Modals and Quasi-modals in English
LANGUAGE AND COMPUTERS: STUDIES IN PRACTICAL LINGUISTICS No 67
edited by Christia...
113 downloads
1842 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Modals and Quasi-modals in English
LANGUAGE AND COMPUTERS: STUDIES IN PRACTICAL LINGUISTICS No 67
edited by Christian Mair Charles F. Meyer Nelleke Oostdijk
Modals and Quasi-modals in English
Peter Collins
Amsterdam - New York, NY 2009
Cover design: Pier Post The paper on which this book is printed meets the requirements of "ISO 9706:1994, Information and documentation - Paper for documents Requirements for permanence". ISBN: 978-90-420-2532-5 ©Editions Rodopi B.V., Amsterdam - New York, NY 2009 Printed in The Netherlands
Contents Preface
ix
1.
Introduction
1
1.1 1.2 1.3 1.4 1.5
1 5 6 7 8
2
Theoretical preliminaries
11
2.1 2.2 2.3
11 12 15 16 17 18 18 20 21 21 21 23 24 25 25 27 28
2.4
2.5
2.6
3
The data An exercise in corpus linguistics Previous corpus-based and corpus-informed studies Diachronic studies of the modals and quasi-modals Regional and stylistic variation
Modality and mood The modal auxiliaries Quasi-modals 2.3.1 Properties of the semi-modals 2.3.2 Properties of the lexico-modals 2.3.3 Grammaticalization and idiomaticity 2.3.3.1 Grammaticalization 2.3.3.2 Idiomaticity 2.3.3.3 Selection of quasi-modals for analysis Modal meanings: approaches and issues 2.4.1 Classifying modal meanings 2.4.2 Polysemy vs monosemy 2.4.3 Negation and ‘temporality’ Further dimensions of modality 2.5.1 Modal strength 2.5.2 Degree of modality 2.5.3 Subjectivity/objectivity General aspects of the relationship between modals and quasimodals
Necessity and obligation 3.1
Must 3.1.1 3.1.2 3.1.3 3.1.4
Deontic must Epistemic must Dynamic must Time reference of situation
29 33 34 35 38 40 41
vi
Contents
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.1.5 Must and negation 3.1.6 Must: regional and stylistic variation Should 3.2.1 Deontic should 3.2.2 Epistemic should 3.2.3 Should and temporality 3.2.4 Should and negation 3.2.5 Special ‘low-degree’ uses of should 3.2.6 ‘Preterite’ should 3.2.7 Should: regional and stylistic variation Ought to 3.3.1 Deontic ought to 3.3.2 Epistemic ought to 3.3.3 Ought to and temporality 3.3.4 Ought to and negation 3.3.5 Ought to: regional and stylistic variation Need 3.4.1 Meanings of need 3.4.2 Need: regional and stylistic variation Have to 3.5.1 Deontic have to 3.5.2 Dynamic have to 3.5.3 Epistemic have to 3.5.4 Time reference of situation 3.5.5 Have to and negation 3.5.6 Have to: regional and stylistic variation Have got to 3.6.1 Deontic have got to 3.6.2 Dynamic have got to 3.6.3 Epistemic have got to 3.6.4 Have got to and temporality 3.6.5 Have got to and negation 3.6.6 Have got to: regional and stylistic variation Need to 3.7.1 Deontic need to 3.7.2 Epistemic need to 3.7.3 Dynamic need to 3.7.4 Need to and temporality 3.7.5 Need to and negation 3.7.6 Need to: regional and stylistic variation Had better 3.8.1 Meanings of had better 3.8.2 Had better: regional and stylistic variation Be supposed to 3.9.1 Meanings of be supposed to
43 43 44 45 46 47 48 48 51 52 52 53 55 56 56 56 57 57 59 59 60 62 63 64 66 67 68 69 70 70 71 72 72 73 73 75 75 75 76 76 77 77 79 80 80
Contents
3.10
3.11
4
3.9.2 Be supposed to: regional and stylistic variation Be to 3.10.1 Meanings of be to 3.10.2 Be to: regional and stylistic variation Be bound to 3.11.1 Meanings of be bound to 3.11.2 Be bound to: regional and stylistic variation
Possibility, permission and ability 4.1
4.2
4.3
4.4
5
vii
May 4.1.1 4.1.2 4.1.3 4.1.4 Can 4.2.1 4.2.2 4.2.3
Epistemic may Deontic may Dynamic may May: regional and stylistic variation
Epistemic can Deontic can Dynamic can 4.2.3.1 Theoretical possibility 4.2.3.2 Ability 4.2.3.3 Dynamic implication 4.2.4 Can: regional and stylistic variation Might and could 4.3.1 Uses of might and could as preterite forms 4.3.1.1 Temporal uses 4.3.1.2 Hypothetical uses 4.3.2 Meanings of might and could 4.3.2.1 Epistemic possibility 4.3.2.2 Dynamic possibility 4.3.2.3 Deontic possibility 4.3.3 May and might: one lexeme or two? 4.3.4 Might and could: regional and stylistic variation Be able to 4.4.1 Meanings of be able to 4.4.2 Be able to: regional and stylistic variation
Prediction and volition 5.1
Will 5.1.1 5.1.2 5.1.3 5.1.4
Epistemic will Dynamic will Deontic will Will: regional and stylistic variation
82 83 83 86 86 87 88 91 92 92 95 95 96 97 98 100 101 101 103 104 104 105 105 105 106 108 109 113 117 117 118 119 119 122 125 126 126 131 134 135
viii
Contents 5.2
5.3
5.4
5.5
5.6
6
Shall 5.2.1 Epistemic shall 5.2.2 Dynamic shall 5.2.3 Deontic shall 5.2.4 Shall: regional and stylistic variation Would 5.3.1 Temporal would 5.3.2 Hypothetical would 5.3.3 Would: regional and stylistic variation Be going to 5.4.1 Epistemic be going to 5.4.2 Dynamic be going to 5.4.3 Deontic be going to 5.4.4 Be going to: regional and stylistic variation Want to 5.5.1 Dynamic want to 5.5.2 Deontic want to 5.5.3 Epistemic want to 5.5.4 Want to: regional and stylistic variation Be about to 5.6.1 Meanings of be about to 5.6.2 Be about to: regional and stylistic variation
Conclusion
135 136 137 137 138 139 140 141 143 143 144 147 148 149 150 152 152 153 153 155 155 156 159
Appendix
163
References
185
Index
191
Preface This book represents the culmination of a long period of interest in modal expressions in English. My first foray into the area was an MA(Hons) dissertation on “Can and may in Shakespeare’s English” (1978). The second followed my participation in the project, directed by Pam Peters of Macquarie University, to compile the Australian Corpus of English, resulting in several publications in which I used data from the corpus to explore the semantics of the modals in Australian English (Collins 1988, 1991a, 1991b). My interest was rekindled yet again by the International Conference on Modality in Contemporary English at the University of Verona in September 2001, following which I determined to undertake a more comprehensive corpus-based study. This book reports the results of the research I have conducted over this period. I wish to thank Christian Mair from the editorial committee for this Rodopi series for his welcome encouragement, Eric van Broekhuizen for his advice on technical matters, Maria Oujo of the University of NSW for her assistance with editing the manuscript, and Pam Peters and Adam Smith for facilitating access to some of the corpora. The work has benefited from discussions with many linguists over the years, including Rodney Huddleston, Pieter Seuren, Robert Eagleson, Edgar Schneider, Geoffrey Leech, Stig Johansson, and Michael Halliday. Last but not least I wish to thank my family for surrendering the dining room table to my piles of books, papers and data. Without your support I could never have completed this project. Peter Collins Sydney, August 2008
Chapter 1 Introduction This book reports the findings of a corpus-based study of the meanings of the modal auxiliaries (or simply ‘modals’) and a set of semantically related ‘quasimodals’ in three parallel corpora of contemporary British English (‘BrE’), American English (‘AmE’) and Australian English (‘AusE’). The study is the largest and most comprehensive yet attempted in this area, based on an analysis of every token of the modals and quasi-modals (a total of 46,121) across the spoken and written data. The selection of corpora was designed to facilitate the exploration of both dialectal variation (across the three World Englishes examined), of stylistic variation (across the spoken and written components of the corpora and their dialogic/monologic and non-printed/printed subcategories respectively), and – albeit more indirectly – of diachronic variation involving the modals and quasi-modals. 1.1
The data
The three corpora used in the study were the British component of the International Corpus of English (‘ICE-GB’), the Australian component of the International Corpus of English (‘ICE-AUS’), and a specially assembled corpus of American English (‘C-US’).1 While ICE-GB and ICE-AUS are members of the collection of parallel million-word corpora of the International Corpus of English (‘ICE’), C-US is designed to fill the gap caused by the non-availability hitherto of an actual ICE-US corpus (see further below). ICE is a collection (not yet complete) of million-word corpora representing national and regional varieties of English which provides a resource for linguists pursuing comparative research into English worldwide (see for example the ICE-based studies reported in the special issue of World Englishes 23: 2, 2004). Prior to ICE the only parallel corpora available for such comparative research were the Brown Corpus of American English (‘Brown’) and the Lancaster/Oslo-Bergen Corpus of British English (‘LOB’), both consisting exclusively of printed material. Initiating the ICE Project in 1988 (in a short notice in World Englishes) the late Sidney Greenbaum acknowledged the contribution of Brown and LOB, and proposed that their scope be extended by representing standard varieties from other countries and national varieties where English is an additional official language, and by including spoken and manuscript English. The project resulting from this initiative has as its long-term goal the preparation of up to twenty one million corpora (with a number of these already complete and available for scholarly use).2
2
Chapter 1
Each ICE corpus samples the English of adults (aged 18 or above) who have been educated through the medium of English to at least the end of secondary schooling. The texts generally date from 1990 to 1994 inclusive; that is, it is during this period that the printed texts were originally published, the spoken texts originally recorded, and the handwritten texts composed. The corpora are intended to be representative of the English used in each country, so a wide range of the social variables that define the population are included (sex, age, and regional background in particular). To ensure comparability, each ICE corpus conforms to a common design, comprising 500 texts, each of 2,000words. The main division is between the primary modes of speech and writing, with 300 spoken texts (600,000 words) and 200 written texts (400,000 words). It should be noted that 50 of the spoken texts are scripted, and thus have a combination of spoken and written attributes, enabling them to provide a bridge between the two modes. The ICE text categories are shown in Table 1.1 below: Table 1.1. Composition of the ICE corpora Spoken (300)
Dialogues (180) Monologues (120)
Written (200)
Non-printed (50) Printed (150)
S1A: Private (100) S1B: Public (80) S2A: Unscripted (70) S2B: Scripted (50) W1A: Student writing (20) W1B: Letters (30) W2A: Academic (40) W2B: Popular (40) W2C: Reportage (20) W2D: Instructional (20) W2E: Persuasive (10) W2F: Creative (20)
The spoken texts are all recorded non-surreptitiously, and in order to eliminate the possibility of distortion of data resulting from self-consciousness the first fifteen minutes of recorded conversations are not used. The dialogues that are private (S1A: face-to-face conversations and telephone calls) differ from those that are public (S1B: class lessons, parliamentary debates, business transactions, etc.) in that the latter are intended for an audience comprising those who are present but not participating in the exchange. Furthermore, while in the public dialogues the topic is known in advance, the private dialogues range over unprepared topics. The unscripted monologues (S2A: commentaries, lectures, demonstrations, etc.) may in some cases involve an element of planning and use of prepared notes, but the scripted monologues (S2B: broadcast news, speeches, etc.) are required to be fully prepared. Nevertheless it is the spoken version of these texts that is used, on the grounds that even a close reading will contain such speech phenomena as false starts, hesitations and pauses.
Introduction
3
The chief distinction in the written component is between non-printed (or, manuscript) texts (W1) and printed texts (W2). The two types differ in their intended readership and mode of composition. The non-printed material (e.g. examination scripts, letters) is typically intended for a single reader, whereas the printed material is produced for a large and unrestricted audience unknown to the writer. The non-printed material is usually not edited by others, whereas the printed material normally complies with a house style and may be edited and revised by different people. The printed texts represent a diversity of registers. The academic, popular and press reportage categories (W2A–W2C) all have a primarily informational function, by comparison with the instructional role of administrative/regulatory writing and publications on skills and hobbies (W2D), the persuasive function of press editorials (W2E), and the creative function of novels and stories (W2F). C-US comprises 196,458 words of spoken and written material in equal proportions to the spoken and written components of the ICE corpora. The C-US texts are taken from two sources. For the spoken component the Santa Barbara Corpus (‘SBC’), which consists of texts recorded in the 1990s and was compiled with a view to inclusion in ICE-US, was selected.3 At the time the present analysis was conducted, only Parts 1 and 2 were available, and the transcript of one of the texts (SBC-006) was missing. The 29 texts used contain a total of 116,458 words (this count determined by stripping out all but orthographic words from the transcripts). Of the texts 23 are dialogic (casual conversations, discussions, meetings, task related interactions, etc.) and six are monologic (three lectures, one with a small amount of interaction, and three sermons). There is unfortunately a degree of generic incomparability between the SBC data used and the spoken ICE data: in the former monologic texts represent approximately 20% of all spoken texts, as opposed to 40% in the ICE corpora.4 For the written component of C-US, a selection of forty of the 2,000-word text categories (80,000 words) were extracted from the Freiburg-Brown Corpus of Written American English (‘Frown’). The Frown corpus, which was built to match the Brown corpus as closely as possible in size and composition, comprises texts published in 1991, three decades after the Brown texts. Both corpora contain one million words in 500 text samples representing press, ‘general’, learned and fictional writing. The selection of Frown texts for the present study was made to match as closely as possible the ICE categories: see Table 1.2. Again there is, inevitably, some incomparability with the ICE corpora, due mainly in this case to the absence of non-printed texts in Frown. For instance the letters in Frown Category G, being printed, are generally less informal than the letters in the ICE corpora, and the texts from Category P of Frown (which contain a large amount of dialogue) resemble, but do not perfectly parallel, the ICE manuscript texts (with their speech-like properties).
4
Chapter 1
Table 1.2. Parallels between C-US and the ICE corpora
Spoken
Written
ICE Text type Dialogue Monologue Total Non-printed
No. of words 360,000 240,000 600,000 100,000
Printed: 200,000 informational
Printed: 40,000 instructional
Printed: persuasive Printed: creative Total
20,000 40,000 400,000
C-US Text type SBC Total Frown G1-3 (Belleslettres, biography, etc.); P1-7 (Romance and love stories) Frown J1-8 (Learned and scientific); Frown F1-8 (Popular lore); A1-4 (Press reportage) Frown H1-2 (Government documents, etc.); E12 (Skills, trades and hobbies) Frown B1-2 (Press editorials) Frown K1-4 (General fiction) Total
No. of words
116,458 20,000
40,000
8,000
4,000 8,000 80,000
All frequencies for C-US were normalized to tokens per one million words, to match those for ICE-GB and ICE-AUS (by dividing the raw frequency count by the number of words in C-US and then multiplying by 1,000,000 (i.e. dividing by 0.196458). In addition, frequencies for the spoken and written subcategories of ICE-GB and ICE-AUS were normalized to tokens per one million words.5 We finish this section by presenting the modal expressions examined in the study along with their raw frequencies (see Table 1.3). The list comprises all the central and marginal modal auxiliaries with the exception of dare and used to (excluded on semantic grounds), and a set of quasi-modals selected on the basis of criteria which are discussed in the next chapter. Each of the expressions has a number of variant forms (as noted in Chapters 3–5) and can/could, may/might, shall/should, and will/would are usually regarded as present/preterite forms of single lexemes.
Introduction
5
Table 1.3. Raw frequencies of the modals and quasi-modals Modals can could may might must need ought to shall should will would Total
1.2
7,663 3,557 2,261 1,499 1,367 56 126 343 2,432 8,505 7,775 35,584
Quasi-modals be able to be about to be bound to be going to be supposed to be to had better have got to have to need to want to Total
889 124 27 2,721 171 371 89 705 2,827 716 1,897 10,537
An exercise in corpus linguistics
The study is an exercise in corpus linguistics. The present section explores the implications of this claim, while the next section overviews previous corpusbased studies of the modals. There is consensus amongst corpus linguists that the use of corpus-derived data enables the researcher to confront the inescapable danger of subjective bias in studies based on introspectively-derived examples. However linguists who make use of corpus data do so in a number of ways, and these tend to be associated with their attitudes towards the nature and status of corpus linguistics. The work of those who use corpus data merely for exemplificatory purposes and in order to determine grammaticality (as in the comprehensive grammars of Quirk et al. 1985 and Huddleston and Pullum 2002, and in the work on the modals by Palmer 1990) is merely an extension of the tradition of using textual data that predates the modern interest in computer corpora (as in Jespersen 1909– 1949, Poutsma 1926–1929). Such ‘corpus-informed’ research may be distinguished from ‘corpus-based’ research, in which the corpus is not simply a source of exemplificatory data but more importantly of frequency data. The latter, typically combined with a commitment to the notion of ‘total accountability’, may influence hypotheses applied to the data, or formulated on the basis of it. Within the reference grammar tradition, Biber et al. (1999) is an example and, within the study of the modals, Coates (1983). More recently, some corpus linguists have espoused the view that corpora can be used to redefine linguistic categories and concepts, and to build up new theories in linguistics. An example is the work of researchers from the University of Birmingham (e.g. Hunston and Francis 2000) on ‘pattern grammar’ that has prompted a reconsideration of the division between grammar and lexis. Proponents, who sometimes refer to their approach as ‘corpus-driven’ (see Römer
6
Chapter 1
2006), regard corpus linguistics as a separate domain of study within linguistics, rather than merely a methodology (e.g. Leech 1992, Kennedy 1998). The present study is corpus-based. Corpus data are used not merely for illustrative purposes but as a source of quantitative information relevant not only to the semantic framework and categorization system adopted, but also to the patterns of regional and diachronic variation examined. The study is not an exercise in corpus-driven linguistics: it seeks to provide a fresh integration of insights from a variety of sources – including semantic theories of modality, work on regional variation in the ‘World Englishes’ tradition, and research on diachronic syntax – but does not pretend to offer a new theory of modality driven by the data. 1.3
Previous corpus-based and corpus-informed studies
Coates (1983) remains the most detailed and widely referred to corpus-based study of the English modals. Data are taken from two British corpora, the one million-word Lancaster corpus (before it was superseded by LOB), and 545,000 words from the Survey of English Usage corpus (representing the spoken, written-to-be-spoken, and manuscript categories). From these Coates prepared “[a] representative sample of each modal in each corpus (…) each sample consisting of approximately 200 cases” (1983: 2). This method allows Coates to examine the frequencies for the meanings and uses of particular modals but not the relative frequencies of the modals. Two studies of the English modals which predate Coates use data from the American Brown corpus: Ehrman (1966) is based on one-third of Brown, Hermerén (1978) on four out of the 15 categories of Brown. There are two studies which provide quantitative information about the modals, the Biber et al. (1999) reference grammar, and Mindt (1995), but the usefulness of the information presented is limited in both cases by the failure of the authors to provide specific frequency figures. While the studies described thus far are corpus-based, there are others that are merely informed by corpus data. The best known of these is Palmer (1990), which makes use of data from the Survey of English Usage corpus “for heuristic and exemplificatory purposes only” (1990: 29). Another significant study which uses the same corpus as a source of illustrative data is Westney (1995). Finally, there are a number of smaller studies of individual modal expressions, or sets thereof, that use corpus data. There is not space to describe these in detail: a brief and selective overview follows. Berglund (1997) examines the ‘expressions of future’ will, shall and be going to in various corpora of BrE (LOB; and the London-Lund Corpus, or ‘LLC’), AmE (Brown), and Indian English (the Kolhapur Corpus). Berglund’s (1999) sociolinguistic study of be going to is based on the British National Corpus (‘BNC’), as is Verplaetse’s (2003) functional study of want to. Facchinetti uses data from ICE-GB in her studies of be able to (2000), can/could (2002) and may (2003). Nokkonen’s (2006) study of need to is based on four British corpora: LOB, LLC, FLOB (the Freiburg-LOB Corpus) and COLT (the Bergen Corpus of London Teenage
Introduction
7
Language). Further quantitative studies, with a diachronic orientation, are described in the next section. 1.4
Diachronic studies of the modals and quasi-modals
Figures presented by Leech (2003), Smith (2003), and Mair and Leech (2006) show that BrE and AmE have seen, in the three decades spanning the early 1960s to the early 1990s, a rise in the frequency of the quasi-modals with a concomitant and related decline in the frequency of the modal auxiliaries.6 Table 1.4 below reproduces Mair and Leech’s figures for written BrE and AmE for the items that are investigated in the present study (listed in alphabetical order), determined by calculating the difference between the frequencies derived from their 1960s corpora (LOB and Brown) and from their 1990s corpora (FLOB and Frown), as a percentage of the former.7 Table 1.4. Changes in the frequency of modals and quasi-modals in recent British and American writing (from Mair and Leech 2006)
can could may might must need ought to shall should will would
Modals BrE +2.2% +2.4% –17.4% –15.1% –29.0% –40.2% –44.2% –43.7% –11.8% –2.7% –11.0%
AmE –1.5%% –6.8% –32.4% –4.5% –34.4% –12.5% –30.0% –43.8% –13.5% –11.1% –6.1%
be going to be to had better have got to have to need to be supposed to want to
Quasi-modals BrE AmE –1.2% +51.6% –17.2% –40.1% –26.0% –17.1% –34.1% +15.6% +9.0% +1.1% +249.1% +123.2% +113.6% +6.3% +18.5% +70.9%
Mair and Leech regret that they are not in a position to provide detailed information about spoken BrE and AmE, given the tendency for cutting edge innovations to be associated with speech rather than writing. Nevertheless they report that a search of spoken corpora of BrE covering a similar period of time shows the trends found for writing to be more pronounced in speech (compare an overall rise of 10.0% of the quasi-modals in British writing with one of 36.1% in British speech, and an overall fall of 9.5% of the modals in British writing with one of 17.3% in British speech). The unavailability of comparable corpora for AmE speech deprives Mair and Leech of the opportunity to provide parallel American figures. However they refer to the results a search of the 4-millionword Longman Corpus of Spoken American English (from the 1990s) which indicated that quasi-modals were 62.5% as frequent as core modals (compared with a figure of 17% for written corpora of AmE and BrE of the same era). This
8
Chapter 1
finding, Mair and Leech conclude, “suggests that, as is often suspected, the spoken American variety of the language is the main driving force of change in this area, as presumably in others, and places the encroachment of semi-modals on the territory of the modals in AmE speech, in frequency terms, beyond doubt” (2006: 328). The present study is synchronic rather than diachronic. Nevertheless a number of the findings have diachronic implications, including the varying frequencies across speech and writing for particular modal expressions, and for different regional varieties. For this reason the frequencies reported in Table 1.4 will be regularly invoked throughout Chapters 3–5. 1.5
Regional and stylistic variation
The availability of matching regional corpora containing both spoken and written material (with identical categories in the case of ICE-GB and ICE-AUS, and similar in the case of C-US) provided the basis not only for a more detailed and comprehensive analysis of the modals and quasi-modals than that offered in previous quantitative studies, but also for the exploration of patterns of dialectal and stylistic variation. The Appendix contains detailed tables for each item with frequencies for their meanings in the spoken texts (subdivided into dialogic and monologic in the ICE corpora) and written texts (subdivided into non-printed and printed) of the three regional corpora. It is widely assumed that there are only minimal grammatical differences between the regional varieties of standard English (cf. Quirk et al. 1985: 18-19). One aim of this study is to test the validity of this assumption and to determine – if it view turns out to be wrong – how extensive the regional differences are and how they are they manifested in the frequency of particular forms and their meanings. What types of innovation or conservatism are reflected in the varying modes of usage? Of the two long-established and influential varieties, BrE and AmE, is one consistently ‘leading the way’ in the changes that are under way with the modals and quasi-modals? And what of the younger Antipodean dialect, AusE? Does it exhibit ‘colonial lag’ (cf. Trudgill 1986: 130), or is there evidence of linguistic individualism, with patterns that are neither clearly British nor American? This study evidenced a higher frequency of quasi-modals and a lower frequency of modals in the American data than in the British or Australian, justifying Leech’s identification of ‘Americanization’ (a shift towards AmE in the frequency of linguistic categories) as a dissemination process that is driving developments in this domain. A second aim is to test Biber et al.’s (1999: 20) claim that “grammatical differences across registers are more extensive than across dialects”. As noted in the previous section, Mair and Leech (2006: 327) report a more exaggerated pattern of change with the modals and quasi-modals in speech than in writing (a steeper fall for the modals and rise for the quasi-modals). The database used for the present study, with its spoken versus written subdivision (and further genre subclasses for ICE-AUS and ICE-GB), facilitated examination of stylistic
Introduction
9
variation. Relevant here, but not testable with the present data, is the possible influence of ‘colloquialization’ (the shift towards acceptability and use of informal linguistic features associated with private conversation in public, official, discourse) on the increasing use in writing of the quasi-modals. The present study provides ample evidence that stylistic factors are at play in developments reported, with the quasi-modals flourishing in speech, their modal counterparts maintaining a penchant for the written word. Furthermore there is a connection between the regional and the stylistic: it is in AmE that the stylistic gulf between semi-modals and modals tends to be most marked, and least in BrE. Notes 1
I wish to thank to Edgar Schneider for suggesting the label ‘C-US’, and Pam Peters for providing me with access to ICE-AUS.
2
See Nelson, Wallis and Aarts (2002) for further information on the design of the ICE corpora, and the ICE website for the latest information on which corpora are available and how to obtain copies: www.ucl.ac.uk/english-usage/ice
3
For more information on SBC, visit: www.linguistics.ucsb.edu/research/sbcorpus.html
4
The (eagerly anticipated) completion of ICE-US will circumvent such problems and provide researchers in the future with the means to make more accurate regional comparisons.
5
Normalized frequencies for the spoken component of C-US were obtained by dividing raw frequencies by 0.116458, and for the written component by dividing raw frequencies by 0.08. For general details of the methodology used see Biber, Conrad and Reppen (1998: 263-264).
6
Another study which contains useful diachronic information, on have to, have got to, and want to in BrE and AmE, is Krug (2000).
7
The figures, from Table 14.3 on p.327 and Table 14.4 in Mair and Leech (2006: 327-328) are based largely on those reported in Leech (2003).
Chapter 2 Theoretical preliminaries This chapter is concerned with the categories and dimensions of modality that inform the study. We begin by defining modality and examining the properties of the class of modal auxiliaries, and those of the more heterogeneous set of quasiauxiliaries. We then move on to the classification of modal meanings, and the dimensions of strength, degree and subjectivity. 2.1
Modality and mood
Modality embraces a range of semantic notions, including possibility, necessity, ability, obligation, permission, and hypotheticality. What they have in common is that they all involve some kind of non-factuality: a situation is represented not as a straightforward fact, as not being known. Some have sought a more positive characterization by tying modal meaning to the fundamental concepts – which constitute the primary concern of modal logic – of necessity and possibility (e.g. Van der Auwera and Plungian 1998, Huddleston and Pullum 2002: 173), so that it is envisaged as involving “a speaker’s judgment that a proposition is possibly or necessarily true or that the actualization of a situation is necessary or possible” (Depraetere and Reed 2006: 269). The reference to “actualization” here is important: as we shall see in Section 2.4.1 many consider there to be two fundamentally different kinds of modality – often labelled ‘epistemic’ and ‘root’ – whose orientation is towards truth and actualization respectively (but which are united in their mutual association with the concepts of necessity and possibility). The grammatical realization of modality via verb inflections is known as ‘mood’. The most distinctive is the subjunctive mood, which is used to refer to a situation in a possible world, and includes both what is traditionally called the ‘present subjunctive’ (the base form of the verb as in God save the Queen and The chairman demanded that the motion be withdrawn) and the ‘past subjunctive’ (essentially were with a first or third person subject as in I wish I were fitter). In English today inflectional systems have but a minor role to play in the expression of modality. The principal means of expressing modality in English is the class of modal auxiliaries and a set of periphrastic expressions, referred to in this book as ‘quasi-modals’, of the type have to, be going to and be to.1 It is these that form the focus of this book. The next two sections are devoted to considering their grammatical properties.
12
Chapter 2
2.2
The modal auxiliaries
As members of the larger class of auxiliaries (which also includes the ‘primary auxiliaries’ be, have, and do), the modal auxiliaries exhibit a number of inflectional and syntactic properties that distinguish them from main, or ‘lexical’, verbs. Let us begin with a provisional list of modals: can, may, will, shall, must, ought to, need, dare and used to.2 The first four are generally assumed to have preterite forms (could, might, would, and should), but as noted below this assumption raises some problematical issues with might and should. Need, dare and used to may be auxiliaries (as in Need/Dare/Used she (to) stay away?) or lexical verbs (as in Did she need/dare/used to stay away?). The modals are, like all auxiliaries, set apart from lexical verbs in the availability of inflectional negative forms with n’t (can’t, mightn’t, mustn’t etc.).3 Dare and used to are excluded from this study on the grounds of their semantic dissimilarity to other modals and quasi-modals, dare expressing the courage of the subject-referent, and used to expressing aspectual rather than modal meaning. Syntactically the modals, like all auxiliaries, differ from lexical verbs in their capacity to be used in the four so-called ‘NICE’ constructions: negation, inversion (of subject and auxiliary), code (post-verbal ellipsis dependent for its interpretation upon previous context), and emphasis (emphatic polarity involving the use of contrastive stress). The constructed examples in (1) below illustrate the use of auxiliary will in the four constructions, contrasting with the impossibility of using the lexical verb like: (1)
a. He will not study. b. Will he study? c. She will study, and so will he. d. They think he won’t study, but he WILL.
*He likes not study. [Negation] *Likes he study? [Inversion] *She likes to study, and so does he. [Code] *They think he doesn’t like studying, but he LIKES. [Emphasis]
The capacity to serve as ‘operator’ in the NICE constructions does not provide sufficient grounds for defining the class of auxiliaries, insofar as there are some lexical verbs that also have this capacity (the copula be as in Is he Malaysian?, and ‘possessive’ have as in He hasn’t any money). 4 In addition to the above-mentioned inflectional and syntactic properties of auxiliaries, the properties listed below are distinctive to the modals.5 Can, may, will, and shall have all these properties, must, ought to, need and dare have most but not all of them, while the quasi-modals have none.
Theoretical preliminaries
13
i. No non-tensed forms The modals are morphologically defective, only having tensed forms (i.e. no bare infinitival or participial forms). The examples in (2), (3) and (4) show that must, will, and can are modals, but the semantically similar quasi-modals have to, be able to, and be going to are not. (2) (3) (4)
I’d prefer to *must/have to wait another month. I resent not *canning/being able to wait another month. I had *would/been going to wait another month.
ii. No person-number agreement The modals exhibit further morphological defectiveness in not having a separate 3rd person singular present tense form like other verbs. This enables them to agree with any kind of subject. Compare for example the agreement restrictions that apply to the quasi-modal have to but not the modal must: He must/has to/*have to wait another month; You must/*has to/have to wait another month; I must/*has to/have to wait another month. iii. Bare infinitival Apart from ought to, the modals take a bare infinitival. There are some lexical verbs that can also take a bare infinitival, but in such cases the infinitive usually follows a direct object. Compare the lexical verbs make and watch in I make/watch them train hard with the modal must in They must train hard. iv. Unreal conditionals In an unreal conditional the first verb of the apodosis must be a modal. This role is served by modal would and could in If he won the lottery, he would/could buy a Ferrari, but cannot be served by quasi-modal was going to or was able to, as shown by the ungrammaticality of *If he won the lottery, he was going to/was able to buy a Ferrari. v. Unreal preterite As Huddleston and Pullum (2002: 107) point out: “The preterites of the modal auxiliaries – could, might, would, should – can be used with the modal remoteness meaning without the grammatical restrictions that apply in the case of other verbs, where it is found in only a small set of subordinate constructions.” Thus the preterite modal could but not the preterite quasi-modal were able to can express non-past hypotheticality in Could/Were you able to you help her? Were
14
Chapter 2
able to can only express this meaning in subordinate clauses such as those in If you were able to help us we would be grateful and We wish you were able to help us. The various properties discussed above may be used to draw a distinction between what Quirk et al. (1985: 137) refer to as ‘central’ modals (can, will, may, shall, must) and ‘marginal’ modals (dare, need, ought to, used to). Even within the central modals we may suggest a distinction between can and will, whose satisfaction of the properties is straightforward, and must, may and shall, for which it is more patchy. In addition to the properties discussed above both can and will have reduced forms (phonologically reduced forms when unstressed and, in the case of will, the clitic ’ll). It is presumably the availability of such forms that accounts in large part for the high frequency of can and will in spoken English (see Sections 4.2.4 and 5.1.4). It is furthermore likely that their syntactic straightforwardness is a factor in the numerical supremacy that they enjoy over the other modals (see Section 1.1), and their diachronic success in resisting incursions into their semantic territory (see Chapter 6). Consider next must, may and shall. Must lacks a preterite form, so the ‘unreal preterite’ property is not applicable, and it is marginal in the apodosis of an unreal conditional (e.g. ?If you caught the 11pm bus you must surely be home by midnight). May is non-prototypical in its lack of a phonologically reduced form and in the impossibility of mayn’t for most speakers, shall in the limited currency of shan’t outside BrE. The most telling difference between may and shall on the one hand, and can and will on the other, is the lack of a clear temporally-based relationship with the preterite forms might and should. For many speakers might no longer represents the preterite form of may, but rather is the present form of a separate lexeme (one which, like must, does not have a preterite): see Section 4.3.3 below. Should has several specialized uses which have no counterpart in shall (see Section 3.2.5). The marginal modals are less straightforward still. Ought to differs from the other modals in its normal requirement of a to-infinitive (which is only rarely relaxed, and only in non-affirmative contexts as in Ought we not apologise? and You ought not complain so much), like must it has no preterite, and it is very marginal in the apodosis of an unreal conditional (e.g. ?If we left at lunchtime we ought surely to arrive before sunset). 6 Need and dare can be either modal auxiliaries or lexical verbs. As modals they are restricted to non-affirmative contexts (e.g. She needn’t/daren’t apply; Need/Dare she apply?) where their only irregular features are an absence of reduced forms and, for auxiliary need, the lack of a preterite counterpart: *Needed she apply?). Used to can also be either a lexical verb or a modal auxiliary, qualifying as the latter formally but not semantically. It is the most marginal member of the modal auxiliary class, with many speakers finding sentences such as They usedn’t to eat meat and Used they to eat meat? unacceptable, and morphologically very anomalous in its lack of present tense and participial forms.
Theoretical preliminaries 2.3
15
Quasi-modals
Modern English has an abundance of periphrastic modal forms, a somewhat loosely-defined grouping formally distinguishable from, but semantically similar to, the modal auxiliaries, and which have been relatively neglected in the literature on modality. The term that is widely used for these expressions is ‘quasi-modals’ (e.g. by Hakutani and Hargis 1972, Chapin 1973, Coates 1983, Perkins 1983, and Westney 1995). Opinions differ as to the nature of the relationship between the modals and quasi-modals. It is generally agreed that a number of quasi-modals serve suppletive roles in the defective morphological paradigms of the modals (e.g. preterite had to in the absence of *musted, and infinitival to be able to in the absence of *to can). It is also generally agreed that there is a close semantic affiliation between a number of quasi-modals and modals, while at the same time it is recognized that there are shades of difference between them (e.g. have to tends to be more ‘objective’ than must; an implication of actualization differentiates was able to from could). Some have attempted to apply semantic distinctions to modals and quasimodals generally, rather than simply to individual pairings. According to Lakoff (1972: 240), it is the presence or absence of speaker involvement that distinguishes the modal from the quasi-modal in the pairs must: have to, may: be allowed to, will: be to, and should: be supposed to. Her suggestion draws support from Larkin (1976) and Palmer (1990: 81). There was no shortage in the data of examples such as (5) and (6), which suggest that Lakoff’s distinction is at least in need of qualification: (5) (6)
You or your partner must be engaged in remunerative work for an average of at least 24 hours a week. (ICE-GB W2D-005 22) You’ll have to give me the money for that ’cause I’m not going to pay for you. (ICE-AUS S1A-017 278)
In (5) the modal must is used without speaker involvement, in the objective statement of a rule, while in (6) quasi-modal have to is used to convey a speakersourced directive. It is furthermore difficult to see how Lakoff’s speaker-relatedness distinction could in any way bear on the actualization implicature that distinguishes many uses of be able to from can, as in (7), where could would not be a suitable substitute for was able to (?Lansbury raised himself up and could address the throng). (7)
Lansbury raised himself up and was able to address the throng through his cell window. (ICE-GB W2B-019 85)
Westney (1995: 11) invokes three criteria for determining whether a complex verb qualifies as a periphrastic (or ‘quasi-modal’ in our terms):
16
Chapter 2
grammaticalization (displaying features which suggest membership of a significant grouping that lacks categorial status), idiomaticity (expressing a meaning that transcends the meanings of its constituent parts), and semantic relatedness to a central modal auxiliary. Semantic relatedness may be established via various tests of equivalence. For example, the utterance in (8) would be compatible with a reinforcing continuation such as he’s got to, and that in (9) with she must, suggesting that they are equivalent in strength. On the other hand, the oddity of she should or she ought to as continuations for (9) suggests that should and ought to differ in strength from have got to. (8) (9)
There is a crisis and he must act now (ICE-GB S1B 056 89) She’s got to get it organised before she goes to hospital for her knee (ICEAUS S1A-016 25)
We discuss grammaticalization and (more briefly) idiomaticity in Section 2.3.3 below, while semantic relatedness is invoked at various points in the following discussion and explored in detail in Chapters 3–5. Within the set of quasi-modals we can distinguish between those which have an auxiliary as their first element, and those which do not (and are in most cases formed with be and a lexical item). Quirk et al. (1985: 137-146) distinguish these as respectively ‘modal idioms’ (had better, would rather, be to, have got to; plus the ‘less common idioms’ had best, would sooner/(just) as soon, may/might (just) as well), and ‘semi-auxiliaries’ (have to, be (un)able to, be about to, be bound to, be going to, be obliged to, be supposed to, be (un)willing to; plus be apt to, be due to, be likely to, be meant to). Quirk et al.’s terms are not at all transparent, so instead of ‘modal idioms’ I shall use ‘semi-modals’ (both because this term is more familiar in the literature and in order to suggest that this class shares more properties in common with the modal auxiliaries than does the second class), and instead of ‘semi-auxiliaries’ I shall use the more selfexplanatory term ‘lexico-modals’. 2.3.1 Properties of the semi-modals The semi-modals had better, would rather, be to, and have got to all express modal meanings. Had better (whose auxiliary element has arguably been reanalyzed as a present tense form, as has happened historically with must and ought) and would rather are both literally comparative idioms, expressing respectively advisability (similar to ought to and should) and the volitional sense of preference. Have got to, which derives from a perfect construction, is semantically similar to the lexico-modal have to (see next section), expressing both obligation and epistemic necessity. Be to expresses both obligation and futurity. The semi-modals have in common that their first element is an auxiliary, and they are like the modals in having no non-tensed forms (compare *We will must/have got to tidy up; *The game has should/been to start soon). They could
Theoretical preliminaries
17
only be considered ‘auxiliary-like’ if we invoked a special construal of the NICE properties, insofar as it is the first element rather than the whole construction that satisfies these properties (compare He hasn’t got to go/*He has gotn’t to go). However with had better and would rather negation may or may not take the auxiliary in its scope (compare They wouldn’t rather/would rather not intervene; They hadn’t better/had better not intervene). The former, according to Quirk et al. (1985: 141), is “typically used in ‘second instance’ contexts (especially in negative questions) where an earlier statement or assumption is being challenged”.7 All of the semi-modals except would rather display ‘voice neutrality’ (compare I had better/am to/have got to/would rather settle the bill ~ The bill had better/is to/has got to/*would rather be settled by me). 2.3.2 Properties of the lexico-modals The lexico-modals are a set of idiomatic expressions expressing modal (and in some cases aspectual) meanings which – except for have to, need to and want to – have be as their first element. The membership of the set is by no means clearcut, and is difficult to delimit in a principled fashion. Quirk et al.’s list (1985: 143) contains be likely to but not be certain/sure to, and be obliged to but not be allowed to. The lexico-modals are modal-like in displaying voice-neutrality (compare Australia is bound to accept the refugees ~ The refugees are bound to be accepted by Australia) and in their compatibility with the existential thereconstruction (e.g. There are bound to be many refugees). A non-modal-like feature of the lexico-modals is the availability of non-tensed forms, making possible combinations of the type had been meant to and seems to be going to. The availability of the non-tensed forms (in examples such as I will have to quit; Many employees are having to quit; The company has had to fire many employees) differentiates the lexico-modal have to from the semantically similar semi-modal have got to (compare *I will have got to quit; *Many employees are having got to quit; *The company has had got to fire many employees). Have to also contrasts with have got to in the availability of a preterite form (compare The company had to/*had got to fire many employees). The lexical verb want to is not included in Quirk et al.’s list, but it is evidently undergoing modalization (for instance, in its common phonological reduction to /wn/, especially in AmE). For further discussion of the quasimodal status of want to see Section 5.5. Need to is somewhat different from the other quasi-modals in the closeness of its relationship to its auxiliary counterpart need. Huddleston and Pullum (2002: 109) treat need and need to as essentially the same verb which can “behave as” either an auxiliary or lexical verb.
18
Chapter 2
2.3.3 Grammaticalization and idiomaticity In this section we discuss two of the criteria that, as noted above, were used to determine the set of quasi-modals examined in the study. 2.3.3.1 Grammaticalization Some quasi-modals have undergone a greater degree of grammaticalization than others. Grammaticalization is the diachronic process by which a periphrastic lexical unit is transformed into a more grammatical one, and typically involves syntactic simplification, phonological weakening, and semantic bleaching and generalization. Within the class of semi-modals we may identify have got to and had better as more grammaticalized, more auxiliarized, than be to and would rather. The phonological evidence for this claim is the availability of reduced forms, a distinctive feature of auxiliary verbs (Huddleston and Pullum 2002: 102). The auxiliary components of had better and have got to may either be reduced to /d/ and /v/ respectively, or omitted altogether (/bet/, /gt/), represented as better and gotta orthographically. A consequence of the elision of the auxiliary in had better is that some speakers reanalyze the lexical element better as an auxiliary, in structures such as We better go, bettern’t we. In the case of gotta, the obscuring of the infinitival in the reduction of the periphrastic form results in the semimodal governing a bare infinitival complement. The fact that be to and would rather have not evolved into single morphophonemic forms suggests a lesser degree of grammaticalization. What is the semantic evidence for grammaticalization with the semimodals? The originally comparative meaning of had better has been largely bleached, the only trace being the implication of ‘ill consequence’ present in some instances (see Section 3.8.1). As Jacobsson (1980: 52) notes, examples of the type in (10) “have not been attested in recent literature”. (10)
You had better murder him than marry him. (Dickens, Bleak House)
The comparative sense of would rather is still salient, however, as in (11): (11)
Simon got fairly close to it but um then he decided he’d rather have the school holidays than spend the school holidays rehearsing. (ICE-AUS S1A-100 273)
According to Krug (1998: 179-180) the obligative sense of have got to has evolved through syntactic renanalysis of possessive constructions (resulting in a fixed ordering of elements) and subsequent semantic bleaching. Less spectacularly, the future arrangement and obligation senses of be to may be assumed to have developed from the futurate use of be (compare There is/is to be
Theoretical preliminaries
19
a board meeting on Friday), the latter via the implication that in appropriate contexts if a speaker refers to an addressee’s future actions s/he is in fact requiring thair actualization. An important aspect of the modalization of the semi-modals is their capacity to express both ‘epistemic’ and ‘root’ meanings (see Section 2.4.1 below), such semantic duality being a hallmark of the central modal auxiliaries. The evidence here does not provide as clear a picture of degrees of grammaticalization as the phonological evidence. Have got to undoubtedly expresses both root (‘obligation’) and epistemic (‘necessity’) meanings (see Section 3.6), while would rather does not, being restricted to the specific root sense of volitional preference. The situation with had better and be to is however somewhat more difficult to interpret. Had better is regarded by most (e.g. Perkins 1983: 63) as expressing only deontic modality, but Mitchell (2003: 145) claims that it can also be used epistemically, as in (12) from the British National Corpus: (12)
This had better be good, I thought grimly as I crossed the road and walked up the cul-de-sac to the Parsonage (BNC, 410)
However Mitchell’s argument, that in such cases the speaker expresses a hope that something is or will be the case, and “Hoping is a type of epistemic volition: a wish that a proposition whose truth is unknown turns out to be true” (2003: 145), is unconvincing: had better in (12) is more plausibly interpreted as deontic, with the speaker anticipating a required outcome (compare the deontic sense of This must/will have to/will need to be good, I thought grimly as I crossed the road). As for be to, its scheduled futurity sense is essentially temporal and it lacks the centrally epistemic use that will has developed (compare They will/?are to be in Paris by now). Morphosyntactically, would rather and be to have some properties that indicate a lesser degree of grammaticalization than have got to and had better. Would rather retains the un-modal-like capacity to take a finite clause complement, as in I would rather (that) you went with someone else). Furthermore, as noted by Huddleston and Pullum (2002: 108) it is possible to negate the would only under certain conditions (e.g. the use of They wouldn’t rather intervene to deny a preceding positive, and of Wouldn’t they rather intervene? to frame a positively-biased negative question). Be to is un-modal-like in exhibiting subject agreement (compare I am/ought to go; You are/ought to go; She is/ought to go). In seeking to determine which members of the set of lexico-modals are the most grammaticalized we can apply similar criteria to those we have used with the semi-modals. We shall argue that have to and be going to are the most advanced in this respect and furthermore that a case can be made for the classification of want to as a quasi-modal. Also showing signs of modalization are be bound to and be supposed to. All five of these expressions are prone to significant phonological reduction, involving not merely the weakening of
20
Chapter 2
infinitival to, but more importantly its incorporation into the preceding word (/hæft, gn, wn, spost, bandt/).8 There is semantic evidence relating to root/epistemic duality for all five items. This evidence is straightforward for have to (see Section 3.5.2) and be supposed to (see Section 3.9.1), even though in the latter case the deontic root sense evolves by implication from the historically prior epistemic meaning, contrary to the well documented fact trend for epistemic meanings to develop historically out of root ones (see, e.g., Traugott 1989). In the case of be going to the epistemic meaning is of the non-central futurity type (and typically so in the case of be bound to: see Section 3.11.1). With want to there is evidence of an incipient epistemic meaning (see Section 5.5.3). 2.3.3.2 Idiomaticity The semi-modals are all idiomatic to the extent that their primary semantic elements (better, rather, got and be) all have non-equivalent uses in other contexts. The question is less straightforward with the lexico-modals. Of the items we have been discussing, have to, be bound to, be going to, be supposed to, and be about to can be regarded as idiomatic in view of the bleaching of possessive content with have, of any literal sense of tying or fastening with bound, of motional meaning with going, of conjectural meaning with supposed, and of locative meaning with about. Be able to is marginal, insofar as the meaning of able in the lexico-modal does not differ greatly from its meaning as a separate adjective. Want to and need to are not idiomatic. There are various tests that are responsive to idiomaticity. 9 One, suggested by Quirk et al. (1985: 144) for identifying the idiomaticity of lexico-modals containing be, is the inability of what follows be to stand at the beginning of a ‘supplementive’ clause. Application of this test using the frame ______ to make a decision, Matilda agreed to stay for dessert suggests that the status of be able to, be bound to, and be about to, as quasi-modals is less clear-cut than that of be going to and be supposed to. A second test is the impossibility of omitting infinitival to in final position. Items which pass this test are be about to, have to, be supposed to, be going to and be bound to (*He has permission to attend, but isn’t about/doesn’t have/isn’t supposed/isn’t going/isn’t bound), while be able to is marginal (?He has permission to attend, but isn’t able). A third test is the resistance of adjectives in lexico-modals to degree modification, which shows that items such as be able to, be bound to, be supposed to and be about to are all idiomatic (*She’s very able/bound/ supposed/about to help us), as opposed to, for example, be likely to or be willing to (She’s very likely/willing to help us).
