Experimental hypotheses are proposed which assert (1) that "thought" and "language" are distinct but both are part of linguistic performance; (2) that "thought" is central, and "language" is a symbolic system that one uses to refer in various ways to what he thinks; and (3) that immediate memory works with utterances and linguistic texts over a relatively short span while longer term memory functions in terms of "thought" or content. In various modes of presentation, the percentage of verbatim text recalled was compared with the percentage of content of "thought" recalled. Results show that once the short-term memory span is exceeded, the Subjects recall significantly less verbatim text that they do substantive content of the text, thus supporting hypothesis (3) and therefore (1) and (2) by inference. The Subjects remembered significantly more content than text regardless of whether they were presented with a prepared text, or were asked to create their own text for subsequent recall. For an earlier version of this paper see ED 031 704. (Author/FWB)