Reconciling Synchrony, Diachrony and Usage in Verb Number Agreement with Complex Collective Subjects [1 ed.] 9780367417154, 9780367815899, 9780367643423

Citation preview

i

Reconciling Synchrony, Diachrony and Usage in Verb Number Agreement with Complex Collective Subjects

This book uses corpus-​based methodologies to investigate the wide variety of factors behind verb number agreement with complex collective noun phrases in English. The literature on collective nouns and their agreement patterns spans an array of disciplines and approaches. However, little of the research conducted to date has focused on the influence of of-​dependents on verb number with relational collective nouns, as in examples such as a bunch of or a group of. Drawing on data from two case studies –​one based on the Corpus of Historical American English (COHA), and the other on the British National Corpus (BNC) and the Corpus of Contemporary American English (COCA)  –​Fernández-​Pena uses statistical modelling to unpack the different morphological, syntactic, semantic and lexical dimensions of the variables affecting verb number agreement with complex collective noun phrases in English. This multidimensional analysis of the significance of of-​dependents in the patterning and contemporary usage of collective nouns offers new insight into and understanding of both synchronic variation and diachronic change. This book is an essential read for scholars of English language variation and change, historical linguistics, corpus linguistics, and usage-​based approaches to the study of language. Yolanda Fernández-​Pena is a postdoctoral fellow in the Department of English, French and German at the University of Vigo, Spain, and a member of the Language Variation and Textual Categorisation (LVTC) research group. She has published in international peer-​ reviewed journals, such as Corpora, Atlantis, RAEL: Revista Electrónica de Lingüística Aplicada and Varieng.

ii

Routledge Studies in Linguistics

22 Systemic Functional Language Description: Making Meaning Matter Edited by J.R. Martin and Y.J. Doran 23 Rarely Used Structures and Lesser-​Studied Languages Insights from the Margins Emily Manetta 24 Externalization Phonological Interpretations of Syntactic Objects Yoshihito Dobashi 25 Approaches to the Study of Sound Structure and Speech Interdisciplinary Work in Honour of Katarzyna Dziubalska-​Kołaczyk Agnieszka Kiełkiewicz-​Janowiak, Magdalena Wrembel and Piotr Gąsiorowski 26 Casting a Minimalist Eye on Adjuncts Stefanie Bode 27 Analysing Scientific Discourse from A Systemic Functional Linguistic Perspective A Framework for Exploring Knowledge Building in Biology Jing Hao 28 The Discourse of Desperation Late 18th and Early 19th Century Letters by Paupers, Prisoners, and Rogues Igor Timmis 29 Reconciling Synchrony, Diachrony and Usage in Verb Number Agreement with Complex Collective Subjects Yolanda Fernández-​Pena For more information about this series, please visit: www.routledge.com/​ Routledge-​Studies-​in-​Linguistics/​book-​series/​SE0719

iii

Reconciling Synchrony, Diachrony and Usage in Verb Number Agreement with Complex Collective Subjects Yolanda Fernández-​Pena

iv

First published 2020 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Taylor & Francis The right of Yolanda Fernández-​Pena to be identified as author of this work has been asserted by her in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-​in-​Publication Data A catalog record has been requested for this title ISBN: 978-​0-​367-​41715-​4  (hbk) ISBN: 978-​0-​367-​81589-​9  (ebk) Typeset in Sabon by Newgen Publishing UK

v

To my grandmothers

vi

vi

Contents

List of figures  List of tables  Acknowledgements  List of symbols and abbreviations  1 Introduction 

ix xi xiii xiv 1

1.1  Aims and scope of the study  4 1.2  The whys and wherefores  6 1.3  Outline of the book  9

2 Complex collective subjects and verb number agreement in English: State of the art 

11

2.1  Complex collective subjects  13 2.1.1.  Lexico-​semantic traits: Lexical profile and quantifying meaning  14 2.1.2.  Morphosyntactic characterisation: Headedness and constituency  23

2.2  Verb number agreement with complex collective subjects  28 2.2.1.  Comprehensive grammars  29 2.2.2.  Syntactically oriented approaches  32 2.2.3.  Cognitive Grammar and the symbolic alternative  39 2.2.4.  Empirical studies  41

2.3  Concluding remarks  49

3 Insights from diachrony: Reconciling form and meaning  3.1  Methodology: Corpora and data retrieval  56 3.2  Analysis of the data  58 3.2.1.  A number of  59 3.2.2.  A group of  64 3.2.3.  A majority of  69 3.2.4.  A bunch of  73 3.2.5.  A couple of  78 3.2.6.  A host of  80

55

vi

viii Contents 3.2.7.  A minority of  84 3.2.8.  Discussion of the results  85

3.3  Summary and final remarks  97

4 Modelling variation in verb number agreement with complex collective subjects in Present-​Day English 

103

4.1  Methodology: Corpora and data retrieval  104 4.1.1.  Corpora  104 4.1.2.  Data retrieval  105

4.2  Variables  108 4.2.1.  Core defining variables  109 4.2.2.  Morphosyntactic variables  109 4.2.3.  Lexico-​semantic variables  115 4.2.4.  Extralinguistic variables  117

4.3  Statistical modelling of the data  117 4.4  Analysis of the data  120 4.4.1.  Morphosyntactic variables  130 4.4.2.  Lexico-​semantic variables  142 4.4.3.  Regional variation: British, American and World Englishes  168

4.5  Summary and final remarks  176

5 Concluding remarks and prospects for future research  References  Index of collective nouns and complex collective noun phrases  Index on subjects 

182 188 204 206

xi

Figures

3.1 Frequency of a number of with and without premodification (N=33,054) and of a MOD number of (N=10,649) in the whole COHA corpus 3.2 Frequency of verb number agreement with a group of in the COHA data (N=470) 3.3 Frequency of a group of with and without premodification (N=12,015) and a MOD group of (N=2,418) in the whole COHA corpus 3.4 Frequency of a MOD majority of in the whole COHA corpus (N=1,063) 3.5 Frequency of verb number agreement with a bunch of in the COHA data (N=76) 3.6 Frequency of a bunch of with and without premodification (N=4,557) and a MOD bunch of (N=662) in the whole COHA corpus 3.7 Frequency of verb number agreement with a host of in the COHA data (N=70) 3.8 Frequency of a host of with and without premodification (N=2,208) and a MOD host of (N=137) in the whole COHA corpus 3.9 Frequency of a minority of with and without premodification (N=527) and a MOD minority of (N=207) in the whole COHA corpus 4.1 Intercept adjustments for the random effect ‘Ncoll’ 4.2 Importance of the variables in the optimal model of the BNC and COCA 4.3 Effect of ‘Det1’ on the predicted probability of plural agreement 4.4 Effect of the interaction ‘type of N2 * number of postmodifiers of N2’ on the predicted probability of plural agreement 4.5 Effect of ‘semantic plurality’ on the predicted probability of plural agreement 4.6 Effect of ‘countability’ on the predicted probability of plural agreement

63 65 67 72 74 76 82 83 86 126 129 131 137 143 145

x

x  List of figures 4.7

Effect of ‘animacy’ on the predicted probability of plural agreement 4.8 Importance of the variables in the optimal models of the BNC and COCA 4.9 Intercept adjustments for the random effect ‘variety’ in GloWbE 4.10 Importance of the variables in the optimal model of GloWbE 4.11 By-​variety importance ranking of all the variables examined in GloWbE

147 170 172 174 175

xi

Tables

3.1

Frequency of verb number agreement with a number of in the COHA data (N=1,609) 3.2 Frequency of verb number agreement with a majority of in the COHA data (N=206) 3.3 Token and normalised frequency (per 10,000 words) of a majority of and the majority of in the whole COHA corpus (N=9,410) 3.4 Summary of the syntactic and semantic parameters of idiomatisation observed in Ncoll-​of-​Npl constructions 3.5 Classification of Ncoll-​of-​Npl constructions according to degree of idiomatisation and quantifier use/​meaning 4.1 Frequency of verb number agreement with the 23 collective nouns 4.2 Mixed-​effects regression model of verb number agreement variation with complex collective subjects (predicted odds are for plural agreement) 4.3 Intercept adjustments for the random effect ‘regional variety’ 4.4 Top ten positive and negative intercept adjustments for the random effect ‘verb base’ 4.5 Output of the mixed-​effects regression model for the fixed effect ‘complexity’ (predicted odds are for plural agreement) 4.6 Frequency of verb number agreement of the seven collective nouns in referential, partitive and pseudopartitive configurations 4.7 Verb-​collexemes distinguishing between plural and singular verb agreement 4.8 Ncoll-​collexemes distinguishing between plural and singular verb agreement 4.9 N2-​collexemes distinguishing between plural and singular verb agreement 4.10 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take plural agreement 4.11 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and mainly human referents

60 70 73 87 90 108 122 124 125 140 150 154 157 159 161 164

xi

xii  List of tables 4.12 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and animate non-​human referents 4.13 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and inanimate referents 4.14 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and variable referents

166 167 168

xi

Acknowledgements

First and foremost, I would like to express my deepest and most sincere gratitude to Prof. Javier Pérez-​Guerra for his invaluable support, encouragement and guidance over these years, especially in the hardest moments. Javier, thanks for sharing your (contagious) passion for research and for so generously taking the time to agree and disagree with me on the puzzling topic of collective nouns for hours on end. The completion of this book would not have been possible but for your insightful review of parts of the manuscript and your boundless patience. For generous financial support, I am grateful to the following institutions: the University of Vigo, the Regional Government of Galicia, the Spanish State Research Agency and the European Regional Development Fund (grant nos. PRE/​2013/​396, FPU13/​01509, ED481B-​2019/​077, FFI2013-​44065-​P, FFI2014-​51873-​REDT, FFI2016-​77018-​P, GPC2014/​060, ED431C 2017/​50, ED431D 2017/​19). During my research stays at the Universities of Leuven and Zurich I had the opportunity to discuss my research with Prof. Kristin Davidse, Prof. Marianne Hundt, Prof. Benedikt Szmrecsanyi, Dr. Katrien Verveckken and Dr. Lieselotte Brems. I would like to warmly thank them for their careful comments and constructive feedback, from which the research reported in this book has greatly benefited. I am also deeply grateful to the team members of my research group, Language Variation and Textual Categorisation (LVTC, University of Vigo), and the research network English Linguistics Circle (ELC) coordinated by Prof. Teresa Fanego. A special word of thanks is owed to Ana Elina Martínez-​ Insua, Nuria Yáñez-​Bouza and David Tizón-​Couto for their kind help and valuable advice. A  heartfelt appreciation goes to Selene Molares-​ Pascual, Evelyn Gandón-​Chapela, Carla Bouzada-​Jabois, Sofía Bemposta-​Rivas, Zeltia Blanco-​Suárez, Francisco Gallardo-​del-​Puerto and Andrea Ruthven for their unconditional support, both academically and personally. My last debt of gratitude is to my family and friends, those who give meaning to my life and lift me up in the hardest moments. Their patience and wholehearted encouragement and support helped give shape to this book far more than they could imagine.

vxi

List of symbols and abbreviations

* ungrammatical example ? marked, hardly acceptable example % grammatical in some dialect(s) only + /​-​/​+? positive, negative and dubious evidence ACAD (academic prose) AgrOP (Object Agreement Phrase) AgrSP (Subject Agreement Phrase) AmE (American English) AusE (Australian English) BNC (The British National Corpus) BrE (British English) CanE (Canadian English) COCA (Corpus of Contemporary American English) COHA (Corpus of Historical American English) Det1 (determiner of the collective noun) Det2 (determiner of the oblique noun) FIC (fiction) GloWbE (Corpus of Global Web-​Based English) iF (semantic component of number) IP (Inflectional Phrase) IrE (Irish English) LF (Logical Form) MAG (magazine) Mod (modifier) N(pl) ((plural) (oblique) noun) N1 (first noun in a binominal noun phrase) N2 (second noun in a binominal noun phrase) Ncoll (collective noun) Ncoll(-​)of(-​)N(pl) (complex collective noun phrase) n.f. (normalised frequency) NF (non-​fiction) NM L (nominal) NN0 (POS-​tag, noun neutral for number) NN1 (POS-​tag, singular noun)

v x

newgenprepdf

List of symbols and abbreviations xv NN2 (POS-​tag, plural noun) NN2.irregular (irregular plural noun) NN2.s (regular plural noun) NP (Noun Phrase) NZE (New Zealand English) OED (Oxford English Dictionary) OR (odds ratio) PDE (Present-​Day English) PF (Phonetic Form) P L (plural) PP (Prepositional Phrase) QUA N T (quantifier) S G (singular) Spec (Specifier) SPOK (spoken) TP (Tense Phrase, non-​finite sentence) uF (morphological component of number) VP (Verb Phrase)

xvi

1

1  Introduction

Received wisdom says that collective nouns in English may show singular or plural agreement patterns based on their inherent notional plurality (Quirk et al. 1985: 757–​759; Biber et al. 1999: 188–​189). A group, for instance, may be interpreted holistically as a unity or a homogenous set, as in the case of (1) below, or as a plurality or aggregate with a distributive reading (most commonly in British English; Quirk et al. 1985: 757–​759; Biber et al. 1999: 188–​ 189; Huddleston & Pullum 2002: 502), as in (2) (Biber et al. 1999: 185): ( 1) [The second group of books] is those written by botanists.1 (2) Under a nearby tree, [a group of children] were having their school lunch. Can we assume, therefore, that this dual conceptualisation is the main driving force behind verb number variation? Examples like (3)  and (4)  suggest not, given the unlikelihood of variable agreement. (3) [A third group of closed back items] consists/​*consist of “antique-​finish” rugs. [BNC: 1985–​1993 EX0 1396] (4) In rural areas, [the number of private BTS houses] was/​*were twice that in the public sector. [BNC: 1985–​1993 K5D 12101] For agreement variation to be possible, the verb must be applicable to either the collective as a whole or the individuals that compose it (Biber et al. 1999: 189). This condition renders the plural patterning in examples (3) and (4) unacceptable, as consist and be can only be used to refer to the group and the number (not to the items and the houses, respectively). Despite wide acknowledgement of the dual interpretation of collective nouns and their flexible verb number patterns, verb number variation with collective nouns is generally believed to be more confined to informal registers, with written language following formal agreement patterns (5) (Quirk et al. 1985: 758–​759; Huddleston & Pullum 2002: 502–​503). Complex collective phrases like (6) and (7) appear to flout that precept, however (Huddleston & Pullum 2002: 501–​504): (5) The committee has not yet come to a decision.

2

2 Introduction (6) [A bunch of hooligans] %was/​were seen leaving the premises. (7) [The majority of her friends] are/​*is Irish. To what extent do these cases differ? The answer lies in the prepositional phrase (henceforth of-​PP). In (5), the necessary conditions to allow for verb number variation are met: committee is a prototypical collective noun that denotes an aggregate of (human) animate entities (see Depraetere 2003: 86–​89 and Corbett 2004: 188), while the verb come allows for a focus on either the group (has come) or the individuals (have come) (see Levin 1998: 105, 2001: 148–​158; Biber et al. 1999: 189 and Depraetere 2003: 102, 116–​ 123). Plural agreement with bunch and majority in (6) and (7) also seems to be motivated by a distributive reading (i.e. by the focus on hooligans and her friends), which is more readily available owing to the explicit reference to the members of the collective in the of-​PP. The availability of a singular reading in these cases is more limited, however, with dual conceptualisation being possible but unlikely in the case of bunch and impossible in the case of majority. The quantifying meaning of bunch and the quantificational interpretation of majority seem to motivate plural verb agreement in these instances (Quirk et  al. 1985: 765; Biber et  al. 1999: 185; Huddleston & Pullum 2002: 503–​ 504). Compare these examples with (8) and (9) below: whereas plural override was clearly obligatory in the case of majority above, the possible collective sense of bunch is observed more clearly in (8). The plural pattern of group in (9) has been explained in terms of the distributive reading of the collective noun or in terms of ‘attraction’ or ‘proximity concord’, as it also occurs with singular verbs (Levin 2001: 140): (8) [A bunch of flowers] was presented to the teacher (Huddleston & Pullum 2002: 503) (9) [A group of visitors] have paid more than £100 a head to watch her kill. (Levin 2001: 140) Attraction refers to the influence of a nominal element other than the head noun (e.g. visitors in (9)) on the number of the verb and, from a psycholinguistic point of view, on the erroneous assignment of verb number (see Bock & Miller 1991; Bock & Cutting 1992; Bock & Eberhard 1993; Bock, Nicol & Cutting 1999). While attraction is the main explanation proposed by some scholars for the agreement patterns of collective nouns that take of-​dependents, such as group in (9) or homologous terms such as flock (see Depraetere 2003: 86), the following examples challenge that assumption: (10) [A batch of reports on the safety of Galecron, that were submitted by Ciba-​Geigy to the World Health Organisation in 1978] have never been published. [BNC: 1985–​1993 B77 283] (11) And the bed looks like [a troop of soldiers] has been over it. [BNC: 1985–​ 1993 HGV 5611]

3

Introduction 3 (12) [A whole flock of AIRPORT PERSONNEL] are now conferring. [COCA: 1994 FIC Mov:OnlyYou] In contrast to (9), in (10) plural verb number is selected even though the plural noun reports is not adjacent to the verb (noun and verb are actually separated by 17 words), which makes it quite unlikely that the plural patterning is the result of attraction only. In (11), by contrast, the lack of intervening material between the plural noun soldiers and the verb does not affect number agreement. Interestingly, in (12) the singular noun personnel co-​occurs with a plural verb, despite the fact that none of the nouns in the subject is formally plural. Examples (13) and (14) are structurally very similar but only the former takes a plural verb. This suggests that the type of entities denoted, which in this case differ in their degree of animacy, should not be dismissed as a potential explanatory factor. (13) [A whole pack of German destroyers] were after them. [BNC: 1985–​ 1993 CEH 3088] (14) Then he ran off down the road like [a pack o’ dogs] was after him. [BNC: 1985–​1993 HUA 356] Animacy is also closely related to the type of collective noun, as not all collective nouns that take of-​PPs (Biber et al. 1999: 249; Huddleston & Pullum 2002: 503) can be used with all types of referents. Aggregates of people may be referred to as groups, bunches or crowds, even as flocks or herds if a hyperbolic meaning is intended; however, references to clumps or shoals of people seem as unlikely as the use of a crowd of objects or a herd of papers. If all the cases presented above may be attributed to the attraction exerted by the noun within the of-​PP, are we to assume that other formal and semantic aspects of the of-​PP and/​or its internal complexity have no bearing on the subject-​verb agreement relationship? It seems unlikely. The following examples highlight the potential effect of presence and type of postmodifier of the of-​PP (15), occurrence of additional preverbal plural nouns (16) and selection or lack of a determiner in the of-​PP  (17): (15) a. The example Winch cites is from the film Violent Saturday in which [a gang of bank robbers] hide from the police [BNC: 1985–​1993 C9B 484] b. [a gang of bank robbers, [masquerading as an unlikely string quartet,]] engages in a battle of wills [BNC: 1985–​1993 A7L 1083] (16) [A crowd of people [of various races, clutching passports and sheaves of official documents,]] were clustered around an office in the foyer of the Questura. [BNC: 1985–​1993 HTT 46] (17) a. suddenly [a bunch of East Coast accounts] appear on her credit report and are not being paid [COCA: 1997 MAG ConsumRep] b. [A bunch of the gifts] were birds: partridges, geese a-​laying, swans a-​swimming, and so forth. [COCA: 2008 FIC Jack&Jill]

4

4 Introduction All of the ideas and observations presented here are in line with an array of recent studies on the factors responsible for subject-​verb agreement patterns of collective nouns. These analyses challenge the traditional explanation of verb number variation in terms of conceptualisation and regional factors, and reveal evidence of the role of syntactic factors, such as distance (Levin 1998, 2001), co-​referential pronouns (Dekeyser 1975: 35–​66; Levin 1998, 1999, 2001) and type of determiner (Levin 1998, 2001); lexico-​semantic variables, such as type and meaning of the verb (Levin 1998, 2001), animacy of the referent (Dekeyser 1975: 35–​66; Levin 1998, 2001) and type of collective noun (Dekeyser 1975: 35–​66; Bauer 1994, 2002; Siemund 1995; Levin 2001; Algeo 2006: 279–​285; Lakaw 2017), as well as variation across other varieties of English (Levin 2001; Hundt 2009b; Smith 2009; Collins 2015) and the diachronic evolution of their verb number preferences (Siemund 1995; Levin 1999, 2001). The role of the of-​PP has been largely bypassed by these studies, however, reduced to a sub-​category of the phenomenon of attraction (Depraetere 2003) or explored as a binary factor to compare the effect of its absence with that of its presence in the subject (Dekeyser 1975: 35–​66), or considered in relation to the quantifying meaning of certain collective nouns when used in complex noun phrases (Smitterberg 2006; Smith 2009; Klockmann 2017: 211–​ 273; Zhang 2017: 53–​ 77; Leclercq & Depraetere 2018). Meanwhile, the formal and lexico-​semantic aspects of this prepositional constituent, its implications and potential interference in the agreement relation, and the quantifying potential of certain collective nouns that take of-​ dependents have remained unexplored. The research reported in this book is an attempt to bridge this gap in the literature.

1.1.  Aims and scope of the study My intention in this study is to uncover the determinants of verb number agreement with collective nouns that take of-​ dependents in English. The research focuses on the role played by the morphosyntactic and lexico-​ semantic features of the of-​PP –​also the collective noun and the noun phrase as a whole  –​in determining subject-​verb agreement. The research is based on the assumption that the of-​PP and its constituent elements are a decisive factor in determining the patterning and present-​day usage of the collective nouns they accompany. My objective is to expand existing research on collective nouns and agreement, and to shed new light on this relatively neglected area of inquiry. In particular, the study will attempt to answer the following research questions: (i) Is there evidence of a diachronic evolution? Have there been any significant changes in relation to complex collective subjects or collective nouns more generally which may have influenced their current verbal patterning and meaning?

5

Introduction 5 (ii) What is the quantifying potential (if any) of complex collective subjects? To what extent does the interaction between the of-​PP and verb agreement contribute to this use? (iii) Is lexis a determining factor? Is verb number agreement affected by the type of verb, type of collective noun or type of noun in the of-​PP? (iv) What determines verb number choice in the case of complex collective subjects: the collective noun, the prepositional phrase or the structure as a whole? (v) To what extent (if at all) do the form and/​or the semantics of the of-​PP and/​or the other elements in the subject affect the use of singular or plural verb number? These questions will be explored from a purely descriptive, usage-​ based perspective and supported by corpus data. Though numerous theoretical frameworks are mentioned and acknowledged throughout the book, the study itself does not adhere to any particular theory, opting instead to describe and explain variation in verb number selection using real (native) data for a set of collective nouns that take of-​dependents. The data are analysed from different morphosyntactic and lexico-​semantic perspectives, both synchronically and diachronically, to ensure a comprehensive, multifaceted, usage-​based account of the topic. Data for the study were extracted from written samples from three of the largest (balanced) corpora of English available. To obtain an in-​depth, large-​ scale lexical perspective, data were selected from the Corpus of Historical American English (COHA, 1810–​2009) (Davies 2010​–). In the absence of comparable data from British sources, data from American English were used to carry out an initial assessment of the extent to which the patterns of agreement of complex collective noun phrases in Present-​ Day English have been influenced by diachronic change and the emergence of quantifying usages of collective nouns. To address the last two questions posed by this research, Present-​Day English (PDE) data were retrieved from the British National Corpus (BNC, 1960–​ 1993) and the Corpus of Contemporary American English (COCA, 1990–​2012) (Davies 2008​–). My survey of PDE focuses mainly on the of-​dependent and its effect on agreement, with additional analysis of the collective noun and the noun phrase subject as a whole. Data selected from the BNC and COCA were scrutinised to assess the influence on verb agreement of morphosyntactic and lexico-​semantic features of the noun phrase subject. This multidimensional analysis also provides useful insight into the third research question by assessing the impact of specifically lexical factors on verb number selection. The complex collective subjects surveyed by this research comprise a singular collective noun (Ncoll), such as crowd in (18) and (19) below, and an of-​PP which contains a noun (N(pl), known as the ‘oblique noun’) that denotes the entities that form the collective. The oblique noun is frequently plural, as in spectators and women below, but is also found as a singular countable

6

6 Introduction (e.g. committee (20)) and uncountable (e.g. mankind (21)). Given the binominal2 structure of these complex collective subjects, I  also refer to them as ‘Ncoll(-​)of(-​)N(pl) (subjects)’. (18) By the end [a crowd of soaked spectators] cheers an ear-​shattering finale that challenges the storm itself. [COCA: 1997 MAG AmHeritage] (19) [A crowd of LUSTY WOMEN] cheer on a STRIPPER IN A FIREMAN’S SUIT. [COCA: 1999 FIC Mov:InShadowOak] (20) [A minority of the Committee] were opposed to this proposal. [BNC: 1985–​1993 GW1 791] (21) Even so, [the majority of mankind] do not find this view practicable or desirable. [BNC: 1975–​1984 EFT 328] The contribution of this study to research on verb agreement with collective noun phrases is its examination of a type of collective noun that has been largely overlooked up to now (Levin 2001; Depraetere 2003; cf. Dekeyser 1975: 35–​66; Reid 1991: 267–​272; Smith 2009; Klockmann 2017: 211–​273; Zhang 2017: 53–​77; Leclercq & Depraetere 2018). Moreover, the study’s more integrationist approach to theory use and its observation and description of samples of natural language from a multifaceted diachronic and synchronic perspective covering an extensive list of potentially explanatory morphosyntactic and lexico-​semantic factors promises to open up new ways of understanding and thinking about the topic.

1.2.  The whys and wherefores Why a corpus-​based study? The research reported in this book uses an eclectic and descriptive perspective, interpreting the data as the natural development of actual language use and variation over time. Corpus Linguistics was deemed the most appropriate methodological framework, as it constitutes the only way to combine real data for language use and change, and a multifaceted examination of variation at the four levels of analysis outlined in the research questions: morphology, syntax, semantics and lexis. In this respect, the methodology for the current study is not completely new, as there are precedents in the literature (see Section 2.2.4). The innovation of this study lies in the following four features: (i) examination of verb number agreement with collective nouns that take of-​dependents only; (ii) use of a four-​layered analysis (i.e. morphology, syntax, semantic and lexis); (iii) statistical testing of an extensive list of variables; and (iv) a purely usage-​ based approach. The research is based on data from both British and American English corpora. Regional variation, though relevant, was considered as just one of multiple potential factors of verb agreement since my analysis in this study is not contrastive and, as Hundt (2009b) observes, variability in the agreement patterns of collective nouns seems to be mainly due to language-​ internal factors rather than regional variety.

7

Introduction 7 Why complex collective noun phrases? One fundamental determinant of verb number choice is the collective noun itself. Research on the topic has demonstrated the relatively free variability of some collective nouns; however, studies also report a growing overall preference for singular agreement patterns since the nineteenth century (see Section 2.2.4). Exceptions to this general trend include nouns that refer to whole communities or groups and those denoting number, especially if they take an of-​PP and (potentially) function as complex determiners or quantifiers, as in (22) and (23) (Dekeyser 1975: 57–​65; Levin 2001: 129–​148). (22) But [a couple of new books] are now making some best-​seller lists. (Levin 2001: 141) (23) [A number of buildings] are, doubtless, destroy’d. (Dekeyser 1975: 56) Cases such as those illustrated in (22) and (23) form the basis of this study. The data set of collective nouns consists of ‘quantifying collectives’ (Biber et  al. 1999: 249) and ‘number-​ transparent’ nouns (Huddleston & Pullum 2002: 503), two types of nominal elements which have a general (potentially quantificational) meaning and combine with a (mainly plural) of-​PP to specify the type of entities comprised by the collective. The set contains 23 singular collective nouns in total: band, batch, bunch, class, clump, couple, crowd, flock, gang, group, herd, host, majority, minority, number, pack, party, rash, series, set, shoal, swarm and troop. Although Biber et al.’s label implies that quantity may be more frequent or prominent than the nouns’ collective sense, all of the items examined in this study are grouped together under the label ‘collective noun’ (henceforth, ‘Ncoll’) on the grounds that all of them refer to a collection of individuals (for a discussion of this term, see Gardelle 2019). The main reason for the selection of nouns was their relational nature: the presence (explicit or implicit) of the prepositional dependent specifying the members of the collective is obligatory, in contrast to more ‘prototypical’ collective nouns such as committee or family (Huddleston & Pullum 2002: 503). For this study, only examples with explicit of-​dependents were considered, based on the impact of of-​ complementation on verb number choice: where the singular collective noun takes an of-​PP (24), plural verbal forms are shown to be the preferred pattern in British and American English (65%), while in the absence of the dependent (25), a significantly higher rate of singular agreement is observed (80%) (see Fernández-​Pena 2017d; see also Dekeyser 1975: 45–​48). (24) On an empty lot between two Harlem streets, [a group of people] arrive for an outdoor party [BNC: 1985–​1994 AHA W_​newsp_​brdsht_​nat_​ arts] (25) If [a group] arrives in the morning …, then the group must be made comfortable whilst waiting to gain access to their rooms. [BNC: 1985–​1994 EA9 W_​commerce]

8

8 Introduction Despite the relational nature of these 23 collective nouns, the set is by no means homogeneous and evidence and opinions regarding their agreement patterns and quantifying potential vary considerably. Only a few are generally accepted as quantifying expressions in English: bunch (Brems 2011: 176–​ 191; Langacker 2013), couple (Dekeyser 1975: 64; Levin 2001: 140), number (Berg 1998; Keizer 2007: 121) and, to a lesser extent, group (Akmajian & Lehrer 1976; Smith, Franck & Tabor 2018), host (Smith 2009; Langacker 2016), majority (Reid 1991; Yuasa & Francis 2003), minority (Levin 2001: 140) and herd (Koptjevskaja-​Tamm 2001). One indication of this quantifying use is the expressions’ preference for plural agreement patterns (i.e. the verb agrees with the plural oblique noun rather than the singular collective noun), the only exceptions in this regard being number (Smitterberg 2006; Langacker 2010) and herd (see Depraetere 2003: 93), which have variable patterns. The heterogeneity within this subset of collective nouns is also observed in the originally quantifying meaning of nouns such as couple (Oxford English Dictionary (OED), s.v. couple I, II.7), majority (OED, s.v. majority n.1 I.3.a), minority (OED, s.v. minority A.3.a) and number (OED, s.v. number I), which contrasts with the more restricted set of collocates of collective nouns such as bunch (things ‘either growing together (as a bunch of grapes), or fastened closely together in any way (as a bunch of flowers, a bunch of keys)’; OED, s.v. bunch n.1 3), host (armed men; OED, s.v. host n.1 1), flock (birds and domestic animals; OED, s.v. flock n.1 2, 3)  and herd (domestic animals or animals ‘feeding or travelling in company’; OED, s.v. herd n.1 1, 2)  in their original collective senses. These collocational restrictions also apply to some of the remaining collective nouns in the set, such as band (armed men, robbers, assassins or musicians; OED, s.v. band n.3 1.a, 4.a), clump (mainly growing plants and micro-​organisms; OED, s.v. clump 2.a, d), shoal (animals, particularly fish; OED, s.v. shoal n.2 1) and troop (mostly soldiers but also animals; OED, s.v. troop 1). The rest (batch, class, crowd, gang, group, pack, party, rash, series, set and swarm) are nouns which have a wider collocational range and mainly variable agreement, and whose quantifying potential is less clear but not impossible (see Keizer 2007: 115–​116; Smith 2009). Why diachronically and synchronically? The synchronic study reported in Chapter 4 deals with the main aim of this book, which is to determine the morphosyntactic and lexico-​semantic variables responsible for verb number agreement variation with complex collective noun phrases (NPs) in Present-​Day English. However, the fact that some of the collective nouns in the study occur more frequently with plural patterns requires further attention, since it immediately introduces the possibility of grammaticalisation and idiomatisation (“a process whereby meaning becomes more opaque, and, syntactically, the phrasal unity becomes more fixed than before”; Akimoto 2002: 17). The role played by plural of-​PPs in the semantic opacity and syntactic fixation found in homologous binominal structures has already been noted by other studies in which collocation with these plural dependents is observed to be one of the factors involved in the desemanticisation,

9

Introduction 9 decategori(ali)sation and recategorialisation of the collective noun and the subsequent shift in headedness from the collective to the (often plural) oblique noun that leads to plural verbal patterns (see Brems 2003, 2004, 2011; Traugott 2008a, 2008b and also Section 2.1.1). Examples of this include cases such as a lot of (Traugott 2008a; Brems 2011) or a bunch of (Brems 2011), originally lexical nouns which have come to be used as quantificational elements in PDE, as illustrated by (26) and (27). (26) the moon had risen and was letting quite [a lot of light] into the bank (Traugott 2008a: 232) (27) I don’t just teach them [a bunch of facts] (Brems 2011: 180) Additionally, some of the 23 collective nouns analysed here show certain colligational preferences in terms of agreement. As will be discussed in detail in Chapter 4 (§§4.1.2; 4.4.2.4.iii), data from both the BNC and COCA show that the majority of items in the set (15) occur more frequently with singular patterns (>52%; band, gang, troop, party, series, crowd, swarm, flock, herd, pack, set, batch, class, rash and clump), while number shows an almost even distribution of singular and plural verbal patterns. The remaining seven nouns (couple, host, majority, minority, bunch, shoal and group) display a preference for plural verb number (>56%). The significance of these observations lies in the high incidence of plural verb number (>80%) with five of the collective nouns in this last group (couple, host, majority, minority and bunch). The diachronic corpus-​based investigation reported in Chapter 3 is an attempt to account for this extremely high rate of plural agreement. The study takes into account existing studies on the syntactic fixation of similar binominal NPs and their subsequent quantifying potential, together with my own recent analyses of the same data set (Fernández-​Pena 2017b). This historical approach to complex collective noun phrases as a potential source for quantifying expressions in English is based on the premise that a colligational preference for plural number may be symptomatic of the idiomatic or even grammaticalised status of the collective noun. The criteria used to select which collective nouns to include were rate of plural agreement and relative frequency in PDE. For this reason, only couple, host, majority, minority, bunch and group were considered, based on their preference for plural verb number (>56%). Shoal was excluded owing to its negligible incidence in the database (0.11%), despite occurring with plural verbs in more than 66% of cases. In the case of number, though it displays only a marginal preference for plural agreement (50.26%), its high frequency in the data set (14.50%), the considerable number of occurrences with a quantifying meaning and my findings from previous research (Fernández-​Pena 2017b) led me to include it in the study.

1.3.  Outline of the book Following on from this general introduction, Chapter 2 reflects upon research into complex collective noun phrases in English and verb number agreement, including comprehensive grammars, formal and cognitive-​functional studies.

01

10 Introduction Chapter  3 reports on the diachronic study carried out using data from the Corpus of Historical American English (COHA) and provides an in-​depth analysis of the collocational profile and agreement patterning of seven complex collective noun phrases from a quantitative and qualitative lexical perspective. Chapter 4 presents the results of the synchronic study based on data from the British National Corpus (BNC) and the Corpus of Contemporary American English (COCA). This analysis focuses in particular on the morphosyntactic, lexico-​semantic and regional factors that condition subject-​verb agreement with complex collective subjects in Present-​Day English. The book ends with a summary of the main findings and conclusions from the study and suggestions for future research.

