<?xml version="1.0" standalone="yes"?>
<Paper uid="M91-1004">
  <Title>MUC-3 LINGUISTIC PHENOMENA TEST EXPERIMEN T</Title>
  <Section position="4" start_page="31" end_page="32" type="metho">
    <SectionTitle>
JOSE PARADA GRANDY, THE BOLIVIAN POLICE CHIEF, TOLD EFE THAT A N
UNIDENTIFIED PERSON STEPPED OUT OF A VEHICLE AND PLACED A
PACKAGE IN ONE OF THE PLANT POTS ON JUAN DE LA RIVA STREET, A FE W
METERS FROM THE U.S. EMBASSY IN DOWNTOWN LA PAZ .
</SectionTitle>
    <Paragraph position="0"> &amp;quot;Jose Parada Grandy&amp;quot; and &amp;quot;the Bolivian Police Chief&amp;quot; are in full appositio n because they each can be omitted resulting in the following acceptable sentences , they each are the subject in those sentences, and all three sentences have the same extralinguistic reference .</Paragraph>
  </Section>
  <Section position="5" start_page="32" end_page="32" type="metho">
    <SectionTitle>
JOSE PARADA GRANDY TOLD EFE THAT AN UNIDENTIFIED PERSON STEPPE D
OUT OF A VEHICLE AND PLACED A PACKAGE IN ONE OF THE PLANT POTS ON
JUAN DE LA RIVA STREET, A FEW METERS FROM THE U .S. EMBASSY IN
DOWNTOWN LA PAZ .
THE BOLIVIAN POLICE CHIEF TOLD EFE THAT AN UNIDENTIFIED PERSO N
STEPPED OUT OF A VEHICLE AND PLACED A PACKAGE IN ONE OF THE PLAN T
POTS ON JUAN DE LA RIVA STREET, A FEW METERS FROM THE U .S .
EMBASSY IN DOWNTOWN LA PAZ .
</SectionTitle>
    <Paragraph position="0"> Partial apposition occurs when the three conditions are not all met . An example of partial apposition not meeting condition (a) appears in test message TST2 MUC3-0100 .</Paragraph>
  </Section>
  <Section position="6" start_page="32" end_page="34" type="metho">
    <SectionTitle>
THE BRAZILIAN EMBASSY IN COLOMBIA HAS CONFIRMED THE RELEASE O F
REDE GLOBO JOURNALIST CARLOS MARCELO WHO WAS KIDNAPPED B Y
COLOMBIAN ARMY OF NATIONAL LIBERATION GUERRILLAS .
</SectionTitle>
    <Paragraph position="0"> The difference between full and partial apposition in this case is trivial requirin g only the addition of a determiner to &amp;quot;Rede Globo journalist&amp;quot; to make the sentence omitting &amp;quot;Carlos Marcelo&amp;quot; acceptable .</Paragraph>
    <Paragraph position="1"> Partial appositives that were omitted from the phenomenon testing were cases of appositives containing &amp;quot;also&amp;quot; and &amp;quot;alias .&amp;quot; These were omitted because of their adverbial nature.</Paragraph>
    <Paragraph position="2"> Another gray area in choosing examples concerns titles . Quirk et al makes th e distinction between apposition and institutionalized titles . The authors show the range from apposition in &amp;quot;critic Paul Jones&amp;quot; to full title in &amp;quot;Mr . Porter&amp;quot; with th e following examples : a. critic Paul Jones the critic Paul Jones (with appositives, a preposed determiner i s normal but not with titles ) (with appositives, postpositio n more normal than preposition whereas the opposite i s allow postposition ) (appositives and most without the prope r determiners ) (most titles and some appositives can be use d  (with titles, a preposed determiner is not normal ) *Porter the mister (postposition with &amp;quot;the&amp;quot; is not allowed here ) *the mister (some titles cannot be used without the proper nouns and with determiners) *mister (vocative) (most titles can be used as vocatives ) (substandard ) In the MUC-3 messages, the appositives and titles are distinguished by the test s above with the cut-off between (3) and (4) . For example, &amp;quot;Colonel,&amp;quot; &amp;quot;Senator,&amp;quot; and &amp;quot;Ambassador&amp;quot; are titles because the following judgments are similar to those fo r &amp;quot;Professor&amp;quot; above: Colonel Heriberto Hernandez *the Colonel Heriberto Hernandez (with titles, a preposed determiner i s not normal unless the noun phrase s are modified restrictively ) ?Heriberto Hernandez the Colonel (with titles that allow postposition , preposition without &amp;quot;the&amp;quot; is mor e normal than postposition with &amp;quot;the&amp;quot; ) the Colonel (appositives and most titles can b e used without the proper nouns an d with determiners ) Colonel (vocative) (most titles can be used as vocatives ) However, &amp;quot;student&amp;quot; and &amp;quot;peasant&amp;quot; are considered appositives because of th e following pattern similar to the pattern for &amp;quot;critic&amp;quot; above :  An attempt was made to limit the appositives used in the testing to those mos t likely to be agreed upon as appositives while still maintaining a reasonable number of examples.</Paragraph>
    <Paragraph position="3"> Construction of the Test Set s The message sentences containing appositives were extracted from th e messages for analysis . The examples were put in a file for distribution to the participants to assist in analysis of their results . This file contained information concerning the categorization of the appositives and the slots affected by the appositioned noun phrases and the entire sentence.</Paragraph>
