Using Linguistic Principles to Recover Empty Categories

Using linguistic principles to recover empty categories Richard CAMPBELL Microsoft Research One Microsoft Way Redmond, WA 98052 USA richcamp@microsoft.com Abstract This paper describes an algorithm for detecting empty nodes in the Penn Treebank (Marcus et al., 1993), finding their antecedents, and assigning them function tags, without access to lexical information such as valency. Unlike previous approaches to this task, the current method is not corpus-based, but rather makes use of the principles of early Government-Binding theory (Chomsky, 1981), the syntactic theory that underlies the annotation. Using the evaluation metric proposed by Johnson (2002), this approach outperforms previously published approaches on both detection of empty categories and antecedent identification, given either annotated input stripped of empty categories or the output of a parser. Some problems with this evaluation metric are noted and an alternative is proposed along with the results. The paper considers the reasons a principle- based approach to this problem should outperform corpus-based approaches, and speculates on the possibility of a hybrid approach. 1 Introduction Many recent approaches to parsing (e.g. Charniak, 2000) have focused on labeled bracketing of the input string, ignoring aspects of structure that are not reflected in the string, such as phonetically null elements and long-distance dependencies, many of which provide important semantic information such as predicate-argument structure. In the Penn Treebank (Marcus et al., 1993), null elements, or empty categories, are used to indicate non-local dependencies, discontinuous constituents, and certain missing elements. Empty categories are coindexed with their antecedents in the same sentence. In addition, if a node has a particular grammatical function (such as subject) or semantic role (such as location), it has a function tag indicating that role; empty categories may also have function tags. Thus in the sentence below, who is coindexed with the empty category *T* in the embedded S; the function tag SBJ indicates that this empty category is the subject of that S: [WHNP-1 who] NP want [S [NP-SBJ-1*T*] to VP] Empty categories, with coindexation and function tags, allow a transparent reconstruction of predicate-argument structure not available from a simple bracketed string. In addition to bracketing the input string, a fully adequate syntactic analyzer should also locate empty categories in the parse tree, identify their antecedents, if any, and assign them appropriate function tags. State-of-the-art statistical parsers (e.g. Charniak, 2000) typically provide a labeled bracketing only; i.e., a parse tree without empty categories. This paper describes an algorithm for inserting empty categories in such impoverished trees, coindexing them with their antecedents, and assigning them function tags. This is the first approach to include function tag assignment as part of the more general task of empty category recovery. Previous approaches to the problem (Collins, 1997; Johnson, 2002; Dienes and Dubey, 2003a,b; Higgins, 2003) have all been learning-based; the primary difference between the present algorithm and earlier ones is that it is not learned, but explicitly incorporates principles of Government- Binding theory (Chomsky, 1981), since that theory underlies the annotation. The absence of rule- based approaches up until now is …

Content truncated to 500 words