White, L. , Delle Luche, C. , Jenkins, H. , Gunning, L. & Floccia, C.
School of Psychology, Plymouth University
The world's languages have been held to comprise a small number of distinct 'rhythm classes'. Although phonetic evidence is weak, patterns of language discrimination by listeners - infants and adults - have been taken as support for perceptually-based classes. For example, young infants have been shown to discriminate between rhythm classes, but not within them: English infants do not distinguish 'syllable-timed' Italian and Spanish nor 'stress-timed' Dutch and German, only distinguishing languages within a class when one of the pair is familiar to them. Furthermore, it is suggested that this apparent early sensitivity to rhythm class has consequences for the first language learner's segmentation of the speech stream into linguistic units. The rhythm class interpretation is challenged, however, by recent evidence showing within-class discrimination by adults exposed to stimuli modified to eliminate segmental information and focus solely on timing. Such results suggest that patterns of language discrimination may be based on gradient variation along prosodic dimensions, rather than on discrete categories. We used a head-turn preference paradigm to test whether English five-month-old infants could discriminate between French and Spanish. Although these are both held to be 'syllable-timed', they differ markedly in the realisation and distribution of strong syllables. We habituated infants to one of the two languages for at least 80 seconds and then exposed them to samples of new speakers, either from the same or the other language. Infants reliably showed differential patterns of looking to new and familiar languages, indicating that they perceived a difference. This within-rhythm class discrimination by young infants strongly suggests that their perceptual behaviour is not determined by intrinsic sensitivity to discrete classes. Rather, we argue that discrimination reflects functionally important, but gradient, differences in prosody, such as the distribution and phonetic marking of lexical stress. Studies to investigate this hypothesis further are ongoing.