poster 173asa tian 624 jkjiatian/posters/poster_173...multi-dimensionalphonetic space for phonation...
TRANSCRIPT
![Page 1: poster 173ASA Tian 624 jkjiatian/posters/poster_173...Multi-dimensionalphonetic space for phonation contrasts. Poster presented at LabPhon13in 2012. [6] Qian, N., (1992). Dangdai wuyu](https://reader035.vdocuments.us/reader035/viewer/2022070902/5f52231adb38ab4f2066b6b6/html5/thumbnails/1.jpg)
www.postersession.com
Methods
Conclusions
References and Acknowledgements
§Participants: 107 native speakers of SH in total. Those born before 1980 (N =52) still maintain phonation contrast.
§Speech Materials: monosyllabic words, 5 onset types (stop, fricative,affricate, nasal, zero) * 2 registers ([+Upper], [-Upper]) * 3 different items§Measures:Pitch: F0; Spectral: H1*-H2*, H2*-H4*, H1*-A1*, H1*-A2*, H1*-A3*;Noise: Cepstral Peak Prominence (CPP); Formant and bandwidth: F1, B1;Electroglottographic: Contact Quotient (CQ), Peak Increase in Contact(PIC), Speed Quotient (SQ).Within-speaker z-score normalizationwas done on each measure.
§Statistics analysis:§Multiple linear mixed-effects models to determine which measures differsignificantly between registers. (based on measurements taken from thethird interval where the contrast is the strongest)§Fixed effects: Reduced model: Gender + Register. Full model: Gender xRegister. Random intercepts: item, speaker. Random slopes: gender by itemrandom slope when it improves the model.§Logistic regression models on acoustic and EGG measures separately todetermine which of the measures are most useful. (based on measurementstaken from the entire syllable)
Results
v Goals of the study• Examine the acoustic and articulatory properties of the non-
modal phonation in Shanghainese• Make comparison with other languages
[1] Cao, J., & Maddieson, I., (1992). An exploration of phonation types in Wu dialects of Chinese. Journal of Phonetics, 20(1), 77-92.
[2] Chao, Y., (1928). Studies in the Modern Wu-Dialects. Peking: Tsing Hua College Research Institute.[3] Gao, J., (2016). Sociolinguistic motivations in sound change: on-going loss of low tone breathy voice in Shanghai Chinese.
Papers in Historical Phonology, Vol 1, 166-186.[4] Jiang, B., & Kuang, J., (2016). Consonant effects on tonal registers in Jiashan Wu. 90th Annual Meeting of LSA.[5] Keating et al., (2012). Multi-dimensional phonetic space for phonation contrasts. Poster presented at LabPhon 13 in 2012.[6] Qian, N., (1992). Dangdai wuyu yanjiu [Studies in the contemporary Wu dialects]. Shanghai: Shanghai Educational Publishing
House.[7] Ren, N., (1992). Phonation types and stop consonant distinctions: Shanghai Chinese. Ph.D. dissertation, The University of
Connecticut.[8] Rose, P., (1989). Phonetics and phonology of yang tone phonation types in Zhenhai. C.L.A.O.[9] Sherard, M., (1972). Shanghai Phonology, Ph.D. dissertation, Cornell University.[10] Xu, B., & Tang, Z., eds. (1988). Shanghai shiqu fangyan zhi [DocumentaDon of the Shanghai dialect in urban districts].
Shanghai: Shanghai Education Press.[11] Yip, M., (1980). The tonal phonology of Chinese. Ph.D. dissertation, Massachusetts Institute of Technology.[12] Zhang, J., & Yan, H. (2015). Contextually dependent cue weighting for a laryngeal contrast in Shanghai Wu. In Proceedings of
the 18th International Congress of Phonetic Sciences.This study is supported by an Upenn faculty research fund to Prof. Jianjing Kuang.
• CPP:
• CQ:
Register Tones(Chao numbers)
[+Upper] 53 34 5
[-Upper] 23 12
v The phonetic properties of the non-modal phonation in Shanghainese:• Lower pitch• Breathier phonation:
• Steeper spectral slope (higher H1-An values). However, H1*-H2*and H2*-H4* makes very little contribution to the contrast.• Higher noise ratio and less periodicity (smaller CPP).• Expanded bandwidth (B1).• Smaller Contact Quotient.
• CPP is the most important acoustic correlates of the non-modalphonation. Various strategies are adopted to make noise. The noisecomponent is potentially also salient in perception.
