Communication on IRC

Elke Hentschel:

1 Introductory remarks

Computer mediated communication is taking over in many areas, and whether we welcome or deplore this development, it is certainly not reversible. It is to be expected that in this case, as in every other one, the means of communication will influence the communication itself. Just as we are compelled to give more minimal verbal responses when talking over the telephone, because our non-verbal feedback cannot be perceived, we must therefore adapt our signals and signs to the communicative possibilities a computer offers to us. It should be interesting to see, therefore, whether and if so, in which ways this kind of communication is different from others.

The following article is based on the author's two years of participation in a special kind of computer mediated communication, the so called IRC, and will try to outline different possible approaches to the analysis and interpretation of this form of communication.

2 The IRC and its users

IRC is an abbreviation that stands for "Internet Relay Chat", a multi user implementation allowing several people to talk simultaneously on the internet. "IRC was developed by Jarkko Oikarinen in Finland in the late eighties, and was originally intended to work as a better substitute for 'talk' on his bulletin board." (1)

In order to chat, IRC users usually enter one or several "channels". The number of people that can enter a channel is not limited, and the same holds true for the number of channels that can be formed over the IRC. On an average evening (European time), there will be at least 20.000 users on 5.000 to 60.000 channels on both sides of the net. (2Communication on IRC takes place in real time, at least under ideal conditions; in reality, however, so called "lags" from a few seconds up to several minutes can occur.

The statistically average IRC user has so far remained a rather unknown specimen, (3but the common prejudice pictures a lonesome human being without friends, not participating in social activities, and spending all their free time in front of the computer. It might seem, in this light, that conversations on IRC represent a special kind of communication led by lonesome individuals who are trying to contact one another in this way because they are too shy for "real world" communication. However, these stereotypes of the average net user have been proved wrong by Nicola Döring (1995), who has analysed interviews she led with N=350 net users (4% female, 96% male, average age 27). In interviewing these people, she had two aims: on one hand, she wanted to see if the stereotypical lonely and shy computer freak really exists. On the other hand, she wanted to submit to an empirical test the wide spread hypothesis that frequent use of computers leads to loneliness, and that the ever growing use of computers will result in a new society of lonely, shy individuals. Döring's results show that quite the contrary is the case: the 350 net users she interviewed showed no signs of loneliness whatsoever.

Only a minority of her sample, about one third, were IRC users. While 67% of the net users never where to be found on the IRC, 7% used it daily. This latter group is commented on by Döring as follows: "Taegliche IRC- Nutzer (n=21) hatten signifikant mehr wichtige Beziehungspartner, mit denen sie auch uebers Netz kommunizierten (12 versus 9) und insgesamt hatten taegliche IRC-Nutzer signifikant mehr Bekannte (48 versus 31, t=2,6, p<0,001) als seltenere IRC-Nutzer" (Döring 1995).

Döring regards IRC as one of the most time consuming net activities. The statistics she gives do not quite support this claim, but according to them, it is at least one of the most time consuming interactive net activities (compared to ftp, for instance, which can take a lot of time, but needs no activities besides the initial one that starts the process):

Time spent
1-10 min
11-20 min
21-30 min
31-60 min
61-120 min
> 120 min


(statistics: Döring 1995)

3 Strategies of communication

Having established the fact that IRC users seem to be completely "normal" people who are quite happy to communicate in the "real world" outside the net as well, it shall now be pointed out in which way communication on IRC is different from all other forms of communication existing (and analysed) up to this point, including other forms of computer mediated communication.

Communication on IRC can be defined as a private, spare-time activity with the following characteristics:

IRC users have only the ascii-code and the ISO Latin-a 8-bit character set at their disposal, which can be used in order to transmit verbal messages and to send commands for a limited number of activities provided by the IRC program itself.  

