The articulation of morphologically complex words

该项目第一个资助期的目标是调查编码形态信息的发音过程，以及它们与词汇熟练程度的关系。我们发现，后缀形态改变了语音编码，甚至在词干中，说话者变得更有能力做到这一点，随着熟练程度的提高。在与VAR项目的一项联合研究中，我们利用自然语言学习的方法对S的词末时长进行了计算模拟。我们的发现表明发音过程是由说话者学习语音信号和形态意义之间的关系的程度来调节的。ART项目的第二阶段旨在通过使用超声波录音复制EMA获得的结果并通过在KEC自发德语对话语料库中调查EMA录音来巩固在项目第一阶段获得的结果。超声波以及对自发语音清晰度的研究对于评估目前结果的生态有效性至关重要，这些结果都是在实验室条件下使用轻微侵入性记录技术获得的。该项目还将扩大其范围，研究大量变化的语言波兰语的发音，并将使用NDL的言语产生的计算模型扩展到例如新西兰英语以及德语的谐音指数。

The objective of the project for the first funding period was to investigate articulatory processes encoding morphological information, and their relation to lexical proficiency. We found that suffixal morphology changes the phonetic encoding even in the stem and that speakers become more capable to do so with increased proficiency. In a joint study with project VAR, we computationally modeled durations of word-final S by means of NDL. Our findings indicate that articulatory processes are modulated by how well the relation between the phonetic signal and morphological meaning is learned by speakers.The second phase of the ART project aims to consolidate the results obtained in the first phase of the project by replicating the results obtained with EMA using ultrasound recordings and by investigating EMA recordings in the KEC corpus of spontaneous German conversations. Ultrasound, as well as the study of articulation in spontaneous speech are essential for assessing the ecological validity of the present results, which were all obtained in laboratory conditions using a slightly invasive recording technique. The project will also broaden its scope by studying articulation in a richly inflecting language, Polish, and by extending the computational modeling of speech production with NDL to, for instance, New Zealand English, as well as to homophonic exponents in German.