LING 542 presentation Meng Liu.IILS 6 - V2

27
PROSODIC FACTORS IN FOREIGN ACCENTS OF L1 ENGLISH LEARNERS OF CHINESE Meng Liu Department of East Asian Language and Cultures University of Illinois Urbana-Champaign

Transcript of LING 542 presentation Meng Liu.IILS 6 - V2

Page 1: LING 542 presentation Meng Liu.IILS 6 - V2

PROSODIC FACTORS IN

FOREIGN ACCENTS OF L1

ENGLISH LEARNERS OF

CHINESE

Meng Liu

Department of East Asian Language and Cultures

University of Illinois Urbana-Champaign

Page 2: LING 542 presentation Meng Liu.IILS 6 - V2

Study Purpose

• The purpose of this study is to explore how

prosodic features contribute to the

perception of foreign accent of L1 English

learners of Chinese with a focus on rhythm,

by using rhythm metrics.

Page 3: LING 542 presentation Meng Liu.IILS 6 - V2

Overview

• Introduction English and Chinese prosody

Factors contribute to foreign accent

•Study1: Perception-foreign accent rating

•Study2: Production-rhythm measurements

•Conclusions

Page 4: LING 542 presentation Meng Liu.IILS 6 - V2

English and Chinese Prosody

• English: syllable-timed (Pike, 1945;

Abercrombie,1967; Grenon & White, 2008)

• Chinese: syllable-timed (Lehiste, 1970; Lin &

Wang, 2007, Arvantiti, 2009)

Page 5: LING 542 presentation Meng Liu.IILS 6 - V2

Foreign Accent

• Jilka (2000): any foreign accent’s major

characteristic could be described as a deviating

realization of corresponding phonological

units that can often be traced back to specific

features of the respective speakers’ native

language.

Page 6: LING 542 presentation Meng Liu.IILS 6 - V2

Factors contribute to Foreign Accent • English

• Segmentals and Syllable structure (Anderson-H. et al. ,

1992)

• Prosody-word duration, pauses and rhythm metrics

(Munro, 1995; Trofovimich & Baker, 2006; Baker, et al., 2011;

Ramus, et al., 1999; Grabe &Low, 2002; Low, et al., 2000;

Dellwo, 2006 )

• Speaking rates (Munro, 1995; Trofovimich & Baker, 2006)

• Chinese

• Vowel formant (Wu, 2011)

• Phrase-initial lengthening/shortening (Yang, 2011)

• Tone (Shih, et al, 2009).

Criticism from experimental

studies: Loukina, et al , 2013;

Arvaniti, A., in press

Page 7: LING 542 presentation Meng Liu.IILS 6 - V2

STUDY 1

Perception

Page 8: LING 542 presentation Meng Liu.IILS 6 - V2

Research question

• How rhythm measures correlate with

different degrees of foreign

accentedness?

• How speech rate correlates with different

degrees of foreign accentedness?

.

Page 9: LING 542 presentation Meng Liu.IILS 6 - V2

Method

• Participants :

• 3 native Mandarin speakers (3 female) from Mainland

China

• Stimuli 6*(8+8)

• 6 informants

• 8 sentences of story reading

• 8 10-second utterances

Procedure

• Accent rating: AVSA , 96 utterances

• Interview: rating criteria

Adaptive visual analogue scale

Page 10: LING 542 presentation Meng Liu.IILS 6 - V2

Statistical analysis • Fa.degree: mean centered.

• Speech rate:

• Syllable/sec.

• Rhythm measurements

• %V ratio=dv.sentence/d.sentence

• VarcoC (s(dc)/ mean(dc))*100

• VarcoV (s(dv)/ mean(dv))*100

• nPIV-V

Rhythm metrics indicate

the durational variability of

consonants and vowels

compared to adjacent

consonants and vowels.

Page 11: LING 542 presentation Meng Liu.IILS 6 - V2

Model Comparison • Linear mixed-effects model

• AIC (Akaike information criterion)

DV Foreign Accent Degree centered

(a-mean(a))/sd(a)

Fixed

effects

Group N, S

Style R,C

Speech rate

VarcoC VarcoV

%V

nPIV-V

Random

effect

Participants

Items

p1,p2,p3

n1c1-s3r8

Page 12: LING 542 presentation Meng Liu.IILS 6 - V2

Results Foreign accent rating : speech style and groups

Estimate Std. Error df t value Pr(>|t|)

(Intercept) -0.73350 0.20219 90.98000 -3.628 0.000471 ***

groups 1.72916 0.05264 90.98000 32.851 < 2e-16 ***

styler 0.10990 0.03407 90.98000 3.225 0.001749 **

speechrate -0.11756 0.02573 90.98000 -4.569 1.53e-05 ***

pv 0.23923 0.25827 90.98000 0.926 0.356756

Group: F(1,91)=1079.20, p<0.001 Style: F (1,91)=10.4, p<0.01

Page 13: LING 542 presentation Meng Liu.IILS 6 - V2

Estimate Std. Error df t value Pr(>|t|)

