Dutch morphologically complex words: The role of morphology in speech production and comprehension

第一阶段的研究单位已记录的痕迹，一个词的形态结构中的声学信号在英语。这个项目的目标是获得更多的信息形态学的作用一般和形态语音痕迹，特别是在语音处理。到目前为止，我们对说话者为什么以及如何产生形态语音痕迹以及听者如何理解形态复杂的单词（有或没有形态语音痕迹）知之甚少。因此，我们无法设计完整的形态处理理论。在第一个子项目中，一名博士生将进行语料研究和实验研究的形态语音痕迹在几个寄存器。此外，这位学生将进行实验研究是否形态语音痕迹可能会影响语音处理，如果是这样，如何。这个子项目的重点是荷兰语，以扩大在语音处理的形态效应的证据，以另一种语言，它将因此提供更多的信息，这些影响之间的相似性，这些语言。第二个子项目调查的作用，形态一般，特别是形态的声学痕迹，在听觉词的理解，通过研究听觉词理解的计算模型需要具备哪些特性才能很好地模拟人类听者对形态复杂词的处理。重点将放在代表两种非常不同类型的架构的计算模型上：DIANA和朴素判别学习。该分项目的目的是模拟现有的几个词汇理解任务数据集中记录的人类听众的行为。

The first phase of the Research Unit has documented traces of a word’s morphological structure in the acoustic signal in English. The goal of this project is to obtain more information about the role of morphology in general and of morpho-phonetic traces in particular in speech processing. So far, we know very little about why and how the speaker produces morpho-phonetic traces and how listeners understand morphologically complex words, produced with or without morpho-phonetic traces. As a consequence, we cannot design full theories of morphological processing.The project consists of two subprojects. In the first subproject, a PhD student will conduct corpus research and experimental research on morpho-phonetic traces in several registers. In addition, this student will conduct experimental research on whether morpho-phonetic traces may affect speech processing, and if so how. This subproject focusses on Dutch in order to extend the evidence on morphological effects in speech processing to another language and it will thus provide more information about the similarity of these effects across languages.The second subproject investigates the role of morphology in general, and of morphological acoustic traces in particular, in auditory word comprehension, by investigating what properties a computational model of auditory word comprehension needs to have in order to well simulate human listeners’ processing of morphologically complex words. The focus will be on the computational models that represent two very different types of architecture: DIANA and Naïve Discriminative Learning. The subproject aims at the simulation of human listeners’ behaviour as documented in several existing datasets of word comprehension tasks.