Theoretical preliminaries
21
2.3.3.3 Selection of quasi-modals for analysis Table 1.3 in Chapter 1 presents the list of those quasi-modals that were selected for analysis in the study on the basis of their association with a modal auxiliary (primarily in the sense of semantic similarity but also, in some cases, in their serving suppletive roles: see Chapters 3–5), their degree of grammaticalization, and their level of idiomaticity. The high frequency quasi-modals have to, have got to, be going, want to and need to, we shall argue, are making inroads into the ‘semantic space’ occupied by their modal equivalents. Of the remaining lowerfrequency quasi-modals, be able to and be about to are less grammaticalized, and be able to is also less idiomatic, but the closeness of their semantic relationship to can and will respectively was considered sufficient grounds for their inclusion. Would rather and may/might as well are discussed, but not analyzed in detail: both have undoubtedly undergone some degree of grammaticalization and idiomaticization, but their meanings are nevertheless largely predictable from those of their auxiliary elements in combination with the comparative sense of the lexical element. 2.4
Modal meanings: approaches and issues
In this section we discuss some approaches to the classification of, and issues in the analysis of, modal meanings. 2.4.1 Classifying modal meanings Many scholars operate with a two-way distinction between ‘epistemic’ and ‘root’ modality (e.g. Coates 1983, and many American linguists), though with differences of terminology. 10 The term ‘root’ suggests that this kind of modality is more basic, an idea that derives support from diachronic evidence that epistemic meanings tend to develop from root meanings via the extension of concepts involving human interaction and properties to the domain of reasoning and judgement (Sweetser’s 1990 ‘subjectification’). Epistemic modality is prototypically concerned with the speaker’s attitude towards the factuality of the situation, the speakers’s judgement of the likelihood that the proposition on which the utterance is based is true, located along a scale ranging from weak possibility (“It may be the case”) to strong necessity (“It must be the case”). In non-protypical instances epistemic modality will be a matter of objective conclusion rather than the speaker’s attitude (as in the objective algebraic necessity expressed by must in If 2x =10, x must be 5). There is not complete consensus on which modals and quasi-modals are exponents of epistemic modality: for example Coates (1983: 244) describes will and shall as “marginally epistemic”, and this in turn follows from their uncertain status, when they indicate futurity, as markers of epistemic modality (see further Sections 5.1.1 and 5.2.1). Similar questions are then applicable to the epistemic status of the quasi-
22
Chapter 2
modals be going to, be to and be about to. In order to maximize the possibility of identifying broad differences between epistemic and non-epistemic modality that can be related to syntactic/semantic, regional and stylistic variables, we shall in this study assume epistemic modality to be a broad overarching system, and certainly one capable of subsuming the subsystems posited by some (e.g. Coates’ 1983: 18-19 parallel inferential and non-inferential scales, and Palmer’s 2001: 8 distinction between epistemic and evidential modality). It is not easy to provide a unitary definition of ‘root’ modality. Palmer (2001: 8) endorses Jespersen’s (1924: 329-31) observation that the modal uses in question are those “containing an element of will” (whereas those involving epistemic modality may be regarded as “containing no element of will”. Bybee and Pagliuca (1985: 63), who use the term ‘agent-oriented’ modalities, describe them as “modalities that predicate conditions of either an internal or external nature on a wilful agent: these are the notions of ability, obligation, desire and intention”. A slightly different focus, on the conditions for actualization rather than on the agent, occurs in Bybee and Fleischman (1995: 6), for whom agentoriented markers “predicate conditions on an agent with regard to the completion of an action referred to by the main predicate”. The most common and arguably important type of root modality is ‘deontic’, which occurs when the factors impinging on the actualization of the situation referred to in the utterance involve some type of authority – as when a person or a set of rules or a social convention is responsible for the imposition of an obligation or a granting of permission (as in You must/may leave at 3pm). However not all cases of root modality are deontic: a possibility or necessity may arise not from an authority, or ‘deontic source’, but rather from general circumstances (as in You can get to the island by ferry) or from properties intrinsic to the subject-referent (as in the ability meaning of Can he speak Chinese, and the volitional meaning of She will iron your shirt). Some writers address the comparative heterogeneity of root modality by recognizing subtypes. For example, Palmer (2001) distinguishes, within what he calls ‘event’ rather than root modality, between ‘deontic’ and ‘dynamic’ modality: the former deriving from an external source and affecting a discourseparticipant, the latter deriving from and affecting an internal source (namely, the subject referent). A slightly different distinction, between ‘participant-external’ and ‘participant-internal’ modality, is posited by Van der Auwera and Plungian (1998). While these categories are broadly similar to Palmer’s deontic and dynamic respectively, they differ in the exclusion of volition (which Van der Auwera and Plungian regard as lying outside the bounds of modality) from the participant-internal category, and in the inclusion of non-deontic root possibility (which Palmer does not regard as a meaning distinct from ability) within the participant-external category. A quite different type of binary distinction, between ‘extrinsic’ and ‘intrinsic’ modality, is advocated by Quirk et al. (1985).11 Extrinsic modality involves “human judgement of what is or is not likely to happen” (1985: 219) and includes epistemic modality along with non-deontic root possibility/necessity, prediction, and ability. Intrinsic modality involves “some kind of intrinsic human
Theoretical preliminaries
23
control over an action” (1985: 221), and covers deontic modality along with volition. Quirk et al. concede that ability involves human control, but argue that more importantly it involves a judgement about the likelihood of actualization. Some writers adopt a tripartite scheme. The division in both Palmer’s earlier (1990) book, and Huddleston and Pullum (2002), is between epistemic, deontic and dynamic modality. The dynamic category includes ability, volition and also non-deontic root (or, circumstantial) meanings. The absence of a larger category that subsumes dynamic and deontic modality (i.e. ‘root’, or Palmer’s 2001 ‘event’ modality) leaves open the possibility that there may be relationships between dynamic and epistemic meanings, just as there are between dynamic and deontic meanings. It is this scheme that is adopted in the present study. 2.4.2 Polysemy vs monosemy There has been a good deal of debate as to whether the meanings expressed by the modals are sufficiently discrete and independent for us to acknowledge that the modals are polysemous, or whether the relationship between the meanings is sufficiently vague for us to assume that each modal is monosemous, with a core meaning that is present in all their uses. The availability of ambiguous examples provides strong evidence that modals are polysemous, as noted by many linguists (e.g. Lyons 1977, Bybee and Fleischman 1995, Palmer 2001, Huddleston and Pullum 2002). In cases where it is impossible to decide, out of context, whether a root or epistemic meaning is intended, the fact that such meanings cannot co-exist provides evidence that that they are discrete. Sometimes ambiguity occurs in contexts where both interpretations are possible and valid, as in (13) below (“you are permitted to receive”; “it is possible that you will receive”). (13)
If you qualify for a War Widow’s Pension you may also get full Unemployment Benefit on your own contributions. (ICE-GB W2D-002 29)
Further support for a polysemy position comes from the association of root and epistemic meanings with such grammatical phenomena as the scope of negation, and from their association with different paraphrases (e.g. “it is possible that” for epistemic possibility, and “it is possible for” for root possibility). Supporters of the monosemy position (e.g. Ehrman 1966, Haegeman 1983, Klinge 1993, Groefsema 1995, Papafragou 2000) point to the occurrence of allegedly indeterminate examples where the different meanings in question are not incompatible, and the difference between them is contextually neutralized (referred to as ‘merger’ by Coates (1983). An example is (14) below, where can is ambivalent between (dynamic) possibility or deontic possibility (‘permission’). (14)
But we can still be friends (ICE-AUS S1A-099 166)
24
Chapter 2
According to the monosemists the occurrence of such indeterminate cases suggests that each modal has an invariant, core meaning, with different uses, or interpretations, being determined by context. The polysemy and monosemy positions are not in practice absolutely mutually exclusive. Those who adhere to a polysemy position will not insist that all the meanings expressed by a particular modal are distinct. Conversely, a monosemy position does not exempt the analyst from seeking to identify different interpretations, but the exercise will proceed from a smaller semantic base and require richer context-dependent analysis than a polysemy one. In the present study we assume a polysemy position, while at the same time acknowledging that the dividing line between deontic and dynamic modality will generally be less determinate than that between epistemic modality and either of these root categories. Furthermore the three primary meanings have subsumed uses which are not always clearly distinguishable. 2.4.3 Negation and ‘temporality’ A sentence with a modal expression can, it is generally accepted, be analyzed into two parts, one representing the modal meaning and the other representing the ‘proposition’ whose truth is being judged or actualization predicated. This division has consequences for negation and for ‘temporality’ in a modal sentence. Negation may bear on the proposition, as in: (15)
The changes in her voice may well not be age (ICE-AUS S1A-035 310)
Here the negation falls within the scope of the modal (“It is possible that the changes in her voice are not age”), and for this reason is described as ‘internal’ by Huddleston and Pullum (2002: 175). By contrast in (16) the negation is ‘external’, bearing on the modal and falling outside its scope (“It is not possible that she is much older than me”). (16)
She can’t be much older than me (ICE-AUS S1A-020 50)
The two types of negation may combine, as in (17), where but has negative force. (17)
I mean you you cannot but see that in the context of Cezanne (ICE-GB S1B-008 157)
The scope of negation is more transparent when there is a lexical expression of modality (as in the paraphrases provided for (15) and (16) above, where the presence of a negative in the subordinate clause indicates internal negation, while one in the superordinate clause indicates external negation). The parallels between such lexical expressions and modal auxiliaries can be exploited as a useful test for determining, in cases of negation with a modal auxiliary, whether the negation is internal or external.
Theoretical preliminaries
25
The distinction between the modal and propositional part of a modal utterance is also important in enabling us to provide a description of the temporal information in the sentence. In the literature there is some inaccuracy in the wording used to account for this. For example Coates (1983: 43) asserts that epistemic must “refers to states and activities in the present or past”, but such reference is made in the proposition rather than by the modal itself. Both the modal and propositional parts are associated with temporal loci and the relationship of the latter to the former may be distinguished as one of ‘anteriority’, ‘simultaneity’ or ‘posteriority’ (Depraetere and Reed 2006: 284). Some of the possibilities for deontic must and have to are exemplified in (18)– (22) (see further Sections 3.1.4 and 3.5.4 below): (18) (19) (20) (21) (22)
No you must meet her (ICE-AUS S1A-018 195) You’ll have to go down (ICE-AUS S1A-016 130) He had to be up at Asquith by six-thirty (ICE-AUS S1A-016 71) Yes, or they must have completed an apprenticeship. (ICE-AUS W2D-002 138) Managers and supervisors must be able to ensure that they will treat the matter confidentially (ICE-AUS W2D-004 161)
In (18)–(20) the obligation is located in the present, future and past respectively. In each case the temporal relationship between the obligation and the situation referred to in the proposition is one of posteriority (e.g. in (18) “it is desirable that you should meet her (at some time in the future)”). It is sometimes stated that deontic modality cannot be associated with an anterior situation (e.g. Depraetere and Reed 2006: 285). While this is true of subjective uses (given the pragmatic impossibility of imposing an obligation on someone, or giving them permission, to do something in the past) it does not apply to general rules or conditions, as shown by (21). A relationship of simultaneity as in (22) is only possible when the modality is of this general kind.
2.5
Further dimensions of modality
In Section 2.4.1 we examined various kinds of modal meaning. In this final section we introduce three further dimensions of modality which are applied at various points in the analysis presented in Chapters 3–5: strength, degree and subjectivity. 2.5.1 Modal strength The dimension of modal strength is defined by Huddleston and Pullum (2002: 175) as “the strength of commitment (prototypically the speaker’s commitment) to the factuality or actualisation of the situation”. It is this concept that provides
26
Chapter 2
the basis for the distinction between the modal concepts of necessity (where the commitment is strong) and possibility (where it is weak). They are logically related in terms of their interaction with negation, as illustrated in (23) and (24). The examples in (23) express epistemic modality (note that mustn’t is not possible for all speakers), while those in (24) express non-epistemic meanings. The abbreviation ‘Poss’ stands for ‘possible’ and ‘Nec’ for ‘necessary’. (23) (24)
It can’t be true [not-Poss] It may not be true [Poss-not] You can’t go [not-Poss] You can not go, if you wish [Poss-not]
= = =
It mustn’t be true [Nec-not] It needn’t be true [not-Nec] You mustn’t go [Nec-not]
=
You needn’t go [not-Nec]
Semantic strength may be affected by pragmatic factors. Compare for example the different strengths of must in (25) and (26). The uncompromising instruction in (25), which doesn’t countenance non-compliance, is consistent with the semantic strength of must. However in (26) the speaker is merely making a suggestion, and accordingly we may regard the strength of must as being pragmatically weakened. (25) (26)
If the ball rolls back onto the path or where the player has interference from his stance or swing, you must re-drop. (ICE-AUS W2D-011 68) If you’re on holiday in France you must visit a chateau (ICE-AUS S2B030 98)
Huddleston and Pullum (2002: 177) identify an intermediate category on the strength continuum which they call ‘medium modality’, associated with should and ought, and lexical modal expressions such as likely, probably and seem. Consider the effect of substituting epistemic should or may for must in the following example: (27)
He must be making an absolute killing. (ICE-AUS S1A-014 158)
That should would be weaker in strength than must is suggested by the possibility of felicitously adding a continuation such as but he may not be to the shouldversion but not to the original with must, suggesting that must is stronger than should. Should, on the other hand, is stronger than may, as suggested by the contrast between the acceptability of He may be making an absolute killing, but it’s not likely and the unacceptability of He should be making an absolute killing, but it’s not likely. Whereas there is a clear semantic difference between internal and external negation with strong and weak modality, medium strength modality is distinctive in the pragmatic closeness of the two negation patterns. Consider:
Theoretical preliminaries (28)
27
Well I’ll find it after I get out of here which shouldn’t be very hard (ICEAUS S1A-090 362)
The negation here is internal, paraphraseable as “It is not likely that it will be very hard”, which is pragmatically equivalent to “It is likely that it will not be very hard”. 2.5.2 Degree of modality While the concept of modal harmony has been invoked in the literature (e.g. by Coates 1983: 45-46, 137-138, 151-152), Huddleston and Pullum (2002: 179-180) are the first to explain it in terms of the concept of degree of modality, which they define as “the extent to which there is a clearly identifiable and separable element of modal meaning” (p.179). A low degree of modality is said to occur when a modal expression exhibits harmony with the larger construction – that is, conveys modal meaning of a similar type and strength – such that its selection might be considered optional. Consider the following: (29) (30)
(31)
It must surely qualify as one of the great symbols of Australia, along with the kangaroo and the koala. (ICE-AUS W2B-030 15) Everyone takes part just by being there so the tradition is upheld and the two Marys are still recognised regardless of what the crowds may believe (ICE-GB S2B-027 157) It’s odd that ah mysticism should have such a bad name (ICE-AUS S2A049 87)
The example in (29) represents the most common type of modal harmony, that involving verb-adverb pairings where the verb and adverb have the same strength, such that omission of the modal would have little impact on the meaning. If we replaced must by may the modal harmony would disappear: the weak strength modal may expressing a different meaning from the strong adverb surely (may being harmonic with perhaps, possibly, and the like). In (30) may merely reinforces the concessive meaning expressed by the larger construction, whose meaning is not significantly different from that of regardless of what the crowds believe. In (31) the so called ‘subjunctive’ should is harmonic with the emotive content of the superordinate clause. In (32) will has low modality, but not on account of modal harmony. Rather, as noted by Huddleston and Pullum (2002: 180), the low degree of modality in such cases is due to the “strong association between will and the temporal concept of futurity”. While will may be omissible in (32) there are comparable examples where it is not, as in (33). (32)
Dr John Meurig Thomas, Fullerian professor of Chemistry at the Royal Institute of Great Britain and director of the Davy Faraday Research Laboratory, will launch Dyfed education authority’s Spacewatch Initiative
28
(33)
Chapter 2 at Gwendraeth Valley Comprehensive School on Wednesday. (ICE-GB W2C-017 93) He’ll be flown to the Camperdown Children’s hospital (ICE-AUS S2B020 133)
2.5.3 Subjectivity/objectivity The distinction between subjectivity and objectivity, which appears in most extended accounts of modality (e.g. Coates 1983, Perkins 1983, Palmer 1990, Westney 1990, Huddleston and Pullum 2002) is of particular importance in this study because, as we shall see in the following section, it is sometimes claimed to systematically differentiate modal auxiliaries from quasi-modals. According to this view the speaker involvement or orientation associated with deontic must in (34) would be replaced by an external, objective orientation if must were substituted by have to: (34)
Then we must stop somewhere for breakfast (ICE-AUS S1A-017 210)
Subjective deontic modality is often regarded as typified by ‘performative’ uses of the modals, as in (35), where must has the illocutionary force of a speaker-initiated directive, by contrast with the objective, externally-initiated, use exemplified in (36), where its force is that of an assertion or report. (35) (36)
You must let me smell it (ICE-GB S1A-041 253) If you are the registered keeper of a vehicle and you change your address or name (on marriage, for example) you must tell DVLA, using the back of the registration document. (ICE-GB W2D-010 40)
Palmer (1990) regards deontic modality as “essentially performative”, observing that: “By using a deontic modal, a speaker may actually give permission (may, can), lay an obligation (must) or make a promise or threat (shall)” (1990: 69). By contrast Coates (1983: 32-33) invokes the notion of performativity only with must, associating it with the stereotypically subjective core of the must’s ‘cline’ of meaning. The subjectivity/objectivity distinction applies to epistemic as well as to deontic modality. Consider the following examples: (37) (38)
(39)
Look at your injuries. You must have really hurt yourself. (ICE-AUS S1A022 140) If there’s a reasonable doubt as to whether there’s a car in front of Mr McGregor’s vehicle that also must point in my submission to a finding of not guilty. (ICE-AUS S2A-068 158) it is common to hear comments made suggesting that Muslims are really clean, that all other religions are deficient or distorted in some way and hence rejected by God, that only Muslims can lead a truly moral life while
Theoretical preliminaries
29
non-Muslims are necessarily amoral and entirely promiscuous, and that Western societies (and indeed all societies that have undergone any process of secularization) must inherently lack any moral foundation whatsoever. (ICE-AUS W2A-013 91) In (37) epistemic must represents the expression of speaker inference or judgment that is regularly associated with epistemic modality. It is considerably weaker than objective epistemic must as exemplified in (38) and (39) (note that there is a greater difference between (37) and its unmodalized counterpart, You have really hurt yourself, than there is between (38) and that also points to a finding of not guilty, and between (39) and Western societies lack any moral foundation whatsoever). Objective epistemic must can be dissociated from the speaker’s opinions and attitudes in two possible ways. It may be alethic (q.v. Lyons 1977: 797, Palmer 1990: 6-7), as in (38), involving a strict semantic necessity, determinable via quasi-mathematical deduction (“If x, then there is no other possible conclusion than y”). Alternatively there may be a contextually determined distancing from the speaker, as in (39): it is not the speaker’s own assessment that is presented but rather one that arises externally and is merely reported.
2.6
General aspects of the relationship between modals and quasi-modals
Is it possible to generalize across the differences between modals and quasimodals in a theoretically-principled and well-motivated fashion? According to Palmer (1990: 10-11) the modals are inherently subjective and it is this property that distinguishes them from quasi-modal expressions. Palmers’s arguments are discussed in some detail by Westney (1995: 54-59), who accepts their general plausibility but at the same time identifies “some complicating factors that may make the whole case less straightforward than it appears” (1995: 55). For example Palmer attributes the fact that preterite forms are either lacking for modals or normally unavailable for expressing past time to their distinctively performative, and hence subjective, nature. The problem here is the assumption that quasi-modals cannot be used subjectively, one that Westney rightly questions and that the data for the present study indicate is open to challenge. For example ’ve got to in (40) is used to impose an obligation, the subjectivity/performativity here being no less apparent than in You must decide or the imperative You decide. (40)
‘You’ve got to decide,’ he said. ‘I can’t tell you what to do.’ (ICE-AUS W2F 002 37)
Palmer further asserts that quasi-modals are not, with the exception of have to, available to express epistemic modality (following from his view that epistemic modality is not normally objective). That this claim is disputable is
30
Chapter 2
suggested by the occurrence of examples such as (41), where the subjective epistemic meaning of ’ve got to is parallel to that of its more formal but apparently semantically identical alternate must in You must be kidding: (41)
Cass’s eyes go wide. “You’ve got to be kidding.” (ICE-AUS W2F-017 82)
A number of writers (e.g. Lakoff 1972, Bouma 1975) relate the subjectivity/objectivity distinction to questions of the ‘source’ and ‘target’ of obligations, permissions, and the like. They lead us to a distinction between modals as carrying a default interpretation in which the speaker is assumed to be the source, and thus associated with greater personal involvement with, or sympathy for, the target than are the quasi-modals. A comprehensive survey of these and other accounts is offered by Westney (1995: 59-67), who suggests that we should exercise caution in assessing their merits in view of the often conflicting evidence presented. Westney’s own view is that while modals are maximally unspecified or unmarked, “the choice of a periphrastic allows a modal predication to be more precisely specified both syntactically and semantically” (1995: 54).
Notes 1
In this book the convention followed in Huddleston and Pullum (2002) of using bold italics to represent lexemes, in abstraction from any of their associated word forms, is adopted in principle. However in practice the use of bold type is restricted to cases where failure to recognize the lexeme versus word form distinction might be a source of confusion (particularly in the present chapter). In contexts where the distinction is of no relevance bold type is avoided on the grounds that it would simply be distracting.
2
Strictly speaking the modal auxiliary lexeme is simply ought, but we shall take it as including the infinitival particle to. This will be consistent with our practice with all the quasi-modals examined which, except for had better, take a to-complement. In the case of need to our practice will serve to distinguish the quasi-modal from the auxiliary, and in the case of have to and be to will distinguish the quasi-modal from other uses of have and be.
3
We accept Huddleston and Pullum’s (2002: 91) analysis of can’t, mightn’t, mustn’t etc. as inflectional forms rather than ‘contractions’. Support for this analysis comes from the fact that they are not always replaceable by their analytic counterparts (e.g. Can’t/*Can not she reach
Theoretical preliminaries
31
it?), and that their phonological form is not always predictable (e.g. won’t, shan’t). 4
Some grammarians (such as Huddleston and Pullum 2002: 92ff) apply the term ‘auxiliary’ to all verbs that satisfy the NICE properties, rejecting the traditional view of auxiliaries as dependents of the following lexical verb in favour of one in which they are analyzed as (catenative) main verbs that take non-finite complements. The issue is not one that that has ramifications for the primarily semantic concerns of this book, and for that reason is not explored in the present discussion.
5
See Huddleston and Pullum (2002: 106-107) for more detailed discussion.
6
There were no corpus examples of ought to used in the apodosis of an unreal conditional or without a to-infinitive).
7
There were no corpus examples of wouldn’t rather or hadn’t better.
8
The last of these, /bandt/, does not appear in Huddleston and Pullum’s (2002: 1616) list of items undergoing such reduction. It is, however, included in Westney’s (1995: 34) list.
9
For a more detailed discussion of these and other tests of idiomaticity for the quasi-modals see Westney (1995: 18ff.).
10
A lucid overview of the issues discussed in this section can be found in Depraetere and Reed (2006).
11
Bolinger (1989) also makes use of this distinction in his study of may.
Chapter 3 Necessity and obligation This chapter deals with the modals and quasi-modals of necessity and obligation. This grouping is smaller than those examined in Chapters 4 and 5, particularly in the terms of the frequency of the auxiliaries, with 3981 modal auxiliary tokens as against 14,980 for the possibility/permission/ability group and 16,623 for the prediction/volition group. Furthermore, of the three groupings, it is the only one for which the quasi-modals (with 4906 tokens) surpass the auxiliaries in frequency, and may be regarded as regularly replacing their auxiliary counterparts. Biber et al. (1999), who likewise note a smaller frequency for the modals of obligation/necessity, suggest as a possible explanation the “general tendency to avoid the face threatening force of expressions with an obligation meaning” (1999: 489). The expressions examined in this chapter are the modals must (embracing the forms must and mustn’t), should (should and shouldn’t), ought to (ought to and oughn’t to), and need (need and needn’t), and the quasi-modals have to (have to, has to, and had to), have got to (have/haven’t got to and has/hasn’t got to), need to (need to, needs to, needed to, and needing to), had better, be supposed to (supposed to preceded by all the inflectional forms of be), be to (all the inflectional forms of be followed by a to-infinitive where the sequence expresses modal meaning), and be bound to (bound to preceded by all the inflectional forms of be).1 Within the set of necessity/obligation modal expressions we may distinguish two broad subgroups in terms of their typical modal strength: the strong forms must, have to, have got to, need, need to, be bound to and be to; and the medium strength forms should, ought to, be supposed to and had better.
34
Chapter 3
Table 3.1. Frequencies of the modal expressions of necessity and obligation Modals must should ought to need Total Quasi- have to modals have got to need to had better be supposed to be to be bound to Total Total
ICE-AUS 613 1,141 36 19 1,809 1,311 332 343 48 47 135 9 2,225 4,034
ICE-GB 675 1,124 80 34 1,913 1,244 339 280 33 99 221 17 2,233 4,146
C-US 402 (79) 850 (167) 51 (10) 15 (3) 1,318 (259) 1,385 (272) 173 (34) 473 (93) 41 (8) 127 (25) 76 (15) 5 (1) 2,280 (448) 3,598 (707)
TOTAL 1,690 (1,367) 3,115 (2,432) 167 (126) 68 (56) 5,040 (3,981) 3,940 (2,827) 844 (705) 1,096 (716) 122 (89) 273 (171) 432 (371) 31 (27) 6,738 (4,906) 11,778 (8,887)
NB In this and all subsequent tables, in columns with two sets of frequencies, raw frequencies are bracketed and tokens per one million words are unbracketed. NB The figures for should include 289 tokens of specialized uses where it’s not a necessity/obligation modal; and for be to 139 tokens of the conditional use. 3.1
Must
While must expresses mainly deontic necessity, or ‘obligation’, it also has a wellestablished epistemic necessity meaning (this being a minor, incipient, meaning with must’s quasi-modal rivals have to and have got to). A third, minor, meaning is dynamic necessity. Table 3.2. Meanings of must Deontic Epistemic Dynamic Indeterminate Total
ICE-AUS 369 185 40 19 613
ICE-GB 391 216 41 27 675
C-US 209 (41) 153 (30) 25 (5) 15 (3) 402 (79)
Total 969 (801) 554 (431) 106 (86) 61 (49) 1,690 (1,367)
% 57.3% 32.8% 6.3% 3.6% 100%
Necessity and obligation
35
3.1.1 Deontic must While deontic must has a default interpretation in which the speaker is identified as the deontic source, as in (1), there is no necessary connection between subjectivity and the use of must. In (2) must is objective, with the source of the obligation (‘the world’) external to the speaker. (1) (2)
If you’re on holiday in France you must visit a Chateau (ICE-AUS S2B030 98) At the United Nations the world agreed that Iraq must withdraw or be driven out of Kuwait (ICE-GB S2B-030 19)
According to Lyons (1977: 832-833) subjective deontic must as in You must open the door – compare the corpus example in (3) – can be used performatively by the speaker (imposing a directive). However, despite the strong compulsion expressed by must in (3), there is not the same degree of directness as would be conveyed by its imperative counterpart (Stop doing that), where the speaker requires immediate compliance. (3)
Then she said oh you must stop doing that (ICE-GB S1A-062 150)
In other cases involving must with a 2nd person subject it would be even less appropriate to invoke the concept of performativity. Subjective deontic must is commonly used in cases where the speaker is not in a position to – or may not even wish to – require actualization, as in the advice in (4), the request in (5), and the exhortation in (6). (4) (5) (6)
You must only do it with your teacher, because you can so easily get into the wrong (ICE-GB S1A-045 61) you must let me photograph your baby for my magazine (ICE-GB S1A039 93) You must meet Forename6 you haven’t met her at all have you (ICE-AUS S1A-018 191)
Objective deontic must is typically found in formal documents containing laws, rules or regulations, as in (7): (7)
A complaint procedure must therefore ensure that both parties are given the opportunity to be heard in a fair and impartial way by a person who is sensitive to the issues and primarily concerned with the effective resolution of the problem. (ICE-AUS W2D-004 75)
Somewhat more abstractly, but still within the realm of human involvement, the deontic source for objective must may be an official position or
36
Chapter 3
view as in (8), or simply an unspecified consideration of what may be considered morally desirable as in (9): (8) (9)
There is this stuffy attitude you know not just in politics but beyond, that somebody must wait another two or three years (ICE-GB S1B-043 54) It would not be in the interests of our troops to do so and they of course must be our prime concern (ICE-GB S2B-014 41)
Deontic must can sometimes be ambivalently subjective/objective, with uncertainty as to the deontic source. For example in (10) it is unclear whether the speaker is giving instructions as a person in authority, or objectively stating rules of the company which employees must abide by. (10)
You should help callers assess the responsibilities and duties of the position, and offer any other relevant information. Once the position has been advertised, you must be available to accept enquiries. (ICE-AUS W2D-001 18)
The present study confirmed the tendency noted by Coates (1983: 34-35) for the subjectivity/objectivity of deontic must to correlate with the person of the subject. The clearest cases of subjective deontic must are those where the subject is you, as in (3)–(6) above. Similarly the clearest cases of objective deontic must are those with a 3rd person subject, as for example in (7)–(9) above. It is important to reiterate that that the correlation between subjectivity/objectivity and the person of the subject is merely a tendency, as examples such as the following show: (11)
(12)
(13)
Northern Building Society has informed us that you must return the Mercantile Mutual Insurance Policy to enable settlement to proceed on the due date, namely 1 September, 1991. (ICE-AUS W1B 326) Depreciation on certain motor vehicles falls under the substantiation rules. For these motor vehicles you must provide proof of cost if they were bought after 30 June 1986. (ICE-AUS W2D-008 122) You must keep them moist (…) That uh bud must not dry out at all. (ICEGB S1B-025 91)
In (11) and (12) must has a 2nd person subject but the speaker is not the deontic source: (11) is used to report an instruction rather than to impose one, while in (13) the 2nd person subject is non-specific plural you (for which the indefinite pronoun one sometimes serves as a more formal alternative) and the obligation originates from an institution. Must in (13) has a 3rd person subject, but the speaker is the deontic source. With 1st person subjects we regularly find both subjectivity and objectivity. The examples in (14)–(17) invite a subjective interpretation: in (14) the speaker engages in insistent self-incitement, (15) instantiates a polite
Necessity and obligation
37
exhortation of self and addressee using inclusive we, and (16) and (17) illustrate the formulaic use of must with a verb of communication in which the utterance effectively realizes the act of admission or statement (a very common use which accounts for 30% of all tokens of (deontic) must with a 1st person subject in the present data): (14)
(15) (16) (17)
It was very comfortable except for not have enough non-smoking places. As there were families with children in the same section it seems obvious they should do something about it. I must write & tell them. (ICE-AUS W1B-009 25) Well Harold mentioned it to me at the time, but I think it was one of those invitations like we must have lunch sometime (ICE-GB S1B-040 7) Yeah I must admit I went home depressed as well (ICE-AUS S1A-024 332) And I must say at the outset that I’m in complete agreement (ICE-AUS S1B-052 91)
In other cases deontic must with a 1st person subject expresses a requirement which, if not externally imposed, at best has a vague source, as in: (18)
Becoming who we are and taking full possession of our own historically conditioned cultural identities – something that we must all attempt if we are to live responsible lives – is, then, of a piece with the practice of anthropology (ICE-AUS W2A-013 165)
The subjective/objective dimension was found not to be reliably quantifiable because of the large number of indeterminate instances. However in view of the tendency for these categories to correlate with the person of the subject, the figures presented in Table 3.3, showing that over two-thirds of deontic musts have a 3rd person subject and less than 10% have a 2nd person subject, suggest that deontic must is more commonly used objectively than subjectively. Table 3.3. Person of subject with deontic must (Frequencies for C-US normalized to tokens per million words) ICE-AUS ICE-GB C-US Total
1st person 70 (19.0%) 85 (21.7%) 67 (32.0%) 222 (22.9%)
2nd person 42 (11.4%) 37 (9.5%) 10 (4.8%) 89 (9.2%)
3rd person 257 (69.6%) 269 (68.8%) 132 (63.2%) 658 (67.9%)
Total 369 (100%) 391 (100%) 209 (100%) 969 (100%)
Deontic must is semantically strong, but it is commonly prone to pragmatic weakening. Its strength – what Sweetser (1990: 54) calls its ‘resistibility’, degrees thereof being determined by the severity of the
38
Chapter 3
consequences for non-fulfilment of the obligation – is closely associated with subjectivity/objectivity. The present data evidenced a tendency for subjective uses to be strong, objective uses to be weak. At the maximum end of the scale of strength we may locate subjective examples like (3) above, where the speaker has sole responsibility for the directive and non-compliance is not countenanced. At the other, weak, end of the scale we may locate an example such as (19), an agentless passive with an unspecified deontic source having no necessary connection with the speaker, where must merely expresses what is thought to be desirable. (19) And I think that they must be encouraged to implement a democratic structure to follow this and not just have yet another military regime (ICEGB S1B-036 72) The correspondence is not perfect, however. For example in (20) the modality is subjective (the obligation being self-imposed by the speaker), but must makes a very weak statement of requirement (as reflected in the fact that, in conventional politeness formulae of this type the degree of modality is low, with the modal being readily omissible). (20)
Uhm I must confess that I’m unrepentant about the poll tax (ICE-GB S1B034 6)
As with subjectivity/objectivity, so with strength, there is predictably a degree of correspondence with the person of the subject: compare the strength of (3), where must has a 2nd person subject, with the weakness of (19), where it has a 3rd person subject. However here the correlations are looser. For example in (2) above, where must has a 3rd person subject, the modality is clearly objective and yet at the same time it is high-strength, with the severity of the consequences for non-compliance spelt out. Again, the 2nd person subject in (6) above is associated with a clearly subjective use of must, but the speech act of polite exhortation is associated with weak resistibility only. 3.1.2 Epistemic must In discussing epistemic must both Coates (1983: 41) and Huddleston and Pullum (2002: 181) refer to the speaker’s ‘confidence’. However it is important to be careful here. As with deontic must, we may distinguish between subjective and objective uses. Objective epistemic must expresses logical certainty, a logical necessity based on what is known, as in: (21)
When you uhm therefore say that you agree with the sentence in practice the hysteric is not infrequently a malingerer too, it must follow from the very terms of that sentence, that sometimes there are hysterics who are not, malingerers too (ICE-GB S1B-070 14)
Necessity and obligation (22)
39
People tend to think that because many of the problems are global, the answer must be global. (ICE-GB W2B-013 46)
In such cases the degree of modality is low and the speaker apparently confident that the conclusion presented is the only one possible. However, it is far more common for epistemic must to be used subjectively (91% of all epistemic musts were subjective) and here we find variation in the amount of confidence shown by the speaker. There is no doubt that we can speak of the speaker’s ‘confident’ inference in many cases, especially those where the grounds for the deduction are spelt out, as in (23) and (24), and those where a harmonic expression such as surely in (25) indicates a low degree of modality and correspondingly strong confidence. (23) (24) (25)
Kim Childs has got about 6 letters this week, her father must be the head of Australia Post. (ICE-AUS W1B-014 85) With all the bits of work you’ve done over the years, your CV must be pretty full? (ICE-GB W1B-001 180) It must surely qualify as one of the great symbols of Australia, along with the kangaroo and the koala. (ICE-AUS W2B-030 15)
However in the majority of subjective cases ‘confidence’ does not seem an appropriate term to use. Semantically strong they may be (insofar as, logically, ‘necessarily p’ entails ‘p’), but there tends to be considerable pragmatic weakening. The presence in the following examples of the (semantically nonharmonic) expressions I suppose, at a guess, presumed, and ‘estimative’ oh provides evidence in support of this claim: (26) (27) (28)
(29)
The shelves must be four foot wide I suppose at least and they’re just they just go up to the roof (ICE-AUS S1A-016 262) At a guess the monkey must have been something like 5ft (1.5m) high standing on its hind legs. (ICE-GB W2B-021 49) I always presumed the child a man has by one woman must be temperamentally different from one he has by another woman. (ICE-GB W2B-004 48) ‘It’s years,’ he said, ‘since I’ve sat on a high stool like this. It must be, oh, eighteen years ago, when Natalie and I lived over in Darlinghurst.’ (ICEAUS W2F-003 56)
Table 3.4 compares Coates’s figures for harmonic expressions such as I’m sure, surely and certain and for ‘hedges’ such as I suppose and I think with those for the present study. In both studies there were more hedges than harmonic expressions, confirming that pragmatic weakening is common with this modal. In the present study ICE-AUS had relatively fewer hedges, perhaps indicating that for Australians must is epistemically weaker than it is for their British and American counterparts.
40
Chapter 3
Table 3.4. Harmonic expressions and hedges with epistemic must Coates Collins
SEU ICE-AUS ICE-GB C-US Total
Harmonic expressions 8 8 8 10 26
Hedges 23 7 13 15 35
Further evidence of the pragmatically weak strength of subjective epistemic must, as noted by Huddleston and Pullum (2002: 181), is to be found in the contrast between the negative must not/mustn’t and its semantically equivalent but pragmatically stronger and more categorical counterpart can not/cannot/ can’t. Consider for example the effect on modal strength of substituting can for must in (30). Rather than an inferred conclusion, the speaker would be felt to be making a statement of what is obvious or perhaps a contradiction of a previous claim. (30)
And secondly he must not be too far off the second person if he doesn’t manage to get second himself (ICE-GB S2B-009 89)
As noted by Coates (1983: 44), and as for epistemic modality generally, epistemic must regularly co-occurs with a range of syntactic features: the perfect aspect as in (27), a stative verb as in (25) and (28), pleonastic/expletive it as in (21), an inanimate subject as in (26), the progressive aspect as in (31) below, and existential there as subject as in (32).2 (31) (32)
Yeah All over him I’m pretty sure he must be sniffing it or something (ICE-AUS S1A-053 11) And there must be a tremendous temptation when you’re when you’ve been labouring away for hours um just to sort of um mark things fairly cursorily (ICE-AUS S1A-032 91)
3.1.3 Dynamic must Dynamic necessity is a minor category with must, accounting for only 6.3% of all tokens (see Table 3.2 above). The clearest cases are those in which must expresses an internal need in the subject referent, as in (33) and (34): (33) (34)
But this realm is arid, sterile, and, ahem, bloodless. The vampires must feed, compulsively, endlessly. (FROWN F01 152) Dugongs are true mammals and must surface to breathe, and so they will inevitably be war casualties (ICE-GB W2B-029 74)
Necessity and obligation
41
These are relatively rare by comparison with cases where the necessity or need for action derives from the force of circumstances, as in the following. (35)
Axon sprouting occurs from the proximal nerve end and must penetrate the fibrous tissue present at the nerve interface. (ICE-GB W2A-026 126)
Dynamic must may not be sharply distinct, especially with a human subject, from deontic must. In (36) and (37) must expresses a circumstantially derived need, involving a habitual time dynamic reading with a non-specific subject. However in each case there is an additional possible reading – albeit one arguably less salient – in which an obligation is understood to be imposed by a deontic source to ensure that the activity occurs, in this case involving a habitual or future situation with a specific subject. (36)
(37)
To get there we must negotiate some of the stormiest oceans in the world deadly icebergs and several hundred kilometres of pack ice (ICE-AUS S2B-035 58) It’s from this land that we must produce all the food all the minerals all the energy and all the potable water for a rapidly increasing population (ICE-AUS S2B-021 46)
Examples such as (36) and (37) may suggest the legitimacy of positing a single root category, as does Coates (1983), who regards both deontic and dynamic must as belonging to a single ‘root’ continuum dubbed ‘obligation’. The closest to an example of dynamic modality discussed by Coates is Clay pots … must have some protection from severe weather (1983: 35), which she describes as having a “very weak” sense of obligation and “minimal” speaker involvement. However examples such as (33)–(35), where there is no element of meaning that could plausibly be associated with speaker involvement, suggest the need to recognize a distinct dynamic category. 3.1.4 Time reference of situation In this section we consider the temporal relationship between the modal meaning and the situation referred to. With deontic must the possibilities differ according to whether the meaning is subjective or objective. Posteriority is possible in both cases (as for example with subjective must in (3) and objective in (8) above), as is simultaneity (as for example with subjective must in (17) and objective in (7) above). However anteriority is not possible with subjective deontic must (since it is not possible pragmatically to oblige someone to do something in the past), although occasional examples are found with objective deontic must used with general requirements as in (38):
42
Chapter 3
(38)
In order to get these credits: the course must have started before you were 21, and you must not have left the course before the beginning of the tax year in which you were 18 (ICE-GB W2D-002 12)
With epistemic must there are no restrictions on the temporal relationship between the modal meaning and the situation. In (39) we have simultaneity and in (40) anteriority: (39)
(40)
But it seems to me that the subject that I teach Yiddish must be pretty high up on the list of subjects about which misinformation abounds and where the true facts are widely ignored (ICE-GB S2B-042 10) I was working, so it must have been a Tuesday or a Wednesday or possibly a Thursday, those were the nights I worked downstairs. (ICEAUS W2F-016 34)
Palmer (1990: 54) reports no examples in his SEU data of must relating to present habitual activity or future situations, while at the same time recognizing that these are fortuitous omissions, given the acceptability of invented examples such as He must travel to London regularly indicating habitual activity and Something must happen next week indicating a future event.3 In the present data there is no shortage of examples (6.1%) of the former, as in (41), and a small number of the latter (1.5%), as in (42). (41)
(42)
Ah the calories that the the mother must consume in breastfeeding and also in the tasks of gathering food while she is nursing the baby mean that there’s nothing left there’s no nourishment left to re- establish ovulation (ICE-AUS S2B-041 172) What with France’s torrential rains, floods and ice, California’s awful earthquakes and our dreadful bushfires, and the wars that are going on here and there over the world, things are not very pleasant these days, are they? They must get better, surely! (ICE-AUS W1B-010 71)
Must itself does not have a preterite form that can locate the modal meaning in past time. For deontic must, the semantic gap is normally filled by had to (see Section 3.5.1 below). This does not of course prevent must from being used in backshift, whether deontic as in (43) or epistemic as in (44), and in contexts where an interior monologue can be hypothesized, whether deontic as in (45) or epistemic as in (46): (43)
Apparently he had some difficulty in persuading the conservative English monks that the ruined choir must be pulled down completely, and even so they kept much of the wall and the eastern transepts with their Norman towers. (ICE-GB W2B-004 48)
Necessity and obligation (44)
(45)
(46)
43
She was born on the 8th which is Roland’s birthday and after trying all afternoon to ring him for that from Montecalim we knew something must be happening. (ICE-AUS W1B-009 40) By dusk I came in view of the spires. I took a room in a public house because next morning I must present myself spruce for business. (ICEAUS W2F-014 49) Refuge in the US or British embassies was not worth thinking about, when both were situated in the middle of town where patrols must surely intercept them. (ICE-GB W2F-015 139)
3.1.5 Must and negation Must normally takes internal negation (i.e. the negation falls within the scope of the modal), as in the deontic example in (47) paraphraseable as “It is necessary that applications received after the cull not be considered”, and the epistemic example in (48) paraphraseable as “It is necessarily the case that he is not too far off the second person”. (47) (48)
But applications received after the cull must not be considered. (ICE-AUS W2D-001 147) And secondly he must not be too far off the second person if he doesn’t manage to get second himself (ICE-GB S2B-009 108)
Negation accounts for 6.0% of deontic and dynamic must tokens, but is extremely rare with epistemic must (with merely three tokens in ICE-GB and none in ICE-AUS or C-US, or 0.5% overall).4 3.1.6 Must: regional and stylistic variation Mair and Leech’s (2006) findings indicate that must has been in substantial decline since the early 1960s, with a frequency difference of 29.0% between LOB and FLOB and 34.4% between Brown and Frown (see Table 1.4 above). A comparison of the frequencies for the individual corpora shows must to be conforming to the trend for AmE to be leading the way in the decline of the modals with C-US having considerably fewer tokens (402 per million words) than ICE-GB (675) and ICE-AUS (613). Not surprisingly, in view of its waning fortunes, must is less popular in speech than writing in all three corpora (with an average speech/writing ratio of 0.6:1 (see Table 3.5). It is with the deontic meaning that the regional differences are the sharpest. As Table 3.2 above shows, Americans have a strong distaste for deontic must, with C-US yielding only slightly more than half as many tokens per million words (209) as British speakers (391) and Australians (369). At the same time the dispreference for deontic must in speech over writing is stronger in C-US (1:3.13) than it is in ICE-GB (1:2.54) or ICE-AUS (1:2.48): see Table 1
44
Chapter 3
(Appendix). Furthermore, examination of the spoken and written subcategories of ICE-AUS and ICE-GB is suggestive of conservatism in the use of deontic must: the frequency of tokens in monologue outweighs that in dialogue by a ratio of 616:378, or 1.6:1, while that in printed texts outweighs that in non-printed texts by a ratio of 1,220:1,100, or 1.1:1.5 Table 3.5. Genre distribution of must Spoken
Written
Total 3.2
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 511 (184) 471 (113) 495 (297) 980 (98) 727 (218) 790 (316) 613
ICE-GB 553 (199) 488 (117) 527 (316) 560 (56) 1,010 (303) 898 (359) 675
C-US
Avg
318 (37)
447
525 (42) 402 (79)
738 563
Should
Coates’s (1983: 60) ‘fuzzy set’ for should has strong obligation at its ‘core’. However in prototypical cases its strength is weaker than that of must, and greater than that of may, as suggested by the contrast between the harmonic combinations must surely, should probably and may possibly. More appropriate, then, is Huddleston and Pullum’s (2002: 177, 186) characterization of should as expressing ‘medium strength modality’. While they distinguish should (and ought to) from both ‘strong’ must, have (got) to and need (to) and ‘weak’ (possibility) can and may, they nevertheless concede that “intuitively it is closer to the strong end than to the weak” (2002: 177). Should expresses mainly deontic modality, with a secondary epistemic meaning (see Table 3.6). No examples were found of should expressing dynamic modality. Even in those instances where should expresses the desirability of an action deriving not from the speaker, or from some moral or legal consideration, but merely from circumstantial expediency, as in (49) below, we understand the action to be recommended by the speaker or by some external body representing the deontic source. (49)
You may need to grip down and adjust the ball position for some shots but the basics of the swing should be the same. (ICE-AUS W2D-011 214)
In addition, should has a number of minor uses – as a ‘quasi-subjunctive’, and as a tentative preterite 1st person variant of would – which are examined in Section 3.2.5.
