<?xml version="1.0" standalone="yes"?>
<Paper uid="P96-1023">
  <Title>INVITED TALK Head Automata and Bilingual Tiling: Translation with Minimal Representations</Title>
  <Section position="10" start_page="174" end_page="175" type="concl">
    <SectionTitle>
7 Language Processing and
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="174" end_page="175" type="sub_section">
      <SectionTitle>
Semantic Representations
</SectionTitle>
      <Paragraph position="0"> The translation system we have described employs only simple representations of sentences and phrases.</Paragraph>
      <Paragraph position="1"> Apart from the words themselves, the only symbols used are the dependency relations R. In our experimental system, these relation symbols are themselves natural language words, although this is not a necessary property of our models. Information coded explicitly in sentence representations by word senses and feature constraints in our previous work (Alshawi 1992) is implicit in the models used to derive the dependency trees and translations. In particular, dependency parameters and context-dependent transfer parameters give rise to an implicit, graded notion of word sense.</Paragraph>
      <Paragraph position="2"> For language-centered applications like translation or summarization, for which we have a large body of examples of the desired behavior, we can think of the task in terms of the formal problem of modeling a relation between strings based on exampies of that relation. By taking this viewpoint, we seem to be ignoring the intuition that most interesting natural language processing tasks (translation, summarization, interfaces) are semantic in nature.</Paragraph>
      <Paragraph position="3"> It is therefore tempting to conclude that an adequate treatment of these tasks requires the manipulation of artificial semantic representation languages with well-understood formal denotations. While the intuition seems reasonable, the conclusion might be too strong in that it rules out the possibility that natural language itself is adequate for manipulating semantic denotations. After all, this is the primary function of natural language.</Paragraph>
      <Paragraph position="4"> The main justification for artificial semantic representation languages is that they are unambiguous by design. This may not be as critical, or useful, as it might first appear. While it is true that natural language is ambiguous and under-specified out of context, this uncertainty is greatly reduced by context to the point where further resolution (e.g.</Paragraph>
      <Paragraph position="5"> full scoping) is irrelevant to the task, or even the intended meaning. The fact that translation is insensitive to many ambiguities motivated the use of unresolved quasi-logical form for transfer (Alshawi et al. 1992).</Paragraph>
      <Paragraph position="6"> To the extent that contextual resolution is necessary, context may be provided by the state of the language processor rather than complex semantic representations. Local context may include the state of local processing components (such as our head automata) for capturing grammatical constraints, or the identity of other words in a phrase for capturing sense distinctions. For larger scale context, I have argued elsewhere (Alshawi 1987) that memory activation patterns resulting from the process of carrying out an understanding task can act as global context without explicit representations of discourse.</Paragraph>
      <Paragraph position="7"> Under this view, the challenge is how to exploit context in performing a task rather than how to map natural language phrases to expressions of a formalism for coding meaning independently of context or intended use.</Paragraph>
      <Paragraph position="8"> There is now greater understanding of the formal semantics of under-specified and ambiguous representations. In Alshawi 1996, I provide a denotational semantics for a simple under-specified language and argue for extending this treatment to a formal semantics of natural language strings as expressions of an under-specified representation. In this paradigm, ordered dependency trees can be viewed as natural language strings annotated so that some of the implicit relations are more explicit. A milder form of this kind of annotation is a bracketed natural language string. We are not advocating an approach in which linguistic structure is ignored (as it is in the IBM translator described by Brown et al. 1990), but rather one in which the syntactic and semantic structure of a string is implicit in the way it is processed by an interpreter.</Paragraph>
      <Paragraph position="9"> One important advantage of using representations that are close to natural language itself is that it reduces the degrees of freedom in specifying language and task models, making these models easier to ac- null quire automatically. With these considerations in mind, we have started to experiment with a version of the translator described here with even simpler representations and for which the model structure, not just the parameters, can be acquired automatically. null</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>