Lack of support for applying extra spacing between Chinese/Japanese and Western text

[xfq]

This issue applies to Chinese and Japanese.

In Chinese and Japanese composition, it is usually recommended to apply extra spacing between Chinese/Japanese han/kana characters and certain other types of text to make the text read better. The gap is typically used between han/kana characters and Latin letters, numbers, or punctuation, and should be smaller than an ASCII space would be.

ASCII spaces are not only too wide, but they spoil the semantics and behaviour of the text.

More:

• Chinese
• Japanese

The GAP (original text)

Currently, browsers do not support this feature, so content authors have to work around the problem by adding spaces manually, but because this produces too wide spaces and inserts characters that are not appropriate into the text, many texts don't have any gap between these runs of text.

In the css-text specification, text-autospace is intended to create 1/8em extra spacing between runs of ideographs/kana and Western letters/numerals.

In order to remediate texts where ASCII spaces have been used, the CSS text-autospace property also has a replace value, which will (without changing the underlying code points) replace the spaces with an appropriately-sized gap during the rendering of the text.

By default space should be added (the default value is normal, which is equivalent to ideograph-alpha and ideograph-numeric.)

Gecko, Blink, and Webkit browsers all fail to add the extra spacing between runs of ideographs and non-ideographic letters.

Priority

Given that, though not ideal, people have been forced to omit the gap or (worse) use ASCII spaces, this is prioritised as an advanced issue, however this is really a basic feature of Chinese/Japanese text, and so needs to be addressed.

Tests & results

text-autospace: ideograph-alpha ideograph-numeric; creates a thin gap between runs of han characters and both ASCII letters and digits in Chinese

text-autospace: ideograph-alpha ideograph-numeric; creates a thin gap between runs of kanji and both romaji letters and digits in Japanese

text-autospace: ideograph-alpha ideograph-numeric; creates a thin gap between runs of hiragana and both romaji letters and digits in Japanese

text-autospace: ideograph-alpha ideograph-numeric; creates a thin gap between runs of katakana and both romaji letters and digits in Japanese

text-autospace: ... replace; replaces ASCII spaces with a thin gap when adding text spacing in Japanese

Action taken

Bug reports: Gecko 1 • Gecko 2 • Blink 1 • Blink 2 • WebKit 1 • WebKit 2 • WebKit 3

GH issues: spacing

Outcomes

As of version 18.4 WebKit browsers now support text-autospace: ideograph-alpha|ideograph-numeric, but they don't yet support the replace keyword. This is problematic because, many websites currently use a space to separate these items, even though it is not ideal.

WebKit also applies no-autospace as the default, contrary to the CSS spec, which expects the value to be normal. There is an ongoing discussion about whether the default behaviour should add or not add gaps.

Gecko is considering support but it is not yet available.

The Blink issue has been marked as fixed on Jul 10, 2025. It is not yet available on Blink browsers.

Unicode proposal: EAST ASIAN AUTO SPACING (Proposal)

text-autospace	Blink	Gecko	WebKit
ideograph-alpha	❌	❌	✅
ideograph-numeric	❌	❌	✅
replace	❌	❌	❌

[xfq]

The first comment in this issue contains text that will automatically appear in one or more gap-analysis documents as a subsection with the same title as this issue. Any edits made to that comment will be immediately available in the document. Proposals for changes or discussion of the content can be made in comments below this point.

Relevant gap analysis documents include:
Chinese • Japanese

[xfq]

@r12a How to add this to the Gap-analysis pipeline?

[macnmm]

The main problem with the workaround is the lack of control in how the widths of these spaces are adjusted. For Japanese spacing, the space glyph in whatever font is likely too wide and doesn't collapse correctly when the line is compressed. For Korean, the space character is always input for word spaces, but again the width is not correct and not contextually controllable (Latin-K versus K-K can be different).

[r12a]

@r12a How to add this to the Gap-analysis pipeline?

It's already there, but only in the editor's version. You need to publish the doc to /TR using echidna once ready. There should be some instructions somewhere for that – i'll check, because things recently changed when new process was added by webreq.

[r12a]

@xfq Could you add a brief description of what the spec has support for (maybe just one sentence will suffice, but it's good to indicate the basis on which this issue is founded, in case the spec changes).

Also, please (again briefly) describe the results of the test you list.

thanks

[r12a]

Btw, i think this issue probably also applies to Thai and other scripts that don't use space as a word-separator, not just C&J. And you should probably create a stub, pointing here for the Japanese gap-analysis doc, so that we don't end up duplicating work.

[xfq]

@r12a How to add this to the Gap-analysis pipeline?

It's already there, but only in the editor's version. You need to publish the doc to /TR using echidna once ready. There should be some instructions somewhere for that – i'll check, because things recently changed when new process was added by webreq.

To clarify, I mean the pipeline, not the gap-analysis document.

[xfq]

@xfq Could you add a brief description of what the spec has support for (maybe just one sentence will suffice, but it's good to indicate the basis on which this issue is founded, in case the spec changes).

Also, please (again briefly) describe the results of the test you list.

Will do. Thank you.

[xfq]

Btw, i think this issue probably also applies to Thai and other scripts that don't use space as a word-separator, not just C&J. And you should probably create a stub, pointing here for the Japanese gap-analysis doc, so that we don't end up duplicating work.

I did consider Thai at the time of writing, but because Thai do use spaces (although not as a word-separator), I didn’t write it. If you are sure that Thai also has this requirement, I can write it in this issue. Or should I create a sealreq issue?