<?xml version="1.0" standalone="yes"?>
<Paper uid="P84-1051">
  <Title>MACHINE TRANSLATION : WHAT TYPE OF POST-EDITING ON WHAT TYPE OF DOCUMENTS FOR WHAT TYPE OF USERS</Title>
  <Section position="5" start_page="236" end_page="237" type="concl">
    <SectionTitle>
III - TYPES OF POST-EDITING
</SectionTitle>
    <Paragraph position="0"> The system used in Luxembourg is still being developped. That means that errors are getting fewer. For instance three years ago verb forms were translated &amp;quot;form to form&amp;quot;, now new rules have been introduced in order to get a past tense for a present tense (or reverse), a passive form for an impersonaI one (or reverse), a.s.o.</Paragraph>
    <Paragraph position="1"> i cf. A.M. Loffler-Laurian, Pour une tvpologie des erreurs dans la traduction automatique, in MULTiLINGUA, 2-2 (1983), 65-78 But at the same time the variety of documents machine trabslated is growing. That means new sources on errors (mainIv vocabulary, but aiso modaiities, structures, a.s.o.).</Paragraph>
    <Paragraph position="2"> Post-editing is always necessary. Until now post-editing has been done by translators who are wishing to do it. The amount of post-editing to do is increasing every day, it becomes obvious that post-editing can't be done just according to somebody's feeling of language and style. There has to be some rules.</Paragraph>
    <Paragraph position="3"> Post-editing is not revision, nor correction, nor rewriting. It is a new way of considering a text, a new wav of working on it, for a new aim. In order to define the characteristics of post-editing, I carried out a study on the two major types of post-editing as they appear in the C.E.C. 2 i. The conventional post-editing (C.P.E.) is supposed to produce a text as similar as possibie to what a human translation would have been, that means a high quality text.</Paragraph>
    <Paragraph position="4"> 2. The rapid post-editing (R.P.E.) is supposed to produce a correct text (on the language level as well as on the level of the meaning) but without taking care of the stvie.</Paragraph>
    <Paragraph position="5"> In the experiment I carried out, time required for post-editing was the only criteria to differenciate these two methods. It appeared that special Iinguistical attitudes were induced by time Iimitation. A statistical survey of C.P.E. and R.P.E. shows the limits between :  I. necessary post-editing, 2. possible post-editing, 3. superfluous post-editing.</Paragraph>
    <Paragraph position="6">  First group includes all post-editing that has to be done to make the text understandabie, clear, readable, exact. Second group inciudes some research in style focused on the adaptation to the communication situation, to the author and to the presumed reader. Third group is post-editing done bY peopIe who didn't want to admit that perfection was not the aim, and that a document that will be read quickIy and thrown away immediateIv ooes not require the same style as a oocument that will be pubiished and largeiv distributed. These people usuaIly could not give out their R.P.E. in the limited time allowed for it.</Paragraph>
    <Paragraph position="7">  In rapid post-editing one has to focus on the central information, and is naturally kept out from the temptation of rewriting the sentence were errors occur. Then the post-editor finds the shortest solution, which is usually the right one. By staying very close to the raw translation, post-editors succeed in giving a good and acceptable translation. Those who, after having post-edited according to the minimal requirements, try to make the text fit better the usual style they know, give us indications to point out the difference between : - a text that is correct according to standard language rules, - a text that obeys the usage rules in use on that level of documents or level of language (some &amp;quot;sub-rules&amp;quot; specific to some specialized fields, authors, situations).</Paragraph>
    <Paragraph position="8"> IV - STYLE, SITUATIONS AND USERS Style in literature is usually defined as the specific way an author writes. Do technical and scientific documents have a specific style ? Many people would agree on the idea that these documents have no style -or have a neutral stYle-. In terms of linguistic features, they can be described as well as any other writing. However the non-apparent aspect of style in informative documents is an important component of their ability to be machine translated. In a novel, the style of the author would be its main value whereas in an informative document, the transparency of style, its leaving the reader unaware of it would be essential. Even more : if style were to be felt, the information would most probably loose some of its accuracy and credibility.</Paragraph>
    <Paragraph position="9"> In every translation situation the author has some information to transmit to a user. Let it be a technical or a political information, a scientific or a social information, the goal may be double : have the reader know more about a question (that relates to didactics), and have the reader react in a specific way to the text. Regarding this second goal, the best style, most adequate, would be the one that would bring the reader to the point the author wanted him. The neutrality of a computerized system is quite fitted to that situation. And the minimal post-editing creates often the best style.</Paragraph>
    <Paragraph position="10"> The users' satisfaction should be the ultimate criterion to evaluate the adequacy of a style.</Paragraph>
    <Paragraph position="11"> Are readers getting used to some new style based on machine translation ? Some people fear for the future of their language: it could evolve uncontrolled because of a new kind of users getting used to some new variety of language induced by a new tool for translation. They fear a loss of some linguistical property. Languages have always been exposed to multiple influences (wars, invasions, economical trends, cultural exchanges, a.s.o.). They are now exposed to technical influences.</Paragraph>
    <Paragraph position="12"> Machine translation is already used by translation services. It will certainly be soon used by private translators (various systems are developped or under development in several countries). It could be used with great profit by linguists and professors to help them think about their own use of language, about the varieties of specialized uses of language, and about the future programmes that could be built up for new generations of students.</Paragraph>
  </Section>
class="xml-element"></Paper>