A Stylistic Analysis of AI-Generated Short Stories: Exploring Lexical, Grammatical, and Figurative Features Using Leech and Short's Framework

Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1211 Liberal Journal of Language & Literature Review T Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 https://llrjournal.com/index.php/11 A Stylistic Analysis of AI-Generated Short Stories: Exploring Lexical, Grammatical, and Figurative Features Using Leech and Short’s Framework Muhammad Fahad MPhil Scholar, Hamdard University Karachi Email: shahFahad[email protected] Dr. Kamran Ali Associate Professor, Hamdard University, Karachi Email: dr[email protected] Farah Naz MPhil Scholar, Hamdard University Karachi Email: farahnaz[email protected] Vol. 3 No. 4 (2025) Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1212 The advent of artificial intelligence (AI) in creative writing has raised important questions about its ability to generate artistic and linguistically sophisticated content and a critical gap remains in understanding the stylistic qualities of AI-generated fiction, particularly short stories. This study aims to fill that gap by providing a systematic stylistic analysis of AI-generated short stories using the framework developed by Leech and Short. The research seeks to explore the lexical, grammatical, and figurative language features of AI-generated narratives, with particular emphasis on four key stylistic categories: Lexical Categories, Grammatical Categories, Figures of Speech, and Cohesion and Context. Through the analysis of a corpus of AIgenerated short stories, this study will evaluate whether AI exhibits a consistent and sophisticated storytelling style or if its output remains formulaic and generic. The findings of this research are expected to offer valuable insights into the potential and limitations of AI as a creative partner in literary production and contribute to a deeper understanding of its linguistic capabilities. Keywords AI-Generated Text, Stylistic Analysis, Short Stories, Lexical Features, Grammatical Features, Figures Of Speech, Narrative Techniques. Introduction: AI-generated text, especially in the form of short stories, has garnered increasing attention for its potential to simulate human-like creativity and narrative skills. From a broader worldview, the intersection of technology and art has always been a subject of both fascination and concern. The question of whether machines can replicate or even surpass human creativity has been a topic of philosophical debate for decades (Turing, 1950). In recent years, however, the conversation has evolved from theoretical discussions to practical applications, as AI systems are now capable of producing text that closely mimics human authorship. Large language models, such as OpenAI's GPT series, have demonstrated the ability to generate coherent and contextually relevant prose, including stories, articles, and essays (Radford et al., 2019). However, despite their impressive technical achievements, AI-generated texts raise important questions about the nature of creativity and authorship. While previous research has focused extensively on the technical capabilities of LLMs, such as their ability to predict and generate humanlike text, there is a conspicuous lack of studies exploring the stylistic and linguistic characteristics of AI-generated fiction (Zaman, Abbasi, & Chandio, 2025). Many AIgenerated texts, including short stories, have been critiqued for their apparent lack of originality and depth, suggesting that they may be more formulaic than creative. Yet, the extent to which AI-generated short stories exhibit a consistent stylistic approach remains underexplored, highlighting the need for a more systematic analysis of these texts. In the context of stylistic analysis, Leech and Short's (1981) framework provides a comprehensive model for examining the linguistic features that define literary texts. Their approach categorizes stylistic elements into four main areas: Lexical Categories, Grammatical Categories, Figures of Speech, and Cohesion and Context. These Abstract Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1213 categories are integral to understanding how language functions within literature to create meaning, mood, and narrative structure. By applying this framework to AIgenerated short stories, this study aims to uncover the specific linguistic patterns that characterize AI’s storytelling style and explore whether AI's output possesses the necessary features of artistic expression or merely reflects a regurgitation of patterns from its training data. The central problem addressed by this research lies in the gap between the technical capabilities of AI and its artistic potential. While much of the existing literature has analyzed AI’s ability to mimic human language in general, there is a notable lack of attention paid to its application in the realm of creative writing (McCormick, 2020). This study, therefore, focuses on a stylistic analysis of AI-generated short stories to answer critical questions about the consistency, sophistication, and originality of AI’s narrative style. By systematically categorizing and analyzing lexical, grammatical, and figurative elements, the research seeks to determine whether AI can produce creative texts that meet the standards of literary style or if its output remains limited by the nature of its training data. Theoretical Framework The theoretical framework adopted for this study is based on Leech and Short’s (1981) Style in Fiction, which offers a robust model for analyzing the stylistic features of literary texts. This framework is particularly