Session 1B

Foreign language aptitude (FLA) refers to the prediction of how well someone is able to learn a foreign language. FLA is typically viewed as a relatively stable cognitive construct, comprised of abilities such as working memory, language analytic ability, and phonetic coding ability. However, with the increasing popularity of complex dynamic systems theory (CDST) to study individual differences in language learning, there has also been a shift in the way that FLA is viewed, with calls for a broader and more dynamic conceptualisation of FLA (Wen, 2021). This PhD project explores FLA from a CDST perspective, inspired by Snow's (1989, 1992) theory of aptitude complexes. Snow conceptualised aptitude as a complex of interactions between cognitive abilities, affective and conative individual differences, and the learning environment. To explore this theory, we are conducting a longitudinal FLA study with adults learning Dutch as a second language (L2) in the Netherlands. Our aim is (1) to better understand the complex interactions between the learning environment and cognitive, conative, and affective components and (2) to assess the importance of each component in relation to language learning gains. 
In this presentation I share initial findings from the ongoing study. We are collecting data online in collaboration with private language schools and university language centres across the Netherlands. We aim to have at least 300 participants. At the start of their Dutch course, participants complete questionnaires about their language background (e.g. first language, length of residence, amount of L2 instruction), conative and affective components (e.g. personality, motivation, anxiety), cognitive abilities (e.g. working memory, inhibition, non-verbal intelligence), and Dutch language proficiency. To measure language learning gains, participants repeat the Dutch language test at the end of their Dutch course. We are conducting a network analysis of the data. Network analysis is a relatively new technique that has been used to model constructs such as intelligence (van der Maas et al., 2006) and cognitive development (Kievit, 2020) from a CDST perspective. By conducting a network analysis of our data, we can visualise interactions between FLA components as a complex adaptive system and assess which components are most closely related to language learning gains. 
References
Kievit, R. A. (2020). Sensitive periods in cognitive development: A mutualistic perspective. Current Opinion in Behavioral Sciences, 36, 144-149. 
Snow, R. (1989). Aptitude-treatment interaction as a framework for research on learning and individual differences. In P. L. Ackerman, R. J. Sternberg, & R. Glaser (Eds.), Learning and individual differences (pp. 13-59). New York: Freeman.  
Snow, R. (1992). Aptitude theory: Yesterday, today, and tomorrow. Education psychologist, 27, 5-32.  
van der Maas, Han L. J., Dolan, C. V., Grasman, Raoul P. P. P., Wicherts, J. M., Huizenga, H. M., & Raijmakers, M. E. J. (2006). A dynamical model of general intelligence: The positive manifold of intelligence by mutualism. Psychological Review, 113(4), 842-861.  
Wen, Z. (2022). Language aptitudes. In T. Gregersen & S. Mercer (Eds.). The Psychology of the language learner. Preprint. New York, NY: Routledge. 

