{ "paper_id": "1989", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T03:26:19.244756Z" }, "title": "Herzlich willkommen in M\u00fcnchen! Welcome to Munich! Welcome to MT Summit II! A: conference chairman I am glad and relieved that so many have responded to our invitation. We are more than 250 participants from all over the world. Thank you for coming! Please do participate actively in the conference in order to make it a success", "authors": [], "year": "", "venue": null, "identifiers": {}, "abstract": "What is the aim of the second MT Summit? Like the first MT Summit in Hakone, Japan, this conference brings together users, research and development experts and government policy makers in order to promote practical MT systems. Why do we need MT? A global economy presupposes worldwide circulation of information. This flow of information is hampered by the language barrier. The information, however, must be available in the national language of the user. Therefore it has to be translated. The demand for translation is very high. In western Europe alone more than 150 million pages were translated last year. There are not enough qualified translators, the costs are enormous. Consequently, MT is the only solution. The use of one language, like English, as a world language is ruled out because we must preserve the linguistic and cultural identity of our partners. We don't want to reduce our national languages to the role of dialects, limited to folkloristic purposes. As I said, the first MT Summit was convened in Japan. This was no accident. Japan has made the greatest effort to develop practical MT systems. Fifteen private companies have developed research prototypes and/or commercial systems. In addition there are projects sponsored by the eminent and the very ambitious attempt to translate even telephone conversations automatically (ATR Interpreting Telephony Research Institute in Kyoto). We had many applications from Japan. Unfortunately we could not provide time slots for the presentation of all the systems. Therefore several systems are only present at the exhibition. In comparison with Japan, Europe invests much less in MT. The EC is funding one big project but the overall amount spent in Europe is far below the Japanese figures. Especially computer makers in Europe and in the USA show far less enthusiasm than their Japanese competitors. This is, of course, reflected in the programme of our meeting and in the accompanying exhibition.", "pdf_parse": { "paper_id": "1989", "_pdf_hash": "", "abstract": [ { "text": "What is the aim of the second MT Summit? Like the first MT Summit in Hakone, Japan, this conference brings together users, research and development experts and government policy makers in order to promote practical MT systems. Why do we need MT? A global economy presupposes worldwide circulation of information. This flow of information is hampered by the language barrier. The information, however, must be available in the national language of the user. Therefore it has to be translated. The demand for translation is very high. In western Europe alone more than 150 million pages were translated last year. There are not enough qualified translators, the costs are enormous. Consequently, MT is the only solution. The use of one language, like English, as a world language is ruled out because we must preserve the linguistic and cultural identity of our partners. We don't want to reduce our national languages to the role of dialects, limited to folkloristic purposes. As I said, the first MT Summit was convened in Japan. This was no accident. Japan has made the greatest effort to develop practical MT systems. Fifteen private companies have developed research prototypes and/or commercial systems. In addition there are projects sponsored by the eminent and the very ambitious attempt to translate even telephone conversations automatically (ATR Interpreting Telephony Research Institute in Kyoto). We had many applications from Japan. Unfortunately we could not provide time slots for the presentation of all the systems. Therefore several systems are only present at the exhibition. In comparison with Japan, Europe invests much less in MT. The EC is funding one big project but the overall amount spent in Europe is far below the Japanese figures. Especially computer makers in Europe and in the USA show far less enthusiasm than their Japanese competitors. This is, of course, reflected in the programme of our meeting and in the accompanying exhibition.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "The emphasis of this conference is on practical MT. It is not a conference where computational linguists and computer scientists discuss special problems in technical jargon. We have three sessions where operational MT systems are presented. Furthermore we have a panel on practical experience in the application of MT systems. At this panel you can ask questions like what results have been achieved so far? Where and how can MT systems be used? Why don't more companies use MT systems? Have these systems been introduced prematurely? Please don't hesitate to ask questions at the panels! We need your active participation at this conference and in the future.