Effect of stimulus range on the identification of distorted /r/ in synthesized adult and child /r-w/ continua

The purpose of this study was to determine if a distorted /r/ category could be identified in /r-w/ continua for adult and child speakers and to examine the effects on distorted /r/ identification of truncating the /r/ range so as to approximate the range of sounds produced by speakers with defective /r/. CV stimuli from /r-w/ continua modeled after adult and child vocal tracts were presented to speech pathology students who were required to identify the stimuli as R, W, or distorted /r/(R?) in full range and truncated range conditions. The results were the following: (1) the ambiguous stimuli were identified as R? considerably below the criterion level established for acceptability in the full range conditions, (2) there was a decrease in R? identifications of the ambiguous stimuli in the truncated range conditions, and (3) there was a shift in the identification of ambiguous child stimuli from R? to W in the truncated range conditions. Along with previous findings, these results demonstrate a considerable influence of the range of stimuli and the number of contrasting stimuli on R? identification and cast doubt on the assumption that distorted /r/ can be distinguished from the /r/ sound it is intended to represent.