Conversational Analysis of Chat Room Talk PHD thesis by Dr. Terrell Neuage  University of South Australia National Library of Australia.

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 http://neuage.co


5.1 Findings of Case Studies 1 - 7

In the first instance my task within each research frame was to examine what each particular methodology could capture and describe within the talk-text as data. Only then could I begin to detect directions within these accumulating sets of features, and so to hypothesise that on-line chat had recurrent or characteristic behaviours and selective techniques, which, while varying across the types of chat sites examined, tended towards the establishment of recognizable “on-line chat” linguistic strategies. By summarizing the most explicit findings in each study, I can now move to compare the seven studies, adding where appropriate observations from five supplementary chatroom studies[1], to show features common to all text-based chat, and generalisable as the “core” discursive modes of Internet chat.

Despite their often incommensurable focus, the range of the theoretical methods used for analysis revealed particular communication features common to all chatrooms. Most of these features are not part of person-to-person off-line talk, and many appear unique to text-based electronic dialogue - although there is evidence that some of these behaviours occur in related CMC-delivery formats, such as SMS.

Returning to the five assumptions, drawn from the CMC literature and from personal experience of IRC, posed at the beginning of the methodology section (3.2), it is now possible to test the Case Study findings against these, and so to construct a series of propositions on the nature of on-line chat: 

·        That language used in chat rooms is more deliberate and calculated than the predominantly “informal” styles might suggest.

·        That conversation within Chatrooms demands a highly sensitized “reading” of texted-talk gambits from participants.

·        That “chat” does not differ from natural conversation in certain key aspects, but does so in others. 

·        That observational study of chatroom conversation can capture adaptations to conversational behaviours.

·        That such work gives a better understanding of how, and why, chatrooms are an important area in which to extend current conversational research theory.

Each case study had three components useful in bringing about such conclusions for chatroom analysis.

Firstly, the linguistic theory and its associated methodology identified key aspects relating to how each text-based set of chat data “worked”.

Secondly, each case study identified features of conversation that were unique to both text-based chatrooms, and to the varying types and functions of such spaces.

Thirdly each case study allowed for the analysis of recurring or “typical” chatroom behaviours, demonstrating elements of communicative activity specific to the theory driving that particular case study. In other words, both general and specialised features were pursued in each case study.

 The primary discoveries in each case study together provided a map of IRC, in both general and specific terms, across a broad spectrum of exemplar behaviours, at least during the sample period, and most likely beyond.

5.1.1 Case Study 1

Case Study One based its analysis on Reader-response theory to show that in on-line chat, both the person writing and the one (or many) reading are co-language-meaning creators. Chatrooms were revealed as an active reading environment where the “reader is left with everything to do…” (Sartre, 1949, p. 176).  In order to engage in conversation the “speaker-writer” first needs to be a “listener-reader”. Yet, as with all Reader-Response research, chat-texts captured for this study illustrated ongoing tensions for users, in relation to the issue of “closure”, or certainty in interpretation. What is left open in chatrooms – more so than in person-to-person conversation - is what later Reader-Response commentators called “preferred readings”: techniques whereby texts are arranged to position readers to receive and interpret them in certain ways which optimize selected understandings and suppress others.  Such texts may construct within themselves “an inscribed reader”, or such a figure and its attendant roles may emerge in “interpretative communities” (Chandler, 2001). But are such positionings found in the “texted” talk of IRC and its user-groups?

Using Reader-Response theory to examine chat in a community of users checking progress of an extreme weather-alert emergency, I found that there are two moments of “reading” that a chat participant carries out in seeking to understand meaning within a chatroom, even before beginning to read the actual utterances of the other chatters. In person-to-person conversation early “readings” of an interlocutor, taken even before we listen to what he or she says, involve viewing the person, their appearance, their posture, body language and the environment  (see Richmond and McCroskey, 1995; Ong, 1993; Goffman, 1981).  Similar work is clearly undertaken in on-line chat.

In chatrooms, firstly, the title of the chatroom is read. Case Study One showed that chatters carried on conversations reflective of the chatroom title, Hurricane Floyd. In other Case Studies with clearly designated topic-related titles I found the same reading techniques used.  Speakers tended to converse about the topic established by the chatroom title. In chatrooms the reader’s response fits the chatroom milieu.  A new utterance may begin a new thread, but there too the response is dependent on the reading. For example in Case Study One turn 107, <SWMPTHNG> inquires <YOU AINT TALKING ABOUT MEX ROOFERS ARE YOU?> in assumed response to turn 99 <EMT-Calvin>: <folks need to be careful for con artest after the storm>. This reading is however still on the same topic of the storm as a thread alongside, which talks about the storm itself – it merely illustrates a different “reading” of the topic. There are indeed very few threads during this conversation that are not directly on  Hurricane Floyd. The chart  below shows that 254 of the utterances in this chatroom are directly on the storm, while 14 turns are about whether Mexican roofers will become involved with rebuilding after the storm seven; are interpersonal  (for example, <your last name wouldn't be Graham would it>); and a small number “drift” from the storm topic onto comments not immediately about the storm, although arguably bearing on the participant’s semi-panicked reaction to it, as well as to their performance within the chat exchange: <VIAGRA AND PRUNE JUICE....DON'T KNOW IF I'M COMING OR GOING.....>, or  <ankash> stating <I gotta go get some Xanax.>. Such lines are not uncommon during even focused and topic-specific chat, and reveal from their “theming” to both topic and interchange relations, the varied “reading” work of participants.



Number of turns in thread

Storm thread



Mexican thread



Personal thread

Turn189 <guest-Beau> Calvin, your last name wouldn't be Graham would it


Chocolate thread

Turn15 <mahmoo> brb.......gotta go get me some chocolate





Reader-response theory takes us further however than just the recognition that topic controls most of the dominant conversational thread-construction. Here, I found that the “writerly-writer” or actively constructing text-talker who initiates a conversational thread, and the “writerly-reader” who responds, are able to move the chat into new avenues, not simply responding in topic-compliant ways to developing conversations, but demonstrating especially “open” and “active” strategies of initiating text and responding to it. The talk remains topic centred, yet works to focus and refocus threads around certain aspects or themes of a topic. This is not just information provision, but creative exchange build around information sharing.

Chat entrants anticipate certain content and behaviours, focused around the chatroom title – but also display tendencies towards adapting rapidly as topic focus shifts and new threads develop, and even a capacity to shift off topic, especially into personalized referential chat. One of the features of reader-response theory as I am using it in chatrooms is thus that it shows how a reader brings certain assumptions to a text, based on the interpretive strategies he/she has brought to a particular community, from other social-cultural contexts (see Gass, Neu, and Joyce, 1995; Blum-Kulka, Kasper, Gabriele, 1989; Rheingold, 1994; Turkle, 1995). The racial tone in Case Study One, displayed toward Mexican roofers, is an example of this. Reader-Response analysis thus reveals inside chat the sort of active, meaning-generating participants considered central in postmodern consumer culture (Lury, 1996; Castells, 1997, 2000). Even where the topic-shifts and socio-cultural attitudes may be directed to conservative or reactionary positions, the claims on reflexive use of communicative technologies and transformational interventions on communicative texts demonstrate Castells’ hypothesis, that the new communicative technologies are necessary to the “project identity” strategies of the postmodern condition. IRC becomes not a trivial pastime, but a key location for social and cultural formation.

How important is the particular chatroom context for the reader-writer interpretive relation?

It is the title of the chatroom that I suggest lures a participant to a particular chatroom. In Case Study One it was the topic of Hurricane Floyd. In Case Study Seven it is baseball and in Case Study Three the title of the chatroom indicates that chat will focus around the pop idol Britney Spears – although in this case, as the analysis suggests, talk focused more into a Britney Spears form of style culture than into direct discussion of the ostensible topic.  It appears then that despite the title as indicator, the chatter has to deal with multiple frames of interpretation, assessing the motivations and attitudes of others in the room.  When in turn 105 of Case Study One <SWMPTHNG> asks <YOU AINT TALKING ABOUT MEX ROOFERS ARE YOU?> the question indicates a moment of direct consensus - checking. <SWMPTHNG> picks up a hint in an earlier posting that there may be an opportunity to redevelop a current thread, and intervenes to “take the floor” in CA terms, in a powerful bid to redirect conversation. Here “context” is both shifting – from hurricane alert information, to discussion of ethnic tension – and not shifting, since <SWMPTHNG> in making this move is assuming that he or she is already culturally contextualised: conversing with a group of like-minded non-Hispanic Americans, who will share his or her views on “Mex roofers”. The “ain’t”, with its appeal to a colloquial repertoire, helps establish that cultural context, and indicates not only a chat entry which has “read” a cultural framing in earlier postings, but which re-inserts its interpretation of that framing, hoping to evoke response in kind.

The chatroom as context appears then to pre-position its users to expect and enact certain behaviours, values and topics. But since this appears to be only partially established through the title and topic selections, chatters also display complex techniques for both signalling and reading back rather less directly expressed aspects of the social and cultural framings brought to the chat. Context is generated in both chat space and real space – and these may or may not align. To this extent it becomes necessary to assess the contribution of the technologisation of chat to its cultural contextual framings, and to take up the findings of Case Study 2.

5.1.2 Case Study 2

Case Study Two examines on-line chat as a form of Computer-Mediated Communication (CMC), with all the special features and characteristics this implies. Computers do not replace but supplement communication. Despite the many obvious influences of the technologisation of on-line talk, communication remains dependant on both the sender of the message and the receiver. Even bots: those elements of on-line practice pre-generated as software and used to automate some on-line messaging functions, are scripted inside the communicative conventions of their language community – and sometimes even of their specialised chatspace.

The many tools available for CMC research conventionally divide the research objects into either asynchronous CMC (e-mails; mailbases; network groups; annotatable webpages; databases and discussion boards) or synchronous CMC (chatrooms and computer-conferencing) – although future studies might well address this division from the perspectives established in studies such as this. While the “liveness” of synchronous chat enables application of such analytical methods as CA, the use of script in “chat” still places interesting limits around the act of communication, and links even the immediacy of IRC to the more stable and enduring CMC forms. Since Computer-Mediated Communication (CMC) is used in business, non-profit organizations, education, and entertainment as well as for personal use, better understandings of how each format works as a communicative act, and of how each suits its wide range of uses, might assist in future selections and development of the various formats, for specialist use. However, as this study has suggested, CMC at this stage still lacks established and specific methodologies to analyse chatroom talk. While this thesis has used several conversational analytical theories, such as Speech Act Theory and Conversational Analysis, as a lens to examine the data in CMC, it has also uncovered in a preliminary way many limitations for analysis, as techniques developed for real-world talk are transferred into electronic forms of communication. Until CMC research moves beyond its current emphasis on pragmatic and developmental studies of user applications, and begins to examine instead the practices of those users in observational, descriptive and analytical ways, “how to” introductions to CMC formats will remain largely at the level of technical glossaries. The most common use of CMC research currently is surveying students and instructors (see Romiszowsk, 1996; Harasim, Hiltz, Teles, & Turoff, 1995; Mason, 1992; Rice, 1990) and tracking e-business supported work coordination (Bowers and Churcher, 1988). CMC is however beginning to be used as a method, as well as a tool, for researching on-line conversation (see Cicognani, 1996, 1997, 1998, 2000; Parrish, 2000; Rheingold, 1993, 1994, 2000; Vallis, 1999, 2001; Turkle, 1982, 1984, 1995, 1996) – but such studies to date work within broad sociological or social-psychological modes. 

To some extent the impacts of CMC on Internet chat are obvious to every user.

Synchronous CMC has its own particular set of difficulties, as I have shown in Case Study Two. Multiple threads of discussion become difficult to follow. Slow Internet connection can mean that the speed of reading and responding cannot be maintained. This results in discussion losing its focus and side discussions (threads) developing. Sometimes participants may simply be slow typists. The result is that what is written is often a response to something written many turns earlier.

