The Phone(cs of Rare Sounds: Produc(on and Percep(on of Aspirated Frica(ves in Sgaw Karen Hugo Salgado [email protected] Jessica Slavic [email protected] Berkeley Linguis6cs Society 40 th Annual Mee6ng February 8, 2014
The Phone(cs of Rare Sounds: Produc(on and Percep(on of
Aspirated Frica(ves in Sgaw Karen Hugo Salgado [email protected]
Jessica Slavic [email protected]
Berkeley Linguis6cs Society 40th Annual Mee6ng February 8, 2014
Overview of Sgaw Karen
2
• Member of Tibeto-‐Burman group of the Sino-‐Tibetan languages (Benedict 1972 )
• Spoken by 1,480,000 people in Myanmar and Thailand (Ethnologue, 2012)
• Relevant phonological inventory: /p t k/, /ph th kh/, around 6 tonemes and 9 vowels
• 2 coronal frica6ve phonemes with contras6ve aspira6on: /sh/ and /s/
• Only 3 languages in the UPSID database of 451
languages feature phonemic /sh/ (Jacques 2011)
Myanmar (Burma)
Thailand
Karen State
Research Overview
• Produc6on of /sh/ – The aspira6on feature on frica6ves behaves differently than aspira6on on stops.
– Aspira6on and frica6on seem to compete for ar6culatory 6me.
– Aspira6on on frica6ves is subject to diminu6on.
• Percep6on of /sh/ – Speakers rely on different perceptual cues for aspirated frica6ves than they do for aspirated stops.
– Perceptual cues for aspirated frica6ves depend on vowel height.
3
Part 1: Produc(on of Aspirated Frica(ves Research Findings • Aspira6on (VOT) on frica6ves is significantly less than stop
aspira6on. • Higher vowels increase stop VOT but decrease frica6ve VOT. • Higher vowels increase frica6on dura6on. • Nega6ve correla6on between frica6on and aspira6on dura6on
(Salgado, Slavic & Zhao 2013). – Frica6on dominates aspira6on dura6on in frica6on-‐lengthening
contexts (i.e. before high vowels).
4
Methodology • 4 na6ve speakers of Sgaw Karen
– 3 female, 1 male; mean age: 28
• Targeted 6 stops [p t k ph th kh] and 2 frica6ves [sh s] • Segments followed by 5 vowel contexts:
– high [i u], mid [e o], and low [a] – represented by 2 tones (tone does not seem to effect aspira6on)
• 220 randomized s6muli presented in the Karen script
5
Acous(c Analysis in Praat
[s] [sh]
Coronal frica6on: turbulent airflow of coronal origin Aspira6on: turbulent airflow of glojal origin (Johnson 1997)
frica6on frica6on aspira6on
6
Aspirated frica(ve VOT very short
7
Stop aspira6on (VOT) significantly greater than frica6ve aspira6on. Frica6ve aspira6on (VOT) near percep6on boundary for stops, Johnson (1997).
High vowels increase frica(on dura(on; frica(on the force behind short VOT
No significant difference between total dura6on of /sh/ and /s/. Aspira6on unstable – subject to diminu6on.
9
Perceptual Implica(ons
• VOT in /sh/is significantly shorter than in aspirated stops.
• VOT in /sh/ is actually closer to that of unaspirated stops.
• Is VOT the main cue to iden6fy aspirated/unaspirated frica6ves in Sgaw Karen?
10
Part 2: Percep(on of Aspirated Frica(ves
Research Findings
• Cues to iden6fy /sh/ are different from cues to iden6fy aspirated stops.
• Speakers rely on mixed cues to dis6nguish /sh/ and /s/ depending on vowel context.
