Word sense awareness is a feature which has not yet been implemented in most Computer-Assisted Language Learning (CALL) environments or in computer-readable resources for pedagogical purposes such as graded word lists. The current study aims to contribute to filling this gap by presenting a word sense disambiguation (WSD) method1 which relies on a tailor-made sense inventory, exploits readily available large language models, and only requires a limited number of prototypical examples sentences as manually curated data. The methodology is evaluated on a set of 74 lexically ambiguous items, with a Spanish language for specific purposes course as the target setting. With weighted F1 scores up to 0.8995, the WSD method shows potential to be applied in real-life CALL scenarios.