Three terms, “gap”, “lapse” and “pause” are used to refer to silences in CA[2]. In chatrooms however, there will never be silences in the proper sense of the word, let alone with the specificity and distinguishability of CA analysis.   If there are silences in real time, the text will simply scroll together to cover these spaces. The CMC technologisation masks what is, in CA terms, significant rupture in “talk”.  Because of the threaded nature of the arrival of chat postings, chat users learn to bridge and to braid: to cross between postings, to reconstruct postings into reciprocal turns. As yet, there is no means of assessing how extended, or how complex, such bridgings and braidings might become – or of registering or measuring their impact on subsequent “replies”. By “breaking open” the technologisation of online turns as I do in many case studies, it becomes possible to examine the space between a person’s turn, and the next time the same person has a turn. I have called the distance between the two turns a “lag” or the distance between speech events of a speaker in a chat situation; a pause between one utterance and another. The unviability of such “lags”, together with the changed perspectives showing up in consequent entries, shows that there is a strong “writerly” form of “reading” in between such frames (see CS 1.2). In face-to-face conversation a conversational lapse or pause can be equated to a listening phase of conversation (see Sacks, 1992). In chat rooms this is a reading phase; interpretive, reconstructive, and wholly significant in the chat process. Without consideration of the lag times, as well as of the intervening utterances, it is impossible to see how much interpretive work is occurring. In de-threaded sequence postings seem incommensurable. Several unconnected themes can develop – and only by consideration of both the time taken in achieving these changed frames, and in shifting focus as new threads intersect and gain attention, can we make sense of the whole contribution. A CA methodology therefore, with its primary focus on relatively immediate conversational responses, even within multilogue circumstances, will need adaptation when dealing with IRC conditions. 

It is also important to locate techniques which will allow analysis of the differences in communicative responses between various Internet communication devices. In discussion groups and e-mails people observably take more time and care with what they write, and are therefore not as immediate in their communication as in Instant Messenger (IM) or chatroom conversations. Users of discussion groups and e-mail may use a spell/grammar check, and plan more consciously before posting their text. There is for instance a more textual format with discussion groups. But while Instant Messenger and chatrooms appear at first sight to be less disciplined and more varied, with the relative spontaneity of casual interchange ignoring many more formal communicative conventions, analysis has shown complex patterns of interpretive and pre-dispositional structuring under way. Messages from the Hurricane Floyd Messages Board, for instance appear more developed textually than the storm-related chatroom utterances – but is this an absolute, or a relative judgement? While IRC postings are far less grammatically formal, they remain as communicatively active and complex.

It is of course possible to postulate that, in the absence of directly reciprocating co-locutors, postings must address an unknown and general audience, in their quest for the specific addressee – and thus the more formalized and “public” mode of expression. In an Instant Messenger chatroom, the contrary is true. Interlocutors – most often established acquaintances, or at least those who are able to establish cultural commonality within the immediate communicative context – form responsive exchanges through their readings of informal, yet nevertheless complex and sophisticated talk-texting repertoires.

I approached this case study with two questions related to Computer-mediated communication: “Do computers change conversation” and “Are Instant Messenger chatrooms closer to off-line-person-to-person conversation than dialogue in a multivoiced chatroom?”  It has certainly become obvious that computers do change conversation, and especially in relation to the suppression of paralinguistic cues, direct address carried by gaze or gesture, tonal emphasis … all of those techniques used in “live” communication to manage the conversational relation. While we have found many emerging CMC techniques being used to replace these physical features, and noted the extraordinary creativity and pace of application in many cases, the informality of the new repertoire: its constitution within practice and its lack of a tailored analytical method, mean that CMC has not yet delivered all of its secrets. Nor can we anticipate that users will cease their creative transformations of the mixed-mode of “texted-talk” into these and other new communicative forms. Already it has become obvious that while CMC has produced and still produces new talk techniques, there is no monolithic regulatory influence being exerted. Practices differ – between chat spaces, between chat participants – even at different moments within a particular chat sequence, as talk-topics shift emphasis, and behaviours adapt. CMC itself has already spun into many different formats, and the talk-texting and speech relations within each have also differentiated. Some patterns appear to cross between CMC technologising practices in different formats. For instance, as with the chat in Case Study One where multithreads branch out from the primary topic of the storm, multiple chat-focus threads are also present in Instant Messenger conversations analysed. 

Yet in Instant Messenger or any two-person-only chatroom there is more opportunity for an organized and familiar turn-taking within communication, and therefore a more immediately meaningful exchange, than in a multi-person chatroom. So how then might the multi-person communicative repertoires of IRC be examined, to assess how participants “manage” the complexities of their flows of talk? Which tools can be used to assess techniques in use by IRC users, to overcome problems posed by CMC technologisation?

5.1.3 Case Study 3

In Case Study Three, using semiotic and pragmatic analysis as my tools of investigation of on-line chat, I particularly wanted to uncover not just how “talk” is accomplished in a chatroom, but how far chatroom “talk” generally may be said to include a broader than usual repertoire of representation, working to “manage” talk relation problems as outlined above, and to compensate the loss of off-line conversational cues. Mihai Nadin (1977) claims that the computer is in itself a semiotic machine, as it is at core a machine that can be programmed to manipulate symbols. Using computers themselves as semiotic generators has an aesthetic appeal to users, because semiotic codes change over time and provide new meanings to old ideas. This seems interestingly close to the sorts of marked creativity the IRC and IM users in particular display in the case studies for this research – although the continuity of these creative “solutions” will communicational problems on-line, with strategies and talk/texting techniques evolved in off-line conversation and reading-writing practices, reduces the implied suggestions that it is the CMC technologisation, and not human communicative ingenuity, which drives these changes.  While users take up and work with some of the special codes and even coding styles CMC systems provide, both the machines and the users develop inside a broader social and cultural context, and source their various communicative pre-dispositions there.

In this case study I focused on the most obvious of the CMC elements of creativity, exploring how the use of non-word representation: emoticons and abbreviations, as well as the “identity” sign-tags or the usernames of the chatters, influenced the turn-takings of the chat-talk (see Crystal 2001; Rivera 2002). 

I chose a chatroom named after a celebrity to firstly discover whether usernames, their “identity” sign-tags, would be reflective of the title of the chatroom. In this case study on “Britney Spears Chat” one chatter did indeed identify as a Britney fan: <baby_britney1>. This identification with the chat-title is consistent with what I have found in the other chatrooms in this thesis, such as in Case Study One, Hurricane Floyd, where there was the username <IMFLOYD>. In Case Study Four on astrology participants used the names “astrochat”, <AquarianBlue>, <TheGods> and <Night-Goddess_>; in Case Study Six, “web 3d animation” there were  <web3dADM> and < Web3DCEO)> and in Case Study Seven, “baseball chat” <MLB-LADY> (major league baseball). Therefore it is evident that usernames can be directly associated with the name-directed topic of the chatroom. When the dialogue is read from the postings of these specific users it is clear that each chatter is indeed interested in the topic of the chatroom:

 <AquarianBlue> in Case Study Four;

10). <AquarianBlue> Nicole 528 is gemini

<web3dADM> in Case Study Six;

10) <web3dADM> just got the Cult3D folks to agree to show up on March 3

<MLBLADY> in Case Study Seven;

6. <MLBLADY>     no clev fan but like wright

But in each of these chatrooms there are also participants, as we saw in each study, identifying against or outside the title-topic convention; contributing postings off-topic; playing with textual form rather than following content threads – even resisting efforts to bring them back on topic. And both within and off topic, we have seen intense moments of creative communicative play, frequently directed more towards the maintenance of communicative relations than to focused engagement with talk topics.

Case Study Two, let us note, centred on inquiry into whether the “playfulness” of on-line chat is a CMC specific impulse. In face-to-face conversation it is clear that people also use an array of semiotic communicative cues: intonation, physical gestures, facial expressions - but with CMC communication semiotic play is restricted to lines of text on a screen as an expressive marker  (Stone, 1995a, p.93) as well as such “characterising” elements as semantically-layered usernames, expressive emoticons or colour selections, and added sound. Semiotic analysis thus enables this study to move beyond a purely linguistic base into examination of the graphical and expressive modes used to compensate, and maybe beyond that, to create meaning in new ways, within the new “conversational” spaces of the chatroom - and particularly so in a chatroom of saturating expressiveness within identity work, as is the case with Britney chat.

In Case Study Three to fully explore this drive to identity performance and exploration, to find out how users extend the actual communicative range of the “language” or coding system used, it was first necessary to examine which communicative functions were actually in use in the Britney Spears chatroom, and to reveal which are dominant and recurrent.

Firstly, it was obvious in this chatroom that chatters employed usernames as signs to give others clues about their identity – or at least about their “preferred identity”, or particular identification with a Britney community. In person-to-person conversation the clues that are given as aspects of identity are personal – indeed, physical. On-line, these are replaced by the sorts of identity markers which demark off-line social or cultural status:  one’s employment or educational level for instance. 

Here, in keeping with the Britney world, user tags are about image and “claiming”, or the image that one wishes to have represent one’s status within the particular social context of the Britney chat group. Each asserts either a relational claim, or one’s desirability as a relational being: <Mickey_P_IsMine>, <JeRz-BaByGurL>, <Pretty_Jennifer>, <baby_britney1>, <IM_2_MUCH_4U>, <AnGeL_GlRL>, <Luvable_gurl15>, <buttercup20031> and <guest-hotgirlz>. These usernames suggest that the chatters, if not actually young girls, are at least identified with a popular teen culture of physicality and cuteness. In real-life <Luvable_gurl15> could be a 58 year old male, but if so he is entirely conversant with the codes and values of the Britney culture – even down to the assertiveness of the orthography: the post-feminist/netchick “gurl” replacing the conventional – and less powerful – “girl”.

Secondly, the title of the chatroom identifies the chatters as interested in the celebrity icon, Britney Spears. The chatroom title alone can provide information on the identity of a participant; for example, in a chatroom such as “Iraq4u”. An adolescent chatroom such as this one is likely to focus discussion on aspects of personal self, as users construct identity around the image and stylized behaviours represented in their idol. A comparison table with a computer software discussion chat shows this to be true in the Britney Spears room. And yet there are distinguishing features beyond the level of topic as well.  Abbreviations were used more extensively; suggesting that adolescent play over identity is also enacted within talk-texting strategies. 

Emoticons too serve a purpose beyond just the saving of time. They are also a marker of informality, and so an “antilanguage”, in Halliday’s sense, indicating a special subcultural group identity, and used to show who is familiar or “up-to-date” with the latest language being used. Of the seven case studies, I have found the highest incidence of abbreviations (30%) and emoticons (6%) in the Britney Spears chatroom (see http://www.geocities.com/picture_poems/thesis/tables.htm for a statistical comparison of the seven chatrooms). In fact the abbreviation for laughing-out-loud “lol” was used fifteen times. In this chatroom frequency counts of specific language forms are indeed revelatory. There were 294 words used within the collected data corpus, with the personal pronoun “I” used the most frequently, (18 times) and “lol” used the next most frequently (15 times). In one sequence “lol” is used nine times in 20 turns, which is more frequent than in any other chatroom examined in this study.  Another form of laughing-out-loud “LMAO” (Laughing my ass off) was used five times.

Firstly then, chatroom semiotics show the specialist communicative skill-level of the participator and whether he or she is in the right communicative arena to continue to be an accepted part of the chat community. Yet identity work of this kind in the Britney Spears chatroom is limited to the user name and the textual input of the chatter. By contrast, in face-to-face conversation, forms of identification are much more extensive and include cues which can reveal personal identity, national identity, occupational identity, corporate identity, gender identity and even religious identity (see Berger, 1998). So the talk-texting and linguistic creativity of these young chatters must achieve high levels of sophistication in order to convey all of the information needed to assert a “Britney” self, and yet remain a distinctive and desirable co-locutor in the “flattened” yet still competitive space of the chatroom. One dimension of chat which seems to become suppressed in these conditions is that of extended reciprocal conversation – those longer threads of debate, information exchange or narrative, which appear in some other chat spaces and cultures. Here, while such narratives of experience for example do exist, they are constantly interrupted by the “social recognition” postings of greetings and farewells, and reactive-expressive turns, working less to cement sociality than to maintain affective role within the chat relation.