Language learning aptitude is one of the most consistent predictors of second language (L2) success (Li, 2015; 2016). Research to date has shown that different components of aptitude play a role at different stages of language learning and in different learning contexts and is dependent on the difficulty of targeted language structures. Higher aptitude seems to be associated with greater success on measures of explicit knowledge testing difficult language structures and on measures favouring implicit knowledge assessing easy language structures (Yalçın & Spada, 2016). Similarly, phonemic coding ability seems to impact learners at the early stages of L2 development, while language analytic abilities seem to impact learners at all levels (Artieda & Muñoz, 2016). Furthermore, studies suggest that phonological short-term memory (PSTM) is a predictor of early language development (Hummel, 2009; Serafini et al., 2016), but its effects in later stages of development seem to be limited. Conversely, measures of the central executive (CE) have been shown to correlate with L2 performance at more advanced levels of proficiency (Kormos & Sáfár, 2008; Linck et al., 2014). 
A major issue in existing research is that many studies have not taken into account the importance of language proficiency or difficulty of language structure, while others have included either a measure of PSTM or a measure of CE, but not measures of both. More specifically, no study to date has examined at what stage of L2 development individual differences in cognitive abilities are important predictors, and at what point they cease to have a major impact. The present study aims to disentangle the effects of language aptitude and working memory capacity on L2 learning by taking all these variables into account. The following research questions are addressed: 
First, to what extent are implicit language aptitude, explicit language aptitude, and working memory capacity associated with adolescent L2 learners' language development? Second, is the role of language aptitude and working memory capacity moderated by the difficulty of the target structure or the level of learners' L2 proficiency?
The study utilises the revised LLAMA (Meara & Rogers, 2019) together with a serial reaction time task (Kaufman et al., 2010) to measure explicit and implicit language aptitude while the automated operation span task (Unsworth et al., 2005) and forward digit span task are used to measure working memory capacity. The English simple past tense, passive, and articles are targeted as language structures of various difficulty. A self-paced reading task is used as a measure of implicit language knowledge, elicited imitation with a word monitoring component (Suzuki & DeKeyser, 2015) is used as a measure of implicit and automatized explicit knowledge, while a gap-filling test is used to gauge explicit language knowledge. The Oxford English Language Placement test and a free oral production task are used to assess language proficiency.
I will present initial data collected from Croatian secondary school students (N = 80) and would like to discuss the merits of different approaches to data analysis for answering the above research questions.

Language learning aptitude has been conceptualised as a construct of cognitive individual differences (ID) that predicts language learning outcomes, with a large amount of empirical evidence from various perspectives in SLA. Given that the aptitude construct is represented by scores from tests, it is crucial to ascertain the reliability and validity of the tests prior to conducting aptitude-learning research, yet this step has been surprisingly neglected to date (cf. Bokander & Bylund, 2019). Existing aptitude batteries face theoretical and methodological challenges that potentially threaten our understanding of the construct. First, aptitude test development has not generally kept pace with changes in frameworks that characterise the aptitude construct. Specifically, tests of working memory (WM) are absent from most existing batteries (except for HiLAB). Second, lack of open access to the batteries severely limits their use, impedes reproducibility and replicability, and constrains our understanding of validity and reliability. And although LLAMA is freely available, item level data are not. 
To address these issues, this project develops an openly accessible web-based battery, Test of Aptitude in Language Learning (TALL), based on the Stage Model (Skehan, 2019) and the P/E Model (Wen, 2019). As a multi-componential battery, TALL measures four facets of aptitude: declarative memory; phonemic coding ability; analytic ability; WM (specifically, phonological short-term memory and executive control capacity). Critically, to the best of our knowledge, it is the only aptitude battery designed to address two important potential confounds in aptitude research to date. (1) It was designed specifically for the L1 Chinese population, as the WM tests use stimuli in the participants' L1 to eliminate possible confounds from knowledge of English. (2) It has two counter-balanced parallel versions (aural and written) that allow us to consider modality effects when measuring aptitude. Additionally, it is administered on a website developed in Java, with data available for download, and scripts and materials openly accessible.
In this doctoral workshop, I will provide an overview of the design of TALL, and a validation schema at the item, subtest, and battery levels, providing the scoring inference, the generalization inference, and the explanatory inference, respectively. The preliminary validation results will be obtained from data from college-level participants (Nmin = 200). In particular, I would like the opportunity to discuss the extent to which, and how, the predictive and external validity of TALL should be validated to improve our understanding of the role of aptitude in learning itself. 
Bokander, L. & Bylund, E. (2019) Probing the internal validity of the LLAMA Language Aptitude Tests. Language Learning, 70(1), 11-47.
Skehan, P. (2019). Language aptitude implicates language and cognitive skills. In Wen, Skehan, Biedroń, Li, & Sparks (Eds.), Language Aptitude: Advancing Theory, Testing, Research and Practice (pp. 56–77). 
Wen, Z. (2019). Working memory as language aptitude: The Phonological/Executive Model. In Wen, Skehan, Biedroń, Li, & Sparks (Eds.), Language Aptitude: Advancing Theory, Testing, Research and Practice (pp. 187–214).