ISCA Workshop on Plasticity in Speech Perception (PSP2005)

Senate House, London, UK
June 15-17, 2005

The Development of Positional Effects on Phonetic Discrimination

Jessica Maye

Northwestern University, USA

Infants' discrimination of speech sounds begins to show native language effects by 12 months of age. For example, Werker and Tees (1984) found that contrasts that are poorly discriminated by adult English speakers but well discriminated by English-learning 6-month-olds become difficult by 12 months. Infants also begin to show awareness of native language phonotactic patterns by 9 months (Jusczyk et al., 1993; Friedierici & Wessels, 1994). The present study examines the development of the combination of these effects: namely, phonotactically induced discrimination difficulty. Some discrimination difficulties arise on the basis of nativelanguage phonotactic constraints. For example, Dupoux and colleagues (1999) found that Japanese listeners have trouble discriminating *ebzo, which is phonotactically illegal in Japanese, from its legal counterpart, ebuzo (Dupoux et al., 1999). Maye (in prep) demonstrated that this difficulty is only true in a particular context: when the difficult part of the contrast is moved to a different context ([zo] and [uzo] were moved to the beginning of the sequence) Japanese speakers had no trouble discriminating. Likewise, a Japanese contrast that is difficult for American English speakers in word-medial position - gudo vs. guro - was discriminated well when the difficult element was moved to a different context. We presented 7- and 9-month-old infants from English-speaking and Japanese-speaking homes with the Japanese gudo~guro contrast. Infants were tested in a habituation paradigm, in which the dependent measure was looking time (LT). Trials were initiated when infants looked at a checkerboard stimulus presented on a computer monitor. On each habituation trial infants heard multiple tokens of gudo, produced by two female native Japanese speakers. Trials ended when the infant looked away from the visual stimulus for more than 2 seconds. When LT decreased to 50% of the initial LT, infants were presented with two test trials on which they heard multiple tokens of guro, produced by the same two speakers. At 7 months both language groups showed significant dishabituation (English, p < .01; Japanese, p < .01), but at 9 months only the Japanese infants discriminated the contrast (English, p = .13; Japanese, p < .05). These results demonstrate that the development of phonotactically based discrimination effects follows a similar time course to the development of both phonetic discrimination and phonotactic patterns. Future research will investigate the possibility that phonetic and phonotactic development arise from a single mechanism that tracks conditional probabilities of phonetic distributions.

Bibliographic reference.  Maye, Jessica (2005): "The development of positional effects on phonetic discrimination", In PSP2005, 252 (abstract).