First International Conference on Spoken Language Processing (ICSLP 90)
This study sought experimental support for the following two proposed collocational constraints for English: (1) that syllable peaks consist of a vowel plus optional sonorant and (2) that the second consonant of a coda must be a coronal. In our experiment, subjects were trained to identify vowel or coda substitutions in words containing only post-vocalic obstruents, then were tested on their ability to identify the same substitutions in words where the first post-vocalic element was a nasal or /I/, which was sometimes treated as part of the nucleus (tense vs. lax vowel) and sometimes as part of the coda (coronal vs. other consonant or cluster). The following significant differences emerged, but only on the two coda tasks: (1) nasals were, in general, more coda-like ('C-sticky') than /I/, (2) nasals and /I/ together were more coda-like after tense vowels than after lax vowels, and (3) they were also more coda-like before single coronal consonants than elsewhere, as expected from the theory.
Bibliographic reference. Derwing, Bruce L. / Nearey, Terrance M. (1990): "Real-time effects of some intrasyllabic collocational constraints in English", In ICSLP-1990, 941-944.