Research has shown that infants perceive many aspects of isolated speech sounds in an adult-like fashion. Little is known, however, about whether infants also organize their perception of continuous speech by segmenting speech into temporally ordered """"""""units"""""""" or segments (e.g., phonetic segments or syllables) resembling those of adults. If infants temporally segment speech at some linguistic level, these segments would be denumerable units. An infant's representation of a particular utterance would therefore contain a particular number of these units. If it could be shown that infants can discriminate among utterances or can match them on the basis of the number of units they contain, then segmental competence could be more justifiably attributed to infants. The experimental method we have developed incorporates a numerical ability known to be present during infancy. We are using this ability as a methodological tool to reveal segmental ability. This method has made it possible for us to collect pilot data which strongly suggest that infants can detect the number of syllables in an utterance. The objective of our proposal is to conduct a more extensive and systematic investigation of infants' segmentation of synthetic and natural speech at the level of syllables and the level of phonetic segments. This investigation will have a twofold significance: (a) It will further delineate the language processing capabilities of infants, and (b) it also will further specify the numerical processing capabilities of infants.