26
Cue-weighting in retroflex and dental perception by Hindi and Tamil listeners Sarah Bakst, UC Berkeley Phonetics and Phonology in Europe University of Cambridge June 29, 2015

Sarah Bakst, UC Berkeley Phonetics and Phonology in Europe University of Cambridge June 29, 2015

Embed Size (px)

Citation preview

Cue-weighting in retroflex and dental perception by Hindi and Tamil listeners

Cue-weighting in retroflex and dental perception by Hindi and Tamil listenersSarah Bakst, UC BerkeleyPhonetics and Phonology in EuropeUniversity of CambridgeJune 29, 2015

Retroflexes and Dentals

210 ms

-1 kHz--2 kHz--3 kHz--4 kHz--5 kHz-

a a a d a210 msThis study considers whether Hindi and Tamil speakers rely on the same acoustic cues to identify retroflexes and dentals.2Burst spectraretroflex

dental

Different distributionsHindiTamilword-initialar fear*word-initial (except borrowings)word-finalpe stomach; pe treeword-final (resonants)p girlword-mediallu ladoo (type of sweet)word-mediala sheepHypothesisPerceptual effect ~ phonological pattern?a. Tamil speakers will be less sensitive to burst cues than Hindi speakers.b. Tamil speakers will be more sensitive to vowel transitions that Hindi speakers.Effect not predicted by motor theory (Liberman 1985) and some versions of direct realism (e.g. Fowler 1986).allow native language to tune the model but5StimuliRecordings of Hindi and Tamil speakers

Isolation of aa and ada sequences

Cross-language and speaker comparison

Retroflex DentalSynthesisSynthesis of aa and ada using the Klatt synthesizerEqualized pitchInterpolated seven step continuum between the two stops.Five continua with altered cues

Synthesized StimuliVowel cues only (VC), no burst.Burst cues only (CV), no initial vowel.Ambiguous burst: VC + step 4 burst.Ambiguous vowel: step 4 vowel transition + CVMismatch: step 1 vowel + step 7 burst -> step 7 vowel + step 1 burst

Predictions for each continuum8ProcedureTwo-alternative forced-choice identification task in Open Sesame in sound-attenuating boothAll stimuli combined and randomizedTen repetitions of all stimuli (420 trials total)Break given just over halfway throughExperiment lasted about twenty minutes9Procedure

Procedure

ParticipantsMostly UC Berkeley students, some members of the community

Native speakers of Hindi (16) or Tamil (17)

$5 or extra credit in intro linguistics classModelingMixed-effects logistic model in R to predict probability of retroflex response.

One model for each continuum: response ~ step*language + 1|subResultsp-level: 0.0083 after Bonferroni correctionsTrending overall language effect:CV (p = .017)Interaction on particular steps:baseline (p = .04, marginal)VC (p = .017)ambiguous consonant (p = .0001)mismatch condition (p < .0001)

*

* * * **

InterpretationTamil speakers slightly more sensitive to burst cues

Hindi speakers slightly more sensitive to vowel cues

Opposite of specific prediction21DiscussionDifferences in number and type of phonotactic systems

Frequency of contrast and functional loadFrequency of minimal pairs or of phoneme.Tamil has more retroflex/plain segment types (stops and liquids).Higher ratio of retroflex to dental stops in Tamil than Hindi22ConclusionSmall effect of language experience on sensitivity to different acoustic cues

Phonetic knowledge is learned rather than innate.

Intended gesture unlikely to be the object of perceptionAcknowledgmentsMany thanks toProf. Keith Johnson for supervising the projectProf. Beth HumeEmily Cibelli, Clara Cohen, Andra Davis, and other members of the UC Berkeley Phonology LabUndergraduate assistants Shannon Foster and Akshayraj Aitha for running subjects and the Linguistics Research Apprenticeship Program for funding the projectSelected ReferencesBest C. T. (1995). A direct realist view of cross-language speech perception, in Speech Percep- tion and Linguistic Experience: Issues in Cross-Language Research, ed Strange W., editor. (Timonium, MD: York Press; ), 171204. Fowler, C.A. (2006) Compensation for coarticulation reflects gesture perception, not spectral contrast. Percept. & Psychophys. 68(2), 161177.Gerrits, E., and M. E. H. Schouten. 2004. Categorical perception depends on the discrimination task. Perception & Psychophysics 66:363376. Holt, Lori L., and Andrew J. Lotto. 2006. Cue weighting in auditory categorization: Implica- tions for first and second language acquisition. Journal of the Acoustical Society of America 119:30593071. Klatt, Dennis H. 1980. Software for a cascade/parallel formant synthesizer. Journal of the Acoustical Society of America 67:971995. Liberman, Alvin M., and Ignatius G. Mattingly. 1985. The motor theory of speech perception revised. Cognition 136. Lisker, Leigh. 1986. Voicing in English: A catalogue of acoustic features signaling /b/ versus /p/ in trochees. Language and Speech 29:311. Mathot, S., D. Schreij, and J. Theeuwes. 2012. OpenSesame: An open-source, graphical exper- iment builder for the social sciences. Behavior Research Methods 44:314324. Morrison, Geoffrey Stewart, and Maria V. Kondaurova. 2009. Analysis of categorical response data: Use logistic regression rather than endpoint-difference scores or discriminant analysis (L). Journal of the Acoustical Society of America 126:21592162. Steriade, Donca. 2001. Directional asymmetries in place assimilation: a perceptual account. In Perception in phonology, ed. E. Hume and K. Johnson. Academic Press. Stevens, Kenneth N., and Sheila E. Blumstein. 1975. Quantal aspects of consonant production and perception: a study of retroflex stop consonants. Journal of Phonetics 3:215233.

Palatography

Tamil

Hindi

TamilTamilmodel