinctions, such contact influence will affect the interpretation of such nouns. They neither have reference nor mass noun status. However, SCE is also in continual contact with SSE, which does have count‑­mass distinctions like any other standard variety. So, instead we have an example of exactly the requi‑ red constructional apparatus for the production of a Count‑­to‑­Mass coercion (a bare non‑­mass nominal occurring as argument of a verb phrase), but it does not take place, since SCE does not have count and mass nominal distinctions. So, on the one hand, the coercion account cannot be used to explain SCE BNCs, and on the other, the coercion account of such construction phenomena, if based on surface‑­syntactic generalisations, cannot be supported with independent evi‑ dence. We may then ask whether the similarity in surface syntactic generalisa‑ tions can yield a common construction‑­type across dialects. Furthermore, the influence of the Chinese language imposes a different world‑­view on the data from Singaporean English, since there is no commitment to number marking on every noun phrase, as discussed in 7.4. Nouns such as horse were perceived to be non‑­individuated and likened to the ‘stuff’ of mass



7.8 Applying the coercion hypothesis to the contact data 

 209

nouns in classical Chinese (Hansen 1983), and they are not considered quanti‑ fiable unless given reference in the discourse. Instead of determiners creating the categories of individuation and replicability, classifiers in Chinese are used to delimit quantities of noun referents and illustrate the existence of part‑­whole dichotomies in reference to entities and objects, rather than the member‑­class distinctions of English noun phrases. This means that in the contact language, Singapore Colloquial English, only specifically‑­referring nouns will be most fre‑ quently marked for replicability, in accordance with their relevance to the dis‑ course, the non‑­specific and generic nouns remaining conceptually equivalent to mass nouns in other dialects. However, conceptual equivalence does not equate to categorial equivalence and SCE remains as transnumeral as Chinese. But it is the domain of conceptual equivalence which is being exploited in the speakers’ search for interlingual identification across contact; this will be dis‑ cussed at greater length in Chapter 8. Szmrecsanyi and Kortmann (2008) refer to a number of Angloversals or uni‑ versals amongst English vernaculars, without specifically isolating features that are unique to contact varieties or new varieties. Amongst them are listed irre‑ gular use of articles, and absence of plural marking on measure nouns, though it would seem that the latter may be found most often with L1 vernaculars. The irregular use of articles is not discussed in detail. However, in the present data, it appears that it is not the use of the article that is at issue, but the deployment of a bare noun form to mark non‑­specificity, since the same form is shared with plural generics, which do not receive indefinite articles in any case. It may be argued that we are not looking at a variety of English at all, but another language.81 In the cases described though, the feature occurs in institutional usage (e.g., 207–209) above; these uses could be understood as representative of the standard variety, a variety said to differ little from other standard varieties of English (Gupta 1994), as noted in Ch. 2. The coercion effects, after all, are held to be a result of the input of incompatible lexical items in the syntax; the lexicon sup‑ plied in the case of contact varieties is from English, as with any other variety, and SCE is not the only variety of English spoken in Singapore, SSE is spoken alongside it. This is where the coercion account comes up against a number of problems: the syntax must determine the interpretation of the lexical items slotted into it, and yet the syntax of contact varieties is not the same as the syntax of a standard variety, albeit with the same lexicon. Nevertheless, the construction inventory is identified by its surface syntactic features (e.g., Verb Phrase constructions), and a

81 As was suggested by an anonymous referee of Constructions and Frames.

210 

 Chapter 7 Bare noun constructions

descriptive approach based on construction‑­merging may be more useful in pro‑ viding a complete theoretical picture (see Chapter 8). As we saw in Chapter 3, unacceptable lexical‑­constructional combinations are rejected on the grounds that they are not extendable either metonymically or metaphorically from the Lexical Prototype Constructions. Thus, even distant conceptual domains in metaphor must be construed to be parallel in some way to their sources, and however distinctly the notion of coercion may be defined in terms of its cross‑­categorial applications, it still cannot be isolated from an asso‑ ciated metonymic or a metaphorical construal: (226–227) may be conceived as metonymically linked to their targets in the lexicon, as noted earlier, for example, by the OBJECT FOR MATERIAL CONSTITUTING THAT OBJECT metonymy of Kövesces and Radden (1998: 31) in which substances are conceived of as parts making up wholes, and the crossing of categorial boundaries is not at issue. It just so happens that the OBJECT source of the metonymy (apple) and the target (apple MATERIAL) are classified as belonging to different noun types. Thus, in lexical terms there is a metonymy (OBJECT FOR MATERIAL CONSTITUTING THAT OBJECT) but in grammatical terms there is a metaphor (A COUNT NOUN IS A MASS NOUN), created by the lexical metonymy. However, the subject in items such as (209), for example, cannot be construed as standing for arbitrary dictio‑ nary material – there is no object‑­for‑­substance lexical metonymy to access. In the same way, the Singaporean examples co‑­occur as bare arguments of the verb phrase, but are neither referential nor maximal, as is required by Michaelis (2003) for the co‑­occurrence of bare nominals in the Verb Phrase Construction. Even for non‑­speakers of Singaporean English, there may be no coercion because there is no plausible link, either metonymical or metaphorical, with the semantics of the source constructions in which bare nouns may be understood as mass nouns. Therefore, even in a dialect in which there are count‑­mass distinctions, such as Standard Singapore English, a construction using pedestrian as a bare noun (e.g., !!You have pedestrian all over your street, you should not drive so fast), regardless of the somewhat macabre connotations, could not be coerced to mass noun status if the metonymical associations are not conventionalised. It cannot, therefore, be argued that the contact data are irrelevant because the dialect does not have count‑­mass distinctions in the first place; the lexical significance of the noun itself cannot be ignored. The grammaticalisation of the indefinite article and the presence of obligatory plural marking in English means that only inde‑ finite mass noun types may fill these slots, and the non‑­referentiality of bare mass noun types is still visible: they cannot occur (felicitously) as subjects of certain sentences, e.g., ??Apple is on your shirt/ ??Milk is on the floor. De Swart



7.8 Applying the coercion hypothesis to the contact data 

 211

and Zwarts (2009) discuss further the argument status of bare nouns as neces‑ sarily referential, at least in English, suggesting that only sentence arguments are constrained by referentiality. The reason that apple in (227) is still accepta‑ ble is because the predicate refers to a timeless, generic or characterising situa‑ tion, and as such the referentiality of apple is affected by the non‑­specificity of the context (there is no spatio‑­temporal grounding). In reference to a specific instance of apple, though, it is not possible to use a bare mass noun in the argu‑ ment slot usually reserved, at least historically, for the topic of the sentence, which must have previous reference in the discourse, i.e., be specific (see, e.g., Shibatani (1991) on the grammaticalisation of sentence subjects from former topics, as mentioned earlier). This illustrates that the bare nouns in (226–227) may be interpreted as non‑­referential (even today) and that there does exist a direct correlation between referentiality and bare mass noun status, even in non‑­contact varieties of English.

7.8.1 A grammatical metaphor We can see from such examples that the constraints on coercion are determined not only by metonymical assimilation to a prototype construction (as discussed in Chapter 3), but also by metaphorical assimilation to the more abstract count‑­ mass categorisations, i.e., A COUNT NOUN IS A MASS NOUN. In mass‑­to‑­count type‑­shifting, the lexical source of the grammaticalising indefinite article (from the numeral one) restricts its range of distribution at later stages of development, and retention in morphological grammaticalisation can explain an unresolved constructional mismatch, e.g.,*a dirt – to use Goldberg’s (2009b) example. But coercion, instead, defies such retention and resolves a mismatch which increa‑ ses the distributional range of a grammaticalising morpheme, e.g., the indefinite article in a beer (Michaelis’ 2003; 2004 example). Since that resolution is only enabled by metonymy (a beer standing for a measurable unit of beer – something which is not conventionalised with dirt to the same extent), then coercion in mass‑­to‑­count type‑­shifting is also seen as just a sabotaged metonymical process, and one which also often results in metaphorical construals of the constructions it occurs in. Thus, for the nominal coercions described above, Table 5 illustrates the con‑ stitutive role of metonymy and metaphor in deriving what is generally known as coercion.

212 

 Chapter 7 Bare noun constructions

Tab. 5: The components and constraints of nominal type‑­shifting. Type‑­shifting

Lexical metonymic source

Retrievable metonymic target

Constructional ­metaphor output

Mass‑­to‑­count

a beer

a (glass) of beer

A MASS NOUN IS A COUNT NOUN

Count‑­to‑­mass

apple

apple material

Mass‑­to‑­count Count‑­to‑­mass

*a dirt (sg.) *pedestrian (pl.)

a ?? of dirt ??pedestrian material

A COUNT NOUN IS A MASS NOUN (Not derived) (Not derived)

As can be seen in Table 5, where there is no metonymy conventionalised, the meta‑ phorical construal of the constructional analogy does not go through (note that although it is necessary to illustrate the links between the source and the target of the metonymy via ellipsis – the recoverable metonymic target is only a possi‑ ble ellipsis, not a necessary one). The only difference between lexical/predicate metonymy and the cases illustrated in the role of coercion is in the category shifts in grammatical types, a function not always associated with classic cases of metonymy. This is not to suggest that such differences distinguish coercion from metonymy, only to suggest that the metonymy involved is most often found abs‑ tractly associated with such grammatical category shifts. In the case of the SCE BNC pedestrian, it does not need mentioning that speakers of SCE are not refer‑ ring to the presence of any kind of pedestrian substance or material (unlike the more classic types of coercion, such as There was rat all over the road –­in Kövec‑ ses and Radden (1998: 51). The constraints illustrated in Table 5 work well for examples such as pedest‑ rian, dictionary, and applicant, etc. in the examples in (192–212). Where there are possible discrepancies, the resolution is not so easy to derive, as in some of the examples in (192–212), such as shirt, jacket, wallet, wine glass, noodle, prawn, etc., where the same noun is used to refer to a count noun as a mass noun conventio‑ nally associated with it, e.g.,:

7.9 Summary 

 213

Tab. 6: The components and constraints of count‑­to‑­mass type‑­shifting, using ambiguous examples in (192–212). Type‑­shifting

Lexical metonymic source

Retrievable metonymic target

Constructional metaphor output

Count‑­to‑­mass

shirt, jacket, wine glass, noodle, prawn

shirt, jacket, glass, noodle, prawn material

A COUNT NOUN IS A MASS NOUN

Count‑­to‑­mass

wallet

??wallet material

(??Not derived)

Count‑­to‑­mass

customer, tour, holiday, hotel room

**customer material, tour material, holiday, hotel room material

(Not derived)

As can be seen in Table 6, some examples from (192–212) may be automatically ambiguous for nominal categorisation in any case, when decontextualised, and may appear not to provide ideal examples on which to base the assumptions illus‑ trated above. In fact, Traugott (2007: 533–4) noted the fact that certain noun cate‑ gories referring to cooked foods, vegetables etc. may be systematically construed as count or mass nouns depending on their contexts, and thus refer equally to a bounded or unbounded quantity, so questioning the likelihood of type‑­shifting in such cases. This need not apply to the examples above, of course, which were advertised commercially as generic items on a list (one would not often find, for example, shirt or noodle being sold as unbounded masses, though their use in the partitive constructions such as Traugott describes may be more plausi‑ ble, e.g., a shred of shirt). At the same time, not all of the examples in (192–209) are ambiguous, and many fall into the same category as pedestrian, dictionary and applicant: there is no metonymy OBJECT FOR MATERIAL CONSITUTING THAT OBJECT and therefore there is no metaphorical relation created either. All this suggests, of course, is that the type‑­shifting hypothesis at least cannot apply to decontextualised examples.

7.9 Summary From the data shown in the present chapter it is clear that a Bare Noun Const‑ ruction exists in SCE which marks referentiality rather than number or definite‑ ness, and that this competes with the specific singular determiner one or plural marking on specific plurals. Plural marking on specific count nouns was conside‑ red to be a counter‑­functional use of plural marking as it appears where it is least needed, i.e., on pre‑­quantified or pre‑­determined NPs. Similar patterns to those of SCE are found in Jamaican and other English creoles.

214 

 Chapter 7 Bare noun constructions

The hypothesis of type‑­shifting and coercion, the only means by which mass nouns and count nouns can switch categorial status in other dialects of English, was tested against the present data, and it was considered that, given an account of coercion as a simple metonymic process (see, e.g., Ziegeler 2007), the data presented from SCE, taken from a number of previous studies of bare nominals in SCE (Ziegeler 2003, 2010, and 2012) revealed that the BNC in Singaporean English cannot be explained completely by hypotheses related to construction coercion and the semantics of mismatch of lexical input into incompatible syn‑ tactic frames. The data illustrate then, not only that such artifacts are intended for standard usage only, but that the differences between count and mass nouns require review, in the face of evidence from contact dialects of English in which both count and mass distinctions and specific‑­non‑­specific distinctions may co‑­ occur. The postulation of a theoretical premise based on such a restricted range of typological scope is surely questionable, and it is clear that the data from contact varieties of English proves that the premise cannot be upheld against independent evidence. BNCs include mass nouns in SCE, but they also embrace a wide range of count noun constructions which are marked as bare due to the absence of referentiality in their discourse functions. The same absence of refe‑ rentiality is obvious in the examples of a number of English‑­based creoles. This is not to suggest that the count noun BNC is ungrammatical in SCE, or ‘belongs’ to another constructicon; what it does suggest is that we should be scrutinising the ­semantics of BNCs in standard varieties more closely to seek out a more universal application of modern construction theory, commensurate with the variational data shown in studies such as the present one. It remains also to seek out the most appropriate means of describing the contact phenomena of BNCs if they are not entirely attributable to substrate features (the creole data also raises this question). This will be one of the aims of the next chapter.

Chapter 8  The Merger Construction: a model of construction convergence 8.1 Introduction From what we have seen so far in the previous four chapters, there is a clear need to explain a number of problems arising from the prolonged contact of two regi­ onal varieties of southeast Asian English, SCE and SSE, in terms of a generalised theory which could be equally applied to other contact situations of a similar nature. The four problem case studies just presented, as noted earlier, differ from other contact phenomena, and indeed, other phenomena previously described in the research on Singapore English, in that they do not all replicate arbritrarily the functions of the model languages with which they were originally in contact (e.g., the southern dialects of Chinese, or Malay) but appear to be motivated, nonethe­ less, by relatively covert linguistic features in the model languages. The features concerned differ in one way or another from related features in the standard subvariety, and yet are in another way, parasitic on the standard sub-variety catego­ ries with which they may be associated. The features summarised in the last four chapters all have one motivation in common, and that is to extend the range of grammatical application of the forms concerned beyond those normally obser­ ved in other varieties. Thus, the reduction of causativity described in Chapter 4 motivates the extension of the transitive construction, the causative component being ignored in the more expanded use of the conventionalised scenario, and in Chapter 5, it is the extension of the adverb ever to non-negative-polarity con­ texts that gives rise to its particular function as a marker of experiential aspect. In Chapter 6, the standard past tense has been extended to be used in a range of contexts in which it may be hypothesised to express precedence and the comple­ tion of an anterior event which continues to hold implication at the moment of speaking. Such an extension can also be associated with the use of the past tense in standard varieties, but it competes with another implicit function associated with past tenses generally, and that is the grammaticalisation of perfectivity or the viewing of an event in its entirety. In Chapter 7, the extension of the range of bare nominal constructions in SCE was seen to apply not only to mass noun referents as in SSE, but also to generics and nominals which have non-specific reference, singular or plural. In all such examples, then, the extended form is generalising on the basis of a semantic link with a similar construction in the ­co-existing, standard sub-varieties.

216 

 Chapter 8 The Merger Construction

Many existing theories of contact deal mainly with contact languages which have often ceased to be in continuous contact with their lexifiers, such as pidgins and creoles, and this factor makes Singapore English a test case for examining more complex theories than those usually associated with such situations. Bao (2010: 792) compares the New Englishes with pidgins and creoles, suggesting that in both situations, there is restructuring of the grammar of the lexifier as a result of contact with local languages. Thus, the influence of the model languages, the southern Chinese, Tamil and Malay substrate, as well as influence from presentday Mandarin and from the (lexical) source language, English (in its adstratum manifestation known as Standard Singapore English), continue to provide a sig­ nificant influence on the development of the vernacular SCE, in such a way that normal processes of restructuring can be observed to differ somewhat from those of typical creole contexts. In the present chapter, the construction-based approach may also be seen to provide an optimum explanation for some types of restructu­ ring described in previous studies. The second section in the present chapter will review some of the background research on grammatical contact phenomena in different situations, in order to determine whether the four case studies provided in the previous chapters may be compared with other recent theories of structural development in contact. Section 8.3 will review the case studies in turn, testing the results against the explanatory tools used in the previous theories of grammatical contact. Section 8.4 will discuss any previous research on the nature of constructions in contact situ­ ations, and Section 8.5 introduces an alternative model, illustrating the way in which the hypothesised theory of Merger Constructions proposes a more viable means of first, explaining the data presented in the case studies, and second, accounting more succinctly for previous theories of contact involving the con­ tinued presence of the lexifier source. It will be hypothesised that the Merger ­Construction is optimal in contributing a role in both explaining variation in mono­ lingual situations, and also in resolving some of the problems outlined in Chapter 3 referring to the constructicons of mixed languages and similar situations.

8.2 Mechanisms of contact construction development 8.2.1 Convergence The title of the present volume refers to the topic of converging grammars, and thus it is appropriate to include a brief resumé of what convergence in contact has been known to suggest in previous accounts. In actual fact, convergence is a term with a fairly indeterminate application in the field of contact, and has been used



8.2 Mechanisms of contact construction development 

 217

by different accounts in different ways. Probably the best summary of its uses can be found in Heine and Kuteva (2005: 9–11), who first compare Aikhenvald’s more generalised (2002:1) definition of languages in contact becoming “more like each other”, with Myers-Scotten’s more complex (2002) definition of convergence as involving both lexical borrowing as well as grammatical morphology, and moti­ vated by sociolinguistic reasons. One of the earliest uses of the term in the sense that Aikhenvald (2002) uses it might be found in Gumperz and Wilson (1971), a study of contact between four Dravidian languages in Kupwar, central India, dis­ cussed in Thomason (2001: 45–6). The convergence between the speakers of the four languages, Kannada, Urdu, Marathi, and Telugu is so intense, according to Gumperz and Wilson, that any sentence translated word for word from one of the four languages produces a perfect sentence in any of the other three languages. In this way, the term is most appropriately used to refer to not only mutual intel­ ligibility, but also to areal phenomena created by contact. Thomason (2001: 125) also uses the term to equate with the phenomenon of isomorphism, suggesting that this concerns convergence in everything except phonology, but that in no circumstances has this kind of merging ever involved a total merger of all the grammatical structure of the languages in contact. She also notes (2001: 89) that convergence, if taken with the definition that two or more languages become like each other, would make almost every contact-induced change a case of convergence. She remarks that there may be more than one way of discussing the use of the term: if one, both, or all languages involved become more similar when in contact; this is a mutual process, not a unidirectional one, and it obviates the need to rely on status-based terms such as Language A and B to refer to the contact situation. The other use is for describing linguistic pheno­ mena which have areal or Sprachbund associations, the source of which is often unknown. To give an example of the first use, Thomason discusses a phrase-final intonation feature used by bilingual child speakers of Turkish and German living in Germany. The feature is expressed in two patterns, one belonging to German, and one belonging to Turkish, both with the same function. The bilingual child­ ren have retained both patterns and differentiate them semantically, so that, alt­ hough the forms can be traced to either Turkish or German as a source language, neither of the two languages has both functions so that the entire pattern belongs to neither language. This is an example of the creativity of convergence. What is emphasised in Thomason’s account is that the term convergence is not usually used to refer to unidirectional convergence, only mutual. Heine and Kuteva (2005) prefer not to adopt the use of the term convergence, especially in its relation to shared areal features, which, according to their accounts, are more likely to have been stimulated by replication by grammaticalisation than any other mechanism (see Johanson 2013 for a critical counter-argument).

218 

 Chapter 8 The Merger Construction

The unusual case described in Gumperz and Wilson (1971) has been analysed more closely by Heine and Kuteva (2005: 210–2) as a collection of common grammati­ calisation features, in which, for the most part, the model languages are the two Indo-European ones, Urdu and Marathi, and the replica languages are the Dra­ vidian ones, Kannada and Telugu. The grammaticalisation cases they describe have taken place over at least three centuries in each case, for example, the deve­ lopment of a polar interrogative marker meaning ‘what’ based on the model of the Indo-European interrogative pronoun, and the crosslinguistically common grammaticalisation process whereby benefactive or dative case markers give rise to purpose markers when they extend from human participants to those that are inanimate. A third example is that of the grammaticalisation (generalisation) of dative postpositions in Kannada to apply to human accusative objects, where in standard Kannada a separate accusative postposition is needed. Matras (2011) also relates convergence to linguistic areas, noting that conver­ gence was originally used to refer to isomorphism in structure across languages. However, amongst his examples of convergence are included those that do not necessarily reflect structural isomorphism in terms of matching morphosyntactic patterns, but instead, demonstrate a convergence of construction characteristics, which may be regarded as pivotal, e.g., (2011: 4): (228) Macedonian: čovek-ot što man-the what ‘the man who arrived’

dojde arrived

(229) Macedonian Turkish: adam ne geldi man what arrived ‘the man who arrived’ Matras (2011: 5) explains that Standard Turkish uses a preposed relative clause expressed as a gerundial (gel-en adam – equivalent, presumably, to ‘the arriving man’), while in the case of Macedonian Turkish, the relative clause is postposed as shown in (229), on the model of Macedonian in (228). The Macedonian Turkish example also lacks marking for definiteness on the head noun, illustrating that convergence is not necessarily a morpheme-by-morpheme process. In this type of convergence, the notion of fusion is invoked; that is, the identification of meaningful elements sharing characteristics that enable a particular gramma­ tical function to be carried across to the replica language, but which does not necessarily include the fusion of all the elements in the model language. This



8.2 Mechanisms of contact construction development 

 219

process may be manifested in the form of a construction, as in the relative clause construction illustrated in (229).82 In the model proposed to follow, the notion of construction sharing in this way will be compared with such examples.

8.2.2 Material and pattern copying It is generally accepted in the majority of the literature on contact languages to date that there are two principle means of transfer taking place between two or more languages in contact. The two principle means are described variously, depending on the sources referred to, as (generally) borrowing on the one hand and on the other, as grammatical calquing, loanshift (Haugen 1950), indirect or morphosyntactic diffusion (Heath 1978; Aikhenvald 2002: 4), interference (Thomason and Kaufmann 1988), congruence (Corne 1999; Mufwene 2001), or structural borrowing (Winford 2003: 12), all the latter terms being cited in Heine and Kuteva (2005: 6). Winford (2005) distinguishes between borrowing (under the Recipient language agentivity) and imposition (under the Source language agentivity), using terms introduced by Van Coetsem (1988). Other terms include fusion (discussed above), and relexification, or relabelling (e.g., Lefebvre 1998, 2001), which takes into account the combined role of the lexicon as well as the grammar in transfer. A further distinction is made by Thomason (2001: 129) between contact-induced changes in which imperfect learning is not related (borrowing, usually involving lexical items) and that in which imperfect learning does play a part (what she prefers to label “shift-induced interference” – usually involving structural transfer). Another distinction along the same lines is made by Matras and Sakel (2007) and that is the notion of copying, divisable into two types, material (equivalent to borrowing) and pattern (structural replication). The concept of replicating the grammar of a contact language is taken to a more comprehensive theoretical coverage in the accounts of Heine and Kuteva (2003, 2005) which discuss contact as a motivation for grammaticalisation, in any field of contact, not just in the development of pidgins and creoles. The processes of

82 Matras (2011) distinguishes constructions from word-forms in that they convey meanings which are more context-dependent and flexible, having meanings that are “derived from a particular configuration of word-forms” (2011: 10), individual word-forms having stable, “attributive” meanings. In this way, his definition of constructions in contact differs slightly from the construction-based approach assumed in the present volume, which follows the principle of any form-meaning relationship.

