Báo cáo khoa học: "translation of russian technical literature by machine"

Chia sẻ: Nghetay_1 Nghetay_1 | Ngày: | Loại File: PDF | Số trang:0

Thêm vào BST

Báo xấu

43
lượt xem 2
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

In order to read scientific or technical Russian, it is necessary to have the meaning of a large number of Russian words stored in the memory. In translating Russian, the corresponding English words must be supplied by the memory accurately and quickly.

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Báo cáo khoa học: "translation of russian technical literature by machine"

[ Mechanical Translation , vol.2, no.1, July 1955; pp. 15-24] translation of russian technical literature by machine* notes on preliminary experiments J ames W. Perry, School of Library Science, Western Reserve University, Cleveland, Ohio The Russian alphabet, the Russian words encountered in then evaluated by persons having scientific background but scientific and technical material and the Russian grammar no knowledge of Russian. d iffer greatly from their English counterparts. In order to T he results obtained were unexpectedly good and justify read scientific or technical Russian, it is necessary to have the conclusion that even this most primitive form of machine the meaning of a large number of Russian words stored in the translation enables persons knowing no Russian to under- memory. In translating Russian, the corresponding English s tand, to a surprising extent, the subject matter of the words must be supplied by the memory accurately and quickly. Russian original. This understanding is far better than would A utomatic electronic equipment can be designed so as to be provided by numerous index entries to the text material. have a memory capacity sufficient for translating Russian In fact, some sentences were understood with complete scientific and technical material. Machine memory, supple- accuracy. mented by appropriate selecting mechanisms, provide the These experiments indicate that a practical, experimental basis for effecting word-by-word translation of Russian. approach to further development of machine translation Preliminary experiments have been performed in which should yield very useful results. The quality of translations machine translation was simulated. One person copied the produced by machine can be greatly improved by designing individual words from samples of Russian text on separate t he machine system so that at least the simpler principles p ieces of paper and the writer took the words at random of Russian grammar are exploited. How to do this to best and supplied separate translations for each word. The text advantage is a problem which will require considerable w as then recreated by restoring the words to the order in experimentation. the Russian original. The crude translation so obtained was in a logical fashion, nevertheless require consid- introduction English-speaking scientists who undertake to learn erable effort to impress on the memory. to read scientific and technical papers in the Finally, the grammar is a source of so many dif- Russian language encounter a number of diffi- ficulties that it often becomes a barrier to learn- culties. The most obvious of these is the alphabet ing to read the language. which consists for the most part of strange, exotic looking letters. Grammar difficulties are not due to a lack of Mastery of the alphabet does little more than logical structure in the Russian language. On the open the door to further difficulties. Although contrary, the basic rules of Russian grammar can, an Indo-European language, Russian is a member to a large degree, be stated in a simple, straight- of the Slavic group. The words that constitute forward fashion. Inflectional endings play a domi- the backbone of the Russian language bear so nating role in Russian grammar; they alone ac- little similarity to corresponding English words count for much of the discouragement one so often that a heavy burden is imposed on the memory encounters. when acquiring the vocabulary needed to read In spite of some strange grammatical features, scientific and technical material. It is true that the basic structure of sentences in Russian and the purely technical and scientific terminology English is similar. Perhaps the most important of modern Russian is, in large degree, derived similarity is the word order, which is so nearly from the same basic words—Latin, Greek, Ger- the same that, once the corresponding English man or French—as are the corresponding English words have been written under the successive terms. However, in adopting words of foreign words in a Russian sentence, very often no rear- origin, the Russian language employs numerous rangement is needed to produce understandable suffixes, which, though used for the most part English sentences and minor rearrangement suf- fices to provide good idiomatic English. *This is a slightly revised version of a paper originally written When the Russian endings are not taken into in September, 1952 and given limited circulation in mimeo- account, a word-by-word translation often proves graphed form. Mr. Perry was then with the Center for Inter- deficient with respect to simple English connec- n ational Studies at M.I.T.
16 j. w. perry tives such as “of” and “to.” In spite of these tive case stands in a dependent relationship to s hortcomings word-by-word translations of another noun. Thus the phrase струйки фонтана R ussian technical material have a surprisingly w as interpreted after machine translation as high degree of intelligibility, as will be evident “little jet fountain” rather than as “a fountain’s from the experiments described below. little jets,” a more appropriate translation, which would have required account to be taken of the experimental method and results fact that фонтана was in the genitive singular I n these experiments, paragraphs were selected case. The writer’s assistants also pointed out that at random from Russian texts on physics, chemis- t he interpretation of the machine translation try and astronomy. The lines in the paragraphs w ould have been simpler if the plural of were numbered as were also the words in each n ounshad been indicated and if it had not been l ine. Each individual word in the Russian text necessary to rely on the context to select those w as copied on a separate piece of paper along nouns which indicate the means or agency used to with the two numbers which identified the line accomplish various actions. Interpreted in and the position of the word in the line. The slips t erms of Russian grammar, this latter observation were then shuffled so as to place them in random means that it would be advisable for machine order. Randomizing the Russian words had the operations to take the instrumental case into purpose of preventing the writer from interpret- consideration.