The Measurement of Meaning in Psychoanalysis by Computer Analysis of Verbal Contexts

IN THIS REPORT I DESCRIBE ONE OF A SERIES of quantitative studies of a single psychoanalysis. My purpose is to demonstrate systematic methods for quantitatively and objectively selecting, reducing, and describing the data in transcripts of audio recordings. My intent throughout these searches for rigorous and relevant measures of meaning has been to reduce to manageable proportions the information contained in approximately one and a half million words spoken in 363 hours of the analysis, to eliminate large numbers of words which were not of immediate interest, and to highlight those which were of interest—in short, to reduce "noise" (error) and amplify "signals."

In the first study (Dahl, 1972), a correlation matrix of 53 variables that occurred in 363 psychoanalytic hours was factor-analyzed. Six major factors were isolated, and estimates of the amount of each factor (factor scores) for each hour were obtained and plotted. These time-plots corresponded strikingly to major events in the analysis. The factor scores were then further combined into two different measures: one of "analytic work" and one of "resistance to analytic work." It was then possible to get an over-all measure of the amount of "analytic work" in each hour. It was also possible to define by an internal criterion (one that emerged out of the data, rather than one imposed by outside