This work examines cue weighting in production and perception of the four-way laryngeal contrast in Hindi and Urdu. Previous work has consistently identified several cues, including prevoicing (duration), aspiration (duration), voice quality, and f0, that are relevant to the contrast, although the phonetic specification of the contrast, and particularly the status of the so-called “voiced aspirates,” remains unclear. In this work, we confirm the importance of prevoicing and aspiration to the contrast overall, but argue that voice quality (murmur or breathy voice) best distinguishes the voiced aspirates in production. In perception, listeners make use of all cues, in line with production patterns. Tokens in which concurrent prevoicing and aspiration are categorically identified as voiced aspirates, indicating that the joint presence of these two cues is sufficient for voiced stop identification and demonstrating the primacy of these features over all of the others tested. At the same time, neither prevoicing nor aspiration is strictly necessary for voiced aspirate identification; a stop token be perceived as a voiced aspirate even when one of these is absent, as long as the breathy voice quality also characteristic of voiced aspirates is present. We attribute the disproportionately large perceptual category space for voiced aspirates to the variability of voiced aspirates in production.

这项工作研究了印地语和乌尔都语中提示音权重的产生以及对四向喉镜对比的感知。尽管对比的语音规范，特别是所谓的“状态”，以前的工作一直一致地确定了与对比相关的几个线索，包括发声（持续时间），愿望（持续时间），语音质量和f0。吸气”，目前尚不清楚。在这项工作中，我们确认了发声和抽吸对于整体对比度的重要性，但认为语音质量（杂音或呼吸声）最能区分生产中的有声抽吸。在感知方面，听众会根据生产模式使用所有线索。将同时发声和误吸归类为浊音的代币，足以用于语音停止识别，并证明了这些功能相对于所有其他测试的首要地位。同时，明确的吸出物识别既不需要发声也不需要吸取。即使不存在停止记号，只要不存在其中之一，只要存在也具有浊音特征的呼吸音质即可。我们将有声抽吸物的不成比例的较大感知类别空间归因于有声抽吸物的生产变化。