[tex-hyphen] Improving hyphenation
wl at gnu.org
Mon Sep 6 08:11:52 CEST 2004
> I don't know if this falls within the charter of the tex-hyphen
> list, but I'd like to humbly point at my old work on hyphen classes,
> which is available at
> Hyphen classes allow to express the fact that, e.g. in the German
> word Au|to|un|fall, it is more desirable to break at the second
> hyphen position than at the other two.
I know this well, but AFAIK there aren't any serious hyphenation
patterns available for testing...
BTW, there was another idea (described somewhere in TUGBoat in the
last years) how to handle compound words, namely to use two passes of
hyphenation, or perhaps even more. The first level uses patterns
which contain only compound hyphenation points (Auto-unfall); if this
doesn't give good results, add hyphenation points for non-compound
words (Au-to and un-fall). There were some other nice properties of
the suggested algorithm which I've forgotten currently.
Has there ever been an implementation of the latter? I'm especially
interested in a comparison of the two methods.
More information about the tex-hyphen