The perception of rhythm in spoken and written language

Rhythm is perceptually salient to the listener. This claim is central to the research project briefly described below: a large-scale investigation of listening, in which the principal issue was how listeners segment continuous speech into words. Listeners must recognize spoken utterances as a sequence of individual words, because the utterances may never previously have been heard. Indeed, listeners' subjective impression of spoken language is that it may be effortlessly perceived as a sequence of words. Yet speech signals are continuous: there are only rarely reliable and robust cues to where one word ends and the next begins. Research on this issue in a number of languages prompted apparently differing proposals. In English, experimental evidence (Cutler & Norris, 1988; Cutler & Butterfield, 1992; McQueen, Norris & Cutler, 1994) suggested that lexical segmentation could be efficiently achieved via a procedure based on exploitation of stress rhythm: listeners assumed that each strong syllable was likely to be the beginning of a new lexical word, and hence segmented speech input at strong syllable onsets. Experimental evidence from French, in contrast (Mehler, Dommergues, Frauenfelder & Segui, 1981; Segui, Frauenfelder & Mehler, 1981) motivated a syllable-by-syllable segmentation procedure. Similar to the French findings were results from Catalan, and, under certain conditions, Spanish (Sebastian, Dupoux, Segui & Mehler, 1992). Cross-linguistic studies with French and English listeners (Cutler, Mehler, Norris & Segui, 1986,1992) established that these contrasting results reflected differences between listeners rather than being effects of the input itself; non-native listeners did not use the segmentation procedures used by native listeners, but instead could apply their native procedure—sometimes inappropriately—to non-native input.