<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1025">
  <Title>A grammar formalism and parser for linearization-based HPSG</Title>
  <Section position="7" start_page="5" end_page="7" type="evalu">
    <SectionTitle>
6 Evaluation
</SectionTitle>
    <Paragraph position="0"> As discussed at the end of section 4.2, it is possible to take a GIDLP grammar and write out the discontinuity. All non-domain introducing rules must be folded into the domain-introducing rules, and then each permitted permutation of a RHS must become a context-free rule on its own - generally, at the cost of a factorial increase in the number of rules.</Paragraph>
    <Paragraph position="1"> This construction indicates the basis for a preliminary assessment of the GIDLP formalism and its parser. The grammar in (13) recognizes a very small fragment of German, focusing on the free word order of arguments and adjuncts in the so-called Mittelfeld that occurs to the right of either the finite verb in yes-no questions or the complementizer in complementized sentences.</Paragraph>
    <Paragraph position="2">  This optimization only applies to epsilon-free grammars. Further work in this regard can involve determining the minumum and maximum yields of each category; some optimizations involving this information can be found in (Haji-Abdolhosseini and Penn, 2003).</Paragraph>
    <Paragraph position="4"> The basic idea of this grammar is that domain compaction only occurs at the top of the head path, after all complements and adjuncts have been found.</Paragraph>
    <Paragraph position="5"> When the grammar is converted into a CFG, the effect of the larger domain can only be mimicked by eliminating the clause and vp constituents altogether. null As a result, while this GIDLP grammar has 10 syntactic rules, the corresponding flattened CFG (allowing for a maximum of two adverbs) has 201 rules. In an experiment, the four sample sentences in (14)  were parsed with both our prototype GIDLP parser (using the GIDLP grammar) as well as a vanilla Earley CFG parser (using the CFG); the results are shown in (15).</Paragraph>
    <Paragraph position="6">  (14) a) Gab der Mann der Frau das Buch? b) dass das Buch der Mann der Frau gab.</Paragraph>
    <Paragraph position="7"> c) dass das Buch gestern der Mann dort der Frau gab.</Paragraph>
    <Paragraph position="8"> d) Denkt der Mann dass das Buch gestern der Mann dort der Frau gab? (15)  Averaging over the four sentences, the GIDLP grammar requires 89% fewer active edges. It also generates additional passive edges corresponding to the extra non-terminals vp and clause. It is important to keep in mind that the GIDLP grammar is more general than the CFG: in order to obtain a finite number of CFG rules, we had to limit the number of adverbs. When using a grammar capable of  The grammar and example sentences are intended as a formal illustration, not a linguistic theory; because of this and space limitations, we have not provided glosses. handling longer sentences with more adverbs, the number of CFG rules (and active edges, as a consequence) increases factorially.</Paragraph>
    <Paragraph position="9"> Timings have not been included in (15); it is generally the case that the GIDLP parser/grammar combination was slower than the CFG/Earley parser. This is an artifact of the use of atomic categories, however. For the large feature structures used as categories in HPSG, we expect the larger numbers of edges encountered while parsing with the CFG to have a greater impact on parsing time, to the point where the GIDLP grammar/parser is faster.</Paragraph>
  </Section>
class="xml-element"></Paper>