intTypePromotion=1
zunia.vn Tuyển sinh 2024 dành cho Gen-Z zunia.vn zunia.vn
ADSENSE
 » 
 » 

Báo cáo khoa học: "THE CONFERENCE ON MECHANICAL TRANSLATION"

Chia sẻ: Nghetay_1 Nghetay_1 | Ngày: | Loại File: PDF | Số trang:9

Thêm vào BST
Báo xấu
44
lượt xem 2
download
 
  Download Vui lòng tải xuống để xem tài liệu đầy đủ

The following report was prepared immediately after the writer's return from the conference. It was written from the viewpoint of an engineer listening to experts in a field far separated from his own. Such judgments as may be found interspersed amongst the reports of individual papers are of an engineering nature, and are not to be construed, as being based upon other than an amateur’s knowledge of linguistic theory.

Chủ đề:

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

Lưu

Nội dung Text: Báo cáo khoa học: "THE CONFERENCE ON MECHANICAL TRANSLATION"

  1. [ Mechanical Translation, vol.1, no.3, December 1954; pp. 47-55] THE CONFERENCE ON MECHANICAL TRANSLATION* Held at M.I.T., June 17-20, 1952 A. C. Reynolds, Jr. International Business Machines Corporation, Endicott, N. Y. The following report was prepared immediately after the writer's return from the conference. It was written from the viewpoint of an engineer listening to experts in a field far separated from his own. Such judgments as may be found interspersed amongst the reports of individual papers are of an engineering nature, and are not to be construed, as being based upon other than an amateur’s knowledge of linguistic theory. Further, they represent only the reporter’s evaluation, not necessarily that of his company as a whole. It is of interest, however, that the writer’s company, The International Business Machines Corporation, has jointly sponsored with Georgetown University a successful demonstration of syntactically correct mechanical translation from Russian into English. The computer employed was the IBM 701, and the programming techniques used were first discussed at the 1952 conference. The concept of mechanical translation origi- ducting his research at M.I.T., acted as the nated in two areas, the first being cryptogra- coordinator of the groups actively interested in phic work conducted by various governments mechanical translations. As part of his work, Dr. during the late war, and the second being the Bar-Hillel prepared a summary entitled “Present successful inauguration and employment of the Interest in Mechanical Translation,” listing the simultaneous translation schemes presently individuals actively working on the application of employed by the UN and other internation con- computers and computer techni-ques to ferences. Broken down into basic essentials, mechanical translation. In 1952 he or-ganized a translation consists of memory scanning for Conference on Mechanical Translation at M.I.T. identification of meaning in two different sym- This report is concerned with providing a bolic systems, called languages, and simultane- precis of the papers and discussions at the Con- ous editing by the translator to convert the syn- ference. tactical relationships of the language being translated to those of the translated language. Session I - June 17, 1952 Of these, the memory scanning is definitely Public Session paralleled in computer techniques. If one to one correlations in meaning existed between The Public Session of the Conference on Me- words of different languages, programming on chanical Translation was announced by invita- existing computers would be completely suc- tions extended by Dr. Yehoshua Bar-Hillel to cessful. Syntactical relationships and shading persons who might be interested in the pro- of meaning by the context of the words makes blems of mechanical translation and, in parti- the problem of mechanization exceedingly diffi- cular to members of the Conference on Speech cult in the absence of a mechanical means of Communication which immediately preceded converting from one syntax to another. the Conference on Mechanical Translation. At Much work was stimulated by a memorandum, the public session papers were not presented, Translation, written by Dr. Warren Weaver of the but short talks were given by each of the five Rockefeller Foundation.which was distri-buted to participants outlining their work in the field and a selected group of linguists, psycholo-gists, their tentative proposals for future work. computer engineers, and philosophers. Dr. Bar-Hillel discussed the need and possi- Dr. Yehoshua Bar-Hillel, acting under a grant bilities for mechanical translation, the need from the Rockefeller Foundation and then con- primarily arising in the fields of science and of diplomacy, for analysis of popular periodi- cals of various countries. Although a person * For a linguist’s view of the same Conference, may be versed in the cultural or popular langu- see MT, Vol. I, No. 2, “Report on the First Con- age of several countries, this does not neces- ference on Mechanical Translation,” Erwin sarily mean that the same individual is capable Reifler, pp. 23-32. A list of participants in the of translating scientific treatises originating in Conference appears on p. 24 of that article.
