Laryngeal mechanisms in speech production and perception: A cross-language study on F0 perturbation

我们使用跨语言声学、电声门图(EEG)和知觉数据研究了喉机制之间的相互作用。我们最重要的成果是对东京日语塞音发音对比的语音描述。我们发现，闭合发音在词首位置不同地产生，但在词中位置保持不同。重要的是，后元音的基频(F0)在很大程度上有助于区分词首位置上的两个塞音系列。这些结果与正在经历类似发展的语言进行了比较。然后，我们考察了母语和非母语的听者对日语发音对比的感知。我们使用合成刺激对讲日语和法语的听众进行了实验。结果表明，闭合发声是两个听者群体的主要线索，但日语听者对F0的敏感度高于法国听者，首尔韩语受试者在这一F0发展方面更为先进，先前的研究也报道了语音质量的作用。我们记录了说韩语的人的脑电数据，发现语音质量的实现具有很大的变异性。唯一的一致性是元音在送气和吐音后比在Fortis塞音后更令人呼吸。总之，我们已经证明，虽然一些自动机制解释了喉部行为之间的相互作用，但喉部对比的产生和感知取决于个体变异性以及语言特有的因素，如音系结构和韵律环境。

We investigated the interplay between laryngeal mechanisms, using cross-linguistic acoustic, electroglottographic (EEG), and perceptual data.Our most important achievement is the phonetic description of the voicing contrast of plosives in Tokyo Japanese. We showed that closure voicing is variably produced in word-initial position but maintained in word-medial position. Crucially, fundamental frequency (F0) of the following vowel contributes largely to the distinction between the two plosive series in word-initial position. These results were compared with languages that are undergoing a similar development.We then examined the perception of the voicing contrast in Japanese by native and nonnative listeners. We conducted experiments using synthesized stimuli, with Japanese- and French-speaking listeners. Our results showed that closure voicing is the primary cue for both listener groups, but the sensitivity to F0 is stronger with Japanese than French listeners.Seoul Korean is more advanced in this F0 development, and previous studies also reported a role of voice quality. We recorded EEG data with Korean speakers and found highly variable realization of the voice quality. The only consistency is that vowel is breathier after aspirated and lenis than after fortis plosives.Altogether, we have demonstrated that while some automatic mechanisms explain the interplay between laryngeal behaviors, the production and perception of the laryngeal contrast depend on individual variability as well as language-specific factors, such as phonological structure and prosodic contexts.