<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Semantic Ambiguity of Spatial Relational Nouns in Japanese</article-title>
      </title-group>
      <contrib-group>
        <aff id="aff0">
          <label>0</label>
          <institution>Department of Management Tokyo University of Science 500 Shimo-kiyoku</institution>
          ,
          <addr-line>Kuki-city, Saitama 346-8512</addr-line>
          ,
          <country country="JP">Japan</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>This paper discusses spatial terms in Japanese. Common nouns such as ue “on/over/above” and naka “inside” are used in Japanese to represent spatial and temporal locations, as front in in front of, or center in in the center of in English. I consider Japanese common nouns that represent spatial locations to be relational nouns that are two-place predicates, one of whose argument slots is filled by the entity represented by the other NP in the NP1-no NP2 construction. Since the corpus data [1] suggest that spatial nouns are often semantically ambiguous among physical, metaphorical, and temporal locations, the unified semantic entry in the Generative Lexicon (GL) [2] proves to be useful for handling the semantic ambiguity.</p>
      </abstract>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>-</title>
      <p>Languages such as Chickasaw in North America use relational nouns to express
locations [3], rather than prepositions such as in, on, under, or between as in English.
In (1a), pakna’ “top” is a relational noun, that follows its possessor chokka’ “house.”
Japanese is another language that expresses locations using relational nouns such as
naka “inside,” ue “on/above,” and shita “under” as in (1b).
(1) a. chokka’ pakna’
house top
“the top of the house (the house’s roof)”
b. mune-no mae-de tenohira-o awase (4179)
chest-GEN front-LOC palms-ACC hold
“Put your palms together in front of your chest”</p>
      <p>
        Mae “front” is a relational noun that does not stand alone semantically; therefore,
it always means “the front of something,” for example, musuko which means “son”
always stands for “someone’s son” (e.g., “Bill’s son”). Naka “inside,” ue “on/above,”
and shita “under” are two-place holders, and nouns such as aida “between” that require
another argument are three-place predicates.
1 The numbers in parentheses indicate the sentence IDs of the output of the data in the Yahoo!
Chiebukuro section of [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ] using ChaKi.NET 1.2b .
[3, 4]
1
(2) a. [[ue“on=top”]] = l xl y[on(y)(x)]
b. [V P [[NP kohi-no ue]-ni] [miruku-o] [V ireru]]
coffee-GEN on-DAT milk-ACC put
“put milk on (the surface of) coffee”
c. [[kohi − no ue“on co f f ee”]] = l x[on(e y.coffee(y))(x)]
2
      </p>
      <p>Ambiguity among Physical, Metaphorical and Temporal
Locations</p>
      <p>Modeling Lexical Ambiguity of Spatial Language</p>
      <p>Formal Semantics
This section formalizes the spatial terms in Japanese. Most of them are two-place
holders except aida “between” which is a three-place predicate.
(4) a. [[mae]] = l x,y[in-front-of(x)(y)]</p>
      <p>b. [[mae]] = l t,t’[before(t)(t’)]
(5) a. [[mune − no mae“in f ront o f the chest”]] = l y.in-front-of(e x.chest(x))(y)
b. [[shuppatsu − no mae“be f ore departure”]]</p>
      <p>= l e’.∃e[before(time(e))(time(e’))&amp;departure(e)]
(6) a. [[ho]] = l x,y[toward(x)(y)]</p>
      <p>b. [[ho]] = l x,y[to(x)(y)]
(7) a. [[(physical)naka]] = l x,y.in(x)(y)</p>
      <p>b. [[(metaphorical)naka]] = l x,y.among(x)(y)
(8) a. [[nabe − no naka“inside the pot”]] = l y.in(e x.pot(x))(y)</p>
      <p>b. [[reshipi − no naka“amongrecipes′′]] = l y.among(e x.recipe(x))(y)
(9) a. [[aida]] = l x,y,z[between(x)(y)(z)]
b. [[aida]] = l x,y,z[among(x)(y)(z)]
c. [[aida]] = l t,t’[t’= during(t)]
(10) a. Ha-to ha-no aida atari-ga chairoku naru-no-desu-ka. (2906)
tooth-and tooth-GEN between vicinity-NOM brown become-GEN-HON-Q
“Have the gaps between your teeth turned brown?”
