<?xml version="1.0" standalone="yes"?>
<Paper uid="C73-2016">
  <Title>LA GATO BEBE LA LECHE DET GATO BEBE LA LECHE</Title>
  <Section position="1" start_page="0" end_page="11" type="metho">
    <SectionTitle>
FRANCISCO MAP.COS
THE AUTOMATIC SYNTACTIC ANALYSIS
AS AN AID IN DICTIONARY MAKING
</SectionTitle>
    <Paragraph position="0"> In this paper we are going to explain some of the problems we have found in our attempts to mechanize the Historical Dictionary (D.H.) of the Spanish language. Our experimental project is a collaboration of the Royal Spanish Academy (R.A.E.) and the Computer Center of the University of Madrid (CCVM).</Paragraph>
    <Paragraph position="1"> There are, indeed, many procedures for making concordances of a text, and those procedures are, in general, very successful ones. We are not going to be concerned with concordance making; nevertheless, it seems obvious to declare that a good concordance system must be in the basis of our researches. We must suppose that in our concordances we get syntactically limited utterances, i.e. that in our concordances we do not have to deal with words belonging to sentences whose verbs are not included in the text given in the concordance. By now, the best method of getting this kind of concordances is to limit them by full stops. So, we get our text divided by full stops or semicolons and we analyze as many sentences as verbs could be included between those punctuation marks. But, and we must say it quickly, we are not dealing with complex sentences yet, on the contrary, we are analyzing rather simple structures of</Paragraph>
    <Section position="1" start_page="0" end_page="11" type="sub_section">
      <SectionTitle>
Subject Verb Direct Object
</SectionTitle>
      <Paragraph position="0"> which means really that we are occupying ourselves with problems of determiners, inflectional endings, agreement or concord, etc., instead of dealing with word order problems. We are making some steps in semantic analysis too.</Paragraph>
      <Paragraph position="1"> Nobody shall expect, then, from our so clearly limitated work, any marvellous discovery. We have just proved that our IBM 7090, assisted by an msi 1401 is able to analyze so simple a phrase as each of the components of the set of experiments you will see.</Paragraph>
      <Paragraph position="2">  162 I. DEL CAMPO- I. GONZALEZ- M a. T. MOLINA- F. MARCOS We have written our programs in SNOBOL, a language specially indicated for the management of linguistic structures. One of the many problems of our slqoBol~ compiler is its inadequacy to give the time used by the computer in performing its task, but we have calculated that for all the lecture, analysis and listing we show, it must be about three minutes.</Paragraph>
      <Paragraph position="3"> Our error message is HA FALLADO, which indicates that something has gone wrong since the beginning. Other error messages are</Paragraph>
    </Section>
  </Section>
  <Section position="2" start_page="11" end_page="11" type="metho">
    <SectionTitle>
LA PALABRA &amp;quot;PALABRA&amp;quot; NO ESTA EN NUESTRA TABLA,
</SectionTitle>
    <Paragraph position="0"> which means that we are using a word that does not exist in our lexicon (Table 1), ESTA FRASE NO FORMA ORACION, i.e. we</Paragraph>
    <Paragraph position="2"/>
  </Section>
  <Section position="3" start_page="11" end_page="11" type="metho">
    <SectionTitle>
RIEGA=VT4- AUX,SUJINTELIGENTE,SUJANIMADO,COMNATU RAL,
</SectionTitle>
    <Paragraph position="0"> have got a set of words without grammatical organization. As grammatical organization we understand our Table 2, &amp;quot;grammar &amp;quot;. EL</Paragraph>
  </Section>
