<?xml version="1.0" standalone="yes"?>
<Paper uid="W96-0311">
  <Title>MORPHOLOGICAL PRODUCTIVITY IN THE LEXICON</Title>
  <Section position="6" start_page="107" end_page="112" type="concl">
    <SectionTitle>
4 Conclusion
</SectionTitle>
    <Paragraph position="0"> For a language with rich morphology, lexical rules can be used for controlled generation of surface forms.</Paragraph>
    <Paragraph position="1"> Inflections and derivations can be seen as word-based (local) operations on the root, and thus be modelled as lexical rules. Phonological alternations in stems can be embedded in the rules as well. Grammatical role changes, type constraints on word subtypes, and noun to NP promotions (as in non-referential objects) control the proliferation of lexical entries. Semantic contribution of inflections seems to be morpheme specific: All derivations take part in semantic composition, but some inflections (such as case and causatives) contribute semantically as well. Most inflections (e.g., person and number markers), however, have grammatical functions only. This is not to say they do not have a semantic form, just that in many cases the form is that of identity. Productive use of derivations is limited by the predictability of the semantic relation of the stem to the affix.</Paragraph>
    <Paragraph position="2"> We have been testing our lexicon design as part of an HPSG grammar for Turkish \[15\]. The grammar development environment, ALE, had to be modified to allow run-time evaluation of lexical rules. Compiling 4cf. \[11, 12\] for a description of these processes. \[6\] is the original work on Turkish that combines finite state morphotactics with morphophonemic alternations.</Paragraph>
    <Paragraph position="3">  out the lexical rules seems to be impractical, since generating every possible form for a large lexicon of roots causes exponential growth in the lexicon. Compilation of all surface forms for a lexicon of only 40 root forms produces around 2800 entries, and takes about 8 minutes on a Sun Sparcstation 10. Run-time execution of rules puts the burden on parsing or generation. We believe that as the lexicons of NLP systems become more comprehensive and open-ended, the trade-off will be resolved in favour of using the lexical rules on demand at the expense of slower performance.</Paragraph>
    <Paragraph position="4"> Acknowledgements: This research is supported in part by grants from NATO Science Division SfS III (contract TU-LANGUAGE), and Scientific and Technological Research Council of Turkey (contract EEEAG-90).</Paragraph>
  </Section>
class="xml-element"></Paper>