Project

Back to overview

Extending the LARA Learning-by-Reading Platform

English title Extending the LARA Learning-by-Reading Platform
Applicant Rayner Emmanuel
Number 204503
Funding scheme Project funding
Research institution Faculté de traduction et d'interprétation Traitement Informatique Multilingue Université de Genève
Institution of higher education University of Geneva - GE
Main discipline Applied linguistics
Start/End 01.02.2022 - 31.01.2023
Approved amount 66'549.00
Show all

Keywords (7)

low-resource languages; open source; evaluation; crowdsourcing; Computer-Assisted Language Learning; reading; sign language

Lay Summary (French)

Lead
LARA (Learning And Reading Assistant ; https://www.unige.ch/callector/lara) est une collaboration open source internationale organisée autour d'un ensemble d'outils gratuits conçus pour aider les gens à construire des textes multimodaux. Les textes LARA contiennent généralement de l'audio et de l'audio et des traductions intégrés et une concordance automatiquement construite. Des projets substantiels ont maintenant été réalisés dans plus d'une douzaine de langues, dont le français, l'allemand, l'italien, l'anglais, l'irlandais, le farsi, l'islandais, le vieux norrois, la langue des signes islandaise, le moyen haut-allemand, le barngarla (langue aborigène australienne), le polonais, le japonais et le mandarin.
Lay summary

L’étape actuelle du projet s'étend sur un an, commençant le 1er février 2022 et se terminant le 31 janvier 2023. Au cours de cette période, l'objectif principal sera de soutenir les nombreuses activités basées sur LARA actuellement en cours. On s'attendra notamment à privilégier les collaborations avec les groupes du Trinity College de Dublin (Irlande ; LARA pour l'enseignement de l'irlandais), de l'Institut Árni Magnússon d'études islandaises (Islande ; LARA pour la littérature islandaise et du vieux norrois), de la Ferdowsi University of Mashhad (Iran ; LARA pour soutenir l'enseignement du farsi comme langue étrangère), Adelaide University (Australie ; LARA pour les langues aborigènes australiennes) et Ruppin Academic Center (Israël ; LARA pour la poésie et les livres d'images pour enfants). Les projets couvrant plusieurs groupes incluent une collaboration pour créer des versions LARA de Le petit prince en plusieurs langues, un projet pour créer des pièces de théâtre LARA et un projet qui utilise des textes LARA pour comparer la qualité de l'audio humain et Text-To-Speech.

Direct link to Lay Summary Last update: 01.02.2022

Responsible applicant and co-applicants

Employees

Name Institute

Associated projects

Number Title Start Funding scheme
177065 A Crowdsourcing Platform for Spoken CALL Content 01.04.2018 COST (European Cooperation in Science and Technology)

Abstract

We propose a 12 month extension of project IZCOZ0/177075, centred around further development of the Learning And Reading Assistant (LARA). LARA, https://www.unige.ch/callector/lara/, is an open source platform designed to facilitate the process of converting texts into an online form which supports non-native readers. A LARA text typically includes integrated audio recordings, translations and an automatically generated concordance, with further options available. The platform consists of three separate layers: a core engine which performs the central conversion functions, a web portal which wraps them for easy use, and a social network, currently undergoing initial testing, which organises the material in a form that connects the members of the online LARA community.This community now contains projects spanning eight countries. High-profile examples include multimodal versions of a series of Farsi textbooks in Iran, which are already being used for teaching in a large language centre; resources to support classroom teaching of Irish in Ireland;high-quality digital versions of Old Norse epic poems used as set texts in schools in Iceland; also in Iceland, interactive documents for teaching sign language; learning materials in Australia, as part of a program to revive a dormant Aboriginal language; the investigation of crowdsourcing methods in Poland; and, in a multinational effort, online versions of plays with integrated audio and translations. Other multinational activities include several large evaluation exercises, one currently ongoing. We have been strongly supported by the enetCollect COST network, and LARA based content development projects have already attracted two rounds of funding in Iceland under national resource programs. An Australian proposal is currently under review for a multi-year project where LARA would play a central role in an Aboriginal language revival program. In general, due to its open source nature, LARA appears to be exceptionally well suited to small and underresourced languages, and fits well into the goals of the recently begun International Decade of Indigenous Languages.Concrete work in the proposed continuation project is divided up under four main headings: supporting ongoing community projects; empirical investigation of the effect of adding the social network layer; documentation, including completion of an ongoing PhD project and the production of an overview book; and the addition of a small amount of new functionality. Specific themes which cross-cut these activities are provision of better logging facilities, necessary to support free development of evaluation projects by community members, and support for various kinds of crowdsourcing functionalities mediated through the social network. An interesting and ambitious new type of functionality involves development of infrastructure to permit conversion of freely available LibriVox audiobooks to LARA form. This would reuse resources developed by the well-known LibriSpeech group, with whom we are currently in initial discussions, and if successful would greatly reduce the effort required to create high-quality LARA texts.The overall intention is to bring LARA to the point where the platform and consortium are in practice self-sustaining. Based on the encouraging progress to date, we believe we are now within a year or two achieving this goal, and the extra effort represented by the project we propose here may well be enough to make the difference.
-