Speech Rate and Segmental Perception or the Role of Words in Phoneme Identification

Abstract

Publisher Summary Suprasegmental properties of speech, for example those related to intonation, rhythmical grouping and speech rate, form one class of factors causing acoustic variability of linguistically invariant units. Studies of the effects of suprasegmental properties of speech on the perception of discrete linguistic units may therefore in principle be used to gain insight into the general problem of how such discrete units are extracted from the variable speech waveform. By way of example, one particular class of such studies, those concerned with the effect of speech rate on phoneme perception, will be examined in some detail. This class of studies is typical of a much wider range of speech perception studies because of the unquestioned assumption that phonemes are immediate and natural response categories in speech perception tasks, and that studying the perception of phonemes is essential to the understanding of speech perception in general. In fact, it can be said that most perception researchers are eagerly trying to find out how phonemes are extracted from the speech waveform. This chapter argues that linguistic processing of speech, and particularly word recognition, is not mediated by phonemes, but that rather phoneme perception as studied in phoneme identification tasks is mediated by word recognition. If this is correct, it leads to interpretation of current data obtained in phoneme identification experiments. More importantly, it suggests that it may be high time to replace phonemes by words as the focus of attention in speech perception research.