Notes 1 Square brackets [ ]‌are used throughout the book to enclose complex collective noun phrases and highlight their internal structure. Italics, underlining and referential indices (e.g. sg /​ pl) are used to emphasise particular elements of the phrase. 2 The term ‘binominal’ is used here in a general sense to refer to a complex noun phrase (subject) containing two nominal elements linked by the preposition of: (Det1) N1 of (Det2) N2 (as in Verveckken 2015). This label has also been used in reference to fully lexical and referential uses of complex noun phrases in contrast to their extended and more idiomatic or grammaticalised usages (e.g. a bunch of grapes vs a bunch of nonsense; as in Denison 2002; Brems 2011). In its more restricted use, the term refers to complex phrases where the two nominals have a predicative relationship, as in a jewel of an island (Aarts 1998; Keizer 2007: 85–​108).

1

2  Complex collective subjects and verb number agreement in English State of the art

Collective nouns are a notoriously slippery concept. Traditionally, they have been defined as nouns that “imply a plural sense without a plural termination” (Bayly 1772: 66) and, as such, may take plural agreeing forms, as in His congregation receive his doctrine (Fogg 1792: 148, my underlined emphasis). Indeed, in anglophone research, verb agreement variation is considered to be the defining trait of collective nouns (Joosten 2006: 73), with this variable patterning attributed to a difference in conceptualisation whereby singular agreement focuses on the group and plural agreement (generally held to be more common in British English) promotes a distributive (i.e. individualistic) reading (Crystal 2000: 76). Recent research, however, has challenged this view by demonstrating that variability in number agreement is constrained by a range of morphosyntactic, lexico-​semantic and regional factors (see Section 2.2.4). Another point of disagreement among grammarians and researchers is the question of how to define the concept of collectivity itself. While some scholars distinguish between collective nouns which have a plural form and those which do not (e.g. crowd(s) vs nobility), with only the former being both morphologically and semantically plural (Nixon 1826: 120), others make a distinction between ‘collective nouns’, which denote a whole and take singular agreement (e.g. the jury consists of twelve persons), and ‘nouns of multitude’, which refer to the individuals and take plural verbs (e.g. the jury (the men on the jury) were divided in their opinions; Nesfield 1898: 9–​10). Inanimate non-​count nouns denoting plurality, such as furniture, and uninflected plural nouns, such as sheeppl or elephantpl, have sometimes been discussed under the label ‘collective’, although inanimate nominal elements are usually disregarded because of their invariable agreement patterns (see Evans & Evans 1957: 99–​100 and Pickett, Kleinedler & Spitz 2005: 94). Few studies to date have attempted to provide a systematic account of this wayward class of nouns. Depraetere (2003: 95) proposes a gradient of prototypicality, in which the most prototypical ‘collective’ is “morphologically singular with … multiple animate (inclusive or generic) reference”, with nouns like cattle in English being closest to the prototype. A more recent proposal by Gardelle (2019: 101; emphasis in the original) challenges earlier definitions and classifications by moving away from the traditional reliance on variable (i.e. ‘hybrid’) agreement and restricting the

21

12  Complex collective subjects and agreement term ‘collective noun’ to count nouns that express at the lexical level “a plurality of units construed as the result of a grouping operation”, as well as a part-​whole relation (collection, crowd, majority, minority, flock). These nouns, which are also characterised by a double layer of conceptualisation, are at the top of Gardelle’s (2019: 189) Scale of Unit Integration, a highly consistent system of semantic pluralities of units which also includes lexical (furniture), morphosyntactic (cat-​s) and discursive (lots of students) means of denoting a plurality of entities. Collective nouns found in complex noun phrases, such as flock, crowd, number, group and bunch, have frequently been termed ‘collective’ on the basis that they refer to groups or collections of people, animals or things (Nixon 1826: 120–​ 121; Jespersen 1909–​ 1949: 93–​ 108; Poutsma 1914: 283–​ 284; Dekeyser 1975: 35–​66; Levin 2001). Not all scholars agree with this analysis, however. In certain studies, the syntactic rigidity and potential quantificational meaning of some of these collectives have led to their exclusion as ‘collective nouns’. Quirk et  al. (1985: 250, 316–​ 317), for example, classify couple, crowd, flock, gang, group, herd, majority, minority and party as ‘collective nouns’, regardless of the implications of their possible occurrence with an of-​ dependent, while nouns such as bunch, series and pack are classed as ‘(quantifying) partitive nouns’. The collective nouns crowd, flock and herd are also included in this last category (1)–​(2): (1) a (large) crowd of people (2) a (huge) flock of birds/​sheep A different case is that of ‘nouns of quantity’, such as number, to which Quirk et  al. (1985: 264) attribute a more functional status as a phrasal quantifier when combined with an indefinite article and an of-​PP (i.e. a number of). Biber et al. (1999: 248) note the lack of a clear boundary between fixed quantifying expressions such as a number of and a couple of and collective nouns that take of-​complementation and denote quantifying meaning: batch of, bunch of, group of, series of, shoal of (referred to as ‘of-​collectives’ by Biber et  al. 2002: 61). According to Huddleston & Pullum (2002: 503), one of the main differences between these ‘quantifying collectives’ (Biber et al. 1999: 249) and ‘prototypical’ collective nouns (e.g. family, committee) is the number transparency of the former, which, in combination with the of-​complement, allows the second noun to determine the number of the NP, as demonstrated by the plural override and the quantifying meaning of the fragment ‘a bunch of’ in (3): (3) [A bunch of hooligans] were seen leaving the premises. As mentioned in Chapter 1, this patterning has sometimes been attributed to the phenomenon of ‘attraction’; that is, production of a plural verb form based on the plurality of the noun in the of-​PP. Depraetere (2003: 86), for example, contends that flock and similar “group terms … are almost uniquely used in combination with an of-​PP” and that their potential variability in agreement

31

Complex collective subjects and agreement 13 “is exclusively based on the principle of attraction or proximity”, and on that basis excludes them from her gradient of ‘prototypical’ collectives. Levin (2001: 140) also attributes the plural patterning of a group of N to attraction, but acknowledges that its variability in number agreement may also be due to the dual conceptualisation of the collective noun. Variability in number agreement has been agreed to have been lost in the highly idiomatic quantifying use of expressions such as a lot of, for example (Traugott 2008a; Brems 2011, 2012). However, the lexical and/​or grammatical status of other expressions, such as a group of, a host of and a majority of, is less clear. Against this, some studies use a less restrictive definition of collective nouns, grouping together words such as majority, number, bunch and couple and elements without of-​complementation or quantificational meaning, such as population and committee (Dekeyser 1975: 35–​66; Levin 2001; cf. Gardelle’s 2019 ‘organised plurality construction’ vs ‘quantificational idiom’). I  consider these words ‘collective’ on the grounds that they can still be used lexically, even though the boundary between their lexical and grammatical status is often fuzzy and uncertain. Throughout this book, therefore, I will use the terms ‘collective(s)’, ‘complex collective noun phrases’ and ‘Ncoll(-​)of(-​)N(pl) (patterns or constructions)’ interchangeably. The purpose of this chapter is to provide a detailed review of the research on complex collective noun phrases in English. The first section looks at their lexico-​semantic (§2.1.1) and morphosyntactic features (§2.1.2), and explores the interplay between their structural configuration and potential for quantification. The second part of the chapter focuses on how the question of the agreement patterns of collective nouns is covered in comprehensive grammars (§2.2.1) and model-​specific syntactic (§2.2.2) and cognitive-​functional (§2.2.3) analyses, including a review of the empirical research literature (§2.2.4). The chapter concludes with a summary of the main points discussed.

2.1.  Complex collective subjects Complex collective NPs are binominal phrases of the type a bunch of ideas or a flock of sheep, in which the first noun is assumed here to be a collective noun (Ncoll) which (i) denotes a collection of entities of any nature, (ii) takes an of-​ dependent (of N) that contains an (often plural) oblique noun (N(pl)) and (iii) may imply some nuance of quantity. These binominal phrases conform to the abstract syntactic configuration ‘(Det1) (Mod) N1 of (Det2) (Mod) N2’ (adapted from Brems 2011: 2), which in English covers a wide range of complex nominal phrases with a variety of functions and meanings: from locative NPs (e.g. the back of the house) and genitives (e.g. a friend of my brother’s), to the so-​ called ‘Type Noun construction’ (TN, e.g. that kind of dog) and ‘Binominal Noun Phrase’ (BNP, e.g. a hell of a problem), to name just a few (for further discussion, see Keizer 2007: 61–​186; Traugott 2008b: 27 and the references cited there). Like the complex collective constructions analysed here, these structures have also been a source of disagreement among scholars regarding number agreement, headedness, constituency and their varying degrees of

41

14  Complex collective subjects and agreement idiomaticity. Complex collective NPs range from more compositional and referential phrases, as in (4) and (5), to more idiomatic configurations, as in (6), which pose problems and conflicts for headedness and number agreement: (4) [a bunch of keys] is placed under the chair in the centre of a large circle. [BNC: 1985–​1993 AM6 802] (5) [A bunch of the other guys] come over [COCA: 2005 MAG RollingStone] (6) [A bunch of cases] are raising those issues. [COCA: 2007 ACAD ABAJournal] Despite their structural similarities, the complex collective NPs above differ in compositionality and meaning: while (4) is referential and (5) ‘partitive’, (6) is what is known as a ‘pseudopartitive’, a less compositional phrase in which the sequence a bunch of modifies or quantifies cases. This chapter aims to demonstrate that the syntactic configuration of complex collective NPs has a crucial effect on their constituency and interpretation, and also, as reported in Chapters 3 and 4, on their functionality and patterning. Although the referential use of complex collective NPs (4) will be mentioned, the chapter will focus on the other two syntactic configurations. The rest of this section examines the main characteristics of the pseudopartitive constructions that distinguish them from partitive and referential uses of complex collective NPs. 2.1.1.  Lexico-​semantic traits: Lexical profile and quantifying meaning Studies on binominal (pseudo)partitives often use the term ‘partitive’ to refer to the particular stage which precedes and gives rise to the purely quantificational use of structures such as a lot of in a lot of love, derived here from a metonymic extension of its original denotation (see Traugott 2008a: 230–​232 and Traugott & Trousdale 2013: 22–​26). The term ‘partitive’, however, is not used consistently in the literature. ‘Partitive’, as the name suggests, denotes partition and, as such, refers to constructions where a part of a larger whole is referred to (as in a bunch of those flowers). Nevertheless, the term has also been used in reference to the of-​PP itself, as a ‘partitive oblique’ (many of the delegates; Huddleston & Pullum 2002: 349), and to the semantics of the first nominal, as in ‘partitive nouns/​constructions’ (e.g. a piece/​bit/​crowd/​bunch of; Quirk et al. 1985: 249–​250; Smitterberg 2006).1 The most controversial use of ‘partitive’ relates to constructions such as (7)–​(8), which differ from purely partitives not only in meaning but also in definiteness. The lack of Det2 has a clear effect on the construction’s semantic (non-​partitive) interpretation. (7) a (large) crowd of people (Quirk et al. 1985: 250) (8) a number of votes (Smitterberg 2006: 254) Though scholars such as Traugott & Trousdale (2013: 23) found the distributional (and apparently also the semantic) differences between definite and

51

Complex collective subjects and agreement 15 indefinite2 of-​PPs “rather minimal” in English (see also Zhang 2017: 69), other authors highlight a range of formal and functional differences. Taking the example of partitive a lot of the delegates and non-​partitive a lot of people, for example, Huddleston & Pullum (2002: 349) note of the former that “the whole NP picks out a subset of the set referred to by the delegates, whereas there is no such subset relation involved in [a lot of people]”. To capture the difference in meaning and form between the two, Selkirk (1977) coined the term ‘pseudopartitive’ (also ‘pseudo-partitive’; a N of N), which has been taken up by numerous scholars, including Brucart (1997), Koptjevskaja-​Tamm (2001, 2009), Stavrou (2003), Alexiadou, Haegeman & Stavrou (2007), Rutkowski (2007), Stickney (2009), Keenan (2013) and Verveckken (2015), to name just a few. These authors observe the same syntactic and semantic differences between a bunch (i.e. ‘subset’) of the flowers and a bunch of (i.e. ‘many’) flowers, which are in turn distinct from the referential use of a bunch (i.e. ‘bouquet’) of flowers. In this study, partitive and pseudopartitive structures are treated as two different syntactic configurations encoding different meanings. The reasons and evidence to justify this approach are explained and discussed in detail below. Lexical profile This first subsection offers a lexical profile of partitive and pseudopartitive constructions, in preparation for the discussion of their quantifying meaning and function in the next subsection. Partitives Partitive meaning may be expressed intrinsically, as in the case of majority, which inherently evokes a totality out of which that majority is extracted (see Brucart 1997: 160–​161 and Section 3.2.3), or structurally, through the use of a complex noun phrase comprising either a numeral/​quantifier or a nominal element, as in (9) and (10), respectively (Selkirk 1977: 311–​312).3 Binominal partitive constructions such as (10), though not uncommon, have received little attention from researchers in this area: (9) Three/​Some of the women4 (10) A bunch of the flowers The embedded NP in a partitive phrase may contain a count noun, in which case the partitive denotes a subset-​set relation (11) (see Huddleston & Pullum 2002: 411), or a non-​count or singular count noun, which favours a part-​ whole interpretation (12): ( 11) a number of the old houses [BNC: 1985–​1993 ECS 1628] (12) a minority of the committee [BNC: 1985–​1993 GW1 791]

61

16  Complex collective subjects and agreement These examples and arguments demonstrate two important characteristics of partitives: first, that in partitives there are two referents with independent reference in the discourse, the subset/​part and the larger set/​whole, and, as a result, the denotation of the first element depends heavily on the second (Keizer 2007: 66). Closely related to this is the presence and definiteness of Det2: the NP following of in a partitive structure is generally agreed to be definite (see Huddleston & Pullum 2002: 349) or, more precisely, referential, meaning that it refers “either to an evoked or inferrable discourse entity or introduc[es] a new entity into the discourse” (Keizer 2007: 69; Leclercq & Depraetere 2018: 344). This revised ‘Partitive Constraint’ explains the ungrammaticality of cases such as (13) (see Keizer 2007: 68–​70): (13) *two of men In (13), men is not referential: it does not evoke or introduce new entities in the discourse and hence it cannot participate in a partitive construction “as it only makes sense to introduce a subset of a larger set if the hearer can be assumed to have (or be able to infer) a mental representation of this larger set” (Keizer 2007: 70). Pseudopartitives The term ‘pseudopartitive’ was coined to account not only for the difference in meaning between a bunch of flowers and a bunch of the flowers, but also for their different syntactic analysis. Selkirk (1977: 302) claims that only the former, unlike the partitive, is comparable to a simple NP formed by a monomorphemic quantifier (or numeral) and a noun (i.e. many flowers), since both indicate an indeterminate quantity. Similarly, in this study, the term ‘pseudopartitive’ is used in reference to examples such as (14) and (15) (Keizer 2007: 112, 116): (14) [A number of members] of staff have the same problem (15) there are [a whole series of things] that one has to consider The same collective nouns found in partitive configurations may also occur in pseudopartitives (e.g. a number of those members vs (14)), with the difference that their function in the latter case is not to denote a subset of the set referred to by the second noun but to quantify over the entities denoted by N2 (Koptjevskaja-​ Tamm 2001: 526–​ 527, 2009: 329–​ 330). Along these lines, Dodge & Wright (2002: 78) (see also Koptjevskaja-​Tamm 2009 and Kim & Lim 2011) explain that these ‘nominal’ quantifiers measure and individuate both mass nouns and plural nouns “to make mass nouns countable [and] to make one countable instance of a multiplex of entities, indicating the relationships between them”, hence their use of the term ‘measure phrases’.5 Pseudopartitive quantifying expressions are often claimed to provide classificatory information about the mass or multiplex substance referred to, serving

71

Complex collective subjects and agreement 17 as “functional equivalents of classifiers” in English (Lehrer 1986, 1987: 95; see also Zhang 2017 for a cognitive-​typological study; cf. Dodge & Wright 2002: 78–​79 on their differences).6 Koptjevskaja-​Tamm (2001: 529) qualifies this assertion by pointing out that nominal quantifiers/​measures … ‘create’ units to be counted for those entities that either do not come in ‘natural units’ (like mass nouns), or come in ‘different units’ (cf. six bunches of carrots, two rows of trees and three fronds of a palm). Classifiers, on the other hand, actualize the semantic boundaries of a given count noun by designating its ‘natural unit’[.]‌ In her study on Spanish binominal quantifying expressions, Verveckken (2015: 78) points to the lexical meaning of the N1 as the source of the unitising meaning and function of the pseudopartitive. Unlike monomorphemic quantifiers (e.g. muchas llamadas ‘many calls’), a pseudopartitive structure “allows to conceptualize a set of N2s at once”, but with the added value that the entities in N2 are “assessed metaphorically by comparing the set of N2 to the typical constellation of N1” (e.g. un alud de llamadas ‘a flood of calls’) (Verveckken 2015: 78). What seems clear is that, in semantic terms, pseudopartitive expressions quantify over the second nominal of the construction by encoding the classificatory nuance of the first nominal element of the configuration. This quantifying pseudopartitive configuration in English is commonly agreed to have its origins in the partitive construction. More specifically, it is associated with the non-​referential use of collective (e.g. group, bunch), measure (e.g. pint, pile), part (e.g. bit, part) and container (e.g. box, bottle) nouns (Akmajian & Lehrer 1976: 411; Rutkowski 2007: 345). The type and semantics of the nominal elements which quantify over mass and plural count N2s in pseudopartitive constructions are varied and it is these issues that determine their occurrence in these constructions (and/​or in partitives), as well as their morphosyntactic behaviour (see Koptjevskaja-​Tamm 2009: 331). This study focuses on three of the five types described in Keizer (2007: 112–​116): quantifier, measure and collection N1s (see also Langacker 2013, 2016): a. quantifier: lot, number;7 b. measure: majority; c. collection: batch, bunch, crowd, group, party, series. ‘Quantifier nouns’ refer to nouns which in the pseudopartitive configuration denote an imprecise and abstract quantity. Alternative classifications use the more transparent label ‘abstract quantity nouns’ (Koptjevskaja-​Tamm 2001, 2009) and distinguish them from ‘number set’ or ‘cardinal’ pseudopartitives, such as dozens/​millions of (Lehrer 1986; Alexiadou, Haegeman & Stavrou 2007: 402). ‘Measure nouns’8 include nouns that express precise or imprecise measures (e.g. metre vs majority) (cf. ‘standard-​measure’ vs ‘dimensional-​ boundary’, foot vs square, in Dodge & Wright 2002). Finally, ‘collection nouns’, as their name reveals, refer to sets of entities. Keizer (2007: 115–​116),

81

18  Complex collective subjects and agreement unlike authors such as Lehrer (1986) and Alexiadou, Haegeman & Stavrou (2007), distinguishes between ‘collection nouns’ and ‘collectives’, the latter being a more general category that includes collection nouns and other nominal elements which may behave semantically and/​or syntactically as plurals, such as university, institute and gentry. These types of N1s are characterised by their cumulative, coextensive and relational nature. The N1s listed above are cumulative in that their denotation is not affected by the addition of another set of entities with the same denotation or, to put it more simply, if a bunch of flowers is added to another bunch of flowers, the denotation of the original bunch remains the same, that is, a collection of flowers (see Alexiadou, Haegeman & Stavrou 2007: 402). Concerning coextensiveness, Langacker (2016: 22; emphasis in the original) claims that in pseudopartitive constructions, N1s share reference with N2, as they constitute “alternate ways of accessing the content for different purposes”, or as he also puts it: [i]‌n terms of actual, real-​world reference …, a group or configuration is the same as its constitutive elements. … And as for a measurement unit, we think of it as being stretched out to cover the measured mass (or conversely). Finally, these N1s are also relational because they require a complement indicating the type of entities to which they refer. Although possible, instances in which they occur without overt complementation are strange and likely to be unacceptable (16) unless the complement is retrievable from the discourse context (17) (Alexiadou, Haegeman & Stavrou 2007: 405–​406): (16) A: What did John buy? B:?John bought a bunch [vs John bought a bunch of flowers]. (17) A: Do you like flowers? B: I buy three bunches every week [i.e. three bunches of flowers]. Whether overt or elided, the second nominal may be a (non-​count) mass (18) or (count) plural noun (a bunch of flowers), but never a singular (count) noun (Alexiadou, Haegeman & Stavrou 2007: 403). Recall that pseudopartitive structures as defined here do not allow for the presence of Det29 (cf. a bunch of the fruit with partitive meaning), which, in turn, means that N2s in pseudopartitives do not have independent referential status.10 Only in combination with the previous material does the N2 establish reference (Keizer 2007: 111). (18) [A small bunch of fruit] falls towards the cloth [BNC: 1985–​ 1993 J1K 339] Structural ambiguity: The referential reading It is important at this point to highlight the potential structural ambiguity11 of the syntactic configuration of (pseudo)partitives: Det1 (Mod) N1 of (Mod)

91

Complex collective subjects and agreement 19 (Det2) N2 (see Akmajian & Lehrer 1976: 404–​408; Selkirk 1977; Alexiadou, Haegeman & Stavrou 2007: 396–​399, 423; Keizer 2007: 112–​116 and Stickney 2009: 178–​ 183). Inspired by Akmajian & Lehrer (1976), Selkirk (1977) elaborates on the structural ambiguity of both partitive and pseudopartitive constructions and examines the structural and semantic differences between their purely quantificational and referential uses.12 She claims that a bunch of flowers may refer either to ‘an amount of flowers’ (quantificational/​ pseudopartitive reading) or to ‘a collection of flowers fastened together’ (referential reading). In the latter interpretation, bunch does not ‘count’ the flowers but instead has an independent referential status, with bunch functioning as the referent and the head of the whole NP and taking the of-​PP of flowers as its dependent. The same explanation applies in the case of the partitive a bunch of the flowers, which Selkirk also treats as an ambiguous phrase, with both a partitive (i.e. subset-​set) and a referential meaning: partitive, when read as a measure phrase (a bunch (of)) and a referential NP (the flowers); referential, when analysed as a noun-​complement structure (a bunch + of the flowers) (Selkirk 1977: 302–​303). I disagree with Selkirk on this point, however, and do not share her dual interpretation and analysis of partitive configurations. The type of determiner of the first noun (Det1) is the key difference between pseudopartitives and referential NPs. A bunch of flowers may mean ‘a bouquet of flowers’ or ‘an indeterminate amount of flowers’; however, only the referential reading of the N1 is flexible as to the number and type of determiner: this/​one/​her bunch of flowers. The quantificational interpretation, in contrast, is only possible with the indefinite article (cf. this/​one/​her bunch of flowers = *‘many flowers’), particularly in cases where the sequence ‘Det1 N1 of’ is already highly grammaticalised, as in (19) and (20) (Keizer 2007: 112): ( 19) There were [a/​*the lot of idiots on the road] weren’t there (20) [A/​*the number of members of staff] have the same problem This lack of flexibility in article selection is closely related to the lexical content of the N1 when used quantificationally. When N1 “is used with minimal lexical content, the indefinite article that precedes it tends to lose its own grammatical features and so become part of a larger [functional] unit” (Alexiadou, Haegeman & Stavrou 2007: 426).13 In fact, it is generally agreed that “purely quantificational pseudopartitive constructions can only take the indefinite article”, particularly those with quantifier nouns (Keizer 2007: 135; see also Traugott & Trousdale 2013: 116). Collection nouns are found with a wider range of determiners, but are also more frequently interpreted referentially, especially when accompanied by a demonstrative (21), a possessive determiner (22) or, in particular, a definite article (23), as this “suppresses [the] quantity reading” in favour of a referential one, even with quantifier N1s (24) (Alexiadou, Haegeman & Stavrou 2007: 431; see also Brems 2011: 192): (21) [This pack of Italian wolves] is large enough [COCA: 1996 ACAD NaturalHist]

02

20  Complex collective subjects and agreement (22) [Our gang of Massachusetts rippers] was born and raised in Holden [COCA: 2007 MAG Skiing] (23) [the batch of documents] is added to the system. [BNC: 1985–​1993 H99 2133] (24) [the largest number of open cases] was 16 [BNC: 1985–​1993 B0W 542] Finally, the type of modifier of N1 is also a differentiating factor because pseudopartitives tend to be premodified by intensifiers (25) or downtoners, which “[modify] the quantificational force of the first noun”, while a referential reading is imposed when N1 is accompanied by an attributive modifier that assigns a property to N1 (or, alternatively, to the whole NP) (26) (Keizer 2007: 138–​140; see also Brems 2011: 191–​201): (25) [a whole series of uhm books that’d been in in a sort of bargain basement and were sort of one pound fifty each] (26) [a later series of rocks] were deposited The discussion in Section 2.1.2 on headedness in the pseudopartitive NP presents additional criteria to help solve the structural ambiguity between pseudopartitive and referential constructions. Quantifying meaning Binominal partitive and pseudopartitive structures are classed as nominal quantifiers (Koptjevskaja-​Tamm 2001: 527; see also Zhang 2017: 69). In addition to monomorphemic quantifiers (many, most), quantification in English may also be expressed structurally by means of both partitive and pseudopartitive NPs expressing partition and indefinite quantities, respectively. The higher degree of functionality of the pseudopartitive over simpler NPs with a monomorphemic quantifier is adduced as the main factor behind the former’s high productivity as a mean of expressing quantification in English (see Brems 2011: 9; Verveckken 2015: 68 and Chapter 3 in this book). The quantification potential of partitives and pseudopartitives is discussed in this section using Langacker’s (2013, 2016) Cognitive Grammar framework. Nouns phrases, or ‘nominals’, as they are referred to in Cognitive Grammar are characterised by the grounding function; that is to say, their “specification of the profiled referent’s epistemic status”, which, for NPs, “is primarily a matter of identification” (Langacker 2013: 65, 2016: 16, 2017a).14 Grounding can thus be broadly compared to the system of determination (Brems 2011: 47). The system of grounding elements for the nominal structure in English comprises articles, demonstratives and quantifiers, which are further subdivided into definite grounding elements, for referents already available in the discourse (i.e. demonstratives and the definite article), and indefinite grounding items, which introduce unidentified instances and comprise the indefinite article, some, sm (unstressed some)15 and, most importantly, quantifiers. This

12

Complex collective subjects and agreement 21 inventory is quite limited yet open to new additions, as the remainder of this section and Chapter 3 will show. English quantifiers can be arranged into two different groups: ‘relative quantifiers’ and ‘absolute quantifiers’ (also known as ‘set’ and ‘scalar’ quantifiers, respectively; Radden & Dirven 2007: 117–​119).16 Relative quantifiers (most, all, some, no, every, each, any) fulfil their grounding function “through a quantitative assessment made relative to a reference mass” (Langacker 1991: 83). In other words, they define the referent as a proportion of the mass (see Davidse 2004), which may be coincident with it (i.e. universal set, e.g. most books) or with just a part of it (i.e. restricted set, e.g. most of my books; Radden & Dirven 2007: 119). Although they have some quantitative meaning, they do not actually specify quantity. Instead, they “indicate degrees of universality within the maximal extension …, for example, most profiles a mass that merely approximates universality within M[aximal E]X[tension]” (i.e. the totality of the instances; Langacker 2013: 66). The set of absolute quantifiers, on the other hand, includes elements such as many, few, much, little and several, along with numerals and, most importantly, complex quantifiers. These quantifiers do specify quantity and that is why Langacker (1991: 83) considers them to be the “ ‘true’ quantifiers that offer a direct description of magnitude” of a mass (whether discrete, as in coins, or continuous, as in water) (cf. Milsark’s 1977 view on universal (i.e. relative) quantifiers as ‘true’ quantifiers).17 In NP-​initial position and in the absence of overt grounding, absolute quantifiers also assume the function of a grounding element (e.g. many apples). Both relative and absolute quantifiers are considered to be “alternate quantifying strategies”: whereas the “universality” in maximal extension of relative quantifiers acts as a kind of “epistemic status and a means of epistemic control”, the epistemic control of absolute quantifiers (when initial) is represented in the form of generalisations which are weaker than those of relative quantifiers because they lack reference to ‘the universality’ (Langacker 2016: 15; emphasis in the original). To understand how this discussion of English quantifiers relates to complex collective NPs, we must recall that Ncoll may occur in partitive or pseudopartitive configurations and that the degree of universality expressed by relative quantifiers may not necessarily involve the totality or the maximal extension. In fact, relative quantifiers often refer to a more limited reference mass, or, to quote Langacker (2016: 18), a “contextually relevant extension”, which may be implicit in the context (27) (i.e. the students who did the exam), or explicit, when stated through a referential of-​PP in a partitive structure (28) (i.e. the totality against which the referent of most is measured): ( 27) It was really a hard exam. Most students failed. (28) Most of her children Partitive structures such as (28) are highly compatible with relative quantifiers, which presuppose a reference mass (e.g. all children). Absolute quantifiers can also be used partitively, as in many of the students. While Langacker does not say so directly, the combination of an absolute quantifier and an of-​PP denoting

2

22  Complex collective subjects and agreement a reference mass seems to involve properties typical of relative quantification, the main reason being that the quantity expressed is assessed in reference to a definite reference mass (Langacker 2016: 18–​20; see also Radden & Dirven 2007: 119–​120). Along these lines, Langacker (2016: 19–​20, 25) discusses partitive structures together with the more idiomatic constructions a bunch of and a lot of in (29) and (30): ( 29) She owns [a bunch of hotels]. (30) [A lot of wine] was consumed. Although both types derive from the same construction (‘nominal + PP-​ modifier’), only in (29) and (30) does of instantiate an identity relation (cf. part-​whole relation in many of the students). Together with the lack of overt grounding of the (count or non-​count) mass N2 and the indefinite Det1, this explains the use of the pseudopartitive configuration as an open-​ended diachronic source for absolute periphrastic quantification (see Langacker 2013: 69–​73, 2016: 24–​29). The proliferation and high productivity of these periphrastic quantifiers seems to be motivated by language users’ need for new and expressive means of denoting quantification (see Brems 2011: 9 and Verveckken 2015: 68). Languages are continuously evolving and, as a result, productive and recurrent constructions can end up losing their full expressive power, leading speakers to devise “alternate ways of saying approximately the same thing” through a process known as ‘renewal’ (Hopper & Traugott 2003: 122; emphasis added). The quantificational system is remarkably sensitive to fashion and renewal, and, in this respect, expressions such as a bunch of N and a heap of N have emerged as a convenient configuration to express quantity (often hyperbolically) on the basis of their original quantitative, measure or size implications (Brems 2011: 9; see also Traugott 2008b: 33). Periphrastic quantifiers cannot be said to have acquired all of the canonical features and properties of those monomorphemic quantifiers, however, owing to the persistence of conceptual and syntactic characteristics of the original lexical meaning in the more conventional quantificational use. For instance, in a bunch of there is an underlying collective meaning (i.e. conceptual persistence), as well as an acceptance of premodification (frequently through intensifying adjectives such as whole; i.e. syntactic persistence), which, far from detracting from its grammatical status and meaning, endows the construction with a higher functionality compared to that of many and much. This approximate equivalence with existing absolute quantifiers enriches the quantifier paradigm with a new element that quantifies entities metaphorically by conceptualising them unitarily or collectively (Brems 2011: 112; Verveckken 2015: 78). The fact that the absolute quantifying meaning of complex collective NPs like a bunch of emerges in the pseudopartitive configuration A (Mod) Ncoll of (Mod) N218 is one of the cornerstones of the historical study presented in Chapter 3, where I argue that only constructions with an indefinite Det1 and a non-​overtly grounded N2 (i.e. without Det2) have the potential to denote absolute quantification and become idiomatic periphrastic quantifiers, and examine

32

Complex collective subjects and agreement 23 the potential for relative quantification in partitive complex collective NPs. The fuzzy boundaries between the quantificational and non-​quantificational uses of complex collective NPs are explored in more detail in the next subsection. 2.1.2.  Morphosyntactic characterisation: Headedness and constituency Determining the head and, in general, the structure of binominal NPs is not an easy task, as both nouns are suitable candidates for headedness. In pseudopartitive and partitive constructions, the structure of the phrase may be simultaneously affected by syntactic, semantic and even discourse factors, which makes it difficult to identify the head. Headedness itself is a complex notion that Keizer (2007: 21, 2019) claims should be viewed as “a matter of degree” and, as such, addressed from a “cluster approach [where the element] fulfilling most criteria wins out”. This section examines the research on headedness in partitive and pseudopartitive phrases, as well as the two opposing views on the constituency of highly grammatical (pseudopartitive) quantifying uses. Headedness Headedness in binominal partitives has received virtually no attention in the literature. One of the few studies is Selkirk (1977: 311), where a bunch of the flowers is claimed to be structurally ambiguous between a purely partitive and a referential reading, as commented above. Selkirk translates this difference in meaning into a different structure with a different head in each case, with the partitives taken as right-​headed (31) and referential phrases as left-​headed  (32): ( 31) [[A bunch of] [those flowershead]] were thrown out on the back lawn. (32) [[A bunchhead] [of those flowers]] was thrown out on the back lawn. Her argument is based on number agreement, pronominalisation and verb selection restrictions (Selkirk 1977: 311). To the best of my knowledge, Selkirk is the only author to analyse the construction in this way. The few references in the literature to binominal partitives argue for a left-​headed analysis, suggesting that they are, by definition, noun-​complement referential constructions, even though semantically they also help to quantify the entities denoted by N2 (i.e. through the part-​whole or subset-​set relation; Jackendoff 1968, 1977; Keizer 2007: 70; Rutkowski 2007). Stickney (2009: 39–​40) categorically rejects this dual interpretation on the basis that the of-​PP is a complement of N1 since: (i) the whole NP, and not N1 alone, may be substituted by the pronoun one (33); (ii) nothing can intervene between the head and its complement (34) and (iii) the of-​PP may undergo extraposition (35):19 (33) *[a plate of the cookies] and one of the cake (Stickney 2009: 40) (34) *[a gallon] from Sicily [of the wine] (Jackendoff 1977: 107) (35) [A lot] has been eaten [of the leftover turkey]. (Jackendoff 1977: 108)

42

24  Complex collective subjects and agreement Interestingly, more recent studies reject this left-​headed analysis: drawing on Martí i Girbau’s (2010) quantifier phrase analysis, Leclercq & Depraetere (2018) support a right-​headed analysis, in which partitives are conceived of as a combination of a quantifier phrase that quantifies over the set denoted by the head N2. While not rejecting the quantifying potential of partitives, to which I attribute a relative quantification function, I share Stickney’s (2009) view of partitives as left-​headed noun-​complement structures. According to Keizer (2007: 149–​151), pseudopartitives, unlike partitives, are best analysed as right-​headed constructions (see also Akmajian & Lehrer 1976; Jackendoff 1977; Selkirk 1977; Alexiadou, Haegeman & Stavrou 2007: 409–​436; Rutkowski 2007; Stickney 2009). The initial nominal, N1, is thus regarded as part of a complex expression that quantifies over the second noun.20 Keizer (2007: 116–​149) also distinguishes pseudopartitives from fully referential (left-​headed) NPs (§2.1.1), on the basis of a series of semantic, syntactic and pragmatic criteria. Of the semantic criteria discussed in Keizer’s study, only the criterion of obligatoriness was found to be reliable in the case of quantifier, measure and collection N1s (Keizer 2007: 117–​120). Her analysis shows, in fact, that even though N2 may be elided (though recoverable) in certain contexts, only N2 can replace the whole NP (36) since the limited semantic content of N1 makes it unable to function independently (Keizer 2007: 118): (36) she’s got a little circle *(of friends that she absolutely clings to) /​(a little circle of) friends that she absolutely clings to The selection restrictions of the verb are revealing: in a referential reading, the verb selects N1, while in a pseudopartitive, N1 is “transparent for selection” so the verb semantically selects N2 (Alexiadou, Haegeman & Stavrou 2007: 398; see also Akmajian & Lehrer 1976: 406; Selkirk 1977: 310). In (37), therefore, N2 “complies with the selection restrictions of the verb (numbers do not have memories, people do)”, whereas in (38) the verb semantically selects number (Keizer 2007: 119):21 ( 37) [a number of people] uh have memories of the Second World War (38) [The true number of vacant posts] was one thousand four hundred With collection nouns, however, verb selection restrictions are an unreliable criterion of headedness, since the coextensiveness of collective nouns with their N2s makes both of them semantically compatible with the verb (Keizer 2007: 120). In relation to morphosyntactic criteria, subject-​verb agreement, the most widely used criterion of headedness (and the most criticised; see Akmajian & Lehrer 1976: 410 and Lehrer 1986: 140–​141), was found to be only partially reliable, as it can only be consistently controlled for when (i)  the pseudopartitive is functioning as subject, (ii) N1 and N2 differ in number, (iii) the verb is inflected for number and (iv) N1 is a quantifier noun (39) (Keizer 2007: 120–​125):

52

Complex collective subjects and agreement 25 (39) [A number of filter samples] were collected on the 2  days of the experiment (40) [The number of work permits] … was eight hundred and sixty-​eight With quantifier nouns (39), the explanation is straightforward: N2 heads the NP, unless they are used referentially (40), in which case N1 is the head. With measure and collection N1s, however, the test is unreliable, since plural agreement could also be determined by a collective or distributive reading of N1, which leaves the issue of headedness unclear (Keizer 2007: 124–​125): (41) [the vast majority of forces in this country] carry out their responsibilities (42) [a series of biochemical reactions] take place Determiner-​head agreement and morphosyntactic locus were found to be unreliable predictors of headedness on the whole: the former, because the determiner invariably agrees in number with N1, even with quantifier N1s (e.g. A *(number of) members have the same problem) (Keizer 2007: 131–​132); and the latter, because in the collective NPs considered in this research, both N1 and most types of N2s may bear inflectional marks. The only exception in this regard are Keizer’s (2007: 126) collection nouns, which may occur in the singular or the plural but take only plural N2s. As regards the criterion of extraposition, Keizer (2007: 126–​ 131; cf. Akmajian & Lehrer 1976: 407; Lehrer 1986: 140) confirms Akmajian & Lehrer’s (1976) and Selkirk’s (1977) observations on the ungrammaticality of extraposing the of-​PP in pseudopartitive phrases (43)–​(44), which constitutes another distinctive trait that sets them apart from partitive (45) (Selkirk 1977: 304) and fully referential (46) structures.22 Though this criterion must be taken with caution since the extraposability of a constituent may be affected by numerous factors, the impossibility of extraposing the of-​PP with quantifier (43), measure and collection (44) pseudopartitives corroborates that N2 is the head if they receive a quantifier interpretation (cf. (46)).23 (43) (44) (45) (46)

*[A large number] have the same problem [of members of staff] *[a series] take place [of biochemical reactions] He gave [a rather large number] to Mary [of his books by famous authors] [A series] occurs [of changes which are histologically described as Wallerian degeneration].

