Non-native rhythm does not block word segmentation in infants

Sundara, M. & Mateu, V.

UCLA Department of Linguistics

Infants' ability to segment words is affected by the prosody of their native language. We tested the extent to which word segmentation is tied to infants' familiarity with the rhythmic properties of their native language. According to Jusczyk et al.'s (1999) rhythm hypothesis, infants should segment words in a rhythmically-similar, but not a rhythmically-different language. Consistent with this hypothesis, monolingual English- and Dutch-learning 9-month-olds can segment two-syllable words in both (rhythmically-similar) languages (Houston et al. 2000). Further, monolingual English- and French-learning 8-month-olds cannot segment two-syllable words in the other (rhythmically-different) language (Polka & Sundara, 2012). These empirical findings are also consistent with an alternate explanation. Two-syllable words in English and Dutch are predominantly trochaic; however, they are likely to be iambic in French. Thus, English-learning infants' attested difficulties in segmenting iambic words (Jusczyk et al., 1999) may also account for their failure to segment two-syllable words in French, but not Dutch. We used the Headturn Preference Procedure to test segmentation of Spanish trochaic CVC.CV words by monolingual English-learning 8.5-month-olds. English and Spanish belong to different rhythm classes; however, disyllabic words of the form CVC.CV are predominantly trochaic in both languages. Infants were first familiarized with two passages (containing either gancho/salsa or gesto/venda, counterbalanced across infants) till they accumulated 45s (60s for Experiment 2) of listening time to each passage. During the test phase, infants were presented all four isolated word lists, two containing previously familiarized words, and two containing novel words. Listening time to the familiarized and novel words were averaged separately and compared statistically. Monolingual English-learning 8.5-mo-olds successfully segmented trochees in Spanish when familiarized with passages for 60s (16/21 infants, t(20)=2.9, p=0.008) but not 45s (7/20 infants, t(19)=-1.3, p=0.2). Thus, the unfamiliarity of the prosodic shape of the target word rather than non-native rhythm blocks word segmentation in infants.