BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//NextToU//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:554759c2-ceaf-49b7-a385-ab10d34482fe@nexttoyou
DTSTAMP:20261105T155147Z
DTSTART:20261105T223000Z
DTEND:20261105T233000Z
SUMMARY:CS Colloquium: Olga Kellert on From Classroom Speech to Evidence-L
 inked AI Analytics of Language-Learning Interaction
LOCATION:ECCR 265
DESCRIPTION:Abstract: What happens after a learner makes an error—and how 
 can AI help researchers and educators examine the exchange that follows? 
 This talk presents two recently submitted studies with my co-authors on t
 ransforming authentic multilingual classroom recordings into reliable\, i
 nspectable analyses of language-learning interaction.\n\nThe main focus i
 s SLA-LENS\, a framework that combines insights from second-language acqu
 isition research with contextual large language model annotation and repr
 oducible analytical reporting. Applied to 40 online Spanish classroom tra
 nscripts containing more than 21\,000 sentence-level instances\, it ident
 ifies learner errors\, instructional feedback\, and learners’ subsequent 
 responses. A deterministic reporting layer summarizes participation\, fee
 dback\, and interaction patterns\, linking results to supporting transcri
 pt excerpts and timestamp. Comparisons of proprietary and open-weight mod
 els show that task-specific QLoRA adaptation improves Qwen3-8B’s exact an
 notation accuracy from 62.4% to 69.8% on an expert-reviewed transcript. O
 verall accuracy does not tell the whole story: models identify teachers’ 
 feedback more reliably than learners’ responses\, and some disagreements 
 arise because an exchange has more than one plausible interpretation. We 
 therefore examine errors that could affect the classroom analysis and lin
 k results to transcript examples\, allowing researchers and educators to 
 check the evidence and use it to inform teaching and curriculum \n\nhttp:
 //next-to-u.app/events/554759c2-ceaf-49b7-a385-ab10d34482fe
URL:http://next-to-u.app/events/554759c2-ceaf-49b7-a385-ab10d34482fe
END:VEVENT
END:VCALENDAR
