RE: [OpenType] Proposal: Ligatures w/ ZWJ in OpenType

From: Asmus Freytag (
Date: Tue Jul 09 2002 - 04:20:22 EDT

At 09:23 PM 7/7/02 -0700, John Hudson wrote:
>At 20:38 7/7/2002, Paul Nelson wrote:
>>Typography should be a higher level handling and should not muck up the
>>backing store of text in any manner whatsoever.
>I agree. The issue for me with the use of ZWJ, at least as far as the
>Latin script is concerned, concerns the specification of ligation in
>documents in which ligation or non-ligation is *not* typographic but is
>germane to the content. Considering a typical content/markup model, it
>seems to me that in such circumstances there is a genuine need to encode
>ligation in the content, rather than in the markup.

So far so good.

>My proposal seeks to distinguish such cases -- exceptional cases, I
>believe -- from those that involve typographic decision making about the
>display of text, and to handle them via a separate mechanism.

While not all languages operate with the same rules, Unicode attempts to
provide for the encoding of text in all languages (as opposed to just the
easy ones for a given script). Let's be careful then, lest the needs for
certain standard orthographies (as opposed to technical usage by scholars)
become ruled out as exceptions that it is somehow not necessary to support.

>>Frankly, I don't see that task as something that would ever be achieved
>>because there is no way to force people to update their software to get
>>the new behavior. It also represents a significant and costly
>>development/testing effort to support this type of behavior.
>Fair enough. I see the use of ZWJ/ZWNJ for the Latin script as something
>that is limited to specific branches of scholarship, e.g. document
>studies, in which the appearance of texts is germane to the content of studies.

OK. Here we go again. There is simply no way that one can 'typographically'
ligate standard German without text (!) based control, since places where
ligatures are prohibited depend on the meaning (i.e. intended content) of
the text, in a way that's similar to hyphenation.

This must not be confused with arbitrary ligation control used in document
studies. Those people need a lot more special features, so that they are
far removed from plain text at any rate.

The use of ZWNJ was not primarily introduced for the academic audience, but
for all (modern) languages where allowing blanket ligation based on
character pairs is orthographically incorrect.


This archive was generated by hypermail 2.1.2 : Tue Jul 09 2002 - 02:33:15 EDT