(Intercept) -0.73350 0.20219 90.98000 -3.628 0.000471 ***

groups 1.72916 0.05264 90.98000 32.851 < 2e-16 ***

styler 0.10990 0.03407 90.98000 3.225 0.001749 **

speechrate -0.11756 0.02573 90.98000 -4.569 1.53e-05 ***

pv 0.23923 0.25827 90.98000 0.926 0.356756 Speech rate:F(1,91)=20.88, p<0.001

Page 14: LING 542 presentation Meng Liu.IILS 6 - V2
Page 15: LING 542 presentation Meng Liu.IILS 6 - V2

STUDY 2

Production

Page 16: LING 542 presentation Meng Liu.IILS 6 - V2

Research question

What are the differences in rhythm between L1

English learners of Chinese and Mandarin

Chinese speakers ?

What are the differences in rhythm concerning

production during reading and spontaneous

speech?

Page 17: LING 542 presentation Meng Liu.IILS 6 - V2

Method

• Informants : • 3 native Mandarin speakers (3 female)

• 3 L1 English learners of Mandarin speakers (3 female)

• 3-4 years of Chinese learning experience

• Intermediate-advanced level

• Stimuli: • Reading story in Chinese ( 8 sentences)

• Answer 8 questions in Chinese

Page 18: LING 542 presentation Meng Liu.IILS 6 - V2

Statistical analysis • File segmenting:

• Force aligner http://www.ling.upenn.edu/phonetics/p2fa/

and manual adjust ment

• Model: Linear mixed model

Dependent Variable VarcoC VarcoV

%V

nPIV-V

Speech rate

Fixed Effect Group N, S

Style Reading,Conv

Random effect Speakers

Items

N1,N2,N3

S1,S2,S3

C1-C8

R1-R8

Page 19: LING 542 presentation Meng Liu.IILS 6 - V2

Estimate Std. Error df t value Pr(>|t|)

(Intercept) 3.5399 0.2632 4.6320 13.450 6.88e-05 ***

groups -1.6193 0.3584 4.0000 -4.519 0.0107 *

rcr 0.2830 0.1279 13.9940 2.212 0.0441 *

Group: F(1,4)=20.42, p<0.05 Style : F (1,14)=4.89, p<0.05

Page 20: LING 542 presentation Meng Liu.IILS 6 - V2

Style: F (1,89)=5.28, p<0.05

Estimate Std. Error df t value Pr(>|t|)

(Intercept) 65.5125 2.5220 6.3500 25.976 1.09e-07 ***

groups 0.8496 3.1748 4.0000 0.268 0.8023

rcr -5.2812 2.2985 89.0000 -2.298 0.0239 *

Page 21: LING 542 presentation Meng Liu.IILS 6 - V2

Style: F (1,89)=50.56, p<0.001

Estimate Std. Error df t value Pr(>|t|)

(Intercept) 64.017 3.555 5.510 18.008 4.17e-06 ***

groups -3.392 4.639 4.000 -0.731 0.505

rcr -19.484 2.740 89.000 -7.111 2.78e-10 ***

Page 22: LING 542 presentation Meng Liu.IILS 6 - V2

Style: F (1,93)=41.53, p<0.001

Estimate Std. Error df t value Pr(>|t|)

(Intercept) 57.549 2.177 93.000 26.438 < 2e-16 ***

groups 3.619 2.513 93.000 1.440 0.153

rcr -16.199 2.513 93.000 -6.445 5.08e-09 ***

Page 23: LING 542 presentation Meng Liu.IILS 6 - V2

Estimate Std. Error df t value Pr(>|t|)

(Intercept) 0.63436 0.02167 6.17400 29.268 7.37e-08 ***

groups 0.02378 0.02725 3.99900 0.873 0.432

rcr -0.02517 0.01623 13.99600 -1.551 0.143

Page 24: LING 542 presentation Meng Liu.IILS 6 - V2

Interview of foreign accent rating

participants • Rating criteria

p1 intonation articulation fluency the way of

expression

p2 tone (mainly),pronunciation

p3 intonation, tones, segment articulation pasues coherent content swallow

Further studies • Segments

• Tone

• Intonation

Page 25: LING 542 presentation Meng Liu.IILS 6 - V2

Conclusions

The study indicates that rhythm metrics is not a powerful tool to distinguish the native and nonnative speech of Chinese

Both the speech style and speaking rate have significant influence on the foreign accent rating score of the student group and native speaker group

Specifically, in spontaneous speech, students were rated with less foreign accent than the reading utterances. The faster the speaking rate, the less foreign accent is being perceived.

%V contributes more to the foreign accent perception than other rhythm metrics, though the statistical model is not significantly different.

Page 26: LING 542 presentation Meng Liu.IILS 6 - V2

Acknowledgements

• Dr. Chilin Shih

• Dr. Jennifer S. Cole

• Dr. Joseph Roy

• EALC TAs

• CHIN 305 participants

• CHIN 404 participants

Page 27: LING 542 presentation Meng Liu.IILS 6 - V2

THANK YOU!

Q & A