Necessity and obligation
45
Table 3.6. Meanings of should Deontic Epistemic Subjunctive = would Indeterminate Total
ICE-AUS 826 134 61 11 109 1141
ICE-GB 721 112 104 57 130 1124
C-US 601 (118) 122 (24) 41 (8) 15 (3) 71 (14) 850 (167)
Total 2,148 (1665) 368 (270) 206 (173) 83 (71) 310 (253) 3,115 (2,432)
% 69.0% 11.8% 6.6% 2.7% 10.0% 100%
3.2.1 Deontic should We have characterized should as having ‘medium strength modality’. However within the section of the scale it occupies it is possible to identify differences, even if maximally strong should does not not rival maximally strong must in strength, and maximally weak should likewise does not rival maximally weak may in weakness. Consider the following examples of stronger should. In (50) the speaker makes a forceful suggestion, in (51) the speaker reports in categorical terms a prohibition that is generally known to apply, and in (52) the speaker makes a suggestion whose strength is clear from the implied seriousness of the consequences of non-fulfilment. (50) (51)
(52)
You should quit (ICE-AUS S1A-036 291) He was actually one of my students but I don’t know anything about supplementaries and the grounds that you apply for them and stuff like that and I from what I know about this place is that you should never talk about something you don’t know (ICE-AUS S1A-032 14) Burmese seem to be lactose intolerant and should not be given milk. (ICEAUS W2D-019 18)
By contrast in (53) the weak strength of the advice offered is indicated by the adjunct maybe, while in (54) there is no sense of obligation or duty in the recommendation reported, merely desirability: (53) (54)
Well maybe you should just let things let him think about what he’s doing first (ICE-AUS S1A-093 214) folklore in Kalgoorlie has it that you should watch a certain taxi driver (ICE-AUS W2B-017 47)
Deontic should may be subjective, indicating what the speaker considers desirable, appropriate or right (as in (50) and (53)), or objective, where the appropriateness or desirability of the course of action described stands independently of the speaker’s endorsement (as in (51), (52) and (54)). One reason for the comparative lack of strength of deontic should, as against must, is the fact that should allows for non-actualization, as in (55)–(57).
46
Chapter 3
Here should is used when it is known that the situation is not or was not actualized, and in these cases criticism is implied. (55) (56)
(57)
Well the book’s about two hundred and fifty pages long and we’re not really doing what we should be doing at the moment (ICE-GB S1A-008 7) I know that I probably I know that I should eat but when and I cook uh considerable quite a large quantity of food and then find that I I don’t feel all that hungry even though mostly uhm I usually skip breakfast and uhm travel on cups of coffee or tea (ICE-GB S1A-059 48) He think he thinks like I should have finished it by now (ICE-GB S1A084 238)
As Palmer (1990: 123) observes, the implication of non-actualization may be associated with the presence of a comparative adverb or adjective, conveying the implication that the subject failed to reach some suggested standard, as in: (58)
And I used to think that she should have had more courage (ICE-GB S1B046 63)
With should referring to the future the implicature is not one of nonactualization but rather of openness or at least neutrality with respect to actualization. Accordingly, in (59) below a continuation such as ‘but it certainly won’t be’ would be possible, but if must were substituted for should such a continuation would produce an unnatural effect. (59)
The budgetary status and viability of the hospital redevelopment should also be examined. (ICE-AUS W2E-006 166)
As Coates (1983: 60) observes, deontic should is often used idiomatically in why-questions, as in: (60)
Why should Australians bother to learn Chinese, Japanese, Arabic or European languages? (ICE-AUS W2E-003 28)
These can be paraphrased as ‘Is x really necessary?’, with the implicit answer being ‘no’. Hence they may convey a tone of impatience on the part of the speaker with respect to the supposed obligation. 3.2.2 Epistemic should As with deontic should, so with epistemic should, a weaker meaning than that expressed by must is normal. Epistemic should is typically subjective, with the speaker indicating a tentative assumption, or assessment of the likelihood of the predication, as in (61) and (62). However occasional examples of objective
Necessity and obligation
47
epistemic should are encountered in the data, as in (63), where the proposition is deduced from known facts: (61)
(62)
(63)
Under Wayne Goss’ Labor state government, Queensland appears to have set a sensible course towards sustainable development and controlled foreign investment which should ensure the state’s continued economic growth into the next century while red tape and bureaucratic intervention remains low. (ICE-AUS W2B-015 19) You should receive notification next week some point telling you whether or whether you haven’t got any money from the fund (ICE-GB S1A-078 12) It would be interesting to look at the Xist levels in X M O mice, if the parental imprint is erased before random X inactivation occurs there should be no difference between Xist expression in X P O and X M O mice. (ICE-AUS W1A-005 48)
As with deontic should, so with epistemic should, we find rhetorical whyquestions, in this case the speaker indicating irritation with an invalid assumption. (64)
Uhm why should the stratigraphic divisions that we’ve established in Britain be of use in Australia or China (ICE-GB S1B-006 168)
As Huddleston and Pullum (2002: 187) observe, there is often a deontic component of meaning in the background with epistemic should. For example should in (64) focuses on the likelihood that British stratigraphic divisions will be useful in Australia or China, which is presented as a false assumption, but there is also at the same time a suggestion that it would not be appropriate or wise to proceed with using these divisions. What Huddleston and Pullum conclude from this is that with should the deontic meaning has primacy over the epistemic. 3.2.3 Should and temporality Deontic should can be associated with a situation that is simultaneous with the time of the modality as in (51), (52) and (55) above, or posterior to it as in (50) and (53) above. When deontic should is used with perfective have the pastness applies to the modality rather than to the proposition. Thus we would paraphrase (57) above as “I was under an obligation to finish it by now”, with the perfect outside the scope of the deontic modality, rather than “I am under an obligation to have finished it by now”, with the perfect within its scope. With epistemic should by contrast there are no temporal restrictions on the proposition. Thus in (65) we have an inference made with respect to a simultaneously present situation, in (66) with respect to a future situation, and in (67) with respect to an earlier situation. A comparison of (57) and (67) reveals
48
Chapter 3
that – as we shall see with auxiliary need (see Section 3.4.4) – the scope of the perfect differs according to whether the modality is deontic or epistemic. Whereas in (57) it is outside the scope of the deontic modality, in (67) it is within the scope of the epistemic modality. (65) (66) (67)
If you’ve used the bevel correctly the marks should have your legs pointing in that direction (ICE-AUS S2A-054 150) At least you should be able to get some to sleep (ICE-AUS S1A-023 140) My big news - I have a full-time permanent job! It’s at the university’s pr PR dept, as Publications Officer (Academic). I’m excited - and getting a little nervous as the day draws nearer. By the time you get this letter, I should have started there! (ICE-AUS W1B-007 170)
3.2.4 Should and negation Should (both deontic and epistemic) normally takes internal negation (where the negative is within the scope of the modality), expressing the speaker’s commitment to the wrongness or undesirability of the proposition. In (68) the speaker urges the addressee to refrain from taking the items in question out of the bag; and in (69) the speaker asserts the probability that no substantial effect will occur. (68) (69)
You shouldn’t take them out of the bag. No taking them out of the bag (ICE-AUS S1A-067 252) Overall, there shouldn’t be any substantial effect either way. (ICE-GB W1A-017 235)
Having said this, there is no clear pragmatic difference between internal and external negation with should, as a medium strength modal (or, as Coates 1983: 64 suggests, as a modal which exhibits merger). (68) above could equally be paraphrased as “It is advisable that you do not take them out of the bag” or as “It is not advisable that you take them out of the bag”; (69) could equally be paraphrased as “It is likely that there will be no substantial effect either way” or as “It is not likely that there will be any substantial effect either way”. 3.2.5 Special ‘low-degree’ uses of should Should has a number of uses which Coates (1983: 67) and Leech (2003: 233) refer to as ‘quasi-subjunctive’, in which it occurs in certain (mainly subordinate) constructions with low-degree modality (that is, with should contributing little discernible modal meaning to the construction). These account for 6.6% of tokens in the corpora. A further use of should which is distinguishable from its major deontic and epistemic uses, is discussed in Section 3.2.6 below. As Table
Necessity and obligation
49
3.6 above shows these uses are considerably more common in ICE-GB than in ICE-AUS and C-US. Huddleston and Pullum (2002: 187) list five types of subordinate construction in which low-degree should may occur. Each is discussed and exemplified below. i.
Mandative
Here should serves as a more informal alternative to a subjunctive and, in the majority of cases, the predicative item in the matrix clause forms a harmonic combination with the modal, as in: (70) (71)
(72)
It is desirable that the robot should be deflected when it is kicked so that the cow is not harmed. (ICE-GB W2A-033 51) A Scotland Yard disciplinary tribunal has recommended that seven police officers, who while off duty were involved in an attack on a man in a pub three years ago should be sacked (ICE-GB S2B-019 55) Are you suggesting are you suggesting that they should carte blanche hand you twenty three million dollars (ICE-AUS S1B-054 176)
In these cases the subjunctive use co-exists readily with the deontic meaning of should (in (70), for example, either “It is desirable that the robot be deflected” or “The robot ought to be deflected”). Should can also combine with a predicative item of strong modality in the matrix clause (insistent in (73) and important in (74)), suggesting that it has undergone grammaticalization in the mandative construction (as noted by Huddleston and Pullum 2002: 187-188). (73)
(74)
ii.
His task is to convince those powers, notably the Soviet Union and China but including France, which have been most insistent that force should be used only if specifically authorised by the UN, that their strict interpretation of legality cannot be a pretext for inaction or appeasement. (ICE-GB W2C-008 116) I’ve got another six days leave but I really think it’s important that I should travel from Rome to Nice and then Duseldorf (ICE-AUS S1B-061 241) Adversative
While there are no corpus examples of should in a lest-clause (as in We were worried lest he should feel left out), the following involves a type of adversative clause.
50
Chapter 3
(75)
They camp on rocky islands in case the ice should suddenly break out (ICE-AUS S2B-029 53)
iii.
Purposive
This category is exemplified in (76): (76)
There was a delib deliberate effort to make it appear surgical and almost consequence free uh in order that that public opinion at home should not should not be eroded (ICE-GB S1B-031 98)
iv.
Emotive
In this use should occurs primarily with predicative items indicating surprise or evaluation, as in: (77) (78)
(79)
v.
It’s interesting that the men should think feel like that because I think that most women would think or look great (ICE-AUS S1A-059 30) And it didn’t strike you as anything odd in in that that the Midland Bank as far as as you say should should just lend money half a million pounds to R C Ward simply on the strength of the value of the property without examining the individual accounts (ICE-GB S1B-061 58) It is perhaps not so surprising that exporters should enthuse about being freed from the ups and downs and the extra costs of doing business in Marks, Francs, Lira and so on, because that would make their job easier. (ICE-GB W2E-009 11) Conditional
According to Huddleston and Pullum (2002: 188), in this use should “expresses slightly greater doubt than the non-modal counterpart” (namely, in the case of (80) and (81) below, “if they somehow reached inside the diplomatic compound” and “if anything is wrong with swimming pool steps”). (80)
(81)
legal representation at a rigged trial was the best they could hope for if they should somehow reach inside the diplomatic compound. (ICE-GB W2F-015 142) but if anything should be wrong with swimming pool steps, as we’ve seen in this case it’s a potential source of injury isn’t it (ICE-GB S1B-067 117)
More common than if-clauses with should in the corpus data are implicit conditional clauses featuring subject-auxiliary inversion (in which should is not omissible), as in:
Necessity and obligation
(82)
(83)
51
I can only hope that I will be able to provide the support, as selflessly as you both have done, to you, should you ever require it. (ICE-AUS W1B013 20) This means that a taxpayer can now appeal to the Administrative Appeals Tribunal should he or she disagree with the commissioner’s decision on whether to allow a particular deduction. (ICE-AUS W2C-020 32)
3.2.6 ‘Preterite’ should Should should be analyzed as the preterite counterpart of shall? The fact that should cannot be used with independent past time meaning (unlike could, would and sometimes might) would imply that the answer is ‘no’. However, as Huddleston and Pullum (2002: 202) note, the possibility of a ‘yes’ answer is suggested by examples such as (84) and (85) involving backshift (in the case of (85) the resolving is located in past time by the semantics of recall), and (86) and (87), where the modal occurs in the apodosis of an unreal conditional. The present tense counterparts of these examples would feature shall in place of should (for example “He is determined that as prime minister he shall have greater control”): (84)
(85)
(86)
(87)
He was determined that as prime minister he should have greater control over policy areas and that key policy initiatives were implemented by the bureaucracy. (ICE-AUS W2A-012 34) I think I recall resolving with Zix that I should wear black tie, or rather, the pink bow tie that she bought for me when we went to a ball in Cambridge, plus the same dinner jacket & matching pants that I wore on that occasion. (ICE-GB W1B-015 35) Had I spent it in some other hostelry, I should now be returning to Oxford with a mind untroubled by any more disquieting burden than my responsibilities as Tutor in Legal History at St George’s College. (ICEGB W2F-011 7) If he had been taught by vigilant professors as he says, the wanderings of fancy would have been restrained, and I should have escaped the temptations of idleness. (ICE-GB W1A-018 95)
In these examples should is substitutible by would (and in fact in (87) alternates with would), and normally occurs with a 1st person subject (example (84), from the Australian corpus, being exceptional in this regard). Of the three corpora examined it is only ICE-GB (with 57 tokens: see Table 3.6 above) that has an appreciable number of tokens instantiating this use of should. The relative paucity of examples in ICE-AUS and C-US suggests that should has been almost completely reanalyzed as a form lexically distinct from shall in AmE and AusE.
52
Chapter 3
One semantic development from the modally remote use of should is its politely tentative, formulaic, use – restricted to 1st person subjects and again substitutible by would – as exemplified in (88)–(90): (88) Well I should think that it will be more than slightly and it will be less than twenty or thirty years time (ICE-AUS S1B-024 29) (89) Oysters I should imagine (ICE-GB S1A-009 296) (90) I should like to help you as much as I can when you come, but unfortunately our flat is too cramped to accommodate more than me and Zix (ICE-GB W1B-015 12) 3.2.7 Should: regional and stylistic variation The findings reported by Leech (2003), and Mair and Leech (2006), indicate that should has undergone a mild decline in both British writing (11.8%) and American writing (13.5%) in recent decades (see Table 1.4 above). According to Leech (2003) this decline has occurred mainly at the expense of the epistemic and ‘minor’ uses of should rather than its deontic meaning. In the present corpora the frequency of the ‘minor’ uses (i.e. ‘subjunctive’ + ‘=would’ in Table 3.6) is markedly smaller in C-US (56) and ICE-AUS (72) than in ICE-GB (161). The frequency of should – like that of must – is substantially smaller in CUS (850 tokens per million words) than it is in ICE-AUS (1141) and ICE-GB (1124). Should is also distributionally similar to must in its greater degree of representation, consistently across the corpora, in writing over speech (1.2:1). This is no doubt one factor in the declining popularity of this modal. Interestingly, deontic should shows a stronger tendency to be associated with the written word (with a writing/speech ratio of 1.21:1) than does epistemic should (1.06:1): see Table 2 (Appendix). Table 3.7. Genre distribution of should Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total 3.3
ICE-AUS 1,208 (435) 821 (197) 1,053 (632) 1,300 (130) 1,263 (379) 1,273 (509) 1,141
ICE-GB 1,111 (400) 942 (226) 1,043 (626) 1,060 (106) 1,307 (392) 1,245 (498) 1,124
C-US
Avg
764 (89)
953
975 (78) 850 (167)
1,164 1,038
Ought to
The marginal modal ought to is generally regarded in the literature as being semantically close to, and often interchangeable with, the central modal should. For example Huddleston and Pullum (2002: 186) claim that “In its most frequent
Necessity and obligation
53
use should expresses medium strength deontic or epistemic modality and is generally interchangeable with ought (+ to)”. Coates (1983: 69) states that ought to “has a Root and an Epistemic meaning, both of which are often synonymous with those of SHOULD.” According to Palmer (1990: 122) “It is not at all clear that (…) English makes any distinction between SHOULD and OUGHT TO. They seem to be largely interchangeable”. Quirk et al. (1985: 227) refer to the “synonymous use” of should and ought to in expressing ‘necessity’ and ‘obligation’. The alternation that is sometimes encountered between should and ought to, as in (91), lends credibility to these claims. (91)
Presumably if we’re going to distribute the students in this way according to pro pro rata according to the student uh distribution in the uh college we ought to be doing the same thing within the faculty should we so a big department has more representation than a small department (ICE-GB S1B-075 82)
While ought to and should are very similar semantically, the two modals are anything but similar in their frequency of use, with should over nineteen times more frequent than ought to in the present study (2432:126). According to Harris (1986) the low frequency of ought to in contemporary English is due to the fact that, unlike have to and need to, it has failed to develop the syntactic properties of a lexical verb (structures such as they didn’t ought to help him are certainly attested in some dialects, but are not found in Standard English), while at the same time its requirement of a to-infinitive has ensured its marginal status as a modal auxiliary. Table 3.8. Meanings of ought to Deontic Epistemic Indeterminate Total
ICE-AUS 33 1 2 36
ICE-GB 74 4 2 80
C-US 51 (10) 0 (0) 0 (0) 51 (10)
Total 158 (117) 5 (5) 4 (4) 167 (126)
% 94.6% 3.0% 2.4% 100%
3.3.1 Deontic ought to Like should, deontic ought to can never be as strong as must, but nevertheless may convey a forceful representation of what the speaker regards as appropriate or right, as in (92), where the speaker refers to his uncompromising position in the matrix clause, and in (93), where it is the strongly prescriptive nature of certain contracts that raises the ire of the speaker. (92)
it’s our belief as we’ve conveyed to the government that there is sufficient constitutional power for the Federal Government to move to put in place
54
Chapter 3
(93)
ah greater protections um for people who ah are HIV affected and I’ve got a very strong view that that ought to happen (ICE-AUS S1B-028 68) Uh and then having got that at the end of the day it’s got to be your decision anyway because I won’t enter into a contract that says you ought to go and do that (ICE-GB S1A-035 123)
At the weak end of the strength continuum for ought to, as for should, are cases that merely suggest that something would be a good idea, as in: (94)
I’ve never done it before partly because all my friends thought it was a stupid idea but following Bernadette’s experience of finding her father and then him dying the following year I thought I ought to track mine down. (ICE-GB W1B-003 116)
Like should, ought to is more commonly subjective than objective, but the proportion of objective cases is higher with ought to. Subjective ought to is exemplified in (95), where the speaker is giving advice authoritatively to the addressee, objective ought to in (96), where generally accepted standards of appropriate behaviour are being invoked. (95)
(96)
There were things he didn’t tell you and things he did tell you but I suggest to you the things he didn’t tell you were stupid things and not really things which you ought to use in your deliberations when you’re considering his evidence (ICE-AUS S2A-063 124) Uhm B deals with failure to do what one ought to do anyway (ICE-GB S2A-069 12)
It is probably the greater incidence of objective tokens with ought to that has prompted claims (Aarts and Wekker 1987; Gailor 1983: 348) that ought to is more often associated with the expression of duty than should (along with the stronger dominance of deontic over epistemic meanings with ought to: see below). As with should there are no clear instances of dynamic modality: the closest we get are deontic cases like (97) where what the speaker considers appropriate is based more on expediency than on moral considerations. (97)
It was a bit too near in time to appreciate having it all set up as if they were still around - dining table set for a banquet with very blue venetian wine-glasses & plates which would not improve one's appetite; afternoon tea tray ready in the drawing room (the only really elegant room) & all his court robes & other uniforms standing about in a headless gathering. I heard a lady say in a worried tone: “They ought to be behind glass or they’ll rot.” (ICE-AUS W1B-009 10)
Necessity and obligation
55
Ought to may convey the same type of implication of non-actualization with present and past time situations that is found with should. For example in (97) above the implication is that the uniforms and other paraphernalia are not behind glass, and in (98) below that the payment was not made: (98)
Your client has stated to various persons that the business has earned in excess of $350,000.00 profit in the last year. As a 50% partner in the business, one half of that profit ought to have been paid to our client and disclosed in his taxation return. (ICE-AUS W1B-20 172)
The present corpus data provide some support for the suggestion of Close (1981: 121) and Gailor (1983: 348-9) that ought to can indicate a lesser degree of likelihood of actualization of the proposition than should.6 In (99), which carries an implication that the obligation has not been fulfilled, and in (100) an implication that it is not being fulfilled, should would not substitute readily for ought to unless it were stressed: And they go public to to expatiate on what they think ought to happen and what has happened (ICE-AUS AUS-S1B 76) (100) and say something a a very short something I may say about teaching itself uh what it is that teachers are or ought to be being taught to do or or encouraged to become (ICE-GB S2A-021 8) (99)
3.3.2 Epistemic ought to With most epistemic tokens there is an additional deontic reading that lies in the background. According to Huddleston and Pullum (2002: 187) this statistical fact suggests the primacy of the deontic meaning with ought to (and should): “there are many cases where the interpretation is purely deontic, but few where it is purely epistemic”. Consider some examples: (101) an alternative view which I happen to notice that Lor Lord Annan was uh expressing the other day in the House of Lords and uh uh uh for what it’s worth I happen to hold myself is uh as a country we we plainly are much less well educated than we ought to be (ICE-GB S2A-021 29) (102) As they glide past the sixty-year-old mark they’re as lively as we imagine twenty-year-olds ought to be (ICE-GB S2A-040 38) In (101) the epistemic meaning (“than it can safely be predicted that we are”) is shadowed by a deontic component (“than we have a duty to be educated to that level”), and likewise in (102) the epistemic meaning (“it is likely that twentyyear-olds are”) is in the foreground and the deontic reading (“it is fitting/appropriate/reasonable for them to be”) is in the background.
56
Chapter 3
3.3.3 Ought to and temporality Ought to behaves like should with respect to temporality. Deontic ought to can be associated with a simultaneous situation, as in (91) above, or a posterior situation as in (92) above, and when it is used with perfective have the anteriority is associated with the modality rather than with the proposition, as in (98) above (= “one half of that profit was supposed to be paid to our client”). Epistemic ought to is temporally unrestricted: (101) and (102) above exemplify simultaneity, (103) below posteriority and (104) below (anteriority): (103) And it’s that prospect of chaos in a nuclear world that ought to I believe concentrate our minds (ICE-GB S2B-047 9) (104) I was the bloke who ought to have been nervous. (ICE-AUS W2F-010 63) 3.3.4 Ought to and negation Like should, ought to normally takes internal negation, though again there is no clear distinction between internal and external negation. Thus in (105) the speaker asserts the desirability of it not being forgotten, but could equally well be construed as asserting the undesirability of it being forgotten: (105) it ought not to be forgotten (ICE-GB S2A-019 103) 3.3.5 Ought to: regional and stylistic variation Ought to is, according to Leech (2003), and Mair and Leech (2006), in severe decline, with a frequency drop of 44.2% from LOB to FLOB, and of 30.0% from Brown to Frown (see Table 1.4 above). Yet it may be premature to consider ought to moribund, at least in all regional varieties. Despite its small numbers in the present study, ought to was found to be considerably more robust in British and American speech than writing (the speech/writing ratios for ICE-GB and CUS respectively being 2.8:1 and 3.8:1). By contrast, the future appears to be considerably bleaker for ought to in ICE-AUS. Not only are its numbers small (36 tokens) but it is also strongly dispreferred in speech over writing (by a ratio of 0.28:1). The frequency of ought to in dialogue bears this message even more clearly: with 6 tokens per million words in ICE-AUS dialogue it is here is about six times (6.3) less popular than it is in monologue and nearly ten times (9.7) less popular than it is in writing.
Necessity and obligation
57
Table 3.9. Genre distribution of ought to Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total 3.4
ICE-AUS 6 (2) 38 (9) 18 (11) 40 (4) 70 (21) 63 (25) 36
ICE-GB 136 (49) 67 (16) 108 (65) 80 (8) 23 (7) 38 (15) 80
C-US
Avg
69 (8)
65
25 (2) 51 (10)
42 56
Need
3.4.1 Meanings of need The extremely small numbers for auxiliary need (56) contrast dramatically with those for its quasi-modal counterpart need to (716), which is discussed in Section 3.7. Semantically, the auxiliary and quasi-modal are similar, covering the same range of meanings (even though these are not represented in quite the same proportions: see Table 3.10 and 3.16 below). As these tables show, dynamic necessity is the main meaning, accounting for close to two-thirds of the tokens for both need and need to. While epistemic necessity is a major meaning for the auxiliary, it expresses deontic necessity less commonly than the quasi-modal. Need and need to have received less attention in the literature than other modal expressions of obligation and necessity, a situation plausibly attributed by Nokkonen (2006) to their relative infrequency and to the challenge presented by their semantics: “as obligation is prototypically felt to come from a source external to the agent and these two markers have been assumed to express internally motivated obligation” (2006: 35). Table 3.10. Meanings of need Dynamic Deontic Epistemic Indeterminate Total
ICE-AUS 12 2 4 1 19
ICE-GB 21 3 8 2 34
C-US 10 (2) 5 (1) 0 (0) 0 (0) 15 (3)
Total 43 (35) 10 (6) 12 (12) 3 (3) 68 (56)
% 63.2% 14.7% 17.6% 4.4% 100%
Is there any semantic distinction between deontic need and need to? If we confine ourselves to non-affirmative uses, a tendency emerges for auxiliary need to to be subjective, as in (106) – by contrast with the typical objectivity of quasimodal need to (compare (174) in Section 3.7) – confirming Perkins’ (183: 63)
58
Chapter 3
suggestion of an analogy with the distinction between the auxiliary/quasi-modal ‘pair’ must/have to. (106) I think if you look inside the magazine it’s Miss Kylie Minogue, uh you you you you can see this on your, copy, needn’t turn it up now, there is a a a sticker which uh draws particular attention to uh this uh uh article (ICEGB S2A-061 090) That we dealing with tendencies, however, is evident from the occurrence of examples of objective need as in (107) (compare the (rarer) subjective use of deontic need to in (175) in Section 3.7). (107) You need not notify DVLA yourself. (ICE-GB W2D-010 64) Epistemic need expresses objective logical necessity, with the speaker making a deduction from the available evidence, as in (108): (108) And er just just because an accident has happened it doesn’t um mean that the ah the person has driven without due care and attention and um due care and attention then due care and attention need not amount to negligence on the part of the defendant (ICE-AUS S2A-070 85) Dynamic need expresses a need that is located in external circumstances, as in (l09): (109) The plate is held at the bottom by a plastic pin which need not be undone as the plate can be flexed out of the way. (ICE-GB W2D-018 9) Like must, have to, and have got to, deontic and dynamic need can refer to a present requirement that is applied to the actualization of a future situation as in (106), a present situation as in (107), or a past situation as in (192) in Section 3.7 below. The same three possibilities are available for epistemic need: applied to a future situation as in (110), a present situation as in (111), or a past situation as in (112). (110) Now it’s true that uh, the worst scenario uh need not necessarily come to pass (ICE-GB S2A-066 48) (111) That need not mean allied tanks and troops going all the way to Baghdad, though if Saddam is determined to fight this war like his other hero, Adolf Hitler, it could come to that. (ICE-GB W2E-002 44) (112) Thus it would appear that acceptance of territory need not have implied service. (ICE-GB W1A-003 27)
Necessity and obligation
59
Whereas with must negation is internal, with need it is external. Thus, in (113) need not, like don’t have to in the preceding clause, is paraphraseable by “it is not necessary for” (whereas it must not be in one lump sum would have involved negation of the proposition: “it is necessary for it not to be”). (113) Now let’s have the good news. You don’t have to plough in a lot of money to make a big difference - and it need not be in one lump sum. (ICE-AUS W2D-012 302) 3.4.2 Need: regional and stylistic variation Table 3.11. Genre distribution of need Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 0 (0) 25 (6) 10 (6) 20 (2) 37 (11) 33 (13) 19
ICE-GB 11 (4) 67 (16) 33 (20) 20 (2) 40 (12) 35 (14) 34
C-US
Avg
0 (0)
14
38 (3) 15 (3)
35 23
According to Leech (2003), Smith (2003), and Mair and Leech (2006) need has declined in both British and American writing (while by contrast need to has increased massively: see Section 3.7.6 below). The frequencies for the present study suggest that AmE is leading the way in the decline of need, which is less than half as popular in C-US (15 tokens) as in ICE-GB (34), with ICEAUS in-between (19). A relevant factor in the dwindling fortunes of need is its dispreference in speech as against writing (the ratio being 43:106, or 0.4:1) 3.5
Have to
The lexico-modal have to significantly outnumbers its closest semantic ‘rivals’ must (by a ratio of 2827:1367, or 2.0:1) and have got to (2827:705, or 4.0:1): see Table 3.1. Deontic necessity is the main meaning with have to, as it is with must, but more dominantly so for the former. However have to does not rival must in the popularity of the epsitemic meaning, a situation that may well change as have to becomes increasingly grammaticalized, with the likely consequence that its epistemic meaning becomes more established via the process of subjectification.
60
Chapter 3
Table 3.12. Meanings of have to Deontic Dynamic Epistemic Indeterminate Total
ICE-AUS 998 295 5 13 1,311
ICE-GB 902 332 2 8 1,244
C-US 1,099 (216) 255 (50) 20 (4) 10 (2) 1,385 (272)
Total 2,999 (2,116) 882 (677) 27 (11) 31 (23) 3,940 (2,827)
% 76.1% 22.4% 0.7% 0.8% 100%
3.5.1 Deontic have to Do deontic must and have to have the same or different senses? That there is considerable overlap between these two items is suggested by examples where they alternate, as in (114) and (115) (where, it may further be noted, the different orderings undermine the possibility that a particular ordering may imply a difference in strength between the ordered items): (114) Would my right honourable friend not agree, that the mark of a single currency is that all other currencies must be extinguished and not merely extinguished but that the capacity of other institutions to issue currency has to be extinguished and that in the case of the United Kingdom would involve this Parliament binding its successors in a way which we have hitherto regarded as unconstitutional (ICE-GB S1B-053 67) (115) While a day visit for reference use will normally be granted immediately, where a special admission ticket has to be made up, particularly for long term use or borrowing, it may not be possible to issue it immediately, and applicants must be prepared to accept a 24/48 hour delay, and may not be able to borrow on their first visit. (ICE-GB W2D-006 136) While such examples may suggest potential semantic equivalence between deontic have to and must, there are nevertheless some differences that tend to differentiate them. According to Westney (1995: 151) deontic have to focuses on “an external, existent obligation that can be perceived or described independently of the speaker”, whereas deontic must serves as a very general marker of obligation with more specific senses such as urgency, irresistibility, and unconditionality being attributable to pragmatic interpretation. Palmer (1990), Coates (1983) and Perkins (1983) are all in agreement that have to contrasts with must in that its distribution is skewed towards the objective (Palmer’s ‘external’) end of the subjective/objective scale. In other words, deontic have to differs from must in its preference for objective meaning, where the deontic source is external to the speaker, as in (116), over subjective meaning, where the speaker is the deontic source, as in (117). (116) Yeah and if you’re a bit older I think and you have a a broader perspective of things then when you have to do an assignment or a presentation your
Necessity and obligation
61
brain is that little bit broader and it makes it easier to do something original and inventive just being older (ICE-AUS S1A-042 231) (117) P.S. It’s your fault that I’m so homesick, if I hadn’t had such a good holiday I would have been glad to get back to Uni PPS Since it’s your fault, you have to fix it and that means lots of LETTERS and/or PICTURES (as the case may be) xox (ICE-AUS W1B-015 97) Admittedly the subjective/objective distinction cannot be consistently applied to instances of deontic have to insofar as it is commonly used, as Leech (1987: 79) observes, to express a general requirement or obligation without specifying the deontic source. Nevertheless there were a sufficient number of clear cases in the present study to confirm the skewed distribution noted above. What implications might this skewing have for the strength of deontic have to? It would appear to suggest that deontic must is stronger than have to, and this view has certainly found support in the literature. For example Sweetser (1990: 540) asserts that “Must has connotations of a directly applied and irresistible force, while have to, ought and need are resistible forces …”. Alexander (1988: 228) asserts that “must conveys more strongly than have to the idea of inescapable obligation”. However there are some, including McCallum-Bayliss (1985), who regard have to as stronger than must. Given such differences of opinion it would seem that strength is a less useful concept for differentiating these items than subjectivity/objectivity. As with must, so with have to, the dimension of subjectivity/objectivity shows correlations with the person of the subject. The strongest correlation is between 3rd person subjects and objectivity, as in (118), where the deontic source is a legal rule. (118) Manson will have to wait five years for another hearing. (ICE-AUS S2B001 208) Even here, however, examples are found where the speaker appears to be the source of the requirement, but they tend to lack the strong sense of compulsion often found with must, as in (119), where the hedge I think indicates pragmatic weakening. (119) He moved in with her but of course that wouldn’t do. It was just a bachelor pad. Now she’s managed to sublet it and they’ve moved into a bigger unit. I had dinner there last week. Alone. Martha isn’t into doordarkening yet. We’re debating the issue. I think Martha has to support me. (ICE-AUS W2F-020 59) With 1st and 2nd person subjects objective modality again predominates, but at the same time subjective modality is relatively more common than it is with 3rd person subjects. Of the 1st person examples below, (120) and (121) are objective (an external source being implied by the speaker’s expressed reluctance
62
Chapter 3
and by the interrogative mood respectively), while (122) and (123) may be analyzed as subjective in the absence of an externally identifiable source. Notice that the use of have to in (121) resembles that of must with verbs of communication, and in fact have to could be substituted by must without any appreciable shift of meaning. (120) and uhm that’s the tradition which reluctantly I have to follow (ICE-GB S1B-047 15) (121) What exactly do I have to wear (ICE-GB S1B-079 89) (122) Although I would love to I have to yes I have to confess an often irking thought of am I really really two pounds less than Kate Hamilton (ICEGB S1A-011 224) (123) I think we have to be careful with the fabric cos there’s so much of it (ICE-GB S1A-086 204) Examples (124)–(126) occur with a 2nd person subject. Whereas (124), which occurs in a bureaucratic text and has non-specific you as subject, is straightforwardly objective, in (125) the strong advice given to the addressee emanates from the speaker. In (126) the deontic source is somewhat ambivalent: the speaker is reporting a boating regulation while at the same time appearing to personally endorse it. (124) I got a a pocket guide to the law and it it explains how how to write your will so and you have to be specific in in the way you write where what goes like who you’re leaving what to (ICE-AUS S1A-055 346) (125) You’ll have to see if dad’ll pick you up afterwards (ICE-AUS S1A-016 153) (126) If you wanted to start on the opposite tack you wouldn’t ho have right of way and if you’re in a close call situation a crossing situation you’d have to give right of way to the other boats that’re on starboard (ICE-AUS S2A-020 35) 3.5.2 Dynamic have to Dynamic necessity is, as Tables 3.2 and 3.12 show, more commonly expressed by have to (22.4%) than by must (6.3%). This meaning is exemplified in (127)– (130), where no deontic source is identifiable, but rather the factors facilitating the activity reside in the situation, as in (127) and (128), or in the subject referent, as in (129) and (130). (127) Trigger’s still at the lead of the sheep and in Australian conditions that would be acceptable but in New Zealand conditions he has to drive (ICEAUS S2A-016 87) (128) Four forty one point four six is the time Haley Lewis has to beat to break her Commonwealth and Australia record (ICE-AUS S2B-016 43)
Necessity and obligation
63
(129) The same thing happens to those who are confined to wheelchairs, or who have to spend long periods in bed (ICE-GB W2B-022 26) (130) They have to keep eating however cnidarians to constantly replenish those dinoflagellates in their tissues (ICE-AUS S2A-025 77) 3.5.3 Epistemic have to The ascendancy of have to over must that is in evidence with root meanings does not extend to epistemic necessity, with tokens of epistemic have to accounting for less than 1% of all tokens (see Table 3.12). Earlier studies of BrE regard epistemic have to as an innovation. According to Hughes and Trudgill (1979: 23) it is used by younger speakers inspired by AmE, and Coates (1983: 57) similarly associates it with the teenage sub-culture. The present study indicates that while epistemic have to is more common in AmE (with 20 tokens per million words in C-US: see Table 3.12) it has become established in AusE and BrE as well. Examples from all three dialects are given below: (131) And so, the molecules are speeding up, they’re getting further and further apart, and taking up more space inside the balloon, so the balloon goes back to its former size and shape. So that has to be what’s happening to the balloons, that are inside this container here. (C-US SBC 27 903-5) (132) And it’s the nature of a new industry that if if it’s not been available in this country before then you don’t have experience in it and we don’t we don’t believe that ah it has to follow that the only people who are qualified to begin a pay television service are those who’re already in television for the last thirty years (ICE-AUS S1B-046 72) (133) I mean another way of looking at that is that if you have a continuous function when you apply it to an interval, it goes to an interval, and since you can get things as big as you like, this thing has to be an interval. (ICE-GB S1B-013 149) One question to be considered is whether the tendency noted above for deontic must to be subjective and have to objective, is also in evidence with epistemic uses. Matthews (1991) presents this as a categorical difference, claiming that: … whereas must reflects the spontaneous performance of an inference with respect to some new evidence, have to and have got to are assertions of objectively or logically necessary inferences. (Matthews 1991: 235) The present data suggest that the situation is less clearcut than Matthews would lead us to believe. Not only can epistemic must be objective (as noted in Section 3.1.2), but epistemic have to can be subjective. Compare the following:
64
Chapter 3
(134) And so, the molecules are speeding up, they’re getting further and further apart, and taking up more space inside the balloon, so the balloon goes back to its former size and shape. So that has to be what’s happening to the balloons, that are inside this container here. (C-US SBC 27 903-5) (135) It would be the largest released, the tallest bull at the Royal Easter Show and this little chap has to be the smallest (ICE-AUS S2B-039117) Example (134) is objective, paraphraseable as “this is the only possible conclusion that can be drawn”, but (135) is subjective, representing a presumption, or inference, made by the speaker. 3.5.4 Time reference of situation Deontic must and have to tend to differ in that must is usually associated with immediate posteriority as in (136), and have to with ‘habitual simultaneity’ as in (137). (136) There is a crisis and he must act now (ICE-GB S1B-056 89) (137) I mean I do know that it does take up time and I do feel that I have to be there for every activity that I set up and I think that’s probably right I think (ICE-GB S1B-078 214) These are merely tendencies. As (117) and (121) above show, have to can be used with an immediately posterior situation and, despite Coates’s (1983: 54) claim that must can never be associated with habitual present meaning, an example such as (138) indicates that the present requirement expressed by must can be applied to an indefinite number of occasions. (138) Brake shoes must always be renewed in sets of four (ICE-GB W2D-018 67) As with deontic must, so with present forms of have to, anteriority is possible (a present requirement for the past fulfilment of an action) even though this is not mentioned in the literature: (139) And the thir And the third one which was a true exit you have to you ha ha have to have gone with a band to get to it (ICE-GB S1A-073 48) With epistemic have to, inferences can be readily made about a present situation as in (135) above, or about a future situation. While there were no corpus examples involving anteriority or posteriority, that these are merely accidental gaps is suggested by the possibility of substituting have to for must in examples such as (40) and (42) above. With must, as we have seen, there are limited possibilities for the time of the modality to be other than present. Have to by contrast has a preterite form
Necessity and obligation
65
had to, as in (140), where it is deontic, expressing a past requirement, and (141), where it is dynamic, expressing a past circumstantial necessity. (140) When the League of Good Men heard about the gatherings they went to the kafeneio of Yiorgos the Apeface to put an end to all the mischief. But when they got there they had to admit that there was no mischief, that everyone was well behaved. (ICE-AUS W2F-018 22) (141) And the fog the smog was so bad that we actually had to take the children home from school. (ICE-AUS S1A-015 130) Had to can also be used in backshift, as in: (142) I was actually hoping to have ah one with familiarname1 as well but um like as in the same type of lunch but ah this did not eventuate as familiarname1 got an attack of calvinistic bad luck and decided he just had to go and um assuage his guilty feelings (ICE-AUS S1A-099 7) In such cases, according to Coates (1983: 57), “it seems likely that HAVE TO is suppletive to MUST (…), functioning as Past for both MUST and HAVE TO”. Suppletion is difficult to prove, however. It could perhaps be argued that when had to is used in conventional formulae with verbs of communication as in (140) (and when it expresses the dynamic ‘inner compulsion’ meaning as in (142)), that this is more likely to be associated with must than have to in present time contexts. Nevertheless present forms of have to are available for these uses and, furthermore, must would itself be possible as an alternative to had to in the indirect speech context in (142). Deontic (and dynamic) had to can be associated with a situation that is present (or, simultaneous) with respect to a past obligation as in (143), past as in (144), or future as in (145): (143) The Inuit dog was traditionally fed seal meat and had to fend for itself for part of the year and so became a good hunter a trait that invariably causes problems for the dog handlers in the Antarctic (ICE-AUS S2B-029 125) (144) If the Court had held that it was an indemnity, the defendant would have had to have paid the loss suffered. (ICE-AUS W1A-015 176) (145) He had to be up at Asquith by six-thirty and do his little bit and then come home again (ICE-AUS S1A-016 71) Have to also differs from must in its capacity to express a future, anticipated, obligation, in the case of (146) one that is associated with a simultaneous situation, and in the case of (147) with a future situation. (146) I think on your bike you’re going to have to wear one for a while, don’t you (ICE-GB:S1A-022 285)
66
Chapter 3
(147) It’s not a good ball though and Van Den Howe will have to backtrack but he keeps the ball in play for Tottenham (ICE-GB S2A-015 141) Depraetere and Reed (2006) do not allow for the possibility of epistemic modality located in the past or future: Epistemic modality by definition entails the making of a judgement about the likelihood that it is true that something is the case. This means that the modality itself must be located at the time of the judgement – either speech time or some implicitly or explicitly evoked speech (or thought) time. (Depraetere and Reed 2006: 286) However there are several examples in the corpora where had to does in fact express a past epistemic judgement, as in (148). (148) He had the worst job in the crew. Apart from being our navigator he also became the easy target of thousands of mosquitoes. This had to be the ultimate test for a can of Aerogard! (ICE-AUS W2F-010 58) 3.5.5 Have to and negation Whereas, as we have seen in Section 3.1.5, when must is negated the negation falls within the scope of the modal (e.g. He mustn’t go yet means “It is necessary for him not to go yet”), with have to the negation is outside the scope of the modal (He doesn’t have to go means “It is not necessary for him to”). Have to takes external negation whether the meaning is deontic as in (149) (“It is not required that you to tell me again”), dynamic as in (150) (“It is not necessary for me to write much”), or epistemic (no corpus examples were found, but note that (151) is logically equivalent to we believe that it doesn’t have to follow that …, meaning “it is not necessarily the case that it follows”). (149) You don’t have to tell me again (ICE-GB S1A-092 87) (150) Maybe not, but it’s not far from it, as I’m really pissed off with writing letters to everyone so usually send postcards instead so I don’t have to write much! (ICE-GB W1B-002 5) (151) And it’s the nature of a new industry that if if it’s not been available in this country before then you don’t have experience in it and we don’t we don’t believe that ah it has to follow that the only people who are qualified to begin a pay television service are those who’re already in television for the last thirty years (ICE-AUS S1B-046 72) The effect of the scopal difference between must and have to with negation is that the latter may serve as a suppletive for the former. Notice in (151) that the contrast the writer desires is not one that could be achieved using must not/mustn’t – hence the use of the lexico-modal:
Necessity and obligation
67
(152) A skilled word processor operator does not have to look at the screen while typing, but what has been entered must be proof-read and this requires concentration to ensure that it is done properly. (ICE-GB W2B033 63) 3.5.6 Have to: regional and stylistic variation Table 3.13. Genre distribution of have to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 1,828 (658) 1,579 (379) 1,728 (1037) 890 (89) 617 (185) 685 (274) 1,311
ICE-GB 1,578 (568) 1,108 (266) 1,390 (834) 1,260 (126) 947 (284) 1,025 (410) 1,244
C-US
Total
2,069 (241)
1,729
388 (31) 1,385 (272)
699 1,313
The decline of must discussed in Section 3.1.6 is matched by an increase in the frequency of have to (of 9.0% in British writing and 1.1% in American writing: see Table 1.4). There are several possible reasons for the differing fortunes of the two items. Syntactically, have to surpasses must in flexibility, occurring in contexts where the auxiliary would be ruled out on the grounds of its defective morphology. Semantically, the more objective obligation that have to expresses in its (dominant) deontic sense, it has been argued by Myhill (1995) and Smith (2003), appeals to speakers seeking a more ‘democratic’, less authoritarian, tenor. Stylistically, as we shall see, have to is more widely attested in speech. In view of the diachronic findings it is not surprising that the popularity of have to should outweigh that of must in the present study. A comparison of the figures for the three regional varieties (see Table 3.13) shows have to following the general trend for AmE to be leading the way in the decline of the modals and the rise of the quasi-modals (see Section 1.4 above), with a must/have to ratio of 1:3.4 in C-US. In both respects, furthermore, BrE is the most conservative (1:1.8 in ICE-GB), with AusE (1:2.1 in ICE-AUS) occupying a position mid-way between AmE and BrE. An examination of the frequencies of must and have to across speech and writing provides insights into their contrasting fortunes. Must is consistently around 40% less popular in speech than writing, in all three dialects. By contrast have to is approximately two and a half (2.47) times more popular in speech than writing. Furthermore the ordering of the dialects in terms of the relative popularity of have to in speech (C-US 5.3:1, ICE-AUS 2.5:1, ICE-GB 1.4:1) matches the ordering as determined by the frequency of have to tokens. The preference for occurrence in speech that we find with have to is even stronger for deontic have to (2.83:1). Similarly the robustness of have to in speech in AmE is even more evident with deontic have to (whose speech/writing
68
Chapter 3
ratio in C-US is 5.5:1). Finally, deontic have to reverses the order of preference displayed by deontic must in the four genre subcategories: deontic have to is preferred in dialogue over monologue by a ratio of 2,506:2,267, or 1.1:1, and in non-printed over printed texts by a ratio of 1,410:1,043, or 1.4:1. 3.6
Have got to
Have got to occurs commonly in reduced form as gotta (gotta accounting for 67.8% of all tokens in the corpora). Further variant realizations are found, such as ’re gotta in (153) (not surprisingly, in view of the grammaticalization that have got to is undergoing: see Krug 1998). (153) They’re a year old now exactly and they’re gotta be two years old before they’re old enough to breed (ICE-AUS S1B-041 200) Have got to and have to are often treated as variants in the literature but they differ in a number of respects, which we mention briefly here but in some cases develop in more detail below. Syntactically, as noted in Section 2.3.1 above, the semi-modal have got to differs from the lexico-modal have to in exhibiting most of the formal properties of the modal auxiliaries: have got to lacks non-tensed forms (compare *to have got to with to have to, *having got to with having to), cannot cooccur with modals (compare *may have got to with may have to), and exhibits the properties of an operator. Furthermore, whereas the preterite form had to is common, had got to is unattested in the present data (see Section 3.6.4 below). Dialectally, AmE display a marked dispreference for have got to over have to in comparison to BrE and AmE (see Section 3.6.6 below). Stylistically, have got to is rare outside of conversation in the corpus (see Section 3.6.6), and thus contrasts with the stylistic neutrality of have to. Semantically, have got to contrasts with have to in its typical subjectivity and in its typical incompatibility with habitual situations (see further Sections 3.6.1– 3.6.4 below). Table 3.14. Meanings of have got to Deontic Dynamic Epistemic Indeterminate Total
ICE-AUS 278 42 8 4 332
ICE-GB 279 51 2 7 339
C-US 153 (30) 15 (3) 5 (1) 0 (0) 173 (34)
TOTAL 710 (587) 108 (96) 15 (11) 11 (11) 844 (705)
% 84.1% 12.8% 1.8% 1.3% 100%
Necessity and obligation
69
3.6.1 Deontic have got to Deontic have got to resembles must in its capacity to be used more often subjectively, encoding a speaker-based statement of requirement, than objectively, encoding an external, existent obligation that is independent of the speaker. In this respect it differs from have to, which as we have noted is predominantly objective. In the following examples objective have got to can readily be substituted by have to, the consequence being more a matter of a shift towards greater formality than any change of meaning: (154) So I’ve been told I’ve got to do a a month o o a month of reading (ICEGB S1A-093 40) (155) That’s the thing you’ve gotta you’ve gotta really say spelt haven’t you (ICE-AUS S1B-020 196) Predictably, however, subjective deontic have got to in the following examples is more readily substitutible by must than it is by have to. In (156) the obligation is self-imposed by the speaker, while (157) exemplifies the ‘performative’ use that is found with must but which Coates (1983: 53) wrongly claims is never possible with have got to. (156) It’s her birthday so I’ve got to try and find think about what to get her (ICE-GB S1A-025 328) (157) “Leigh, why? Why do you keep latching on to men like that? You’ve gotta stop” (ICE-AUS W2F-017 39) Have got to also has in common with must that it is often used in quasiformulaic locutions with verbs of communication where the speaker appears to be the deontic source, as in: (158) Right Yeah Well I mean Well OK I can understand that but um you’ve gotta remember that um I mean this image of the of the rolling drunk you know the drunken yobbo is I mean it’s an it’s an extreme (ICE-AUS S1A053 76) As with must and have to, so with have got to, the tendency towards deontic subjectivity is strongest with 2nd person subjects as in (157) and (158) above, and weakest with 3rd person subjects as in (159), where the deontic source that is explicitly invoked is the constitution. (159) Remember we we’re we’re faced with a constitution that says that it’s got to be um prevention and settlement of um industrial disputes extending beyond the limits of one state so we’ve got to satisfy the requirement the requirement of interstateness um (ICE-AUS S1B-010 221)
70
Chapter 3
With 1st person subjects the requirement is equally likely to be objective or subjective. In (160) the deontic source is speaker-external, while in (161) it is speaker-internal, with have got to expressing ‘self-compulsion’ or ‘selfobligation’ (Leech 1987: 77): (160) Yeah I’ve got to go to the dentist this afternoon (ICE-AUS S1A-024 61) (161) David I’ve got to ask you this I mean with this is a brand new show for yourself and obviously I wha the reports the reviews have been just fantastic (ICE-AUS S1B-044 261) According to Westney (1995: 127, 151) a further difference between have got to and have to is that the former conveys a sense of immediacy or urgency. However such a sense is elusive and seems not to be in evidence in an example such as (162), where have got to and have to alternate, displaying not only comparable modal strength but also a comparable degree of immediacy or urgency. (162) Oh I’ve got to I’ve got to ring up I have to find out ’cause they’re supposed to call her back today from the interview that she went on Monday (ICE-AUS S1A-013 137) 3.6.2 Dynamic have got to Like deontic have got to, dynamic have got to can alternate freely with have to, as in: (163) and so the rest of the heat that’s got to get to back out into space to balance this uh process uhm has to get away as heat energy (ICE-GB S2A-043 97) Dynamic have got to may be similar in meaning to need (to) in its capacity to express some need that is intrinsic to the subject-referent as in (163), or it may, as have to regularly does, express a need imposed by external circumstances as in (164). (164) I mean you you’ve got to have a big powerful machine to run it or plenty of RAM (ICE-GB S1A-029 242) 3.6.3 Epistemic have got to As Table 3.14 shows, epistemic necessity is a minor meaning for have got to, just as it is for have to, and especially in BrE.7 Like must, epistemic have got to tends to be subjective, as in:
Necessity and obligation
71
(165) Ah, he’s got to have some stashed away somewhere He’s gotta, with that sort of empire he’s gotta have, name a figure, five ten twenty fifty million dollars stashed away somewhere (ICE-AUS S1B-049 79) (166) Loose shirts over jeans has got to be a sort of, temporary prejudice (ICEGB S1A-054 161) Nowhere is this subjectivity more apparent than when have got to combines with a progressive construction, often with an emotive overlay, as in (167). The popularity of this combination is compatible with Melrose’s (1983) claim that have got to is used for expressing affective, or emotionally-loaded, modality. (167) Cass’s eyes go wide. “You’ve got to be kidding.” (ICE-AUS W2F-017 82) 3.6.4 Have got to and temporality Temporally, have got to tends to be closer to must in its characteristic association with immediate posteriority, as in the deontic example in (168), the dynamic in (169) and the epistemic in (170), than it is to have to, which is more often associated with ‘habitual simultaneity’. (168) I’ve got to go and see to the dinner in a minute (ICE-GB S1A-007 219) (169) They’ve got to average sixty-one point five for the last five laps to get them near the qualifying time (ICE-GB S2A-007 66) (170) Not much coming in. Something’s gotta give here sooner or later (ICEAUS S1A-014 90) This is not to deny the possibility of have got to expressing habitual meaning, despite the fact that Coates (1983: 54) disallows this. As with must, occasional examples are found, as in (171): (171) Oh they’re terrible. They’ve always got to know where you are. (ICEAUS S1A-093 88) Present epistemic have got to can be used with a present situation as in (165)–(167) above, or a future situation as in (170). While there were no examples of anteriority, that this is an accidental gap is suggested by the possibility of substituting have got to for must in (40) above. Have got to is also like must, and different from have to, in having no independent past use (had got to) in any meaning. A plausible suggestion for this gap, with respect to epistemic have got to, is offered by Westney (1995: 148): “in contrast to had to, it lacks any suggestion of actuality, and this would be particularly odd for an epistemic use, which in past reference would naturally suggest actuality”.
