Re: ZWJ, ZWNJ and Markup languages.

From: Asmus Freytag (t) <>
Date: Fri, 27 Nov 2015 20:14:40 -0800
On 11/27/2015 5:42 PM, Martin J. Dürst wrote:
On 2015/11/28 04:55, Plug Gulp wrote:

The Unicode standard 8.0 states in chapter 23, section titled "Cursive
Connection and Ligatures"(printed page #814, PDF page #850) that:

"The zero width joiner and non-joiner characters are designed for use
in plain text; they should not be used where higher-level ligation and
cursive control is available. (See Uni-code Technical Report #20,
“Unicode in XML and Other Markup Languages,” for more information.) "

I went through TR#20 and did not find any mention that ZWJ and ZWNJ
are not suitable for use with markup languages. On the contrary, ZWJ
and ZWNJ are listed in TR#20 under section 4 titled "Format Characters
Suitable for Use with Markup".

So are ZWJ and ZWNJ characters suitable for use with markup languages
such as HTML and XML?

They are indeed suitable for use with markup languages. They are so suitable that they are already provided as entities in RFC 2070, which is now historic, and from there on through HTML 4.0 and onwards. Please see

I'm not sure why Unicode 8.0 has the text it has; at the least, this should be toned down somewhat to say "they may be replaced by higher-level ligation and cursive control mechanisms if available".
Thanks for finding this!

The main reason for this is that these characters apply at a single point; creating markup such as <zwj/> and <zwnj/> would not give any advantages over &zwj;/&zwnj;.

Markup is at its best when it can be applied to nested spans of text. It is not inconcievable that something like <do_not_ligate_inside>...
</do_not_ligate_inside> could occasionally be useful, but I have difficulties immagining a use case of the top of my head.

I'll file a bug report with the content of this email.

The whole thing started with the mistaken idea that ligation was independent of orthography.

It is not, though the examples are going away (as modern usage is being adapted to make it "easier" on software).

When you typeset Fraktur (which, according to Unicode, is merely a different style of rendering Latin text), the rules require that when you ligate "st" it must not be across word boundaries inside compound nouns.

However, there are many examples of compound words that have the same letters, but different internal division. So, you need to know where the intended division is and then mark it with zwnj.

That was definitely not appreciated from the outset, because in English ligation does not intersect with orthography.

That's my take on the root cause of all this strong language on the use of zwj / zwnj.

Received on Fri Nov 27 2015 - 22:15:51 CST

This archive was generated by hypermail 2.2.0 : Fri Nov 27 2015 - 22:15:52 CST