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Besides operational systems there will be presentations of systems under development. We intend to give you a global view and have included systems from Europe, China, Japan and America. Of special interest are the multilingual projects: EUROTRA, the project of the European Community which should lead to a system translating between the nine languages of the EC., and ODA, a project of an MT system for Japan and its neighbouring countries. ODA intends to translate between Chinese, Indonesian, Malaysian, Thai and Japanese. In the panel on new directions in MT systems we will try to look a bit into the future. We use natural language for representing, storing and processing information. If one really believes that the layman, the casual user, will be able, one day, to interact naturally with the computer, then natural language is the optimal instrument. MT is only one application in the area of Human-Computer-Interaction, but it is one of the most difficult and challenging certainly. Research on MT can lay the basis for future high-level knowledge and information processing.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Research on MT is very expensive. Therefore we have to find ways to share the research and development costs in the precompetitive phase. To give one example: every MT system which translates from German into another language will need an electronic dictionary of German. Why don't we pool our resources and develop a common German machine-readable dictionary? Or if we think in terms of Europe '92, why don't we share the costs of electronic dictionaries for the languages of the European Community? We must cooperate. Or, to quote La Rochefoucauld, the famous French philosopher and moralist of the 17th century: \"It would foolish to try to be smart alone\". Which form such a cooperation should take is one topics of the panel on governmental views of MT. The political decision makers, however, should not only discuss administrative measures. One of the main reasons why MT is not more advanced and not used more widely lies in our educational system. The study of language at the universities and at the school for interpreters is part of the humanities. In the humanities computers are still very often rejected on ideological grounds. When they are used, they have a purely instrumental function, they don't serve as a tool for creative research. A few weeks ago the Wissenschaftsrat, the highest body which advises the German Federal Government on questions of university education, recommended that every university student (irrespective of what he studies) should take a four-hour semester course in computer science. I don't know of any German university which already fulfills this requirement. A translator certainly needs more than a four-hour course, but at the moment he does not even get that. In addition all students must have access to adequate computers. The more a user of an MT system knows about language, computers and computational linguistics, the more he can profit by the system. Most potential users today have an unrealistic picture of what a computer can do with natural language.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Another angle from which we have to approach MT is technical writing. MT is not for poetry or fiction. -Syntactically and semantically political speeches also fall under fiction. -MT is for technical texts. Now who writes technical texts? Where do you learn to write good technical texts? The better a text is written, the easier it is to translate, for human translators as well as for machines. Two steps are necessary. On the one hand, courses on technical writing should be established at university level, preferably at technical universities where the future technical writer can at the same time study the subject he will write about later. On the other hand, more research is necessary on technical language and sublanguages in general. How can one produce texts which do not contain ambiguities, or at least only very few ambiguities? Is it necessary to restrict the number of syntactic constructions and the words which can be used in order to avoid ambiguities? Can we make MT fully automatic by controlling the input and still produce texts which don't bore the people for whom they are written? For me these are fascinating questions? Why shouldn't they also fascinate some students and professors of German, especially if the student finds a job afterwards?", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "At the beginning of our talk we said that a global society, a world society, needs a free flow of information and that it is translation which makes this flow of information possible. Johann Wolfgang von Goethe, a Weltb\u00fcrger, a citizen of the world, par excellence clearly recognized this fundamental task of translation when he wrote: \"Denn was man auch von der Unzul\u00e4nglichkeit des \u00dcbersetzens sagen mag, so ist und bleibt es doch eines der wichtigsten und w\u00fcrdigsten Gesch\u00e4fte in dem allgemeinen Weltverkehr\". (\"Kunst und Altertum\", Artemis, Gedenkausgabe, Band 14, S. 932). Translated literally: Whatever one may say about the shortcomings of translation, it is and remains one of the most important and respected occupations in world communication, trade and traffic. (Following Herder, Goethe uses 'Verkehr' in the sense of communication, trade and traffic).