Categorial unification grammars

Abstract

Categorial unification grammars (CUGs) embody the essential properties of both unification and categorial grammar formalisms. Their efficient and uniform way of encoding linguistic knowledge in well-understood and widely used representations makes them attractive for computational applications and for linguistic research. In this paper, the basic concepts of CUGs and simple examples of their application will be presented. It will be argued that the strategies and potentials of CUGs justify their further exploration in the wider context of research on unification grammars. Approaches to selected linguistic phenomena such as long-distance dependencies, adjuncts, word order, and extraposition are discussed.

Gapping as constituent coordination

“…The existence of left node raising in SVO languages, together with the related dependency of so called "forward" and "backward" gapping on VSO and SOV word-order discussed below, is in fact one of the strongest pieces of confirmatory evidence in favour of the present proposal to base the theory of grammar on these two combinatory operations. It is in respect of these constructions that the theory should be contrasted with other closely related function-oriented and unification-based theories, such as those advanced by Kartunnen (1 986), Uszkoreit (1986), Joshi (1987), Zeevat et al (1987), and Pollard and Sag (1987).…”

“…They can be represented computationally by a single data structure uniting syntactic type and semantic interpretation in unification-based implementations like those of Zeevat, Klein & Calder (1986), Pareschi & Steedman (1987), and Pareschi (1989) (Cf. Karttunen 1986, Uszkoreit 1986, Shieber 1986, and Wittenburg 1986 for related approaches.) The categories can be represented in full in an expanded notation (which will be used as sparingly as possible), in which each elementary category is associated with a term representing its interpretation using the symbol ":".…”

Structure and Intonation

“…The category ( S \ N P ) / N P could be regarded as both a syntactic and a semantic object, as in the unification-based categorial grammars of Karttunen 1986, Uszkoreit 1986, Wittenburg 1986, and Zeevat et al 1986. (See Steedman 1990a for an explicit expression of combinatory categorial grammars in unification-based terms, uniting syntax and semantics in this way.)…”

Chinese Treebanks and Grammar Extraction

“…Some common collocations found by word-to-word relations Word bi-gram statistics is often the major information for constructing language model (Yuan et al, 1997;Manning et al, 1999 Table 3. Words frequently co-occurred with proper names Grammar rule extraction is the major usage of Treebanks (Uszkoreit, 1986). Not only sentential/phrasal patterns but also their probabilities of usages can be derived as exemplified in Table 4.…”