Nasal coarticulation and sound-change: a real-time MRI study

点击翻译按钮获取中文摘要

Speech production is highly context-dependent because the sounds of speech overlap and influence each other in time. This is also true for sound sequences in which a vowel is followed by a nasal consonant in fluent speech: the velum lowering gesture that is associated with the nasal stop is initiated during the vowel, i.e. before the oral constriction for the nasal stop is achieved. Thus, in many languages, vowels are typically more nasalised before nasal than before oral consonants. In some languages like French, this type of phonetic variation has evolved historically into a nasal-oral contrast in the vowel combined with a deletion of the following nasal consonant. Fundamental to explaining the phonologisation of nasalisation is an understanding of how a nasal consonant's two primary articulators – the velum and the nasal consonant's oral constriction – become disassociated from each other, as in e.g. French main (‘hand’) /mɛ̃/ from Latin manus, in which the alveolar contact for the /n/ presumably initially lenited and was then lost as nasalisation in the vowel increased. The overall purpose of this project is to determine the phonetic conditions that could cause nasalisation to become increasingly associated with an oral vowel in a nasal context. Building on the fruitful results of the preceding proposal, this project will focus more specifically on the inter-gestural reorganisation of the tongue and the velum when phonetic conditions are modified, especially if the segment and word duration is decreased during speech due to different speech modes (read speech vs. spontaneous speech) or due to controlled speech rates (moderate vs. fast speech). The analysis will primarily focus on highly frequent function words containing post-vocalic nasals in which reduction is more likely than in less frequent content words. Real-time MRI will be used for recording up to 60 speakers of Standard German to track the movement of the velum in relation to other articulators, in particular the tongue tip. German is appropriate for such a study both because there are no known sound changes in progress involving nasals and because nasals occur in such a wide range of different contexts. For data acquisition, a high temporal resolution of 80 fps (12.5 ms) will be used, which is particularly advantageous for capturing details of small, fast tongue-tip movements whose analysis forms a key part of the renewal proposal. For the purposes of analysing sound change from time-varying speech data, the renewal proposal will make use of functional principal component analysis (FPCA), a method that expresses time-varying shapes of signals as a set of principal components which are time-varying functions. This project will lead to a deeper understanding of how the inter-gestural reorganisation of the velum and the tongue tip is affected by different speaking styles, sound contexts and word frequency, and how such variation provides the conditions for sound change.