MONO91 Other Potentially Related Research chapter Mary Elizabeth Stevens National Bureau of Standards In terms of our present concern, however, we shall select only a few examples. "By automatic content analysis is meant the use of computer programs to detect or select content themes in a sentence-by-sentence scanning of text or verbal protocols". 1/ The interest of psychologists in machine techniques to assist in the analysis of linguistically- given materials, as in propaganda analysis, probably precedes at least in sophistication if not by ci[OCRerr]ate, that of documentalists or of machine specialists interested in library and information problems. 3/. The "General Inquirer" program developed by Stone et al, - is an example of question-answering techniques based upon selective extractions from natural language text. It involves the use of a master vocabulary consisting of words previously selected by an investigator as being likely to be content-indicative in a body of material to be processed, together with his pre-established indications of the categories he expects their occurrence should predict. It is to be noted that this is a custom-tailored set of categories and of clue-word lists associated with each, manually pre-established. Text is now processed in such way that each word is looked up and, if it appears in the master vocabulary, it is tagged with identifiers of the categories for which it is presumably predictive. A subsequent "Tag Tally" routine then counts the tag frequencies to deter- mine for which categories the input material has high or low scores, and these in turn can be compared with expected norms. This type of program has been applied to such varied materials as suicide notes, folk tales from different cultures, reports of field workers, recordings of group dis- cussions as in supervisory-leadership training sessions, and protocols for various psychological tests. 4/ Interesting variations developed by Jaffe and others 5/ involvc the use of non-verbal as well as verbal clues as content-indicators, specifically, time- sequence patterns recorded along with the words spoken in client-therapist sessions. At the meeting of the Association for Computational Linguistics and Machine Translation held in Denver, August, 1963, Jaffe reported findings indicative of positive correlation between the structure of temporal and lexical patterns in dialogue and suggested applica- tions to automatic abstracting or indexing by the use of the time-sequence patterns as clues to high information-value areas. 1/ 2/ 3/ 4/ 5/ [OCRerr]ord, Jr. b 1963 [498], p.3. See, for example, Jaffe 1952 [297], Hart and Bach, 1959 [256], POQ4, 1959 [475], the latter covering the proceedings of a conference held in 1955. Stone and Hunt, 1963 r576[OCRerr]J; Stone et al, 1962 [575j. See Ford, 1963 [498], p. 8. See for example, Cassotta, et al, 1964 L[OCRerr]o4J Jaffe, L2"43t0 L[OCRerr][OCRerr](j 137