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.
Here we add useful commands.
echo "Nu lea." | apertium -d . sme-smn
The program actually goes through 9 steps in order to make the translation (cf. this picture of the overall process, if you are a visual person).
In order to see what goes on, we may stop on any of the 9 steps:
We thus write echo "Nu lea." | apertium -d . sme-smn-morph
etc. instead of only sme-smn, to get the output we want.
With all the tags the output may be messy. A practical advice
is to output one word on each line:
echo "Nu lea." | apertium -d . sme-smn-morph | tr ' ' '\n'
Here is the documentation for these modes
(Regression tests are tests ensuring things do not get worse).
To run the tests, write:
t/update-latest t/regression-test
The number at the bottom indicates the number success/test.
To add tests, go to these pages, click “edit” and format as for the other sentences:
Pending tests (just add tests here)
Regression tests (Add tests that succeed here)
Cf. also the documentation.
How to use the MT analysers and generators automorf (for analysis) and autogen (for generation)
Analyse North Saami:
echo ja|hfst-proc -a sme-smn.automorf.hfst
^ja/ja<cnjcoo>$
Generate North Saami:
echo '^ja<cnjcoo>$'|hfst-proc -d sme-smn.autogen.hfst
ja
Analyse Inari Saami:
echo já|hfst-proc -a smn-sme.automorf.hfst
^já/já<cnjcoo>$
Generate Inari Saami:
echo '^já<cnjcoo>$'|hfst-proc -d sme-smn.autogen.hfst
já