The target theory of vowel production suggests that vowels are characterized by formant values called targets. It is assumed that these targets are fixed with respect to speaking rate and stress pattern. In the work described here we attempted to investigate this issue. We dynamically modelled formant time series to estimate unreached targets; then we compared their variability with the formant frequency variability for the [a] in the context [iai] for different speaking rate and stress pattern pairs. Below a certain speaking rate, our results were consistent with the hypothesis that the [a] is characterized in the (Fi,F2) plane by a fixed target with respect to speaking rate and stress pattern.