Universal Semantic Tagging for English and Mandarin Chinese

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 5554–5566 June 6–11, 2021. ©2021 Association for Computational Linguistics 5554 Universal Semantic Tagging for English and Mandarin Chinese Wenxi Li1, Yiyang Hou1, Yajie Ye2, Li Liang1∗and Weiwei Sun3 1Department of Chinese Language and Literature, Peking University 2Wangxuan Institute of Computer Technology, Peking University 3Department of Computer Science and Technology, University of Cambridge {liwenxi,yiyang.hou,liangli15,yeyajie}@pku.edu.cn ws390@cam.ac.uk Abstract Universal Semantic Tagging aims to provide lightweight unified analysis for all languages at the word level. Though the proposed an- notation scheme is conceptually promising, the feasibility is only examined in four Indo– European languages. This paper is concerned with extending the annotation scheme to han- dle Mandarin Chinese and empirically study the plausibility of unifying meaning represen- tations for multiple languages. We discuss a set of language-specific semantic phenom- ena, propose new annotation specifications and build a richly annotated corpus. The cor- pus consists of 1100 English–Chinese paral- lel sentences, where compositional semantic analysis is available for English, and another 1000 Chinese sentences which has enriched syntactic analysis. By means of the new an- notations, we also evaluate a series of neural tagging models to gauge how successful se- mantic tagging can be: accuracies of 92.7% and 94.6% are obtained for Chinese and En- glish respectively. The English tagging perfor- mance is remarkably better than the state-of- the-art by 7.7%. 1 Introduction Developing meaning representations across differ- ent languages plays a fundamental and essential role in multilingual natural language processing, and is attracting more and more research interests (Costa-jussà et al., 2020). Existing approaches can be roughly divided into three categories: the crosslingual1 approach focuses on lending seman- tic annotation of a resource-rich language, such as English, to an under-resourced language (Wang et al., 2019; Blloshmi et al., 2020; Mohiuddin and Joty, 2020); the interlingual approach attempts to ∗This author is now working in Tencent. 1 The terminology in the literature is quite diverse—the usages of “crosslingual”, “interlingual” and “multilingual” vary from author to author. provide a unified semantic framework for all lan- guages (Abend and Rappoport, 2013; White et al., 2016; Ranta et al., 2020); the multilingual ap- proach aims at developing comparable but not nec- essarily identical annotation schemes shared by different languages (Bond and Foster, 2013; Baker and Ellsworth, 2017; Pires et al., 2019). In line with the interlingual approach, Universal Semantic Tagging (UST; Bjerva et al., 2016) de- velops a set of language-neutral tags (hereafter re- ferred to as sem-tag) to annotate individual words, providing shallow yet effective semantic informa- tion. Semantic analyses of different languages utilise a same core tag set, but may also employ a few language-specific tags. Figure 1 presents an example. English I/PRO had/PST repaired/EXT my/HAS watch/CON ./NIL German Ich/PRO hatte/PST meine/HAS Arm- banduhr/CON repariert/EXT ./NIL Italian Ho/NOW riparito/EXT il/DEF mio/HAS orologio/CON ./NIL Chinese 我/PRO 把/OBJ 我/PRO 的/MOD 手 表/CON 修/EXT 好/EXT 了/PFT 。/NIL Figure 1: An example of parallel sentences and their sem-tags. PRO: anaphoric …

Content truncated to 500 words