> No idea. I prefer to slowly add new words as I go through the lessons so
> I can always learn stuff that's relevant or already known.
I suppose it wouldn't be hard to acquire some kind of translation
database; I'll post back on the mailing list if I find anything
suitable with a translation script/tool to make it work with the sfc
database format, as a resource.
> That'd be great, but writing the rules and keeping it generic sounds
> annoying. Maybe some sort of tagging system [nn,st] for noun, strong
> declension or similar would work. But you'd still have to specify what
> to do based on the tags, and judging from the few languages I've learned
> that could be quite a lot... Changing the end, or the middle, etc. The
> end is obviously simple enough (just put the base in the word and let
> tags handle the rest), but if the stem changes as well... Bleh. If you
> do write a patch, just send it to me ;)
Yeah at first I'm thinking only support for simple ending changes, but
as it progresses it could grow to encompass stem changes, etc. At
first I'm interested in West/North Germanic languages, but really
anything could work theoretically. I like the idea of tags, you can
have word types (noun, adjective, verb, adverb, preposition,
conjuction, etc), and then combined inflection type markers ("strong
feminine 1", "common", "strong", "irregular"; but shortened), as you
showed. A separate character (like '-') in conjunction with defined
inflection rules should be enough for tacking on affixes, but stem
changes could just have some psuedo-regular expression markup to
denote what changes where, or simply list the other forms for
irregulars.
I guess the alternative is just to enumerate everything once and tag
it (generating each case with endings), but I suppose that would be
just as much work and larger files, plus it leaves updating as a
difficult task. The nice thing about doing it the first way is it
pretty much leaves the extended inflection handling as an optional
component that one can blatently ignore if they so desire (besides
stripping the stem marker ('-')).
I'll have to work with it a bit though and see what actually turns out
to be sane.
--
Samuel Baldwin - logik.li
