<?xml version="1.0" standalone="yes"?>
<Paper uid="A83-1029">
  <Title>COMPUTER-ASSISTED TRANSLATION SYSTEMS: The Standard Design and A Multi-level Design</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
I THE STANDARD DESIGN FOR A COMPUTER-
ASSISTED TRANSLATION SYSTEM.
</SectionTitle>
    <Paragraph position="0"> The standard design for a computer-assisted translation system consists of three phases: (A) data entry of the source text, (B) machine translation of the text, and (C) human revision of the raw machine translation. Most machine translation projects of the past thirty years have used this design without questioning its validity, yet it may not be optimal. This section will discuss this design and some possible objections to it.</Paragraph>
    <Paragraph position="1"> The data entry phase may be trivial if the source text is available in machine-readable form already or can be optically scanned, or it may involve considerable overhead if the text must be entered on a keyboard and proofread.</Paragraph>
    <Paragraph position="2"> The actual machine translation is usually of the whole text. That is, the system is generally designed to produce some output for each sentence of the source text. Of course, some sentences will not receive a full analysis and so there will be a considerable variation in the quality of the output from sentence to sentence.</Paragraph>
    <Paragraph position="3"> Also, there may be several possible translations for a given word within the same gramatical category and subject matter so that the system must choose one of the translations arbitrarily. That choice may of course be appropriate or inappropriate. It is well-known that for these and other reasons, a machine translation of a whole text is usually of rather uneven quality. There is an alternative to translating the whole text -na~nely, &amp;quot;selective translation,&amp;quot; a notion which will be discussed further later on.</Paragraph>
    <Paragraph position="4"> Revision of the raw machine translation by a human translator seems at first to be an attractive way to compensate for whatever errors may occur in the raw machine translation. However, revision is effective only if the raw translation is already nearly acceptable. Brinkmann (Ig8O) concluded that even if only 20% of the text needs revision, it is better to translate from scratch instead of revising.</Paragraph>
    <Paragraph position="5"> The author worked on a system with this standard design for a whole decade (from 1970 to 1980). This design can, of course, work very well. The author's major objection to this ~esign is that it must be almost perfect or it is nearly useless.</Paragraph>
    <Paragraph position="6"> In other words, the system does not become progressively more useful as the output improves from being 50% correct to 60% to 70% to 80% to 90%.</Paragraph>
    <Paragraph position="7"> Instead, the system is nearly useless as the output improves and passes some threshold of quality. Then, all of a sudden, the system becomes very useful. It would, of course, be preferable to work with a design which allows the system to become progressivelv more useful.</Paragraph>
    <Paragraph position="8"> Here is a summary of objections to the standard design: WHY COMPUTATIONAL LINGUISTS 00 NOT LIKE IT: Because even if the algorithms start out &amp;quot;clean&amp;quot;, they must be kludged to make sure that somethino comes out for every sentence that goes in.</Paragraph>
  </Section>
class="xml-element"></Paper>