This paper casts doubt on the existence of phonological ambisyllabicity and attempts to find its phonetic substances both temporally and spectrally. An intervocalic consonant becomes ambisyllabic when preceded by a stressed lax vowel, a theoretical apparatus to make the preceding stressed light syllable heavy by linking the intervocalic consonant to both preceding coda and following onset. In the experiment, temporal durations of ambisyllabic and non-ambisyllabic intervocalic consonants were compared, where consonants were varied with obstruents, nasals, and liquids. For a spectral measurement, lateral tokens were investigated because lateral /l/is the only consonant which shows a maximum allophonic difference between onset and coda (clear-[l] and dark-[ɫ]). Their F2-F1 values were compared among the laterals in ambisyllabic, non-ambisyllabic, word-initial onset, and word-final coda positions. Results showed that ambisyllabic and non-ambisyllabic consonants were not significantly different in duration for all three categories and that F2-F1 values of the laterals were not significantly different between ambisyllabic and non-ambisyllabic positions. This is not consistent with the phonological analysis. They were both distinctive from onset or coda laterals. This indicates that there might be a new allophony occurring in the intervocalic position where the preceding syllable is, whether its vowel is tense or lax, stressed.
1. Introduction
2. Previous approaches to ambisyllabicity
3. Hypotheses
4. Experiment
5. Conclusion