Tuesday, June 5, 2012

Anchoring pronunciation in CALL assessment and classroom


Clip art: Clker
Clip art: Clker
One of the problems with machine-based assessment of pronunciation, particularly of vowel quality and vowel quality setting, is individual variability. Global assessment of pronunciation or intelligibility, such as that employed by the iBT TOEFL, is well established and reliable. In this 2011 doctoral dissertation "An anchoring technique was developed to account for speaker intrinsic pronunciation differences and to allow for meaningful comparisons of vowel pronunciation." That procedure basically solves the problem by creating an efficient model of the individual's vowel system that serves as a point of reference for assessment. The parallel in the (haptic-integrated) classroom is the pedagogical movement pattern, accompanied by articulation of a sound and the felt sense of body resonance. What is important to recognize is that although initially the learner may not have the articulation within acceptable range, the PMP and resonance, along with repeated practice and grounded listening (doing the PMP as you "replay" an example that you have heard, with our without repeating it orally)--and feedback in the classroom by instructor and peers--should do the job. And in the classroom, often just seeing the manner in which the PMP is done by a learner reveals the "problem" with the accuracy of the vowel, for instance. Correctly aligning the PMP (a very indirect and nonthreatening action in the classroom) often almost immediately produces a more approximate vowel articulation. So, "Anchor(ing)s a-weigh, eh!"

1 comment:

  1. Hat tip to R. Smith for pointing me to that CALL study! I'll follow up later with another post on how the EHIEP system can be integrated into CALL instruction as well. It was initially designed in 1992 with just that in mind, suggested by a teenage virtual game designer in Japan.

    ReplyDelete