Finally, in relation to pragmatic criteria, Keizer (2007: 132–​135) explores discourse reference and pronominalisation, finding that anaphoric pronouns refer to N2 in pseudopartitives (47) and N1 in referential structures (48) (Keizer 2019: 350): ( 47) I invited [a number of students] to the party. They all accepted. (48) I don’t know [the exact number of students] in our department, but it must be very large.

62

26  Complex collective subjects and agreement The criteria surveyed in this subsection suggest that pseudopartitive phrases are headed by N2 when interpreted with a purely quantifying meaning and by N1 when interpreted referentially. These parameters are not unequivocal or conclusive, however (see Keizer 2007: 149–​151, 2019: 350). The following discussion on the question of constituency will attempt to shed more light on the behaviour of these constructions. Constituency The debate surrounding the headedness of pseudopartitive and partitive phrases is inextricably linked to the lack of consensus regarding the role of of and the constituency of the whole phrase. While in partitives of is viewed by some as the head of the prepositional phrase (Huddleston & Pullum 2002: 412; Stickney 2009: 44), others see it as a transformationally inserted grammatical formative (Akmajian & Lehrer 1976; Jackendoff 1977; Selkirk 1977; cf. Jackendoff 1968). In the case of pseudopartitives, in contrast to the functional role identified by some studies (e.g. Jackendoff 1977; Selkirk 1977; Löbel 1989; Corver 1998; Keizer 2007; Rutkowski 2007; Stickney 2009), others focus on the lexical status of of as the head of a PP that complements N1 (Huddleston & Pullum 2002: 351–​352). In line with this lexical reading, Langacker (2016: 25, see also 2010: 41–​43) points to the elision of of when N2 is omitted and the impossibility of introducing material between of and N2 to propose a compositional analysis of even highly grammatical pseudopartitives, as in (49): (49) [[a lot]nml [of wine]pp]nml24 A: How much wine was consumed? B: A lot (*of). *She bought a lot of (10 gallons, to be precise) cheap red wine. Langacker (2010: 42; emphasis added) contends that the boundary should “[fall] between a lot and the of-​phrase” as [i]‌ts components are still recognizable and make definite contributions to the meaning of the whole [and] it retains a bit of flexibility and participates in patterns indicating that a lot is itself a nominal consisting of article and noun, and that of heads a modifying prepositional phrase: a lot; a whole lot; an awful lot; lots; something of which I have a lot.25 Based on this argument, only entrenchment of the coalesced form alotta may be said to involve syntactic reanalysis of ‘a lot of’ into [alotta]quant [N2]. Otherwise, Langacker (2016: 23) rejects the existence of two different structures, referring instead to “functional reorganization”, which involves “the superimposition of an additional layer of semantic functions” on top of the semantic functions of a grammatical structure (termed ‘recategorization’ by Langacker 2017b: 389–​ 399). The dual lexical and quantifier interpretation of complex collective NPs

72

Complex collective subjects and agreement 27 such as a group of and a lot of is thus argued to derive from the coextension of the collective and its constitutive entities and the size implications of the former, which eventually lead to the addition of a functional quantifying layer (see Section 2.2.3 for the effect of this on verb agreement). Constituency-​based approaches, conversely, propose “competing analyses in the form of alternate constituency hierarchies” (Langacker 2016: 21–​22). Based on the view of language as a network of form-​meaning pairings called ‘constructions’, Traugott & Trousdale’s (2013: 22–​ 26) constructionalist approach offers two separate analyses: the original head-​modifier construction and the quantifier construction. The grammaticalisation (see Hopper & Traugott 2003) of the quantifier meaning and function of a lot of or, to use Traugott & Trousdale’s (2013) term, the ‘constructionalisation’ of the new quantifying form-​meaning pairing, has its origins in the metonymic extension of the construction’s initially collocationally restricted use (i.e. with concrete inanimate nouns only), which denotes “an object (usually made of wood) by which persons were selected” (Traugott 2008a: 230). The subsequent expansion of the expression’s collocational range to ‘groups of people or things’ and its frequent co-​occurrence in its extended use with plural of-​PPs triggered a “pragmatic ‘invited inference’ of quantity” and a certain ambiguity between its referential and potentially quantifying readings (50) (Traugott & Trousdale 2013: 26). The plausibility of the quantity inference, the frequency of the structure with plural of-​PPs and a later extension in collocational range to abstract nouns such as power (in (51)) have prompted the progressive bleaching or generalisation of the lexical meaning of lot (termed ‘desemanticization’ by Brinton 1996: 54), as well as the “loss of morphosyntactic features indicative of its lexical class” (referred to as ‘decategori(ali)zation’ in Bybee 2015: 129; see also Brinton & Traugott 2005: 107 and Traugott & Trousdale 2013: 116): (50) I bought a lot of clothes of the shopman (Brems 2011: 212) (51) He is only young, with a lot of power. (Traugott & Trousdale 2013: 25) As Traugott & Trousdale’s analysis shows, lot has undergone a decrease in syntagmatic variability, collocating only with the indefinite article in this new quantifying function and losing its role as head of the NP. However, it has also acquired both the status of a quantifying expression, or ‘degree modifier’ (Traugott 2008a: 230–​232), and a grammatical meaning that denotes ‘large quantities’ (Traugott & Trousdale 2013: 25–​26). There has been no loss of meaning, therefore, but a “redistribution” via pragmatic strengthening and recategorialisation, its meaning being thus “metaphorically transferred to [the quantificational] domain” while the structure “acquires properties typically associated with quantifiers” (Verveckken 2015: 63, emphasis in the original; see also Brems 2011: 111–​112 and Heine, Claudi & Hünnemeyer 1991: 233–​ 238 on recategorialisation). The syntactic and semantic reanalysis (referred to as ‘neoanalysis’ in Traugott & Trousdale 2013: 25) of the original referential construction into a quantifying structure is illustrated in (52):

82

28  Complex collective subjects and agreement (52) a lot Head

of land (for sale) Modifier [N [of N]]

>

a lot of land/​love Modifier Head [[N of] N]

The possible coalescence of a lot of and a bunch of in their new quantifying sense into the reduced forms alotta (also a lotta and alott of) and buncha, together with their lack of control of number agreement (e.g. there were [a lot of fine beasts]), has been adduced as evidence of the ‘constructionalisation’ of their quantifying meaning and function (Traugott & Trousdale 2013: 211).26 The coexistence in Present-​ Day English of the new construction with the (infrequent) original lexical use of lot results in ‘layering’ (Hopper & Traugott 2003: 124; Traugott 2008a: 234), a phenomenon also found in analogous constructions, such as a bunch of, a load of and a pile of (Brems 2011).27 Unlike lot, however, which can also be used as a free adverbial, as in they had to excavate quite a lot (termed ‘free degree adjunct’ in Traugott 2008a: 232), not all of these constructions have followed the same process of constructionalisation or continued to develop new grammatical meanings. In fact, given their lexical content and the functionality which that content confers, it seems reasonable to assume that some may never reach the same level of grammaticalisation (see Brems 2011: 232). Chapter 3 elaborates on this question by analysing the emergence of the quantifying meaning of different complex collective noun phrases.

2.2.  Verb number agreement with complex collective subjects ‘Agreement’ refers in a broad sense to the relationship between two linguistic items by means of the instantiation of the same grammatical or semantic feature (e.g. number). Despite the very few remnants of the rich morphological system of Old English (e.g. singular-​plural distinction in the noun, third person singular in verbs), agreement variation is far from infrequent in Present-​Day English, particularly with (complex) collective NPs. This is reflected in grammar books and usage guides dating back to the eighteenth century, where this variation in number agreement has been the object of much debate and criticism (see Sundby, Bjørge & Haugland 1991: 140, 157). Some of the so-​called ‘prescriptive grammars’ (or ‘grammars of errors’) from the eighteenth and nineteenth centuries called for “purity, precision and perspicuity”, condemning as ‘harsh’, ‘improper’ and ‘ungrammatical’ examples of plural verb agreement with complex collective NPs, as in (53), and corrected examples such as (54) (Sundby, Bjørge & Haugland 1991: 1, 140; see also Dekeyser 1975: 35–​42, 1996 and Bjørge 1989): (53) [A flock of sheep] were driven to market. (54) [A swarm of bees] make a pleasing spectacle. (‘makes’) In modern (twentieth-​and twenty-​ first-​ century) usage guides,28 agreement dilemmas with collective nouns are sidestepped with the recommendation to

92

Complex collective subjects and agreement 29 include the phrase ‘the members of’ in cases such as the jury is considering or the jury are considering (Taggart 2010: 37). In relation to complex collective NPs, approaches range from compiling frequent usages and clues to solve the dilemma (e.g. the favouring the singular and of-​PP the plural (55); O’Conner 1996: 25–​26) and insisting on the importance of consistency with a given (notional or grammatical) number across all its agreeing forms (56) (Stilman 1997: 215–​216), to accepting variability in verb agreement (57) (Evans & Evans 1957: 22) and advising the reader about elegance, suitability and preferred use (58) (Pickett, Kleinedler & Spitz 2005: 288): (55) The majority is in charge. Still, [a majority of voters] are unhappy. (56) *[The group of ticketholders] was furiously demanding refunds of their money. (57) There is [a pile of books] on the table /​[A pile of books] were on the table (58) [The majority of the voters] live (*lives) in the city In the early twentieth century, three milestone comprehensive historical grammars stand out, following the new philological tradition to the study of grammar: Jespersen’s A modern English grammar on historical principles (1909–​1949), Poutsma’s A grammar of Late Modern English (1914) and Kruisinga’s Handbook of Present-​Day English (1932). Precursors of the three canonical reference grammars discussed below, these seminal works of descriptivism already noted the relationship between type and conceptualisation of the collective noun and variation in verb agreement. 2.2.1.  Comprehensive grammars The three canonical works examined in this section are Quirk et al. (1985), Biber et al. (1999) and Huddleston & Pullum (2002). Each of these grammars provides a comprehensive view of the phenomenon referred to as ‘agreement’ in Huddleston & Pullum (2002) and ‘concord’ in both Quirk et  al. (1985) and Biber et  al. (1999). The main difference between their accounts lies in the scope of their descriptions: Huddleston & Pullum (2002) refer to subject-​ verb agreement in number and person but also discuss gender and pronominal agreement, among other issues; Biber et  al. (1999) tackle the topic from a narrower point of view, briefly discussing pronominal agreement but focusing on the prototypical relationship of subject-​verb agreement; finally, Quirk et al. (1985: 755) contemplate concord from the more general perspective of a relationship between any two linguistic items “such that one of them displays a particular feature (eg plurality) that accords with a displayed (or semantically implicit) feature in the other”. ‘Simple agreement’ (Huddleston & Pullum 2002: 499) or ‘grammatical concord’ (Quirk et  al. 1985: 757) requires that the grammatical number of the subject determine the number of the verb, being overtly marked only in the third person singular of lexical verbs in the present tense and the past of the verb to be. This rule is strictly followed only in formal contexts, however,

03

30  Complex collective subjects and agreement thanks to “the sanction of teaching and editorial tradition” (Quirk et al. 1985: 766). Uses that override grammatical agreement on, for example, semantic grounds are thus relegated to colloquial English. These deviations from grammatical agreement are discussed in all three grammars as ‘complications’ (Biber et al. 1999: 180–​187), ‘departures’ (Huddleston & Pullum 2002: 500–​501) or ‘difficulties over concord’ (Quirk et  al. 1985: 757–​759), where they address the variability in verb number agreement of (complex) collective NPs that forms the core of the discussion in the remainder of this chapter. These non-​ prototypical cases of agreement may be broadly classified according to the aspect of the subject NP that prompts the deviation: its form and/​or meaning (i.e. notional concord) and the distance between the subject and the verb (i.e. proximity concord). Notional concord As already mentioned, English often relies on the semantic or notional number of the subject (instead of its grammatical number) to assign number to the verb, a phenomenon commonly known as ‘notional concord’ (Quirk et  al. 1985: 757; Biber et al. 1999: 187) or ‘semantically motivated override’ (Huddleston & Pullum 2002: 501–​504). The prototypical example of notional concord in English involves collective nouns, whose semantic plurality allows them to take plural verbal forms. In this regard, these three grammars concur that notional agreement with these nouns reflects a difference in conceptualisation: plural agreement places the focus on the members of the collective (59), whereas with singular agreement, the collective is taken as a unit (60) (Biber et  al. 1999: 188–​189): (59) The Catholic flock –​who constitute one third of Malawi’s population –​ are tired of dividing their loyalties. (NE WS ) (60) It is alleged that the flock is infected with Salmonella Typhimurium. (N EW S ) Biber et al. (1999: 188) observe that the most common collective nouns prefer singular agreement, though also identify collective nouns that favour the plural pattern, which, as Huddleston & Pullum (2002: 502) contend, “is unquestionably fully grammatical in Standard English, and … generally recognised by the usage manuals”. Quirk et al. (1985: 758) remark, however, that plural agreement is preferred mainly in speech (see also Huddleston & Pullum 2002: 502), with singular concord preferred in written texts and in cases of doubt, particularly in British English. Regional variation is another consideration here. The distribution of singular and plural patterns of agreement in this context is erratic across the different varieties of English. As far as the main varieties are concerned, Biber et  al. (1999: 19, 188–​189) note that singular concord is the most common pattern, although they acknowledge that almost any collective in the British variety may collocate with plural verbal forms, while American English

13

Complex collective subjects and agreement 31 “generally treats singular collective nouns as singular” (Quirk et al. 1985: 758; see also Huddleston & Pullum 2002: 502). This variation is claimed to be further constrained by the meaning of the verb. In particular, Biber et al. (1999: 189) point out that: [f]‌or plural concord to be available, the meaning of the verb must clearly be applicable to individual members of the group. Thus singular concord only is found in cases such as: The committee comprises/​consists of/​has eight members. Nevertheless, counterexamples and even variation within the same context are far from uncommon (Biber et al. 1999: 189). In fact, the patterns of agreement of complex collective noun phrases, particularly those that denote quantity, are argued to show considerable variability. For instance, the structures a number/​ bunch/​group/​series of are found to take plural agreement when followed by a plural noun (61), but singular agreement when the collective is conceived of as a homogeneous unit (62) (Biber et al. 1999: 185): (61) Under a nearby tree, [a group of children] were having their school lunch. (N EW S ) (62) [The second group of books] is those written by botanists. (ACA D ) Huddleston & Pullum (2002: 349, 501–​ 504) explain that nouns such as number, bunch, group and majority are ‘number-​transparent nouns’: nouns bleached of their original meaning which thus allow “the number of the oblique to percolate up to determine the number of the whole NP”, as in (61). It is the explicit (though sometimes elided) and often plural of-​complement taken by these nouns that triggers an obligatory and semantically motivated override, either singular (63) or plural (64). While they admit the existence of borderline number-​transparent nouns that may take singular agreement with plural of-​complements (65), Huddleston & Pullum (2002: 501–​504) restrict verb number variation mainly to ‘prototypical’ collective nouns (i.e. those not requiring an overt of-​complement), where variation entails a difference in conceptualisation (66) (cf. (63) and (64)): (63) [Heaps of money] has/​*have been spent. (64) [A number of spots] have/​*has appeared. (65) %The fact [the overwhelming majority of Americans] doesn’t want the President impeached does not necessarily mean that that would be the right decision. (66) The committee has/​have not yet come to a decision. Proximity concord Notional agreement is closely related to the principle of proximity (also known as ‘attraction’), which “denotes agreement of the verb with a closely preceding

23

32  Complex collective subjects and agreement noun phrase in preference to agreement with the head of the noun phrase that functions as subject”, as illustrated in (67) (Quirk et al. 1985: 757; see also Biber et al. 1999: 189–​190). The interaction between proximity concord and notional agreement is what makes a sentence like (67) acceptable (i.e. ‘only his own supporters agree with him’), albeit only in informal, colloquial contexts (Quirk et al. 1985: 757; see also Biber et al. 1999: 189–​190; cf. Huddleston & Pullum 2002: 500–​501 on processing errors): (67) ?No one except his own supporters agree with him. This interaction is further complicated by increasing distance (long postmodifiers, intervening material) between the subject and the verb, as distance is found to reinforce the use of plural verbal patterns (Quirk et al. 1985: 757). This is the case of complex collective subjects such as (68). Quirk et al. (1985: 764–​765) argue that verb number is determined by notional agreement in these cases, but observe that type of construction is also a factor: complex quantifier expressions such as (68) and (69) favour the use of plural agreement, owing to their notional equivalence to many and most (see also Biber et  al. 1999: 190): (68) [A (large) number of people] have applied for the job. (69) The majority are Moslems. 2.2.2.  Syntactically oriented approaches This section takes a detailed look at Corbett’s canonical model of agreement, a theory-​ independent approach based primarily on syntactic (i.e. formal) agreement. Numerous linguists have also analysed from this perspective the role of agreement in the configuration of the sentence and in the relationship between the verb and its subject. Generative Grammar, whose mechanisms and approaches are explored in Section 2.2.2.2, is another major proponent of this view. 2.2.2.1.  Corbett’s canonical model Corbett’s canonical approach to agreement defines, demarcates and analyses the phenomenon and its components independently of any pre-​established theoretical points of view (see Corbett 2003: 162–​ 164). Instead, Corbett (2006: 2–​4) conceives of agreement as an operation which affects different linguistic areas, from syntax and semantics to morphology, lexicology and even pragmatics. According to his model, agreement cannot be taken as a purely syntactic or semantic phenomenon: while semantics may determine the type of agreement relation in certain cases (e.g. notional agreement with collective nouns), ultimately syntax determines the syntactic domains in which semantic agreement is possible, with the morphology and lexicon of each language limiting the possible overt manifestations of agreement features (Corbett

3

Complex collective subjects and agreement 33 2006: 2–​3).29 It is for these reasons that Corbett (2006: 8–​9) opts for a more ‘neutral’ position, taking as his starting point the prototypical examples of formal (or syntactic) agreement “where there is an overt controller and an overt target where the features of both match exactly”, “work[ing] then ‘outwards’ to include the less canonical cases” (Corbett 2003: 162, 192). This ‘canon’, however idealistic or unrealistic, constitutes a fixed point “from which occurring phenomena can be calibrated”, thus enabling the accommodation of the different patterns of agreement on a gradient scale of canonicity while also accounting for the cross-​linguistic variation of the phenomenon (Corbett 2006: 9). The basic tenet of Corbett’s (1994: 55) model is the idea that agreement requires the “covariance or matching of features between two separate elements”, or, in other words, a systematic sharing of features: “as one element varies so will the other” (Corbett 2006: 4). Regardless of their nature and the variability imposed by the features, two components or elements are always present in an agreement relation: the controller and the target (Corbett 2006: 4–​5). The ‘controller’ (also referred to as the ‘trigger’ or ‘source’) is the item which determines the agreement relation. Controllers are usually nominal, i.e. nouns or (complex) noun phrases. Canonical controllers control “a consistent agreement pattern” (e.g. apple in (70)), unlike ‘hybrid’ controllers (including collective nouns), which may control different feature values on different targets, as shown in (71) (Corbett 2006: 11–​12, 157–​165, 207, 2015): (70) thissg applesg tastessg good (71) Thissg/​*Thesepl committee, which hassg/​who havepl decided …. Itsg/​ Theypl …. The ‘target’ is the element whose form is determined by the feature specifications of the controller. The category of canonical targets prototypically includes verbs and adjectives, the former being the most frequent within the clausal domain (Corbett 2006: 5, 40–​53). Canonical targets show obligatory agreement marking, which may be expressed through concatenative inflectional morphology (e.g. bound affixes like –​s in (70) above), with suppletive and pronominal forms (e.g. It/​They in (71)) thus considered less canonical (Corbett 2006: 15). The other elements involved in an agreement relationship are the domain, the features and the conditions (Corbett 2006: 4–​5). The domain refers to “[t]‌he syntactic environment in which agreement occurs”, which in canonical agreement is local and asymmetric. This asymmetry is apparent in the terms ‘controller’ and ‘target’ and, even more clearly, in the fact that (i)  the target adapts its morphology to the features of the controller (not the other way round) and (ii) the controller contributes to the semantic interpretation of the construction (as in the notional agreement found in the example The committee have decided; Corbett 2006: 19–​23, 155).30 As regards the locality of the domain, “the smaller the structural distance between controller and target the more canonical is the instance of agreement” (Corbett 2006: 21).

43

34  Complex collective subjects and agreement I will return to the question of domains later in this section. The features are the elements directly reflected in the relationship which indicate by means of different values “in what respect there is agreement” (e.g. ‘number’). The most canonical features are lexical (i.e. available in the lexicon) and assigned formally (i.e. the marking on the target depends exclusively on the marking on the controller; Corbett 2006: 23–​24). Finally, the conditions refer to “factors (like word order) which have an effect on agreement but are not directly reflected” (Corbett 2006: 4). In canonical agreement, no conditions are specified (Corbett 2006: 26). The interaction of all of these components and the resulting agreement relation is represented in (72) (Corbett 2006: 5): (72)

domain controller the system

target works

condition

feature: number value: singular

Canonical and non-​canonical agreement The criteria of Corbett’s theory may be summarised in three general principles which “converge on the notion of ‘canonical agreement’ ” (Corbett 2006: 26–​27): Principle I: Canonical agreement is redundant rather than informative Principle II: Canonical agreement is syntactically simple Principle III: The closer the expression of agreement is to canonical (i.e. affixal) inflectional morphology, the more canonical it is as agreement Canonical agreement is therefore found where a perfect formal matching exists between the controller and the target with respect to a particular feature and value in a given domain (Principle I), agreement is as simple as possible (Principle II) and inflectional morphology is used (Principle III). Canonical agreement is by no means common among the world’s languages, however. One of the issues contributing to this low frequency is precisely the redundancy indicated in Principle I. The formal instantiation of the same feature in both target and controller leads to redundancy of information, which is what makes this phenomenon worthy of attention: after all, if matching were not formal and relied on semantics only, it would not be necessary to account for it as a distinct phenomenon (i.e. the matching would be always predicted from the semantic specifications of the controller and the target; Corbett 2006: 27). Finally, redundancy is an important resource for speakers to facilitate comprehension of their message by “allowing the hearer to keep track of the different referents in a discourse” and marking constituency (Corbett 2006: 275; see also Levin 2001: 26).

53

Complex collective subjects and agreement 35 Canonical agreement concerns instances which show a formal matching between the features of the target and those of the controller (also referred to as ‘syntactic/​formal/​grammatical agreement’ or ‘agreement ad formam’), as in (73) (Corbett 2006: 155). (73) [a couple of researches] has displayed that lactose intolerance could be improved [COCA: 2008 ACAD CollegeStud] Agreement is therefore less canonical when there is consistency between verb number and the meaning of the controller (i.e. ‘notional agreement’; also referred to as ‘semantic/​logical agreement’, ‘agreement ad sensum’ or ‘synesis’; Corbett 2006: 155).31 This deviation from the canon fulfils an important function of agreement by providing a different interpretation of the same entity or event (Corbett 2006: 275). This is the case of collective nouns and complex collective NPs, as in (74), which Corbett (2006: 222) acknowledges also “give rise to agreement choices”: (74) [A couple of works] are particularly informed by recent work in semiotics. [COCA: 1992 ACAD AmerStudies] Variability in agreement patterns in these and other cases is, however, subject to the so-​called ‘Agreement Hierarchy’. The ‘Agreement Hierarchy’ Corbett (1979, 2006: 207) observes that the availability and acceptability of alternative agreement patterns are constrained by the locality of the domain, as reflected by the position of the target on the ‘Agreement Hierarchy’ (75): (75) attributive > predicate > relative pronoun > personal pronoun Corbett’s hierarchy (1979: 203) ranks the different agreement targets in terms of their distance from the controller to predict the “possibility and relative frequency of semantic as compared to syntactic agreement”. The four syntactic domains it comprises differ in locality: while attributive targets are restricted to the NP, the scope of pronouns extends beyond the clause. The layout of the hierarchy reflects the permeability of each domain to semantic agreement (i.e. to the possibility of deviating semantically from canonical (syntactic) agreement): moving rightwards along the hierarchy increases monotonically the distance between the controller and the target and, with it, the likelihood of semantic overrides of agreement (Corbett 1979: 204). If semantic agreement is possible in one position, therefore, it will also be available in all of the other positions to its right (see examples (76)–​(78); Corbett 1979: 204, 2006: 207): (76) This/​*These committee sat late. (77) The committee has/​have decided.

63

36  Complex collective subjects and agreement (78) The committee, which has/​who have decided … It/​They … The claim that syntactic distance increases the likelihood of semantic agreement (Corbett 1979) with collective nouns has been examined and corroborated by studies like Levin (2001) (see Section 2.2.4) and is also tested as part of the analysis reported in Chapter 4. 2.2.2.2.  Generative Grammar Within the Generative framework, ‘core agreement’ is defined as the “morphological marking on the central verb of a clause that reflects some features of a noun phrase in that clause” (Baker 2013: 607). An agreement relation thus comprises a verbal element containing agreement features, such as number or person (e.g. shows [+singular] [3rd person] in (79)), which match similar features in a nominal element within the same sentence (e.g. set [+singular]): (79) [this second set of photos] shows the casting in an intermediate stage. [COCA: 1995 MAG Inc.] While extensions of core agreement may include agreement between the verb and its (in)direct object(s) (Baker 2013: 607), agreement is generally restricted to the verbal domain. For this reason, some generativists use ‘concord’ rather than ‘agreement’ to refer to the agreement relation between a determiner (or a modifier) and an NP, as in this and set in (79) above. From ‘Inflection’ to ‘Agree’ The agreement operation did not become prominent in Generative Grammar until the publication of Chomsky’s (1981) Lectures on government and binding. According to this framework, the relation between the subject and the verb is mediated by a functional ‘Inflectional Phrase’ (IP) which is headed by the functional category ‘Inflection’ (also ‘Infl’ and ‘I’) and contains the syntactic features necessary for a sentence to be grammatical. Later reformulations of Chomsky’s proposal included the subdivision of the IP into ‘AgrSP’ (Subject Agreement Phrase), ‘TP’ (Tense Phrase) and ‘AgrOP’ (Object Agreement Phrase), whose functional heads host the agreement features of the subject (AgrS), finiteness/​ case (T) and the agreement features of the object (if any) (AgrO), respectively (see Pollock 1989; Belletti 2001: 484–​488 and Boeckx 2006: 2). Within this framework, subjects are claimed to be generated in the specifier (Spec) position of the verb phrase (VP), since lexical verbs can only assign ‘thematic roles’ (i.e. agent, patient) to their arguments when the argument is their sister in the projection (i.e. either at the V’ or VP levels;32 see Haegeman 2006: 191–​195 and Cook & Newson 2007: 82–​85). Until the Minimalist Program (Chomsky 1995), it was agreed that lexical elements entered derivation as “a bare root”: subjects moved to Spec of TP to acquire case33 and then to Spec of AgrSP to

73

Complex collective subjects and agreement 37 acquire number and person (i.e. ‘φ/​phi-​features’; Epstein et al. 2013: 500; see also Boeckx 2006: 5 and Baker 2013: 611–​612).34 The advent of the Minimalist Program brought about a major simplification of the “theoretical and descriptive apparatus used to describe language”, including the agreement operation (Radford 2004: 9).35 Previously required operations, such as feature assignment and movement, were replaced by ‘feature-​checking’ or ‘feature valuation’ of the uninterpretable features of the verb (i.e. person and number) against those of the subject (i.e. person and number in (pro)nominal elements being interpretable; Radford 2004: 284–​ 291, 483; Hornstein, Nunes & Grohmann 2005: 317). This valuation of (now inherent) features of the new ‘Agree’ operation,36 where agreement is a precondition for the movement of the subject, thus ensures that only interpretable features (i.e. those that contribute to the semantic interpretation in the semantic component of language) eventually enter the derivation, with uninterpretable ones being eliminated via feature-​checking/​valuation (Baker 2013: 612–​614; Bošković 2013: 118).37 Accommodating semantic agreement The mechanisms of the Generative framework outlined above are strictly formal and, therefore, perfectly accommodate formal relations of agreement such as those found in (80). Nonetheless, in English (particularly British English),38 semantic agreement (81) is common (Smith 2017: 824): (80) The committee is not making decisions right now. (81) This/​*These committee are deciding on a solution. The question is how to apply Agree and feature valuation operations to examples such as (81), where the number feature of the verbal target differs from that of the subject (which should have valued the feature number of the verb), bearing in mind that a “mismatch of features cancels the derivation” (Chomsky 1995: 371). How, moreover, can this framework account for the asymmetry between the number of the subject and the verb, and the ungrammaticality of a similar asymmetry within the nominal domain? Attempts to resolve this apparent conflict can be divided into two groups: (i) those that propose a lexical difference between singular-​ agreeing collective NPs (80) and their plural-​agreeing counterparts (81), and (ii) those that explain the restrictions on plural agreement in terms of the structural location of the plural feature and the nature of the agreement (Smith 2017: 850–​855).39 The first group includes studies such as Sauerland (2004a, 2004b), who claims that plural-​agreeing collective NPs comprise a plurality operator (i.e. Γ-​1) that makes them definite, even when they take indefinite articles (i.e. ‘hidden definites’). On the other hand, den Dikken (2001) proposes a null pronoun (i.e. pro) as the head of a plural-​agreeing collective NP, which he describes as ‘pluringular’. Both Sauerland’s definite NP and den Dikken’s pronominal-​ headed NP explain why plural agreement with collective NPs is not accepted

83

38  Complex collective subjects and agreement in contexts such as (82), as definite and pronominal NPs do not usually occur in existential constructions.40 (82) *There are a committee (holding a meeting) in the room. (den Dikken 2001: 32) The second group includes Sauerland & Elbourne (2002) (also Elbourne 1999) and Smith (2017). Sauerland & Elbourne (2002) introduce a new feature into the agreement operation, ‘mereology’, to indicate whether the noun is conceived of as having one or more than one member.41 This feature, however, is only accessible to verbal and anaphoric agreement (83), since determiner-​ noun agreement can only be based on conventional number (see Smith 2017: 824): (83) This government is corrupt. They are nothing but crooks. Smith (2017) sidesteps the difficulties encountered by Sauerland & Elbourne’s in attempting to demonstrate their assumption of mereology by opting for a two-​fold conception of the number specification in nouns that comprises a semantic component (iF) and a morphological component (uF). Smith (2017: 836) claims that in collective NPs (CNPs) there is a divergence between the values on the uF and the values on the iF. CNPs are morphologically singular because their uF is singular, while their iF is valued as plural, accounting for their plural properties. This is illustrated in (84) (adapted from Smith 2017: 836), which explains cases of subject-​verb agreement such as (85), where agreement targets at iF, which is plural, so that this is the number valued in the verbal target (Smith 2017: 824): (84)

Number specification for collective noun phrases

ϕ number

[uF:singular]

[iF:plural]

(85) The committee[uF:sg, iF:pl] decide who is hired. This assumption accounts for why plural agreement with collective nouns is more restricted than formal agreement and is not allowed in existential constructions. Since iF is only available at LF (Logical Form) and not at PF (Phonetic Form) (Smith 2017: 836–​838), agreement with iF can only occur before PF and only if iF c-​commands the target, as in (85), where the nominal controller committee c-​commands the verbal target (i.e. “a constituent X c-​commands its sister constituent Y and any constituent Z that is contained

