Chat Logs as Data Source

1
Chat Logs as Data Source

• Saturday Nov. 22, 2003
• The 29th Annual JALT Conference
• Shizuoka, Japan
• AV Hall

2
Chat Logs?

• Chat logs are the text created when people
  chat. They are records of the online discourse
  between the participants.
• Chat is computer mediated communication in
  synchronous form.
• Chat is usually written in CGI or Java.
• Chat logs are text (.txt) files kept on the
  server.

3
Chat Logs

• Chat logs are kept in database format.
• They usually include the date and time of post,
  character name, channel, and message.
• These are all in delimited format.
• The chat log can easily be imported into database
  programs such as MS Excel, Access, or other
  similar databases.

4
Why?

• Chat logs are in delimited form.
• Logs can be globally searched.
• No transcription necessary.
• No overlaps.
• No incoherent (uncaught) speech.
• Genuine, communicative.

5
What is Chat

• When using chat, it needs to be defined.
• (Lets compare.)

6
Speech

• Speech is transitory.
• Speech gives little chance for reflection.
• Mistakes require repairs such as reiteration,
  explanations, or rephrasing.
• For data taking, speech requires transcription.
• Speech overlaps, and is often unintelligible.

7
Writing

• Writing is permanent.
• Writing gives sufficient time for processing.
• Mistakes can be eliminated without the reader
  ever knowing about them.
• Writing doesnt require transcription.
• Writing allows sufficient time for preparation.

8
(Asynchronous) Chat

• Bulletin Boards E-mail
• Similar in nature to writing.
• Allows global searching.
• Is easily reproduced and instantly disseminated.
• Can be transferred easily to other formats.

9
(Synchronous) Chat

• Communication is in near real-time.
• Mistakes require recasts, explanations.
• Allows little time for preparation.
• It is transitory for the interlocutors
  (Chatters).
• Allows little time for processing.
• Like Speech

10
Synchronous Chat

• Logs are permanent.
• More time to reflect.
• Doesnt overlap or become unintelligible.
• Encoding uses typing skills.
• Decoding uses reading skills.
• Not Speech

11
Synchronous Chat

• More time to reflect than speech.
• Encoding requires typing skills.
• Decoding requires reading skills.
• Record is permanent.
• Doesnt require transcription.
• Like Writing

12
Synchronous Chat

• Real-time communication.
• Interlocutors see mistakes.
• Little time for preparation.
• Little time for processing.
• Not Writing

13
Synchronous Chat

• Unique form of communication.
• Has unique advantages over other communication
  forms.
• Social constraints and programmed time limits
  allow little preparation.
• Close relation to interlanguage.
• Data is easily collected and analyzed.

14
Is It Valid?

• Not Writing.
• Not Speaking
• Unique form of Communication

15
Experiment

• Four University Speaking Classes
• Used chat extensively.
• Analyzed resulting chat logs.
• Compared them with field notes and current
  research.
• Noted similarities between classroom speech and
  chat data.
• Charted progress of related on-line quizzes.

16
Conclusion of Experiment

• Chat is unique.
• Chat logs CAN be a valid source of interlanguage
  data.
• Chat logs are easily analyzed.
• Definitions of chat genres are necessary.
• Guidelines are necessary for maintaining
  construct validity.

17
Definitions

• Three branches of chat
• Moderated Chat
• Unmoderated Chat
• Language Classroom Chat

18
Moderated Chat

• Breaks down in 3 genres.
• Interview Chat
• Information Chat
• Technical Chat

19
Interview Chat

• Requires a Host, or Moderator.
• Requires a special guest.
• Is large scale.
• Questions from the audience go through a separate
  channel to the moderator.
• Moderator poses interview questions to the guest,
  controls the protocol.
• Audience is anonymous, and has little chance to
  participate.
• Trend away from anonymity.

20
Information Chat

• Main guest usually acts as moderator.
• Purpose is similar to classrooms.
• Questions are posed directly.
• Usually very technical.
• Sessions are often posted for future reference.
• Audience is expected to participate, and may be
  called on.
• Rarely anonymous.
• Includes on-line distance learning sessions.

21
Technical Chat

• A member of the main guests often acts as loose
  moderator.
• Usually a technical group discussion.
• Audience may freely participate, but is not
  required to do so.
• Log is often posted, or conclusions posted to FAQ
  forum.
• Rarely anonymous.

22
Unmoderated Chat

• Two genres
• Service Chat
• Social Chat

23
Service Chat

• Usually assistance using or finding a product.
• Conversational.
• Rarely posted.
• Rarely fully anonymous.
• Includes peer tutoring, collaboration.

24
Social Chat

• Stereotypical chat session.
• Normally, but not always anonymous.
• Usually follows a theme or topic.
• Rarely if ever posted for future reference.
• Mainly social in nature, little substance.
• Protocols are agreed upon by the interlocutors.

25
Language Classroom Chat

• Meta-chat sessions.
• Can include any format moderated or
  un-moderated.
• Physical environment controlled and monitored.
• Main purpose is to learn a second language.
• Students expect to use formal grammar rules of
  target language.

26
Guidelines for Data Validity

• In an experiment, fully define the programs used,
  anecdotally and practically.
• This includes the chat sessions.
• Guard against cheating.
• For interlanguage data use only Language
  Classroom Chat session logs.

27
Cheating

• Using dictionaries or grammar helpers.
• Spoofing.
• Copy Pasting.
• Help from an expert.
• Unprocessed chunks.

28
Allowed and Encouraged

• Mistakes.
• Requests for help.
• Recasts, explanations.
• Participation from the entire class.
• Humor.
• Self-expression

29
Allowed but Not Encouraged

• Help from a friend at similar level.
• Trolling.
• Non-participation.

30
Technical Issues

• Chat rooms are usually written in CGI or Java
  languages.
• CGI programs reside on the server, not your
  computer.
• You must have CGI permissions.
• Java chat programs look better and are more
  functional.
• CGI chat programs are easier to create and
  maintain.
• Free chat rooms on sponsored servers cannot be
  logged.
• There are many good freeware chat programs.
• You will have to learn some programming, but it
  isnt necessary to be an expert.
• Do your homework before asking expert advice.

31
Thank you!

• Any questions?
• ????????!
• This presentation can be found at
• http//ww5.tiki.ne.jp/sorensen/personal/resume.ht
  m

32

• There are two kinds of language teachers.
• Those that use CALL
• - and those that will.
• (Commonly heard at CALL NSIG meetings)