DLLA Closing event

We take the opportunity of this closing event to present and discuss some of our methods for keylog analysis (session 1), learner difficulty detection and feedback generation (session 2), some possible uses of Large Language Models (session 3) and current  projects in the field (session 4) .

PROGRAMME

NOVEMBER 30 AM (room 720 and on zoom

https://u-paris.zoom.us/j/85257310750?pwd=TVd0Vm1UNWIybEpMQU1KOWFWQ08rUT09

ID : 852 5731 0750
password 611552)

SESSION 1: Keylog data analysis

9h30 welcome

10h 00  Helen  Yannakoudakis (King’s College, London) and Nicolas Ballier  (IUniversité Paris Cité) introduction

10h 15 Nicolas Ballier demo of the on-line data capture interface

10h 30 Georgios Velentzas (KCL / Kinhub / Sorbonne Université ) and DLLA Team : the DLLA dataset : Prolific data collection

Andrew Caines (Cambridge, ALTA) CEFR grading of the dataset

11h 00 Erin Pacquetet (UB) Pre-processing of keystroke analysis for linguistic data

11h30 Cyriel Mallart (Rennes 2) , Keylogs API : Reconstructing and measuring typing behaviour through bursts and revisions

Discussion :  Using Keylog to predict learner level ?

12h 30 LUNCH

SESSION 2 AI for learner difficulty detection and scaffolding/visualisation 

14h -15h00 poster session

Patrick Li (Rennes2 )  Exploring verb-noun collocations in learner English
Cyriel Mall: A new learner language data set for the study of English for Specific Purposes
Thomas: Exploring a new Grammatico-functional Type of Measure as part of a Language Learning Expert System
Marek Rei (Imperial College, tbc)

15h 00 Georgios Velentzas (KCL / Kinhub / Sorbonne Université ) Exploratory Visual Analysis of Keystrokes (demo)

15 h 30 Discussion : keylogs for learning analytics?

16h00  NB : the microsystem approach : profiling learners

16h30 dicussion

17h 00 end

 DECEMBER 1st  (room 720 and on zoom

https://u-paris.zoom.us/j/84452291431?pwd=eHkxZ2tNeC9rdzRINksvL1RBYlVzQT09

ID : 844 5229 1431
password : 958470)

SESSION 3  LLM FOR LANGUAGE LEARNING
discussants : Guillaume Wisniewski & Benoît Crabbé  (both Université Paris Cité, LLF)

9h00 Welcome

9h 30 introduction

9h 45 Guillaume Wisniewski  : What does an LLM do and what can be expected?

10h 15  Bernardo Stearns (NGUI) : Training language models to predict second language learners’ masked sentences

10h 45 discussion  : how to fine-tune LLMs to reproduce spelling errors ?

11 h 15 Nicolas Ballier  OpenAI’s whisper models with learner data : diagnosing segmental errors?

11h45 discussion
discussants : Guillaume Wisniewski & Benoît Crabbé  (both Université Paris Cité, LLF)

12h 30 LUNCH

SESSION 4 CURRENT PROJECTS AND NEXT STEPS: AI and Education for Modern Languages in Universities : next plans?

14h plenary talk: Thierry Olive (CNRS/ Poitiers)  : A psycholinguistic perspective on keylogs and AI

15h 00  Rita Burgo (KCL) Potential visualisation applications for keylogs analytics

15h30 discussion

16h00 Coffee break

16h30 Thomas Gaillat, Interfacing MOODLE CMS with data analytics microservices

17h00  Round table: DEEP LEARNING FOR LANGUAGE ASSESSMENT in Circle U and beyond

 We invite presentations of current current  projects in the field using AI for language learning.

18 h 00 closing