93

Complex collective subjects and agreement 39 within Y”; Radford 2004: 440).42 Otherwise, iF is not accessible to Agree and therefore cannot value the number feature of the verb. Consequently, the ungrammaticality of (86) lies in the fact that it is the ‘dummy’ pronoun that c-​commands the verb and not the iF (Smith 2017: 846): (86) *There are a committee[uF:sg, iF:pl] In contrast, the uF can occur at the PF and this in turn means that the target does not have to be c-​commanded by uF to value the number feature of the verb, making formal agreement less restricted than semantic agreement, hence the grammatical status of There is a committee[uF:sg, iF:pl].43 2.2.3.  Cognitive Grammar and the symbolic alternative A number of scholars have highlighted the drawbacks and limitations of explaining agreement relations on the sole basis of syntactic criteria on the grounds that the semantic component underlying this syntactic operation is of paramount importance when it comes to accounting for mismatches and deviations, the variable agreement patterning of collective nouns being a case in point (see Morgan 1972; Barlow 1988, 1992, 1999; Dowty & Jacobson 1988 and Reid 1991, among others). Alternative approaches to syntactocentric accounts range from purely semantic accounts, such as Dowty & Jacobson (1988) and Reid (1991), to those which partly rely on syntactic mechanisms such as the Feature-​Merging or Unification-​Based framework (i.e. Generalised Phrase Structure Grammar, Lexical Functional Grammar and Head-​Driven Phrase Structure Grammar), which advocate for a non-​derivational “integrated theory of natural language syntax and semantics” (Pollard & Sag 1994: 1; see also Pollard & Sag 1987, 1988; Wechsler & Zlatić 1998a, 1998b, 2000 and Kathol 2000). Cognitive Grammar represents a more radical departure from purely formalist approaches, framing agreement and complex collective NPs in semantic terms as symbolic units of meaning. The remainder of this section looks in more detail at the contributions of one of the leading exponents of Cognitive Linguistics, Ronald Langacker. The foundations of Cognitive Grammar reside not in the autonomy of grammar but in the integration of lexicon and semantics based on the symbolic nature of linguistic structure (Langacker 2009: 1). Language is thus conceived of as a set of symbolic units “having both semantic and phonological import” (schematised as [X /​x]), which range from more specific (i.e. words, phrases, etc.) to more abstract symbolic structures or ‘schemas’ (used to represent grammatical patterns) (Langacker 1990: 290–​294, 2009: 3). These symbolic units, however, are not completely autonomous from semantics, as “serving a specifiable grammatical function is perfectly consistent with being meaningful” (Langacker 1990: 308). Agreement markers can pose a challenge for a semantic account of grammar, given their grammatical nature and functions. Langacker (1990, 1991, 2008, 2009) maintains the compatibility of their functional nature with their semantic

04

40  Complex collective subjects and agreement implications by classifying agreement markers not as features copied from one linguistic item or percolating from one to another, but as meaningful symbolic units that have their own meaning and make a semantic contribution (x and x’ in the quotation below). Agreement markers may be required or optional, and partially or wholly redundant, and may serve particular grammatical functions; however, these factors are not inconsistent with their symbolic nature and meaning (Langacker 1990: 307): In cognitive grammar, agreement markings are viewed as being predications in their own right. … Suppose an agreement pattern has the basic form A+x … B+x’, where x is an inflection representing some property of A, and x’ agrees with x (though they need not be formally identical); that is, even though x’ is manifested on B, its selection is in some way determined by x and thus reflects an aspect of A. Redundancy of agreement is thus conceived of as an instantiation of “multiple symbolization” where “information about some entity is symbolized by more than one component structure within the same symbolic assembly and thus has multiple manifestations in a single complex expression” (Langacker 2008: 188). Put simply, when two elements agree, each of them provides the composite structure they form with semantic contributions which may or may not overlap with the information provided by other elements in the rest of the complex construction. Subject-​verb agreement is understood as a relational predication in which the verb singles out the subject as its trajector (i.e. “figure within the profiled relationship”, Langacker 1990: 296). In (87), what identifies a flock as the subject “is not any particular constituent structure, but rather the fact that its profile44 corresponds to the trajector of … the clausal head”, that is, fly, thus dispensing with movement and transformational rules (Langacker 1990: 296). The contrast in verb number therefore constitutes a meaningful symbolic unit which makes an independent contribution to the whole composite structure since the use of singular or plural verbal forms corresponds to a particular emphasis on “either the trajector’s unitary nature or its internal multiplicity” (Langacker 2009: 52). On that basis, variation in verb agreement with collective nouns as in (87) and (88) is motivated by a change in their conceptualisation “to highlight either the individual or the collective aspect of the motion” (Langacker 2009: 52), as well as in the way in which the subject links up with the profiled process: (87) The faculty has/​have accepted the new curriculum. (88) [A flock of geese] was/​were flying overhead. In (88), flock and geese refer to the same entities (i.e. coextensiveness). They are “two conceived entities that are objectively coincident”, the main difference being “a matter of which aspect of the referent is singled out for profiling at

14

Complex collective subjects and agreement 41 the composite structure level”: either its unitary nature (89) or its multiplicity (90) (Langacker 2009: 53; emphasis in the original). In other words, Cognitive Grammar accounts for agreement variation with complex collective subjects in terms of a metonymic alternation or shift in profile resulting from possible alternative interpretations (Langacker 2009: 52):45 (89) [A flock of geese] was flying overhead, shaped like a V. (90) [A flock of geese] were flying overhead, flapping their wings in unison. In example (89), the only referent which may be profiled is that of N1, as evinced by the singular verb and the applicability of shaped like a V only to the whole flock. In (90), in contrast, the plural pattern places the focus on geese, which evinces that the size and quantity specifications inherent in flock have been given prominence and, as a result, a flock of has been taken as an absolute quantifier, with geese heading the NP (Langacker 2010: 46–​56; §2.1.1). This metonymic shift is optional, however, in the case of N1s which still have a lexical interpretation (e.g. flock, bunch, pack) and impossible in the case of N1s which have undergone extreme grammaticalisation, such as a lot of (*a lot of zebras is color blind; Langacker 2010: 56). In both instances, the referential and quantificational interpretations of the NP are associated with the same syntactic structure, with the latter taking the form of a superimposed additional layer of semantic function (Langacker 2016: 23; §2.1.2). 2.2.4.  Empirical studies To my knowledge, the first corpus-​based study on the agreement patterns of collective nouns is Nixon’s (1972), followed shortly afterwards by Dekeyser’s (1975: 35–​66) more specific analysis of collectives that take of-​ dependents.46 Since the turn of the century, interest in verb number variation with collective nouns has increased considerably, as evidenced by the large number of publications on the matter in response to Bauer (1994: 61–​66), Siemund (1995) and Algeo (1988). These three studies, though pioneering and important, merely describe the agreement preferences of a set of collective nouns in British and American English. The contentious issue of their variable agreement patterns was not addressed in depth until Levin (2001; see also 1998, 1999, 2006), who took the research a step further and carried out a comprehensive study to uncover the intralinguistic (syntactic, semantic and lexical) factors and extralinguistic variables responsible for verbal and pronominal agreement patterns with collective nouns.47 Levin’s findings, together with those of related empirical studies, on agreement variation with (complex) collective NPs, gave rise to most of the variables scrutinised in the synchronic study reported in Chapter 4. This subsection provides an overview of the main empirical works on the subject of agreement variation with collective nouns, with studies divided according to whether they tackle intralinguistic

24

42  Complex collective subjects and agreement (morphosyntactic and lexico-​semantic) or extralinguistic factors. A few more theoretical references are interspersed for convenience. Intralinguistic factors I: Formal complexity Extensive research has been carried out on the question of NP complexity in English, particularly with regard to how to assess complexity and what complexity means in practical terms (see Schäpers 2009; Berlage 2014; Schaub 2016). In this review, I focus on the overt formal complexity48 of the phrase, which refers to the complexity that is overtly manifested in morphological and syntactic patterns (see Bisang 2009: 34, 2014: 129, 2015).49 The complexity of an NP can be measured on the basis of several formal factors: either quantitatively, using metrics such as length, structure and depth of embedding of its nodes;50 or qualitatively, using proxies such as the combination of the nodes and the nominal or sentential status of the NP (Schäpers 2009: 53–​56; Berlage 2014:  6–​18). Structural and syntactic complexity: Length and structure Traditionally, length has been assumed to be the “simplest possible index of surface complexity” (e.g. Smith & van Kleeck 1986: 390). Longer structures lead to a higher degree of complexity and, therefore, tend to be harder to process (see Frazier 1985: 172; Smith 1988: 272; Ferreira 1991: 226–​227; Wasow 1997: 94; Yaruss 1999: 330 and Gibson 2000: 98). Locality is a key factor, therefore, as local dependencies incur a lower processing load, whereas distance between two given linguistic units (say a head and its dependent) increases the cognitive complexity of the structure considerably, leading to greater integration costs (Gibson 1998: 11–​13; Cuyckens, D’hoedt & Szmrecsanyi 2014: 198). Opinions vary regarding the relation between length in number of words and syntactic and phrasal nodes as proxies of NP complexity: while some argue that complexity may be affected by both measures independently (see Wasow 2002: 35 and, for further references, Berlage 2014: 7–​14) and that their relative strength depends on the construction itself (Berlage 2014: 10, 250–​251), others maintain that structural complexity and length are highly correlated (Hawkins 1994: 74; Wasow 1997: 93, 2002: 31–​32; Yaruss 1999: 339; Szmrecsányi 2004: 1037–​1038; Shih & Grafmiller 2011). Empirical studies on collective NPs use both word and node count (also syntactic boundaries) as a predictor of agreement variation. Nixon (1972), for instance, measures the distance in number of words between the collective noun and the agreement form, finding that only agreeing pronouns, not verbal forms, favour high rates of plural agreement. Levin (1998, 1999, 2001: 92–​ 102), in contrast, reports a positive correlation between plural agreement and increasing distance as measured in words and, particularly, syntactic

34

Complex collective subjects and agreement 43 boundaries between the collective and both its verbal and pronominal agreeing forms. The of-​PP accompanying the collective, together with other intervening plural phrases, is adduced as one of the main factors responsible for this trend, as significance is reached after two-​word intervals between the collective and the verbal or pronominal target (Levin 1998, 2001: 92–​102). The finding that distance measured in syntactic boundaries affects verb agreement variation relates to two of the theories presented above: ‘Agreement Hierarchy’ and ‘attraction’.51 Levin’s data corroborate the Agreement Hierarchy prediction of a correlation between the locality of the agreement domain and the likelihood of semantic agreement overriding grammatical agreement (Corbett 1979; Section 2.2.2.1) by showing that targets that tend to be more distant from their controllers, as is the case of pronouns with cross-​clausal and even cross-​ sentential scope, are more prone to plural verb agreement with singular collective controllers (Levin 1998, 2001: 105–​110; see also Nixon 1972) than local targets such as determiners, which may even favour formal agreement with the verb (see Levin 1998: 105–​106, 2001: 121–​125; also Strang 1969: 107). Levin (2001: 91–​102) confirms the preference for plural agreement with increasing syntactic distance across both written and spoken genres, leading him to conclude that the increase in plural agreement should be attributed to the independence that verbal and pronominal targets acquire with distance, and not to the constraints of short-​term memory or the primacy of semantic memory over syntactic memory. Findings from the realm of psycholinguistics, particularly from experimental studies on attraction errors, also support the hypothesis of interference by complexity and distance in the agreement operation. Complex NP subjects in English are highly susceptible to attraction where the first nominal is singular and the second one (i.e. the ‘local’ or ‘oblique’ noun) is plural. This interference is frequently attributed to the proximity of the oblique noun and the verb (91) (Bock & Cutting 1992: 118; see also Bock & Miller 1991; Bock & Eberhard 1993 and Bock, Nicol & Cutting 1999). However, when syntactic distance between the oblique noun and the verb increases, attraction errors still prove significant (92) (Franck, Vigliocco & Nicol 2002: 372; see also Vigliocco & Nicol 1998). ( 91) *The report of the destructive fires were accurate (92) *The threat to the presidents of the company are serious Some studies on collective NPs use attraction to account for plural agreement patterns (93), despite their dual interpretation and variable agreement patterning (Levin 2001: 140; see also Depraetere 2003: 86): (93) [A group of visitors] have paid more than £100 a head to watch her kill.

4

44  Complex collective subjects and agreement Psycholinguistic studies corroborate the strong preference for plural verbal patterns with complex collective NPs, which they attribute to the distributive reading favoured by the of-​PP: “the more conceptually plural a phrase is, the more it should tend to activate a plural verb form” (Haskell & MacDonald 2003: 768; see also Bock, Nicol & Cutting 1999: 342 and Bock et  al. 2006: 76).52 Qualitative dimension of NP complexity Noun phrase complexity depends not only on its structural configuration but also on the combination of the nodes within the NP and, in particular, on the categorial status of the NP. These qualitative factors may in fact be more decisive predictors of complexity, at least in combination with quantitative parameters (Berlage 2014: 1–​6). Premodification of English NPs tends to be relatively common only in written registers, with multiple premodification (three or more modifiers) observed very infrequently owing to its low degree of explicitness and high density of content (Quirk et  al. 1985: 1321–​1322, 1338; Biber et  al. 1999: 588–​589, 597). Postmodifiers, by contrast, though often quite complex, are normally less condensed and therefore more explicit (Biber et al. 1999: 588). Their degree of explicitness, Quirk et al. (1985: 1243) claim, varies according to the structural type of modifier in each case: finite relative clauses (94) are more explicit than non-​finite clauses (95) owing to their overt tense marking, while non-​finite clauses are more explicit than phrases which lack a verbal form (96). (94) the number of people who suffer to this extent [BNC: 1985–​ 1993 ADE 1086] (95) the number of people suffering from flu [BNC: 1985–​1993 K5E 329] (96) the number of people with diabetes [COCA: 2012 ACAD PracticeNurse] Schäpers (2009: 55) stresses the importance of the typology of the phrasal nodes, recalling that “sentential or clausal complementation has always been regarded as belonging to the more complex structures in language”. In fact, sentential dependents are considered to be computed separately and are therefore less likely to interfere in agreement operations and favour a lower rate of attraction effects than phrasal constituents (Bock & Miller 1991; Bock & Cutting 1992; Nicol 1995). In an attempt to reflect the higher complexity of clausal constituents, Berlage (2014: 14–​18, 41–​42) adapts Ross’ (2004: 351) scale of ‘nouniness’ (97), which ranks NP constructions from less to more noun-​like. Berlage’s adaptation uses the same hierarchy as the scale of NP complexity, with the simplest (most noun-​like) elements at the bottom and the most sentential and, therefore, most complex phrases at the top:

54

Complex collective subjects and agreement 45 (97) Ross’ (2004) scale of nouniness

Berlage’s (2014) adaptation

a.  that = that-​clauses most complex that Max gave the letters to Frieda NP+non-​finite  clause b.  for to = for NP to V X for Max to have given the letters to Frieda c.  Q: embedded questions how willingly Max gave the letters to Frieda d.  Acc Ing = [NP+Acc] V+ ing X gerundial construction Max giving the letters to Frieda e.  Poss Ing = NP’s V+ ing X Max’s giving the letters to Frieda f.  Action nominal [Max’s/​the] giving the letters to Frieda g.  Derived nominal NP+PP [Max’s/​the] gift of the letters to Frieda h. Noun least complex spatula Note that non-​finite constituents (d)  are taken to be simpler than finite sentential constituents (a), as “tensed verbs incur a larger memory cost than untensed verbs, because of the additional discourse requirements of tense” (Gibson 1998: 37).53 Intralinguistic factors II: Morphological (un)markedness and complexity The predominance of agreement conflicts and attraction in complex NPs where N1 is singular and N2 is plural is attributed to the so-​called ‘markedness effect’. Since “the singular is typologically unmarked and the plural is typologically marked” (Croft 2003: 89), the marked morphology of plural oblique nouns in the complex NP is more likely to prevail over the morphologically singular head noun and thus interfere in the agreement relation to favour plural verb number.54 The formal markedness of N2 is directly related to explicitness and complexity in that plural number is morphologically and semantically more complex than singular number (Bock & Cutting 1992: 103; see also Moravcsik 1988: 92), which, crucially, makes N2s not only grammatically more explicit but also cognitively more complex (Rohdenburg 2003: 223). In English, singular verbal forms carry overt inflection and therefore represent the more explicit linguistic choice (Rohdenburg 1996: 155). Despite their formal explicitness, in semantic terms they lack a specific reference (i.e. they do not

64

46  Complex collective subjects and agreement necessarily denote ‘one’), which is why they have been taken to be the default alternative (even with collective nouns) when there is no clear evidence to use the plural verbal form (Levin 2006: 327; see also Bock & Cutting 1992: 110 and Depraetere 2003: 124). Along these lines, the ‘Complexity Principle’ (Rohdenburg 1996, 2003) predicts that, where the singular-​plural number contrast in the verb is taken as a constructional variant, complex environments promote higher rates of singular verb number, as the explicitness of the singular verbal form facilitates the processing of a complex subject. Rohdenburg (1996: 155) even suggests that the inflected (i.e. –​s) verbal form “has a special affinity with longer subject expressions” since the longer the subject is, the longer it will take to recognise, integrate and therefore process the subject-​verb relation. Few scholars have explored the impact of the morphological manifestation of N2 plurality (regular, irregular or non-​overt) on verb agreement. Contrary to Bock & Eberhard’s (1993: 80) findings and the cognitive salience or “linguistically atypical” form of irregular plural nouns, Haskell & MacDonald (2003: 771) point to a higher impact of regular plural N2s on erroneous plural agreement (i.e. attraction) than irregular plural N2s, although it is unclear whether this is due to the different types of morphological marking or a lower conceptual plurality of irregular plural nouns: for example, the family of rats vs the family of mice. Intralinguistic factors III: Lexical factors Animacy The interaction between animacy and agreement is reflected in the so-​called ‘Animacy Hierarchy’ (simplified here in (98); see Comrie 1989: 185 and Corbett 2004: 56), in which referents are ranked into three categories from left to right, in decreasing order of degree of animacy. The hierarchy reflects the correlation between the animacy and conceptualisation of the referent: while animate nouns are much more readily conceived of as individuals, inanimate entities are more likely to be viewed as undifferentiated masses (Comrie 1989: 189). (98) human > animate/​animal > inanimate Accordingly, the hierarchy predicts a monotonic decrease in variation in verb number towards the right end of the scale, with animate nouns showing a greater likelihood of singular-​plural number variation (99) (recall that agreement is further constrained by the Agreement Hierarchy in relation to type of target). Though “slightly odd” in Cruse’s estimation (1986: 177), Dekeyser (1975: 53–​56) and Levin (1998, 2001: 127–​129) observe that plural agreement with inanimate referents is possible if the collective noun takes an of-​dependent, has a numerical meaning (e.g. majority or couple) or is frequently associated with human referents, as in (100):

74

Complex collective subjects and agreement 47 (99) [the majority of the population] are implicitly absolved from responsibility [BNC: 1985–​1993 CAF 1227] (100) The berries … could be one of the nightshade family, which are potentially highly toxic. (Levin 2001: 128) Animate/​animal referents (101), however, are found to be an in-​between category (Cruse 1986: 177; Corbett 2004: 188–​189), showing a lower likelihood of semantic agreement than human referents:55 (101) Here a family of mice are beset not only by two vicious cats but by a one-​eyed farmer (Levin 2001: 128) Verb type As discussed earlier (Section 2.2.1), verb number variation with collective nouns is constrained by the type of verb used and in particular by its compatibility with a holistic or distributive reading of the collective (cf. Depraetere 2003: 114). In general terms, Levin (2001: 149–​155) finds that verbs which focus on the composition (102) or size (103) of the collective disallow plural agreement in British English. Verbs of speaking, such as say or insist, and verbs that denote mental states or actions (e.g. believe (104), play) are claimed to show greater variation (see also Depraetere 2003: 118–​119): (102) [The second set of items], Part 2 of the DELV-​ST, consists of/​(*consist of) seven morphosyntactic, four syntactic, and six nonword repetition items [adapted from COCA: 2012 ACAD LanguageSpeech] (103) Britain’s population has increased by a tenth (Levin 2001: 150) (104) Meanwhile, [a host of university officials] believe the potential rewards are vital to economic well-​being. [COCA: 2000 MAG TechReview] It is worth noting that singular agreement is possible in virtually all cases in which plural agreement is used, except when ‘plural-​inducing’ words tilt the balance in favour of plural agreement (Levin 2001: 151–​152): (105) [A herd of young kids] were chasing each other [COCA: 2007 FIC Analog] Except in the above-​ mentioned cases, verb type and meaning are not considered to be the ultimate determinants of number agreement. What do seem to be determining are the individual patterning preferences of each collective noun, as discussed below (Levin 2001: 147–​158, 2006; see also Dekeyser 1975: 35–​66 and Depraetere 2003: 102). Type of collective noun The main trends observed in the literature regarding the individual patterning of collective nouns indicate a preference for singular agreeing forms with

84

48  Complex collective subjects and agreement political, military and decision-​making bodies (e.g. committee or army) and plural agreement with nouns that (i)  identify whole communities or groups (e.g. crew), (ii) are used as almost invariable plurals (e.g. police) and, most importantly, (iii) denote number, especially if they take an of-​ PP and can function as complex quantifiers (106) (see Dekeyser 1975: 57–​65 and Levin 2001: 129–​148; see also Reid 1991: 269–​270 on semantic weight). (106) [A number of artists] attempt to fight off the challenge of this abrasive colour, not unsuccessfully. [BNC: 1985–​1993 CF6 265] Other nouns show variable agreement patterns, although the rate of collective nouns accepting variable concord seems to have been on the decrease since the nineteenth (Dekeyser 1975: 35–​66; Lakaw 2017) and, especially, the twentieth century (Bauer 1994: 61–​66; Siemund 1995: 365–​369; Levin 1999, 2001: 86–​ 90). The stronger preference for singular patterns, even in the British variety, has been attributed to (i) an “American-​led revival” of the historically “older latent pattern” (Collins 2015: 29; see also Depraetere 2003: 112–​113; cf. Bauer 1994: 61–​66); (ii) a possible “parallel long-​term development” of both varieties (Hundt 2009a: 30); (iii) the use of the singular as the default option in the absence of clear semantic and pragmatic reasons to justify plural patterning (see Depraetere 2003) and (iv) the grammaticalisation of “[t]‌he number feature of the nouns” (Levin 2001: 163; see also Dekeyser 1975: 64–​66; Siemund 1995: 366–​369 and Bock et al. 2006). In the case of Ncoll-​of-​N structures, grammaticalisation operates differently in that it is not the number of Ncoll that is conventionalised but the whole Ncoll-​of sequence, which is reanalysed as a complex quantifier of N, with the result that N becomes the new head of the complex NP (as discussed in Section 2.1.2). Evidence for this reanalysis is found in the agreement with N (instead of Ncoll), which is one of the tests used by scholars (e.g. Berg 1998; Smitterberg 2006; Smith 2009; Riveiro-​ Outeiral & Acuña-​ Fariña 2012; Leclercq & Depraetere 2018; Smith, Franck & Tabor 2018) to detect a quantifying use of binominal (partitive) phrases. Extralinguistic factors: Textual and regional factors External factors such as text type, style and medium also show an impact on the agreement patterns of collective nouns. In terms of medium, for instance, Levin (2001: 70–​78) observes higher rates of plural agreement in speech than in writing, particularly with pronominal agreeing forms in American English (cf. Dekeyser 1975: 42–​57). This observation is closely linked to stylistic variation and the type of text involved, as more formal genres and styles are found to conform more readily to the expected ‘correct’ singular patterning than more informal texts (Levin 2001: 78–​86). The extralinguistic factor that has garnered the most attention, however, is regional variation. Levin (2001: 60–​70) confirms the general consensus in the literature that collective nouns take higher rates of singular agreement in

94

Complex collective subjects and agreement 49 American English than in the British variety (see also Algeo 2006: 279 and Trudgill & Hannah 2008: 73), observing also that Australian English adopts an intermediate position between British and American. The examination of other varieties of English, the so-​called ‘World Englishes’, which has a precedent in Bauer’s (1988) experimental study on New Zealand English, is proving a very prolific line of research. There have been numerous studies carried out on inner-​circle varieties, especially New Zealand English (Vantellini 2003; Hundt 2009b; Lindquist 2009; Smith 2009) and Australian English (Hundt 2009b; Lindquist 2009; Smith 2009; Collins 2015), but also Irish English (Sand 2008) and Canadian English (Lindquist 2009), and outer-​ circle varieties such as Singapore English (Hundt 2006; Sand 2008), Philippine English (Hundt 2006; Lindquist 2009), Jamaican English (Sand 2008; Jantos 2009, 2010), Indian English (Sand 2008; Lindquist 2009), Hong Kong English (Lindquist 2009; Wong 2017: 53–​69) and African varieties of English (Sand 2008; Lindquist 2009). Despite extensive research, however, the literature on this topic still lacks a systematic, comprehensive contrastive study to compare findings and establish some generalisations about variation in agreement patterns with collective nouns.

2.3.  Concluding remarks From the arguments presented thus far, it should be clear that complex collective NPs can adopt different structural configurations in English (referential, partitive and pseudopartitive) and that these structures serve different functions and convey different meanings. While referential and partitive phrases are mainly conceived of as left-​headed, fully fledged noun-​complement structures, pseudopartitives are better viewed as right-​ headed phrases. Partitives are analysed as expressing relative quantification through the part-​ whole or subset-​set relation they denote, whereas pseudopartitives express an abstract/​ indefinite quantity of the mass or (count) plurality referred to by N2, making them the perfect locus for the creation of new absolute periphrastic quantifiers. The results reported in the following chapters corroborate these observations. Chapter  3 delves into the quantifying potential of pseudopartitive collective NPs and demonstrates how this configuration has contributed to the diversification of the quantifier paradigm of the English language. The terms ‘binominal/​periphrastic quantifier/​construction’ and ‘Ncoll of N’ are used throughout, despite the lower degree of syntactic flexibility and combinatorial freedom of pseudopartitive and highly idiomatic quantifier constructions.56 My review of the literature in this area reveals that while number agreement variation with complex collective NPs is mentioned in comprehensive grammars, the discussion is framed in rather vague terms. More specific studies on agreement patterns fail to establish a consistent, homogeneous definition and description of these relational collective nouns and sometimes exclude them altogether. The model-​ specific syntactic and cognitive approaches described in this chapter take into account variation in number agreement with Ncoll-​of-​N subjects, yet none of them reflects the full extent of the factors

05

50  Complex collective subjects and agreement and dimensions involved. The contribution of my research thus lies in its use of four levels of analysis to explain verb number variation with complex collective NPs: morphology, syntax, semantics and lexis. The studies reported in Chapters  3 and 4 will explore the extralinguistic and morphosyntactic and semantic intralinguistic factors established by previous empirical research (§2.2.4) in order to model verb number agreement variation from a multifaceted, usage-​based perspective.

Notes 1 ‘Partitive’ has also been used in reference to ‘quality partition’, as in a kind of paper, and ‘measure partitives’, as in a mile of cable (Quirk et al. 1985: 249, 251) or three feet of (that) wire (Kim 2002). 2 Note that Traugott & Trousdale (2013: 23) use ‘indefinite’ in reference to the embedded NP in a lot/​bit/​shred of a N (vs a piece of the pie with a “definite NP2”) and ‘pseudopartitive’ to refer to the whole complex NP, which is argued to be the binominal phrase from which the grammatical quantifiers a lot/​bit/​shred of have developed. According to the definition of ‘pseudopartitive’ adopted in this book, a lot/​bit/​shred of a N would be considered both syntactically (noun-​complement structure) and semantically (part-​whole relation) partitive. 3 Partitive nouns such as piece and bit are not considered here, as they fall out of the scope of this investigation. 4 Advocates of an analysis of partitives such as three/​some of the women as binominal phrases with an elided N1 include Jackendoff (1977), Huddleston & Pullum (2002) and Keizer (2007, 2017). 5 Selkirk (1977: 302) also uses the term ‘measure phrase’ in reference to pseudopartitives. ‘Measure phrase’, however, mainly refers to noun phrases that describe standard measurements, such as three feet long (Jackendoff 1977: 137–​ 141), and in Generative Grammar to the functional category projected by the N1 in pseudopartitives (see Stavrou 2003). ‘Measure nouns’ such as inch, metre and percentage are also related to pseudopartitives since they often occur as N1 in this pattern (see Keizer 2007: 113). 6 Craig (1999: 70) notes that European languages are not classifier languages (i.e. languages “with overt systems of nominal categorization realized by sets of classifiers”) but still exhibit classification phenomena, such as measure expressions in English. 7 Keizer (2007: 139, 2019: 349) classifies number and majority as ‘quantifier’ and ‘measure’ nouns respectively, but at some points discusses number as a measure noun and majority with collection nouns. 8 Brems (2003, 2004) uses the term ‘measure noun’ and later on ‘size noun’ (Brems 2010, 2011: 2, 2012) to refer to a “nominal expression that describes size or shape, implying a measure, such as a bunch of, heaps of, a bit of, a jot of, etc.”, which contrasts with Keizer’s (2007) quantifier, measure, part and collection categories. 9 The (in)definiteness of N2 in pseudopartitives contributes to their definition with respect to recursivity or recursion: while partitive structures are “infinitely recursive” (based on Stickney’s 2009: 61 left-​headed analysis, e.g. A crate of those boxes of the big red cartons of Bessie’s milk), recursion with purely quantificational pseudopartitives results in an ungrammatical outcome in *a number of a lot of people (Keizer 2007: 147).

15

Complex collective subjects and agreement 51 10 In the absence of a determiner, some pseudopartitive configurations may even dispense with of: e.g. Can I borrow a couple (of) sheets of paper? (Selkirk 1977: 308). 11 Structural or syntactic ambiguity is understood here as “ambiguity in which the variant readings of a sentence involve identical lexical units; the ambiguity is thus necessarily a matter merely of the way the elements are grouped together” (Cruse 1986: 66). 12 The referential or ‘literal’ interpretation of complex NPs has also been called the ‘noun-​complement’ reading, since N1s function as fully lexical nouns, that is, heading the phrase, determining agreement, establishing the main referent and taking the of-​PP as its complement (Keizer 2007: 117). 13 Stavrou (2003: 332) and Alexiadou, Haegeman & Stavrou (2007: 465–​466) claim that the difference between a pseudopartitive and a referential construction (N + of-​complement) is a simple matter of low vs high degree of lexicality of N1 (i.e. what they call ‘semi-​functionality’ or ‘semi-​lexicality’), dismissing that the ambiguity between them may be attributable to structural factors. 14 On profile, grounding and quantification in the nominal domain, see Langacker (2004, 2016, 2017b: 329–​365) and Davidse (2004). 15 Stressed some promotes a subset-​ set reading (i.e. some vs not others; e.g. SOME[stressed] men walked in, but not others), while unstressed some or sm does not (e.g. There are some[unstressed, sm] women on the staff). See Milsark (1977: 18–​19). 16 Milsark (1977) proposes a classification which includes not only quantifiers but also the rest of the English determiners, as he claims they all denote quantification. His categories, established on the basis of their acceptability in existential sentences, are ‘weak/​cardinal quantifiers’ (e.g. a, any, sm, lots of, etc.) and ‘strong/​ universal quantifiers’ (e.g. the, all, every, most, etc.). This classification correlates with Langacker’s ‘absolute’ and ‘relative’ quantification, except for the fact that only Milsark acknowledges the systematic absolute-​ relative polysemy with no, any, some, many, and so on. See Brems & Davidse (2003) and Davidse (2004) on Milsark’s claim of the quantificational import of determiners. 17 Absolute quantifiers differ from relative quantifiers in that the former profile the grounding relation whereas the latter profile the grounded entity: while many in the many cars profiles a relation in which the quantifier denotes a certain portion of the scale, most in most cars designates the entity profiled by the noun (see Langacker 2013: 67). Further differences include the fact that relative quantifiers, as (intrinsically) grounding elements themselves, are mutually exclusive with other grounding items (*the all cars), while absolute quantifiers are compatible with other grounding elements (the many/​few/​three cars) and may also act as grounding elements (many apples) or clausal predicates (Our problems are many/​few/​several/​ three) (Langacker 2013: 67). 18 Cf. Langacker (2016: 20) on N1 also being grounded by quantifiers or occurring in the plural form in absolute quantifier constructions. 19 This is also supported by the discourse relevance encoded in the N2’s determiner (i.e. N2 = discourse relevant set): “If the determiner is encoding discourse-​relevant information then it will necessarily project a full D[eterminer]P[hrase]  –​which prevents N2 from being able to head the construction” (Stickney 2009: 181). 20 For a discussion on two types of pseudopartitives in Dutch that differ in headedness, see Vos (1999). 21 The structural ambiguity is clearer with container N1s such as bottle in a bottle of wine broke: for example, a bottle of wine alone is structurally ambiguous and may

25

52  Complex collective subjects and agreement therefore quantify over the wine or refer to the container; however, the potential ambiguity in this instance is cancelled by the verb, as only the bottle, not the wine, may be broken (Akmajian & Lehrer 1976: 406). 22 Lehrer (1986: 141–​145) notes that pseudopartitives also reject preposing the of-​PP (e.g.?[Of animals] we saw [a herd]). 23 Unlike partitive and referential phrases, in pseudopartitives, an embedded complement may undergo extraposition: e.g. [A number of answers] were given [to your argument] vs ?*[A number of the commentaries] have appeared [on Anne’s latest book] (Selkirk 1977: 306, 309). This conforms to the constraint on NP-​cyclical nodes, whereby “[n]‌o element may be extraposed more than one cycle up from the cycle containing it” (Akmajian & Lehrer 1976: 396). 24 See Langacker (2010: 54, 2013: 79) on the semantic attenuation and still functional potential of a in grammaticalised complex NPs such as a lot of. 25 See Yuasa & Francis (2003: 197–​198) and Francis & Yuasa (2008: 50) for a similar explanation based on a multi-​modular approach to grammar. 26 Further periphrastic quantifiers in PDE include a load of, a whiff of, a flicker of (Brems 2011), a deal of (Traugott & Trousdale 2013), a bit of and a shred of (Brems 2011; Traugott & Trousdale 2013). 27 The grammaticalisation of originally binominal noun phrases into quantifiers is observed cross-​linguistically: for example, un aluvión de (‘a flood of’) in Spanish (Verveckken 2012, 2015, 2016; Delbecque & Verveckken 2014); tas d’/​de (‘a heap of’) (Brems 2015) or la/​une foule (‘the/​a crowd’) in French (Tristram 2015); ein bisschen X (‘a bit of’) or ein wenig X (‘a little of’) in German (Neels & Hartmann 2018), to cite just a few examples. 28 I consulted the usage guides of the HUGE (Hyper Usage Guide of English) database, which comprises references from 1770 to 2010. The selection was made on the basis of (i) problem tags ‘agreement’ or ‘concord’, (ii) availability of the reference, (iii) representativeness of the whole twentieth and the early twenty-​first centuries and (iv) language variety (British and American). 29 Semantic interference in the agreement operation ultimately depends on the strength of the morphological system of a given language (Berg 1998: 61; see also Acuña-​ Fariña 2009, 2016b and Riveiro-​Outeiral & Acuña-​Fariña 2012): when syntactic and semantic principles compete in the resolution of number conflicts (e.g. verb agreement with (complex) collective subjects), the morphological attrition of the English language system makes it more vulnerable to semantics than rich morphological paradigms with “high frequency of syntactically based agreement processes” of languages like Spanish and German. 30 Theoretically, this asymmetry is reflected in the ‘feature-​ copying’ or ‘feature-​ checking’ operations within the Generative framework and by ‘anchoring’ in Head-​ Driven Phrase Structure Grammar (see Corbett 2006: 114–​116). 31 Note that the use of the terms ‘semantic’, ‘notional’ or ‘logical’ does not imply that ‘syntactic’, ‘formal’ or ‘grammatical’ agreement may not be semantic as well (see Levin 2001: 27–​28). 32 Already within the Government and Binding and Principles and Parameters models, the so-​called ‘X-​Bar Theory’ predicts that lexical elements heading a constituent (X)  project their properties onto two further levels: the ‘bar level’ (X’) and the ‘phrasal level’ (XP or X’’), to accommodate complements and specifiers, respectively. 33 Case is relevant to the agreement operation because it “reflects some aspect of its grammatical relationship to the central verb” (Baker 2013: 608) and all NPs must

35

Complex collective subjects and agreement 53 be assigned case in order to be well formed and grammatical (on ‘Case Filter’, see Hornstein, Nunes & Grohmann 2005: 26–​32 and Cook & Newson 2007: 75). 34 X-​Bar Theory predicts that functional heads (e.g. T and AgrS) can only agree with their specifiers (Boeckx 2006: 5; Baker 2013: 611–​612). 35 In view of the lack of a unified theory within the Minimalist framework and the different analyses proposed in the literature on agreement, here I follow Radford (2004), Hornstein, Nunes & Grohmann (2005), Boeckx (2006), Baker (2013), Bošković (2013) and Epstein et al. (2013). 36 For current theoretical debates on Agree, see Smith, Mursell & Hartmann (2020). 37 For the application of the Agree operation within the nominal domain, see Belletti (2001), Coene & D’hulst (2003: 1–​33) and Corver (2013). 38 See Smith (2017: 855–​ 860) for an account of regional variation within the Generative framework. 39 To my knowledge, there are no references to agreement resolution with complex collective subjects. 40 For further contexts where plural agreement with collective NPs is restricted (e.g. the reconstruction of a raised collective NP and its predicate reading), see Smith (2013, 2017); see also den Dikken (2001). 41 For a discussion on mereology, see Sauerland & Elbourne (2002: 289–​296). 42 For a full account of semantic agreement and the two-​fold conception of Agree (i.e. AG R EE-​L IN K and AG R EE-​C OPY ), see Smith (2017: 833–​849). 43 For further evidence in support of this proposal, see Smith (2017: 834–​835). 44 The ‘profile’ is the substructure which is focalised in a given context from the conceptual base of a given linguistic structure or relation (Langacker 1990: 5). 45 Examples that lack overt inflection determining the nature of the trajector (e.g. A flock of geese flew overhead) leave the subject-​ verb relation indeterminate (Langacker 2009: 52). 46 Previous attempts to systematise agreement patterns with collective nouns include Pooley (1934), Strang (1966), Bailey (1987) and Fries (1988). 47 Levin (2001) includes some collective nouns with of-​ PPs and excludes other examples on the basis of their bias for plural agreement; his distinction between these and other (non-​complex) collective NPs is not consistent, however. 48 For an overview of overt complexity types other than formal complexity, see Schäpers (2009: 49–​56), where she also discusses ‘cognitive complexity’ (i.e. acquisition of grammatical morphemes and how the meaning of words affects their acquisition) and ‘processing complexity’ (i.e. how language is processed and understood syntactically and lexically). 49 There is also a hidden side of complexity: morphosyntactic structures frequently leave certain aspects of their intended meaning unexpressed, while others, though apparently quite simple on the surface, may be counterbalanced by a heavy reliance on pragmatic inferring. Both sides of complexity are independent but not mutually exclusive (for further discussion, see Bisang 2009, 2014, 2015). 50 For a detailed account of both theoretically based (e.g. Frazier’s 1985 ‘nonterminal-​ to-​terminal ratio’) and applied complexity metrics (e.g. ‘mean length of utterance’ MLU and ‘mean clause utterance’ MCU), as well as her own metric ‘Linear Complexity Analysis’ (LCA), see Schäpers (2009: 57–​74). For an alternative metric based on the number of grammatical embeddedness indicators, see Szmrecsányi (2004). 51 Levin (2001: 102–​104) posits attraction as a special case of distance and another factor favouring plural patterning (never ‘performance errors’) with collective  nouns.

