Conversational Analysis of Chat Room Talk PHD thesis by  Dr. Terrell Neuage  University of South Australia National Library of Australia. Thesis full text availalbe from the University of South Australia library

THESIShome ~ Abstract.html/pdf ~ Glossary.html/pdfIntroduction.html/pdf  ~ methodology.html/pdf  ~ literature review.html/pdfCase Study 1.html/pdf~ 2.html/pdf~ 3.html/pdf~  4.html/pdf~ 5.html/pdf~  6.html/pdf~  7.html/pdf~ discussion.html/pdf  ~ conclusion.html~ postscipt.html/pdf~ O*D*A*M.html/pdf~ Bibliography.html/pdf~  911~ thesis-complete.htm/~ Terrell Neuage Home Appendixes  1, 2, 3, 4, 5, 6, 7.  DATA ~ Case Study   1 ~ 2 ~ 3 ~ 4 ~ 5 ~ 6 ~ 7 ~ These links are from early notes and not the final edits which are in the published version available at the University of South Australia only. Not all links are active due to changing domains. Home page see /

Methodology Wednesday, 20 February 2002

Development of a protocol of a transcription methodology

Types of Chatrooms                                                                                                               

Grammatical patterns in chatroom 'talk' including use of abbreviations and emoticons

How roles change in a chatroom 'talk'

Patterns of interaction in chatroom 'talk'

How moods are established or/and changed in chatrooms

The use of abbreviations and emoticons in chatrooms


These hypothesis can not be answered using quantitative analysis, as there is not a way at this time to know who is in what chatroom. I discuss the problems associated with attempting to answer these hypothesis in the Methodology Chapter and suggest areas for further research.

No Terrell, you outline them here and discuss them in your methods chapter. These are problems you had to face and had to deal with to devise a method of analysing chat. Your method is a significant contribution to research but it doesn’t belong in your conclusions chapter. Your conclusions should summarise what you concluded as a result of your study.

Major theoretical studies have examined conversation as interaction between participants with conversation understood as spoken communication. One primary characteristic of conversation is that it is fully interactive - at least two people must participate in it, and they exchange messages in a real-time basis. Participants take turns in exchanging these messages, so conversation is fundamentally a sequential activity (Nofsinger 1991: p.3).<SPAN style="mso-spacerun: yes">  </SPAN>However, on-line sequential activity is rare.<SPAN style="mso-spacerun: yes">  </SPAN>Conversation is often similar to bumper cars in a side show amusement park.<SPAN style="mso-spacerun: yes">  </SPAN>Dialogue seemingly bumps and weaves often without any reason for its existence. There is a sense that participants are "thinking out loud". In a chat room, turn taking has to be isolated in order to assemble conversation into meaning.  What I will attempt to do with my methodology is to elaborate on the theories and methods of empirical research that already exist in both conversational analysis theories on the Internet and on Internet-based communities ­ such as diverse types of chatrooms. Finding how internal meaning is transmitting is a primary concern of chat room conversation. How are words or objects (using emoticons) linked to create a semantic chain to produce an identifiable and answerable sequence?

Chatrooms have limitations that conversations in which physical speech is produced do not have. Talk in chatroom is limited to short phrases. Rarely will there be more than several words written at a time by a 'speaker'. Looking at a sampling of a dozen Chatrooms and hundreds of entrances I found that there was an average of 7.08 words per turn. Within that sampling 25 percent of words consisted of two letters, and 20 percent consisted of three letter words. Eighty-three percent of words used in chatroom conversations were five letters or less. The way we will communicate will change and is now changing. As we are faced with more choices and more to do all the time communication will become more concise or the speaker will be left behind.

There are millions of chat rooms catering to all possible human interaction.  A majority of chat rooms become seemingly stuck in the ‘hello’ or ‘anyone want to chat privately’ categories.  The chat rooms I am analysing are rich in turn-taking and developed conversation.  This chapter on ‘storm’ a study in chat room linguistics during an emergency is my starting point in working with an on-line linguistic.

<![if !supportEmptyParas]><![endif]><o:p></o:p>The purpose of my methodology is to discover the structures of chat room language. Internet communication is rapid conversation.<SPAN style="mso-spacerun: yes">  </SPAN>It is rarely ‘frozen’ as it is when the chat is saved to examine.<SPAN style="mso-spacerun: yes">  By developing an "analytical framework" to study chatroom conversation, I will identify differences between casual conversation and information-seeking dialogues.  This first study, ‘Storm’, is mainly of the information-seeking dialogues framework.

