Font Size: a A A

Effects Of Speech Rate On The Acoustic Correlates Of Speech Rhythm

Posted on:2019-01-17Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y ZhangFull Text:PDF
GTID:1315330545979571Subject:English Language and Literature
Abstract/Summary:PDF Full Text Request
The existence of two rhythmic categories,stress-and syllable-timing,has been an active area of phonetic and phonological research on rhythm.However,empirical studies have failed to show evidence for a strict division into these two types of rhythm.This debate received renewed interest with the advent of rhythm metrics,formulas that seek to quantify consonantal and vocalic variability and use this quantification to classify languages rhythmically,such as %V and ΔC(the percentage of vocalic intervals in speech and the standard deviation of consonantal duration).Despite the popular acceptance of these metrics for rhythmic classification,these metrics actually fail to classify most languages.One factor that has not received enough attention in this debate is speech rate.Do differences in speech rate affect rhythmic metrics,and if so,how?The present study,therefore,has two main goals:(i)to find out the effects of speech rate on rhythm in different languages,and(ii)to assess the reliability of the different rhythmic measures that have been proposed in the literature,based on a comparison of existing descriptions with new data.Two experiments were conducted,involving British English,Shanghai Wu and Beijing Mandarin(the control group),and also of Shanghai-accented English and Beijing-accented English(the control group).Five Shanghai Wu speakers and four British English native speakers,and four Beijing Mandarin speakers(the control group)were recorded at three different speech rates(slow,normal and fast).We first tested the performance of the metrics for rhythmic classification in normal speech,and then tested the reliability of the metrics when speech rate was changed.On the one hand,the statistical analyses showed that speech rate has clear effects on rhythm.There is a significantly negative correlation between speech rate scores and the metrics data.Together with %V and ΔC,consonantal and vocalic PVIs(pairwise comparison indices for measuring consecutive consonantal and vocalic intervals)as well as consonantal and vocalic varcos(the normalization for ΔC and ΔV)exhibit a remarkable decrease as speech rate changes overall.This effect is different for slowerthan-normal speech than for faster-than-normal speech.The range of metrics like %V,consonantal PVI,consonantal varco,and ?C tended to be larger for the normal-to-fast changes than for slowed speech.In addition,consonants and vowels were affected differently by differences in speech rate,namely,consonantal metrics exhibit a larger range than that vocalic ones when the speech rate changes.Remarkable drops were found for consonantal PVI,consonantal varco,and ?C,compared with the vocalic ones.On the other hand,the metrics for British English and Beijing Mandarin(the control group)showed more consistency at different speech rates.Since these latter two languages are consistently classified as stress-timed and syllable-timed,respectively,the other languages failed to be rhythmically classified.The main conclusion is that metrics to classify speech rhythm should crucially be sensitive to speech rate.Other findings support the rejection of a strict division of language types as either syllable-timed or stress-timed.The different behavior of consonants and vowels under different speech rates deserves attention in further research,as well as phonetic/phonological effects such as elision,related to syllable structure.
Keywords/Search Tags:speech rhythm, rhythmic classification, speech rate, rhythm metrics
PDF Full Text Request
Related items