Detecting Dependencies between Semantic Verb Subclasses and Subcategorization Frames in Text Corpora

Detecting Dependencies between Semantic Verb Subclasses and Subcategorization Frames in Text Corpora Victor Poznadski, Antonio Sanfilippo SHARP Laboratories of Europe Ltd. Oxford Science Park, Oxford OX 4 4GA { vp, aps} @prg.oxford.ac.uk Abstract We present a method for individuating dependencies between the semantic class of predicates and their associated subcategorization frames, and describe an implemen- tation which allows the acquisition of such dependencies from bracketed texts. 1 Introduction There is a widespread belief among linguists that a predicate's subcategorization frames are largely determined by its lexical-semantic properties [23, 11, 12]. Consider the do- main of movement verbs. Following Talmy [23], these can he semantically classified with reference to the meaning components: MOTION, MANNER, CAUSATION, THEME (MOVING ENTITY), PATH AND REFERENCE LOCATIONS (GOAL, SOURCE). Lexicalization patterns which arise from identifying clusters of such meaning components in verb senses can be systematically related to distinct subcategorization frames. 1 For example, the arguments of a verb expressing directed caused motion (e.g. bring, put, give) are normally a causative subject (agent), a theme direct object (moving entity) and a directional argument express- ing path and reference location (goal), e.g. (1) Jackie will bring a bottle of retsina to the party CAUSER THEME PATH GOAL However, a motion verb which is not amenable to direct external causation [13], will typically take a theme subject, with the possible addition of a directional argument, e.g. (2) The baby crawled (across the room) Co-occurrence restrictions between meaning components may also preempt subcategoriza- tion options; for example, manner of motion verbs in Italian cannot integrate a completed path component and therefore never subcategorize for a directional argument, e.g. (3)*Carlo ha camminato a casa Carlo walked home 1Following Levin [12] and Sanfilippo [18], we maintain that valency reduction processes (e.g. the causative-inchoative alternation) are semantically governed and thus do not weaken the correlation be- tween verb semantics and subcategorization properties. 82 These generalizations are important for NLP since they frequently cover large sub- classes of lexical items and can be used both to reduce redundancy and elucidate sig- nificant aspects of lexical structure. Moreover, a precise characterization of the relation between semantic subclasses and subcategorization properties of verbs can aid lexical dis- ambiguation. For example, the verb accord can be used in either one of two senses: agree or give, e,g. (4) a The two alibis do not accord Your alibi does not accord with his b They accorded him a warm welcome Accord is intransitive in the agree senses shown in (4a), and ditransitive in the give sense shown in (4b). The manual encoding of subcategorization options for each choice of verb subclass in the language is very costly to develop and maintain. This problem can be alleviated by automatically extracting collocational information, e.g. grammar codes, from Machine Readable Dictionaries (MRDs). However, most of these dictionaries are not intended for such processing; their readership rarely require or desire such exhaustive and exacting precision. More specifically, the information available is in most cases compiled manually according to the lexicographer's intuitions …

Content truncated to 500 words