From Wall-of-Text to Graphic Novel: Can Generative AI Enhance the Pleasure Reading Experience of English as a Second Language Readers?

Abstract

This study explores an innovative approach to enhancing the reading experience for English as a Second Language (ESL) readers by leveraging generative AI to transform traditional narratives into graphic novels. Traditional pleasure reading is often perceived with frustration and, in today’s fast-paced world, there seems to be little time for it. Yet its benefits, especially for ESL readers, are well-documented, from improving language skills to boosting cultural understanding. Graphic novels are increasingly recognized as helpful tools to aid ESL readers' approach literature through the combination of text and visuals. Given the growing influence of generative AI, we investigated whether generative models could transform a short-story into a graphic novel, and whether this AI-generated graphic novel could deliver the same advantages as traditional ones in terms of comprehension, engagement and enjoyment. To address these questions, GPT-4 and Midjourney were used to automatically transform the story “Flight” into a graphic novel via a LangChain prompt-engineering pipeline. To test the efficacy of the generated graphic novel, 80 proficient ESL participants were recruited, and a pre-test/post-test control group design with post-experimental disclosure was employed. The pre-test ensured comparable English baselines, while the post-test assessed whether comprehension, narrative engagement, and enjoyment of the graphic novel (treatment group) were higher than those of the text-only format (control group). Data was analyzed using descriptive statistics, quantitative analysis, and qualitative thematic analysis. Surprisingly, results showed no significant difference between groups for comprehension, enjoyment, or overall engagement, with the control group exhibiting significantly higher emotional engagement. Post-experimental disclosure, which focused on the treatment group’s opinions and perceptions of AI in creative fields, revealed that 70% of participants detected AI involvement due to image texture and inconsistencies. Moreover, being comfortable with illustrating narratives using AI significantly predicted the belief that using AI as in this study is useful, while frequent exposure to generated images significantly decreased its likelihood. Additionally, no significant difference was found in the enjoyment of the graphic novel before and after participants were informed about AI’s role in its creation. These findings reveal nuanced insights into how ESL learners perceived and benefited from the AI-generated graphic novel, requiring further exploration of the interplay between AI-generated content and the pleasure reading experience. Additional interpretations and implications of using generative AI to streamline the transformation of narratives into graphic novels are discussed, together with study limitations.