A corpus-based study of phonological free variation in

0 downloads 0 Views 15MB Size Report
E) Usage-based (frequency, analogy) factors ... Well-known exceptions: surveys of pronunciation ... Some problems of pronunciation research by written.
A corpus-based study of phonological free variation in English Jose A. Mompean University of Murcia, Spain [email protected]

International Conference on English Pronunciation: Issues & Practices Université de Savoie, Chambéry. June 3-5 2009

1

Contents 1

Introduction

2

An empirical study

3

2.1

Method

2.2

Results

Discussion/Conclusions

2

1. Introduction • What is phonological free variation? A) two or more phonemes can be found in the same position in a lexical item e.g. again /ǩ'gen/-/ ǩ'geǺn/ B) a phoneme does or doesn’t occur in the same position in a lexical item e.g. often /ɑǢfǩn/-/ɑǢftǩn/ C) a given item exhibits two (or three) stress patterns e.g. dispute (n.) /dǺɑspju:t/-/ɑdǺspju:t/

3

1. Introduction • What is phonological free variation? In citation forms, independent of speech processes * e.g. and /ænd/-/ǩnd, ǩn, n, ǩm, ǩŋ…/ * e.g. academic /ɕækǩɑdemǺk/ but academic year /ɕækǩ(ɕ)demǺk ɑjǺǩ/ No change in meaning or grammatical category * import (n.) /ɑǺmpǤət/ - import (v.) /ǺmɑpǤət/ * bow (n.) /bǩȚ/ - bow (n.) /baȚ/ (a weapon) (forward end in vessels)

4

1. Introduction • Why does it occur? A) Ongoing sound changes sure /ȓȚǩ/ → /ȓǤə/ B) Other phonetic factors February /ɑfebruǩri/ -/ɑfebjuǩri/ (‘Febuary’) C) Sound-spelling regularisation tendencies nephew /ɑnevju:/ → /ɑnefju:/

5

1. Introduction •

Why does it occur?

D) Sound-spelling deregularisation tendencies Bolognese /bǢlǩɑneǺz/-/bǢlǩɑnjeǺzeǺ/ E) Usage-based (frequency, analogy) factors incomparable /ǺnɑkǢmpǩrǩbǩl/-/Ǻnkǩmɑpærǩbǩl/ applicable /ɑæplǺkǩbǩl/-/ǩɑplǺkǩbǩl/ (cf. reɑpair → reɑpairable, reɑly → reɑliable)

6

1. Introduction • Previous studies: Little empirical attention in English (a marginal phenomenon?) Well-known exceptions: surveys of pronunciation preferences carried out by written questionnaire (e.g. Shitara, 1993; Wells, 1995, 1999, 2008)

7

1. Introduction • Previous studies: Some problems of pronunciation research by written questionnaire Some variables difficult to control for (e.g. accent) Respondents’ failure to report on their actual usage accurately

8

2. Empirical study Need for the analysis of spoken data Data Through corpus analysis and/or elicited speech

Spoken corpus-based study of phonological free Results variation in English Aims: Find out… rates of speakers producing the variants rates of occurrence of instances of ‘free variants’

9

2.1 Empirical study: method Data/Speakers News archives from the BBC Learning English website (section Words in the News) years 1999-2008

Results

1339 ‘words in the news’’ files

10

2.1 Empirical study: method Data/Speakers Criteria: a) Only newsreaders’ speech b) Only BBC English/RP speech analysed c) Only speech produced by identified speakers Results

ca. 200,000 words analysed 318 RP speakers (222 males, 96 females).

11

2.1 Empirical study: method Procedure Corpus created by copying written versions onto another document and creating a database Pre-search of items and check for occurrence Results Ca 50 items. General selection of lexical items/sets:

a) with 10 or more occurrences in the corpus b) produced by more than 10 speakers

12

2.1 Empirical study: method Procedure Specific words listened to by two independent judges In case of divergent opinions, a third judge listened to the items Dubious instances inspected with SFS/WASP, developed at UCL

