Word use of outpatients with a personality disorder and concurrent or previous major depressive disorder

Abstract

In a recent study, Rude, Gortner, and Pennebaker (2004) found word use to be related to depression and vulnerability to depression in the essays of college students. We sought to replicate and extend these findings in a clinical sample. Written essays of 304 psychiatric outpatients with a personality disorder and a mixed psychiatric profile on DSM-IV axis-I and 108 healthy controls were examined with word count software. Data on the tendency to be discrepant about the current self compared to a more ideal self were also gathered. We found that psychiatric outpatients in general used more words referring to the self and negative emotion words and fewer positive emotion words, compared to healthy controls. However, word-use proved unrelated to depression specifically. Actual-ideal self discrepancies were related to patient status and to current depression. Contrary to our hypothesis, these discrepancies did not correlate with the use of words referring to the self. We conclude that the negative content and self-focus of written essays and high levels of discrepancy reflect a negative thinking style that is common to a range of psychiatric disorders rather than being specific to depression.

Introduction

Depression is one of the most common forms of psychopathology, affecting one in four women and one in six men sometime during their lives (Kessler et al., 2005). Due to its nature and tendency to be recurrent, depression has significant impact on many personal, social, and health-care domains (Burcusa & Iacono, 2007). Major clinical and research efforts have sought to understand the features associated with depression. One way to understand depression is provided by cognitive theory, originally proposed by Aaron Beck (1967). This theory postulates that maladaptive cognitive patterns underlie depressive symptoms, particularly in interaction with stressors. The theory includes as a core hypothesis that depressed and depression prone individuals have negative thinking styles, driven by so called “depressogenic schemata” (Beck, 1987). Beck defined schemata as stored bodies of knowledge that are enduring, automatic and characteristic for a person's cognitive organization. In addition, Beck hypothesized that depressogenic schemata do not affect mood until activated by stress. This pattern, which is often referred to as the vulnerability component in Beck's theory, has indeed been observed in many studies (Burcusa and Iacono, 2007, Miranda et al., 1990), although not in all (e.g. Coyne, 1992). Moreover, it is as yet not fully understood whether depressogenic schemata play a causal role in the etiology of depression, or whether they are merely correlates (Monroe, Slavich, Torres, & Gotlib, 2007). Another debate pertains to the question of whether depressogenic schemata are truly specific to mood disorders. The ‘cognitive content-specificity hypothesis’ predicts that each diagnosis on Axis-I and Axis-II is associated with a predominant thinking style (Beck, 1987). This hypothesis, however, is questioned by for instance Lamberton and Oei (2008), who conclude that depressogenic schemata reflect a common factor in a range of psychiatric disorders.

A second cognitive theory on depression speculates that, in addition to negatively toned automatic thoughts, depressed individuals have an enduring tendency to focus on the ‘self’ (Pyczsinski & Greenberg, 1987). Self-focus is proposed to function as a coping mechanism by which an individual tries to reduce discrepancies between the ‘real-self’ and a more ‘ideal-self’ (Robinson & Alloy, 2003). However, when attempts to reduce negative discrepancies fail, an increased self-focus magnifies negative emotions and constitutes a disposition to depression (Higgins, 1987, Pyczsinski and Greenberg, 1987).

Of particular relevance for these perspectives are recent findings by Rude et al. (2004). They conceptualized cognitive distortions in terms of word use, and offer the only documented data to date on differences in writing styles between depressed, formerly depressed and never depressed individuals. These authors asked their participants (currently, formerly, and never depressed students) to write an essay about going to college. Essays then were subjected to linguistic analysis using the Linguistic Inquiry and Word Count software (LIWC) (Pennebaker, Francis, & Booth, 2001). Rude and colleagues predicted that depressed students would use words of negative valence and the word ‘I’ more often in their essays as compared to formerly and never depressed students. In addition, they expected that previously depressed students would use the word ‘I’ increasingly while progressing through their essay. Both predictions were confirmed. The finding that depressed students used words of negative valence and the word ‘I’ more often, as compared to previously and never depressed students was interpreted as evidence for the existence of negative schemata and an enhanced self-focus in depressed students. The finding that previously depressed students used the word ‘I’ increasingly was interpreted as activation of latent depressogenic schemata while writing a self-relevant essay.

Notwithstanding these remarkable results, the study by Rude et al. (2004) had several limitations. Firstly, they used a sample of relatively healthy, young (mean age = 18.3 years), and well educated subjects. Secondly, they neglected the critical factor discrepancy between ideal-self and actual-self that theoretically is required to make self-focus maladaptive and instead only focused on use of the word ‘I’. Thirdly, only depressive symptoms were assessed, hence predictions regarding the specificity of the depressive thinking style could not be tested. We therefore aimed: (1) to replicate the Rude et al. (2004) results by contrasting clinical and nonclinical samples, (2) to elucidate the self-focus hypothesis further by including a validated current-ideal self discrepancy measure, and (3) to assess whether manifestations of negatively toned self-schemata are specific to depression. We expected depressogenic schemata and an enhanced self-focus to be evident throughout the essays of depressed subjects and to be gradually activated in the essays of formerly depressed patients, relative to never depressed psychiatric patients and healthy controls. To test this, we divided our patients, all of whom had at least one personality disorder (PD), into patients who (1) were currently depressed, (2) had a history of depression, and (3) had never been depressed. This division of patients was based on the current or lifetime presence of a depressive disorder and was made regardless of the presence of an additional anxiety disorder. In extending the Rude et al. findings, we expected use of the word ‘I’ to be positively related to self-ideal discrepancies and depression severity. This expectation followed from the observation by Rude et al. (2004) that depressed individual use the word ‘I’ more often while writing an essay and from the work by Higgins (1987) and Pyczsinski & Greenberg (1987). Pyczsinski & Greenberg (1987) report that the dispositional tendency to be highly self-focused is positively correlated to depression; Higgins (1987) proposed that self-ideal discrepancies signify the absence of positive outcomes and the presence of negative outcomes, which is related to dissatisfaction, sadness, and depression.

To further test the cognitive content-specificity hypothesis, we changed the way the sample was divided. We divided the sample of individuals with a PD into those with (a) a comorbid current and/or previous mood disorder, (b) a comorbid anxiety disorder, and (c) no additional current or lifetime diagnosis of depressive or anxiety disorder. The rationale for combining the previously and the currently depressed group was that both groups are thought to endorse depressogenic schemata and in this regard thus form a homogenous group (see for example Hamilton & Abramson, 1983). We expected that the word use of the currently depressed group would reflect the presence of depressogenic schemata to a higher degree, relative to the other two groups.

Computerized word counts (such as the LIWC) are sometimes considered to be of limited psychological relevance because this software does not consider context and therefore is unable to grasp a communicated message (Alpers et al., 2005). We recognized this to be a potential flaw. In order to gain concurrent validity for the LIWC as a method to study psychological states and for our study, we attempted a comparison of LIWC performance with human ratings. These ratings were compared with LIWC dimensions equivalent to the features on which the essays were judged by 2 independent humans.