<?xml version="1.0" standalone="yes"?>
<Paper uid="C90-3003">
  <Title>Backwards Phonology</Title>
  <Section position="11" start_page="0" end_page="0" type="concl">
    <SectionTitle>
11 Conclusion
</SectionTitle>
    <Paragraph position="0"> The question has been asked, &amp;quot;What is so good about Koskenniemi's two-level phonology?&amp;quot; The answer is that it allows one to write reversible, nonprocedural descriptions of phonological phenomena with much more accuracy than does the conventional unilevel formalism. The point I have stressed here is the reversibility. From a computational point of view, this represents a step forward. There are no published accounts of reversible grammars written in a unilevel formalism so far as I know and there are many written in two-level rules. Koskenniemi's proposal was made with computation ill mind as opposed to linguistic theory. It may, in the long run, have an impact on linguistic theory. It definitely has had a large impact on computational morphology.</Paragraph>
  </Section>
class="xml-element"></Paper>