Computer support for second language learners

1
Computer support for second language learners
free text production-Initial Studies-

• O. Knutsson, T. Cerratto Pargman
• K. Severinson Eklundh
• Royal Institute of Technology
• Stockholm - Sweden

2
Outline

• Introduction
• Background
• Theoretical Framework
• User Study
• The computer program Granska
• Research questions
• Data collection and methods
• Preliminary findings
• Discussion

3
Introduction

• Interest in the use of computer support for
  learning Swedish as a second language
• Focus on the use of computer-language tools for
  writers who can fluently write and speak in their
  mother tongue
• Goals
• to study how writers develop their writing
  practices in the context of learning Swedish as a
  second language
• to contribute to improving the design of existing
  language tools for writing in learning contexts

4
Writing in the acquisition of second language

• Writing turns speech and language into objects of
  reflection and analysis (Vygotsky, 1962 Luria,
  1976)
• Far from transcribing speech, writing creates the
  categories in terms of which we become conscious
  of speech (Olson, 1995)

5
Language tools for second language writers

• Computer language programs supporting free text
  production available in Swedish have been
  developed for native speakers
• the grammar checker in Microsoft Word (Lingsoft)
• the research prototype Scarrie (Uppsala
  University)
• the prototype GRANSKA (Royal Institute of
  Technology)
• Most of the computer aided language learning
  programs available on the market rarely analyze
  learners written or spoken productions

6
Second-language learning processes

• The acquisition and development of a second
  language is regarded as a complex processes
  requiring the interplay of motivation, identity,
  context, culture, intellectual competence
  (Sjögren, 1996)
• Second-language learning is viewed as a
  combination of spontaneous, inductive learning
  with systematic, deductive learning strategies
  (Laurillard, 1993)

7
A developmental perspective on the use of
language tools

• Language tools are viewed as artifacts that
  become instruments through the writers activity
  (Rabardel, 1995)
• Language errors are a source for the
  understanding of how writers make sense and
  construct a new symbolic system (Scott, 2001)
• Written feedback is an important resource for the
  writers language understanding and construction
  of a new symbolic system (Cohen and Cavalcanti,
  1990)

8
Pilot study on the use of Granska in
second-language writing environments

• Aims
• to study how the grammar checker, Granska,
  should be adapted to second language writers
  needs
• to develop a method for assessing the use of
  Granska in a naturalistic environment

9
Granska - a Swedish Grammar Checker

• It provides different functions such as grammar
  checking and proofreading, linguistic editing
  functions, language rules and help system
• It supports detection, diagnosis and correction
  of language errors in the writers revision
  process
• It combines statistical and rule-based methods

10
(No Transcript)
11
Research questions

• Does Granska support second language writers
  revision process?
• What parts of Granska are most important to
  improve and develop further?
• Which are the research methods suitable for
  studying second language writers free text
  production?

12
Data collection
User Age Native language Language level Time of study No. of texts Length of writing Graded texts
A 34 Spanish Advanced 6 months 18 2240 words 4
B 37 Spanish Intermediate 4 months 16 244 words 4
C 34 German Advanced 2 months 1 190 words 1
13
Method

• Focus on free text production during the revision
  process
• Instructions to the users
• Use Granska whenever you want and when you think
  it will help you
• Save the original text and the final version
  revised with Granska
• Analysis of users judgment of Granskas alarms,
  detections, diagnoses and correction proposals
• Analysis of interviews with second language
  teachers

14
Example two versions of the same text for the
study of users actions

• Version 1
• Hon skulle komma hit och träffas oss för att
  prata om våra gemensamma intresse. Diagnosis
  Om våra syftar på intresse är det
  kongruensfelProposals vårt gemensamma
  intresse     våra gemensamma intressenVersion
  2
• Hon skulle komma hit och träffas oss för att
  prata om våra gemensamma intresser. Diagnosis
  Okänt ord Proposals intressen     intressera
  

15
Teachers views of errors and written feedback

• The type of errors depends much on the level of
  the language reached
• Common errors are syntactical errors, word
  order, verb inflection, agreement and use of
  prepositions
• Different approaches on written feedback
• Immediate written feedback and support for
  drafting processes could become useful for writers

16
Error type Detections Detections Diagnoses Diagnoses Corrections Corrections
Error type No. of judge-ments Mean value No. of judge-ments Mean value No. of judge-ments Mean value
Typographical 0 - 0 - 0 -
Orthographical 18 4,6 18 3,9 16 4,2
Morphosynt. 4 4,0 4 3,8 3 3,7
Syntactical 13 3,5 13 3,2 8 4,5
Lexical 0 - 0 - 0 -
Semantic 0 - 0 - 0 -
Pragmatic 0 - 0 - 0 -
Style 0 - 0 - 0 -
All errors 35 4,1 35 3,6 27 4,2
17
Preliminary findings cont

• Users repaired spelling errors without feedback
  from the program
• Users followed Granskas advice if correctional
  proposal was provided
• Users mentioned to be satisfied with the
  programs correction proposals
• Users could not understand some of the diagnoses
  presented when correction proposal was not
  provided

18
Discussion

• When should we start to adapt/redesign Granska
  from what we know from the users studies ?
• Which types of users should we focus on ?
• How should we improve methods for collecting and
  analyzing writers free text production?

19
Information about the project

• www.nada.kth.se/theory/projects/xcheck/
• www.nada.kth.se/theory/projects/granska/demo.html
• Welcome! to contact us
• tessy_at_nada.kth.se
• knutsson_at_nada.kth.se