Having established such high degrees of symbolic or creative-linguistic play, it becomes important with this chat culture to examine more carefully how this specific talk-texting repertoire works.  Pragmatics as a lens of conversational analysis in chatrooms (Ayer, 1968; Pierce, 1980) can reveal a socially embedded reading of chat “talk”. Pragmatics helps to focus on how the various communicative items in chatrooms; emoticons, abbreviations and misspelled words as well as chat utterance sentence structures (CUSS) are used within an on-line linguistic society. Pragmatics in chatrooms starts from the observation that people use on-line language to accomplish certain kinds of acts, broadly known as speech acts (Speech Act Theory is discussed in Case Study Four). Studies by Simeon J. Yates (1996) have shown that the language used in interactive speech in chatrooms more closely resembles spoken than written language, especially in the interpersonal respect (including use of personal pronouns). As I have shown, in Britney Spears chat, Table 8 - http://www.geocities.com/picture_poems/thesis/table8.htm (also on the CD) “I” has been used 18 times in the chat, the most used word in the whole chat. 

Writing (or text-talking) back to a previous utterance in a synchronous conversational situation in chatrooms leads to a pragmatic re-contextualization of the use of the sorts of double-loaded semiotic expression discussed in Case Study Three. It is how the signs are read which provides meaning, and entices, or provokes, other participants to either continue building an utterance into a thread, or begin a new thread – including responses to its graphic or creative-abbreviation load.  In Case Study Three there are several utterances that do not become threads, as they evoke no comment on them. For example neither of the following utterances have a response.

23. <baby_britney1> do any guys wanna chat?


27.<SluGGie> need to fix my hair..


Despite the direct question/invitation in posting 23, and the focus on a Britney-culture preoccupation with physical appearance in posting 27, neither turn is answered. The sorts of creative play with chat-semiotic loadings which we have seen above appear more likely to evoke reciprocal posting, when otherwise powerful conversational and communicative strategies such as direct invitation or topic and contextual focus, do not. Even those postings which access and reproduce the contextual “antilanguage” or specialist codes, with the conventional attitudinal and behavioural signifiers in place, do not always succeed in chat. In these next two turns <Mickey_P_IsMine> similarly receives no response - but responds to him or her self in turn 64.

56. <Mickey_P_IsMine> Ahh i got a retest tomrrow mi failin math lol..and i think science

64.  <Mickey_P_IsMine> which i duno how im failin science


The casual texting, including colloquialism (“dunno”), spelling lapses “tomrrow “, and “mi” for “im” = “I’M”) – even the “lol” abbreviation – code into the established styles of group talk – yet seemingly without sufficient creativity to gain notice. While responding to abbreviations and emoticons and colloquial forms and specialized lexical terms shows a commonality of understanding amongst those who are chatting, this appears not enough in itself to command a reply.   Commonality is clearly indicated when  <Paul665> in turn 44 asks <Jen> to give details on his or her self, and it is evident that to evoke a response  <Paul665> must assume that Jen knows the abbreviation “asl”.

44. <Paul665> Ok Jenn asl

  <Pretty_Jennifer> responds:

51. <Pretty_Jennifer> 15/f/fl u?


But while we can clearly see that here the codes are exchanged in perfect reciprocity, what we cannot do is calculate with certainty why this exchange succeeds, while others fail. The gambit is not as directive as in <baby_britney1>’s direct question in posting 23, so that we are left with an interesting possibility that the direct question works less effectively in this chat context than the coded-abbreviated “asl” convention: perhaps a signal of <Paul665>’s chat-credentials and comparative “cool” – while <baby-britney1> may be showing too much real-world social desperation and push. But it is impossible to be certain. Maybe chatters were attending to other surrounding threads as posting 23 arrived. It is at such points that textual analysis, no matter how multi-layered, begins to fail, and only ethnographic or observational work can succeed.

5.1.4 Case Study 4

Since Case Study Three therefore raises the question of whether the conversation in each chatroom varies in its focus in relation to talk techniques, and not just in topic focus, this study moves to consider which talk forms are evident in chat, and whether variability in given chat spaces can be detected – and perhaps even predicted, from the “chat community” present. Case Study Four used Speech Act Theory to identify dominant types of speech activity in a single chat space. While IRC chat makes application of Speech Act Theory difficult, because of the indeterminacy of the “response”, it is still possible to categorise postings within the speech act repertoire, and, where threaded exchanges are evident, to evaluate the success or “felicity conditions” of an utterance. It remains difficult to assess  how much of the intentional load of a chat utterance is carried by para-linguistic elements such as emoticons or abbreviations, codings shown as of immense communicative significance in previous case studies. Given the frequency of use and rapid assimilation of these elements into on-line communication in various media, it is important to attempt at least a preliminary investigation of their “speech act” role.

Direct Speech Acts

In chat there are clear examples of direct speech acts being deployed, and in quite conventional ways:

Speech Act






conveys information; is true or false

(Case Study Four) 11) <Nicole528> im a Gemini

(Case Study One) 10) <guest-MoreheadCityNC> NO she's near 10th & Gville Blvd




elicits information

(Case Study Four) 2) dingo42 nicole wahts your sign ??

(Case Study Four) 17) <AquarianBlue> your meeting her judy? when?

(Case Study Four) 32) <Night-Goddess_> anyone cool in here?

(Case Study - 911) 182) Brazilian report: some one know any new about the manhattan situation ???

Orders and Requests


causes others to behave in certain ways

(Case Study Five) 47) <scud4> bwitched stop scrollin in here

(Case Study One) 123) <Zardiw> smptthing................go back to your SWAMP


Direct speech acts that use performative verbs to accomplish their ends expand the three basic types shown: statements, requests and commands (as shown below).


Case Study One

37. <EMT Calvin> well folks im signing off here


Case Study Six

49. <Brian> r u talking about blaxxun and shout3d implimentations or something else

Orders and Requests

See the CD “911.doc”

296. <MissMaca> Brazillian Report: Iknow it was a building %&#%head. Give up on the %&#%ing nuke's ok!!!>.

Indirect Speech Acts

Indirect forms in chat are dominated by a generalized activation formation, which masquerades as a question addressed to the entire chat community:

Case Study One

74. <guestTom> does anyone know where floyd isnow>.

Case Study One

125. <guest kodiak> does anyone know why UNCC has not closed>.

Case Study One

162. <guestEZGuest367> Anyone know if I should worry about daughter in west NC?>.

The form has even evolved its own abbreviation:

Appendix “911” 370.

<England> n e one know of other active new york chat rooms?>.

The first four postings are clearly in the form of questions, but equally clearly are not inquiries about issues the chatter can anticipate will be answered by an expert “knower”. Thus the speech act is in itself indirect, as we can see by examining possible answers. Most of the time, the answer “yes, I do” to any of these four questions would be an uncooperative response. The normal answers we would expect in real life talk would be “Yes, the Weather Channel tells us that Hurricane Floyd is passing over North Carolina now”; <UNCC is closed because of the storm>; <if your daughter is in the eye of the storm you should be worried>; <another active New York chatroom is at http://www.superglobe.com/chat/>. Because of the anonymity of the chat situation, each response depends upon what could be called a “validation” format: the use of an indirect statement or reported speech from another context: “The weather channel tells us that…” A simple “yes” answer that responded to the literal meaning would usually be taken for an uncooperative answer in actual social life. For example “Yes, I do”, would be heard as “Yes, I do, but I'm not necessarily going to tell you where the storm is, why UNCC is closed or the location of other active chatrooms in NY”. So the five examples above function as indirect questions, more accurately coded as “I want you to tell me where the storm is now”, “I would like to know whether UCC is closed yet”, or “Please tell me of some other New York chatrooms so that I can move to them” and the chatroom participants are clearly able to interpret this function, and respond appropriately. In other words, despite the added indirection of chat speech act formation, chat continues. But this means that very complex speech act relations are concealed beneath the quick-form exchanges of IRC – across a range of chat communities. Indirect speech acts appear to be in heavy use.

The key question for this Case Study and this chatroom

“What is a successful speech act in a chatroom?” thus appears to require consideration of the more than usual loadings of indirect speech acts inside a non-physical and multilogue talk community.

Austin and Searle claim that the speech act is the basic unit of meaning and force, or the most basic linguistic entity, with both a constative and a performative dimension. They both accept that there are illocutionary acts and perlocutionary acts, using Speech Act Theory as their theoretical foundation and analysing the data by message length, distribution, message links, and interaction. Speech Act Theory is based on the notion that what people say is consistent with what they do (Howell-Richardson; Mellar, 1996). Such a definition indicates that we should examine those zones in which chat “unravels” some of the regulatory functions hypothesized in speech act theory. Distribution roles, or those aspects of speech working to direct talk relations and to control its performative dimensions, are problematic within the generalized speech relations of chat: one explanation of the sorts of indirect strategies outlined above – and maybe of the retreat into saturating expressives and relational work.

In part this indeterminacy which bedevils speech act analysis in chat rests in the technologisation and “de-threading” of the format. Speech Act Theory cannot categorise all utterances in a chatroom, with certainty – and it may be that the confusion and chaos that new users so frequently report of the chat experience relates to this indeterminacy, in relation to off-line talk. Yet at the same time regular chat users do manage their talk successfully.

Speech Act Theory can be used to examine features common to all chatrooms.  In particular it can help establish interconnections within the threads of conversation. Unlike face-to-face conversation, where a person appears to respond to the most recent statement in a conversation, in a chatroom the utterance can be a continuation of someone else’s utterance - or it can be on a new topic, with the hope that someone else may join in. The example below shows three unrelated utterances, but all are either continuations of a thread or the initiation of a new thread:

30)  <judythejedi> i don't think so..she's bringing amtrack down maybe

31) <Nicole528> whats your sign dingo?

32) <Night-Goddess_> anyone cool in here?


Because of the technologisation of chat there are no markers to segregate or “direct” this conversational traffic. Chat participants must then de-code the speech acts, and re-connect threads into logically sequential strands. Since posting 30 relates to an earlier posting, only those participants already threaded into that particular chat will respond – unless of course a new chatter asks directly “You don’t think what? She who? Amtrak down to where? Why only ‘maybe’”?  Since such a response would be an interruption of an implied co-locutor relation, it is unlikely to occur. Posting 31 creates a similar “directedness”, signing it with the user name “dingo” – the sole participant invited to reply. So it is no surprise that of these three consecutive postings, it is 32, the generalized and indirect question/invitation form, which succeeds. Following <Night-Goddess_>’s utterance <anyone cool in here?> a thread develops that plays across the issue of  whether anyone is “cool” in this room – and incidentally provides a possible answer to the role of posting 34 from <AquarianBlue>.

32)     <Night-Goddess_> anyone cool in here?