3.1 Non-verbal IRC activities

3.1.1 Territorial behaviour

IRC allows for a number of non-verbal activities that mostly deal with the division and distribution of virtual space. Conversation on IRC is mostly, although by no means exclusively, lead on so-called channels, i.e. virtual rooms where users meet. Anyone can create one or several channels and invite others to join them there in order to talk. Channels are not steady; they exist only as long as they are being used, and they vanish without a trace as soon a s the last user leaves them. This is one of the reasons why some IRC'ers employ "bots" to keep their channel up while they are away. Bots (short for: robots) are programs that simulate a client and stay on IRC while their owners are away; they a re quite common, although they are frowned upon by many IRC servers and even used to be strictly forbidden by the majority of them the past. Most bots are restricted to making sure that the channel exists even if there is no real person on it, and to the safeguarding of the channel "op" (operator status), which they give to their owners or to other privileged persons when they come to the channel. Often they safeguard the ops on their channel not only positively, i.e. handing them over to certain users, but negatively as well, by de-oping everybody who has been given an op by one of the human users of the channel but whose address is not in their memory. Moreover, the majority of them react to certain words, for instance obscene expressions, by non-verbal activities that only op-holders can perform: by "kicking" and "banning".

A kick will throw a person from the channel, and a ban will make it impossible for the banned user to enter the channel again. The following example, taken from chan nel #berlin2, shows a combination of kick and ban done by a user whose nick is Joringel (5):

The reason for the kick-and-ban-activity shown above is probably quite clear and needs no further comment. What is important about this, however, is the fact that beyond any doubt, kicks and bans are territorial activities; the only difference is that the territory in question is virtual rather than real. The consequences are, however, exactly the same as in real space: while kicked users almost always want to get back on the channel, the kick may have the effect of a pedagogical measure insofar as the incriminated behaviour will have to be changed. Bans, however, are much more serious; while kicks have in my experience even been used in order to flirt, bans are always considered a very serious and unfriendly act. Banned users often invest a lot of time and energy in order to get back on the channel they are banned from, mostly with the intention of kicking and banning those who have banned them. The results of this are sometimes virtual wars, that can in some cases consume an enormous amount of time and energy. Their aim is always the occupation of foreign territory, although it is, as has been pointed out, virtual rather than real. "War" is not my personal metaphor for these events. The programs that are used in these encounters, so-called scripts, are called "war scripts", some of which use the term even in their names (for instance PhoenixWar). The author of this article had the opportunity to watch IRC wars with mutliple channel takeovers and unfriendly behavior of all sorts, especially during the war in the Balkans, when for instance users of #Bosnia or #Croatia took over channel #Serbia, and vice versa.

Takeovers are a such a wide spread nuisance on IRC that some IRC-servers even warn their users that they are very seriously disapproved of and will, if detected, lead to their beeing banned from the server. (6However, as takeovers require some rather technical knowlegde, they will not be explained here in any detail. The point to be made is only that we definitely find territorial behavior on IRC, even though it is hostile.

The following example illustrates the general behavioral pattern behind these IRC wars; what we see here is a person joining the channel #serbia only in order to tell its users that they should be thrown out:

Besides takeovers, there are hostile means of employing verbal messages, the so called "flooding". Flooding sends so many messages to a user or a channel that any other exchange of messages is rendered impossible. The following example shows part of such an action on #serbia, although by no means all of it, as this assault went on for a much longer period of time. Although it might appear otherwise, there is only one user harassing the others on the channel: the users that seem to join the channel in order to flood it are so called clone bots, small bots that can be created by war scripts. The aim is, as one can easily see, to fill up all the space available with nonsensical, or as in the following example, unfriendly messages (the Slovenian phrase translates as: 'Slovenia has defeated Serbia'), thereby making it impossible for others to communicate: etc. (this went on and on)

Besides these rather unfriendly methods of territorial behavior in the virtual space on IRC, there are of course several positive ones, and it is "normal" communication on IRC that will be dealt with in the following sections.

3.1.2 Body language and facial expressions

Most body language and facial expressions are verbalized on IRC. There is however one way of signalling emotions analogically: emoticons. Smileys and frownies are wide-spread and well known to everybody who uses the net; there are long lists of them spread by email, and there is no need to go into detail. (7) In everyday IRC communication, only three to four of them are used regularly, and one finds them frequently in their "shortened" version (i.e. lacking their noses):

In order to emphasize the emotion in question, the smiley's mouth is often duplicated, cf. for instance: for a hearty laughter


for deep sadness.

