Text generation and recall by high-knowledge and low-knowledge individuals

Abstract

Text generation and recall within a particular subject matter domain (baseball) was studied in relation to domain knowledge. A problem-solving model for text generation and comprehension was developed. The model assumes a PLAN and two major PLAN components, a PROBLEM REPRESENTATION and a CONTENT REPRESENTATION. In Experiment 1, each high-knowledge (HK) and low-knowledge (LK) individual generated a domain-related text and recalled its contents 2 weeks later. Results indicated: (a) HK individuals generate text that is more detailed in CONTENT REPRESENTATION than do LK individuals; (b) HK recall was superior on measures related to the sequential “flow” of actions and changes in game states. This finding was attributed to the more extensive generation of detailed information by HK individuals, with such information making the problem solution path more discriminable. In Experiment 2, HK recall was shown to be superior to LK recall for HK-generated text while HK recall was superior to LK recall for LK-generated text on relatively few measures. The findings support the idea that domain-related knowledge has a similar influence upon the processes of text generation and comprehension, that such processes may be viewed in terms of a problem-solving framework, and that HK—LK performance differences are due to differences in the respective problem spaces and to differences in the ability to monitor the selected solution path.