33)<judythejedi> hi night

34)/\32 <AquarianBlue> hmmmmmmm

35)/\32 <judythejedi>everyone is cool here

36)/\32 <Nicole528> is cool lol

37)<poopaloo> 10ty judy

38)/\32 <Nicole528> is cold too

39)<sara4u> I LOVE YOU TO MUCH.......ACARD

40)<jijirika>is back

41) <tazdevil144> cool


For this speech act to be completed there needs to be an understanding of what <Night-Goddess_> means by “being cool”. The speech community within the room chooses in interesting ways to respond by playing across the semantics of the term “cool” – yet in doing so, indicates an understanding of the indirectness of her speech act strategy. As <Nicole528> and <poopaloo> evaluate and reward the claim from <judythejedi> that all the chatters in this space are cool, and <tazdevil> extends the game by using the term to express pleasure that <jijirika> has rejoined the chat, each understands not only the “surface” codes, or display techniques which sign “cool” chat expertise: “lol”, and “10 ty”, but also the indirection of <Night-Goddess>’s speech act. This is not a directed question. As its “anyone” address formula shows, it is an invitation to talk. But specifically, in its address to not only a chat community, but to a known and familiar group (note <judythejedi>’s diminutive tag-name response: “night”) it creates a speech act which is less a general question than an assertion of communality. In effect, it says something like: “Hello to all my old friends: I’m ready to be as active in chat as usual” – and those chat friends react entirely appropriately.  Responses demonstrate “cool”, in chat terms, with a mix of community affirmation:

<judythejedi> everyone is cool here>

appreciation of the communality:

<Nicole528> is cool lol>

and the sort of metatextual play across chat conventions which establishes the cachet of cool on-line:

<Nicole528> is cold too>

No surprise then that the thread is continued for several more turns before a new thread is begun. The original utterance serves not to elicit specific answers, but to evoke the sorts of talk which on-line chat promotes, and which is distinctive to its form: reflexive, linguistically aware, communally directed, generalized and inclusive/exclusive, fast-paced, and multi-threaded: 

49) \32 10c. <Night-Goddess_> I is not cool

50) \49 5l. <judythejedi> yup

51) \49 6j. <Nicole528> really

52) \4910d. <Night-Goddess_> I is awsome

53) \496k. <Nicole528> yes your cool

54) \465m. <judythejedi> lol..i know prncess

55) \476l. <Nicole528> cool dingo

56) \521c. <gina2b> coolfool

Is there then sufficient evidence to assert that in its Speech Acts, on-line chat is predominantly relational – working more on its communal elements through generalization of its modalities, than on its performative or illocutionary acts? To test this requires assessment of chat in a strongly topic-directed chatroom – one in which we might anticipate task and topic oriented talk. Case Study 5 takes up the analysis of chat in an Astrology chatroom, in which many chatters appear to already know one another – therefore appearing less reliant on self-assertion or community formation.

5.1.5 Case Study 5

 If there is a preponderance of relational talk-texting in chat rooms, by examining a chatroom with a predominance of markedly short turn-taking sequences and a clear and consistently central foucs on topic, it may be possible to discover whether even in the rapidly scrolling conversation of on-line chat, there is enough time and appropriate “speech act” work establishing a communication community amongst the chatters present.

Talk in text-based chat is as fleeting as its off-line equivalent. Text disappears as it scrolls by. The participant gets one opportunity to read the text, after which time it cannot be retrieved – at least not without time out for back-scrolling – during which period postings continue to amass. This capacity I have called “fleeting text”.  On-line fleeting text affects discursive connectiveness. There is a counter-intuitive distinction here between talk and text. Conventional spoken language is also dynamic, fleeting, and irreversible communication, but printed language breaks the strictures of time and leads to permanence. The two together in an on-line environment contain elements of both – what has been said can be “revisited”, as long as the chatroom is showing previous turn takings. My data cannot show evidence that users do check back to re-establish threads, but the co-presence of postings onscreen, even while fleeting and constantly mobile, does encourage longer consideration than in talk. 

Thread-framing is in itself a major phenomenon in chatrooms. A posting appears to “begin” and “end” because it arrives on the receivers’ screens inside an individual text-box.  These framed pieces of conversation are of course not necessarily sequential. Threads twist around, stop and start, and several may arrive at one time, in a seemingly chaotic fashion. What then is the relationship between the seeming coherence of a single chat utterance, and its equally contained surrounding utterances?

We have already seen that the apparent commensurability of utterances, each framed in the same spatial convention, is an illusion. Immediately consecutive utterances are often unrelated, or at least out of sequence – and many remain so. Further, because this form of visual framing is the only contribution to the communicative regulation of texted-talk by its technologisation, users themselves must work instead at the level of language alone – including of course both verbal and visual elements – to construct meaningful communication. 

At the linguistic level the “threading” which constructs meaningful conversational exchanges across and between these individual and flattening visual frames also must read back possibilities for response. It is this form of “framing” which gives a starting and finishing point to a thread, and turns it from an artificial sequence of random utterances to a meaningful conversation. Since there are no visual codings contributed by the CMC technologisation to mark a new or ending thread, that decision too must be made by the chat participants; read back from the speech act possibilities. Curiously, in many cases the originator of a thread is also the last “voice” seen in that particular thread. In the example below, <Night-Goddess_> begins a new thread by asking whether there is <anyone cool in here?>.  The topic is also ended by <Night-Goddess_> 20 turns later, with the comment: <I is awesome>.

32) <Night-Goddess_> anyone cool in here?

49) <Night-Goddess_> I is not cool

52) <Night-Goddess_> I is awsome


Because this topic had centred so clearly upon the word “cool”, this transformation – “cool” becomes “awesome” – ends the potential for wordplay, and so terminates the frame. But to sense this termination chat participants must be able to “read” and respond beyond the level of conversational turn-taking exchange – the CA level. By reading speech act intent in utterances, and seeing <Night-Goddess> “switch off” the topic cue at this point, collocutors can indeed note a frame termination – and they move on accordingly.

The initial framing of a thread can thus determine – or at least work towards determining – its success and duration.  But in the case above, as already noted there is a particularly consensual group in communication. This community of astrology followers appears to be regular collocutors on-line, and know one another’s behaviours. How far then is this, the cooperative communication of a friendship group, as opposed to a specific communicative behaviour of on-line communities generally; a feature of “chat”, rather than of this one example of “astrochat”?

One way to examine this is to check for deliberate interventions: “policing” of chat posts. If there is hostility shown in a chatroom, or as shown in Case Study One, an attitude such as racism, (in this case towards Mexican roofers) will other speakers contribute to the thread in like manner, supportively, as in the astrochat sequence? Here there is clear evidence that such threads can be very deliberately de-railed, and comments such as <SWMPTHNG>’s stopped by others.  A different speaker can and will end a thread, indicating a multi-chatter frame (see Tannen, 1998; Bays, 2000). Since to do so they must however also “read” the frame – understand the intent of the utterance – the termination/transformation intervention still acts as evidence for the power of talk-text framing. So clear is the framing intent (or re-framing intent) of some postings to some collocutors, that they move to end a posting – or at least, to re-direct it. And indeed, without such framing a thread could continue indefinitely. Framing is what completes the thought in chatroom discourse but it is also what enables groups to maintain focus. How then does each participant enact these interventions and responses within a given frame? What additional problems for analysis of chat exchange does its online practice present?                                                                                                                                                                                                                                                                                                                                                                                               

5.1.6 Case Study 6

Using conversation analysis (CA) in chatrooms helped me discover how communication on-line regulates its exchanges. While the “capturing” of data is different in chatrooms from that used to research face-to-face conversation there are similarities in the analysis process. Traditionally, CA researchers audio record a session and discuss from a printed readout “what happened” in the conversational exchanges. In the example below from such a taped session[3] the time between turns and the pauses in the conversation are noted – not an element that can be considered in on-line chat, or at least not in those chatrooms which do not mark the time of arrival of each utterance – and even then, given the packet-switching technology, this does not reflect the times of entry for a given posting. Some aspects conventionally of communicative import in CA are therefore not available for analysis in chat. In CA for instance most work is done with two or three people speaking. In the example below two people are having a phone conversation. This one-on-one speech relation, or its close approximation within a small group, has contributed many of the techniques and features of CA method.

To an extent, the features identified by CA in small-group or dyadic talk relations can also describe chatroom interactions. Conversational analysis of chatroom talk shows for instance examples of adjacency pairs and turn-taking conventions common in CA-analyses of natural talk. But both the capacity for multilogue and the technologisation of the talk, through text and through CMC, create new complexities inside the talk relations. One primary difference, as this case study and others have shown, is the interjection of conversation before a thought is complete, due to the tendency to use the enter button “mid utterance”. A second disctinctive difference arises with the often lengthy periods between utterances, filled with other streams of talk. Offline “natural” conversation offers talk techniques not possible in online chat. In examples A and B below we see clear indications of turn-taking, and the development of a conversation. In A however there are interruptions (for example in turn 45), impossible in chatroom turn-taking.


A CA transcription from tape recording

B Web 3D Chat  on CD at 6a.doc


Utterances are mostly complete turns in chatrooms, with the only breakage in a particular utterance being made by the user at the time of the utterance – for only if they press the enter button does the utterance become broken. In the chatroom turns 21-24 below (column B) <Leonard> makes two utterances that are different thoughts, but because they are entered sequentially without anyone making an utterance between the two thoughts <web3dADM> is left to answer them both, as different thoughts, sequentially after <Leonard>’s entrances.

21) <Leonard> Anyone used Xeena?

22)  <Leonard> 3D just arrived today

23) <web3dADM>   no it's on my list

24) <web3dADM>   ahhh great Len

In a face-to-face conversation one would assume that <web3dADM> would respond to <Leonard>’s question, <Anyone used Xeena?> with the utterance <no it's on my list> and then to <Leonard>’s <3D just arrived today> with <ahhh great Len>, ordering the conversation differently:

21) <Leonard> Anyone used Xeena?

23) <web3dADM> no it's on my list

22)  <Leonard> 3D just arrived today

24) <web3dADM>   ahhh great Len

If in fact utterances 21 and 22 had been offered in sequence in a natural conversation, it is also likely that <web3dADM> would reverse the response sequence, offering his expressive and evaluative response before his explanation  – in effect replying to 22 before 21:

21) <Leonard> Anyone used Xeena?

22) <Leonard> 3D just arrived today

24) <web3dADM>   ahhh great Len

23) <web3dADM>   [no] it's on my list [too]

Online however, <web3dADM> could have been typing in <no it's on my list> at the same time as <Leonard> was typing in <3D just arrived today> - or even before, since we do not know the relative distances travelled through the system, or the traffic-flow conditions encountered by the packet-switching .[4]

According to conversation analysis, turn-taking is integral to the formation of any interpersonal exchange. Online however, the conventions of turns are very much modified. Chat participants appear conversant and comfortable with the new regulatory demands. Unless lurking, the participants in chatrooms demonstrate their knowledge of the particular chat conventions of the chat-site they are visiting in order to be accepted or rejected by others in the chatroom.

The signalling of one’s status as an insider is for instance especially important in establishing dominance. In the chatroom I used for this case study with its expert topic of computer animation, it is clear that <web3dADM> is the leader or moderator, not only because of the abbreviation for administrator (ADM) behind the <web3d> part of the username, but because <web3dADM> provides answers to questions people ask in the chatroom regarding the chatroom itself. The status of this participant is thus marked in various ways, but key among them is this specificity of interrelational role – a feature marked by particular forms of  turn-taking as identified in CA, but also by thje chat-specific conditions of a combined general and individualised set of response-relations.

The underpinnings of CA, sequential organization, turn-taking and repair, and how they can depict interactional competence, are therefore useful in reading chatroom talk. However, the circumstances of chatroom technologisation demand adaptations to CA protocols, to enable analysis of conversational relations occurring in de-threaded sequences. Unlike face-to-face conversation the sequential organization of a given chat exchange needs to be separated from what else is being enacted in the chatroom. The isolating of pairs in the chat is difficult if there are many people chatting and the text is scrolling at a rapid rate.  In finding adjacent pairs in Case Study One for instance the conversation had to be re-threaded.  What is revealed below is that there is a turn-taking strategy present between <lookout4110> and <Werblessed>, but each utterance has several turns in between.









Who is in Wilm. right now?




Im 50 Miles west of Wilm.




How ya holding up Werblessed?




So far just strong wind gusts and lots of rain.. Over 8 inches so far..




Have the winds been strong?




