Reading segments is not reading words: Comment on Kawamoto et al. (2014)

Word production involves computing the sound of a word from its concept or visually presented form. To perform such process multiple operations are required, including lexical access, word form encoding, and articulation. With regard to articulation, a critical issue is understanding which is the minimal planning unit (henceforth MPU) to address articulatory programs and start motor implementation. Kawamoto and colleagues (2014, QJEP) have recently proposed that the MPU consists in the word initial segment. The segment as MPU hypothesis is winsome, but it is difficult to substantiate. In this commentary, we argue that, in addition to some methodological concerns, at least three issues may challenge this hypothesis. A first challenge to the segment as MPU hypothesis is how to reconcile it with the typical articulatory behavior shown by readers. A second issue questioning the generalizability of the segment as MPU hypothesis is the phonetic realization of stress in polysyllabic words, which in many languages constitute a great part of the lexicon. A third issue deals with the opacity of the orthography-to-phonology mapping. Overall, our observations suggest that the segment as MPU hypothesis is more an exceptional behavior than the behavior usually occurring at the phonology-to-phonetics interface of the reading system. We show that several findings suggest that the phonology-to-phonetics interface adopts units larger than the segment. The latter proposal can account for a bulk of phenomena such as those related to orthography-to-phonology opacity, stress assignment, and co-articulation.