  2. A.C.REYNOLDS, JR. 48 the same countries. This is due to the well among members of the same group. The ma- known fact that each scientific discipline cre- chine must be capable of resolving idiomatic, ates its own jargon, assigning very specific contextual, and syntactic ambiguities if human meanings to common words of the language, editing is to be kept at a minimum and maximum these meanings being peculiar to the particular intelligibility is to be achieved. Dr. Helmer science itself. There is, therefore, a need for discussed schemes that have been tentatively translators who are capable of making mean- investigated by the Rand Corporation for sol-ving ingful interpretations, not only in the more pop- this problem. His conclusion is that high speed ular writings, but also in specific areas of general purpose computing machines will be able scientific research. The volume of material to handle the main translation task. appearing in popular periodicals is appalling in Dr. Andrew D. Booth, Director, The Electro- its magnitude and complete scanning of a par- nic Computer Section, Birkbeck College, Uni- ticular nation’s output is virtually impossible versity of London, discussed the popular mis- as long as human translators must be relied conceptions covered by the question, “How in- upon. He concluded that it is in these areas telligent can a machine translator be ?” The that mechanical translation is capable of mak- conclusions necessarily were that “intelligence” ing a major contribution to society. as applied to machines involves a complete mis- Prof. Leon Dostert, Director of the Institute of understanding both of intelligence and of ma- Languages and Linguistics, Georgetown Uni- chines. No intelligence is required, on the part versity, Washington, D. C., spoke on the sub- of the machine at least, in mechanical transla- ject of human translation versus machine trans- tion. lation. Prof. Dostert drew on his experience in Dr. James W. Perry, Center of International setting up the translation system employed at the Studies, M.I.T., discussed machine techniques Nuremburg trials in Germany and in working and index searching and translation. The basis with IBM in the development of the of Dr. Perry’s talk was the index searching ma- simultaneous translation system used at the UN chine developed by IBM to solve the problem of and other international conferences. In discuss- scanning vast amounts of information and ex- ing this problem, he made the statement that, tracting certain specific items. He discussed except in the very specialized areas discussed the development of coding on punched cards in by Dr. Bar-Hillel, there is no shortage of hu- order to employ a machine at maximum effici- man translators, owing apparently to the fact ency. He concluded on the basis of his acquain- that the current workload is regulated by their tanceship with existing machines and machine availability. The contribution a machine can techniques that mechanical translation was not make is in the processing of the vast amount of only feasible but far closer to realizations than material that is currently not even being touch- possibly the audience recognized. ed in the specialized fields. He described sys- A period of discussion from the floor followed tems employed in setting up efficient simul- the presentation of the talks. There was general taneous translation systems and also rapid agreement on the part of both the panel and the printed translations in international gatherings. audience that mechanical translation was feasi- These systems were remarkably similar in ble. It was interesting to note that the computer their organization to machine organization for engineers present presented all of the difficul- computer application. He confessed that he ties standing in the way of producing a mech- came to the Conference as a sceptic. (Later in anical translator from the engineering stand- the Conference he became convinced that me- point; the linguist, from his standpoint; and the chanical translation would be possible.) psychologists and philosophers from the stand- Dr. Olaf Helmer, Director of Research, Math- point of their respective disciplines. Each ematical Division, Rand Corporation, Santa agreed, however, that, if the other two groups Monica, California, discussed the structure of the did their work, we could in the near future pro- problem of mechanical translation. Mean- duce adequate and intelligible machine pro- ings of particular words and phrases may be grammed translations. idiomatic or may be changed or modified by Session II - June 18, 1952 the context in which they appear. Further, each Chairman - Dr. Leon Dostert group of languages has its own syntactical re- lationships which are peculiar to the group,and Prof. Erwin Reifler.Far Eastern and Russian most frequently also vary in minor details Institute, University of Washington, Seattle,
