Soviet developments in machine translation: Russian sentence analysis.

Т. М. Николаева · Mech. Transl. Comput. Linguistics · 1958

THE ANALYTIC PART of MT in from Russian is a system of routines that work out the grammatical and syntactic features of the words needed for into another language. The entire analysis of Russian breaks down into two large, independent parts: a dictionary and a grammatical analysis. Each word in the sentence to be analyzed is examined in the dictionary, after which the word with the appropriate lexical information passes on to the grammatical analysis. This information indicates the part of speech to which the word belongs, the characteristics of that part of speech, and finally the specific morphological properties of the word. For example, inherent features of the noun are: gender, membership in one of three declensions, relationship to the category of animateness, as well as stem type. Characteristics of the verb are conjugation, quality of stem, and the possibility of being transitive. We have divided words having specific derivational characteristics into separate groups. An indication of the number of a group will constitute the dictionary information about the derivational properties of the word to be analyzed. Our dictionary is peculiar in that each word includes only information about its grammatical nature and place in the grammatical system of the language as a whole. This is a special type of dictionary, one differing substantially from the familiar kinds. Our dictionary lacks information about the semantic side of the word, that is, about its particular meaning. This is explained by the unusual role played in our practical work by Russian, which serves as an intermediary language. Since the analytic part of from Russian in our work is the same for into any language whatsoever, it would be useless to give the translation of a word in the Russian dictionary, inasmuch as it forms the content of the synthetic part of the dictionaries for the various languages.

Read the paper · More papers on PaperTik