Work

Toward Equitable Access to Technical Training via Multilingual Conversational Agents

Year: 2026
Type: article
Abstract: The global technical skills gap disproportionately affects non-English-speaking populations who encounter substantial language barriers when attempting to access digital training resources. Multilingual conversational agents (MCAs), powered by advanced natural language processing (NLP) technologies, offer a promising pathway toward democratizing technical education across linguistic communities. This paper presents a mixed-methods investigation into the design, deployment, and evaluation of an MCA system tailored for vocational and technical skill acquisition in multilingual environments. Drawing on transformer-based multilingual language models, recurrent encoder-decoder architectures, and adaptive task-oriented dialogue management frameworks, the proposed system was assessed across four language groups — Spanish, Arabic, Mandarin Chinese, and Hindi — in simulated technical training scenarios covering information technology fundamentals and industrial safety protocols. Quantitative results indicate statistically significant improvements in learning outcomes, task completion rates, and user engagement compared to English-only baseline systems. Qualitative findings further reveal that language alignment between learner and agent substantially reduces cognitive load and increases perceived system credibility. This research contributes a validated architecture and evaluation framework for practitioners and policymakers committed to equitable access to technical education through intelligent conversational systems.
Language: English
Open Access status: hybrid
OpenAlex ID: W7153479617
FWCI:
Cites: 27
Cited by:
Related to: 10