基于逻辑回归模型的send类与格交替的多因素分析

A multifactorial analysis on send-type dative alternation based on logistic regression model

本研究基于COCA语料库,采用语料库量化方法,通过逐步逻辑回归分析,建立语料库概率模型,探究口语语体中send类与格交替使用的制约因素及其影响效应,研究结果显示:在口语语体中,send类与格变体的使用受多种因素的影响;同时,在口语语体中,受事可及性、与事可定性并不具备显著影响效应,与事可及性、受事可定性、与事代词性、受事代词性、与事有生性、受事有生性、受事具体性以及长度差异对send类与格变体的选择具有显著影响效应;影响大小排序为:与事代词性>受事代词性>受事有生性>受事可定性>与事有生性>长度差异>与事可及性>受事具体性。

Based on COCA,this study adopts the corpus-based quantitative method. A corpus probabilistic model is established in order to explore the factors influencing the use of send-type dative alternation in spoken language by means of stepwise logistic regression analysis. The results show that the use of send-type dative alternation in spoken language is affected by many factors;in the meantime,accessibility of theme,definiteness of recipient,animacy of theme,animacy of recipient and length difference have no significant effect,while accessibility of recipient,definiteness of theme,pronominality of recipient,pronominality of theme and concreteness of theme have significant effect on speakers’ choice on variants of send-type dative alternation and present a hierarchy of these factors,that is,pronominality of recipient > pronominality of theme>accessibility of recipient>concreteness of theme>definiteness of theme.