In the previous section, we have investigated various aspects of automatic instruction. In the design of feedback, we have excluded the 3D-tongue movement using the OPG device in the current system because it is still technologically at the proto-type stage. Instead, contrastive visual and auditory feedback, and formant-tracking (spectrographic representation) feedback has been incorporated into the front-end modules of our CAPL system. It was already introduced in section 2.4 and section 3.3. But these kinds of information are never to meet the needs of learners: ``how to correct pronunciation problems''. Most learners require the feedback instruction in more detail, especially that which is linked to their articulatory movement. For this purpose, we propose our novel methods of self-corrective feedback on articulation.