suited to the task of examining AIgenerated short stories, as it categorizes linguistic elements into four key areas that are essential for understanding narrative style: lexical categories, grammatical categories, figures of speech, and cohesion and context. Lexical categories focus on the choice and use of words, considering aspects such as diction, connotation, and word frequency, which contribute to the tone and mood of a narrative (Leech & Short, 1981). Grammatical categories, on the other hand, highlight sentence structure, syntactic variety, and the complexity of grammatical constructions. This dimension is crucial for understanding how AI constructs its narratives and whether it exhibits the flexibility and complexity typically found in human-authored texts (Leech & Short, 1981). Figures of speech encompass a variety of rhetorical devices, including metaphors, similes, and personification, which are vital in creating imagery, evoking emotions, and enriching the narrative voice (Leech & Short, 1981). Lastly, cohesion and context deal with how different parts of the text connect, both within sentences and across larger narrative structures. This category is particularly relevant for assessing how AIgenerated stories maintain coherence, manage narrative shifts, and create a cohesive world for the reader (Leech & Short, 1981). By applying this comprehensive framework to AI-generated short stories, this study aims to explore the extent to which AI-generated narratives demonstrate stylistic sophistication, or whether they remain formulaic and derivative. The approach is designed to assess how well AI can construct meaning and artistic expression, evaluating its potential as a creative partner in storytelling. Problem Statement The creation of AI-generated text has raised significant questions about its artistic and linguistic capabilities, particularly within the realm of creative writing. While much of the existing research on AI language models (Bender et al., 2021; Radford et al., Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1214 2019) has focused on their technical performance and ethical implications, little attention has been paid to the stylistic characteristics of AI-generated fiction. It remains unclear whether AI can generate text that exhibits a sophisticated, consistent style, or if its output is limited to a generic amalgamation of its training data. The problem, therefore, lies in the absence of a systematic, linguistic analysis of AI’s creative outputs, particularly short stories, which are often criticized for lacking the depth and originality found in human-authored works (McCormick, 2020). Given this gap in the literature, this study aims to apply Leech and Short’s (1981) framework to AI-generated short stories to assess the stylistic features that define their narrative structure. The research will address whether AI-generated stories demonstrate meaningful stylistic choices that contribute to a coherent and engaging narrative or if they merely replicate surface-level patterns without exhibiting genuine artistic expression. By systematically analyzing the lexical, grammatical, and figurative elements of these texts, the study will provide valuable insights into the creative potential and limitations of AI in the domain of narrative fiction. Research Objectives This study aims to provide a detailed and systematic stylistic analysis of AI-generated short stories using Leech and Short’s (1981) framework. The general objective is to evaluate the dominant lexical, grammatical, and figurative language features in a corpus of AI-generated short stories. Specific objectives include: To identify and categorize the dominant lexical, grammatical, and figurative language features in AI-generated short stories. To analyze the use of stylistic devices related to four main categories: Lexical Categories, Grammatical Categories, Figures of Speech, and Cohesion and Context. Research Question The research questions guiding this study are: What are the predominant lexical and grammatical patterns in the selected AIgenerated short stories? How does the AI employ figures of speech (e.g., simile, metaphor, irony) and narrative techniques related to context and point of view? Literature Review The intersection of artificial intelligence (AI) and creative writing, particularly in the generation of fiction, has been an area of growing interest. As AI models such as large language models (LLMs) like GPT-3 and GPT-4 become increasingly adept at producing coherent and contextually relevant text, there is a noticeable shift from technical assessments toward evaluating the artistic and stylistic quality of their output. Despite these advances, research in this domain remains sparse, particularly when it comes to the stylistic analysis of AI-generated short stories. The existing literature primarily addresses the technical capabilities, ethical concerns, and linguistic accuracy of AI-generated text (Bender et al., 2021), but there is limited work exploring the narrative and stylistic dimensions of AI’s creative fiction output. Stylistic analysis, as a discipline within linguistics, has a long history of examining the language features that define literary texts. Leech and Short’s (1981) framework for stylistic analysis offers a comprehensive approach, categorizing stylistic elements into four major areas: lexical categories, grammatical categories, figures of speech, Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1215 and cohesion and context. This framework has been widely used in literary studies to understand how language creates meaning in literature. However, its