45

54  Complex collective subjects and agreement 52 Humphreys & Bock (2005) and Bock et al. (2006) find that collective heads constitute an intermediate category between singular and plural heads in terms of their elicitation of plural verb forms and that their co-​occurrence with plural dependents increases the rate of plural agreement significantly: e.g. the sailor/​sailors/​crew with the peacekeeping force(s). 53 For the interaction of complexity and clausal status with clause depth, see de Haan (1989, 1991). 54 ‘Markedness effect’ (referred to as ‘mismatch asymmetry effect’ in Barker & Nicol 2000: 100) has been widely observed in experimental studies (e.g. Bock & Miller 1991; Bock & Cutting 1992; Bock & Eberhard 1993; Vigliocco & Nicol 1998; Bock, Nicol & Cutting 1999; Bock, Carreiras & Meseguer 2012). For a discussion on formal and semantic markedness, see also Battistella (1990: 36); Matthews (1991: 234–​236); Blevins (2000) and Levin (2006: 325–​329). 55 Animacy was found not to interfere in agreement when N1 and N2 differ in degree of animacy (e.g. The speech of the authors vs The author of the speeches; Bock & Miller 1991). 56 This decision is in keeping with Verveckken’s (2015) use of ‘binominal quantifier’ to refer to structures that were originally binominal but are now grammaticalised quantifying expressions, such as un montón de (‘a heap of’) in Spanish.

5

3  Insights from diachrony Reconciling form and meaning

This chapter reports the results of a qualitative and quantitative diachronic study on the historical evolution of a selection of complex collective noun phrases designed to determine the impact of lexical issues on their agreement patterns. My hypothesis is that some complex collective constructions exhibit colligational and collocational preferences which constrain their potential verb number variation and signal their idiomatic or even grammaticalised status as periphrastic quantifying expressions. The main aim pursued by this case study is to explore to what extent agreement patterns and variation are constrained and determined by the lexical choice of the collective noun (Ncoll) and to assess the degree of idiomatisation (i.e. syntactic fixation and semantic opacity) of each of the complex collective NPs (Ncoll of N(pl)) in Present-​Day English. The study investigates the following research questions: (i) Is there evidence of a diachronic evolution? Have there been any significant changes in relation to complex collective subjects or collective nouns more generally which may have influenced their current verbal patterning and meaning? (ii) What is the quantifying potential (if any) of complex collective subjects? To what extent does the interaction between the of-​PP and verb agreement contribute to this use? (iii) Is lexis a determining factor? Is verb number agreement affected by the type of verb, type of collective noun or type of noun in the of-​PP? As explained in Section 1.2, this study focuses on a set of seven collective nouns with high rates of plural agreement and/​or relative frequency in PDE: number, group, majority, bunch, couple, host and minority. In light of the findings from previous studies on the grammaticalisation of binominal noun phrases (reported in Chapter 2; see also Fernández-​Pena 2017b), the study was further restricted to the occurrence of these seven collective nouns in mainly indefinite noun phrases (i.e. a Ncoll of N), the only configuration which has been shown (as confirmed here) to be susceptible to grammaticalisation (Traugott & Trousdale 2013: 116; see also Keizer 2007: 135–​137). This small data set has the additional advantage of enabling an in-​ depth individual quantitative and qualitative diachronic analysis that will offer valuable insights into

65

56  Insights from diachrony the constructions’ collocational and colligational patterns in PDE. It is also important to note that this constitutes a first approach to the topic and, as such, the conclusions made here cannot be extrapolated to all Ncoll of N(pl) or contexts of use other than subject position. The chapter is organised as follows. Section 3.1 describes the corpora used and the data retrieval process. Section 3.2 presents the data and discussion of the results obtained for the sample of structures surveyed: a number of (§3.2.1), a group of (§3.2.2), a majority of (§3.2.3), a bunch of (§3.2.4), a couple of (§3.2.5), a host of (§3.2.6) and a minority of (§3.2.7). In Section 3.2.8, these results are used to provide a classification of the seven constructions and the quantificational potential of each one is discussed. The chapter concludes with a summary of the main findings and some final remarks.

3.1.  Methodology: Corpora and data retrieval The data were retrieved from one of the largest electronic historical corpora of the English language currently available, the Corpus of Historical American English (COHA) (Davies 2010–​), which comprises 406,232,024 words from fiction and non-​fiction texts (books, magazines and newspapers) published between 1810 and 2009 and which can be accessed online through the Brigham Young University interface.1 Additional historical corpora were also surveyed to obtain data for British English: A Representative Corpus of Historical English Registers (ARCHER 3.2), the Penn Parsed Corpus of Early Modern English (PPCEME) (Kroch, Santorini & Diertani 2004), the Penn Parsed Corpus of Modern British English (PPCMBE) (Kroch, Santorini & Diertani 2010) and the Corpus of Late Modern English Texts, version 3.0 (CLMET3.0) (De Smet, Diller & Tyrkkö 2013), all covering the period 1500–​1999. American data from earlier periods were also consulted, drawn in particular from the Corpus of Early American English (CEAL, 1690–​1920) (Höglund & Syrjänen 2016). Unfortunately, none of these sources provided enough results to permit a robust quantitative analysis and, therefore, I have only been able to establish firm conclusions regarding trends in American English in COHA.2 The data were obtained in various steps, the first of which was to search for complex collective subjects containing: (i) an article (indefinite or definite): although my focus is on indefinite NPs, their definite counterparts were also retrieved for the purposes of comparison and illustration; (ii) a singular collective noun: bunch, couple, group, host, majority, minority and number; (iii) an of-​PP consisting of the preposition of, an optional element (e.g. a possible determiner and/​or premodifier) and either a morphologically marked plural noun (e.g. boys, things) or the non-​morphologically marked plural noun people,3 both referred to as Npl henceforth; (iv) an inflected verbal form.

75

Insights from diachrony 57 Only subjects in which the two nouns bear different number values (e.g. a bunchsg of grapespl) were considered, as this is the only syntactic context in which the status of the collective as a head or quantifier can be consistently assessed (Brems 2011: 129). In relation to the of-​PP, the decision to explore only cases in which the complement of of is formed by a noun on its own or preceded by one word (usually a determiner or a modifier) was taken in an attempt to avoid more complex syntactic configurations and thus ensure the viability of an in-​depth, fine-​grained analysis. Since COHA is only POS-​tagged and not parsed, to retrieve instances that meet these criteria, the following search patterns were used:4 (i) ‘(a/​the) (bunch/​couple/​group/​host/​majority/​minority/​number) of (*) (*.[NN2]/​people) *.[(VBZ/​VBDZ/​VDZ/​VHZ/​VVZ)]’ for singular  verbs; (ii) ‘(a/​the) (bunch/​couple/​group/​host/​majority/​minority/​number) of (*) (*.[NN2]/​people) *.[(VBR/​VBDR/​VD0/​VH0/​VV0)]’ for plural  verbs. In these patterns, ‘NN2’ stands for morphologically marked plural nouns. ‘VV0’ stands for the base form and ‘VVZ’ stands for the third person singular of all lexical verbs except those which may be lexical or auxiliary, which are retrieved with the following tags: ‘VBR’ and ‘VBZ’ for the present forms of to be, ‘VBDR’ and ‘VBDZ’ for its past forms, ‘VH0’ and ‘VHZ’ for to have, and ‘VD0’ and ‘VDZ’ for to do.5 The parentheses do not belong to the query syntax; they are used here for clarification purposes to represent the different options used within each search pattern. After pruning the results and discarding examples not valid for the analysis, the total number of hits retrieved from COHA was 4,776. The default metadata provided by the online interface of COHA (i.e. date, genre, source) were incorporated into the database as three individual variables. The list of factors categorised was completed by variables related to: (i) the whole NP: type of structure, determiner and collective noun; (ii) the of-​PP: type (if any) of determiner or modifier of the second noun (i.e. Npl); (iii) the verb: type of verb, number value and status (i.e. auxiliary, semi-​modal or main verb); (iv) the period: date of publication was arranged into four periods: 1810s–​ 1850s, 1860s–​1900s, 1910s–​1950s and 1960s–​2000s. As a second step, to obtain additional data on aspects such as frequency of the complex collective noun phrases over time or premodification patterns, I  used the unfiltered absolute frequencies provided by the corpus statistical tools (164,639 hits) (i.e. the data were not manually pruned and so included syntactic contexts in which the collective noun was not necessarily in subject position). Henceforth, these data are referred to as ‘data from the whole COHA corpus’ to distinguish them from the data retrieved from the corpus and analysed manually (‘COHA data’ or ‘data (retrieved) from COHA’). In the

85

58  Insights from diachrony case of the frequency of the constructions, the raw data were normalised to 10,000 words per period.6 The parameters controlled for as possible indicators of the syntactic fixation of the structures relate to the following collocational restrictions: (i) premodification patterns: in quantifier uses, premodification is restricted to quantificational elements which normally reinforce the quantifying meaning evoked by the construction (e.g. a whole bunch of studies; Brems 2011: 194–​201; see also Keizer 2007: 138–​140); (ii) specialisation with the indefinite article: decategorialisation of the nominal status of Ncoll applies only in indefinite constructions (Traugott & Trousdale 2013: 116); (iii) verb agreement: when a Ncoll of is used quantificationally, verb number is plural, as it is determined by Npl; plural agreement is thus taken as an indicator of the potentially quantifying use of a Ncoll of (though this does not preclude categorically the possible occurrence of semantic agreement with a referential use of the collective noun) (see Keizer 2007: 120–​126 and Brems 2011: 129); (iv) the presence and type of determiner of both Ncoll (i.e. Det1) and Npl (i.e. Det2): whether Det1 is definite or indefinite and whether Det2 is present or not, which helps to distinguish broadly between (a) referential, (b) partitive and (c) pseudopartitive structures (Verveckken 2015: 48–​51): a. the Ncoll of (Mod)7 Npl  the group of (young) girls b. a Ncoll of Det2 Npl  a group of the (young) girls c. a Ncoll of (Mod) Npl  a group of (young) girls The focus of this study is (c), which has the potential to develop an idiomatic meaning as a periphrastic absolute quantifying expression and, secondarily (b), where Ncoll has been observed to denote relative quantification (Section 2.1.1). Examples of referential uses of Ncoll-​of-​Npl constructions are also interspersed in the discussion for the purposes of comparison and illustration. As regards semantic opacity, the 4,776 instances retrieved from COHA were analysed individually in order to determine whether they convey a comparable grammatical/​ quantificational meaning to that of grammaticalised constructions, such as a lot of, or whether, by contrast, the collective noun retains its original lexical meaning.

3.2.  Analysis of the data This section examines historical data for Ncoll-​of-​Npl constructions in order to determine the extent to which their evolution in the recent history of English has affected and, specifically, constrained their patterns of agreement. To the best of my knowledge, few diachronic studies have been done on this topic (Dekeyser 1975: 35–​66; Smitterberg 2006; Brems 2010; Shao, Cai & Trousdale 2019), making this study a necessary contribution to a still under-​researched  area.

95

Insights from diachrony 59 3.2.1.  A number of The noun number has different senses, all of them based on its intrinsic numerical meaning: it refers primarily to enumeration (i.e. the ‘precise sum or aggregate of a collection of individual things or persons’), as in (1), and to the graphic representation of that quantity, along with other more specialised meanings (e.g. a telephone number or an issue of a periodical/​magazine; OED, s.v. number I): (1) [The number of fools] is infinite. [1797 R.  SOUTHEY Lett. from Spain xxiv. 439] Number also has a more collective-​ like meaning denoting things that are collected together, groups of people or, less specifically, aggregates (particularly large aggregates) of individuals (not necessarily counted) (OED, s.v. number II). This last sense is shown in (2), which reveals that, as early as the seventeenth century, a number of was used in a similar fashion to homologous constructions such as a lot of: apart from plural verb number selection, the meaning of the construction seems to allow for a clearly quantificational reading denoting an indefinite number of entities (herbs, in this case): (2) Water-­Lilly..hath a Root in the Ground; And so have [a Number of other Herbs] that grow in Ponds. [1626 F. BACON Sylua Syluarum §567] Strang (1966: 79) observes and discusses this same patterning, later confirmed by Dekeyser’s (1975: 59) corpus-​based study, where he, too, points to the complementary distribution of agreement patterns and article selection with number: The trend is for a number (of) to pattern with the plural, as 1839 Ainsworth J 235 A number of farming-​men were passing and repassing about their various occupations, and for the number (of) to pattern with the singular, as 1836 Lane M 135 The number of persons who enjoy this distinction has become very considerable. Reid (1991: 280–​283) and Smitterberg (2006: 265) also corroborate this alternation. Along these lines, Langacker (2010: 56) maintains that plural agreement is compulsory when number is “invoked only for purposes of assessing other, more tangible entities”, as in the indefinite expression a number of in (3). The number of, on the other hand, is not compatible with a quantificational reading owing to the definite article and the singular verbal form (see Akmajian & Lehrer 1976: 411 and Keizer 2007: 121): (3) [A number of quantifiers] (*remains/​remain) to be described. (Langacker 2010: 56)

06

60  Insights from diachrony It has been frequently argued that this complementary distribution of agreement patterns stems from the multifunctional nature of number as both a lexical noun (the number of) and a quantifier-​like noun (a number of) (for detailed evidence, see Klockmann 2017: 238–​245). There is extensive literature on this pattern alternation, including case studies on the quantifying use of a number of, as well as a number of more general references to the topic. The most relevant of these include: Selkirk (1977), Lehrer (1986, 1987), Sinclair (1989), Berg (1998), Keizer (2007: 121), Acuña-​Fariña (2016a), Zhang (2017: 66–​75) and Leclercq & Depraetere (2018). The data obtained from COHA for this study provide historical evidence for the claims and observations found in this earlier work. To start with syntactic fixation, the patterns of verbal agreement with a number of reveal a clear predominance of plural verb number as early as the beginning of the nineteenth century, as examples (4)  and (5)  and Table  3.1 illustrate: (4) [A number of eminent scientists] are active in promoting closer tics[sic] between scholarship and religion [COHA: 1958 MAG SatEvePost] (5) On one side [a number of birds] perch in a single, rather flimsy, tree [COHA: 1997 MAG Antiques] The occurrence of singular verbal forms with a number of was found to be negligible, limited to just one instance per period, with the exception of the period 1910s–​1950s. It follows from these data that a number of was largely fixed in terms of number agreement patterns by the nineteenth century, with verb number not being determined by the singular number of the noun number. This supports the decategorialisation or loss of lexical function of number, its ‘quantificational’ function (as claimed by some scholars) and the subsequent predictable development of a number of into an idiomatised/​grammaticalised structure. In order to qualify as a quantifier construction, however, a number of must exhibit not only a particular form but also a particular meaning (see Traugott & Trousdale 2013: 17). The results of a qualitative analysis of the examples retrieved from COHA showed that in most cases the lexical meaning of number (i.e. the ‘precise sum or aggregate of a collection of individual Table 3.1 Frequency of verb number agreement with a number of in the COHA data (N=1,609) Verb number

singular plural Total

Period 1810s–​1850s

1860s–​1900s

1910s–​1950s

1960s–​2000s

1 (0.79%) 126 (99.21%) 127

1 (0.30%) 329 (99.70%) 330

6 (1.19%) 500 (98.81%) 506

1 (0.15%) 645 (99.85%) 646

16

Insights from diachrony 61 things or persons’; OED, s.v. number I.1.a) has been bleached in favour of a quantificational reading denoting ‘a large or considerable … collection or aggregate of persons or things, not precisely reckoned or counted’ (OED, s.v. number II.8, 9, 10; emphasis in the original), as illustrated in (6)–​(8): (6) [A number of papers] were very soon sent in; some with names, some anonymously. [COHA: 1886 FIC MortalAntipathy] (7) From this, [a number of important consequences] follow [COHA: 1990 NF ThinkingSociologically] (8) [A number of stories] were based not on off-​the-​record conversations but on public statements and documentation by U.N.  inspectors. [COHA: 2005 NEWS WashPost] These examples illustrate that a number of denotes a quantificational meaning in reference to an indefinite quantity of entities or individuals, similar to the grammatical meaning of the conventionalised quantifying expression a lot of. The historical data thus support the desemanticisation and subsequent semantic opacity which is attributed to a number of in the literature. It should be noted, however, that this semantic change must have taken place long before the temporal span covered by this study since, according to the data presented here, it was already well established in late Modern English.8 These arguments are further supported by the comparative study of the trends observed for the construction’s definite counterpart, the number of. According to the parameters of idiomatisation considered here, syntactic fixation and semantic opacity should only affect the indefinite structure (Traugott & Trousdale 2013: 116). The data retrieved for the number of confirm this prediction, with singular verb number patterns predominating (>94%) over much more marginal plural verbal patterns (1.30103

guys (3), keys (2) kids (3) tourists (2), friends (2), teenagers (2), fans (2), computers (1), hands (1), rugs (1), salesmen (1) –​

>3 >2 >1.30103

minority

bunch

>2

>1.30103 repulsion

repulsion

repulsion

reference (e.g. courses, shares, studies, systems). These results suggest that, despite its preference for human entities, majority may co-​occur with inanimate oblique nouns as well. While there is no a priori explanation for its repulsion38 of men and kids, the fact that birds was also repelled highlights the seemingly infrequent occurrence of majority with non-​human entities (note that none of its collexemes2 refers to animals). As a final remark, recall that majority showed the strongest association with plural agreement in the collexeme analysis (266.23). The analysis of its collocates is thus consistent with those findings in that most of its collexemes are human, a type of referent which has been demonstrated to favour plural agreement patterns. As regards number, Table  4.10 shows that number of people is the most frequent collocation. However, even though number commonly attracts this noun with human reference, it also repels human referents such as men, students and Americans. This finding seems to be in keeping with its general preference for inanimate collexemes: of the remaining 21 collocates, 15 have inanimate referents (responses, houses, seats, works, conditions, calls, chains, characteristics, colours, dwellings, permits, themes, options, schools and

3 6 1

Modelling variation in verb agreement 163 transactions) and the other six refer to either animate referents (partners, cells, collectors, scholars and researchers) or both animate and inanimate entities (e.g. organisations). Group and minority were found to be relatively similar in terms of collocational preferences. Group showed a strong association with animate, mainly human referents: of the 22 significant collexemes detected, 17 refer to human entities (youngsters, nuns, boys, leaders, people, men, students, agents, democrats, volunteers, prisoners, subjects, investors, pilots, nurses, protesters and skiers), while only one makes reference to animals (dolphins). Likewise, minority occurs almost exclusively with oblique nouns with human referents, the exception being companies, which may also denote inanimate reference. While group was found to collocate with more inanimate collexemes (notes, elements and paintings) and with a noun that may be both animate and inanimate (firms), these were more marginal collocations. These observations are in line with findings in the literature regarding its productivity and flexibility as to the type of referent it denotes (Biber et al. 1999: 249). People, which is one of the most significant collexemes of group, was observed to be significantly repelled by minority: although my data attest 11 tokens of minority of people, the extremely high frequency of people (894 tokens) compared to the number of times it occurs with minority may account for this finding of repulsion. Finally, for couple, host and bunch, more than half of their (significant) collexemes are not reported in Table 4.10 (44, 35 and 29, respectively) owing to the fact that they occur only once in my database. In any case, no clear trends could be discerned from these instances. Focusing then on the rest of their collexemes, the results reveal a clear difference between couple and bunch, on the one hand, and host, on the other: host takes almost exclusively inanimate referents (factors, angels, laws, questions, programmes, events, issues, countries), with the exception of agencies, which may have both inanimate and animate human referents. This trend is further supported by its repulsion of people (they co-​occur only three times in my database), which confirms that host is very unlikely to be found in reference to human and, in general, animate referents, in contrast to its lexical use (‘army’). It follows from this finding that, despite the increase in singular agreement in the twentieth century attested in COHA (§3.2.6), it seems probable that the examples of host in the BNC and COCA illustrate quantifying uses of the collective noun, as this would explain its strong association with both inanimate oblique nouns and plural verbal patterns. Couple and bunch, by contrast, were very frequently associated with animate, particularly human entities. In fact, their strong collexemes all denote human referents, that is, kids, people, ladies, friends, doctors, guys, tourists, teenagers, hands.39 Nevertheless, a few inanimate referents occur with considerable frequency with bunch as well, as in the case of computers, rugs and keys, the last of these being the prototypical collocate of the lexical meaning of this collective noun. Compared to keys, however, the rest of the collocates are less easily conceivable in terms of a physical bunch. In these cases, it seems

4 6 1

164  Modelling variation in verb agreement Table 4.11 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and mainly human referents collexeme1 collexeme2 band

party

troop

crowd

gang

coll. strength

musicians (4), adventurers (2), windows (2) thieves (2), drivers (2), volunteers (2) travellers (2), farmers (2), advisers (1) believers (1), chefs (1), criminals (1), devotees (1), dinosaurs (1), fighters (1), hunters (1), looters (1), natives (1), pensioners (1), pickers (1), pilgrims (1), rebels (1), thinkers (1), warriors (1) people (1)

>3 >2 >1.30103

firefighters (2), tourists (3), riders (2) men (5), anglers (2), soldiers (2), travellers (2) astronomers (1), inspectors (1), aides (1), faces (1), gulls (1), Indians (1), journalists (1) people (6)

>3 >2 >1.30103

monkeys (4), soldiers (2) chefs (1), rapists (1), rebels (1) peasants (1), businessmen (1), schoolchildren (1), creatures (1), horses (1) –​

>3 >2 >1.30103

onlookers (6), fans (6) locals (4), bystanders (2), spectators (2), strangers (2) reporters (2), teenagers (3) visitors (2), kids (3), men (5), citizens (2) –​

>3

robbers (3), boys (6), youths (4), pickpockets (2), ruffians (2), workmen (2) thieves (2) men (4), teenagers (2), backers (1), cats (1), dealers (1), fighters (1), pickers (1), pots (1), rats (1), brothers (1), terrorists (1) people (1)

>3

repulsion

repulsion

repulsion

>2 >1.30103 repulsion

>2 >1.30103 repulsion

that bunch may be used instead with a more general collective or even abstract quantity meaning. In the interests of clarity, the collexemes of the singular pattern have been arranged in separate tables according to their collocational preferences in terms of animacy. Table  4.11 displays the results of the collocations which more clearly denote human referents, which are associated with subjects with the collexeme1 band, crowd, gang, party and troop.40 Band and party collocate mainly with human referents (e.g. band of musicians/​warriors, party of tourists/​soldiers). Their collexemes differ from those of other collectives in that they reveal an implied nuance of ‘common

5 6 1

Modelling variation in verb agreement 165 goal’ or ‘common characteristic(s) of the individuals’: thieves, volunteers, riders, hunters, pilgrims. The idea that these nouns do not simply denote a group of human individuals in general but, instead, carry the more specific nuance of ‘common purpose or feature’, as illustrated in (94) and (95), is reinforced by their repulsion of people. (94) On the edge of the crowd [the inevitable party of Japanese tourists] busily video the scene [BNC: 1985–​1993 A8W 178] (95) [another band of opportunistic thieves] comes ashore in search of provisions. [COCA: 2010 NEWS CSMonitor] This is also the case with troop, which takes mainly human collexemes (e.g. soldiers, chefs, peasants), as its lexical meaning refers to a body of soldiers (96). (96) In 1849 [a troop of soldiers] was ordered to northeast Texas to protect settlers [COCA: 2003 NEWS Houston] As for gang and crowd, the collexemes of the latter were found to be exclusively human, while with the former some exceptions were observed, such as cats, pots and rats. It is worth noting that, as in the case of band, gang co-​ occurs with a series of collexemes with negative connotations, such as robbers, ruffians, thieves and terrorists (see Biber et al. 1999: 249), which indicates that both seem to imply a certain pejorative nuance. The next set of collocations to be discussed involves collective nouns which take oblique nouns that are mainly associated with animate but non-​human referents: clump, flock, herd and swarm. Table 4.12 reports the absolute frequencies of the NplS with each of the collective nouns and their covarying collexeme strength. This set follows Biber et  al.’s (1999: 249) findings. Accordingly, clump occurs solely with collexemes that refer to plants (e.g. leaves, shrubs, trees) and micro-​organisms (cells), while the other three collectives show a strong association with non-​human collexemes (e.g. flock of birds/​geese/​eagles, herd of elephants/​goats/​whales, swarm of bees/​mosquitoes/​crows). Nevertheless, there is also a considerable number of human referents associated with these three collectives, particularly in the case of swarm, as a result of the metaphorical extension of their lexical meaning to denote a large group but with hyperbolic effect, as illustrated in (97) (see Biber et  al. 1999: 249). Interesting in this regard is the repulsion of people by flock and swarm, which once again indicates that neither of these two nouns tends to denote aggregates of people without any additional intended meaning. Despite their collocation with plural animate referents, note that all four nouns (like those discussed below) show a strong preference for singular agreement (§4.4.2.4.iii). (97) [a swarm of screaming girls] surrounds us, they are pulling at him, “Stop! Stop!,” he is flashing a knife [COCA: 1994 MAG HarpersMag]

61

166  Modelling variation in verb agreement Table 4.12 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and animate non-​human referents collexeme1 collexeme2

coll. strength

clump

leaves (2), cells (2) shrubs (1) trees (1), plants (1) –​

>3 >2 >1.30103 repulsion

flock

birds (19), geese (9), pigeons (7), ducks (3), cranes (2), doves (2), pelicans (2), seagulls (2), starlings (2), turkeys (2) newcomers (2) bumps (1), commuters (1), crows (1), eagles (1), titles (1) people (1)

>3

elephants (11), cows (7), goats (3), animals (4), horses (3), wildebeests (2), whales (2) –​ dinosaurs (1), feet (1), reporters (1) –​

>3

bees (9), flies (4), mosquitoes (3) –​ girls (2), crows (1), hands (1), helicopters (1), pedestrians (1), technicians (1), experts (1), points (1) people (1)

>3 >2 >1.30103

herd

swarm

>2 >1.30103 repulsion

>2 >1.30103 repulsion

repulsion

The vast majority of Ncoll-​of-​Npl subjects with inanimate referents have as their collexeme1 batch, rash, series or set (see Biber et al. 1999: 249). In fact, only a few of their collexemes are animate (e.g. batch of Republicans, series of figures, rash of babies). The results from the covarying collexeme analysis are reported in Table 4.13. As before, the lexical meaning of the collexemes1 affects the type of collexemes2 accepted. For instance, the strongest association of batch, set and series is with nouns that refer to things that can be conceived of or treated as units or lots (e.g. batch of eggs/​papers, set of rules/​items, series of tests/​studies), which may be applied in a more general sense even to groups of humans when treated as ‘sets’, as in (98). (98) while [a second batch of prisoners] was flown under cover of darkness to Cuba. [COCA: 2002 NEWS AssocPress] Rash is the only collective noun in this set that does not repel people; nevertheless, it does not buck the general trend since it attracts only one human collexeme, that is, babies (99). Rash also seems to be the only collective noun in the set that is more strongly associated with negative collexemes. Note, in

7 6 1

Modelling variation in verb agreement 167 Table 4.13 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and inanimate referents collexeme1

collexeme2

coll. strength

batch

eggs (4), units (3), cars (3) bills (2), monitors (2), papers (2), recruits (2) prisoners (2), reports (2), machines (2), apartments (1), archaeologists (1), Britons (1), chemicals (1), coolies (1), criminals (1), documents (1), labels (1), planets (1), potatoes (1), pots (1), qualifications (1), Republicans (1), trainees (1) people (1)

>3 >2 >1.30103

values (6), rules (5), terms (4), variables (5), coins (3), measurements (3), teeth (3), tools (3), questions (5), numbers (3) drawings (3), antlers (2), attitudes (2, commands (2), interfaces (2), observations (2), panels (2), sources (2), standards (2), activities (2), features (2), instructions (2), policies (2) items (3), leaves (2), relationships (2), actions (2), genes (2) people (7)

>3

analyses (6), concerts (3), tests (4), photographs (3) battles (2), disasters (2), pumps (2), works (6), studies (6), events (3), compounds (2), lectures (2), movements (2), paintings (2) models (3), figures (2), relationships (2), actions (2), governments (2), reports (2), machines (2) people (1)

>3 >2

thefts (3) –​ babies (1), bumps (1), killings (1), magazines (1), mergers (1), attacks (1), incidents (1), loans (1), accidents (1) –​

>3 >2 >1.30103

set

series

rash

repulsion

>2

>1.30103 repulsion

>1.30103 repulsion

repulsion

this regard, the collexeme strength of thefts (i.e. >3) and its co-​occurrence with killings, attacks, incidents and accidents. (99) At home, [a rash of babies] piles up behind the viewing glass of maternity hospitals [COCA: 2006 FIC Bk:TimeOurSinging] The final collocational patterns examined are those of class and pack. The reason for discussing them separately is based on the types of collexemes that they take, as no clear trend could be discerned in either case. The results are presented in Table 4.14.

8 6 1

168  Modelling variation in verb agreement Table 4.14 Most significant collexeme combinations of Ncoll-​of-​Npl subjects that take singular agreement and variable referents collexeme1

collexeme2

coll. strength

class

drugs (14), solutions (6), graders (4), functions (3) models (4), submarines (2), beneficiaries (2), landowners (2) businessmen (2), freshmen (2), students (9), genes (2), objects (2), subjects (2), issues (2), variables (2) people (15)

>3 >2

dogs (7), wolves (6), coyotes (5), cigarettes (5), cards (4), hounds (3) batteries (2), bottles (2) girls (2), destroyers (1), rats (1) –​

>3

pack

>1.30103 repulsion

>2 >1.30103 repulsion

The data show that class co-​occurs with both human (e.g. landowners, businessmen) and inanimate (e.g. drugs, issues) collexemes. This is also the case with pack, which, in contrast, may be used to refer to groups of inanimate entities (e.g. cigarettes, batteries), animals (e.g. dogs, coyotes) and even humans (e.g. girls). The findings reported in this section corroborate the close relation between animacy and verb number agreement. The distinctive collexeme analysis shows that oblique nouns with a stronger association with the plural verbal pattern have mainly human referents, while collexemes of the singular variant show more variation and no clear preferences in terms of animacy. The covarying collexeme analysis reveals that more than half of the 22 collective nouns frequently co-​occur with animate oblique nouns, both human and non-​human: band, bunch, couple, crowd, gang, minority, party, clump, flock, herd, swarm and troop. A smaller set of five collective nouns (batch, host, rash, series and set) was found to collocate more commonly with inanimate referents, while the remaining five nouns (class, group, majority, number and pack) were more flexible in their collocational range and, thus, more freely associated with both animate and inanimate referents. 4.4.3.  Regional variation: British, American and World Englishes Regional variation has been widely attested and accepted as one of the main points of divergence in the agreement patterning of collective nouns (§2.2.4). For example, collective nouns have been observed to occur with plural agreeing forms more frequently in British English than in American English (Quirk et al. 1985: 19; Biber et  al. 1999: 19, 188; Levin 2001: 60–​70; Algeo 2006: 279; Trudgill & Hannah 2008: 72). The results outlined in Section 4.4 confirm this claim, showing a stronger preference for plural verbal forms in British English than in the American variety. I  return to this issue in this section in

9 6 1

Modelling variation in verb agreement 169 an attempt to assess in more detail the impact of regional variation on verb number agreement with complex collective NPs. To do this, I examined the data set from the BNC and COCA from a different perspective and incorporated data from the six inner-​circle varieties of English represented in the Corpus of Global Web-​Based English (GloWbE). Previous corpus-​ based studies on inner-​ circle Englishes have observed diatopic differences in the agreement patterns of collective nouns. Australian and New Zealand English, which are observed not to differ substantially in their agreement preferences (Hundt 2009b), have been described as an in-​ between category with respect to British and American English: although they tend to favour singular agreement more frequently than the British variety, overall they allow much more variation than American English (Bauer 1988; Levin 2001: 60–​70; Collins 2015). Nevertheless, intralinguistic factors seem to have a greater influence: Hundt (2009b), for example, observes that in British, Australian and New Zealand English, agreement varies depending on the collective noun (for further individual patterns in British, American, Australian, New Zealand, Canadian and Irish English, see Vantellini 2003; Sand 2008 and Lindquist 2009). To the best of my knowledge, there are no precedents in the literature for the exploration of verb number agreement variation with complex collective noun phrases only, except for Smith’s (2009) study on the so-​called ‘Non-​Numerical Quantifiers’ (e.g. (a) lot(s) of, (a) bunch(es) of, a host of) in British, Australian and New Zealand English. The research reported in this section aims to bridge that gap. Starting with the trends observed in the BNC and COCA, the by-​variety adjustments to the intercept of the model reported in Section 4.4 signal a stronger preference for plural agreement in the British variety, whereas American English shows the opposite pattern. However, these adjustments only provide information about how the two varieties differ in their baseline preferences for plural agreement with respect to the intercept; the model does not indicate whether British and American English behave differently with respect to the different levels of the fixed effects. For exploratory purposes, I re-​ran the model (§4.3) with ‘regional variety’ as a fixed effect and included interactions for this predictor with the rest of the variables in the fixed-​effects structure. The results showed the differences between the two varieties to be minimal. The only divergence in the odds of plural agreement was found with singular oblique nouns (NN1), inanimate oblique nouns and the oblique noun people. Singular obliques (vs overtly marked plural nouns) were found to decrease the odds of the plural in American English significantly compared to British English (OR: 0.27, 95% Confidence Interval: 0.08–​ 0.91, p=0.035), while inanimate oblique nouns and people increased the odds of plural agreement in the American variety compared to British English by a factor of 1.61 (95% CI: 1.09–​2.38, p=0.016) and 1.91 (95% CI: 1.12–​3.25, p=0.017), respectively, vis-​à-​vis their reference levels (i.e. animate human and overtly marked plural oblique nouns). While the first observation comes as no surprise, the findings for animacy and the type of oblique run counter to assumptions of an overall lower preference for plural agreement in American English.