  3. THE CONFERENCE ON MECHANICAL TRANSLATION 49 Washington, presented the first two papers of form of post-editing in order to resolve the the morning session entitled, “Mechanical ambiguities inherent in this relationship be- Translation with Pre-editing,”and “Writing for tween languages. Dr. Bar-Hillel is much con- Mechanical Translation.” cerned with the tremendously increased de- The first paper concerned itself with the fact mands in terms of machine storage capacity that syntactical relationships differ amongst which this situation implies. It is, however, languages. For ease in programming on a me- not quite so grave as appears on the surface, chanical translator, a source language should since particularly in scientific writings, a vast be arranged according to the syntax of the tar- number of one-to-one correlations do exist. get language (language into which the material (The subject of glossaries to handle the sci- is being translated). Where this is not possible entific translations was covered in a later ses- due to the fact that the syntax is inseparable sion of the conference.) from the actual word form (such as the dative The fourth paper, “Model English for Mech- case in Latin) certain keys, such as capital let- anical Translation” was presented by Prof. ters or diacritical marks, can be inserted as Stuart C. Dodd, Director, Washington Public recognizable signals for a machine whose input Opinion Laboratory, University of Washington, is a print scanning device. Pre-editing then Seattle. Dr. Dodd’s paper concerned itself with would imply the use of a human editor to re- the standardization of English syntax as a arrange the source language insofar as possible in means of simplifying the use of English either accordance with the syntax of the target lan- as a source language or as a target language. A guage, and secondly, employment of various in- model language, as defined by Dr. Dodd, means serted signals to notify the machine of syntacti- any language in which the rules of syntax have cal arrangements inseparable from the word been regularized, and in which familiarity of form. words is a governing criterion. The specific The second paper, on “Writing for Mechanical rules used in regularizing a language are item- Translation,” would necessitate the training of ized in the paper. The examples employed by all writers, and more particularly their secre- Dr. Dodd indicate that regularizing, that is, taries,in the required conventions for arrange- constructing a model language, impaires but ment of an article for translation into a given very slightly the readability and understanda- language. The discussion of these two papers bility of the subject matter. In English, at indicated that the use of a pre-editor, rather least, regularizing leads only to a certain than educating all authors and all secretaries in quaintness of expression somewhat similar to techniques of writing for mechanical transla- the sentence structure employed by the Quakers. tions, is far preferable. As a matter of fact, a No attempts have been made as yet to regu- person skilled in keyboard operation could be larize languages other than English, but at readily trained to insert syntactical recognition least for the Romance languages it seems on signals at the time of keying the text into the first view that such regularization can be ac- machine. This, of course, also holds for the complished. preparation of a manuscript for machine scan- The particular rules of importance to Mech- ning. anical Translation are: one word order; one Dr. Yehoshua Bar-Hillel presented a paper on meaning for each word; and one form for each Mechanical Translation employing a post-editor. word. Since a one-to-one correlation does not exist The experience gained in using model langu- between meanings of words expressing essen- age at the Washington Public Opinion Labora- tially the same idea in various languages, if a tory indicates clearly that regularization of a machine operates on a comparison basis only, language minimizes the points brought out by or even if it is capable of computing syntactical Dr. Bar-Hillel. The discussion showed that the relationship, a multiplicity of words in the tar- conference was in substantial agreement that get language can be derived for any single word regularization by use of the concepts of a model of the source language. For a particular sen- language is feasible and directly applicable to tence, say of 10 words length, this can easily the problems of mechanical translation. In result in possible combinations of words in the particular, so far as the machines to be em- target language extending to several thousands of ployed are concerned, the machine men present more or less meaningful combinations. It is felt that it could be a decided advantage in re- necessary, therefore, to incorporate some ducing the complexity of equipment required.
