Content Details | NUML Online Research Repository

Content Details

Back to Department Listing

Title TRACING CONCEPTUAL METAPHORS, CONSTRUAL OPERATIONS, AND FRAME SEMANTICS IN CHATGPT AND HUMAN LANGUAGE: A COGNITIVE LINGUISTICS PERSPECTIVE

Author(s) Muhammad Naeem

Abstract This study investigated the cognitive mechanisms that shape the linguistic output of ChatGPT and human participants. Using the frameworks of Conceptual Metaphor Theory, Construal Operations, and Frame Semantics, the research explores whether and how AI-generated text reflects the cognitive strategies typically employed by human writers. This study adopted a mixed-methods approach. The researcher collected 100 essays written in response to CSS examination prompts, 50 composed by human participants and 50 by ChatGPT. The data underwent both statistical and qualitative analysis, using NLP tools, Corpus-Based software and cognitive semantic frameworks. The analysis revealed that both ChatGPT and human participants exhibit metaphorical reasoning and conceptual structuring, but they differ in the depth, coherence, and cultural grounding of their expressions. ChatGPT showcased an impressive ability to replicate human-like metaphors and frames but lacked the experiential and emotional anchoring which was evident in human responses. The findings indicated that while ChatGPT can produce syntactically better and metaphorically rich language, its expression lacks the embodied experience, cultural nuance, and intentional construal found in human linguistic expression. The study concluded that ChatGPT approximates human-like conceptual structures through probabilistic modeling rather than through genuine cognitive understanding. These insights contribute to ongoing debates in cognitive linguistics, artificial intelligence, and computational language modeling, and offer insightsfor both theoretical as well as practical inquiry regarding AI and language.

Type Thesis/Dissertation

Faculty Languages

Department English

Language English

Publication Date 2025-12-15

Subject

Publisher

Contributor(s)

Format

Identifier

Source

Relation

Coverage

Rights