The following example shows the use of this form of emphasis by a person on IRC. This IRC user, Vesna, uses verbal as well as nonverbal signs in order to show her amusement. It is interesting to note how she makes use of the ASCII-code in order to spell her name in "cyrillics":

3.2 The verbal side of IRC communication

3.2.1 Verbal code representing verbal code: the lingo

As has been pointed out before, the only signs available are those of the ASCII code (and the ISO Latin-a 8-bit character set); there are no special signs like umlaut or other diacritics. The problems arising because of this are solved differently from language to language. German IRC users write ae, oe, ue and ss for ä, ö, ü and ß; Serbian IRC'ers just write the basic letter without the diacritic sign; Russian users (mostly calling from the USA) make use of the English transcription of the Russian letters, and the Japanese use special ANSI escape control sequences to represent the Kanji signs, and so on.

The lingo that is transmitted in this way is very colloquial. Orthographic rules are not considered as important, and commas are only used if necessary in order to make oneself understood. This can be shown by the following German example:


(Rough translation: <banshee> where's Jori???? <Eugen> tnx4op <boswell> gone! <Snip> banshee: off <banshee> that I can see as well... <Snip> hehehehehehehe <banshee> and where is he gone??? *wonder* * Snip helps whereever he can.... <Snip> ban: well, he left <banshee> what the heck - with the cd ??? <boswell> ban: what cd?)

What happens here is most interesting: we find a sort of phonetical spelling which takes into account all the contractions that are typical for colloquial German, but not - yet - allowed to be written down. It is by no means logical, however, that the contraction of, for instance, in + das > ins is accepted, while mit + der > mitter is not. The German IRC'ers apply a way of spell ing that probably will sooner or later be "official" as well. The same holds true for the abbrevation of the modal particle denn, which in spoken language mostly occurs as n; the indefinite article eine(e) can occasionally b e found as 'n(e) even outside IRC. The wassnfuerne of the example would be was denn für 'ne in "official" spelling, but wassnfuerne certainly more correctly mirrors the way the language is actually spoken.

Whatever conclusions one might be inclined to draw from these findings, one thing is obvious: the way IRC'ers use and spell their language shows very realistically the way in which the language is used in actual everyday conversation, and the way speakers feel about morphological borders. The latter is true even in languages like Serbian, where we find a more or less phonological spelling anyway. What we can see here, for instance, is the fact that most speakers have come to consider the negator ne as part of the verb rather than as a free morpheme (the negator is marked):

(Roughly: 'It doesn't mean anything, but if it is impossible for you not to do it...')

We might conclude, therefore, that IRC provides a means of getting first hand information about speakers attitudes towards the phonetical, phonological and morphological - and of course syntactic - facts of their languages, and therefore enables insights in actual processes of language change.

3.2.2 At the border verbal/nonverbal: how to prevent turn taking

In face-to-face communication, there are many ways for the speakers to show that they desire to keep the turn and go on speaking: from body language to prosod ic markers, there are numerous signs that indicate that the speaker has not yet finished. (8) The same holds true for communication over the telephone, where even though body language and facial expression cannot be used, there are still all sorts of par alinguistic signs possible. And if all else fails and yet one wants to keep the turn to oneself, one can simply go on speaking without pausing.

However, not only are nonverbal and paralinguistic signs unemployable on the IRC, even this last resort of talking on and on cannot be used. One of the reasons for this is the fact that the number of signs the buffer will take is limited to approximately four rows of text. But even more important: other than in the "talk" program, IRC'ers don't s ee what others type unless the message is "sent" (that is, the return key is pressed). Even if one could go on writing endlessly, nobody would be able to notice that there is a flow of intended communication. The only visible effect on the chann el would be silence, which by itself could be interpreted as a temporary wish not to say anything, or as a private conversation going on by means of "msgs" (9), or as a result of a lag. A person's silence on the channel might lead to questions like "are you there?" or to a "ping" (10) in order to see if there is a lag, but never to the assumption that someone is trying to communicate.