Results

13

2.2 Empirical study: results Weak vowel/diphthong in non-stressed syllables FIACE

FIACIAL /ǩ, Ǻ/

/ǩ, Ǻ/

14%

/aǺ/

86%

/aǺ/

/eǺ/

100%

HURRICAE

DECADE /ǩ/

0%

/ǩ/

6% 85%

/eǺ/

93% 7%

LPD 08 /ǩ/: 40% /eǺ/: 60%

14

2.2 Empirical study: results Weak vowel/diphthong in non-stressed syllables /dɪ-, dǩ-/ DIRECT (v.)

DIRECTOR

58%

42%

DIRECT (adj.) DIRECTLY

/daǺǩ-/

71%

29%

42%

39%

58%

61%

LPD 1998 “direction”/dɪ-, dǩ-/: 30%, /daǺǩ-/: 69%

15

2.2 Empirical study: results Sound-spelling correspondences ECOOMIC /ˌi:k-/

47%

/ˌek-/

53%

LPD 1988 /ˌi:k-/: 62% /ˌek-/: 38%

MUSLIM /Ț/

100%

/Ȝ/ /z/ /s/

91% 9%

LPD 1998 /Ț/: 70% /z/: 89%

/Ȝ/: 30% /s/: 11%

16

2.2 Empirical study: results Sound-spelling correspondences BEIJIG /Ȣ/ /

ASIA /ȓ/ /

38%

/ȴ/ /

/Ȣ/ /

62%

/sk/

80% 20%

89%

LPD 1998 2008

SCHEDULE /ȓ/

11%

Asia

/ȓ/: /Ȣ/:

49% 36% 51% 64%

schedule /ȓ/:

70% /sk/: 30% (+1973, 65%)

17

2.2 Empirical study: results Sound-spelling correspondences AGAI /e/ /eǺ/

AGAIST 85%

15%

/e/ /eǺ/

/aǺ/

34%

EITHER

EITHER /i:/

66%

/i:/

15% 85%

/aǺ/

23% 77%

18

2.2 Empirical study: results Sound change: Decline and disappearance of /ʊə// POOR /Țǩ/

32%

/Ǥə/

68%

SURE /Țǩ/ /Ǥə/

LPD

1988 1998 2008

/pȚǩ/

43% 45% 26%

/pǤə/

57% 55% 74%

LPD /ȓȚǩ/

6% 94%

1998 54%

/ȓǤə/

46% (Born since 1973 60%)

19

2.2 Empirical study: results Sound change: Yod coalescence SITUATIO /tj/

83%

SCHEDULE

/ȴ/

TUESDAY

/tj/

17%

/ȷ/

/dj/

LPD /tj/: 64%, 1998 /ȷ/: 35%

/ȷ/

LPD /dj/: 79% 1998 /ȴ/: 21%

30% 70%

/dj/ /ȴ/

60% 40%

PRODUCE (v.) 60% 40%

20

2.2 Empirical study: results Phonetic factors: dissimilation LPD 1998

FEBRUARY /ɑfebr-/ /ɑfebj-/

/ɑfebr-/: 61%, /ɑfebj-/: 39%

58% 42%

/-ˈsǩʊȓi-/

ASSOCIATE (v.)

/-ˈsǩȚsi-/

ASSOCIATIO

/ǩɕsǩȚȓiˈeǺȓǩn/ /ǩɕsǩȚsiˈeǺȓǩn/

7% 93% 0% 100%

21

2.2 Empirical study: results Word Stress: competing variants COTRIBUTE /ɑkǢn-/

/ɑkǺl-/

22%

/-ɑtrǺb-/

KILOMETRE

78%

/-ɑlǢm-/

/-ɑtrǢ-/

75% 25%

LPD 98 /ɑkǢn-/ : 40%, /-ɑtrǢ-/ : 60%

74%

CIGARETTE

COTROVERSY /ɑkǢn-/

26%

/-ɑret/

50%

/ɑsǺg-/

50%

LPD 88 /-ɑret/ : 85% /ɑsǺg-/: 15% 22

2.2 Empirical study: results Word Stress: disappearing/rare variants Also for

COMPLEX /ɑkǢm-/

DETAIL 100%

/-ɑpleks/ DISPUTE (n.) /ɑdǺ-/ /-ɑspju:t/

/ɑdi:-/ (100%) /-ɑteǺəl/ (0%) ELECTORAL

6% 94%

/-ˈlekt-/ (100%) /-ˈtǤər-/ (0%)

23

3. Discussion/Conclusions Asia

Contribute

In general, results reveal many ‘preferences’ similar to those found in the LPD pronunciation polls But they also reveal interesting data for certain specific items (e.g. accent-specific)

LPD 08 BBC LPD 90/00

/ˈ…/: 41%

/ʃ/: 36%

/.ˈ../: 59%

/ʒ/: 64%

/ˈ…/: 24%

/ʃ/: 11%

/.ˈ../: 82%

/ʒ/: 89%

/ˈ…/: 27%

/ʃ/: 49%

/.ˈ../: 73%

/ʒ/: 51%

Controversy LPD 00 BBC BBC

/ˈ…/: 40%

/.ˈ../: 60%

/ˈ…/: 73% /.ˈ../: 27%

24

3. Discussion/Conclusions Limitations of the present study a) age differences cannot be investigated b) under-representativeness of certain items c) number of speakers/instances difficult to control for Need for further evidence/studies

25

3. Discussion/Conclusions Directions for future research a) Enlarge the corpus (obtain more representativeness of specific items, study diachronic evolution) b) Conduct further independent or coordinated: -pronunciation polls -corpus-based studies -controlled experimental tasks

26

27