Is there a multi-channel extension to VXML?

Hi VXMLers,

I've been searching W3C and reading papers on multimodal UI looking for
ways to extend VXML 2.1 for multi-channel grammar and interpretation (e.g.,
by adding gesture channels).

Two questions for you:

1. Is there any candidate spec or informal work that extends vxml in this
way? I've read several papers on speech and gesture simultaneous
recognition, and all assume multi-channel grammars so that one channel can
disambiguate others. (The only W3C spec in this space I'm aware of is
SCXML, but it seems to require each channel to be fully disambiguated.)

2. Is there an official dtd or schema for the vxml 2.1 data element to
allow the ecmaxmltype attribute? It would be great to use the w3c validator
with such a resource (instead of creating my own).

Cheers,
David Pautler