The effect of native language on the second language vowel variability

Abstract

Korean learners of English must create four vowel categories for English (/i, ɪ/ and /ɛ, æ/) in relation to two similar native categories (/i/ and /ɛ/). It is hypothesized that new categories should be easier to learn than similar ones (Flege, 1994), but it is unclear whether the English L2 vowels are similar or new. The degree of similarity between the four English vowels and the two Korean vowels was examined using the distribution metrics (i.e., ellipse overlap, cross-entropy, and Gaussian Mixture Model) as well as Euclidean distance in F1/F2 space. The L2 spoken corpus included 100 repetitions of words in both Korean and English spoken by 37 Korean L2 learners (20 female). Preliminary results indicate that the English (L2) high vowel pair was more overlapped with Korean /i/ than the low vowel pair with Korean /ɛ/, especially for male speakers. For the English (L2) low vowel pair, female speakers showed less overlap but higher variability along F1 direction than male speakers. This demonstrates that the similarity between Korean and English vowels is characterized by the distribution as well as the distances between the vowel categories. Acoustic results will be further compared with identification by native English speakers.

Full text

The Effect of native language (L1) on the second language (L2) vowel variability
How much ‘new’ vowels differ from ‘similar’ vowels?
Jaekoo Kang1, D.H. Whalen1 2 3, Hosung Nam2 4
1The CUNY Graduate Center, 2Haskins Laboratories, 3Yale University, 4Korea University
jkang@gradcenter.cuny.edu dwhalen@gc.cuny.edu hnam@korea.ac.kr
3pSC26 176th ASA Meeting & 2018 Acoustics Week in Canada
Victoria, BC, Canada, 5-9 November 2018
MAIN GOALS Results
Discussion and Conclusions
•Within-vowel variability was not a function of ‘new’ vs. ‘similar’ vowel
pairs.
• → No significance both in F1 and F2 (DV: Standard Deviation)
• Vowel categorical differences (/i/ or /ɛ/ group) affected the distributional
similarity of ‘new’ and ‘similar’ vowel pairs.
• → Significant result in F1 (/ɛ/ Group) and F2 (/i/ Group)
(DV: Bhattacharyya Coefficient, Adjusted Mutual Information)
References
Acknowledgements
Methods
• Are ‘new’ vowels more variable than ‘similar’
vowels?
• Are ‘new’ vowels less overlapped and less
similar than ‘similar’ vowels?
•Compare the formant distribution of ‘new’
and ‘similar’ L2 English vowels spoken by
Korean native speakers.
• A spoken corpus of Korean native speakers (27
speakers,16 ♀), highly proficient in English.
• 2 Korean front vowels in hi, hɛ and 4 English
vowels in heed, hid, head, had). 100 repetitions
each, measured at mid-point.
Standard
Deviation
Bhattacharyya
Coefficient
(overlap)
Adjusted
Mutual
Information
(similarity)
‘Similar’ vs. ‘New’ vowels
• L2 learners typically have difficulty in acquiring
“new” vowels which do not belong to their native
phonetic categories (SLM, PAM, PAM-L2).
• Proficient L2 speakers often differentiate them by
the Euclidean distance.
• However, how the formant distribution differs
remains unclear besides the central tendency.
•[1] Flege, J. E., Munro, M. J., & MacKay, I. R. (1995). Factors affecting strength of perceived foreign
accent in a second language. The Journal of the Acoustical Society of America, 97(5), 3125-3134.
•[2] Flege, J. E., Bohn, O. S., & Jang, S. (1997). Effects of experience on non-native speakers' production
and perception of English vowels. Journal of phonetics,25(4), 437-470.
•[3] Best, C. T., & Tyler, M. D. (2007). Nonnative and second-language speech perception. Language
Learning & Language Teaching (Vol. 17, pp. 13–34). Amsterdam: John Benjamins Publishing Company.
•[4] Yang, B. (1996). A comparative study of American English and Korean vowels produced by male and
female speakers. Journal of Phonetics, 24(2), 245–261.
/i/ /i/
/ɪ/
Korean English
‘Similar’
/i/ Group
/ɛ/ /ɛ/
/æ/
Korean English
‘Similar’
/ɛ/ Group
Work supported by NIH grant DC-002717 to Haskins Laboratories.
*** p<.0005
** p<.005
* p<.05
P2FA Manual
correction
Formant
tracking
Pre-
processing
Compute
EucDist
Compute
Shape Metrics
Statistical
analysis
!" #$%&'# ( %)(*
+
+
,-.
!" #$%&0 for no overlap
!" #$%&1 for entire overlap
/01#$%& '#($2 345 #)($2*
#(%)(*
+
,$6
789 #$%&0 for chance level
789 #$%&1 for identical
distribution
*For simplicity only MI was shown.
•Are ‘new’ vowels more variable than ‘similar’ vowels?
No
L1 /i/
L2 /i/
L2 /ɪ/
L1 /i/
L2 /i/
L2 /ɪ/
n.s.
n.s.
L1 /ɛ/
L2 /ɛ/
L2 /æ/
L1 /ɛ/
L2 /ɛ/
L2 /æ/
n.s.
n.s.
•Are ‘new’ vowels less overlapped and less similar in
distribution than ‘similar’ vowels?
F1 ~ Word + (1|Speaker).
F2 ~ Word + (1|Speaker).
Yes
F1 ~ Group*Pair + (1|Speaker).
F2 ~ Group*Pair + (1|Speaker).
/i/ Group /ɛ/ Group
**
***
**
**
/i/ Group /ɛ/ Group
# : & ';<=)><$?<*
@
<AB
Gaussian Mixtual Model (GMM)
with Expectation Maximization
and model selection based on BIC
[1,2]
[1,2,3]
[2]
/i/ /ɛ/
/ɛ/ /i/
/ɪ/
/ɪ/
/ɛ/
/æ/
/i/
/æ/
L1 L2
L1 L2
Less variable ← → More variable Less variable ← → More variable
(width)
(height) (height)
(width)
Less overlap ←→ More overlap Less overlap ←→ More overlap
Less similar ←→ More similar Less similar ←→ More similar
Tendencies
only