PSYCHOLINGUISTICS AND LARGE LANGUAGE MODELS IN PROFILING: NEW HORIZONS IN BEHAVIORAL ANALYSIS
DOI:
https://doi.org/10.31108/3.2025.9.6Abstract
The aim of this study is to assess the role of psycholinguistics and large language models (LLMs), such as GPT-3, in profiling, specifically their ability to expand the possibilities for behavioral analysis based on speech patterns. Psycholinguistics is an effective tool for creating psychological profiles of individuals through language analysis. The integration of LLMs into this process significantly enhances the accuracy and speed of analyzing large volumes of text, uncovering emotional, cognitive, and social behavioral markers.
The research methods employed include a mixed approach combining both qualitative and quantitative speech analysis. The data sample consisted of 200 text fragments sourced from social media, emails, and interviews. For analysis, both LLMs (GPT-3) and traditional psycholinguistic methods (LIWC) were used. GPT-3 was applied for the automated analysis of speech patterns, and the results were compared with those obtained through LIWC. Statistical testing was also conducted to examine the differences in marker frequencies.
The results indicated that GPT-3 identified a higher frequency of emotional, cognitive, and social markers compared to LIWC, although these differences were not statistically significant (p > 0.05). LLMs demonstrate a significant advantage in processing speed, making them an effective tool for analyzing large amounts of data. Additionally, GPT-3 exhibited a higher level of accuracy in identifying emotional and cognitive markers, particularly in situations where traditional methods may not provide precise results due to limitations in contextual analysis.
The key findings of the study suggest that GPT-3 is a powerful tool for psycholinguistic analysis and profiling, capable of enhancing the accuracy of psychological profile creation in fields such as criminal investigations and human resource management. However, ethical considerations regarding data privacy and potential limitations in the accuracy of LLMs must be taken into account. Future research should focus on integrating LLMs with other analytical methods to develop more effective and ethical profiling solutions.
Keywords: psycholinguistics, large language models, GPT-3, profiling, behavior analysis, speech patterns, emotional markers, cognitive processes
Accepted: 20.10.2024
Reviewed: 28.11.2024
Published: 30.04.2025
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Shymko Vitalii Шимко Віталій
This work is licensed under a Creative Commons Attribution 4.0 International License.
Creative Commons Copyright Notice
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
