Russian to English machine translation with simple logical processing

Abstract

A SKILLED polyglot accomplishes translation from one language to another by translating thoughts; the source-language material is read and digested, and then the thought of the material is formed into target language sentences and recorded. The unskilled translator would translate first on a word-for-word basis, and would then make improvements in the output by consideration of the context of the words. The ultimate in machine translation would be that of translation by thought, as the accomplished polygot would translate, but several fundamental problems remain to be solved before anything of this sort may be accomplished. For instance, there is a great spread between the amount of storage in the human brain and the greatest amount of storage available in any present-day computer. The brain is estimated to have a total storage capacity of 1015 to 1020 bits. The largest of computers stores only about 109 bits, or not over one millionth as much storage as the human brain. A second limitation is that of present knowledge of language. Linguistic knowledge concerning many problems which are vital to machine translation is rather lacking. Thus it would seem that in the present elementary state of the machine translation art the attempt to perform translations should be made more as the unskilled translator does, by basing the translations on the word-for-word translation, and making such improvements as possible from a consideration of the context of the words in the sentence.