Báo cáo khoa học: Contextual Analysis in Word-for-word MT

The difficulty is not so much one of word order as of syntactic and semantic ambiguity of individual words. Regardless of the treatment of the problem of inflected forms, for example, it is impossible in the majority of instances to identify the grammatical case of Russian nouns. In addition to syntactic ambiguity, multiple equivalents must be assigned to a large percentage of words (to an estimated 45% of the running words in a physics text). The chief disadvantage of word-for-word MT, then, is its prolixity: the reader is confronted with a burdensome multiplicity of potential equivalents (syntactic and semantic) for...
Nội dung trích xuất từ tài liệu:
Báo cáo khoa học: " Contextual Analysis in Word-for-word MT" [Mechanical Translation, vol.3, no.2, November 1956; pp. 40-41]Contextual Analysis in Word-for-word MTKenneth E. Harper, Slavic Department, University of California, Los Angeles general, there are no serious problems of syn-EXPERIMENTS with word-for-word MT of tax which cannot be resolved by reference to theRussian scientific literature have given results g rammatical features of pre- or post-words.which, except for such limited purposes as in-dexing, are far from satisfactory. The diffi- b) Semantic clarification. The correctculty is not so much one of word order as of English equivalents of most of the glue wordssyntactic and semantic ambiguity of individual (especially prepositions and conjunctions) canwords. Regardless of the treatment of the pro- be found only through contextual analysis. Theblem of inflected forms, for example, it is im- programming of such analysis should be basedpossible in the majority of instances to identify on the observed behavior of these words in ac-the grammatical case of Russian nouns. In tual conditions. Thus, the meaning of the con-addition to syntactic ambiguity, multiple equi- junction i, which has at least four equivalentsvalents must be assigned to a large percentage (and, but, also, even) can be pinpointed in moreof words (to an estimated 45% of the running than 90% of all occurrences by simple referencewords in a physics text). The chief disadvantage to the grammatical category of contiguous words;of word-for-word MT, then, is its prolixity: the the pronoun-adjective ikh, meaning (of) theirreader is confronted with a burdensome multi- or (of) them, can be similarly resolved. Itplicity of potential equivalents (syntactic and should be stressed that completely unpredict-semantic) for several words in each sentence. able and unexpected relationships can be found The chief cause of this ambiguity is the fact between structural context and meaning, andthat each word is examined in isolation, as a that the barest kind of routine comparison re-discrete item. The human translator operates sults in a high (although not absolute) degree ofwith the tremendous advantage of something accuracy in the determination of meaning.called context. Broadly speaking, context Non-structural clarification of meaning takessignifies environment: surrounding words, sen- several forms. In the first place, techniques oftences, and even the subject area itself. Investi- MT lexicography need to be developed, i.e., thegation shows that restricted contextual analysis, science of choosing the best cover-all targetperformed routinely, can resolve most of the language equivalent from a group of relativelyproblems of ambiguity. Remarkable clarifica- synonymous equivalents, and the selection oftion is attained even when the comparison of a equivalents based on observed behavior, rathergiven ambiguous word x is limited to the imme- than upon the evidence of a dictionary. (Thus,diately contiguous word in the sentence (the in the area of physics the Russian izmeneniepre-x or post-x word). Without attempting to may always be found to equate with change,rearrange the word order of the Russian sen- although Brays technical dictionary lists ninetence, one can obtain the following by compari- fairly distinct meanings.) In effect, what isson of each ambiguous word with the coded needed are true ideoglossaries, based on actual,grammatical features or semantic class of con- rather than potential, behavior.tiguous words: The application of contextual analysis offers a) Syntactic clarification. The ambiguity of great potentialities for semantic clarification.case forms in nouns can be reduced to an insig- Operating again on the basis of observation, wenificant percentage, and proper English equiva- can construct and code word classes whichlents can be supplied in the form of English pre- cause contiguous words to behave in a predict-positions as demanded by the genitive, dative, able manner. ...