Re: Sonified Debugger vs. Screenreader Question
- From: "Andreas Stefik" <stefika@xxxxxxxxx>
- To: programmingblind@xxxxxxxxxxxxx
- Date: Wed, 21 Nov 2007 21:32:08 -0800
Sina asks:
Are you going to be doing a goms analysis
Andreas says:
Not exactly. I've built a custom compiler, debugger, etc, that
integrates speech based audio into the very heart of the compiler
itself. The key advantage, as you might imagine, is that the speech
audio can give hordes of information that visual studio can never come
close to duplicating, as even the add-in architecture can't connect
super deeply into the compiler (for technical reasons with the way
visual studio's compiler/debugger architecture works).
So, while GOMS analysis is largely, if I recall, keystroke level
empirically derived analysis about the time it would take to complete
tasks, our analysis is more about how effective the audio is for
helping you program. I think GOMS estimates are built on sighted folks
as well, so I have no idea how they apply to the blind community.
(With braille keyboards typically, correct?) Have you seen a paper on
this Sina?
In other words, we are not doing a GOMS analysis of non-sighted
programmer activities, but someone publishing this would be a super
valuable research project. It could be a great contribution to the
literature as well, as sighted folks like me have near zero data on
the keystrokes non-sighted programmers actually use in practice, let
alone how they compare to the traditional GOMS techniques for
analyzing keystrokes.
Sina said:
At first, when you mentioned video taping it, I immediately thought of a
cognitive JogThrough as introduced by Rowlands, D.E. and Rhodes, D.G.
Andreas said:
Oh, yes, that's definitely one way to do it. The video tapes in our
case are mostly so we can go back and figure out what, inevitably,
went wrong, AKA they are more qualitative analysis than anything super
formal. My phd advisor though has done some of that insanely time
consuming analysis before. Several of my friends/colleagues have
participated on those kind studies and man, you aren't kidding, it's
like several months of work for a single study, pretty nuts. For
whatever reason though, I've always gotten lucky and not had to do it.
They always have me writing some piece of a compiler instead.
We have worked on designing measures of how well people can comprehend
audio for programming, but those measures are also insanely time
consuming as well. No free lunch I guess. Our next study should be
pretty straightforward, though, no super fancy analysis. I think we're
just going to measure things like "time to complete task" and stuff
like that. Easy.
Hope that answers your questions!
Andreas
__________
View the list's information and change your settings at
//www.freelists.org/list/programmingblind
Other related posts:
