Faculty Members Publication
AI-Integrated Multimodal Framework: Text-to-Image Generators for Enhancing English Metaphor Comprehension
Conference Presentation
3rd TESOLBD DIU International Conference, Daffodil International University, December 2025
Traditional Bangladeshi ESL pedagogy often fails to foster instinctive metaphor comprehension among first-year students. This study introduces an AI-integrated multimodal approach, employing text-to-image generators to visualize abstract metaphors within culturally relevant frameworks. Using a within-subjects design, students experienced metaphor instruction both with and without AI-generated visuals. Bloom’s Taxonomy-based tests measured cognitive outcomes, while self-assessments gauged affective responses. Statistical analysis revealed that AI-supported instruction significantly enhanced comprehension and learner confidence. The integration of AI-generated imagery demonstrates a promising pedagogical innovation, uniting cognitive
Collaborative Composition of Bangla Free Verse Poetry with Generative AI: Gauging the Future of Cyborgian Creative Capacity
-
Conference Presentation
-
International Conference on Teaching English Literature and Interdisciplinarity”, East West University, July 2023
The paper explores the potential and limitations of composing Bangla Free Verse Poetry in collaboration with Generative Artificial Intelligence (AI). Firstly, it analyzes the linguistic structure of both Free Verse Poetry and Bangla Poetry to gain insight into the linguistic nature of Bangla Free Verse Poetry. Then the study employs an experimental methodology to collect and analyze data, providing a quantifiable information pool for its proposed hypotheses. Using the findings as a springboard, the study investigates the difficulties of instructing a predictive text generation algorithm to generate Bangla Free Verse Poetry with genuine meaning. It underlines the level of difficulty to map Bangla poetry using Natural Language Programming (NLP) because of its intricacy, structure, depth of emotion, and semantic complexity. Detailed discussion regarding the mechanics behind predictive text generation used by generative AI systems is emphasized, highlighting their shortcomings in expressing the rich emotional range present in poetry. The research also highlights the requirement for generative AI models to acquire a profound awareness of their own emotional states and the capacity to abstractly convey them into human language. To do this, the AI system must simulate emotional states and use abstract reasoning techniques to create emotionally resonant replies. It also emphasizes the value of each reader's unique perception of poetry as well as their neurological programming, which is based on the nature vs. nurture debate. Ultimately, the paper tries to predict the potential and limitations of creative writing endeavors humans can take on utilizing Generative AI. It emphasizes the need for Generative AI systems to integrate emotional intelligence, linguistic proficiency, and creative expression more deeply, which may lead to far more ambitious creative projects than poetry generation.
Keywords: Free Verse Poetry, Bangla Poetry, Generative AI, Cyborgian Creative Capacity (CCC), Natural Language Processing (NLP)
Implementing Suggestopedia to Address Second Language Learning Challenges Among Trauma Afflicted Adult ESL Learners
Poster Presentation
First Runner Up, “Ruptures and Resilience: English Studies in the Now”, North South University, November 2022
This research explores the implementation of the Post-Suggestopedic Method in addressing language learning challenges among young adult learners who have experienced trauma. The study investigates how this innovative educational approach, which builds upon the principles of suggestopedia by incorporating contemporary psychological insights and pedagogical techniques, can effectively support and enhance the second language (L2) acquisition process for this specific learner demographic. Through a combination of qualitative and quantitative methodologies, the research aims to evaluate the effectiveness of the Post-Suggestopedic Method in creating a supportive learning environment that not only facilitates language learning but also addresses the unique needs of trauma-affected individuals, thereby contributing to the field of language education and trauma recovery.