application to AI-generated texts, especially short stories, is largely unexplored. While there has been significant focus on how AI models can generate grammatically correct and semantically meaningful text (Radford et al., 2019), questions remain about whether AI can develop a distinctive narrative style akin to human authors. Previous studies on AI and creative writing have highlighted the potential of AI to generate text that mimics human authorship (McCormick, 2020). Early works, such as the creation of AI-generated poems and short stories, have revealed that while AI can produce coherent narratives, they often lack the depth, complexity, and stylistic coherence of human-written literature. Researchers have argued that AI-generated stories tend to follow a formulaic structure, influenced by the vast dataset from which the AI learns (Zaman, Wasim, & Chandio, 2025). For instance, McCormick (2020) suggests that AI lacks the true creativity necessary for literary fiction, often producing predictable outcomes due to its reliance on data patterns rather than original thought processes. A key concern in AI-generated creative fiction is the question of originality. AI systems are trained on extensive corpora of text from a variety of sources, which raises the issue of whether AI can truly create new, unique literary styles or if its work is simply an amalgamation of existing patterns. This concern is echoed by Bender et al. (2021), who caution that large language models may perpetuate biases inherent in their training data, leading to a form of "stochastic parroting" rather than genuine creative output. The question of whether AI-generated texts can exhibit the stylistic sophistication of human writers remains central to understanding the limitations and potential of AI in literary production. Recent studies have begun to address some of the stylistic features of AI-generated text, but these are often focused on general language use and the coherence of the narrative rather than specific literary stylistics. For example, AI models have been shown to effectively generate grammatically accurate sentences and employ syntactic structures that mimic human writing (Vaswani et al., 2017). However, less attention has been given to the subtler aspects of style, such as lexical choice, tone, and the use of figures of speech. This gap is crucial because literary style involves not only grammatical correctness but also the nuanced use of language to evoke mood, theme, and meaning (Leech & Short, 1981). Figures of speech, such as metaphors, similes, and irony, play an important role in shaping the narrative voice and emotional resonance of a story. The use of these devices in AI-generated texts has been largely overlooked in existing research. However, some studies suggest that while AI can generate these devices, they are often used in a mechanical or predictable manner (McCormick, 2020). This raises questions about whether AI can develop a true "voice" or if its use of stylistic devices is simply formulaic, drawn from patterns observed in its training data. Given the increasing role of AI in creative fields, there is a clear need for more focused research on the stylistic features of AI-generated fiction. While previous research has addressed the technical and ethical dimensions of AI’s capabilities (Bender et al., 2021; Radford et al., 2019), little attention has been paid to understanding how AI’s linguistic choices affect the artistic quality of its generated texts. This gap is particularly important for evaluating AI as a creative partner in literature, as the stylistic features of its output directly influence its potential for Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1216 meaningful artistic contribution. Leech and Short’s (1981) framework offers a robust methodology for exploring these features systematically. By applying this framework to a corpus of AI-generated short stories, this study aims to fill the gap in the literature by providing a detailed analysis of the lexical, grammatical, and figurative elements that characterize AI’s narrative style. Understanding these stylistic features is essential for assessing whether AI can produce texts that possess the depth, creativity, and individuality associated with human authorship, or if its work remains confined to the replication of pre-existing patterns. Methodology This study adopted a qualitative research design. The primary methodology was Stylistic Analysis, systematically guided by the analytical checklist provided in Leech and Short's Style in Fiction. Research Philosophy The interpretivist philosophy emphasizes understanding and interpreting the meanings and patterns within human (or linguistic) expressions rather than measuring them quantitatively. It values subjective interpretation and contextual analysis. Guided by this perspective, the present study has adopted an interpretivist research philosophy to explore and interpret the stylistic features of AI-generated short stories, focusing on how language constructs meaning and creativity within these narratives. Data Collection Tools Step 1: AI Text Generation Platform ChatGPT has been used as the primary tool for generating short stories, with a focus on creating stories that convey a moral lesson. Step 2: Prompt Design A single prompt was crafted to guide the generation of short stories. This prompt encouraged the creation of narratives with a clear moral message, without imposing specific stylistic constraints. Sampling Technique A purposive sampling technique was employed to select stories generated by the AI for analysis. Specifically, 10 stories were selected