0 7 1

170  Modelling variation in verb agreement

Figure 4.8 Importance of the variables in the optimal models of the BNC and COCA

To complement these observations, I  fitted separate mixed-​effects regression models for each variety and random forests on their optimal models to determine which predictors were more important for agreement variation in each case (Figure 4.8). The models on which they are based showed a good fit: classification accuracy (C-​index) was 0.921 for the model with data from the BNC (N=2,107) and 0.918 for COCA (N=3,097); predictive accuracy was approximately 85% in both cases (baseline: 63% in the BNC, 57% in COCA); explanatory power (conditional R2) amounted to 0.69 (vs marginal R2 around 0.43) and condition number kappa scored 6.88 for the BNC and 6.12 for COCA. The C-​index for the random forests was an outstanding 0.95 in the British data and 0.94 in the data for the American variety, and predictive accuracy was approximately 85% (baseline: 63% in the BNC, 57% in COCA). Moving on to the ranking of variable importance, Figure 4.8 reveals several differences between the two varieties. First, given its substantially higher importance in the model reported in Section 4.4, it comes as no surprise that the most important predictor of verb agreement variation in both varieties is the type of collective noun (‘N1’). The main differences among the rest of the variables were observed in the upper part of the ranking: oblique noun and type of oblique noun scored higher in the American variety, followed by ‘animacy’, while in British English, the type of determiner of the collective and the lexico-​semantic factors ‘semantic plurality’ and ‘verb base’ followed the collective noun in the ranking, together with type of oblique noun. Determiner, type of verb and semantic plurality ranked slightly lower in COCA, just after ‘animacy’, a variable which was excluded from the optimal model of the BNC as it was found not to be a highly determining predictor. As regards the rest of the predictors, it bears pointing out that the number of postmodifiers and premodifiers (which was found not to be significant in the main model of this

1 7

Modelling variation in verb agreement 171 research) were observed to improve the goodness-​of-​fit of the model only in the case of British English, although the importance of the number of premodifiers is negligible, as shown in Figure 4.8. To expand my analysis of diatopic variation in the patterns of agreement of complex collective subjects, I conducted a corpus-​based study with data from the parsed version of the Corpus of Global Web-​Based English (henceforth, GloWbE).41 The original corpus (Davies 2013) contains almost two billion words obtained from written texts from websites and blogs from 20 different varieties of English.42 The data were retrieved using the dependency syntax query of the Dependency Bank interface of the University of Zurich. The aim of the query was to recall prepositional phrases headed by the preposition of and attached to a noun that functions as subject (i.e. each of the 23 collective nouns). The tag ‘NN’ was specified to avoid tagging errors of the collective noun. Up to 3,000 hits per collective noun were considered. Given their more frequent occurrence in complex collective NPs, for this investigation I decided to consider only regular and irregular plural oblique nouns and the formally unmarked plural noun people. The parser (precision approx. 90%, recall approx. 85% with subjects; Lehmann & Schneider 2009, 2012) identified over 32,000 examples of complex collective subjects.43 After manual pruning (using the same criteria as in §4.1.2), the number of instances in the data set amounted to 8,742. In preliminary investigations, Ncoll-​of-​N subjects were observed to favour plural verbal forms in more than 55% of cases in nine of the 20 varieties of English in GloWbE: Jamaican, Nigerian, Irish, Tanzanian, British, Sri Lankan, Kenyan, Pakistani and Ghanaian English. A set of ten varieties were found to pattern quite evenly in the singular and the plural (46%–​54%): Canadian, Australian, South African, New Zealand, Singapore, Malaysian, Indian, American, Bangladeshi and Philippine English. Only Hong Kong English favoured singular agreement in more than 55% of cases (Fernández-​ Pena 2017a). The diatopic diversity of GloWbE made it necessary to restrict the scope of the study to avoid any potential interference from the substrate languages that are in contact with the varieties of English from the outer circle. For this reason, the research reported in the remainder of this section is based on data from only the six ‘native’ (Kachru 1985) inner-​circle varieties: Irish, British, Canadian, Australian, New Zealand and American English. The new restricted data set comprises 5,478 instances and 21 collective nouns (shoal was not attested and clump was excluded because of its categorical singular agreement: >95%; see Section 4.1.2). Replicating for the most part the procedure described in Section 4.3, a mixed-​effects regression analysis and a random forest with the variables of the optimal regression model were fitted.44 The summary statistics confirmed the good fit of the mixed-​effects regression analysis: the C-​index indicated excellent discrimination (0.869), with a predictive accuracy of 80.47% (baseline: 54%). Collinearity was not problematic (condition number kappa = 8.72) and the variance accounted for by the whole model was only 0.45 (conditional

2 7 1

172  Modelling variation in verb agreement

Figure 4.9 Intercept adjustments for the random effect ‘variety’ in GloWbE

R2 vs marginal R2: 0.11). The predictive accuracy of the random forest was 83.66% (baseline: 54%) and the classification accuracy indicated outstanding discrimination power: C-​index  0.921. Starting with the results of the agreement preferences of the six inner-​ circle varieties, the intercept adjustments for ‘variety’ in the regression model (Figure 4.9) reveal notable variation in the baseline agreement preferences of the different varieties. Only the varieties of English spoken in Ireland (IrE) and Great Britain (BrE) display a strong association with plural verb agreement (positive adjustments), which was notably more marked in the case of Irish English. The negative intercept adjustments in the other four Englishes indicate a preference for singular patterns. Among these, Canadian English (CanE) and New Zealand English (NZE) show very low adjustments, while the Australian (AusE) and, particularly, the American (AmE) varieties more clearly disfavour plural verbal patterns with complex collective subjects. These results are largely consistent with prior literature. First, they corroborate the widely discussed contrast between American English, where collective nouns are observed to pattern mainly in the singular, and British English, where number variation is more common (Levin 2001: 60–​70; Algeo 2006: 279; Trudgill & Hannah 2008: 72). They confirm the findings for the BNC and COCA (§4.4) and demonstrate that these diatopic differences also apply to complex collective subjects. In the case of GloWbE, however, a greater difference between the two varieties was predicted: the higher degree of informality of the corpus was expected to result in greater variation in verb number and a higher incidence of plural agreement, particularly in the variety which is more accepting of variation, British English.

3 7 1

Modelling variation in verb agreement 173 The intercept adjustments were also consistent with previous research in that both Australian and New Zealand English were shown to represent an in-​ between category with a baseline preference for the singular, albeit not as strongly as in the American variety (Bauer 1988; Levin 2001; Hundt 2009b; Collins 2015). My research has only found a couple of studies on Irish and Canadian English (Sand 2008; Lindquist 2009); however, the limited data set of collective nouns in each case precludes any conclusions in relation to either. As a final remark on the patterns of the six inner-​circle varieties, it is worth noting again the potential influence of genre and register on these results. The data from the more formal written British and American corpora and the results from the more informal setting in GloWbE both point to a contrast between baseline preferences for number agreement in British English and American English. It remains to be seen whether data from more formal corpora for each of the other four varieties support these observations, particularly in the case of Irish and Canadian English. The determining factors of verb number agreement variation as attested by the optimal regression analysis are displayed in Figure  4.10, ranked in decreasing order of importance. For the inner-​circle Englishes as a whole, the collective noun (‘N1’) was once again shown to outstrip the other variables as the most important predictor of verb number variation. In contrast to the data from the BNC and COCA, however, Figure 4.10 shows that in GloWbE the most important determinants of agreement variation are all lexico-​semantic factors: ‘N1’, ‘partition’, ‘verb base’ and ‘animacy’. The determining effect of the collective noun and ‘partition’ (i.e. its syntactic configuration) is yet another indication of the quantifying potential of complex collective NPs that I have argued throughout (Chapter 3; §4.4.2.4). The rest of the predictors scored very low in importance. The morphosyntactic variables that measure the complexity of the NP, both quantitatively and qualitatively, showed a marginal importance in comparison with the lexico-​ semantic factors. The lowest position in the ranking is occupied by ‘type of N2’, which indicates that the three types of oblique nouns considered (regular, irregular and people) had little influence on verb number choice. Finally, the low importance value of ‘(regional) variety’ suggests that, despite the baseline preferences observed above, there may not be great differences among the six inner-​circle Englishes as to which factors constrain agreement variation with complex collective subjects. Mindful, however, of the contrasting results obtained from the model based on data from the BNC and COCA (§4.4) and the mixed-​effect regression model and random forests discussed at the beginning of this section, I decided to run random forests for each inner-​circle variety separately, considering in this case not the variables of the optimal model but the whole list of (17 random and fixed) variables surveyed in the study of GloWbE.45 The classification accuracy (C-​index) of the forests was greater than 0.95 in all cases, with a predictive accuracy range of 81%–​92% (baselines: approx. 52%–​64%).46 The results are displayed in Figure 4.11. The forests did not diverge significantly in the order of the variable importance values. The collective noun (‘N1’) was the most important predictor

4 7 1

174  Modelling variation in verb agreement

Figure 4.10 Importance of the variables in the optimal model of GloWbE

in all six Englishes, followed by ‘partition’, as in the random forest of the optimal model (Figure 4.10), and ‘Det1’, which is inextricably linked to partition and was excluded from the model owing to their collinearity. The fact that these three predictors are determinants of verb number agreement in the six varieties highlights the essential role of individual agreement patterns, the constraints and fixation involved in certain syntactic configurations of the NP (i.e. pseudopartitive) and the potential quantifying use of complex collective noun phrases. A second set of mainly lexico-​semantic predictors also proved to be of relative importance: ‘animacy’, ‘verb base’ and, to a lesser extent, ‘N2’. However, the values for these factors were more marginal on the whole compared to the predictors discussed above. In fact, high scores were only observed in two varieties, Irish and British English, which were also the only ones with a wider range of important variables. The rest of the factors considered in the random

5 7 1

Modelling variation in verb agreement 175

Figure 4.11 By-​variety importance ranking of all the variables examined in GloWbE

6 7 1

176  Modelling variation in verb agreement forests were of little or no importance in the varieties of English from Canada, New Zealand, Australia and America. The only exception was the ‘premodifier of the Ncoll’, which was found to be as important as ‘N2’ in all of the Englishes except American and which is also involved in the fixation and quantifying use of complex collective NPs (Chapter 3). Finally, it is worth noting that, in contrast to the general trends for all six varieties observed in Figure 4.10, the complexity of the collective subject in British and particularly Irish English appears to play a role in agreement variation. The by-​variety random forests confirm that the importance of the predictors of agreement variation with complex collective subjects is similar in the six inner-​circle varieties. Except for the three most important factors, the collective noun (‘N1’), ‘partition’ and ‘Det1’, where there was almost no variation, the rest of the predictors showed similarly low values, which precluded further discussion. This study does corroborate, however, that regional variation is not the most important factor for agreement variation with complex collective subjects. In line with Hundt’s (2009b) findings in relation to prototypical collective nouns, this analysis of diatopic variation points to intralinguistic factors as more decisive predictors of the verb number patterns of complex collective subjects.

4.5.  Summary and final remarks The aim of the study reported in this chapter was to identify the morphosyntactic and lexico-​semantic factors that determine verb agreement variation with complex collective subjects in Present-​Day English, focusing in particular on the role of the of-​dependent. A generalised linear mixed-​effects regression analysis was run with data from the BNC and COCA with the aim of answering the third, fourth and fifth research questions of the study, which relate to the extent to which the morphosyntactic and lexico-​semantic characteristics of the of-​PP and the elements involved in the agreement relation affect verb number choice. To complement the research and offer a wider picture of the situation, the lexical and diatopic dimensions were also examined through analysis of collocational preferences and restrictions using collostructional analysis, on the one hand, and regression analyses of the agreement patterns of complex collective subjects in six inner-​circle Englishes, on the other. This multifaceted approach confirmed the impact on the agreement operation of both lexico-​semantic factors (especially the collective noun) and morphosyntactic variables. The results from the statistical model of the data from the BNC and COCA demonstrated that the patterns of agreement of complex collective subjects are conditioned by the type of determiner, countability, animacy, semantic plurality and morphological number of the oblique, as well as by the complexity of the of-​ PP. Complex collective subjects were found to conform to the Animacy Hierarchy, as non-human referents, which are less readily conceived of as aggregates of individuals, are significantly less likely to opt for plural verbal forms in comparison with ​human referents. Semantically singular and uncountable oblique nouns also

71

Modelling variation in verb agreement 177 showed a very low likelihood of plural agreement, while semantically plural and countable N2s allowed for greater variation. As regards the type of N2, the data confirmed the so-​called ‘markedness effect’ as singular oblique nouns were shown to be significantly less likely to occur with plural verbal forms in comparison with regular plural nouns in syntactically simple contexts. The finding that irregular plural nouns favour more plural agreement than regular obliques contradicted previous studies. I  attribute this to the contrastiveness of the nouns’ morphological plurality. Unmarked morphology, in contrast, was not associated in any way with significant changes in the odds of plural agreement, which confirms the negligible impact of morphological unmarkedness on agreement variation. In fact, morphologically unmarked nouns (NN0), including the semantically plural noun people, showed a significant decrease in the likelihood of plural agreement with the increasing syntactic complexity of the noun phrase. These observations, which go against previous research on collective nouns and do not conform to the Agreement Hierarchy, are explained here in terms of the ‘Complexity Principle’: the fact of a noun’s morphological number not being overtly marked (i.e. less explicit) results in a higher probability of its taking a more explicit singular verb number in syntactically more complex contexts. In terms of complexity, only syntactic complexity in terms of number of postmodifiers was shown to have a higher impact on agreement variation. Structural complexity in number of words was not found to be a comparable proxy of NP complexity: not only were the results less conclusive, but they contradicted both previous findings and the effect of syntactic complexity. This means that only syntactic complexity was shown to be highly determinant of verb number agreement with complex collective subjects. In fact, most of the predictors that measured NP complexity (number and length of the premodifiers, clause depth of the NP and number of morphologically (un)marked nouns in the postmodifier) were discarded from the model for not improving its goodness-​of-​fit. The exploration of the type of postmodifier and its noun/​clause-​like status using a separate model revealed some influence by clausal postmodification on the reduction of the odds of plural agreement. The results coincide with previous studies, although the scarcity of data in relation to more complex domains makes it difficult to draw any categorical conclusions other than the stronger determining influence of syntactic complexity. The last morphosyntactic factor relates to the determiner of the collective. The data confirmed the observations and claims made in Chapter 3, where the definite article was found to reduce the odds of plural agreement significantly compared to the indefinite article, which is, in turn, the only determiner that facilitates a quantifying use of complex collective noun phrases. With the mixed-​effects model, I also explored by-​item and by-​variety variability in the data through random factors. The regression analysis confirmed that most of the variance in the data is accounted for by the collective noun, followed by the oblique noun and the verb. A series of collostructional analyses were used to examine these lexical factors more closely, producing further evidence of the collocational and colligational restrictions highlighted in

8 7 1

178  Modelling variation in verb agreement Chapter 3. The most important findings concerned the interaction between the animacy of the referent and plural verb number, and the strong association of bunch, couple, host, majority and minority with the plural, as further evidence of their quantifying potential. By-​variety variability confirmed the predictably lower likelihood of plural agreement with collective NPs in American English. In an attempt to provide a wider picture of the diatopic variation of verb number agreement, further regression analyses were carried out on the data from the BNC and COCA and on the six inner-​circle varieties of English in GloWbE. Some divergence in the importance of the predictors was found between the British and the American varieties, although, overall, the differences were not shown to alter the effect of the variables on modelling agreement variation significantly. As for the other World Englishes, it was found out that, with minor differences, agreement is constrained to a greater extent by lexico-​ semantic factors, confirming that intralinguistic variables play a stronger determining role than regional variation. The main contribution of this analysis lies in its investigation of an under-​ researched area where comprehensive descriptions of the topic from a multidimensional perspective, not to mention statistical analyses, are scarce (Dekeyser 1975; Levin 2001; Smith 2009). The study offers a multifaceted, robustly statistical analysis of the morphosyntactic, lexico-​semantic and diatopic dimensions of the phenomenon of verb number agreement with complex collective NPs. The conclusions and insights gained will, I hope, serve to broaden and inform our understanding of the topic, and open up new pathways for research in the future.

Notes Access to the interface at www.natcorp.ox.ac.uk. 1 2 Access to the interface at www.english-​corpora.org/​bnc. 3 Further details of the spoken and written components at www.natcorp.ox.ac.uk/​ docs/​URG.xml?ID=BNCdes. 4 Information retrieved from www.english-​corpora.org/​coca. 5 Further information at http://​corpus.byu.edu/​coca/​help/​spoken.asp. 6 Access to the interface at www.english-​corpora.org/​coca. 7 Further information on the BNC at www.natcorp.ox.ac.uk/​docs/​URG/​BNCdes. html#spodes and on COCA at http://​corpus.byu.edu/​coca/​help/​spoken.asp. 8 A preliminary study using data from the spoken components of the BNC and COCA revealed that, of the 23 collective nouns analysed, only majority and number (and, to a lesser extent, couple and group) showed high token frequencies in subject position in spoken English; the other 19 collective nouns showed very low token frequencies (mostly 5, which amounted to 58% of the data retrieved). The study showed that, in British English, verbs with

1 8

Modelling variation in verb agreement 181 a frequency of occurrence in the plural of 60% or higher belonged to semantic types closely associated with human subjects or referents: ‘possession’ (get, give), ‘send/​carry’, ‘existence’ (live, gather), ‘communication’ (say, ask). In contrast, in verbs that were less frequently found in the plural (i.e. 3 = p2 = p1.30103 = p3 = p2 = p1.30103 = p1.30103) were included in the same category. 39 Hands was coded as ‘human’ because the referent of the whole NP is human: A bunch of the new hands were sitting together in the yard commiserating when Jackie Tiptoe limped up. [BNC: 1985–​1993 AEB 1869]. 40 As in the case of couple, host and bunch above, instances where collexeme2 occurs only once in the database are not reported here, even if their collexeme strength is statistically significant. 41 The parsed version of GloWbE may be accessed through the Dependency Bank interface at the University of Zurich. Further information at www.es.uzh.ch/​en/​ Subsites/​Projects/​dbank.html. 42 Fernández-​Pena (2017a) surveys corpora of a number of different varieties from the International Corpus of English (ICE) family (see http://​ice-​corpora.net/​ice), including Canada, New Zealand, Ireland, Great Britain, India, Hong Kong, Jamaica, Philippines and Singapore. However, none of the varieties surveyed provides enough results to support a robust quantitative and contrastive analysis. 43 For further information on the Pro3Gres Parser, see https://​files.ifi.uzh.ch/​cl/​ gschneid/​parser. 44 Given that only plural oblique nouns were considered in this study, the variables ‘formal variation’, ‘countability’ and ‘semantic plurality’ were not included in the model. The presence of plural and formally unmarked nouns in the postmodifier and the embeddedness of the subject were not examined either. 45 Total number of instances of each variety: 1,697 American English, 461 Australian English, 1,819 British English, 604 Canadian English, 487 Irish English and 410 New Zealand English. 46 Classification (C-​index) and predictive accuracy per variety: 0.971 and 89.16% (baseline: 51.62%) in American English, 0.969 and 87.85% (baseline: 52.28%) in Australian English, 0.952 and 85.10% (baseline: 57.78%) in British English, 0.971 and 92.05% (baseline: 52.48%) in Canadian English, 0.964 and 81.93% (baseline: 64.27%) in Irish English, and 0.966 and 89.51% (baseline: 51.95%) in New Zealand English.

2 8 1

5  Concluding remarks and prospects for future research

This book is the culmination of a comprehensive multifaceted corpus-​based investigation into verb number agreement with complex collective noun phrases in English. The aim of the research was to model verb number agreement variation with a set of relational collective nouns that take of-​dependents (e.g. bunch, set, majority, herd). The study used a multidimensional analysis of the morphosyntactic and lexico-​semantic factors that intervene in and influence the choice of number in the main verb, focusing in particular on the morphosyntactic and lexico-​semantic characterisation of the so-​called ‘oblique noun’ (e.g. a bunch of flowers, a shoal of fish). Two case studies were carried out, the first of which explored the diachronic dimension of the phenomenon from the perspective of grammaticalisation, using data retrieved from the Corpus of Historical American English (COHA, 1810–​2009). The second analysis was based on data extracted from written samples of Present-​Day English from the British National Corpus (BNC, 1960–​1993) and the Corpus of Contemporary American English (COCA, 1990–​2012). This combined synchronic-​diachronic approach not only served to expand the data set and the analysis but also gave the research a historical perspective, leading to more robust, generalisable results. The research was addressed from a descriptive, multifaceted perspective, using a Corpus Linguistics methodology but no specific theoretical framework. The main purpose of this approach was to formulate a usage-​based account of the phenomenon, based primarily on the results of the multidimensional analysis of the diachronic and synchronic morphosyntactic and lexico-​semantic determinants surveyed. To my knowledge, no previous research has provided such a large-​scale diachronic and synchronic corpus-​based study, representing the intersection of four levels of analysis and testing a comparable extensive set of variables with regression and collostructional techniques. The comprehensiveness of the study’s scope and approach, coupled with the scarcity of prior research on the of-​dependent (see Dekeyser 1975: 35–​66; Reid 1991: 267–​ 272; Smith 2009; Klockmann 2017: 211–​273; Zhang 2017: 53–​77; Leclercq & Depraetere 2018), makes this a valuable contribution to the field and will, I hope, serve to open up new horizons of inquiry. The research reported in the book was driven by the research questions presented in Chapter 1. These questions addressed issues such as which of the

3 8 1

Concluding remarks and future research 183 elements in the complex collective subject determines verb number, to what extent the formal and/​or semantic characterisation of the of-​dependent has a determining influence on agreement and the interaction between verb number selection and lexico-​semantic factors such as the type of collective noun and/​or its quantifying potential. The contribution of this research lies in the numerous conclusions drawn from the two corpus-​based studies carried out, the most important of which are outlined below. The diachronic investigation in Chapter  3 explored the role played by lexical factors in verb number agreement variation with complex collective subjects in order to gauge the potential idiomatic quantifying use of these subjects and determine the influence of that use on their present-​day patterning. The results showed positive signs of syntactic fixation in terms of restricted premodification patterns, indefinite article selection and plural verb number agreement for a number of, a bunch of and a couple of and, to a lesser extent, for a host of and a group of. Because of the different meaning they convey, a majority of and a minority of were concluded not to have a highly idiomatic status. As concerns quantification, all seven of the Ncoll-​of-​Npl structures scrutinised were shown to have developed, to varying extents, relative and absolute quantifying readings. Relative quantification (i.e. denotation of quantity with respect to a reference mass) was expressed either structurally, with the partitive configuration, or lexically, as in the case of majority and minority. Absolute quantification, in contrast, emerged in the highly idiomatic and syntactically more fixed pseudopartitive configuration (a Ncoll of Npl), in keeping with findings reported elsewhere. In this type of structure, collective nouns serve as periphrastic absolute quantifiers (i.e. comparable to several or many) that denote indefinite quantification of the entities referred to by Npl, which in turn assumes the role of head of the NP and controller of agreement. The emergence of their absolute quantifying use was interpreted as a response to the general need for new expressive means of conveying quantification, also in line with the literature. The functionality and productivity of these complex collective subjects was attributed to the persistence of certain nuances of their original collective meaning, which represents an added value to their quantifying uses that enriches the paradigm of indefinite quantifying expressions in English. The statistical model used for the synchronic study in Chapter  4 found verb number agreement variation with complex collective noun phrases to be determined by both lexico-​semantic and morphosyntactic factors. Variation was constrained when the collective took a determiner other than the indefinite article or an uncountable (vs countable) and/​or a semantically singular (vs semantically plural) oblique noun, and when it referred to a non-​human (vs human) referent. The interplay between the formal (un)markedness of the oblique noun and the complexity of the noun phrase yielded results consistent with the markedness effect and the Complexity Principle. The former was attested in the significantly lower preference for plural agreement with singular oblique nouns in syntactically simpler contexts in comparison to regular plural nouns and, particularly, irregular plural obliques, which showed a significantly

4 8 1

184  Concluding remarks and future research higher likelihood of plural agreement owing to the contrastiveness of their marked morphology. The Complexity Principle was supported by the finding that morphologically unmarked nouns (NN0 and people) were significantly less likely than regular plural nouns to pattern with the plural in the most complex subjects. Their formal unmarkedness was adduced as the main explanation for this: in the absence of a morphological marker for number highlighting the inherent plurality of the collective NP, the most complex domains constrain the use of the plural in favour of the more explicit singular verb number, which in turn facilitates the processing of the sentence. Chapter 4 also examined lexically specific and variety-​specific variability. In the case of the former, both the regression model and the collostructional analyses indicated the important role of lexical factors, particularly the collective noun, on verb number agreement. The collocational preferences of the collective noun confirmed a strong association with plural agreement and the consequent potential for quantification of all of the collective nouns surveyed in Chapter 3, with the exception of number and group, whose still frequent use as lexical nouns resulted in a greater attraction to the singular pattern. As regards the oblique noun and the verb, the data revealed significant interaction between verb number, animacy and lexical choice. In fact, plural agreement was shown to be more strongly associated with animate, particularly human, oblique nouns and verbs that require or are more compatible with animate (human) subjects. Finally, as regards regional variation, my research corroborated the general consensus in the literature about the lower likelihood of plural agreement with collective nouns in American English compared to the British variety. Chapter 4 provided further statistical evidence of regional variation based on data from the six inner-​circle varieties of English in the Corpus of Global Web-​Based English (GloWbE), confirming that, despite differences between the varieties in terms of their baseline verb number preferences, agreement variation with complex collective subjects is more strongly predicted by intralinguistic factors, especially the type of collective noun. The conclusions summarised here highlight the core contribution of this research in revealing and confirming the interplay between morphology, syntax, semantics and lexis in the agreement patterns of complex collective NPs. The multifaceted approach ensured a comprehensive analysis and assessment of the subject that will expand the literature in this area. Before concluding this volume, let us recap briefly on the answers obtained for each of the research questions: (i) Is there evidence of a diachronic evolution? Have there been any significant changes in relation to complex collective subjects or collective nouns more generally which may have influenced their current verbal patterning and meaning? (ii) What is the quantifying potential (if any) of complex collective subjects? To what extent does the interaction between the of-​PP and verb agreement contribute to this use?

5 8 1

Concluding remarks and future research 185 The evidence from the diachronic study in Chapter 3 corroborated the impact of diachrony by showing that the changes (or lack of change) observed in relation to each of the seven collective nouns analysed over the two-​century period have had a determining effect on their quantifying potential. The study was specifically designed to identify any significant variation over time that might explain their agreement patterns in PDE. The results revealed that all seven collective nouns may be used with a quantificational reading, but that the frequency and idiomaticity of that use differs substantially. Plural agreement was strongly associated with co-​occurrence with a plural oblique noun in the of-​PP since the plurality of the oblique noun more readily promotes a distributive reading and, with it, a plural verbal pattern and a quantifying reading of the fragment a Ncoll of. Unlike previous findings on the conventionalisation of the number of the collective noun, I contend here that what is conventionalised is not the collective noun but the whole a Ncoll of sequence, which contributes to widening the range of absolute quantifiers and confirms the productivity of this diachronic source of periphrastic quantification in English. (iii) Is lexis a determining factor? Is verb number agreement affected by the type of verb, type of collective noun or type of noun in the of-​PP? The positive answer to this question is implicit in the results for the previous two. It is important to point out, however, that, in addition to the type of collective noun, the type of oblique noun and verb were also shown to have a significant effect on the complex collective subject and its verbal patterning. The study in Chapter  4 revealed significant animacy-​ related collocational preferences with the oblique noun and the verb, which help to explain the main trends with respect to verb number agreement. The strong association of particular collective nouns with particular oblique nouns and plural verbal agreement led to two important conclusions: first, that the collective nouns examined do not constitute a homogeneous group, as agreement patterns and variation are strongly dependent on the type of collective noun; and, second, that complex collective NPs may acquire and express quantifying meaning structurally in both partitive and pseudopartitive configurations, with only the latter leading to the idiomatic absolute quantifying use that is susceptible to grammaticalisation. (iv) What determines verb number choice in the case of complex collective subjects: the collective noun, the prepositional phrase or the structure as a whole? My research showed that there is no univocal answer to this question. Although the collective noun plays an essential role in agreement variation, the data from the synchronic study revealed that other aspects, such as complexity of the noun phrase, type of determiner, countability, morphological and semantic number of the oblique noun, animacy of the referent and type of verb, have a significant constraining effect on the agreement patterns of the complex

6 8 1

186  Concluding remarks and future research collective subjects. The results indicate that verb number is determined by the combined effect of the simultaneous interaction of formal, semantic and lexical factors, except where the degree of idiomatisation of the collective noun constrains variation. This conclusion demonstrates and justifies the robustness, necessity and suitability of the multifaceted approach adopted here. (v) To what extent (if at all) do the form and/​or the semantics of the of-​PP and/​or the other elements in the subject affect the use of singular or plural verb number? In formal terms, one crucial factor of agreement variation was the interaction between the degree of formal markedness of the oblique noun and the complexity of the of-​PP, while in lexico-​semantic terms, verb number was shown to be affected by the type of collective noun and the animacy of the referent, and the interaction between them. Additional factors included the dependents of the oblique noun, the countability of the latter and the verb. The findings obtained from this research thus corroborated my theory regarding the determining role of this prepositional dependent and the need to explore its effect on verb agreement. By way of conclusion, I would like to propose a few directions for future research. In the first place, a great source of data for continued exploration of the interplay of formal complexity, semantics and agreement are the agreement shifts observed within clauses depending on their target (i.e. verb or pronoun) and across clauses and sentences (i.e. co-​referential pronouns) (e.g. [This group of fans] does not see themselves as being disloyal to the Whalers; they see the team’s owner as being disloyal to them; COCA: 2007 ACAD SportBehavior). The analysis of structural and syntactic complexity would also benefit from further scrutiny of non-​overt or hidden complexity in contexts where the lower degree of explicitness of, for example, non-​clausal dependents, translates into higher cognitive complexity (e.g. [the number of people [ill with AIDS]] vs [the number of people [who are mentally ill]]). To obtain further insights into the idiomatic and even grammaticalised use of some complex collective NPs, future research could explore complex collective phrases in syntactic environments other than subject position (1) and with other agreement targets, such as determiners (2). (1) I think the power look is [a bunch of nonsense] [COCA: 1993 MAG Essence] (2) [Those couple of minutes] were mercifully short [COCA: 2009 ACAD Humanist] Another aspect of the topic that would reward further research is that of complex collective subjects in contexts where trade-​offs between formal and semantic agreement are common, such as existential there-​constructions (particularly involving there’s) (3)  and cleft constructions (4), as well as in contexts in which the potential influence of the number of the oblique noun is

7 8 1

Concluding remarks and future research 187 cancelled by the inversion of the canonical word order (5) or by the ellipsis of the of-​dependent  (6). (3) There’s [a large number of things], and most of them in museums [COCA: 2002 SPOK NPR_​ATCW] (4) what you have is [a bunch of young Republicans who were not there long enough] [COCA: 1992 SPOK ABC_​Brinkley] (5) there in his fist was [a bunch of wet, fleshy flowers] [BNC: 1985–​1994 HTN W_​fict_​prose] (6) Few women in Ash Creek work for wages; [a number] sell home products [COCA: 1992 ACAD AnthropolQ] Research into these questions could also benefit greatly from psycholinguistic testing, to complement the results obtained in this study with experimental data from acceptability judgement and elicitation tasks with native speakers of English. More generally, the potential of regional variation in British, American and other World Englishes remains to be explored in more depth and detail. Finally, it is worth recalling that this investigation was deliberately based on written material, since the lack of comparable samples of spoken data precluded the exploration of agreement in this setting. Complex collective NPs are not particularly frequent in spoken data, as noun phrases tend to be more complex in written registers (Schäpers 2009: 101–​119; Berlage 2014: 32; Schaub 2016: 262) and the use of written sources has the advantage of identifying more stable trends (Bauer 1994: 61; Tristram 2014: 354). Nevertheless, the exploration of spoken material could reveal incipient patterns of usage that have not yet permeated more formal registers and which could therefore offer further support for the prospectively emerging quantifying use of some of the collective NPs examined here. Spoken data, and in general text type variation, could constitute the perfect complement to the insights gained from written registers, which would help to expand the literature on this phenomenon and indicate potential applications in areas of Linguistics such as Second Language Acquisition. Despite the inevitable limitations of scope in this research, its exploration of the significance of the collective noun, the role played by lexical collocational preferences and restrictions and the formal (un)markedness of the oblique noun and its interaction with the (syntactic) complexity of the of-​PP, demonstrating the continuous interplay between morphosyntactic and lexico-​ semantics factors in the resolution of verb number agreement with complex collective subjects, represents an important contribution to this field of study.

