orientations and IMWA

[Stuart Thiessen]

True. This is where you tell the computer that these are the usual  
assumptions. Then you program the exceptions into the system so the  
computer prepares to do the usual, but checks its "database" of  
exceptions to see if one applies before displaying the assumed  
symbol. That is why we do have to research how the system is used to  
understand where the assumptions are true and where the exceptions  
appear. That can help us know how to "train" the computer input  
process appropriately as Sandy suggests.

Thanks,

Stuart

On May 5, 2007, at 6:09, Charles Butler wrote:

> For orientations, sometimes one instance is not enough, you might  
> need three for front, back, and middle, then the rest are rotations  
> and mirrors, but when the thumb switches sides, or goes into the  
> middle pointing toward or away from the reader, there is only so  
> much that a computer can be "taught" to do, and it requires a human  
> eye to ensure that you actually have a properly oriented symbol  
> that is clear. I think we can do with 3 with the rest as rotations,  
> flops, and shadings.
>
> The old keyboarding system had a lot of positionings assumed, but  
> we have discovered with using the IMWA point and click that  
> "assumed" positioning is not fine motor enough for points of  
> contact.  We have a rather large corpus of signs now to start doing  
> some experiments with computerized assumptions, but we are still  
> learning.
>
> And once you add dance and ballet, one can assume nothing, as a  
> dancer can move the arms and legs into any number of positions that  
> simply cannot be assumed.
>
> Sandy Fleming <sandy at scotstext.org> wrote:
>
> Another example is with the number of different orientations  
> offered by
> the IMWA. As others discussed recently, we only need a fraction of  
> these
> - one instance of each shape is enough.
>
> Sandy (now wondering if this email sounds less calm than intended :)
>
>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/sw-l/attachments/20070505/5436aa5f/attachment.htm>