Previous work using the RSVP procedure has shown that the reproduction probability of "nuclear" elements of a sentence was higher than that of "secondary" or "modifiers" elements. In order to reduce the importance of the preparation of verbal response on this task a new procedure which combines the RSVP method of presentation with a RT probe measure was used in three experiments. The results obtained in these experiments suggest that the observed differences between nuclear and secondary elements are not due to a preparation of verbal response but a selection process which takes place during the perceptual integration of a sentence. Semantic and syntactic parameters play independent roles in the perceptual integration process. © 1982 North-Holland Publishing Company.