Finding Phrases Rather Than Discovering Collocations: Searching Corpora for Dictionary Phrases
Debra S. Baddorf, Martha W. Evens · 1998
This paper describes our attempts to find information about phrases and their syntactic variants for inclu-sion in a computer lexicon. We started with a list of thirty phrases from the British Collins English Dictio-nary. After making a few phrase modifications to accom-modate American usage, we searched for occurrences in the Gutenberg corpus, the Wall Street Journal (1987, 1988, and 1989) and the Department of Energy techni-cal abstracts from the ACL-DCI CDROM. Finding syntactic variants of phrases forced us to al-low variations in word order, tense and number, and to be flexible in looking for the smaller words of a phrase. We had to use flexible matching techniques to handle insertions or changes in adjectives, adverbs, and prepo-