- From: Al Gilman <asgilman@access.digex.net>
- Date: Mon, 2 Feb 1998 13:21:45 -0500 (EST)
- To: tdowling@ohiolink.edu (Thomas Dowling)
- Cc: w3c-wai-ig@w3.org
Floodgate Warning. to follow up on what Thomas Dowling said: > Being new to the list, I don't want to cover ground that has > been gone over before, but thought I'd chip in another > perspective on abbreviations and acronyms. AG:: Welcome, indeed. You are filling a gap in our knowledge base. > In the library world, there's a standard reference work called the > _Acronyms, Abbreviations, and Initialisms Dictionary_. As I recall its > introductory materials, it sets out definitions in which acronyms and > initialisms are both subsets of the larger body of abbreviations; as has > been pointed out, acronyms are pronounced as words (e.g. NASA and UNICEF) > and are created from the [usually] first letters of the words in a phrase. > Initialism is the term given to first-letter abbreviations that are > pronouned letter by letter (IBM and FBI). AG:: Great. The good news is that there already is in standard library parlance a semantic model that meets the needs of the screen reader. The bad news is that it is jargon, not English. So it doesn't meet the needs of the author community. The distinction between acronyms and initialisms is, in linguistic terms, "not controlled by a preponderance of the speakers of English." In this business, we need to take such a populist standard for what English is. My vocabulary is somewhere in the upper tail of English speakers and I don't recall encountering it before. It's not in my desktop dictionary. So this terminology which is commonplace among librarians is not yet a solution to "how do we get authors to distinguish in the way needed to guide pronunciation?" In a companion post I suggested it is not totally a failure if the browser or screen reader has to guess. It is even better if the spell-checker does some guessing regarding when to challenge the author for an expansion or pronunciation for an unfamilar word. For the acronym RAM, if you don't mark it as an acronym, the people who suffer are the people who don't know anything about computers, not the people using text-to-speech. Because if the screen reader ignores the acronym status of this term and says "ram" the listener will hear the right message. If we tie markup to "failure to reach audience X" we can expect authors to be responsive. We can use prognostic guessing as a way to limit the burden on authors. If the spell checker includes rules for which abbreviations are likely to cause pronunciation problems, then the author will not be asked to mark all acronyms with markup. If we ask the author to manually add markup for all acronyms used, we will get outright rebellion and we won't be able to do much more than be remembered as martyrs. In terms of how much we bug the author, we should be savvy. The author has a notion of his/her intended audience. Start by getting the things marked up that the author needs to reach the audience that he/she already believes in. If the audience is a general audience and not just computer geeks, then the RAM acronym may lead to confusion and not just 'title="Random Access Memory"' but also a link to the RAM entry in "A Child's Garden of Computers" is appropriate. The spell-check dictionary is the place to distinguish between pronounceable acronyms and initialisms that should be spelled or expanded to words in speaking. Find a market where this function is backed by demand with money. The TelePromTer market which does prompting texts for high-profile speakers in politics and the media is where one might expect this distinction to be a requirement backed with money. How do they handle this. Do they have an existing markup that they ask word-processing tools to provide, or do we have the capability to solve a problem for them with HTML? In summary: The distinction between acronyms and initialisms begs the question of how do we get authors to notice and do something effective. We should ask authors for explicit markup only in the case where the prevailing text to speech methods fail. This means putting a somewhat cautious algorithm in the author tool based on algorithms that parallel those used in screen readers. Integration with spell-checking is a promising way to get this check actually used. Al Gilman
Received on Monday, 2 February 1998 13:22:20 UTC