Based on structural differences between spoken and written language, the purpose of this paper was to investigate whether spoken and written communication imply a different representation in reporting an experienced dream. In fact, the clausal-dynamic quality of the former and nominal-synoptic quality of the latter, with the consequent differences in length, cohesion and density, could enhance/reduce the perceptual character and narrative structure of report features often considered in order to assess sleep mentation. In particular, we wondered whether, after eliminating all the elements responsible for the more conspicuous quantitative differences across the two forms (false starts, repetitions, digressions, redundancies), we would obtain two equivalent ways of reporting dreams. Three hundred and two (101 males and 201 females) subjects, raging in age from 18 to 40 years, participated as volunteers in the study and were asked to complete a dream diary daily for 14 days, by tape-recording their own dreams and putting them down in writing. The reports were analyzed by a psycholinguistic system. Results indicated that the report modality is able to affect the dream experience representation, conditioning the figurative (re)translation carried out by the tester: written forms show a loss of hallucinatory information and a non-complete correspondence also of the bizarreness features with respect to the spoken texts. Partialling out for length eliminated any difference between spoken and written dream reports on bizarreness; on the other hand, controlling the sequence effect of the report did not introduce changes. Methodological and theoretical implications were discussed.