Therefore, if one wants to keep their turn on the IRC, one must take refuge in othe r resources. Contrary to other forms of communication, what one does is not talking on and on without pause, but interrupting ones own sentences - typically in places where it is quite clear that they are not yet complete. The following example, again taken from the channel #serbia, shows a small part of a very typical IRC monologue, broken up in tiny pieces in order to keep others from taking the turn:

original monologue

rough translation

<AGX> naravno
of course
<AGX> ovde imamo neke saljivdzije
there are some jokers here
<AGX> ali vama imam samo jedno da kazem
but I have to tell you just one thing
<AGX> sada je 3:50 ovde
it's now 3:50 here
<AGX> ujutro
in the morning
<AGX> ja svaki dan ovde sedim
I'm sitting here every day
<AGX> i cuvam kanal
and protecting the channel
<AGX> i neko ce reic:
and someone will say:
<AGX> sta? da neces medalju?
what? don't you want a medal?
<AGX> da neces mozda da te poljubimo u dupe?
or would you like us to kiss your behind?
<AGX> e pa reci cu vam nesto
well, I'll tell you something
<AGX> mozda je tesko poverovati
maybe it's hard to believe
<Peca> agx:to za dupe, to necemo:)
(Peca:) as for your behind, we won't! :)
<AGX> ali eto
but here you are
<AGX> tako je...
that's how it is...
<AGX> zaiste ima neko
there is really someone
<AGX> ko se zrtvuje
who sacrifies himself
<AGX> i ko ceni nesto
and who valuates something
<AGX> sto se zove sloboda
which is called freedom
<AGX> i mada je rtesko da se u to veruje
and although it's rare that someone believes in this
<AGX> to ipak postoji
this nevertheless exists
<AGX> mozda je tesko da se poveruje da ima boga
maybe it's hard to believe that God exists
<AGX> daleko od toga da sam ja bog
of course I am not God
<AGX> ali on postoji
but he exists
<AGX> i tu je...
and is here...
<AGX> neki misle
some think
<AGX> da ce
that [they] will

and so on. (The content of this monologue is probably hard to understand without context, but it is of little interest, since the communicative technics applied here can be seen even without understanding the language).

3.2.3 Verbalization of nonverbal signs

It is not only smileys and frownies that are used in order to comment on the verbal messages exchanged. Besides them, a who le system of meta-comments has established itself. These meta-comments, of course, differ greatly from language to languag, but are, however, quite the same in principle.

One of the ways to comment on ones own attitude is provided by the IRC program itself: the command /me leads to third-person-utterances about oneself. We have already encountered one of them in one of the examples above, where the user Vesna commented on her laughter:

But this is not the only way to comment on ones utterances. The other way is to incorporate the comment in ones utterance and to mark it by asterisks. In order to verbalize a smiley, for instance, one would simply write: *grin*, as in the followi ng German example: Another meta-comment of this kind has already occured in the earlier example that had been used in order to illustrate kick and ban: It is interesting to see how the user in question fuses the two German words Hände ('hands') and reiben ('to rub') together in order to use them as one comment. The same holds true for the following example, which is made up from gefährlich = 'dangerous' and guck(en) = 'to look': But there are, of course, genuine one-word-comments as well, one of which has already occured in a previous example: The principle of using only the verb stem is not restricted to German or English, as the following typical example shows: (Tesiti means 'to comfort'; -iti is the infinitive morpheme.)

Onomatopoetics can be used in the same way; cf. for instance the following example, where the user Joringel comments on the fact that he is drinking a glass of wine:

There are even complete dialogues based exclusively on the verbalization of nonverbal reactions, as the following example shows:

3.2.4 Paralinguistic markers

Besides the nonverbal signs of spacial behavior, body language and facial expression, there remains the whole paralinguistic range of communicative signs to be made up for: short pauses and temporary silence, voice pitch and prosodic markers. As has been pointed out before, pauses and silence cannot be used as communicative signs on IRC, because it would not be possible to interpret th em as such. There is, however, one way to make up for short pauses:

Voice pitch and prosody, as they are used in order to express emphasis, are expressed in many ways. One of the most frequent is reduplication, as shown in the following examples: It is interesting to see that only two of these examples mirror the real life speech behavior: it would be quite natural to lengthen the a in ganz (gaaaanz ruhig!), and it would be possible, if maybe somewhat less natural, to use a long a in the name Julia, even though the lengthening of the u would seem more common. But there is no way to reduplicate the g as in arggggggg; and nobody in a real life conversation would repeat the number "30" 24 times (the conversation which this example is taken from concerned the 30. birthday of one of the users on the channel). So, while reduplication does occur in real life conversation, and might even be a frequent phanomenon, it is used in quite a different way on IRC: it does not take into account the phonetical and practical aspects of spoken language, bu t rather makes exclusive use of the graphemes.