81

References

A.  Primary sources ARCHER 3.2 = A Representative Corpus of Historical English Registers version 3.2. 1990–​1993/​2002/​2007/​2010/​2013/​2016. Originally compiled under the supervision of Douglas Biber & Edward Finegan at Northern Arizona University and University of Southern California; modified and expanded by subsequent members of a consortium of universities. Current member universities are Bamberg, Freiburg, Heidelberg, Helsinki, Lancaster, Leicester, Manchester, Michigan, Northern Arizona, Santiago de Compostela, Southern California, Trier, Uppsala, Zurich. Examples of usage taken from ARCHER were obtained under the terms of the ARCHER User Agreement. Davies, Mark. 2004–​. British National Corpus (from Oxford University Press). https://​ www.english-​corpora.org/​bnc Davies, Mark. 2007–​ . TIME Magazine Corpus: 100  million words, 1920s–​ 2000s. https://​www.english-​corpora.org/​time Davies, Mark. 2008–​. The Corpus of Contemporary American English (COCA): One billion words, 1990–​2019. https://​www.english-​corpora.org/​coca Davies, Mark. 2010–​. The Corpus of Historical American English (COHA): 400 million words, 1810–​2009. https://​www.english-​corpora.org/​coha Davies, Mark. 2011–​. Corpus of American Soap Operas: 100  million words. https://​ www.english-​corpora.org/​soap Davies, Mark. 2011–​. Google Books Corpus. (Based on Google Books n-​grams). https://​ www.english-​corpora.org/​googlebooks. Based on: Jean-​Baptiste Michel, Yuan Kui Shen, Aviva Presser Aiden, Adrian Veres, Matthew K. Gray, The Google Books Team, Joseph P.  Pickett, Dale Hoiberg, Dan Clancy, Peter Norvig, Jon Orwant, Steven Pinker, Martin A. Nowak & Erez Lieberman Aiden. 2011. Quantitative analysis of culture using millions of digitized books. Science 331(6014). 176–​182. Davies, Mark. 2013. Corpus of Global Web-​Based English: 1.9 billion words from speakers in 20 countries (GloWbE). https://​www.english-​corpora.org/​glowbe Davies, Mark. 2015. Hansard Corpus. Part of the SAMUELS project. https://​www. english-​corpora.org/​hansard Davies, Mark. 2015. The Wikipedia Corpus: 4.6  million articles, 1.9 billion words. Adapted from Wikipedia. https://​www.english-​corpora.org/​wiki Davies, Mark. 2016–​. Corpus of News on the Web (NOW): 10 billion words from 20 countries, updated every day. https://​www.english-​corpora.org/​now Davies, Mark. 2016–​. Corpus of Online Registers of English (CORE). https://​www. english-​corpora.org/​core

9 8 1

References 189 De Smet, Hendrik, Hans-​Jürgen Diller & Jukka Tyrkkö. 2013. The Corpus of Late Modern English Texts, version 3.0. https://​perswww.kuleuven.be/​~u0044428. Leuven: KU Leuven. Höglund, Mikko & Kaj Syrjänen. 2016. Corpus of Early American Literature. ICAME Journal 40(1). 17–​38. Kroch, Anthony, Beatrice Santorini & Ariel Diertani. 2004. The Penn-​Helsinki Parsed Corpus of Early Modern English (PPCEME). Department of Linguistics, University of Pennsylvania. http://​www.ling.upenn.edu/​hist-​corpora/​PPCEME-​RELEASE-​2/​ index.html Kroch, Anthony, Beatrice Santorini & Ariel Diertani. 2010. The Penn Parsed Corpus of Modern British English (PPCMBE). Department of Linguistics, University of Pennsylvania. http://​www.ling.upenn.edu/​hist-​corpora/​PPCMBE-​RELEASE-​1/​index. html Love, Robbie, Claire Dembry, Andrew Hardie, Vaclav Brezina & Tony McEnery. 2017. The Spoken BNC2014: Designing and building a spoken corpus of everyday conversations. International Journal of Corpus Linguistics 22(3). 319–​344. Oxford English Dictionary, 3rd edn. online. https://​www.oed.com (17 May, 2020.) The British National Corpus, version 3 (BNC XML Edition). 2007. Distributed by Oxford University Computing Services on behalf of the BNC Consortium. 100 million words. 1960s–​1993. http://​bncweb.lancs.ac.uk The Strathy Corpus of Canadian English (STRATHY): 50 million words, 1970–​2010. https://​www.english-​corpora.org/​can

B.  Secondary sources Aarts, Bas. 1998. Binominal noun phrases in English. Transactions of the Philological Society 96(1). 117–​158. Acuña-​Fariña, Juan Carlos. 2009. The linguistics and psycholinguistics of agreement: A tutorial overview. Lingua 119(3). 389–​424. Acuña-​Fariña, Carlos. 2016a. (A few) psycholinguistic properties of the NP. Functions of Language 23(1). 120–​141. Acuña-​ Fariña, J. Carlos. 2016b. Opportunistic processing of language. Language Sciences 57. 34–​48. Akimoto, Minoji. 2002. Two types of passivization of ‘V+NP+P’ constructions in relation to idiomatization. In Teresa Fanego, Javier Pérez-​Guerra & María José López-​ Couso (eds.), English historical syntax and morphology: Selected papers from the 11 ICEHL, Santiago de Compostela, 7–​11 September 2000, vol. 1, 9–​22. Amsterdam & Philadelphia: John Benjamins. Akmajian, Adrian & Adrienne Lehrer. 1976. NP-​like quantifiers and the problem of determining the head of an NP. Linguistic Analysis 2(4). 395–​413. Alexiadou, Artemis, Liliane Haegeman & Melita Stavrou. 2007. Noun phrase in the Generative perspective. Berlin: Mouton de Gruyter. Algeo, John. 1988. British and American grammatical differences. International Journal of Lexicography 1(1). 1–​31. Algeo, John. 2006. British or American English? A handbook of word and grammar patterns. Cambridge: Cambridge University Press. Baayen, R. H. 2008. Analyzing linguistic data: A practical introduction to statistics using R. Cambridge: Cambridge University Press.

0 9 1

190 References Bailey, Charles-​James N. 1987. Marginalia on singulars and plurals in English. Arbeiten aus Anglistik und Amerikanistik 12(1). 3–​11. Baker, Mark C. 2013. Agreement and case. In Marcel den Dikken (ed.), The Cambridge handbook of Generative syntax, 607–​654. Cambridge: Cambridge University Press. Barker, Jason & Janet Nicol. 2000. Word frequency effects on the processing of subject-​ verb number agreement. Journal of Psycholinguistic Research 29(1). 99–​106. Barlow, Michael. 1988. Unification and agreement. Stanford, CA: CSLI Publications. Barlow, Michael. 1992. A situated theory of agreement. New York: Garland. Barlow, Michael. 1999. Agreement as a discourse phenomenon. Folia Linguistica 33(2). 187–​210. Bartoń, Kamil. 2019. MuMIn: Multi-​ Model Inference. R package version 1.43.6. https://​CRAN.R-​project.org/​package=MuMIn. Bates, Douglas, Martin Mächler, Ben Bolker & Steve Walker. 2015. Fitting linear mixed-​ effects models using lme4. Journal of Statistical Software 67(1). 1–​48. Battistella, Edwin L. 1990. Markedness: The evaluative superstructure of language. Albany, NY: State University of New York Press. Bauer, Laurie. 1988. Number agreement with collective nouns in New Zealand English. Australian Journal of Linguistics 8(2). 247–​259. Bauer, Laurie. 1994. Watching English change: An introduction to the study of linguistic change in standard Englishes in the twentieth century. New York: Longman. Bauer, Laurie. 2002. An introduction to international varieties of English. Edinburgh: Edinburgh University Press. Bayly, Anselm. 1772. A plain and complete grammar of the English language; to which is prefixed the English accedence: With remarks and observations on a short introduction to English grammar. London. Belletti, Adriana. 2001. Agreement projections. In Mark Baltin & Chris Collins (eds.), The handbook of contemporary syntactic theory, 483–​ 510. Oxford: Blackwell Publishers. Berg, Thomas. 1998. The resolution of number conflicts in English and German agreement patterns. Linguistics 36(1). 41–​70. Berlage, Eva. 2014. Noun phrase complexity in English. Cambridge: Cambridge University Press. Biber, Douglas, Susan Conrad & Geoffrey Leech. 2002. Longman student grammar of spoken and written English. Harlow: Longman. Biber, Douglas & Bethany Gray. 2011. Grammatical change in the noun phrase: The influence of written language use. English Language and Linguistics 15(2). 223–​250. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. London: Longman. Bisang, Walter. 2009. On the evolution of complexity:​Sometimes less is more in East and mainland Southeast Asia. In Geoffrey Sampson, David Gil & Peter Trudgill (eds.), Language complexity as an evolving variable, 34–​ 49. Oxford: Oxford University Press. Bisang, Walter. 2014. Overt and hidden complexity –​Two types of complexity and their implications. Poznań Studies in Contemporary Linguistics 50(2). 127–​143. Bisang, Walter. 2015. Hidden complexity – The neglected side of complexity and its implications. Linguistics Vanguard 1(1). 177–​187. Bjørge, Anne Kari. 1989. ‘I poor man is singing’: Subject-​verb concord in eighteenth-​ century English grammar. In Leiv E. Breivik, Arnoldus Hille & Stig Johansson (eds.), Essays on English language in honour of Bertil Sundby, 47–​65. Oslo: Novus Forlag.

1 9

References 191 Blevins, James P. 2000. Markedness and agreement. Transactions of the Philological Society 98(2). 233–​262. Bock, Kathryn, Manuel Carreiras & Enrique Meseguer. 2012. Number meaning and number grammar in English and Spanish. Journal of Memory and Language 66(1).  17–​37. Bock, Kathryn, Anne Cutler, Kathleen M. Eberhard, Sally Butterfield, J. Cooper Cutting & Karin R. Humphreys. 2006. Number agreement in British and American English: Disagreeing to agree collectively. Language 82(1). 64–​113. Bock, Kathryn & J. Cooper Cutting. 1992. Regulating mental energy: Performance units in language production. Journal of Memory and Language 31(1). 99–​127. Bock, Kathryn & Kathleen M. Eberhard. 1993. Meaning, sound and syntax in English number agreement. Language and Cognitive Processes 8(1). 57–​99. Bock, Kathryn & Carol A. Miller. 1991. Broken agreement. Cognitive Psychology 23(1).  45–​93. Bock, Kathryn, Janet Nicol & J. Cooper Cutting. 1999. The ties that bind: Creating number agreement in speech. Journal of Memory and Language 40(3). 330–​346. Boeckx, Cedric. 2006. Agreement systems. Amsterdam & Philadelphia: John Benjamins. Bolinger, Dwight. 1980. Couple: An English dual. In Sidney Greenbaum, Geoffrey Leech & Jan Svartvick (eds.), Studies in English Linguistics for Randolph Quirk, 30–​41. New York & London: Longman. Bošković, Željko. 2013. Principles and Parameters theory and Minimalism. In Marcel den Dikken (ed.), The Cambridge handbook of Generative syntax, 95–​121. Cambridge: Cambridge University Press. Brems, Lieselotte. 2003. Measure Noun constructions: An instance of semantically-​ driven grammaticalisation. International Journal of Corpus Linguistics 8(2). 283–​312. Brems, Lieselotte. 2004. Measure noun constructions: Degrees of delexicalization and grammaticalization. In Karin Aijmer & Bengt Altenberg (eds.), Advances in Corpus Linguistics. Papers from the 23rd International Conference on English Language Research on Computerized Corpora (ICAME 23), Göteborg 22–​26 May 2002, 249–​ 265. Amsterdam: Rodopi. Brems, Lieselotte. 2010. Size noun constructions as collocationally constrained constructions: Lexical and grammaticalised uses. English Language and Linguistics 14(1). 83–​109. Brems, Lieselotte. 2011. Layering of Size and Type Noun constructions in English. Berlin: Mouton de Gruyter. Brems, Lieselotte. 2012. The establishment of quantifier constructions for size nouns: A diachronic case study of heap(s) and lot(s). Journal of Historical Pragmatics 13(2). 202–​231. Brems, Lieselotte. 2015. Some binominal size noun constructions in English and French: A contrastive corpus-​based perspective. Nordic Journal of English Studies 14(1). 90–​115. Brems, Lieselotte & Kristin Davidse. 2003. Absolute and relative quantification: Beyond mutually exclusive word classes. Belgian Journal of English Language and Literatures. New Series 1. 49–​60. Brezina, Vaclav. 2018. Statistics in Corpus Linguistics: A practical guide. Cambridge: Cambridge University Press. Brinton, Laurel J. 1996. Pragmatic markers in English: Grammaticalization and discourse functions. Berlin: Mouton de Gruyter.

2 9 1

192 References Brinton, Laurel J. & Elizabeth Closs Traugott. 2005. Lexicalization and language change. Cambridge: Cambridge University Press. Brucart, José Mª. 1997. Concordancia ad sensum y partitividad en español. In Manuel Almeida & Josefa Dorta (eds.), Contribuciones al estudio de la lingüística hispánica: Homenaje al profesor Ramón Trujillo, vol. 1, 157–​183. Barcelona: Montesinos. Bybee, Joan. 2015. Language change. Cambridge: Cambridge University Press. Chomsky, Noam. 1981. Lectures on government and binding: The Pisa lectures. Dordrecht: Foris. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: The MIT Press. Coene, Martine & Yves D’hulst. 2003. Introduction: The syntax and semantics of noun phrases. Theoretical background. In Martine Coene & Yves D’hulst (eds.), From NP to DP, vol. 1: The syntax and semantics of noun phrases, 1–​33. Amsterdam & Philadelphia: John Benjamins. Collins, Peter. 2015. Diachronic variation in the grammar of Australian English: Corpus-​based explorations. In Peter Collins (ed.), Grammatical change in English world-​wide, 15–​42. Amsterdam & Philadelphia: John Benjamins. Comrie, Bernard. 1989 (1981). Language universals and linguistic typology: Syntax and morphology. Chicago, IL: The University of Chicago Press. Cook, V. J. & Mark Newson. 2007 (1988). Chomsky’s Universal Grammar: An introduction. Oxford: Blackwell. Corbett, G. G. 1979. The agreement hierarchy. Journal of Linguistics 15(2). 203–​224. Corbett, Greville G. 1994. Agreement. In R. E. Asher (ed.), The encyclopedia of language and Linguistics, vol. 1, 54–​60. Oxford: Pergamon Press. Corbett, Greville G. 2003. Agreement: The range of the phenomenon and the principles of the Surrey Database of Agreement. Transactions of the Philological Society 101(2). 155–​202. Corbett, Greville G. 2004. Number. Cambridge: Cambridge University Press. Corbett, Greville G. 2006. Agreement. Cambridge: Cambridge University Press. Corbett, Greville G. 2015. Hybrid nouns and their complexity. In Jürg Fleischer, Elisabeth Rieken & Paul Widmer (eds.), Agreement from a diachronic perspective, 191–​214. Berlin: Mouton de Gruyter. Corver, Norbert. 1998. Predicative movement in pseudopartitive constructions. In Artemis Alexiadou & Chris Wilder (eds.), Possessors, predicates and movement in the determiner phrase, 215–​258. Amsterdam & Philadelphia: John Benjamins. Corver, Norbert. 2013. Lexical categories and (extended) projection. In Marcel den Dikken (ed.), The Cambridge handbook of Generative syntax, 353–​424. Cambridge: Cambridge University Press. Craig, C. A. 1999. Classifier languages. In Keith Brown & Jim Miller (eds.), Concise encyclopedia of grammatical categories, 70–​75. Oxford: Elsevier. Croft, William. 2003 (1990). Typology and universals. Cambridge: Cambridge University Press. Cruse, D. A. 1986. Lexical semantics. Cambridge: Cambridge University Press. Crystal, David. 2000 (1984). Who cares about English usage? London: Penguin. Cuyckens, Hubert, Frauke D’hoedt & Benedikt Szmrecsanyi. 2014. Variability in verb complementation in Late Modern English: Finite vs. non-​finite patterns. In Marianne Hundt (ed.), Late Modern English syntax, 182–​ 204. Cambridge: Cambridge University Press. Davidse, Kristin. 2004. The interaction of quantification and identification in English determiners. In Michel Achard & Suzanne Kemmer (eds.), Language, culture, and mind, 507–​533. Stanford, CA: CSLI Publications.

3 9 1

References 193 de Haan, Pieter (see Haan) Dekeyser, Xavier. 1975. Number and case relations in 19th century British English: A comparative study of grammar and usage. Antwerp: De Nederlandsche Boekhandel. Dekeyser, Xavier. 1996. Concord of number in Lindley Murray’s Grammar: The dialectics between prescription and usage. In Ingrid Tieken-​Boon van Ostade (ed.), Two hundred years of Lindley Murray, 193–​205. Münster: Nodus. Delbecque, Nicole & Katrien Verveckken. 2014. Conceptually-​ driven analogy in the grammaticalization of Spanish binominal quantifiers. Linguistics 52(3). 637–​684. den Dikken, Marcel (see Dikken) Denison, David. 1998. Syntax. In Suzanne Romaine (ed.), The Cambridge history of the English Language, vol. 4: 1776–​1997, 92–​329. Cambridge: Cambridge University Press. Denison, David. 2002. History of the sort of construction family. Paper presented at the Second International Conference on Construction Grammar, University of Helsinki, 7 September 2002. http://​www.humanities.manchester.ac.uk/​medialibrary/​llc/​files/​ david-​denison/​Helsinki_​ICCG2.pdf (20 September, 2018.) Depraetere, Ilse. 2003. On verbal concord with collective nouns in British English. English Language and Linguistics 7(1). 85–​127. Dikken, Marcel den. 2001. “Pluringulars”, pronouns and quirky agreement. The Linguistic Review 18(1). 19–​41. Dodge, Ellen & Abby Wright. 2002. Herds of wildebeest, flasks of vodka, heaps of trouble: An Embodied Construction Grammar approach to English Measure Phrases. In Julie Larson & Mary Paster (eds.), Proceedings of the 28th Annual Meeting of the Berkeley Linguistics Society, 75–​86. Berkeley, CA: Berkeley Linguistics Society. Dowty, David & Pauline Jacobson. 1988. Agreement as a semantic phenomenon. In J. Powers (ed.), Proceedings of the 5th Annual Eastern States Conference on Linguistics (ESCOL), 95–​108. Elbourne, Paul. 1999. Some correlations between semantic plurality and quantifier scope. In Nancy Hall, Masako Hirotani & Pius Tamanji (eds.), The North East Linguistic Society (NELS) 29. 81–​92. Amherst, MA: GLSA. Epstein, Samuel D., Hisatsugu Kitahara, Miki Obata & T. Daniel Seely. 2013. Economy of derivation and representation. In Marcel den Dikken (ed.), The Cambridge handbook of Generative syntax, 487–​514. Cambridge: Cambridge University Press. Evans, Bergen & Cornelia Evans. 1957. A dictionary of contemporary American usage. New York: Random House. Fält, Gunnar. 1972. Tres problemas de concordancia verbal en el español moderno. Uppsala: Almqvist & Wiksell. Fernández-​Pena, Yolanda. 2017a. Exploring the patterns of verbal agreement of collective noun-​headed subjects in a parsed corpus of Present-​Day English. In Chelo Vargas-​Sierra (ed.), AESLA 2016. (EPiC Series in Language and Linguistics, vol. 2). 325–​334. Fernández-​Pena, Yolanda. 2017b. ‘These points stated, a number of problems remain’: A corpus-​based analysis of the idiomatisation of collective noun-​based constructions. In Sebastian Hoffmann, Andrea Sand & Sabine Arndt-​ Lappe (eds.), Exploring recent diachrony: Corpus studies of lexicogrammar and language practices in Late Modern English. (Studies in Variation, Contacts and Change in English 18). Helsinki: VARIENG. Fernández-​ Pena, Yolanda. 2017c. Verbal agreement with collective nominal constructions: Syntactic and semantic determinants. Atlantis 39(1). 33–​54.

4 9 1

194 References Fernández-​ Pena, Yolanda. 2017d. Verbal agreement with collectives taking of-​ dependents: Syntactic and structural complexity as determinant factors. In Sofía Bemposta-​Rivas, Carla Bouzada-​Jabois, Yolanda Fernández-​Pena, Tamara Bouso, Yolanda J. Calvo-​Benzies & Iván Tamaredo (eds.), New trends and methodologies in applied English language research III: Synchronic and diachronic studies on discourse, lexis and grammar processing, 109–​130. Bern: Peter Lang. Fernández-​Pena, Yolanda. 2017e. Patterns of verbal agreement with collective nouns taking plural of-​dependents: A corpus-​based analysis of syntactic distance. Corpora 12(2). 207–​241. Fernández-​Pena, Yolanda. 2017f. Sujetos colectivos y concordancia verbal: Un estudio de frecuencias, semántica y sintaxis. RAEL: Revista Electrónica de Lingüística Aplicada 16(1). 70–​84. Ferreira, Fernanda. 1991. Effects of length and syntactic complexity on initiation times for prepared sentences. Journal of Memory and Language 30(2). 210–​233. Fogg, Peter Walkden. 1792. Elementa anglicana; or, The principles of English grammar displayed and exemplified in a method entirely new. In two volumes, vol. 1. Stockport. Francis, Elaine J. & Etsuyo Yuasa. 2008. A multi-​modular approach to gradual change in grammaticalization. Journal of Linguistics 44(1). 45–​86. Franck, Julie, Gabriella Vigliocco & Janet Nicol. 2002. Subject-​verb agreement errors in French and English: The role of syntactic hierarchy. Language and Cognitive Processes 17(4). 371–​404. Frazier, Lyn. 1985. Syntactic complexity. In David R. Dowty, Lauri Karttunen & Arnold M. Zwicky (eds.), Natural language parsing: Psychological, computational, and theoretical perspectives, 129–​189. Cambridge: Cambridge University Press. Fries, Udo. 1988. The crew have abandoned the ship: Concord with collective nouns revisited. AAA: Arbeiten aus Anglistik und Amerikanistik 13(2). 99–​104. Gardelle, Laure. 2019. Semantic plurality: English collective nouns and other ways of denoting pluralities of entities. Amsterdam & Philadelphia: John Benjamins. Gelman, Andrew & Jennifer Hill. 2007. Data analysis using regression and multilevel/​ hierarchical models. Cambridge: Cambridge University Press. Gibson, Edward. 1998. Linguistic complexity: Locality of syntactic dependencies. Cognition 68(1). 1–​76. Gibson, Edward. 2000. The dependency locality theory: A distance-​based theory of linguistic complexity. In Alec P. Marantz, Yasushi Miyashita & Wayne O’Neil (eds.), Image, language, brain: Papers from the First Mind Articulation Project Symposium, 95–​126. Cambridge, MA: The MIT Press. Grafmiller, Jason. 2018. JGmermod: Custom functions for mixed-​effects regression models. R package version 0.2.0. Gries, Stefan Th. 2007. Coll.analysis 3.2a. A program for R for Windows 2.x. Gries, Stefan Th. 2013 (2009). Statistics for Linguistics with R. Berlin: Mouton de Gruyter. Gries, Stefan Th. 2015. The most under-​used statistical method in Corpus Linguistics: Multi-​level (and mixed-​effects) models. Corpora 10(1). 95–​125. Gries, Stefan Th. & Anatol Stefanowitsch. 2004a. Extending collostructional analysis. A corpus-​based perspective on ‘alternations’. International Journal of Corpus Linguistics 9(1). 97–​129. Gries, Stefan Th. & Anatol Stefanowitsch. 2004b. Covarying collexemes in the into-​ causative. In Michel Achard & Suzanne Kemmer (eds.). Language, culture, and mind, 225–​236. Stanford, CA: CSLI Publications.

5 9 1

References 195 Guy, Gregory R. 1988. Advanced varbrul analysis. In Kathleen Ferrara, Becky Brown, Keith Walters & John Baugh (eds.), Linguistic change and contact, 124–​136. Austin, TX: University of Texas Department of Linguistics. Haan, Pieter de. 1989. Postmodifying clauses in the English noun phrase: A corpus-​ based study. Amsterdam: Rodopi. Haan, Pieter de. 1991. On the exploration of corpus data by means of problem-​oriented tagging: Postmodifying clauses in the English noun phrase. In Stig Johansson & Anna-​Brita Stenström (eds.), English computer corpora: Selected papers and research guide, 51–​65. Berlin: Mouton de Gruyter. Haan, Pieter de. 1993. Noun phrase structure as an indication of text variety. In Andreas H. Jucker (ed.), The noun phrase in English: Its structure and variability, 85–​106. Heidelberg: Winter. Haegeman, Liliane. 2006. Thinking syntactically: A guide to argumentation and analysis. Oxford: Blackwell. Harrell, Frank E., Jr., with contributions from Charles Dupont & many others. 2019. Hmisc: Harrell miscellaneous. R package version 4.3-​0. https://​CRAN.R-​project. org/​package=Hmisc. Haskell, Todd R. & Maryellen C. MacDonald. 2003. Conflicting cues and competition in subject-​verb agreement. Journal of Memory and Language 48(4). 760–​778. Hawkins, John A. 1994. A performance theory of order and constituency. Cambridge: Cambridge University Press. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. Grammaticalization: A conceptual framework. Chicago, IL: The University of Chicago Press. Hilpert, Martin. 2014. Collostructional analysis: Measuring associations between constructions and lexical elements. In Dylan Glynn & Justyna A. Robinson (eds.), Corpus methods for semantics: Quantitative studies in polysemy and synonymy, 391–​404. Amsterdam & Philadelphia: John Benjamins. Hopper, Paul J. & Elizabeth Closs Traugott. 2003 (1993). Grammaticalization. Cambridge: Cambridge University Press. Hornstein, Norbert, Jairo Nunes & Kleanthes K. Grohmann. 2005. Understanding Minimalism. Cambridge: Cambridge University Press. Hothorn, Torsten, Peter Bühlmann, Sandrine Dudoit, Annette Molinaro & Mark J. Van Der Laan. 2006. Survival ensembles. Biostatistics 7(3). 355–​373. Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Humphreys, Karin R. & Kathryn Bock. 2005. Notional number agreement in English. Psychonomic Bulletin & Review 12(4). 689–​695. Hundt, Marianne. 2006. The committee has/​have decided …: On concord patterns with collective nouns in inner-​and outer-​circle varieties of English. Journal of English Linguistics 34(3). 206–​232. Hundt, Marianne. 2009a. Colonial lag, colonial innovation or simply language change? In Günter Rohdenburg & Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 13–​37. Cambridge: Cambridge University Press. Hundt, Marianne. 2009b. Concord with collective nouns in Australian and New Zealand English. In Pam Peters, Peter Collins & Adam Smith (eds.), Comparative studies in Australian and New Zealand English: Grammar and beyond, 207–​224. Amsterdam & Philadelphia: John Benjamins. Hyper Usage Guide of English (HUGE). 2014. Leiden University. http://​huge.ullet.net.

6 9 1

196 References Jackendoff, Ray S. 1968. Quantifiers in English. Foundations of Language 4(4). 422–​442. Jackendoff, Ray S. 1977. X’ syntax: A study of phrase structure. Cambridge, MA: The MIT Press. Jantos, Susanne. 2009. Agreement in educated Jamaican English: A corpus investigation of ICE-​Jamaica. Freiburg: Albert Ludwig University of Freiburg PhD dissertation. Jantos, Susanne. 2010. Agreement in educated Jamaican English: A corpus-​ based study of spoken usage in ICE-​Jamaica. In Heidrun Dorgeloh & Anja Wanner (eds.), Syntactic variation and genre, 305–​331. Berlin: Mouton de Gruyter. Jespersen, Otto. 1909–​1949. A modern English grammar on historical principles, vol. 2: Syntax (first volume). London: George Allen & Unwind. Joosten, Frank. 2006. Why club and lingerie do not belong together. A plea for redefining collective nouns. In Georges Kleiber, Catherine Schnedecker & Anne Theissen (eds.), La relation partie-​tout, 73–​88. Leuven: Peeters. Juul, Arne. 1975. On concord of number in Modern English. Copenhagen: Nova. Kachru, Braj B. 1985. Standards, codification, and sociolinguistic realism: The English language in the outer circle. In Randolph Quirk & H. G. Widdowson (eds.), English in the world: Teaching and learning the language and literatures, 11–​30. Cambridge: Cambridge University Press. Kathol, Andreas. 2000. Agreement and the syntax-​morphology interface in HPSG. In Robert D. Levine & Georgia M. Green (eds.), Studies in contemporary Phrase Structure Grammar, 223–​274. Cambridge: Cambridge University Press. Keenan, Caitlin. 2013. ‘A pleasant three days in Philadelphia’: Arguments for a pseudopartitive analysis. University of Pennsylvania Working Papers in Linguistics 19(1). 87–​96. http://​repository.upenn.edu/​pwpl/​vol19/​iss1/​11 (20 September, 2018.) Keizer, Evelien. 2007. The English noun phrase: The nature of linguistic categorization. Cambridge: Cambridge University Press. Keizer, Evelien. 2017. English partitives in Functional Discourse Grammar: Types and constraints. Glossa: A Journal of General Linguistics 2(1). 16. Keizer, Evelien. 2019. Noun phrases. In Bas Aarts, Jill Bowie & Gergana Popova (eds.), The Oxford handbook of English grammar, 335–​ 357. Oxford: Oxford University Press. Kim, Jong-​Bok. 2002. On the structure of English partitive NPs and agreement. Studies in Generative Grammar 12(2). 309–​338. Kim, Jong-​Bok & Jooyoung Lim. 2011. Measure noun phrases in English: Usages in a corpus and textbooks. Studies in Modern Grammar 66. 47–​69. Kim, Jong-​Bok & Peter Sells. 2015. English binominal NPs: A construction-​based perspective. Journal of Linguistics 51(1). 41–​73. Klockmann, Heidi. 2017. The design of semi-​ lexicality: Evidence from case and agreement in the nominal domain. Utrecht: University of Utrecht PhD dissertation. Koptjevskaja-​Tamm, Maria. 2001. “A piece of the cake” and “a cup of tea”: Partitive and pseudo-​ partitive nominal constructions in the Circum-​ Baltic languages. In Östen Dahl & Maria Koptjevskaja-​ Tamm (eds.), Circum-​Baltic languages, vol. 2: Grammar and typology, 523–​ 568. Amsterdam & Philadelphia: John Benjamins. Koptjevskaja-​Tamm, Maria. 2009. “A lot of grammar with a good portion of lexicon”: Towards a typology of partitive and pseudo-​partitive nominal constructions. In Johannes Helmbrecht, Yoko Nishina, Yong-​Min Shin, Stavros Skopeteas & Elisabeth Verhoeven (eds.), Form and function in language research: Papers in honour of Christian Lehmann, 329–​346. Berlin: Mouton de Gruyter.

7 9 1

References 197 Kruisinga, Etsko. 1932 (1911). A handbook of Present-​Day English. Part II: English accidence and syntax 3. Groningen: P. Noordhoff. Kuhn, Max, with contributions from Jed Wing, Steve Weston, Andre Williams, Chris Keefer, Allan Engelhardt, Tony Cooper, Zachary Mayer, Brenton Kenkel, the R Core Team, Michael Benesty, Reynald Lescarbeau, Andrew Ziem, Luca Scrucca, Yuan Tang, Can Candan & Tyler Hunt. 2019. caret: Classification and regression training. R package version 6.0–​84. https://​CRAN.R-​project.org/​package=caret. Labov, William. 1969. Contraction, deletion, and inherent variability of the English copula. Language 45(4). 715–​762. Lakaw, Alexander. 2017. Diachronic shifts in agreement patterns of collective nouns in 19th-​ century American and British English. In Sebastian Hoffmann, Andrea Sand & Sabine Arndt-​Lappe (eds.), Exploring recent diachrony: Corpus studies of lexicogrammar and language practices in Late Modern English. (Studies in Variation, Contacts and Change in English 18). Helsinki: VARIENG. Langacker, Ronald W. 1990. Concept, image and symbol: The cognitive basis of grammar. Berlin: Mouton de Gruyter. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, vol. 2: Descriptive application. Stanford, CA: Stanford University Press. Langacker, Ronald W. 2004. Remarks on nominal grounding. Functions of Language 11(1). 77–​113. Langacker, Ronald W. 2008. Cognitive Grammar: A basic introduction. Oxford: Oxford University Press. Langacker, Ronald W. 2009. Investigations in Cognitive Grammar. Berlin: Mouton de Gruyter. Langacker, Ronald W. 2010. A lot of quantifiers. In Sally Rice & John Newman (eds.), Empirical and experimental methods in cognitive/​ functional research, 41–​57. Stanford, CA: CSLI Publications. Langacker, Ronald W. 2013. The indefinite article in complex quantifiers. In Mike Borkent, Barbara Dancygier & Jennifer Hinnell (eds.), Language and the creative mind, 65–​81. Stanford, CA: CSLI Publications. Langacker, Ronald W. 2016. Nominal grounding and English quantifiers. Cognitive Linguistic Studies 3(1). 1–​31. Langacker, Ronald W. 2017a. Grounding, semantic functions, and absolute quantifiers. English Text Construction 10(2). 233–​248. Langacker, Ronald W. 2017b. Ten lectures on the elaboration of Cognitive Grammar. Leiden: Brill. Leclercq, Benoît & Ilse Depraetere. 2018. Verbal agreement with partitive noun phrases. Studia Linguistica 72(2). 340–​361. Lehmann, Hans Martin & Gerold Schneider. 2009. Parser-​based analysis of syntax-​ lexis interactions. In Andreas H. Jucker, Daniel Schreier & Marianne Hundt (eds.), Corpora: Pragmatics and discourse. Papers from the 29th International Conference on English Language Research on Computerized Corpora (ICAME 29). Ascona, Switzerland, 14–​18 May 2008, 477–​502. Amsterdam: Rodopi. Lehmann, Hans Martin & Gerold Schneider. 2012. BNC Dependency Bank 1.0. In Signe Oksefjell Ebeling, Jarle Ebeling & Hilde Hasselgård (eds.), Aspects of Corpus Linguistics: Compilation, annotation, analysis. (Studies in Variation, Contacts and Change in English 12). Helsinki: VARIENG. Lehrer, Adrienne. 1986. English classifier constructions. Lingua 68(2/​3). 109–​148. Lehrer, Adrienne. 1987. English quantifiers from noun sources. In Ross Steele & Terry Threadgold (eds.), Language topics: Essays in honour of Michael Halliday, vol. 2, 95–​105. Amsterdam & Philadelphia: John Benjamins.

8 9 1

198 References Levin, Beth. 1993. English verb classes and alternations: A preliminary investigation. Chicago, IL: University of Chicago Press. Levin, Magnus. 1998. On concord with collective nouns in English. In Antoinette Renouf (ed.), Explorations in Corpus Linguistics, 99–​112. Amsterdam: Rodopi. Levin, Magnus. 1999. Concord with collective nouns revisited. ICAME Journal 23.  21–​33. Levin, Magnus. 2001. Agreement with collective nouns in English. Lund: Lund Studies in English. Levin, Magnus. 2006. Collective nouns and language change. English Language and Linguistics 10(2). 321–​343. Levshina, Natalia. 2015. How to do Linguistics with R: Data exploration and statistical analysis. Amsterdam & Philadelphia: John Benjamins. Lindquist, Hans. 2009. Corpus Linguistics and the description of English. Edinburgh: Edinburgh University Press. Löbel, Elisabeth. 1989. Q as a functional category. In Christa Bhatt, Elisabeth Löbel & Claudia Maria Schmidt (eds.), Syntactic phrase structure phenomena in noun phrases and sentences, 133–​158. Amsterdam & Philadelphia: John Benjamins. Martí i Girbau, Núria. 2010. The syntax of partitives. Barcelona: Autonomous University of Barcelona PhD dissertation. Matthews, Peter H. 1991 (1974). Morphology. Cambridge: Cambridge University Press. Milsark, Gary L. 1977. Toward and explanation of certain peculiarities of the existential construction in English. Linguistic Analysis 3(1). 1–​29. Moravcsik, Edith A. 1988. Agreement and markedness. In Michael Barlow & Charles A. Ferguson (eds.), Agreement in natural language: Approaches, theories, descriptions, 89–​106. Stanford, CA: CSLI Publications. Morgan, Jerry L. 1972. Verb agreement as a rule of English. In P. M. Peranteau, J. N. Levi & G. C. Phares (eds.), Papers from the Eighth Regional Meeting, Chicago Linguistic Society (April 14–​16, 1972), 278–​286. Chicago, IL: Chicago Linguistic Society. Neels, Jakob & Stefan Hartmann. 2018. Reduction or expansion? A bit of both. A  case study on the development of German degree modifiers. In Evie Coussé, Peter Andersson & Joel Olofsson (eds.), Grammaticalization meets Construction Grammar, 137–​168. Amsterdam & Philadelphia: John Benjamins. Nesfield, J. C. 1898. English grammar: Past and present. London: Macmillan. Nicol, Janet L. 1995. Effects of clausal structure on subject-​verb agreement errors. Journal of Psycholinguistic Research 24(6). 507–​516. Nixon, Graham. 1972. Corporate-​concord phenomena in English. Studia Neophilologica 44(1). 120–​126. Nixon, H. 1826. The English parser; being a complete and original system of English parsing, with examples and models. London. O’Conner, Patricia T. 1996. Woe is I: The grammarphobe’s guide to better English in plain English. New York: G. P. Putnam’s Sons. Persson, Gunnar. 1989. On the semantics of collective nouns. In Bengt Odenstedt & Gunnar Persson (eds.), Instead of flowers: Papers in honour of Mats Rydén on the occasion of his sixtieth birthday, August 27, 1989, 179–​ 188. Umeå: University of Umeå. Pickett, Joseph P., Steven Kleinedler & Susan Spitz (eds.). 2005. The American heritage guide to contemporary usage and style. Boston, MA: Houghton Mifflin. Pollard, Carl & Ivan A. Sag. 1987. Information-​based syntax and semantics, vol. 1: Fundamentals. Stanford, CA: CSLI Publications.