72
Chapter 3
3.6.5 Have got to and negation Negation of have got to is rare in the corpus. As with have to, negation is external, applying to the predication rather than the modality: the paraphrase “it is not necessary for” is thus applicable to the following two examples. (172) I just want somewhere where I haven’t got to worry about living, where someone will look after me. (ICE-AUS W2F-011 75) (173) That means we haven’t got to keep looking (ICE-AUS S1A-023 15) 3.6.6 Have got to: regional and stylistic variation Table 3.15. Genre distribution of have got to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 739 (266) 217 (52) 530 (318) 10 (1) 43 (13) 35 (14) 332
ICE-GB 789 (284) 167 (40) 540 (324) 80 (8) 23 (7) 38 (15) 339
C-US
Avg
266 (31)
445
38 (3) 173 (34)
37 281
As we have seen in Section 3.5, the overall frequency of have got to is one quarter that of have to and one half that of must in contemporary English. For this quasi-modal the diachronic profile that is offered in Table 1.4 is somewhat unrevealing. It is based on written English and thus does not reflect the most sweeping change that is under way; namely, as Krug (2000: 63), Smith (2003), and Leech (2003) all note, a strong increase in informal spoken usage. As Table 3.15 indicates, in the present data have got to is overwhelmingly more common in speech than it is in writing (by a ratio of 1,366:111, or 12.3:1), and furthermore almost four times more common in dialogue than monologue in the British and American corpora (1,500:381, or 3.93:1). It is highly likely that the traditional stigma attached to got, especially in more formal styles, has had a role to play in this generic imbalance. Have got to bucks the trend found with most of the ‘major’ ascendant quasi-modals (compare have to, be going to, want to) for AmE to be the most innovative of the dialects and to have the highest frequency of tokens in the present corpora. In fact C-US lags well behind the other two corpora, with only about half as many tokens as each (C-US 173, ICE-AUS 332, and ICE-GB 339).
Necessity and obligation 3.7
73
Need to
As noted in Section 3.4.1 above, the numbers for quasi-modal need to (716) are significantly larger than those for auxiliary need (56). Like its auxiliary counterpart, need to expresses mainly dynamic necessity, but differs from it in the relative proportions of deontic tokens (28.9% for need to and 14.7% for need) and epistemic tokens (2.5% for need to and 17.6% for need). Table 3.16. Meanings of need to Dynamic Deontic Epistemic Indeterminate Total
ICE-AUS 217 99 6 21 343
ICE-GB 158 96 11 15 280
C-US 305 (60) 122 (24) 10 (2) 36 (7) 473 (93)
Total 680 (435) 317 (219) 27 (19) 72 (43) 1,096 (716)
% 62.0% 28.9% 2.5% 6.6% 100%
3.7.1 Deontic need to Consider the basic difference between must, have to and need to. In a sentence with a 1st person subject as in I must/have to/need to get ready, must would tend to suggest self-obligation arising from a personal sense of duty, have to an obligation arising from a source external to the speaker, and need to a compulsion arising from within the speaker. It is from this sense of internal compulsion that the deontic meaning of need to derives. Statements pertaining to an addressee’s needs come – via indirect illocutionary force – to serve the role of recommendations or exhortations. As Smith (2003: 260) observes, need to “can acquire the force of an imposed obligation, but – something which does not apply to the other markers – the writer or speaker can claim than the required action is merely being recommended for the doer’s own sake”. In Section 3.4.1 above it was suggested that in non-assertive contexts deontic need to tends to be objective, by contrast with the typical subjectivity of deontic need. Of the following two examples, then, that with objective need to in (174) is more typical than that with subjective need to in (175). (174) If your contribution is £1.20 or more you do not need to fill in ST (V) (ICE-GB W2D-001 84) (175) You don’t need to bother (ICE-GB S1A-057 64) Affirmative deontic need to is not, however, resistant to subjectivity. What we find here is the same tendency for modal strength and subjectivity/objectivity to correlate with the person of the subject, as we have found with deontic must and have to. Strong subjectivity is most likely when need to occurs with a 2nd person subject. In contexts where there is an apparent authority structure the utterance will have the force of a directive, as in:
74
Chapter 3
(176) I think you need to focus now, on one aspect, whether it’s impacts whether it’s just temperature whether it’s rainfall (ICE-GB S1B-007 216) (177) And the rest of you need to gather around. One or two of you can lay on the ground. (ICE-AUS S2A-060 50) Slightly weaker are cases where the speaker is merely making a suggestion or giving advice, as in: (178) Maybe you need to try to do that through your GP locally (ICE-GB S1A062 117) (179) then you need to have your teeth extremely thoroughly cleaned, as soon as possible (ICE-GB S1A-087 197) With 1st person singular I as subject, need to far more commonly expresses dynamic than deontic modality. When deontic, it may be objective as in (180), where compliance with legal regulations is at issue, or subjective as in (181), where it serves as an indirect directive. (180) On 17 Nov I rang to ask progress and was informed by Mr Bromet that he was preparing his report for Head Office. I asked how long it would take to get a decision and he said it could take a year but, in any case, it would not help me as I needed to take my own legal action. (ICE-AUS W1B-023 83) (181) I need to know exactly where the car is going to be (ICE-GB S1B-080 71) Instances of need to in the self-hortatory use that is found with deontic must as in (14) above are rare. An example is (182): (182) And worse now it appears that territorial gains made by the Serbians in particular and to some extent I need to say the Croatians through a through aggression will actually now be legitimized (ICE-AUS S1B-051 103) Deontic need to can be used with 1st person plural we to convey a speakerderived directive as in (183) and as a rhetorical marker as in (184): (183) I think this is the first action that needs to be taken and we need to take it very soon (ICE-GB S2A-031 43) (184) And what we need to recognise is that in the in um covering some of the options there are degrees of formality (ICE-AUS S1B-011 7) With 3rd person subjects deontic need to is typically objective as in (185), where the deontic source is an institutional requirement:
Necessity and obligation
75
(185) Could you also inform me whether individual members receive the journal or whether they need to be journal subscribers as well. (ICE-GB W1B-028 148) 3.7.2 Epistemic need to Clear examples of epistemic need to, as in (186), are rare, and like epistemic need express objective logical necessity: (186) Of course the chaos when the Supreme Being was discovered tied up and concussed on the floor would be indescribable, but surely they would need to be more than just lucky to win much more time out of mere chaos? (ICE-GB W2F-015 69) 3.7.3 Dynamic need to Dynamic need to, but not need, may express a need that is intrinsic to the subjectreferent, an ‘internal compulsion’ (Nokkonen 2006: 62), as in (187) and (188), or one that is located in external circumstances, as in (189): (187) They need to sort of let their hair down (ICE-GB S1A-048 138) (188) So where we may need to have three or four meals a day the ah the crocodiles only need to eat once or twice a week or a month um or even go much longer than that without eating (ICE-AUS S2A-057 20) (189) occasionally you need to crown a tooth which is perfectly, you know perfectly sound from the aspect and there’s no reason for an x-ray but, they just insist on having one (ICE-GB S1A-088 12) 3.7.4 Need to and temporality Like must, have to, and have got to, deontic and dynamic need to can refer to a present requirement that is applied to the actualization of a future situation as in (176–179) above, a present situation as in (174), (185) and (187–189) above, or a past situation as in (190) below: (190) Non-contributory means that you don’t need to have paid National Insurance contributions to qualify. (ICE-GB W2D-005 3) When the modality is in the past there is a contrast between need to, whose preterite is formed inflectionally, as in (191), and need, which has no preterite but can express a past necessity in construction with perfect have, as in (192) below. The meanings here are not identical, differing in their actualization implicatures. In (192) we understand that the situation, of ‘worrying’, was
76
Chapter 3
actualized, but there is no such implicature in (191), as evidenced by the possibility of a continuation such as and so she didn’t. (191) I had taught myself strength as another man might have taught himself ballroom-dancing, had learned strength as a way of dealing with my weakness. But Lilian seemed to have been born with unbreakable will: it was not something she needed to learn. (ICE-AUS W2F-012 51) (192) However, she need not have worried. I was just a normal schoolboy with normal schoolboy habits and a propensity for mishap, as evidenced by the day I came a cropper in tar at Mentone. (ICE-AUS W2F-013 42) A further temporal possibility available to need to, but not need, is the expression of a future necessity, as in: (193) We may need to see him. We -- we w- will need to see him again. (C-US SBC 18 82-4) 3.7.5 Need to and negation Need to, like need, takes external negation. Thus, in (194) do not need to is paraphraseable as “it is not necessary for”: (194) You do not need to show your Registration Document when relicensing with a reminder form. (ICE-GB W2D-010 111) 3.7.6 Need to: regional and stylistic variation According to Leech (2003), Smith (2003), and Mair and Leech (2006) while need has declined sharply in both British and American writing, need to has enjoyoyed a spectacular increase (of 249.1% in British writing and 123.2% in American writing: see Table 1.4 above). According to Smith (2003), the rise of the quasi-modal has been more pronounced in assertive than non-assertive contexts (where it might be argued that it is entering into competition with must and have to). There are several possible reasons for the contrasting diachronic fortunes of the two items. One is the greater syntactic flexibility of need to, which is not like auxiliary need restricted to non-affirmative contexts and lacking in preterite and non-tensed forms. Another reason may be that the trend towards increasing informality in English, especially in public modes of discourse (Fairclough 1992) favours the quasi-modal. Whereas need is, as noted in Section 3.4.2, strongly disfavoured in speech over writing, need to is preferred in speech (with a speech/writing a ratio of 1310:786, or 1.66:1. Furthermore need to is more common – albeit marginally – in dialogue than in monologue in ICE-AUS and ICE-GB, by a ratio of 662:608, or 1.08:1 (see Appendix Table 7).
Necessity and obligation
77
The frequencies for the present study suggest that AmE is leading the way in the rise of need to, as it is in the decline of need. Need to is almost twice as popular in C-US (473 tokens per million words) as in ICE-GB (280), with ICEAUS in-between (343). Table 3.17. Genre distribution of need to Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total 3.8
ICE-AUS (107) (101) 347 (208) 320 (32) 343 (103) 338 (135) 343
ICE-GB (131) (45) 293 (176) 240 (24) 267 (80) 260 (104) 280
C-US
Avg
670 (78)
437
188 (15) 473 (93)
262 365
Had better
With its modest frequency (89 tokens in the present corpora, and its numbers reportedly declining (Mair and Leech 2006: 328) semi-modal had better is a minor item. As Mitchell (2003) points out, had better has in common with the modal idioms may/might as well that their etymological background involves the notion of comparison, and that as a result of grammaticalization and semantic bleaching this notion has been attenuated. The extent of grammaticalization of had better is evident in the proportion of instances where the auxiliary had is dropped (20.2%), as in the second two instances in (195): (195) ‘No, I’d better talk to her.’ Better face the bloody music, whatever it is, she thought, but when she picked up the phone, her hand was shaking. ‘Listen, bitch,’ hissed the thin, hoarse voice, ‘y’ better get round ’ere quick an’ bail out y’ precious boyfriend.’ (ICE-AUS W2F-004 127) 3.8.1 Meanings of had better In Section 2.3.3.1 above we noted, and rejected, Mitchell’s (2003: 145) claim that the quasi-modal had better has developed an epistemic sense. As a consequence had better is here regarded as essentially monosemous, a deontic expression (so the title of this subsection is admittedly slightly misleading). It is via semantic bleaching that had better earns its place in the modal system. It is typically used with mitigated directive force representing a type of subjective deontic modality, with a meaning best described as ‘advisability’ (q.v. Jacobsson 1980: 52). There was only one example of had better in the corpora where its literally comparative sense is still salient (as reinforced by the following comparative clause):
78
Chapter 3
(196) Actually today I’m nursing a very bad hangover so I decided I had better stay at home rather than throw up on the Metro! (ICE-GB W1B-009 33) A number of writers have suggested that the advice or warning that is conveyed by had better is often accompanied by an implication that the speaker has good grounds for the directive, insofar as a less desirable alternative is understood to exist. For example Palmer (1990: 82) suggests that had better implies an “or else” consequence as part of its basic meaning, claiming that the speaker “is fairly firm about his advice with the implication that unpleasant consequences may follow if it is not taken”. Similarly Perkins (1983: 64) suggests that non-fulfilment “entails adverse consequences” and Edmonson et al. (1977: 297) claim that the basic meaning of had better involves a threat for the subject. This implication is in fact quite rare (accounting for only 4.5%, of tokens in the corpora). (196) above is an example, and others are provided below. In (197) the speaker reports the advice that she has been given to start a family on the grounds that she is approaching the age when it may be difficult to do so. In (198) the undesirable alternative, that “no one’ll believe her unless she’s taken a photograph of the three sisters” is stated explicitly. (197) My main thing ’cos we want to start a family and I know I’m I’m no spring chicken any more and everyone keeps saying hey you’d better get on with it you know (ICE-AUS S1A-046 231) (198) Um well I thought we’d better go and do the touristy bit because no one’ll believe her unless she’s taken a photograph of the three sisters (ICE-AUS S1A-057 320) In the majority of cases, however, there is little justification for invoking the notion of an adverse consequence. In (199) had better merely conveys a deontic sense that is similar to that of should (which could be readily substituted for it). Notice that had better is harmonic, as is should, with perhaps. (199) B: Perhaps I’d better give her a ring A: Perhaps you’d better Yeah That’s probably a really good idea (ICEAUS S1A-075 120, 121) Like should, had better typically has medium strength, as in (199), where it would be possible for the advice to be politely declined. However it may be pragmatically strengthened, as in (200), in which a forceful imperative is reported, reinforced by the swear word bloody. Here the strength of had better is stronger than that of should, approaching that of performative must, which it resembles in not countenancing non-actualization (neither You must turn the tape around nor You’d better turn the tape around could felicitously be followed by but I don’t suppose you will, unlike You should turn the tape around). (200) I said better turn the bloody tape round (ICE-AUS S1A-020 194)
Necessity and obligation
79
Perkins (1983: 63) wrongly claims that had better is objective: “it is objective in that the deontic source is not (directly) identifiable as the speaker”. In fact, as Huddleston and Pullum (2002: 196) observe, it is “generally subjective, giving the speaker’s judgement as to the best course of action”. Quantitative support for this claim comes from the high proportion of instances (61.8%) with a 2nd person subject, which we have seen with other modals to correlate strongly with subjectivity, and a 1st person subject (34.8%). Only 3.4% of tokens have a 3rd person subject. Had better is consistently used in the data with reference to specific future events, as reinforced by the temporal adjunct right now in (201), and appears incapable of being used with the general present reference that is common with should.8 Notice that had better could not be felicitously substituted for should in (202): (201) In that case you’d better switch off right now (ICE-AUS S1B-030146) (202) You should always look carefully at your map afterwards so you can see what’s what’s around that area (ICE-AUS S1A-056 292) Finally, had better normally takes internal negation, whether the negative follows better, as in (203), or whether it is it precedes it (as in You hadn’t better let Jo get hold of this – there were no corpus examples of the latter), both more plausibly paraphraseable as “it’s advisable that you do not let Jo get hold of this” than “it’s not advisable that you let Jo get hold of this”. (203) You’d better not let Jo get hold of this (ICE-GB S1A-030 282) 3.8.2 Had better: regional and stylistic variation As Table 3.18 shows, the numbers for had better are healthiest in ICE-AUS, perhaps a reflection of its relatively greater vitality in speech (with a speech/writing ratio of 2.4:1, as against 1.1:1 in C-US and 1.0:1 in ICE-GB). The much higher frequency of this quasi-modal in dialogue as against monologue (4.3:1) and in non-printed as against printed texts (3.0:1) is predictable from the dominance of its subjective deontic meaning. Table 3.18. Genre distribution of had better Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 89 (32) 29 (7) 56 (39) 10 (1) 27 (8) 23 (9) 48
ICE-GB 53 (19) 4 (1) 33 (20) 100 (10) 10 (3) 33 (13) 33
C-US
Avg
43 (5)
47
38 (3) 41 (8)
31 40
80
Chapter 3
3.9
Be supposed to
Be supposed to is a medium strength lexico-modal with semantic affinities to should and ought to. Mair and Leech’s (2006: 328) figures show it to be on the rise, especially in British writing: see Table 1.4 above. It may well be that its increasing numbers are occurring at the expense of ought to, with which it is most similar. 3.9.1 Meanings of be supposed to Table 3.19. Meanings of be supposed to Deontic Epistemic Dynamic Indeterminate Total
ICE-AUS 26 18 2 1 47
ICE-GB 46 44 6 3 99
C-US 66 (13) 51 (10) 5 (1) 5 (1) 127 (25)
Total 138 (85) 113 (72) 13 (9) 9 (5) 273 (171)
% 50.5% 41.4% 4.8% 3.3% 100%
Whereas the typical pattern with modal expressions, certainly for the modal auxiliaries, is for epistemic senses to derive from historically prior deontic ones, with this lexico-modal it is the epistemic meaning that is prior. As Table 3.19 indicates, be supposed to expresses predominantly deontic and epistemic modality, and very occasionally dynamic, these meanings being illustrated respectively in (204) (“I’m obliged to”), (205) (“he’s thought/alleged to be”), and (206) (“what is determined by circumstances/nature”). (204) But you were saying I’m supposed to be encouraging you (ICE-GB S1A075 11) (205) That boy, he’s supposed to be awesome. (C-US SBC 02 27-29) (206) When they usually run, and, fish weren’t running this year, you know, it’s like everywhere. Nothing’s doing what it’s supposed to, anymore, anywhere. (C-US SBC 04 480-2) The senses are not sharply distinct, and pose challenges to analysis. In (207), for example, an epistemic reading (“it is supposedly/it is thought that it is”), and a deontic reading (“it is under an obligation to be”) are both possible, though arguably the former is the more salient. Similarly possible, in (208), are both the epistemic sense of something that is alleged or assumed to be the case and the deontic sense of something that is appropriate or required by custom.
Necessity and obligation
81
(207) Although the monarchy is supposed to be part and parcel of British life people of all sorts can envisage Britain without a monarchy (ICE-GB S2B-032 19) (208) Yeah and so strange the things that people (unclear word) about religion (unclear words) for something that’s supposed to make you behave well to the people around you treat them with respect and dignity and caring and kindness (ICE-GB S1A-084 191) While be supposed to is similar to both should and ought to in strength, its objectivity gives it a closer affinity with ought to than with the typically subjective should. Be supposed to is consistently objective, with the expectation or requirement arising from a source external to the speaker. Despite Edmonson et al.’s (1977: 289-90) claim that it is possible for deontic be supposed to to be subjective, there are no clear examples in the corpora (not surprisingly perhaps, since, as Westney 1995: 178 notes, these might be construed as “conversationally misleading”). Be supposed to also resembles ought to in its conversationally-derived implication of non-fulfilment, an implication which is clear in epistemic cases with a past predication as in (209) and deontic cases with the modality in the past as in (210): (209) He’s the guy who is supposed to have left (ICE-GB S1A-008 266) (210) The idea was you know we were supposed to do all these graphs and stuff (ICE-GB S1A-008 52) The temporal flexibility of be supposed to suggests that it may operate as a suppletive to medium strength modals like should and ought to. Both the modality and the proposition can be associated with a variety of temporal domains. With epistemic be supposed to both the modality and proposition can be present or past, but not future. In (205) above both the modality and the proposition are present, while in (209) the modality is present and the proposition past. In (211) the modality is past and the proposition present, and in (212) both are past: (211) Shem was the uhm (unclear word) Semitic and Ham was supposed to be the descendant of Africans and Japeth’s the European (ICE-GB S1A-053 241) (212) I mean, this destruction was supposed to have occurred in the eighteen eighties (ICE-AUS S2A-035 180) With deontic be supposed to the modality can be present or past, the proposition, present, past or future. For example in (204) above and (213) below both are present, while in (210) the modality is past and the proposition future (with respect to the modality). In (214) the modality is past and the proposition present. In (215), even though the perfect is associated syntactically with the
82
Chapter 3
complement of supposed, it applies semantically to the modality (“my results were required (PAST) to come out today (FUTURE)”). With deontic be supposed to the two syntactic options are available, but the perfect option is the only one available for deontic should and ought to (my results should/ought to have come out today). (213) I’m afraid I think the I think I think the uhm that chicken is supposed to be a bit juicier (ICE-GB S1A-022 76) (214) The photographs came to life as she talked. (This was called a voiceover.) They were supposed to have all the gloss and the false authority of a soft drink commercial. (ICE-AUS W2F-020 11) (215) And apart from that I mean my results are supposed to have come out today (ICE-GB S1A-093 237) Huddleston and Pullum (2002: 187) observe that should and ought to normally take internal negation, and claim that there are no “equivalent items taking external negation” (which they attribute to the lack of a clear pragmatic difference between external and internal negation with medium strength modality). Arguably, however, be supposed to fills the gap in the paradigm, Consider: (216) But I’m not supposed to be saying anything (ICE-GB S1A-017 337) (217) The thing is you’re not supposed to do it without your teacher (ICE-GB S1A-045 56) It is certainly possible to construe (216) and (217) as having internal negation like should and ought to (“expected/obliged not to”). However examples of this type are often felt to express more than medium strength modality. The basis for this is probably that they may alternatively be regarded as having a conventionally established interpretation involving negative raising and external negation (q.v. Westney 1995: 180). 3.9.2 Be supposed to: regional and stylistic variation Table 3.20. Genre distribution of be supposed to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 92 (33) 21 (5) 63 (38) 30 (3) 20 (6) 23 (9) 47
ICE-GB 200 (72) 42 (10) 137 (82) 50 (5) 40 (12) 43 (17) 99
C-US
Avg
146 (17)
115
100 (8) 127 (25)
55 91
Necessity and obligation
83
As Mair and Leech’s (2006) figures in Table 1.4 indicate, be supposed to has undergone a marked increase in recent British writing (113.6%), and a mild increase (6.3%) in recent American writing. Be supposed to is marginally more frequent in C-US than in ICE-GB (and quite unpopular in ICE-AUS), but at the same time relatively more robust in spoken BrE than spoken AmE (with a speech/writing ratio in ICE-GB of 137:43, or 3.18:1, as against 146:100, or 1.46:1, in C-US). Given the semantic similarities between be supposed to and ought to, it may be conjectured that the increasing numbers of the lexico-modal are occurring at the expense of the marginal modal. 3.10
Be to
Be to is a semantically strong semi-modal, like be bound to. In addition to its deontic, epistemic and deontic meanings it has special uses in the protasis of conditional constructions. It is on the grounds of its deontic meaning that be to is included in the present chapter, rather than in Chapter 4 on the basis of its canlike dynamic meaning, or Chapter 5 on the basis of its will-like temporal meaning. Table 3.21. Meanings/uses of be to Deontic Epistemic Dynamic Conditional Indeterminate Total
ICE-AUS 46 39 5 40 5 135
ICE-GB 26 86 18 79 12 221
C-US 31 (6) 15 (3) 5 (1) 20 (4) 5 (1) 76 (15)
Total 103 (78) 140 (128) 28 (24) 139 (123) 22 (18) 432 (371)
% 23.8% 32.4% 6.5% 32.2% 5.1% 100%
3.10.1 Meanings of be to Be to expresses strong deontic necessity. It can be strongly subjective, comparable to deontic must, as in (218), which reports an aggressive command: (218) and afterward he had kicked her out of his car, telling her she was ugly, and that he had never loved her, and that she was to remember his name, because he was a man who was going places. (SBC Frown P04 145) More commonly however it is objective, as in (219) and (220), where it is semantically close to objective have to (with which it alternates in (219)), and in (221), where the deontic source is an institutional regulation and be to is close to deontic shall.
84
Chapter 3
(219) Liz and John had to get dressed and Liz was to ring the house we were going to. John had to get the truck ready. I was to finish shutting my cases, then wake Stuart up. (ICE-AUS W2B-004 16) (220) We note that you are to discuss your overall condition with him upon your next attendance. (ICE-AUS W1B-023 65) (221) (6) The Council may appoint any other person who is neither a student nor a member of staff of the University to be a member of the Council and the person, on being appointed, is to be taken to be an appointed member of the Council in addition to the members appointed under subsection(4). (ICE-AUS W2D-005 186) With strong deontic be to negation is internal, as in (222) (“it is stipulated that they not be consulted”): (222) and the peoples of Europe are not to be formally consulted at any point, by referendum or otherwise. (ICE-GB W2E-001 29) When deontic be to takes a passive complement as in (223), or an active complement with a passive interpretation as in (224), its strength may be less, comparable to that of deontic should. (223) And I think the strength of the police authorities is something which is to be applauded (ICE-GB S1B-033 23) (224) Federal Environment Minister Ros Kelly says the public is not always to blame for the country’s water problems (ICE-AUS S2B-004 248) Even weaker in strength is the use of be to to express dynamic (theoretical) possibility with passive complements. In (225), (226) and (227) be to is similar in meaning and strength to dynamic can, with which it alternates in (227). (225) This is the dying society of “The Dead”. One of non-questioning being replaced by another intellectual one. The evidence is to be found at several points through the story. (ICE-AUS W1A-013 113) (226) If there is no abrupt change, but rather a gradual transformation, in uncial letter forms, it is perhaps in the use of uncial that a break or change is to be observed. (ICE-GB w2A-008 031) (227) Therefore an unconformity is to be found in the bottom of a rock unit and can be identified by the varying dip or erosive nature between the two beds. (ICEGB W1A-020 091) Whereas, as we have seen, deontic be to takes internal negation, dynamic be to takes external negation (is not to be found = “it is not possible for it to be found”: there were no corpus examples).
Necessity and obligation
85
Be to has some further uses which are more temporal than modal. It may refer to an event that is planned or scheduled, whether the plan/schedule is in the present as in (228), or in the past as in (229). (228) And fines are to be related for the first time to the offender’s income, so that the jails will be less crowded with those who will not pay because they cannot. (ICE-GB W2C-007 53) (229) It appears that the network manager, which was to be sited in the Box Office itself will now be in the general office along with the third terminal. (ICE-GB W1B-021 48) Here the clause containing be to does not indicate whether or not the situation was actualized (although non-actualization can be implicated contextually as in (229) above, or by means of the perfect aspect as in (230). (230) The cost of the project was to have been about £9,000m, with much of the money coming from the private sector. (ICE-GB W2E-008 046) Actualization is, however, entailed in the “future in the past” use (Palmer 1990: 165, Huddleston and Pullum 2002: 206) referring to an event that was known to occur subsequent to other events in the past. In this use, exemplified in (231) and (232), be to is equivalent to would. (231) By the time Samuel Stead came to Sydney, gold had paved the way for a boom in the Australian economy that was to last for thirty years. (ICEAUS W2B-003 20) (232) Sweetheart was then quite a local identity, although she was to be caught by the rangers a few months later and moved to a reserve (ICE-AUS W2F-010 51) Finally, be to has some special uses in conditionals. In the protasis of an open conditional, as in (233), it may carry a sense of purpose, while in the protasis of a remote conditional, as in (234), it simply reinforces the remoteness of the condition (compare if Done’s work broke the mould of its commercial reputation). (233) Here the Government has recognised that schoolteachers will have to educate children about the environment if future generations are to solve global problems (ICE-GB S2B-022 82 2) (234) Still, Capon does indicate that if Done’s work was to break the mould of its commercial reputation, he may be exhibited by leading galleries. (ICEAUS W2B-010 160) Opinions differ as to whether the futurity or deontic use of be to is the more basic. According to Quirk et al. (1985: 143) be to is to be regarded as
86
Chapter 3
“expressing futurity, with varied connotations of ‘compulsion’, ‘plan’, ‘destiny’, etc, according to context”. By contrast, according to Perkins (1983: 69) is to indicates that circumstances which involve “some conscious organization” are disposed towards the occurrence of an event. In this account the futurity use is an incidental consequence of the fact that determined actions necessarily take place in subsequent time. 3.10.2 Be to: regional and stylistic variation Table 3.22. Genre distribution of be to Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total
ICE-AUS 33 (12) 83 (20) 53 (32) 250 (25) 260 (78) 258 (103) 135
ICE-GB 94 (34) 271 (65) 165 (99) 180 (18) 347 (104) 305 (122) 221
C-US
Avg
52 (6)
90
113 (9) 76 (15)
225 144
According to Mair and Leech (2006: 328) be to is in decline in both written BrE and written AmE, and especially the latter, where it is both less frequent, and declining more rapidly. In the present study the difference between the two Englishes is even more marked, with almost three times as many tokens in ICEGB as in C-US (221:76, or 2.90:1). The relative unpopularity of this item in speech is most likely a factor in its declining fortunes (the ratio of tokens in speech/writing being 270:590, or 1:2.18). One reflection of the greater vitality of be to in BrE is the comparatively milder dispreference for this expression in speech in that dialect (the speech/writing ratio in ICE-GB being 165:305, or 1:1.84, as against 52:113, or 1:2.17, in C-US). The relative dispreference for be to in speech is even greater with Australian speakers, with tokens in speech outstripped by those in writing by a ratio of 53:258, or 1:4.86. 3.11
Be bound to
The lexico-modal be bound to is a very minor item in frequency terms, but included in the study on semantic grounds: it expresses strong deontic and epistemic necessity and is sometimes claimed to be a suppletive for must.
Necessity and obligation
87
Table 3.23. Meanings of be bound to Deontic Epistemic Indeterminate Total
ICE-AUS 2 7 0 9
ICE-GB 4 12 1 17
C-US 0 (0) 5 (1) 0 (0) 5 (1)
Total 6 (6) 24 (20) 1 (1) 31 (27)
% 19.4% 77.4% 3.2% 100%
3.11.1 Meanings of be bound to Like must, have to, have got to, need and need to, the lexico-modal be bound to expresses strong modality. It is used both deontically and epistemically, though the latter is considerably more common (see Table 3.23 above). Deontic be bound to, which conveys a strong objective obligation, is exemplified in (235) and (236): (235) I’m bound to say there are a whole series of things that one has to consider when one’s examining matters to do with the leadership of the Conservative party (ICE-GB S1B-043 008) (236) In my submission Your Worship you’re bound to act on uncontradicted evidence if that evidence is credible (ICE-AUS S2A-068 104) According to Palmer (1990: 55-56) epistemic be bound to operates as a kind of suppletive for must when future time reference is involved. It is true that substitution of must for ’s bound to in (237) would result in unacceptability. (237) For for the shoppers so that there you know there may not be as many spaces there but if he’s going at half past six he’s bound to get one (ICEAUS S1A-068 38) However the notion that be bound to is a suppletive item is problematical. As Coates (1983: 42-43) has noted, not only is be bound to rare, but also epistemic must can at times be used with reference to a future situation (as we have seen in Section 3.1.4 above). Furthermore, be bound to can be used with a present situation, as Palmer himself (1990: 55) observes, and as the habitual present example in (238) confirms, as well as a past situation as in (239). (238) You’re bound to make mistakes early on and the instructor is there to help you put them right (ICE-GB S2A-054 102) (239) She’s bound to have lost it (ICE-GB S1A-093 230) Epistemic be bound to is similar to must in expressing strong modality, but whereas epistemic must has a sense of conclusion (suggesting the likelihood that the speaker is drawing the most obvious conclusion), epistemic be bound to carries a sense of inevitability (as reflected in the fact that epistemic will could
88
Chapter 3
readily serve as a substitute in (238) and (239), whereas must would be impossible in (238) and would express conclusivity in (239). The sense of inevitability carried by epistemic be bound to presents problems for the view that it is a suppletive item: inevitability is not the same as conclusivity and epistemic be bound to tends to be more readily substitutible by epistemic will than it is by must. 3.11.2 Be bound to: regional and stylistic variation Table 3.24. Genre distribution of be bound to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 6 (2) 13 (3) 8 (5) 10 (1) 10 (3) 10 (4) 9
ICE-GB 19 (7) 13 (3) 17 (10) 10 (1) 20 (6) 18 (7) 17
C-US
Avg
0 (0)
8
5 (1) 5 (1)
11 10
Be bound to enjoys greater representation in BrE than in the other two dialects and greater popularity in more ‘conservative’ genres. It is preferred more in writing than in speech (by a ratio of 33:25, or 1.32:1), marginally more in monologue than dialogue (by a ratio of 26:25, or 1.04:1), and more in printed than non-printed writing (by a ratio of 30:20, or 1.5:1).
Notes 1
There were no tokens of oughtn’t to in the corpora.
2
Pleonastic it is not listed by Coates (1983).
3
Both Palmer (1990: 55) and Coates (1983: 42-43) point out that bound to is available for expressing posteriority with epistemic necessity: see further Section 3.11.
4
Coates’s (1983: 46) claim that negative epistemic must does not exist, and that the missing form in the must paradigm is supplied by can’t, is thus wrong. The absence of tokens in ICE-AUS and C-US would appear to be accidental. Collins (1991: 156) presents the following example from his 225,000-word corpus of AusE: He mustn’t have wanted the coupons because he came up and give them to me.
Necessity and obligation
89
5
It was not possible to determine if the same situation obtains in C-US, because the Santa Barbara texts are not subclassified on the dialogic/monologic dimension.
6
Westney (1995: 170) makes the interesting point that as you should know is natural as an opener when a speaker is giving information or advice, but not as you ought to know, because the latter impolitely suggests, more strongly than should, that the addressee doesn’t know something that it would be appropriate for him/her to know.
7
These figures provide some support for Palmer’s (1990: 56) suggestion that You must be kidding is more likely in BrE than You’ve got to be kidding.
8
Westney (1995: 183) notes that past reference is possible, as in You’d better have left by the time I get there (there were no examples in the present corpora). However, note that even here the advised departure is (to be) in the future with respect to the time of utterance.
Chapter 4 Possibility, permission and ability The two modals of possibility, can and may, share a high level of semantic overlap, so it is not surprising that there has been a good deal of attention paid to the relationship between them in the literature (e.g. Lebrun 1965, Duffley et al. 1981, Dirven 1981, Collins 1988, Bolinger 1989, Klinge 1993, Coates 1995, Groefsema 1995). Perhaps more than any other of the modal expressions examined in this book, can and may raise challenging questions as to the preferability of a polysemy or monosemy position. In analyzing their meanings, as is the practice elsewhere in this book, we adopt a polysemy approach, assuming that each modal has a number of independent meanings (a position which, as noted in Section 2.4.2 above, is supported by the availability of instances which, in abstraction from context, are ambiguous between an epistemic and non-epistemic meaning; e.g. She may go may mean “She is permitted to go” or “It is possible that she will go”). At the same time it must be conceded that polysemy is less clearly in evidence with can, whose permission, (root) possibility and ability meanings are not always readily distinguishable, than it is with may. Despite this difference it is not necessary to infer that can is monosemous, as do Leech and Coates (1980) who gather the various subsenses into a single gradient unified by the notion of ‘inherency’, with one extreme (weakest inherency) implying “the circumstances which enable p to happen are independent of the participants in p”, and the other (strongest inherency) “the circumstances which enable p to happen are inherent to the performer of p” (1980: 83). This chapter examines can and may, along with their preterite counterparts could and might (including the negative forms can’t, cannot, mayn’t, couldn’t, and mightn’t). Also discussed is the lexico-modal be able to (able to and unable to preceded by all the inflectional forms of be).1 As Table 4.1 below shows, the frequency of can in the corpora is considerably greater than that of may (7763:2261, or 3.4:1), and that of could outstrips might (3357:1499, or 2.2:1). There are, furthermore, some striking regional differences, may being more popular in ICE-GB than in the other two corpora (see further Section 4.1.4), and might less popular in C-US than in the other two corpora (see further Section 4.3.4).
