Jul 2026· International Conference on Conversational User Interfaces· 0 citations· 42 references
Computer Science
TL;DR
Foxy, a proactive LLM-driven Conversational Agent designed to assist Italian children aged 8–11 with dyslexia during morphological training, acts as a specialized tutor that provides scaffolding, limits topic drift, and reduces hallucinations via a verified lexical knowledge base.
Abstract
Developmental dyslexia involves persistent difficulties in word-level reading and decoding, requiring sustained linguistic practice that is difficult to maintain without supervision. Although Generative AI offers personalized support, standard Large Language Models (LLMs) often lack the pedagogical and therapeutic knowledge required for linguistic intervention. We present Foxy, a proactive LLM-driven Conversational Agent designed to assist Italian children aged 8–11 with dyslexia during morphological training. Through a modular prompt-orchestration framework and an event-driven architecture, Foxy acts as a specialized tutor that provides scaffolding, limits topic drift, and reduces hallucinations via a verified lexical knowledge base. The system was refined through expert-led co-design and evaluated in a pilot study with 18 educators, and therapists. Results indicate positive perceptions of usability, usefulness, and appropriateness, with recognition of Foxy’s motivational benefits. These findings suggest Foxy is a promising tool for dyslexia intervention, pending wider empirical validation.
Abstract—Students with dyslexia, dysgraphia, dyscalculia, attention-deficit/hyperactivity disorder, autism-related support needs, and language-processing difficulties often encounter barriers in standardized digital and classroom instruction. This article presents NeuroAdapt-RAG, a responsible Large Language Model (LLM) framework that combines learner profiling, curriculum-grounded retrieval-augmented generation (RAG), adaptive feedback, multimodal accessibility, progress analytics, and human oversight. The framework converts a learner's academic level, preferred modality, support requirements, mastery state, and interaction history into controlled prompt instructions. Relevant curriculum passages are retrieved from an approved knowledge base and supplied to the LLM to reduce unsupported generation. A safety layer checks age appropriateness, instructional relevance, privacy exposure, and potentially harmful content before delivery. The proposed architecture supports simplified explanations, guided practice, text-to-speech, speech-to-text, teacher/parent dashboards, and auditable learner progress. Functional validation is organized around authentication, profile creation, retrieval, personalization, accessibility, feedback, security, and latency. Because controlled learner-outcome data are not yet available, this study deliberately reports a reproducible evaluation protocol and result template rather than unsupported efficacy claims. The framework provides a practical foundation for inclusive AI tutoring while preserving educator authority and ethical safeguards.
Keywords— adaptive learning, inclusive education, large language models, learning disabilities, personalization, retrieval-augmented generation, responsible AI.
Gummadi Meenakshi, P. Priya· International Scientific Jou...· 0 citations
Generative artificial intelligence (GenAI) has rapidly expanded from text-based assistance to voice-enabled dialogue, automated oral feedback, and multimodal speaking assessment. These developments appear to address a persistent problem in second-language (L2) education: learners need frequent, low-risk opportunities to speak, but teachers cannot always provide individual interaction and feedback at scale. However, the availability of an apparently fluent conversation partner does not establish that durable speaking development, fair assessment, or transfer to human communication will follow. This critical integrative review examines three questions: what learning and affective outcomes are associated with GenAI-supported L2 speaking practice; what limitations weaken its pedagogical value; and what implementation conditions support responsible use. Targeted searches of Google Scholar, publisher platforms, open research repositories, and citation chains produced an analytical corpus of 11 core publications, supported by 14 theoretical, methodological, assessment, feedback-literacy, and ethics sources published or retained for interpretation through July 2026. Evidence was appraised for contextual clarity, task alignment, outcome validity, feedback transparency, and the strength of claims. The synthesis indicates that GenAI can increase practice volume, support rehearsal, provide scenario-based interaction, and reduce fear of immediate human judgement. Adaptive prompts and rapid feedback may also support vocabulary retrieval, discourse organisation, pronunciation awareness, and willingness to communicate. Yet the evidence remains dominated by small samples, short interventions, self-report measures, prototype studies, and technical benchmarks. Recurring risks include inaccurate or overconfident feedback, accent and speech-recognition bias, unnatural interaction, dependency, privacy concerns, unequal access, and weak transfer evidence. The review proposes the SPEAK framework: Structured tasks, Progressive intelligibility-focused feedback, Equity and ethics, Agency and anxiety-sensitive practice, and Knowledgeable teacher oversight. GenAI should therefore be used as a structured rehearsal and feedback resource within teacher-designed speaking pedagogy, not as an autonomous replacement for human interaction or professional judgement.