  <Section position="4" start_page="11" end_page="11" type="metho">
    <SectionTitle>
TABLA2 \[GRAMATICA)
DET SUST=SN
</SectionTitle>
    <Paragraph position="0"/>
    <Paragraph position="2"> agree with the article in the subject (or in the object) &amp;quot;; LA PALABRA &amp;quot;PALABRA &amp;quot; NO ES ARTICULO, SUSTANTIVO NI VERBO, the word we are dealing with is not in our grammar, which only includes</Paragraph>
  </Section>
  <Section position="5" start_page="11" end_page="11" type="metho">
    <SectionTitle>
AUTOMATIC SYNTACTIC ANALYSIS IN DICTIONARY MAKING 163
</SectionTitle>
    <Paragraph position="0"> articles, substantives, and verbs; NO HAY COHERENCIA ENTRE SUJETO Y VERBO &amp;quot;there is no semantical agreement between subject and verb &amp;quot;, i.e., the features of the subject are not those demanded by the verb; LA PALABRA &amp;quot;PALABRA&amp;quot; ESTA MAL COLOCADA, meaning that we have a word out of place (referring always to our grammar); EL SUSTANTIVO NO CONCUERDA</Paragraph>
  </Section>
  <Section position="6" start_page="11" end_page="11" type="metho">
    <SectionTitle>
CON EL AR TICULO &amp;quot;there is no equivalent among the grammatical
</SectionTitle>
    <Paragraph position="0"> features of the article and those of the substantive &amp;quot;, and NO HAY</Paragraph>
  </Section>
  <Section position="7" start_page="11" end_page="11" type="metho">
    <SectionTitle>
COHERENCIA ENTRE COMPLEMENTO Y VERBO &amp;quot; the se-
</SectionTitle>
    <Paragraph position="0"> mantic features of the object are not those demanded by the verb &amp;quot;.</Paragraph>
    <Paragraph position="1"> Our success message is LA FRASE ESTA BIEN CONSTRUIDA (&amp;quot; the sentence is a grammatical one &amp;quot;); if there has been a semantic disagreement in the analysis, but not a grammatical one we get A</Paragraph>
  </Section>
  <Section position="8" start_page="11" end_page="31" type="metho">
    <SectionTitle>
PESAR DE ELLO LA FRASE ESTA BIEN CONS TRUIDA (in
</SectionTitle>
    <Paragraph position="0"> spite of our signalled semantic disagreement the sentence is well built).</Paragraph>
    <Paragraph position="1"> With this innovation we are trying to research on the domain of apparent incoherences like metaphors. For instance, we establish that the verb HABLAK &amp;quot;to speak &amp;quot; requires the feature + HUMAN in the subject, so if we get EL PERRO HABLO &amp;quot;the dog spoke&amp;quot; our message got will be &amp;quot;there is no semantic agreement between subject and verb &amp;quot;: &amp;quot;dog&amp;quot; is --HUMAN; but, in spite of that, we do not stop our analysis and at the end we obtain &amp;quot;in spite of that the sentence is well built &amp;quot;, which assures us that the sentence is grammatical to a lesser degree than another one with total agreement.</Paragraph>
    <Paragraph position="2"> Our grammar is like this: S &gt; SN SP (we admit a difference between SP and ST/', but it is irrelevant at this step of our job, so we make SP ~ SV)  The computer makes substitutions beginning at the left side; if there is no agreement it emits the corresponding error message, and if this disagreement is of a grammatical kind it stops.</Paragraph>
    <Paragraph position="3">  A part of our lexicon is included in the listing, in which it may be seen that the first feature belongs to the grammatical analysis, and the rest to the semantic one. So PEDP~O is a substantive, masculine, singular, on the grammatical side, and animated, intelligent, on the semantic side. The verb COME &amp;quot;eats&amp;quot; is a transitive verb, which needs an animated subject and a solid object. It seems fair to declare that till now we were much more concerned about syntactic problems than about those of morphology. So, we operated with verbs in the third person singular. Now we are trying to build a morphology which will permit us to apply our analysis to a broader field.</Paragraph>