The Leveling Effect: AI-Assisted Narrative Production and the Future of Professionalism in Creative Industries
-
Conference Presentation
-
The Nation and Its Imagination: Culture and Creative Industries in Bangladesh, International Conference, Independent University of Bangladesh, March 2026
- The proliferation of large language models (LLMs) in creative writing has raised critical questions about authorial authenticity, skill differentiation, and the future of literary production. This double-blind experimental study investigates whether AI-assisted narrative endings reduce perceptible quality differences between professional and amateur fiction writers—a phenomenon termed the “leveling effect.” Employing a 2×2 mixed-factorial design with Latin Square counterbalancing, 120 readers evaluated narrative endings across 10 short stories (5 professional, 5 amateur) in both human-original and GPT-4o-augmented conditions. Results were analyzed using Linear Mixed-Effects Models (LMM) and Signal Detection Theory (SDT) to assess both quality perception and readers’ ability to detect AI intervention. Preliminary findings suggest that AI augmentation disproportionately elevates amateur writing quality (Δ Mamateur = +1.42 vs. Δ Mprofessional = +0.38 on narrative transportation scales, p < .001), while simultaneously reducing stylistic distinctiveness. Readers demonstrated above-chance AI detection accuracy (d’ = 0.74) but with significant false-alarm rates, indicating perceptual ambiguity in hybrid human-AI texts. This study contributes to computational creativity scholarship by providing empirical evidence for AI’s role as a “stylistic equalizer” and raises urgent questions about authorship attribution, literary pedagogy, and the preservation of authorial voice in increasingly AI-mediated creative industries.
Keywords: artificial intelligence, narrative generation, computational creativity, authorship, literary quality assessment, human-AI collaboration
The Craft of Sublimation: A Comparative Study on the Portrayal ofSuffering in Derek Walcott and Kamau Brathwaite’s Poetry
The practice of channeling energies derived from suffering into an act of healing, leading to apositive transformation is not new in the world of literature. Walcott and Brathwaite both have partaken in this practice. The suffering in Walcott and Brathwaite’s case is caused, largely, due to owning a diasporic identity. However, along with it, also weighs heavy the history of a wounded past. They have channeled this suffering into a creative force by producing works that sublime the angst that is not simply exclusive to them but flows through the majority of the Caribbean population. Walcott and Brathwaite have justified their suffering of identity crisis through helping a nation heal its wounds, and in certain contexts, making it transcend its former self to a healthier form. However, they have crafted unique styles to achieve that end. The paper attempts to highlight the differences as well as similarities between the methods and perceptions of these two poets in this regard
(PDF) The Craft of Sublimation: A Comparative Study on the Portrayal of Suffering in. Available from: https://www.researchgate.net/publication/378262986_The_Craft_of_Sublimation_A_Comparative_Study_on_the_Portrayal_of_Suffering_in [accessed Feb 17 2024].
Undergraduate and Graduate English Majors’ Attitudes toward Native and Non- Native English Accents and Their Role in Identity Formation (Accepted in the 19th Volume of Spectrum, Journal of Department of English, University of Dhaka)
This study examines the attitudes of undergraduate and graduate English majors toward ‘native’
and ‘non-native’ English accents and explores how these attitudes influence their identity
formation. It aims to determine participants’ perceptions of the status and importance of different
English accents, their self-assessment of their own accent, and the relationship between accent
and identity. Employing an explanatory sequential mixed-methods approach, the study integrates
a questionnaire administered to 105 students (54 first-year undergraduates and 51 M.A. graduates)
at a public university in Bangladesh, complemented by focus group discussions analysed through
grounded theory of the constructivist design. Quantitative data are analysed using descriptive
statistics and Levene’s t-test to identify differences between the two groups of participants, while
qualitative data are examined through manual coding. Findings show that participants generally
reject the notion of a ‘native’ English accent as the definitive ideal and do not regard a ‘non-
native’ English accent as inferior. Nevertheless, quantitative data indicate that participants
perceive a ‘native’ English accent as socially prestigious, suggesting a desire to acquire it.
Qualitative discussions reveal deeper power dynamics rooted in colonial history, with participants
expressing a preference for developing a 'standard regional accent' that preserves their linguistic
heritage and aligns with local vocal norms, rather than aspiring to emulate a 'native’ English
accent.
Flipped classroom: beyond the lecture
An online newspaper article that explores the efficacy of inverting the conventional approach of teaching in the classroom.
Authors: Shaharima Parvin and Md Arshad Shahriar
Publication date: 2026/4/21
Source: New Age
Link: https://www.newagebd.net/post/opinion/293686/flipped-classroom-beyond-the-lecture
Conference Paper Presentations
- Paul, H. (2026), "Investigating Self-Regulated Learning in Bangladeshi ELT Programs," paper presented at the 30th NELTA International Conference 2026, Kathmandu, Nepal, 30-31 January.
- Raisa, J.T. Paul, H. (2025) "AI Tools in Fostering Critical Thinking: Exploring Perceptions of Bangladeshi Undergraduate Students," paper presented at the 3rd TESOLBD-DIU International Conference 2025, Dhaka, Bangladesh, 5-6 December.

