Re: Supporting multiple affixes for Agglutinative Languages in OpenOffice's Hunspell

Hi,

My answer for a similar question (Hunspell issue 2413299):

"Affix in Hunspell means affix combinations, too. For example, the Hungarian
word "kutyátlaníthatatlanságaitokéiért" from the stem "kutya"
generated by the -talaníthatatlanság derivative affix combination
(talan-ít-hat-atlan-ság) and the -aitokéiért inflectional affix
combination (-a-i-tok-é-i-ért), so two "affixes" are enough for complex
morphology, too:

$ hunspell -d hu_HU -m
kutyátlaníthatatlanságaitokéiért
kutyátlaníthatatlanságaitokéiért st:kutya po:noun ts:NOM
ds:tAlAn_LESS_adj ds:Ít_TRANSITIVE_vrb ds:hAt_MODAL_vrb ds:tAlAn_LESS_adj
ds:sÁg_ABSTRACT_noun ts:NOM is:PLUR is:POSS_PL_2 is:POSSESSEE is:PLUR
is:CAUS/FIN

Average Hunspell dictionary of an agglutinative language can contain 10-20
thousand affix rules. Unfortunately, there is no standard tool yet to
generate these combinations from simple n-fold descriptions (like in the
dictionaries of the two-level rule compilers). Hungarian affix combinations
are generated by nested m4 macros."

The Basque Hunspell dictionary was generated from a description for
two-level phonological rule-compiler, see
www.lrec-conf.org/proceedings/lrec2008/pdf/274_paper.pdf.

The Hunlex tool may help to describe and generate Hunspell
dictionaries for agglutinative languages:
http://wiki.apertium.org/wiki/Hunmorph

Regards,
László


> From: Edmond Kachale <[email protected]>
> Date: 2010/9/6
> Subject: [l10n-dev] Supporting multiple affixes for Agglutinative
> Languages in OpenOffice's Hunspell
> To: dev <[email protected]>, issues <[email protected]>
> Másolatot kap: issues <[email protected]>, Kevin Scannell
> <[email protected]>
>
>
> One of the languages I am working on is agglutinative in nature. The most
> afflicted sentence constituent is verb, which can carry up to 20 affixes
> (both prefixes and suffixes).
>
> I wonder if anyone knows how to enable this feature in OpenOffice's
> hunspell. I would like to support multiple affixes as follows:
> PFX + PFX + PFX + PFX + VERB + SFX + SFX + SFX + SFX + SFX
> e.g. ndi + kana + dza + ku + lemb + ets + er + a + ni + tu
> => ndikanadzakulembetseranitu (=I would have registered for you)
>
> I have tried using COMPLEXPREFIXES flag and continuation classes (as in PFX
> G 0 dza/ABCE .) but this makes it understand only two-fold affixes.
>
> I might be missing something, but I would like to support n-fold for an
> abrtriary 0 <= n <= 20.
>
> I'm using this reference:
>
>   - http://manpages.ubuntu.com/manpages/dapper/man4/hunspell.4.html
>   - http://pwet.fr/man/linux/fichiers_speciaux/hunspell
>
> ---
> Edmond
> Software Developer | Baobab Health Trust (http://www.baobabhealth.org/) |
> Malawi
>
> Cell:  +265 999 465 137 | +265 881 234 717
>
> *"Many people doubt open source software and probably don’t realize that
> there is an alternative… which is just as good.." -- Kevin Scannell*
>