14th International Congress of Phonetic Sciences (ICPhS-14)
San Francisco, CA, USA
Vowel-Consonant-Vowel (V1CV2) sequences are synthesized using a vocal-tract synthesizer. A deliberately simple phoneme concatenation by interpolation of their target area functions is employed to derive the time-varying area function, which is fed into the synthesizer. The sequences involve three vowels, /i/, /a/, and /u/, and stops. First, a uniform tube having a single constriction section at an appropriate place is used as the target area for the consonant. Listening tests indicate that when V2 is /u/, the synthesized [k] is not always satisfactory. Second, in order to improve the phonetic value of [k], we modify the consonantal target to take into account for the "anticipatory" effects of [u], such that the lip section during the k-closure is already rounded. After the modification, [k] is judged highly intelligible by all the listeners. It is concluded then that the libialization is not anticipatory coarticulation, but the auditorily required articulation for [k].
Bibliographic reference. Maeda, Shinji (1999): "Labialization during /k/ followed by a rounded vowel is not anticipation but the auditorily required articulation", In ICPhS-14, 41-44.