The Faculty of Teacher Training at the Universitat de València hosted CASLL 2026 on 14–15 May — a new conference dedicated to corpus research on school learner language. Beth and Lisa were among the invited speakers.
Their presentation, Modelling coherence in L2 English at German schools: The role of cohesion and complexity, tackled a question that sits at the heart of writing instruction: what makes a text feel coherent to a reader? Using 163 argumentative essays from the YGLE corpus — roughly 26,000 words written by students in grades 10–12 across three German states — they measured cohesion, lexical complexity, and syntactic complexity through automated tools (TAACO, TAALES, TAASSC). Human raters assessed coherence holistically, and regression modelling tested which textual features actually predict those judgments.
The conference also featured a keynote from the MULTIWRITE project (Research Council of Norway), presenting longitudinal data on student writing across Norwegian, English, and third languages. A timely reminder that learner corpora are most powerful when they cross linguistic and disciplinary boundaries.