updates from Clear Language (outcomes only)

[jspellman]

Added Editor note from Outcomes section of google doc and Outcomes 1-23

https://raw.githack.com/w3c/wcag3/clear-Language-aug23/guidelines/index.html#clear-language

[rachaelbradley]

Summary:

This Pull Request includes the Outcomes from the clear language subgroup.

Maturity Level

Exploratory

Action Needed:

• Review content
• Add a thumbs up if you agree with the general direction, thumbs down if you do not, considering (eyes) if you have concerns but are generally ok with this going forward.
• Add comments to PR or Google Doc (Note: We will lock this thread at 8 pm eastern on Monday August 21st.)

Content for Review:

• Code in PR
• Github Preview
• Google Document

[u9000-Nine]

I generally like the intent of this criterion, but I have two main concerns about its general direction right now:

1. I feel like it is nonfiction-specific. I feel like many of its criteria have this problem, but it is most evident with things like a limit on the number of paragraphs per section. In my experience, fiction works often benefit from smooth transitions to guide readers through the story, and clearly defined short sections might make it harder to follow (or at least less entertaining).
2. I feel like some criteria overly restrict the author's voice. I think most do not have this problem, but some do beyond a reasonable amount. As an example, I feel like 2.3.2's "Unnecessary words or phrases" would remove a lot of the personality that can shine through in an author's voice, especially considering that (in my experience) these words don't cause a problem most of the time.

[johnfoliot]

2.3.1 Word Comprehension - Outcome: Uncommon words:

I currently have some significant concerns related to Internationalization and the requirement for "word lists". While the draft points to different lists for english words, I cannot locate similar resources that would be the equivalent for other common languages. Is the AG intending to create these lists as part of this requirement? If not, who will? (I also have a secondary concern if/when two similar but not identical lists emerge from different sources: which list would be considered 'normative'? What about differences in spelling? - i.e., in Canada and the UK it's colour, not color, theatre versus theater, etc.)

What, (if anything) would the role of spell check dictionaries play here?

The introductory section also notes, "Internationalization: Several outcomes for Clear Language will need conditional tests by language. Our subgroup will build tests and methods after we get feedback on our current approach for WCAG 3. " I have concerns related to scale here, as how many conditional language tests are envisioned, and who will author them? To get a sense of the scale concern, consider the following list:

• English – 1.452 billion speakers
• Mandarin Chinese – 1.118 billion speakers
• Hindi – 602 million speakers
• Spanish – 534 million speakers
• Arabic – 372 million speakers
• French – 300 million speakers
• Bengali – 265 million speakers
• Russian – 258 million speakers
• Portuguese – 223 million speakers
• Swahili (also known as Kiswahili) – 200 million speakers
• Indonesian – 200 million speakers
• Urdu – 170 million speakers
• Japanese – 150 million speakers
• German – 135 million speakers
• Punjabi – 125 million speakers
• Farsi (Persian) – 110 million speakers
• Javanese – 98 million speakers
• Italian – 85 million speakers
• Turkish – 85 million speakers

(source: https://lingua.edu/the-most-spoken-languages-in-the-world/)

2.3.1 Word Comprehension - Outcome: Ambiguous pronunciation:

I note references to COGA pattern 4.4.7 and SC 3.1.6 Pronunciation, however I do not see any reference to the emergent work coming from APA-Pronunciation TF.

2.3.2 Sentence structure - Outcome: Tense

Here again I have concerns related to Internationalization. Not all languages have tense: tenseless languages including Burmese, Dyirbal, most varieties of Chinese, Malay (including Indonesian), Thai, Yukatek (Mayan), Vietnamese and more. See also: https://en.wikipedia.org/wiki/Grammatical_tense#Possible_tenses

2.3.2 Sentence structure - Outcome: Number of ideas in a sentence; Outcome: Unnecessary words or phrases(*); Outcome: Double negatives

These proposed outcomes currently feel very Western-centric, and may not be applicable in all languages. At a minimum, a limiting 'conditional' (such as "in languages that support..." or similar) may be helpful here. (*)Who determines which words are "unnecessary"?

2.3.5 Numerical concepts - Outcome: Supplements to aid understanding of numerical concepts

This feels very ambiguous. When I review COGA pattern 4.4.6 it states "I need words rather than numbers and numerical concepts."

Referencing the list of common languages I provided earlier, how would this requirement be applied to that content? Would this meet the expected outcome?