  4. A.C.REYNOLDS, JR. 50 Session III - June 18, 1952 Oswald proposed to take advantage of this fact Chairman - A. C. Reynolds, Jr. by the use of what he termed micro-glossaries. These micro-glossaries would be constructed Prof. Victor A. Oswald, Department of Ger- on the basis of the words most commonly used manic Languages, University of California, Los in specific subjects of interest; one such glos- Angeles, presented the first paper entitled sary being constructed for each subject to be “Word-by-Word Translation.” Prof. Oswald translated. Mechanically, this means that two and Dr. Harry D. Huskey, Assistant Director, memories would be employed in a machine; one, National Bureau of Standards Institute for Nu- a most used general vocabulary for the langu- merical Analysis, University of California, Los ages being processed; and two, a specific mi- Angeles, jointly conducted experiments in the cro-glossary to assign specific meanings to translation of a text in mathematics and another words that would otherwise have a multiplicity in brain surgery from German into English. The of meaning; that is, if all their fields of usage investigation by Dr. Oswald indicated that word- were to be considered simultaneously. The con- by-word translation from German into English cept of a micro-glossary and the use of block- was a virtually impossible task, chiefly because by-block syntactic recognition in the machine of the fact that German “articles” are also met with favor from all the participants in the “words.” Also, German sentence structure is conference. The linguists appeared certain such that word-by-word translation from Ger- that block-by-block syntactic analysis of sen- man into English becomes virtually meaning- tences could be accomplished and likewise were less. Initial investigation resulted in a pub- in agreement as to the reduction of ambiguity lished report entitled, “Proposals for the Me- in the meaning of a word when only one field of chanical Resolution of German Syntax Patterns.” interest was to be considered. The engineers Although word-by-word translation seemed present fully recognized the advantage to be impossible, breaking of the German sentence gained from the reduction in size of memories into a block-by-block formation, in which each growing out of the micro-glossary concept. block has a certain specific syntactical func- Dr. Yehoshua Bar-Hillel presented the next tion, was far more profitable. Regularization paper on “Operational Syntax.” No proposal of the German language and other languages of had yet been presented to the conference re- similar structure thus appears to be dependent garding a means of programming a machine for upon such block-by-block analysis. The “Pro- recognizing syntactic connections. Dr. Bar- posals” indicate that machines can be instructed Hillel, examining this problem as a problem in to recognize syntactic connection upon this ba- symbolic logic, has discovered certain rela- sis. tionships that exist within the syntax of sen- The second major consideration for block-by- tence structure. Further, he has discovered block translation is the problem of recognizing that these can be quite readily symbolized in and interpreting the meaning-bearing words the form of symbolic fractions. A simple mul- within a block. Syntactic connections will al- tiplication of the fractions, which results in the most infallibly identify the word function and cancellation of like quantities in the numerator hence function recognition can be programmed. and denominator, results in a unique symbol Linguistic research, particularly that conducted indicative of the functions of the word block so by Prof. William E. Bull, Department of Spanish, analyzed. Use of this analysis permits ready University of California, Los Angeles, (also a recognition of word blocks functioning as nouns, participant at the conference) shows clearly verbs, adjectives, or adverbs. that the only meaning-bearing forms that can be The identification results in the ability to re- isolated are nouns, verbs, adjectives, and possi- arrange the syntax of the source language into bly adverbs. In general, of these classes, nouns the syntax of the target language. This is a are by far the most useful and used bearers of simple arithmetic operation that can be readily meaning. No system yet proposed will solve the programmed on a machine. The investigations problem of multiple significance of the meaning- to date have been preliminary, but indicate that bearing words. However, within a specific sub- the field is limited only by the number of lan- ject, a meaning-bearing word in general has guages which it would be profitable to so ana- only one specific meaning. This fact can be lyze. utilized to advantage in mechanical translation This was a completely new concept to the lin- in which the criterion of meaning is determined guists of the conference who had intuitively felt by the subject matter being considered. Dr. that such a structure did exist but without the
