<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1407">
  <Title>The Organization of the Lexicon The Polysemy of Grow and Disambiguation</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
3 For computational treatment
</SectionTitle>
    <Paragraph position="0"> Finally, I would like to make comments on the computational treatment of the polysemy of grow.</Paragraph>
    <Paragraph position="1"> The following table gives an idea of which readings are similar syntactically and semantically and of what is required to disambiguate the polysemy.</Paragraph>
    <Paragraph position="2"> For the computational treatment of disambiguation, we would not need to exhaust all the features involved in argument structure for each meaning of a verb, but to have sufficient rules of elimination of other readings would suffice.</Paragraph>
    <Paragraph position="3"> Following is a table showing relationships among the meanings of grow, which is organized in terms of grammatical and thematic roles.7  has a distintive construction by having the OBJECT, and is easy to distinguish from the others. In addition, constructions such as His love grew cold and I am growing to hate him are not included in this table because these are typical of the rea ding of 'grow_change', and are easy to detect syntactically. 8 &amp;quot;=F&amp;quot; means that in this reading the LOCUS element has a more specific role, FOCUS.</Paragraph>
    <Paragraph position="4"> The construction in the form of SUBJECT + GROW appears in both readings of 'grow_proto' and 'grow_increase'. Both readings are distinguishable from each other by the semantics of the SUBJECT noun phrases. The SUBJECT noun phrase in the reading of 'grow_proto' requires an entity referring to a living thing, while that in the reading of 'grow_increase', an abstract entity.</Paragraph>
    <Paragraph position="5"> Three readings, 'grow_proto', 'grow_appear' and 'grow_increase' take the form of SUBJECT noun phrase + GROW + P9 + LOCUS noun phrase; in particular the last reading in the form of SUBJECT + GROW + in + FOCUS noun instead. As the LOCUS in the reading of 'grow_proto' denotes the PLACE feature and the others do not, the reading of 'grow_proto' can be distinguished from the other two. Between the remaining two, the reading of 'grow_appear' takes the LOCUS requiring a physical entity whereas the reading of 'grow_increase', an abstract entity.</Paragraph>
    <Paragraph position="6"> The readings of 'grow_incease' and 'grow_change' may appear in the form of SUBJECT + GROW + to + GOAL or in the form of SUBJECT + GROW + from + SOURCE + to + GOAL. They are distinguishable because the GOAL and SOURCE in the reading of 'grow_increase' include features of QUANTITY while those in the reading of 'grow_change' do not. The readings of 'grow_proto' and 'grow_appear' take the form of SUBJECT + GROW + from + SOURCE, but while the SOURCE in the reading of 'grow_appear' refers to an abstract entity, that in the reading of 'grow_proto' does not. As mentioned above, the two readings are also distinguishable in terms of the semantic features of the SUBJECT noun phrases because the SUBJECT of the reading of 'grow_proto' requires a feature indicating a living thing while that of the reading of 'grow_appear' should not.</Paragraph>
    <Paragraph position="7"> As discussed above, once rules of elimination are prepared, it would not be difficult to disambiguate the polysemy of a word, because the polysemy is based on the extension of meanings that do not overlap and the contexts that make the extensions 9 &amp;quot;P&amp;quot; stands for preposition.</Paragraph>
    <Paragraph position="8"> viable should not overlap either. The contexts in the case of verbs are the argument structures.</Paragraph>
  </Section>
  <Section position="5" start_page="0" end_page="0" type="metho">
    <SectionTitle>
4 Implications
</SectionTitle>
    <Paragraph position="0"> Interestingly, the hypothesis of the relationship between the primary sense and other extended senses in this study differs from that in Tyler and Evans' (2001) study on the polysemy of the preposition over. This study claims that the basic or primary sense of grow has a combination of individual connotations such as change of state, increase and appearance, and that grow acquires a new sense extended from each individual connotation by allowing a different argument structure. On the other hand, the primary sense of over proposed in Tyler and Evans 'represents a highly idealized abstraction from our rich recurring experience of spatial scenes' and 'other distinct senses instantiated in the polysemy network for over result from pragmatic strengthening, i.e. reanalysis and encoding.' (p. 762) It would be an open question whether this difference is due to different subject matters, i.e. between the verb and the preposition, or due to different approaches, i.e. between computationally motivated and psycholinguistic approaches.</Paragraph>
    <Paragraph position="1"> However, both studies have a common underlying thesis that the lexicon is highly motivated exhibiting regularities. This is in contrast to the view generally held by Minimalist linguists that the lexicon is not organized by general principles (Chomsky  The advantage of the analysis of the polysemy of grow proposed in this study is that it is somewhat similar to analyses we find in the entries for grow in dictionaries. This suggests that it might be possible to use analyses of lexical entries listed in tr aditional dictionaries for computational purposes.</Paragraph>
  </Section>
class="xml-element"></Paper>