* ing the meaning of the word in the light of the In spite of these limitations—and other less ob- context. After this had been done by an assistant v ious ones—the rough translations exhibited a who knew no Russian, the writer supplied one, or high degree of intelligibility. To establish this if necessary more than one, English word as a point, two of the writer's assistants who had had translation for each Russian word on an individual training in physics (Miss Patricia Fergus) and basis without knowing how the Russian sentences chemistry (Mrs. Anna M. Reid) were requested had been worded. This operation of translating to edit the rough translation produced by simu- individual words one by one could be accom- lated machine operations so as to indicate how plished by an appropriately designed automatic they would interpret its meaning. The results of electronic machine in whose memory units a t heir editorial interpretations are presented in Russian-English dictionary in properly encoded the pages which follow, along with a rather literal form had been recorded. translation of the Russian text prepared by the The numbers on the slips were next used to sort a uthor as a check. the individual words back into the original order ( work slips arranged in order are reproduced discussion of results below in an appendix). The English words were The practical usefulness of machine translation then copied off to produce the equivalent of a is, of course, the most important point we have to machine translation. consider. As is evident from the results, such In the all important step of supplying an English translation, even in a primitively simple form, translation for individual Russian words, no con- provides an astonishing degree of insight into sideration was given to inflectional endings, with Russian technical and scientific material. Such exception of certain irregular verb forms whose insight is more than sufficient to allow decisions frequent occurrence would justify their being to be made as to the pertinency of a document to i ncluded in the dictionary as separate entries. a given study. At the very least, therefore, ma- T he participles of verbs were also treated as c hine translation provides a basis for selecting t hough they were separate dictionary entries. out documents to be investigated in further detail. No consideration was given to case endings of nouns, pronouns and adjectives, nor to the tense *K. E. Harper documents this conclusion in his paper "The e ndings of verbs. This means, first of all, that Mechanical Translation of Russian—A Preliminary Report," no distinction was made between the singular and Modern Language Forum, Vol. 38, No. 3-4, pages 12-29 (Sept.- Dec. 1953). See also his chapter "A Preliminary Study of p lural of nouns. Furthermore the translation Russian," in Machine Translation of Languages, Ed. by Locke, provided no hint that a Russian noun in the geni- W. N. and Booth, A. D., Technology Press and John Wiley and Sons, 1955 (New York), pages 66-85.
russian technical literature 17 SAMPLE I — PHYSICS direct translation of russian original machine translation J . W. Perry E dited by Miss Patricia Fergus PIEZOELECTRICAL AND THERMOELECTRICAL PHE- PIEZOELECTRICAL AND THERMOELECTRICAL PHENOM- NOMENA. The polarization of a crystalline dielectric E NON. Polarization of a crystalline dielectric can m ay occur not only under the influence of the o ccur not only under the action of an electrical e lectric field, but in the case of certain crystals f ield but in the case of certain crystals (a number ( from the group not possessing a center of sym- o f which do not possess center of symmetry) polar- metry) the polarization may be caused by me- i zation can be caused by mechanical and also by c hanical, and also even by thermal action. The t hermal action. Electrical polarization of a crystal, e lectrical polarization of a crystal, when caused c aused by its tension or compression is called by its being under tension or compression, is called p iezoelectrical effect and polarization taking the piezoelectric effect, and polarization, occurring p lace during a change in temperature is called o n change of temperature, is called thermoelec- t hermoelectrical effect. trical effect.
j. w. perry 18 direct translation of russian original machine translation J. W. Perry E dited by Mrs. Anna M. Reid The saccharification of cellulose is beginning to be Saccharification of cellulose begins to employ employed in technology. For this purpose, waste technique. For that, the waste products of wood products of wood-working plants are heated under processing plants are heated under pressure with pressure with 0.1% solution of H2SO4; the syrup a 0.1% sulfuric acid solution. The syrup thus ob- obtained in this way is processed into alcohol. tained may be converted on to wine alcohol. Ac- According to another process the saccharifica- cording to other processes, saccharification may be t ion is carried out in the cold by the action of accomplished by cold action of very strong hydro- very strong (sp. gr. 1.21) hydrochloric acid. After chloric acid (sp. gr. 1.21). After removal of the removal of the acid there remains a solid product, a cid, the solid product remaining is used as a which is used as a feed stuff. food material.
russian technical literature 19 head comet. direct translation of russian original machine translation J . W. Perry E dited by Miss Patricia Fergus In Fig. 12 are plotted the parabolas, along which On Fig. 12 a parabola is drawn according to which bodies move when ejected with a velocity of 10 a body moves, thrown with the velocity of 10 m/sec at angles of 15°, 30°, 45° and 60° to the verti- m/sec and making angles of 15°, 30°, 45°, 60° with cal. Thus are distributed a fountain's little jets, the vertical line. Thus a little jet fountain is being when they are ejected in all directions from point thrown out in all directions from point A. Deflect- A. The envelope of deflection of all these little ing all these little jets, plotted on the graph, the jets has been plotted on the sketch as a dotted dotted line also forms a parabola. This is, in fact, line, and it is also a parabola. And this is in fact the outline of the head comet. the contour of the head of a comet.