  5. THE CONFERENCE ON MECHANICAL TRANSLATION 51 tools of symbolic logic had been unable to iso- work required in such pre-editing and post- late the essential features that lead to the ex- editing. ceedingly simply arithmetic operations. The Assuming that syntactic considerations could engineers immediately recognized the extreme be solved by such an analysis as that proposed advantages and the simplicity of the computing by Dr. Bar-Hillel, the work of translation would loops necessary to give the machine the ability be very greatly facilitated by the use of special- to recognize word block functions and pro- ized glossaries concerned with the specific sub- grammed reorganization of sentence structure. ject matter of the material being translated. Prof. William N. Locke, Department of Mo- (Dr. Oswald terms this type of glossary a mi- dern Languages, M.I.T., presented the third cro-glossary, and the analysis that leads to it, paper on “Mechanical Translation of Printed micro semantical investigation.) and Spoken Material.” This paper was pre- The data obtained from every sort of linguis- sented orally only, no copies having been made tic frequency count when arranged according to for distribution. descending numbers forms a monotonic descend- Prof. Locke is interested in the potentiality of ing curve. The words of highest frequency drop using voice input to produce either a voice out- quite abruptly; words of medium frequency start put or a printed output. He drew on work that flattening out; and words of highly specialized has been conducted at the Bell Laboratories, at meaning that are used but seldom cause the the Haskins Laboratories, at M.I.T., and else- curve to approach the horizontal axis asympto- where on the analysis of speech and the recog- tically. The upper segment of the curve con- nition of the components that form the spoken tains the words which are usually found in the word. It appears at the present time that 8 such normal or everyday vocabulary of a language, components uniquely determined a sound. Re- and contains about 80 per cent of the actual cognition of these 8 elements leads to the iden- volume of the material. Unfortunately, these tification of one sound to the exclusion of all terms consist mainly of articles which convey other sounds. It was Prof. Locke’s contention but little meaning; the meaning-bearing forms, that a machine could be built to recognize these and in particular the nouns, are represented by 8 components and give a unique output (phoneme). the tail of the curve. All languages exhibit this The phoneme so constructed could be used with characteristic curve. Thus, in order to find other phonemes to locate a specific unit within those words conveying the major meaning in any the memory whose meaning in the target langu- text, we are concerned with the tail of the curve age would be the same as the meaning in the rather than the large grouping of words occur- source language. This of course pre-supposes ring at the beginning of the curve. Considering the utilization of the philosophy in construct- that this particular section of the curve is re- ing memories as outlined in the previous pages presentative of a micro-glossary of a specific of the conference. subject in the language, the words of this sec- The discussion of Prof. Locke’s paper was tion in general will have one and only one mean- completely speculative since devices capable of ing. so analyzing sounds are not yet in existence and To verify this assumption, Dr. Oswald ana- it appears that it will be sometime in the future lyzed nearly a hundred papers in German on the before such an art can become a science. subject of brain surgery. Technical nouns were abstracted from the first article. Additional Session IV - June 19, 1952 nouns were added from the second article, and Chairman - Dr. A. Don Booth so through the complete series of texts em- ployed. Each succeeding text was chosen from Dr. Victor A. Oswald presented the first pa- a different field of brain surgery. The amazing per, entitled “Microsemantics.” This paper fact developed that after the fourth article, the continued the analysis that Dr. Oswald had pre- glossary derived covered an average of 80 per sented on the preceding day in his discussion of cent of all the technical nouns in each succeed- word-by-word translation. He was now con- ing article. From this, he constructed a micro- cerned with the fact that, in general, editing of glossary that he considers representative of the the subject material would be required both be- field of brain surgery in the German langu- fore translation, in the source language, and age. after translation, in the target language. The A similar glossary of non-technical nouns problem is to simplify as much as possible the was also compiled from the same series of
  6. A.C.REYNOLDS, JR. 52 articles. The frequency curve of the non-tech- adequately convey to a post-editor the necessary nical nouns was the same as that of the techni- raw material to be employed in producing a cal nouns. In other words, the brain surgeons polished translation. Dr. Oswald was congratu- are not only compelled to choose their technical lated by the group for his work and analysis of nouns from a limited vocabulary, but their pat- this phenomena. tern of communication is so limited by practice Prof. William E. Bull, Department of Spanish, and convention that even the range of non-tech- University of California, Los Angeles, presented nical nouns is predictable. the second paper entitled “Frequency Problems We may generalize, although perhaps danger- in Mechanical Translation.” Prof. Bull’s inves- ously, that the same phenomenon will appear tigation in Spanish literature paralleled the in- in all