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "I hope that MT Summit II will succeed in promoting translation in general and machine translation in particular. we all hope that the MT SUMMIT II will be interesting for you and for the organizers. In Hakone, I announced our intention to hold this convention in our old German capital of Berlin; but our efforts in this respect were not successful; I am therefore happy to welcome you to the Bavarian capital -which is in fact frequently called Germany's secret capital and -with regard to the weather -shows its best side, today.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "The Federal Minister for Research and Technology supports work in the fields of MT and computer linguistics for three reasons: 1) Machine translation systems require computer linguistics research and development in the analysis of the source language, in transfer and Interlingua problems and in the generation of the target language. R&D results must lead to software developments, which can then help to realize the translation. Analysis, transfer or Interlingua and generation are directly connected with basic linguistic questions and the related basic research. Computer linguistics, software development and basic linguistic research are a complex R&D field. The object of research is the still enigmatic structure of human language and of individual languages in the field of tension between practical handling in translation and theoretical knowledge and profound understanding. It seems to be expedient therefore that at the MT SUMMIT II, MT suppliers and leading computer linguists together attempt to present and explain to the translators the state-of-the-art.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Considering the increasing internationalization and globalization of scientific, economic and cultural developments, the BMFT feels it is his task to make a supportive contribution to this R&D field.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "2) The BMFT supports R&D, it does not, however, support commercial MT systems because this is the task of industry, which is also responsible for commercialization. Provision of government support in this field would always contort competition, In addition, translators and their representatives fear that the use of MT systems might take their jobs away. This fear is understandable -against the background of a general reservation about computers -but, in my view, it is irrelevant at the present time. And it is very unlikely to materialize in the future for the following reasons:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "a) The demand for translations is increasing in many areas at an above-average rate, in particular with regard to long technical texts and manuals. b) Machine translation systems will always be developed further in order to make ever better translations, which will not, however, be perfect for a long time into the future. Pre-and post-editing of such machine translations will always call for the translator. However, translators -and rightly so I think demand a certain quality standard of machine translations, since pre-and postediting must not, of course, take longer than direct translation by the human translator.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Here, the complex semantic problems, which have not yet been solved satisfactorily play an essential role; the human translator will always be better than any machine. For this very reason, it seems important to me that MT systems be used to support translation. The exhibition provides you with quite a comprehensive survey of systems which are either already on the market or in the development stage.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "3) The BMFT also provides support for scientific and technical information including its dissemination via electronic data bases. Owing to the existence of worldwide telecommunications networks, this information market is a world market par excellence, in which buying and selling of information is a matter of seconds. This world market demands that data bases be offered in the English language; it has become almost general practice for German information products such as the Beilstein data base and physics or mathematics bibliographic data bases offered in the STN Network. The German language, which was once the leading language science, has been replaced by the English language after the First and Second World Wars. Of course, we also face translation problems in this field when establishing or using such data bases, and here MT systems might acquire a new significance. However, to put it somewhat pointedly, the computer and the English language have a particular affinity for each other, which the German language and the computer do not have. For this reason and on account of the scientific, economic and cultural position of the Federal Republic of Germany, another concern of the Federal Government is that the German language be included in MT systems, at least in the language pair German-English, but in future preferably also in other language pairs.