Student Diversity

1
Student Diversity Academic Writing Project
(SDAW)

• Prof. Lucas Introna
• Lancaster University
• Dr. Edgar Whitley
• London School of Economics

2
Student Diversity and Academic Writing Project

• HEFCE-funded FDTL 5 project
• 2005-2008
• Collaboration between the London School of
  Economics and Lancaster University
• Research and Development

3
SDAW project structure

• Four sub projects
• A Country visits (India, Greece, China)
• B UK fieldwork
• C Detection Software
• D Dissemination of results and development of
  resources

4
SDAW project aims

• To inject timely and topical research results
  into the debate about the way international
  students are recruited, prepared and taught and
  how plagiarism can be deterred
• To develop evidence based resources to support
  the different stakeholders in dealing with
  plagiarism as an integral part of teaching and
  learning practice in a holistic manner

5
Plagiarism Detection SystemsPilot work with
Turnitin

• Student Diversity Academic Writing Project
  (SDAW)

6
The Turnitin Pilot project

• When / what / why
• Key questions
• Preliminary results
• Some conclusions and implications
• Next steps

7
When / what / why

• In Summer and Michaelmas term of 2005
• Overall objective
• to learn about the actual behaviour of Turnitin
  system through detailed systematic trials
• Why Turnitin?
• Used be 5000 institutions (12 million students
  and educators) worldwide. 
• 50,000 papers submitted per day  
• Used in over 80 countries
• Turnitin crawler has downloaded over 9.5 billion
  Internet pages and updates itself at a rate of 60
  million pages per day.

8
Key questions

• What is the actual level of coverage of Turnitin?
• How much must one change text before it become
  unrecognisable by Turnitin?
• When faced with paraphrasing difficult text would
  writing strategies of non-native speakers be more
  recognisable to Turnitin?

9
Actual level of coverage
Electronic Sources Used for the data collection

• Total of 103 fragments were submitted to
  Turnitin
• 47 were found (similarity index gt 25)
• 45.6 of fragments were found

10
Actual level of coverage
11
How much must one change text

• Resequencing
• Will always recognise
• Short fragments (lt 30 words)
• Where you change is very significant
• Long fragments (gt 30 words)
• Significant changes required

12
(No Transcript)
13
(No Transcript)
14
Paraphrasing difficult text / non-native
15
Some conclusions and implications

• Coverage of Turnitin database much less than
  expected
• Small fragments recognition unexpected
  (implication for non-native paraphrasing)
• Faced with difficult text non-native speakers
  will tend to use as is significant phrases from
  the source

16
Next steps

• Turnitin Experiments
• Large scale coverage experiment (automated?)
• Extending experiments with text changes
• Large scale paraphrasing experiment (50 100)
• Institutional research
• Interviews with staff using it
• Review of institutional policies