technical fields of a restricted nature. vestigations of Dr. Oswald. Running texts in The micro-glossary was employed in pro- Spanish literature, which employed a general gramming translations on the SWAC in coopera- vocabulary rather than a restricted vocabulary, tion with Dr. Harry D. Huskey, Assistant Di- verify in detail the existence of the same phe- rector, National Bureau of Standards Institute nomenon in general language as occurred in for Numerical Analysis, University of Califor- the restricted field of brain surgery, but Prof. nia, Los Angeles. The translations so obtained Bull stressed that low frequency, unpredictable conveyed the meaning of the original article terms often carry critically important mean- with correlations of meaning better than 90 per ing. cent, on the assumption that the problems of Prof. Bull exhibited numerous slides showing syntax and contextual modification had pre- the frequency counts of words, the frequency viously been solved. Even without this assump- occurrence of particular parts of speech, and tion, the translated articles, when presented to the frequency counts of words within the classi- a specialist in the field, in the raw un-edited fication of a particular part of speech. He dis- form, conveyed the major portion of the mean- cussed in some detail the problem of deter- ing of the original article in the original langu- mining syntactic connections in Spanish sen- age. tences. He also discussed the type of work and The discussion that followed the paper clearly the type of personnel required to extend know- showed that the linguists working in other lan- ledge in this field not only for Spanish but also guages than German were in complete agree- for other languages of interest. ment as to the ease with which such micro- Prof. Bull's paper was in part abstracted glossaries could be constructed. The engineers from a monograph not yet published. There- and scientists, from their knowledge of techni- fore, he did not present a written paper to the cal articles in their respective fields, indicated participants of the conference, and this ma- that the size of micro-glossaries in these fields terial is at present unavailable. would be as small in comparison to the com- Substantially, Prof. Bull’s paper was a veri- plete vocabulary of a language as Dr. Oswald fication of the work of Dr. Oswald and indicated postulated. All agreed that the use of such the fruitfulness of this approach to the problem micro-glossaries would enormously reduce the of Mechanical Translation. A discussion of the amount of memory required in a translating means required to further extend the investiga- machine. tions showed clearly that the analysis could be In particular, the discussion centered on iso- facilitated by the use of punched cards. Such lation of nouns as the major meaning-bearing mechanization can enormously increase our words of a language. A rough analysis was knowledge of language structure, whereas the made of the language being used around the present handwritten and hand-sorting techniques table, and it was quite evident that in general are far too slow to materially aid in the solu- verbs employed in conveying meaning through tion of the problems of mechanical translation. speech are in the present tense and in the vast Prof. Bull accepted the suggestion that he in- majority of cases the verb is a form of the verb vestigate the possibilities of employing punched “to be.” Since information is adequately con- cards as a means of extending the scope of his veyed by speech, it seemed reasonable to the research. participants that a translation which would ig- The third paper was presented by Prof. Erwin nore tenses and concentrate on nouns which - Reifler and was entitled “General Mechanical in newspaper parlance - convey the who, what, Translation and Universal Grammar.” Prof. when, where, and how, of a statement, would Reifler has inaugurated a new school of linguis-
  7. THE CONFERENCE ON MECHANICAL TRANSLATION 53 tic investigation which is currently known as collaboration with Dr. Oswald. The problems “Comparative Semantics.” Prof. Reifler has encountered were, to a certain extent, peculiar to been investigating languages in order to dis- the SWAC, which was the machine available cover such patterns of verbally conveying for the translation. The basic problems were meaning, underlying the actual words and syntax the construction of a vocabulary for entry into of a language, as are common to all languages. the machine, the derivation of a system of ad- Such a structure could form a “universal gram- dressing to find particular units in the memory, mar.” and the syntactic programming to obtain cor- Mechanical translation poses the following rect sentence structure in the output of the ma- question: “Is it possible to solve the problems chine. These problems are basic to any ma- of Mechanical Translation in such a way that chine translation. In general, the design of the one and the same preparation of the code text machine will govern the type of programming may serve for a Mechanical Translation into required. The use of two types of memories