<SPAN style="FONT-FAMILY: Arial">The methodology I propose to pursue for the textual analysis within this project is a mixture of several approaches to linguistic studies. As what I am proposing to do includes several fields of study, as shown below, I have to be clear at all times that what I am doing is at core a linguistic study. And within that framework discover the impact of text upon real events.  My approach to this study therefore, differs from a psychological or sociological approach to the use of language. The psychologist asks why we have conversation the way we do. Sociological conversation analysis asks us instead how we do the conversation. Linguists ask, "How is language structured to enable us to do conversation" (Eggins & Slade 1997, p.7). By extending this 3rd, linguistic approach into electronic interactions I can retain for my study a focus on evolving practices within a sphere still loosely considered textual rather than talk-based. In other words, I anticipate the possibility of being able to capture emergent conventional patterns of use within Internet chat behaviour, as my original contribution to this field of study.<o:p></o:p></SPAN><SPAN style="FONT-FAMILY: Arial"><![if !supportEmptyParas]>

In developing a transcription system[1] to accommodate and "capture" IRC multilogue, I will use symbols to indicate: interaction between participants, change of topic, and introduction. The example below is a generic format.  Each case study will have a different method for analysis of data.  This is because I am using different theories and looking for different interpretations of meanings in each chat room.

I will analyse chatrooms using

Reception and Reader - Response Theory and Reader Theory’ (Umberto Eco (1979, 1986, 1995), J. Kristeva (1980), Michael Payne (1993).) See Case Study 1

Speech Act Theory (Jurgen Habermas (1989), John Rogers Searle (1965, 1969, 1976), Deborah Schiffrin (1987), Terry Winograd (1986). See Case Study 2

Discourse Analysis (Norman Fairclough (1989, 1995), Bakhtin, See: Case Study 3

Conversation Analysis (Diana Slade and Suzanne Eggins (1997), Donald Allen and Rebecca Guy (1974), John Austin (1962), Erving Goffman (1959), H Sacks (1974), E. Schegloff (1974), Deborah Tannen (1989). See Case Study 4

Semiotics and Pragmatics (Chandler, Barthes, Halliday, Saussure, M. A. K. Halliday (1978), S.C. Levinson), Nofsinger (1991). See Case Study 5

Linguistic schools of thought: (See: Case Study 6

Computer Mediated Communication including: Electronic Communicated Analysis, Computational Linguistics and Text and Corpus Analysis (Charles Ess (1996), Michael Stubbs (1996) See Case Study 7

(note - *Verbal Messages

>>Information Theory - Claude Shannon & Warren Weaver

>>Meaning - I. A. Richards

>>Coordinated Management of Meaning - W. Barnett Pearce & Vernon Cronen

>>Symbolic Interactionism - George Herbert Mead

*Nonverbal Messages

>>Expectancy Violations Theory - Judee Burgoon

>>Semiotics - Roland Barthes



The individual methods for each study are at:

Case study 1 storm

Case study 2 Astrology 'chat'

Case study 3 General chat

Case study 4 ’Web3d computer modeling 'chat'

Case study 5 ‘Britney Spears Chat’

Case study 6 'baseball chat'

Case study 7


In the below example of interaction between participants I will indicate retrograde speech referencing, as "speakers" can only refer to what has already been said. For example, in the multilogue below, the text in 1 is not answered until 4. Indicating this interaction will be coded 4 1. For a new topic/thread the # symbol will indicate the change. For example 'speaker' in text 5 jumps into an already existing conversation and may be changing the topic - it will depend on what follows whether '1love's' change will begin a new thread or will be ignored. To indicate this change it will be coded #5. This will be demonstrated more in example III and IV. To indicate a speaker not speaking to a known participant, such as 5 'speaking' to 'curtis' who is not in the immediate conversation, I will write 5-?. Greetings to a new participant will be represented by *. The codes will be in brackets following the text. [ #]. Note that the numbers represent the line of text, not the speaker.


1. Janis> Through now I know we are part of the universal plan to exist on the third dimension, but why was there such a plan for us to exist in the first place. [#]

2. dammit>(Singapore) hi janis [*]

3. steven> hi janis, dammit! [*] Just wanted to dropped in home-- after splatter painting my consciousness throughout the multi-verse for eons, it is nice to be here! [ 3Ù 1]

4. steven> Janis, I see this no thing, some thing is like a pendulum/fulcrum swing. Tell me more about lexigrams--sounds fascinating! [4Ù 1]

5. 1love> curtis, thanks for your photo, [*] this mustard seed looks all golden to me! My photo is on its way, just got the pics back today. [#]

The above would be coded thus: #1, 3Ù 1, 4Ù 1, #5.