    <Paragraph position="4"> We reproduce here a listing of one of our experiments.</Paragraph>
    <Paragraph position="6"/>
  </Section>
  <Section position="9" start_page="31" end_page="31" type="metho">
    <SectionTitle>
AUTOMATIC SYNTACTIC ANALYSIS IN DICTIONARY MAKING 167
</SectionTitle>
    <Paragraph position="0"/>
    <Paragraph position="2"/>
    <Paragraph position="4"/>
  </Section>
  <Section position="10" start_page="31" end_page="31" type="metho">
    <SectionTitle>
ANALISIS GRAMATICAL DE LA FRASE
PEDRO RIEGA FLORES
SUST RIEGA FLORES
SN RIEGA FLORES
SN VT+AUX FLORES
SN GV FLORES
8N GV SUST
SN GV SN
SN SV
SN SP
O
*-*LA FRASE ESTA BIEN CONSTRUIDA***
ANALISIS GRAMATICAL DE LA FRASE
PEDRO BEBE VINO
SUST BEBE VINO
SN BEBE VINO
SN VT+AUX VINO
SN GV VINO
* LA PALABRA VINO NO ESTA EN NUESTRA TABLA*
AUTOMATIC SYNTACTIC ANALYSIS IN DICTIONARY MAKING 173
ANALISIS GRAMATICAL DE LA FRASE
EL GATO BEBE LECHE
DET GATO BEBE LECHE
DET SUST BEBE LECHE
SN BEBE LECHE
SN VT+AUX LECHE
SN GV LECHE
SN GV SUST
SN GV SN
SN SV
SN SP
O
***LA FRASE ESTA BIEN CONSTRUIDA***
ANALISIS GRAMATICAL DE LA FRASE
PEDRO COME EL PAN
SUST COME EL PAN
$N COME EL PAN
$N VT+AUX EL PAN
SN GV EL PAN
SN GV DET PAN
SN GV DET SUST
SN GV SN
SN SV
SN SP
O
***LA FRASE ESTA BIEN CONSTRUIDA***
</SectionTitle>
    <Paragraph position="0"> 174 I. DEL CAMPO- I. GONZALES- M a. T. MOLINA- F. MARCOS</Paragraph>
  </Section>
  <Section position="11" start_page="31" end_page="31" type="metho">
    <SectionTitle>
ANALISIS GRAMATICAL DE LA FRASE
EL PERRO COME PAN
DET PERRO COME PAN
DET SUST COME PAN
SN COME PAN
SN VT+AUX PAN
SN GV PAN
SN GV SUST
</SectionTitle>
    <Paragraph position="0"/>
  </Section>
  <Section position="12" start_page="31" end_page="31" type="metho">
    <SectionTitle>
***LA FRASE ESTA BIEN CONSTRUIDA***
ANALISIS GRAMATICAL DE LA FRASE
EL GATO COME EL PAN
DET GATO COME EL PAN
DET SUST COME EL PAN
SN COME EL PAN
SN VT+AUX EL PAN
SN GV EL PAN
SN GV DET PAN
SN GV DET SUST
SN GV SN
SN SV
SN SP
O
***LA FRASE ESTA BIEN CONSTRUIDA***
AUTOMATIC SYNTACTIC ANALYSIS IN DICTIONARY MAKING 175
ANALISIS GRAMATICAL DE LA FRASE
EL PERRO RIEGA LAS FLORES
DET PERRO RIEGA LAS FLORES
DET SUST RIEGA LAS FLORES
SN RIEGA LAS FLORES
SN VT+AUX LAS FLORES
SN GV LAS FLORES
SN GV DET FLORES
SN GV DET SUST
SN GV SN
SN SV
SN SP
O
*NO HAY COHERENCIA ENTRE SUJETO Y VERBO*
*A PESAR DE ELLO LA FRASE ESTA BIEN CONSTRUIDA*
ANALISIS GRAMATICAL DE LA FRASE
EL GATO BEBE LAS LECHE
DET GATO BEBE LAS LECHE
DET SUST BEBE LAS LECHE
SN BEBE LAS LECHE
SN VT+AUX LAS LECHE
SN GV LAS LECHE
SN GV DET LECHE
*EL SUSTANTIVO NO CONCUERDA CON EL ARTICULO EN EL COMPLEMENTO*
</SectionTitle>
    <Paragraph position="0"> 176 I. DEL CAMPO- I. GONZALES- M a. T. MOLINA- F. MARCOS</Paragraph>
  </Section>
  <Section position="13" start_page="31" end_page="31" type="metho">
    <SectionTitle>
ANALISlS GRAMATICAL DE LA FRASE
LA GATO BEBE LA LECHE
DET GATO BEBE LA LECHE
*EL SUSTANTIVO NO CONCUERDA CON EL ARTICULO EN EL SUJETO*
</SectionTitle>
    <Paragraph position="0"> Even if we are at this moment at the preliminary steps, we hope that in the near future we shall be able to analyze more complicated sentences, helping this way to lemmatization by distinguishing, for instance, CANTO substantive, &amp;quot;song &amp;quot;, from CANTO, verb, &amp;quot;I sing &amp;quot;, or the two possibilities of ESPERABA,&amp;quot; I hoped &amp;quot;,&amp;quot; he hoped&amp;quot;</Paragraph>
  </Section>
class="xml-element"></Paper>