Gusts up to 60-65 so far its starting to pick up a bit.. Only gonna get stronger Between now and midnite

The first number in the “between utterances” column is the number of turns since the previous utterance was addressed, and the second number is the number of turns since the last utterance by the same speaker. The complexity of the posting relation is apparent. After these three sets of turn-takings <lookout4110> and <Werblessed> no longer interact directly. <lookout4110> contributes more utterances, concluding at turn 164, and <Werblessed>’s final utterance in this segment is at turn 180. In other words, given the multiple threads available for response in on-line chat, threads form and reform, as participants shift focus. But the degree to which such shifts are driven by the complexities of the multilogue is hard to evaluate – another feature which CA is unable to address, and which may require a more ethnographic inquiry to assess.

CA is however able to help the online chat analyst consider some aspects of conversational breakdown – for instance, repair, a standard part of normal conversation. Natural conversation is rich in examples of breakdown – a feature which CA analysts often find disruptive to other programs of their analysis:

When we consider spontaneous speech (particularly conversation) any clear and obvious division into intonational-groups is not so apparent because of the broken nature of much spontaneous speech, including as it does hesitation, repetitions, false starts, incomplete sentences, and sentences involving a grammatical caesura in their middle (Cruttenden, 1986, pg. 36).

In chatrooms, where utterances are mostly posted complete, this experience of breakdown at first sight seems less of a problem. But chat-repairs do come about, due to two primary causes. The first is introduced when a word is typed incorrectly – for instance, when <IroquoisPrncess> says <hey Judy did a get my car in the link thingy>. While “car” is a proper word, it is wrongly entered when “card” is intended and confuses the meaning, since interlocutor <judythejedi> does not associate the word “car” with the utterance-topic, leaving <IroquoisPrncess> to correct the error. Here the error is text and CMC related: clearly a typing error, and a feature which in natural conversation would be corrected immediately. In speech it would be enacted as a mispronunciation, or a mishearing – its conncetion probably cued by a quizzical glance or facial frown.  Here the interlocutor, <judythe jedi>, directly addresses the need for repair. In texted talk, there is no other cue for repair.The second repair error however is less techno-conversational, than CMC technological. Owing to pressing the enter key early, dividing his utterance, <Leonard> leaves a curious suspension in his exchange with <brian>. Has <brian> pre-empted a reply in advance of all the information, because he senses that the utterance object introduced by “this” must be “spring”? Does <Leonard> enter “spring” while <brian> is entering his own utterance, or because he thinks if <brian> has all the information he may change his response? Because we have no information available on the timing of the utterances we are unable to analyse the interaction further – an interesting example of chat’s technologisation defying CA principles on repair.

From Case Study Four

From Case Study Six

57) <IroquoisPrncess> hey Judy did a get my car inthe link thingy

63) <judythejedi> car in the link?

66) <IroquoisPrncess> card

40) <Leonard> I will be offereing it on-line through Digital University sometime this

41) <brian> can't make it

42) <Leonard>  spring


Are there then instances of chat which require more than the sorts of extended CA repertoires discussed here, for examination of the full range of utterance behaviours and conversational techniques? Are the chat participants examined above displaying both interesting instances of the language-use pressures of chat, and conscious attempts to redress these? Are there other techniques of talk or text analysis which can help both identify and explain some of these communicative behaviours? One issue raised in CA work on chat is the need for a more finely-focused examination of word-selection and word-ordering in utterances – and especially in such self-conscious moments as those occurring around instances of repair. In a final pass over the chat-room communicative experience, this study used current approaches to grammatical analysis, to assess how far chat already displays ways to use and/or depart from standard text or talk grammar conventions.

5.1.7 Case Study 7

This case study examined baseball chat, a talk-community likely to use high degrees of informality in grammatical formations, to assess whether the functioning of grammar in chatroom communication could be shown to be the same as, or different to, that evident in text or talk. Do common grammatical conventions – such as word order, sentence structure, question formation, hold up in on-line chat? Do baseball-chatters on-line use the same specialist formations as their off-line brethren? Are there any new constructions evident?

Language in a chatroom certainly proved to be altered by its users, both deliberately and by mistake. Formal sentence structure conventions become less evident, as abbreviation and graphic elements arise to meet the speed-entry demands of the chat technology and its new communicative ethos.  Compound forms arise, with the informality of spoken language, but enacted in the sorts of textual play and creativity otherwise seen in communicative genres such as poetry, or advertising. The grammar of chatrooms, if it is done intentionally, is developing a highly sophisticated form of prose that is semantically and semiotically  innovative and daring.

Below, <CathyTrix-guest> in turn 108 creates the utterance <2blech>. Such combinations of numerals and letters have no place or “utterability” in spoken conversation – yet in this chatroom, at this moment, inside this thread, the utterance communicates.  The “2” refers to an earlier request for chatters to press the “3” key if they like the New York Yankees baseball team. <CathyTrix-guest> emphases his or her dislike of the Yankees by pressing a different key from the “3” suggested, confirming it with the comment: “blech” - not conventionally a meaningful word, but one used colloquially as an onomatopoeic representation of the act of vomiting.  The turn thus communicates something like “I don’t like the Yankees, they make me sick, I would only score them at a rate of 2”. The economy, the creativity and the expressiveness of the utterance overturn the conventions of a more formal sentence construction, without losing communicative power. But at the same time, they demonstrate a linguistic and grammatical formatting not available or possible in speaking about baseball.




if you like the yanks press 3








got it




















2I hate the Yankees




don't have a 3
















yankees s-ck




im removing that # now




you wish


Similar concision in chat utterances operates as both efficiency forced by the required typing speeds, and a stylistic marker of on-line competency. In turn 77 of this chat <MLB-LADY> enters a question: “dd any see the atanta score”. A formally grammatical rendering would produce the form: “did anyone see the Atlanta score?” While the third spelling error: atanta for Atlanta, is likely to be a simple typing error, the suppression of the vowel from “did” and the lack of capitalisation for the proper noun “Atlanta” are both conventions of on-line use.

Similar effects are achieved by the use of single letters or numerals in place of whole words: u – you, 4 – for, r –are, c – see, 2 – to.  In posting 128 of Case Study Seven <BLUERHINO11> refers to <dhch96>by using the letter “d” – an abbreviation of a user-tag which works as both familiarity (“may I call you “d”, <dhch96>?) and as on-line efficiency.

In chatrooms, grammar is thus a developing protocol, reacting to both the demands of the rapid scrolling of the conversational threads, and to the creative demands of establishing on-line communicative competence. Common grammatical and orthographic principles are applied differently in chatrooms. In society generally, we use grammar to judge people in terms of social status, regional origins, and educational level. In chatrooms the rules have to some extent already changed. A person may be judged by how efficiently he or she types, and by the familiarity they are able to display with on-line chat conventions, such as abbreviations, graphics integration, and the capacity to respond to creative utterances in kind – to continue the stylistic directions of a thread, as well as its content or semantic load – and that may well mean “reading” and writing back the sorts of grammatical adjustments outlined in Case Study Seven. There does indeed appear to be evidence that on-line chat is activating new elements in the communicative repertoire.

5.2 Unique features of chatrooms

Overall, the case study sites have then been able to display not only communicative complexity inside the chat utterances, but complexity resolving into specific on-line chat techniques. Electronic chat is no longer only one small communication exercise among many, sharing most of the communicative styles of natural conversation or equivalent text forms (such as for instance the memo), but an important and distinctive form of communication, establishing its own regulatory systems and practices. Internet text-based chat is already changing as a technology, with the increasing use of webcams, multimedia and 3D Graphics-based chat communities[5] and the ability to use voice instead of only text. New applications of text-based chat are appearing with the availability of wearable computers[6], including miniature PCs, personal digital assistants (PDAs), cellular phone watches, cognitive-radios[7], and electronic performance support systems (EPSS)[8]. Such devices will enable people to access information via networks anytime, even while out walking. But as this occurs, it will in turn force adaptations to the sorts of on-line communicative practices revealed in this study, and others. From the discussion of the seven primary chatrooms in the case studies and several secondary chatrooms I have found that there are common, “core” elements, present on all web-based chat sites, as well as specialist elements on specialist sites – and further, that these elements are not limited to a special lexis, as might be expected in such relatively new communication contexts, but extend to the full range of communicative behaviours.

This study has shown too that chatrooms place particular limitations on communication, producing unique communicative strategies which not only mark them as communicative locations and cultures, but are consciously deployed by users to demonstrate competence and status within on-line community.  In summary, moving from Case Study to Case Study, the following communicative features already mark on-line chat:

Author as reader, reader as author (Case Study 1)

On-line, as talk text generates, the “reader” and the “author” can be the same person at the same time.  The listening and response phases of face-to-face conversation are less separable on-line, where the formulation of a reply is dependent upon a high-demand interpretation or “reading” of prior postings – including their formatting, recognition of which is required for reciprocal expression, which lifts a participant’s status within the chatroom. Without this capacity to process postings at speed, and to reply creatively and in like mode, chat participants become less successful in on-line communication. To be a powerful on-line “author” is also to be a competent on-line “reader”.

Chatroom titles as communicative-community controls (Case Study 1)

The title of a given chatroom often fails to indicate what is actually discussed. On-line communities, like casual conversationalists in the off-line world, very often redirect their communicative focus – and sometimes permanently, with consensual groups setting up regular meetings in spaces no longer very relevant to their topics.  This “drift” in topic direction demonstrates once again the focus produced within on-line chat on communicative technique, with chat very often more directed towards features of its own communicative repertoires than to pre-determined topics.

Multiple-Authorship in different chatrooms (Case Study 2)

It is difficult in face-to-face conversation to carry on two or more conversations at the same time, but in chat communication it is possible to open two or more screens on one’s monitor, in order to chat in several chatrooms at the same time. This can be expanded to having conversations in different locations at the same time, for example speaking with someone in Australia at midnight there and someone in New York in the early afternoon, New York time. And within a given chatroom, it is also possible to maintain multiple conversational threads, responding to different topic-focused chat relations as the relevant postings appear.  On-line, communicative “authorship” thus distantiates from the actual “author”, in quite formal communicative ways – well in advance of any conjecture as to identity experimentation or concealment. Chat is markedly “presentified”, in Lefebvre’s terms: that is, attending always to the response happening NOW, as well as displaying a strong interest in and skill with presentational aspects of communication – but without arriving at Derrida’s postulation of “presence” in speech: that authorizing validation of communication which is conventionally thought of as originating in the physical being of a speaker. This suggests that the curious and much-remarked physical absence of chatters from the relations they establish is over-compensated through such practices as multiple simultaneous engagement in chatrooms and chat strands, and in the excessively conscious attention to chat utterance forms.

Avatars (Case Study 3)

Avatars are graphic or textual representatives of the speaker, based on how the chatter identifies him or herself. The avatar could be an animal, cartoon, celebrity or any object. An avatar is the chatter at the time of textual engagement. Again, its created character both distantiates and characterizes a chat participant, acting to position them in the larger chat community in a preferred way. The persona thus also becomes a part of the communicative intent, adding to the complexity of chat techniques.

Emoticons (Case Study 3)

Using a series of keyed characters to indicate an emotion, such as pleasure [:-) J] or sadness [:-( L] chatters are able to communicate beyond the “word”, giving faster communication. Some emoticons are becoming universal – even carrying the same meaning in different languages.  The first and most used emoticon is the smiley[9]. Emoticons re-deploy the keyboard repertoire, adding expression to a communicative form denied the expressive techniques of gesture, facial expression or vocalization. Once again however they have already established themselves as a layer of communicative competence, used not only to add nuance (acting for instance as mitigators or intensifiers) but to demonstrate creativity and “wit” in interchanges.

Threads and Discontinuity (Case Study 4)

Because conversational threads disconnect in on-line chat, as the posting sequences react to the technologisation of the IRC software and not to interpersonal turn-relations, all chat participants must both accept and learn to negotiate discontinuities in their postings and those of others. The ability to focus on topic and to build even multilogue discussion under these circumstances has already established itself across many types of chatroom – so much so that common elements of practice are already evident from chatroom to chatroom. Often even very extended sequences of intervening text do not appear to deter thread focus, while chatters are also able to respond to sequences which “de-thread” as postings arrive in inappropriate order; i.e. sequences dictated more by typing speeds or transport efficiency than by the logic of the topic development. This particular form of “repair” work appears to pose few problems for chatters.