220 

 Chapter 8 The Merger Construction

grammaticalisation in contact have been outlined in Chapter 5 and will be reitera­ ted briefly below. Other terms used to describe the two means of transfer include Haugen’s (1950) importation (borrowing) versus calque, as well as ‘substance’ compared with ‘form’ (Gołab 1956, 1959) and Johanson’s (1992) use of the terms global copying and partial copying, the latter three sources mentioned in Matras and Sakel (2007). Matras and Sakel’s (2007) study, like Thomason (2001), regards contact as taking place on a micro-level, involving individuals, as well as on a broader scale involving entire communities, thus using the term contact somewhat less discriminatingly than Heine and Kuteva (2005) had done, whose studies focused more closely on the role of contact grammaticalisation in language areas. In Backhus, Doğruöz and Heine (2011), micro-level contact is conside­ red in most cases equivalent to nonce replication and is said not to guaran­ tee change, or adoption and conventionalisation by a linguistic community, a process which can take place over generations of speakers. However, in both Matras and Sakel (2007: 848) and in Matras (2011) it is made clear that locallevel convergence amongst bilinguals must be followed up by propagation of the innovation within a community of speakers for a change to take place in the replica language, and the examples of local-level replication of features by bilinguals were not necessarily intended to infer a change would take place; rather, they were intended to explain the progress from the individual bilin­ gual speaker’s exploitation of their total repertoire of bilingual constructions at their disposal to potential propagation across the community of speakers. As such, they present an important contribution to studies of actuation in change through contact. The differences between material replication and pattern replication are shown in the following examples. The first type of replication (MAT, or material) is illustrated in an example from the Sinti dialect of Romani, a language in contact with German, in which a German particle hin is transferred wholesale into the contact language of Sinti Romani, as a material replication (Matras and Sakel 2007: 846): (230) a. Sinti Romani: me dža u I go-1SG b. German: ich geh e I go-1SG ‘I go [there]’

hin DIR hin DIR



8.2 Mechanisms of contact construction development 

 221

In another case, however, the German particle is readily translated by a Sinti Romani form: (231) a. Sinti Romani: me ker au I make-1SG

b. German: Ich mach e I make-1SG ‘I open the door’.

o DEF.M die DEF.F

vuder door Tür door

pre up

auf up

The example in (231a) is one of pattern replication (PAT), where the polysemous form of the verb-particle, auf, which functions also as a preposition meaning ‘up’ in German, can serve as the pivot from which the Sinti Romani form is triggered as a loan translation, using a preposition of equivalent meaning, pre. In the first case (230), the result is a direct borrowing of the form, as there is no translational equivalent in the model language from which to provide a pivot. Such distinctions served as the basis for Matras and Sakel’s (2007) theoreti­ cal approach to grammaticalisation in contact, allowing them to build on earlier research by Nau (1995), who suggests that loan translations involve the recogni­ tion of a certain polysemy between concrete and abstract senses in words in the model language for correspondences to be made in the replica language. They also cite Keesing (1991), whose studies on Melanesian Pidgin are equally signifi­ cant, suggesting that speakers of the replica language are able to identify lexemes in the superstrate that carry the same grammatical meanings as functionally-­ parallel lexemes in the substrate languages (Matras and Sakel 2007: 833–4). Such studies had highlighted the importance of a polysemous grammatical structure in the model language which was identifiable by the speakers of the contact lan­ guage in order for them to create new functions. The type of identification descri­ bed is thus useful to resolving the questions associated with grammaticalisa­ tion studies of contact, and the hypothesis of recapitulating historical routes of grammaticalisation in contact (as noted also by Matthews and Yip 2009). What it means is that the more semantically abstract a grammatical form becomes, the less likely it will be accessed by contact speakers for replication in contact. The strategies of identification of lexical polysemies in contact in this way are usually associated with polysemous grammatical items in the model language. However, since contact speakers may also identify polysemies in the replica language from which to base parallel grammatical functions, the process does not belong to the substrate or model language alone, and is ultimately a strategy of semantic

222 

 Chapter 8 The Merger Construction

analogy, as indicated in Matras (2010: 69). What is an important factor in the process is the identification and matching of functions and forms, in other words, construction matching. Matras and Sakel noted the presence of an additional motivation in the strat­ egy of pivot-matching: the recognition of the potential for certain morphosyntactic structures in the replica language to serve as the appropriate patterns for repli­ cation (2007: 853). This is something which is not so readily explainable, and is alluded to in a study by Matthews and Yip (2009). Although the present chapter does not deal especially with grammaticalisation as a mechanism of contact cons­ truction development, the need to search for polysemous items in pattern-copying is germane to the hypotheses to be proposed later. Thus, crucial to the present ana­ lysis is the definition of a pivot, the nucleus of interlingual conceptual exchange. The bilingual speaker creates new constructions on the basis of the isolation of pivotal features, as described in Matras (2010), taking into account the full range of bilingual linguistic apparatus at his or her disposal. Importantly, the pivot is not only a semantic element but may be represented as an abstract morphosyntactic device as well (Matras and Sakel 2007: 856), e.g., (Matras 2010: 71): (232) German; age 6:0, addressing both parents, commenting on their ­conversation (which is conducted in German): Was redet ihr über? what talk2.PL you.PL about ‘What are you talking about?’ German: Worüber/Über was redet ihr? Matras (2010) describes the utterance as a hybrid or a blend, using English prepo­ sition-stranding and German morphosyntax, and deconstructing the construction to isolate the pivotal feature of preposition-stranding (a highly conventionalised feature in this particular construction context). The pivot then becomes a source for the selection of a matching form in the speaker’s inventory of context-appro­ priate forms, an inventory that includes not only words but combinatory rules as well, and the result is an innovative construction conceived in a contact situation (2010: 72). Such constructions will be compared below with the results of the case studies in the preceding chapters, in the proposal of a Merger Construction model.

8.2.3 Grammaticalisation The theory of contact grammaticalisation proposed by Heine and Kuteva (2003, 2005) has probably become one of the milestones in research on contact linguistics,



8.2 Mechanisms of contact construction development 

 223

though it has not been without its critics, as one of the most difficult concepts for many researchers to embrace has been the replication of historical patterns in the model language in what appears to be a condensed diachronic period of contact. In Heine and Kuteva’s theory, replication in contact usually has been demonstra­ ted to reveal replication of stage-by-stage grammaticalisation changes, not simply the shifting of the source and the target of the grammaticalised item, something which they refer to instead as polysemy copying (Heine and Kuteva 2005: 100). The basics of their theory of contact grammaticalisation have been outlined in Chapter 5.4, and will not be reiterated here, apart from testing the two models, ordinary and replica grammaticalisation, against the case studies in Chapters 4, 6, and 7 (see below). What has not been discussed so far, is what is actually the model language in cases of contact, and whether it is always the substrate or contact speaker’s L1. Replication of historical patterns usually refers to the replication of historical patterns of the substrate model languages, in that entire pathways of grammaticalisation may be replicated anew in the language of contact. However, more recent studies on the New English varieties show that features which are common to more than one New (contact) English variety may be replicating stages of the pre-contact replica language itself (see Ziegeler 2014a). For example, the similarities in the development of the future modal will in SCE and in Old English both point to the fact that there is a high frequency of uses found with generic con­ structions or habituals (Ziegeler 2006). In the case of SCE, the frequency is marked by comparison to that of present-day British English, which is extremely low (less than 5% for habituals found in the International Corpus of English-GB). Similar findings are associated with other New Englishes, such as Indian English and Trinidadian English. The recapitulation of historical routes in this way may not be exact, but it is not without reason that other studies such as Mufwene (2008) have insisted that grammaticalisation in creoles proceeds in a way little different from the way it proceeds in the evolution of other languages (2008: 172). The same observation has been made by Heine and Kuteva (2005: 241–2). Mufwene (2008) also criticises the position taken by Bruyn (1996, 2009) that cases of grammaticalisation in contact are only ‘apparent’ grammaticalisation, or a kind of local relexification or calquing. Bruyn’s claims are built on data mainly from Sranan, an English-lexified language that arose rapidly in Suriname, with a number of West African substrate languages such as Fon and Ewe (Gbe members of the Kwa group) and Kikongo (Bantu). The short period of English colonisation meant that English was a target language for only a brief period of time. On the basis of a number of examples, including the article system and the use of ben as an anterior marker (lexified from the participle been), Bruyn maintains that the appropriate grammaticalisation pathways known to be associated with these items are not necessarily followed through in Sranan, and that in the case of ben,

224 

 Chapter 8 The Merger Construction

a reanalysis has taken place using material with no apparent lexical beginnings, e.g., (2009: 329): a ben jeri, a ben sa (233) effi if 3SG PAST hear 3SG PAST FUT ‘If he had heard (it), he would have come’ [C18-SchDict15]83

komm come

However, grammaticalisation has already been defined in Hopper and Traugott 1993[2003]) to involve not only the use of lexical sources, but also of material which may be already grammaticalised being used in even more grammatical functions. Bruyn (2009) subscribes to Detges’ (2000) proposal that items like ben should be considered as reanalyses without grammaticalisation, rather than grammaticalised contact variants. Mufwene (2008: 164) questions this approach, suggesting that ben functioning as an anterior marker may have developed like any other case of grammaticalisation, regardless of whether it may be considered lexical or grammatical, since it is found in many language contact situations inclu­ ding that of AAVE. Mufwene complains: Why should a language such as Sranan, which is identified as an English creole, be expec­ ted not to have patterned some of its grammatical constructions on the “lexifier” itself in order for these to count as internally motivated? Does any normal speaker ever learn words of a language without paying attention to their patterns of usage, although its “acquisi­ tions” need not be perfect ..? (2008: 164).

Any approach to contact grammaticalisation must of necessity include such attention, as it cannot be the case that contact speakers are only attempting to replicate the grammatical structures of their first languages. The traditional ­literature on contact linguistics in general, but especially that associated with colonial domains of contact, is replete with examples of transfer, as if the spea­ kers in a contact setting were adamantly resistant to adopting anything but the phonology of the language with which they come into contact. Evidence of borro­ wing is used to amplify such positions. However, it was never emphasised in Heine and Kuteva’s original studies that the case of grammaticalisation was the only means by which transfer should take place across contact. In more recent work, Heine and Nomachi (2013: 84–9) vouch for a series of diagnostics as determinants that grammaticalisation across contact is the most likely scenario. These include differences in the rate of ­grammaticalisation of a specific feature of two languages in contact; e.g., the Slavic

83 Bruyn’s (2009) examples are derived from a range of diachronic sources, including S ­ chumann (1783).



8.2 Mechanisms of contact construction development 

 225

languages in closest contact with German have grammaticalised the numeral ‘one’ in their languages to become an indefinite article – this is grammaticalised further in languages such as Upper Sorbian than in languages such as Ukranian, further from the German border (see Heine and Kuteva 2010 for more detailed discussion of this evidence as an implicational scale). Other diagnostics include rare grammaticalisation cases: a case which is crosslinguistically uncommon, for example the use of a lexical source meaning ‘take’ for a future tense marker in Russian Romani dialects, where many languages tend to use sources such as ‘go to’. There is evidence though, that such sources are found in Ukrainian and some Belarusian and Russian dialects, indicating that a contact-induced gram­ maticalisation in Russian Romani is highly likely. Another diagnostic is paired ­grammaticalisation – in cases where two neighbouring languages which are not related or only remotely related may share the same two grammaticalisation processes for one function, e.g., the marking of future tense in Bulgarian and in Romani languages in contact with Bulgarian, the Romani languages replicating not just the positive form in Bulgarian, but also the negative future tense which uses a different lexical source altogether. The data supplied in Heine and Kuteva’s studies (2003, 2005, 2010) is compellingly illustrative of the point they are ende­ avouring to uphold, and that is that of contact as a motivation for grammaticali­ sation more than grammaticalisation as a motivation for contact; i.e., to provide evidence for the general behaviour of external factors in grammaticalisation as much as internal ones. Heine and Kuteva (2010) also mention restructuring as an alternative to contact grammaticalisation, but they do not discuss it further. It is, however, listed by them as a form of contact-induced linguistic transfer. The factor of transfer, again, seems unavoidably bound up with the field of contactinduced grammaticalisation. But there appears to be little research on the possi­ bility of contact-induced grammaticalisation without transfer. This leaves open a much wider field of future endeavour.

8.2.4 Equivalence, and other constraints One illustration of the fact that Heine and his colleagues do not subscribe unswer­ vingly to the thesis that grammaticalisation is the only explanation for contact transfer is in their discussions of polysemy-copying (Heine and Kuteva 2005: 100). This process is otherwise known as grammatical calquing, and is also the process referred to by Bruyn (1996, 2009) in her identification of apparent grammaticalisa­ tion. However, Heine and Kuteva distinguish contact-induced grammaticalisation from polysemy – copying in the fact that grammaticalisation in contact involves an

226 

 Chapter 8 The Merger Construction

intermediate series of stages, it is unidirectional, and it reveals differences in the pace or the rates of grammaticalisation across the languages in contact. Matthews and Yip (2009) also observe that contact grammaticalisation, like any other gram­ maticalisation, can exhibit evidence of persistence, divergence and layering (see Hopper 1991). Although differences between ordinary contact-induced grammaticalisation and replica grammaticalisation have been outlined in Chapter 5, it is worth reitera­ ting that the main differences are that in the former, grammaticalisation takes place using a crosslinguistically universal lexical source (e.g., already to express perfec­ tive aspect, as described in Chapter 5), while in the latter, the grammaticalisation path of the replica language recapitulates that of the model language, using the same lexical sources (examples include the use of the Location Schema [X is after Y] to form the ‘hot-news’ perfect in Irish English, which goes through the same stages of development to form the perfect aspect in the model language, Irish). In some cases, though, the lexical source of grammaticalisation in contact is less transpa­ rent, and this situation gives rise to the notion of equivalence across contact (Heine and Kuteva 2005: Chapter 6.1; Heine 2013). Equivalence is sometimes described as isomorphism, though as we have seen above, the term isomorphism is sometimes used in place of convergence, which is a more complex notion altogether, involving linguistic areas. The examples illustrating cases of equivalence are not always, it seems, completely isomorphic, as well, though it is clear that what is established is often a functional equivalence between constructions without necessarily a trans­ parent lexical or morphosyntactic equivalence. The example described by Heine and Kuteva (2005: 220) as a “paradigm example” does just that: it establishes func­ tional equivalence only. It concerns the use of infinitive markers in some languages being considered as corresponding to nominalisation markers in another language (e.g., they note Johanson’s (1998) example of Persian infinitives perceived as corre­ sponding to the verbal noun in Southern Azerbaijanian). Similar correspondences can be observed in Singapore Colloquial English in which the sentence final discourse marker one is treated as equivalent to the sentence-final nominaliser – de in (Mandarin) Chinese (Bao 2009), where there is no similar structural or lexical source in English to copy, and speakers of SCE have selected a pronominaliser as the nearest approximation to the functions they wish to convey, e.g., (234), where the pronoun one is perceived as equivalent to the emphatic marker de in Mandarin (Bao 2009: 340): (234) SCE: I always use microwave one! Mandarin Chinese: wo zongshi yong weibolu de English: I always use a microwave!



8.2 Mechanisms of contact construction development 

 227

Bao (2009) isolates this one example as the only example in which one does not modify a noun phrase; however, speakers of SCE appear, nevertheless to have established equivalence between a discourse-marking function of emphasis in Mandarin and SCE. In other words, one has been extended in its nominalising uses (e.g., the blue one) to those of marking clause-final discourse emphasis, appa­ rently quite unrelated functions. However, it is not entirely clear that the nomi­ nalising function is completely unrelated to the use of the SCE example in (234), as a grammaticalisation path of nominaliser-to-stance marker has been identified across many Chinese dialects by Yap and Matthews (2008). Gil (2003: 18) has also referred to the clause-final function of one in SCE, associating it with a function known as reification, in which a grammatical marker (normally used for other functions) assigns to an entire construction a meaning equivalent to ‘thing’. As Heine and Kuteva (2005: 221) note, such equivalences are not without an interpre­ tation as replica grammaticalisations, and furthermore, speakers are likely to pay attention to semantic or functional parameters in establishing equivalence, rather than morphosyntactic ones, even though any structural or semantic feature can serve as the basis for establishing it (Heine and Kuteva 2005: 229). However, in the early stages of replica grammaticalisation, such correspondences may sometimes appear to have no semantic justification for their selection by speakers in contact, and selections may be specific to the contact situation to which they refer. According to Heine (2013) equivalence may take place simply by the speakers of the contact language selecting an item on which to base a replication, or other­ wise by modifying existing material in the replica language to suit their functio­ nal needs. Many of the examples given by Heine (2013) are of lexical equivalen­ ces, and in some cases, these involve the augmentation of senses found in the model language by additional polysemies associated with the replica language, such as the sense ‘manage’ associated with the verb run in English, which has been added to the the meaning of the verb courir in Guernésiais French spoken on the Channel island of Guernsey; it was never part of the meaning of the verb in standard French. Equivalences, according to Heine (2013), can be represented by function-to-function correspondences, additions of functions or meanings, losses of functions or meanings, partial correspondences, and generalisations in contact, such as the Los Angeles Yiddish English bilinguals’ use of the subordi­ nating conjunction when to cover both the meanings of temporal conjunction and conditional conjunction, based on the Yiddish model language which does not have separate conjunctions for these functions. Partial correspondences include the replication of the Romance and Germanic possessive perfects by languages in contact with such languages, e.g., in Celtic, Slavic or Basque languages, the category is only partially replicated and does not yet extend to intransitive verbs or inanimate subjects (Heine 2013: 13).

228 

 Chapter 8 The Merger Construction

8.2.5 Relexification and systemic transfer The mechanism of polysemy-copying is discussed in Matthews and Yip (2009) as related also to the mechanisms of relexification discussed by Lefebvre (1998, 2001) ­ ppropriate for the fact that in such processes, subsequent to identification of the a equivalence in the lexifier language, the entire range of grammatical functions is transferred wholesale to the replica language with no intervening stages of ­grammaticalisation necessary. Lefebvre (2001) refers to Muysken’s (1981) defini­ tion of relexification as a process of vocabulary substitution, and then adds that: “A ­relexified lexical entry thus has the semantic and syntactic properties of that of the lexicon it has been copied from and a phonological representation derived from a phonetic string in the lexifier language.” (Lefebvre 2001: 374) She goes on to relabel the term relexification as relabelling. At first sight, observations of relabelling and relexification appear hopelessly simplistic and give rise to the impression of a need to penetrate the problems of contact more deeply. The work of Lefebvre and her colleagues, though, is centred around the study of a French creole, Haitian Creole, and for that reason alone, does not necessarily need to apply to the same conditions as are found in contact situa­ tions on the German-Polish border, for example. One important difference, as noted earlier, is in the length of contact: creoles typically demonstrate a situation of brief contact and rapid development of a replica language (around 70 years for Haitian Creole, according to Lefebvre 2001), while border languages in Europe may sustain prolonged periods of contact enduring over several centuries. As such, the prognostics for grammaticalisation are more tangible in the latter cases. There is also the question of what is meant by a target language: the work of relabellists conveys the sense of perpetuating the substrate at all costs; according to Lefebvre (2009: 303n), “... identifying in the superstrate language an appropriate phonetic string to provide a new phonological representation for a substrate lexical entry”. However, as Mufwene (2008) frequently pointed out, such a mechanism gives little credit to the acquisitional capacities of the contact language speakers, nor to their creative abilities in developing an entirely new language as symbolic of their own independent identity. The problems of relexification are, nevertheless, at the heart of the problems of mixed construction inventories across contact: if the replica language is nothing but a relabelled phonological representation of the model language, often the substrate, then how is the construction inventory described when it has the mor­ phosyntax of one language and the lexicon of another? The present chapter aims to resolve such problems, but crucial to Lefebvre’s studies is the proviso that relabelling must be driven by semantic factors (e.g., 2009: 282), and is not just a transfer of morphosyntactic structure with a new phonological representation.



8.2 Mechanisms of contact construction development 

 229

For this reason, she claims that functional categories such as case marking cannot be relabelled. Preferred targets for relabelling include determiners, pro­ nouns, tense, mood and aspect markers instead. And if there are no phonologi­ cal forms in the superstrate from which to relabel the substrate categories, the category in the creole remains phonologically null, or, unexpressed, according to Lefebvre (2009: 282). The following example illustrates the way in which the morphosyntactic structure of Haitian Creole reflects that of the major substrate language, Fongbe, while exploiting French lexical material: (235) tab la, m te aste table DEF.ART I ANT buy ‘the table I bought …’ (cited from Koopman 1982: 174)

a DEF

It should be noted that in Haitian Creole, determiners are post-nominal and a is considered an allomorph of la. The source for the definite marker la is taken to be the French deictic adverb là, as in the following French example (Lefebvre 2009: 285):84 (236) Cet homme -là vient DEM man [+deic] come ‘This/that man just arrived.’

d’arriver arrive

As such, it contributes its semantics to the functional needs of the Haitian Creole definite determiner, as well as its syntactic positioning. However, Aboh and DeGraff (2014) point out that the post-nominal syntactic positioning aligns with one of the principle substrates also, Gungbe, and a comprehensive study of the nominal determiner system in Haitian Creole suggests that the features were not attributable to relexification, but to a combination of both the Gbe substrate lan­ guages and the French lexifier (2014: 233). Further research into the possibilities of relexification is therefore necessary, but the emphasis in Lefebvre’s studies for a semantic link between the lexifier and the substrate languages is significant to the present study. The same constraints on relexification have been taken up by Bao (2005, 2009, 2010, in press), and Bao and Lye (2005) in the model of systemic substratum transfer in Singapore English. Bao (2005: 253) illustrates a case of partial overlap with the aspect system of (essentially Mandarin) Chinese, but not an exact repli­ cation, though he notes that at the same time, features in transferral form a cluster of related category groupings. He reiterates Lefebvre’s claims for a lexical entry at

84 Lefebvre (2009) does not supply glosses for the abbreviations in the examples given.

230 

 Chapter 8 The Merger Construction

the core of transfer, which carries with it all the semantic and syntactic properties of the substrate as well. The approach is illustrated roughly as in the following example (fr. Lefebvre 2001: 374): (237) ORIGINAL LEXICAL ENTRY [phonology]i [semantic feature]k [syntactic feature]n

LEXIFIER LANGUAGE [phonetic string] j used in specific ­semantic and pragmatic contexts NEW LEXICAL ENTRY [phonology] j or [ø] [semantic feature]k [syntactic feature]n

Bao (2005) emphasises that, although relexification simplistically involves using the phonological form of the lexifier to express the semantic or syntactic features of the substratum, it is constrained by the overlap of the semantic properties of the original lexical entry with those of the lexifier (2005: 254). The same con­ straints, in many ways, are at the basis of contact-induced grammaticalisation (evidenced by a gradual period of development), as well as Matras and Sakel’s (2007) accounts of pivot-matching. It will be seen below that the role of semantic overlap is also crucial to understanding the merger of construction-types across contact, but that the overlap of semantic properties in Merger Constructions involves contact between the contact replica language and the lexifier instead. Bao (2005) provides an example of the transfer of an entire grammatical sub­ system in his illustration of the transfer of the existential and universal quanti­ fiers yǒu (= ‘have’) and dōu (‘all’, ‘also’) from Mandarin, which are realised in the functions of the verb got and the adverb also in SCE, e.g., (2005: 255–6): Singapore Colloquial English: (238) a. Existential got: Got at least one time everyone happy ‘There was at least once that everyone was happy’.

b. Mandarin Chinese existential yǒu: Yǒu liǎng-men kè wǒ xiǎng course I want have two-CL ‘There are two courses I want to take’.

(239) c. Universal also: Everything also I want. ‘I want everything’.

xuǎn take



8.2 Mechanisms of contact construction development 

 231



d. Mandarin Chinese universal quantifier dōu: wǒ méi yàng dōngxi dōu yào I every kind thing all want ‘I want everything’. Not only the quantifier uses, but the full range of these markers is transferred into SCE, including the locative and aspectual uses of yǒu and the additive uses of dōu.85 The example of the transfer of quantifiers in this way is seen as pro­ viding evidence for the transfer of an entire grammatical sub-system. When the entire sub-system is not transferred, it is considered to be constrained by what Bao labels the Lexifier Filter, by which “morphosyntactic exponence of the transferred system conforms to the (surface) structural requirements of the lexical-source language” (Bao 2005: 258). This is basically a formalisation of the null phonological parameters that were mentioned by Lefebvre (2001, 2009) for transferred items that could not meet the semantic requirements of the lexifier, as discussed above, with additional consideration of the parameters of structural isomorphism. For example, already as an aspect marker in Singlish conforms to the requirements of English in appearing clause-finally, rather than post-verbally (perfective le, its Mandarin equivalent, may appear in either position).86 However, as far as the Mandarin imperfective marker zhe is concerned, it has no realisa­ tion in SCE as its aspectual functions expressing durative simultaneity are not represented morphosyntactically in the same way in SSE or any other variety, or at least using quite different structures, e.g., conjunctive clauses (they sat and talked, etc.). In this way, Bao believed that the Lexifier Filter overrrides the trans­ fer of such a function from the substrate, and that null phonological representa­ tion means it does not exist at all in the contact language. Bao’s (2010) explanation is that what the Lexifier Filter rules out, in many cases, are less frequently used functions, and he provides the example of give vs. kena as markers of adversative passives, e.g., as in (31) in Chapter 2, repeated below as (240): (240) The dog give the boy kick ‘The dog was kicked by the boy’. 

(Bao and Wee 1999: 5)

85 Not all of these functions are represented in standard Mandarin Chinese, as noted by Bao (2005). 86 Recent observations reveal that SCE already is beginning to occupy immediate post-verbal position, e.g., I ate already my lunch (Bao Zhi Ming, p.c., Jan. 2014).

232 

 Chapter 8 The Merger Construction

The give-passive is marked by an extremely low acceptability ratio compared with kena in surveys using native-speaker informants, as noted in Chapter 2, and Bao concludes that such results are caused by the concomitent, low frequency of use of the form (there are no examples of it in the International Corpus of English-­ Singapore Corpus at all). If acceptability judgements may be taken as an indi­ cation of the speakers’ selective intuitions in replicating equivalent functions in contact, the results are quite revealing. At the same time, it still needs to be ascertained whether another verb form, such as let, has been selected instead as the equivalent of an adversative passive verb in the substrate. If so, it removes entirely the need to seek out direct lexical translation equivalents in the lexifier. Bao’s studies are particularly insightful to the field of contact linguistics since, like the studies in the present volume, they focus on a situation of contact which is not matched by many of the creoles described in earlier literature; it is a situation in which the lexifier continues to be used alongside the contact variety with probably the same speakers, in many cases, using it as an alter­ native to Singapore Colloquial English. The situation has been described as diglossic, as in Chapter 2, but with specific, often conflicting sociopolitical targets in the community of either publicly eliminating the contact dialect as an example of stigmatised and undesirable, non-standard usage, or of maintaining it as a lone emblem of harmonious sovereign identity in an otherwise ethnically-mixed and culturally-diverse, expanding social framework. As noted earlier, also, the substrate Chinese languages are losing ground to the more dominant standard Mandarin Chinese, not exactly a substratum language in the classical contact context, but a language actively encouraged (initially via the education system) as an alternative to the use of standard English, and it is this language which is now providing the majority of the morphosyntax to be replicated in Singapore Colloquial English. Thus, the New English variety spoken in Singapore is on an almost equivalent standing to the New Mandarin variety used there as well, with the exception that the genetic origins of Mandarin are much closer to the original substrates than those of English. As such, the sociolinguistic context is somewhat unique, and supplies a different perspective on many of the contact phenomena which arise in previous literature. In spite of such uniqueness, however, the ways in which the case studies outlined in Chapters 4–7 will be dealed with must also accord with contemporary theoretical findings in contact linguistics: the remainder of the present chapter will explain how.

8.3 The case studies in the present volume From the literature on contact discussed above, it would seem that there is little cause to attempt to describe the nature of the data supplied in the present case



8.3 The case studies in the present volume 

 233

studies by yet another theoretical device, but at the same time there is a clear need for such a description for at least one reason: the contact data referred to in the majority of the principle studies on language contact, pidgins, creoles, and bilingualism in general, refers mainly to the transfer of concepts or struc­ tural formations (whether construction-types, or simply one-to-one morpholo­ gical pairings is referred to). It is not often the case that features arise in contact to provoke the curiosity of researchers working in such areas which appear not to be the result of substrate transfer. The case studies described in the present volume, for the most part, may then fall into the category of non-transferred features, though in the case of the ever-construction discussed in Chapter 5, it is clear that the motivation for its emergence was on the basis of a transferred functional need in the (substrate) model languages. For the other three, they need to be given more consideration, since the functions they describe appear to have no parallel in the pre-contact replica language. They are features which have received little attention in the literature on SCE to date, mainly for the fact that their origins are not always as transparent as in the case of directly trans­ ferred features. However, in explaining their presence using the description of a merger construction hypothesis, their role in the complex contact machinery of linguistic adaptation observed in Singapore English becomes infinitely more perspicuous. There are several factors so far observed in the previous theoretical approa­ ches outlined above which combine to justify a merger contruction analysis of contact: (i) the involvement of a conceptual or morphosyntactic pivot upon which analogies can be made across the languages of contact (e.g., Matras and Sakel 2007); (ii) the suggestions proposed by some accounts of contact, e.g., Mufwene (2008), that speakers of contact languages are not only attempting to pattern the structures of the new language of the models of the substrates, that the lexifier may equally be considered a model; (iii) the problems of relexification, relabel­ ling and similar accounts (e.g., Lefebvre 1998, 2001), in which the construction must necessarily be formed from a mix of two different languages; and (iv) Bao’s (2005) emphasis on semantic conflicts between the original lexical entry of the substrate item being relabelled and the lexifier. These four points emerge from the contact literature as critical for the development of a new analysis in terms of a merger between the construction of the contact replica language and the cons­ truction of the pre-contact replica language from which it takes its phonological form, for the following reasons: they highlight the need for a closer investigation of the way in which the form of a construction can be attributed to more than one function across contact, and the reasons that speakers of a contact language are able to exploit and reanalyse the lexicon of the pre-contact replica language to suit their particular functional needs. But first, the four case studies will be tested against the theoretical approaches just discussed.

