<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-2102">
  <Title>Design of a Lexical Database for Sanskrit</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> The computational linguistic tools should be modular, with an open-ended structure, and their evolution should proceed in a breadth-first manner, encompassing all aspects from phonetics to morphology to syntax to semantics to pragmatics to corpus acquisition, with the lexical database as a core switching structure.</Paragraph>
    <Paragraph position="1"> Proper tools have to be built, so that the analytic structure is confronted to the linguistic facts, and evolves through experimentally verifiable improvements. The interlinking of the lexicon, the grammatical tools and the marked-up corpus is essential to distill all linguistic information, so that it is explicit in the lexicon, while encoded in the minimal way which makes it non-redundant.</Paragraph>
    <Paragraph position="2"> We have argued in this article that the design of an hypertext interface is useful to refine the structure of the lexicon in such a way as to enforce these requirements. However, such a linguistic platform must carefully distinguish between the external exchange formats (XML, Unicode) and the internal logical structure, where proper computational structures (inductive data types, parametric modules, powerful finite-state algorithms) may enforce the consistency invariants.</Paragraph>
  </Section>
class="xml-element"></Paper>