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "These are the main ideas behind our promotion activities, which will be explained in detail in the Panel II. In conclusion, I wish the SUMMIT II every success and continuation in two years. Thank you. Meine Damen und Herren, noch ist es viel zu fr\u00fch, in der maschinellen \u00dcbersetzung das Ende jener Sprachverwirrung zu erkennen, die mit dem Turmbau zu Babel ihren Anfang nahm. Auch die Sch\u00fcler an unseren Schulen sollten sich nicht zu fr\u00fch freuen und glauben, sie k\u00f6nnten das \u00dcbersetzen bald ebenso den Maschinen \u00fcberlassen wie jetzt schon h\u00e4ufig das Kopfrechnen. Ich w\u00fcnsche Ihnen, meine Damen und Herren, jedenfalls nicht, da\u00df sie eine maschinell ins Englische \u00fcbersetzte Fassung meines Gru\u00dfworts lesen m\u00fcssen.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Vielleicht aber kann ich Sie eines Tages auf Bayerisch begr\u00fc\u00dfen, und eine Maschine \u00fcbersetzt es Ihnen simultan ins Hochdeutsche, Englische oder welche Sprache auch immer Sie gerne h\u00e4tten. In diesem Sinn w\u00fcnsche ich Ihrer Tagung einen erfolgreichen Verlauf und Ihnen dar\u00fcber hinaus noch gen\u00fcgend Zeit, den Charme und die Sch\u00f6nheit Bayerns und seiner Landeshauptstadt kennenzulernen und auf sich wirken zu lassen.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [], "bib_entries": {}, "ref_entries": { "FIGREF0": { "text": "in conveying to you the best wishes of the Federal Government and of the Federal Minister for Research and Technology, Dr. Heinz Riesenhuber;", "uris": null, "num": null, "type_str": "figure" }, "TABREF0": { "content": "
wirtschaftliche Bedeutung auf der Hand. Man braucht sich
Anrede! F\u00fcr die Einladung zum 2. Gipfeltreffen f\u00fcr maschinelle \u00dcbersetzung danke ich Ihnen herzlich. Es ist mir eine gro\u00dfe Freude, die inter-national f\u00fchrenden Experten auf diesem in die weite Zukunft weisenden Gebiet hier in M\u00fcnchen zu begr\u00fc\u00dfen. Ich tue dies auch im Namen der Bayerischen Staatsregierung und speziell im Namen des Bayerischen Staatsministeriums f\u00fcr Wissenschaft und Kunst. Da\u00df Sie, beispielsweise nur vorzustellenEin Politiker verbindet mit den Fortschritten der maschinellen meine Damen und Herren, M\u00fcnchen als Tagungsort gew\u00e4hlt haben, werte \u00dcbersetzung zum Beispiel die Erwartung, da\u00df das umfangreiche ich als Zeichen der gro\u00dfen Bedeutung, die der bayerischen Schriftgut der EG k\u00fcnftig rechtzeitig auch in deutscher Fassung Landeshauptstadt als Forschungs-und Industriestandort inzwischen vorliegt. Die Freude dar\u00fcber w\u00e4re allerdings getr\u00fcbt, wenn die Texte zukommt. dann noch holpriger und schwerer verst\u00e4ndlich ausfielen,als sie es
bisher schon der Fall ist.
Durch maschinelle \u00dcbersetzung sollen Texte und sogar gesprochene
Sprache schneller, als dies bisher m\u00f6glich ist, und zugleich gut Vielleicht lie\u00dfe sich durch maschinelle \u00dcbersetzung auch ein Trend \u00fcbersetzt werden. Das Anliegen, die Produktivit\u00e4t beim \u00dcbersetzen zu aufhalten, der im wissenschaftlichen Publikationswesen unseres steigern, wird angesichts der bestehenden Sachlage \u00fcberaus Landes zu beobachten ist. Immer h\u00e4ufiger wird n\u00e4mlich in Englisch verst\u00e4ndlich. So hei\u00dft es in einer Pressemitteilung zu dieser und nicht in der Muttersprache ver\u00f6ffentlicht. Ich jedenfalls w\u00fcrde Veranstaltung, da\u00df 1986 allein in Westeuropa ca. 100 Mio Seiten mich freuen, wenn dem Deutschen als Wissenschaftssprache wieder mehr \u00fcbersetzt wurden und weltweit jedes Jahr viele Milliarden ausgegeben werden, um Informationen in einer anderen Sprache zu vermitteln. Geltung verschafft werden k\u00f6nnte,
Wie kaum ein anderer ist der Forschungszweig \"Maschinelle Wachsende internationale Verflechtungen machen es immer \u00dcbersetzung\" auf interdisziplin\u00e4re Zusammenarbeit angewiesen. F\u00fcr selbstverst\u00e4ndlicher, mit Partnern aus verschiedenen k\u00fcnftige Erfolge auf dem Gebiet der maschinellen Texterfassung ist Sprachgemeinschaften zu kommunizieren, Dies gilt seit jeher f\u00fcr es unerl\u00e4\u00dflich, da\u00df Informatiker, Linguisten, Psycholinguisten und Wissenschaft und Wirtschaft, aber in zunehmendem Ma\u00df auch f\u00fcr die sicherlich noch eine Reihe weiterer Vertreter verschiedener Politik, man denke etwa an die fortschreitende Einigung Westeuropas mit dem Fernziel einer Europ\u00e4ischen Union, Fachdisziplinen intensiv zusammenarbeiten.