many different languages?” The existence of seems desirable – the first having short access a universal grammar would most assuredly time and the second, which will contain words assist in the solution of this problem if such a of infrequent use, having a longer access time. grammar could be shown to exist. To date, the The arithmetic operations required for the con- science of linguistics states that no such uni- struction of the correct sentence structure will versal grammar exists, but linguists do speak of be dependent upon the arithmetic devices pro- language universals. In particular, many vided with the machine. The complexity of the highly interesting cases of parallel develop- machine, if a machine is constructed for the ment in the evolution of the expression of sole purpose of mechanical translation, will be meaning amongst structurally unrelated langu- a function of the degree of accuracy required in ages do exist. The universals may be used to the translation. This in turn will be dependent re-adjust the language structure to form what upon the allocation of time for pre-editing the Prof. Reifler terms “adjusted model target material for machine input and post-editing of languages.” This is in line with the recommen- the machine output. dation that Prof. Stuart C. Dodd presented in The second paper was presented by Mr. J. W. his paper on “Model English.” Use of the ad- Forrester, Director of Digital Computer Labo- justment clearly simplifies the mechanical ratory, M.I.T., on the subject of “Problems of translation problem and the engineering re-quired Storage and Cost.” for its solution. This also was presented in the form of a talk, The discussion of the paper reinforced the no written material being distributed. conclusions of the discussion on Prof. Dodd’s Mr. Forrester presented no cost items that paper. It is encouraging to note that where are not known to computers and business ma- Prof. Dodd has restricted his considerations to chine engineers. His major purpose was to in- English and hypothesized extension to other dicate to the linguists present the cost of the languages, Prof. Reifler, working from a com- machine that they were proposing. Techniques pletely different viewpoint and another purpose employing magnetic drums, magnetic tapes, and in mind, arrived at the same conclusions as to electrostatic storage devices singly and in com- the feasibility of regularizing a language and bination with one another were presented for further demonstrated our ability to regularize consideration. The most economical array con- major language groups of the world. sists of an intermediate memory and computing unit of low access time and a large scale mem- Session V - June 20, 1952 ory of long access time. The cost of the ma- Chairman - Prof. Wm. E. Bull chine is dependent on the same considerations as listed by Dr. Huskey. The third paper was presented by Dr. A. Don- Dr. Harry D. Huskey, Assistant Director, ald Booth, Director, Computation Laboratory, National Bureau of Standards Institute for Nu- Birkbeck College, London. The title was merical Analysis, University of California, Los changed from that listed in the program to Angeles, presented the first talk, “Basic Ma- “Some Methods of Mechanized Translation,” chine Operations in Mechanical Translation.” which was written in collaboration with Dr. R. No paper was prepared for distribution to the H. Richens of the Biological Laboratories of the members of the conference. University of London. General principles of Dr. Huskey reviewed the problems encoun- mechanical translation, as scheduled and pro- tered in programming German translations in
  8. A.C.REYNOLDS, JR. 54 grammed on the computer built by Dr. Booth IBM. Fundamentally, the same conclusions as for the University of London, were discussed. to memory and access times were reached by The use of punched card machinery was com- Prof. Perry as had been previously derived by pared with the use of an automatic digital com- the other participants in the conference. puter. Time comparisons were worked out that Session VI - June 20, 1952 favored the use of the automatic digital compu- Chairman - Prof. Wm. E. Bull ting machinery by a time ratio of at least 7 to 1. Examples of translations in the field of genetics The closing session of the conference was from Albanian, Danish, Dutch, Finnish, French, devoted to a consideration of organization for German, Hungarian, Indonesian, Italian, Latin, future research. A seven-man committee was Latvian, Norwegian, Polish, Portuguese, Ruma- organized at this session to act as coordinators nian, Spanish, Swedish, Turkish, Arabic, and and consultants for further work. The commit- Japanese were given. Usable translations in tee is composed of Dr. Yehoshua Bar-Hillel, each of these cases, despite the limited storage as chairman; Prof. Leon Dostert, secretary; available with Dr. Booth’s computer, were ob- and Dr. Olaf Helmer, Dr. Harry D. Huskey, tained. Post-editing was necessary in all cases, Prof. Erwin Reifler, and Mr. A. C. Reynolds, however, to produce a readable, although not Jr., as members. Dr. A. Donald Booth was necessarily