We present a method for the analysis of motor equivalence on two French corpuses, evidenced by articulatory modelling. This processing enables us to make out the individual actions of each degree of freedom of the vocal tract. We intend to define the phonetic types we are studying with a combination of degrees of freedom recruited. Such a characterisation makes it possible to account for coarticulatory processes in a coherent way. This method brings to light compensatory strategies, striking strategies of preservation of vocalic configurations in some consonants by one speaker only, and a large divergence in the strategies used by the two speakers.