Systematic review resulting in two complementary outputs:
Literature- and web-based systematic review of available technologies
Polizzi, D., A. Ferraresi, S. Bernardini, C. Cervini, M. Miličević Petrović, and G. Palmieri. Forthcoming. “Re-Evaluating AI Chatbot Effectiveness: A Systematic Review of Dialogue-Based CALL Research.” ALLiED: Journal of Applied Linguistics and Languages in Educational and Digital Settings.
Link: https://serena.sharepress.it/AL/article/view/13677/13705
A set of Guidelines for the evaluation of dialogue systems, jointly developed with CAST and grounded in the principles of Universal Design for Learning (UDL). These Guidelines provide a structured and operational framework to assess DS in terms of engagement, representation, and action and expression, with particular attention to the needs of students with disabilities and specific learning disorders.
Chatbot Accessibility Guidelines
Raffi, F., and L. Fedeli. 2026. Chatbot Accessibility Guidelines - Chatbot accessibility assessment within the Universal Design for Learning. Zenodo.
Link: https://doi.org/10.5281/zenodo.18184801
A richly annotated corpus of written interactions between learners and DS, to support subsequent corpus-based analyses and to provide a reusable resource for the research community. The resulting dataset represents an innovative contribution, as one of the first learner corpora specifically dedicated to human–machine interaction in EFL contexts.
Protocols are available here:
Link: https://doi.org/10.5281/zenodo.19049030
UNITE Zenodo Community:
Link: https://zenodo.org/communities/unite/
An adaptation of the Louvain error tagging taxonomy (Granger et al. 2022) within the UNITE project. The original taxonomy, which includes 54 error categories, was extended to address the specific characteristics of the UNITE corpus (Ferraresi et al. 2026), consisting of learner interactions with AI chatbots.
The tagset was released as part of the UNITE corpus package
Link: https://bellatrix.sslmit.unibo.it/noske/public/#dashboard?corpname=unite_corpus
UNIOR unit developed and implemented an innovative annotation framework (DIS-TAG), designed to detect and categorise discursive features related to implicit assumptions about users’ abilities. The annotation revealed recurrent patterns in chatbot language, such as highly standardised politeness, consistent positive reinforcement, and uniformly confident responses, alongside implicit assumptions related to mobility, sensory perception, and cognition.
While not overtly exclusionary, these patterns reflect underlying normative frameworks that may not fully account for user diversity. Importantly, these findings contrast with users’ perceptions of DS as neutral and reliable, highlighting the need to foster critical AI literacy. This strand resulted in the release of both the annotated corpus and the DIS-TAG tagset, which constitutes a novel, reusable resource for the analysis of inclusivity in AI-generated discourse.
Teaching materials to support the effective and inclusive integration of DS in English as a Foreign Language (EFL) teaching and learning, both in classroom settings and for autonomous learning.
Miličević Petrović, M., Cervini, C., Paone, E., Palmieri, G., & Ferraresi, A. (2026). Chatting with Chatbots in L2 English: The UNITE Guidelines. Zenodo.
Link: https://doi.org/10.5281/zenodo.19169244
Polizzi, D., A. Ferraresi, S. Bernardini, C. Cervini, M. Miličević Petrović, and G. Palmieri. Forthcoming. “Re-Evaluating AI Chatbot Effectiveness: A Systematic Review of Dialogue-Based CALL Research.” ALLiED: Journal of Applied Linguistics and Languages in Educational and Digital Settings.
Link: https://serena.sharepress.it/AL/article/view/13677/13705
Raffi, F., and L. Fedeli. 2026. Chatbot Accessibility Guidelines - Chatbot accessibility assessment within the Universal Design for Learning. Zenodo.
Developed in collaboration with: CAST - Until learning has no limits
Fedeli, L., and F. Raffi. 2025. “Digital Innovation in English Language Teaching: The Role of AI Chatbots for Social Sustainability.” The Journal of Language and Teaching Technology 7: 35–45. New Brunswick, NJ: Rutgers University.
Ferraresi, A., A. Mongibello, F. Raffi, and S. Bibauw, eds. Forthcoming. “AI and Foreign Language Learning: Bridging Classroom Practice and Research”. Special issue of journal ALLiED: Journal of Applied Linguistics and Languages in Educational Digital Settings.
Miličević Petrović, M., Cervini, C., Paone, E., Palmieri, G., & Ferraresi, A. (2026). Chatting with Chatbots in L2 English: The UNITE Guidelines. Zenodo.
The initial phase of the project (M1–M9) was devoted to the overview of available DS and to the preparation of experimental protocols. This phase required more time than originally anticipated, due to the need to reorient the technological focus from “traditional” chatbots to LLM-based systems, which had become dominant shortly before the start of the project. During this phase, UNIBO concentrated on experimentation with emerging tools, while UNIMC focused on refining the inclusivity framework and validating data collection instruments.