<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1018">
  <Title>Sydney, July 2006. c(c)2006 Association for Computational Linguistics Polarized Unification Grammars</Title>
  <Section position="6" start_page="142" end_page="143" type="concl">
    <SectionTitle>
4 Conclusion
</SectionTitle>
    <Paragraph position="0"> The PUG formalism is extremely simple: it only imposes that combining two structures involves at least the unification of two objects. Forcing or forbiding more objects to combine is then entirely controled by polarization of objects. Polarization wil thus guide the process of combination of elementary structures. In spite of its simplicity, the PUG formalism is powerful enough to elegantly simulate most of the rule-based formalisms used in formal linguistics and NLP. This sheds new light on these formalisms and allows us to bring to the fore the exact nature  of the structures they handle and to extract some procedural mechanisms hiden by the formalism.</Paragraph>
    <Paragraph position="1"> But above all, the PUG formalism allows us to write separately several modules of the grammar handling various structures and to put them together in a same formalism by synchronization of the grammars, as we show with our translation of LFG. Thus PUGs extend unification grammars based on feature structures by allowing a greatest diversity of geometric structures and a best control of resources. Further investigations must concern the computational properties of PUGs, notably restrictions allowing polynomial time parsing.</Paragraph>
  </Section>
class="xml-element"></Paper>