The Scoring System for Latent Structure (SSLS) was used to test three hypotheses regarding the degree of self-participation in dreams reported during sleep onset (SO) and rapid eye movement (REM) sleep. These hypotheses were that (a) the ratio of interactive to associative sentences would be significantly greater in REM than in SO; (b) the level of Ego activity would be significantly greater in REM than in SO; and (c) the ratio between interactive sentences with Ego not present and interactive sentences with Ego present would be significantly different in REM and in SO. None of these hypotheses was confirmed. However, the following significant differences were found: (a) the ratio between sentences with Ego substituted and the total number of interactive sentences was greater in REM than in SO (p less than 0.05); (b) the ratio between sentences with Ego substituted and sentences with Ego present was greater (p less than 0.05) in REM than in SO; (c) the ratio between sentences with Ego inserted and sentences with Ego substituted was greater in SO than in REM (p less than 0.01); and (d) the relative incidence of defective Ego was greater (p less than 0.01) in SO than in REM. These data were interpreted in terms of psychodynamic models of dream formation.