Chapter 4
92
Table 4.1. Frequencies of the modal expressions of possibility and permission Modals
Quasimodals Total
4.1
may can might could Total be able to
ICE-AUS 881 3,378 695 1,590 6,544 387
ICE-GB 1,218 3,565 702 1,622 7,107 434
C-US 825 (162) 3,665 (720) 520 (102) 1,757 (345) 6,765 (1,329) 346 (68)
Total 2,924 (2,261) 10,608 (7,663) 1,917 (1,499) 4,969 (3,557) 20,417 (14,980) 1,167 (889)
6,931
7,541
7,111 (1,397) 21,583 (15,869)
May
As Table 4.2 shows, the epistemic possibility meaning of may is dominant (accounting for 79.0% of tokens), with deontic and dynamic possibility being at best minor meanings. As we shall see in Section 4.2, while can expresses similar meanings to may, the proportions are quite different, with can being dominantly dynamic. Table 4.2. Meanings of may Epistemic Deontic Dynamic Indeterminate Total
ICE-AUS 651 78 101 51 881
ICE-GB 1,023 70 60 65 1,218
C-US 636 (125) 56 (11) 76 (15) 56 (11) 825 (162)
Total 2,310 (1,799) 204 (159) 237 (176) 172 (127) 2,924 (2,261)
% 79.0% 7.0% 8.1% 5.9% 100%
4.1.1 Epistemic may Epistemic may can be either subjective or objective, though the latter possibility is quite rare, and rarely mentioned in the literature. Coates (1983: 134) is an exception, but she nevertheless claims not to have found any instances in her data. The examples in (1) and (2) illustrate the subjective use, expressing the speaker’s lack of knowledge as to whether or not the proposition is true, and assessment of it as merely a possibility. In (1) the speaker’s lack of confidence in the truth of the proposition is reflected in the harmonic combination with I s’pose, and in (2) with I dunno. (1)
Ah we’re not talking about mutualism so much now except that of course they don’t have to eat as much algae as they would normally do because they have maintained the chloroplasts and I s’pose that may be some
Possibility, permission and ability
(2)
93
marginal significance in a mutualistic sort of way to the algae that they are feeding on (ICE-AUS S2A-025 111) I dunno. They may have (ICE-AUS S1A-073 102)
Verstraete (2001) claims that epistemic modality cannot be objective, invoking as an argument the resistance of epistemic modals to interrogatives (a resistance which presumably derives from the orientation of interrogatives towards the interlocutor rather than the speaker) and to if-clauses (the resistance here deriving from the suspension of the speaker’s commitment to the proposition). However instances of objective may do occur, where the estimation is one that is entertained more generally. In (3) the impersonal extraposition with It’s thought as matrix clause, and in (4) the existential construction there are suggestions …, indicate that the judgement is not limited to the speaker but rather on public record, as it were. (3) (4)
It’s thought the man may have committed suicide (ICE-GB S2B-016 134) The nature of the mutation in sugary maize has not been characterized biochemically, but there are suggestions that it may involve a debranching enzyme. (ICE-AUS W2A-038 164)
Epistemic may has a concessive use, which accounts for 4.9% of all epistemic may tokens and which, according to Coates (1983: 136), serves to ‘soften’ the speaker’s assertion. However, it is more accurately interpreted as involving a type of pragmatic strengthening (see further Section 2.5.2 above) in which the speaker concedes the truth of the proposition, rather than expressing a lack of confidence in it. Thus in (5) below the clause containing concessive may is equivalent to the unmodalized clause “although the timing is uncertain”, and similarly in (6) to “although we believe the rulers of this period and their advisers were misguided and deluded in …”). (5) (6)
The timing may be uncertain but the outcome is absolutely certain (ICEGB S2B-005 116) However misguided and deluded we may believe the rulers of this period and their advisers to have been in seeking to resolve conflicts by force of arms, the relative weakness of diplomatic alternatives must always be borne in mind. (ICE-GB W2A-010 17)
The time of the situation with epistemic may can be past, as in (7), present as in (8) (including ‘general present’: see below), or future as in (9). (7)
Instead, it seems that relying on the free market in health may have brought nemesis on those who deny the right of all to health and trust to the impersonal forces of the market to protect them. (ICE-GB W2A-019 70)
94 (8) (9)
Chapter 4 We can’t eat beef that may be contaminated with BSE and cheese that may be full of lysteria. (ICE-GB W2B-014 26) One person who thinks it is not, and who suspects we may be in for a flip of the current system (and therefore the climate) into a new pattern is Wallace Broecker. (ICE-GB W2B-025 45)
The use of may with a general present time situation as in (10) and (11) below raises challenging issues of classification. (10)
(11)
In severe cases, the condition may seriously interfere with the child’s schooling and special arrangements may have to be made. (ICE-GB W2B023 24) The aura only lasts a few moments before the convulsion begins but may be long enough to enable the patient to get himself into a safe and comfortable position so that he will not hurt himself by falling to the floor. (ICE-GB W2B-023 40)
The meaning of may here is somewhat ambivalent between epistemic modality (paraphraseable by “it is possible that”) and dynamic modality (paraphraseable by “it is possible for”), and this has prompted some writers (e.g. Coates 1983, Vihla 2000) to characterize cases such as those in (10) and (11) as involving neutralization or ‘merger’. They are nevertheless here analyzed as epistemic on the grounds that the main focus of attention is on the speaker’s uncertainty as to whether or not, at any given moment, a situation whose potential for occurrence is not in doubt will be actualized (and may be paraphrased by “it is possible that … will”). Further support for this position is to be found in the incidence of harmonic combinations with epistemic expressions, such as the occurrence of may in an extraposed that-clause with it is possible as matrix clause as in (12), and with the epistemic adjunct possibly as in (13): (12)
(13)
Their primary role is likely to be in the removal of nutrients (nitrogen and phosphorus), although in the absence of light it is possible that they may behave heterotrophically and play a small part in BOD removal. (ICE-GB W2A-021 22) they may possibly increase the capacity of an individual organism to track or avoid change. (ICE-GB W1A-009 56)
Finally, the compatibility between may and intensifying adverb well, exemplified in (14), contrasts with the resistance of can to it: (14)
The physical mechanisms governing intra-plate earthquakes are not wellunderstood although they may well be associated with previous faulting activity (Johnston and Kantor, 1990). (ICE-AUS W2A-033 25)
Possibility, permission and ability
95
As Table 4.3 indicates, the frequencies for may used with future, present and general present situations were quite similar, with past situations less common. C-US is slightly out of step with the other two corpora, with relatively more present tokens and relatively fewer past and general present. Table 4.3. Time reference with epistemic may Future ICE-GB 291 (28.4%) ICE-AUS 187 (28.7%) C-US 35 (28.0%) Total 513 (28.0%)
Present 267 (26.1%) 174 (26.7%) 60 (48.0%) 501 (27.8%)
Past 120 (11.7%) 66 (10.1%) 5 (4.0%) 191 (10.6%)
General pres 345 (33.7%) 224 (34.4%) 25 (20.0%) 594 (33.0%)
Total 1023 (100%) 651 (100%) 125 (100%) 1,799 (100%)
4.1.2 Deontic may Leech (2003: 232-234) notes that while epistemic may has increased in frequency in recent decades, there has been a marked decline in the frequency of deontic may (mainly in speech). In the present data it is the least common of the meanings of may (see Table 4.2 above). Deontic may is prototypically, but not more frequently, subjective, with the speaker as the deontic source as in (15), or – more commonly – the addressee in questions as in (16) and in conditionals as in (17). The rarity of subjective examples is reflected in the incidence of somewhat formal, formulaic, tokens of the type in (16) and (17). (15) (16) (17)
You may use my desk. Well wait a minute, it’s a royal mess, isn’t it. (C-US SBC-019 138-140) Oh who’s he meeting there may I ask (ICE-AUS S1A-058 175) Let me develop the point if I may Jonathan (ICE-GB S1B-043 119)
In the corpora, examples of this type are surpassed in frequency by those in which the deontic possibility is objective, found predominantly in the statement of rules and regulations in bureaucratic and administrative writing, as in: (18)
Your local Training Officer can assist you in understanding and meeting your obligations. Any problems that arise may be referred to the Vocational Training Board. (ICE-AUS W2D-002 258)
4.1.3 Dynamic may Epistemic possibility is a homogeneous semantic category, concerned with verification, with the question of whether a situation has been, is, or will be, actualized). By contrast, dynamic possibility is a rather heterogeneous category, as reflected in the plethora of terms that have been proposed in attempts to
96
Chapter 4
generalize across its range of subsenses. None of these terms seems capable of capturing the full range of dynamic senses. For instance the notion of ‘potentiality’, as invoked by Klinge (1993), does not encompass dynamic can as used with verbs of perception which, we argue in Section 4.2.3.2 below, often denotes actualization rather than merely potentiality. The further notions of ‘inherency’ (Leech and Coates 1980) and ‘intrinsicness’ (Bolinger 1989) cannot be convincingly applied to dynamic can in its ‘circumstantially possible’ use (see Section 4.2.3.1), since the orientation of can here is to external rather than internal factors. Dynamic possibility is, we have seen, a minor meaning for may but a major one for can. Accordingly the discussion in the present section will be quite brief, with a more detailed account reserved for dynamic can in Section 4.2.3. May is used to express two types of dynamic possibility. The first, ‘theoretical possibility’ (a term suggested by Leech 1987), involves a potentiality for action that resides in the external situation. In this use may is often associated with a greater degree of formality than can. Examples follow: (19)
(20)
it’s not necessary uh for uh me to dwell at length upon, the Civil Evidence Act uh or uh the uh Rules i in regard to, uh what evidence may be uh, adduced (ICE-GB S2A-063 17) The role of the convenor is to direct the conduct and activities of the Selection Committee so it may find the most efficient applicant for the vacant position. (ICE-AUS W2D-001 67)
The second use, ‘dynamic implication’ (a term suggested by Palmer 1990), involves a potentiality for action that is the basis for an implied directive speech act. This category differs from theoretical possibility to the extent that it requires an expansion of the semantic framework into pragmatics. The literal meaning of may in the following example is dynamic, but a more satisfactory interpretation requires reference to its directive illocutionary force. (21)
And you may remember that the organisations the republics that were in the Soviet Union competed in the recent Winter Olympics under the title Commonwealth of Independent States (ICE-AUS S2A-027 40)
4.1.4 May: regional and stylistic variation Figures supplied by Mair and Leech (2006) indicate that may has been undergoing a decline in frequency in recent decades, one more marked in AmE (32.4%) than BrE (17.4%): see Section 1.4. This regional difference is reflected, in the present study (see Table 4.4), in the striking difference in overall frequencies for may between ICE-GB (1218 tokens) on the one hand and C-US (825 tokens per million words). ICE-AUS, with 881 tokens, patterns more closely with C-US than ICE-GB.
Possibility, permission and ability
97
The comparative unpopularity of may in speech over writing (by an overall ratio of 3.3:1 across the three corpora) reflects its declining fortunes in Modern English. The findings presented in Table 4.4 suggest that this decline is most advanced in C-US, where the speech vs writing gap is the greatest (3.7:1, by comparison with 3.2:1 for ICE-AUS and 3.1:1 for ICE-GB). A comparison of the frequencies for subvarieties in ICE-AUS and ICE-GB confirms the greater conservatism of the latter for the written component. The printed vs non-printed ratio for ICE-GB, based on normalized frequencies, is 2257:1420, or 1.6:1, compared with 1427:1690, or 0.8:1 for ICE-AUS (see Appendix Table 12). The present study provides some support for Coates’s (1995) claim that “may is the chief exponent of epistemic possibility in British English, but is less common in American English, where may has connotations of formality” (p.64). Not only are there more tokens of epistemic may per million words in ICE-GB (1023) than in C-US (636), but the notion that epistemic may has connotations of formality in AmE is supported by the finding that the ratio of written to spoken tokens in C-US is weighted more towards the (generally more formal) written categories (4.2:1) than is the case for ICE-GB (2.2:1), and also for ICE-AUS (2.7:1): see Appendix Table 12. Table 4.4. Genre distribution of may Spoken
Written
Total 4.2
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 350 (126) 658 (158) 473 (284) 1,690 (169) 1,427 (428) 1,493 (597) 881
ICE-GB 544 (196) 846 (203) 665 (399) 1,420 (142) 2,257 (677) 2,048 (819) 1,218
C-US
Avg
395 (46)
511
1,450 (116) 825 (162)
1,664 975
Can
Can contrasts with may in the dominance of the dynamic possibility meaning, across all three dialects (see Table 4.5). The popularity of deontic can is not even across the corpora. According to Coates (1995: 64) “can is less commonly used to express permission in American English; may is the normal exponent of permission” (p.64). However the findings of the present study suggest otherwise: not only are there relatively more tokens of deontic can per million words in CUS than in the other two corpora, but as Table 4.2 above indicates, deontic may is relatively dispreferred in C-US.
98
Chapter 4
Table 4.5. Meanings of can Dynamic Deontic Epistemic Indeterminate Total
ICE-AUS 2,729 353 24 272 3,378
ICE-GB 2,953 279 32 301 3,565
C-US 2,906 (571) 417 (82) 56 (11) 285 (56) 3,665 (720)
Total 8,588 (6,253) 1,049 (714) 112 (67) 858 (629) 10,608 (7663)
% 81.0% 9.9% 1.1% 8.1% 100%
4.2.1 Epistemic can The status of can as a modal that can express epistemic possibility is a matter of some controversy. According to some (e.g. Collins 1988, Huddleston and Pullum 2002: 180) can may serve as a marker of epistemic possibility, albeit restricted to non-affirmative contexts, as in (22) and (23): (22)
(23)
There are plenty of funds available to educate women about osteoporosis thanks to an exhaustive public health campaign there can’t be a woman alive in Australia unaware that calcium rich foods help prevent the disease of bone deterioration. (ICE-AUS W2B-012 100) How can it be an intrusion of privacy if the newspapers are being presented with the stories from the main participants on a silver plate (ICE-AUS S2B-004 159)
For others (e.g. Coates 1983) can is merely a suppletive form for must which supplies the missing non-affirmative form in the epistemic must paradigm. Interestingly, in a subsequent publication Coates (1995) suggests that can is beginning to develop a ‘genuinely’ affirmative epistemic use in AmE, a development which she correctly observes is in conformity with the “historical pattern of epistemic meanings developing from root meanings” (Coates 1995: 62). Coates quotes an example she heard at a symposium at the University of New Mexico in 1992: we hope this coding system can be useful (to other linguists working in the field). An example such as the following from ICE-GB shows that the emergence of assertive epistemic can is not limited to AmE. (24)
B: If if girls have uhm I’m going to show my ignorance in front of the tape here. If girls have black eyebrows but really blond hair is there something odd there does that A: I think some guys do as well B: Oh it it’s not it doesn’t mean they’ve dyed it then does it A: I think they can have but I’m not sure. I think they’re just darker, the hair’s just dark. It may mean their hair is dyed actually (ICE-GB S1A-041 46)
Possibility, permission and ability
99
Like epistemic may, epistemic can is typically subjective, as in (25) and (26), where the inferencing on which the speaker’s assessment of possibilities is based is spelt out respectively in the because-clause and in the preceding coordinated clause. (25)
(26)
No it can’t be hundred percent wrong ’cause the program um if you don’t have your exact time of birth you set it to oblique orbit of zero for the time and PM for the hour (ICE-AUS S1A-064 259) Oh I’ve only been back a few minutes anyway but the school hasn’t phoned so he can’t be too bad (ICE-AUS S1A-096 95)
Like epistemic may, epistemic can can be used with a past situation as in (27), present as in (28), or future as in (29). With future time situations, epistemic may is common while can is rare, though Huddleston and Pullum’s (2002: 182) claim that the latter is “barely possible” is an overstatement, as shown by the naturalness of (29). (27) (28) (29)
No they can’t have done it (ICE-GB S1B-071 34) It’s just uh it’s just it’s so convoluted and all the rest of it that it can’t possibly be right (ICE-GB S1A-005 32 Well I waited for the weather to break and I kept looking out thinking oh its it’s going to you know this can’t last all day (ICE-AUS S1A-100 7)
As we have noted in Section 4.1.1, the use of may with a general present time situation (as in (10) and (11) above) presents some classificatory difficulties. Apparently similar examples with can are found, as in (30): (30)
But it is a reminder that pain can be the one reality in one’s life. It can be the only thing that anyone even the Son of Man can think about. (ICE-GB S2B-028 75)
However, there is a semantic distinction, admittedly subtle, between may and can in these cases, one determined by the boundary which separates epistemic and dynamic modality. May falls on the epistemic side of the boundary, expressing the speaker’s uncertainty as to the serial actualization of a situation over a period of time (“it is possible that … will”). On the other hand can, which falls on the dynamic side, focuses merely upon the potential for occurrence of a situation (“it is possible for … to”). This distinction has been discussed in the literature, with a variety of labels applied to relevant senses of may and can: ‘factual possibility’ and ‘theoretical possibility’ (Leech 1987); ‘indeterminacy’ and ‘contingency’ (Van der Auwera 1986); ‘extrinsic possibility’ and ‘intrinsic possibility’ (Bolinger 1989); and ‘possibility’ and ‘potentiality’ (Klinge 1993). Some evidence for a ‘different meanings’ position was presented in Section 4.1.1 above. A merger interpretation might be argued to draw support from examples such as (31) and (32), where the alternation between may and can
100
Chapter 4
is suggestive of semantic closeness between may and can. However it is not necessary to assume that two meanings have merged into one in such cases. It is more plausible to maintain that the meanings of may and can are in fact distinguishable – as epistemic and dynamic respectively – and that there are certain contexts which are compatible with both meanings. Casting further doubt on the validity of a merger interpretation is the strong tendency for can to precede may in alternations of the type in (31) and (32): if one were to assume that may and can expressed the same meaning in such alternations, then it might be expected that their relative ordering would be random. (31)
(32)
There can be a similar problem with satellite dishes, where several “footprints” may overlap a particular geographical area. (ICE-GB W2D014 29) It can lead to a disproportionate preoccupation for example with esoteric tax issues which may affect only a handful of people invariably the already-advantaged as compared with the unattended deserts of popular ills particularly in fields of poverty law and tribunals (ICE-GB S2A-039 90)
4.2.2 Deontic can It has been argued that deontic may and can are semantically distinct (generally by writers whose allegiance is to a monosemy approach, and who therefore assume – surely counterintuitively – that the same meaning can never be expressed by different modals). For Vanparys (1987: 232) the difference is that may is subjective and can objective: “May is used to perform acts of granting permission, while can is used to state that someone has permission”. Groefsema (1995: 68) is of the view that “when I ask you May I smoke in here, I make my smoking solely dependent on your permission, whereas when I ask Can I smoke in here, I communicate that your permission is only one factor under consideration”. Duffley et al. (1981) characterize may as expressing the ‘virtual’ giving of permission, involving some external permitter, whereas with can it is intrinsically possessed by the permittee. Bolinger (1989: 7) differentiates can and may in terms of ‘intrinsic’ and ‘extrinsic’ possibility. In his opinion “Can refers to what a person, thing, or situation is endowed with, whether naturally in place (physical, mental) or implanted (authorized, permitted); it invokes what is immanent, inherent. May refers to the external, to what transcends the entity or situation.” It is certainly true that deontic can is more often objective than subjective, as in (33): (33)
As a race official, John cannot place any bets on the Cup. But the rest of Australia wagered an amazing $71.4 million with the TAB last Cup day. (ICE-AUS W2B-002 44)
Possibility, permission and ability
101
Nevertheless there is no shortage of examples in the present data in which deontic can and may are semantically parallel, even though they differ in their frequency of occurrence and often in formality: (34)–(36) below match (15)–(17) above. (34) (35) (36)
You can come back for a second helping (ICE-AUS S1A-004 257) Uh can I pick the boxes up please (ICE-GB S1A-068 184) If I can partly misquote John and say yes indeed it adds to the gaiety of the country (ICE-GB S1B-024 26)
4.2.3 Dynamic can
In Section 4.1.3 above we identified two uses of dynamic may, ‘theoretical possibility’ and ‘dynamic implication’. Both are found with can, as is a third use that is not available for may, ‘ability’. Ability involves a potentiality for action that lies within the capacities of the subject-referent. Table 4.6 below presents figures for these dynamic possibility uses, not only for can but also may. The theoretical possibility use is the most common of the three dynamic uses, accounting for 77.2% of all dynamic tokens of may, and 65.8% of can. The ability use, which is not found with may, is expressed by just over one quarter of dynamic possibility (28.7%) cans, and is less popular in ICE-AUS than it is in the British and American corpora. The dynamic implicature use is a minor one for can (accounting for 5.6% of its dynamic uses), and relatively more common in ICE-GB, but quite common with may (22.8%), though comparatively rare in ICEGB. Table 4.6. Dynamic possibility with can and may
Theoretical possibility Ability Dynamic implication Total
can may can can may can may
ICE-AUS 1,869 75 703 157 26 2,729 101
ICE-GB 1,895 52 862 196 8 2,953 60
C-US 1,883 (370) 56 (11) 896 (176) 127 (25) 20 (4) 2,906 (571) 76 (15)
Total 5,647 (4,134) 183 (138) 2,461 (1,741) 480 (378) 54 (38) 8,588 (6,253) 237 (176)
4.2.3.1 Theoretical possibility In the clearest cases of theoretical possibility there is explicit mention of the enabling circumstances, as in (37) and (38):
102 (37)
(38)
Chapter 4 Eucalypts that grow on nutrient-poor soils, however, cannot do this because the lack of nutrients limits their growth. (ICE-AUS W2B-024 190) It’s great to have my own bolt-hole where I can do as I like without having to fit around anyone else (ICE-GB W1B-013 54)
There are two recognizable subtypes of theoretical possibility, ‘existential’ and ‘rational’. Existential modality (to use the term suggested by Palmer 1990 and adopted by, amongst others, Facchinetti 2002, and Huddleston and Pullum 2002) lies on the borderline between modality and aspectuality, and is found with can but not may. Its characteristic feature is an implicit existential quantifier, which involves either set membership as in (39) (“Some of these movements are quite large”) or characteristic behaviour as in (40) “She sometimes acts in a pushy way”). (39)
(40)
These movements can be quite large. Sorensen et al., (10) indicate that movements of 50 mm are not uncommon, and in extreme cases, movements as high as 100 mm have been recorded. (ICE-AUS W2A-034 40) She can be pushy (ICE-AUS S1A-045 181)
Harmonic combinations – particularly with the adjunct sometimes – are common, as in (41) and (42): (41)
(42)
Construction can now begin with the usual format of resistors first, although it can sometimes be an advantage to fit the IC sockets at an early stage to assist in the less obvious location of other components. (ICE-GB W2B-032 73) UUmm and you’ve got it and they don’t get it back till the Sunday and they can really make life tough and that can sometimes tempt children to steal it back (ICE-AUS S2A-039 184)
Existential possibility can is commonly encountered in scientific discourse, as in (43) and (44): (43) (44)
The interaction of an incident ultrasonic beam with a defect can cause a number of different responses. (ICE-GB W2A-036 20) Being labelled “maternally” or “paternally derived” can affect the degree to which the gene is expressed. (ICE-AUS W1A-005 23)
The second subtype of theoretical possibility, rational modality (to again use Palmer’s 1990 term), is similar in some respects to objective deontic possibility, the difference being that with the latter there is an identifiable personal or institutional deontic source, whereas with rational modality actualization is licensed by more abstract factors pertaining to what is generally
Possibility, permission and ability
103
considered to be reasonable, legitimate, societally/culturally acceptable, and the like, as in (45) and (46): (45)
(46)
this will bring us to some questions about what can reasonably be expected from organised government, operating within a framework of nations, in which nationhood, is regarded as incomplete, unless it entails sovereignty (ICE-GB S2B-048 13) But who can deny it’s brought entertainment and enjoyment to countless hundreds of people (ICE-AUS S1B-036 227)
4.2.3.2 Ability The ability meaning of can is commonly treated as distinct from its other meanings. The justification for the present analysis of ability as a subcategory of dynamic possibility is its association with the notion of potentiality, albeit one which lies in the subject-referent. Ability can normally requires an animate subject with agentive function, as in (47) and (48), but inanimate subjects can also be found, as in (49) and (50): (47) (48) (49)
(50)
and uh about the age of four most children can actually speak to virtually the same standard uh that a adult can speak (ICE-GB S1B-003 54) And he can play chess and I can’t (ICE-AUS S1A-076 147) The features for feeding paper that are built into a printer depend on the use for which it was designed. The two types of paper these printers can handle are cut sheet and/or continuous forms. (ICE-AUS W2B-038 31) Modern launch vehicles have more efficient engines and can launch a heavier payload: typically as much as two per cent of their launch weight. (ICE-GB W2B-035 71)
The ability meaning may be close to actualization, though it may not be realized. Compare the following two examples. In (51) we understand that the record-breaking in question is yet to be achieved, while in (52) the evidence for the subject-referent’s ability is past actualization (there is little difference between (52) and its unmodalized counterpart, You speak French). (51) (52)
Can Haley Lewis break the Commonwealth record I don’t think she can (ICE-AUS S2B-016(C):49) You you can speak French. Oh well you’ve studied French and I’m studying German and you’re studying Japanese. (ICE-AUS S1A-037 217)
With verbs denoting perception and cognition there is an especially low degree of modality. Thus in (53)–(56) it would make little difference if the modal were absent (I see the Prime Minister; I hear the crowds …; I remember vividly …; I don’t understand …).
104 (53) (54)
(55) (56)
Chapter 4 And now I can see the Prime Minister, John Major (ICE-GB S2A-019 30) Outside I can hear the crowds cheering as Her Majesty the Queen arrives with the Duke of Edinburgh to be greeted by the Minister of Glasgow Cathedral Dr William Morris one of her chaplains (ICE-GB S2A-020 118) I can remember vividly my first day at primary school (ICE-GB S1A-076 103) No I can’t understand any of this at all (ICE-GB W1B-003 170)
4.2.3.3 Dynamic implication As noted in Section 4.1.3 above, with the category of dynamic implication the modal is used in the formulation of an indirect speech act. The utterances in the following examples are more than merely statements of dynamic possibility: the illocutionary force of (57) is offer, request in (58), suggestion in (59), and instruction in (60). (57) (58) (59) (60)
Well I I can write on your behalf (ICE-AUS S1B-073 272) Can you pass the s the sour cream please (ICE-GB S1A-071 141) Perhaps we can talk about that on another occasion (ICE-GB S1B-055 65) Well, I’ll pour the ladies’, you can pour the men’s (ICE-AUS S1A-004 116)
As noted by Facchinetti (2002: 237), dynamic implication is common in instructional and didactic registers, as in (61) and (62): (61)
(62)
So in fact the far side of this temple here, and there you can just make out some, uh, stony parts which are in fact the remains of outdoor altars because you offer sacrifices sacrifices of huge animals like oxen which have just been, just been killed (ICE-GB S2A-024 94) OK this one has obviously got its body wall cut up so I’m not gonna do another one now this afternoon but you can look at this one (ICE-AUS S2A-052 10)
4.2.4 Can: regional and stylistic variation Mair and Leech’s (2006) diachronic figures, as presented in Table 1.4 above, indicate that, by contrast with may, can has been relatively stable in recent decades, enjoying a small increase in British writing (2.2%) and suffering a small decline in American writing (1.5%). However, as the figures in Table 4.7 indicate, the three corpora vary in the relative popularity of can in their spoken and written components. The figures suggest greater vitality for this modal in AmE (the ratio of tokens in speech to writing in C-US being 1.9:1) than in AusE (ICE-AUS 1.5:1) and BrE (ICE-GB 1.2:1). Interestingly, closer inspection of the figures for subvarieties of speech and writing in ICE-GB and ICE-AUS (not
Possibility, permission and ability
105
possible for C-US) suggests greater vitality for can in BrE than AusE. The extent to which can’s popularity in dialogue exceeds that in monologue is greater in ICE-GB (4275:3117, or 1.4:1) than in ICE-AUS (3908:3829, or 1.02:1). A similar conclusion is suggested by a comparison of the ratio for can in nonprinted as against printed texts (ICE-GB 3780:3000, or 1.3:1; ICE-AUS 2570:2650, or 0.96:1). Table 4.7. Genre distribution of can Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total 4.3
ICE-AUS 3,908 (1407) 3,829 (919) 3,877 (2,326) 2,570 (257) 2,650 (795) 2,630 (1,052) 3,378
ICE-GB 4,275 (1,539) 3,117 (748) 3,817 (2,287) 3,780 (378) 3,000 (900) 3,195 (1,278) 3,565
C-US
Avg
4,525 (527)
4,073
2,413 (193) 3,665 (720)
2,746 3,536
Might and could
It has been our practice thus far to devote each section to a single modal expression. However, because as preterite modal forms could and might share certain uses which normally differentiate them from can and may respectively, it is preferable to treat them together. 4.3.1 Uses of might and could as preterite forms Two broad uses may be distinuished, both of which serve to distance the situation from the reality of the present moment: temporal and hypothetical. 4.3.1.1 Temporal uses Temporal could and might may both serve either to identify a past time situation, or as a backshifted preterite. Could occurs readily in the past time use, as in (63). Might, on the other hand, has been claimed to be obsolete in this use. For example, according to Joos (1964: 187) might “never has real-past meaning in modern English; its remoteness is that of unreality.” This claim is wrong: might is not obsolete, though certainly rare, and largely restricted to formal and literary genres, as in (64). (63)
After two meals from the Small Palace Garden, Ella began to make dinner for the film-makers. She could tolerate fish and chips once a week, but would go no further along that path. (ICE-AUS W2F-020 9)
106 (64)
Chapter 4 His Honour would lightly touch the arm of a young waitress who was passing, he would put his hand on her wrist, she would feel constrained to bend close and listen to his whispered request, so that, with their foreheads almost touching, for a moment the two would seem to be lovers. He might even lift the fall of her hair from her cheek and hold it gently aside in a manner both suggestive and dotted line well, judicious, quaintly formal, certainly irreproachable, in order to murmur his order privately. (ICEAUS W2F-016 71)
Preterite could and might may also be used as backshifted counterparts of present tense can and may, as in (65) and (66). Here they are non-deictic, the past time reference being identified by the matrix verb. (65)
(66)
He told the C B I conference in Glasgow that Britain could protect its own interests within the European Community without having to protest about threats to national sovereignty (ICE-GB S2B-007 14) Virgo Lonergan had started life as Virginia. After seventeen years of being asked if Santa Claus existed, and five of smutty puns, she decided she might as well go the whole hog. (ICE-AUS W2F-008 26)
Backshifted could and might are not restricted to indirect reported speech (notice that the matrix clause in (66) above expresses a reported thought), and may be used to represent interior monologue, as in (67): (67)
How could she wait, with ever-increasing anticipation, for her dead child to come back to her (ICE-GB W2F-020 152)
4.3.1.2 Hypothetical uses Two subclasses may again be identified, which we shall refer to as ‘unreal’ and ‘weak’. The unreal use is found in the apodosis of unreal conditionals (with the protasis expressed as in (68), implicit in a construction other than an if-clause as in (69), or textually implied as in (70)), and in the complement of wish as in (71). Hypothetical could and might are used for talking about a counterfactual situation (for example in (68) it is presupposed that Mr Gorbachev does not have the implied level of respect, in (70) that the discussion has not taken place, and in (71) that the speaker cannot drive): (68)
(69)
In the meantime, Mr Gorbachev might achieve greater respect if he dropped the pretence that his republics would enjoy equality - let alone sovereignty. (ICE-GB W2E-008 22) An effective judicial control of the preparation of serious criminal cases for trial could greatly reduce the opportunities of police wrongdoing at that stage or at the very least expose such wrongdoing before trial (ICEGB S2B-037 113)
Possibility, permission and ability (70) (71)
107
There are many parts of the climate system that we could discuss uhm (ICE-GB S2A-0143 22) I felt so guilty not driving you on Frid on Saturday though I wish I could drive (ICE-AUS S1A-094 79)
Some unreal tokens are used to express what Huddleston and Pullum (2002: 201-202) refer to as ‘unactualized possibility’, which they regard as a distinct use. In (72), for example, we understand that the adoption of a different attitude did not occur, even though it was logically possible: (72)
So uh if it had been put like that you might have taken a different attitude (ICE-GB S1B-069 158)
The weak use of could and might has arisen via pragmatic extension from the unreal use. In the absence of an implicit condition the ‘irrealis’ meaning has receded, enabling the modal to superimpose a component of tentativeness or diffidence onto the meaning expressed by its present tense counterpart. In (73), for example, the illocutionary force associated with might, a tentative marker of dynamic possibility, is that of polite suggestion. (73)
It might be an idea if you were to spend some time with each head of department here, let them show you their areas of work (ICE-GB S1B-053 101)
Weak might serves most frequently as a diffident marker of epistemic possibility, a use that accounts for 59.5% of all tokens in the three corpora. AmE leads the way in the development of this use, which accounts for 63.0% of tokens in C-US, followed closely by AusE (62.5% in ICE-AUS) and BrE (53.8% in ICEGB). Moreover, as discussed in Section 4.3.2.1 below, a further development has taken place, with the diffidence associated with might eroding to the point that it has come to express a degree of likelihood comparable to that of epistemic may. In addition to this, as we shall see, could appears to be undergoing a similar semantic development, with a weak epistemic use evolving from the unreal use via bleaching of irrealis meaning, and subsequent shedding of its tentativeness marking. As indicated in Table 4.8, the temporal use is less common than the hypothetical use with both modals, in all three corpora, but relatively moreso with might (5.3:1) than with could (1.5:1). Of the three varieties AusE evidences the strongest relative popularity of hypothetical might over past might (6.3:1), but the least for hypothetical could over past could (1.3:1). This finding could be interpreted as suggesting that the process by which may and might are becoming increasingly distinct from one another as present and past inflectional forms of a single morpheme is more advanced in AusE than in the other two varieties (see further Section 4.3.3).
108
Chapter 4
One striking difference between might and could concerns the relative proportions of the unreal and weak hypothetical uses. In fact there is a degree of complementarity between the two preterite forms. With might the weak use is almost five times more popular (4.9:1) than the unreal use. By contrast with could it is the unreal use that is more popular than the weak use, by a broadly similar margin (4.4:1). The contrast is more pronounced in C-US than in the other two corpora (with a weak to unreal ratio of 5.7:1 with might, and an unreal to weak ratio of 5:1 with could). Table 4.8. Uses of might and could as preterite forms Past might
could
Unreal Hypothetical Weak Total Indeterminate Total Past Unreal Hypothetical Weak Total Indeterminate Total
ICE-AUS 93 81 503 584 18 695 686 659 230 889 15 1,590
ICE-GB 118 120 447 567 22 702 598 691 320 1,011 13 1,622
C-US 87 (17) 61 (12) 346 (68) 407 (80) 25 (5) 519 (102) 692 (136) 870 (171) 173 (34) 1,043 (205) 20 (4) 1,909 (345)
Total 298 (228) 262 (213) 1,296 (1,018) 1,558 (1,231) 65 (45) 1,916 (1,499) 1,976 (1,420) 2,220 (1,521) 723 (584) 2,943 (2,105) 48 (32) 5,121 (3,557)
4.3.2 Meanings of might and could The semantic profiles of might and could are broadly comparable to those of their present tense counterparts, may and can. Like may, might is dominantly epistemic. The main differences relate to the minor meanings of deontic possibility (which is extremely rare with might) and dynamic possibility (which is twice as popular with might as with may). Similarly, both could and can are dominantly dynamic, but the minor meaning of epistemic possibility is extremely rare with can, and that of deontic possibility with could.
Possibility, permission and ability
109
Table 4.9. Meanings of might and could
might
could
Dynamic Deontic Epistemic Indeterminate Total Dynamic Deontic Epistemic Indeterminate Total
ICEAUS 122 7 534 32 695 1,197 61 235 97 1,590
ICEGB 120 5 534 43 702 1,220 40 280 82 1,622
C-US
Total
%
76 (15) 0 (0) 412 (81) 31 (6) 520 (102) 1,430 (281) 41 (8) 188 (37) 97 (19) 1,756 (345)
318 (257) 12 (12) 1,480 (1,149) 106 (81) 1,917 (1,499) 3,847 (2,698) 142 (109) 703 (552) 276 (198) 4,969 (3,557)
16.6% 0.6% 77.2% 5.5% 100% 77.4% 2.9% 14.1% 5.6% 100%
4.3.2.1 Epistemic possibility In both their temporal and hypothetical uses might and could may express epistemic meaning. Epistemic could is not restricted – as can is largely (see Section 4.2.1 above) – to non-affirmative uses. Consider first the temporal uses, the past time use of epistemic could in (74)–(76) with reference respectively to a present situation (“it was not possible that there were”), a past situation (“it was possible that it had been”) and a future time situation (“it was possible that it would”) and the backshifted use of epistemic could in (77)–(79), again respectively with a present, past and future time situation: (74)
(75) (76)
(77)
(78) (79)
By Jean’s calculation, there couldn’t possibly be many more left before someone summoned sufficient daring to enter Forla’s presence. (ICE-GB W2F-015 60) But it could as easily have been a picnic at Cedar Creek Falls that he saw, or the railway cutting, how would I know? (ICE-AUS W2F-016 58) Another tree surgeon was unwilling to report in case he was proved wrong as it could do anything including collapsing under its own weight. (ICEAUS W1B 126) In the interim Ken Jack had been busily sabotaging our efforts by nugatorially ringing all and sundry and telling them we couldn’t possibly have enough water in the pond, it had taken five days last time and we had only forty-hours left. (ICE-AUS W1B-005 68) Dee reflected grimly that it was too bad that the other half of the worried duo couldn’t have been Hal. (ICE-GB W2F-006 155) He said the Federal Government's approach could lead to reduced investment (ICE-AUS W2C-008 38)
110
Chapter 4
There are no tokens in the corpora of past time epistemic might but the full set of possibilities occurs with backshifted might. In (80) it occurs with a present situation, in (81) with a past situation, and in (82) with a future situation. (80)
(81) (82)
She was terrified by her anger. It was like a tall, black wave rearing, threatening her dotted line she could lose her footing, drown in it, and she never knew where it might be waiting to rise. (ICE-AUS W2F-020 61) Yes I thought I might have got a chance today but there wasn't a lot of time (ICE-AUS S1A-016 124) and uh, so, we thought we might be OK this year cos I’m sending my son (ICE-GB S1A-087 114)
Epistemic might can also be used with general time situations, as in the past time example in (64) above and the backshifted example in (83): (83)
Apart from creating an impression that Rajiv could not meet the American president without straight away reporting back to Moscow, the Moscow stop-over gave rise to worries that he might be giving Mikhail Gorbachev advance information of an American posture at Geneva. (ICE-GB W2B011 26)
In this use might, like its present counterpart may, is close to the somewhat indeterminate borderline between epistemic modality (paraphraseable by “possible that”) and dynamic modality (paraphraseable by “possible for”). Semantically similar examples with could are found, as in (84), but the position adopted here – as argued above for can – is that they fall on the dynamic (existential) side of the borderline (see Section 4.3.2.2 below). (84)
During my banking or property days August could be a very busy month indeed (ICE-GB W2B-012 31)
Hypothetical (unreal) could and might can both express epistemic possibility (“it is possible that NP would ...”), with reference to a present situation in (85) and (86), a past situation in (87) and (88), and a future situation in (89) and (90). (85)
(86)
That’s funny cos I was reading somewhere that historically if you said someone has left his coat you could mean male or female (ICE-GB S1A053 319) You know Rachel you might have a chance if you if you got that accepted that that pavilion was sacrosanct to the men most probably because they want to be on their own (ICE-GB S1B-021 39)
Possibility, permission and ability (87)
(88)
(89)
(90)
111
It’s a photographer's paradise I could have spent the rest of my life well the rest of my Saturdays of my life wandering around here and I I would never have run out of material (ICE-AUS S2A-048 91) It’s an interdisciplinary field which if you if you ah occasionally read the Higher Education Supplement in The Australian you might’ve noticed is even now traversed by controversy and that’s exactly appropriate (ICEAUS S2A-037 37) Well I mean our s our corpus is could become very large indeed, if our plans go forward for the for the raising of the funding for the language centre ww are talking about very big money (ICE-GB S1B-076 11) Or if she was eighteen she might have a sexual relationship of her own (ICE-GB S1A-031 57)
The weak use of epistemic might derives historically from its ‘irrealis’ use, but insofar as there is no longer any requirement of an explicit or even implicit condition it is preferable to regard it as real rather than unreal. It is sometimes maintained (e.g. by Hermerén 1978, Palmer 1990) that epistemic might and may differ in terms of the degree of likelihood they express, with might lower on the scale. This view would seem to draw support from examples such as (91), where the speaker’s confidence about possible market penetration appears to increase with the switch from might to may, suggesting that might is being used as a more diffident marker of epistemic possibility than may. (91)
It is increasingly a question of how we can penetrate market places in the Asia Pacific region or other parts of the world where our products might, oh oh and indeed our services um may be desired (ICE-AUS S2A-028 51)
Coates (1983: 152), however, finds epistemic may and might in her data to express similar degrees of likelihood: “MIGHT, in my data at least, does not seem to express a more tentative meaning than MAY”. The data in the present study provide support for Coates’ claim, yielding numerous cases of epistemic may and might being used alternately without any apparent difference in diffidence, as in (92) and (93). (92)
(93)
there’s no doubt so it seems to me uh that for a considerable time, he may well have hoped uh that uh some at least, of these uh witnesses from Gibraltar who gave evidence at the inquest, uh might’ve been, in the case of those whose identity is not known, identified and might have been uh uh brought able to be brought here to give oral evidence (ICE-GB S2A063 68) I think Mrs Thatcher might possibly have done a lot of things uh that would’ve been very different stimulated the private sector to a substantial extent in the fourth term and I think sooner or later the Government may actually have to say that (ICE-GB S1B-039 53)
112
Chapter 4
Notice too that might, like may, is compatible with the strengthening adverb well: (94)
That’s some way short of it but it might well be an improvement on the lead (ICE-GB S2A-007 18)
In fact the present data in fact suggest that, even though epistemic may outnumbers its might counterpart (2310:1480 tokens per million words, or 1.6:1), and even though the semantic territory of may is more strongly dominated by present epistemic meaning (83.5%) than is that of might (60.2%), there are signs that might is beginning to threaten may as the primary exponent of epistemic possibility. One piece of evidence for this claim is that epistemic may and might sometimes figure in repairs, and in such cases it is usually may that is repaired to might, as in: (95)
I dunno Advice may be might be too harsh a word (ICE-AUS S1B-003 250)
Another is that present epistemic might has become the less stylistically unmarked of the pair: epistemic might occurs more commonly in speech than in writing (1721:1262, or 1.4:1), while epistemic may occurs far more commonly in writing than speech 3856:1270, or 3.0:1). It would seem that there is a more salient difference between present epistemic may and might in formality than in diffidence.2 A similar historical process to that which applied to might has seen ‘irrealis’ could also develop a weak epistemic sense. In fact, according to Leech (1987: 120-1), it is difficult to see any difference between could and might in sentences such as: There could be trouble at the Springboks match tomorrow; The door might be locked already; Our team might still win the race. His view is apparently shared by Quirk et al. (1985: 233), who claim that “could and might have the same meaning and both express the epistemic possibility associated with may”. Coates (1983: 167), however, finds epistemic could more tentative than might: “It does not seem implausible to suggest that, while might is becoming the main exponent of Epistemic possibility in every day spoken language and no longer expresses a more tentative meaning but is in most contexts synonymous with may, could is filling the gap left by might and is the new exponent of tentative Epistemic possibility.” Gresset (2003: 82) concedes that could is being used increasingly in “apparently epistemic or epistemically-oriented contexts”, but nevertheless argues that this does not necessarily entail that they are synonymous. According to Gresset, not only does could express a higher degree of probability than might, but it can never be purely epistemic (insofar as, in his view, the possibility expressed invariably derives from the subject referent’s properties or characteristics). As persuasive as the arguments of Coates and Gresset may seem, there is no shortage in the data of cases where epistemic could and might alternate freely,
Possibility, permission and ability
113
with no apparent difference in their meanings or in the level of likelihood expressed, as in (96) and (97): (96)
(97)
Failures that can be traced to excessively high or low temperatures might be beyond the scope of your maintenance contract and could result in a bill approaching half the original cost of the damaged equipment. (ICEGB W2B-033 14) The market has reacted to good news but, over the period of the conflict, something awful could happen and the market might react downwards. (ICE-GB W2C-012 86)
In the present study only 11.7% of coulds expressed present epistemic possibility, indicating that this meaning is as yet less established with could than it is with might (60.2%). Curiously, while the expansion of this meaning with might is more advanced in AmE (64.7%) and AusE (64.2%) than it is in BrE (55.6%), the reverse is the case with could, with BrE (13.1%) ahead of AusE (11.4%) and AmE (6.4%). Epistemic might and could can, like epistemic may (see Section 4.1.1), be used concessively. In such cases the degree of modality is low: in (98) and (99) the meaning would not differ significantly if the modal were omitted (“even if he is innocent”, “whatever that means”). (98)
(99)
He’s obviously a criminal and he should go down and and their arguments are basically there’s too much crime in society and if we let the buggers get away with the things even if he might be if if even if he might be innocent if he didn’t do this he’s done something else (ICE-AUS S1A-095 195) Now, how does that fit with our observation that sometimes when we hear people speak there are bits missing, whatever that could mean (ICE-GB S2A-030 136)
4.3.2.2 Dynamic possibility In both their past time and hypothetical uses could and might parallel can and may in their dynamic uses. Given the number of uses, and in some cases subtypes of these, associated with the dynamic meaning, we shall not attempt to provide a full set of examples. Consider firstly theoretical possibility. In the most central cases the potentiality lies in the physical situation, an appropriate paraphrase being “have an opportunity to”, expressed by past temporal could and might in (100) and (101) respectively, by backshifted temporal could in (102), and by unreal hypothetical might and could in (103) and (104) respectively.
114
Chapter 4
(100) I realised fairly quickly what I had done and was able to get Stuart to grab the stuff on my lap, so I could put my head behind the newspaper better and my leg down without anything else sliding. (ICE-AUS W2B-004 63) (101) Such a glittering vehicle as this might occasionally be seen elsewhere in the city, but never in their lives had these ragged, undernourished people seen a pale blue Jaguar roadster in their dirty alleys, the gold insignia of General Forla glinting malignantly as it passed. (ICE-GB W2F-015 32) (102) He said they didn’t have any water in wh in whi in which they could bathe (ICE-GB S1B-038 102) (103) This is of course we know now a a rather simplistic kind of view of how his uh ends might have been achieved (ICE-GB S2A-026 59) (104) If many more views of the object were available the shape could be more easily established (ICE-GB W2A-036 57) The rational subtype is exemplified with past temporal might in (105), and with hypothetical could and might in (106) and (107). In (105) the adjectival modifier reasonable, and in (107) the adverb reasonably, provide harmonic reinforcement. (105) It landed me in some small adventures such as no reasonable AngloPrussian might expect (ICE-GB S2B-025 2) (106) A high level of population/employment growth, high labour mobility and successful labour market programs has ensured that Australia has avoided the structural unemployment problems experienced in Europe. It could be argued however that the high unemployment rate may in itself create structural unemployment. (ICE-AUS W1A-008 62) (107) Um what are Australia's top priorities in public expenditures which might reasonably be directed in the next decade to research and development (ICE-AUS S2A-028 29) The implicitly quantificational, existential, subtype is exemplified with past temporal could in (108) and unreal hypothetical could in (109), where the proposition is understood to apply to a subset of members of the relevant population (“Some Augusts were very busy”, “There are some individuals it would not be possible to hypnotise …”). (108) During my banking or property days August could be a very busy month indeed (ICE-GB W2B-012 31) (109) Some individuals couldn’t be hypnotized and you couldn’t produce that, but to a large proportion of the population you could hypnotise them and say to them under hypnosis, your left arm is paralysed (ICE-GB S1B-070 85) The ability use is found almost exclusively with could: the only instances of ability might appearing archaic, as in (110):
Possibility, permission and ability
115
(110) Frantic now, not thinking how he might control his rage, he ran from room to room, scattering sundry possessions from the dressing-table, bathroom shelves, kitchen cupboards (ICE-GB W2F-008 115) As observed above, the ability use differs from the other dynamic uses we have discussed to the extent that the potentiality in question is located within the subject-referent. As with ability can, the basis on which such potentiality is predicated of the subject-referent is prior actualization, as in (111). (111) But I mean it’s it yeah it’s even at a higher level than that though because I mean Pete could do basic things on a computer but it wasn’t enough (ICEGB S1A-005 189) However this is not a necessary condition: in (112) we are led to understand that Weary has not previously swum across lake Eacham. (112) Weary has never been able to resist a challenge, especially one involving extraordinary physical effort, and he accepted a bet that he couldn’t swim across Lake Eacham and back, a distance which he calculated to be about a kilometre and a half. (ICE-AUS W2B-005 25) The lexical verbs in this use may denote not only acquired skills as in (111) and (112), but also innate capacities, as in (113), in which case non-human and even inanimate subjects are possible: (113) The Combi couldn’t do more than about twenty miles an hour (ICE-AUS S1A-059 149) The potentiality sense is associated not only with temporal uses as in (111), and (112), but also hypothetical, as in (114) and (115). (114) For instance, the same system viewing a 60 km by 60 km area from a low orbit could transmit data to give 10 m resolution, but used in geostationary orbit to cover a whole hemisphere would have to be restricted to 5 km resolution using the same data transmission rate. (ICE-GB W2A-037 58) (115) Frantic now, not thinking how he might control his rage, he ran from room to room, scattering sundry possessions from the dressing-table, bathroom shelves, kitchen cupboards (ICE-GB W2F-008 115) When ability could is used with verbs of perception as in (116), and cognition as in (117), the degree of modality is low, as evidenced by the fact that could could typically be readily omitted (“They heard the phone ringing”; “Dr Hewson said he understood Mr Bradbury’s frustration”).
