Multiword expressions are a key problem for the development of large-scale, linguistically sound natural language processing technology. This paper surveys...

Much recent research on the design of natural language lexicons has made use of nonmonotonic inheritance networks as originally developed for general knowledge representation purposes in Artificial Intelligence. DATR is a simple, spartan language for de ning nonmonotonic inheritance networks with path/value equations, one that has been designed specifically for lexical knowledge representation. In keeping with its intendedly minimalist character, it lacks many of the constructs embodied either in general purpose knowledge representation languages or in contemporary grammar formalisms. The present paper shows that the language is nonetheless sufficiently expressive to represent concisely the structure of lexical information at a variety of levels of linguistic analysis. The paper provides an informal example-based introduction to DATR and to techniques for its use, including finite state transduction, the encoding of DAGs and lexical rules, and the representation of ambiguity and alternation. Sample analyses of phenomena such as inflectional syncretism and verbal subcategorisation are given which show how the language can be used to squeeze out redundancy from lexical descriptions.

This paper introduces an approach to modeling the interpretation of semantically underspecified logical metonymies, such as John began the book. A distinctive feature of the theory presented is its emphasis on accounting for their behavior in discourse contexts. The approach depends on the definition of a pragmatic component which interacts in the appropriate manner with lexicosyntactic information to establish the coherence of a discourse. The infelicity of certain logical metonymy constructions in some discourses is shown to stem from the non-default nature of the lexicosyntactically determined interpretation for such constructions. The extent of the influence of contextual information from the discourse on the interpretation of logical metonymies is therefore constrained by the lexical properties of the constituents of the metonymies. Contextually-cued interpretations are shown to be unattainable when indefeasible lexical information conflicts with these interpretations.

This paper develops within HPSG a model of grammar with two syntactic levels, valence lists and argument structure, at which sentences may have different representations: syntactically ergative and Western Austronesian languages are distinctive by allowing different prominence orderings between the valence lists and argument structure, while forms like passives and causatives have nested argument structure lists. While binding theory and related phenomena have traditionally been described in terms of surface grammatical relations or configurations, we demonstrate that binding theory is actually correctly described in terms of argument structure configurations. Such an approach generalizes nicely over accusative and ergative constructions, correctly predicts binding patterns with causative and passive verbs, and supports the lexicality-preserving account of passives and causatives a...

This paper proposes a modification of HPSG theory – Sign-Based Construction Grammar – that incorporates a strong theory of both selectional and constructional locality. A number of empirical phenomena that give the appearance of requiring nonlocal constraints are given a principled, localist analysis consistent with this general approach, which incorporates certain insights from work in the tradition of Berkeley Construction Grammar, as exemplified by Fillmore et al. (1988), Kay and Fillmore (1999), and related work. 1

The question we address in this paper is whether `Lexical Rules' deserve their grand status, a status that is often conveyed by a special purpose formalism and/or a separate component, one that may even be external to the lexicon proper. We will argue that they do not and that a lexical knowledge representation language that is as expressive as it needs to be for other lexical purposes will, ipso facto, be expressive enough to encode `Lexical Rules' internally as lexical rules. Such internal encoding is not only possible but also desirable since `Lexical Rules' will then automatically acquire other characteristics which are now standardly associated with common or garden lexical rules, including inheritance, generalization by default, and the ability to relate lexical information from different levels of linguistic description. We give examples of what we take to be instances of common or garden lexical rules and then show how the same formal machinery provides for the statement of a P...

In this paper we show to what degree the countability of English nouns is predictable from their semantics. We found that at 78 % of nouns’ countability could be predicted using an ontology of 2,710 nodes. We also show how this predictability can be used to aid non-native speakers to determine the countability of English nouns when building a bilingual machine translation lexicon. 1

Lexical knowledge is increasingly important in language engineering. However, it is a difficult kind of knowledge to represent and reason with. Existing approaches to formalizing lexical knowledge have used languages with limited expressibility and in particular they have not addressed the context-dependent nature of lexical knowledge. Here we present a framework, based on default logic, called the dex framework, for capturing context-dependent reasoning with lexical knowledge. Default logic is a first-order logic offering a more expressive formalisation than inheritance hierarchies: (1) First-order formulae about words can be inferred; (2) Preferences over formulae can be based on specificity, reasoning about exceptions, or explicit priorities; (3) Contexts can be reasoned with as first formulae; and (4) Contexts can be derived as default inferences. In the dex framework, the word for which further information is sought called the query word. The context for a query word is derived fr...

In this paper we explore how compositional semantics, discourse structure, and the cognitive states of participants all contribute to pragmatic constraints on answers to questions in dialogue. We synthesise formal semantic theories on questions and answers with techniques for discourse interpretation familiar from computational linguistics, and show how this provides richer constraints on responses in dialogue than either component can achieve alone. 1 Introduction Understanding dialogue involves a complex interaction between the semantics of the utterances, the discourse context of which they are part and the cognitive states of the participants. In an effort to begin to spell out these connections precisely, we focus here on the role of questions and answers in dialogue---more specifically, how-questions. Like many researchers (e.g., Hobbs 1985, Polanyi 1985), we represent a discourse context as a structured object consisting of propositions related by discourse relations. How-...

This paper deals with a number issues having to do with locality in natural language. Locality of selection is the problem of delimiting what syntactic and semantic information lexical items select. Related issues include the proper analysis of idiomatic expressions, control of overt pronominals, and cross-linguistic