NOTE ON THE RETENTION OF CONNECTED DISCOURSE

Two sentences were constructed having roughly the same content but different predictability ( P -) scores. One or other of the sentences was presented auditorily once only to different groups of subjects who were instructed to listen carefully, without however, being informed of the learning character of the experiment. The groups were then tested for retention after one of the following intervals: IM, 3 min., 2 hours, 20 hours. The sentences were found to differ in memorability (amount recalled) in IM but not in LTM. The reproductions were also analysed with regard to synonym and non-synonym substitutions and the implications for the chunking hypothesis discussed.