Chapter 4
116
(116) As they drove into the garden, they could hear the phone ringing through the empty house. (ICE-AUS W2F-004 122) (117) Dr Hewson said he could understand Mr Bradbury’s frustration but he must accept that he and the party organisation do not make the policies of the parliamentary party. (ICE-AUS W2C-004 28) The currently actualized sense may be associated not only with the temporal uses as in (116) and (117), but also hypothetical as in (118). (118) They’ve got all the cheeses and they’ve got ah eggs and brus bread and bread rolls and pitta bare bread um and all the nuts of course as well as all the beautiful fruit that you could imagine (ICE-AUS S1A-092 93) Dynamic implication, found mainly with could, is a pragmatic extension of the unreal hypothetical use, with the utterance being used to perform a speech act other than mere assertion: for instance, offer as in (119), and request as in (120): (119) I could uhm get you that other book when I stay at Hilda’s if you’re near U C (ICE-GB S1A-053 167) (120) In your letter to me you say that “it is not the ownership of the NRMA that is under review.” Could you explain that to me please? (ICE-AUS W1B026 86) Might can be used to implicate a suggestion, as in (121), and is regularly used to do so in the idiom might as well, as in (122): (121) Now if if that is the form that your publication is is going to take um then one of the things that you might also think about as an adjunct to your address is the use of audio-visuals or at least visuals. (ICE-AUS S1B-003 220) (122) I used to just always heat the house because it was always if you if your house was cold the kids get sick and you pay it out in medical bills anyway if so you might as well just forget about it just heat your house (ICE-AUS S1A-046 15) Dynamic implicature is typically found with 1st and 2nd person subjects, but 3 person subjects are also possible, as in (123): rd
(123) It would be advantageous if the initial clean could occur the previous weekend 20/21 July. (ICE-GB W1B-028 29)
Possibility, permission and ability
117
4.3.2.3 Deontic possibility In the present corpora all tokens of deontic possibility, or ‘permission’, might were subjective, and generally found in questions as in (124), and in conditionals as in (125), with the addressee as the deontic source: (124) My oh why I don’t lose sl I don’t lose sleep over it might I say (ICE-AUS S2B-047 181) (125) If I might just turn to that ah and just point out where I think that the the defect is (ICE-AUS S1B-057 28) Permission could, by contrast, yielded both subjective tokens as in (126) and objective as in (127). (126) Steve Kayland, could I ask you to comment here (ICE-AUS S1B-021 181) (127) Anderson states the British imperialism was a cultural policy of anglization anglicisation and stresses the incompatibility of empire and nation. Australians could not become Governors-general until much later on and these positions were held by the English English. (ICE-AUS W1A007 27) 4.3.3 May and might: one lexeme or two? Could regularly serves as a temporal past form and is uncontroversially regarded as the preterite counterpart of can. The evidence for a comparable inflectional relationship between may and might is somewhat less compelling. As we have seen, might is rarely temporal past (see Section 4.3.1.1). In the epistemic meaning (see Section 4.3.2.1) the level of likelihood expressed by ‘weak’ might is no longer systematically distinguishable from that expressed by may. Conversely, may has assumed functions once the sole province of might, as suggested by examples – encountered more commonly in the Australian data than in the British or American – of the type in (128) where may is used in backshift, and in (129) where it is used in an unreal conditional. (128) The Sandanistas thought that the invasion of Grenada may help their prospects for the 1990 election which resulted from a peace process initiated by the Central American governments. (ICE-AUS W1A-014 171) (129) Queenslanders may be much better off if there was simply a moratorium on new Government services and programs until the economy picks up. (ICE-AUS W2C-002 197) It would appear that the relationship between may and might is considerably less regular than that between can and could, supporting Huddleston and Pullum’s (2002: 202) claim that for many speakers today “may has been reanalyzed as
118
Chapter 4
lexically distinct from might: they are forms of different lexemes, both present tense forms.” 4.3.4 Might and could: regional and stylistic variation Mair and Leech’s (2006) findings indicate that might is in decline both in BrE writing (with a decrease of 15.1% in the period spanning the early 1960s to the early 1990s) and AmE writing (4.5%): see Table 1.4 above. However a comparison of its frequencies in speech and writing suggests that might’s future may not be as bleak as that of its present tense counterpart may. As noted in Section 4.1.4 above, may is over three times more common in writing. By contrast, might is more popular in speech than writing (by a ratio of 2115:1619, or 1.3:1). There are some striking regional differences: in the Australian corpus might is more than twice as common in speech as writing (875:425, or 2.05:1), but in C-US marginally more common in writing (539:507, or 1.06:1). The apparent association of might with more conservative, formal, usage in the American corpus is consistent with its relatively smaller overall numbers. Table 4.10. Genre distribution of might Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total
ICE-AUS 958 (345) 750 (180) 875 (525) 240 (24) 487 (146) 425 (170) 695
ICE-GB 872 (314) 525 (126) 733 (440) 520 (52) 700 (210) 655 (262) 702
C-US
Avg
507 (59)
705
539 (43) 520 (102)
540 639
By comparison with might, could is more frequent in C-US than in the other two corpora. Mair and Leech’s (2006) finding that could has undergone a 6.8% decline in recent AmE is perhaps reflected in its more modest popularity in speech over writing in C-US (1.02:1) than in ICE-GB (1.13:1) or ICE-AUS (1.18:1). Table 4.11. Genre distribution of could Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 1,883 (678) 1,325 (318) 1,660 (996) 1,540 (154) 1,467 (440) 1,485 (594) 1,590
ICE-GB 1,925 (693) 1,358 (326) 1,698 (1,019) 1,370 (137) 1,553 (466) 1,508 (603) 1,622
C-US
Avg
1,767 (206)
1,708
1,738 (139) 1,757 (345)
1,577 1,656
Possibility, permission and ability 4.4
119
Be able to
With a total of 889 tokens in the three corpora, the lexico-modal be (un)able to does not threaten the numerical supremacy of can/could, whose combined total is 11,220 tokens. 4.4.1 Meanings of be able to There are conflicting opinions in the literature regarding the semantic relationship between can and the lexico-modal be able to. According to some, be able to can only express ability. For example Hermerén (1978: 83) claims that “be able to will be used as a paraphrase of can (in the ability sense)”, and Facchinetti (2000: 118) that “Of the three values conveyed by can – possibility, ability, and permission – the modal appears to be paraphrased by use of be able to only with the value of ability”. By contrast Coates (1983: 124) is of the view that be able to covers the range of meaning associated with can. The truth lies somewhere inbetween: be able to can certainly express more than merely ability, and yet at the same time there are several uses of can that cannot be served by be able to. The lexico-modal cannot be used for subjective deontic possibility, dynamic existential possibility, present actualized ability with ‘private’ verbs of perception and cognition, or epistemic possibility, and thus could not readily be substituted for can in the following: (130) She she she came to the point where she said you know you don’t have to go you can stay al alright (ICE-AUS S1A-064 86) (131) it’s her it’s her duty to be on the door every night but um yeah it can be a bit of a nuisance sometimes (ICE-AUS S1A-074 125) (132) Yes you said I can understand that (ICE-AUS S1A-006 279) (133) It can’t possibly be faster on the back (ICE-AUS S1A-005 277) However, as the following examples show, apart from these restrictions, be able to can be used across the range of meanings expressed by can. With the exception of the two dynamic uses illustrated in (131) and (132) above, all the other dynamic uses of can are possible with be able to. Thus, the lexico-modal expresses ability in (134), circumstantial potentiality in (135), rational possibility in (136), and dynamic implication (here with the illocutionary force of a suggestion) in (137). In (138) it expresses objective deontic possibility: (134) And I’m sure even the affirmative team once they’d eventually nutted out the on part would be able to work out how to turn the bloody thing off (ICE-AUS S1B-036 219) (135) I just presumed that you were able to weigh yourself at Brin’s (ICE-GB S1A-011 183) (136) However, not all is rosy. Jill Matthews in 1988 (1986?) was able to complain about the far greater number of male contributions to the
120
Chapter 4
Australian bicentenary project than women, and also to state that, in these volumes, the role of women generally remained confined to childcare and food production, and providing an audience while men got on with whatever real business. (ICE-AUS W1A-018 92) (137) I said on the ’phone, we have not reached any decision about this and it would be very helpful if we were able to discuss it further. (ICE-GB W1B-017 80) (138) It is not clear whether a publisher is able to control the making of a machine readable copy of the edition by exercise of the published edition copyright. (ICE-AUS W1B-17 60) As Table 4.12 indicates, the dynamic uses are dominant with be able to (representing 90.4% of all tokens, compared with 81.0% for can and 77.4% for could), with only a handful of deontic cases. Table 4.12. Meanings of be able to Ability Theor poss Deontic Indeterminate Total
ICE-AUS 139 209 18 21 387
ICE-GB 204 182 23 25 434
C-US 158 (31) 163 (32) 10 (2) 15 (3) 346 (68)
Total 501 (374) 554 (423) 51 (43) 61 (49) 1,167 (889)
% 42.9% 47.5% 4.4% 5.2% 100%
Both Facchinetti (2000) and Aijmer (2004) attempt to distinguish can and be able to on the basis of an implication of actuality carried by the latter, often paraphraseable by ‘manage to’ or ‘succeed in’. Thus in (139), where could would sound quite unnatural, was able to conveys the subject-referent’s successful achievement of its victory (“the Labor Government managed to achieve/ succeeded in achieving its victory”). (139) But now it’s time now it’s time for the Student Movement to ask the Federal Government to return that favour in kind because there’s no doubt that it was due to the support the support from groups such as the Student Movement that the Labor Government was able to achieve its victory (ICE-AUS S2A-040 90) However, this claim is problematical, or at least in need of extensive qualification. For one thing, as we have seen in Section 4.2.3.2 above there are uses of can (most notably with verbs of perception and cognition) where we can speak of actualization. Furthermore the difference is far more salient when we compare the preterite forms could and was/were able to than it is when we compare the present forms can and am/is/are able to. In (140) the present generalizations about the abilities of the human body feature a number of relevant expressions including not only can and be able to but also ability, able-bodied
Possibility, permission and ability
121
and capable, with no apparent difference in implied actuality. (140) And it also makes you so aware of your own ability, and other people’s abilities, and how the body works whereas, uhm when you’re uhm working with a very able-bodied people you become very blasé about it I think and you forget, how complicated the body is and how, wonderful it is to be able to move certain parts and ways of doing it and the various different ways you can do it and even when you can’t actually, uh you aren’t capable of moving how you can get around these problems and become capable of moving (ICE-GB S1A-002 45) On the other hand, as noted by Palmer (1990: 90): “In the past tense BE ABLE TO is used to indicate ‘actuality’ in environments in which CAN cannot be used”. Thus be able to is virtually obligatory in assertive contexts with a single action as in (141), and is preferred to could when habitual or repeated actions are referred to, as in (142). (141) Because I had been carrying the tape recorder with me, I was able to record the authentic sounds of the of the various technicians and therapists at work, when I went to the National Hospital for check ups. (ICE-GB W2B-001 55) (142) However there have been highly hierarchical society pastoralists groups such as the Mongols who, through intense military organization were able to conquer a huge area of land, maintaining an empire that negotiated trade routes between the East and the West. (ICE-GB W1A-011 57) This restriction does not apply to non-assertive contexts, however, presumably because these carry no implication of actuality. Thus could not is possible as a replacement for the semantically negative were unable to in (143): (143) Done first saw The Cabin when he was 14, began renting it in the early eighties, but was unable to buy it until 1987. (ICE-AUS W2B-010 198) It is perhaps the availability of preterite was/were able to for indicating actualization, where could is unable to do so, that explains the different proportions of present/preterite forms for be able to (134 preterite tokens as against 124 present: see Table 4.13) and can (7663 present tokens as against 3557 preterite).
122
Chapter 4
Table 4.13. Be able to and tense Present Preterite Non-tensed Total
ICE-AUS 49 55 283 387
ICE-GB 67 67 300 434
C-US 41 (8) 61 (12) 244 (48) 346 (68)
Total 157 (124) 183 (134) 827 (631) 1,167 (889)
% 13.5% 15.7% 70.9% 100%
4.4.2 Be able to: regional and stylistic variation Of the three corpora ICE-GB has the largest number of tokens (434) and C-US the smallest (346), with ICE-AUS in-between (387). Interestingly, however, CUS is the only corpus in which be able to is more frequent in speech than writing (by a ratio of 412:250, or 1.6:1, compared to 343:453, or 0.75:1, for ICE-AUS and 380:515, or 0.73:1, for ICE-GB), perhaps suggesting that despite its modest numbers in C-US the lexico-modal is faring comparatively well. Table 4.14. Genre distribution of be able to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 325 (117) 371 (89) 343 (206) 650 (65) 387 (116) 453 (181) 387
ICE-GB 381 (137) 379 (91) 380 (228) 800 (80) 420 (126) 515 (206) 434
C-US
Avg
412 (48)
378
250 (20) 346 (68)
406 389
Notes 1
Even though, strictly speaking, able and unable are separate lexemes they are here treated as belonging to the same lexico-modal on the grounds of the close relationship between not able to and unable to.
2
Trousdale’s (2003) study of non-standard Tyneside English reveals trends far more advanced than in the present study. In this variety might has become the main marker of epistemic possibility, with 91.2% of might tokens expressing this meaning (as against 79.2% for standard BrE in the present study). A more striking finding in the Trousdale study is that epistemic possibility accounted for only 8.8% of may tokens (whereas in the present study it seems to be more a matter of might catching up to may
Possibility, permission and ability
123
– 79.0% of whose tokens expressed epistemic possibility – than a case of may declining).
Chapter 5 Prediction and volition While necessity and possibility are firmly established domains of modality, logically related to each other and involving clearly identifiable epistemic and deontic meanings, by contrast the prediction/volition domain discussed in this chapter is less central. It includes epistemic uses of a temporal nature which may involve a minimal component of modal meaning. Furthermore, the dominant type of non-epistemic meaning, involving the subject-referent’s volition, has received little attention in the literature on modality, a situation explained by Krug (2000) as follows: Due to the progressive grammaticalization of the will future since Middle English, no central modal has ‘desire’ as its central notional domain any longer, even though some volitive traces can be found in will and would. (Krug 2000: 117) Volition has, furthermore, been associated with different types of modality: not only dynamic, as in the present study, but also deontic (e.g. Traugott 1989: 38) and agent-oriented (Bybee et al. 1994: Chapter 6). The expressions examined in this chapter are the modals will (embracing the forms will, ’ll and won’t, and the preterite forms would, ’d and wouldn’t) and shall (shall, shan’t), and the quasi-modals be going to (going to preceded by all the inflectional forms of be, along with gonna and gunna), want to (want to, wants to, wanted to, wanting to, plus wanna), and be about to (about to preceded by all the inflectional forms of be). As Table 5.1 shows, this group is dominated by will and would, which together account for 75.2% of all tokens across the three corpora. Shall is by contrast a minor modal (shown by Mair and Leech 2006 to be in severe decline: see Section 1.4 above), with an average frequency 27.4 times smaller than that of will. The quasi-modals be going to and want to are both substantially more popular in C-US than in the British and Australian corpora, a finding that is compatible with Mair and Leech’s (2006) finding that they have undergone a massive increase in AmE writing in recent decades (see Section 1.4). Although will is not under threat from want to in terms of overall frequency, a comparison of the volitional tokens (2,296 for will and 2,137 for want to: see Table 5.3 and 5.14 below) suggests that with respect to this meaning the quasi-modal may be challenging the modal.
126
Chapter 5
Table 5.1. Frequencies of the expressions of prediction and volition Modals
Quasimodals
Total
ICE-AUS will 3,868 shall 100 would 3,585 Total 7,553 be going to 1,191 want to 1,039 be about to 63 Total 2,293 9,846
ICE-GB 3,861 223 3,404 7,488 1,056 858 54 1,968 9,456
C-US 3,950 (776) 102 (20) 4,001 (786) 8,053 (1,582) 2,413 (474) 1,425 (280) 36 (7) 3,874 (761) 11,926 (2,343)
Total (8,505) (343) (7,775) (16,623) (2,721) (1,897) (124) (5,022) (21,645)
Avg 3,893 142 3,663 7,698 1,553 1,107 51 2,717 10,409
NB For C-US figures are normalized to tokens per one million words; raw figures in parentheses.
5.1
Will
Will is primarily an exponent of epistemic modality, this meaning accounting for 59.2% of all tokens across the three corpora, approximately twice as many as for dynamic modality (30.4%), while deontic modality was a minor meaning in the data (1.7%). Table 5.2. Meanings of will Epistemic Dynamic Deontic Indeterminate Total
ICE-AUS 2,662 945 68 193 3,868
ICE-GB 2,523 1,044 84 210 3,861
C-US 2,128 (418) 1,563 (307) 51 (10) 209 (41) 3,950 (776)
Total 6,913 (5,603) 3,552 (2,296) 203 (162) 612 (444) 11,679 (8,505)
% 59.2% 30.4% 1.7% 5.2% 100%
5.1.1 Epistemic will There are two main uses of epistemic will, which we shall discuss in turn. One is the uncontroversially epistemic use in which will is found with present and past situations, and labelled ‘central-epistemic’ by Huddleston and Pullum (2002: 188), and ‘predictability’ by Coates (1983: 169). The other use, where the status of will as temporal or modal has been subject to debate in the literature (e.g. Lyons 1968: 310, Palmer 1990: 163), is found with future time situations. The use is labelled ‘futurity’ by Huddleston and Pullum (2002: 188), and ‘prediction’ by Coates (1983: 169). We shall adopt Coates’ terms ‘predictability’ and ‘prediction’.
Prediction and volition
127
In (1) and (2) predictability will conveys the speaker’s confidence in the truth of the proposition, based on evidence and knowledge of, respectively, a present situation and a past situation: (1)
(2)
In other words, referring to a particular instance of the the total idea, a home truth or a a particular type of truth then it will have those properties (ICE-AUS S2A-022 92) It has to be said that some of the reason for this is that S&P rate only the top tier of companies in Europe, so the highly publicised troubles that have hit smaller companies don’t show in the European statistics, while they will have been a major factor behind the increase in the US downgradings. (ICE-GB W2C-013 55)
Predictability will predominantly involves subjective modality, but occasional objective examples are found, as in: (3)
However, the positive side of C3 is connected to the supply via R11. Because its negative end is now connected to a low, C3 will charge towards 12V. When C3 has charged sufficiently, the input to IC1c will be a logic 1, causing its output to switch low. (ICE-AUS W2D-013 76)
In such cases there is a low degree of modality: the evidence for the factuality of the proposition is so compelling that it would be possible to substitute is for will be (“the input to IC1c is a logic 1”) with little change of meaning. Predictability will is semantically strong, comparable in strength to epistemic must. In (4) epistemic will and must alternate, and it would be possible to substitute one for the other without altering the strength of the speaker’s claims: (4)
Douglas Hurd will doubtless be grateful for that endorsement not least because of the pressure he must be feeling this morning. (ICE-GB S2B009 87)
Nevertheless must and will are dissimilar in other ways, as Huddleston and Pullum (2002: 189) observe: Must conveys the idea of conclusion, and is often used in explanations: Ed’s late – he must have overslept. With central-epistemic will it is more a matter of assumption or expectation, very often with a suggestion of future confirmation, as in: I can’t tell you what the word means but it will be in the dictionary. (Huddleston and Pullum 2002: 189) The second, ‘prediction’, use of will has prompted a good deal of debate concerning its status as a marker of future tense or an epistemic modal (e.g. Declerck 1991: 8-13, Huddleston 1995, Larreya 2000). It is excluded from the
128
Chapter 5
epistemic category by Palmer, who nevertheless concedes that: “Where there is reference to future action, it is difficult, and sometimes impossible, to distinguish epistemic WILL from the WILL of futurity” (1990: 57). Palmer’s case for maintaining a distinction between futurity will and epistemic will rests primarily on examples such as (5) and (6) which, in his view, involve factual assertions that lack any element of speaker judgement (Palmer 1990: 163). (5) (6)
It’ll be Christmas soon. (ICE-GB W2F-016 115) ’Cos she’s she’ll be seventeen after August so seventeen’s normal but most people are seventeen turning eighteen and she’s sixteen turning seventeen (ICE-AUS S1A-036 75)
It is certainly true that the degree of modality is low here, with will seemingly little more than a mere marker of futurity (as indicated by the fact that will could be readily omitted: “It’s Christmas soon”; “she’ll be seventeen after August”). While the epistemic modal component may be minimal, it is nevertheless not completely absent: even in the case of (5) and (6) the prediction could be prevented from coming true by an unforeseen eventuality, such as the death of the subject-referent in the case of (6). As Huddleston and Pullum (2002: 190) observe, there is an intimate connection between futurity and modality: “our knowledge about the future is inevitably much more limited than our knowledge about the past and the present, and what we say about the future will typically be perceived as having the character of a prediction rather than an unqualified factual assertion”. In the present study the will of ‘prediction’ is treated as belonging to epistemic modality rather than being merely a marker of futurity. A further indication of the low degree of modality associated with prediction will is its compatibility with a wide range of modal adjuncts of different strengths, such as hopefully in (7), possibly in (8), and most likely in (9), which provide more specific modal meaning in the absence of a distinctive epistemic component expressed by the modal. (7)
(8) (9)
The following notes provide guidance on some of the conditions placed on these arrangements, and hopefully will serve to avoid misunderstandings. (ICE-GB W2D-006 118) Depending on who comes, you’ll possibly need to bring sleeping bags and I hope you don't mind sleeping on the floor. (ICE-GB W1B-004 53) you’ll most likely strengthen your defenses (C-US Frown G03 118)
The case for regarding prediction will as expressing epistemic modality is based not merely on the modal overlay that accompanies instances of its use, but also on the co-occurrence patterns that it shares with predictability will. For example, both predictability will and prediction will may collocate with the epistemic adjunct probably, as in (10) and (11) respectively:
Prediction and volition (10)
(11)
129
The bait will probably be stored in a fridge in the same medium in which it was supplied from the maggot farm (invariably sawdust). (ICE-GB W2D017 52) The sun is out, just a spasm, but by this arvo it will probably be slightly warmer. (ICE-AUS W1B-003 21)
Predictability will and prediction will may both be used with the perfect aspect, as in (12) and (13): (12)
(13)
Well if it’s any sort of amendment at all and the proposer doesn’t accept it then it can’t go straight to the vote until uhm the amendment has been li discussed because there won’t have been a speech against the motion by by b by by by the very nature of the fact you cannot move it until there’s been a speech against it (ICE-GB S1A-068 26) So in just three years with Richmond Jackson will’ve played just nineteen games (ICE-AUS S2B-016 183)
In (14) and (15) both are used with the progressive aspect: (14)
(15)
Okay And the events that they’ll be talking about will be occurring in the here and now whereas with decontextualized language the entities are likely to be absent from the material situation or setting or they're likely to be generic kind of entities (ICE-AUS S2A-046 64) His lead up months haven’t been too bad and he’ll be running on at the end of the race (ICE-AUS S2A-017 131)
In (16) and (17) both are used with the passive voice: (16) (17)
It will gradually be incorporated into the soil by earthworms and other soil creatures. (ICE-GB W2D-016 29) At the end of the seventy five days the licence will be allocated to us I believe (ICE-AUS S1B-046 215)
In (18) and (19) both are used with non-agentive verbs: (18)
(19)
A good quality pencil has a bonded lead, usually etched with acid, that will not shatter inside the wood if the pencil is dropped, nor slide out in use. (ICE-GB W2D-016 29) These are just a few of the visible benefits that will come about as chemical farming methods are gradually abandoned. (ICE-GB W2B-027 46)
In (20) and (21) both are used with stative verbs:
130
Chapter 5
(20)
Even the most moderately priced unit these days will have a zoom facility, which ideally should give a bottom picture section of about thirty metres. (ICE-AUS W2D-015 77) They won’t like that, and if you feel like me ... you couldn’t face being stripped by Manangans anyhow. (ICE-GB W2F-015 163)
(21)
In (22) and (23) both are used in the existential there-construction: (22)
(23)
And within any phrase there will be an internal structure consisting of the main word, usually referred to as the head and the subordinate words or the dependent words (ICE-AUS S2A-022 31) Don’t forget next weekend is um Anzac weekend though so there’ll be a lot of people away (ICE-AUS S1A-023 323)
Finally, in (24) and (25) both are used in the it-extraposition construction: (24)
(25)
In other words, if you look around the languages of the world, it will probably be true that if they have nouns and if they have verbs and if they have adjectives, the most common sorts of roles that these words play will be to refer to things in the case of nouns, to refer to actions in the case of verbs, to refer to descriptions in the case of adjectives (ICE-AUS S2A-022 205) Uh it’ll probably be rather nice actually actually trying to match that up (ICE-GB S1A-086 100)
Prediction will is semantically strong (like predictability will but unlike other epistemic modals that may be used with reference to future time situations, such as may, might, should, could). Note the contrast between epistemically weak might and semantically strong will in the following example: (26)
You might be admired for your red tie, but you will be derided for your seventies attitudes. (ICE-AUS S2A-044 34)
Epistemic will may also, as both Palmer (1990: 138) and Huddleston and Pullum (2002: 191-192) observe, convey a conditional meaning: more precisely, it may be used in the apodosis of conditional constructions, where it has a stronger sense than weak epistemic may. The time of the situation may be present or past (with the will of predictability) as in (27) and (28), or future (with the will of prediction) as in (29). In this conditional use will is usually objective, and low in modality (note that the will-clause in (27) is a repair for the incomplete, and unmodalized, you actually see them in the papers). (27)
And if you keep your eyes open you actually you’ll see them in the papers as as as facts (ICE-GB S1A-063 167)
Prediction and volition (28)
(29)
131
But if you take those patients on the weekly regimen who finish their treatment at this point in time here they will have received a much higher overall dose than patients on the three-weekly regimen at the same point in time (ICE-GB S2A-035 70) The organisation ARK has said that sea-level will rise by one metre if present pollution levels and conditions continue (ICE-GB S1B-007 199)
The figures in Table 5.3 suggest that speakers use epistemic will far more commonly to make predictions and predictability judgements about a third party (80.0%) than about either themselves (8.7%) or their addressee(s) (11.2%). By contrast with epistemic will, dynamic will occurs most commonly with a 1st person subject (72.2%), speakers presumably finding themselves more often in a position to describe their own volition than that of others. Table 5.3. Person of the subject with will 1st Epistemic 489 (8.7%) Dynamic 1,658 (72.2%) Deontic 0 (0.0%) Indeterminate 145 (32.7%) Total 2,292 (26.9%)
2nd 630 (11.2%) 51 (2.2%) 149 (92.0%) 30 (6.8%) 860 (10.1%)
3rd 4,484 (80.0%) 587 (25.6%) 13 (8.0%) 269 (60.6%) 5,353 (62.9%)
Total 5,603 (100%) 2,296 (100%) 162 (100%) 444 (100%) 8,505 (100%)
Whereas modals in the other semantic groups enter into contrasts of internal and external negation, such as that between mustn’t and needn’t (see Section 3.4.1), epistemic will does not enter into such contrasts. In the case of both predictability won’t as in (30) and prediction won’t as in (31) the negation is internal (“it can be predicted that not-p”), and there are no forms available in which the modality rather than the proposition is negated. (30) (31)
your English guide thing is is sure to have that won’t it (ICE-AUS S2B032 74) Oh well ours won’t be till September as usual (ICE-GB S1A-087 123)
As Huddleston and Pullum (2002: 192) point out the behaviour of epistemic will with respect to negation is a reflection, along with the fact that as noted above it is compatible with modal adjuncts of differing strengths, of its status as a lowdegree modal. 5.1.2 Dynamic will Dynamic will expresses the potential for an activity or event deriving, characteristically, from the subject-referent’s volition. In (32) there is a contrast between the first instance of won’t, which expresses this meaning (“I am unwilling/reluctant to take the case”), and the second, which occurs within a
132
Chapter 5
clausal complement to the noun likelihood with the non-agentive verb come (up with) and expresses epistemic prediction (“I can be predicted not to”): (32)
I’m not saying I won’t take the case, but you’ve got to be prepared for the likelihood that I won’t come up with anything. (ICE-AUS W2F-008 73)
A distinction may be drawn within the volitional will category, one comparable in some respects to the temporally-determined distinction between (future) prediction and (non-future) predictability, between ‘intention’ and ‘willingness’ (Coates 1983: 169-176). Intentional will, exemplified in (33) and (34), focuses upon a future event that is planned, promised, threatened, etc. (Coates’ ‘intention’), whereas with willingness will, exemplified in (35) and (36), the focus is upon the present state of mind of the subject-referent. (33) (34)
(35)
(36)
All right Go on Don’t be shy I won’t bite you yet but my God I will soon (ICE-AUS S2A-056 37) We have submitted Purchasers Requisitions to the Vendor. We will undertake the usual searches as soon as we receive your advices that finance has been approved. (ICE-AUS W1B-021 36) Uh and then having got that at the end of the day it’s got to be your decision anyway, because I won’t enter into a contract that says you ought to go and do that (ICE-GB S1A-035 123) Will you please explain to me the meaning of the phrase “Currently, NRMA’s profits are “locked up” ” used in answer to L.G. Norman’s letter? (ICE-AUS W1B-026 78)
The strength of the volition with willingness will may be heightened, so that it expresses determination or insistence. In the most typical cases will occurs with a 1st person subject, is stressed and is not contractible to ’ll, as in (37) below. (37)
I certainly will yes (ICE-GB S1A-100 37)
Willingness will may often be glossed by periphrastic expressions such as be willing to, and be prepared to. In (38) the alternation between will and be willing to is suggestive of semantic closeness: (38)
But there’s a lot of people you get who who won’t accept that aren’t willing to argue (ICE-GB S1A-084 117)
However as Huddleston and Pullum (2002: 192) warn, the auxiliary differs from the periphrastic expressions in several respects. For one thing, the auxiliary expresses stronger modality. In (34) above, for example, We will undertake the usual searches entails the actualization of the searches, which explains why it would sound odd to add but we may not manage to do so. Compare the acceptability of We are willing to/are prepared to undertake the usual searches
Prediction and volition
133
but we may not manage to. Also suggestive of a semantic difference between dynamic will and the periphrastics is the absence of a contrast with dynamic will, as with epistemic will, between internal and external negation. In (35) for example the negation type is unclear (though it may be preferable to regard it as internal, insofar as it is our non-entering into a contract that is entailed rather than our lack of volition), by contrast with the clear difference between We are not willing to enter into a contract and We are willing not to enter into a contract. Finally, whereas with dynamic will the time of the volition can only be present, with the periphrastics it can also be past (was willing to, was prepared to, etc.) or future (will be willing to, will be prepared to, etc.). According to Huddleston and Pullum (2002: 193) the volitional meaning of will is to be interpreted as a futurity-related implicature, deriving from the assumption that a future situation is subject to the will of subject-referent. They note, furthermore, that with a 1st person subject, as in (33) and (37) above, volition may be associated with an additional implicature of commitment, such that the speaker’s speech act might be thought of as a promise or an undertaking. However, the distinctive co-occurrence patterns that we have noted above to be displayed by epistemic will suggest that dynamic and epistemic prediction will are in fact readily distinguishable. The difference between them is perhaps sharpest in cases where the volitionality of dynamic will is reinforced by negativeness, as in (35) above and (39) below, or by its use in a closed interrogative with the illocutionary force of a request, as in (36) above and (40) below: (39) (40)
Those who won’t ask don’t get (ICE-GB W2E-003 57) If you cannot agree will you please telephone this office before sending any further demands. (ICE-GB W1B-023 121)
According to Coates (1983: 176) the dynamic uses of will that we have discussed above always occur with an animate subject. However in the present data there were examples of will used with an inanimate subject (generally with respect to situations concerned with the satisfaction of the speaker’s desires or needs), as in: (41) (42)
it won’t print because it keeps saying feed paper (ICE-GB S1A-077 6) when I go to bed at night I I have to st my arm won’t straighten you see (ICE-GB S1A-052 73)
Dynamic will may be used with reference to an habitual or typical activity that is ascribable to the disposition or insistence of the subject-referent. Huddleston and Pullum (2002: 194) label this use ‘propensity’, describing it as concerned with either “characteristic or habitual behaviour of animates” as in (43) and (44), often with an attendant suggestion of the speaker’s disapproval or resignation, or “general properties of inanimates” as in (45) and (46). Palmer (1990: 136-137) distinguishes these as two different subtypes which he calls
134
Chapter 5
respectively ‘habit’ (“concerned with habitual (or better, ‘typical’) behaviour”) and ‘power’ (“volition applied to inanimate objects”). (43) (44)
(45) (46)
There are those who will see all this activity as a sign of chaos, but it is absolutely the opposite. (ICE-AUS W2C-017 107) a person with co-dependent behaviour is one who either stays in relationships they don’t want to be in and are destructive for them because they’re too scared of getting out of them or will never allow themselves to be close enough to have a relationship because they’re too afraid of the pains and things involved with it. (ICE-AUS S1B-025 186) Remember too that heavy rain will raise creek and stream levels and may render them impassable; (ICE-AUS W2D-018 32) Hand stroking will remove any loose hairs and shredding is also minimal. (ICE-AUS W2D-019 28)
Examples of this type are to be distinguished from those involving habituality where the speaker makes an inference about the predictability of an activity based on its regular occurrence and the disposition of the subject-referent is not at-issue, as in (47) and (48). These I have classified as epistemic rather than dynamic.1 (47)
(48)
Wide departures from this value will invariably lead to malfunction which can be either (1) insidious, hidden, with the system apparently working normally or (2) catastrophic with total hardware failure with, perhaps, disk drives not starting, components breaking down and sometimes even smoke! (ICE-GB W2B-033 13) With an aperture setting of f2 the plane of sharpness will be rather shallow, in some cases only a matter of a few centimetres. (ICE-AUS W2D-016 54)
5.1.3 Deontic will Will is occasionally used deontically, with a 2nd person subject and the speaker understood to be the deontic source. Huddleston and Pullum (2002: 194) identify this use as deriving via implicature from futurity: “if I predict your agentive actions (or someone else’s) in a context where I have the authority to require them, I will be understood as tacitly invoking that authority”. Examples follow: (49) (50) (51)
I’ll withdraw that above your right eye. You’ll say he headbutted you. (ICE-AUS S1B-067 88) You’ll clear the bikes (ICE-GB S2A-054 35) And I said yeah and she said oh you’ll be quiet and I said yes so um um yeah we had to go in and print in there because it just won’t feed and it tried to do it last Tuesday (ICE-AUS S1A-021 52)
Prediction and volition
135
5.1.4 Will: regional and stylistic variation Table 5.4. Genre distribution of will Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total
ICE-AUS 3,942 (1,419) 4,763 (1,143) 4,270 (2,562) 4,240 (424) 2,940 (882) 3,265 (1,306) 3,868
ICE-GB 3,978 (1,432) 3,579 (859) 3,818 (2,291) 4,540 (454) 3,720 (1,116) 3,925 (1,570) 3,861
C-US
Avg
4,173 (486)
4,087
3,625 (290) 3,950 (776)
3,605 3,893
Will has very similar frequencies across the three corpora (see Table 5.4). On average it is marginally more common in speech than writing (by a ratio of 1.1:1, and ratios in the three corpora of 1.31:1 for ICE-AUS, 0.97:1 in ICE-GB and 1.15:1 in C-US). As Table 5.2 shows, the dominant epistemic and dynamic meanings are not evenly distributed across the corpora, the ratio of epistemic to dynamic tokens being 2.4:1 in both ICE-AUS and ICE-GB, but 1.4:1 in C-US. However, comparison of the speech vs writing frequencies for dynamic will (see Table 17, Appendix) reveals that it is equally popular in the two modes in C-US (1,589:1,525) in the other two dialects it is approximately twice as popular in speech as it is in writing (1,263:492 in ICE-AUS, and 1,298:662 in ICE-GB). This finding, viewed in conjunction with the greater popularity of want to in CUS than in ICE-AUS and ICE-GB (see Section 5.6.2), may suggest that – at least in speech – dynamic will is succumbing to the incursion of want to more rapidly in AmE than it is in the other two varieties. 5.2
Shall
Shall is traditionally associated with will, but its frequency is strikingly smaller, the ratio of tokens being 343:8,505 (or 1:24.8). While they express similar meanings the proportions are quite different (compare Tables 5.2 and 5.5): epistemic modality/futurity is the main meaning for will, but minor for shall; and deontic modality is the main meaning for shall, but minor for will. The negative form shan’t is rare, with only four tokens in the data, three of them in ICE-GB. Table 5.5. Meanings of shall Epistemic Dynamic Deontic Indeterminate Total
ICE-AUS 1 48 49 2 100
ICE-GB 14 90 116 3 223
C-US 31 (6) 41 (8) 31 (6) 0 (0) 102 (20)
Total 46 (21) 179 (146) 196 (171) 5 (5) 425 (343)
% 10.8% 42.1% 46.1% 1.2% 100%
136
Chapter 5
5.2.1 Epistemic shall Both Palmer (1990: 162-163) and Huddleston and Pullum (2002: 195) claim that shall is never found in the central epistemic use with non-future reference, and Palmer presents this claim as supporting evidence for his treatment of future will and shall as non-epistemic. However examples can be found, as in (52): (52)
Like those on the home front in earlier wars we shall often be imperfectly informed of what is happening, and this too puts our patience to the test. (ICE-GB W2E-007 22)
Furthermore shall diplays a number of the co-occurrence patterns that we have observed with epistemic will (Section 5.1.1). These include compatibility with an epistemic adjunct as in (53), use with the progressive aspect as in (54), use with a non-agentive verb as in (55), and use with a stative verb as in (56): (53) (54) (55) (56)
I shall probably look in at the College once or twice during the autumn, and hope to see you then. (ICE-GB W1B-014 69) I understand that I shall be using this under my own responsibility. (ICEGB W1B-017 115) As we shall discover, the concept of child abuse is an extremely elusive one and means different things to different people. (ICE-GB W2B-017 26) He asks God to disperse these mists, which blot and fill My perspective still as they pass Or else remove me hence unto that hill Where I shall need no glass (ICE-GB W1A-018 60)
Like epistemic will, epistemic shall may express conditional consequence, in the apodosis of a conditional construction, as in: (57)
And if she recommends my book after that I shall be very surprised (ICEGB S1B-025 127)
Contemporary speakers, the data suggest, pay little heed to the traditional prescriptive rule, that in referring to the future shall should take a 1st person subject and will a 2nd or 3rd person subject: 8.7% of epistemic will tokens in the corpora had a 1st person subject (see Table 5.3 above), while 28.6% of epistemic shalls had a non-1st person subject (see Table 5.6 below), as in:2 (58)
my heart’s desire is that my neighbour Ivan’s donkey shall die tomorrow morning (ICE-GB S2B-047 77)
Prediction and volition
137
Table 5.6. Person of subject with shall Epistemic Dynamic Deontic Indeterminate Total
1st 15 (71.4%) 146 (100%) 69 (40.4%) 4 (80%) 234 (68.2%)
2nd 1 (4.8%) 0 (0%) 0 (0%) 1 (20%) 2 (0.6%)
3rd 5 (23.8%) 0 (0%) 102 (59.6%) 0 (0%) 107 (31.2%)
Total 21 (100%) 146 (100%) 171 (100%) 5 (100%) 343 (100%)
NB Figures represent raw frequencies 5.2.2 Dynamic shall Dynamic volitionality is the second most common meaning of shall, as it is with will. Voltional shall always takes 1st person subject in the data, and typically expresses intentionality as in (59), but occasionally also willingness as in (60). (59) (60)
Everything in Turkey is very cheap - which reminds me Cath, I shall send you some Turkish money. (ICE-AUS W1B-012 17) Well I shan’t see her. (ICE-GB S1A-090 103)
5.2.3 Deontic shall Shall has several deontic uses, the most central to this type of modality being that which Huddleston and Pullum (2002: 194) describe as “constitutive/regulative”, used with a 3rd person subject, and normally found in legal documents, regulations, and the like, as in (61) and (62): (61)
(62)
Once an issue has been considered and appropriately addressed by Council, relevant papers shall no longer be considered “Council-inConfidence” material. (ICE-AUS W2D-009 304) Pursuant to subsection 128(2) of the Act, the Commission Chairperson may at any time, and shall wherever the Council does not have a Chairperson, convene a meeting of the Regional Council. (ICE-AUS W2D-009 34)
There are two further uses of shall which are appropriately identified by Huddleston and Pullum (2002: 194) and Palmer (1990: 74) as deontic. In the first of these the meaning may appear to be simply dynamic: however the speaker does not merely indicate a readiness to carry out the activity, but rather undertakes an obligation or gives a guarantee to do so (as in You shall have it tomorrow). While there were no corpus examples with a 2nd person subject, formulaic addressee-oriented examples of the type in (63) were counted as belonging to this type: (63)
We shall see (ICE-GB S1A-098 29)
138
Chapter 5
The second non-central deontic use of shall is in ‘direction questions’ as in (64) and (65), including interrogative tags as in (66), normally with a 1st person subject. Such questions are used by the speaker to consult the addressee’s wishes, prompting Coates (1983: 188) to treat this use as volitional. However the undertaking that the speaker makes to perform the activity suggests that they are more plausibly regarded as deontic. Notice that the most likely answer is not declarative but rather imperative (e.g. Yes, do). (64) (65) (66)
Shall I do something civilized, like clear the table? (C-US SBC 03 1221) Shall I write it down for you? (ICE-GB S1B-080 21) I’ll go and get it, shall I? (ICE-GB W2F-004 198)
Sometimes the sense of obligation is less salient: the speaker may seek the addressee’s advice, rather than consulting their wishes, as in (67). (67)
Yeah I’m I’ve been on the same position because I’ve been thinking you know shall I push them for fulltime or shall I push them for a pay increase you know (ICE-AUS S1A-061 99)
Like will you, shall we may be used with directive force, as in (68), where it is used in a tag question attached to a 1st person imperative: (68)
Let’s stop for the moment, shall we? (ICE-GB S1A-001 51)
A similar kind of illocutionary force is associated with the use of shall in the formula shall we say, used to introduce an expression that the speaker expects the addressee to accept, even if provisionally: (69)
He did all sorts of things and he became a shall we say suburban Australian archaeological hero. (ICE-AUS S1A-026 61)
5.2.4 Shall: regional and stylistic variation Leech (2003) and Mair and Leech (2006) found shall to have suffered a massive decrease in frequency between 1961 and 1991/2 in AmE and BrE writing (of almost identical proportions: 43.8% from Brown to Frown, and 43.7% from LOB to FLOB): see Table 1.4 above. There were some striking differences between the dialects in the frequency of shall: the dispreference for this modal was relatively less pronounced in ICE-GB (with 223 tokens per million words) than in ICEAUS (100) and C-US (102). The numbers for shall in the American corpus would have been considerably less had it not been for their high frequency in one, religious, text (SBC-020).
Prediction and volition
139
Table 5.7. Genre distribution of shall Spoken
Written
Dialogue Monologue Total Non-printed Printed Total
Total
ICE-AUS 72 (26) 17 (4) 50 (30) 230 (23) 157 (47) 175 (70) 100
ICE-GB 256 (92) 163 (39) 218 (131) 300 (30) 207 (62) 230 (92) 223
C-US
Avg
112 (13)
127
88 (7) 102 (20)
164 142
A comparison of the frequencies for shall across the corpus subcategories provides further insights into the nature of its survival in BrE and AusE, as reflected in its degree of robustness in speech. As Table 5.7 shows, shall is relatively more popular in speech in ICE-GB than it is in ICE-AUS (the speechto-writing ratio in the two corpora being 0.9:1 and 0.3:1 respectively). The skewing in C-US noted above resulting from the high number of tokens in a single text invalidates meaningful comparisons with C-US on this dimension. 5.3
Would
Like preterite could and might, would has both temporal and hypothetical uses, albeit in quite different proportions. While the overall ratio of past would to hypothetical would is 29.5%:70.5%, the proportion of past would is lower in speech (25.2%), and higher in writing (35.9%). This is not a surprising finding in view of the relative popularity of past tense forms generally in such written genres as news and fiction, as against conversation (q.v. Biber et al. 1999: 456). Table 5.8. Uses of would as a preterite Spoken
Past Hypothetical Total Written Past Hypothetical Total Indeterminate Total
ICE-AUS 915 (549) 2,990 (1,794) 3,905 (2,343) 1,025 (410) 1,728 (691) 2,753 (1101) 141 3,585
ICE-GB 787 (472) 2,863 (1,718) 3,650 (2,190) 970 (388) 1,770 (708) 2,740 (1096) 118 3,404
C-US 1,374 (160) 3,262 (380) 4,637 (540) 975 (78) 1,800 (144) 2,775 (222) 122 (24) 4,001 (786)
Avg 1,025 3,038 4,064 990 1,766 2,756 127 3,663
Apart from the absence of a deontic use, would shares with will a strong preference for epistemic over dynamic modality (see Table 5.9).
