Abstract
Psycholinguistic research has established that words interact within the mental lexicon during both processing and learning. In spite of this, many Computer-Assisted Language Learning (CALL) systems treat second language (L2) vocabulary learning as the memorization of "vocabulary facts", and employ spaced-repetition algorithms designed to optimize the formation and maintenance of individual memory traces. The Knowledge- Learning-Instruction (KLI) framework provides guidelines as to what kind of knowledge components involve which learning processes, and how they are best taught. We reconsider the position of L2 vocabulary learning in the KLI framework, in light of extensive evidence of interaction and transfer effects in L2 vocabulary learning. We argue that L2 vocabulary learning involves the acquisition of generalisable skills. We further validate this claim with evidence from research into novel approaches to L2 vocabulary teaching. These novel approaches align with the instruction al recommendations made by the KLI framework for teaching complex rules, not facts, yet they yield significant improvements in L2 vocabulary acquisition. Finally, on the basis of these findings, we advocate for the use of skill-based learner models in order to optimize L2 vocabulary learning in CALL applications.
