DataDriven South Asian Language Learning

1
Data-Driven South Asian Language Learning

• SALRC Pedagogy Workshop
• June 8, 2005
• J. Scott Payne
• Penn State University
• jspayne_at_psu.edu

2
Corpus-Based Approaches to L2 Instruction

• Traditional Paradigm
• present gt practice gt produce
• Data-Driven Learning
• observe gt hypothesize gt experiment

3
What is a corpus?

• "A corpus is a body of text assembled according
  to explicit design criteria for a specific
  purpose" (Atkins Clear, 1992 p.5).
• John Sinclair (1991) on the development of the
  field of Corpus Linguistics
• "Thirty years ago when this research started it
  was considered impossible to process texts of
  several million words in length. Twenty years ago
  it was considered marginally possible but
  lunatic. Ten years ago it was considered quite
  possible, but still lunatic. Today it is very
  popular."

4
Comparing Genre

• Different genre of corpora
• Journalistic
• Conversational (formal, informal, etc.)
• Literary
• Scientific
• Academic
• Second language learner

5
Analyzing Corpora

• Qualitative techniques
• Concordances or keyword-in-context
• Quantitative techniques
• Frequencies analysis of individual words and
  collocations
• Lexical diversity
• Lexical density

6
What is Data-driven Learning (DDL)?

• Application of tools (concordancers) and
  techniques from corpus linguistics in the service
  of language learning.
• Concordances as a tool for developing
  instructional exercises.
• Places raw linguistic primary source material
  in the hands of learners - learners as
  researchers.
• Learners have the opportunity to discover
  language rules by themselves.

7
DDL Examples

• Tim Johns DDL website

8
Research on DDL

• Vocabulary acquisition
• Vocabulary acquisition improved through the use
  of concordances (Steven, 1991 Cobb, 1997).
• Horst and Cobb 2001 found that of four tools
  supplied to students for the learning and
  acquisition of vocabulary (traditional bilingual
  and monolingual dictionaries, online dictionaries
  and concordances) , after monolingual
  dictionaries, use of concordances was more
  indicative of learning gains.
• Writing instruction.
• Cobb 2004 used Lextutor to help students correct
  their own writing errors. Only 8 of students
  indicated that the concordance had helped them.

9
Corpus Tool for Data-Driven Learning

• KWICionary - http//conic.la.psu.edu/kwic/
• OCAT - http//conic.la.psu.edu/ocat/

10
Activity

• Construct a corpus in your language selecting
  texts from the Web.
• Explore the corpus using the concordance query.
• Generate at least one activity that you could use
  in the classroom this summer.

11
Learner Corpora

• Criteria for Learner Corpora
• Continuous stretches of discourse, not isolated
  sentences or words, containing both erroneous
  and correct uses of the language.
• Resulting from authentic activity classroom or
  naturalistic interactions.
• Explicit design criteria meta-data about the
  learners background, setting, level of
  proficiency, etc.

12
Learner Corpus Typology

• Monolingual ltgt Bilingual
• General ltgt Technical
• Synchronic ltgt Diachronic
• Written ltgt Spoken

13
Analyzing Learner Corpora

• Contrastive Interlanguage Analysis
• Compares NS and NNS language
• Can highlight features of non-nativeness in
  learner writing and speech (e.g. under and
  overrepresentation of words, phrases, and
  structures)

14
Corpus-Based Assessment of Language Development

• Goal to construct an evolving performance-based
  linguistic profile for individual learners.
• Elements of a learner linguistic profile
• detailed diagnostic analysis of linguistic
  features lexical inventory, morphology, syntax,
  and a variety of discourse properties including
  coherence and cohesion devices.
• It could encompass a comparison between learner
  performance and native speaker usage and a
  comparison of learner performance with other
  learners.
• allow for the introduction of a genuinely
  longitudinal approach to language assessment
  because it will enable teachers and researchers
  to track individual learner development over
  time for any relevant linguistic features.