Title: A Multifactorial Analysis of Dative Alternation Based on Corpus-based Probabilistic Models: An Example of “Send”

Abstract: This study is a multifactorial analysis to explore the factors influencing the use of send-type dative alternation. Corpus-based probabilistic models are established after extracting and annotating the corpus data in semantic, pragmatic and syntactic aspects. The results show that in spoken English, significant factors influencing the use of send-type dative alternation include pronominality of recipient, pronominality of theme, animacy of recipient, definiteness of theme and definiteness of recipient. There are interactions among three factors, i.e., pronominality, animacy and definiteness. The relative importance of the predictors is ranked as: pronominality of recipient > length difference > animacy of recipient > pronominality of theme > definiteness of theme > definiteness of recipient > accessibility of recipient > accessibility of theme > animacy of theme > concreteness of theme. The choice of construction variants is proved to be systematic and probabilistic, and allostruction model theory is provided with further evidence. Also, corpus-based models established in this study possess good fitting ability and accuracy. The use of corpus-based probabilistic models can be regarded as a new attempt, having implications in the construction grammar study of dative alternation.

Keywords: dative alternation, send-type dative verbs, Tree & Forest model, naïve discriminative learning model

Authors: Yongxue Yuan, School of Foreign Studies, Jiangnan University, Wuxi, Jiangsu, China; Jianxue Li, Professor, School of Foreign Studies, Jiangnan University, Wuxi, Jiangsu, China.