Language Design, Computers and Brains
Abstract
The processing of natural languages by computers and the programming of computers to simulate some aspects of language learning offer insights into those aspects of language design which allow successful processing to take place as well as those properties of natural languages which make their computer-based analysis difficult. In essence, these factors represent the differences between the two information processors involved in all such endeavors, the biological one, the human brain, of which natural languages are an evolutionary product, and the digital computer. The properties of language which allow computer processing of language data are those of structure, specifiable by a grammar which captures systematically the redundancy in language organization and its striving towards a reasonable degree of efficiency. The difficulties encountered in language processing by computers center around ambiguity, which can be either structural or semantic. The theory of markedness is described which posits a systematic use of semantic ambiguity for the sake of efficient communication. Finally, a project conducted at Brown University is described, in which a computer program simulating a child learner acquires an interesting subset of English grammar from a parent program which knows the language. Originator- supplied keywords include: Foreign language, Information theory, Formal grammars, Redundancy, Ambiguity, Polysemy, Markedness, Language learning, Language translations, Sequential vs. Parallel Processing.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 29, 1985
- Accession Number
- ADA152819
Entities
People
- Henry Kucera
Organizations
- Brown University