Discontinuity, i.e. popup ads or ads amongst the turn-takings (Case Study 4)

One form of stop in the flow of conversation in chatrooms is caused by advertisements that are auto-inserted at regular places amongst turn-takings. Different chatrooms will have varying spaces for their ads, some having an ad appear every five turns, others displaying ads that appear to randomly pop-up in the midst of the chat. These interruptions also appear to be no problem to chat participants, who remain focused on their threads. It appears that intervening postings of this kind are dealt with not as chat, but as otherwise-framed text, which does not “interrupt” the texts of talk.

Chatroom graffiti (Case Study 5)

The messages conveyed through the work of graffiti artists are often highly political and deliberately aggressive, positioned in public spaces most likely to attract notice and force response. Some on-line participants go from chatroom to chatroom, leaving messages but not participating in chatroom conversation: I refer to this as chatroom graffiti. Perhaps because their postings appear to chatters as utterance rather than as “otherwise-framed” text, these postings are more likely to evoke negative response – especially if repeated.

Fleeting text (Case Study 5)

Chat, despite its textual base, is still a synchronous communication form, yet lacks the permanency of asynchronous texted message systems. Thus, despite its texted format, it shares more features with talk than with prose – among them the tendency to “patrol” or work positively and negatively to maintain the specific features of the communicative forms and relations present in a given chatroom. This drive to include and exclude utterance forms, utterances and utterers is evident in different degrees and different ways in different spaces and chat modes, but does mark a communal sense of control over chat, and a regulation of what is and is not acceptable or preferred behaviour.

Lurking (Case Study 6)

 Lurking is one behaviour which may not be welcomed in chatrooms. Some chatrooms do not show inactive chatters in the room and therefore the lurker is even more hidden from view. A lurker is able to read and observe behaviour in a chatroom without making any contribution – but since chat is by definition a participatory activity, lurking defies all aspects of the communicative act, with even the “reading” which we might anticipate as being carried out by a lurker being inactive by virtue of its failure to connect with the “w/reading” of texted chat which is signaled in properly configured response postings. Since chat status is judged by the relevance and creativity and format-matching of one’s postings, lurking is so low status as to attract derision and censure – or at the very least, nervousness. 

Collaborated-Selves (Case Study 6)

MUDs and MOOs are collaborative, networked environments where the MOO and MUD consists of a number of connected rooms. Chatters create a “combined self”, partly fictionalized but partly built on his or her own chat capacities and skills, in order to create a space or story or thread in the chatroom. It is the MUD and MOO experience which signals most clearly the continuity-separation aspect of chat identity on-line, where the skills required to chat with authority and efficacy – elements continuous with our off-line expectations of a “present” or authorizing self from which “expression” can flow – can be shown to be fictionally deployed, in the service of an on-line character role. This insight drives a further wedge between identity and chat-skills: that is, it establishes the distance that exists on-line between whatever roles and statuses a chat participant may be accorded in real life, and those established through their skills at on-line chat. It is here that the special chat codings enter the scene, providing a repertoire of possibility across which chat experts can play, to establish their on-line credibility.

Spelling, Abbreviations and Grammatical errors as on-line “norms” (Case Study 7).

Abbreviations and grammatical errors are not only accepted but also dominant in on-line chat, for two primary reasons. Firstly the speed of “speech” in a chatroom does not provide time for writing out what can be abbreviated, leading to forms such as “btw” for “by the way”. Once this is established as commonplace however, it becomes a marker of expertise. High-statused chatters – those whose postings gain attention – display creative innovation and application of such compounds, abbreviations and grammatico-orthographical reformations.  Moments of reciprocation between chatters all displaying command of these new conventions become peak moments of on-line chat, showing the degree to which chat conventions themselves are a major element of on-line community identity, and have become central to chat as a communicative form. 

Long gaps between asking and answering in turn-takings, with other turn-takings in between – equivalent to the listening phase in a conversation (Case Study Two)

If chat-community is established in the formal conventions of chat “style”, “w/readers” or entrants to a chat space who seek to participate must work to establish the repertoires in play; the level of skills required to intervene, and the likely acceptability of their own postings, in terms not just of ideas and opinions – semantic issues – but of their capacity to reciprocate in kind at the formal level. But other elements of chat skill are also demanded. The length of gap between turns, and the ability to locate and follow discontinuous threads, also place a premium on chatroom experience. For many new chat users this threading complexity is baffling. Its difficulty is often dependent on, firstly, how many people there are in the chatroom, and secondly the number of turn-takings offered and taken up – by one or by many participants. For example, in the “911” chat I have referred to in this study, there were as many as 45 turns in a minute – sometimes two entries for the same second – which leaves little time to construct those turns. Below there are seventeen turns in one minute.





Pete: Let kill all Palestian terrorist´s greetings from Finland ps:morjens Will kuis panee






oscar: that's not shute will!!!!



MissMaca: hikacked planes, and flew 3 planes into the pentagon.



mike: I think so, miss maca.



sascha: hallo from germany



Hello: How many building are still up in NY



1Bone!!: Whats up in NY???????????



damaged: no then we get a world wore 2



dolly: our news says five planes now



1Bone!!: I'm from germany too!



novyk: who's the author of this ... ??? Anyone know there ???



sascha: 3



Will: Pete: Siinähän se



sascha: the 3rd world wore



1Bone!!: %&#% 3. Worldwar?!?!



oscar: hello 1 bone, where are you from?


Of these eleven chatters who “spoke”, only three had more than one turn in that minute. <1Bone!!:> had four utterances in this minute:






1Bone!!: Whats up in NY???????????



1Bone!!: I'm from germany too!



1Bone!!: %&#% 3. Worldwar?!?!




The degree to which this chatter also manages to engage other postings, all within this very tight time frame, suggests on-line experience – as does the heavy use of keyboard expressives and “stuttered” repetitions as intensifiers. <1Bone!!> is able to drive multiple conversations right across the crowded chatroom, to follow up on postings, but also to present a coherent and even passionate political engagement – even permitting  a distraction: “I’m from Germany too!” as he/she notes Sascha’s posting. This occupancy of close to 25% of this set of postings renders this chatter a dominant force at this moment.

Chat technologisation and turn-taking disruption: anticipating discourse

As in face-to-face chat there are sometimes instances when an unexpected utterance occurs. With the de-ordering that can occur within the delayed response of entry and posting, curious effects can arise. In the thread above, <!Bone!!> has an utterance arrive on the site only one second after <sascha>, at line 44 introduces the phrase and so the concept: “world war”. Without the time=entry evidence, <1Bone!!>’s posting looks like a response-turn: reaction either to the suggestion of war, or perhaps to the misspelling: “world wore”. But the single second of elapsed time makes this impossible. <1Bone!!>’s other turns arrive at about 10-15 second intervals – about the time it takes to read, respond, enter and have a posting arrive. What we have is not a response turn – a dialogue – but two independent chatters arriving at the same conclusion at the same moment.



sascha: the 3rd world wore



1Bone!!: %&#% 3. Worldwar?!?!




Repeated utterances with little or no content e.g. “hello”, “anyone want to chat” (see Case Study One).

In chat terms these are phatic communicative entries: ritual exchanges, signaling presence in an otherwise un-indicatable context. Greetings have become very quickly established as a formal necessity in chatrooms, and a round of greetings is considered a requirement for entry into existing chat threads, or the launching of new ones – anything less is interruption. Unacknowledged greetings thus become signs that a chat group is unwilling to admit more members: a hint to either await a suitable thread to enter, or to go away. Repeated greetings from the same individual thus read as intrusive – or perhaps as desperate. Unless such a potential chat participant can move to establish the requisite codes of credibility through the “display” features of their postings, they are less and less likely to receive response and be admitted to chat exchange.

Short conversational utterances

In almost all cases, talk in chatrooms is limited to short phrases. Rarely will there be more than several words written at a time by a “speaker”. Counting the words of hundreds of entries in my seven chatrooms (see table below) I found an average of 5.82 units per turn; including words, abbreviations, and emoticons. Within that sampling 25 percent of words consisted of only two letters, and 20 percent consisted of three letter words. Using CMC or the computer as the tool for an electronic content analysis, introduced in Case Study Two, I found that eighty-three percent of words used in chatroom conversations consisted of five letters or less.

1)     Purpose chatroom (Hurricane Floyd) Avg. 7.17/per turn

2)     Instant Messenger (two-person conversation)  11.32/per turn

3)     Celebrity chat  Avg. 4.2per turn

4)     Astrology – purpose chat Avg. 3.5//per turn

5)     No topic chat - Avg. 3.2/per turn

6)     Topic (3D animation) chat  Avg. 4.4/per turn

7)     Topic – baseball chat - Avg. 6.7  /per turn

The above table shows that users of multi-voiced chatrooms, whether they are working with a stated topic or not, produce fewer utterances than users in a chatroom with only two people speaking, as in an Instant Messenger environment. The Instant Messenger chat that I “captured” had 11.32 words per turn compared to other chatrooms that averaged 3.2; 3.5; 4.2; 4.4; 6.7 and 7.17 words per turn.

This implies that more is said when only two people are in a chatroom. With several voices seemingly all speaking, it is difficult, unless one is a very fast typist, to respond before someone else does. The “reading” time on a busy board, allied to the waiting time to have your own turns attended to with a directed response, cuts back on the ratio of postings from each participant.

On-line chat and intimacy: public conversation and personal expressiveness.

Many of the findings of the uniqueness of chatrooms can be seen in the table below which highlights differences between asynchronous on-line communication (chatrooms) and synchronous electronic formats (e-mail, discussion groups).



time-bound conversation – or real-time communication

on-going conversation – not necessarily the same day

must arrange a specified time to participate to meet

can communicate any time

can interact only with those presently on-line

can interact with people not presently on-line

fast and free-flowing conversation may be hard to follow (much chat is very informal and relaxed)

slow paced conversation allows more time for understanding and formulating thoughts (more opportunity for formal, thoughtful discussion)

multiple conversations occurring simultaneously may be difficult to follow

conversations are usually arranged by topics

one-to-one (IM) allows for individual conversation; IRC is “public” chat

private conversation on a one-to-one basis in e-mail, but not on noticeboards

messages are fleeting; can't be referred to later except if saved; scrolling back to capture past comments means missing ongoing talk

messages are permanent for later reference


Chatrooms display many of the features of off-line “friendship” gatherings and their talk-formats, including the necessity to display “notable” qualities in the talk performance, to be noticed within the group; to meet the norms of the particular group in order to be an acceptable group member; to know the codes, preferred topics, and specialized locations of chat types, and to be prepared to “meet” and talk regularly, to keep these skills honed and updated. On-line chat appears to demand much the same commitment to sociality as its off-line equivalent.

Chat-types have however already differentiated within the IRC community generally, and can be further defined by the following chat-behavioural categories[10]:

1. Initiating messages which successfully stimulate a new discussion.

Chatters begin discussional threads with the anticipation that others will continue. Continuity stops if no one responds.

2. Initiating messages which fail to stimulate further discussion,

If no one responds, a chatter may attempt to re-introduce the thread, but if no one responds then the thread dies, unless someone else reintroduces it.

3. Continuing messages which cause further discussion.

Responding successfully requires the sorts of w/readerly sensitivity to issues and form which enables chatters to create utterances suited to the group norms – or if possible, extending them further, in the right ways. Responses which simply approve or confirm are acceptable; for instance indicating approval in chat-abbreviation form: “lol” or “J” – but the most responded to are those postings which move a thread forward, whilst also displaying chat-form expertise and creativity.