from the narratives created by the AI. Data Analysis The data analysis followed a systematic, iterative process using the Leech and Short framework. Initially, the entire corpus was familiarized through repeated readings. Then, each story was coded based on four levels: lexical categories (word choice), grammatical categories (sentence complexity), figures of speech (metaphors, similes, etc.), and cohesion and context (narrative perspective and speech presentation). Recurring patterns and notable deviations within the codes were identified. These patterns were compared across stories generated by the same and different AIs, leading to a comprehensive stylistic profile of the narratives. Finally, the identified features were interpreted to understand the nature and sophistication of AI's creative Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1217 language use. Story 1 “The Garden of Second Chances" Aspect Details Lexical Categories (Key Terms) Ayan, Mr. Rahim, garden, help, flowers, promise Grammatical Categories (Sentence Structure) Simple, Compound Figures of Speech (Examples) Metaphor (garden as a second chance), Irony (garden's transformation), Personification (the garden 'coming to life') Cohesion and Context (Narrative Perspective) Third-person omniscient (narrator knows Ayan’s and Mr. Rahim's thoughts) Moral/Theme Small acts of kindness can revive what the world has given up on. Story 2 “The Lost Wallet" Aspect Details Lexical Categories (Key Terms) Wallet, money, honesty, reward, peace Grammatical Categories (Sentence Structure) Simple, Compound Figures of Speech (Examples) Irony (Sara’s temptation vs. the man’s gratitude), Metaphor (money represents a life saved) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Sara’s actions and internal conflict) Moral/Theme Doing the right thing may not always be easy, but it always brings peace and blessings. Story 3 “The Boy Who Returned the Rain" Aspect Details Lexical Categories (Key Terms) Hamza, rain, nature, village, bird, land, clean, hope Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Personification (the bird’s glow), Metaphor (rain as nature's reward for care), Imagery (descriptions of the village) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Hamza’s actions and thoughts) Moral/Theme When you care for nature, nature cares for you in return. Story 4 “The Clockmaker’s Promise" Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1218 Aspect Details Lexical Categories (Key Terms) Clock, watch, honesty, trust, promise Grammatical Categories (Sentence Structure) Simple, Compound Figures of Speech (Examples) Metaphor (the watch as a symbol of trust), Irony (the broken clock vs. the man’s honesty) Cohesion and Context (Narrative Perspective) Third-person omniscient (focusing on Yusuf’s internal struggle and Raza’s perspective) Moral/Theme Honesty may be difficult, but it always earns respect in the end. Story 5 "The Artist Who Couldn’t See Colors" Aspect Details Lexical Categories (Key Terms) Mira, colors, art, feelings, emotion Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Personification (painting as an experience), Metaphor (painting as a medium of emotion) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Mira’s emotional journey and artistic process) Moral/Theme Limitations cannot stop you if you turn them into strengths. Story 6 “The Teacher Who Stayed Late" Aspect Details Lexical Categories (Key Terms) Faris, teacher, lessons, belief, effort, education Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Metaphor (teaching as nurturing), Irony (Faris’ struggle to succeed despite his efforts) Cohesion and Context (Narrative Perspective) Third-person omniscient (focusing on Mrs. Hania’s thoughts and actions) Moral/Theme Sometimes, one person’s belief in you can rewrite your whole future. Story 7 “The Child With the Empty Jar" Aspect Details Lexical Categories (Key Terms) Ilham, king, honesty, seed, truth Grammatical Categories Simple, Compound Liberal Journal of Language & Literature Review Print ISSN: 3006-5887 Online ISSN: 3006-5895 1219 Aspect Details (Sentence Structure) Figures of Speech (Examples) Metaphor (empty jar as honesty), Irony (the other children’s plants were false) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Ilham’s honesty and the king’s recognition) Moral/Theme Honesty shines even when the world rewards only results. Story 8 "The Boy Who Saved the Library" Aspect Details Lexical Categories (Key Terms) Imran, library, books, community, effort, challenge Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Metaphor (books as friends), Imagery (vivid descriptions of the library’s state and activity) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Imran’s determination and actions) Moral/Theme One determined effort can protect what matters to everyone. Story 9 “The Friend Who Didn’t Give Up" Aspect Details Lexical Categories (Key Terms) Ayan, Zoya, friendship, school, hardship, support Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Metaphor (friendship as a lifeline), Irony (Zoya’s struggle vs. Ayan’s determination to help) Cohesion and Context (Narrative Perspective) Third-person limited (focusing on Ayan’s perspective and Zoya’s struggles) Moral/Theme True friendship means standing by someone when they cannot stand by themselves. EStory 10 "The King and the Three Bags of Rice" Aspect Details Lexical Categories (Key Terms) King, rice, kindness, honesty, leadership Grammatical Categories (Sentence Structure) Simple, Compound, Complex Figures of Speech (Examples) Metaphor (rice as a symbol of generosity), Irony (Haris and Sami’s deception vs. Rafat’s selflessness)