Speech Units Workshop

The idea that languages can be classified according to different rhythms has been a topic of research at least since Pike (1945). Three rhythmic classes have been considered, based on 3 (hierarchically organized) rhythmic units: the (trochaic) foot/stress unit, the syllable and the mora. The present talk will explore how rhythmic classes might play a role in early language acquisition. In a first part, I will present work on word-form segmentation, that is, the ability to extract potential word-forms from fluent speech. Many studies have shown that this ability emerges around 6 months of age in many languages, and have established that infants use a combination of cues that partially mark word boundaries. While some of these cues might be language-general (transitional probabilities), many of them are language-specific (rhythmic cues, phonotactic cues, knowledge of known words…). I will present work on the developmental trajectory of early word segmentation in French, a syllable-based language, in reference to what is known for early word-form segmentation in stress-based English. The focus will be on the use of language-specific rhythmic units in segmentation. In a second part, I will present data on cortical entrainment, exploring whether the adult and infant brain process differently rhythmic units (stress-, syllable- and mora-based units), comparing 3 linguistic populations with native languages of different rhythmic type: English, French and mora-based Japanese respectively. Taken together, these studies will highlight both the commonalities and the specificities of the emergence of word-form segmentation across languages, and the possibility of a default syllable-based rhythm.