4. Continuing messages which create branching threads.

A thread can have several thread nodes branching from the root branch, which will then have an overall topic but with sub-discussions. For example in Case Study One there is the main thread of Hurricane Floyd with several branching threads that are still about the storm but a different aspect of it – such as the discussion about Mexican roofers or a thread about sizes of buildings.

As my research dealt with the formal aspects of on-line chat, it did not attempt to explore how the users felt about their time on-line. Studies have been done that show that a majority of chatters “felt like they could jump right in and chat”, or that “chat discussions are too superficial”, or that “chat went too fast because he or she could not keep up with the conversation”, or that “14 out of 15 felt a moderator was needed” [11]. My own research has not identified what people think, but is still able to show that users can indeed “jump right in and chat” – but that most in fact consider the prior postings before doing so. To “write” is to “read” first.

Are these then the major features of on-line chat across all domains, all languages, and into the future? Certainly the technologisation of this form of talk appears to have spread across language groups and cultural behaviours.

Chatrooms currently provide one of the most universal forms of communicating. By late 2002 there were 4206 Internet cafes in 140 countries[12] and wherever there is an internet café there is the opportunity to chat on-line. In the Middle East for instance there are many chatrooms available and most have translating software for the language of the chatters to be translated into the user’s native language. On the chat-server, http://www.chatinternational.com  the following chatrooms were currently available (as of December 2002):

Afghanistan (5)

Armenia (5)

Azerbaijan (5)

Bahrain (4)

Iran (9)

Iraq (5)

Israel (9)

Jordan (4)

Lebanon (7)

Pakistan (15)

Syria (7)

Turkey (9)


The universality of chat-styles can be demonstrated by examining a chatroom on the Iraq-Net domain, which has similarities to the chatrooms in all of my case studies. Since this is a JavaScript chatroom the log could not be captured as text, but is “snapshotted” direct from its webpage.

(Iraq-Net chatroom on the day the US invaded Iraq – March 2003)

The formatting of chat entries is immediately recognizable, even when in Arabic script, as is the convention of name-tagging – right-to-left, even in a left-to-right texting language such as Arabic. The list of users on-line to the right indicates the fusion of cultural representations available: Anglo or Arabic names in Roman script (<basil> or <Haedar>); Arabic coded into Roman script with accent markers – not reproducible in the Word Processing package I am using for this discussion: see tags 5 and 7 in the list. At the same time, within these selections, participants are able to code their tags for expressive effect – not only in the overt US aggression of a “GI Joe” on such a site, but possibly also in the interesting number of “Alexanders” who appear to be registered to this site. Already it is possible, even in such limited sampling and in the presence of a dual cultural context, to see chatters playing to position their contributions in the sorts of ways introduced in discussions above.

Even where chat participants enter from different language and cultural contexts, IRC conventions are observable.

Lebanon-based chatroom

On this Lebanon-based chatroom, which has an instant translator, the speaker is not demonstrating good command of English. But common abbreviations are used that would be found in any English-speaking chatroom, such as <how r u> - and the emoticon < :) > is used in standard form. Even in the dual-language situation, where threads cross in scripts as well as in topics, chatters build response relations in familiar ways:

Soso’s careful attempt to suggest that Moz “serch after help” for his violence finally devolves into reciprocal personal abuse: “they have to blow u from the world”.

This study has shown that on-line chat communities do take on social agendas as much as they would in person-to-person meetings. Communities of practice can be communities marked by acceptable and non-acceptable behaviours registered at the level of the doubled speech of chat, with its semiotic loadings of meaning and familiarity. In Case Study One it was apparent that there was an ease among the speakers in discussing Mexican roofers in the midst of a discussion of a national emergency. In Case Study Seven the baseball chatroom has a community of practice where the participants are comfortable with their specialised sports talk. Here the participants have not developed an in-depth discussion or a site-specific set of codes - but there are the same practices of greetings, abbreviations and quickly accelerating shifts from mitigation to abuse, as seen across all case studies.  Topic and situation it seems, do not prevail against the standard features of on-line chat behaviour.

5.3 Research Questions and answers

Having revealed them both a tendency towards community-specific chat behaviours and at least the foundations for “chat universals”, it is time to revisit the research questions which orginally drove this project. How have they contriubted to, or contrained the findings?  The five initial focus areas for this study were as follows:

  1. Is meaning communicated in chatrooms?
  2. How is turn-taking negotiated within chatrooms?
  3. Are issues of cultural sensitivity as relevant as in face-to-face talk?
  4. How is electronic chat reflective of current social discourses?
  5. Will chat become a universally understood language?

Added to these mixed and incommensurable questions were an equally multi-level listing of my then-current assumptions on online communication:

  1. That people create a different textual self for the chatroom environment that they are in
  2. That conversation within chatrooms will change how we come to know others
  3. That ‘chat’ does not differ from natural conversation
  4. That observational studies of chatroom conversation can capture some of the adaptations of conversational behaviours
  5. That this work gives us a better understanding of how and why chatrooms are an important area in which  to create a new conversational research theory


Having completed the seven differently-focused case studies designed to investigate these issues, it is now possible to see the quite distinctive directions these questions raise and the concomitant ways in which equally distinctive “clusters” of research focus have proven to have arise. The studies move from the fine focus of what can now be seen as technological and methodological questions (turn-taking; meaning-making; observational study) to a comparative emphasis cultural sensitivity in chat and in real-life talk; chat as reflective of real-life discourses; chat in comparison to natural conversation to the “postulatory” emphasis of much broader questions (chat as a useful area for new conversational research theory; chat as a new universal language). My own preliminary thinking indicated a three part study program, moving from existing linguistic-based observational and analytical methods, to an empirical evidence-founded description of actual online “talk” practice, and so to a deeper and richer set of hypotheses relating to online “chat” practices and behaviours. The study has thus begun the first stage of a methodological design for the study of chat – and perhaps of its future technologisations. The ODAM or Online Discourse Analysis Method proposed at the outset has evolved across the seven constitutive Case Studies:

  • CS1  Reader-response

Meaning-making depends on interpretation

Interpretation depends on Habitus and e-Habitus

So the study moves to

  • CS2 Technologisation

CMCs contribute new connective problems (gaps/pauses) and selections (bridgings/ braidings).

IM is relatively familiar (like conversation):

IRC is complex

So the study moves to         

  • CS3 Pragmatics/Semiotics

How talk is managed and represented online




Pragmatics                                                             Semiotics

Regulatory online cues/codes           Graphic play

command more response                    Creative play dominant      

So the study moves to

  • CS 4 Speech Act Theory:

Which talk forms occur in IRC?

Very indirect forms common, to keep relations OPEN.

Is IRC primarily relational?

So the study moves to

  • CS 5 Discourse Analysis:

Does the “relational” work online construct a communication community?




Consensual semiotic play                                   Speech-Acts-as initiation/termination

                                                                                                of “threads”

So the study moves to

  • CS 6 Conversational Analysis:

How online communication regulates exchanges. Turn-taking and repair evident, but more complex than in real life conversation

So the study moves to

  • CS 7 Grammar:

How do word selection and order contribute to chat?

Technologisation, creative play, force new “grammars”.



Concision       online competency                     familiarity in online community  

                                                                                                relational work

Building inwards from the broad user-perspectives of Reader Response, examining chat postings as actively received and interpreted “wreaderly” communication, the ODAM has cut four deep and “rich descriptive” wedges from a multi-dimensional, multi-leveled set of chat actions. Beginning simply as an empirically driven aim: to cut into actual instances of chat practice using any existing research methods which could examine how online talk “works”, the study can now be seen to offer in the first instance, a set of interlocking research tools, any or all of which can be picked up, critiqued and re-applied, to be improved upon in future studies, by future researchers.

A second “wedge” or cut from the research findings however establishes that a key direction in linguistic research methodologies: the drive towards establishing the “regulatory” or rules-and-systems elements behind language  use, is indeed given a different spin within online talk. Here it proves possible, again and again, from method to method – across the seven case studies “speech communities”, to reveal tightly regulated, recurrent and systematic talk practices, variant from those observable offline, even where there are equivalent interest or topic groups. Wedge 2 indicates those already established online practices which constitute the difference, and even expertise of online chat. They are what suggests that it may well be on its way to constituting its own “speech community/ies”.

But it is wedge 3: those descriptive features which reveal a markedly “open” or “creative” set of communicative behaviours online, which reveal how chat is being constituted. Here the evidence of complexity, semiotic and graphic play, consistent relational focus and creative expertise introduces the dynamic energy of online communication, favouring members and strategies and expertise which reveal skill and creativity and fast-paced interpretive responsiveness. Wedge 3 practices lead us on to the discourse-under-formation of Wedge 4: a discourse demanding continual enactment of familiarity, consensual strategies, relational work, and what CA would call “category maintenance” – of an exclusively “communal” kind.

Online chat, regardless of topic or the specifics of a participant group, appears directed to community itself. Not quite un-agentic as it dis-connects from action, it becomes meta-agentic: more about how to operate than about “what to do”. It is thus, contrary to most contemporary public and media accounts, richer in value than in projects. It is a discourse largely about itself.

5.4 Summary

After analysis of seven different locations for and modes of Internet chat, this study can be used to suggest that in the chatrooms captured and analysed for the period 1995 to 2001 there is evidence for a new genre of interactive, conversational writing, or “talk-texting”. While awaiting (and perhaps assisting in) the evolution of new methods of analysis for this hybrid communicative form and technologically transitional format, this study has tested a broad range of existing text and speech based analytical techniques, to uncover what we can know of how Internet chat forms currently operate. This genre – or set of genres - must then be regarded as historical and time bound, because the technology of delivery is in itself already changing; for example to include images and sound, so that communication within chatrooms is no longer simply text-based.

Nor is this transience within the format the sole aspect marking the ephemerality of chat. Chatters themselves know that their text may be lost forever; and yet ideas, offerings in creative prose, experiments with personal and social identity, debates and discussions and inquiries and statements are being written, posted and lost from moment to moment: communicative effort that in other more conventional writing genres would be saved, reflectively reassessed and elaborated on. On chatsites text is speech – with all of the misdirection, rapidity of onward flow, focus on the inter-relational, and lack of attention to permanence experienced in speech communities. It is surely significant that at the very moment that this attempt to capture and catalogue at least some of the behaviours of this communicative genre was being prepared for the processes of printing and binding, a major service for the activity of chat was, without warning, curtailed. In September 2003 Microsoft announced the closure of its IRC services.

While IRC services of various types remain available to users, and it seems likely, given the use of chat in various functions from education to industrial design and conferencing, that the genres will in some form prevail, a central moment of chat as a social activity is passing. This document may then, as it has so often suggested, be already on its way to being an historical study. It is important therefore to note that, despite the wide variations in chat purpose and performance found in the seven case studies used here, chat has in its short life evolved a solid central repertoire of communicative techniques. Each case study revealed some unique talk-texting features, but the primary outcome of each of the case studies proved that there were more common features in chat spaces and styles than differences. 

There is a new genre of “text-based conversation” text – that found in chatroom postings.  The chief characteristics of this genre include recognising how users create a distinctive, but site and talk-category regulated, “textual self” for each chatroom environment they enter. Conversation within chatrooms, without all the cues of previous forms of conversation, changes how we come to know and interact with others, so that new cues based on written conversation become as important as the physical ones which we rely on now. Observational study of chatroom conversation can capture some of the adaptations of standard conversational behaviours to the demands of on-line chat. Observation, description and analysis of chat, using existing analytical methodologies from both text and speech traditions, lets us take a first step towards recognition and analysis of new, hybrid, communicative forms. But it is already possible to uncover a consistency and replicability in findings across chat types and sites, which suggests that chatroom conversation has certain features which make it different from off-line, person-to-person conversation, including the following standard features:

1.                                          That the author or “speaker” role can be complex, requiring a rapid mixing of the reader and writer roles, as well as the capacity for multiple simultaneous engagement in a number of conversational threads – even using multiple log-on identities.