20 j. w. perry literature. O bviously, such further investigation may re- quire the services of a skilled translator to assure Another point to be remembered when estimating that obscure—though important—points are not the value of a machine translation is its useful- misunderstood. ness to a human translator as a rough draft from which he can prepare a completely accurate trans- The first example (see page 17) provides an in- lation of documents whose importance warrants stance in which misunderstanding regarding an such attention. A rough draft prepared by ma- i mportant point crept into the machine trans- c hine translation can save much time and effort lation. In editing Sample I (Physics), Miss Fergus o n the part of human translators. made the first sentence read “Polarization of a c rystalline dielectric can occur not only under The crude examples of machine translation pre- the action of an electrical field but in the case sented above were produced with only a minimum of certain crystals (a number of which do not pos- of use of Russian grammar, namely the addition sess center symmetry) polarization can also occur of a parenthetical notation—e.g. “noun,” “verb,” by mechanical and also by thermal action.” The “ adj.”—to an English word to indicate the part italicized parenthetical statement is somewhat of speech of its Russian counterpart. Such gram- erroneous and would be better translated by “from matical identification can be readily accomplished the group not possessing a center of symmetry.” i n machine translation, as the Russian language The error was the result of the rather uncommon is so constructed that it is easy to distinguish use of the Russian word число to mean “group” between nouns, verbs, adjectives and other parts i nstead of “number.” To eliminate this type of of speech. The young ladies who edited the crude error, some of the rarer meanings of words would translations remarked that it would have been h ave to be included in the machine output. helpful if more grammatical notations could have C lose inspection of the other examples of ma- been included. chine translation reveals similar misunderstand- Many possibilities of exploiting the Russian gram- i ngs, which do not, however, invalidate our mar to improve the quality of machine translation previous conclusion that machine translation can await exploration. In particular, the elaborate p rovide an astonishing degree of insight into Russian system of inflectional endings provides a Russian scientific and technical material. wide range of leads to the structure and meaning As already noted, machine translation could serve of Russian sentences. When investigating these the very useful purpose of facilitating selection possibilities, the most practical approach would of documents pertinent to a given subject or prob- be to establish by experimentation which features lem. It is possible to imagine a system which would of grammar can be most advantageously incor- i ndex Russian material without translating it porated into a machine translation system.* and in this way provide a basis for machine search- I t is perhaps obvious that advantage is gained ing by recently developed automatic equipment. when the time and effort involved in using the To set up such a system, a list of key Russian o utput of a translative machine are decreased, w ords and phrases would have to be drawn up but the expense of increased complexity of design and these encoded so as to constitute an indexing a nd increased maintenance cost must be borne system. The translating machine, when it en- in mind. It would be easy to go beyond the point countered a key word or phrase would perform of diminishing returns in developing elaborate ma- two operations simultaneously. One would be the chines and elaborate machine translating methods, t ranslation of the word or phrase into English, w hich might produce translations of better lit- the other the encoding of the key word or phrase erary quality, but might fail to provide a prof- so as to convert it into an index entry appropriate i table return on the increased investment. for machine searching operations. Once such a system was set up, it would permit a large volume of Russian material to be analyzed and correlated *Much work has been done in this direction since the present without the help of persons having the scientific paper was originally written. See especially Oettinger, A. G., and linguistic training necessary to read and A Study for the Design of an Automatic Dictionary, Harvard u nderstand Russian scientific and technical thesis 1954, also Harper, op. cit.
russian technical literature 21 A good starting point for investigating the pos- advisable before undertaking to design a trans- s ibilities of exploiting Russian grammar to lating machine. However, such an investigation, improve machine translation might be furnished in order to remain within the realm of the prac- by the more than 700 example sentences which tical, should take into account the limitations which the writer used to illustrate the different imposed by the present state of development of points of grammar in his book Scientific Russian, automatic electronic equipment. Interscience Publishers, New York, 1950. conclusion Certain news reports may have given the mis- Preliminary experiments indicate that it is pos- leading impression that digital electronic equip- sible to apply machine methods advantageously ment already in existence would be well suited to the problem of translating Russian scientific for translating scientific and technical Russian. a nd technical material. Even the crude trans- Discussions with experts in digital electronic lation produced without systematic exploitation machines indicate on the contrary that present o f the Russian grammar provide a surprising machines would be grossly inefficient if used for degree of insight into the subject matter of scien- translating but that techniques and sub-assem- tific and technical material. An important prob- blies used in constructing digital computers can lem awaiting investigation is how best to exploit doubtless be used to construct a practical trans- the possibilities inherent in the Russian grammar lating machine. Further investigation of the while still remaining within the realm of the eco- m ethodology of machine translation appears nomically feasible. appendix—work slips from sample III (The numbers refer to the arrangement on the original Russian page where the first line contained eight words and the last, only one.)
j. w. perry 22
russian technical literature 23
24 j. w. perry