Phonetic properties of the non-modal phonation in Shanghainese register contrast Jia Tian, Jianjing Kuang
Department of Linguistics, University of Pennsylvania, [email protected];; [email protected]
• H1*-A1*
v Tonal registers• Related to pitch difference: Pitch range is divided into two
halves (Yip, 1980): [+Upper] vs. [-Upper].• Related to phonation difference: Non-modal phonation is
associated with the [-Upper] register (Chao, 1928; Sherard,1972; Qian, 1992, etc).
• Instrumental studies in the early 1990s found non-modalphonation:
• More recent studies found the loss of non-modal phonation:
• F0:v Shanghainese (SH)• Spoken in Shanghai, one of the largest cities in China.• Northern Wu dialect.• Five tones (Xu and Tang 1988).
v Tones in SH (Yip, 1980):
Acoustics `17 Boston, 25-29 June 2017
Introduction
Underlines denote short toneswith glottal stop in coda.
Studies Subjects Measures Findings
Zhang and Yan2015
Younger speakers(mean age 25)
H1*-H2*, H1*-A1*,H1*-A2*, H1*-A3*,Cepstral PeakProminence (CPP)
Younger speakersgenerally do notshow phonationdifferencebetween registers.Gao finds thatmany olderspeakers do notproduce breathyphonation either.
Gao 2016 12 young (aged20-30) and 10elderly (aged 60-80) speakers
H1-H2, H1-A1, H1-A2, CPP, F1, OpenQuotient (OQ)
Studies Subjects Measures FindingsCao andMaddieson 1992
Speakers bornin 1950s and1960s.
H1-H2, H1-A1,Airflow/pressureratio (AF/AP)
The [-Upper]register isbreathier.
Ren 1992 H1-H2, H1-A1.
Gender effect n.s. .
§Selected individual measures
• Other non-modal phonations (whisper, whispery, growl)were found in Zhenhai, a closely related Wu dialect (Rose,1989).
Female Male
First third Middle third Final third First third Middle third Final third
−0.5
0.0
0.5
1.0
Mea
n H
1*−A
1* (z−s
core
)
RegisterUpper
Lower
Female Male
First third Middle third Final third First third Middle third Final third
−1
0
1
Mea
n C
PP (z−s
core
)
RegisterUpper
Lower
Female Male
First third Middle third Final third First third Middle third Final third−1.5
−1.0
−0.5
0.0
0.5
1.0
Mea
n F0
(z−s
core
)
RegisterUpper
Lower
Female Male
First third Middle third Final third First third Middle third Final third
−1.5
−1.0
−0.5
0.0
0.5
Mea
n C
Q_H
T (z−s
core
)
RegisterUpper
Lower
• H1*-H2*:
**Female Male
First third Middle third Final third First third Middle third Final third
−0.6
−0.3
0.0
0.3
Mea
n H1
*−H2
* (z−
scor
e)
RegisterUpper
Lower
n.s. n.s.
*
Female speakers show larger F0 difference.
Gender effect n.s. .
** *
Strongest contrast in the middle third.Gender effect n.s. .
* *
Gender effect n.s. .
§Individual variation: different strategies to make noise in SH
More open glottis & more noisein the Lower register.
More restricted glottis & more noisein the Lower register.
Voiceless.Rose’s “whisper”.Also more noise.
§Cross-linguistic comparison: relative importance of acoustic measures
Male Female
1940 1960 1980 20001940 1960 1980 2000
−0.5
0.0
0.5
1.0
Speaker Year of Birth
H1*−A
1* (z−s
core
)
RegisterUpper
Lower
Male Female
1940 1960 1980 20001940 1960 1980 2000−1.5
−1.0
−0.5
0.0
Speaker Year of Birth
CPP
(z−s
core
)
RegisterUpper
Lower
• PIC:Female Male
First third Middle third Final third First third Middle third Final third
−0.75
−0.50
−0.25
0.00
0.25
0.50
Mea
n PI
C (z−s
core
)
RegisterUpper
Lower
*
• SH, Southern Yi & White Hmong are tonal. • Gujarati is not tonal. • SH and Southern Yi are register languages. • Spectral measures contribute the most in Southern Yi, Gujarati & White Hmong.
• Noise measure (CPP) is the most salient in SH.
• Noise measure contributes little in Southern Yi’s tense lax register contrast.
•Upper
•Lower
• Speaker 1 • Speaker 2 • Speaker 3
Only male speakers show PIC distinction.