Presentation of Adaptive Software at CLIL 2010 Conference

Text corpora and web services

for automated content generation

Rintse van der Werf

Edia – Education Technology, www.edia.nl/en

Ton Koenraad www.koenraad.info

Utrecht University of Applied Sciences, TELLConsult

Overview

• CLIL issues and technology

• Adaptive software

– Using text corpora

– Adaptive content generation

• The smart newsreader software

• Towards CLIL applications

A quick scan of issues,

concerns & needs in CLIL

• Windows on CLIL (20 country profiles, Maljers et al., 2007)

• The International CLIL Research Journal (ICRJ)

• Criteria for CLIL Learning Materials ((de Graaf et al., 2009; Mehisto, forthcoming)

• Introducing the CLIL-Pyramid (Meyer, forthcoming)

• Vienna CLIL Teacher Ed. Master theses

• Google searches

CLIL in EU context: some findings

• Diversity: target groups, aims, programmes, ..

• Growth (numbers, sectors)

• Need for customised materials

• Emergent CLIL methodology: scaffolded, but autonomous,

student-centred learning (March et al., 2007)

• More learning skills development needed

• Teacher availability + CLIL competences

• Inadequate productive skills: writing (Vollmer, 2008)

• Increased use of Internet based resources

• Limited publications on practice & research of ICT-use

Documented ICT use in CLIL

• Generic tools for materials development

• Additional resources : YouTube, websites, podcasts, wiki, blog

• Tools for scaffolding webbased input:

TEXTblender (POOLS –T Project)

• Tools for consulting and annotating video interviews:

Backbone Project

• Tasks involving Internet consultation:

WebQuest (Koenraad & Westhoff, 2003; Luzon,

2009)

• CMS platforms to:

- organize blended & distance CLIL learning: VLEs, ALI-CLIL

- community building and content sharing: CCN, e-CLIL, BEP

• More…?

It is time for

more passion

between CLIL and CALL:

TECLIL: Technology Enhanced CLIL?

CLIL Materials!

• [..] the availability of materials has been an ongoing issue in

Finnish CLIL. It is clearly difficult and time-consuming for

teachers to find suitable materials for content and language

teaching that would be in accordance with the national

curriculum and suitable for the students’ language level.

(Marsh et al., 2007)

• CLIL is currently gaining considerable momentum and it is

being integrated into curricula all across Europe. However,

there is still a lack of appropriate teaching materials and a

comprehensive and integrative CLIL methodology has yet to

be developed. (Meyer, forthcoming)

CLIL Materials Issues:Availability, quality, equal access (e.g. special needs), diversity

(content areas, target groups, language levels)

• Armenia, Belgium, Bulgaria, (lack)

• Czech Republic, Germany (adaptation needed)

• Estonia, (teacher made materials)

• France (localisation needed)

• Hungary (quality of translated content)

• Austria (content available locally, but copyright issues)

• Norway (lack of suitable textbooks, national curriculum)

• Poland (textbook import, translation)

• Slovakia (local adaptation & elaboration of imported textbooks)

• Spain (lack of resources, regional diversity)

Input Materials

• […]The (imported) textbook used was too difficult for pupils of average

and below-average ability.

“When the pupils have to tackle work on their own, they will not show any

progress unless they can fully comprehend what they are asked to do.

Also, if the pupils’ intrinsic motivation is low, providing books which have a

high level of prose difficulty is more likely to lead to non-comprehension

and frustration.” (Sollars, 1988)

• […] Words need to be understood and learned within the contextual

setting provided by the subject matter. This means that a basic level of

general English proficiency is not sufficient for successful content learning.

It seems that not enough is being done in the classroom in order to

ensure that learners grasp the relevant register.

Farrell and Ventura (1998); Farrugia (2003)

Adaptive personalised software

• Applications for (non CLIL)

– L1 Dutch, L2 Dutch, L2 German and L2 English

• Strongly rooted in scientific research

(vd Werf & Vermeer, 2008)

(Hootsen et al., 2007)

• Moving towards method integration

– Word lists, specific learning goals

• New possibilities for content integration

– In L1 and in L2

Comprehensible input

• not too difficult yet enough opportunities for

learning.

– SLA: Krashen: i+1

– Vygotsky: Zone of proximal development

• Operationalisation?

– Activate previous knowledge

– Nation, 1993: 90% of lemmas in a text known

– Textbook sequencing

Input for CLIL

• Linguistic and non linguistic

• Has a cognitive and language level

• Must be comprehensible, yet challenging

enough to provide opportunity for learning

• Focus on text comprehension

– Vocabulary size

– Knowledge of the ‘world’

Text Corpora

• As a source of textual (lesson) materials

– Representative

– Web As a Corpus

• As a source for linguistic analysis

– Frequency information

– Keyword analysis

– Part of Speech tagging

– Information analysis (clustering)

Adaptive personalised software

• Selection of textual materials

– Comprehensible, target word list

• Attention on relevant aspects

– Language and/or content

– Related to learning goals

• Adding help and guidance

– Web services such as TTS, online dictionaries

• Automated task generation

– Cloze tasks, dictate

Smart Newsreader application

• Web mining of news articles

– Dutch corpus size: 936000 texts,

– German corpus size: 235000 texts

– English corpus size: 195000 texts

• Selection

– Text coverage (% of known lemmas)

– Unknown words are learning goals

• Adding help with unknown words

– TTS, dictionary, contexts, morphology

Smart Newsreader application

• Generating tasks

– Cloze, drag and drop, dictate, open questions

• Monitor usage

– Words read, help asked, task results

• Give feedback

• Update model of the user

– Profficiencies, preferences, interests

Towards CLIL reader(s)

• Specific purpose text corpus

– Economics, history, osmosis, etc.

– Manual creations and web crawlers

• Text analysis

– Wordlists

• General vocabulary

• Subject terminology

• Academic words

• Text selection

– Comprehensible subject and language input

Adaptive Tools &

CLIL materials quality criteria & methodologies

• Rich, authentic, multimodal content input at appropiate level (i + 1)

• Scaffolded input provision (just in time help)

• Lexical approach (concepts in context)

• Academic Language Proficiency (focus on form: register features

e.g. collocations)

Reading – Writing integration (Loranc-Paszylk, 2009)

• Development of (language) learning skills (learner as researcher)

• Learner-centred, safe environment, learner autonomy

• Meaningful repetition, (immediate) feedback

Adaptive Tools & CLIL issues

• Assessment of learning

• Data collection for research

• Teacher education & development: personalised, self-access training of both language & register

and content terminology