D. Dasanayake· International journal of res...· 0 citations
Despite the persistent teacher-fronted pedagogy in low-resource university contexts, EFL learners continue to struggle with autonomy and speaking opportunities. This study therefore investigates the extent to which integrating AI-driven conversational agents can foster autonomous EFL learning among Beninese university students. A mixed-methods quasi-experimental design was adopted with 86 second-year EFL students divided into an experimental group using a ChatGPT-based conversational agent for 12 weeks and a control group receiving traditional instruction. Data were collected through pre/post oral proficiency tests, learner autonomy questionnaires based on Benson (2013), and semi-structured interviews. Quantitative data were analyzed using ANCOVA, while thematic analysis was applied to qualitative data.The experimental group showed statistically significant gains in oral fluency, lexical complexity, and learner autonomy compared to the control group (p < .001, d = 0.87). Qualitatively, learners reported increased motivation, reduced speaking anxiety, and enhanced self-regulation. The AI agent provided 24/7 personalized feedback, thereby compensating for large class sizes and limited teacher contact time. The findings confirm that AI- driven conversation al agents constitute a viable, scalable pedagogical tool for promoting learner autonomy in low-resource EFL contexts.
Assogba Evariste Kottin, Coffi Martinien Zounhin Toboula, Arlette J. Viviane Hounhanou· International Journal of Adv...· 0 citations
This quasi‐experimental study examines the impact of short‐term prompting instruction on the communicative performance of Czech learners of English in human–chatbot interaction, conceptualizing their prompting behavior as a context‐specific manifestation of pragmatic competence. The study was conducted in two phases (September 2024 and May 2026) and included four groups of participants with varying chatbot experience. Participants with (experimental groups) and without (control groups) prior prompting instruction completed 20 communicative tasks of increasing complexity, during which four indicators of communication effectiveness were recorded: task completion rate, number of conversational turns, user satisfaction, and frequency of chatbot misunderstandings. The findings indicate that prompting instruction improves communication efficiency, particularly in higher‐complexity tasks, as reflected in fewer conversational turns and significantly reduced misunderstandings in human–chatbot interaction, alongside a modest, non‐significant advantage in task completion. More importantly, it facilitates a shift from the inference‐based pragmatic strategies typical of human–human communication toward more explicit forms of contextualization and communicative guidance. The inclusion of two data‐collection phases further enabled a comparison of interaction patterns across different stages of public familiarity with chatbot technology, suggesting that growing familiarity may encourage users to rely more heavily on established interaction routines while applying grammar and spelling rules less consistently. The study therefore contributes to the emerging discussion of chatbot pragmatics by demonstrating how Gricean communicative principles are adapted in AI‐mediated interaction and by highlighting the role of prompting instruction in facilitating this adaptation.
Alexey Tymbay· International Journal of App...· 0 citations
Grammatical accuracy in spontaneous second language (L2) speech remains difficult to develop, particularly in contexts with limited opportunities for sustained oral practice. This study investigates whether structured, memory-enabled artificial intelligence (AI) speech coaching supports the development of grammatical accuracy among Armenian EFL learners.
A 14-week longitudinal mixed-methods design was employed with 64 undergraduate students, supplemented by a non-randomized comparison group (n = 13). Learners engaged in weekly AI-supported speaking tasks, and grammatical accuracy was measured as errors per 100 words across three time points.
Repeated-measures ANOVA revealed a significant reduction in grammatical errors over time (ηp² = .45), with the strongest improvements observed in rule-governed categories such as tense–aspect, subject–verb agreement, and word order, and more limited progress in articles and prepositions. Moderate-proficiency learners demonstrated the most consistent gains, while the comparison group showed no significant improvement.
The findings suggest that AI-mediated feedback can support grammatical development when embedded within structured pedagogical frameworks, while highlighting the importance of instructional guidance to prevent overreliance on automated corrections. This study contributes longitudinal evidence on AI-supported grammatical development in spontaneous speech and provides a fine-grained analysis of category-specific improvement.
L. Sargsyan· Teaching English as a Second...· 0 citations
A simulation-based textual analysis of prompt design evaluates a frontier large language model as a tutor across 60 scripted sessions on a single topic and point to dynamic, dialogue-aware prompting alongside explicit SRL scaffolding.
Kendall Hartley, Fabiola Sáez-Delgado, Javier Mella-Norambuena· Future Internet· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
