Using language to detect and change attachment style
Decades of research on attachment style has resulted in a plethora of assessment techniques and an imprecise taxonomy of styles. Beginning with an assumption that interpersonal connections are revealed in the words used to discuss those relationships, the current studies were designed to investigate the efficacy of a language-based approach to discriminating attachment styles. Because recent research has shown that small words such as pronouns are psychologically important (e.g., Pennebaker & Stone, in press), Study 1 explored the patterns of word use that best predicted attachment styles in a sample of Adult Attachment Interviews (AAI; George, Kaplan & Main, 1985/1996) of married couples (N=214). A multinomial logistic regression using theoretically-selected word categories resulted in correct categorizations of 82.3% of cases (secure 82.4%, dismissing 87.6%, and preoccupied 62.5%). Applying the resulting equations to the AAI resulted in correct classification of 50% of cases. Predicting to a different criterion, the Adult Attachment Scale (AAS; Collins & Read, 1990), provided a slightly more effective set of categorizations (58%). There were theoretically relevant differences among attachment styles in the use of pronouns and emotional words. Study 2 (N=201) was designed to apply the model derived in Study 1 to a sample of college students using the AAS as a criterion. In addition, Study 2 employed a writing intervention designed to change the attachment styles of insecurely attached students. Participants in the experimental condition wrote five essays about their family experiences across a semester in response to a prompt assembled from AAI questions. The Study 1 model remained a significant predictor of attachment style in Study 2, correctly classifying 51% at Time 1 and 61% at Time 5. Participants in the writing intervention condition did not change attachment styles as a function of condition, but there were some beneficial effects of condition on selected adjustment variables including coping and social integration. The inability of the writing intervention to influence attachment style change is discussed in the context of the social nature of attachment style.