Figure 5.
A multi-stage parsing process. 1) As the target vowel is heard, it is identified, and its effects as well as those of speaker are partialed out of F1 and F2. 2) This can be used to make inferences about the upcoming consonant. 3) When the consonant is heard and identified, it's effects on F1 and F2 of the target vowel are partialed out. 4) This permits an accurate prediction of the context-vowel.