Corpus-based Method for Automatic Identification of Support Verbs for Nominalizations

Corpus-based Method for Automatic Identification of Support Verbs for Nominalizations Gregory Grefenstette Simone Teufel Rank Xerox Research Centre Universit£t Stuttgart 38240 Meylan, France Institut ffir maschinelle Sprachverarbeitung grefen©xerox.fr D 70174 Stuttgart 1 s imone~ ims. uni- stuttgart, de Abstract Nominalization is a highly pro- ductive phenomena in most lan- guages. The process of nominaliza- tion ejects a verb from its syntactic role into a nominal position. The original verb is often replaced by a semantically emptied support verb (e.g., make a proposal). The choice of a support verb for a given nomi- nalization is unpredictable, causing a problem for language learners as well as for natural language pro- cessing systems. We present here a method of discovering support verbs from an untagged corpus via low-level syntactic processing and comparison of arguments attached to verbal forms and potential nom- inalized forms. The result of the process is a list of potential support verbs for the nominalized form of a given predicate. 1 Introduction Nominalization, the transformation of a ver- bal phrase into a nominal form, is possible in most languages (Comrie and Thompson, 1990). Nominalizations are used for a vari- ety of stylistic reasons: to avoid repetitions of a verb, to avoid awkward intransitive uses of transitive verbs, in technical descriptions where passive is commonly used, etc. Though, as a result of nominalization, the original verb is ejected from its syntactic po- sition, it often retains many of its thematic roles. The original agents and patients can reappear as genitival or adjectival modifiers of the nominalized predicate. In the syntactic place of the original verb can appear a seman- tically impoverished verb. The semantically impoverished verb, often called a support verb, to be used with a nom- inalized predicate structure is unpredictable. Allerton (1982)[p. 76] writes: Perhaps the most serious prob- lem for these structures is that there is no constant selection of the empty verb: sometimes we find have, sometimes take, sometimes give, and rarely pay; sometimes we have a choice between two or more empty verbs e.g. have/take a look ... We have little choice but to record such irregularities in the lexicon ... For this reason, the collocational choice of a support verb for a given nominalization is a difficult problem for language learners, as well as for natural language processor implemen- tation. We present here a method of deriving prob- able support verbs for nominalized predicates from corpora using low-level syntactic anal- ysis and simple frequency statistics over its results. This automatic procedure may be looked upon as an aid to lexicographers, as an independent extraction tool, or as a verifica- tion of lexical collocation information stored in a machine-based lexicon. Since nominalized predicates can semanti- cally drift over time to become concrete nouns having lost all their thematic role, our method first attempts to distinguish true nominaliza- tions from concrete uses of the nominalized surface form. This is done by comparing ap- proximations to the argument/adjunct struc- tures of verbal predicates to those of candi- date …

Content truncated to 500 words