ABSTRACT
Affricates have been observed to be problematic in phonological acquisition and disordered speech across languages, due to their relatively complex spatial and temporal articulatory patterns. Remediation of difficulties in the production of affricates requires understanding of how these sounds are typically produced. This study presents the first systematic articulatory and acoustic investigation of voiceless geminate affricate /ʧ/ in Kannada (a Dravidian language), compared to the palatal glide and the voiceless dental stop. Ultrasound data from 10 normal speakers from Mysore, India revealed that /ʧ/ is produced with the tongue shape intermediate between the palatal glide and the dental stop, and with the laminal constriction at the alveolar ridge. The observed articulatory differences are reflected in acoustic formant patterns of vowel transitions and stop/affricate bursts. Altogether, the results show that the Kannada consonant in question is an alveolopalatal affricate, supporting some of the previous descriptive phonetic accounts of the language and raising questions for further research on normal and disordered speech. The results and our survey of literature also suggest that affricates in South Asian languages tend to be phonetically variable and historically unstable compared to other consonant articulations.
Acknowledgments
The authors are grateful to two anonymous reviewers and the volume co-editor James Scobbie for helpful comments and suggestions. The authors also thank Paul Arsenault, Midula Kasim, Mahesh B. V. M., Tim Bressmann, R. Manjula, Keren Rice and S. R. Savithri, as well as to the audience of the 2015 Summer Phonetics/Phonology Workshop at the University of Toronto.
Funding
This work was supported by grants from All India Institute of Speech and Hearing and Social Sciences and Humanities Council of Canada.
Declaration of interest
The authors report no conflicts of interest.
Notes
1 The linguistic situation in India is very complex, with bilingualism and multilingualism being the norm rather than an exception in many locations (Bhatia & Ritchie, Citation2013). In primary and secondary schools in Karnataka, the medium of instruction is either Kannada or English, with Hindi (the national language), Tamil or Telugu (neighbouring state languages) occasionally being taught as additional school subjects. Influences of these languages on the Kannada production of our participants cannot be discounted, yet these can be considered part of the normal linguistic behaviour of the population.
2 The item atta was also used as a control in a study of Kannada retroflexes (Kochetov et al., Citation2014), data for which was collected during the same experimental session.
3 The segmentation of the words with /j/ was less straightforward, due to the continuous nature of the vowel-glide transitions, and therefore duration results for ayya (/ajːa/) should be considered with caution.
4 Release duration for English /ʧ/ was on average 74 (8) ms for the female and 103 (9) ms for the male speaker. Burst COG values were 4123 (266) Hz for the female and 3316 (137) Hz for the male; the frication COG values were 3573 (204) and 3264 (126) Hz, respectively. Relative intensity values for frication were −1.17 (2.16) dB for the female and 2.20 (1.89) dB for the male speaker.