Priyankoo Sarmah and Charles Redmon (2013)
Paper presented at Acoustics 2013 in New Delhi
Earlier work on Korean and two languages in Northeast India, Bodo and Rabha, suggested that this cross-linguistic set could be a good test case for the organization of vowel spaces as all three languages (including 3 dialects of Korean) exhibit similar systems (6 vowels for Bodo and Rabha, 7 for Korean) with a high-central vowel of some sort. In particular we wanted to test the prediction from dispersion theory that vowels will show a tendency toward maximal separation of categories in formant space, and since the other 5-6 (non-high-central) vowels are well separated in all three languages, answering this question crucially hinged on the placement of the high-central vowel in that space. What we found was clear separation of the high-central vowel in Bodo along the F2 dimension (that is, between /i/ and /u/), but for Rabha and Korean the placement of the high-central vowel was not so clear, and generally tended to appear fronted with some overlap with /i/ (in Rabha the situation is more complicated, as speakers produce these vowels anywhere from a high back unrounded vowel to a high front lax vowel). These results complicate expectations from dispersion theory, and suggest articulatory constraints may interact with perceptual constraints to yield a vowel space which is less than optimal from a dispersion-theoretic perspective.