On Mon, Jul 18, 2011 at 12:28 PM, Jörg Rhiemeier <[log in to unmask]> wrote:
> Many
> languages also have words that are just plain irregular, and
> those are usually among those that are used most.

Unless I am misunderstanding "irregular" it seems to me those could
still be handled by glossing. If the irregularities have to do with
declension or conjugation there's nothing that would prevent, for
example, "be.1SG" from being a different dictionary entry than
"be.3PL". The glossing itself would have to reflect the irregularities
and carry all that hand-entered information in order to choose the
correct irregular form from the dictionary.

Remember that what I am proposing is not an "intelligent" translation
program, but simply and automated version of a dictionary-driven
search and replace. The structure of the gloss and the structure of
the final "translation" would necessarily be identical, because the
program would not be capable of changing the sentence _structure_ in
any way.

Also, a given gloss would be unique to the conlang in question. That
same gloss could not be used to produce a "translation" into a
different language.