W3C

– DRAFT –
Pronunciation TF

10 August 2026

Attendees

Present
PaulG
Regrets
-
Chair
-
Scribe
matatk

Meeting minutes

Agenda Review, Membership & Announcements

<PaulG> with my new work responsibilities, I can't spend much time on this TF for the time being

Minutes from last time: https://www.w3.org/2026/04/13-pronunciation-minutes.html

<PaulG> matatk: there may be renewed interest due to some work in authoring tools

[DRAFT] ARIA Counter Proposal: https://github.com/w3c/pronunciation/wiki/%5BDRAFT%5D-ARIA-Counter-Proposal

PaulG: The ARIA proposal solves a couple of things, but not others, such as break/pause. We could pull back to just going with the ARIA approach as standalone.

PaulG: I'd broken out a way to say for example whether something should be read as ordinal, cardinal, or characters (e.g. ten)

PaulG: The ARIA breaking part required there to be a string that caused the break. You would have to include something in there to cause breaks (could use a non-breaking space, though).

PaulG: ...So a little hacky but could be done. Emphasis and Whisper were more forced/kludgy.

PaulG: Say-as, substitution, interpret, were pretty strong

matatk: We talked last time about the importance of determinism.

PaulG: Yes, ETS research backing up learners being more successful when they heard consistent pronunciations. But they put pause/break as an equal requirement to that, and weren't able to accept half measures.

PaulG: I think if we went with the ARIA proposal we'd get stuck on pause/break again and would need to make peace with that. Pause/break would still be heuristic until another solution is found

matatk: Concern about whether we would preclude fixing that if we adopted the AIRA approach for other parts (needs research).

matatk: (explains how Adapt symbols are intended to work - ruby, Unicode, font mapping)

matatk: How could we use this for pronunciation?

PaulG and matatk: What about lang=ipa (or something like)

*group looks into this*

PaulG: It's an alphabet, not a language, but maps to all llanguages of which we know.

matatk: Does it do break/pause?

PaulG: There's are various types of prosodic boundary indicators.

PaulG: May not be enough for some of the cases we need.

matatk: This has the potential to be visible in more situations for more users, rather than the AIRA approach which is invisible to non-AT users.

PaulG: Could render conditionally based on different interactions.

PaulG: Would then need to revive the proposed type attr on <ruby> e.g. type=ipa, lang=en-gb

matatk: Could we also do (as per some education cases) 'English as spoken in the 1920s?'

PaulG: We discussed an approach for 'leisure' in en-gb and en-us; what about 'color' and 'colour' - we have a matrix of markup examples. Somewhere there'd need to be a map. Can't be in the UA as it has to be author-controlled. Needs thought.

Other Business

Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Maybe present: matatk

All speakers: matatk, PaulG

Active on IRC: matatk, PaulG