The Identification and Discrimination of Synthetic Vowels

Abstract

A series of thirteen two-formant vowels was synthesized and used as the basis of labelling and discrimination tests with a group of English-speaking listeners. The sounds varied only in F1/F2 plot and the resulting vowel qualities were such that listeners found no difficulty in assigning each sound to one of three phonemic categories, those of the vowels in bid, bed and bad. The results of the tests were compared with those previously obtained in experiments involving the consonant phonemes /b, d, g/. It appears from the data that the phoneme boundaries in the case of the three vowel phonemes are less sharply defined than in the case of the stop consonants. The labelling functions for the vowels show a gradual slope and the discrimination functions do not show any marked increase in sensitivity to change in the region of the phoneme boundaries. It is clear also that the listeners were able to discriminate differences very much smaller than would need to be distinguished simply in order to place vowels in the appropriate category. The results show further that the effect of sequence or acoustic context in the perception of vowels is very considerable. In all the aspects examined in these experiments, the perception of synthetic vowels is found to be different from that of synthetic stop consonants. These differences lend some support to the hypothesis that the degree of articulatory discontinuity between sounds may be correlated with the sharpness of the phonemic boundaries that separate them.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 320 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.