The purposes of this study were to determine the nature of human semantic memory and to obtain knowledge usable in the future development of computer systems that can converse with people. The work was based on a computer model which is designed to comprehend English text, relating the text to information stored in a semantic data base that is structured like the human brain. The study compared the way information was organized within the computer model with the way humans retrieve information from their own memory structures. The results show that a network model is a viable representation of numerical as well as semantic information. The specific experimental results are summarized in the annotated bibliography in the second section of the document. (Author/TS)