ERIC Number: ED281385
Record Type: RIE
Publication Date: 1986-Oct
Reference Count: 0
Digging in the Dictionary: Building a Relational Lexicon To Support Natural Language Processing Applications.
Evens, Martha; And Others
Advanced learners of second languages and natural language processing systems both demand much more detailed lexical information than conventional dictionaries provide. Text composition, whether by humans or machines, requires a thorough understanding of relationships between words, such as selectional restrictions, case patterns, factives, and other kinds of verb implicature. For verbs, we need to know whether they are action or stative, performative or not, and what kinds of complements they take. It is important to know whether an adjective is non-predicating, non-attributive, action, or stative, For nouns, we need relations like taxonomy, part-whole, membership, and modification, and also attributes like count, mass, human, and animate. This paper discusses these and other kinds of lexical information found only implicitly, if at all, in most commercial dictionaries. (MSE)
Descriptors: Adjectives, Computational Linguistics, Dictionaries, Difficulty Level, Discourse Analysis, English (Second Language), Form Classes (Languages), Grammar, Information Needs, Language Processing, Learning Processes, Lexicography, Nouns, Second Language Learning, Semantics, Verbs, Vocabulary Development
Publication Type: Reports - Evaluative; Opinion Papers; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: National Science Foundation, Washington, DC.
Note: Paper presented at the Computers in Language Research and Language Learning Conference (Urbana, IL, October 25-26, 1986).