SFST, HFST and XFST and things like them can do this and in ways that are
not inefficient for the program (and in fact part of the intended use). If
you have an exact example, I might be able to show you what you'd do.
Otherwise, the example you've provided should be easy.
I was a little curious about why you were asking with respect to Bribri, so
I looked up some examples of morphology and phonology in the language, and
it seems like it'd be a fun challenge given how tonal it is, and that
there's nasal harmony (ooh!). I have no doubts you'd work something out for
the problems you'd face in that language with one of these programs,
assuming that everything in the program is in order as it should be. With
HFST anyway, the community is fairly supportive and if you hang around in
the IRC channels you can get fairly quick help and diagnose whether
potential bugs are in the software or just understanding how the software
works.
Ryan
2011/9/17 Sofia Flores <[email protected]>
> Ok, what about SFST?, do you know if it is better, in a linguistically
> way, than lttolbok with the alignments to the corresponding
> morphemes?
>
> Sofia
>
> 2011/9/17 Jimmy O'Regan <[email protected]>:
> > On 17 September 2011 17:26, Sofia Flores <[email protected]>
> wrote:
> >> Hello
> >>
> >> I want to build a morphological parser for bribri (Amerindian
> >> language from Costa Rica). I have tried lttoolbox, but my question is
> >> if is possible to build aligned morphological dictionaries,
> >> so the correspondence between the surface form and the lexical form of
> >> a word is aligned at the character level, ej:
> >>
> >> (h,h) (o,o) (u,u) (s,s) (e,e) (s,<n>), (θ,<pl>)
> >>
> >> instead of
> >>
> >> houses, house<n><pl>
> >
> > Sure:
> >
> > <e>
> > <p><l>h</l><r>h</r></p>
> > <p><l>o</l><r>o</r></p>
> > <p><l>u</l><r>u</r></p>
> > <p><l>s</l><r>s</r></p>
> > <p><l>e</l><r>e</r></p>
> > <p><l></l><r><s n="n"/></r></p>
> > <p><l>s</l><r><s n="pl"/></r></p>
> > </e>
> >
> > ...but that's extremely inefficient, and the task is performed by
> > lttoolbox internally, with the added bonus that it finds the optimal
> > alignment itself. The <i> element effectively says 'each contained
> > character maps to itself', which is usually what you want.
> >
> > --
> > <Sefam> Are any of the mentors around?
> > <jimregan> yes, they're the ones trolling you
> >
> >
> ------------------------------------------------------------------------------
> > BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA
> > http://p.sf.net/sfu/rim-devcon-copy2
> > _______________________________________________
> > Apertium-stuff mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/apertium-stuff
> >
>
>
> ------------------------------------------------------------------------------
> BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA
> http://p.sf.net/sfu/rim-devcon-copy2
> _______________________________________________
> Apertium-stuff mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>
------------------------------------------------------------------------------
BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA
http://p.sf.net/sfu/rim-devcon-copy2
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