b. Geinojin-no aida-de hayat-teiru daietto-shokuhin (427)
entertainer-GEN among-LOC popular-PROG diet-food
“The diet food popular among TV entertainers”
c. Koko sukagetsu-no aida (3201)
this a few months-GEN period
“during these few months”
3.2</p>
      <p>Lexical Ambiguity in the Generative Lexicon
Contrary to the previous section which listed two-way or three-way ambiguous lexical
entries, the GL [2] has the means to provide unified lexical entries for a single
spatial term, due to its elaborate lexical semantic information. In particular, the Lexical
Conceptual Paradigm (LCP) [4, 2] is a powerful tool for resolving semantic ambiguity.</p>
      <p>The formal quale in GL contains ontological information. In (11), coffee is a drink
according to its formal quale, and its higher ontological category is a physical entity,
which implies that ue “on” is interpreted physically. The unification profess is described
in the following manner:
(11)
COFFEE
 ARG1 = x DRINK 
ARGSTR = D-ARG1 = y HUMAN
 D-E1 = e1 PROCESS 
 FORMAL = LIQUID( x ) 
QUALIA = TELIC = DRINK ACT( e1 , y , x )
UE “ON” KOHI-NO UE “ON COFFEE”</p>
      <p>ARG1 = x PHYSICAL OBJECT
AQURGASLTIAR== [FAORRGM2A=Ly=POHNY(SIeC1AxL, OxB,JEyCT)]AQURGASLTIAR==</p>
      <p>
D-E1 = e1 STATE 
</p>
      <p>
ARG1 = x PHYSICAL OBJECT
ARG2 = y COFFEE 
D-E1 = e1 STATE 
[FORMAL = ON( e1 , x , y )] </p>
      <p>Mae “in front/before” is lexically ambiguous between physical and temporal
locations. Lexical ambiguity calls for a meta-entry, that is, the LCP, which is a Cartesian
product of the different concepts represented by a single lexical item [2, 5] as in (12).
For example, book is a Cartesian product of a physical entity and the information
contained within it, thus, (13a,b) are both grammatically correct.
(12) mae.lcp = {location.time, location, time }
(13) a. The book is on the table.</p>
      <p>b. That book was right. An earthquake did happen as it had predicted.
(14)
MAE
 “AQIURNGAFSLRTIOARNT==OF CHEST”</p>
      <p>
        Argument structure also needs to have metaentries since mae “front/before” and
aida “between/among/during” combine with different types of semantic arguments.
4 Conclusion
In this paper, spatial language in the form of “NP1-GEN NP2” constructions in Japanese
was taken from [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ] and classified into literal, temporal, and metaphorical meanings.
Spatial terms are semantically ambiguous relational nouns. Lexical meta-entries in the
GL effectively handle the semantic ambiguity of the most common spatial nouns.
      </p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          <article-title>[1] BCCWJ: Balanced Corpus of Contemporary Written Japanese, BCCWJ2009 edition</article-title>
          .
          <source>The National Institute of Japanese Language</source>
          (
          <year>2009</year>
          ) [2]
          <string-name>
            <surname>Pustejovsky</surname>
            ,
            <given-names>J.:</given-names>
          </string-name>
          <article-title>The Generative Lexicon</article-title>
          . MIT Press, Cambridge (
          <year>1995</year>
          ) [3]
          <string-name>
            <surname>Lillehaugen</surname>
            ,
            <given-names>B.D.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Munro</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          :
          <article-title>Prepositions and relational nouns in a typology of component part locatives (</article-title>
          <year>2006</year>
          ) [4]
          <string-name>
            <surname>Pustejovsky</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Anick</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          :
          <article-title>On the semantic interpretation of nominals</article-title>
          .
          <source>In: Proceedings of COLING-1988</source>
          , Budapest (
          <year>1988</year>
          ) [5]
          <string-name>
            <surname>Carpenter</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          :
          <article-title>The Logic of Typed Feature Structures</article-title>
          . Cambridge University Press, Cambridge (
          <year>1992</year>
          )
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>