The effect of segment duration on the perceptual integration of nasals for adult and child speech

Abstract

It has been hypothesized that the acoustic properties within a temporal domain of 10 to 30 ms of boundaries between speech sounds contain significant information on the phonetic features of segments, and that these cues are perceptually integrated by the auditory system [Stevens, Phonetic Linguistics: Essays in Honor of Peter Ladefoged (Academic, London, 1985)]. The purpose of the current research was to examine the effects of stimulus duration adjacent to speech sound boundaries on the perceptual integration of place of articulation of nasals before and after disruption of the abrupt changes in spectra between the murmur and transition. In experiment I, three children, aged 3, 5, and 7 years, and an adult female and male produced consonant-vowel (CV) syllables consisting of [m] and [n] in four vowel contexts, [i ae u a]. Approximately 25-ms segments of the murmur and vowel transition adjacent to the speech sound boundary were digitally removed from these productions. Intervals of silence ranging from 0 to 2000 ms, which can potentially perturb integration processes, were inserted between these segments. The stimuli were then presented to adult listeners for the identification of the nasal. The main findings revealed a consistent decline in identification with gap durations up to 150 ms across speakers and vowel context. However, the adult labial feature was resistant to perceptual change as a function of gap duration. This result appeared to relate to formant transition duration, and not to response bias. In experiment II, stimuli with durations shorter than those in experiment I were further analyzed for adult speakers. The main finding was a quantification of the acoustic segment duration needed for perceptual integration of the murmur and vowel transition. Across both experiments, the results reveal a decline in the identification of both alveolar and labial nasals within a time interval mediated by short-term auditory memory, and that the duration of the acoustic segment needed for perceptual integration is longer for [n] than [m].

Oscillatory neural network for pattern recognition: trajectory based classification and supervised learning

“…The inspiration for the experiments comes from by speech generation; frequency transitions are present in speech signals and are believed to be one of the fundamental features involved in word recognition (Gulick et al 1989). For example, it has been demonstrated in several experiments with frequency transitions over a short time interval, such as those found in consonant-vowel boundaries, that the 10 ms portion of the waveform prior to the stationary vowel segment contains the crucial information for proper syllable detection (Ohde and Ochs 1996). For example, it has been demonstrated in several experiments with frequency transitions over a short time interval, such as those found in consonant-vowel boundaries, that the 10 ms portion of the waveform prior to the stationary vowel segment contains the crucial information for proper syllable detection (Ohde and Ochs 1996).…”

“…For example, it has been demonstrated in several experiments with frequency transitions over a short time interval, such as those found in consonant-vowel boundaries, that the 10 ms portion of the waveform prior to the stationary vowel segment contains the crucial information for proper syllable detection (Ohde and Ochs 1996). For example, it has been demonstrated in several experiments with frequency transitions over a short time interval, such as those found in consonant-vowel boundaries, that the 10 ms portion of the waveform prior to the stationary vowel segment contains the crucial information for proper syllable detection (Ohde and Ochs 1996). Also, Stevens et al (1992) demonstrated the importance of the spectral changes between the consonant and vowel for use in classification.…”

Perception of the [m]-[n] distinction in consonant-vowel (CV) and vowel-consonant (VC) syllables produced by child and adult talkers

“…Several acoustic properties have been shown to contribute to the perception of place of articulation in adult and child speakers' production of nasal consonants ͑Kurowski and Blumstein, 1984;Repp, 1986;Repp and Svastikula, 1988;Ohde and Perry, 1994;Ohde and Ochs, 1996͒. The most distinctive acoustic properties of consonantvowel ͑CV͒ and vowel-consonant ͑VC͒ syllables include a low-frequency murmur representing sound passing through the nose and a relatively abrupt change in amplitude and spectrum at the juncture between the consonant and the vowel.…”

“…Results from studies on nasal consonants indicate that the perceptual mechanisms involved in processing spectral change may include early integration of physiological processes mediated by short-term adaptation ͑Delgutte, 1980; Delgutte and Kiang, 1984͒ and/or late integration of cognitive processes mediated by long-term auditory memory ͑Repp, 1987; Ohde and Perry, 1994;Ohde and Ochs, 1996͒. The adaptation effect in humans may involve the murmur in CVs adapting the following vowel onset by attenuating low resonance frequencies and enhancing second and third formant transitions, where distinctive place of articulation cues are located. The acoustic metrics that have been proposed to account for the perception of spectral discontinuities in nasal consonant+ vowel syllables include relational cues residing in the spectral change between the murmur and the vowel ͑Kurowski and Blumstein, 1987;Seitz et al, 1990;Kurowski and Blumstein, 1993͒ and the relational property of the combination of separately processed murmur and vowel spectra ͑Harrington, 1994͒.…”

“…The most distinctive acoustic properties of consonantvowel ͑CV͒ and vowel-consonant ͑VC͒ syllables include a low-frequency murmur representing sound passing through the nose and a relatively abrupt change in amplitude and spectrum at the juncture between the consonant and the vowel. Although listeners can identify place of articulation beyond chance from just the murmur ͑Kurowski and Blumstein, 1993͒, perceptual identification may be particularly accurate when based on tens of milliseconds of sound energy surrounding the point of spectral discontinuity between the consonant and vowel ͑Ohde, 1994; Ohde and Perry, 1994;Ohde and Ochs, 1996͒. In a study comparing the perception of nasals from murmur and vowel segments produced by three children and two adults, Ohde ͑1994͒ found that identification of place of articulation was more accurate from vowel segments than from murmur segments across the five speakers.…”