234 

 Chapter 8 The Merger Construction

8.3.1 Transitivity and the conventionalised scenario construction The conventionalised scenario constructions discussed in Chapter 4 as false or ‘fake’ transitive constructions, as noted earlier, carry the syntax of a transitive construc­ tion, but the covert semantics of a causative-resultative construction, in a similar way to the examples listed as conventionalised scenarios by Goldberg (1995). Like the conventionalised scenarios described by Goldberg (1995), their implicit cau­ sative semantics have been ignored or overlooked in the creation of a short-cut, metonymic link with what is superficially a transitive construction, but has the underlying semantics of an indirect causative construction. In the constraints on conventionalised scenarios listed in Chapter 4, it appears that examples such as (241) I cut my hair rate as prototypical CS constructions, since they fulfill the requirements of refle­ xive causativity: the subject is both a goal and source of the affected event, the subject is in a possessor-possessee relationship with the object NP, and direc­ tion of the causative change of state always reverts towards the causer, and never away from the causer source. In addition, a CS must be alternatively construable as a transitive construction. Thus, while (241) may be plausibly reinterpreted as a transitive construction, (242) She cut his hair would present no reason to qualify as a CS. The accommodation of CSs into Singa­ pore Colloquial English, though, has been seen to be potentially explosive, accor­ ding to previous research as well as the results of the study in Chapter 4. Thus, in Chapter 4, we find evidence of constructions such as (243) I will remove it by surgery (244) If it is a back tooth I’ll just pull it out (245) Take another new one [photo of myself] in which the direction of the action is away from the source, and their alternative construal as ordinary transitives is highly implausible. These have been classed as CSs in the study for categorisation purposes, though they do not entirely fulfill the requirements of the CS construction as listed in (73), Chapter 4. They do share another common feature, though, and that is that the source or causer of the event is also an affected recipient, or is affected in some way by the event. It was concluded in Chapter 4 that such a construction may represent an unli­ kely transfer from the model languages of Chinese (causative constructions can be found in Malay and Tamil, as noted in Chapter 4). In terms of the mechanisms described above, it may be possible to describe the construction as a convergence



8.3 The case studies in the present volume 

 235

of the characteristics of a Chinese construction with reduced causativity over the identical construction frame of an English transitive construction, in the same way as Matras (2011) has described for relative clauses in Macedonian Turkish. The correspondences are not all that clear, though, as the Macedonian Turkish example bears no relation with a similar construction in the lexifier Turkish, but rather, with Macedonian, whereas in the present case, the convergence is between Chinese, the contact variety, SCE, and the original lexifier, English. In terms of a grammaticalisation analysis, it could be argued that since the bǎ-construction is described as a pre-transitive marker in many accounts, and it is frequently trans­ lated as a simple transitive construction in English, we have the pattern from which grammaticalisation of transitivity may proceed. However, from the examp­ les supplied by Li and Thompson (1981[1989]), for example, it is clear that the bǎ-construction functions in the same way as an argument s­ tructure construction of Goldberg (1995), in that it frequently co-occurs with a locative, ditransitive or resultative complement. In other words, it retains its causative origins. It is not clear that a parallel analysis can be applied to the CS in Singapore English, but a more intensive study might prove that it could be. It still remains doubtful whether the construction could be considered as unequivocably transferred, when there seems to be a persistent need to express causativity in the bǎ-construction option, as well as the ‘go’-auxiliaries which are also optionally found in the Hokkien and Cantonese examples in Chapter 4. These constructions therefore have their ­parallels in the causative-resultative construction instead. By the same token, it is doubtful that such constructions could be considered as ‘equivalents’ of constructions in Chinese, as they tend to copy a morphosyn­ tactic pattern rather than a semantic or functional one, if at all. That the morpho­ syntactic pattern is only optionally used to translate the causative-resultative in English is another matter. Similarly, the mechanism of relexification or relabel­ ling could not apply to the case of conventionalised scenarios either, as it is not clear precisely what they would relabel. The transfer of a whole system is not relevant, though the absence of subject-verb selection restrictions as in (243–245) might be of importance if it is considered that the construction forms one of the topic-prominent constructions transferred wholesale as an entire system from Chinese to SCE, such as Bao and Lye (2005) point out. However, other studies on topic-prominence (e.g., Ziegeler 2008) have revealed that although the selec­ tion restrictions between the topic and the main verb are being ignored, there is sometimes number and person agreement on the main verb, which indicates that the pre-verbal nominal is not always just a topic.87 The description of such a

87 The notion of ‘topic-predicate’ constructions has thus been mentioned also (Mark Donohue, p.c.).

236 

 Chapter 8 The Merger Construction

construction in contact terms therefore becomes difficult, and it is only possible to define it by analogy to constructions with similar form-meaning relationships. We shall return to this problem again in section 8.5.

8.3.2 The experiential aspect construction The prospect of a contact construction analysis for ever assumes the more fun­ damental definition of a construction as a (symbolic) form-meaning pairing of any sort, which can be reducible to a single word, or morpheme, according to Croft (2001), for example. As discussed in Chapter 3, similar definitions of constructions abound in the works of, for examples, Langacker (2005a) who describes constructions as simply pairings of form and meaning, and Bergen and Chang (2005) who refer to the multi-dimensional aspects of a construc­ tion with respect not only to pairings of phonology and meaning, but also to pairings of morphology, syntax, pragmatics and discourse with meaning. With a construction ­definition as fundamental as a word-meaning pairing under many accounts, there is little reason not to include the experiential marker ever in a construction-based analysis. The question is whether experiential ever can be analysed by any other kind of contact mechanism. It must be remembered that experiential aspect in the major substrate languages is represented by verb forms, with meanings totally different from those conveyed by ever, for example, in the Hokkien example (110) in Chapter 5, repeated below as (246): (246) Goa bat khi Jit-pun I EVER go Japan ‘I’ve been to Japan (before)’ As noted in Chapter 5, the lexical meaning of the verb bat in Hokkien was ‘to be able, to know, by experience’, and therefore one of the polysemies associated with a verb of ability has been grammaticalised. There is no lexical similarity between such verbs and the English adverb, ever, nor between the Mandarin experiential verb, guò or the cognate Cantonese verbs gwoh/kwo and ever either. Thus it may be concluded that the selection of ever to fulfill such functions cannot have been the result of calquing, relexification, or pattern copying either. Bao (2005) has discussed the possibilities of ever in his model of total systemic trans­ fer (of Chinese aspectual functions into SCE), and notes that it reveals the domi­ nance of the Lexifier Filter over System Transfer, so that when the two operations are in conflict, it is always the Lexifier Filter that wins out over System Transfer; i.e., that the transferred feature must confirm to the requirements of the lexifier.



8.3 The case studies in the present volume 

 237

In the case of ever, it is the syntactic positioning that reveals the work of the Lexi­ fier Filter – pre-verbal, as with the use of the same adverb in standard varieties of English. Bao has little to say about the actual selection of ever as an experien­ tial marker in SCE, except to note that it demonstrates an aspectual function of marking the end of a given state at the time of reference, unlike already, which marks the beginning of one (2005: 242). The function of marking the end of a given state at reference time is not, though, necessarily a semantic feature of the experiential present perfect in English, so it cannot be for that reason that it was selected by speakers of SCE to replicate the function of experiential aspect in their model languages. Given that its only relation to the English use of ever is its syn­ tactic positioning, and that English ever is, for the most part, restricted in use to negative-polarity contexts, the selection of ever looks even more mysterious. The possibility of a replica grammaticalisation of the Mandarin experiential marker guò, derived from a lexical source verb meaning ‘pass’, is ruled out due to the differences in lexical source forms. The diachronic development of ever has been investigated in Chapter 5, and it would seem that there is no confirmed evidence for an experiential function in the evolution of standard English ever either. The prospect of an ordinary contact-induced grammaticalisation may be more likely, though, if one considers the additional data from Modern Dutch, in which the negative polarity uses of ooit ‘ever’ have been found to extend to an adverb meaning ‘once’. In this way, it may be hypothesised that ever has been selected because it follows a universal strategy of grammaticalisation of marking experiential aspect, much like already to mark the perfective in SCE, which is clearly an areal feature in Southeast Asian languages (see Haspelmath et al. 2005). On the other hand, the intermediate stages of grammaticalisation are not all that clear, and Dutch is not a contact language. A final possibility, however, would be to consider ever as an example of equi­ valence, in that it establishes only functional equivalence between the replica language and the model language. In some way, it could be considered an arbit­ rary correspondence relation, having no semantic links with the function it spe­ cifies at all. However, this would raise the question why the same form has been selected to grammaticalise the same function in Dutch, a non-contact language. It is possible that semantic correlations are not clearly apparent because ever may be only at an early stage of development into an aspect marker. The syntactic positioning of ever may not be the only correspondence shared across contact, if Heine and Kuteva (2005: 221) indicate that speakers are more likely to pay atten­ tion to semantic and functional parallels than morphosyntactic ones in establi­ shing equivalence. It is hypothesised, along with the conclusions established in Chapter 5, that there were in fact semantic reasons for selecting ever, but they are unlikely to have been affected by the guidance of the model language alone;

238 

 Chapter 8 The Merger Construction

the semantic parameters must be shared across contact. This involves pivot-­ matching, as discussed by Matras and Sakel (2007).

8.3.3 The past tense construction It has been repeatedly claimed in the literature on constructions and construction grammars that constructions are multi-layered entities and one construction may consist of any number of combining constructions, (e.g., Boas 2010; Croft 2001). Within such taxonomies the Past Tense Construction has been isolated in Chapter 6, a construction type manifested in English by the -ed suffix or the irre­ gular forms with which it is associated. There appears to have been little mention of it as a distinct construction type in the mainstream literature on constructions, though Croft (2013) does mention the existence of a Morphological Verb construc­ tion, defined by its occurrence in the Tense-Agreement Inflectional Construction (which is manifested in English as the alternation ø/-s or the Past -ed morpheme, understood as tense distinctions). Given the hierarchical nature of his classifi­ cations though, it is viable to isolate the Past Tense Construction as a construc­ tion type in itself, combinable with any English finite verb, and possibly, one might add, occurring within a construction-hierarchy level of finiteness. Croft, though, adheres to the form-driven nomenclature for introducing new construc­ tion labels, and cites Bybee’s (1985) similar approach that the abstract concept of [past tense] has no existence beyond its inflectional representation in linguistic form across languages. Under such an analysis, we may now proceed to consider again the data in Chapter 6. If the Past Tense Construction is therefore nothing but an entity derived from its inflectional representation across languages, then there is no doubt that it may be treated the same way across two different dialects of English. Earlier research by Goldberg (2003) claims that no two constructions can be the same across two languages unless there has been a diachronic reason or the two languages have been in contact in their histories. The evidence supports a case for mutual const­ ruction-sharing in SCE, regardless of the typological nature of the model language or substrate. In the case of SCE, the model language is a recent influence, but the situation is no different from the historical models of contact that contributed to what we know as standard English today, for example, French, Anglo-Saxon, or Old Norse. The only difference is therefore chronological, and in the survival or conventionalisation of features across time. Thus, it could not be argued that there is one past tense construction for SCE and another for SSE (or any other standard variety). Goldberg’s (2003) solution to the occurrence of identical const­ ructions in this way is to label them construction-types rather than constructions



8.3 The case studies in the present volume 

 239

per se. Such labelling would account for many crosslinguistic similarities, but would not necessarily always explain the interchange between a contact replica language construction and a pre-contact standard language construction, one of the major problems of the present study. The differences can only be reconciled by means of some form of construction sharing. In the contact literature reviewed so far, there is no apparent explanation for the functions found for the past tense construction (PTC) in Chapter 6, in which past tense is used for present tense habitual aspect, anterior uses and performative uses. There is no apparent pattern-copying, since the uses of past tense for present time reference is not found in the model languages. In the work of Ho and Platt (1993) it was concluded that past tense, as summarised in Chapter 6, copies only perfective lexical (or grammatical) aspect in SCE; i.e., it only partially expresses the past tense of standard varieties of English, but has only partly replicated the functions of marking perfectivity in SCE also. As such, it fulfils a possible func­ tion of only partial equivalence. In terms of Ziegeler’s (2000) relative grammati­ calisation patterns, it is hypo- (under) -grammaticalised compared with standard English varieties. But the extension of its uses to domains which would standardly require a verb unmarked for past cannot be explained as any form of transfer from the model, as discussed in Chapter 6. There is no morphology to relexify, since the model languages do not have a tense-marking system, and hence no syste­ mic transfer either. It offers no explanation in terms of fusion, convergence or cal­ quing, which would require a model function to replicate. A tentative explanation might relate one of the functions, that of marking habitual presents, to a universal crosslinguistic pattern in which they are found with perfective marking in some languages, as discussed in Chapter 6.3, but that would not account for the other uses described in Chapter 6, nor would it account for the absence of past marking on past habituals. The absence of any form of transfer mechanism leaves open the only possibility: that the past tense construction used in present tense contexts (the PFP) may be nothing more than an innovation.88 However, it remains to deter­ mine on what grounds it was selected as an innovation; this will be illustrated below. Innovations are not uncommon in the history of many languages (see, e.g., Lehmann 1995), but little research, to present knowledge, has been devoted to investigating the likelihood of innovation as due to contact.

88 There is similarity between the functions of the Past-for-Present in SCE and the historical functions of the (positive affirmative) auxiliary do in the history of English (see, e.g., Ziegeler 2004). The existence of such parallels is speculative and need not inspire a recapitulation grammaticalisation account (Ziegeler 2014a), though further research might prove interesting.

240 

 Chapter 8 The Merger Construction

8.3.4 The bare noun construction The last of the case studies, the Bare Noun Construction (BNC) described in Chapter 7, also leaves room for doubt as to the involvement of mechanisms of transfer in the contact situation. It has been isolated as a construction-type for the reason that its zero-marking for number or definiteness actually grammaticalises a function in SCE: the function of marking non-referentiality or non-specificity, as discussed in Chapter 7 (see, e.g., Bybee 1994 on the grammaticalisation of zero, in which zero grams (grammatical morphemes) are considered to have semantic substance in opposition to marked categories). In the case of the BNC, the absence of overt marking for number on non-specific nouns is semantically salient, and the absence of it carries automatic inferential value that the meaning associated with overt number marking was not intended. Thus, a BNC construc­ tion carries not only lexical meaning, but also the information that the referent is non-specific. The link between the overt marking for specificity and that for number in SCE has been related to a form of modality on such NPs, in which number-marking carries the most salient semantic feature in representing exis­ tence in SCE (Ziegeler 2012). In terms of transfer mechanisms, though, it was demonstrated in Chapter 7 that SCE Bare Noun Constructions only partially capture the categories that the same constructions represent in Mandarin Chinese, in which even referential or specific NPs may be found unmarked for number or referentiality. Gil (2003) emphasises this in indicating that in the SCE example in (247), the noun apple may be understood as singular or plural, definite or indefinite, and quantified or unquantified. It may also refer to an unmeasured amount of chopped or mashed apple, to several apples, or a single whole apple. (247) Geraint eat apple He goes on to note that the translational equivalents of examples like (247), in the more frequently-used Chinese substrates such as Hokkien and Cantonese are differentiated from those of Mandarin Chinese only in the fact that singularity of nouns is marked by a classifier following a demonstrative pronoun, e.g., Canto­ nese go35 lap5 piŋ 11-guo35 and Hokkien hit4 liap21 pheŋ 22-ko53, both glossed as ‘DEM:DIST CLF apple’ (2003: 11), whereas the classifier is optional in Mandarin (nà pínggǒu, ‘this/that apple’), even though the number remains singular. Stan­ dard Malay epal does not have obligatory marking for definiteness or number either. The ambiguity regarding number and definiteness seen in examples such as (247), then, might be difficult to relate precisely to a particular model language: SCE does not have a classifier system, and as we have already seen in Chapter 7, the same indeterminacy regarding number or definiteness produces Bare Noun



8.3 The case studies in the present volume 

 241

Constructions in many creole languages as well, which do not have Chinese dia­ lects as their principle substrate or model languages. It is questionable, therefore, whether there is actually a model language for such constructions in SCE. On the other hand, they might not be definable as innovations like the present habitual use of the past tense described above. The categories of specific-non-­specific distinctions are described by Bickerton (1981) and represent a classic category of many contact languages and creoles. As such, they could be described as a case of ordinary contact-induced grammaticalisation, as per Heine and Kuteva (2003, 2005), in which the indefinite article evolves, crosslin­ guistically, from a numeral ‘one’ source. However, such a case would only explain the emergence of specific nominal determiners; what needs to be explained is the grammatical correlation between non-specific reference and the BNC (also a correlation associated with English bare mass nouns). And yet the BNC is iden­ tifiable as a distinct construction type whenever it appears (even in present-day standard English, see, e.g., the accounts of De Swart and Zwarts (2009). It has been demonstrated also in Chapter 7 that the BNC cannot be explained by the construction coercion theories associated with other forms of bare noun construc­ tions in standard varieties of English. There is no material copying, since there is nothing to copy – it is morphosyntactically a zero item. There is also nothing to grammaticalise, diachronically (in spite of Bybee’s 1994 use of the term gramma­ ticisation to indicate that it merely carries a grammatical label in the system). The BNC, furthermore, relexifies no morphology from the model languages, though in SCE, it partially replicates what is expressed as a BNC in the Chinese dialects. As such, it could be considered as marking equivalence. But partial correspon­ dences, from the examples supplied by Heine (2013), appear to involve some morphosyntactic form; the BNC has no morphosyntactic realisation at all.89 The four cases presented above: the CS construction, the Experiential ever Construction, the Past Tense Construction, and the Bare Noun Construction, all appear to require a stronger explanation than what has been offered in the recent contact literature. The reasons that such items require further explanation is that they are not always seen as transferred from any model (e.g., the past tense construction used for present habituals), or that they cannot be explained with respect to model language features. They comprise features of contact relatively

89 The question arises again of whether it can be considered a construction type if a definition of a construction is a form-meaning pairing. However, the unmarked lexical input to a BNC is a form-meaning pairing in itself, and the zero form of the construction carries a grammatical salience over and above the combined lexical construction type, as well as the Common Noun Construction type with which it may assimilate.

242 

 Chapter 8 The Merger Construction

­ naffected, in most cases, by the morphosyntax or the semantics of substrate or u model constructions. Furthermore, they all are seen to overlap in some way with replica language systems, but not to share the same distributional limitations of the equivalent constructions in their replica language models. The CS construc­ tion in SCE is generalised far beyond its limitations in other varieties of English, to a point where it appears can no longer be alternatively interpreted as a simple transitive construction; the use of experiential ever exaggerates the minimiser function of negative-polarity ever to apply to a category not even available in the replica language. The Past Tense Construction in SCE is generalised to apply to almost any non-progressive aspect, as if in recognition that the absence of a pro­ gressive imperfective entails the need for a perfective marker anywhere it does not occur; the Bare Noun Construction takes the form of a bare mass noun in the replica language and extends the form to apply to any noun, count, mass, defi­ nite or indefinite, as long as it is non-referential. The question is what licenses such apparent over-exploitation of existing replica language material? In order to understand the processes illustrated by such examples, it is necessary to consider the ways in which constructions merge in contact.

8.4 Previous studies of contact constructions The field of contact constructions is a field of construction theory which has barely been approached so far in the literature, to present knowledge, which is some­ what surprising in view of the challenges raised in previous mainstream theories of constructions. Croft (2001), for example, frequently states that constructions are language-specific, and Traugott (2007) maintained that they are impervious to variation and time. Goldberg (2003), as noted above, was more reticent in the matter, suggesting that two constructions may be the same across two languages if there has been contact in the histories of the languages (see 8.3.3). One of the objectives of the present volume is to reveal that the claims for language-specific constructions may have their limitations in the field of contact, and at the same time, it is necessary to describe the way in which variation in crossdialectal con­ struction inventories may be accounted for. According to Pietsch (2010: 119), the notion of interlingual identification dis­ cussed by Weinreich (1953) leads readily to the positing of a construction as an entity of mental convergence in the mind of a bilingual speaker, whose efforts to reach assimilation create perceived analogies and cognitive links between construction schemata in both languages. Pietsch uses earlier studies relating to retentionist theories of contact-induced grammaticalisation (e.g., Pietsch 2009) to explain the presence of three construction types in Hiberno (Irish) English. At



8.4 Previous studies of contact constructions 

 243

the basis of his claims is that despite the apparent surface structural differences between the Irish substratum constructions and the English constructions used at the time of the development of the contact variety (between the 17th and the 19th centuries), the conceptual similarities in terms of the schematic source con­ structions that were employed to express a range of functions were surprisingly close. He uses three case studies to demonstrate the construction-based similari­ ties, e.g., the Prepositional Aspect Construction:90 (248) She is after coming home; I am upon/for/about going; the Medial Object Perfect construction, e.g., (249) I have my dinner eaten and another construction in which a subject pronoun appears in a non-finite con­ struction, e.g.,: (250)  James came over and borrowed 3 pounds of [sic.] me and he going to the ­diggings (‘ ... while he was going ...’) (examples from Pietsch 2010). Pietsch also notes that (250) also provides an example of the subordinating use of the conjunction and, another frequentlyused Hiberno-English construction. In all three cases, there is a clear correspondence with a construction already available in the British English of contact at the time, which Pietsch believes has been co-opted to serve a replicating function in the model language, Irish, and then to have become grammaticalised further than it was in British English due to the prevailing influence of the Irish substratum. Surviving (random inter­ net) examples from present-day British English include the following (Pietsch 2010: 12): (251) If there’s a better way to obtain this info, then I’m for trying something else. The Medial Object Perfect survives today as a less grammaticalised construction known as the conclusive perfect (cited in Kirchner 1952; and Brinton 1994): (252) I have it all figured out. The Subject Pronoun of Non-finite Constructions can also still be found in British English today, classified as the nominative absolutive construction (Pietsch 2010: 22): (253) He being your best friend, you should trust him.

90 The construction types are labelled with upper case initials here for greater clarification.

244 

 Chapter 8 The Merger Construction

Pietsch’s construction types build a case for a merger construction of one kind, as they are formed on the premise that a small number of relatively marginal construction types were co-opted by the bilingual speakers and identified as representative of an English schema in use at the time. The process is one of formal assimilation to the English pattern, rather than formal replication. The contact machinery focuses more on the replication of a construction at the level of its semantics and function; e.g., in the Medial Object Perfect, there is a morphosyntactic difference between the way in which the perfect is expressed in Irish, and that of the Medial Object Perfect of British English at the time, which was less frequently used than the auxiliary perfect, according to Pietsch (2010). However, the semantics are noted as of matching identity in the use of a possessive schema of grammaticalisation (Heine 1997), in either language. The possessive schema in Irish, which falls into one of Heine’s (1997) Location schemas of expressing possession, is articulated in the following example (Pietsch 2010: 18): (254) tá an bád diolta is the boat sold ‘I have sold the boat’

agam at.me

The source schema, which generalises as ‘X is at Y’ is crosslinguistically common, being found in languages such as Russian and Turkish according to Heine (1997), and conveys the same message as ‘Y has X’, the Action Schema in languages using a verb of possession. What such parallels implicate is that even though lexical source schema constructions in the grammaticalisation of possession may differ from one language to another, in a situation such as this, the gammaticalisation of the medial object perfect is based on the identification of a conceptual means of expressing possession across both languages in contact. Thus, in such shared construction-types, the pivot-matching can take place at any semantic level; in the present case, the most schematic possible. Whether such examples can be used to resolve the problems at hand in the present study is another matter. It is clear from the research on contact transfer and change that meaning equivalence plays an inestimable role in the creation of the replica language. In the case of Hiberno English, and the functional ana­ logies cultivated in the bilingual’s contact language, the speaker’s needs to seek out an interlingual means of identifiying a model language formula are clearly satisfied in the grammaticalisation of features retained from the point of histo­ rical contact. The studies deal with model forms which are different from the replica forms, and replica forms which can be traced back to a model form in the substrate on the basis of construction semantics. Pietsch’s Medial Object Perfect



8.4 Previous studies of contact constructions 

 245

construction descriptions are based on a semantic identification between the Irish passival perfect construction and the Hiberno-English construction, which is partially identical to the construction found in earlier stages of British English, the conclusive perfect. Partial identity is based on sharing at a higher schematic level of semantics in the two constructions, the level of possession semantics, and grammaticalisation proceeds from that pivotal point onwards. However, in the present volume as noted earlier, not all of the case studies deal with features of transferral which can be traced to a model construction in the wide range of substrate languages. The past tense construction has no equivalent in the langu­ ages originally spoken in Singapore, for example, and the reason for the curious occurrences of PFP present habitual and performative uses discussed in Chapter 6 must therefore be explained with reference to the replica language alone. The problem of which languages to attribute contact constructions to is approa­ ched in the work of Höder (2014) in his categorisations of contact constructions as diasystemic construction grammars. In the study he focuses on the contact dia­ lects of Low German-High German bilinguals living in Northern Germany, mostly around Hamburg, Bremen and northern Lower Saxony. The High German acqui­ red by the speakers of Low German around the middle of the 20th century is not the standard High German spoken elsewhere, but a variety known as Northern High German, having a somewhat lower social prestige than standard German. Höder (2014) claims that it represents an expansion of the diasystemic connec­ tions between Low German and High German. Höder (2014: 141) also resorts to Weinreich’s (1964) notion of interlingual identification to explain the construction of such diasystems, a term first used by Weinreich (1954), while at the same time admitting that the establishment of diasystems is arbitrary to some extent since it is not always possible to perceive in a contact situation from which language a construction originated. Höder’s approach to contact constructions differs considerably from Pietsch’s. In the first instance, not only morphosyntactic phenomena but phonological con­ structions are treated as part of a diasystem; for example, the diaconstruction /h_s/ ‘house’ which has two phonological realisations, according to dialect, one with a diphthong and the other using a monophthongised variant. The phono­ logically-headed diaconstruction, furthermore, contains information relevant to the nature of the functions served by the form /h_s/, such as Noun, Affix, in the case of its plural form, as well as a phonological description of the pluralisation of the noun (Noun + Umlaut-/әr/), and a semantic description appropriate to the originally phonologically-headed construction form, such as referent, plural (2014: 142), and more. Pietsch (2010), on the other hand, uses the contact construc­ tion formulae only to describe differences in semantic content which are crucial to

246 

 Chapter 8 The Merger Construction

his hypothesis of semantic integration proposed explaining the presence of vari­ ation in Hiberno-English. Another construction model of Höder (2014) is the diaconstruction [DEM­ PART1PREP2], a construction type which in standard German is manifested as a unified phonological word, a demonstrative pro-particle and a preposition, e.g., darüber. In Low German the demonstrative particle and the preposition are found in discontinuous positions in the clause, not always adjacent to each other, and when they are adjacent, pronunciation is not continuous. Thus, the following differences may appear (Höder 2014: 144): (255) Standard German a. Darüber DEMPARTabout Low German

müsst must.IND.PRS.2PL

b. Dor mööt ji you DEMPART must-PRS.PL   ‘You have to think about that’

ihr you

över about

nachdenken think nadenken think

The Low German pattern is copied into the Northern High German dialect, using the lexicon of standard German (Höder 2014: 146): Northern High German: c. Da müsst DEMPART must-PRS.2PL ‘You have to think about that’

ihr you

über about

nachdenken think

In (255c) there is a clear case of construction-convergence: the Northern High German construction bears the syntax of one dialect (Low German) and the lexicon of another (standard German). The question again arises whether it is identified as a standard German construction, or a Low German one. The process resembles very closely the pattern-copying strategies outlined by Matras and Sakel (2007), but which Höder terms a syntactic calque, and which he maps in a model demons­ trating the way in which the syntactic schema is generalised over two distinct construction types. Such a generalisation would constitute a type of construction merger of a kind, though it is not clear from the model exactly what is being shared and what is distinct in the three dialects concerned. Some of the diaconstructions are represented as phonologically-headed, or at least conveying phonological information (as well as functional and syntactic information), while others appear to be syntactically-headed, and convey only syntactic information (e.g., the Aspec­ tual Pseudo-Coordination Construction) with dialectal lexical inserts at the lowest hierarchical level. Unlike Pietsch’s (2010) more simplified, linear models, they are



8.5 The Merger Construction Model 

 247

composed of a complex informational machinery representing what is considered to be distinctive in the dialects. But it is not entirely clear in this model at which level are found the semantic pivots of interlingual identification mentioned by Höder as responsible for the types of strategies used by the bilingual speakers in contact in replicating the structures of their model languages.

