Recent change in the English verb phrase: findings from a spoken corpus

1. Jill Bowie Survey of English Usage, UCL 24 November 2011 j.bowie@ucl.ac.uk Recent change in the English verb phrase: findings from a spoken corpus

2. 1. Changing VP project at SEU ‘The changing verb phrase in present-day British English’ • Corpus study of recent grammatical change in the VP in spoken British English • PI: Bas Aarts (director of SEU) • Researchers: Joanne Close, Jill Bowie • With collaboration of Sean Wallis (Senior Research Fellow, SEU)

3. Recent grammatical change • New field of research, compared with long-term change • Availability of large electronic corpora • Leech et al. 2009: ‘Brown quartet’ • printed written English corpora • BrE: LOB 1961, FLOB 1991 • AmE: Brown 1961, Frown 1992 • Need for work on spoken English • Spoken language primary • Changes likely to be found first in spoken language • DCPSE corpus covers similar time period to Brown quartet

4. Survey of English Usage Founded 1959 by Randolph Quirk Other well-known linguists at the Survey have included: • Sidney Greenbaum, Geoffrey Leech, Jan Svartvik, David Crystal, and many others Well-known work on English linguistics • e.g. A Comprehensive Grammar of the English Language (1985) by Quirk, Greenbaum, Leech & Svartvik Pioneering work in English corpus linguistics • original ‘Survey Corpus’ compiled by Quirk • later computerised as LLC • ICE-GB compiled by Sidney Greenbaum, 1990s data

10. DCPSE:Diachronic Corpus of Present-day Spoken English • Orthographically transcribed spoken BrE • Approx. 400,000 words each from • the LLC (London-Lund Corpus, or ‘Survey Corpus’) • the ICE-GB corpus • Matching text categories (wide range of genres) • Allows comparison over time: • LLC: samples from 1958-1977 • ICE-GB: 1990-1992 • Fully parsed, with tree diagrams

11. A tree diagram

12. FTFs (Fuzzy Tree Fragments) A search facility provided by the ICECUP software. A single-node FTF query to find present-tense perfect auxiliaries:

13. A more complex FTF ...

14. ... and a matching tree structure

15. Findings of VP projectfor spoken BrE, 1960s to 1990s Significant changes found over this period, e.g.: • progressive construction: increased frequency • This is making too much of a racket. • core modal verbs: decreased frequency overall • dramatic falls for must, may, shall • falls for would, could, should • increase for will • perfect construction: to be discussed • introduction to perfect • 3 phases of investigation

16. 2. The perfect: introduction Perfect auxiliary HAVE + verb in past participle form. Typically expresses anteriority (pastness relative to a reference point). Subtypes according to inflectional form of HAVE: • Present perfect, e.g. I’vedone a bit of writing before • Past perfect, e.g. andI was staying in this sort of bedsitter place that she’dbooked up for me • Perfect infinitive, e.g. the virus seems to have cleared up; I must have told you that • ing-participial perfect, e.g. and not having heard from anybody I thought I’d better check up

17. The present perfect More complex than expressing anteriority ... • situation occurs within (or continues through) a timespan leading up to the present • typically involves ‘current relevance’, connection to present • contrasts with simple past, i.e. morphologically marked past tense (which involves time wholly in past rather than connected to present) Examples: • He’s lost his job vs He lost his job (last year) • They’ve known him since 1983 vs *They knew him since 1983 These complexities have led to varying analyses of the perfect, e.g. in terms of aspect or (secondary) tense.

18. Why investigate recent change in perfect? Longer-term change • grammaticalisation process from OE to ModE: rapid increase in frequency of present perfect, mainly at expense of past tense, up to 1600, then tailing off; from 1800 to PDE, clear fall in AmE, and some evidence of fall in BrE (Elsness 1997) • contrast with some other European languages where grammaticalisation process has continued, pushing the perfect further in direction of replacing past tense, e.g. S German Reports of new uses of present perfect • e.g. narrative use, increasing use with certain temporal adjuncts Regional variation: BrE vs AmE • present perfect more frequent in BrE than AmE

19. 3. The perfect: initial investigation Initial research questions • Is there evidence of recent change in the overall frequency of the perfect construction in spoken British English? • Is the same pattern of change seen for the different subtypes of the perfect (present, past, and non-finite)? Data retrieval • Single-node FTF searches for category AUX with type feature ‘perf’ and tense-form features ‘pres’, ‘past’, etc. • Had to screen out instances of stative idiom HAVE got ‘have’, e.g. he’s got two kids (FTF with present perfect auxiliary followed by got).

20. Frequencies (pmw) of perfect auxiliaries

21. % change in pmw frequencies

22. Findings • Overall, there is a statistically significant decline in the pmw frequency of the perfect construction from LLC (1960s-70s) to ICE-GB (1990s). • The different subtypes of the perfect show different patterns over time: • The present perfect shows no significant change in pmw frequency. • The past perfect and infinitive perfect both show statistically significant declines in pmw frequency.

23. 4. The perfect: further testing • Why the declines in past perfect and perfect infinitive? • Do they reflect genuine changes in use? Or are there other possible explanations? (1)Perfect infinitive: decline in possible contexts of occurrence? • Most (88%) occur after a modal: I must have told you that • Other context is after to: The virus seems to have cleared up • Modal auxiliaries have declined significantly in frequency • Could that explain the decline in perfect infinitive? (2)Both forms: less talk about past time, due to chance sampling factors?

24. Perfect infinitives in modal contexts • The proportion of perfect infinitives decreases significantly in the modal context (and also in the to context): there is a trend of decline independent of the decline in modals.

