Aim/Purpose
This study explores how free AI conversational applications support beginner English as a Foreign Language (EFL) learners’ speaking development, confidence, and self-regulated practice in Omani higher education. It specifically investigates how AI-mediated speaking practice influences fluency, vocabulary growth, anxiety reduction, and learner autonomy outside the classroom.

Background
Although AI-assisted language learning has expanded rapidly, beginner-level learners remain underrepresented in the literature, as most studies focus on intermediate or advanced users who can sustain longer AI interactions. For beginners, speaking anxiety, limited vocabulary, and fear of classroom judgement often restrict oral participation. Free conversational AI applications may offer psychologically safe, accessible, and scalable speaking spaces that lower affective barriers while promoting repeated practice and self-regulated learning, particularly in resource-limited higher education contexts.

Methodology
This mixed-methods case study followed 21 beginner-level undergraduate EFL learners at a private Omani higher education institution over fourteen weeks across two academic terms. Students used five free conversational AI applications—TalkPal, LingUp, Speakey, SpeakEasy, and Langua—for guided out-of-class speaking practice, supported by orientation sessions and weekly teacher follow-ups. Quantitative data were collected through pre- and post-oral rubric assessments and a validated learner questionnaire (Cronbach’s α = 0.87). Qualitative data were generated through focus group interviews and analysed thematically.

Contribution
The study contributes a beginner-focused perspective to AI-enhanced speaking pedagogy by proposing the Triadic Model of Conversational AI–Supported Speaking Development, which integrates cognitive feedback, self-regulated learning, and affective growth. It extends affective-filter and self-regulation perspectives into AI-mediated speaking development while demonstrating the pedagogical value of cost-free tools in inclusive and low-resource higher education settings.

Findings
Findings revealed significant gains in fluency and coherence (Δ = +0.73, p < .001) and lexical resource (Δ = +0.66, p < .001), alongside moderate gains in grammar and interactive communication. Learners also reported in-creased confidence, reduced anxiety, and stronger willingness to communicate. Qualitative findings highlighted recurring themes of improved fluency, enhanced vocabulary, gamified motivation, emotional safety, and growing learner autonomy. Students particularly valued the non-judgmental nature of AI conversations and the ability to practice repeatedly at their own pace.

Recommendations for Practitioners
Free AI conversational tools can serve as effective “after-class speaking partners” for beginner learners. Their impact is strengthened when paired with teacher orientation, regular follow-ups, and structured self-regulated learning routines. Embedding these tools into blended language support models may extend oral practice opportunities without increasing teacher workload.

Recommendations for Researchers
Future research should investigate larger beginner populations across multiple institutions and compare free versus premium AI conversational applications. Longitudinal and experimental designs are needed to examine sustained gains in speaking proficiency, confidence, and learner autonomy over time.

Impact on Society
The study demonstrates how freely accessible AI tools can democratize speaking practice for beginner EFL learners, particularly in contexts where learners may lack opportunities for authentic English interaction. By reducing anxiety and increasing confidence, AI conversational systems can enhance communication readiness, inclusion, and equitable access to language support.

Future Research
Future work should explore culturally responsive AI conversation design, including local Omani topics and dialect-sensitive scaffolds, while examining teacher readiness for AI-supported speaking instruction. Comparative studies across Gulf higher education contexts would further strengthen transferability.