8.5 The Merger Construction Model The two contact construction models described above are similar in the way in which they both deal with contact material in which a syntactic pattern is copied from the model language into the replica language using material from the replica language; in both cases, resembling instances of Matras and Sakel’s (2007) PAT strategy. In Pietsch (2010) there is in addition, a clear case for contactinduced grammaticalisation, as described by Heine and Kuteva (2003, 2005). The processes do not relate to arbitrary relexification, since the linguistic mate­ rial selected from the replica language is crosslinguistically a common source, in many cases, as in the possessive schemas for expressing perfect aspect in Hiberno English. In the case of the German dialects, it is unclear whether or not grammaticalisation is involved – more diachronic evidence would be needed to verify this. The two cases are not apparent as cases of equivalence-setting, as there is a semantic justification for the selection of replica material in the Hiberno-English case, and in the German dialects clearly a loan translation. The role of Systemic Transfer is also not relevant to the two cases of construc­ tion-sharing described above. Thus, the two theories of contact-constructions so far have been created with a view to either (a) explaining replication of model language transferrals in what is not obviously a loan translation, in Hiberno-English; or (b) simply explaining the replication of a loan translation as a construction-based entity (in the case of Northern High German). Although elegantly applied to their respec­ tive data sources, neither of the two models can be appropriated to the present four cases, for which the problems of explanation are quite different: (i) explai­ ning the interlingual identification which resulted in the selection of linguistic material from the replica language to express functions not transferred from the model language (as in the case studies in Chapters. 4, 6, and to some extent, 7), or (ii) explaining the interlingual identification motivating selection of replica language material to grammaticalise an apparently arbitrary equivalence (as in Chapter 5, where only the function is transferred). In the present four cases there is not necessarily a correspondence with any formal or semantic model at all, and as noted above, some cases may be considered to be innovations in contact.

248 

 Chapter 8 The Merger Construction

But what requires explanation is not just the selection of linguistic material from the pre-contact replica language (English) but the selection of entire construc­ tions (form-meaning pairs) in the pre-contact replica language to be exploited in the contact replica language (Singlish) in order to frame new constructions which have never existed in the pre-contact replica language, or to extend the distributional constraints of existing constructions in the pre-contact replica language. The result is an expansion of the functional inventory of the replica language, but without necessarily increasing the construction inventory beyond that of the pre-contact replica language. The hypothesis of the Merger Const­ ruction Model allows for the possibility of semantic continuity at the point of contact, and thus of construction-sharing to explain the anomalies illustrated in the previous chapters. It will also ensure the parsimony of a single constructicon in a mixed language situation in which the sub-parts of the single construction frame are derived from different source languages. We shall consider each of the case studies in turn.

8.5.1 The Transitive Merger-Construction It should be noted that the term transitive construction has been applied to the case of the Conventionalised Scenario (CS) construction mentioned earlier, since it is the transitive construction that applies semantic constraints on what may be generalised as a CS. It was seen in examples such as (243–245) that the CS is restricted in its distribution to domains which can be alternatively construed as transitive constructions, and (243–245) violate the restrictions, which is why they are unacceptable as CSs in other varieties of English. The historical evolution of the transitive construction, as discussed in Chapter 4, suggests that the transitive construction (TC) has extended its distribution to include not just constructions with agents performing actions on patients, but almost any semantic phenomena which can be assimilated. The CS construction therefore assimilates to a TC by the fact that not only direct causativity over transitive events is expressed in its semantic definitions, but even indirect causativity, as an implicit extension of the semantics of a TC. However, in the process of its generalisation, the TC is affected by lexical constraints associated with its source construction prototype: those of direct causativity, where a human agent assumes control over an action affecting an object (see, e.g., Hopper and Thompson 1980). For this reason, only CSs which are potentially ambiguous between direct causativity and indirect cau­ sativity at the same time can be admitted into the construction frame of a CS. The SCE CS then exploits the fact that indirect causativity can now be condensed into the syntax of a transitive construction, and extends the new distribution



8.5 The Merger Construction Model 

 249

unconstrainedly, ignoring the mediating influence of lexical retention from the Lexical Prototype Transitive Construction, its source construction. Since transiti­ vity is often expressed indirectly in the model languages (e.g., the bǎ construction in Mandarin), the morphosyntactic parallels with a CS construction, as noted in 8.3.1, are not always present as a model, though there are conceptual parallels. The bǎ construction also condenses and compacts the notion of reflexive cau­ sativity in its semantics, in which transitivity emerges as the bleached meaning of a former serial-verb construction in earlier diachronic periods (as discussed above). That is, the causative serial construction semantics are also being ignored in the Mandarin construction. Because of the dubious likelihood of transferral from the substrate models of contact emphasised in the present account, in the Merger Construction Models to be described below, the pivot-matching strategy, unlike that of other accounts of contact and interlingual identification, is viewed rather as a reanalysis or reinter­ pretation of existing functions in the pre-contact replica language, i.e., (other vari­ eties of) English. The Merger Construction is thus regarded as a problem-solving device rather than a descriptive device, consisting of a merger between the precontact replica language and the contact replica language. In the present study, the pre-contact replica language has been equated with SSE (Standard Singapore English), since this sub-variety is in constant contact with present-day SCE, as mentioned in Chapter 2. SSE, as noted in more recent diglossic studies such as Alsagoff (2010) (see Chapter 2), is also shaped by constant international contact in a global environment now dominated by vastly-expanded means of international communication by comparison to those of the mid-20th century, and the influ­ ences on what was once the pre-replica English contact language are much more diverse than in the early stages of contact. Thus, the material selected for reinter­ pretation in a Merger Construction Model is now likely to reflect a more neutral, international usage of English. The converging grammars are then the converging grammars of SSE and SCE, and it will be seen that the classical model of a construction as a formmeaning pairing can be ranged across both grammars. The local contact langu­ ages such as Mandarin Chinese or Malay, and the original substrates which also include Hokkien and other southern Chinese dialects, are of minimal importance in such a model. The merger is a merger of two sub-varieties of English spoken in Singapore, SSE and SCE, and the SCE construction variant is expressed as a reformulation of the same construction in SSE, rather than a completely new constructicon. The point of interlingual identification or pivot-matching is isolated as part of the meaning level of the construction inventories of both dialects, in the assumption of a construction as being a fusion of not just form and meaning, but also function. In accordance with what was discussed in

250 

 Chapter 8 The Merger Construction

Chapter 3.2.4, the model of constructions most similar to the present one is that of Langacker (2005b), in which the integrative elements are not form and meaning, but phonology and meaning. However, the model will be expanded to illustrate that the same phonology across two dialects in contact can produce different but related functions, using the same syntactic form, illustrated in the const­ ruction composition below. Phonological sharing is thus a fait accompli, which varies according to the specificity of the lexical needs. The shared meaning element is necessarily a polysemy, of which one sense element is selected to provide semantic continuity across contact, to convey the pivot-matching link that underlies the merger of the two constructions. In Figure 8, in the Transitive Merger-Construction, the pivot-matching link connecting the constructions in the two dialects is clearly that of causativity, as illustrated below:

Form

Function

SSE

SCE

NP V NP

NP V NP

↕

↕

↕

Indirect causatives (CSs)

Transitivity

Meaning

↕

SSE/SCE Causativity

Fig. 8: The Transitive Merger-Construction.

As seen in Figure 8, while the formal pole of a construction composition is inevi­ tably identical in both dialects in contact, it has no immediately obvious relation with any formal structure in the substrate or model languages of SCE, and the formal signifia are taken on face value as equivalent if it is logically assumed that the selection of such formal elements for expressing the functions required in SCE was not without plausible reason. Thus, for the transitive construction, its distri­ bution over indirect causative functions as well as the direct causative functions of a transitive construction in SSE is evidence of its expansion, still preserving at



8.5 The Merger Construction Model 

 251

a more schematic level the pivotal semantics of inherent causativity.91 The SCE construction shares the construction on the basis of reinterpreting the semantics of the TC in which the structure used for expressing direct causativity extends to indirect causativity as well. Transitivity has not been included as a function of the same syntactic form in SCE, due to the prevailing influence of topic-prominence, in which the subject role is not grammaticalised to the same extent as in ordinary transitive constructions, but even if it were, it would not affect the Merger Const­ ruction distributional constraints in any way. It could be argued that not all transitive constructions in today’s English contain causativity as part of their meaning pole (e.g., the examples: The tent sleeps six and The hotel forbids dogs discussed in Chapter 3). But as noted in that chapter, the present-day transitive construction is a product of the generalisation of the construction semantics to the point of ‘bleaching’, where only the word order of the syntax remains, regardless of the semantics of the former case roles that it replaced. It is not necessarily the case that Singlish adopts the syntactic frame of the construction alone, since the same extension of the CS to a transitive framework is not unconditional in the standard variety. The use of the transitive construction in the extension of the CS construction in Singlish is motivated by the need to prioritise the action of the event over its result, which is usually priori­ tised in the counterpart indirect, causative-resultative constructions. This in turn is enabled by the identification of causativity as a conceptual element, a semantic pivot linking the two construction types.

8.5.2 The Experiential ever Merger-Construction The establishment of ever as a construction-type, as noted above, needs less ­introduction than the SCE Conventionalised Scenario, as unlike the latter, ever does not refer to entire clausal material, but only a single adverb as its formmeaning pairing. In addition, there is no ambiguity with the syntactic form of any other construction-type, since it is found in different linguistic environments from those of ever in standard and other varieties of English. However, on the other

91 This is not to suggest, of course, that CSs are not found in standard varieties as well – the data in Chapter 4 shows that they are. However, the expansion of the transitive construction from expressing direct causativity to indirect causativity in the CS construction is not as generalised in SSE as it is in SCE. The adversative CS construction in standard varieties (e.g., I cut my hand) is not without causativity either – it differs from other CSs only in the fact that the causativity is unintentional.

252 

 Chapter 8 The Merger Construction

hand, it shares similarities with the transitive construction in that it appears to be reinterpreting or reanalysing a function of a construction already in existence in SSE, towards a wider distributional range of uses. It was mentioned above that it might be analysed as an equivalence form, according to Heine (2013) and Heine and Kuteva (2005); as such it conveys a substrate model language function but with no transferral of any morphosyntactic material, nor of any lexical meaning, in this case. It was simply selected to serve a functional purpose in the model language which does not exist in the pre-contact replica language. There is little point, nevertheless, in describing the equivalent function in the model language as part of a Merger Construction Model for experiential ever. The transfer is functional only and provides no insight as to the selection of the form from the local standard language. On the other hand, evidence from Modern Dutch suggests that the function to which negative polarity ever has been exten­ ded in expressing experiential aspect may have its counterparts in other langu­ ages, and thus the possibility of a universal grammaticalisation path might not be ruled out altogether (as noted earlier, Dutch, it must be remembered, is not a contact language). An overview of the grammaticalisation paths of other negative polarity items might be necessary in order to determine whether such a shift is a universal crosslinguistic phenomenon or not. For the moment, if the use of the adverb is viewed as a reanalysis of the standard English adverb ever, a minimi­ sing marker of emphatic possibility restricted to negative-polarity contexts, it may by logical equivalences and interdefinitions become reinterpreted as a marker of minimal necessity, so that the meanings ‘at least one (possible) time’ shade into ‘at least one (necessary) time’ (experiential functions) in the context of negation. This analysis has been used to describe the mechanics of a semantic back-forma­ tion in Chapter 5, and has been parallelled in the interchange of modal meanings of necessity with those of possibility in the history of the modal verbs in English and in German (see, e.g., Diewald 1999, and Goossens 1987). But never co-exists happily with negative-polarity ever in both dialects, and while such logical equivalences may seem obvious in explaining a back-formed shift from never to experiential ever, they do not explain the reason that negativepolarity ever can continue to be used by SCE speakers, polysemously, alongside its experiential function in the aspectual system. It may appear that the alter­ native functions of ever will restrict the creation of a new function in SCE, quite unrelated to negative polarity. But if we consider again the observations made by ­Chappell (2001) that one of the crucial defining characteristics of experiential aspect markers in the Chinese languages most likely to have formed the substrate for SCE is the factor of repeatability, it is apparent that such a semantic feature may indeed serve as the pivot-matching point on which speakers of SCE formed analo­ gous functional links with the meanings of the experiential aspect constructions

8.5 The Merger Construction Model 



 253

in the Chinese dialects. It is hypothesised in Chapter 5 that the semantic feature of repeatability was most likely a pragmatic inference arising from the gravita­ tion of ever towards negative-polarity contexts, in which it serves the function of a minimiser, and hence is lower-bounded on a scalar dimension. It is this semantic feature which has licensed the back-formation of never, to create a pivot-matching process enabling the convergence of the SCE and SSE functions of ever, as maintai­ ned in Chapter 5. The semantic link of repeatability is thus shown to provide con­ tinuity and yield the Experiential ever Merger-Construction, as shown in Figure 9.

Form

Function

Meaning

SSE

SCE

adverb

adverb

↕

↕

↕

minimising quantifier

↕

experiential aspect

SSE/SCE repeatability

Fig. 9: The Experiential ever Merger-Construction.

Note that both the minimising quantifier functions of ever and the experiential aspect functions share the same inferential meanings of repeatability. The two functionally-differentiated constructions are thus seen as one single Merger ­Construction in the contact of the two dialects, and the functions of the model languages are satisfied accordingly.

8.5.3 The Past Tense Merger-Construction The use of the past tense in the Past-for-Present functions in Chapter 6 places quite a different viewpoint on the hypothesis of a Merger Construction. Rather than being motivated by a functional correspondence in the model languages, the PFP dis­ cussed in Chapter 6 has no functional equivalence in any of the contact langua­ ges. Such a situation would possibly invoke the argument that the PFP should be

254 

 Chapter 8 The Merger Construction

considered a construction type distinct from any other Past Tense Construction. And yet we are still faced with the problem of accounting for the morphosyntactic repli­ cation of lexifier material. It was claimed in Ho and Platt (1993), as noted earlier, that the use of the past tense was restricted to lexically perfective verb types and grammatically perfective aspectual contexts (e.g., the past habitual was unmarked for past). However, such uses are only found where there is past time reference in any case. The enigma of the PFP cases described above is that they are not replica­ ting anything and do not form part of the subsystem of tense-marking in standard varieties nor do they replicate in parallel the aspectual system of the model langua­ ges. As noted earlier, they can only be ascribed to an innovation in contact. On the other hand, it is necessary to explain why such an innovation emerged. The use of the progressive aspect in English is a marked category, while the nonprogressive aspects are unmarked in the simple form of the verb, present or past. The simple form of the present-tense verb is distributed across a range of functions, including habituals, performatives, play-by-play narrative uses (e.g., as in sports commentaries and cooking demonstrations), and planned future uses, as noted in Chapter 6. It was hypothesised that SCE speakers may feel a need to mark such functions in opposition to the progressive, to create the sense of an action viewed as completed but with relevance extending into the moment of speaking; i.e., it can be functioning to mark what could be described as an implicational perfective function. Such contexts are seen as perfective by default, since they are not imper­ fective. However, marking them as perfective does not replicate in any way the aspectual system of the substrate languages on which SCE was modelled, and in this way the extension to present-tense contexts must be considered an innovation. It could be claimed that such phenomena are the result of mass hypercorrection on the part of speakers who do not universally mark verbs for past tense, as suggested by Bao Zhiming (p.c.); however, the frequency and the variety of contexts in which the past tense is found in SCE suggest that whether or not the feature started out as a hyper-correction, it has spread beyond the idiolectal distribution which would mark it as such initially, and therefore requires a plau­ sible functional explanation for its frequent presence in the system. The explana­ tion offered in Chapter 6, that it marks precedence of an action or event licenses a pivot-matching strategy involving the semantics of the past tense in the standard language of contact. However, precedence implies the likelihood of implications for the future, as well as potential recurrence; it is for this reason that the function has been termed an implicational perfective. The most likely functional domain in which such a pivot-matching strategy may have taken place is in the domain of present habituals, which can be regarded as assertable on the basis of evidence of repeated past occurrences. The Past Tense Merger Construction can therefore be illustrated as in Figure 10.

8.5 The Merger Construction Model 



Form

Function

SSE

SCE

past tense

past tense

↕

↕

↕

temporal reference

Meaning

 255

↕

(implicational) perfectivity

SSE/SCE precedence

Fig. 10: The Past Tense Merger-Construction.

Thus, in Figure 10, it can be seen that a marker of temporal reference in the standard sub-variety has been exploited to serve the function of marking the perfective present aspect as an extension of its earlier use in marking perfectives in the past, in SCE, and that the justification for interlingual identification was made on the basis of the semantic feature of precedence, common to both functions. In expressing a present habitual past tense offers evidential justification in the form of precedent events, just as for single-event past tense expressions, often indicated by the perfective marker already. The function of marking past temporal reference with past tense verbs refer­ ring to single events entails the same precedence of action. The only problem may lie in the use of the past tense on stative verbs, which do not so readily entail an eventprecedence associated with non-statives. However, statives are not often marked for past in SCE, as noted in Chapter 6, since they are non-punctual and imperfective in their lexical semantics. Figure 10 illustrates, then, that speakers of SCE have expan­ ded and analogised the Past Tense Construction which was restricted to punctual aspect marking to accommodate present habituals, anteriors and performatives on the basis of the semantic factor of event-precedence alone.

8.5.4 The Bare Noun Merger-Construction The last remaining construction, the BNC, probably presents the most lucid case for a Merger Construction. It was argued in Chapter 7 and above that the Bare Noun Construction actually serves a function in its own right, as a zero-marked construction type, found in nouns which are non-specific and thus unmarked for

256 

 Chapter 8 The Merger Construction

number, singular or plural, for example (179b) in Chapter 7, from Alsagoff and Ho (1998), repeated below as (256): (256) She got car or not? The same zero-marked means of expressing the absence of individuation in the standard sub-variety, in the mass noun type, is recruited to justify the absence of quantification in count nouns in the contact variety, SCE. In this way, as in all the other cases discussed, the contact sub-variety analogises a new functional appli­ cation for an observed formal criterion in the syntax of the standard sub-variety, reanalysing it in accordance with the construction types of the substrates. As with the other three cases, in order for such an analogy to be made, spea­ kers of SCE must search for a linking semantic domain accessible to both dialects, at which the two construction meanings may be perceived to merge. Since this type of reanalysis may take place by a process of deductive inferencing from form to function, the reinterpretation requires that the mass noun semantics share some common semantic feature with the non-referential count noun. It is hypo­ thesized that the semantic level relevant to the bare nominal is one expressing unboundedness of the referent, both in the case of a mass noun and in the case of a non-referential count noun in SCE. SSE, on the one hand, marks the conceptual (configurational) unboundedness of a mass term referent, while SCE marks the quantificational unboundedness of the same count noun referent. Thus, unboun­ ded quantification and unbounded form are seen as analogically related through the same morphosyntactic representation, a Bare Noun Merger-Construction, expressed as for the other three cases, over three interconnected levels of const­ ruction integrity, form, function and meaning, as illustrated in Figure 11:

Form

Function

SSE

SCE

bare NP

bare NP

↕

↕

↕

indefinite mass noun

Meaning

Fig. 11: The Bare Noun Merger-Construction.

↕

non-specific noun

SSE/SCE unboundedness



8.5 The Merger Construction Model 

 257

From the point of view of constructions and coercion, then, the Bare Noun Merger-Construction, consisting of two contributing constructions, the mass-noun construction and the non-specific noun construction, parallels the processes of type-shifting referred to in Chapter 7, in which the argument of the verb which is not marked lexically or grammatically for reference (as are pronouns), shifts categorial status from a count to a mass noun. The merger construction supports both constructional analyses in the sharing of a substructural level of schematic meaning which is common to both SSE mass nouns and SCE non-specific nouns: mass nouns are by default non-referential when occurring as bare verb arguments, just like non-specific nouns, which also occur as bare verb arguments. The two contributing constructions of bare nominal morphosyntactic representation are linked by a common, intra-constructional layer, a shared level of nominal meaning of unboundedness (of the lexical input) which is manifested functionally as a mass noun in one dialect and simply a non-referential one in another. The reason that such a construal is possible is that the bare mass noun type actually possesses as part of its inherent semantics a feature [-boundedness], emerging in opposition to the semantics of the indefinite article construction which has generalised the earlier referentiality distinctions of its complement to express meanings of poten­ tial, non-referential individuation as well. In the bare plural, for example (209) in Chapter 7, repeated below as (257): (257) Electronic dictionary is not allowed [in exams] the potential is not for individuation, but for quantification, as the number of electronic dictionaries referred to is either not known or irrelevant. Potential plural quantification is expressed in the Plural Construction as discussed in Chapter 7, in which a single morpheme, the affix -s serves to mark both known and unknown quantification in SSE and other standard dialects. However, the BNC exploits the form of a mass noun to express the distinction between potential and actual quantification in SCE, where configurational unboundedness is ana­ logised to quantificational unboundedness as well. In addition, the studies discussed in Chapter 7 on count-mass coercion in constructions are undermined by contact data which reveal that there is semantic unification at a more schematic level, and that coercion is relevant only at the level of a superficial categorial discrepancy. In terms of the requirements of the Verb Phrase construction for mass noun lexical arguments discussed in Chapter 7, the semantics of a mass noun are the same as those of a bare, non-specific nominal in SCE. Although the (bare) mass noun type functions as a lexical entry into the Verb Phrase construction, it might also be noted that its form, a bare nominal, has grammatical significance in itself, in marking the absence of refe­ rentiality, and this grammatical significance applies across dialects, and in SCE in which referentiality distinctions are ranked as more salient than count-mass

258 

 Chapter 8 The Merger Construction

distinctions. Therefore the SCE data provides an exception to the coercion ruling: the bare nominal need not only be a mass noun in order to function as the lexical argument of the Verb Phrase construction.

