Inferring Depression and Its Semantic Underpinnings from Simple Lexical Choices.

Area of Science:

• Psycholinguistics
• Clinical Psychology
• Computational Linguistics

Background:

• Spatial demonstratives (e.g., 'this' vs. 'that') typically signal physical proximity.
• Emerging research suggests demonstrative choice also reflects psychological or emotional proximity.
• Depression is associated with heightened self-focus and maladaptive self-referential processing.

Purpose of the Study:

• To investigate if the Demonstrative Choice Task (DCT) can identify semantic characteristics predictive of depression.
• To explore the utility of DCT-based models for inferring depressive states and individual differences.

Main Methods:

• Two independent cross-sectional studies with 775 and 879 participants, respectively.
• Participants completed the DCT, and depression severity was measured using the Patient Health Questionnaire (PHQ-9).
• Classification models were built using DCT data to predict PHQ-9 scores.

Main Results:

• DCT-based classification models accurately captured semantic characteristics of experiential states linked to depression.
• These models outperformed baseline approaches in predicting self-reported depression symptom severity.
• Replicated previously observed semantic patterns of negative affect associated with depression.

Conclusions:

• The DCT paradigm effectively measures semantic features of psychological experiences relevant to depression.
• This task provides a novel, potentially simple method for assessing depression-related semantic characteristics.
• Findings suggest the DCT can map individuals within a semantic space, offering insights into depression heterogeneity.