From: Asmus Freytag (asmusf@ix.netcom.com)
Date: Sun Mar 06 2005 - 21:57:01 CST
At 11:57 AM 3/6/2005, Doug Ewell wrote:
>Word has a language attribute for runs of text, used for spelling and
>grammar checking. Could this be used as a starting point for ligation
>as well? That might handle the Irish and Scandinavian and Turkish and
>Azeri cases.
Actually, the word fjord is widely used as a loan woard.
>German and other languages where ligation depends on syllable or subword
>boundaries might apply dictionary lookup, and provide an easy undo or a
>confirmation stage (yellow squiggly line?) for ambiguous cases like
>Wachstube.
This is not unreasonable, as automatic hyphenation has that issue as well,
but compounds are notoriously productive, so the use of dictionaries will
fail to capture ambigous cases. The (in-)famous one for hypehenation is
"Urinstinkt", which is a nice compound that unfortunately shares all its
letters, in that order, with a not so nice sentence.
>I'm not pretending this is trivially simple, or black-and-white. I'd
>just like something better than "no ligation ever, under any
>circumstances."
The amount of work needed to provide correctly globalized softare, and,
in some languages, the amount of work for the user to get the feature to
work will push back. If both of these were trivial concerns, we would
undoubtedly see all office memos and e-mail correctly ligated already.
A./
This archive was generated by hypermail 2.1.5 : Sun Mar 06 2005 - 21:57:31 CST