8.6 Summary The present chapter has endeavoured to reveal a number of significant aspects of the study of contact linguistics and mechanisms of transfer across contact. It has reviewed a selection of studies of contact transfer, and from what they demons­ trate, it appears that, with the exception of grammaticalisation theory, there are few differences in the ways in which various researchers approach the question of transfer of grammatical or morphosyntactic material across contact. Perhaps the most concise means of describing contact transfer is found in Matras and Sakels’ (2007) account of pattern and material copying, and the implications of an accom­ panying pivot from which matches may be made in material copying, a pivot which can refer to any semantic or morphosyntactic form, construction or otherwise. Also discussed are the somewhat vaguer notions of convergence, and isomorphism, and it is felt that none of these processes are of relevance in the present cases, except perhaps as an extended theory, in which convergence includes not only conver­ gence with the substrate, but with the local standard (or lexifier) material as well. The notion of a pivot-match has been extended to apply to such cases also. The pos­ sibility of relexification is difficult to uphold in cases in which the transfer from the model language is not transparent: even in the last case (8.5.4), the level of transfer of a zero-construction is questionable, especially given the fact that the same refe­ rentiality distinctions are found in creoles, as discussed in Chapter 7. The only pos­ sible category of transfer might be the absence of number-marking, but one may ask, can an absent category be transferred, and then relexified? There is no clear parallel with any of Lefebvre’s (1998, 2001) cases from Haitian Creole. Bao (2005, 2009, 2010) has extended the relexification theory to one of total systemic trans­ fer, but does not account for the difficulties of interlingual identification associated with items that are not so obviously perceived as associated with grammatical fea­ tures of Chinese, for example, the Past Tense Construction and its PFP functions. Bao (2010: 801) uses the term misanalysis to refer to the way in which the creator-developers of SCE have reinterpreted the auxiliary get as got, by mistake as it were, in constructions such as (258) I got wash my hands (‘I have washed my hands’). However, it would be unjustified to suggest that spea­ kers of SCE were simply erroneously reinterpreting similar construction-types in

8.6 Summary  

 259

SSE on the basis of the ‘correct’ functions to which they are assigned in SSE. It is clear that the construction types developed by analogy and the sharing of an intra-construction level of meaning goes further than that, in the establishment of a system which can economise, exploiting the inventory of constructions and accommodating alternative functions and shared polysemies for the same formal structure. As such, the Merger Construction cuts across the dialect distinctions which may be laid open to question in an ideal construction grammar. In this regard, it is seen as prefereable to describe the two contributing const­ ructions to a Merger Construction as co-existing within the same functional field, as alternates available to speakers to select according to their own needs. Given such situations, what appears like a Merger Construction may be better described in SCE as a convertible construction, i.e., a single construction type with two alter­ nating facets, which can be exploited for more than one purpose. This type would not be unlike what Cappelle (2006) has termed allo-constructions. ­Cappelle argues against the necessity for syntactic structures to be always derived from one another; at the same time, he questions the likelihood of a construction grammar that rigorously insists on the absolute autonomy of each individual construction. He maintains that the possibility of the two construction types in examples such as She took away my breath and She took my breath away being stored twice in the lexicon without representation at a level of semantic identicality “... lacks psychological plausibility” (2006: 13). However, Cappelle’s (2006) allo-constructions are distinct from the present cases because, although they allow variant realisations of a single construction type, the composition also requires an element of under-specification (of word order, in the case of his verb-particle constructions). They also differ from the present situation of Merger Constructions in that his constructions refer first, to alternative surface syntactic representations (they are two manifestations of different word orders), and second, they are not the result of contact conver­ gence. There is no convergence of grammars, no fusion of any semantic subst­ ratum. In the case of the Transitive Merger-Construction, for example, there is simply a common semantic element of causativity by which the two construc­ tion types, the CS construction and the transitive, may be identified to share the same syntax. In the Experiential ever Merger-Construction, although the process of derivation may be via logical equivalences, the selection of the formal means of expressing experiential aspect was made on the basis of a common seman­ tic feature of repeatability. In the Past Tense Merger-Construction, the common semantic feature linking both the SSE functions and the SCE extended functions of PFPs was the feature of precedence of action. And in the Bare Noun MergerConstruction, it was a feature of unboundedness that was isolated by speakers of SCE in order to build a common construction-type in contact. That is, one

260 

 Chapter 8 The Merger Construction

semantic feature of each of the construction-types has been extracted and fused into the pivot creating a relation with an existing construction-type in the original lexifier language or in SSE, without increasing the number of constructions in the replica language inventory. This ensures that there is only one replica language, whatever functions its forms may involve, and that the constructions of the local standard variety, in terms of form-function-meaning correspondences, are shared with those of the local replica language where there is formal identity. Similarly, the contact constructions of Pietsch (2010) and Höder (2014) are not appropriate to the present model either, as they account only for the merger of grammatical or pattern copying across contact, from the model language to the replica language. As such, they only describe, or amplify in more precise terms what has been observed as deriving from the substratum languages in either case. The nature of substratum transfer can be described in multifarious fashion, and there is no limit to the different ways of explaining how a feature from a model language was translated to become a feature in a replica language. Such accounts are equally important in their assumptions of a unitary construction model for the types of data they describe, but they rarely explain the selection basis for interlingual identification, often more difficult to determine, in the absence of clear functional parallels with substrate structures. The replica gram­ maticalisation account of Heine and Kuteva (2005) is insightful, as it reveals the ways in which copying may follow the same diachronic stages as the model lan­ guage has followed. But that is not the entire story: there are features of contact which appear to be renalyses of standard contact features and yet there is much less transfer evidence on which to base their appearances. The hypothesis of the Merger Construction has taken a common factor resorted to in the transfer accounts, that of interlingual identification, or semantic links, pivot matches, and common polysemies, as the most important feature in resolving perceived analogical associations with lexifier functions. In this way, it readily accounts for the difficulties of accommodating diverse types of dialectal variation in contact, and offers a parsimonious explanation for the presence of contact constructions in any situation.

Chapter 9  Concluding remarks In his colourful account of the development of Singlish as an Asian English Variety (AEV), Ansaldo (2010) adheres to the thesis that although English was one of the languages in the multilingual ecology where Singlish first emerged, it was by no means a significant player in contributing to the typological basis of the contact language grammar. Rather, it was the Sinitic languages and Malay, he felt, that provided the essential structural dynamics upon which Singlish was born. In explaining his position, Ansaldo uses the examples of structural features such as topic‑­prominence, reduplication, and zero‑­copula to illustrate the way in which a newly‑­developing contact language selects from a feature pool in the linguistic environment to structure the new language. Because of the presence of a feature pool, it is always the features that have the highest type‑­frequency in the multilin‑ gual context of communication that win out in the end (2010: 508) – both Sinitic and Malay, for example, have topic‑­prominence, and thus this feature has a higher frequency of usage in the environment. The role of English itself, he believes, is therefore minimal according to this account, though not excluded altogether. The use of substrate and adstrate grammatical strategies to restructure a new language of contact is not surprising, and clearly in the case of Singlish, the most prominent grammatical features are also the most prominent features of the lan‑ guages spoken alongside Singlish, as Ansaldo points out. However, there is still a need to account for features of the grammar of contact which are not so readily transparent in the grammars of the substrate and adstrate languages. Such needs have been one of the main objectives of the present volume, since, although they appear to have been motivated in some cases by substrate or local contact influen‑ ces, their appearance in the grammar of Singlish is not always as predictable as a substrate account would permit. Furthermore, it throws into disarray any attempt to order the grammatical constructions of the newly emerging contact language on the basis of the construction bias of another language of contact. It is for this reason that the volume has collected a number of cases under a unified construction‑­based approach: to explain a series of unsystematic features of Singlish that appear not be readily accountable by any traditional methodology of contact linguistics at all. It may be argued that there are, nevertheless, associations to be made between Singlish, the replica language, and the (at least 10) southern Sinitic or Malay languages such as Hokkien, Cantonese, Hakka and Teochew which have provided Singlish with the main framework of its construction inventory. At the same time, as discussed in Chapter 8, any such associations are obscured

262 

 Chapter 9 Concluding remarks

by the more perspicuous appearance of Merger Constructions as the results of analogies on the part of the replica language speakers in approximating the construction inventory of the local standard language, whether the former sub‑ strate languages might have played a motivating role or not. In the (undeniably) multilingual ecology of Singapore today, the selection of a particular gramma‑ tical construction is not so likely to have been influenced by earlier languages of contact which are now spoken much less frequently than at the time Singlish first started to develop, and the choices of structural replication are perhaps more limited than at that time as well. What is most visible in today’s mul‑ tilingual ecology is a more restricted feature pool from which speakers have selected constructions of transferral, and a more systematic, conventionalised contact language situation than ever before; i.e., one in which the four main, official languages become the objects of replication, rather than the languages of a colonial past. This does not entail that features of earlier grammars will not become fos‑ silised or entrenched over a period of time into the more regularised patterns of Singlish that we know today. The structuring of the constructicon of Singlish is as much a legacy of the contact of yesteryear, as it is a product of today’s language contact input. It is for this reason that the case studies in the present volume have been selected: they reflect not so much the grammars of the mul‑ titude of languages formerly spoken in the region, in the vibrant, historical melting‑­pot context of trade and rapid post‑­colonial expansion. Instead they illustrate the speaker’s desire to manipulate the current linguistic situation in accommodating Singlish within the diverse mosaic of today’s international English usage. Thus, although certain of the features discussed are unavoidably traceable to the influence of the local languages, e.g., the Bare Noun Construc‑ tion, discussed in Chapter 7, they do not replicate piecemeal any earlier subst‑ rate model constructions into the new contact language. The evidence of Merger Constructions reveals, rather, a level of compromise and equilibrium between the replica language and the local standard English, indicating the extended developmental stages of the new variety, so that as far as some constructions are concerned, the multilingual feature pool of contact selection described by Ansaldo (2009, 2010) includes English itself. In these compromise construc‑ tions, English is not just a donor language to shape the lexicon, but also a true construction source. The construction source is then reinterpreted to conform to the conceptual needs of the former substrate grammatical systems, as well as those of the local standard variety of English. The present volume has also held a theoretical responsibility in terms of what it may contribute to the field of construction‑­based approaches. In the



Concluding remarks 

 263

Introductory chapter, a number of questions were raised which formed the research rationale for its production. These included the following: If it has been demonstrated in the more fundamentalist construction literature that const‑ ructions follow precise, form‑­function compositional correspondences, with the result that it is the form of the construction type that may ultimately determine its meaning, at least at a certain level ..., then how can the mixed ­language situations [ ....] cope with a form‑dicta‑ ted construction semantics?

As noted in Chapter 3, most general theories of construction grammar posit a form‑­dictated semantics for a construction‑­type, based on the expansion of the lexicon to include idiomatic chunks which are largely undecomposable, such as kick the bucket, or the way‑­construction (as in I knitted my way across the Atlan‑ tic, Goldberg’s (1995) example). The extension of the lexicon to include such items meant that any syntactic cluster could now be considered a candidate for construction‑­hood on account of its unanalysable semantic composition. As we saw in Chapter 1 and Chapter 3, the understanding of construction semantics as based exclusively on the syntax alone is difficult to reconcile with variation‑­ based approaches to constructions: what is a construction in one dialect is not the same construction in another, though there is nothing in the syntax that can possibly indicate the presence of more than one semantic interpretation from the same syntactic string. Thus, a transitive construction in one dialect is not one in another, in which it represents an indirect causative construction, a conventio‑ nalised scenario, as illustrated in Chapter 4. The conventionalised scenario con‑ struction in Singlish, though, is generalised beyond its usual limits in standard varieties and is not restrained by its parallel interpretability as a plausible tran‑ sitive construction. Such constraints are not obviously derivable from the syntax of a construction alone, since the CS and the transitive construction share the same form. However, the transitive construction is still dictating the semantics to a certain extent; it is not entirely disassociated from a transitive construction of Standard Singapore English, in that both constructions merge to share the common semantic feature of inherent causativity. The same question could be asked of the results of the other three case studies: the ever construction discussed in Chapter 5 illustrates clearly how the form‑­dictated construction semantics can be accommodated across a shared meaning component at which the constructions of the local standard variety and those of Singlish or SCE are seen to converge. Thus, although ever is a negative‑­ polarity adverb in SSE and other varieties of English, in Singlish is extends its usage to experiential functions through the semantic pivot of the sense of repea‑ tability, implicit in the semantics of its minimiser adverbial function as well as in its experiential aspectual function. In Chapter 6, the form of the past tense

264 

 Chapter 9 Concluding remarks

construction in SCE, extended to what in SSE are considered to be situations requiring of present tense rather than past tense, functions to mark implicational perfectivity, an aspectual function similar to that formerly served by the preterite‑­ present forms of Germanic and Old English. In expressing the perfectivity of an event with potential implications for recurrence, the semantic attribute of prece‑ dence is identified in the standard usage as the pivot upon which to justify the use of a SCE past tense marker in present‑­tense habituals, performatives, and anterior clauses. The construction semantics of the SSE past tense and the SCE past tense are merged by this attribute, and the two construction types may be considered as one shared across dialects. In the same way, the Bare Noun Construction in Chapter 7 shares the same construction form component as the mass noun in the standard system which distinguishes count from mass nouns. This is made possible in that the semantic components of the two construction types are linked by the shared meaning pivot of unboundedness –­of configurational form in the mass noun and of quantifi‑ cation in the non‑­specific noun of SCE. Thus, again, the form may be plausibly perceived to dictate the meaning of the construction at a more profound level of analysis than is obvious at first glance. The Bare Noun Construction thus emerges at a more schematic hierarchical level, and the construction is considered as belonging to both dialects. Thus, it would seem that there is no ‘mixed’ language situation in the case of Singapore English, and the effects of other languages of contact spoken in the region, either in the past or currently, need not always be relevant in determining a construction inventory across contact. The evidence points to (i) a necessity to dis‑ tinguish between meaning and function in the composition of a construction; and (ii) a sharing of constructions at a higher schematic level than is usually the case in studies in which the function of a construction is not so readily discernible from the meanings associated with it. Although some of the meanings discussed are clearly, pragmatically implicit, such as the implicit sense of repeatability shared between the experiential aspect and the minimiser functions of ever, they are nevertheless conventionalised sufficiently to allow speakers of the new contact language to be able to identify them as justification on which to build the construction inventory. Ever is a Merger Construction in Singlish, regardless of its functions, and we are looking at the same construction as SSE ever, merely used in different ways, rather than a different construction for a different dialect. From such evidence we can go on to answer some of the other research ques‑ tions raised in the first chapter, including the following: [ ] the most searching question in the face of construction theory is how to describe the constructions of a language in the case of contact varieties of that language which exhibit



Concluding remarks 

 265

the syntax of one language (usually the substratum, or model language) and the lexicon of another. Do they possess only the constructions of the lexifier, or the constructions of the language which supplies the (morpho)syntax? Furthermore, how is the form‑­meaning relationship described in contact situations?

We have seen from the above examples and in Chapter 8 the way in which the form‑­meaning relationship is described in contact situations in which there is no obvious transfer of any morphosyntactic structure in to the new contact lan‑ guage; this was discussed in Chapter 8 and above as one of the reasons for the selection of case studies. The cases shown are exceptional in this way and cannot be so readily attributed to former substrate features. In a classic situation of contact, as discussed by protagonists of restructuring and relexifying theory, e.g., Lefebvre (1998, 2001) or proponents of total systemic transfer (e.g., Bao 2005), it would appear that part of the constructions of contact, the morphosyntax, is derived from the substrate, while another part of it, the lexicon, is derived from the lexifier. If the lexicon is understood to contain the majority of the meaning component of a construction, then the meaning is from one language and the morphosyntax is from another.91 The present study differs from such approaches as it downplays the role of the substrate in the kinds of processes by which relexi‑ fication or restructuring can be explained, and questions the arbitrary selection of lexifier material to fulfil the functions of the former substrate languages. It could be questioned, though, how to accommodate clear cases of substrate transfer within such a framework. To take a fairly frequently‑­cited example occur‑ ring in the total systemic transfer of the aspectual system of Chinese discussed by Bao (2005), the adverb already, it has been clearly demonstrated in studies such as Matthews and Yip (2009) that already is a case of the ordinary contact‑­induced grammaticalisation discussed by Heine and Kuteva (2005). As noted in Ch. 5, it is frequently found as a conceptual source for the expression of perfectivity in the Southeast Asian region (in which it has been described as an example of a lamitive by Dahl (2006), as well as in other languages such as Yiddish, Jewish‑­American dialects, and Afrikaans (according to the entry for already in the Oxford English Dictionary Online (updated Sept. 2012). In other words, it is a universal source for the grammaticalisation of perfectivity, and need not be considered any diffe‑ rently in cases of contact. Given such evidence, it is not unreasonable to expect

91 This is not to suggest, though, that the constructions of a contact language comprise only of the morphosyntax and the lexicon as the exponents of a form‑­meaning correspondence. As noted in Chapter 3, for Croft (2001) and Traugott (2014: 89), the form pole of a construction can refer to syntax, morphology or phonology, while the meaning pole may refer to semantics, pragmatics or discourse function.

266 

 Chapter 9 Concluding remarks

that the adverb contains as part of its semantic composition a sense element that lends itself most readily to reinterpretation as an aspectual marker; this element could possibly be derived from the meanings of inchoativity associated with its use (Van der Auwera 1993: 627), in which it expresses the discontinuation of a prior negative state (= ‘not still not’), and is thus associated with grammatical perfectivity (see, e.g., Comrie 1976). Thus, even in cases which, on the surface, appear to represent a combinatory, mixed language construction dilemma, the morphosyntax of the substrate model is only represented in functional transfer, while the lexifier supplies the form most likely to be selected in any situation in which perfectivity requires grammaticalisation. Thus, linking the two contribu‑ ting languages of contact is an underlying semantic feature which forms an iden‑ tifiable pivot for the builders of the new contact language to develop a functional category in the replica language. The perfective aspectual use of already merges quite harmoniously with the adverb of Standard Singapore English, due to the shared meaning component of inchoativity. Other mixed construction types may be more difficult to dismiss. The topic‑­ comment information structure of many constructions in Singlish is transferred wholesale from the syntax of the local substrate languages, as noted by Bao and Lye (2005) e.g., as in Platt et al.’s (1984: 125) example, (24) in Chapter 2, repeated below as (259): (259) You go by metre, you have to pay. In (259), the subordinate clause functions as the topic of the construction, a covert conditional construction in which there is no subordination but the two clauses are simply juxtaposed. This kind of information structure is typical of the Sinitic languages as well as Malay. Thus the question whether such a construction is representative of the languages which supply the morphosyntax, or the language which supplies the lexicon is a question for further consideration. However, it is not the case that the same way of expressing conditionals is not available to standard varieties of English either, and constructions with the same syntax as in (259) are not exclusive to Sinitic languages (e.g., in English warning signs, such as You break it, you pay for it, or You smoke, you die). This does not mean to suggest that constructions are merged in all cases at the point of contact. As noted in Chapter 3, Trousdale (2012: 193) pointed out that lexicalisations are also a type of constructionalisation (Traugott 2014 prefers the use of the term contentful constructionalisation, as against procedural, or mor‑ phosyntactic). Needless to say, the discourse markers of SCE, borrowed directly and material‑­copied into the contact language from the various local languages of the substrate, are not mergers of anything. The particle lah discussed in Chapter 2, or the borrowed verb kena (= the adversative got) are simply added to the



Concluding remarks 

 267

construction inventory of SCE and serve to enrich the dialect. There is no question of the morphosyntax of one language and the lexicon of another – or the form of one language and the meaning of another. They contain both the form and the meaning of another language. The constructicon of Singlish is thus far from categorically equivalent to that of Standard Singapore English, if lexical loans and discourse particle constructions are taken into consideration. However, such examples are not a problem for a contact constructicon insofar as the presence of Merger ­Constructions are concerned. The problem for analysis in the present volume has been that of constructions which appear to bear elements of more than one language of contact, and their significance to a theory of constructions which states that all constructions are language‑­specific and variety‑­specific. In this way the following objectives proposed in Chapter 1 have been met: ... to develop the theory of constructions in such a way that it readily accounts for situ‑ ations of variation, contact or otherwise [ ... ] to bring to light a fundamental problem of construction‑­based approaches not so far researched at any depth, and to use the findings of such research to explain certain grammatical phenomena of a contact dialect which have been left relatively under‑­researched for too long.

The objectives above have been approached in the four case studies presen‑ ted; they have brought to light a fundamental problem of construction‑­based approaches –­that of accommodating ostensibly mixed construction types into the constructicons of contact languages. In the first case study, that of transi‑ tive constructions discussed in Chapter 4, the research study examined the frequency of the Conventionalised Scenario (CS) construction, which, it had been noted, was extended beyond its plausibility limits in Standard Singapore English and other standard varieties. The study investigated first the reasons why this should be so, and it was initially concluded that CS constructions are delimited by their parallel plausibility as ordinary transitive constructions, and that they expressed a reflexive causativity. They were considered in that study to be reduced, metonymical extensions of the causative‑­resultative construction: X causes Y to be V‑­ed. As such, many of the examples cited in the literature and produced by speakers in the surveys conducted of the feature in SCE could not be admissable as either ordinary transitive constructions or as reflexive causatives. And yet they were used, and appeared with a much higher frequency than in the control studies of British English. The role of substratum languages was conside‑ red to be an influence in the underlying presence of topic‑­prominent information structure, which permits more freedom in the selection of pre‑­verbal nominals in the dialect, and a relaxation of the agentive constraints usually holding between subject and verb. As a result, the construction generalises beyond its usual cons‑ traints, but is nevertheless linked to the transitive construction form by a shared

268 

 Chapter 9 Concluding remarks

semantic component of (general) causativity, common to both prototypical tran‑ sitives and the indirect causative‑­resultative construction that the CS stands for. The Transitive Merger Construction then provides the form‑­meaning correspon‑ dences for two construction functions at the same time. In the Ever Merger Construction discussed in Chapter 5, the construction con‑ sists of a single item, and adverb which in SSE functions as a minimiser of emphatic possibility in negative polarity contexts, and in SCE has an additional function as an experiential aspect marker, referring to events that have occurred at least once, a function which is not marked in SSE by any specific morphology but in the subst‑ rate languages of contact, is represented most often by a verb (in Mandarin Chinese with the meaning of ‘pass’ or ‘cross’). The relation between the substratum experi‑ ential verbs and the use of ever was found to defy any kind of syntactic association with the English adverb beyond that of supplying a motivating need to include the function in the Singlish constructicon. Previous accounts of a back‑­formation from the negative form never were considered to fall short of a total explanation since the paraphrasing of never as ‘not once’ (and consequently, the back‑­formation of ever as ‘once’) did not satisfactorily account for examples of experiential ever used alongside the adverb once, nor did they account for the use of negative‑­polarity ever used alongside experiential ever in the same dialect. It was believed that the expe‑ riential use of ever was derived from the minimser use in negative polarity contexts, where a meaning of emphatic possibility is reinterpreted as minimal necessity in positive contexts expressing the occurrence of an event at least once within a given time span. The semantic feature of repeatability, implicit in its minimiser function, is shared by both the negative polarity uses and the experiential uses, and com‑ prises one of the identifiable characteristics of experiential markers crosslinguisti‑ cally, according to Chappell (2001). This characteristic enables the experiential ever to be linked as a single, merger construction with the minimiser ever, providing a unified, form‑­meaning construction relation across both functions. The Past Tense Merger Construction allows a similar expansion of its most common functions to mark past time reference. Chapter 6 investigated the unusual use of the past tense in Singlish appearing in present time reference habitual aspect, performative uses, and as a marker of anteriority in subordinate clauses. These three uncharacteristic uses were isolated for the fact that they could not be explained by any substrate causes or influence (with, perhaps, the exception of the anterior use which was shared, in any case, with the aspect marker already –­ see Bao (2005). The marking of present habituals for past was compared against a survey of habitual marking in other languages in which perfective aspect was some‑ times found to alternate with imperfective aspect marking, depending on whether the aspectual focus was on the grammatical aspect or the lexical aspect of the verbs used in habituals, and it was found that the latter was the case in Serbo‑­Croatian and Bulgarian. In some of the Chinese dialects, such as Hokkien, habitual aspect



Concluding remarks 

 269

was often found to be marked by a verb u meaning ‘have’, indicating an aspectual function, but this form was not calqued into SCE and its function did not extend to include performatives and anteriors, as in Singlish. The study covered a range of different examples taken from the Flowerpod Corpus, selected for their wide lexical variety rather than their statistical significance. However, it was hypothesised that the sheer range of functions and variety of lexical contexts in which the extended Past Tense Construction appeared could not be attributable to idiomatic usage or mass hypercorrection alone, and that the functional reason for employing the past tense in such environments was due to a need to express the precedence of an event, a meaning component common to both the regular use of the past tense and the extended use shown in the examples. It was this meaning component of the construction which had been identified as the pivot on which the Past Tense Merger Construction had been created, a construction type which had taken some element of the SSE parallel construction and analogised it to new functional domains. The final case study of the Bare Noun Merger Construction in Chapter 7 also explains a grammatical phenomenon which has been left under‑­researched for too long. The frequent appearance of the construction in referring to non‑­specific items in lists of merchandise, or non‑­specific entities in traffic signs and other public notices and advertising genre had been previously observed in Singapore English in a former study (Ziegeler 2003). More recently, the data had been assem‑ bled to provide evidence against the hypothesis of construction coercion (via type‑­shifting) from count‑­to‑­mass nouns in a later study (Ziegeler 2010). Chapter 7 collected the findings of both studies and raised the question whether we are looking at two different dialects with different constructicons. Although evidence from Chinese was considered to have played a motivating role in the emergence of the Bare Noun Construction in SCE, in the fact that, like Singlish, it is transnu‑ meral, in other ways, the correlations with the Singlish system are not replicated exactly, as both specific and non‑­specific nouns in Chinese may be left bare, while in Singlish it is usually non‑­specific nouns that take the form of a Bare Noun Con‑ struction. The question was also raised whether this construction is actually a construction after all – since there is no form with which to pair the meaning (of non‑­specificity). However, it was concluded that even zero‑­constructions have a grammatical salience in being left unmarked (this was also noted by Bybee 1994, in referring only to grammatical morphology). In addition, the use of coercion as an explanation for the Bare Noun Construction was constrained by the need for a retrievable metonymic target, e.g., !!pedestrian material or !!customer material from which non‑­specific pedestrian or customer in the examples listed could be derived as the source of a mass noun. Thus it was concluded that Bare Noun Cons‑ tructions could survive without a type‑­shifting explanation, and that, at the same time, speakers exploited the form of a mass noun to share its parallel semantics of unboundedness (of configuration for the bare mass noun, and of quantification

270 

 Chapter 9 Concluding remarks

for the non‑­specific noun). In this way, the two dialects of SSE and SCE were seen to share the same construction. Chapter 8 collected the four cases studies and subjected them to the ­examination of a number of current theories of contact, mainly including those that deal with what may potentially produce a merger construction having the (morpho)syntax of one language and the lexicon of another. Some of these theo‑ ries included, for example, relexification (Lefebvre 1998, 2001) and total Systemic Transfer (Bao 2001, 2005); the chapter also reviewed the numerous different expla‑ nations of convergence in the literature, as well as the use of grammaticalisation as an explanatory tool in contact (Heine and Kuteva 2003, 2005). Although gramma‑ ticalisation is useful for explaining certain features of contact which may be found to either have universal sources, or to be replicating the developmental paths of the model languages, the explanation of grammaticalisation has not been pursued deeply enough in the present volume to reveal its relevance to the descriptive data so far presented. It is possible that further research may find that ordinary contact grammaticalisation may explain the extension of the adverb ever from a minimiser to an experiential marker, for example, as similar uses were found in Dutch. If so, it may be hypothesised that merger constructions give rise to grammaticalisation processes in some cases. Furthermore, a more intensive survey of the development of the numeral ‘one’ across languages may reveal that Singlish is at an earlier stage of development than SSE in terms of how it grammaticalises countability on nouns, and that the Bare Noun Construction is just a stage along the way. The data in the present volume have not, for the most part, been put to the test of diachronic considerations, and this are leaves ample scope for future investiga‑ tion, perhaps combining the study of merger contact constructions with historical constructionalisation (Pietsch 2010 has already begun to work in such domains, but in a slightly different way). The Merger Construction does not attempt to over‑­ claim its explanatory capacity in this respect. It is merely a descriptive tool to account for a collection of unusual features of contact in an Asian English dialect which are difficult to explain otherwise, for the fact that they are manifested, not as replicating a similar model language construction, but as extended uses of the local contact standard language which was once their lexifier. The Merger Construction is not simply an analogy, but is a language‑­builders’ expression of constructional parsimony in a context in which many construction types, as form‑­meaning pai‑ rings, may otherwise be duplicated beyond necessity. It is in this respect that the hypothesis of a Merger Construction has aimed to maintain its two‑­fold objective of contributing to the development of modern, construction‑­based theory as much as employing construction‑­based theory to contribute to the explanations required of the case studies presented in the preceding chapters.

