Reply to O’Seaghdha et al.: Primary phonological planning units in Chinese are phonemically specified

Abstract

Whereas O’Seaghdha et al. (1) generally agree with our argument that phoneme-sized segments are implicated in Chinese language production (2), they raise two critical points. First, they argue that our main event-related potential (ERP) finding (i.e., ERP modulation resulting from initial phoneme repetition) may reflect phonological connectivity rather than functional involvement of phonemes during speech planning. We believe, however, that our ERP effects cannot be accounted for via phonological connectivity within the proximate unit framework that is espoused by these researchers. Recall that in their framework (3), the phonological encoding stage is operationalized exclusively in terms of atonal syllables, which are indivisible units without any internal structure (ref. 3, p. 297). Phonemes are not activated until a later (“penultimate”) stage at which the tone and segmental context of an atonal syllable are specified, which directly precedes articulation of speech (which begins approximately 1,000 ms after stimulus onset in our study). Whereas the temporal location of our ERP effects (200 to 400 ms after picture onset) is highly compatible with current timing estimates of phonological encoding in Western languages (4), it is rather earlier than phonological connectivity effects could be expected. Theoretically, our findings are compatible with a model in which primary phonological planning units in Chinese are phonemically specified syllables (i.e., Chinese lexical representations accessed during phonological encoding are syllabified, phonemically specified representations). In English, primary phonological planning units are phonemes and representations accessed during phonological encoding are not syllabified.