• VOT used to iden6fy /sh/ in low vowel contexts. • Vowel quality used to iden6fy /sh/ in high vowel contexts. – c.f. aspirated affricates (Clements & Kha6wada, 2007)
11
2 tests to determine perceptual cues
• VOT con(nuum test – Aspirated stops and frica6ves – 3 VOT Levels:
• short (near 0 ms) • medium (50 ms)/(28ms for /sh/) • long (100ms)/(54ms for /sh/)
– 2 Vowel Heights: /a/ and /i/
• Conflic(ng cues test – Cross-‐splice stops/frica6ves
and vowels of different categories
– 2 Vowel Heights: /a/ and /i/
12 12
Rarity of aspirated frica(ves explained?
• Difficult to produce: VOT in aspirated frica6ves is significantly shorter than VOT in aspirated stops.
• Aspira6on and frica6on compete for ar6culatory 6me.
• VOT in frica6ves is further shortened before high vowels.
Rarity of aspirated frica(ves explained?
• Difficult to perceive: VOT is the main cue to iden6fy aspirated/unaspirated stops.
• Percep6on of aspirated frica6ves depends on mixed cues: VOT in low vowel contexts, vowel quality in high vowel contexts.
Future Research
• What quali6es in the following vowel are relevant for aspira6on percep6on? – Breathy voice? (H1, H2) – Vowel length
• Are there cross-‐linguis6c allophonic rules that treat aspira6on differently in varying vowel contexts?
20
Thank you!
• We are deeply grateful for the generous assistance of our Na6ve Speaker Consultants.
• Special thanks to David Mora Marín, Jennifer Smith and Ellioj Moreton.
• Special thanks to our UNC colleagues in P-‐Side and K-‐Side.
21
References Blevins, J. (2004). Evolu2onary phonology: The emergence of sound pa;erns. Cambridge: Cambridge
University Press.
Clements, G.N., & Kha6wada, Rajesh. (2007). Phone2c realiza2on of contras2vely aspirated affricates in Nepali. Proceedings of the 16th Interna6onal
Congress of Phone6c Sciences (ICPhS 16). Chaida, A., & Nirgianaki, E., & Fourakis, Marios (2009). Temporal characteris6cs of Greek frica6ves.
Paper for the 9thInterna6onal Conference on Greek Linguis6cs, Chicago, Illinois.
Ethnologue hjp://www.ethnologue.com/show_language.asp?code=ksw Jacques, G. (July 01, 2011). A panchronic study of aspirated frica6ves, with new evidence from Pumi.
Lingua, 121, 9, 1518-‐1538.
Jongman, A., Wayland, R., & Wong, S. (January 01, 2000). Acous6c characteris6cs of English
frica6ves. The Journal of the Acous2cal Society of America, 108, 3, 1252-‐63. Johnson, K. (1997). Acous2c and auditory phone2cs. Cambridge, Mass: Blackwell Publishers. Kim, H. (January 01, 2001). A phone6cally based account of phonological stop assibila6on
Phonology, 18, 1, 81-‐108. Kuan-‐Yi, C., & Li-‐Mei, C. (June 01, 2008). A Cross-‐Linguis6c Study of Voice Onset Time in Stop
Consonant Produc6ons. 215-‐231. Salgado, Slavic & Zhao. (2013). The produc6on of aspirated frica6ves in Sgaw Karen. Studies in the Linguis2c Sciences: Illinois Working Papers. 148-‐161. Yavas, M. (2009). Factors influencing the VOT of English long lag stops in Interlanguage phonology, in M. Watkins, A. Reuber & B. Bap6sta (Eds.) Recent
Research in Second Language Phone2cs/Phonology: percep2on and produc2on. Cambridge Scholars Publishing. 244-‐255.
22
Conclusions
• Difficult to produce: VOT in aspirated frica6ves is significantly shorter than VOT in aspirated stops.
• Aspira6on and frica6on compete for ar6culatory 6me. • VOT in frica6ves is even shorter before high vowels. • Difficult to perceive: VOT is the main cue to iden6fy
aspirated/unaspirated stops. • Percep6on of aspirated frica6ves depends on mixed cues:
VOT in low vowel contexts. Vowel Quality in high vowel contexts.
23