References Aboh, Enoch, O. & Michel DeGraff. 2014. Some notes on bare noun phrases in Haitian Creole and Gungbe. A transatlantic Sprachbund perspective. In: Tor A. Åfarli & Brit Mæhlum (eds.), The Sociolinguistics of Grammar, 203–236. Amsterdam & Philadelphia: John Benjamins. Aikhenvald, Alexandra. 2002. Language contact in Amazonia. New York: Oxford University Press. Alsagoff, Lubna. 2007. Singlish: negotiating capital, culture and identity. In: Viniti Vaish, S. Gopinathan & Yongbing Liu (eds.), Language, Capital, Culture. Rotterdam: Sense Publishers, 25–46. Alsagoff, Lubna. 2010. English in Singapore: culture, capital and identity in linguistic variation. World Englishes 29. 336–348. Alsagoff, Lubna & Ho Chee Lick. 1998. The grammar of Singapore English. In: Joseph A. Foley, Thiru Kandiah, Bao Zhi‑­Ming, Anthea Fraser Gupta, Lubna Algasoff, Ho Chee Lick, Lionel Wee, Ismail S. Talib & Wendy Bokhorst‑­Heng. English in new cultural contexts. Reflections from Singapore, 127–151. Singapore: Oxford University Press [Singapore Institute of Management]. Ansaldo, Umberto. 2009. Contact languages. Ecology and evolution in Asia. Cambridge: Cambridge University Press. Ansaldo, Umberto. 2010. Contact and Asian varieties of English. In: Raymond Hickey (ed.), The Handbook of language contact, 498–517. Oxford: Blackwell. Backus, Ad, A. Seza Doğruöz & Bernd Heine. 2011. Salient stages in contact‑­induced grammatical change: Evidence from synchronic vs. diachronic contact situations. Language Sciences 33. 738–752. Bailey, Beryl. 1966. Jamaican Creole syntax. A transformational approach. Cambridge: Cambridge University Press. Baker, C.L. 1966. Definiteness and indefiniteness in English. Illinois: University of Illinois Masters Thesis. Bakker, Peter. 2000. Rapid language change: Creolization, intertwining, convergence. In: Colin Renfrew, April McMahon & Larry Trask (eds.), Time depth in historical linguistics, 585–620. Cambridge: The McDonald Institute. Bakker, Peter & Maarten Mous (eds.). 1994. Mixed languages: 15 case studies in language intertwining. Amsterdam: IFOTT. [Distribution: HAG Publications/Holland Academic Graphics, The Hague]. Bao, Ming‑­Shun. 1971. Shuo Wen Bu Shou Tong Shi [Explanations of the radicals in Shuo Wen]. Taipei: Author’s own publication. Bao, Zhi Ming. 1995. Already in Singapore English. World Englishes 14. 181–188. Bao, Zhi Ming. 2001. The origin of empty categories in Singapore English. Journal of Pidgin and Creole Languages 16. 275–319. Bao, Zhi Ming. 2003. Social stigma and grammatical autonomy in nonnative varieties of English. Language in Society 32(1). 23–46. Bao, Zhi Ming. 2005. The aspectual system of Singapore English and the systemic substratist explanation. Journal of Linguistics 41. 237–267. Bao, Zhi Ming. 2009. One in Singapore English. Studies in Language 33. 338–365.

272 

 References

Bao, Zhi Ming. 2010a. A usage‑­based approach to substratum transfer: the case of four unproductive features in Singapore English. Language 86(4). 792–820. Bao, Zhi Ming. 2010b. Must in Singapore English. Lingua 120. 1727–1737. Bao, Zhi Ming. In press. The making of Singapore English. Cambridge: Cambridge University Press. Bao, Zhi Ming & Lye Hui Min. 2005. Systemic transfer, topic prominence and the bare conditional in Singapore English. Journal of Pidgin and Creole Languages 20. 269–291. Bao, Zhi Ming & Lionel Wee. 1999. The passive in Singapore English. World Englishes 22. 5–13. Bencini, Giulia M.L. & Adele Goldberg. 2000. The contribution of argument structure constructions to sentence meaning. Journal of Memory and Language 43. 640–651. Bergen, Benjamin K. & Nancy Chang. 2005. Embodied Construction Grammar in simulation based language understanding. In: Jan‑­Ola Östman & Mirjam Fried (eds.), Construction grammars. Cognitive grounding and theoretical extensions, 147–190. Amsterdam & Philadelphia: John Benjamins. Bickerton, Derek. 1981. Roots of language. Ann Arbor: Karoma Publishers Inc. Boas, Hans. 2003. A constructional approach to resultatives. Stanford: CSLI Publications. Boas, Hans. 2005. Determining the productivity of resultatives: A reply to Goldberg and Jackendoff. Language 81. 448–464. Boas, Hans. 2008. Determining the structure of lexical entries and grammatical constructions in Construction Grammar. Annual Review of Cognitive Linguistics 6. 113–144. Boas, Hans. 2010. Comparing constructions across languages. In: Hans Boas (ed.), Contrastive studies in construction grammar, 1–19. Amsterdam & Philadelphia: John Benjamins. Boas, Hans. 2011. Coercion and leaking argument structures in Construction Grammar. Linguistics 49. 1271–1303. Boas, Hans. 2013. Cognitive construction grammar. In: Graeme Trousdale & Thomas Hoffmann (eds.), The Oxford handbook of construction grammar. Oxford: Oxford University Press. [Pre‑­print version.] Bodman, Nicholas. 1955. Spoken Amoy Hokkien. Kuala Lumpur: Charles Grenier & Son Ltd. Bolton, Kingsley. 2002. Chinese Englishes: from Canton jargon to global English. World Englishes 21. 181–199. Brinton, Laurel. 1994. The differentiation of statives and perfects in Early Modern English: The development of the conclusive perfect. In: Dieter Stein & Ingrid Tieken Boon van Ostade (eds.), Towards a Standard English 1600–1800, 135–170. Berlin & New York: Mouton de Gruyter. Brisard, Frank. 1997. The English tense system as an epistemic category: the case of futurity. In: Marjolijn Verspoor, K.D. Lee & Eve Sweetser (eds.), Lexical and syntactical constructions and the construction of meaning, 271–286. Amsterdam and Philadelphia: John Benjamins. Brown, Adam. 1999. Singapore English in a nutshell. Singapore: Federal Publications. Bruyn, Adrienne. 1996. On identifying instances of grammaticalization in creole languages. In: Philip Baker & Anand Syea (eds.), Changing meanings, changing functions: Papers relating to grammaticalization in contact languages, 29–46. London: University of Westminster Press. Bruyn, Adrienne. 2009. Grammaticalization in creoles: Ordinary and not‑­so‑­ordinary cases. Studies in Language 33. 312–337. Budge C. 1986. Variation in Hong Kong English. Melbourne: Monash University PhD thesis. Butler, Susan. 1999. A view on standards in South‑­East Asia. World Englishes 18. 187–198. Bybee, Joan. 1985. Morphology. A study of the relation between meaning and form. Amsterdam & Philadelphia: John Benjamins.

References 

 273

Bybee, Joan. 1994. The grammaticalisation of zero: asymmetries in tense and aspect systems. In: William Pagliuca (ed.), Perspectives on grammaticalization, 235–254. Amsterdam & Philadelphia: John Benjamins. Bybee, Joan. 1995. The semantic development of past tense modals in English. In: Joan Bybee & Suzanne Fleischman (eds.), Modality in grammar and discourse, 503–517. Amsterdam & Philadelphia: John Benjamins. Bybee, Joan. 2003. Mechanisms of change in grammaticization: the role of frequency. In: Richard Janda & Brian Joseph (eds.), Handbook of historical linguistics, 602–623. Oxford: Blackwell. Bybee, Joan. 2013. Usage‑­Based Theory and exemplar representations of constructions. In: Graeme Trousdale & Thomas Hoffmann (eds.), The Oxford handbook of construction grammar, 49–69. Oxford: Oxford University Press. Bybee, Joan & William Pagliuca. 1987. The evolution of future meaning. In: Anna G. Ramat, O. Carruba & G. Bernini (eds.), Papers from the seventh International Conference on Historical Linguistics, 109–122. Amsterdam & Philadelphia: John Benjamins. Bybee, Joan, Revere D. Perkins & William Pagliuca. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Cappelle, Bert. 2006. Particle placement and the case for “allostructions”. Constructions SV1–7/2006. www.constructions‑­online.de. urn:nbn:de:0009–4‑­6839, ISSN 1860–2010 (accessed 19 May 2014). Chao, Y.R. 1968. A grammar of spoken Chinese. Berkeley & Los Angeles: University of California Press. Chappell, Hilary. 1992. Towards a typology of aspect in Sinitic languages. Zhongguo Jingnei Yuyanji Yuyanxue [Chinese Languages and Linguistics] Vol. 1: Hanyu Fangyan [Chinese Dialects]. 67–106. Taipei: Academia Sinica. Chappell, Hilary. 2001. The experiential perfect as an evidential marker in Sinitic languages. Pre‑­publication version of: A typology of evidential markers in Sinitic languages. In: Hilary Chappell (ed.), Sinitic grammar: Synchronic and diachronic perspectives, 56–84. Oxford: Oxford University Press. Chappell, Hilary. 2001 (ed.). Sinitic grammar: Synchronic and diachronic perspectives. Oxford: Oxford University Press. Chaudenson, Robert. 1977. Toward the reconstruction of the social matrix of creole language. In: Albert Valdman (ed.), Pidgin and creole linguistics, 259–276. Bloomington, In.: Indiana University Press. Chen, Jason Yingjie. 2008. Already in Singapore Colloquial English. National University of Singapore, Faculty of Arts and Sciences academic exercise. Cheng, Robert L. 1979. Taiwanese u and Mandarin you. In: Cheng Liang Wei, Tsao Feng Fu & Tsai Mei Hui (eds.), Taiwan Minnan Yu Yufa Lun Wen Xuan Ji (2) [Selected papers on Taiwanese Southern Min grammar (2)], 170–209. Hsinchu: National Tsing Hua University. Cheng, Robert L. 1997. Tai, Huayu de Jiechu Yu Tongyi Yu de Hudong [Taiwanese and Mandarin structures and their developmental trends in Taiwan II: Contrasts between Taiwanese and Mandarin and restructuring of their synonyms]. Taipei: Yuan Liu Publishing Co. Claridge, Claudia. 2007. Constructing a corpus from the web: message boards. In: Marianne Hundt, Nadja Nesselhauf & Carolin Biewer (eds.), Corpus linguistics and the web, 87–108. Amsterdam: Rodopi. Colleman, Timothy & Bernard de Clerk. 2011. Constructional semantics on the move: On semantic specialization in the English double object construction. Cognitive Linguistics 22. 183–209.

274 

 References

Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Corne, Chris. 1995. A contact‑­induced and vernacularized language: How Melanesian is Tayo? In: Philip Baker (ed.), From contact to creole and beyond, 121–48. London: University of Westminster Press. Corne, Chris. 1999. From French to creole: The development of new vernaculars in the French colonial world. London: University of Westminster Press. Crewe, William. 1977[1984]. Singapore English and Standard English: Exercises in awareness. Singapore: Eastern Universities Press. Cristofaro, Sonia. 2004. Past habituals and irrealis. In: Yuri A. Lander, Vladimir A. Plungian & Anna Yu Urmanchieva (eds.), 256–272. Irrealis and irreality. Moscow: Gnosis. Croft, William. 1990. Possible verbs and the structure of events. In: S.L. Tsohatzidis (ed.), Meanings and prototypes: Studies in linguistic categorization, 48–75. London: Routledge. Croft, William. 2000. Explaining language change. An evolutionary approach. Harlow: Longman. Croft, Wiliam. 2001. Radical Construction Grammar. Syntactic theory in typological perspective. Oxford: Oxford University Press. Croft, William. 2003. Lexical rules vs. constructions. A false dichotomy. In: Hubert Cuyckens, Thomas Berg, René Dirven & Klaus‑­Uwe Panther (eds.), Motivation in language. Studies in honor of Günter Radden, 49–68. Amsterdam & Philadelphia: John Benjamins. Croft, William. 2013. Radical Construction Grammar. In: Graeme Trousdale & Thomas Hoffmann (eds.), The Oxford handbook of construction grammar, 211–232. Oxford: Oxford University Press. Dahl, Östen. 1985. Tense and aspect systems. Oxford: Blackwell. Dahl, Östen. 1996. The marking of the episodic/generic distinction in tense‑­aspect systems. In: Gregory Carlson & Francis J. Pelletier (eds.), The generic book, 412–425. Chicago & London: University of Chicago Press. Dahl, Östen. 2006. Relativism and universalism from the perspective of areal typology and grammaticalization theory. Plenary talk presented at the 39th Annual Meeting of the Societas Linguistica Europaea, University of Bremen, 30 August–2 September. Davidse, Kristin. 1992. Transitivity/ergativity: the Janus‑­headed grammar of actions and events. In: M. Davies & L. Ravelli (eds.), Advances in systemic linguistics, 105–135. London and New York: Pinter. Davydova, Julia. 2013. Detecting historical continuity in a linguistically diverse urban area. The present perfect in modern Singapore English. In: Joana Duarte & Ingrid Gogolin (eds.), Linguistic superdiversity in urban areas: Research approaches, 193–226. Amsterdam: John Benjamins. DeCamp, David. 1971. Toward a generative analysis of a post‑­creole speech continuum. In: Del Hymes (ed), Pidginization and creolization of languages, 349–370. Cambridge: Cambridge University Press. Dennys, Nicholas B. 1878a. Handbook of Malay Colloquial, as spoken in Singapore: Being a series of introductory lessons for domestic and business purposes. London: Trubner and Co. Dennys, Nicholas B. 1878b. “Pidgin” English. Journal of the Straits Branch of the Royal Asiatic Society 1. 168–174. De Swart, Henrietta & Joost Zwarts. 2009. Less form – more meaning: why bare singular nouns are special. Lingua 119. 280–295. Deterding, David. Tenses and will/would in a corpus of Singapore English. In: David Deterding, Low Ee Ling & Adam Brown (eds.), English in Singapore. Research on grammar, 31–37. Singapore: McGraw Hill.

References 

 275

Detges, Ulrich. 2000. Two types of restructuring in French creoles: A cognitive approach to the genesis of tense markers. In: Ingrid Neumann‑­Holzschuh & Edgar W. Schneider, (eds.), Degrees of restructuring in creole languages, 135–162. Amsterdam & Philadelphia: John Benjamins. Diewald, Gabriele. 1999. Die Modalverben im Deutschen. Grammatikalisierung und ­Polyfunktionalität.Tübingen: Max Niemeyer. Eckert, Penelope. 2008. Variation and the indexical field. Journal of Sociolinguistics 12. 453–476. Ferguson, Charles A. 1959. Diglossia. Word 15. 325–340. Fillmore, Charles. 1967. The syntax of pre‑­verbs. Glossa 1. 91–125. Fillmore, Charles. 1988. The mechanisms of ‘Construction Grammar’. Proceedings of the Berkeley Linguistics Society 14. 35–55. Fillmore, Charles, Paul Kay & Mary Kay O’Connor. 1988. Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64. 501–538. Fleischman, Suzanne. 1989. Temporal distance: A basic linguistic metaphor. Studies in Language 13. 1–50. Fodor, William A. & Ivan A. Sag. 1982. Referential and quantificational indefinites. Linguistics and Philosophy 5. 355–398. Foley, William A. & Robert D. Van Valin Jr. 1984. Functional syntax and universal grammar. Cambridge: Cambridge University Press. Fried, Mirjam (to appear). Construction grammar. Forthcoming in A. Alexiadou & T. Kiss (eds.), Handbook of syntax (2nd ed.). Berlin: Walter de Gruyter. Gast, Volker & Johan van der Auwera. 2012. What is contact‑­induced grammaticalization? Evidence from Mayan and Mixed‑­Zoquean languages. In: Björn Wiemer, Bernhard Wälchli & Björn Hansen (eds.), Grammatical replication and borrowability in language contact, 381–426. Berlin & New York: Mouton de Gruyter. Geurts, Bart, in press. Specific indefinites, presupposition, and scope. To appear in R. Bäuerle, U. Reyle & T.E. Zimmerman (eds.), Presupposition and discourse. Gil, David. 2003. English goes Asian: Number and indefiniteness in the Singlish noun‑­phrase. In: F. Plank (ed.), Noun phrase structure in the languages of Europe, 467–514. Berlin & New York: Mouton de Gruyter. Givón, Talmy. 1973. Opacity and reference in language: an inquiry into the role of modalities. In: John Kimball (ed.): Syntax and semantics 2, 95–122. New York: Academic Press. Givón, Talmy. 1978. Definiteness and referentiality. In: Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik (eds.), Universals of human language. Vol. IV, 291–330. Stanford: Stanford University Press. Givón, Talmy. 1981. On the development of the numeral ‘one’ as an indefinite marker. Folia Linguistica Historica II/1. 35–55. Givón, Talmy. 1989. Mind, code and context. Essays in pragmatics. Hillsdale: Erlbaum. Givón, Talmy. 1994. Irrealis and the subjunctive. Studies in Language 18. 265–337. Givón, Talmy & Lynne Yang. 1994. The rise of the English get‑­passive. In: Barbara Fox & Paul J. Hopper (eds.), Voice, form and function, 119–149. Amsterdam & Philadelphia: John Benjamins. Goldberg, Adele. 1995. Constructions. A construction grammar approach to argument structure. Chicago: University of Chicago Press. Gołąb, Zbigniew. 1956. The concept of isogrammatism. Biuletyn Polskiego Towarzystwa ­Jezykoznawczego 15. 1–12.

276 

 References

Gołąb, Zbigniew. 1959. Some Arumanium‑­Macedonian isogrammaticisms and the social background of their development. Word 15. 415–435. Goldberg, Adele. 2000. Patient arguments of causative verbs can be omitted: the role of information structure in argument distribution. Language Sciences 34. 503–524. Goldberg, Adele. 2002. Surface generalizations: An alternative to alternations. Cognitive Linguistics 13. 327–356. Goldberg, Adele. 2003. Constructions: a new theoretical approach to language. Trends in Cognitive Sciences 7. 219–224. Goldberg, Adele. 2006. Constructions at work. The nature of generalization in language. Oxford: Oxford University Press. Goldberg, Adele. 2009a. The nature of generalization in language. Cognitive Linguistics 20. 93–127. Goldberg, Adele. 2009b. Coercion and the partial productivity of constructions. Paper presented at the Third International Conference of the French Cognitive Linguistics Association (AFLiCo3), University of Paris 10, 28 May. Goldberg, Adele. 2013. Constructionist approaches. In: Graeme Trousdale & Thomas Hoffmann (eds.), The Oxford handbook of construction grammar, 15–31. Oxford: Oxford University Press (pre‑­print version). Goldberg, Adele. To appear. Tuning in to the verb particle construction in English. Forthcoming in Léa Nash & Pollet Samvelian (eds.), Syntax and semantics. Goldberg, Adele & Ray Jackendoff. 2004. The English resultative as a family of constructions. Language 80. 532–568. Goossens, Louis. 1987. Modal shifts and predication types. In: Johan van der Auwera & Louis Goossens (eds.), Ins and outs of predication, 21–37. Dordrecht: Foris. Grice, H. Paul. 1975. Logic and conversation. In: Peter Cole & Jerry Morgan (eds.), Speech acts [Syntax and Semantics 3], 113–128. New York: Academic Press. Gumperz, John J. & Robert Wilson. 1971. Convergence and creolization: a case from the Indo‑­Aryan/Dravidian border. In: Dell Hymes (ed.), Pidginization and creolization of languages, 151–167. Cambridge: Cambridge University Press. Gupta, Anthea Fraser. 1991 Almost a creole: Singapore Colloquial English. California Linguistic Notes 23 (1). 9–21. Gupta, Anthea Fraser. 1992a. Contact features of Singapore Colloquial English. In: Kingsley Bolton & Helen Kwok (eds.), Sociolinguistics today. International perspectives, 323–345. London: Routledge. Gupta, Anthea Fraser. 1992b. The pragmatic particles of Singapore Colloquial English. Journal of Pragmatics 18. 31–57. Gupta, Anthea Fraser. 1994. The step‑­tongue. Children’s English in Singapore. Clevedon: Multilingual Matters. Gupta, Anthea Fraser. 1998. The situation of English in Singapore. In: Joseph A. Foley, Thiru Kandiah, Zhi‑­Ming Bao, Anthea Fraser Gupta, Lubna Algasoff, Chee Lick Ho, Lionel Wee, Ismail S. Talib, & Wendy Bokhorst‑­Heng. English in new cultural contexts. Reflections from Singapore, 106–126. Singapore: Oxford University Press [Singapore Institute of Management]. Gupta, Anthea Fraser. 2006. Singlish on the web. In: Azirah Hashim & Norizah Hassan (eds.), Varieties of English in South East Asia and beyond, 19–37. Kuala Lumpur: University of Malaya Press.

References 

 277

Hagège, Claude. 1993. The language builder: An essay on the human signature in linguistic morphogenesis. Amsterdam and Philadelphia: John Benjamins. Hansen, Chad. 1983. Language and logic in ancient China. Ann Arbor: University of Michigan Press. Haspelmath, Martin. 1987. Transitivity alternations of the antiaccusative type [Arbeitspapier Nr. 5]. Cologne: University of Cologne Institute of Linguistics. Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Clarendon. Haspelmath, Martin. 1999. Why is grammaticalization irreversible? Linguistics 37. 1043–1068. Haspelmath, Martin, Matthew S. Dryer, David Gil & Bernard Comrie. 2005. The world atlas of language structures. Oxford: Oxford University Press. Hassan, Abdullah & Ainon Muhammed. 1994. Kamus Ingerris‑­Melayu [English‑­Malay dictionary]. Kuala Lumpur: Utusan Publications. Haugen, Einar. 1950. The analysis of linguistic borrowing. Language 26. 210–231. Heath, Jeffrey. 1978. Linguistic diffusion in Arnhem Land. (Australian Aboriginal Studies Research and Regional Studies, 13). Canberra: Australian Institute of Aboriginal Studies. Heine, Bernd. 1992. Grammaticalization chains. Studies in Language 16. 335–368. Heine, Bernd. 1993. Auxiliaries. Cognitive forces and grammaticalization. New York: Oxford University Press. Heine, Bernd. 1997. Cognitive foundations of grammar. New York: Oxford University Press. Heine, Bernd. 2002. On the role of context in grammaticalization. In: Ilse Wischer & Gabriele Diewald (eds.), New reflections on grammaticalization, 83–101. Amsterdam & Philadelphia: John Benjamins. Heine, Bernd. 2004. On genetic motivation in grammar. In: Günter Radden & Klaus‑­Uwe Panther (eds.), Studies in linguistic motivation (Cognitive Linguistics Research 28), 103–120. Berlin & New York: Mouton de Gruyter. Heine, Bernd. 2013. Isomorphism and formulas of equivalence in language contact. In: Guangshun Cao, Hilary Chappell, Redouane Djamouri & Thekla Wiebusch (eds.), Breaking down the barriers: Interdisciplinary studies in Chinese linguistics and beyond (Festschrift for Prof. Alain Peyraube). Taiwan: Institute of Linguistics, Academia Sinica. Pre‑­print version. Heine, Bernd & Mechtild Reh. 1984. Grammaticalization and reanalysis in African languages. Hamburg: Helmut Buske. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. Grammaticalization: A conceptual framework. Chicago: University of Chicago Press. Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Heine, Bernd & Tania Kuteva. 2003. On contact‑­induced grammaticalization. Studies in Language 27. 529–72. Heine, Bernd & Tania Kuteva. 2005. Language contact and grammatical change. Cambridge: Cambridge University Press. Heine, Bernd & Tania Kuteva. 2010. Contact and grammaticalization. In: Raymond Hickey (ed.), Handbook of language contact, 86–105. Oxford: Blackwell. Heine, Bernd & Motoki Nomachi. 2013. Contact‑­induced replication. Some diagnostics. In: Martine Robbeets & Hubert Cuyckens (eds.), Shared grammaticalization. With special focus on the Transeurasian languages, 67–100. Amsterdam & Philadelphia: John Benjamins. The Helsinki Diachronic Corpus of English Texts (electronic version). Department of English, University of Helsinki.

278 

 References

Hewson, John & Vit Bubenik. 1997. Tense and aspect in Indo‑­European languages. Theory, typology, diachrony. Amsterdam & Philadelphia: John Benjamins. Himmelmann, Nikolaus. 2004. Lexicalization and grammaticalization: Opposite or orthogonal? In: Walter Bisang, Nikolaus P. Himmelmann & Bjorn Wiemer (eds.), What makes ­grammaticalization ‑­a look from its fringes and its components, 21–42. Berlin & New York: Mouton de Gruyter. Ho, Mian Lian. 1986. The verb phrase in Singapore English. Melbourne: Monash University Phd. thesis. Ho, Mian Lian & John Platt. 1993. Dynamics of a contact continuum. Singaporean English. Oxford: Clarendon. Ho, Mian Lian & Irene Wong. 2001. The use of ever in Singaporean English. World Englishes 20. 79–87. Ho, Seng Ong. 1964. Methodist schools in Malaysia: Their record and history. Petaling Jaya: Board of Education. Höder, Steffen. 2014. Constructing diasystems. Grammatical organisation in bilingual groups. In: Tor A. Åfarli and Brit Mæhlum (eds.), The sociolinguistics of grammar, 137–152. Amsterdam & Philadelphia: John Benjamins. Hoeksema, Jack. 1998. On the (non)loss of the polarity sensitive Dutch ooit. In: Richard M. Hogg & Linda van Bergen (eds.), Historical linguistics 1995. Selected papers from the 13th International Conference on Historical Linguistics, Manchester, August 1995. Volume 2: Germanic linguistics, 101–114. Amsterdam & Philadelphia: John Benjamins. Hoffmann, Thomas & Graeme Trousdale. 2011. Variation, change and constructions in English. Cognitive Linguistics 22. 1–23. Hopper, Paul J. 1987. Emergent grammar. Proceedings of the Berkeley Linguistics Society 13. 139–57. Hopper, Paul J. 1991. On some principles of grammaticization. In: Elizabeth C. Traugott & Bernd Heine (eds.), Approaches to Grammaticalization, Vol.1, 17–35. Amsterdam & Philadelphia: John Benjamins. Hopper, Paul J. & Janice Martin. 1987. Structuralism and diachrony: the development of the indefinite article in English. In: A.G. Ramat, O. Carruba & G. Bernini (eds.), Papers from the seventh International Conference on Historical Linguistics, 295–304. Amsterdam & Philadelphia: John Benjamins. Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56. 251–299. Hopper, Paul & Elizabeth Closs Traugott. 2003 [1993]. Grammaticalization. Cambridge: Cambridge University Press. Horn, Laurence R. 2000a. Any and ever: Free choice and free relatives. In: Adam Z. Wyner (ed.), Proceedings of the 15th Annual Conference of the Israel Association for Theoretical Linguistics, 71–111. University of Haifa. Horn, Laurence R. 2000b. Pick a theory (not just any theory). Indiscriminatives and the free choice indefinite. In: Laurence Horn & Yasuhiko Kato (eds.), Negation and polarity. Syntactic and semantic perspectives, 147–192. Oxford: Oxford University Press. Horn, Laurence R. 2001[1989]. A natural history of negation, 2nd edn. Stanford: CSLI publications. Horn. Laurence R. 2007. In press. Histoire d’*O: Lexical pragmatics and the geometry of opposition. In: Jean‑­Yves Béziau & Gilbert Payette (eds.) New Perspectives on the Square of Opposition, 393–426. In press, Peter Lang.

