Proceedings of the ... International Congress of Phonetic Sciences. International Congress of Phonetic Sciences最新文献

This study uses multitaper spectral analysis to examine the differences in consonants produced by patients who present with different dentofacial disharmonies (DFD) including severe overbites (Class II), underbites (Class III) and anterior open bites. Previous studies have found that patients with these malocclusion types all produce sibilants and plosives with increased spectral center of gravity and increased spectral spread relative to controls. This result is puzzling since some DFD groups differ from controls in opposite ways. To better understand the articulatory basis of these differences, we apply several spectral shape measures and find that all groups of DFD patients produce /s ʃ t tʃ/ with mid-frequency spectral peaks that are less prominent than those of the control group, but peak frequency measures are largely the same across all groups. This indicates that the DFD patients differ more in sibilant noise source than front cavity size.¹.

A common indicator of speech production disorders in children is a reduced ability to articulate complex syllables. Clinical studies of syllabic complexity of child speech have traditionally relied on phonetic transcription by trained listeners to characterize deviations in phonotatic structure. The labor-intensive nature of transcribing, segmenting, labeling, and hand-counting syllables has limited clinical analysis of large samples of continuous speech. In this paper, we discuss the use of a computer-assisted method, Automatic Syllabic Cluster Analysis, for broad transcription, segmentation, and counting syllabic units as a means for fast analysis of differences in speech precision when comparing children with and without speech-related disorders. Findings show that the number of syllabic clusters per utterance is a significant indicator of speech disorder.

In background noise, the amplitude fluctuations of speech commonly provide for momentary glimpses of high intensity portions of speech, predominantly from vowels. Previous investigations have provided glimpses of consonants or vowels to determine the perceptual contribution of different speech acoustics to sentence intelligibility. The present study investigated the consistency of perceptual contributions across eight American English dialects for a group of listeners from the southern United States. Results demonstrated that sentences preserving predominant vowel acoustics were consistently more intelligible across dialects for all participants. The significant contribution of vowels does not appear dependent on familiarity with properties of dialectal variation but may represent the preservation of more general acoustic cues important for sentence recognition. Acoustic analyses of temporal amplitude modulation suggest important cues present during vowels and highlight gradient differences across dialects associated with intelligibility.

This study examines the effects of determiner reduction and coarticulation on the perceived naturalness of resynthesized shock-the-geek (V-the-N) sequences. The determiner, equally spaced between monosyllabic V and N, was manipulated in 3 experiments along a 7-step continuum: (1) duration varied from 0.25x the original duration to 4x this duration; (2) amplitude varied from 55 dB to 85 dB; (3) schwa formants varied from completely overlapped with the vowel in V to completely overlapped with the vowel in N. Listeners rated V-the-N sequences with reduced duration and intensity and more anticipatory coarticulation more favourably than sequences with increased duration and intensity and more preservatory coarticulation. These results are consistent with a listener preference for the production of supralexical chunks that adhere to morphosyntactic rather than metrical structure.

The goal of the present study was to test the effects of working memory on speech production. Twenty American-English speaking adults produced syntactically complex sentences in tasks that taxed either verbal or spatial working memory. Sentences spoken under load were produced with more errors, fewer prosodic breaks, and at faster rates than sentence produced in the control conditions, but other acoustic correlates of rhythm and intonation did not change. Verbal and spatial working memory had very similar effects on production, suggesting that the different span tasks used to tax working memory merely shifted speakers' attention away from the act of speaking. This finding runs contra the hypothesis of incremental phonological/phonetic encoding, which predicts the manipulation of information in verbal working memory during speech production.

Thirty-two native American-English speaking children and adults reliably reproduced a non-contrastive segmental length distinction encoded in a set of monomorphemic nonce words. However, participants were unable to explicitly identify the contrast in a same/different judgment task. The results support the view that grammatically irrelevant, lexically specific temporal patterns can be learned and are represented in the lexicon.