ABSTRACT

The experiment, therefore, tests the two assumptions jointly: that the memory for text is propositional, and that necessary nonexplicit textual information is inferred during reading. The material for the present experiment consisted of pairs of paragraph surface structures which were both derived from the same propositional base. The reading time results for short paragraphs did not differ significantly for the explicit and implicit versions, though the latter contained one less proposition or 14" less information than the explicit version. The memory level identified by Jarvella, R. J. may be much more like the acoustic short-term memory buffer that keeps reappearing in list-learning work than the memory for surface features that one were forced to postulate. Memory for surface features is, however, subject to strong interference and is quickly lost; hence, the superiority of explicit sentences is also lost and subjects must rely upon their propositional memory, which is, presumably, equal in the cases studied for explicit and implicit test sentences.