摘要
依存方向是语序类型计量研究的重要指标。为探究基于特定句法关系依存方向进行跨语言语序类型区分的有效性及可行性,本研究使用跨语言平行树库及《世界语言结构图集》对20种语言中的高频句法关系开展计量调查。结果表明:1)主语、宾语和修饰语是跨语言平行树库中最为高频的句法关系;2)根据主语和宾语句法关系依存方向指标确定的优势语序与《世界语言结构图集》中的优势语序高度一致,体现了基于高频句法关系依存方向进行类型学研究的可能性。同时,以语言谱系分类为参照,主语和宾语句法关系的依存方向能够对语序类型进行粗粒度区分,而修饰语句法关系的依存方向则可对语序类型进行细粒度区分;3)随机森林算法对修饰语句法关系的重要性进行了预测与验证,进一步说明其在展现语言类型特征中的重要作用。上述结果证明了基于特定句法关系依存方向进行语序类型研究的可行性,并为基于依存语法框架进行应用语言学、自然语言处理等研究提供新视角与新参考。
Dependency direction is an important indicator for quantitative word-order typology.The present study aims to examine the feasibility of cross-linguistic typological classification based on the dependency directions of specific syntactic relations in 20 cross-linguistic parallel treebanks.The findings suggest that 1)In cross-linguistic studies,subj(subject),comp:obj(complement:object)and mod(modifier)are the most weighted syntactic relations;2)The dominant word orders determined by the dependency directions of main syntactic relations are close to what are retrieved from the World Atlas of Language Structures(WALS).More importantly,based on the dependency direction of mod(modifier),a fine-grained genealogical classification along a typological continuum can be achieved for all 20 languages;3)The random forest algorithm confirms the importance of mod(momidifier)in demonstrating the typological features across languages.These findings display the potential of utilizing the dependency directions of specific syntactic relations in typological studies,thereby contributing to the fields of applied linguistics,natural language processing,and beyond,under the framework of dependency grammar.
作者
阎建玮
刘海涛
YAN Jianwei;LIU Haitao
出处
《语言文字应用》
CSSCI
北大核心
2023年第2期79-90,共12页
Applied Linguistics
基金
教育部人文社会科学重点研究基地重大项目“数据驱动的外语能力发展研究”(22JJD740018)
中央高校基本科研业务费青年科研创新专项“基于语料库的跨语族语言语序共性研究”的资助。
关键词
依存方向
语序类型
句法关系
计量语言学
dependency direction
word-order typology
syntactic relations
Quantitative Linguistics
