摘要
语言学家现已建立了不同精细度的语义角色分类体系,有些体系已应用到为自然语言处理服务的语义知识工程中。然而由于基于不同的语言学理论,这些语义资源所设立的语义角色类别和数目差异较大,很难做到信息共享。本研究回归语言学本体研究,从各类语义角色的界定与命名、双重语义角色以及“一句一格”原则等三个方面对语义角色研究中的根本分歧展开论述,提出对语义角色的界定应同时考虑形式和意义两个方面,将语义角色的界定及鉴别建立在谓词的基本句法构造之上。
Linguists have proposed various sets of semantic roles with different levels of granularity, some of which have been applied to the annotation of large corpora. However, for lack of unified criteria that are generally acceptable, existing semantic resources are rather subject in defining and identifying semantic roles, hindering information sharing. This paper revisits the theoretical debates on semantic roles and discusses the definitions of some problematic semantic roles, dual semantic roles and one-case-per-clause principle. We propose arguments to consider both form and meaning when defining and identifying semantic roles and take the basic syntactic constructions as the formal criterion.
出处
《现代语言学》
2018年第5期764-771,共8页
Modern Linguistics
基金
教育部2015年度人文社科规划基金项目(项目编号:15YJA740071)的阶段性研究成果
河南理工大学博士基金项目(项目编号:SKB2015-13)的资助。