The 44th Annual Meeting of the Chicago Linguistics Society
24 April 2008
The Phonetic Space of Phonological Categories in
Heritage Speakers of Mandarin
Charles B. Chang, Erin Haynes, Russell Rhodes, and Yao Yao
University of California, Berkeley
[email protected], [email protected], [email protected], [email protected]
Outline
1.
Background
2.
Methods
3.
Results
4.
Discussion
5.
Conclusions
Background


This study compares production of both
Mandarin and English in heritage speakers of
Mandarin to that of native Mandarin speakers
and that of native English speakers learning
Mandarin as a foreign language.
Heritage speakers of Mandarin (narrow definition):
people who have had exposure to Mandarin in their
family but have shifted to primarily using English
3
Background

A few studies have examined the phonological
competence of heritage speakers:



Au et al. (2002) and Knightly et al. (2003): heritage
speakers of Spanish have a phonological advantage
over late learners (VOT, degree of lenition, and
accent ratings).
Oh et al. (2002, 2003): heritage speakers of Korean
exhibit rather native-like production (VOT and
accent ratings).
Godson (2003): heritage speakers of Armenian show
influence in their Armenian vowels from English, but
only for Armenian vowels close to English vowels.
4
Research Questions


Do heritage speakers maintain contrasts in both
the heritage language and the dominant language?
If so, do they maintain the contrasts to the same
degree as native speakers?
Do heritage speakers maintain contrasts between
segments of the heritage language and similar
segments of the dominant language?
5
Previous Findings

In previous work connected with the current
study (Chang et al. 2008), we investigated
heritage speakers’ fricative production, and
found that


Native speakers and late learners most likely
collapse /ʃ/ and /ʂ/, while heritage speakers
tend to keep the two sounds apart.
There is a correspondence in heritage speakers
between linguistic performance and amount of
exposure to the heritage language.
Current Study

In this study, we focus on back vowels and
stop consonants.


Experiment 1: Five vowels are investigated.
English /u/, English /o(w)/, Mandarin /u/,
Mandarin /o(w)/, and Mandarin /y/.
Experiment 2: Four stop categories. English
voiceless (aspirated) stops (/p/, /t/, /k/) and
voiced stops (/b/, /d/, /g/); Mandarin aspirated
stops (/ph/, /th/, /kh/) and unaspirated ones
(/p/, /t/, /k/)
Outline
1.
Background and research questions
2.
Methods
3.
Results
4.
Discussion
5.
Conclusions
Methods

Participants

18 speakers total




Questionnaire


5 native speakers of Mandarin
8 heritage speakers of Mandarin
5 late learners of Mandarin
Speakers’ status determined based on a language background
questionnaire
Recordings


All items recorded in a sound-proof booth (at 48 kHz, 16 bps)
Pre-amp????, AKG C420 head-mounted condenser
microphone
9
Methods

Stimuli

Exp 1:



Exp 2



English: 11 /CuC/ words. 10 /Co(w)C/ words
Mandarin: 10 /Cu/ words. 7 /Co(w)/ words. 3 /Cy/ words.
English: 12 stop-initial CVC words. 2 per stop category.
Mandarin: 12 stop-initial words. 2 per stop category.
Both experiments



Segmental context is matched across language if possible.
Falling tones are chosen for Mandarin words if possible.
E.g. tote / tòu 透
10
Methods

Stimuli (cont’d)

Presentation of stimuli


words read off of index cards
 English words written in English orthography
 Mandarin words written in Mandarin orthography (traditional and
simplified characters) and romanization (pinyin and BoPoMoFo)
 all words written and read in isolation
words read in 8 blocks
 4 Mandarin blocks
 4 English blocks
 block consisted of reading all of the words from a given language
 words randomized before each block
11
Methods

Acoustic measurements




All measurements were performed in Praat (Boersma
& Weenink 2008).
In Experiment 1, average values of F1, F2, and F3
were measured over the whole duration of the vowel.
In Experiment 2, VOT values were measured for
word-initial stops.
Analysis of data

Statistical analysis was performed using the Wilcoxon
matched pairs signed-rank test.
12
Outline
1.
Background and research questions
2.
Methods
3.
Results
4.
Discussion
5.
Conclusions
Results

