<?xml version="1.0" standalone="yes"?>
<Paper uid="C96-1031">
  <Title>GramCheck: A Grammar and Style Checker</Title>
  <Section position="3" start_page="0" end_page="175" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Grammar checking stelnmed as a logical application from forlner attelni)ts to natural language. ulMerstanding by comtmters. Many of the NLU systems developed in the 70's indu(le(l a kind of error recovery Inechanisln ranging flom the treatment only of spelling e.rrors, PARRY (1)arkin son c't al., 1977), to tile inclusion also of incomplete int)ut containing some kind of ellipsis, LAD-DEll,/LIFEll (Hendrix et al., 1977).</Paragraph>
    <Paragraph position="1"> The interest in the 80's begun to turn considering grammar checking as an enterprise of its own right (Carbonell &amp; Hayes, 1983), (Ilayes &amp; Mouradian, 1981), (Heidorn et al., 1982), (.lensen at al., 1983), though many of the approaches were still in I;t1(: NLU tradition ((\]harniak, 198a), (Granger, 1983), (Kwasny &amp; Sondheimer, 1981), (Weischedel &amp; Black, \]980), (Weisehedel &amp; Sondheimer, 1983). A 1985 Ovum report on nal;llral language applications (.lohnson, 1985) already identifies grammar and style checking as one of the seven major apt)lications of NLP. Currently, every project in grammar checking has as its goal the creation of a writing aid rather than a robust man-machine interface (Adriaens, 1994), (llolioli ctal., 1992), (Vosse, 1992).</Paragraph>
    <Paragraph position="2"> Current systems dealillg with grammatical deviance have be(m inainly involve(t in the integi~&gt; don of special techniques to detect and correct, when possible, these, deviances. In some case.s, these have be.en incorporated to traditional parsing techniques, as it is the case with feature relaxation in the context of unification-based formalisms (Bolioli et al., 1992), or the addition of a set of catching error rules si)ecially handling the deviant constructions (Thurlnair, 1990). In other eases, the relaxation component has heen included as a new add-in feature to the parsing algoril,hm, as in the IBM's PLNLI' aI)proach (Heidorn et al., 1982), or in the work developed for tim Tra.nslator's Workbench t)roject using the METAL MT-system (TWB, 1992).</Paragraph>
    <Paragraph position="3"> Besides, an increasing concern in current projects is that of linguistic relevance of the analysis t)erformed by the grammar correction system.</Paragraph>
    <Paragraph position="4"> In this sense, the adequate integration of error detection and correction techniques within mainstream grammm&amp;quot; formalisms has l)een addressed by a nunl|)er of these projects (\[Iolioli eta/., 1992), (Vosse, 1992), ((\]enthia.l ctal., t992), (O(~uthial et al., 1994).</Paragraph>
    <Paragraph position="5"> l~bllowing this concern, this paper presents resuits fl'om the project GramCheck (A Grammar and Style Checker, MLAP93-11), flmded by the CEC. GramCheck has developed a grammar checker demonstrator for Spanish and Greek native writers using ALEP (ET6/1, 1991), (Simpkins, 1994) as the NLP development platform, a client-server architeeUlre as implenmnted in the X Windows system, Motif as the 'look ~md fe, el' interface and Xminfo as the kllowh!dge t)ase, storage format. Generalized use of extensions to the highly typed and unifi(:ation based formalism imi)Iemented in ALEP has been 1)erformed. These extensions (called Constraint Solvers, CSs) are nothing but pieces of PR()I,OG code l)erforlning different l)oolean and relational operations over feature wdues. Besides, GramCheck has used ongoing results Dora LS-GRAM (LRE61029), a project alining at the implementation of middle  coverage ALEP grammars for a number of European languages.</Paragraph>
    <Paragraph position="6"> The demonstrator checks whether a document contains grammar errors or style weaknesses and, if found any, users are provided with messages, suggestions and, for grammar errors only, autoinatic correction(s).</Paragraph>
    <Paragraph position="7"> 2 Brief grammar error typology for Spanish The linguistic statements made by developers of current grammar checkers based on NLP ted&gt; niques are often contradictory regarding the types of errors that grammar checkers must correct automatically. (Veronis, 1988) claims that native writers are unlikely to produce errors involving morphological features, while (Vosse, 1992) acce.t)ts such morpho-syntactic errors, in spite of tile fact that an examination of texts by the author revealed that their appearance in native writer's texts is not frequent. Both authors agree in characterizing morpho-syntactic errors as a sainple of lack of competence.</Paragraph>
    <Paragraph position="8"> On the other hand, an examination of real texts produced by Spanish writers revealed that they do produce morpho-syntactic errors I . Spanish is an inflectiolml language, which increases the possibilities of such exrors. Nevertheless, other errors related to structural configuration of the language ark: produced as well.</Paragraph>
    <Paragraph position="9"> Errors found fall into one of the following subtypes, assuming that featurization is the technique used in t)arsing sentences: 1. Mislnatching of features that do It(it affect representational issues (intra- or intersyntactic agreement on gender, number, per-son and cask' for categories showing this phenonmnon). These mismatchings produce rto~?,- St'l&amp;quot;ttct~K/'ttl e7&amp;quot;l'O~'S.</Paragraph>
    <Paragraph position="10"> 2. Mismatching of fl:atures which describe certain representational properties for categories, as wrong head-argument relations, word order and substitution of certain categories. These mismatchings produce structural errors.</Paragraph>
    <Paragraph position="11"> Table 1 shows rite percentages of diflhrent types of errors tbund in the corpus. Punctuation errors must be considered as structural violations, while for style weaknesses, it depends on its subtype.</Paragraph>
    <Paragraph position="12"> Errors at the lexical level are diflh:ult to classit~y and itios(; of them must regaMed as spelling rather than gralnlnar errors. The nulnl)er of erroIs identiffed in this corpus is 543. These statisti(:s couhl 1The corims used contains nearly 7(I,000 words including text fragments from literature, newspal)(:rs , technical and administradw: documentation. It has been provided to a large extent by GramCheck pilot user, ANAYA, S.A.</Paragraph>
    <Paragraph position="13"> be regarded as a representative average of the fi'equency of errors/mistakes oc(:urring in Spanish texts.</Paragraph>
    <Paragraph position="14">  &amp;quot;Errors at the h'xical level includ(' tyl)ing errors, word segmentation (ai no vs sirl.o), and cognitive errors (onccavo (pmtitivc) vs nndd.cirno (ordinal).</Paragraph>
    <Paragraph position="15"> A presupposition adopted in the project led to the idea that violations at the featm'e level can be capl:ured by means of the relaxation of the possibly violated features while violations at (;he level of configuration may not be relaxed withou(; raising unpredictable parsing results, thus being candidates for the implementation of explicit rules encoding such incorrect structures.</Paragraph>
    <Paragraph position="16"> Under this view, a comprehensive gralnlnar checker must make use of both strategies, called in the literature feature or constraint relaxation and error anticipation, respectively.</Paragraph>
    <Paragraph position="17"> However, given the relevance of features ill the encoding of linguistic information in TFSs, SOIIlO strllcttlral errors (;ail t)e re.analyzed as agrcelilent errors in a wide sense (as feature nfisma.tching violal;ions rather than structural ones). This allows the implementatioll of a uniforn) apt)roach to grammar ('orrcction, thus a.voitting explicit rules for ill-formed illpUt. This paI)er describes such ilnpleinentatioI~ for both Ilon-sl;rnctural and strltt:tura.I violations.</Paragraph>
  </Section>
class="xml-element"></Paper>