more intelligible translation. placed on the committee as the European re- The fourth paper was presented by Prof. Wm. presentative. E. Bull and was concerned with the possible In the organization for future research, the future effect of the concept of mechanical trans- conferees were asked to what degree they were lation on the teaching of foreign languages. Prof. interested in future work and in which areas Bull stated that the concept of mechanical trans- they wished to participate. lation necessitates a completely new approach to Dr. Booth will continue with the work he has the problem of language teaching. An analogy already started with Dr. R. H. Richens at the was drawn between a machine into whose mem- University of London. ory a vocabulary had not been incorporated and a Prof. Bull is interested in the field of linguis- student into whose brain such a vocabulary tic problems of translation and as part of his must also be introduced. The approach in research activity will continue with his study of teaching syntactic connections to both the ma- the Spanish language. He is not concerned with chine and to the student in terms of the pro- mechanical translation as such, but recognizes gramming required to obtain syntactically cor- the necessity for, and the value of, his linguis- rect constructions from the memory storage was tic work in reaching this goal. discussed. Prof. Bull reached the conclu- Dr. Dodd will continue his work in the studies sions that the same considerations that govern the of regularizing languages and determine the de- choice of vocabulary and the use of inter-mediate gree of extension possible in languages other and large scale memories in the ma-chine could than English. be advantageously incorporated into the teaching Prof. Dostert intends to work actively, of languages as well as the design through the Institute of Languages and Linguis- of machines for mechanical translation. tics, Georgetown University, in the derivation Dr. Louis N. Ridenour was unfortunately un- of principles for the use of machines in trans- able to attend the conference, and his paper on lation. “Learning Machines” was not presented. Dr. Olaf Helmer stated that the Rand Corpor- In his place, Prof. James W. Perry, Research ation is interested from the theoretical view- Associate, Center for International Studies, point, but in all probability at the present time M.I.T., presented a paper on “Machine Techni- will confine itself only to theoretical work as ques for Index Searching and for Machine Trans- secondary to its work on computers. lation.” This paper was an elaboration of the Dr. Huskey had no comment other than that he talk that Prof. Perry presented at the opening would continue to collaborate with Prof. Oswald. public session of the conference. To a con- Prof. Oswald is interested in extending the siderable extent, the concepts in the paper were concept of micro-glossaries and in the study of based on Prof. Perry’s experience in setting up syntactic relations. He intends to continue work coding and indexing systems for hand-sorted in the programming of translation for machines. punched cards, and also on his experience with Prof. Reifler is extremely interested in de- the library-cataloging machine developed by monstrating the existence of universals in gram-
  9. THE CONFERENCE ON MECHANICAL TRANSLATION 55 mar, and in applying these universals to the vitally interested in this problem and, like Dr. problem of mechanical translation. Wiesner, if a concrete proposal for such a Dr. Bar-Hillel will continue his basic re- translation and subsequent demonstration could search in symbolic logic and its applications to be formulated, the Department of Defense would the field of mechanical translation. be prepared to give financial backing. Dr. Jerome B. Wiesner, speaking for the Mr. Reynolds stated that IBM was interested M.I.T. staff present, stated that the research in the application of its present punched card laboratory at M.I.T. is very much interested in techniques and its computers to this problem the application of computer techniques to the and as such would participate on the basis of problem of mechanical translation and that if a exchange of theoretical information with the concrete program was formulated, financial members of the conference. support could quite conceivably be forthcoming The conference closed on a note of optimism from the Research Laboratory. regarding the potentialities now known to be Mr. Duncan Harkin of the Department of De- physically present in the concept of mechanical fense stated that the Department of Defense was translation.
ADSENSE

CÓ THỂ BẠN MUỐN DOWNLOAD

LV.26: Bộ 320 Luận Văn Thạc Sĩ Y Học
320 tài liệu
1247 lượt tải
THÔNG TIN
TRỢ GIÚP
HỖ TRỢ KHÁCH HÀNG
Theo dõi chúng tôi

Chịu trách nhiệm nội dung:

Nguyễn Công Hà - Giám đốc Công ty TNHH TÀI LIỆU TRỰC TUYẾN VI NA

LIÊN HỆ

Địa chỉ: P402, 54A Nơ Trang Long, Phường 14, Q.Bình Thạnh, TP.HCM

Hotline: 093 303 0098

Email: support@tailieu.vn

Giấy phép Mạng Xã Hội số: 670/GP-BTTTT cấp ngày 30/11/2015 Copyright © 2022-2032 TaiLieu.VN. All rights reserved.

 

Đồng bộ tài khoản
2=>2