25. Tensed, past-marked (TPM) VPs • Tensed VPs marked for past either morphologically or with perfect, or both: • Past non-perfect e.g. he saw it • Present perfect e.g. he has seen it • Past perfect e.g. he had seen it • Modal + perfect infinitive e.g. he may have seen it • Excludes • Present non-perfect VPs • Non-tensed VPs (this group includes to + perfect infinitive, and -ing participle perfect)

26. TPM VPs as proportion of all VPs

27. % change as proportion of all VPs

28. % change as proportion of TPM VPs

29. Combined graphs for % changes

30. Findings • The categories past perfect and modal + perfect infinitive decline significantly as a proportion of TPM VPs. • This shows that declines in past perfect and perfect infinitive are not due simply to differences in rates of reference to past time. • In addition, the category present perfect increases significantly as a proportion of TPM VPs.

31. 5. Past perfect: digging deeper Main use: past in past (anteriority to past reference point) • ... but when I met him she’d already left him Questions: • Are people increasingly choosing simple past instead of past perfect? I.e. is there a process of replacement under way? Is the distinction gradually being lost? • Are there contexts where it makes little difference? i.e. contexts of possible alternation between the two forms. Problem: • How can we identify instances of ‘non-occurrence’ of the past perfect? I.e. how can we find instances where it could have been used but was not, among 25,000-odd instances of the simple past?

32. Past in past: temporal & relative clauses Temporal clauses with after, as soon as, before, etc. • I met him a lot after my father had died … • They sent one to my mother after she died or something Relative clauses • We asked for the documents which they had mentioned • So I just took some of the tablets you gave me Past perfect vs simple past: proportion of past perfect declines significantly over time in temporal and relative clauses. Suggests possible replacement process.

33. Past in past: main clauses Past in past already expressed in preceding context • Sarah had left the kitchen roll in the sitting room and Chicken had methodically shredded it • I’dphoned the day before and the doctor said it wasn’t necessary ‘Before just then’ / ‘before now’ contexts • Oh I hadn’t realized that • I didn’t realize that Past perfect vs simple past: proportion of past perfect declines significantly over time in main clauses. Suggests possible replacement process.

34. Other uses of past perfect • Backshifting of past ‘original’ (c.15%) • Oh actually Dad asked me if Sarah had phoned me on Sunday • cf. And you said you first got in touch with A S Consultants before Christmas • Modal remoteness + pastness (c.10%) • It would’ve been nice if I had known • cf. I was bored. Well I would have been if I did the last year [i.e. of study in a particular subject] Past perfect vs simple past: no sig. difference in proportion of past perfect in relevant structural contexts: • DO clauses with higher past tense VPs (poss. backshifting context) • If-conditional clauses (poss. modal remoteness context)

35. Possible explanations for declines of past and infinitival perfect • American influence? • AmE lower frequencies than BrE (Elsness 1997) • ? further decline in use of past perfect in AmE (Gorrell 1995) • Increasing tendency to simplify verb phrase? • But not general to all auxiliary types, cf. increasing frequency of progressive • Contrast with present perfect (not declining): lower-frequency categories more likely to suffer loss?

36. 6.Further issue: genre variation • Frequencies of grammatical forms can vary considerably across different genres • Patterns of change over time may also vary by genre • DCPSE includes a wide range of spoken genres • Study of recent change in core modal auxiliaries, broken down by genre • can/could, may/might, must, shall/should, will/would • Baselines for measuring modal frequencies • pmw, per VP, per tensed VP • Does it make a difference?

37. VP density across genres • Does VP density (pmw frequency of VPs) vary in DCPSE over time, across different spoken genres? • Does tensed VP density vary over time across different spoken genres? • What are the implications of these findings for the study of current change in the VP?

38. Tensed VPs pmw, by genre

39. 30% 15% 20% 9% 8% b discussions 6% assorted spont legal x-exam prepared sp 4% 10% 2% 0% 0% Total -2% -10% telephone commentary b interviews formal f-to-f parliament informal f-to-f -20% -17% -30% -22% -21% % change in tensed VPs pmw, by genre

40. 7. Concluding points Important in corpus studies of recent change to: • check output of computerized searches • make careful choice of baseline for measuring frequency (pmw, per VP, per tensed VP, ...) • test alternative explanations of apparent changes in use (e.g. chance sampling factors) • consider structural contexts of occurrence & whether these may have changed over time • consider genre variation & possibility of different patterns of change in different genres

41. References and websites Aarts, B., Close, J., Leech, G. & Wallis, S. (eds.) (forthcoming 2012) The English verb phrase: corpus methodology and current change. Cambridge UP. Bowie, J., Wallis, S. & Aarts, B. (forthcoming 2012) The perfect in spoken British English. In Aarts et al. (eds.). Elsness, J. (1997) The perfect and the preterite in contemporary and earlier English (Topics in English Linguistics, 21). Mouton de Gruyter. Gorrell, R. (1995) The future of past tenses. English Today 44 (v.11 no.4), 25–27. Hundt, M., & N. Smith (2009) The present perfect in British and American English: has there been any change, recently? ICAME Journal 33, 45–63. Leech, G., M. Hundt, C. Mair, & N. Smith (2009) Change in contemporary English: a grammatical study. Cambridge UP. Websites: Survey of English Usage: www.ucl.ac.uk/english-usage VP project: www.ucl.ac.uk/english-usage/projects/verb-phrase/ BNC: www.natcorp.ox.ac.uk/ AmE corpora (Mark Davies): corpus.byu.edu/