W3C

– DRAFT –
WCAG3 Implied Meaning Subgroup

03 sep 2024

Attendees

Present
JohnRochford, julierawe, kirkwood, Laura_Carlson, LenB, Makoto, Rain
Regrets
Jan, Mike
Chair
julierawe
Scribe
julierawe

Meeting minutes

https://docs.google.com/document/d/1P7fOyEPVlqf1aXuJY0SO9LeC-E7EZllg/edit#heading=h.r8flmepsu94v

<Frankie> prrsent+

We are reviewing updated draft from the top

John R asked Makoto if Japanese includes onomatopia words like "plop"

Makoto said yt

yes

julierawe Reminder that we are trying to come up with universal guidance and will include language-specific tests

<Rain> Determine if the text requires specific background knowledge that may not be universal to understand it.

<Rain> Above as a potential best practice or guidance to offer people to consider, but not require specifically because too broad

JohnRochford Previously in our work we referenced common-word lists

JohnRochford We could say if your text uses common words or phrases?

<kirkwood> i didn’t want to put into notes (link shared in chat) but there is much discussion about large language models and implied meaning in ai

JohnRochford How will people recognize if they are using clear language?

JohnRochford We could have a list. In the working draft from 2023, we have a long list

<kirkwood> implied meaning in ai language models is under discussion here and may be useful for us: https://www.aimodels.fyi/papers/arxiv/do-language-models-capture-implied-discourse-meanings

<JohnRochford> https://docs.google.com/document/d/1JKJ32DK_A82ahsQ_k0RVPc-ZYpX5uuKtuGOLdo2OqyY/edit#heading=h.4bh5hmxp8y5q

JohnRochford I don't see any practical way to determine all of the possible meanings that a word can have in every language

JohnRochford If you don't focus on the list of common words, and you just craft a technique, maybe this will be helpful

Rain I think having a framework to craft the techinque will help us

JohnRochford I suggest the way we crafted the common words draft could be helpful

<kirkwood> do we need a method to determine if there is implied meaning?

JohnRochford How can we help people determine if they have implied meaning?

kirkwood That's the big question to me

Rain We are helping people, not doing it for them. We're giving them things they can try. We're not guaranteeing their success.

Rain We can say, here are some things that can help you in this process. A set of suggestions or offerings.

JohnRochford I agree. What Kirkwood and I are talking about is how we can help people recognize if they're using literal text or not.

Rain I think we need to be more suggestive, not proscriptive

<kirkwood> so is it that we need to make a method to determine if there is a significant implied meaning that affects reasonable understanding of content

kirkwood Defining non-literal language is a start.

Frankie I'm having trouble understanding what we're debating.

Frankie Common words depends on the space I'm working in.

<kirkwood> +1

Rain When we're asking authors to determine if they have implied meaning or not, what would be helpful for you?

Rain There is always a level that is context-specific to the space. Style guides are incredibly important to making those decisions.

Sorry, that was Frankie

<kirkwood> +1 to Frankie

Frankie A lot of tech writers are using AI tools to go through their writing for a specific reading level

Frankie We'd have a list of don't-use-these-words

<kirkwood> I feel we should scrap the concept of a common words list

Frankie I'm not sure how helpful a list of common words would be

Rain I agree. What we need to do is transition into what we're putting in the document.

Rain One technique could be reference a style guide, but what if you don't have a style guide?

<kirkwood> is using ai a technique ?

<Laura_Carlson> Can the language used be understood exactly as stated, according to the definitions attached to the words involved?

Makoto WCAG 2.2 has a success criterion about using unusual words. It said you have to explain only if you think the intended users are not familiar with the words. Don't need to explain jargon to professionals who use that jargon. Up to the website to determine.

<kirkwood> Mkota had a good point “unusual for the intended user of website”

LenB Sports websites or car websites or technology websites—they all have their own way of speaking.

LenB An attorney website will have very different language than a sports website.

<kirkwood> well said, good example

<Rain> Those are great examples

julierawe Let's keep the focus of this outcome on implied meaning rather than jargon. A sports site or an attorney site will use a lot of jargon, but if those sites use metaphors, sarcasm, etc, that's what this outcome should focus on.

julierawe Let's all review the margin comments in the draft before our next meeting, thanks.

Minutes manually created (not a transcript), formatted by scribe.perl version 229 (Thu Jul 25 08:38:54 2024 UTC).

Diagnostics

No scribenick or scribe found. Guessed: julierawe

Active on IRC: Frankie, JohnRochford, julierawe, kirkwood, Laura_Carlson, LenB, Makoto, Rain