The construct of communicative competence (CC), i.e., the capacity to produce context appropriate language (Bachman, 1990) has been broadly embraced as a key component of L2 development in assessment settings (e.g., Brindley, 2013; McNamara, 1996; Mislevy et al., 2002; Morrow, 1979, 2018). However, the implementation of the construct in language assessment has been limited due to issues related to consistency and standardisation (Harding et al., 2023) as well as requirements to adhere to frameworks and demands of the industry (Brunfaut & Harding, 2020). This study aims to address this challenge by proposing the construct of Linguistic Adaptability (LA) as a component of CC which can be observed and analysed with corpus methods. LA refers to the ability to adapt to the demands of communicative contexts through the observation of variation of language production across tasks and modes. Specifically, the study addresses the following questions: (i) what is the effect of spoken and written task on L2 output in the Aptis Corpus, and (ii) to what extent is task- based variation affected by L2 proficiency? The study takes a corpus approach to the analysis of L2 assessment output using the Aptis Corpus, a collection of L2 test-taker data from the British Council Aptis General Test (British Council, 2020) developed in collaboration with Lancaster University. A range of linguistic features were analysed based on their frequency variation across tasks and modes. Variation was further explored using concordance analysis in order to examine the links between variation and task context and results were compared across proficiency levels to investigate connections between LA and L2 proficiency development. The study found systematic patterns of variation for a range of linguistic features, suggesting that LA may be observable through corpus analysis. In addition, interactions between proficiency level and task-based variation provide some support for the importance of LA in L2 development. The findings of the study are discussed with reference to previous studies of SLA and LTA, and suggestions are made for the implementation of LA analysis in assessment contexts.
Emil Tangham Hazelhurst (Thu,) studied this question.