ChinGram: A TRALE Implementation of an HPSG Fragment for Mandarin Chinese

ChinGram: A TRALE Implementation of an HPSG Fragment of Mandarin Chinese Stefan M¨uller Freie Universität Berlin Stefan.Mueller@fu-berlin.de Janna Lipenkova Freie Universität Berlin Janna.Lipenkova@fu-berlin.de Abstract In this paper, we present our effort in the de- velopment of a HPSG grammar for Chinese. We present the basic notions of the HPSG framework, review existing theoretical anal- yses and implementations of Chinese gram- mar fragments in HPSG and present a range of deep linguistic analyses that are part of our own implementation. 1 Introduction This paper presents a grammar fragment of Chi- nese which is built in the framework of HPSG (Pollard and Sag 1994) and implemented in the grammar development system Trale (Meurers et al. 2002; Penn 2004). The grammar is one of the grammars that are developed in the CoreGram project (M¨uller 2013a). Apart from the Chinese grammar, which will be documented in M¨uller and Lipenkova (In Preparation), there are smaller frag- ments of several languages and larger fragments of German, Persian, Danish, and Maltese (see M¨uller (2013b) for details on size). These grammars share a common core und hence crosslinguistic generalizations are captured. We see the advan- tages of the HPSG framework for a formal analy- sis of Chinese as follows: * The work reported in this paper was supported by the grant ChinGram MU 2822/5-1 by the Deutsche Forschungsgemeinschaft. ** The following abbreviations are used: • HPSG sign features: HD: head; SS: synsem; IND: index • Tree arc symbols: Arg: argument daughter; Spr: speci- fier daughter; H: head daughter; NH: non-head daugh- ter; Adj: adjunct daughter • Glosses: CL: classifier; ATTR: attributive particle de; LOC: localizer particle • HPSG provides a range of powerful formal tools for the description of linguistic ex- pressions which are embedded into the log- ical framework of Typed Feature Structure Logic (Carpenter 1992) and allow a seam- less implementation in logical programming paradigms. • HPSG makes restricted use of a-priori theory-internal statements about the empiri- cial properties of linguistic signs. Since Chi- nese phenomena often cannot be explained using the terminology and assumptions of the Western linguistic tradition, HPSG provides us with a ‘neutral’ framework for the for- malization of language-specific phenomena based on which more general principles can be derived. • In contrast to most formal theories, HPSG is not a syntax-driven framework. That is, there is no central syntactic component from which a Phonological Form and a Logical Form is derived. Instead, the different lev- els of linguistic representation – phonology, syntax, semantics, pragmatics – have equal weight. This is especially beneficial for Chi- nese, which has a poor morphological sys- tem and exhibits a high degree of surface ambiguity. The use of a powerful semantic- pragmatic component with fine-grained defi- nitions of semantic types and selectional re- strictions and preferences thus helps disam- biguation. In the following, we first introduce the basic feature architecture and formal tools of the gram- mar formalism. Then, we review existing work in HPSG and grammar development for Chinese. Finally, we describe the theoretical and empirical …

Content truncated to 500 words