During my PhD research↑ (.)
I transcribed sixTEen↑ hours of face-to-face interaction
for talk (.) errors (.) laughters=sniffs pauses (.)
facial expressions=gaze movements and voice qualia (.)
like PItch, fre:quency energy
(.) and their variations.
Mining human data is an extremely complex and painful exercise. However I guess one can eventually reach a decent level of empirical precision and also get to contemplate a number of interesting insights on ‘errors’ and their propagation.
More to follow…