In this paper, several studies about infants' ability to discriminate languages between birth and 4-5 months were reviewed. In the rhythmic class acquisition hypothesis, infants' initial sensitivity to rhythmic classes would allow them to specify the common rhythmic properties of their native rhythmic class, and from this they would develop an associated metrical segmentation procedure. In addition,several studies about American infants' ability to extract word-like units from fluent speech were reviewed. Finally, several studies about Japanese infants' sensitivity to the typical rhythm pattern based on morae of Japanese baby words were reviewed.