• English – one point four five two billion speakers
• Mandarin Chinese – one point one one eight billion speakers (one point one hundred and eighteen??)
• Hindi – six hundred and two million speakers

For COGA requirements addressing Dyscalculia needs, will that suffice? My understanding of this condition is that it is more than just recognizing and processing actual numerals, but rather that at a higher order, "People who have dyscalculia struggle with numbers and math because their brains don't process math-related concepts like the brains of people without this disorder." (source: https://my.clevelandclinic.org/health/diseases/23949-dyscalculia#:~:text=What%20is%20dyscalculia%3F,of%20people%20without%20this%20disorder. Emphasis mine.)

Additionally, the current outcome states, "...such as weather forecasts that pair low temperatures with descriptions like “very cold.”..." which again is conditional: for residents of Carribean Islands, 5 Celsius would likely be considered "very cold", yet for residents of Northern countries (Norway, Sweden, Canada, etc.), 5 C is Cool, but hardly considered "very cold". I find this to be a very poor example.

[patrickhlauke]

I currently have some significant concerns related to Internationalization and the requirement for "word lists".

a very general +1 on this aspect

[rachaelbradley]

I think this area is one of the most difficult areas to work out but I also think there is value in continuing to work on it. Some of these outcomes may only be supported by assertions. Some, as is noted, will only apply under certain conditions. I think it is worth moving forward with these as exploratory in order to identify possible test and assertions.

My primary question at this stage is whether the grouped topics should each be their own guidelines rather than merged together into a single guideline:

• Word comprehension
• Sentence structure
• Implied meaning
• Chunking information
• Numerical concepts
• Typography

[mattgarrish]

1. I feel like it is nonfiction-specific.

I'd go even further and say it sounds great for things like manual writing, scientific papers, etc., where being clear and succinct is vital, but probably goes too far in restricting the voice and expression in many other common types of non-fiction, too. Do opinion writers for news outlets have to explain every idiom they use? Even saying that every paragraph has to start with a topic sentence assumes essay-like writing.

I want to like this, but I can't see how all of what can be published on the web can be expected to meet it. I'm also not sure it can be scoped to content where it would be most beneficial.

[jspellman]

I generally like the intent of this criterion, but I have two main concerns about its general direction right now:

1. **I feel like it is nonfiction-specific**. 
2. **I feel like some criteria overly restrict the author's voice**.

These are valid points, but they are assuming that the outcomes will be required like WCAG2 success criteria are. At this stage, it is too early to assume that will be the case. Our goal at this exploratory stage is to identify the user needs and outcomes that will address them.

Our writing process is to include these concerns as editor's notes so that people know that we are going to address these issues.

[mattgarrish]

It's also not clear to me what constitutes a "long" document (or video). When do you transition from short to medium to long? Is there going to be a word count, paragraph count, or some other objective means of deciding?

To use a news example again, could a publisher argue only long-form journalism is long and everything else is short?

[johnfoliot]

...but they are assuming that the outcomes will be required like WCAG2 success criteria are.

Is this saying then that some Outcomes will be optional? That's certainly how I read the response. (Question: where has that Working Group decision been documented?)

If that is the case, has there been any discussion/investigation around expected adoption rates? The sad reality is that for many large organizations, they will do what they MUST, but will often leave the SHOULDs and MAYs behind for a myriad of reasons (we have close to 15 years of experience with AAA Success Criteria in the 2.x family, and scant evidence of adoption at scale of ANY of those SC; I will suggest for that very reason - because they are NOT mandated.)

If indeed the intent is to offer "optional" Outcomes, I strongly urge the conscientious use of RFC 2119 MUST, SHOULD, MAY language in authoring the Outcomes, to reduce any ambiguity. [Ref: https://www.w3.org/TR/wcag-3.0/#normative-requirements]

Picking a random example, here are 3 variants of one of the proposed Outcomes:

• Outcome: Acronyms and abbreviations: Content authors MUST provide expanded versions of acronyms and abbreviations using one of the techniques in SC 3.1.4 Abbreviations. See COGA pattern 4.4.1.

• Outcome: Acronyms and abbreviations: Content authors SHOULD provide expanded versions of acronyms and abbreviations using one of the techniques in SC 3.1.4 Abbreviations. See COGA pattern 4.4.1.

• Outcome: Acronyms and abbreviations: Content authors MAY provide expanded versions of acronyms and abbreviations using one of the techniques in SC 3.1.4 Abbreviations. See COGA pattern 4.4.1.

Which one would it be?

[johnfoliot]

My primary question at this stage is whether the grouped topics should each be their own guidelines rather than merged together into a single guideline

Opinion: granularity is better for reducing ambiguity, so yes grouped topics should each be their own guidelines.

[u9000-Nine]

jspellman wrote:

These are valid points, but they are assuming that the outcomes will be required like WCAG2 success criteria are. At this stage, it is too early to assume that will be the case. Our goal at this exploratory stage is to identify the user needs and outcomes that will address them.

