On Thu, Aug 17, 2006 at 08:25:03AM +0200, Steve Bennett wrote:
On 8/17/06, Andrew Dunbar hippytrail@gmail.com wrote:
No problem. Linguists create grammars by interviewing people who speak them and it apparently works best if these informants can't analyse their language grammatically. We're going very well so far! (-:
Hence the quote "Believing everything an informant says in his language, and nothing he says about it."
I had another thought, would it be possible to hack the nap parser such that a small set of magic words like d''a etc are parsed differently? If there's only a small number of special cases and they're not ambiguous (ie, no one would ever really want a d followed immediately be an a in italics...), then that might work? Or with javascript :)
IIRC, Steve, wasn't a large part of the problem that the parser would puke and die when presented with a *link* to such a page?
Cheers, -- jra