Voice pitch as such, that is shouting or screaming, is mostly represented by capital letters; cf:

If there are no other contextual markers, this way of spelling is automatically interpreted as speaking at the top of ones voice, as the following example illustrates: It is quite clear that the user Darko was not intending to speak at the top of his voice (he is, after all, only asking nibrs how much time something takes), but rather had only hit the caps lock accidently. He is nonetheless admonished by nibrs not to shout. This is by no means the only example of such a reaction to caps lock that I have encountered.

Reduplication and capital letters can of course be combined (and again we see that the reduplication is exclusively used as a graphic marker and mirrors no phonetic reality whatsoever):

Other means of marking emphasis are rather rare. Inversion and underlining are considered annoying by many users, and spacing is very rare; so far, I have found only one e xample: Two other means of marking something are provided by the program itself: beep and banner. While beeping - that is, sending an acoustic signal - is mostly used only in order to get the attention of a particular IRC'er who might temporarily be in another window, and is almost never sent to a whole channel (where it almost certainly would not be welcome), banners are found rather frequently, even though they, too, are not always welcome. Cf. the following example, where a banner was used in order to greet the author of this article when she joined the channel:

3.2.5 Singing

It has been pointed out before that capital letters are used to mark shouting. Depending on the context, however, it can mean another activity as well, which requires a good bit of vocal power: singing. Cf. the following example, taken from channel #croatia:

                        STRANGE IS THE DARKNESS, NO TRACE OF YOU, ONLY WILDERNESS                         Vitez: are you producing poetry?                         YOU MEAN EVERYTHING TO ME                         jasam: no... I'm only reproducing it... momentarily on my walkman

The conversation shows that there was a small lag; Vitez goes on singing two more lines before he sees the question about his production of poetry ? which was of course asked in order to elicite an answer like "no, I'm only singing a song" - and then answers it.

It might seem strange to sing in this way, typing the song and sending it to the channel, but it is anything but rare. People are not only singing alone, but also together, joining in by typing the next line of text.


3 Conclusion

This short survey has by no means covered all the areas and layers of communicatio n on IRC, but it is to be hoped that it has succeeded in conveying the fascinating range of possibilities that hide behind such a seemingly poor and rudimentary mean of communication, deprived of almost all nonverbal and paralinguistic signs. It was meant to show that new, specific codes develop under such circumstances, and can make up for, if not all, at least for the greater part of the missing metacommunicative signs. IRC interaction covers almost every aspect of real life interaction, and allows even wars to be fought and love to be made (there is something called "net sex"). Apart from that, IRC conversations allow a glimpse of the phonetic and phonological, morphologica and syntactical structure of everyday language, seen in the way the n ative speakers themselves perceive it.



1 "Undernet", . [back]

2 Unfortunately, a split has developed between the European and the American IRC, so that there are now actually two IRCs - not counting the "Undernet" and similar additional IRC domains. [back]

3 Even the general IRC statistics on have not been updated since 1994; cf. [back]

4 Exept for the IRC itself, there are several approaches to the analysis of computer mediated communication. A very good survey of papers related to this subject could be found at (link no longer active) [back]

5 This example illustrates one other thing: IRC'ers do not use their real names, but instead give themselves nicknames, so called nicks. [back]

6 Cf. for instance the welcome message of


7 However, anyone interested in detail should refer to: [back]

8 Cf. for instance Burgoon et al. (1989). [back]

9 "Msgs" (short for messages) can be exchanged between IRC'ers in private; the command /msg <nick> sends the message exclusively to the user with the given nick. The whole communication process is invisible for everybody else. As msgs are private and so-to-speak confidential, they are not quoted in this article; all the examples given were public messages on channels. [back]

10 The command /ping <nick> sends a signal to the other users and measures the time it takes to travel there and back; it therefore shows if there is a lag between two users. [back]

11 Pustionja should really be pustinja ('desert', 'wilderness'); the misspelling is clearly a "typo". All examples are original parts of log files, and none of the misspellings or other errors have been corrected. [back]


Burgoon, Judee K., et al. (1989): Nonverbal communication. The unspoken dialogue. New York.

Caldwell, Barrett S. / Lilas H. Taha (1993): Starving at the banquet: Social isolation in electronic communication media.

Döring, Nicola (1995): Isolation und Einsamkeit bei Netznutzern? Oeffentliche Diskussion und empirische Daten.