Study of spoken Sentence Grammar Integrating Syntactic and Phonological Processing

我们从自然语言处理和形式语法的角度研究了一种集语义、句法、形态和语音处理为一体的口语语法。随着研究的深入，我们发现，一方面基于EEG和fMRI的句子加工过程中的大脑活动观察，另一方面基于动态句法的建模是符合我们的目的的。因此，我们在研究的后半部分强调了这一方法论。基于这些方法，我们确认了日语句子是递增处理的假设，预测了改变当前输入的单词或短语。

We have investigated a grammar for spoken sentences which integrates semantic, syntactic, morphological, and phonological processing from the standpoint of natural language processing and formal grammar. As our research went on, we found that observations of brain activities during sentence processing based on EEG and fMRI on one hand and its modeling based on Dynamic Syntax on the other are appropriate for our purpose. Therefore we emphasized this methodology in the latter half of our research period. Based on these methods, we confirmed our hypothesis that Japanese sentences are processed incrementally, predicting the word or phrase that come alter the current input.