Auditory nerve fiber representation of cues to voicing in syllable-final stop consonants.

Abstract

Acoustic cues to the identity of consonants such as /d/ and /t/ vary according to contextual factors such as the position of the consonant within a syllable. However, investigations of the neural coding of consonants have almost always used stimuli in which the consonant occurs in the syllable-initial position. The present experiments examined the peripheral neural representation of spectral and temporal cues that can distinguish between stop consonants /d/ and /t/ in syllable-final position. Stimulus sets consisting of the syllables /hid/, /hit/, /hud/, and /hut/ were recorded by three different talkers. During the consonant closure interval, the spectrum of /d/ was characterized by the presence of a low-frequency "voice bar." The closure interval for the voiceless consonant /t/ was longer and lacked a voice bar. Most neurons' responses were characterized by discharge rate decreases at the beginning of the closure interval and by rate increases that marked the release of the consonant closure. Exceptions were seen in the responses of neurons with characteristic frequencies (CFs) below approximately 0.7 kHz to syllables ending in /d/. These neurons responded to the voice bar with discharge rates that could approach the rates elicited by the vowel. The latencies of prominent discharge rate changes were measured for all neurons and used to compute the length of the "encoded closure interval." The encoded interval was clearly longer for syllables ending in /t/ than in /d/. The encoded interval increased with CF for both consonants but more rapidly for /t/. Differences in the encoded closure interval were small for syllables with different vowels or syllables produced by different talkers.