Articles in L2 English (“a”/“the”) are notoriously difficult to acquire, especially for learners with article-less, [-art], L1s. The purpose of this dissertation is to identify the potential causes of learner difficulty with articles and to suggest an effective teaching intervention.

To identify the learner problem, I conducted a learner corpus-based study of English article use by L2 learners with four typologically distinct first languages (L1s): German and Brazilian Portuguese, both [+art], Chinese and Russian, [-art]. I coded and analysed 660 written scripts from the Education First Cambridge Open Database to investigate several semantic and morphosyntactic factors, such as specificity and prenominal modification. The key finding is the differential effect of specificity on definite and indefinite articles: learners tend to associate specificity with “a”, which results in article omission with non-specific indefinite singulars and overuse of “a” with specific indefinite mass nouns. Prenominal modifiers further contribute to perceived specificity leading to article overuse with modified indefinite mass nouns. However, in definite contexts, prenominal modifiers are associated with increased article omission.

Drawing on Input Processing theory, I argue that learner difficulties might stem from the inappropriate processing of articles in the input, which leads to learners making incorrect mappings between form (articles) and meaning (identifiability). In addition, learners at different levels are subject to cognitive constraints of varying strength, which can make it difficult or impossible to attend to morphosyntactic information. I designed a Processing Instruction Intervention containing activities that force correct article processing. Seventy pre-intermediate to upper-intermediate learners, half [-art] and half [+art], participated in a three-week-long online intervention experiment with randomised control-group pre-test post-test design. The pre-test contained a timed grammaticality judgment task (GJT) and an elicited imitation (EI) task. Mixed-effects regression modelling revealed a significant improvement in article accuracy on the timed GJT for the Experimental group but not for the Control group for both [-art] and [+art] L1s, which was maintained by the delayed post-test (3 months later). The results of the EI were less robust but showed a similar trend of improvement for the Experimental group but not for the Control group, which was mediated by participants’ education level and potentially linked to working memory effects in younger participants. Overall, Processing Instruction appears to be an effective and practical way of addressing common difficulties with articles in learner English.