重音感知：以法語、波斯語及西班牙語使用者為例

Abstract

NA

Previous research has shown that perception is conditioned by native language experience in that listeners become attuned to the speech signal properties that are most relevant to their language. This effect has been found in segmental features (e.g. Goto 1971; Dupoux et al. 1999) as well as suprasegmental features (e.g. Wang et al. 1999). A classic example the latter is that native French speakers have difficulty in perceiving stress contrasts (Dupoux et al. 1997), whereas Spanish speakers are able to perceive stress contrasts as accurately as segmental contrasts (Dupoux et al. 2008). One difference between these two languages is that syllable prominence in French is predictable, whereas Spanish has lexically contrastive stress. The difficulty with perceiving stress has been termed stress “deafness” (Dupoux et al. 1997). Similar findings have been demonstrated in other languages with predictable stress, including Polish and Finnish (Peperkamp and Dupoux 2002). The aim of this work is to evaluate stress “deafness” in Persian and French listeners and compare it to previous findings of Spanish and French listeners. In particular, this work investigates the acoustic and perceptual cues of the three languages. The purpose of this research is to evaluate the importance attributed to each of the perceptual correlates of the acoustic cues of prominence to determine if the differences in first language affect the degree of stress “deafness”. French and Persian were chosen because they are comparable in terms of placement and predictability of stress, and listeners of both languages have demonstrated stress “deafness” (Dupoux et al. 1997; Rahmani et al. in prep), yet differ in terms of acoustic cues. In French, in phrase-final words, the final syllable carries a primary stress, which is realized with longer duration and higher intensity (Jun and Fougeron 2002). On the other hand, Persian stress involves both intensity and pitch as acoustic cues (Mahootian 1997). Four main acoustic cues to prominence identified in the literature include fundamental frequency, amplitude, duration and vowel quality (Sluijter 1995). These acoustic cues correspond to different perceptual cues: pitch, volume, length and timbre, respectively. As not every language utilizes the same acoustic cues in production, it is assumed that speakers of different languages listen for stress differently as well. In other words, the perception of stress may depend on the acoustic cues used in one’s native language. In fact, Frost (2011) provides experimental evidence that English and French native speakers listen for stress differently, indicating linguistic transfer. Thus, in order to understand perceptual differences in stress and the potential phonetic underpinnings of stress “deafness”, acoustic cues must be considered. Two experiments were designed in order to investigate the how different acoustic cues of stress may be perceived differently based on L1 background. The first experiment tested perception on a phonetic level. An AX experiment presents two three-syllable nonce word tokens with the same stress position. However, the cues available to the listen vary by the three aforementioned conditions (pitch, duration or both pitch and duration). Additionally, an identification task tested the phonological level of perception. In this task, participants were asked to identify which syllable of a trisyllabic nonce word is stressed (or more prominent). Available stress cues include: pitch, duration or both pitch and duration. The goal of this experiment was to determine if listeners are more sensitive to different stress cues or stress in different locations. Though the predicted language effects were found in these studies, they did pattern according to the predictions. Speakers of French and Persian, two languages which have shown stress “deafness” in the aforementioned studies, performed significantly different in ?many aspects. In fact, French speakers outperformed Persian speakers and performed similarly to the speakers of Spanish. These results motivate additional research questions that could inspire further research into how L1 phonetic features affect perception of phonological features.