|Clip art: Clker|
It calls into question the whole idea of how we respond to evolving interlanguage forms in teaching. If, for example, you simply model the correct response for a learner and then provide a "good" when something close is offered in return, according to the research, you may have just further confused the learner. If, on the other hand, you are able to demonstrate or repeat the error for the learner first and then go on to provide the model, clearly indicating the "distance" or difference, you may have accomplished more. (Many experienced pronunciation instructors do just that, in fact.)
In haptic pronunciation work, it is relatively easy to visually (and haptically) model off target rhythm, stress, intonation and some consonant problems. In initial instruction, if you have a homogenous group and have some basis in phonetics, it is a great idea to begin with a visual/haptic walk through of at least the vowel system of the L1. (Even better if you understand the basic intonation or pitch movement patterns of the L1 as well.)
Best case, do it haptically. Otherwise, at least watch your gratuitous "Goods!"