Mean formant values of /u/, by speaker group
**
*
**
**
*
**
*
*
**
*
**
**
*
*
**
**
*
14
Results

Mean formant values of /ow/, by speaker group
**
*
**
**
*
**
*
*
**
*
**
**
*
*
**
**
*
15
Results

Distinctions made between fricatives, by speaker:
(1-3 = native, 4-9 = heritage, 10-12 = learners)
/ʂ/-/ʃ/
/ɕ/-/ʃ/
/ʂ/-/ɕ/
1 2 3 4 5 6 7
- - / - + / +
/ + / + + + +
+ + + + / + /
8 9 10 11 12
/ - + - + + + + +
+ + + + +
16
Outline
1.
Background and research questions
2.
Methods
3.
Results
4.
Discussion
5.
Conclusions
Discussion

The results of Experiment 1 show that:



Mandarin back vowels are more to the back than
English back vowels for all speakers.
In both languages, native Mandarin speakers’
back vowels are more to the back than HL
speakers’ and L2 learners.
????HL speakers tend to maintain the largest
difference in F2 between English back vowels
and Mandarin back vowels.
Discussion

Similar results were found in the fricative
experiment.


Most native speakers and most late learners
collapse Mandarin /ʂ/ and English /ʃ/.
Most heritage speakers keep /ʂ/ and /ʃ/ apart
on one or both spectral measures.
Discussion

The results of Experiment 2 show that:



All speakers distinguish voiceless and voiced
stops in English, and aspirated and unaspirated
stops in Mandarin.
Most native Mandarin speakers and HL speakers
have longer VOT for Mandarin aspirated stops
than English voiceless (aspirated) stops.
Less than half of the L2 learners show the same
distinction.
Outline
1.
Background and research questions
2.
Methods
3.
Results
4.
Discussion
5.
Conclusions
Conclusions


Our results suggest that heritage speakers tend
to be better at maintaining contrasts between
two “similar” categories in two languages.
Two possible explanations:



Early exposure to both languages makes heritage
speakers better at hitting the two targets.
Early-acquired categories interact with each
other and are dissimilated.
The first hypothesis is more supported by our
vowel data and VOT data.
Conclusions

Our results also suggest that there is a
correspondence in heritage speakers between
linguistic performance and amount of
exposure to the heritage language.
native speakers
most advanced
heritage speakers
intermediate
heritage speakers
late learners
least advanced
heritage speakers
Thank you!
Acknowledgements:
Sharon Inkelas
Keith Johnson
all speaker participants
participants in a seminar on phonological learning (UCB, Fall 2007)
UC Berkeley Linguistics
24
Selected References
Au, Terry K., Leah M. Knightly, Sun-Ah Jun, and Janet S. Oh. 2002.
Overhearing a language during childhood. Psychological Science 13(3): 238243.
Boersma, Paul, and David Weenink. 2008. Praat: Doing phonetics by
computer. http://www.praat.org.
Godson, Linda. 2003. Phonetics of Language Attrition: Vowel Production and
Articulatory Setting in the Speech of Western Armenian Heritage Speakers. PhD
dissertation, University of California, San Diego.
Knightly, Leah M., Sun-Ah Jun, Janet S. Oh, and Terry K. Au. 2003.
Production benefits of childhood overhearing. Journal of the Acoustical
Society of America 114(1): 465-474.
Ladefoged, Peter. 2005. Vowels and Consonants, 2nd edition. Malden, MA:
Blackwell Publishing.
Oh, Janet S., Terry K. Au, and Sun-Ah Jun. 2002. Benefits of childhood
language experience for adult L2 learners’ phonology. In B. Skarabela et al.
(eds.), Proceedings of the 26th Annual Boston University Conference on Language
Development, Vol. 2: 464-472. Somerville, MA: Cascadilla Press.
Oh, Janet, Sun-Ah Jun, Leah Knightly, and Terry Au. 2003. Holding on to
childhood language memory. Cognition 86(3): B53-B64.
25
Results

Mean F1 frequency
26
Results

Mean F2 frequency
27
Results

Mean F3 frequency
28
Descargar

Slide 1