Edit comment LC-1192 for Accessibility Guidelines Working Group

Quick access to

Previous: LC-887 Next: LC-1562

Comment LC-1192: Comment Shortname:; Commenter: Al Gilman <Alfred.S.Gilman@IEEE.org>; Message-id:; Section of the document concerned:

or; Refinement of the section concerned:; Status:
assigned to; Type:; Resolution status:
Response drafted Resolution implemented Reply sent to commenter; Response status:
No response from Commenter yet Commenter approved disposition Commenter objected to disposition
Commenter's response (URI):; Comment:
This clause is unenforceably vague. NLP looks at the term in context and can always produce a vector of meanings weighted with probabilities. This language gives no standard as to how well one has to have isolated one meaning before considering "meaning can be determined without [further help e.g.] pronunciation." Are you OK if the right meaning is the maximum-likelihood estimate? If it is more than 80% likely? If no other meaning is as close as half as likely? What?

Proposed Change:

merge into 3.1.3, as the relevant mechanisms are interchangeable.; Related issues: 972
(space separated ids); WG Notes: [TEAMB] MU: For Japanese, single Han character can have multiple reading and screen readers cannot always read it correctly. Especially, Han character used in the name of person or place has unfamiliar reading. Also there are Han characters that the reading is difficult or unfamiliar to most users. That is why we need this SC. Though I think "reading" is better word than "pronunciation" for Japanese. This differs from "definition" issue in 3.1.3. Discussed in the 10 August 2006 telecon: resolution: Christophe to look at 1192 for resolutions and keep open http://www.w3.org/WAI/GL/2006/08/10-wai-wcag-minutes.htm Response from Christophe, Jan 17, 2007: Al Gilman proposes to merge SC 3.1.3 and SC 3.1.6 because the relevant mechanisms are interchangeable. However, SC 3.1.6 also covers words or phrases that are not necessarily used in an unusual or restricted way. Chinese has a number of common characters that have more than one pronunciation in Mandarin Chinese; sometimes the meaning is the same, but sometimes it is not. (I don't have examples right now but I can dig them up later; I started learning Mandarin Chinese in September.) Dutch also has homographs (words that have the same spelling but different pronunciations because they have different meanings); sometimes the only difference is words stress (examples at http://nl.wikipedia.org/wiki/Homograaf - in Dutch). Such words are not necessarily words "used in an unusual or restricted way", so that is an argument we could use against merging SC 3.1.3 and SC 3.1.6. However, Al Gilman also says that the current SC 3.1.6 is vague: natural language processing software can look at the context of a term and pick the most likely pronunciation. (I doubt that this is true for all languages in use on the Web.) In the 10 August 2006 survey (http://www.w3.org/2002/09/wbs/35422/teamb081006/results#xlc1192) I proposed a rewording that could address this vagueness: "A mechanism is available for identifying specific pronunciation of words where meaning *is ambiguous* without pronunciation." (i.e. replace "meaning cannot be determined" with "meaning is ambiguous"). This is consistent with the intent of the SC. I would remove the second paragraph (about maximum-likelihood estimates) of the current proposed response are replace it with a reference to the rewording if the WG accepts it. I think that it is also possible to flesh out the technique "Providing the pronunciation immediately following the word (future link)", but some of this is language dependent. For Japanese, I suppose one would use hiragana in a "furigana technique" (http://en.wikipedia.org/wiki/Hiragana http://en.wikipedia.org/wiki/Furigana). This partly overlaps with the current Ruby technique. For Chinese, one could use Pinyin (http://en.wikipedia.org/wiki/Pinyin). Pinyin is used for reading instruction in early school years in China. Discussed in the 01 February 2007 telecon: resolution: accept LC 1192 as amended (add "knowing the" to 3.1.6) http://www.w3.org/WAI/GL/2007/02/01-wai-wcag-minutes.html {not-accept} DONE Change SC 3.1.6 to: A mechanism is available for identifying specific pronunciation of words where meaning is ambiguous without knowing the pronunciation. Internal WD updated 22 March 2007 Response: http://lists.w3.org/Archives/Public/public-comments-wcag20/2007Jun/0108.html; Resolution: SC 3.1.3 and SC 3.1.6 apply to different sets of words. SC 3.1.3 applies only to words used in an unusual or restricted way, while SC 3.1.6 applies to any words where information about pronunciation is needed. We have revised SC 3.1.6 to read, "3.1.6 Pronunciation: A mechanism is available for identifying specific pronunciation of words where meaning is ambiguous without knowing the pronunciation." Chinese has a number of common characters that have more than one pronunciation in Mandarin Chinese; sometimes the meaning is the same, but sometimes it is not. Dutch also has homographs; sometimes the only difference is words stress (examples at http://nl.wikipedia.org/wiki/Homograaf). For Japanese, there may be no need to specify the meaning of a name written using Han characters, but there may be a need to provide the pronunciation. While the same mechanism may be used to satisfy both SC 3.1.3 and SC 3.1.6, weaker mechanisms may suffice. There are sufficient techniques for SC 3.1.6 that would not be sufficient for SC 3.1.3, e.g., providing the pronunciation immediately following the word.; (Please make sure the resolution is adapted for public consumption)