GiellaLT provides an infrastructure for rule-based language technology aimed at minority and indigenous languages, and streamlines building anything from keyboards to speech technology. Read more about Why. See also How to get started and our Privacy document.
.xml to .lexc generation in the main/langs/LANG/src/morphology/stems
directory allows for reusing of lemma:stem:continuation information with other important dimensions of a given language. The same xml file might be used as a source file for the NDS reader’s assistant, enabling storage of source language to target language word pairs for multiple target languages. It might also be used, at least in the initial work, with: ICALL (Oahpa environment); Constraint Grammar; Rule-based translation (Apertium), and text-to-speech applications.
If you want to utilize .xml to .lexc generation in the main/langs/LANG/src/morphology/stems
directory, there are few items to bear in mind:
main/langs/LANG/src/morphology/Makefile.am
filemain/langs/LANG/src/morphology/stems/...lexc
files.In the main/langs/LANG/src/morphology/stems/
directory, you may wish to change from simple .lexc work to generating these .lexc files from analogous .xml files. This will mean that comments should be moved out of the main/langs/LANG/src/morphology/stems/...lexc
files and into perhaps the main/langs/LANG/src/morphology/root.lexc
or the main/langs/LANG/src/morphology/affixes/...lexc
files.
Once comments have been moved out of the main/langs/LANG/src/morphology/stems/...lexc
files, we are ready to declare generation targets and sources in the
main/langs/LANG/src/morphology/Makefile.am
file. There are three fields of declaration: (1) Giellatekno lexc sources, maintained source lexc files;
(2) generated lexc sources, these produced with xsl transformation, and (3) Giellatekno xml sources, the part-of-speech oriented database. If you are just beginning work with a language, your main/langs/LANG/src/morphology/Makefile.am
file might look something like the following:
START
# Set this to the names of all other source lexc files
GT_LEXC_SRCS=\
stems/abbr.lexc \
stems/acro.lexc \
stems/exceptions.lexc \
stems/punct.lexc \
stems/adjectives.lexc \
stems/adverbs.lexc \
stems/nouns.lexc \
stems/verbs.lexc \
affixes/adjectives.lexc \
affixes/adverbs.lexc \
affixes/nouns.lexc \
affixes/verbs.lexc
# Set this to the names of all generated lexc files, if any
GENERATED_LEXC_SRCS=
# Set this to the names of all source xml files, if any
GT_XML_SRCS=
END
The changes you want to make in the declaration will include the removal of the files:
stems/adjectives.lexc \
stems/adverbs.lexc \
stems/nouns.lexc \
stems/verbs.lexc \
from the
GT_LEXC_SRCS=\
declaration, and their declaration in
GENERATED_LEXC_SRCS=\
Please, note the commented new lines.
Additionally, the xml files must also be declared, resulting in the following:
START
# Set this to the names of all other source lexc files
GT_LEXC_SRCS=\
stems/abbr.lexc \
stems/acro.lexc \
stems/exceptions.lexc \
stems/punct.lexc \
affixes/adjectives.lexc \
affixes/adverbs.lexc \
affixes/nouns.lexc \
affixes/verbs.lexc
# Set this to the names of all generated lexc files, if any
GENERATED_LEXC_SRCS=\
$(srcdir)/stems/adjectives.lexc \
$(srcdir)/stems/adverbs.lexc \
$(srcdir)/stems/nouns.lexc \
$(srcdir)/stems/verbs.lexc
# Set this to the names of all source xml files, if any
GT_XML_SRCS=\
stems/adjectives.xml \
stems/adverbs.xml \
stems/nouns.xml \
stems/verbs.xml
END
A few notes:
As the .xml files become more extended and specific to your every need, you may choose to establish feeder .lexc files with simple lexc content. If you want to do this, there are various solutions to choose from. One used in kpv is the main/langs/LANG/src/morphology/stems/nouns_newwords.lexc
file.
Remember to declare the new file both in GT_LEXC_SRCS= of the main/langs/LANG/src/morphology/Makefile.am
file and the new lexicon in the main/langs/LANG/src/morphology/root.lexc
file, perhaps N_NEWWORDS would be a working lexicon name.
So far the xml format has only been used by the AKU project. The xml files used in AKU make use of a mutual xsd description of the xml structure, and the xsl conversion features simple Finnish glossing in the generated .lexc source files.
Here are a couple of further points:
<r xsi:noNamespaceSchemaLocation="$GTHOME/giella-core/schemas/fiu-dict-schema.xsd"
xml:lang="mrj" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:xs="http://www.w3.org/2001/XMLSchema"
xmlns:hfp="http://www.w3.org/2001/XMLSchema-hasFacetAndProperty"
xmlns:fn="http://www.w3.org/2005/02/xpath-functions">
src/morphology/stems/
is enough - they will then be automatically
converted to lexc, and from there on to the lexical fst as usual.Use of the xml format outside the AKU project may require some adaption or generalisations of the present structure.
The xsd provides a possibility for semantic classes to be added within mg elements. These semantic classes correspond to semantic classes used in the oahpa.no/davvi ICALL environment.