Vibecoding as Method: A Structured Workflow for AI-Assisted Research Instrument Development in Applied Linguistics

Abstract

This methods tutorial reframes 'vibecoding' - the practice of building functional software through natural-language interaction with AI systems (e.g., Anthropic's Claude) without conventional programming - as vibecoding-as-method: a structured workflow for designing and developing research instruments in applied linguistics. Situated within a design-based research (DBR) framework, vibecoding-as-method is conceptualized as a mechanism for accelerating iterative research instrument design through prompt-mediated development and rapid AI prototyping. Using corpus-based data-driven learning (DDL) as an illustrative context, and demonstrated through the development of VetLingua, a vibecoded DDL instrument for veterinary domainspecific language learning, the tutorial outlines a stepwise workflow encompassing corpus preparation, feature operationalization, prompt design, interface specification, backend generation, deployment, and (planned) extensions to analytics and evaluation. Across the vibecoding-asmethod workflow, prompts, constraints, and system outputs are treated as methodological artefacts requiring documentation and justification, emphasizing transparency, replicability, and researcher oversight. Representing a first articulation that requires validation across further contexts, we present vibecoding-as-method as a potentially scalable and accessible approach to research instrument development - one with the potential to extend methodological repertoires for technology-mediated research in applied linguistics and other disciplines.