Adolescence has been identified as an important period for the development of self-concept [1]. Developing advanced cognitive abilities such as abstract thinking, enables young people to construct more complex representations of who they are, i.e. their ‘self-concept’[2, 3] and to hold complex mental images of themselves that include both positive and negative (e.g. “I am kind”; “I am ugly’” Hards, Fisk, Ellis & Reynolds, 2019). The development of the self-concept interacts with observable changes in mental health and well-being, including difficulties that frequently emerge during adolescence, e.g., major depression and social anxiety. For example, negative self-evaluation is one of the most frequently reported symptoms of depression amongst young people [4].

Depressive disorders, as well as sub-threshold symptoms of depression, are common during adolescence [5, 6]. A diagnosis of major depression during adolescence is associated with a range of immediate and long-term adverse consequences including an increased risk of suicidal thoughts and behaviours [7], difficulties in education and employment [8], and relationships [9]. The cognitive model of depression [10] suggests that pervasive and sustained negative beliefs about the self, the world and the future (i.e. the ‘cognitive triad’) increase vulnerability to depression and after the onset of depression maintain low mood. Evidence-based treatments for depression in young people are moderately successful [11, 12] but often fail to engage young people. These treatments are adapted from treatments developed for adults and thus may not integrate or tackle aspects depression that are characteristic of adolescents and which therefore are more salient and relevant to them, and more effective.

Negative self-evaluation is a symptom of major depression, a central feature of the cognitive model of depression and is highly characteristic of depression in young people. There is a well-established relationship between low self-esteem (i.e., a more negative self-evaluation) and depression, and low self-esteem predicts future depression symptoms [13, 14]. However, few treatments, including Cognitive Behaviour Therapy [15] explicitly address this symptom. Qualitative research with depressed adolescents suggests that they identify negative self-evaluation as a “big part of depression” and that it is not targeted enough in treatment [14]. Thus, based on feedback from young people and results of previous studies, it is important to better understand the nature of self-evaluation in adolescents and its relationship with depression. This understanding may then be used to adapt psychological models and treatments to the specific experiences and needs of adolescents who have depression.

Self-evaluation can be measured in different ways, focusing on different aspects of the construct. Self-esteem is typically measured using the well-established self-report Rosenberg Self Esteem scale [16]. The RSE is quick and standardised and provides a global assessment of self-esteem on a continuum from positive to negative. Self-evaluation can also be examined as an information processing bias. Self-referential processing refers to the process of how we use perceptions of our self, to guide the evaluation and interpretation of new information [17]. Thus ‘negative self-referential processing’ refers to the tendency to easily associate negative perceptions about oneself (e.g., “annoying”) and difficulties associating positive evaluations (e.g., “smart”; [18]. This is a well examined cognitive bias and studies have shown that compared to participants without a history of depression, formerly depressed patients show negatively biased self-referential processing when in a negative mood state [19, 20]. Similarly, studies have shown that following a negative mood induction, adolescents recall significantly more negative self-descriptions than after a neutral mood induction [21]. Thus this bias appears to be easily triggered (i.e. by a brief mood induction) and long lasting (i.e. in people with a history of depression). Individuals who have negative self-referent processing tend also to also have other information processing biases, including attention and memory biases [19,20,21]). This co-occurrence of different information processes biases may also predict the onset of future depression episodes and thus constitute a potentially modifiable vulnerability factor [23].

Studies examining negative self-referential processing typically use the Self Referential Encoding Task (SRET) task [22, 23]. In this task participants are presented with a series of adjectives and are asked to rate the words for whether they are self-descriptive. Next, they are given a surprise recall task and asked to recall as many of the adjectives as they can. Using this task, depressed adolescents showed a negative self-referential bias. However, they also endorsed and remembered positive pro-social self-referential words at the same rate as non-depressed adolescents, suggesting that important aspects of positive self-evaluation may remain intact during depression [24]. However, research has identified a specific limitation of using the SRET task with adolescents. A significant number of participants did not recognise or understand some of the words included in the SRET, i.e. ‘feeble’ and ‘pitiful’. This highlights the need to use stimuli that are designed for adolescents and reflect changes in how language is used over time. An alternative strategy to assess self-evaluation in adolescents would be to use adjectives that they themselves have generated to ensure these self-evaluations are familiar.

Anderson (1966) produced a set data norms for 555 descriptive words that have been used to study a range of psychological phenomena including gender stereotypes [25], human attribution processes (self/other judgmental tasks; [26] attachment styles and communication [27], depression [28,29,30] and self-evaluation in adolescents. For example, one study used trait-words selected from Anderson’s data norms to examine the development of self-evaluation and the impact of social comparison on the valence of self-evaluation across adolescents aged 9–25 years [31]. Other studies have also used trait adjectives from Anderson’s date norms to examine self-appraisal in young adults and healthy adolescents [32]. Anderson’s normative data is clearly flexible and widely used. However, as data were derived from adults and not from adolescents, there is a clear need to create an adolescent specific set of norms that is based on words generated by and used by adolescents rather than adults. Making this resource open-access is important as this could then be used alongside and compliment Anderson’s word-lists. One way in which self-evaluations could be collated is using the Twenty Statements Test (TST; 25), this measure invites participants to generate their own adjectives to describe themselves, in response to a standard prompt, “ I am …….”

Only one study has used an open-response measure to elicit adolescent self-evaluation and measure associations with depression [33]. In this study, the TST was used, however authors coded statements as either ‘positive’, ‘negative’ or ‘neutral.’ A single score was then derived to reflect the overall valence of self-evaluation. Thus, linguistic properties of adolescent self-evaluation which may reflect important nuances in respect to depression remain unexplored. Thus, it may be useful to apply computer linguistic analysis such as Linguistic Inquiry and Word Count (LIWC; [34] as this program codes words into pre-defined categories i.e., ‘positive emotion’ and ‘negative emotion’ and includes sub-categories such as ‘anxiety’ and ‘anger’. The application of LIWC in depression research has demonstrated potential in reflecting the biases typically characterised by depression such as in memory [35] and language [36]. Specifically, research has found that depressed adults tend to recall fewer positive emotion ‘words’ than healthy controls when asked to recall personal experiences (autobiographical memories; [35]. Other literature has also shown increased use of first-person singular pronouns in depressed participants compared to never-depressed controls [37]. Therefore, LIWC may be a useful tool to further explore the specifics of self-evaluation in respect to adolescent depression, given that different components of self-evaluation can be measured separately e.g., positive emotion, negative emotion, anxiety and anger.

The aim of the current study was two-fold; firstly, to examine the specific content of self-evaluation and present a bank of self-referential words generated by healthy adolescents and those with elevated symptoms of depression. This bank of data could be used alongside and complement Anderson’s existing data norms. Secondly, to use linguistic analysis (LIWC) to test the hypothesis that when describing themselves, adolescents with elevated symptoms of depression will use more negative emotion, anxiety, anger and sadness related words and fewer positive emotion words than healthy adolescents.