References 

 279

Hudson, Richard A. 1986. Sociolinguistics. Cambridge: Cambridge University Press. Hussain, Zakir. 2006. How to use ‘aiyah’? Look it up online, lah. The Straits Times 10 February 2006. 1–2. Israel, Michael. 1998. Ever: polysemy and polarity sensitivity. Linguistic Notes from La Jolla 19. 29–45. Israel, Michael. 2004. The pragmatics of polarity. In: Laurence Horn & Gregory Ward (eds.), Handbook of pragmatics, 701–723. Oxford: Blackwell. Iwata, Seizi. 2002. Does MANNER count or not? Manner‑­of‑­motion verbs revisited. Linguistics 40. 61–110. Janson, Tore 1984. Articles and plural formation in creoles: change and universals. Lingua 64. 291–323. Jespersen, Otto. 1966. Negation in English and other languages. (2nd ed.). Copenhagen: E. Muksgaard. Johanson, Lars. 1992. Strukturelle Faktoren in türkischen Sprachkontakten. Stuttgart: Franz Steiner Verlag. Johanson, Lars. 1998. Code‑­copying in Irano‑­Turkic. Language Sciences 20. 325–337. Johanson, Lars. 2013. Isomorphic processes: grammaticalization and copying of grammatical elements. In: Martine Robbeets & Hubert Cuyckens (eds.), Shared grammaticalization. With special focus on the Transeurasian languages, 101–109. Amsterdam & Philadelphia: John Benjamins. Kaltenböck, Gunther, Bernd Heine & Tania Kuteva. 2011. On thetical grammar. Studies in Language 35. 848–893. Kasher, A. & Gabbay D.M. 1976. On the semantics and pragmatics of specific and non‑­specific indefinite expressions. Theoretical Linguistics 3. 145–190. Kaufmann, Stefan. 2005. Conditional predictions. Linguistics and Philosophy 28. 181–231. Kay, Paul. 1984. The kind of/sort of construction. Proceedings of the Berkeley Linguistics Society 10. 157–171. Keesing, Roger M. 1991. Substrates, calquing and grammaticalization in Melanesian Pidgin. In: Elizabeth Traugott & Bernd Heine (eds.), Approaches to Grammaticalization, vol. 1, 315–342. Amsterdam & Philadelphia: John Benjamins. Kemmer, Suzanne & Arie Verhagen. 1994. The grammar of causatives and the conceptual structure of events. Cognitive Linguistics 5. 115–156. Kirchner, Gustav. 1952. Die zehn Hauptverben des Englischen im Britischen und ­Amerikanischen. Halle/Saal. Niemeyer. Kövecses, Zoltán & Günter Radden. 1998. Metonymy: developing a cognitive linguistic view. Cognitive Linguistics 9. 37–77. Kratzer, Angelika. 1995. Scope or pseudoscope? Are there wide‑­scope indefinites? In: S. Rothstein (ed.), Events and grammar, 163–196. Dordrecht: Kluwer. Krifka, Manfred, Francis Jeffrey Pelletier, Gregory N. Carlson, Alice ter Meulen, Godehard Link, & Gennaro Cherchia. 1995. Genericity: an introduction. In: Gregory N. Carlson & Francis J. Pelletier (eds.), The Generic Book, 1–24. Chicago: University of Chicago Press. Kuteva, Tania. 2013. Sentence‑­final what in Singapore English: what kind of grammaticalization is it? Invited seminar presented at the Université Sorbonne Nouvelle Paris 3, Institut du Monde Anglophone, 31 May 2013. Kwan‑­Terry, Anna. 1978. The meaning and source of the ‘la’ and the ‘what’ particles in Singapore English. RELC Journal 9. 22–36.

280 

 References

Kwan‑­Terry, Anna. 1992. Towards a dictionary of Singapore English ‑­issues relating to making entries for particles in Singapore English. In: Anne Pakir (ed.), Words in a cultural context, 62–72. Singapore: Unipress. Lakoff, George. 1987. Women, fire, and dangerous things. Chicago: University of Chicago Press. Lakoff, George & Mark Johnson. 1980. Metaphors we live by. Chicago: University of Chigago Press. Lakoff, Robyn. 1971. Passive resistance. Proceedings of the Chicago Linguistics Society 7. 149–162. Lambrecht, Knud. 1986. Pragmatically motivated syntax: presentation cleft constructions in spoken French. Chicago Linguistics Society 22 (2). 115–126. Langacker, Ronald W. 1982. Remarks on English aspect. In: Paul J. Hopper (ed.), Tense‑­aspect: Between semantics and pragmatics, 265–304. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald W. 1987. Foundations of cognitive grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1991. Foundations of cognitive grammar. Vol. 2: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 1997. Generics and habituals. In: Angeliki Athanasiadou & René Dirven (eds.) On conditionals again, 191–122. Amsterdam & Philadelphia: John Benjamins. Langacker, Ronald W. 2001. The English present tense. English Language and Linguistics 5. 251–272. Langacker, Ronald W. 2005a. Construction grammars: Cognitive, Radical, and less so. In: Francisco José Ruiz de Mendoza Ibañez & M. Sandra Peña Cervel (eds.), Cognitive linguistics. Internal dynamics and inter‑­disciplinary interaction, 101–159. Berlin: Mouton de Gruyter. Langacker, Ronald W. 2005b. Integation, grammaticization, and constructional meaning. In: Mirjam Fried & Hans Boas (eds.), Grammatical constructions: Back to the roots, 157–189. Amsterdam & Philadelphia: John Benjamins. Lefebvre, Claire. 1998. Creole genesis and the acquisition of grammar: The case of Haitian Creole. Cambridge: Cambridge University Press. Lefebvre, Claire. 2001. The interplay of relexification and levelling in creole genesis and development. Linguistics 39. 371–408. Lefebvre, Claire. 2009. The contribution of relexification, grammaticalisation, and reanalysis to creole genesis and development. Studies in Language 33. 277–311. Lehmann, Christian. 1992. Word order change by grammaticalization. In: M. Gerritsen & Dieter Stein (eds.), Internal and external factors in syntactic change, 395–416. Berlin & New York: Mouton de Gruyter. Lehmann, Christian. 1995. Thoughts on grammaticalization. München: Lincom Europa. Originally published as: Lehmann, Christian. 1982. Thoughts on grammaticalization. A programmatic sketch. Vol. I. [Arbeiten des Kölner Universalienprojekts 48]. Köln: Institut für Sprachwissenschaft der Universität. Lehmann, Winifred P. 1976. From topic to subject in Indo‑­European. In: Charles N. Li (ed.), Subject and topic, 445–456. New York: Academic Press. Leimgruber, Jacob R.E. 2009. Modelling variation in Singapore English. University of Oxford doctoral thesis. Leimgruber, Jacob R.E. 2012. Singapore English: An indexical approach. World Englishes 31. 1–14.

References 

 281

Leiss, Elisabeth. 2008. The silent and aspect‑­driven patterns of deonticity and epistemicity. A chapter on diachronic typology. In: Werner Abraham & Elisabeth Leiss (eds.), Modality‑­aspect interfaces. Implications and typological solutions, 15–41. Amsterdam & Philadelphia: John Benjamins. Lenci, Alessandro & Pier Marco Bertinetto. 2000. Aspects, adverbs, and events. Habituality vs. Perfectivity. In: James Higginbotham, Fabio Pianesi & Achille C. Varzi (eds.), Speaking of events, 245–286. Oxford: Oxford University Press. Ler, Soon Lay Vivien. 2006. A relevance‑­theoretic approach to discourse particles in Singapore English. In: Kerstin Fischer (ed.), Approaches to discourse particles, 149–166. Amsterdam: Elsevier. Leuschner, Torsten. 1996. ‘Ever’ and universal quantifiers of time: observations from some Germanic languages. Language Sciences 18. 469–484. Levin, Beth. 1993. English verb classes and alternations. Chicago: University of Chicago Press. Li, Charles N. & Sandra A. Thompson. 1976a. Development of the causative in Mandarin Chinese: interaction of diachronic processes in syntax. In: Masayoshi Shibatani (ed.), The grammar of causative constructions [Syntax and Semantics 6], 477–492. New York: Academic Press. Li, Charles N. & Sandra A. Thompson. 1976b. Subject and topic: a new typology of language. In: Charles N. Li (ed.), Subject and topic, 457–489. New York: Academic Press. Li, Charles N. & Sandra A. Thompson. 1989[1981]. Mandarin Chinese: A functional reference grammar. Berkeley: University of California Press. Li, Charles N. & Sandra A. Thompson. 1994. On ‘middle voice’ verbs in Mandarin. In: Barbara Fox & Paul J. Hopper (eds.), Voice, form and function, 231–246. Amsterdam & Philadelphia: John Benjamins. Lim, Lisa. 2004. Sounding Singaporean. In: Lisa Lim (ed.), Singapore English: A grammatical description, 19–56. Amsterdam & Philadelphia: John Benjamins, Lim, Lisa. 2007. Mergers and acquisitions: on the ages and origins of Singapore English particles. World Englishes 26: 446–473. Lim, Lisa & Joseph Foley. 2004. English in Singapore and Singapore English. In: Lisa Lim (ed.), Singapore English: A grammatical description, 1–18. Amsterdam & Philadelphia: John Benjamins. Loh, Jia Peng. 1998. Tense and aspect in Singapore English. National University of Singapore Academic Exercise. Low, Ee Ling & Adam Brown. 2003. An introduction to Singapore English. Singapore: McGraw Hill. Lu, Wendy L. 1991. The semantics of Taiwanese u. Rice University MA dissertation. Matras, Yaron. 1998. Convergence vs. fusion in linguistic areas. Paper presented at the Annual Meeting of the Deutsche Gesellschaft für Sprachwissenschaft, Halle, 5–7 March. Matras, Yaron. 2010. Contact, convergence, and typology. In: Raymond Hickey (ed.), Handbook of language contact, 66–85.Oxford: Blackwell. Matras, Yaron. 2011. Explaining convergence and the formation of linguistic areas. In: Osamu Hieda, Christa König &Hirosi Nakagowa (eds.), Geographical typology and linguistic areas: with special reference to Africa, 143–160. Amsterdam & Philadelphia: John Benjamins. Matras, Yaron & Peter Bakker (eds.). 2003. The mixed language debate. Theoretical and empirical advances [Trends in Linguistic Studies]. Berlin & New York: Mouton de Gruyter. Matras, Yaron & Jeanette Sakel. 2007. Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31. 829–865.

282 

 References

Matthews, Stephen. 2010. Language contact and Chinese. In: Raymond Hickey (ed.), Handbook of language contact, 757–769. Oxford: Blackwell Reference Online. http://www.blackwell‑ reference.com/public/book.hyml?id=g9781405175807_9781405175807 (accessed 2 November 2014). Matthews, Stephen & Virginia Yip. 1994. Cantonese. A comprehensive reference grammar. London: Routledge. Matthews, Stephen & Virginia Yip. 2009. Contact‑­induced grammaticalization: a view from bilingual acquisition. Studies in Language 33. 366–395. Matthewson, Lisa. 1999. On the interpretation of wide‑­scope indefinites. Natural Language Semantics 7. 79–134. Michaelis, Laura. 2003. Headless constructions and coercion by construction. In: Elaine Francis & Laura Michaelis (eds.), Mismatch. Form‑­function incongruity and the architecture of grammar, 259–310. Stanford: CSLI Publications. Michaelis, Laura. 2004. Type shifting in construction grammar: an integrated approach to aspectual coercion. Cognitive Linguistics 15:1–67. Mønnesland, Svein. 1984. The slavonic frequentative habitual. In: Casper de Groot & Hannu Tommola (eds.), Aspect bound, 53–76. Dordrecht: Foris. Mufwene, Salikoko. 1981. Non‑­individuation and the count/mass distinction. Papers from the 17th Regional Meeting of the Chicago Linguistics Society. 221–238. Mufwene, Salikoko. 1990. Transfer and the substrate hypothesis in creolistics. Studies in Second Language Acquisition 12. 1–23. Mufwene, Salikoko (ed.). 1993. Africanisms in Afro‑­American language varieties. Athens: The University of Georgia Press. Mufwene, Salikoko. 1996. The Founder Principle in creole genesis. Diachronica 8. 83–134. Mufwene, Salikoko. 2001. The ecology of language evolution. Cambridge: Cambridge University Press. Mufwene, Salikoko. 2008. Language evolution, contact, competition and change. London & New York: Continuum. Muysken, Pieter C. 1981. Half‑­way between Quechua and Spanish: The case for relexification. In: Arnold R. Highfield & Albert Valdman (eds.), Historicity and variation in creole studies, 52–79. Ann Arbor MI: Karoma. Muysken, Pieter & Norval Smith (eds.). 1986. Substrata versus universals in creole genesis. Amsterdam & Philadelphia: John Benjamins. Myers-Scotten, Carol. 2002. Contact linguistics. Bilingual encounters and grammatical outcomes. Oxford: Oxford University Press. Nau, Nicole. 1995. Möglichkeiten und Mechanismen kontaktbewegten Sprachwandels ‑­unter besonderer Berücksichtigung des Finnischen. Münich: Lincom. Newbrook, Mark. 1987. Aspects of the syntax of educated Singaporean English. Frankfurt: Peter Lang. Newbrook, Mark. 1997. Malaysian English: status, norms, some grammatical and lexical features. In: Edgar Schneider (ed.), Englishes around the world 2:.Caribbean, Africa, Asia, Australasia. Studies in honour of Manfred Görlach, 229–256. Amsterdam & Philadelphia: John Benjamins. Nöel, Dirk. 2007. Diachronic construction grammar and grammaticalization theory. Functions of Language, 14. 177–202. Ng, Chye Ping. 1999. Experiential aspect in Colloquial Singapore English. Dept. of English Language and Literature, National University of Singapore Honours thesis.

References 

 283

Ooi, Vincent, Peter K‑­W Tan & Andy K‑­L Chiang. 2008. Analyzing personal weblogs in Singapore English: the Wmatrix approach. www.helsinki.fi/varieng/journal/volumes/02/ooi_et_al/ (accessed 10 March 2009). OED Online = Oxford English Dictionary Online, 3rd edition (updated September 2012). http://www.oed.com.ezproxy.univ‑­paris.fr (accessed 7 November 2014). Panther, Klaus‑­Uwe & Linda Thornburg. 2000. The EFFECT FOR CAUSE metonymy in English grammar. In: Antonio Barcelona (ed.), Metaphor and metonymy at the crossroads. A cognitive perspective [Topics in English Linguistics 30], 215–231. Berlin & New York: Mouton de Gruyter. Patrick, Peter. 2004. Jamaican Creole morphology and syntax. In: Bernd Kortmann, Edgar W. Schneider, Clive Upton, Rajend Methrie & Kate Burridge (eds.), Handbook of varieties of English. Vol. 2: Morphology and syntax. Berlin & New York: Mouton de Gruyter (draft version). Pietsch, Lukas. 2009. Hiberno‑­English medial‑­object perfects reconsidered: A case of ­contact‑­induced grammaticalisation. Studies in Language 33. 528–568. Pietsch, Lukas. 2010. What has changed in Hiberno‑­English: Constructions and their role in contact‑­induced change. Language Typology and Universals 63. 118–145. Petruck, Miriam. 1996. Frame semantics. In: Jeff Verschueren, J‑­O. Östman, J. Blommaert, & C. Bulcaen (eds.), Handbook of pragmatics, 1–13. Amsterdam & Philadelphia: John Benjamins. Platt, John T. 1975. The Singapore English speech continuum and its basilect ‘Singlish’ as a ‘creoloid’. Anthropological Linguistics 17. 363–374. Platt, John T. 1977. The subvarieties of Singapore English and their sociolectal and functional status. In: William J. Crewe (ed.), The English language in Singapore, 83–95. Singapore: Eastern Universities Press. Platt, John T. 1978. The concept of a ‘creoloid’. Exemplification: Basilectal Singaporean English. Papers in pidgin and creole studies, Series A, No. 54: 53–65. (Canberra: The Australian National University). Platt, John T. 1987. Communicative functions of particles in Singapore English. In: Ross Steele & Terry Threadgold (eds.), Language Topics: Essays in Honour of Michael Halliday, vol. 1., 391–401. Amsterdam & Philadelphia: John Benjamins. Platt, John T. 1989. The nature of indigenized Englishes: Interference, creativity, universals. Language Sciences 11. 395–407. Platt, J.T. 1991. Social and linguistic constraints on variation in the use of two grammatical variables in Singapore English. In: Jenny Cheshire (ed.), English around the world: Sociolinguistic perspectives, 376–387. Cambridge University Press. Platt, John T. & Ho Mian Lian. 1980. Discourse particles in Singaporean English: substratum influences and universals. World Englishes 2. 215–221. Platt, John T. & Heidi Weber. 1980. English in Singapore and Malaysia: Status, features, functions. Kuala Lumpur: Oxford University Press. Platt, John T., Heidi Weber & Mian Lian Ho. 1984. The New Englishes. London: Routledge & Kegan Paul. Poplack Shana, Sali Tagliamonte & Ejike Eze. 2000. Reconstructing the source of Early African American English plural marking: a comparative study of English and Creole. In: Shana Poplack (ed.), The English history of African American English, 73–105. Malden, Massachusetts: Blackwells.

284 

 References

Queller, Kurt. 2003. Metonymic sense shift: Its origins in hearers’ abductive construal of usage in context. In: Hubert Cuykens, René Diren & John R. Taylor (eds.), Cognitive Approaches to Lexical Semantics, 211–241. Berlin & New York: Mouton de Gruyter. Quirk, Randolph & Wrenn C.L. 1957. An Old English grammar. London & New York: Methuen. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey & Svartvik. Jan. 1985. A comprehensive grammar of the English language. London: Longman. Radden, Günter & Zoltán Kövecses. 1999. Towards a theory of metonymy. In: Klaus-Uwe Panther & Günter Radden (eds.), 1999. Metonymy in language and thought [Human Cognitive Processing 4], 17–59. Amsterdam & Philadelphia: John Benjamins. Rickford, John R. 1996. Copula variability in Jamaican Creole and African American Vernacular English: A reanalysis of De Camp’s texts. In: Gregory R. Guy, Crawford Feagin, Deborah Schiffrin & John Baugh (eds.), Towards a social science of language: Papers in honor of William Labov. Vol. 1. Variation and change in language and society [Current Issues in Linguistic Theory 127], 357–372. Amsterdam & Philadelphia: John Benjamins. Roberts, Sarah J. 2000. Nativisation and the genesis of Hawaiian Creole and diffusion. In: John Rickford & Suzanne Romaine (eds.), Creole genesis, attitudes and discourse: Studies celebrating Charlene J. Sato, 45–70. Amsterdam & Philadelphia: John Benjamins. Romaine, Suzanne. 1999. The grammaticalization of the proximative in Tok Pisin. Language 75. 322–346. Rosch, Eleanor H. 1978. Principles of categorization. In: Eleanor Rosch & Barbara B. Lloyd (eds.), Cognition and categorization, 27–48. Hillsdale, NJ: Lawrence Erlbaum. Ruiz de Mendoza Ibañez, Francisco José & Ricardo Mairal Usón. 2008. Levels of description and constraining factors in meaning construction: an introduction to the Lexical Constructional Model. Folia Linguistica 42. 355–400. Sag, Ivan, Hans Boas & Paul Kay. 2012. Introducing Sign‑­Based Construction Grammar. In: Hans Boas & Ivan Sag (eds.), Sign‑­based construction grammar, 1–29. Stanford: CSLI Publications. Sasse, Hans‑­Jürgen. 2002. Recent activity in the theory of aspect: accomplishments, achievements, or just non‑­progressive state? Linguistic Typology 6. 199–221. Schneider, Edgar W. 2003. The dynamics of new Englishes: from identity construction to dialect birth. Language 79. 233–281. Schumann, Christian L. 1793. Neger‑­Englishes Wörter‑­Buch [Dutch‑­English dictionary]. Editio tertia. Ms Paramaribo (Moravian Archives Paramaribo 12/648). Sharma, Devyani. 2005. Language transfer and discourse universals in Indian English article use. Studies in Second Language Acquisition 27. 535–566. Sharma, Devyani. 2009. Typological diversity in New Englishes. English World‑­Wide 30. 170–195. Shibatani, Masayoshi. 1991. Grammaticization of topic into subject. In: Elizabeth Closs Traugott & Bernd Heine (eds.), Approaches to grammaticalization, vol. 2, 93–133. Amsterdam & Philadelphia: John Benjamins. Shibatani, Masayoshi. 2002. Introduction. Some basic issues in the grammar of causation. In: Masayoshi Shibatani (ed.), The grammar of causation and interpersonal manipulation [Typological Studies in Language 48], 1–22. Amsterdam & Philadelphia: John Benjamins. Singler, John V. 1989. Plural marking in Liberian Settler English, 1820–1980. American Speech 64. 40–64.

References 

 285

Singler, John V. 1990. The impact of decreolization upon T‑­M‑­A: Tenselessness. mood and aspect in Kru Pidgin English. In: John V. Singler (ed.), Pidgin and creole tense‑­mood‑­aspect systems, 203–230. Amsterdam & Philadelphia: John Benjamins. Singler, John V. 1991. Social and linguistic constraints on plural marking in Liberian English. In: Jenny Cheshire (ed), English around the world: Sociolinguistic perspectives, 544–561. Cambridge: Cambridge University Press. Smith, Carlota. 1991. The parameter of aspect. 1st edition. Dordrecht: Kluwer. Stefanowitsch, Anatol. 2003. A construction‑­based approach to indirect speech acts. In: Klaus‑­Uwe Panther & Linda Thornburg (eds.), Metonymy and pragmatic inferencing [Pragmatics and Beyond New Series 113], 105–126. Amsterdam & Philadelphia: John Benjamins. Stewart, Michele M. 2011. The expression of number in Jamaican Creole. Journal of Pidgin and Creole Languages 26. 363–385. Sudipa, I. Nengah. 1986. The acquisition of English by Balinese students. MA thesis, Monash University. Szmrecsanyi, Benedikt & Bernd Kortmann. 2008. Vernacular universals and Angloversals in a typological perspective. In: Markku Filppula, Juhani Klemola & Heli Paulasto (eds.), Vernacular universals and language contacts, 33–53. London: Routledge. Tagliamonte, Sali, Shana Poplack & Ejike Eze. 1997. Plural marking patterns in Nigerian Pidgin English. Journal of Pidgin and Creole Languages 12.103–129. Talmy, Leonard. 2000. Toward a cognitive semantics. Vol .I: Concept structuring systems. Cambridge, Mass.: The MIT Press. Teow, A.L. 1988. Modal auxiliaries in Singaporean English. National University of Singapore Faculty of Arts and Sciences academic exercise. Thomason, Sarah G. 2001. Language contact: An introduction. Edinburgh & Washington, DC: Edinburgh University Press & Georgetown University Press. Thomason, Sarah G. & Terrence Kaufmann. 1988. Language contact, creolization, and genetic linguistics. Berkeley, Los Angeles & London: University of California Press. Thomson, John T. 1864. Glimpses into life in Malayan lands. London: Richardson and Company. (Re‑­issued by Singapore: Oxford University Press, 1991.) Todd, Loreto. 1979. Some day been dey: West African Pidgin folktales. London: Routledge & Kegan Paul. Tomasello, Michael. 2000. Do young children have adult linguistic competence? Cognition 74. 209–253. Tongue, Ray K. 1974[1979]. The English of Singapore and Malaysia. Singapore: Eastern Universities Press. Traugott, Elizabeth C. 1988. Pragmatic strengthening and grammaticalization. Proceedings of the Berkeley Linguistics Society 14 (Parasession): 406–416. Traugott, Elizabeth Closs. 2007. The concepts of construction mismatch and type‑­shifting from the perspective of grammaticalization. Cognitive linguistics 18. 523–557. Traugott, E.C. 2008. Grammaticalization, constructions and the incremental development of language: suggestions from the development of degree modifiers in English. In: Regine Eckhardt, Gerhard Jäger & Tonjes Veenstra (eds.), Variation, selection, development: Probing the evolutionary model of language change, 219–250. Berlin & New York: Mouton de Gruyter.

