<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-1012">
  <Title>Generating Determiners and Quantifiers in Hebrew</Title>
  <Section position="4" start_page="91" end_page="92" type="intro">
    <SectionTitle>
2 Previous Work
</SectionTitle>
    <Paragraph position="0"> Descriptive studies of the syntax of the Hebrew noun phrase do not generally distinguish between different functions of determiners but focus rather on the syntactic structure of the NP.</Paragraph>
    <Paragraph position="1"> Glinert \[7\] adopts a functional perspective, more appropriate to the needs of a generation system, and identifies a general pattern of the NP that we use as a basis: \[partitive determiner amount head classifiers describers post-det/quant qualifiers\] In this pattern, the slots can be filled by any word that fits the function: partitive expressions identify a subset of the set referred to by the rest of the NP; amount expressions identify the quantity or number, exact or inexact, of the thing referred to; determiners identify the thing referred to more or less precisely or specifically. The general picture of the input specification is that of a reference set from which a subset is identified by different complementary means. The function of each word realizing this input specification determines where it appears in the syntactic structure.</Paragraph>
    <Paragraph position="2"> Yizhar and Doron \[3\] \[12\] distinguish between two sets of determiners, that they call D and Q quantifiers. The distinction is based on syntactic features, such as location, ability to be modified, ability to participate in partitive structures, requirement to agree in number and gender with the head. This distinction is used to explain cooccurrence restrictions, the order of appearance of D vs Q quantifiers and the recursive structure of D determiners: D determiners can be layered on top of other D determiners. A single Q quantifier can occur in an NP and it remains attached closest to the head.</Paragraph>
    <Paragraph position="3"> As discussed in the introduction, the distinction D/Q is not sufficient to explain some restrictions on the cooccurrence of several Ds in the same NP, and does not help to predict the restrictions on occurrence of determiners in different definiteness and partitive contexts.</Paragraph>
    <Paragraph position="4"> In \[i\] we have, therefore, refined the D/Q classification using functional criteria: we map the Q quantifiers to the &amp;quot;amount&amp;quot; category defined by Glinert, and the D set is split into the partitive and determiner categories - each with a different function. Of these, only partitive are recursive.</Paragraph>
    <Paragraph position="5"> Another controversial issue within Hebrew studies is whether determiners are types of nouns \[2\] \[6\]. This is an important issue to explain in a uniform manner why constructs like smixut (noun compounding) are used for certain determiners and why certain determiners determine the number and gender of the whole NP. We have implemented some of these constraints by defining lexical features for determiners which determine this behavior.</Paragraph>
    <Section position="1" start_page="91" end_page="92" type="sub_section">
      <SectionTitle>
2.1 Determiner Sequence in Generation
</SectionTitle>
      <Paragraph position="0"> Our goal is to define a set of functional features to include in the input to a generator. These features must provide enough information to construct a wide variety of determiner constructions both in Hebrew and in English. For the comparison with English, we based our work on the SURGE generation grammar \[5\].</Paragraph>
      <Paragraph position="1"> The determiner sequence in SURGE is based on \[8\] and has the following pattern: \[pre-det (of) deZ deictic2 ordinal cardinal quantifier (describer classifier head)\].</Paragraph>
      <Paragraph position="2"> Where pre-det can be any one of all, both, half, multipliers and fractions, det is a deictic determiner, deicticg is an adjective from a restricted class that expresses the anaphoric status of the thing referred to (e.g., above, same, different); and quantifier expresses the amount or quantity of the thing referred to.</Paragraph>
      <Paragraph position="3"> In \[4\], we gathered a minimal set of 24 features to define the determiner sequence of the NP in English. Complex systems in SURGE implement the cooccurence restrictions among these features, defniteness and partitive constructs.</Paragraph>
      <Paragraph position="4">  When studying the Hebrew determiner system, we found that this set of features had to be extended and refined. As a consequence, both the implementations of SURGE and of HUGG have been made compatible with the same set of features, which we motivate in the rest of the paper.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>