| While most psycholinguists today acknowledge that prosodic and syllabic structure are part of the linguistic representations used to produce speech, how this structure is represented is still very controversial. Some psycholinguists have hypothesized that units corresponding to syllables are activated and concatenated in speech production (e.g., Fujimura and Lovins, 1978; Levelt 1989). Producing the word 'carpet', for example, would involve the activation of two units: one corresponding to the syllable 'car', and the other corresponding to the syllable 'pet'. According to one version of this hypothesis (Crompton, 1981), the language system would contain a listing of all the syllables of the language. Producing the words 'carpet' and 'Carmen' would involve retrieving from this syllable library the same unit 'car'.; It is possible, on the other hand, to hypothesize linguistic representations that consist of phoneme not syllable units, but that still represent prosodic and syllabic structure. According to a current linguistic approach, autosegmental phonology, linguistic representations consist of phonological segments that are structured into larger constituents by suprasegmental units corresponding to syllables and feet. For example, the word 'carpet' would consist of six phonemes linked to two structural syllable units.; A priming paradigm has been used to gather evidence for the two hypotheses above. It has been found that RTs to a target word are faster when it is preceded by a prime having the same prosodic and syllabic structure, even if prime and target share no phonemes. Sharing phonemes results in further priming. No difference is found, however, if the shared phonemes form a syllable unit or if they are shared across a syllable boundary. These results do not support a model of speech production that involves the activation of syllable units, like the one proposed by Crompton. On the other hand, they are consist with a model that distinguishes between segmental and suprasegmental units or, more generally, with any model of speech production that honors a distinction between phoneme-sized units on one hand, and the structure that orders and organizes these units on the other. Models along these lines have been recently proposed by Dell (1988), Stemberger (1985) and Shattuck-Hufnagel (1987). |