    <Paragraph position="4"> The appositives were categorized as postposed versus preposed and simple versus complex . An example of a postposed appositive is &amp;quot;Jose Parada Grandy, th e Bolivian Police Chief&amp;quot; and an example of a preposed appositive is &amp;quot;Rede Glob o journalist Carlos Marcelo .&amp;quot; The subdivision of the appositives according to thei r complexity was done subjectively based on internal structure and the context . Both &amp;quot;Jose Parada Grandy, the Bolivian Police Chief&amp;quot; and &amp;quot;Rede Globo journalist Carlo s Marcelo&amp;quot; were considered simple . Any complexity in an example, such as conjunction within the appositive, a missing comma, or a comma inside double quotes, pu t that example in the complex category. Probably the most complex appositioned nou n phrase in the corpus was in apposition to &amp;quot;peasants&amp;quot; in TST2-MUC3-0036 . The misspelling &amp;quot;Colonal&amp;quot; is part of the message .</Paragraph>
  </Section>
  <Section position="7" start_page="34" end_page="34" type="metho">
    <SectionTitle>
THE PEASANT COMMUNAL ASSOCIATION, ACC, CONTINUES TO DEMAND TH E
RELEASE OF PEASANTS BARTOLO RODRIGUEZ, WHO WAS CAPTURED ON 2 7
JANUARY, AND [NAME INDISTINCT] CAPTURED ON 2 FEBRUARY B Y
TROOPS OF COLONAL ORLANDO MONTANO OF THE 6TH INFANTRY BRIGADE .
</SectionTitle>
    <Paragraph position="0"> The most important and difficult activity in constructing phenomena tests is t o determine the individual slots that could only be filled from the phrase containin g the phenomenon being tested. The slots that could only be filled by the informatio n in the appositioned noun phrases as well as in the sentences containing those appositioned noun phrases were noted . The configuration option files for the scoring system were constructed to score just those slots directly affected by the presence o f an appositive . Slots that could have been filled from any other phrase/sentence no t containing an appositive were excluded from the scoring . This step in the test construction is the most likely point where human error can intrude .</Paragraph>
    <Paragraph position="1"> For the purposes of running the &amp;quot;minimal pair&amp;quot; test, a modified version of th e message file was produced . The messages were altered to contain simple sentence s expressing the equivalence of the appositioned noun phrases in cases where th e appositioned noun phrases directly affected at least one slot in the template fill . The appositive no longer appeared in the original sentence . For example ,</Paragraph>
  </Section>
  <Section position="8" start_page="34" end_page="35" type="metho">
    <SectionTitle>
THE BRAZILIAN EMBASSY IN COLOMBIA HAS CONFIRMED THE RELEASE O F
REDE GLOBO JOURNALIST CARLOS MARCELO WHO WAS KIDNAPPED B Y
COLOMBIAN ARMY OF NATIONAL LIBERATION GUERRILLAS .
</SectionTitle>
    <Paragraph position="0"> The &amp;quot;minimal pair&amp;quot; test was voluntary because it required a separate run o f the data extraction systems on the modified messages .</Paragraph>
    <Paragraph position="1"> The scoring of the appositive tests was diluted somewhat by the allowance i n the scoring guidelines for partial credit to be given when the key contains a complete proper name and the response contains only the identifying part of the name . It was typical of the appositioned noun phrases that they were the place where th e full name of the person was introduced with only part of the name being used fro m then on for reference . A previously undetected bug in the scoring system cause d one template not affected by apposition to be scored instead of another template tha t was affected by apposition . For phrases, only 2 slots out of a possible 66 slots (3%) were potentially affected ; for sentences, 9 slots out of a possible 198 slots (4 .5%) were potentially affected.</Paragraph>
  </Section>
  <Section position="9" start_page="35" end_page="35" type="metho">
    <SectionTitle>
HYPOTHESES
</SectionTitle>
    <Paragraph position="0"> The intent of the testing was to discover whether the scoring isolated the phenomenon of apposition . Each of the following hypotheses was proposed and tested i n order to uncover evidence of isolation of the phenomenon .</Paragraph>
    <Paragraph position="1"> Hypothesis 1 . The systems should score differentally on the appositive s (both phrasally and sententially) than they did on the overall testing .</Paragraph>
    <Paragraph position="2"> Hypothesis 2 . The systems should score higher on the simple r appositives .</Paragraph>
    <Paragraph position="3"> Hypothesis 3 . The systems should score differently on the postpose d and preposed appositives . It was not possible to hypothesize which score would be higher. Although postposed appositives are more prototypical and have indications they are appositives such as comma s or dashes, preposed appositives lend themselves to treatment a s</Paragraph>
  </Section>
class="xml-element"></Paper>