NAME

       Lingua::GA::Gramadoir - Check the grammar of Irish language text

SYNOPSIS

         use Lingua::GA::Gramadoir;

         my $gr = new Lingua::GA::Gramadoir;

         my $errors = $gr->grammatical_errors( $text );
         foreach my $error (@$errors) {
               # process $error appropriately
         }

DESCRIPTION

       This module contains the code for segmentation, spell checking, part-of-speech tagging, and grammar
       checking used by "An Gramadóir", an open-source grammar and style checker that can be used with vim,
       emacs, OpenOffice, or through a command-line interface.  An Gramadóir is intended as a platform for the
       development of sophisticated natural language processing tools for languages with limited computational
       resources.

       The Perl code contained in this module is generated automatically from a higher-level representation of
       the grammatical rules and should not be edited directly.  Anyone interested in helping improve the
       lexicon or the rule sets should download the developers' pack from the An Gramadóir web site:
       <http://borel.slu.edu/gramadoir/>.

CONSTRUCTOR

       new %PARAMS
           Constructs an instance of the grammar checker and loads the lexicon into memory.  It should only be
           called once.   Options may be specified by passing a hash containing any of the following keys:

           fix_spelling => 0

           Suggest replacements for misspelled or unknown words.

           use_ignore_file => 0

           Read a file containing words to be ignored when checking spelling and grammar.

           unigram_tagging => 1

           Resolve ambiguous part of speech according to frequency.  This should be set to false only for
           debugging purposes because the pattern matching for grammatical errors relies on complete
           disambiguation.

           interface_language => ""

           Specify the language of output messages (not necessarily the language of the text to be checked).
           With the default value, Locale::Maketext attempts to determine the correct language to use based on
           things like your environment variables.

           input_encoding => 'ISO-8859-1'

           Specify the encoding for all texts passed to one of the module's exported functions.   There is no
           currently no way to change the encoding of the data returned by the exported functions (always
           encoded as perl strings).

METHODS

       get_sentences TEXT
           Splits the input TEXT up into sentences and returns a reference to an array containing the sentences.

       tokenize TEXT
           Splits the input TEXT up into orthographic words and returns a reference to an array containing the
           words.

       spell_check TEXT
           Returns a reference to an array containing the misspelled words appearing in the input text.

       all_possible_tags WORD
           Takes the input WORD and returns it with (XML-style) markup indicating all of its possible parts of
           speech.

       add_tags TEXT
           Takes the input TEXT and returns a reference to an array of sentences with (XML-style)
           *disambiguated* part-of-speech tags.  Does not do any grammatical rule checking.

       xml_stream TEXT
           Takes the input TEXT and returns it as well-formed XML (encoded as perl strings, not utf-8) with full
           grammatical markup.   Error messages are not localized.  This function should only be exported for
           debugging/development purposes.  Use "grammatical_errors" (which is basically "xml_stream" plus some
           whittling down) as an interface with other programs.

       grammatical_errors TEXT
           Returns the grammatical errors in the input TEXT as a reference to an array, one error per element of
           the array, with each error given in a simple XML format usable by other applications.  Error messages
           are localized according to locale settings as determined by Locale::Maketext.

SEE ALSO

       •   <http://borel.slu.edu/gramadoir/>

       •   Locale::Maketext

       •   perl(1)

BUGS

       The grammar checker does not attempt a full parse of the input sentences nor does it attempt to exploit
       any semantic information.  There are, therefore, certain constructs that cannot be dealt with correctly.
       See <http://borel.slu.edu/gramadoir/bugs.html> for a detailed discussion and specific examples.

AUTHOR

       Kevin P. Scannell, <kscanne@gmail.com>.

COPYRIGHT AND LICENSE

       Copyright (C) 2004, 2005, 2007 Kevin P. Scannell

       This library is free software; you can redistribute it and/or modify it under the same terms as Perl
       itself, either Perl version 5.8.2 or, at your option, any later version of Perl 5 you may have available.