, Volume 2, Issue 9, pp 437-442

Recopition memory for syntactic and semantic aspects of connected discourse


This study investigates the pattern of retention of syntactic and semantic information shortly after comprehension of connected discourse. Ninety-six Ss listened to 24 taped passages and, after each passage, heard one recognition test sentence which was either identical to a sentence that had occurred in the passage, or was changed in some slight way. The Ss responded “identical” or “changed,” rated their confidence, and classified changes as “meaning” or “form.” Two independent variables were manipulated: (1) The relationship between the original sentence in the passage and the test sentence. The test sentence was (a) semantically changed, (b) changed from active to passive voice or vice versa, (c) formally changed in other ways that did not affect the meaning, or (d) unchanged. Each sentence appeared in all change types. (2) The amount of interpolated material between the original and test sentences was zero, 80, or 160 syllables of connected discourse which was a continuation of the passage. Each S heard passages representing all levels of each variable. All combinations of particular passages, relationship of original and test sentence, and amount of interpolated material were tested.

When the test sentence was heard immediately after the original, retention was high for all test types. But after 80½160 syllables, recognition for syntactic changes had dropped to near chance levels while remaining high for semantic changes. Even when the meaning of a sentence was remembered, formal properties that were not necessary for that meaning were forgotten very quickly. The results suggest that the original form of the sentence is stored only for the short time necessary for comprehension to occur. When a semantic interpretation has been made, the meaning is stored. Thus the memory of the meaning is not dependent on memory of the original form of the sentence.

Based upon a dissertation submitted to the University of California at Berkeley in partial fulfillment of the requirements for the Ph.D. degree. The research was supported in part by a National Science Foundation Predoctoral Fellowship. The author is indebted to Dan l. Slobin for suggestions and criticism. Some of the data were presented at the April, 1066 meeting of the Eastern Psychological Association meeting in New York, New York.