Predicting adult language learning success in its very initial stages.

Thumbnail Image
Issue Date
Journal Title
Journal ISSN
Volume Title
Adult language learning success in its very initial stages: the effect of learning condition and individual learner variables in three language tests across five L1s The VILLA project, Varieties of Initial Learners in Language Acquisition (Dimroth et al., 2013; VILLA Field Manual, 2022) managed to find a format in which the very initial stages of adult language acquisition could be investigated. The researchers recruited one single teacher, teaching a carefully devised curriculum for both learning conditions (form- vs. meaning-based) to all of the thoroughly selected participants with no prior knowledge of the language (Polish), thus ensuring uniformity in entry level, input, and exposure duration. In this longitudinal experiment (two weeks), 162 adult learners of five different project countries with five different L1s, recruited from universities of the Netherlands (Nijmegen), the UK (York), France (Paris), Germany (Osnabrück) and Italy (Pavia) took part in a Polish language course in which they were exposed to 14 hours of monolingual input in 10 sessions. End 2021, integrated databases became available (Villa Manual 2022), finally enabling to answer two general research questions: 1. Do the meaning- and form-based learning conditions give different language learning successes? 2. Can individual learner differences predict language learning success? We selected tests that had been administered at a minimum of two time intervals, and that covered different language domains: Word Recognition, Grammaticality Judgment and Phoneme Discrimination. We applied linear mixed effects regression. We found clear progress both in Word Recognition and Grammaticality Judgment, but no significant effect of learning condition. For Phoneme Discrimination, testing a phonological distinction present in Polish, but not in the L1s of the learners, we found no significant effects at all. The answer to research question 1 is a clear “no”. We had a large set of individual learner variables related to executive functions (Digit Span, Letter Number Sequencing, Flanker) , Perceptual Preference (Barsch), Language Aptitude (Llama B (Word Learning sills), Llama D (Phonological Recogntion) and Llama F (Grammatical Inferencing)), personality (NEO FFI-3 and ISALEM-97), and motivation (AMTB). We found that only Llama D, Llama F, Raven, and motivation were successful predictors for progress in two of the three language tests. None of these variables had any influence on the results in the Phoneme Discrimination, although several participants could discriminate the unknown sounds in question. Returning to our second research question, the answer is a qualified “yes”. It seems that although learner variables can have predictive value, they are not solely responsible for learning success. In the last part of our presentation, we will address possible theoretical implications of our results, taking into account potential differences between countries, but in particular the learnability of various linguistic elements and overall features, and the variability between language components. Dimroth, C., Rast, R., Starren, M., & Watorek, M. (2013). Methods for studying the acquisition of a new language under controlled input conditions: the VILLA project. Eurosla Yearbook, 13(1), 109–138. VILLA Field Manual (2022), team villa researchers. Wissenschaftliche Schriften der WWU Münster.
Faculteit der Letteren