Phoneme Similarity in English and Mandarin Chinese Language Production

Abstract Despite general agreement on the main stages of language production, the stage of phonetic encoding and articulation remains understudied. To shed light on this neglected component, more specifically the interaction of phonological units and articulatory features, this thesis studied the effects of initial phoneme similarity in controlled word-pair production experiments. I comprehensively examined the cost of different feature type differences in word onsets in parallel English and Mandarin Chinese word pair production experiments while also varying phonological contexts (presence or absence of shared rhyme segments). Monosyllabic word pairs were produced using a cued word-order procedure that requires the generation of phonology immediately before speaking. I predicted phoneme similarity costs for all features in both languages. The rhymed context should have more cost than the unrhymed context but keep the same pattern across features. Competition indexed by slower speech initiation latencies and increased errors was present in both rhymed (e.g., birch-perch, ?(bi3)-?(pi2)) and unrhymed (e.g., bog-pouch ?(bu4)-?(pa4)) related conditions in both languages. Surprisingly, for unrhymed word pairs, place was the most influential feature in both languages. For rhymed word pairs, all features were important, place remained the most robust effect in Chinese, but voice was the strongest feature in English. Error patterns were broadly consistent with the response time data. I conclude that the phoneme similarity effect operates both in a syllable-centric (Chinese) as well as a phoneme-centric (English) phonological system, suggesting a limited role for earlier production stages.