Separating the redundancy of voicing from nasality in American English

This paper reports the results of a study designed to separate the voicing redundancy from nasality in American English and to determine their individual saliencies. Sixty-seven subjects judged the similarity of nonnasal pairs versus nasal pairs of CV and VC nonsense syllables on a 5-point magnitude-estimation task. The 80 stimuli were constructed to control and to demand subjects' cognitive strategies for coping with redundancy rather than experimental manipulation of the data. They were developed from the consonantal repertoire of / b d g m n p t k / paired with the vowels / i a u /, totaling 72 pairs of dyads. Eight additional stimuli of similar structure, testing multifeature and identity contrasts, were included to monitor the validity of the experimental procedure. Results indicated that (a) speech redundancy can be circumvented cognitively, (b) nasality was more salient (different) than voicing, and (c) a recency effect was found. These findings are discussed in reference to existing data.