Yes, I am aware of that. My point in bringing this up is that user needs for fiction and nonfiction differ, and that right now, this only addresses nonfiction yet applies to both.

Our writing process is to include these concerns as editor's notes so that people know that we are going to address these issues.

I thought that changed with the move to GitHub?

[u9000-Nine]

johnfoliot wrote:

2.3.1 Word Comprehension - Outcome: Uncommon words:

I currently have some significant concerns related to Internationalization and the requirement for "word lists". [...]

What, (if anything) would the role of spell check dictionaries play here?

The introductory section also notes, "Internationalization: Several outcomes for Clear Language will need conditional tests by language. Our subgroup will build tests and methods after we get feedback on our current approach for WCAG 3. " I have concerns related to scale here, as how many conditional language tests are envisioned, and who will author them?

2.3.2 Sentence structure - Outcome: Number of ideas in a sentence; Outcome: Unnecessary words or phrases(*); Outcome: Double negatives

These proposed outcomes currently feel very Western-centric, and may not be applicable in all languages. At a minimum, a limiting 'conditional' (such as "in languages that support..." or similar) may be helpful here. (*)Who determines which words are "unnecessary"?

Might this be helped with something more generic which uses automatic tools or possibly professionals in each language? I'm thinking something along the lines of the analysis GNU's Style and Diction programs do. Requiring a thoughtful review rather than specific attributes would also cover other document types where user needs differ, such as fiction documents, scientific journal articles, and fields with standardized terminology.

2.3.5 Numerical concepts - Outcome: Supplements to aid understanding of numerical concepts

This feels very ambiguous. When I review COGA pattern 4.4.6 it states "I need words rather than numbers and numerical concepts." [...]

What about languages where specific numbers and broad ranges share the same word? For example, in toki pona, "mute" can mean "many", "more than two", "twenty", and "more than 19", and "ali" is both "100" and "all".

PS: GitHub apparently supports inputting <span lang="foo">, so we can accessibly markup other languages if needed.

[bruce-usab]

Opinion: granularity is better for reducing ambiguity, so yes grouped topics should each be their own guidelines.

I also favor that grouped topics should each be their own guidelines. My opinion is more from the practical perspective of breaking the work into smaller more manageable pieces.

Nothing is particularly hard if you divide it into small jobs.

If after some time, some guidelines end up being very duplicative, it is not much of a lift to re-combine them. Clear Language has been challenging, but maybe some of that is because it is such a large grouped topic?

[jspellman]

...but they are assuming that the outcomes will be required like WCAG2 success criteria are.

Is this saying then that some Outcomes will be optional? That's certainly how I read the response. (Question: where has that Working Group decision been documented?)

The Working Group has not decided how to do conformance. That's why we are working on Guidelines, so we can develop real examples to test and evaluate the various conformance models that have been proposed. I meant that people should not assume that outcomes are the same as success criteria. It's natural that people would assume that conformance would work the same way, but we should all be careful not to make assumptions.

[lauracarlson]

I currently have some significant concerns related to Internationalization and the requirement for "word lists".

a very general +1 on this aspect

[dbjorge]

General direction

• +1 that I favor keeping the granular outcomes
• +1 to the general feedback that many of these feel too specific to non-fiction/instructional content (especially 2.3.2 and 2.3.4)
• It feels confusing to me that all of these things labeled "outcomes" are all written in command tense - they are all describing an action authors must/should/may take, not the outcome of that action. This feels inconsistent both with the common definition of the word "outcome" and with the definition of Outcome provided by the first sentence of 3.3.1 ("Outcomes are verifiable statements")

[2.3.1] Outcome: Uncommon words

• I'm concerned that "words" isn't the right direction for the idea behind this requirement:
  • I don't think this should apply to most person/location names
  • I worry it may create confusion about whether things like acronyms, abbreviations, numbers, URLs, etc count as "words"
  • A literal reading suggests that definitions should be per-word, but per-term is often more appropriate for multi-word terms
• +1 to concerns about word lists. Others have already mentioned internationalizability and fiction vs non-fiction concerns. I agree with those, and also would be concerned about how to avoid word lists referenced by standards becoming out of date (if we end up with suggested "default" lists per language) and how to avoid creating a de-facto requirement that you MUST link an accessibility statement to meet the outcome (if we don't include a "default" list per language)

[2.3.2] Outcome: Voice

• This requirement concerns me because it has especially subtle/subjective internationalization impact. Different languages/cultures have different rules for whether active vs passive voice is used to indicate politeness, and to what degree; I'm concerned it would be untestably subjective to decide exactly when a rule like this should be enforced.

[2.3.4] Outcome: White spacing

• This outcome feels redundant to me; what exactly would this handle that wouldn't already be covered more specifically by either "Outcome: Paragraph length" or "2.3.6 Typography"?