286 

 References

Traugott, Elizabeth Closs. 2012a. Grammatical constructionalization: Rethinking ­grammaticalization in the light of constructionalization. Seminar presented in collaboration with Graeme Trousdale, University of Santiago de Compostela, October 17th. Traugott, Elizabeth Closs. 2012b. Constructional changes and constructionalization. Seminar presented in collaboration with Graeme Trousdale, University of Santiago de Compostela, October 16th. Traugott, Elizabeth Closs. 2014. Towards a constructional framework for research on language change. In: Sylvie Hancil & Ekkehard König (eds.), Grammaticalization ‑­theory and data [Studies in Language Companion Series 162], 87–105. Amsterdam & Philadelphia: John Benjamins. Traugott, Elizabeth Closs (to appear). Towards a coherent account of grammatical ­constructionalization. Traugott, Elizabeth Closs & Richard Dasher. 2002. Regularity in semantic change. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs & Graeme Trousdale. 2013. Constructionalization and constructional changes. Oxford: Oxford University Press. Trousdale, Graeme. 2008a. Constructions in grammaticalization and lexicalisation: Evidence from the history of a composite predicate construction in the history of English. In: Graeme Trousdale & Nick Gisborne (eds.), Constructional approaches to English grammar, 33–67. Berlin & New York: Mouton de Gruyter. Trousdale, Graeme. 2008b. Words and constructions in grammaticalization: The end of the English impersonal construction. In: Susan Fitzmaurice & Donka Minkova (eds.), Studies in the history of the English language IV: Empirical and analytical advances in the study of English language change, 301–326. Berlin & New York: Mouton de Gruyter. Trousdale, Graeme. 2010. Issues in constructional approaches to grammaticalization in English. In: Katerina Stathi, Elke Gehweiler & Ekkehard König (eds.), Grammaticalization. Current views and issues, 51–71. Amsterdam & Philadelphia: John Benjamins. Trousdale, Graeme. 2012. Grammaticalization, constructions and the grammaticalization of constructions. In: Kristin Davidse, Tine Breban, Lieselotte Brems & Tanja Mortelmans (eds., in collaboration with Bert Cornillie, Hubert Cuyckens & Torsten Leuschner), Grammaticalization and language change. New reflections, 167–198. Amsterdam & Philadelphia: John Benjamins. Trousdale, Graeme & Thomas Hoffmann (eds.). 2013. The Oxford handbook of construction grammar. Oxford: Oxford University Press Trudgill, Peter & Hannah, Jean. 2002[1982]. International English: A guide to varieties of Standard English. London: Arnold. Tsao, Feng‑­Fu. 1995. Tan lun Taiyu u de wu zhong yongfa jiji wen de guanxi [On the five functions of Taiwanese u and other related problems]. Zhong Guo Yu Wen Yanjiu 11. 155–167. Van Coetsem, Frans. 1988. Loan phonology and the two transfer types in language contact. Dordrecht: Foris. Van der Auwera, Johan. 1993. Already and still: beyond duality. Linguistics and Philosophy 16. 613–653. Van Rooy, R. 1999. The specificity of indefinites. Ms., University of Amsterdam. Verkuyl, Hank J. 1993. A theory of aspectuality. The interaction between temporal and atemporal structure. Cambridge: Cambridge University Press.

References 

 287

Von Heusinger, Klaus. 2002. Specificity and definiteness in sentence and discourse structure. Journal of Semantics 19. 245–274. Wee, Lionel. 2004. Reduplication and discourse particles. In: Lisa Lim (ed.), Singapore English: A grammatical description, 105–126. Amsterdam & Philadelphia: John Benjamins. Weinreich, Uriel. 1964[1953]. Languages in contact. Findings and problems. Den Haag: Mouton. Weinreich, Uriel. 1954. Is a structural dialectology possible? Word 10. 388–400. Winford, Donald. 2003. An introduction to contact linguistics. Oxford: Blackwell. Winford, Donald. 2005. Contact‑­induced changes. Classifications and processes. Diachronica 22. 373–427. Wong, Jock. 2004. The particles of Singapore English: a semantic and cultural interpretation. Journal of Pragmatics 36. 739–793. Wong, Wee Kim (ed.). 2011. Census of population 2010. Statistical release 1: Demographic characteristics, education, language and religion. Singapore: Department of Statistics. Wu, Jiun-Shung. 2008. Terminability, wholeness, and the semantics of experiential guo. Journal of East Asian Linguistics 17. 1–32. Yap, Foong‑­Ha & Shoichi Iwasaki. 2003. From causatives to passives: a passage in some East and Southeast Asian languages. In: Eugene H. Casad & G. Palmer (eds.), Cognitive linguistics and non‑­Indo‑­European languages [Cognitive Linguistics Research 18], 419–445. Berlin: Mouton de Gruyter. Yap, Foong Ha & Stephen Matthews. 2008. The development of nominalizers in East Asian languages. In: Maria José López‑­Couso & Elena Seoane (eds.), Rethinking ­grammaticalization: New perspectives [Typological Studies in Language 76], 309–341. Amsterdam & Philadelphia: John Benjamins (pre‑­print copy). Yeom, J.‑­I. 1998. A presuppositional analysis of specific indefinites: Common grounds as structured information states. New York: Garland. Ziegeler, Debra P. 1995. Diachronic factors in the grammaticalization of counterfactual implicatures in Singaporean English. Language Sciences 17. 305–328. Ziegeler, Debra P. 2000a. Hypothetical Modality. Grammaticalisation in an L2 Dialect [Studies in Language Companion Series 51]. Amsterdam & Philadelphia: John Benjamins. Ziegeler, Debra P. 2000b. A possession‑­based analysis of the ba‑­construction in Mandarin Chinese. Lingua 1100. 807–842. Ziegeler, Debra P. 2003. On the zero‑­plural in commercial Singaporean English. In: David Deterding, Adam Brown & Ee‑­Ling Low (eds.), English in Singapore: Research on grammar, 48–57. Singapore: McGraw Hill. Ziegeler, Debra P. 2004. Reanalysis in the history of do: A view from construction grammar. Cognitive Linguistics 15. 529–574. Ziegeler, Debra P. 2006. Interfaces with English aspect. Diachronic and empirical studies. Amsterdam & Philadelphia: John Benjamins. Ziegeler, Debra. 2007. A word of caution on coercion. Journal of Pragmatics 39. 990–1028. Ziegeler, Debra P. 2008a. Grammaticalisation under Control: Towards a functional analysis of same‑­subject identity‑­sharing. Folia Linguistica 42. 401–451. Ziegeler, Debra. 2008b. Propositional aspect and the development of modal inferences in English. In: Werner Abraham & Elisabeth Leiss (eds.), Modality‑­aspect interfaces. Implications and typological solutions, 43–79. Amsterdam & Philadelphia: John Benjamins. Ziegeler, Debra P. 2010. Count‑­mass coercion, and the perspective of time and variation. Constructions and Frames 2. 33–73.

288 

 References

Ziegeler, Debra P. 2012a. Towards a composite definition of nominal modality. In: Werner Abraham & Elisabeth Leiss (eds.), Covert patterns of modality, 343–383. Cambridge: Cambridge Scholars Series. Ziegeler, Debra. 2012b. On the interaction of past tense and potentiality in Singaporean Colloquial English. Language Sciences 34. 229–251. Ziegeler, Debra P. 2014a. Replica grammaticalisation as recapitulation: the other side of contact. Diachronica 31. 106–141. Ziegeler, Debra P . 2014b. Are constructions dialect‑­proof? The challenge of English variational data for construction‑­grammar research. Handbook of world Englishes. DOI: 10.1093/ oxfordhb/9780199777716.013.31 Ziegeler, Debra P. To appear. Historical replication in contact grammaticalisation. Ziegeler, Debra P. & Sarah Lee. 2006. Causativity reduction in Singapore English. English World‑­Wide 27. 265–294. Ziegeler, Debra P. & Sarah Lee. 2009. A metonymic analysis of Singaporean and Malaysian English causative constructions. In: Klaus‑­Uwe Panther, Linda L. Thornburg & Antonio Barcelona (eds.), Metonymy in grammar and discourse [HCP 25], 291–322. Amsterdam & Philadelphia: John Benjamins.

Corpora used CLMET(EV): The Corpus of Late Modern English Texts (Extended Version), compiled by Hendrik de Smet. https://perswww.kuleuven.be/~u0044428/clmet.htm. Helsinki Corpus: Helsinki Corpus of English texts, Diachronic Part. Compiled by Matti Rissanen, Merja Kytö, Leena Kahlas‑­Tarkka, Matti Kilipiö, Saara Nevanlinna, Irma Taavitsainen, Terttu Nevalainen, Helena Raumolin‑­Brunberg. Available from The Oxford text Archive. ICAME International Computer Archive of Modern and Medieval English. http://icame.uib.no/. The ICE Corpus: The International Corpus of English: http://ice‑­corpora.net/ ice/avail.htm. ICE‑­GB [=The ICE Corpus (Great Britain): The International Corpus of English: http://ice‑­corpora. net/ ice/avail.htm.] ICE‑­IND [= The ICE Corpus (India): The International Corpus of English: http://ice‑­corpora.net/ ice/avail.htm.] ICE‑­SIN [= The ICE Corpus (Singapore)): The International Corpus of English: http://ice‑­corpora. net/ ice/avail.htm.]

Index Aboh, Enoch 231 acrolect 21, 24 adstrate 7, 17, 40, 204, 261 agentivity 11, 79, 102n, 111–112, 219 Aikhenvald, Alexandra 217, 219 Aktionsart 32, 144, 146 allo-constructions 259 already 30, 129, 156, 170, 172–173, 175, 231, 231n, 265, 268 Alsagoff, Lubna 22, 24, 26, 28–29, 31, 34–35, 39, 168, 183, 195, 249, 256 analogy 51–52, 64, 78, 92, 177, 212, 222, 256, 259, 270 animacy-definiteness hierarchy 200 Ansaldo, Umberto 2, 7, 17, 21n, 24, 127, 261–262 anterior(ity) aspect/clause 11–12, 30, 120–121, 148, 152–153, 156, 158–159, 160, 162–164, 167–168, 170, 174n , 175, 178, 215, 268 Aorist 152 areal (features) 30, 129, 217, 237 Aristotelian Square of Oppositions 138 back-formation 11, 140, 252–253, 268 Backus, Ad 100 Bakker, Peter 2, 23 Bao, Zhi Ming (Zhiming) 1–2, 15–17, 24, 30, 33–36, 96, 117, 120, 126–127, 172, 216, 229–231, 235–237, 265–266, 270 basilect 22, 24, 26, 39, 283 Bencini, Gulia M.L. 88 benefactive (construction, roles) 86, 92–93, 111n, 218 Bergen, Benjamin K. 43, 236 Bickerton, Derek 17, 21, 39, 184, 189, 241 Bioprogram (Bickerton’s) 183 bleaching 63, 78, 90–91, 251 Boas, Hans 42–44, 47–53, 74, 78, 238 Bolton, Kingsley 17 borrowing 14, 23–25, 157n, 217, 219–221, 224, 277 Brinton, Laurel 244 Brisard, Frank 143

Brown, Adam 38, 67, 93, 120 Bruyn, Adrienne 127, 223–225 Butler, Susan 23 Bybee, Joan 32, 32n, 41, 52, 61, 70n, 112, 144, 149–150, 163, 170, 172, 240, 269 calque 23, 24, 35, 38, 121, 220, 240, 269 Cameroon English Pidgin 189 Cantonese 15, 20, 37–38, 95, 97, 123, 153, 203, 236, 240, 261, 282 Cappelle, Bert 259 caused motion (construction) 8–9, 42, 50, 52, 52n, 57, 58, 69, 74, 87 Chappell, Hilary 15, 121, 123, 126, 126n, 139, 141–142, 176, 252, 268 Changsha 123 Chaozhou dialect (see Teochew) Chaudenson, Robert 16 Cheng, Robert L. 152–153 Chiang, Andy 154, 156 Claridge, Claudia 154, 197 coercion 67–68, 68n, 69, 74, 181–182, 184, 204–212, 214, 257–258, 269, 272, 287 Cognitive Construction Grammar 53–54, 272 Cognitive Grammar 55, 57, 280 Colleman, Timothy 42, 42n Communicative Pressure 127–128 Comrie, Bernard 32n, 35, 144–145, 150–151, 151n, 172, 175, 266, 277 congruence 219 construct 48n, 63 constructicon 42, 46, 61, 74, 214, 216, 249, 262, 267–269 constructionalisation 55, 61–62, 62n, 63, 66, 73, 266 contact dialect 2, 4, 7, 9, 12, 67, 105, 127, 181, 200, 232, 245 continuity, semantic 7, 8, 91–92, 117, 248, 250 continuum, sociolectal 22, 27, 39 convergence 4, 7, 30, 38, 117, 131, 215–218, 220, 226, 234–235, 239, 242, 246, 253, 258–259, 271, 276, 281 conversational implicature 92, 141

290 

 Index

convertible construction 259 copying – global 220 – material 24, 36, 219, 241, 258 – partial 220 – pattern 36, 95, 219, 222, 236, 239, 246, 258, 260 – polysemy 223–225, 228 Corne, Chris 219 creoloid 6, 17, 21, 21n, 283 Crewe, William 20, 31, 283 Cristofaro, Sonia 149–150 Croft, William 3, 43–46, 48, 54, 57, 62, 65, 71, 80, 236, 238, 242, 265 countability 13, 182–184, 200, 270 Cultural Orientation Model (COM) 26–28 Dahl, Östen 30, 129, 150–151, 172, 265 Davidse, Kristin 86, 286 Davydova, Julia 18 DeCamp, David 21 definiteness 183, 186, 188, 195, 200–203, 240, 275 DeGraff, Michel 231 DeSwart, Henrietta 67, 210, 241 Deterding, David 31, 287 Detges, Ulrich 224 diasystemic construction grammar, diaconstructions 245–246, 278 diglossia, diglossic 1, 9, 17, 22, 24, 26–28, 31, 40, 232, 249, 275 discourse particle 24, 37, 37n, 38, 197n, 276, 287 ditransitive 8, 42, 48n, 50, 52n, 61 Dutch (Modern) 90, 133, 138–139, 142, 237, 252 East African English 40 equivalence 13, 225–228, 237, 239, 247, 252, 277 evidential 121, 123, 126, 126n, 139, 176, 176n, 273 feature pool 261–262 Ferguson, Charles 22, 275 Fillmore, Charles 41, 49, 51, 73, 186

frame (semantics) 49–52, 55, 214, 283 free-choice (item) 118–119, 136, 140–141, 278 free variation 31–32 Fried, Mirjam 8, 43, 48, 53–55, 57, 272, 280 fusion 218–219, 239, 249, 259, 281 Fuzhouese 123 Gan 123 Gast, Volker 130 German 64, 133, 177, 185, 188, 199, 217, 220–222, 225, 227–228, 245–247, 252 Gil, David 195, 200–201, 227, 240 Givón, Talmy 12, 89n, 127n, 144, 144n, 149, 184–185, 202 Goldberg, Adele 2–3, 8, 10, 43–44, 47, 47n, 49–52, 57–58, 63, 64–65, 77–78, 80–82, 100, 234, 242, 272 grammaticalisation 7–8, 14n, 33, 40–42, 60–61, 63–67, 72–73, 77, 107–108, 112–113, 115, 141–142, 145, 173, 189–190, 199, 218, 225–226, 228, 235, 237, 240, 247, 265, 273, 280, 283 – accelerated 127–128, 131 – contact/replica 7, 118–119, 127, 127n, 128–131, 137, 142, 173, 217, 220, 221–227, 230, 237, 247, 265, 270 – hyper- 33, 113, 239 – hypo- 113 Guerts, Bart 184–186 Gumperz, John J. 217–218 Gupta, Anthea 6, 9, 15–20, 22–25, 27, 39, 209, 271 Hagège, Claude 127 Haitian Creole 228–229, 280 Hakka 15, 123, 153, 261 Haspelmath, Martin 74, 80, 129, 136, 237 Hawaiian Creole 188, 197, 284 Heine, Bernd 6–7, 13, 40, 63, 66, 100, 118, 127–129, 160, 172, 198–199, 217–220, 222–227, 238, 241, 244, 247, 252, 260, 265, 270 Hewson, John 150, 174 Hiberno English (see Irish English) Himmelmann, Nikolaus 61

Index 

Ho, Mian Lian 9, 11–12, 19, 30–32, 37, 117, 119–125, 137, 145–148, 154–155, 183, 197, 239 Höder, Steffen 1, 13, 245–247, 260 Hoeksema, Jack 138 Hokkien 15, 19, 27, 36–38, 94–97, 119, 123–127, 152, 157n, 197n, 236, 240, 261, 268, 272 Hong Kong English 183, 272 Hopper, Paul 39, 42, 56, 65–66, 83, 85, 108, 112, 126, 169, 206, 224, 226, 248, 275 Horn, Laurence 133, 136n, 137, 137n, 138, 141 Hudson, Richard 4, 59 hyperanalysis 72, 203 Idealized Cognitive Model 82 immanent aspect 174 implicational perfective 254–255, 264 importation 220 Indian English 35, 40, 94n, 198, 198n, 199, 223, 284 individuation 187, 200–201, 206, 209, 256–257, 282 inducee-conflation 81 Inheritance Links 8, 52, 52n, 109–110, 110n innovation 12, 220, 239, 241, 247, 254 interlingual identification 242, 245, 247, 249, 255, 258, 260 Irish 131, 226, 243–244, 245 Irish English 35, 226, 242 irrealis 32, 144, 144n, 149, 150, 160, 173–174, 174n, 175–179, 274 isomorphism 217–218, 220, 231, 258, 277 Israel, Michael 118, 118n, 119, 132, 132n, 133, 141 Jamaican Creole 21, 187–189, 197, 283 Jespersen, Otto 133, 137 Johanson, Lars 217 Kaltenböck, Gunther 37 Kay, Paul 41–43, 73, 275 Keesing, Roger M. 221 Kejia dialect (see Hakka) Kemmer, Suzanne 89 Kövecses, Zoltán 82, 108–109, 212

 291

Krifka, Manfred 176 Kru (creole) 150 Kuteva, Tania 6, 39–40, 118, 125, 127–129, 190, 198–199, 217–220, 222–223, 225–227, 238, 241, 252, 270, 277, 279 Kwan, Terry, Anna 37n, 38–39 Lakoff, George 82 Lakoff, Robyn 90, 92 lamitives 30, 128, 266 Langacker, Ronald W. 55–57, 59, 74, 112, 172–174, 176, 236, 250 layering 126, 226 Lee, Sarah 10, 79 Lefebvre, Claire 123, 219, 228–231, 233, 265, 270 Lehmann, Christian 61, 108, 239 Lehmann, Winifred 111 Leimgruber, Jacob 15–16, 22, 26–28 Leiss, Elisabeth 176–177, 188, 287–288 Leuschner, Torsten 132–133, 136, 286 Levin, Beth 52 Lexical Construction Model 49n lexical diffusion 70n, 78 Lexical Prototype Construction 69, 71–72, 210 lexifier 4, 6, 30, 36, 58–59, 128, 131, 216, 224, 228–229, 230–233, 236, 265 – filter 30, 33, 121, 231, 236–237 Li, Charles 80, 108, 112–113, 153, 171, 202–203, 235, 280 Liberian Settler English 190, 285 Lim, Lisa 15–16, 19, 37, 37n, 38, 197n, 287 lingua franca 10, 16–17, 20, 26, 128, 183 loanshift 219 loan translation 156, 221–222, 247 Low, Ee Ling 67, 93, 120, 274 Macedonian 218 Macedonian Turkish 218, 235 macro/meso-construction 60–61, 64 macro-roles 84 Malay 7, 15–18, 36–38, 94–95, 97–98, 119, 124–127, 135, 201, 216, 234 – Baba 94 – Bazaar 16, 19, 37 Malayalam 150

292 

 Index

malefactive 36, 93 Mandarin 1, 7, 16, 19n, 20, 27–28, 33–38, 94–96, 108, 119–120, 123–124, 146, 152, 172, 183, 202, 226, 229–231, 231n, 232, 240, 249, 281 Matras, Yaron 2–4, 7, 130–131, 218–219, 219n, 220–222, 230, 233, 235, 238, 246–247 Matthews, Stephen 15, 129–130, 153, 203, 226–228, 266, 287 Media Lengua 23 medial perfect 131, 244 mesolect 21–22, 188n metaphor 70, 108, 182, 210–13 metonymy 8–9, 67, 69–70, 79, 82, 90, 108–110, 114, 138, 182 , 205, 208, 210–214, 269, 279, 283–284 , 288 – grammatical 82 Michaelis, Laura 12, 63, 67–68, 195, 204–208, 210–211 micro-construction 60–61 middle voice 111 mini constructions 49–50, 52–54, 71, 78 minimizer 11, 119, 138, 141, 242, 253, 263–264, 268 mismatch 67–69, 204–205, 208, 211, 214, 282 mixed language 2, 19, 23, 248, 264, 266, 271, 281 modal (verb) 31–33, 40, 45, 149–150, 162–163, 177, 179, 223, 252, 273 model language 6, 8, 13, 58, 128, 130, 131, 172–173, 218–219, 221, 223, 226, 228, 233, 238–239, 241, 247, 252, 260, 270 Mufwene, Salikoko 29, 36, 127, 183, 187–188, 202, 219, 223–224, 228, 233 Muysken, Pieter 36 Nau, Nicole 14, 221 negative polarity 11, 118–119, 122–123, 125, 132–134, 136, 137n, 138–141, 214, 236, 242, 252–253, 268 Newbrook, Mark 18–19, 25, 29 New English(es) 4, 216, 223, 283–284 Nigerian Pidgin English 190, 193, 201, 285 Noël, Dirk 65–66, 71 nominal modality 181, 302

Old English 64, 132–133, 138, 174, 174n, 176, 189, 223 Ooi, Vincent 154, 156 Override Principle 205, 207 Panther, Klaus-Uwe 82, 90, 108, 114, 274, 277, 284–285, 288 Patrick, Peter 188, 188n passive 36, 36n, 88–89, 89n, 90–91, 98, 107, 109–111, 114, 130, 231–232, 272 performatives 12, 171–173, 174n, 175–176, 178, 239, 245, 254–255, 264, 268–269 pidgin/China Pidgin English 16–17, 17n, 39, 127–128, 189–190, 201, 216, 221, 223, 274 Pietsch, Lukas 1, 13, 131, 242–245, 247, 260, 270 pivot 141–142, 221–222, 223, 247, 258, 260, 263, 266 pivot-matching 130–131, 142, 222, 230, 244, 249–250, 252–254 Platt, John 1, 5, 9, 11, 15, 17, 19, 20–21, 23–24, 29–30, 34, 37, 39, 143, 145–148, 170, 197, 239 polysemy 7, 14, 14n, 50–51, 130, 131, 221–223, 225, 227–228, 250, 259, 260 Poplack, Shana 190, 201, 285 precedence 12, 178, 216, 254–255, 259, 263, 269 preterite-presents 12, 176, 177 pre-transitive marker 96, 235 Principle of Maximised Motivation 77 Principle of Relevance 145 progressive (aspect) 40, 68–69, 94n, 120, 149, 172, 173–174, 242, 254, 284 Proto-roles 77 punctual aspect/verbs 12, 31, 145–146, 162, 255 quantifier 25, 119, 132–133, 136–137, 137n, 138, 142, 183, 201, 203, 231, 253 Queller, Kurt 140 Radden, Günter 9, 82, 108–109, 205, 210, 213, 274

Index 

Radical Construction Grammar 46, 56–57, 62, 274 reanalysis 62, 117, 137, 140, 142, 224, 249, 252, 256, 277 Recipient language 219 referentiality 136, 159, 182, 184, 186, 190, 194–195, 200, 202–203, 207–208, 211, 213–214 reflexive, reflexive causativity 88, 89n, 92, 106–107, 234, 249, 267 reification 227 relabelling (see relexification) relexification 219, 224, 228–230, 234, 236–237, 247, 258, 265, 270, 280 repeatability 121n, 123, 135, 141, 252–253, 259, 263–264 replacement, lexical/syntactic 108 replica language 6–7, 13–14, 59, 128, 130, 220–223, 226–228, 230, 239, 242, 242–248, 260 – pre-contact 6, 233, 248–249, 252 restructuring 14, 216, 225, 265, 275 resultative/causative-resultative 50, 79–83, 88, 88n, 89–91, 108–110, 110n, 171, 235, 267, 276 retention, lexical 34, 249 retentionist (approach) 39, 131, 242 Romaine, Suzanne 39 Ruiz de Mendoza Ibañez, Francisco 71, 280 Sag, Ivan 42–43, 184, 275 Sakel, Jeanette 131, 219–222, 230, 233, 238, 246–247, 258 Sasse, Hans-Jürgen 32, 144 selection restrictions 87, 112, 114, 206, 235 Serbo-Croatian 151–152, 268 serial verb 107–108, 249 Schneider, Edgar 5, 19, 275 Shanghainese 123 Sharma, Devyani 40, 198, 198n Shibatani, Masayoshi 80, 111, 113, 211, 295 Singler, John 150, 190 Sinti Romani 220–221

 293

Slavic 144, 150–151, 224, 227 Smith, Carlota 32n, 36, 144–145 Source language 36, 59, 75, 118, 131, 141–142, 216, 219, 231, 248 Southern Min dialect (see Hokkien) Speak Mandarin Campaign 20, 94 Sranan 223–224 Stefanowitsch, Anatol 110 Stewart, Michele 187–188 subjectivity 185 Subject-Predicate construction 47, 54, 113, 206, 208 substrate 6–7, 13–14, 19, 24, 28–30, 33–34, 40, 94–95, 117, 123, 128–129, 142, 198–200, 221, 228–230, 232–233, 261–262, 265–266, 268 Superventive Relations 83 systemic transfer 172, 228, 236, 239, 247, 258, 265, 272 Szmrecsanyi, Benedikt 209 Tagliamonte, Sali 190, 201, 283 Talmy, Leonard 81, 83–84, 89 Tamil 7, 16, 18–19, 94–95, 98, 150, 199, 216 Tan, Peter K. W. 154–156 Teochew 15, 94, 261 Thomason, Sarah G. 23, 217, 219–220 Thornburg, Linda 82, 90, 108, 114, 288 Tomasello, Michael 74 tone 37, 37n, 38 topicalisation, topic-prominence 4, 11, 35, 79, 111–114, 211, 235, 235n, 251, 261, 266–267, 272, 281, 284 transitivization 80–81 transnumerality 198, 203, 209, 269 Traugott, Elizabeth 3–4, 8, 39, 41–43, 45, 54, 57, 59, 60–61, 62–66, 71n, 72–73, 117, 205, 213, 224, 265n, 266 Trousdale, Graeme 3, 8, 45, 59, 63–66, 74, 77, 272–274, 276, 278, 286 Trinidadian English 223 Turkish 80, 137, 217–218, 235, 244 unboundedness 256–257, 259, 264, 269 Undergoer (role) 83, 83n, 84–85, 88, 106, 107

294 

 Index

Van der Auwera, Johan 130, 266 verb-specific constructions 3, 52, 71 voice-conversion 88, 91 Von Heusinger, Klaus 185–186 Wee, Lionel 36–37, 231, 271 Weinreich, Uriel 242, 245 Winford, Donald 219

Wong, Jock 37–39 Wong, Richard 95 Yap, Foong Ha 88, 95n, 98, 125, 227 Yip, Virginia 129–130, 153, 203, 226–228, 266 Ziegeler, Debra 32–33, 42, 69–70, 113, 145, 174, 190–191, 199, 214, 240, 269