140
Chapter 5
Table 5.9. Meanings of would
Dynamic Epistemic Indeterminate Total
ICEAUS 1,210 1,903 472 3,585
ICE-GB C-US
Total
%
702 2,299 403 3,404
2,518 (2,031) 7,068 (4,765) 1,404 (979) 10,990 (7,775)
22.9% 64.3% 12.8% 100%
606 (119) 2,866 (563) 529 (104) 4,001 (786)
5.3.1 Temporal would Past time would may be dynamic, expressing volition, as in (70) and (71), and propensity, as in (72) and (73). (70)
(71) (72)
(73)
Since the cook would by no means of persuasion be induced to set foot outside the house Pritchard wandered off by herself. (ICE-GB W2F-005 70) It was really funny. She like she looks around and this woman wouldn’t help us wouldn’t serve us or anything (ICE-AUS S1A-009 97) Whenever it was put to him that India tilted towards the Soviet Union in the great ideological divide, he would quote his mother. (ICE-GB W2B011 33) Occasionally he would proffer advice, I would decline it, he would insist, and I would go off to the editor, Mr Trelford, who would almost always support me. (ICE-GB W2B-015 50)
Like past time could, volitional would is normally restricted to non-affirmative contexts with a single event. Thus it would not be possible to change the polarity of (70) (*Since the cook would by means of persuasion …), or to substitute would for wouldn’t in (71). Propensity would differs from volitional would in that it involves an habitual situation, applying to the typical behaviour of the subjectreferent. Epistemic modality is represented by the future in the past use, found mainly in narrative genres, where we understand the event to have occurred subsequent to the time referred to by would but before the time of the utterance, as in (74). (74)
But the the area at the front here uh where in later Greek temples pieces of sculpture will be placed that’s that’s nowhere near as useful a shape as it would be later on (ICE-GB S2A-024 89)
The temporal use of would far more commonly involves backshifting. In (75) and (76) backshifted would expresses respectively volition and futurity.
Prediction and volition (75)
(76)
141
One breeder said that any serious breeder would not sell a kitten like the latter, although the former condition is controllable. (ICE-AUS W2D-019 107) Bob Hawke was still Prime Minister at that stage and I knew that Hazel would be constrained to a certain extent in what she could write because of that public office that she held (ICE-AUS S2A-041 55)
5.3.2 Hypothetical would As with could and might, so with would, we may distinguish between the hypothetical and weakened tentative uses. Hypothetical would is found in the apodosis of unreal conditional constructions, as in (77), but also occasionally in the clausal complement of wish, as in (78). (77) (78)
So it would only count if I was doing a coherent major in literature (ICEAUS S1A-015 267) I wish the stores would open earlier. (C-US SBC-011 698-700)
Hypothetical would typically expresses epistemic meaning, as in (77) and (78) above, but volitional examples are also sometimes found, as in (79). (79)
If you could get them to me I would be d deeply appreciative. (C-US SBC021 76-8)
More often than not the protasis of the conditional construction is unexpressed, implicit in the context, as in (80) where it might be interpreted as something like “if I had to deal with them”, and in (81) “if you owned one”, and in the volitional example in (82) “if I didn’t have to”. (80) (81)
(82)
I mean I didn’t grow up with young children so I would be hopeless with young children (ICE-AUS S1B-015 243) Feeding Breeders feed their cats a wide range of foods from commercial cat food and biscuits to grated cheese, cottage cheese, yoghurt and lean meat, with added calcium for kittens. It would cost about $5 to $8 a week to feed Cornish Rex. (ICE-AUS W2D-019 68) This town stinks. I would never live here. It’s a dump. (ICE-AUS W1A010 233)
The tentative use of hypothetical would, like that of hypothetical could, might and should, involves such features as diffidence, tact, politeness and unassuredness. In (83) and (84) epistemic would is less assured or forthright than epistemic will be or unmodalized is would be: (83)
he’d be a bit younger than Jeff wouldn’t he (ICE-AUS S1A-005 5)
142
Chapter 5
(84)
That ’d ’d be about it I think, cos the imaging of planetaries is (ICE-GB S2A-058 120)
In (85) and (86) it is used to perform the indirect speech act of requesting (literal questions of this type about a person’s wishes being conventionally interpreted as requests). (86) is even more syntactically indirect than (86) making it even more diffident. (85)
(86)
We refer to further demands for 1989/90 issued on the 11th June and would be grateful if you refer to our letter of the 17th June and also would you please liaise with the Inspector as I do believe there is a repayment which is sufficient to cover any outstanding amounts for 1989/90. (ICEGB W1B-023 120) Would you care to elaborate (ICEAUS S1B-016 209)
Similarly, in (87) suggest is used performatively, the only difference between would suggest and suggest being the component of diffidence introduced by tentative would. (87)
and what I would suggest is that we make you an appointment to go and see one and talk it through (ICE-GB S1A-078 100)
Example (88) illustrates the formulaic phrase would/’d like to, a less abrupt and confronting expression of volition than want to insofar as it allows that it may not be possible to actualize the event expressed in the to-complement. (88)
I’d like to answer that in a s in a slightly different way (ICE-GB S1A-001 118)
The same phrase can also be used to formulate indirect speech acts, as in the polite request in (89). (89)
Would you like to say grace (ICE-AUS S1A-004 47)
Tentative would is often used to reduce the speaker’s level of confidence in the truth of the proposition, as in (90) and (91). (90)
(91)
I mean I’d ra I would imagine that that’s what people from the country would bring to cousins in the city wouldn’t you think a a a nice fat goose (ICE-GB S1B-014 106) No I wouldn’t think the market’s so slow at the moment (ICE-AUS S1A008 34)
Prediction and volition
143
5.3.3 Would: regional and stylistic variation A comparison of the frequencies for would across the three corpora (see Table 5.4) indicates that it is more robust in the American data not only in terms of its overall frequency but also in terms of its relative popularity in speech over writing. In all three corpora would is more frequent in speech, but proportionately moreso in C-US (1.7:1) than in ICE-AUS (1.4:1) and ICE-GB (1.3:1). This finding is compatible with that of Mair and Leech (2006) that would has undergone a milder decline in written AmE (6.1%) than it has in written BrE (11.0%) in recent decades (see Table 1.4 above). It may also be noted that AmE differs from the other two varieties in a further respect. As the figures in Table 5.8 above show, while the proportion of past woulds is comparable across the written components of the three corpora (ICE-AUS 37.2%, ICE-GB 35.4%, and C-US 35.1%), C-US has a higher proportion of past woulds in speech (29.6%) than the others (ICE-AUS 23.4%, ICE-GB 21.6%). Table 5.10. Genre distribution of would Spoken
Written
Dialogue Monologue Total Nonprinted Printed Total
Total 5.4
ICE-AUS 2,114 (1,761) 2,800 (672) 4,055 (2,433) 324 (324)
ICE-GB 4,672 (1,682) 2,408 (578) 3,767 (2,260) 353 (353)
2,760 (828) 2,880 (1,152) 3,585
2,637 (791) 2,860 (1,144) 3,404
C-US
Avg
4,809 (560)
4,210
2,825 (226) 4,001 (786)
2,855 3,663
Be going to
Be going to – a lexico-modal whose frequency (2,721 tokens) is just below that of the most common lexico-modal in the present data, have to (2,827) – is an idiomatic expression which derives historically from a progressive construction with a literally motional sense, but whose motional and progressive senses have been bleached in contemporary English. The grammaticalization of the idiom is reflected in the incorporation, in informal varieties, of the infinitival to into a morphological compound that is typically represented in informal writing as gonna, sometimes gunna. There are 1,689 tokens of be going to in the three corpora (as against 1,032 tokens of gonna/gunna). There is immense variation across the corpora in the magnitude of the difference between the frequencies of the full form versus the compound (1,047:9 in ICE-GB, 574:617 in ICE-AUS, and 68:406 in C-US). While such a degree of variation is presumably indicative of divergent transcription practices and cannot be taken as representative of the
144
Chapter 5
actual occurrence of the full and reduced forms in the three varieties, it is nevertheless suggestive of dialectal variation, with the compound favoured most in AmE. Semantically be going to is broadly similar to will, though favouring dynamic meanings comparatively more and epistemic meanings comparatively less (see Tables 5.2 and 5.11). Table 5.11. Meanings of be going to Epistemic Dynamic Deontic Indeterminate Total
ICE-AUS 659 446 4 82 1,191
ICE-GB 562 408 11 75 1,056
C-US 1,217 (239) 957 (188) 25 (5) 214 (42) 2,412 (474)
Total 2,438 (1,460) 1,811 (1,042) 40 (20) 371 (199) 4,659 (2,721)
% 52.3% 38.9% 0.9% 8.0% 100%
5.4.1 Epistemic be going to It is generally assumed that with be going to the predication always refers to a future event or state (Coates 1983: 198, Huddleston and Pullum 2002: 211). However there is evidence in the corpora that be going to is not in fact restricted to being merely a marker of futurity (a reflection of the modalization that it is undergoing). In (92) and (93) be going to expresses general present reference, or ‘predictability’ (with always providing harmonic reinforcement in (92) and I think in (93) compatible with the epistemic meaning). Substitution of will would not affect the meaning in either case. (92)
(93)
Even though we’ve got this wretched document we’re talking about there’s always going to be an Asterix book by the bedside or something like that (ICE-GB S1A-013 188) I think that there’s going to be incompetence in every profession (ICE-GB S1B-030 86)
There are nevertheless a number of differences between epistemic will and be going to, which require us to qualify Westney’s (1995: 189) claim that “it is impossible to point to any distinction between epistemic uses of will and be going to, and it is tempting to see the periphrastic as simply a rare, markedly colloquial alternative to the modal”. The first thing to notice is that examples such as (92) and (93) are considerably less common with be going to than they are with will. Secondly, when be is present tense be going to generally differs from will in carrying an implicature of immediacy, typically being used with situations that are on the point of occurring or are already in train. This implicature is undoubtedly derived from the “current orientation” (Palmer 1990: 144) associated with present tense forms of be: the future is referred to from the
Prediction and volition
145
standpoint of the present, thereby suggesting that there are features of the present situation that are determining future events. In (94) and (95), for example, will would not have the same implicature of immediacy that is found with be going to: in (94) the imminence of rain is indicated by current weather conditions, while in (95) the woman is already experiencing the symptoms which lead her to believe that she is about to faint. (94) (95)
No no um no it’s that it looks as if it’s gonna rain and it’s been like this for a few days you know (ICE-AUS S1A-090 106) So we sat down and started to eat and after a minute, maybe two, she said, “I feel really dreadful”, and I looked at her and she was ashen. … And she said, “I think I’m going to faint.” (ICE-AUS W2C-013 81)
Sometimes an adverb such as (temporal) just or suddenly is used to reinforce this sense of immediacy, as in (96) and (97), where will would sound extremely unnatural: (96)
(97)
And all you do is just tip it out of its pot Okay and you can see it just ready and it’s just gonna spill over the edge there for you (ICE-AUS S2A053 104) the play is going to take a sudden surprise and what they’re suddenly going to observe is going to be slightly different from what they might have first thought (ICE-GB S1B-019 120)
The current orientation of epistemic be going to (with present forms of be) is most apparent when, as in (94)–(97), the future time of the infinitival complement is close or relatively close, but is also found with situations that are not temporally near, as in (98) and (99), where the speaker’s prediction is made on the basis of current trends. (98)
(99)
We believe that the demand for hospice care is going to increase in the future and we want to know more about it to help us formulate our policy. (ICE-GB W2F-004 78) Arguably, in the next few years there is going to be a new boom period. (ICE-GB W1A-014 49)
A third difference between epistemic be going to and will is that, unlike epistemic will, be going to can occur readily in conditional clauses, as in (100)– (102). Here again the present orientation is reflected in the fact that the most likely interpretation in (100) is one where the speaker has already been performing poorly in exercises, in (101) where the distribution of students is already being practised, and in (102) where England’s opponents have already experienced difficulties in the scrum feeds.
146
Chapter 5
(100) So in the end I just thought this is ridiculous this is only the starting-point and if I’m going to make such a meal out of every exercise I’ll never complete the course (ICE-GB S1A-064 62) (101) Presumably if we’re going to distribute the students in this way according to pro pro rata according to the student uh distribution in the uh college we ought to be doing the same thing within the faculty should we so a big department has more representation than a small department (ICE-GB S1B-075 82) (102) If they’re going to be pushed around on England’s put-in to the scrum they’re going to have a very uncomfortable afternoon (ICE-GB S2A-002 40) Epistemic will could not be substituted for be going to in such cases, and in fact is very rare in conditional protases. An example is (103) below, where the prediction made is not one that involves the speaker’s subjective judgement but rather one involving an objective prediction grounded in the present (“If it is the case that you will be entitled …”). However (100)–(102) do not lend themselves to such an objective interpretation. (103) If you do keep on working you may be able to get Unemployment Benefit at pension rate when you are out of work if you will be entitled to a Retirement Pension on your own (or your late husband’s) contributions when you do retire. (ICE-GB W2D-002 37) A number of the familiar co-occurrence patterns that are associated with epistemic modality occur with be going to. These include compatibility with epistemic adjuncts as in (104), use with the progressive aspect as in (105), use with the passive voice as in (106), use with non-agentive verbs as in (107), use with stative verbs as in (108), use in the existential-there construction as in (109), and use in the it-extraposition construction as in (110): (104) Well maybe I’m going to win the football pools this weekend (ICE-GB S1A-067 336) (105) Well that was before he knew that he was going to be playing it at all these places next week (ICE-AUS S1A-016 59) (106) So so really you’re addressing a group of of practitioners who are going to be be involved in this site that you’ve been researching (ICE-AUS S1B003 268) (107) I came in from the holidays sort of wondering where I was going to get the energy to do anything at work (ICE-AUS S1A-065 55) (108) Well I made a big bloomer today because our school’s going to be a centre of creative ah arts (ICE-AUS S1A-065 2) (109) He’s pushing Doncaster all the way but Doncaster’s having none of it and certainly there’s going to be no doubt about the winner (ICE-GB S2A-012 83)
Prediction and volition
147
(110) Um it’s going to be a bit more expensive to do that than to ah basically here um than keeping the shower and well the basin of any bath around would be about you know into this position. (ICE-AUS S1B-080 313) Epistemic be going to most commonly takes a 3rd person subject (see Table 5.12 below), but less often than epistemic will (see Table 5.3 above), and, by contrast, takes a first person subject twice as often as epistemic will. A more general question that is raised by the figures in Table 5.12 is whether be going to is as ‘personal’ as will? If we base our answer on that meaning of the two modal expressions which has by far the highest percentage of 1st person subjects (the dynamic meaning), as Table 5.12 indicates dynamic be going to (for which 64.8% of tokens have a 1st person subject) is marginally less personal than dynamic will (with 72.2%). By contrast, dynamic be going to is more comfortable with a 2nd person subject (15.2%) than dynamic will (2.2%). Table 5.12. Person of subject with be going to (Figures represent raw frequencies) Epistemic Dynamic Deontic Indeterminate Total
1st 257 (17.6%) 675 (64.8%) 0 (0%) 84 (42.2%) 1,016 (37.3%)
2nd 159 (10.9%) 158 (15.2%) 6 (30.0%) 38 (19.1%) 361 (13.3%)
3rd 1,044 (71.5%) 229 (22.0%) 14 (70.0%) 57 (28.6%) 1,344 (49.4%)
Total 1,460 (100%) 1,042 (100%) 20 (100%) 199 (100%) 2,721 (100%)
5.4.2 Dynamic be going to Be going to can be volitional, but it generally conveys the weaker sense of intention rather than willingness. Thus in (111) I’m not going to post this can be appropriately paraphrased by “I don’t intend to” whereas I won’t post this in the same context would be paraphraseable by “I refuse to”. (111) I am not going to post this until I get prints of my photos to send you all, so I will update you about her then. (ICE-AUS W1B-008 36) As Huddleston and Pullum (2002: 212) note, the contrast described here is sharper in the preterite. Thus in (112) the refusal that would be expressed by would not would be less appropriate than the non-intentionality expressed by was not going to if it subsequently transpired that the speaker allowed himself to be cheated of Lillian. (112) NATURE HAD DEALT Lillian a nasty blow in making her a female, but I was not going to be cheated of her. She was a chip off the old block in
148
Chapter 5 every respect but one, and I was going to make sure that one flaw did not spoil the rest of her. (ICE-AUS W2F-012 5)
When be is preterite, the focus is on the intention that obtained previously, rather than the intended activity, and thus was/were going to do not entail actualization of the activity. This is the case in (114) for example, where there is no entailment that the anticipated purchase actually occurred (and the coordinated clause in fact strongly suggests that it probably did not). (113) Cause I was going to go and buy things for you and I but I thought no (ICE-AUS S1A-022 74) While be going to normally expresses intentionality it is not, as Coates (1983: 199-200) implies, restricted to doing so. There are corpus examples which suggest that be going to can in fact, like will (and shall), express strong willingness, especially when negative as in (114) (where an additional factor favouring a willingess interpretation is the multiple situation rather than singulary future situation), and when be is preterite as in (115) (where an additional factor is the implication conveyed by decision in the matrix clause that the speaker’s will has been consciously exercised). (114) But I can’t I’m not going to buy something inferior if I want something nice to drink (ICE-GB S1A-047 68) (115) And uhm I guess I made a decision that I wasn’t going to I wasn’t going to be hurt again (ICE-GB S1A-072 215) 5.4.3 Deontic be going to Coates’ (1983: 202) suggestion that the deontic meaning of be going to arises via pragmatic specialization (compare Huddleston and Pullum’s 2002: 194 analysis of deontic will discussed in Section 5.1.3 above) seems plausible: “the speaker makes an assertion about the addressee’s future activities which implies he (the speaker) intends to see they are carried out.” In the following examples the addressee, who may be referred to not simply by 2nd person you as in (116) but also by 1st person-inclusive we as in (117) and even a 3rd person NP as in (118), is placed under an obligation by the speaker to actualize the predicated activity. (116) You’re going to try and be bit earlier (ICE-GB S1A-099 115) (117) Now with your hand still on the front brake we’re going to put the bike on its side stand (ICE-GB S2A-054 47) (118) I gave her a look which caused her to shrivel somewhat. ‘Yes, Norah,’ I said, with an exaggerated show of patience. ‘I am sure it was just the ticket for you. But no daughter of mine is going to be offered mere emasculated fragments.’ Norah did not answer this, but went on sipping her tea and smoothing the fine hairs on her forearm. (ICE-AUS W2F-012 13)
Prediction and volition
149
In some cases we find that deontic be going to has undergone a further extension, with a third party – rather than the addressee – as the deontic target, as in (119): (119) The flowers were the perfect cover, so the press were unable to take photographs of me. (We had bills to pay; they weren’t going to get useful photographs for nothing.) I put both arms around the flowers and hid my face behind them. (ICE-AUS W2B-004 57) 5.4.4 Be going to: regional and stylistic variation The finding of the present study that be going to was more than twice as frequent in the American corpus as in the Australian or British corpora (see Table 5.13) is compatible with the diachronic findings reported by Leech (2003), and Mair and Leech (2006): see Section 1.4 above. Both studies noted an enormous increase in the frequency of be going to (51.6%) in American writing between 1961 and 1991/2, and by contrast a small decline in British writing.3 In the present study the distribution of be going to across speech and writing was skewed strongly towards the former (by a ratio of 9.9:1), as Table 5.13 shows. This finding is compatible with Leech’s (2003) discovery, in a follow-up analysis of be going to in spoken British data (based on a selection of texts from the Survey of English Usage corpus and ICE-GB), of a sharper increase in popularity there than in writing, prompting him to conclude: “The hypothesis that be going to has been increasing in frequency as a consequence of grammaticalization appears to be supported for spoken British English, though not for written British English” (Leech 2003: 232). Table 5.13. Genre distribution of be going to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 2,261 (814) 1,242 (298) 1,853 (1112) 200 (20) 197 (59) 198 (79) 1,191
ICE-GB 2,161 (778) 863 (207) 1,642 (985) 300 (30) 137 (41) 178 (71) 1,056
C-US
Avg
3,821 (445)
2,439
363 (29) (474)
246 1,220
What these findings suggest is the influence of two factors in the growth of be going to. One factor, as suggested by the findings for speech and writing, is colloquialization, with the attested upsurge of this quasi-modal in American writing being influenced by its robustness in speech. Note in this regard that, as the figures in Table 5.13 indicate, be going to is more frequent in the typically more informal manuscript genres than in the more formal printed genres. A second factor is ‘Americanization’, with AmE leading the way in the rise of be
150
Chapter 5
going to. Not only is the frequency of be going to significantly higher in C-US than in the other two corpora, as already noted, but its relative popularity in speech over writing is also stronger in C-US (10.5:1) than in ICE-AUS (9.4:1) and ICE-GB (9.2:1). 5.5
Want to
It was suggested in Section 2.3.3 above that, even though want to is not conventionally regarded as a quasi-modal, a case can be made for such a classification. As Krug (2000: 117ff) observes, the assumption by want to of morphological and semantic features associated with modal auxiliaries is indicative of modalization. According to Bolinger (1980: 297): “(t)he moment a verb is given an infinitival complement, that verb starts down the road of auxiliariness”. This observation is especially apposite when there is morphological incorporation of the infinitival to into a single compound form, as found commonly found in speech with want to, and represented orthographically in informal styles as wanna. In this respect want to (with 320/2,177, or 14.7%, of tokens so reduced) has not progressed as far as be going to (with 1,032/2,721, or 37.9%). As is the case with be going to/gonna, the three regional varieties differ greatly in the ratio of full form to compound tokens (855:3 in ICE-GB, 841:198 in ICE-AUS, and 161:119 in C-US). Again, while the magnitude of the differences may indicate that the same conventions were not applied consistently during transcription of the spoken data, it is nevertheless suggestive of dialectal variation, with the compound favoured most in AmE. Semantically, as noted by Krug (2000: 147-151) there is some evidence that want to is undergoing modalization/auxiliarization in the emergence of modal senses additional to its dominantly volitional meaning. These meanings, deontic and epistemic, whose modest frequencies suggest that their development may be relatively recent, are discussed in Sections 5.4.2 and 5.4.3 below. There are, in addition to the formal and semantic arguments for counting want to as a quasi-modal, some functional arguments. As Verplaetse (2003: 156) observes, “convincing functional correspondence of WANT TO/WANNA and both central modals and semi-auxiliaries in the modal field can be found in the negotiation of referential content in spoken discourse”. In (120) want to, be going to and will alternate within the same volitional domain, even if they differ in the shades of meaning expressed: (120) I’m actually going to be using the words so um, while I don’t really want to apologise to you for doing that, ’cause I don’t think that’s um the sensible thing to do, I’m um gonna tell you that I will be using the words to describe what I want to talk about (ICE-AUS S2A-026 22) Another functional argument concerns the tentative, deferential, use of the preterite wanted to. It has been observed by many (e.g. Coates 1983: 118, Palmer 2001: 203-204) that preterite modal forms may be used to make an utterance
Prediction and volition
151
more diffident or polite than it would be with a present modal form (compare Could I ask …? and Can I ask …?: see Sections 4.3.1.2 and 5.3.2). It has furthermore been noted (e.g. by Bybee 1995: 503-517, Krug 2000: 155, Veplaetse 2003: 157) that the same effect can be achieved with preterite wanted to (compare I wanted to ask … and I want to ask …). In (121) and (122) wanted to is used in a present context to refer to the subject-referent’s present wish. Significantly, Palmer (2001: 204) observes that this tense usage, possible with “notionally modal” wanted to, is “not found with other types of verb”. (121) Now Paul we’ll get to you in a second but I wanted to ask ah you two guys ah (ICE-AUS S1B 035 52) (122) Oh look that’s fine. Look I knew I was too inundated this week anyway but um you know because we did talk about it I wanted to you know let you know I hadn’t forgotten (ICE-AUS S1A-096 49) The final argument concerns the role of frequency evidence in the study of grammaticalization. Krug (2000: 118-141) notes that while want to was very rare in the Early Modern English period it has increased greatly in frequency in recent times. This is confirmed by Mair and Leech’s (2006: 328) finding of an 18.5% increase in written BrE and 70.9% in written AmE over the period 1961–1991 (see further Section 5.5.4 below). In a study of 1st person volition with will, be going to, and want to in the BNC, Verplaetse (1999) noted that will accounted for approximately two-thirds of the expression of volition, the remaining third being shared almost equally by want to and be going to. In the present study volitional (‘dynamic’) want to (with 2,137 tokens across the three corpora, and for all persons: see Table 5.14 below) is almost as frequent as volitional will (2,296 tokens), each almost twice as frequent as volitional be going to (1,980 tokens).4 These frequencies suggest that the status of will as the primary modal exponent of volition may be under challenge from want to. Table 5.14. Meanings of want to Dynamic Deontic Epistemic Indeterminate Total
ICE-AUS 1021 12 4 2 1,039
ICE-GB 843 8 6 1 858
C-US 1,390 (273) 15 (3) 15 (3) 5 (1) 1,425 (280)
Total 3,254 (2,137) 35 (23) 25 (13) 8 (4) 3,322 (2,177)
% 98.0% 1.1% 0.8% 0.2% 100%
Finally, the figures in Table 5.15 suggest that want to is more ‘personal’ than the other expressions in the prediction/volition group. 41.9% of want to tokens have a 1st person subject, as against 26.9% for will and 37.3% for be going to. If we take 1st person and 2nd person subjects together, the differences are magnified (want to 68.5%, will 37.1%, and be going to 50.6%).
152
Chapter 5
Table 5.15. Person of subject with want to (Figures represent raw frequencies) Dynamic Deontic Epistemic Indeterminate Total
1st 905 (42.3%) 4 (17.4%) 3 (23.1%) 1 (25.0%) 913 (41.9%)
2nd 555 (26.0%) 17 (73.9%) 7 (53.8%) 0 (0%) 579 (26.6%)
3rd 677 (31.7%) 2 (8.7%) 3 (23.1%) 3 (75.0%) 685 (31.5%)
Total 2,137 (100%) 23 (100%) 13 (100%) 4 (100%) 2,177 (100%)
5.5.1 Dynamic want to In Modern English want to is the item most consistently associated with the expression of the dynamic (more specifically, volitional) modality, with a meaning comparable to, though arguably not as strong as, that of willingness will. Thus in (123) there is no implication that the “making a game of it” will be actualized, as there would be if we were to substitute ’ll for want to. Even more pointed is the contrast between the preterite didn’t wanna and its modal auxiliary counterpart wouldn’t in (124), the latter paraphraseable by “refused to”. The original would allow a continuation such as “but I managed to”, but this would be infelicitous with the stronger wouldn’t. (123) Um I mean it depends whether you want to make a game of it or not (ICEAUS S1A-008 109) (124) you know, I didn’t wanna embarrass her (C-US SBC 10 586) 5.5.2 Deontic want to As noted above the development of a deontic use of want to provides one piece of evidence that this catenative lexical verb is undergoing auxiliarization/ modalization. It is found most commonly with a 2nd person subject, as in (125), (126), and (127). (125) A: Do you want tap water or this B: Just normal water A: It’s it’s Spa B: Solpadeine is is uh A: What You want to use the tap water then (ICE-GB S1A-043 143) (126) So you wanna think about what’s my hypothesis and how am I gonna how am I gonna attempt then test that hypothesis and what’s the best way to do it (ICE-AUS S1B-019 129) (127) Ah ere r we’re coming out of it. We’ve turned the corner and ah I believe we don’t wanna go to a dog eat dog survival of the fittest society. (ICEAUS S1B-029 52)
Prediction and volition
153
It is maintained by some (e.g. Aarts and Aarts 1995: 178) that the sense is volitional in such cases, with the speaker’s wishes being ‘projected’ onto the addressee. However a deontic interpretation is consistent with the apparent issuing of a directive or at least strong recommendation by the speaker in (125) and (126) (an interpretation supported by the didactic context in (126)), and the objective statement of an obligation in (127) (“there is an obligation on us”). Such an obligational interpretation is motivated by implicature (as we have argued is the case with deontic will in Section 5.1.3 and deontic be going to in Section 5.4.3), with want to analyzed as comparable in meaning and modal strength to deontic should and ought to. 5.5.3 Epistemic want to Is want to in the process of developing the same type of root/epistemic semantic duality that is characteristic of the modal class? There is some evidence that it is, but the epistemic meaning of want to is rather more elusive than its deontic meaning. Bolinger (1980: 297) suggests an example (Pick it up with both hands. You wanna be careful it doesn’t fall.), but this appears to be deontic rather than epistemic. Westney (1995: 32) concedes that it is “unclear” whether want to can be epistemic, but nevertheless offers a plausible (constructed) example which he describes as “perfectly possible in a colloquial context”: They want to be pretty stupid if they believe everything he says. Krug (2000: 150) presents an example overheard from a Californian female shop assistant (Customer: Do you have coolers? Assistant: Coolers? They wanna be on one of the top shelves somewhere. They only arrived this morning.). The corpora yielded a small number of tokens: in (128) and (129) the speaker is making an inference about a situation rather than describing the subject-referent’s willingness to actualize it. (128) Lorraine Mullen can hardly lift her legs in third place the thirty seven year old Kiwi. She would have been used to all these hill running. Once of New Zealand, twenty two miles of the hardest run you’d ever wanna run. (ICEAUS S2A-001 127) (129) H Well maybe th that’s the reason why they’re leav I mean hopefully if they wanted a franchise certainly but if they’re retiring one they’re retiring which ? Well they’d wanna they’d wanna be. H Being being encouraged, (ICE-AUS S1B-018 199) 5.5.4 Want to: regional and stylistic variation Two recent diachronic studies have confirmed the growing popularity of want to, particularly in AmE. In his comparison of the press and fiction categories of LOB/Brown and FLOB/Frown, Krug (2000: 135) notes spectacular increases in
154
Chapter 5
the American corpora, leading him to conclude that “while the rise of the new volitional modal probably did not originate in the US, the change obviously caught on more rapidly here than in Britain”. Similarly, Mair and Leech (2006) report a spectacular increase in the popularity of want to (of 70.9%) in American writing between 1961 and 1991/2, with a smaller though still substantial increase of 18.5% in British writing (see Table 1.4). As Table 5.16 shows, in the present study want to is more than 50% more popular in the American corpus than in the British (1.7:1), and almost 50% more popular than in the Australian corpus (1.4:1). Table 5.16. Genre distribution of want to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 1,889 (680) 808 (194) 1,457 (874) 430 (43) 407 (122) 413 (165) 1,039
ICE-GB 1,542 (555) 542 (130) 1,142 (685) 720 (72) 337 (101) 433 (173) 858
C-US
Avg
1,966 (229)
1,522
638 (51) 1,425 (280)
495 1,389
Want to displays stylistic as well as regional variation. In the present study it was preferred in speech over writing (albeit less overwhelmingly than be going to (see Section 5.4.4 above) by a ratio of 3.1:1, a finding uncannily in line with Krug’s (2000: 136) claim that it is “approximately three times more common in spoken English”. This finding supports the credibility of Krug’s view that “spoken performance data are influencing the written medium towards a greater use of this lexeme” (2000: 136); in other words, that colloquialization has played an important role in the frequency gains experienced by want to in contemporary English. The findings for person (see Table 5.15 above) indicate that there is a certain degree of complementarity between the modal and semi-modal. 2nd person subjects are considerably more common with dynamic want to (accounting for 26.0% of all tokens) than with dynamic will (2.2%), while dynamic want to selects a 1st person subject less commonly (42.3%) than does dynamic will (72.2%). Insofar as volition is more transparently and consistently expressed by want to than will, this difference may be explained in terms of speakers’ greater preparedness to make overt reference to their addressee’s volition than to their own. A further inference that may be drawn from this patterning is that the incursion that want to seems to be making into the semantic territory of dynamic will is not occurring haphazardly, but rather in a way that is setting up a certain degree of distributional complementarity.
Prediction and volition 5.6
155
Be about to
5.6.1 Meanings of be about to Be about to is a low frequency item whose semantic similarity to be going to warrants a brief entry at this point. As Table 5.17 shows, however, its epistemic (futurity) sense is dominant, no deontic meaning and a small number of dynamic tokens (see further below). Table 5.17. Meanings of be about to Epistemic Dynamic Indeterminate Total
ICE-AUS 59 3 1 63
ICE-GB 52 2 0 54
C-US 36 (7) 0 (0) 0 (0) 36 (7)
Total 147 (118) 5 (5) 1 (1) 153 (124)
% 96.1% 3.3% 0.7% 100%
In view of the semantic similarities between be about to and be going to it is not surprising that they display similar behaviour in subject selection (as Table 5.18 indicates, both strongly preferring a 3rd person subject and with 1st person preferred over 2nd person). Table 5.18. Person of subject with be about to (Figures represent raw frequencies) Epistemic Dynamic Indeterminate Total
1st 26 (22.0%) 2 (40%) 0 (0%) 28 (22.6%)
2nd 4 (3.4%) 1 (20%) 0 (0%) 5 (4.0%)
3rd 88 (74.6%) 2 (40%) 1 (100%) 91 (73.4%)
Total 118 (100%) 5 (100%) 1 (100%) 124 (100%)
Like be going to, be about to always locates the situation in future time when be is present tense. However the sense of immediacy carried by be about to is even stronger than it is with be going to. For instance the use of be about to in (130) suggests the imminence of the communication, and in (131) of the landing. (130) Just a moment I think he was about to say something else (ICE-AUS S1B065 299) (131) I think I can hear the sound of an aircraft, uh in the distance, and I think they’re about to come into land. (ICE-GB S2A-008 103) Collocation with temporal just is very common, as in (132) and (133):
156
Chapter 5
(132) We were just about to have tea and he wanted a packet of chips (ICE-AUS S1A-048 391) (133) I was just about to tuck into it and I noticed this great fly soaked in tomato sauce (ICE-GB S1A-055 222) Be about to is less grammaticalized than be going to (whose auxiliarization is in evidence phonologically in the appearance of gonna and semantically in the development of its volitional sense). Nevertheless, there are several examples which suggest that be about to may have started to undergo a similar semantic development. In (134), for example, it expresses intentionality, and in (135) the stronger volitional sense of refusal: (134) He laughed. You’re not about to let me go on that tea business are you? (ICE-AUS W2F-001 72) (135) I wanted to tell you this because we were talking about it when I saw you and I wanted you to realise that I was am not about to be taken for a ride (ICE-GB W1B-005 49) These non-epistemic senses are rare and only found with negation. In (135) there is another possible interpretation, involving a use of be (not) about to that was first commented on some 40 years ago as an innovation in AmE, and glossed as “the actor is not the sort of person from whom such a deed can be expected” (Joos 1968: 24). This development appears to be comparable to that which resulted in the ‘propensity’ use of will (see Section 5.1.2 above), involving an assessment of a person’s character deriving from their wilful behaviour. 5.6.2 Be about to: regional and stylistic variation Table 5.19 indicates that be about to is less popular in C-US than in the other two corpora, and this may be associated with its comparative unpopularity in speech, by contrast with its greater popularity in speech in both ICE-AUS and ICE-GB. Table 5.19. Genre distribution of be about to Spoken
Written
Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 47 (17) 113 (27 73 (44) 50 (5) 47 (14) 48 (19) 63
ICE-GB 56 (20) 83 (20) 67 (40) 40 (4) 33 (10) 35 (14) 54
C-US
Avg
60 (1)
67
75 (6) 36 (7)
53 51
Prediction and volition
157
Notes 1
Note, in this regard, that used to – excluded from this study on the grounds that it expresses aspectual rather than modal meaning – is used to express characteristic or habitual behaviour in the past, without any suggestion that the possibility of occurrence of the situation is attributable to properties of the subject-referent.
2
These figures show that Huddleston and Pullum’s (2002: 189) claim that shall “always has deontic meaning with 2nd/3rd person subjects” is inaccurate.
3
Leech’s (2003) percentage figure for British writing is 3.1%, while Mair and Leech’s (2006) is 1.2%.
4
In fact there are probably more tokens of volitional want to than will, given that the figure of 2,296 tokens for dynamic will includes those representing the propensity subclass.
Chapter 6 Conclusion This book has examined the meanings of the English modal auxiliaries, and a set of semantically-related quasi-modals, in three parallel corpora representing BrE, AmE, and AusE. A tripartite classification scheme was adopted, distinguishing between epistemic, deontic and dynamic modality, and the analysis enriched via the concepts of modal strength, modal degree, and subjectivity/objectivity. Also addressed was the interaction between modal expressions and negation, and between modal expressions and temporality. A number of vexing issues have also been discussed, including the questionable classification of might and should as preterite forms, and the modal status of futurity will. Patterns of regional and stylistic variation have been explored, and the findings found to be compatible with Mair and Leech’s (2006) conclusion that AmE is in the box seat of change in the rise of the quasi-modals and the decline of the modals. The findings further suggest that BrE is the most conservative of the three regional varieties, with AusE occupying a middle position, appearing to distance itself from both the innovativeness of the Americans and conservativism of the British. 1 Consider firstly the quasi-modals examined. The overall frequency for the quasi-modals examined in C-US (6,500 tokens per million words) is strikingly larger than that for ICE-AUS (4,905) and ICE-GB (4,625). Mair and Leech suggest that the rise of the quasi-modals has been particularly marked in spoken English, as we might expect given the tendency for innovations to spread rapidly in informal spoken genres before becoming more broadly established in the language. The quasi-modals’ preference for occurrence in speech over writing is overwhelmingly greater in the American corpus (with a speech/writing ratio of 9,505:2,196, or 4.33:1) than in ICE-AUS (6,511:2,514, or 2.59:1) and ICE-GB (5,806:2,883, or 2.01:1). On both measures, then, it is AmE that emerges as the most progressive, BrE as the least, with AusE in between. Consider secondly the modals, for which, in view of the attested decline of the class, a paucity of numbers in a particular variety may be interpretable as a sign of advanced change. Here the regional differences are less pronounced. BrE emerges, as the variety with the largest number of tokens (16,508 in ICE-GB), as the most conservative. AusE has the least number of tokens (15,906 in ICEAUS), but is followed closely by AmE (16,136 in C-US). Interestingly if we restrict the count to just those modals that might be regarded as being in competition with the quasi-modal ‘big players’ (have to, need to, be going to and want to), namely must should, need, will and shall, the ordering again mirrors that for the quasi-modals (C-US 5,370 > ICE-GB 5,777 > ICE-GB 5,997), with AmE the most innovative and BrE the most conservative. Speech/writing ratios reveal further symmetries. BrE has the highest proportion of modals in writing
160
Chapter 6
(16,672:16,460, or 1.01:1, in ICE-GB), AmE the least (14,279:17,439, or 0.82:1, in C-US), and AusE in-between (14,545:16,846, or 0.86:1, in ICE-AUS). Equipped with this broad picture of American innovativeness, British conservativism and Australian independence from both, let us review the findings for individual modal expressions. Consider first the expressions of necessity and obligation, beginning with the contrasting fortunes of must and have to. Given the declining numbers of must and increasing numbers of have to (noted in Section 1.4 above) it is not surprising that have to should outstrip must in the three corpora. The degree of difference (and therefore the degree to which the trend may have advanced) is considerably greater in C-US. The ordering across the varieties (C-US 3.44:1 > AusE 2.13:1 > BrE 1.84:1) reflects the same American domination as that determined by the frequency of have to tokens across the varieties (C-US 1,385 >AusE 1,311 > BrE 1,244). These findings suggest that at least one important factor driving the popularity of have to in AusE and BrE may be ‘Americanization’. Furthermore, as noted in Section 3.5.6, have to is considerably more popular in speech than in writing (by a ratio of 2.47:1), contrasting strikingly with must, which is almost twice as popular – 1.65:1 – in writing. This finding, in combination with the evidence provided by Mair and Leech (2006) that have to has been on the rise in recent British and American writing, suggests that another possible factor is ‘colloquialization’ (the drift into other genres of – and increasing acceptability of – features associated with colloquial speech). The result for AmE is particularly notable, with the frequency of have to being more than five times greater in speech, and the by-now-familiar ordering of dialects (CUS 5.33:1 > ICE-AUS 2.52:1 > ICE-GB 1.35:1). A further factor in the encroachment of have to on the semantic territory of must is suggested by the contrast between the typically subjective deontic meaning expressed by must and the typically objective kind expressed by have to. Have to may thus present itself an attractive option for speakers seeking a more ‘democratic’ and less overtly authoritative modal expression than must (see further Section 3.5.6 above). Again it is AmE that appears to be in the vanguard of change: the deontic meaning of have to is more dominant, and that of must less dominant, in C-US than in the other two corpora (see Tables 3.2 and 3.12). Have got to is semantically similar to have to, but syntactically dissimilar (compare for example to have to/*to have got to; may have to/*may have got to). Its regional and stylistic distribution is also quite different from that of have to. Have got to bucks the trend for quasi-modals to be most numerous in AmE (with C-US in fact having the least number of tokens of the three corpora). It also displays a preference for occurrence in speech over writing (12.3:1) that is more extreme than that for any other quasi-modal examined, and probably attributable to the likelihood that in speech it will attract less attention than in writing from prescriptivists concerned with censuring the use of got. While should may have sufficient numbers to compete with have to in the semantic field of deontic necessity, it differs from the quasi-modal in its typical subjectivity and weaker strength. Table 1.4 indicates that should has undergone a
Conclusion
161
decline in recent British and American writing. It seems likely that this decline has occurred in speech as well, given the smaller numbers for should in speech as against writing in the present study (see Table 3.7). Furthermore the relatively modest number of shoulds in C-US (850 per million words) suggests that AmE may be leading the way in its decline. Ought to is semantically very close to should, but its numbers are both extremely small by comparison and in serious decline. In Section 3.3 it was suggested that the main reason for the unpopularity of ought to is that, despite continuing to require a to-infinitive like have to and need to, it has failed to develop non-auxiliary syntactic properties in Standard English. Need and need to are, like should and ought to, semantically alike though strikingly different in frequency. Mair and Leech’s (2006) figures (see Table 1.4) reveal a recent massive rise for need to, complemented by a sharp decline for need (the latter presumably attributable to the syntactic inflexibility of the auxiliary, which is restricted largely to negative clauses). Once again it is AmE leading the way and BrE the most conservative, the frequencies for need to being C-US 473 > ICE-AUS 343 > ICE-GB 280, and for need C-US 15 > ICE-AUS 19 > ICE-GB 34. Again we find correlations with speech/writing patterns of distribution: it is in C-US that need to has the strongest preference for occurrence in speech (3.56:1), and need the weakest (0.0:1). Why is need to expanding so rapidly? One possible explanation is to be found in the attractive option offered by its deontic use (which has developed as an extension of its intrinsically dynamic sense) of enabling the speaker to formulate a requirement that at the same time acknowledges and endorses the subject-referent’s needs. Such a sense is not expressed by any of the other expressions of deontic necessity. There is less to say about the possibility/permission/ability group, given that the only quasi-modal that we have considered is be able to (and no diachronic figures are supplied for this item by Mair and Leech). Though semantically similar to can in its capacity to express a range of dynamic meanings, be able to differs from the modal in the implication of actuality that it conveys. The large numbers for can (surpassed only by those for will/would) along with its diachronic stability (see Table 1.4) suggest that it is not under threat from be able to. May expresses a similar range of meanings to can, but the proportions differ greatly, may being dominantly epistemic, can dynamic. The findings of the study are in line with Mair and Leech’s finding that may has suffered a stronger decline in AmE writing than BrE writing (see Table 1.4) with C-US evidencing the smallest number of tokens (C-US 825 < ICE-AUS 881 < ICE-GB 1218), along with the greatest degree of unpopularity in speech (C-US 0.27:1 < ICEAUS 0.31:1 < ICE-GB 0.32:1). Finally, consider the prediction/volition group, which is dominated numerically by will/would, but also contains the two rapidly growing quasimodals be going to and want to. There are some subtle semantic differences between be going to and will which should make us wary of any suggestion that the former is intruding upon the semantic territory of the latter. While epistemic
162
Chapter 6
will typically locates a situation in future time, epistemic be going to is restricted to doing so. Furthermore epistemic be going to conveys a sense of immediacy (albeit not as strong as that conveyed by be about to) that differentiates it from will. While dynamic will may express willingness or intentionality, dynamic be going to tends to restricted to the latter. Mair and Leech’s (2006) frequencies, presented in Table 1.4, document starkly contrasting the fortunes of be going to in AmE writing (a rise of 51.6%) and BrE writing (a small decline of 1.2%). The findings of the present study are in keeping with this diachronic picture: be going to is more than twice as frequent in the American corpus as it is in the others (CUS 2,413 > ICE-AUS 1,191 > ICE-GB 1,056), with speech/writing ratios reflecting the same ordering of the regional varieties: C-US 10.52:1 > ICE-AUS 9.35:1 > ICE-GB 9.22:1. Want to is a predominantly volitional expression, with a meaning comparable to that of dynamic will. Given its spectacular upsurge in popularity in American writing, as reported by Mair and Leech (2006), it was not surprising to find the frequency for want to in C-US outstripping that in the other corpora: CUS 1,425 > ICE-AUS 1,039 > ICE-GB 858. Meanwhile want to in AmE evidenced a strong preference for occurrence in speech over writing, albeit one not quite as strong as that in the Australian corpus: ICE-AUS 3.52:1 > C-US 3.08:1 > ICE-GB 2.63:1. Shall is traditionally associated with will, even though its primary use in Modern English is as a deontic modal found mainly in constitutive/regulative contexts. Its frequency is strikingly smaller than that of will, and rapidly declining. In the present data shall was found to be considerably more robust in the British corpus (where it is relatively popular in speech) than in ICE-AUS and C-US.
Notes 1
I am grateful to Christian Mair for pointing out that a potentially fruitful avenue for further research would be the diachronic implications of the pragmatic/politeness dimensions of modality (as reflected in cases such as I would think so and This would seem to suggest). It may be, for instance, that instead of the picture of British conservatism that emerges from the present essentially quantitative study, we find speakers of BrE to be more prolific users of modal hedges than speakers of AmE and AusE. In turn such a finding might invite the conclusion that instead of having three Englishes at different diachronic-developmental stages, what we have is a fairly stable system of choices in which the available options are used differently across the varieties for essentially pragmatic reasons.
