Introduction

Finding out the relations between utterance pattern and semantic meaning has long intrigued researchers in computational linguistics，computer speech processing and related fields.As many claimed that linking regularities across languages of such type of relations suggest that the interface between syntactic and semantics is part of architecture of language.For example，there was a claim that there are two types of intransitive verb，the unergative verbs and unaccusative verbs，the former has its syntactic pattern of D-structure with subject and no object or has an external argument but no direct internal argument，and the latter has a D-structure object but without a subject or have a direct internal argument but no external argument：

Unergative verb：NP［vpV］

Unaccusative verb：［vpV NP/CP］

Finding out the authentic utterance patterns is also importance step in word spotting techniques as the speech recognition's task is to recognised real time spoken language，the man-made utterance discounts the reliability in using these utterance patterns drew from nonauthentic data in speech recognition tasks.

In order to achieve the above two goals，this investigation was organised as in the following order：（1）using a concord software to list all utterances a typical word such as a verb，noun，adjective or adverb appears；（2）identifying and segmenting each word and phrase including idioms and usages from each utterance；（3）parsing and tagging each utterance to retrieve each word's syntactic status；（4）listing each word's utterance pattern according to their frequency.