AI Reveals How Psychology Terms Shift in Popular Media

A new analysis using artificial intelligence techniques has uncovered significant differences in how psychological concepts are understood in scientific research versus popular media. Researchers applied distributional semantics s to Russian-language corpora, revealing that terms like burnout and depression undergo semantic shifts when moving from academic journals to online platforms. This work demonstrates how computational tools can track changes in meaning across different communicative contexts, offering insights into the divergence between professional and public discourse on mental health topics.

The study found that scientific texts emphasize ological and clinical terminology, while popular science materials foreground everyday experience and therapeutic practice. For instance, in scientific discourse, the concept of burnout is linked to terms such as psychological resources, symptomatology, and diagnostic constructs, including words like fatigue, resilience, self-esteem, and clinical markers. In contrast, popular science discourse frames burnout through personal narratives, emotions, and everyday situations, with associations like work, person, feeling, and psychotherapy. Similarly, depression in scientific contexts is surrounded by medical and clinical concepts such as anxiety, distress, and PTSD, whereas in popular contexts, it is associated with more general terms like why, work, and feeling.

To conduct this analysis, the researchers compiled two corpora: a scientific corpus of approximately 300 research articles from Russian psychology journals, totaling 767,543 tokens, and a popular science corpus from online platforms, totaling 1,199,150 tokens. After preprocessing steps including OCR recognition, lemmatization, and removal of stop words, the corpora were analyzed through frequency analysis, clustering, and identification of semantic associations. Figures 1 and 2 show the top 500 most frequent words in each corpus, highlighting vocabulary differences, with scientific texts featuring terms like sample and scale, while popular texts include words like service and session.

, Detailed in figures 3 through 8, indicate clear semantic shifts. Clustering analysis showed that scientific texts are well-grouped, whereas popular science texts are more dispersed, possibly due to variations in text length and style. Semantic associations for key concepts were extracted, revealing that scientific discourse maintains precision with terms related to psychological characteristics and clinical markers, while popular discourse leans toward experiential and narrative elements. For example, the scientific definition of burnout aligns with components like emotional exhaustion and reduction in personal achievements, as reflected in associations, whereas popular contexts lack such specificity.

These matter because they highlight a gap between professional understanding and public perception of psychological concepts, which could impact how mental health issues are discussed and addressed in society. The shift from precise terminology to generalized meanings in popular media may lead to misunderstandings or oversimplifications of complex conditions. This study confirms the effectiveness of distributional semantics s for identifying such transformations, offering a tool for researchers to monitor language evolution in various fields.

However, the study has limitations, including its focus on Russian-language sources, which may not generalize to other languages or cultural contexts. The automatic extraction of named entities from texts, particularly from Telegram channels, resulted in the loss of some concepts, indicating potential issues with short or syntactically varied texts. Additionally, the analysis relies on predefined corpora, and future work could expand to include more diverse sources or longitudinal data to track semantic changes over time.