Appendix NOTE: In all tables raw frequencies are bracketed, and frequencies normalized to tokens per one million words are unbracketed. Table 1. Must in ICE-AUS, ICE-GB, and C-US ICE-AUS Dialogue 164 (59) Spoken Monologue 333 (80) Total 232 (139) Deontic Non-printed 760 (76) Written Printed 513 (154) Total 575 (230) Total 369 (369) Dialogue 17 (6) Spoken Monologue 38 (9) Total 25 (15) Dynamic Non-printed 80 (8) Written Printed 57 (17) Total 63 (25) Total 40 (40) Dialogue 317 (114) Spoken Monologue 88 (21) Total 225 (135) Epistemic Non-printed 120 (12) Written Printed 127 (38) Total 125 (50) Total 185 (185) Indeterminate 19 (19) Total 613 (613)
ICE-GB C-US 214 (77) 283 (68) 242 (145) 112 (13) 340 (34) 707 (212) 615 (246) 350 (28) 391 (391) 209 (41) 17 (6) 21 (5) 18 (11) 9 (1) 50 (5) 83 (25) 75 (30) 50 (4) 41 (41) 25 (5) 297 (107) 167 (40) 245 (147) 189 (22) 150 (15) 180 (54) 173 (69) 100 (8) 216 (216) 153 (30) 27 (27) 15 (3) 675 (675) 402 (79)
Avg
195
513 323
17
63 35
220
133 185 20 563
164
Appendix
Table 2. Should in ICE-AUS, ICE-GB, and C-US. Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Epistemic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Subjunctive Non-printed Written Printed Total Total Dialogue Spoken Monologue Total = would Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS ICE-GB C-US Avg 911 (328) 658 (237) 521 (125) 633 (152) 755 (453) 748 (389) 532 (62) 678 790 (79) 550 (55) 980 (294) 923 (277) 933 (373) 830 (332) 700 (56) 918 826 (826) 721 (721) 601 (118) 821 158 (57) 142 (51) 125 (30) 67 (16) 145 (87) 112 (67) 103 (12) 120 190 (19) 200 (20) 93 (28) 83 (25) 118 (47) 113 (45) 150 (12) 127 134 (134) 112 (112) 122 (24) 123 25 (9) 67 (24) 83 (20) 79 (19) 48 (29) 72 (43) 34 (4) 51 190 (19) 200 (20) 43 (13) 137 (41) 80 (32) 153 (61) 50 (4) 61 (61) 104 (104) 41 (8) 69 19 (7) 106 (38) 8 (2) 38 (9) 15 (9) 78 (47) 9 (1) 34 0 (0) 40 (4) 7 (2) 20 (6) 5 (2) 25 (10) 25 (2) 18 11 (11) 57 (57) 15 (3) 28 109 130 71 (14) 103 1,141 (1,141) 1,124 (1,124) 850 (167) 1,038
Appendix
165
Table 3. Ought to in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epistemic Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 6 (2) 38 (9) 18 (11) 40 (4) 60 (18) 55 (22) 33 (33) 0 (0) 0 (0) 0 (0) 0 (0) 3 (1) 3 (1) 1 (1) 2 (2) 36 (36)
ICE-GB 133 (48) 54 (13) 102 (61) 70 (7) 20 (6) 33 (13) 74 (74) 0 (0) 13 (3) 5 (3) 0 (0) 3 (1) 3 (1) 4 (4) 2 (2) 80 (80)
C-US
Avg
69 (8)
63
25 (2) 38 51 (10) 53
0 (0)
2
0 (0) 0 (0) 0 (0) 51 (10)
2 2 1 56
166
Appendix
Table 4. Need in ICE-AUS, ICE-GB, and C-US
Spoken Dynamic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epistemic Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 0 (0) 17 (4) 7 (4) 10 (1) 23 (7) 20 (8) 12 (12) 0 (0) 0 (0) 0 (0) 10 (1) 3 (1) 5 (2) 2 (2) 0 (0) 8 (2) 3 (2) 0 (0) 7 (2) 5 (2) 4 (4) 1 (1) 19 (19)
ICE-GB 6 (2) 42 (10) 20 (12) 20 (2) 23 (7) 23 (9) 21 (21) 3 (1) 4 (1) 3 (2) 0 (0) 3 (1) 3 (1) 3 (3) 0 (0) 21 (5) 8 (5) 0 (0) 10 (3) 8 (3) 8 (8) 2 (2) 34 (34)
C-US
Avg
0 (0)
9
25 (2) 10 (2)
23 14
0 (0)
1
13 (1) 5 (1)
7 3
0 (0)
4
0 (0) 0 (0) 0 (0) 15 (3)
4 4 1 23
Appendix
167
Table 5. Have to in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Dynamic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Epistemic Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS ICE-GB C-US Avg 1,389 (500) 1,117 (402) 1,321 (317) 946 (227) 1,362 (817) 1,048 (629) 1,649 (192) 1,353 610 (61) 800 (80) 400 (120) 643 (193) 453 (181) 683 (273) 300 (24) 479 998 (998) 902 (902) 1,099 (216) 1,000 414 (149) 442 (159) 242 (58) 163 (39) 345 (207) 330 (198) 369 (43) 348 280 (28) 440 (44) 180 (60) 300 (90) 220 (88) 335 (134) 88 (7) 214 295 (295) 332 (332) 255 (50) 294 6 (2) 0 (0) 13 (3) 0 (0) 8 (5) 0 (0) 34 (4) 14 0 (0) 3 (1) 0 (0) 3 (1) 0 (0) 6 (2) 0 (0) 2 5 (5) 2 (2) 20 (4) 9 13 (13) 8 (8) 10 (2) 10 1,311 (1,311) 1,244 (1,244) 1,385 (272) 1,313
168
Appendix
Table 6. Have got to in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Dynamic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epistemic Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 636 (229) 167 (40) 448 (269) 10 (1) 27 (8) 23 (9) 278 (278) 81 (29) 38 (9) 63 (38) 0 (0) 13 (4) 10 (4) 42 (42) 11 (4) 13 (3) 12 (7) 0 (0) 3 (1) 3 (1) 8 (8) 4 (4) 332 (332)
ICE-GB 672 (242) 113 (27) 448 (269) 50 (5) 17 (5) 25 (10) 279 (279) 94 (34) 50 (12) 77 (46) 30 (3) 7 (2) 13 (5) 51 (51) 6 (2) 0 (0) 3 (2) 0 (0) 0 (0) 0 (0) 2 (2) 7 (7) 339 (339)
C-US
Avg
232 (27) 382
38 (3) 29 153 (30) 240
26 (3)
166
0 (0) 15 (3)
36
9 (1)
8
0 (0) 5 (1) 0 (0) 173 (34)
1 5 4 281
Appendix
169
Table 7. Need to in ICE-AUS, ICE-GB, and C-US Dynamic Spoken
Written Total Deontic Spoken
Written Total Epistemic Spoken
Written Total Indeterminate Total
ICE-AUS Dialogue 178 (64) Monologue 271 (65) Total 215 (129) Non-printed 220 (22) Printed 230 (66) Total 230 (88) 220 (217) Dialogue 94 (34) Monologue 108 (26) Total 100 (60) Non-printed 90 (9) Printed 100 (30) Total 98 (39) 99 (99) Dialogue 3 (1) Monologue 8 (2) Total 5 (3) Non-printed 0 (0) Printed 10 (3) Total 8 (3) 6 (6) 21 (21) 343 (343)
ICE-GB 214 (77) 75 (18) 158 (95) 140 (14) 177 (49) 170 (63) 527 (158) 122 (44) 96 (23) 112 (67) 70 (7) 73 (22) 73 (29) 96 (96) 14 (5) 8 (2) 12 (7) 10 (1) 10 (3) 10 (4) 11 (11) 15 (15) 280 (280)
C-US
Avg
429 (50)
267
125 (10) 305 (60)
175 351
189 (22)
134
25 (2) 122 (24)
65 106
0 (0)
6
25 (2) 10 (2) 36 (7) 473 (93)
14 9 24 365
170
Appendix
Table 8. Had better in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 89 (32) 29 (7) 65 (39) 10 (1) 27 (8) 23 (9) 48 (48)
ICE-GB 53 (19) 4 (1) 33 (20) 100 (10) 10 (3) 33 (13) 33 (33)
C-US
Avg
43 (5)
47
38 (3) 41 (8)
31 40
Appendix
171
Table 9. Be supposed to in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epist Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 56 (20) 8 (2) 37 (22) 20 (2) 13 (4) 15 (6) 28 (28) 36 (13) 8 (2) 25 (15) 10 (1) 7 (2) 8 (3) 18 (18) 1 (1) 36 (47)
ICE-GB 119 (43) 13 (3) 77 (46) 10 (1) 10 (3) 10 (4) 50 (50) 75 (27) 25 (6) 55 (33) 40 (4) 30 (9) 33 (13) 46 (46) 3 (3) 80 (99)
C-US
Avg
94 (11)
69
25 (2) 66 (13)
17 48
43 (5)
41
75 (6) 56 (11) 5 (1) 51 (10)
39 40 3 56
172
Appendix
Table 10. Be to in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Epistemic Non-printed (Temporal) Written Printed Total Total Dialogue Spoken Monologue Total Dynamic Non-printed (Possibility) Written Printed Total Total Dialogue Spoken Monologue Total Conditional Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS 8 (3) 0 (0) 5 (3) 70 (7) 120 (36) 108 (43) 46 (46) 3 (1) 29 (7) 13 (8) 120 (12) 63 (19) 78 (31) 39 (39) 0 (0) 4 (1) 2 (1) 10 (1) 10 (3) 10 (4) 5 (5) 22 (8) 50 (12) 33 (20) 50 (5) 50 (15) 50 (20) 40 (40) 5 135 (135)
ICE-GB 14 (5) 21 (5) 17 (10) 10 (1) 50 (15) 40 (16) 26 (26) 11 (4) 108 (26) 50 (30) 100 (10) 153 (46) 140 (56) 86 (86) 8 (3) 25 (6) 15 (9) 20 (2) 23 (7) 23 (9) 18 (18) 61 (22) 100 (24) 77 (46) 50 (5) 93 (28) 83 (33) 79 (79) 12 221 (221)
C-US
Avg
26 (3)
16
38 (3) 31 (6)
62 34
0 (0)
21
38 (3) 15 (3)
85 47
0 (0)
6
13 (1) 5 (1)
15 9
26 (3)
45
13 (1) 20 (4) 5 (1) 76 (15)
49 46 144
Appendix
173
Table 11. Be bound to in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epistemic Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 0 (0) 8 (2) 3 (2) 0 (0) 0 (0) 0 (0) 2 (2) 6 (2) 4 (1) 6 (3) 10 (1) 10 (3) 10 (4) 7 (7) 0 (0) 9 (9)
ICE-GB 6 (2) 0 (0) 3 (2) 10 (1) 3 (1) 5 (2) 4 (4) 11 (4) 13 (3) 12 (7) 0 (0) 17 (5) 13 (5) 12 (12) 1 (1) 17 (17)
C-US
Avg
0 (0)
2
0 (0) 0 (0)
2 2
0 (0)
6
13 (1) 5 (1) 0 (0) 5 (1)
12 8 0 10
174
Appendix
Table 12. May in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Epistemic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Dynamic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS ICE-GB C-US 283 (102) 508 (183) 542 (130) 758 (182) 387 (232) 608 (365) 275 (32) 1,380 (138) 1,160 (116) 937 (281) 1807 (542) 1,048 (419) 1,355 (658) 1,163 (93) 651 (651) 1,023 (1,023) 636 (125) 39 (14) 0 (0) 50 (12) 25 (6) 43 (26) 10 (6) 17 (2) 180 (18) 50 (5) 190 (57) 163 (49) 188 (75) 180 (54) 163 (13) 76 (15) 101 (101) 60 (60) 8 (3) 3 (1) 33 (8) 25 (6) 18 (11) 12 (7) 77 (9) 60 (6) 100 (10) 203 (61) 177 (53) 168 (67) 158 (63) 25 (2) 70 (70) 56 (11) 78 (78) 51 (51) 65 (65) 56 (11) 881 (881) 1,218 (1,218) 825 (162)
Avg
423
1189 770
23
177 79
36
117 68 57 975
Appendix
175
Table 13. Can in ICE-AUS, ICE-GB, and C-US ICE-AUS Dialogue 3,025 (1,089) Spoken Monologue 3,196 (767) Total 3,093 (1,856) Dynamic Non-printed 2,080 (208) Written Printed 2,217 (665) Total 2,183 (873) Total 2,729 (2,729) Dialogue 578 (208) Spoken Monologue 238 (57) Total 442 (265) Deontic Non-printed 220 (22) Written Printed 220 (66) Total 220 (88) Total 353 (353) Dialogue 39 (14) Spoken Monologue 25 (6) Total 33 (20) Epistemic Non-printed 0 (0) Written Printed 13 (4) Total 10 (4) Total 24 (24) Indeterminate 272 (272) Total 3,378 (3,378)
ICE-GB 3,503 (1,261) 2,629 (631) 3,153 (1,892) 3,300 (330) 2,437 (731) 2,653 (1,061) 2,953 (2,953) 422 (152) 113 (27) 298 (179) 130 (13) 290 (87) 250 (100) 279 (279) 47 (17) 38 (9) 43 (26) 20 (2) 13 (4) 15 (6) 32 (32) 301 (301) 3,565 (3,565)
C-US
Avg
3,460 (403) 3,235
2,100 (168) 2,312 2,906 (571) 2,863
653 (76)
464
75 (6) 417 (82)
190 350
86 (10)
54
13 (1) 13 56 (11) 37 285 (56) 286 3,665 (720) 3,536
176
Appendix
Table 14. Might in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Dynamic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Epistemic Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS 86 (31) 225 (54) 142 (85) 50 (5) 107 (32) 93 (37) 122 (122) 8 (3) 4 (1) 7 (4) 0 (0) 10 (3) 8 (3) 7 (7) 817 (294) 483 (116) 683 (410) 190 (19) 350 (105) 310 (124) 534 (534) 32 (32) 695 (695)
ICE-GB 111 (40) 113 (27) 112 (67) 140 (14) 130 (39) 133 (53) 120 (120) 6 (2) 0 (0) 3 (2) 10 (1) 7 (2) 8 (3) 5 (5) 703 (253) 379 (91) 573 (344) 350 (35) 517 (155) 475 (190) 534 (534) 43 (43) 702 (702)
C-US
Avg
69 (8)
108
88 (7) 76 (15)
105 106
0 (0)
3
0 (0) 0 (0)
5 4
404 (47)
553
425 (34) 412 (81) 31 (6) 520 (102)
403 493 35 639
Appendix
177
Table 15. Could in ICE-AUS, ICE-GB, and C-US
Spoken Dynamic Written Total Spoken Deontic Written Total Spoken Epistemic Written Total Indeterminate Total
ICE-AUS ICE-GB C-US Avg Dialogue 1,461 (526) 1,531 (551) Monologue 954 (229) 900 (216) Total 1,258 (755) 1,278 (767) 1,443 (168) 1,326 Non-printed 1,170 (117) 1,200 (120) Printed 1,083 (325) 1,110 (333) Total 1,105 (442) 1,510 (453) 1,413 (113) 1,343 1,197 (1197) 1,220 (1,220) 1,430 (281) 1,282 Dialogue 92 (33) 61 (22) Monologue 50 (12) 33 (8) Total 75 (45) 50 (30) 60 (7) 62 Non-printed 70 (7) 30 (3) Printed 30 (9) 23 (7) Total 40 (16) 25 (10) 13 (1) 26 40 (40) 41 (8) 47 61 (61) Dialogue 203 (73) 231 (83) Monologue 250 (60) 346 (83) Total 222 (133) 277 (166) 172 (20) 224 Non-printed 220 (22) 60 (6) Printed 267 (80) 360 (108) Total 255 (102) 285 (114) 213 (17) 251 280 (280) 188 (37) 234 235 (235) 97 (97) 82 (82) 97 (19) 92 1,590 (1,590) 1,622 (1,622) 1,756 (345) 1,656
178
Appendix
Table 16. Be able to in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Ability Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Theoretical Non-printed possibility Written Printed Total Total Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS 128 (46) 125 (30) 135 (76) 160 (16) 173 (47) 170 (63) 149 (139) 181 (63) 229 (52) 200 (115) 420 (39) 193 (55) 250 (94) 220 (209) 8 (3) 8 (2) 8 (5) 70 (7) 20 (6) 33 (13) 18 (18) 21 (21) 387 (387)
ICE-GB 161 (58) 238 (57) 200 (115) 300 (29) 213 (60) 235 (89) 214 (204) 192 (64) 125 (30) 165 (94) 500 (45) 160 (43) 245 (88) 197 (182) 17 (6) 13 (3) 15 (9) 0 (0) 47 (14) 47 (14) 23 (23) 25 (25) 434 (434)
C-US
Avg
189 (21)
175
125 (10) 163 (31)
177 175
206 (23)
190
125 (9) 173 (32)
207 197
17 (2)
13
0 (0) 10 (2) 15 (3) 346 (68)
17 17 389
Appendix
179
Table 17. Will in ICE-AUS, ICE-GB, and C-US ICE-AUS ICE-GB C-US Dialogue 1,350 (486) 1,519 (547) Spoken Monologue 1,133 (272) 967 (232) Total 1,263 (758) 1,298 (779) 1,589 (185) Dynamic Non-printed 850 (85) 1,710 (171) Written Printed 373 (112) 313 (94) Total 493 (197) 663 (265) 1,525 (122) 1,044 (1,044) 1,563 (307) Total 955 (955) Dialogue 92 (33) 111 (40) Spoken Monologue 25 (6) 63 (15) Total 65 (39) 92 (55) 77 (9) Deontic Non-printed 120 (12) 210 (21) Written Printed 57 (17) 27 (8) Total 73 (29) 73 (29) 13 (1) 84 (84) 51 (10) Total 68 (68) Dialogue 2,278 (820) 2,125 (765) Spoken Monologue 3,333 (800) 2,342 (562) Total 2,700 (1,620) 2,212 (1,327) 2,267 (264) Epistemic Non-printed 3,140 (314) 2,350 (235) Written Printed 2,427 (728) 3,203 (961) Total 2,605 (1,042) 2,990 (1,196) 1,925 (154) Total 2,662 (2,662) 2,523 (2,523) 2,128 (418) Indeterminate 183 (183) 210 (210) 209 (41) Total 3,868 (3,868) 3,861 (3,861) 3,950 (776)
Avg
1,383
894 1,187
78
53 68
2,393
2,507 2,438 201 3,893
180
Appendix
Table 18. Shall in ICE-AUS, ICE-GB, and C-US
Spoken Deontic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Dynamic Written
Dialogue Monologue Total Non-printed Printed Total
Total Spoken Epistemic Written Total Indeterminate Total
Dialogue Monologue Total Non-printed Printed Total
ICE-AUS 25 (9) 4 (1) 17 (10) 120 (12) 90 (27) 98 (39) 49 (49) 44 (16) 13 (3) 32 (19) 110 (11) 60 (18) 73 (29) 48 (48) 0 (0) 0 (0) 0 (0) 0 (0) 3 (1) 3 (1) 1 (1) 2 (2) 100 (100)
ICE-GB 139 (50) 75(18) 113 (68) 160 (16) 107 (32) 120 (48) 116 (116) 108 (39) 79 (19) 97 (58) 100 (10) 73 (22) 80 (32) 90 (90) 3 (1) 8 (2) 5 (3) 40 (4) 23 (7) 28 (11) 14 (14) 3 (3) 223 (223)
C-US
Avg
26 (3)
52
38 (3) 31 (6)
85 65
34 (4)
54
50 (4) 41 (8)
68 60
52 (6)
19
0 (0) 10 31 (6) 15 0 (0) 102 (20) 142
Appendix
181
Table 19. Would in ICE-AUS, ICE-GB, and C-US ICE-AUS Dialogue 1,314 (473) Spoken Monologue 1,179 (283) Total 1,260 (756) Dynamic Non-printed 1,470 (147) Written Printed 1,023 (307) Total 1,135 (454) Total 1,210 (1,210) Dialogue 3,000 (1,080) Spoken Monologue 1,554 (373) Total 2,422 (1,453) Epistemic Non-printed 160 (160) Written Printed 967 (290) Total 1,125 (450) Total 1,903 (1,903) Indeterminate 472 (472) Total 3,585 (3,585)
ICE-GB C-US 1,108 (399) 517 (124) 872 (523) 618 (72) 500 (50) 430 (129) 448 (179) 488 (47) 702 (702) 606 (119) 2,817 (1,014) 2,388 (573) 2,645 (1,587) 3,761 (438) 2,610 (261) 1,503 (451) 1,780 (712) 1,562 (125) 2,299 (2,299) 2,866 (563) 403 (403) 529 (104) 3,404 (3,404) 4,001 (786)
Avg
917
690 839
2,943
1,489 2,356 468 3,663
182
Appendix
Table 20. Be going to in ICE-AUS, ICE-GB, and C-US ICE-AUS ICE-GB C-US Avg Dialogue 928 (334) 872 (314) Spoken Monologue 338 (81) 317 (76) Total 692 (415) 650 (390) 1,537 (179) 960 Dynamic Non-printed 90 (9) 90 (9) Written Printed 73 (22) 30 (9) Total 78 (31) 45 (18) 113 (9) 79 408 (408) 957 (188) 604 Total 446 (446) Dialogue 6 (2) 17 (6) Spoken Monologue 0 (0) 17 (4) Total 3 (2) 17 (10) 34 (4) 18 Deontic Non-printed 0 (0) 0 (0) Written Printed 7 (2) 3 (1) Total 5 (2) 3 (1) 13 (1) 7 11 (11) 25 (5) 13 Total 4 (4) Dialogue 1,189 (428) 1,139 (410) Spoken Monologue 800 (192) 442 (106) Total 1,033 (620) 860 (516) 1,898 (221) 1,264 Epistemic Non-printed 90 (9) 180 (18) Written Printed 100 (30) 93 (28) Total 98 (39) 115 (46) 225 (18) 146 562 (562) 1,217 (239) 813 Total 659 (659) Indeterminate 82 (82) 75 (75) 214 (42) 124 Total 1,191 (1,191) 1,056 (1,056) 2,413 (474) 1,553
Appendix
183
Table 21. Want to in ICE-AUS, ICE-GB, and C-US Dialogue Spoken Monologue Total Dynamic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Deontic Non-printed Written Printed Total Total Dialogue Spoken Monologue Total Epistemic Non-printed Written Printed Total Total Indeterminate Total
ICE-AUS 1,856 (668) 792 (190) 1,430 (858) 430 (43) 500 (120) 408 (163) 1,021 (1,021) 22 (8) 8 (2) 17 (10) 0 (0) 3 (2) 5 (2) 12 (12) 6 (2) 8 (2) 6 (4) 0 (0) 0 (0) 0 (0) 4 (4) 2 (2) 1,039 (1,039)
ICE-GB 1,511 (544) 533 (128) 1,120 (672) 700 (70) 337 (101) 428 (171) 843 (843) 19 (7) 4 (1) 13 (8) 0 (0) 0 (0) 0 (0) 8 (8) 8 (3) 4 (1) 6 (4) 20 (2) 0 (0) 5 (2) 6 (6) 1 (1) 858 (858)
C-US
Avg
1,923 (224) 1,491
613 (49) 483 1,390 (273) 1,085
26 (3)
19
0 (0) 15 (3)
2 12
9 (1)
7
25 (2) 10 15 (3) 8 5 (1) 3 1,425 (280) 1,107
184
Appendix
Table 22. Be about to in ICE-AUS, ICE-GB, and C-US
Spoken Epistemic Written Total Spoken Epistemic Written Total Indeterminate Total
ICE-AUS 47 (17) 104 (25) 70 (42) 50 (5) 40 (12) 43 (17) 59 (59) Dialogue 0 (0) Monologue 4 (1) Total 2 (1) Non-printed 0 (0) Printed 7 (2) Total 5 (2) 3 (3) 1 (1) 63 (63)
Dialogue Monologue Total Non-printed Printed Total
ICE-GB 56 (20) 79 (19) 65 (39) 30 (3) 33 (10) 33 (13) 52 (52) 0 (0) 4 (1) 2 (1) 10 (1) 0 (0) 3 (1) 2 (2) 0 (0) 54 (54)
C-US
Avg
9 (1)
48
75 (6) 36 (7)
50 49
0 (0)
1
0 (0) 0 (0) 0 (0) 36 (7)
3 2 0 153
References
Aarts, F. and H. Wekker (1987), A contrastive grammar of English and Dutch. Leiden: Nijhoff. Aijmer, K. (2004), ‘The semantic path from modality to aspect: Be able to in a cross-linguistic perspective’, in: H. Lindquist and C. Mair (eds), Corpus approaches to grammaticalization in English. Amsterdam: John Benjamins. 57-78. Alexander, L. G. (1988), Longman English grammar. London: Longman. Berglund, Y. (1997), ‘Future in Present-day English: corpus-based evidence on the rivalry of expressions’, ICAME Journal, 21: 7-19. Berglund, Y. (2000), ‘Gonna and going to in the spoken component of the British National Corpus’, in: C. Mair and M. Hundt (eds), Corpus linguistics and linguistic theory. Amsterdam: Rodopi. 35-49. Biber, D., S. Conrad and R. Reppen (1998), Corpus linguistics: investigating language structure and use. Cambridge: Cambridge University Press. Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999), Longman grammar of spoken and written English. London: Longman. Bolinger, D. (1980), ‘WANNA and the gradience of auxiliaries’, in: G. Brettschneider and C. Lehmann (eds), Wege zur Universalienforschung: Sprachwissenschaftliche Beitrage zum 60. Geburtstag von Hansjakob Seiler. Tübingen: Gunter Narr. 292-299. Bolinger, D. (1989), ‘Extrinsic possibility and intrinsic potentiality: 7 on MAY and CAN+1’, Journal of Pragmatics, 13: 1-23. Bouma, L. (1975), ‘On contrasting the semantics of the modal auxiliaries of German and English’, Lingua, 37: 313-339. Bybee, J. and W. Pagliuca (1985), ‘Cross-linguistic comparison and the development of grammatical meaning’, in: J. Fisiak (ed.), Historical semantics. Historical word-formation. Berlin: Mouton. 59-83. Bybee, J., W. Pagliuca and R. Perkins (1994), The evolution of grammar: tense, aspect, and modality in the languages of the world. Chicago: Chicago University Press. Bybee, J. L. (1995), ‘The semantic development of past tense modals in English’, in: J. L. Bybee and S. Fleischman (eds), Modality in grammar and discourse. Amsterdam: John Benjamins. 503-517. Bybee, J. L. and S. Fleischman (1995), ‘Modality in grammar and discourse: an introductory essay’, in: J. L. Bybee and S. Fleischman (eds), Modality in grammar and discourse. Amsterdam: John Benjamins. 1-14. Chapin, P. G. (1973), ‘Quasi-modals’, Journal of Linguistics, 9: 1-9. Coates, J. (1983), The semantics of the modal auxiliaries. London/Canberra: Croom Helm.
186
References
Coates, J. (1995), ‘The expression of root and epistemic possibility in English’, in: J. L. Bybee and S. Fleischman (eds), Modality in grammar and discourse. Amsterdam: John Benjamins. 55-66. Coates, J. (1995), ‘Root and epistemic possibility in English’, in: B. Aarts and C. Meyer (eds), The verb in contemporary English: theory and description. Cambridge: Cambridge University Press. 145-156. Collins, P. (1988), ‘The semantics of some modals in contemporary Australian English’, Australian Journal of Linguistics, 8: 233-258. Collins, P. (1991a), ‘Will and shall in Australian English’, in: S. Johansson and A-B. Stenström (eds), English computer corpora: selected papers and research guide. Berlin: Mouton de Gruyter. 181-199. Collins, P. (1991b), ‘The modals of obligation and necessity in Australian English’, in: K. Aijmer and B. Altenberg (eds), English corpus linguistics: studies in honour of Jan Svartvik. London: Longman. 145-165. Collins, P. (2005), ‘The modals and quasi-modals of obligation and necessity in Australian English and other Englishes’, English World-Wide, 26 (3): 249273. Collins, P. (2007a), ‘Modality across World Englishes: the modals and semimodals of prediction and volition’, in: C. Butler, R. Hidalgo Downing and J. Lavid (eds), Functional perspectives on grammar and discourse: in honour of Angela Downing. Amsterdam: Benjamins. 447-468. Collins, P. (2007b), ‘Can/could and may/might in British, American and Australian English: a corpus-based account’, World Englishes, 26(4): 474491. Davidsen-Nielsen, N. (1986), ‘Modal verbs in English and Danish’, in: D. Kastovsky and A. Szwedek (eds), Linguistics across historical and geographical boundaries: in honour of Jacek Fisiak on the occasion of his 50th birthday. Vol.1: Linguistic theory and historical linguistics. Berlin: Mouton de Gruyter. 1183-1194. Declerck, R. (1991), Tense in English: its structure and use in discourse. London: Routledge. Depraetere, I. and S. Reed (2006), ‘Mood and modality in English’, in: B. Aarts and A. McMahon (eds), The handbook of English Linguistics. Oxford, Blackwell. 269-290. Dirven, R. (1981), ‘Pragmatic forces associated with can- and may-sentences’, Folia Linguistica, 13: 145-215. Duffley, P., S. Clarke and W. Hirtle (1981), ‘May, can, and the expression of permission’, Canadian Journal of Linguistics, 26: 161-202. Edmonson, W., J. House, G. Kasper and J. McKeown (1977), A pedagogic grammar of the English verb. Tübingen: Narr. Ehrman, M. (1966), The meanings of the modals in Present-Day American English. The Hague: Mouton. Facchinetti, R. (2000), ‘Be able to in Present-Day British English’, in: C. Mair and M. Hundt (eds) Corpus linguistics and linguistic theory. Amsterdam: Rodopi. 117-130.
References
187
Facchinetti, R. (2002), ‘Can and could in contemporary British English: a study of the ICE-GB corpus’, in: P. Peters, P. Collins and A. Smith (eds), New frontiers of corpus research. Amsterdam: Rodopi. 229-246. Facchinetti, R. (2003), ‘Pragmatic and sociological constraints on the functions of may in contemporary British English’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 301-327. Fairclough, N. (1992), Discourse and social change. Cambridge: Polity Press. Gailor, D. (1983), ‘Reflections on should, ought to, and must’, English Language Teaching Journal, 37: 346-349. Gotti, M. (2003), ‘Shall and will in contemporary English: a comparison with past uses’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 267-300. Gresset, S. (2003), ‘Towards a contextual micro-analysis of the non-equivalence of might and could’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 81-99. Groefsema, M. (1995), ‘Can, may, must and should: a Relevance Theoretic account’, Journal of Linguistics, 31: 53-79. Haegeman, L. (1983), The semantics of ‘will’ in present-day British English: a unified account. Brussels: Paleis der Academiën. Hakutani Y. and C. H. Hargis (1972), ‘The syntax of modal constructions in English’. Lingua, 30: 301-332. Harris, M. (1986), ‘English ought (to)’, in: D. Kastovsky and A. Szwedek (eds), Linguistics across historical and geographical boundaries: in honour of Jacek Fisiak on the occasion of his 50th birthday. Vol.1: Linguistic theory and historical linguistics. Berlin: Mouton de Grutyer. 347-358. He, A. W. and S. Tsoneva (1998), ‘The symbiosis of choices and control: toward a discourse-based account of can’, Journal of Pragmatics, 29: 615-637. Hermerén, L. (1978), On modality in English: a study of the semantics of the modals. Lund: CWK Gleerup. Huddleston, R. (1995), ‘The case against a future tense in English’, Studies in Language, 19: 399-446. Huddleston, R. and G. K. Pullum (2002), The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Hughes, A. and P. Trudgill (1979), English accents and dialects. London: Arnold. Hunston, S. and G. Francis (2000), Pattern Grammar. A corpus-driven approach to the Lexical Grammar of English. Amsterdam: John Benjamins. Jacosson, B. (1980), ‘On the syntax and semantics of the modal auxiliary had better’, Studia Neophilologica, 52: 47-53. Jespersen, O. (1924), The philosophy of grammar. London: Allen and Unwin. Jespersen, O. (1909-1949), A modern English grammar on historical principles (7 vols.) London: George Allen and Unwin, Copenhagen: Munsgaard. (Republished, Heidelberg: Carl Winter; London: George Allen and Unwin.)
188
References
Jespersen, O. (1958 (1914), A modern English grammar on historical principles VII. Syntax, vol. 1. London: Allen and Unwin, Copenhagen: Ejnar Munsgaard. Kennedy, G. (1998), An introduction to Corpus Linguistics. London: Longman. Klinge, A. (1993), ‘The English modal auxiliaries: from lexical semantics to utterance interpretation’, Journal of Linguistics, 29: 315-357. Krug, M. (1998), ‘Gotta – the tenth central modal in English? Social, stylistic and regional variation in the British National Corpus as evidence of ongoing grammaticalization’, in: H. Lindquist, S. Klintborg, M. Levin, and M. Estling (eds), The major varieties of English. Växjö: Växjö University Press. 177-191. Krug, M. (2000), Emerging English modals. A corpus-based study of grammaticalization. Berlin: Mouton de Gruyter. Lakoff, R. (1972), ‘The pragmatics of modality’, Papers from the 8th Regional Meeting of the Chicago Linguistic Society. 229-246. Larkin, D. (1976), ‘Some notes on English modals’, in: J. D. McCawley (ed.), Syntax and semantics. Vol. 7: notes from the linguistic underground. New York: Academic Press. 387-398. Larreya, P. (2000), ‘Modal verbs and the expression of futurity in English, French and Italian’, Belgian Journal of Linguistics, 14: 115-129. Larreya, P. (2003), ‘Irrealis, past time reference and modality’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 21-45. Lebrun, Y. (1965), Can and may in Present-Day English. Brussels: Presses Universitaires de Bruxelles. Leech, G. (1987), Meaning and the English verb, 2nd. edn. London: Longman. Leech, G. (1992), ‘Corpora and theories of linguistic performance’, in: J. Svartvik (ed.), Directions in Corpus Linguistics. Proceedings of Nobel Symposium 82, Stockholm, 4-8 August 1991. Berlin: Mouton de Gruyter. 105-122. Leech, G. (2003), ‘Modality on the move: the English modal auxiliaries 19611992’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 223-240. Leech, G. and J. Coates (1980), ‘Semantic indeterminacy and the modals’, in S. Greenbaum, G. Leech and J. Svartvik (eds), Studies in English Linguistics for Randolph Quirk. London: Longman.79-90. Lyons, J. (1977), Semantics. Cambridge: Cambridge University Press. Mair C. and G. N. Leech (2006), ‘Current changes in English syntax’, in: B. Aarts and A. McMahon (eds), Handbook of English linguistics. Oxford: Blackwell. 318-342. Matthews, R. (1991), Words and worlds: on the linguistic analysis of modality. Frankfurt: Lang. McCallum-Bayliss, H. (1985), The modal verbs: universal/univocal lexical items. PhD Dissertation, Georgetown University. Ann Arbor: University Microfilms International.
References
189
Mindt, D. (1995), An empirical grammar of the English verb: modal verbs. Berlin: Cornelsen Verlag. Mitchell, K. (2003), ‘Had better and might as well: on the margins of modality?’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 129-149. Myhill, J. (1995), ‘Change and continuity in the functions of the American English modals’, Linguistics 33: 157-211. Myhill, J. (1997), ‘Should and ought: the rise of individually oriented modality in American English’, English Language and Linguistics, 1 (1): 3-23. Nagle, S. (1988), ‘Quasi-modals, marginal modals, and the diachrony of the English modal auxiliaries’, Folia Linguistica Historica, 9 (2): 93-104. Nelson, G. (2003), ‘Modals of obligation and necessity in varieties of English’, in: P. Peters (ed.), From local to global English: proceedings of Style Council 2001/2. Macquarie University: Dictionary Research Centre. 2532. Nelson, G., S. Wallis and B. Aarts (2002), Exploring natural language: working with the British Component of the International Corpus of English. Amsterdam: John Benjamins. Nokkonen, S. (2006), ‘The semantic variation of NEED TO in four recent British English corpora’, International Journal of Corpus Linguistics, 11: 29-71. Papafragou, A. (2000), Modality: issues in the semantics–pragmatics interface. Amsterdam: Elsevier Science. Palmer, F. (1990), Modality and the English modals, 2nd edn. London: Longman. Palmer, F. (2001), Mood and modality, 2nd edn. Cambridge: Cambridge University Press. Perkins, M. (1983), Modal expressions in English. London: Frances Pinter. Poutsma, H. (1926-1929), A grammar of late Modern English. Groningen: Noordhoof. Quirk, R., S. Greenbaum, G. N. Leech and J. Svartvik (1985), A comprehensive grammar of the English language. London: Longman. Römer, U. (2005), Progressives, patterns, pedagogy. A corpus-driven approach to English progressive forms, functions, contexts and didactics Amsterdam: John Benjamins. Smith, N. (2003), ‘Changes in the modals and semi-modals of strong obligation and epistemic necessity in recent British English’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 241-266. Sweetser, E. (1990), From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, E. (1989), ‘On the rise of epistemic meanings in English: an example of subjectification in semantic change’, Language, 65: 31-55. Tregidgo, P. S. (1982), ‘Must and may: demand and permission’, Lingua, 56: 7592. Trousdale, G. (2003), ‘Simplification and redistribution: an account of modal verb usage in Tyneside English’, English World-Wide, 24: 271-284.
190
References
Trudgill, P. (1986), Dialects in contact. Oxford: Basil Blackwell. Van der Auwera, J. (1986), ‘The possibilities of may and can’, in: D. Kastovsky and A. Szwedek (eds), Linguistics across historical and geographical boundaries: in honour of Jacek Fisiak on the occasion of his 50th birthday. Vol.2: Descriptive, contrastive and applied linguistics. Berlin: Mouton de Gruyter. 1067-1076. Van der Auwera, J. and V. A. Plungian (1998), ‘Modality’s semantic map’, Linguistic Typology, 2: 79-124. Vanparys, J. (1987), ‘Toward a pragmatic approach to modality: the case of permissive can and may’, in: J. Verschueren (ed.), The pragmatic perspective. Selected papers from the 1985 International Pragmatics Conference. Amsterdam: John Benjamins. 229-238. Verplaetse, H. (1999), ‘The expression of volition in political interviews Reconsidered’, Belgian Essays on Language and Literature, 91-117. Verplaetse, H. (2003), ‘What you and I want: a functional approach to verb complementation of modal WANT TO’, in: R. Facchinetti, M. Krug and F. Palmer (eds), Modality in contemporary English. Berlin: Mouton de Gruyter. 151-189. Verstraete, J-C. (2001), ‘Subjective and objective modality: interpersonal and ideational functions in the English modal auxiliary system’, Journal of Pragmatics, 33: 1505–28. Vihla, M. (2000), ‘Epistemic possibility: a study based on a medical corpus’, in: J. Kirk (ed.), Corpora galore: analyses and techniques in describing English. Amsterdam: Rodopi. 209-224. Walton, A. L. (1991), ‘The semantics and pragmatics of CAN’, Linguistische Berichte, 135: 325-345. Westney, P. (1995), Modals and periphrastics in English: an investigation into the semantic correspondence between certain English modal verbs and their periphrastic equivalents. Tübingen: Max Niemeyer Verlag.
Index Aarts, F. 9, 153 ability 91-123 actualization 15, 35, 103, 116, 120, 148 non-actualization 45-46, 55, 78 adversative 49-50 Aijmer, K. 121 ambiguity 23 Americanization 8, 149 auxiliary: semi-auxiliary 16 primary auxiliary 12 backshifting 106 be able to 119-122, 178 be about to 155-156, 184 be bound to 86-88, 173 be going to 143-150, 182 be supposed to 80-83, 171 be to 83-86, 172 Berglund, Y. 6 Biber, D. et al., 5, 6, 9, 33, 140 Bolinger, D. 91, 96, 100, 150, 153 British National Corpus 19 Bybee, J.L. et al., 22, 23, 125 can 97-105, 175 Chapin, P.J. 15 Coates, J. 5, 6, 15, 21, 24, 25, 27, 28, 29, 36, 38, 39, 40, 41, 42, 43, 44, 46, 48, 53, 60, 63, 64, 65, 69, 71, 88, 91, 92, 93, 94, 96, 97, 98, 112, 113, 119, 126, 132, 133, 138, 144, 148, 149, 151 Collins, P. 91, 98 colloquialization 9, 149 concessive 93, 113 conditional 50-51, 85, 117, 130, 141 Conrad, S. 9 corpus linguistics: corpus-based 5-6 corpus-driven 5-6 corpus-informed 5-6 could 105-119, 177
Depraetere, I. 11, 25, 66 diachronic change 7-8 Dirven, R. 91 Duffley, P. et al., 91, 100 dynamic implication 96, 104, 116 Edmonson, W. et al., 78, 81 Ehrman, M. 6, 24 emotive 50 Facchinetti, R. 7, 77, 102, 105, 119, 121 Fairclough, N. 77 Francis, G. 5 Freiburg-Brown Corpus 3-4 gonna 143 grammaticalization 16, 18-20, 21, 59, 68, 77, 143, 151, 157 Gresset, S. 113 Groefsema, M. 24, 91, 100 had better 77-79, 170 Haegeman, L. 24 Hakutani, Y. 15 Hargis, C.H. 15 harmonic 39, 92-93, 94 have got to 68-72, 160, 168 have to 59-68, 160, 167 hedge 39-40 Hermerén, L. 6, 112, 119 Huddleston, R. 11, 13, 17-19, 23, 24, 26-28, 30, 38, 40, 44, 47, 49, 50, 51, 53, 55, 79, 82, 85, 98, 99, 102, 107, 118, 126-128, 130, 131, 133-138, 144, 148, 149, 157 Hughes, A. 63 Hunston, S. 5 idiomaticity 16, 21 indirect speech act 104, 142 inherency 96 International Corpus of English 1-3, 9 irrealis 111-112 Jacobsson, B. 18 Jespersen, O. 5, 22
192
Index
Kantor, 95 Kennedy, G. 6 Klinge, A. 24, 91, 96, 100 Krug, M. 18, 68, 72, 125, 150, 151, 154, 155 Lakoff, R. 15, 30 Larkin, D. 15 Lebrun, Y. 91 Leech, G. 6-9, 43, 48, 52, 56, 59, 61, 70, 72, 76, 77, 80, 83, 86, 91, 95, 96, 97, 100, 105, 113, 118, 119, 125, 139, 143, 149, 151, 154, 157, 159, 160-162 Longman Corpus 7-8 Lyons, J. 126 Mair, C. 7-9, 43, 52, 56, 59, 76, 77, 80, 83, 86, 125, 139, 143, 149, 151, 154, 157, 159, 160-162 mandative 49 may 92-97, 161, 174 Matthews, R. 64 McCallum-Bayliss, H. 61 might 105-119, 176 Mitchell, K. 77, 78 modal harmony 27 modality 11 agent-oriented modality 22 degree of modality 27-28, 39, 127-128 deontic modality 22, 23 dynamic modality 22, 23 epistemic modality 11, 19, 20, 21, 23 root modality 11, 19, 20, 21, 22 strength of modality 25-27, 33 modalization 17, 19, 144, 150, 152 modals: central modals 14 lexico-modals 16-17 marginal modals 14 modal idioms 16 quasi-modals 11, 15-17 semi-modals 16-17 monosemy 23-24 mood 11
subjunctive mood 11, 27 must 35-44, 160, 163 Myhill, J. 67 necessity 33-89 need 57-59, 161, 166 need to 73-77, 161, 169 negation 24-25, 27-28, 43, 48, 56, 66-67, 72, 76, 79, 84, 88, 131, 133 Nelson, G. 9 ‘NICE’ constructions 12, 17, 31 Nokkonen, S. 7 non-assertive 73, 121 objectivity 28-29, 35-38 obligation 33-89 ought to 52-57, 165 Palmer, F. 5, 6, 15, 22, 23, 28-30, 42, 46, 53, 60, 71, 78, 85, 87, 88, 96, 102, 103, 112, 121, 126, 128, 130, 134, 136, 138, 145, 151 Papafragou, A. 24 paraphrase 23 performative 28, 35, 69, 142 Perkins, M. 15, 19, 28, 58, 60, 78, 79, 86 permission 91-123 Plungian, V.A. 11, 22 polysemy 23-24 possibility 91-123 existential possibility 102, 114 rational possibility 102-103, 114 theoretical possibility 96 potentiality 96, 103, 115 Poutsma, H. 5 pragmatic: pragmatic strengthening 93 pragmatic weakening 27, 37-38, 39-40 predictability 126-127 prediction 125-157 propensity 133 purposive 50 Pullum, G.K. 5, 11, 13, 17-19, 23, 24, 26-28, 30, 38, 40, 44, 47, 49, 50, 51, 53, 55, 79, 82, 85, 98, 99,
Index 102, 107, 118, 126-128, 130, 131, 133-138, 144, 148, 149, 157 Quirk, R. et al., 5, 8, 14, 16, 17, 20, 23, 113 reanalaysis 117-118 Reed, S. 11, 25, 66 Reppen, R. 9 resistibility 37-38 rhetorical question 47 Römer, U. 6 Santa Barbara Corpus 3-4, 9 semantic bleaching 18-19, 20, 77 shall 135-139, 162, 180 should 44-52, 160-161, 164 quasi-subjunctive should 44, 48 preterite should 51-52 Smith, N. 59, 67, 72, 73, 76 subjectification 21, 59 subjectivity 28-29, 35-38 suppletion 15, 87, 98 Sweetser, E. 21, 37, 61
193 temporality 24-25, 41-43, 47-48, 56, 64-66, 71, 75 Traugott, E. 125 Trudgill, P. 8, 63 unreal 106 unreal conditional 13 unreal preterite 13-14 Van der Auwera, J. 11, 22, 100 Vanparys, J. 100 Verplaetse, H. 7, 93, 151, 152 Vihla, M. 94 voice-neutrality 17 volition 125-157 Wallis, S. 9 wanna 150 want to 150-154, 162, 183 Westney, P. 6, 15, 16, 28-30, 55, 60, 70, 72, 79, 81, 83 will 126-135, 161, 179 would 139-143, 161, 181