Augmented analysis: Using large language model tools for linguistic research

About the authors:

Natalia N. Belozerova, Dr. Sci. (Philol.), Professor, Professor of the Department of Applied and Theoretical Linguistics, Institute of Social Sciences and Humanities, University of Tyumen, Tyumen, Russia; n.n.belozerova@utmn.ru https://orcid.org/0000-0003-0892-7380

Anna V. Gabrielyan, Master Student (Applied Linguistics), Department of Applied and Theoretical Linguistics, Institute of Social Sciences and Humanities, University of Tyumen, Tyumen, Russia

stud0000220942@study.utmn.ru, https://orcid.org/0009-0000-1604-9720

Mikhail A. Voronkov, Master Student (Applied Linguistics), Department of Applied and Theoretical Linguistics, Institute of Social Sciences and Humanities, University of Tyumen, Tyumen, Russia

stud0000263093@study.utmn.ru, https://orcid.org/0009-0004-1523-2135

Polina A. Zimens, Master Student (Applied Linguistics), Department of Applied and Theoretical Linguistics, Institute of Social Sciences and Humanities, University of Tyumen, Tyumen, Russia

stud0000262142@study.utmn.ru, https://orcid.org/0009-0006-5573-8450

Abstract:

The article analyzes preliminary results of testing “intra-model” generative artificial intelligence tools for the primary processing of linguistic and discourse units during the “Language Models” course for first-year students of the Master’s program in Applied Linguistics. The testing utilized both classic NLP tools, introduced into practice in the 1950s–1970s, and transformer tools such as self-attention and multi-head attention, as well as rule-based reasoning model tools (knowledge graph, chain-of-thought). During testing, the names of these tools were entered into the query text (prompt). Such prompts were used to validate classic explanatory linguistic models, ranging from sign models to Jacques Derrida’s deconstruction, in the analysis of a range of heterogeneous texts. The analysis was conducted in two modes: instrumental and expert, both sequentially and in an integrated manner, depending on the students’ analytical preferences. The following language models were used when working with NLP, transformer, and reasoning tools: DeepSeek R1, Perplexity AI, ChatGPT 4o, GigaChat, and Mistral AI3. Analysis of the results revealed that the “reasoning” model DeepSeek R1, which incorporates chain-of-thought and knowledge graph mechanisms, proved to be the most comprehensive. All analytical procedures undertaken using the in-model tools of generative language models essentially represent combined human verbal text processing. The model results, as well as the authors’ own expert analysis using the parameters of classical explanatory linguistic models, created a basis for subsequent research into linguistic and discourse units with a minimal percentage of subjectivization of the results.

References:

Belozerova, N. N. (2010). The Real World and the Virtual World: Two Ecological Systems? Publishing house of Tyumen State University. [In Russian]

Gadame, H. (1988). Truth and Method: Fundamentals of Philosophical Hermeneutics. Progress. [In Russian]

Heidegger, M. (2011). Time and being. Metaphysics, (1), 205–224. Retrieved July 7, 2026, from https://www.intelros.ru/readroom/metafizika/metf-1-2011/13450-vremya-i-bytie.html [In Russian]

Alfino, M. (1991). Another look at the Derrida-Searle debate. Philosophy & Rhetoric, 24(2), 143–152. https://philpapers.org/rec/ALFALA

Austin, J. L. (1975). How to Do Things with Words. Harvard University Press. ISBN 0-674-41152-8.

Bass, A. (2020). “Signature Event Context”: Margins of philosophy (Derrida 1982 [first published 1972]). In Jacques Derrida: Basic Writings (pр. 105–134). Routledge.

Bornedal, P. (2020). Deconstructive vs pragmatic: A critique of the Derrida–Searle debate. The European Legacy, 25(1), 62–81. https://www.tandfonline.com/action/showCitFormats?doi=10.1080/10848770.2019.1652039.

Chandler, D. (1994). Semiotics for Beginners. Retrieved from July 13, 2026, from https://share.google/Md73S8LEmRIkvrPeB 

De Beaugrande, R.-A., & Dressler, W. U. (1981). Introduction to Text Linguistics. Longman. Vol. 1. ISBN 0582554861.

Derrida, J. (1982). Signature event context. In Margins of Philosophy (translated by A. Bass) (pр. 307–330). University of Chicago Press. ISBN 0226143252.

Farrell, F. B. (1988). Iterability and meaning: The Searle-Derrida debate. Metaphilosophy, 19(1), 53–64. https://doi.org/10.1111/j.1467-9973.1988.tb00701.x

Greimas, A. J. (1976). Pour une theorie de modalite. Langages, Année 43, 90–101. Retrieved from July 13, 2026, from https://www.persee.fr/doc/lgge_0458-726x_1976_num_10_43_2322 [In French]

Kenaan, H. (2002). Language, philosophy and the risk of failure: Rereading the debate between Searle and Derrida. Continental Philosophy Review, 35(2), 117–133. https://doi.org/10.1023/A:1016583115826.

Moati, R. (2014). Derrida/Searle: Deconstruction and Ordinary Language. Columbia University Press. http://hdl.handle.net/10722/255135.

Raffel, S. (2011). Understanding each other: The case of the Derrida–Searle debate. Human Studies, 34(3), 277–292. https://doi.org/10.1007/s10746-011-9189-6.

Searle, J. R. (1969). Speech Acts. Cambridge University Press.

Searle, J. (1977). Reiterating the differences: A reply to Jacques Derrida. Glyph, (1), 198–208. https://www.academia.edu/15932757/Reiterating_the_Differences_A_Reply_to_Derrida_by_John_R_Searle

Shakespeare, W. Sonnets. Sonnet 144. In Shakespeare: Collection of Works (p. 15554). (Shakespeare-Riverside, p. 1775).

Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30. https://doi.org/10.48550/arXiv.1706.03762.