[TzviyaSiegman]

I think that this is a good structure, but we have to be extremely cautious about language and personal taste/choice. As many have commented, there is a bit of Western and English slant. Another aspect that seems to have crept in is that so much of accessibility is about enabling the user to choose their own path. This is difficult to standardize, especially with the use of multiple languages.

[johnfoliot]

The Working Group has not decided how to do conformance. That's why we are working on Guidelines, so we can develop real examples to test and evaluate the various conformance models that have been proposed.

With respect, I didn't ask about Guidelines, Conformance, Methods, or Tests, I asked about Outcomes, which is what this PR is about (and titled: updates from Clear Language (outcomes only)).

Definitions taken from the current Draft:

• Guideline High-level, plain-language content used to organize outcomes.

• Outcome Result of practices that reduce or eliminate barriers that people with disabilities experience. See Outcomes.

  • 3.3.1 Outcomes Outcomes are verifiable statements that allow testers to reliably determine if the content being evaluated satisfies the user needs identified in the Guideline.

• Test Mechanism to evaluate implementation of a method.

• Method Detailed information, either technology-specific or technology-agnostic, on ways to meet the outcome as well as tests and scoring information.

...but we should all be careful not to make assumptions.

I am NOT making any assumptions, on the contrary I am seeking clarification about Outcomes (The result of practices that reduce or eliminate barriers that people with disabilities experience), to avoid any assumptions/presumptions.

You previously wrote: "Our goal at this exploratory stage is to identify the user needs and outcomes that will address them." However, you also wrote, "...they are assuming that the outcomes will be required like WCAG2 success criteria are."

So I am confused: logic would suggest that if (some) Outcomes are (potentially) NOT required, it would then seem that meeting some user needs would also be (potentially) NOT required, based upon your articulated goal of identifying user needs and outcomes.

(I will note as well that there is a difference between a NEED and a WANT (a.k.a. PREFERENCE); for example, when it comes to Acronyms and Abbreviations, I think we can safely say that all users would NEED the expansion of the Acronym or Abbreviation at least once per page/screen; some users however would likely PREFER the expansion NOT be on every instance of the Acronym or Abbreviation per page/screen - a nuanced distinction not currently noted in the draft language. This type of ambiguity is also applicable to Uncommon words (once or every time?) and Ambiguous pronunciation (once or every time?)).

Respectfully then, I will re-phrase and re-pose the question: is the current thinking today that some Outcomes will potentially NOT be required?

[rachaelbradley]

Summary 21 August 2023

1. There is general support for breaking this guideline apart for the exploratory phase. It can be recombined later.

2. There was a conversation around internationalization of outcomes under clear language. This topic applies to other areas such as text spacing. Chairs have started a Github discussion with COGA and Internationalization to begin to address this issue at Possible Internationalization approach #17

3. It is worth noting that many tests for these outcomes may be conditional or handled through assertions. Concerns about subjectivity and personal preference will need to be addressed regardless of the test or assertion used. Additional issues are captured by topic below.

Word comprehension

• Concerns about if and how “unnecessary words or phrases” would apply to fiction, scientific articles and other genres and styles. Some other areas may raise similar concerns.
• Concern about how to test and apply word lists.
• Note to include work from APA-Pronunciation Task Force.
• Suggestion to consider “terms” instead of “words” to better incorporate acronyms and multiword “terms”

Sentence structure

• Concern raised that the outcome on Voice would be particularly subjective

Implied meaning

• None at this time

Chunking information

• Questions about what is a “short” vs “long” document or video
• Concern that white spacing may be redundant

Numerical concepts

• Terms like warm or cool which may help clarify numerical concepts may vary by individual

Typography

• This should be included in the text appearance and semantics guideline but we can point to your work in this area

4. Other conversations on the thread related to conformance and will be discussed further after the exploratory guidelines are complete.

5. Another question was raised about outcome phrasing. This can be opened as an issue for discussion after we go through the guideline and outcome exercise and have more knowledge as a group.

[eadraffan]

2.3.1 Word Comprehension - Outcome: Uncommon words:

I agree with John Foliot's comments and others about the internationalisation concerns and word lists etc plus the work that will be needed to ensure statements / patterns etc are applied in a way that does not compromise what was said about fiction versus non-fiction

[rachaelbradley]

Notes from 22 August 2023 Meeting

Support for requesting Clear Language be split into separate Guidelines.

Additional notes from conversation

• Can we identify reference dictionaries as a way to make clear words more objective?
• Can we do this in a way that supports future advances in AI that could potentially generate plain language versions?
• Add issue about writing outcomes to support movement in support from author to AT or user agents as tech evolves.