91

References 199 Pollard, Carl & Ivan A. Sag. 1988. An information-​based theory of agreement. Stanford, CA: CSLI Publications. Pollard, Carl & Ivan A. Sag. 1994. Head-​driven Phrase Structure Grammar. Stanford, CA: CSLI Publications. Pollock, Jean-​Yves. 1989. Verb movement, Universal Grammar and the structure of IP. Linguistic Inquiry 20(3). 365–​424. Pooley, Robert C. 1934. Subject-​verb agreement. American Speech 9(1). 31–​36. Poutsma, H. 1914. A grammar of Late Modern English for the use of continental, especially Dutch, students, Part II: The parts of speech, Section I, A: Nouns, adjectives and articles. Groningen: P. Noordhoff. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. R Core Team. 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://​www.R-​project.org. Radden, Günter & René Dirven. 2007. Cognitive English grammar. Amsterdam & Philadelphia: John Benjamins. Radford, Andrew. 2004. Minimalist syntax: Exploring the structure of English. Cambridge: Cambridge University Press. Reed, Ann. 1996. Partitives, existentials, and partitive determiners. In Jacob Hoeksema (ed.), Partitives: Studies on the syntax and semantics of partitive and related constructions, 143–​178. Berlin: Mouton de Gruyter. Reid, Wallis. 1991. Verb and noun number in English: A functional explanation. New York: Longman. Riveiro-​ Outeiral, Sara María & Juan Carlos Acuña-​ Fariña. 2012. Agreement processes in English and Spanish. A  completion study. Functions of Language 19(1). 58–​88. Rohdenburg, Günter. 1996. Cognitive complexity and increased grammatical explicitness in English. Cognitive Linguistics 7(2). 149–​182. Rohdenburg, Günter. 2003. Cognitive complexity and horror aequi as factors determining the use of interrogative clause linkers in English. In Günter Rohdenburg & Britta Mondorf (eds.), Determinants of grammatical variation, 205–​249. Berlin: Mouton de Gruyter. Ross, John Robert. 2004 (1973). Nouniness. In Bas Aarts, David Denison, Evelien Keizer & Gergana Popova (eds.), Fuzzy grammar: A reader, 351–​422. Oxford: Oxford University Press. Röthlisberger, Melanie. 2018. Regional variation in probabilistic grammars: A multifactorial study of the English dative alternation. Leuven: University of Leuven PhD dissertation. Rutkowski, Paweɫ. 2007. The syntactic structure of grammaticalized partitives (Pseudo-​ partitives). University of Pennsylvania Working Papers in Linguistics 13(1). 337–​ 350. http://​repository.upenn.edu/​pwpl/​vol13/​iss1/​26 (20 September, 2018.) Sand, Andrea. 2008. Angloversals? Concord and interrogatives in contact varieties of English. In Terttu Nevalainen, Irma Taavitsainen, Päivi Pahta & Minna Korhonen (eds.), The dynamics of linguistic variation: Corpus evidence on English past and present, 183–​202. Amsterdam & Philadelphia: John Benjamins. Sauerland, Uli. 2004a. A team, definitely. Snippets 9. 11–​12. Sauerland, Uli. 2004b. A comprehensive semantics for agreement. Manuscript. https://​ core.ac.uk/​download/​pdf/​14511769.pdf (20 September, 2018.) Sauerland, Uli & Paul Elbourne. 2002. Total reconstruction, PF movement, and derivational order. Linguistic Inquiry 33(2). 283–​319.

02

200 References Schäpers, Uta Katharina Elisabeth. 2009. Nominal versus clausal complexity in spoken and written English: Theory and description. Frankfurt am Main: Peter Lang. Schaub, Steffen. 2016. The influence of register on noun phrase complexity in varieties of English. In Christoph Schubert & Christina Sanchez-​Stockhammer (eds.), Variational text linguistics: Revisiting register in English, 251–​270. Berlin: Mouton de Gruyter. Schlüter, Julia. 2005. Rhythmic grammar: The influence of rhythm on grammatical variation and change in English. Berlin: Mouton de Gruyter. Selkirk, Elisabeth. 1977. Some remarks on noun phrase structure. In Peter W. Culicover, Thomas Wasow & Adrian Akmajian (eds.), Formal syntax, 285–​316. New  York: Academic Press. Shao, Bin, Yingying Cai & Graeme Trousdale. 2019. A multivariate analysis of diachronic variation in a bunch of NOU N : A Construction Grammar account. Journal of English Linguistics 47(2). 150–​174. Shih, Stephanie & Jason Grafmiller. 2011. Weighing in on end weight. Paper presented at the 85th Annual Meeting of the Linguistic Society of America, Pittsburg, PA, 9–​11 January. Siemund, Rainer. 1995. ‘For who the bell tolls’ –​Or, why Corpus Linguistics should carry the bell in the study of language change in Present-​Day English. AAA: Arbeiten aus Anglistik und Amerikanistik 20(2). 351–​377. Sinclair, J. M. 1989. Uncommonly common words. In Makhan L. Tickoo (ed.), Learners’ dictionaries: State of the art, 135–​152. Singapore: SEAMEO Regional Language Centre. Sinclair, John. 1991. Corpus, concordance, collocation. Oxford: Oxford University  Press. Smith, Adam. 2009. Non-​numerical quantifiers. In Pam Peters, Peter Collins & Adam Smith (eds.), Comparative studies in Australian and New Zealand English: Grammar and beyond, 159–​181. Amsterdam & Philadelphia: John Benjamins. Smith, Carlota S. 1988. Factors of linguistic complexity and performance. In Alice Davison & Georgia M. Green (eds.), Linguistic complexity and text comprehension: Readability issues reconsidered, 247–​279. Hillsdale, NJ: Lawrence Erlbaum. Smith, Carlota S. & Anne van Kleeck. 1986. Linguistic complexity and performance. Journal of Child Language 13(2). 389–​408. Smith, Garrett, Julie Franck & Whitney Tabor. 2018. A self-​organizing approach to subject-​verb number agreement. Cognitive Science 42(S4). 1043–​1074. Smith, Peter W. 2013. Collective (dis)agreement: On a 3/​4 pattern of British English collective NPs. In Enrico Boone, Martin Kohlberger & Maartje Schulpen (eds.), Proceedings of ConSOLE XX, 2012, 229–​253. Leiden: University of Leiden. Smith, Peter W. 2017. The syntax of semantic agreement in English. Journal of Linguistics 53(4). 823–​863. Smith, Peter W., Johannes Mursell & Katharina Hartmann (eds.). 2020. Agree to agree: Agreement in the Minimalist Programme. Berlin: Language Science Press. Smitterberg, Erik. 2006. Partitive constructions in nineteenth-​century English. In Merja Kytö, Mats Rydén & Erik Smitterberg (eds.), Nineteenth-​century English: Stability and change, 242–​271. Cambridge: Cambridge University Press. Speelman, Dirk, Kris Heylen & Dirk Geeraerts. 2018. Introduction. In Dirk Speelman, Kris Heylen & Dirk Geeraerts (eds.), Mixed-​effects regression models in Linguistics, 1–​10. Cham: Springer. Stavrou, Melita. 2003. Semi-​ lexical nouns, classifiers, and the interpretation(s) of the pseudopartitive construction. In Martine Coene & Yves D’hulst. From NP to DP, vol. 1: The syntax and semantics of noun phrases, 329–​353. Amsterdam & Philadelphia: John Benjamins.

1 0 2

References 201 Stefanowitsch, Anatol & Stefan Th. Gries. 2003. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8(2). 209–​243. Stickney, Helen. 2009. The emergence of DP in the partitive structure. Amherst, MA: University of Massachusetts PhD dissertation. Stilman, Anne. 1997. Grammatically correct: The writer’s essential guide to punctuation, spelling, style, usage and grammar. Cincinnati, OH: Writer’s Digest. Strang, Barbara M. H. 1966. Some features of S-​V concord in Present-​Day English. In Ilva Cellini & Giorgio Melchiori (eds.), English studies today. Fourth series, 73–​87. Rome: Edizioni di Storia e Letteratura. Strang, Barbara M.  H. 1969 (1962). Modern English structure. London: Edward Arnold. Strobl, Carolin, Anne-​ Laure Boulesteix, Thomas Kneib, Thomas Augustin & Achim Zeileis. 2008. Conditional variable importance for random forests. BMC Bioinformatics 9(307). http://​www.biomedcentral.com/​1471–​2105/​9/​307. Strobl, Carolin, Anne-​ Laure Boulesteix, Achim Zeileis & Torsten Hothorn. 2007. Bias in random forest variable importance measures: Illustrations, sources and a solution. BMC Bioinformatics 8(25). http://​www.biomedcentral.com/​1471–​ 2105/​8/​25. Sundby, Bertil, Anne Kari Bjørge & Kari E. Haugland. 1991. A dictionary of English normative grammars: 1700–​ 1800 (DENG). Amsterdam & Philadelphia: John Benjamins. Szmrecsányi, Benedikt M. 2004. On operationalizing syntactic complexity. In Gérald Purnelle, Cédrick Fairon & Anne Dister (eds.), Le poids des mots: Actes des 7es Journées Internationales D’Analyse Statistique des Données Textuelles, 1031–​1038. Louvain-​la-​Neuve: Presses Universitaires de Louvain. Szmrecsanyi, Benedikt. 2019. Register in variationist linguistics. Register Studies 1(1).  76–​99. Szmrecsanyi, Benedikt, Douglas Biber, Jesse Egbert & Karlien Franco. 2016. Toward more accountability: Modelling ternary genitive variation in Late Modern English. Language Variation and Change 28(1). 1–​29. Taggart, Caroline. 2010. Her ladyship’s guide to the Queen’s English. London: National Trust. Tagliamonte, Sali A. 2006. Analysing sociolinguistic variation. Cambridge: Cambridge University Press. Tagliamonte, Sali A. & R. Harald Baayen. 2012. Models, forests, and trees of York English: Was/​were variation as a case study for statistical practice. Language Variation and Change 24(2). 135–​178. Traugott, Elizabeth Closs. 2008a. Grammaticalization, constructions and the incremental development of language: Suggestions from the development of Degree Modifiers in English. In Regine Eckardt, Gerhard Jäger & Tonjes Veenstra (eds.), Variation, selection, development: Probing the evolutionary model of language change, 219–​250. Berlin: Mouton de Gruyter. Traugott, Elizabeth Closs. 2008b. The grammaticalization of NP of NP patterns. In Alexander Bergs & Gabriele Diewald (eds.), Constructions and language change, 23–​45. Berlin: Mouton de Gruyter. Traugott, Elizabeth Closs & Graeme Trousdale. 2013. Constructionalization and constructional changes. Oxford: Oxford University Press. Tristram, Anna. 2014. Diachronic change in verbal agreement patterns with majorité. Transactions of the Philological Society 112(3). 344–​366.

2 0

202 References Tristram, Anna. 2015. L’accord sujet-​verbe en français contemporain: Une étude de cas –​ la/​une foule. Revue Romane 50(2). 191–​221. Trudgill, Peter & Jean Hannah. 2008 (1982). International English: A guide to varieties of Standard English. London: Hodder Education. Vantellini, Laura. 2003. Agreement with collective nouns in New Zealand English. New Zealand English Journal 17. 45–​49. Verveckken, Katrien. 2012. Towards a constructional account of high and low frequency binominal quantifiers in Spanish. Cognitive Linguistic 23(2). 421–​478. Verveckken, Katrien. 2015. Binominal quantifiers in Spanish: Conceptually-​driven analogy in diachrony and synchrony. Berlin: Mouton de Gruyter. Verveckken, Katrien. 2016. Binominal quantifiers in Spanish: Syntagmatic and paradigmatic analogy in interaction. Language Sciences 53(part B). 114–​135. Vigliocco, Gabriella & Janet Nicol. 1998. Separating hierarchical relations and word order in language production: Is proximity concord syntactic or linear? Cognition 68(1). B13–​B29. Vos, Riet. 1999. A grammar of partitive constructions. Tilburg: University of Tilburg PhD dissertation. Wasow, Thomas. 1997. Remarks on grammatical weight. Language Variation and Change 9(1). 81–​105. Wasow, Thomas. 2002. Postverbal behavior. Stanford, CA: CSLI Publications. Wechsler, Stephen & Larisa Zlatić. 1998a. Agreement in discourse. In Proceedings of the Conference on the Structure of Non-​Narrative Texts, University of Texas, Austin. http://​citeseerx.ist.psu.edu/​viewdoc/​download;jsessionid=8884C99C7 8B7BBFC2DE75D8F37757E57?doi=10.1.1.6.8246&rep=rep1&type=pdf (20 September, 2018.) Wechsler, Stephen & Larisa Zlatić. 1998b. A theory of agreement and disagreement. In Benjamin K. Bergen, Madelaine C. Plauché & Ashlee C. Bailey (eds.), Proceedings of the Twenty-​Fourth Annual Meeting of the Berkeley Linguistics Society: General Session and Parasession on Phonetics and Phonological Universals, 280–​291. Berkeley, CA: Berkeley Linguistics Society. Wechsler, Stephen & Larisa Zlatić. 2000. A theory of agreement and its application to Serbo-​Croatian. Language 76(4). 799–​832. Willemse, Peter. 2007. Indefinite possessive NPs and the distinction between determining and nondetermining genitives in English. English Language and Linguistics 11(3). 537–​568. Wong, May. 2017. Hong Kong English: Exploring lexicogrammar and discourse from a corpus-​linguistic perspective. London: Palgrave Macmillan. Yaruss, J. Scott. 1999. Utterance length, syntactic complexity, and childhood stuttering. Journal of Speech, Language, and Hearing Research 42(2). 329–​344. Yuasa, Etsuyo & Elaine J. Francis. 2003. Categorial mismatch in a multi-​modular theory of grammar. In Elaine J. Francis & Laura A. Michaelis (eds.), Mismatch: Form-​function incongruity and the architecture of grammar, 179–​227. Stanford, CA: CSLI Publications. Zaenen, Annie, Jean Carletta, Gregory Garretson, Joan Bresnan, Andrew Koontz-​ Garboden, Tatiana Nikitina, M. Catherine O’Connor & Tom Wasow. 2004. Animacy encoding in English: Why and how. In Bonnie Webber & Donna Byron (eds.), Proceedings of the 2004 ACL Workshop on Discourse Annotation, 118–​125. East Stroudsburg, PA: Association for Computational Linguistics.

3 0 2

References 203 Zhang, Xu. 2017. English Quasi-​ Numeral Classifiers: A corpus-​ based cognitive-​ typological study. Bern: Peter Lang. Zuur, Alain F., Elena N. Ieno, Neil J. Walker, Anatoly A. Saveliev & Graham M. Smith. 2009. Mixed effects models and extensions in ecology with R. New York: Springer.

4 0 2

Index of collective nouns and complex collective noun phrases

band 7–​9, 109, 126–​127, 149, 158, 164–​165, 168 batch 7–​9, 12, 17, 109, 126–​127, 149, 158, 166, 168 bunch 2–​3, 7–​9, 12–​13, 15, 17, 19, 31, 41, 55–​57, 68, 73–​75, 77–​78, 82, 91, 95, 97–​98, 100n16, 101n25, 102n32, 109, 126–​127, 130, 149, 151, 157–​158, 163–​164, 168, 178, 180n34, 181n40, 182; a bunch of 9, 12, 14–​15, 19, 22, 28, 31, 74–​78, 81, 86, 91–​92, 98–​99, 102n28, 169, 183; the bunch of 75, 77; (a) buncha 28, 102n28 class 7–​9, 109, 126, 149, 158, 167–​168 clump 3, 7–​9, 107–​108, 147, 149, 165, 168, 171 couple 7–​9, 12–​13, 46, 55–​57, 78–​80, 90–​91, 95, 98, 99n2, 100n17, 101n19–​21, 102n32, 107–​108, 127, 146–​147, 149, 151, 157–​158, 163, 168, 178, 178n8, 180n31, 181n40; coupla 102n28; a couple of 12, 64, 78–​80, 84, 86, 90–​92, 98–​99, 100n18, 101n20, 102n28, 183; the couple of 80 crowd 3, 5, 7–​9, 11–​12, 14, 17, 109, 126–​127, 149, 158, 164–​165, 168 flock 2–​3, 7–​9, 12, 40–​41, 109, 126, 149, 158, 165, 168 gang 7–​9, 12, 109, 126, 149, 158, 164–​165, 168 group 1–​3, 7–​9, 12, 17, 31, 55–​57, 64–​66, 68, 71, 73, 75, 90–​91, 95–​98, 100n10–​11, 102n32, 109, 126–​127, 149, 151–​152, 157–​158, 163, 168, 178n8, 184; a group of 13,

27, 31, 65–​68, 81, 86–​87, 90–​92, 98–​99, 183; the group of 65–​66 herd 3, 7–​9, 12, 109, 126, 149, 158, 165, 168, 182 host 7–​9, 55–​57, 80–​82, 91, 95, 97–​98, 102n32, 109, 126–​127, 149, 151, 157–​158, 160, 163, 168, 178, 181n40; a host of 13, 81–​83, 86–​87, 91–​92, 98–​99, 169, 183; the host of 81 majority 2, 7–​9, 12–​13, 15, 17, 31, 46, 50n7, 55–​57, 69–​73, 75, 87, 90, 93–​95, 97–​99, 99n2, 100n12–​13, 101n27, 102n28, 102n32, 109, 126–​127, 146, 149–​153, 156–​158, 160–​162, 168, 178, 178n8, 182–​183; a majority of 13, 69–​71, 73, 87, 89–​90, 92–​93, 98, 100n15, 102n27, 183; the majority of 69–​73, 85, 89, 93, 98, 100n15 minority 7–​9, 12, 55–​57, 84–​85, 90, 93–​95, 98–​99, 101n26, 102n27–28, 102n32, 109, 126–127, 149, 151, 153, 157–158, 163, 168, 178, 183; a minority of 84–​85, 87, 90, 92, 98, 102n27, 183; the minority of 84, 101n26 number 1, 7–​9, 12–​13, 17, 24, 31, 50n7, 55–​57, 59–​62, 70–​71, 73, 75, 91–​92, 95–​98, 99n2, 109, 126–​127, 149–​152, 156–​158, 162, 168, 178n8, 184; a number of 12, 31, 59–​65, 70, 86, 91–​92, 98–​99, 100n8, 183; the number of 59–​62, 65, 96–​97, 151 pack 7–​9, 12, 41, 109, 126, 149, 158, 167–​168

5 0 2

Index of collective nouns and complex collective noun phrases 205 party 7–​9, 12, 17, 109, 126, 149, 158, 164, 168 rash 7–​9, 109, 126–​127, 149, 158, 166, 168 series 7–​9, 12, 17, 31, 109, 126–​127, 149, 158, 166, 168

set 7–​9, 36, 109, 126, 149, 158, 166, 168, 182 shoal 3, 7–​9, 12, 109, 126–​127, 147, 157, 171 swarm 7–​9, 109, 126–​127, 149, 158, 165, 168 troop 7–​9, 109, 126–​127, 149, 157–​158, 164–​165, 168

6 0 2

Index of subjects

Agree 37, 39, 53n36–​37, 53n42 agreement: Agreement Hierarchy 35, 43, 46, 132–​133, 135–​138, 177; canonical 33–​35; formal 1, 32–​33, 35, 38–​39, 43, 52n31, 130, 132, 186; grammatical 30, 35, 43, 52n31, 133; notional 30–​33, 35; semantic 32, 35–​37, 39, 43, 47, 53n42, 58, 133, 135–​136, 145–​146, 186; syntactic 32–​33, 35, 52n31 ambiguity 27, 51n13, 68, 79, 84, 90, 107, 149; structural 18–​20, 51n11, 51n21, 89 animacy 3–​4, 46, 54n55, 95–​96, 116, 128, 130, 145–​146, 155, 160, 164, 168–​170, 173–​174, 176, 178, 184–​186; Animacy Hierarchy 46, 145–​146, 160, 176; animate 46–​47, 95–​96, 124, 145–​146, 153, 155–​156, 160–​161, 163, 165–​166, 168, 184; (animate) human 2, 46–​47, 95–​96, 116, 128, 145–​146, 155, 160–​166, 168–​169, 176, 181n36, 181n39, 183–​184; (animate) non-​human 95, 116, 128, 145–​146, 160, 162, 165, 168, 176, 183; inanimate 11, 27, 46, 95, 116, 145–​146, 155–​156, 160–​163, 166, 168–​169 article 19–​20, 56, 59, 71–​72, 87, 93, 100n14, 100n17, 131; definite 19–​20, 59, 61, 66, 73, 77, 84–​85, 89, 93, 109–​110, 116–​117, 130–​131, 149, 177, 180n35; indefinite 12, 19–​20, 27, 37, 58, 66, 87, 93, 109–​110, 116, 128, 130–​132, 149, 152, 177, 180n35, 183 attraction 2–​4, 12–​13, 31, 43–​46, 53n51, 64, 74, 141 binominal 6, 8, 10n2, 14–​15, 17, 20, 23, 65–​66, 71, 82, 98; (noun) phrase 9, 13,

23, 48, 50n2, 50n4, 52n27, 55, 144; quantifier 49, 54n56, 62, 69 bleaching 27, 68, 90, 98 coextension 27, 90–​91; coextensiveness 18, 24, 40, 78, 81, 90, 149 collection 7, 12–​13, 18–​19, 25, 50n8, 64, 91, 151, 160, 180n34; collection noun 17–​19, 24–​25, 50n7 collexeme analysis 148, 153, 156–​159, 161–​162, 166, 168, 180n33 collocational range 8, 27, 77–​78, 81, 95, 98, 168 collostructional analysis 147–​148, 153, 157, 176–​177, 180n33, 184 complexity 3, 42–​45, 53n48–​50, 54n53, 112, 114, 120, 128–​129, 133, 136–​142, 173, 176–​177, 183, 185–​186; Complexity Principle 46, 138, 141, 177, 183–​184; structural 42, 135–​136, 138–​139, 177, 186; syntactic 133, 135–​139, 142, 177, 186–​187 concord 29–​30, 36, 48, 52n28; grammatical 29; notional 30; proximity 2, 30, 32 constructionalisation 27–​28 controller 33–​35, 38, 43, 70, 183 count 12, 15, 17–​18, 22, 49, 95; countability 128, 144, 146, 176, 181n44, 185–​186; non-​count 11, 15, 18, 22, 95; (un)countable 5–​6, 111, 115–​116, 128, 144, 176–​177, 180n34, 183 decategorialisation 9, 27, 58, 60, 65, 80, 91, 98 desemanticisation 8, 27, 61, 68, 71, 77, 90–​91 determiner 3–​4, 7, 19, 25, 36, 43, 51n10, 51n16, 51n19, 56–​58, 88, 101n27,

7 0 2

Index of subjects 207 106, 109–​110, 113, 115, 128–​133, 147–​148, 170, 176–​177, 183, 185–​186 distance 4, 30, 32, 35–​36, 42–​43, 53n51, 114, 128, 133, 135–​136, 138 English(es) 105, 172, 174, 176; American English 5–​7, 30, 41, 48–​49, 56, 64, 96, 100n8, 103–​105, 121, 124, 132, 168–​169, 171–​173, 178, 181n45–46, 184, 187; Australian English 49, 64, 169, 171, 173, 181n45–​46; British English 1, 6–​7, 11, 30, 37, 41, 47, 56, 64, 80, 100n8, 103–​105, 121, 124, 132, 168–​174, 176, 179n15, 180n36, 181n45–​46, 187; Canadian English 49, 169, 171–​173, 181n45–​46; inner-​circle 124, 169, 173, 176; Irish English 49, 169, 171–​174, 176, 181n45–​46; New Zealand English 49, 169, 171–​173, 181n45–​46; World Englishes 49, 124, 178, 187 entrenchment 26, 66, 80, 91 explicitness 44–​46, 138, 186 extension 27, 77–​78, 81, 95; metaphorical 91, 98, 165; metonymic 14, 27, 78, 92, 98 fixation 66, 72, 80, 91, 174, 176; syntactic 8–​9, 55, 58, 60–​62, 66, 68–​70, 72–​73, 75–​77, 79–​80, 82–​83, 85–​87, 98, 146–​148, 180n34, 183 grammaticalisation 8, 27–​28, 41, 48, 52n27, 55, 64–​65, 69, 73, 77, 79, 95, 101n25, 182, 185 grounding 20–​22, 51n14, 51n17 headedness 9, 13–​14, 20, 23–​26, 51n20, 103 idiom: idiomatic 9, 10n2, 13–​14, 22, 49, 55, 58, 62, 68–​70, 80, 87, 91–​92, 95, 98–​99, 101n20, 147, 149, 155, 183, 185–​186; idiomaticity 14, 90, 185; idiomatisation 8, 55, 61–​65, 68–​71, 73, 81, 85, 87, 89–​92, 94–​98, 102n28, 132, 146, 148, 186 lexical 5, 12–​13, 15, 19, 26, 28, 34, 36–​37, 52n32, 55, 61, 68, 89, 97–​98, 118–​119, 121, 125, 128, 147–​149, 151, 176–​177, 183–​184, 186–​187; function 60, 70; interpretation 26, 41; meaning 17, 22, 27, 58, 60–​61, 65, 68, 73–​74, 78, 87, 89–​90, 92–​93, 98–​99,

163, 165–​166; noun 9, 51n12, 60, 81, 88, 97–​98, 117, 132, 149–​152, 156, 158, 184; reading 26, 89, 130; use 10n2, 28, 61, 68, 77, 89, 98, 100n11, 152, 158, 163 locality 33, 35, 42–​43 marked 29, 45, 56–​57, 99n3, 106, 111, 115, 133–​136, 138, 142–​143, 147, 157, 169, 177, 184; markedness 45, 54n54, 133–​134, 138–​139, 142, 183, 186–​187; markedness effect 45, 54n54, 134, 177, 183; unmarked 106, 111, 114, 120, 128, 135, 137, 143–​144, 171, 177, 179n16, 181n44, 184; unmarkedness 135, 138–​139, 142, 177, 183–​184, 187 measure 17, 22, 25, 50n6, 65, 91; noun 17, 24–​25, 50n5, 50n7–​8; phrase 16, 19, 50n5 modification 112, 141–​142, 179n19; postmodification 112–​113, 137, 139, 141–​142, 177, 180n32; premodification 22, 44, 57–​58, 62, 66, 71–​72, 75–​76, 80, 82–​83, 85, 87, 110, 112, 120, 129, 180n32, 183 modifier 20, 36, 44, 57, 62, 72, 75, 98, 100n7, 100n11, 106, 112–​113, 140, 179n17, 179n19; postmodifier 3, 32, 44, 113–​114, 119–​120, 128–​129, 133–​134, 136–​139, 141–​142, 170, 177, 181n44; premodifier 56, 62, 66, 72, 75, 80, 82–​83, 85, 100n7, 101n21, 106, 112, 120, 170–​171, 176–​177 noun: mass 16–​18, 22, 111, 115, 134; number-​transparent 7, 31; oblique 5, 8–​9, 13, 43, 45, 88, 95, 106–​107, 109–​112, 114–​115, 121, 124, 128–​129, 132–​148, 153, 155, 157–​163, 165, 168–​171, 173, 176, 179n15, 180n34, 181n44, 182–​187 number: grammatical 29–​30; morphological 112, 134–​135, 176–​177, 185; notional 29–​30, 115; semantic 30, 115, 142–​144, 185 of-​: of-​collective 12; of-​ complement(ation) 7, 12–​13, 31, 127; of-​dependent 2, 4–​7, 12–​13, 41, 46, 88, 103, 105, 107, 134, 146, 176, 182–​183, 187; of-​phrase 64, 69, 79–​80, 112–​113; of-​PP 2–​5, 7–​8, 12, 14–​15, 19, 21, 23, 25, 27, 29, 43–​44, 48, 51n12, 52n22, 53n47, 55–​57, 64,

8 0 2

208  Index of subjects 69, 74, 93, 95, 98, 103–​104, 106, 135, 151, 176, 184–​187 partition 14, 20, 50n1, 84, 120, 147–​148, 153, 158, 173–​174, 176 partitive 14–​21, 23–​24, 26, 49, 50n4, 50n1–​2, 68–​69, 71, 74, 78, 88, 93–​95, 99, 116, 132, 148–​149, 151–​152, 183, 185; construction 14–​17, 19, 23, 62, 88, 95, 105, 152; meaning 15, 18–​19, 71, 73, 90, 93, 95, 151–​152; noun 12, 14, 50n3, 101n20; (noun) phrase 15, 20, 23, 26, 48–​49, 52n23, 99, 148, 152; reading 23, 69, 71, 88, 101n27; structure 15–​16, 20–​22, 25, 50n9, 58, 68, 71, 80, 82, 85, 88, 92, 95, 149, 151–​152; us(ag)e 14, 68–​69, 71, 73, 151 persistence 22, 97, 99, 183; conceptual 22; syntactic 22 plurality 1, 11–​12, 37, 46, 49, 80, 106, 135, 138, 144, 177, 184–​185; conceptual 46, 133–​134; notional 1; semantic 12, 30, 107, 128, 134–​136, 138, 142, 144, 146, 170, 176, 181n44 pseudopartitive 14–​26, 49, 50n2, 50n5, 50n9, 51n10, 51n20, 52n22–​23, 62, 68, 74, 78, 87–​89, 93, 95, 116–​117, 148–​149, 151–​153, 174, 183, 185; construction 14–​15, 17–​20, 23, 49, 51n13, 62, 71, 89, 94–​95, 99; (noun) phrase 19–​20, 23, 25–​26, 49, 132, 148, 150; structure 15, 17–​18, 20, 58, 68, 71, 80, 82, 85, 87–​88, 92–​93, 99, 102n27, 149, 153; use 68, 71, 130 quantification 20, 22, 51n14, 51n16, 64, 71, 87, 89, 91–​92, 97, 99, 101n26–​27, 153, 155, 183–​184; absolute 22, 51n16, 88–​89, 92, 153, 183; periphrastic 22, 185; potential 20, 87; relative 22–​24, 49, 51n16, 58, 88, 92–​93, 101n27, 151, 153, 183 quantificational 9, 13, 22, 50n9, 51n16, 58, 60, 65, 68, 77, 79, 91, 99, 101n20, 132, 151; interpretation 2, 19, 41, 61, 130; meaning 7, 12–​13, 58, 61, 69, 71, 74–​75, 87, 89–​92, 95, 98, 102n27; reading 19, 59, 61, 68–​69, 77–​78, 87–​90, 185; use 14, 19, 22–​23, 68, 75, 89, 149 quantifier 12, 15, 20–​22, 24–​25, 27, 49, 50n2, 51n16–​18, 52n27, 57, 60, 65, 68–​69, 74, 78–​80, 90–​91, 94,

99, 101n21, 101n27, 107, 109–​110, 116–​117, 130, 144, 149, 180n35; absolute 21–​22, 41, 49, 51n17–​18, 62, 78, 87, 89, 92–​94, 99, 102n28, 153, 183, 185; binominal see binominal quantifier; complex 7, 21, 32, 48, 153; interpretation 25–​26, 66, 132; meaning 27, 73, 89, 99; monomorphemic 16–​17, 20, 22; nominal 16, 20, 144; noun 17, 19, 24–​25, 50n7; periphrastic 22, 49, 52n26, 62, 66, 71, 88–​90, 95, 99, 102n28, 183; relative 21, 51n17, 71, 92–​94, 99; use 58, 65, 77 quantifying 12, 17, 27–​28, 65, 98, 130, 132, 180n34; collective 7, 12, 150; expression 8–​9, 12, 16–​17, 27, 54n56, 55, 58, 61, 70, 74, 77, 81, 87, 94–​96, 98, 102n28, 127, 150, 156, 160–​161, 183; function 27–​28, 65, 78; interpretation 95, 99, 117, 130, 148–​149; meaning 2, 4, 8–​9, 12, 22, 26, 28, 58, 64, 71, 78, 80–​81, 89–​91, 95, 98–​99, 100n11, 107, 151, 158, 185; potential 4–​5, 8–​9, 24, 49, 55, 101n26, 127, 149, 157, 159, 173, 178, 183–​185; reading 27, 88, 93, 132, 156, 183, 185; us(ag)e 5, 8, 13, 23, 48, 58, 60, 64, 69, 74, 78–​79, 84, 99, 100n16, 126–​127, 130, 132, 147, 149, 158–​159, 163, 174, 176–​177, 180n34, 183, 185, 187 quantity 7, 13, 16–​17, 20–​22, 27, 31, 41, 49, 59, 61–​62, 68, 75, 78–​79, 84, 88, 92–​94, 99, 130, 151, 164, 183 reanalysis 48, 74, 77, 161; syntactic 26–​27; semantic 27 recategorialisation 9, 27, 91, 98 referential 14, 16, 18–​19, 21, 49, 58, 87–​88, 97, 116–​117, 132, 148–​149, 151–​152; construction 20, 23, 27, 51n13; interpretation 41, 51n12, 77, 132, 152; meaning 19, 89; (noun) phrase 14, 19, 23–​24, 49, 52n23, 148, 151; reading 19–​20, 23–​24, 27, 61, 117, 130, 149; structure 25, 58, 88, 149, 151; use 10n2, 14–​15, 19, 58, 61, 65, 88–​89, 95, 101n26, 130, 132, 150–​151, 153, 156, 158 semantic opacity 8, 55, 58, 61, 68, 85, 87, 98, 146 target 33–​35, 37–​39, 43, 46, 132–​133, 135–​136, 186

9 0 2

Index of subjects 209 uncountable see count unmarked see marked; unmarkedness see marked variation: diatopic 171, 176, 178; formal 110, 120, 134, 181n44; regional 6, 30, 48, 53n38, 105, 121, 124, 168–​169, 176, 178, 184, 187

variety 4, 30, 48–​49, 52n28, 103, 117, 121, 126, 169–​174, 176, 181n42, 181n45–​46, 184; American 104, 121, 168–​170, 172–​173, 178; Australian 172; British 30, 48–​49, 169, 178, 184; inner-​ circle 49, 120, 169, 171–​173, 176, 178, 184; outer-​circle 49; regional 6, 69, 105, 119, 124, 128, 169, 173

0 1 2