2.                                          That chatrooms use an on-line-specific adapted language which incorporates semi-graphic elements such as emoticons, a specialist “anti-language” of abbreviations, an expressive range of self-selected “tailored” settings involving font colours and styles, and the deployment of pre-formed phrases and ikons as representative of the author.

There is, above all else, an intensified emphasis in chat practice, on the fleeting nature of this texted conversation, since the Internet is itself an unstable, and even experimental, place. This set of studies of contemporary on-line chat behaviours has produced above all else, a foregrounding of the complex, interactive nature of on-line conversation, it demands upfront attention to inter-relational aspects of the talk-texting exchange, signalled in the complex braiding structure of the conversational threads and the inherent discontinuity of talk-exchanges introduced by the technology of the posting software. It is, in itself, a braided study, at the level of description, theorisation, case selection, methodology, and even of presentational design.

And that is, in the final analysis, the nature of the research object: Internet chat. It is likely to illustrate a tendency to conitnaul change – and one issuing ongoing challenges to researchers.

5.5 Future Research

Electronic communication is becoming an established form of communication. However, there are many areas within electronic and online communication which remain unexamined, yet which are undeniably generating new forms of communicative behaviour – and which have potential to feed back into further developments of the Computer Mediated Communication technologies and applications available to today’s and perhaps tomorrow’s communicators.

Among these experientially new social forms of communication evident in online chat, are some curiously invisible forms of communicative practice, qualitatively new and outside the scope of even the broad range of communicative methods of data-capture and analysis used in this study.  Research into silence in a chatroom, referred to as lurking (see in this thesis) has not been fully explored. In person-to-person communication, silence does have readable meaning. A participant’s silence in “natural” conversation is observable to both other participants and to analysts. It literally “speaks”, as a conscious act of non-participation. In electronic communication without visual cues, we cannot fully know the purpose of a person’s silence – and in the rapid stream of other conversational postings and responses, may not even notice it. What then is the social or relational impact of online silence? And beyond this more “absolute” silence, what of the uses of lag-times in active participation? Is there for instance an acceptable time lag between chats entries? If a participant is a slow typist, or considers a response for a length of time – or conducts multi-stranded exchanges and so is slower to each response, does this alter the communicative relation? How long can a response gap stretch, before it becomes too difficult to re-connect? In Instant Messenger chats there is a notice that appears that reads the “respondent is writing a reply,” but in multivoiced chatrooms it is impossible to know whether a person is slow in responses, otherwise occupied, or is actively “lurking” for a reason.

The impact of participation in casual electronic chat on privacy is another area of research that is still under formulation. While this research shows that chat has tendencies towards the establishment of casual and even intimate social relations, the literature suggests that many participants consider this non-proximate and non-physical social relation to be a secure space in which to interact with a broader than usual range of others, and to test out various ideas, behaviours, and even personae.. Attitudes to online security have however altered after aspects of the 9/11 events were connected to the capacities of the Internet to offer ease of international communication to terrorist groups. Subsequent security measures taken in the US to detect terrorist activity online may mean that chatters become more careful with their “talk”. In a Harris Poll conducted in April 2002[13] the following findings indicate that the US public, which had actively favoured monitoring of Internet communications by their government, is turning back towards an unregulated system:

  • Law enforcement monitoring of Internet discussions in chat rooms and other forums: favored by 55%, down from 63%; and
  • Expanded government monitoring of cell phones and email, to intercept communications: now favored by only 44% and opposed by 51%.

Will chatrooms remain an open sphere of communication, or have they lost their “innocence” as a place of play and experimentation?

Research into similarities between chatroom and mobile phone messaging (and image exchange) would seem to be an inviting field of study, with Internet based and phone based codes (especially of abbreviations for instance) appearing to converge. Are they in fact the same? And if differences exist, what might explain them?  Study into how mobile phone text-messaging is used to convey meaning in place of a voice message on mobile phones would help to show whether messaging conveyance is as effective with the abbreviations and emoticons used in phone text as speaking. It would also provide some interesting guidance on the possible communicative impact of moving to voice-activation on the Internet – and on some of the ways to interlink aural and text systems. Text-messaging is as short as chatroom text, but is more accessible – a rapid disseminator of the short-form texted message into new communities of users.  SMS was launched commercially for the first time in 1995 and by 2002 there were one billion SMS per day exchanged globally (December 2002)[14]. It may prove that my predictions in this study that IRC will be a short-lived technology, may in part be wrong – if SMS and mobile telephony become heir to the form.

Finally, this research raises questions in relation to the “global” or universal use of electronic and online translation software, offering instantaneous contact between speakers of different languages. With electronic chat becoming global, whether online or on a mobile phone, the need to exchange rapid messages across language barriers becomes more pressing.  But how accurate are the translation devices that are used for online communication? Online translators are available from services such as

 http://www.worldlingo.com who offer “WorldLingo Chat,” giving one the ability to chat instantly in ten languages; or Alta Vista’s Babel at http://world.altavista.com/ while at http://www.freetranslation.com/ there is Instant Multilingual Messaging for American On Line Instant Messenger and SMS Translators that gives translations from one’s mobile phone. But how accurate are the translated messages? More importantly, how can one use abbreviations in this environment and still be understood? The examples of the two phones above are full-sentence-translated - but what happens with typically shortened chat writing?  Imagine the message: Will U wed me @ Gretna tomorrow pls darling?  Translated into Dutch on Alta Vista’s Babel it comes back as Zal U wed me @ morgen pls darling Gretna? Would the receiver get the message correct? The translator at WorldLingo.com translates it differently: U wed me @ zal morgen pls darling Gretna whilst freetranslation.com interprets it as Wiedde wens U mij @ Gretna morgen pls lieveling? All three translations are different, with different meanings. If something as short and simple – yet as socially crucial! - as this message is translated incorrectly, what is needed to exchange meaning in international electronic devices? Can translation between languages also accommodate an online code of abbreviations which is informal and non-standardised – and to date, unrecorded?

Further research into online discursive communication will undoubtedly be driven by rapidly changing technologies as it becomes more intensified, more complex, more globalised, subtler and far more widespread.

But no matter the design outcomes, or the decisions taken technologically, or the platforms chosen for communicative exchange, we can be sure that users themselves, across an ever increasing range of language forms, will respond to these new “chat” formats in ways just as lively and variable; just as practically directed to communication, yet displaying just as much experimentation and pleasure, as the Internet chat participants captured here.




 911 chat (chat data is on the CD 911.html)

Afghanistan chat (chat data is on the CD afgan.htm)

Bondage chat (chat data is on the CD bondage.htm)

CNN News chat on 911 (chat data is on the CD CNN.htm)

Christian chat (chat data is on the CD christian_chat.htm)                                     


[2] Some of the definitions used in CA can serve as a starting point to describe what happens in between these turns. Three terms in common CA practice are gap, lapse and pause. A gap does not “belong” to anyone. It is a place of transition. A gap is a silence; the speaker has stopped speaking, and the next speaker “self selects”. In chatrooms this silence may be occupied by others reading the chat. 

When there is a silence, the next speaker has not been selected, and no one self selects, we have a “lapse”. It is only possible to distinguish a gap from a lapse after the event. Again in chatrooms, the next speaker may already be writing the response, reading the previous response, or there may simply be a silence in the same sense as the CA definition. 

A pause is silence when the current speaker has selected the next speaker and stopped talking, but the next speaker is silent. A pause is also silence that occurs within a participant’s turn. A pause "belongs" to the person currently designated speaker.


[3] This is a page from several pages of a CA workshop held on Fridays in 2002 at the State University of New York at Albany.

[4] For example in the very crowded 911 chat during the World Trade Centre destructions there were 644 turns and 4833 words of spoken text covering 80 minutes or an average of 8.05 turns per minute. Often there were utterances logged at the same second.



tippybond: can someone field me to another other chats for ny



Gary: i woke up to this and i just cant belive it....my heart goes out to all who have been injured


[5] Active Worlds, a Virtual-Reality experience, lets users visit and chat in 3D worlds that are built by other users. Viewed 12-2002, http://www.activeworlds.com/

ATMOSPHERE, with Adobe® Atmosphere™. With Atmosphere, users add a third dimension to their Web experience by creating realistic and immersive environments that offer a revolutionary approach to content, navigation, community, and communication.

Viewed 12-2002, http://www.adobe.com/products/atmosphere/

EXCITE CHAT, Text-based and graphics-based chat, events, and web content. Viewed 12-2002, http://www.excite.com/

HABBO HOTEL, Graphics-based chat where the user visits different hotel rooms or creates his or her own room. Viewed 12-2002, http://www.habbohotel.com/habbo/en/

Moove German-created 3D visual chat program. Viewed 12-2002, http://www.moove.com/

A continually updated list of other 3D chatrooms are at

http://www.thescarletletters.com/Blah/LipSync.html Viewed 12-2002.

[6] Mann (1997) suggests five characteristics of a wearable computer:

(1.) it may be used while the wearer is in motion;

(2.) it may be used while one or both hands are free, or occupied with other tasks;

(3.) it exists within the corporeal envelope of the user, ie, it should be not merely attached to the body but becomes an integral part of the person's clothing

(4.) it must allow the user to maintain control;

(5.) it must exhibit constancy, in the sense that it should be constantly available.

Mann, S. (1997) Conveners report of CHI '97 Workshop on Wearable Computers, Personal Communication to attendees. Viewed 12-2002 at http://www.bham.ac.uk/ManMechEng/IEG/w1.html

[7] Cognitive radio, a radio that is programmable to send messages on its own is part of the array of devices for wireless providers, for voice and data communication for the fourth-generation, or 4G, wireless services beginning in 2004. Viewed 12-2002 http://www.techextreme.com/perl/story/20731.html

[8] Electronic Performance Support System Viewed 12-2002 http://wearables.gatech.edu/EPSS.asp

[9] There are two claims for the origins of the smiley. One is that in 1972 journalist Franklin Loufrani created a simple concept for France soir and other European newspapers. He displayed icons to communicate news and especially good news. He gave this original icon the name of Smiley. It was published for the first time on Jan 1st 1972. Under Loufrani's supervision, SMILEY quickly spread across the world, easily crossing political, social and economic boundaries with his ever-increasing vocabulary of instantly recognizable emotions. (See The Smiley World at http://www.smileyworld.com/). The other claim for the origin of the smiley is that artist Harvey Ball created the first “smiley face” around December 1963 for one of his clients. He designed a yellow pin with the smiley face. This pin was handed out to company employees and clients and soon became a big hit. In a short time the “smiley face” appeared on all sorts of products. By the end of the 60s “smiley” had spread around the world (see World Smile Corporation at http://www.worldsmile.com/).  World Smiley Day was proclaimed for October 03, 2003.

[10] See, four possible types of message posted to a mailing list McElhearn, 2000, and Gruber, 1996.

[11] The results cited are from a survey on Assessing Student Learning Outcomes online at http://www.csusm.edu/acrl/imls/Q3Report.htm Sited online October 21, 2000. Other online surveys and viewers responses include; Test of an Internet virtual world for teen smoking cessation online at http://www.trdrp.org/PageGrant.asp?grant_id=2423; Hispanics in the U.S. 16 years of age and older, 38 percent are using the Internet on a regular basis, according to a new study released by AHAA and 43% are using the Internet for chatrooms,

http://www.ahaa.org/Mediaroom/Roslow%20Research%20Study.htm. INTERNET USE AND THE SELF CONCEPT:  LINKING SPECIFIC USES TO GLOBAL SELF-ESTEEM College freshmen at a mid-sized university in the mid-Atlantic of the USA were surveyed on chatroom behaviour - http://www.uiowa.edu/~grpproc/crisp/crisp.8.1.html

[12] Cybercafes worldwide are added constantly to at http://www.cybercafes.com/ Cited November 30, 2002

[13] THE HARRIS POLL® #16, April 3, 2002 online at


[14] See A Brief History of UK Text online at http://www.text.it/mediacentre/default.asp?intPageID=567