<?xml version="1.0" standalone="yes"?>
<Paper uid="P81-1005">
  <Title>REPRESENTATION</Title>
  <Section position="3" start_page="0" end_page="23" type="intro">
    <SectionTitle>
(7) E
</SectionTitle>
    <Paragraph position="0"> coronal anterio l syllabi anterior I~ high 2/-&amp;quot; high I strident ~ back J The representation provides a language in which to express hypotheses. The task is to find statements in this language to express the data. Thus the representation implicitly defines the search space. The search space is restricted by the following constraint on the 'distance' between a UR and its pronounciations. Every feature specification in the UR must be present in a 'corresponding' segment in at least one of the phonetic forms. Consider, for example, morpheme i from (3) above: it ham three pronounciations \[sarap\], \[sarab\], \[sarav\].  This constraint restricts its possible URs to /sarap/, /sarah/, /sarav/, /saraf/.</Paragraph>
    <Paragraph position="1"> Even If\] does not appear in any of the pronouciations of this morpheme, its +continuant specification occurs in Iv\] and its -voice specification occurs in \[p\]; its other feature specifications are common to \[p\], Cb\], Iv\]. This constraint is weaker than the &amp;quot;strong alternation condition&amp;quot; (cf. \[4\]), which would restrict the final UR segment to be /p/, /b/, or /V/o The term &amp;quot;alternation&amp;quot; will be important of the discussion below; here \[p\] vs. \[b\] vs. Iv\] is an alternation.</Paragraph>
  </Section>
class="xml-element"></Paper>