The above dialogue was take off of the 'Time Traveller' web page:

Here the "out of step" narrative of the multilogue is clear. An attempt for instance to schematise the interconnections of the 4 speakers would include retrograde as well as forward directions - and include some references not in the current "dialogue box" (or "multilogue" box). To show how contributors and readers manoeuvre within such a system of exchanges I will need to develop a protocol model similar to CA to diagnose speech and to find how readers and writers understand, interact and continue. There are several models to build upon but I will use the pluri-semantic model of Kress and van Leeuwen and O'Toole (cf. Kress and van Leeuwen 1990, 1996, O'Toole 1994) in Eggins and Slade's work (1997, p.49). The pluri-semantic model is outlined below, giving three main approaches to analysing casual conversation: ideational, interpersonal, textual.

Types of meaning


Examples: above Ex. I


Meanings about the world, representation of reality (eg. topics, subject matter)

Conversation, expressions; the universal plan - #1


Meanings about roles and relationships (eg. status, intimacy, contact, sharedness between interactants)

4Ù 1 share meanings

5-? Relationships undefined

2Ù 1 greetings/contact

3Ù 1,2 contact/greetings

3Ù 1 shared meanings through metaphysical 'talk'



Meanings about the message (eg. foregrounding/salience; types of cohesion)

1 positioning the conversation ideologically

3 continues metaphysical meaning of 1

5 breaks own conversation into two (re. Photos) by inserting text about mustard seed.

(Schema modified from Eggins and Slade 1997, p.49)

One of the areas I am interested in researching is how, within chatrooms the original discourse changes. I aim to isolate and analyse the 'departure points' from original topics. Does the person come into the chatroom with an alternative motive? Is the topic becoming boring and in need of shift? And who are the people who are speaking? Some people have a link to their 'homepage' which may contain more information about the person, but as Daniel Chandler says in his "Personal Home Pages and the Construction of Identities on the Web" ( .

..the created 'textual self' is how the author wishes others to see them. "The medium of web pages offers possibilities both for the 'presentation' and shaping of self which are shared either by text on paper or face-to-face interaction.

This suggests that the 'textual self' can present itself as a less constructed "reality" in the constructed exchange of On-line presentation. But whether 'textual selves' operate the same in chatrooms and IM as they do in one's homepage needs to be researched before a conclusion can be known. I hypothesize that people create a different 'textual self' for each environment they are in, and that we should not continue to regard all electronic textual practices as equal.

Like other areas of the Internet, chatrooms too have etiquette, and rules of cybersense are continuously evolving. Jill and Wayne Freeze point out in their book Introducing WebTV,

..that what is written is not always what is meant. A fair amount of meaning relies on inflection and body language. It is best to clarify a person's intentions before jumping to conclusions or getting defensive. (p. 135).

"Rules" are however already established in IRC - for instance, the convention that capitals imply shouting. Other, more subtle conventions also are developing, as well as abbreviated "talk" (see notes on 'abbreviations in chatrooms' 10).



An article on analyzing tape recordings. Time estimates for CHAT & CLAN " If the goal of the transcription is basically to "get the words right", I would figure about 15 hours of transcription per hour of..."

CHATROOM ABREVIATIONS – a much larger database of emoticons and abbreviations are saved at:

{{{{{{}}}}}} cyberhugs. {{{{{Terrell}}}}}}}

:) a smiley face denotes that you are joking or happy. there are many variations on this such as :-) ;-) :0 so keep smiling

:( a sad face. this too has variations of despair that can be added such as :(~~ for crying and :P~~ giving someone the raspberry

>:) at its mildest is someone who is mischievous and at its worst...a horny devil.

0:) an angel...

<s> is a tiny smile

<S> a huge smile

<g> is a grin and to make it a bigger one, use a capital g.

<eg> evil grin

<weg> wicked evil grin?

<vweg> very wicked evil grin for those people who are not faint of heart.

<bfg> big freaking grin


<giggle> <tee hee>

<gag> <choke> <hack> <cough>

LOL laughing out loud

lmao laughing my ass off

pita pain in the ass

btw by the way

fyi for your information

imo in my opinion

iyo in your opinion

brb be right back

bbiab be back in a bit

bbl be back later

afk away from keyboard

^5 high five and means you're congratulating someone on a comment they

[1] Another transcription method of Internet chat is shown at: