Abstract:

The perception of voicing, manner, and place of articulation was investigated in a series of gating experiments in which subjects identified English consonants in six different duration conditions. The stimuli were created from natural speech consonant tokens excised from C[a] syllables. The consonants were then gated to include 10% to 50% of the total consonant duration from the onset, to which vowel babble was appended in order to make listeners less sensitive to the absolute duration of the stimuli. In experiment 1, subjects were presented with eight consonants [p, t, b, d, f, s, v, z] and results were analyzed in terms of correct identification for the following features: voiced or voiceless, stop or fricative, and labial or alveolar. In experiment 2, subjects were presented with two mutually exclusive sets of stimuli (subsets of the stimuli from experiment 1) such that the information for a particular feature was given (e.g., voicing given in the sets [p, t, f, s] and [b, d, v, z]). A comparison of the results of the two experiments will show whether certain feature information facilitates the perception of any other feature. Results will be discussed in relation to a psychological model of speech perception.