Grunds\u00e4tzlich sind zwei Wege gangbar, um Sprachgrenzen zu
\u00fcberwinden. Man konnte sich zum einen auf eine f\u00fcr alle verbindliche
Sprache einigen und auf diese Weise das Problem der \u00dcbersetzung
umgehen. Dieser Weg wurde und wird beschritten -ich erinnere an die
Rolle des Lateinischen in vergangener Zeit oder auch an die ganz
\u00e4hnliche Funktion, die das Englische gegenw\u00e4rtig erf\u00fcllt.
Bei allen Vorteilen, die eine einheitliche Sprache bietet, bleiben
doch gravierende Nachteile bestehen. Wer sich einer fremden Sprache
bedienen mu\u00df, ist h\u00e4ufig nicht in der Lage, mit der von der Sache
her gebotenen Differenziertheit und Subtilit\u00e4t zu formulieren, sich
dem Gespr\u00e4chspartner mitzuteilen. Mi\u00dfverst\u00e4ndnisse sind dann
unausweichlich.
Auf lange Sicht wird die Dominanz einer einzelnen Sprache aber auch
kulturelle Unterschiede einebnen. Gerade im Hinblick auf die
Einigung Europas, das, wie ich meine, zu Recht stolz ist auf seine
kulturelle Vielfalt, sollte deshalb die Bedeutung und das Gewicht
der verschiedenen Nationalsprachen erhalten, ja gest\u00e4rkt werden.
Damit stehen wir wieder vor dem Problem der \u00dcbersetzung. Sie ist der
zweite Weg, Sprachbarrieren zu \u00fcberwinden.
Was die maschinelle \u00dcbersetzung anbetrifft, so liegt ihre
", "text": "Gru\u00dfwort des Staatssekret\u00e4rs im Bayerischen Staatsministerium f\u00fcr Wissenschaft und Kunst, Dr. Thomas Goppel, MdL, anl\u00e4\u00dflich des 2. Gipfeltreffens zur maschinellen \u00dcbersetzung (Machine Translation Summit II) am 16.08.1989 in M\u00fcnchen. , da\u00df die Lieferung eines \"fertigen\" Produkts an einen ausl\u00e4ndischen Empf\u00e4nger verz\u00f6gert wird, nur weil die \u00fcbersetzte Fassung der Bedienungsanleitung noch nicht vorliegt. Unter \u00f6konomischem Aspekt erscheint das untragbar. Eine leistungsf\u00e4hige maschinelle \u00dcbersetzung w\u00e4re deshalb f\u00fcr ein exportorientiertes Industrieland wie die Bundesrepublik Deutschland ein Faktor von \u00fcberragender wirtschaftlicher Bedeutung -w\u00e4re, weil in diesem technologischen Neuland die Grenzen, wie mir scheint, manchmal noch flie\u00dfend sind. Ich bin allerdings sehr neugierig und lasse mich gern eines besseren belehren.Einen Beitrag zur F\u00f6rderung dieser interdisziplin\u00e4ren Kooperation darf man sich z.B. von dem Lehrstuhl f\u00fcr Computerlinguistik erwarten, der an der Universit\u00e4t M\u00fcnchen neu geschaffen wurde. Das Besetzungsverfahren ist allerdings noch nicht abgeschlossen. Des weiteren k\u00f6nnen sich im Rahmen einer Kooperation zwischen der Firma Siemens und der Universit\u00e4t M\u00fcnchen derzeit zwei Linguisten bei Siemens weiterbilden.", "num": null, "type_str": "table", "html": null } } } }