Perception and production of misarticulated (r)

Twelve children who consistently misarticulated consonant [r] and five children who correctly articulated [r] were recorded while repeating sentences which differed only in a single (r)-(w) contrast. All (r) and (w) productions were spectrographically analyzed. Error productions were judged for their similarity to [w]. Each child identified all of the recorded sentences via a picture-pointing task. Misarticulated [r] was identified as (w) at above chance levels only by the children who did not misarticulate [r]. The subject groups did not differ in their perception of correctly articulated (r) and (w) phones. Children whose misarticulated [r] phones were judged to be (w)-like were most likely to misperceive their own productions of (r). Children whose misarticulated [r] productions were characterized by higher second formant frequencies were better able to identify their productions of (r). Results suggest that a subpopulation of children who misarticulate [r] may mark it acoustically in a nonstandard manner.