Dissertations and Theses
Date of Award
2024
Document Type
Dissertation
Department
Psychology
First Advisor
Eric Fertuck
Second Advisor
Diana Diamond
Third Advisor
Erhard Mergenthaler
Keywords
differentiation-relatedness, computer text analysis, psychoanalytic/psychodynamic therapy, technology in psychotherapy research and training, language and psychotherapy, attachment
Abstract
Self-definition and relatedness are considered two important processes for psychological development. Self-definition refers to an individual’s ability to maintain a stable, agentic, integrated sense of self that is differentiated from others. Relatedness refers to one’s ability to establish and maintain mutually reciprocal, stable, interpersonal relationships. Research shows that impairments in self-definition and relatedness are connected with various forms of personality and psychiatric disorders. Assessing for these impairments may allow clinicians and researchers to identity relevant factors that lead to therapeutic change. One way to assess for self-definition and relatedness is by understanding an individual’s description of self and others, which may reveal experiences of autonomy, dependency, intimacy, as well as emotions and thoughts in relation to themselves and others. The Differentiation-Relatedness Scale (D-RS) assesses for the level of differentiation-relatedness (DR) in open-ended descriptions of self and significant others. Specifically, self- and object representations are rated along ten levels ranging from poor differentiation and lack of boundaries between self and other, to higher levels of differentiation and relatedness that represent an integrated and stable construction of self and others in reciprocally attuned relationships. However, applying the D-RS requires raters to undergo extensive reliability training that is a time-consuming and expensive commitment.
Our pilot study aim was to develop and validate a new computerized DR measure that will be a less resource intensive, efficient, and empirical method to assess for levels of DR. We used the Marker Approach to transform the manualized D-RS into a computerized DR measure (CDR). Previously employed in two computerized development studies, the Marker Approach identifies linguistic features or word markers that indicate a particular psychological state or capacity, in this case, DR speech located in self-other descriptions elicited in the Object Relations Inventory (ORI). This study used three archival ORI datasets scored for D-RS. ORI text samples are randomly allocated into the Development and Test Corpora. In the development phase, we aimed to identify unique word markers pertaining to high and low DR speech in the Development Corpus which then formed the High and Low CDR dictionaries used for validation analysis in the Test Corpus.
Results: The dictionary development analysis successfully yielded a final list of n = 34 unique word markers in the High CDR dictionary and n = 35 in the Low CDR dictionary. The subsequent validation results provided support for criterion validity of the High CDR dictionary but not the Low CDR dictionary. The High CDR dictionary consistently showed the strongest significance findings in its ability to correlate with the D-RS as well as distinguish between High and Low Test Corpora, and as such was selected as the new CDR measure. Results varied within clinical and non-clinical sub-samples in the Test Corpus. Specifically, the High CDR dictionary did not show significant findings and was unable to distinguish between high and low DR within the clinical sample. It was, however, able to produce similar though weaker ability to distinguish between high and low DR within the non-clinical sample.
We were also able to show partial support for construct validity through assessing for the relationships between the new CDR measure and existing computerized personality functioning measures of reflective functioning (CRF) and referential activity (CRA). We aimed to produce evidence that the CDR is able to capture a unique aspect of personality functioning specific to the DR construct thus adding proving its utility as a new measure. Correlations between the CDR, CRF, and CRA measures were higher than hypothesized, suggesting a stronger similarity shared among the three measures. However, the CDR was able to consistently show the strongest relationships with D-RS scores in comparison to the CRF and CRA measures, thus proving its added utility as a new measure that is able to detect the DR construct. Taken together, these findings provide preliminary criterion and construct validity of the new CDR measure.
Recommended Citation
Wong, Liang Yi, "The Development of a Computerized Text Analysis Measure of Differentiation Relatedness: A Pilot Study" (2024). CUNY Academic Works.
https://academicworks.cuny.edu/cc_etds_theses/1171