汉语变体中分析型致使构式变异研究——多分类逻辑斯蒂回归建模

Chinese Analytic Causative Constructions and Their Lectal Variation:A Multinomial Logistic Regression

以往对于汉语分析型致使构式的研究较少关注其在语言变体维度上的变异。本研究在认知社会语言学的框架下,基于跨变体语料库,使用多分类逻辑斯蒂回归考察了影响致使词“使”、“令”、“让”选择的句法、语义、语言变体等因素。研究发现:1)“令”构式用于关系从句的概率显著高于其他小句类型,其结果谓词倾向于使用形容词,被使者常充当感事。2)相较于“令”,“使”和“让”都更偏好及物动词作为结果谓词、排斥充当感事的被使者。3)当被使者为定指或充当受事时“使”的概率显著提高;而“让”更倾向于充当施事的被使者,当被使者为客体时“让”的概率显著降低。4)“语言变体”与“小句类型”之间存在显著的交互效应:中国大陆地区汉语中“使”和“让”在不同小句类型中的概率差异更大;中国台湾地区汉语和新加坡汉语中“小句类型”对“使”和“让”概率的影响较小,且“让”的概率总是高于“使”;而三个变体在“令”的用法上未表现出显著差异。本文展示了认知社会语言学将语言内部和外部因素整合在同一框架内进行考察的优势,并在更多数据和参项的基础上丰富了汉语分析型致使构式与汉语变体变异的研究。

Lectal variation in Chinese analytic causative constructions has not yet been investigated in detail.In the framework of Cognitive Sociolinguistics,this paper employs a multinomial logistic regression model to examine the syntactic,semantic and lectal factors that may affect the alternation of Chinese analytic causative constructions with shi,rang and ling based on regiolectallybalanced corpora.Our analyses show that:1)the causee of ling-constructions tends to be an experiencer,followed by adjectives,and ling-constructions are significantly more likely to appear in a relative clause.2)Compared with ling,both shi and rang are more favored when the effected predicate is a transitive verb and more disfavored than ling when the causee is an experiencer.3)The probability of shi significantly increases when the causee is definite or serves as a patient,whereas rang favors the causee whose semantic role is an agent;The probability of rang significantly decreases when the causee is a theme.4)There is a significant interaction between Variety and ClauseType,i.e.the influence of ClauseType is strongest in Mainland Chinese,with shi and rang preferring different clause types,whereas in Taiwan Chinese and Singapore Chinese,rang is always favored.The usage of ling shows no significant cross-variety variation.This study not only shows the benefit of Cognitive Sociolinguistics for synthetically analyzing language internal and external factors,but also contributes to the research on Chinese analytic causative constructions and variation in Chinese varieties by analyzing large-scale data and incorporating considerable predictors.