On the locus of the syllable frequency effect in speech production

The observation of a syllable frequency effect in naming latencies has been an argument in favor of a functional role of stored syllables in speech production. Accordingly, various theoretical models postulate that a repository of syllable representations is accessed during phonetic encoding. However, the direct empirical evidence for locating the syllable frequency effect at this level, rather than at the phonological or motor programming levels, is scarce. To investigate the origin of this effect, we conducted six experiments involving immediate and delayed production, with or without an interfering task (articulatory suppression). Previous evidence from psycholinguistic and short-term memory studies allows the working hypothesis that this interfering task disrupts phonetic processing, while leaving phonological encoding relatively intact. Experiments 1 and 3 showed a syllable frequency effect in immediate pseudo-word production and picture naming, respectively. Experiments 2 and 4 required delayed naming (participants produced the items after a short delay, upon presentation of a response cue). The delay was or was not filled with articulatory suppression. The syllable frequency effect was not observed in standard delayed naming. By contrast, it was observed when the delay was filled with articulatory suppression. The effects for words and pseudo-words were highly similar. This pattern of results is interpreted as evidence that syllable frequency affects the stage of phonetic encoding. This interpretation is consistent with the previously postulated hypothesis that phonetic encoding involves the retrieval of syllable sized representations.