Re: Ligaturing (was: Yet Another I-Dotting Proposal)

From: Martin J. Dürst (mduerst@ifi.unizh.ch)
Date: Mon Aug 18 1997 - 09:49:40 EDT


On Thu, 14 Aug 1997, Otto Stolz wrote:

> So, German text requires
> - either language tagging + lingustic analysis,
> - or ZWNJ characters in all places where ligatures are forbidden
> (but who will be willing to type them in, consistently?),
> for correct ligaturing.

Well, who is willing to type in all the ligature characters correctly?
I guess the probability that a sequence of "f" and "i" ligates is around
50% for German. It's definitely difficult to decide which case is more
frequent.

What we can forsee is something like correct ligaturing being part
of a spell checker. For unknown cases, or ambiguous cases, the user
gets to select one of two variants (or more of them in the case of
an f-f-i sequence). How the system encodes these variants in the text
is rather irrelevant to the user.

Regards, Martin.



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:36 EDT