音節結構與音調於母音長度感知之影響

Abstract

先前的研究指出，聲調差異會影響母音長度的感知，聽者會將口語產出長度較長的聲調感知為較短的聲調(Gussenhoven & Zhou, 2013)。因此，許多學說相繼提出不同的理論來解釋母音的感知與口語產出長度之關係，例如矯枉過正(hyperarticulation; Yu 2010)、感知補償性機制(compensatory listening strategy; Gussenhoven 2007)，以及聲調群聚(tone crowding; Sundberg 1973)。不僅如此，先前的研究發現，由於聲調與阻塞音有無送氣的差異在華語中會造成語意上的不同，而在荷蘭語中則不會，因此以華語為母語的使用者相較於以荷蘭語為母語的使用者而言，在感知母音的長度時更容易被口語產出之長度差異與音首差異影響。其顯示了母語之間的聲調及音段差異皆會影響母語使用者對於母音長度的感知(Gussenhoven & Zhou, 2013)。 此篇研究以兩項感知評聽實驗(perceptual rating experiments)來探討台灣閩南語與台灣華語使用者對母音長度的感知情況。結果顯示，聲調的感知長度無法完全僅以單一學說來解釋，而是與口語產出之長度相似(mimicking effect)以及與聲調群聚多寡(tone crowding; Sundberg 1973)相關。除此之外，音段差異也會影響母音長度的感知。當音首為鼻音([m])時，母音感知長度會比音首為阻塞音([b])時來得長(達顯著差異)，顯示母音前的子音若為響音([m])時，會影響使用者對於母音長度的感知。當母音後帶有音尾時，母音感知長度會比母音後無音尾時還要長(亦達顯著差異)，顯示音尾的語音長度也會影響使用者對於母音長度的感知。 綜言之，本論文探討台灣閩南語與台灣華語之聲調與音段方言差異如何影響使用者對於母音長度的感知，而基本頻率、音首與音尾的差異皆為影響的因素。

Previous studies have shown that the shape of pitch contours has an influence on perceived vowel duration in that listeners tend to perceive vowels as shorter even when they are produced longer (Gussenhoven & Zhou, 2013). Several hypotheses have thus been proposed to account for the correlation between the perceived and produced vowel lengths, including hyperarticulation (Yu 2010), compensatory listening strategy (Gussenhoven 2007), and tone crowding (Sundberg 1973). Furthermore, Gussenhoven & Zhou (2013) have also reported that vowels of different duration steps and with different onsets (i.e., aspirated vs. unaspirated stops) were perceived as more different by Chinese listeners, whose native language uses pitch and [+spread glottis] contrastively, than by Dutch listeners, suggesting an effect of tonal and segmental contrasts in one’s native language on one’s perception. In this study, two perceptual rating experiments were used to investigate whether tonal and segmental differences would influence the perceived vowel duration by listeners of different native languages (i.e., Taiwanese Southern Min vs. Mandarin Chinese). Findings from these experiments showed that the pitch patterns are not readily explainable by a single strategy. Instead, perception often mimics actual production length (i.e., mimicking effect) and syllables with more tones are in general perceived as longer than syllables with fewer tones (i.e., a tone crowding hypothesis). Segmental differences are shown to also affect vowel perception length. Vowels with nasal [m] onset were perceived significantly longer than those with voiced obstruent onset [b], suggesting that vowels with a sonorant onset (i.e., [m]) may contribute to the perception of vowel duration. Participants gave higher rating scores to vowels with codas than to vowels without codas, suggesting that coda duration is counted towards perceived vowel duration as well. An explanation from processing these stimuli in the auditory mode is provided. Taken together, the results of this thesis show that vowel perception length may depend on fundamental frequency, onset type, and coda type.