The impact of computational methods on Albanian literature analysis

This study explores the integration of computational emotional analysis methods in the teaching of Albanian literature and their effect on students' analytical skills. Using a mixed-methods approach, 31 graduate students in Albanian Language and Literature combined traditional textual analysis with distant reading, applying the NRC Emotion Lexicon through Perplexity AI. They analyzed Ismail Kadare's Chronicle in Stone and Ervin Nezha's The Dead Live in the Mirror, evaluating the educational potential of integrating literary interpretation with computational analysis. Methodological note: Given the lack of natural language processing tools for Albanian, texts were translated into English for computational analysis, introducing translation bias recognized as a significant limitation. Additionally, pre- and post-intervention surveys (n=31) measured attitudes and self-reported changes in competencies, using instruments validated and adapted to the Albanian academic context. Results: Initial findings reveal that despite technical difficulties (77% without computational experience), 68% considered emotional analysis "moderately useful" for literary interpretation. The traditional approach identified fear, anxiety, and melancholy, while computational analysis detected fear (35%) and sadness (25%) in Kadare, as well as trust/belief (30%) in Nezha. Inter-rater agreement on emotional preservation in translation reached κ = 0.72, considered substantial. Implications: This pilot study demonstrates the potential of hybrid pedagogical approaches in literature for low-resource languages, though it requires validation with larger samples and control groups. It constitutes the first systematic research on computational emotional analysis in Albanian literary education and offers a framework for future studies in similar contexts. Limitations: The small convenience sample (n=31), translation-dependent methodology, lack of control group, single-institution setting, and short-term evaluation limit the generalizability of results. Future research requires larger samples (n>64 for medium effects), Albanian-specific tools, and long-term follow-up.