- From: Al Gilman <asgilman@access.digex.net>
- Date: Wed, 9 Dec 1998 14:42:53 -0500 (EST)
- To: gardner@physics.orst.edu (John Gardner)
- Cc: jongund@staff.uiuc.edu, nir@nirdagan.com, w3c-wai-gl@w3.org, raman@adobe.com, gardner@physics.orst.edu
The expression complexity of visual symbolic language for mathematics breaks the complexity management guidelines for spoken information. I see a lot of equations in print with too many productions in them to be spoken comprehensibly in one sentence. It's like one needs to sic a composition teacher on them and get the argument restated as a sequence of shorter sentences. Another way to state this is: "Here is another domain, like translating programs between programming languages, where a purely syntactic transformation may yield less-than-desirable results." The risk associated with expression complexity is greater in speech than in print. Loss of tracking kicks in at a lower expression complexity. Articulating some mathematical utterance as speech may work better if a whole article is compiled, and not just the text (including math symbol structures) transliterated. The shape of the compile process would be something like a) build a knowledge base from the whole article including symbolic exhibits and verbal voice-over b) segment it into feasible verbal paragraphs for readout from the knowledge base, c) code-generate the latter into fluent verbal language. Or tolerable. Al
Received on Wednesday, 9 December 1998 14:43:05 UTC