This method uses sound spectrograms as feedback. This technique is widely used for clinical purposes in the course of language acquisition by many researchers. The spectrographic representations of speech by the model native speaker and the learner are displayed together with the formant-tracking results (see section 2.2.4 for more details). By the visual examination of spectrograms, the learners can observe frequency, amplitude, and bandwidth of formants with particular attention to beginnings and ends of formant tracks and to formant spacing and translations. Formant tracks generally provide clues to the movement of speech organs from the preceding vowel position to the consonant position and from the consonant position to the following vowel. In fact, more detailed articulatory information is to be found in the adjacent vowels than in the consonants themselves[22]. However, it has two substantial disadvantages when applied to the automatic instruction. First, the learners with no knowledge of speech acoustics have difficulty in reading and interpreting the visualized formant tracks. Second, it is hard to correct articulation behavior directly from formant tracks, since there is often no detailed correspondence between them.
