Signed and spoken languages are expressed in two different modalities. The main question of the present work is to know whether this difference of modality leads to differences of structure. Particularly, we are interested in questions raised by a sign language phonology, and in the possibility to take account of the semantic dimension at a low level. I will show the inadequacy of the structural equivalences postulated in classical phonological studies on sign languages, particularly the " sign = word " equivalence. These theoretical problems have implications on the choices made in protocols of experimental studies. Finally, this premise of structural equivalence is maintained by transcription practices of sign language corpora at all levels of analysis, and these practices don't allow us to account for the meaning-form relation in sign languages.
