Växjö University School of Humanities English Department END185
Spring term 2006
Supervisor: Ibolya Maricic
I r R elevant and C haotic or
I ndeed R elatively C ooperative?
A Gricean comparison of chatroom and face-to-face interaction
Elisabeth Hals
Abstract
Chatroom conversations often elicit an initial impression of chaos. This is probably chiefly due to disrupted adjacency sequences, but also a result of the language being rich in non-standard
linguistic forms and grammar. This study explores chatroom conversations with reference to Grice’s (1975) cooperative principle and the maxims that accompany it, and compares them to real life conversations. The aim is to see whether they differ from real life conversations to the extent expected, and whether these differences give rise to any compensational strategies to ensure successful communication. The results reveal a slightly higher amount of maxim undermining in the chat room than in the real life conversations, but not as high as expected.
Accordingly, few compensational strategies need be adopted. It is suggested that the main explanation for these findings is that chatroom users have adapted their conversation patterns to the medium.
Keywords: Computer-mediated communication, cooperative principle, face-to-face conversation, Grice, IRC, maxims of conversation, relevance
Contents
1. Introduction ... 2
1.1 Aim and scope ... 2
1.2 Research questions ... 3
2. Literature review ... 4
2.1 Theoretical background... 4
2.1.1 The maxim of quantity ... 5
2.1.2 The maxim of relation ... 6
2.1.3 The maxim of manner ... 7
2.1.4 The maxim of quality ... 8
2.2 Previous research... 9
3. Material and method... 10
3.1 Data description... 10
3.2 Method ... 16
4. Results ... 18
5. Discussion ... 23
5.1. Undermining the quantity maxim ... 24
5.2 Undermining the relation maxim ... 29
5.3 Undermining the manner maxim... 33
5.4 Undermining the quality maxim ... 35
6. Conclusion... 37
References ... 41
1. Introduction
Anyone who has ever entered an online chatroom has probably been struck by an initial impression of total chaos. Deviant spellings and sentences without any perceivable structure, or context-related content, seem to flourish to an extent previously not encountered in any other medium. There is apparently no organisation in terms of turn-taking, as messages seem to pop up on the screen in a more or less random order. Does this mean that there is linguistic anarchy ‘in there’, or does communication in chatrooms actually display some kind of structure?
The language used in chatrooms shares many features with that of text-messages on mobile phones in that it is rich in non-standard linguistic forms and grammar (Crystal, 2001:
165; Thurlow & Brown, 2003: 6-7, 15-16). However, as language is a tool used for communication it would go against common sense to expect that the messages exchanged in such communicative situations are entirely incomprehensible. The users obviously recognise that a certain level of intelligibility is necessary (Thurlow & Brown, 2003: 16). As Herring notes, “if CMC [computer-mediated communication] were seriously incoherent, users would not flock to the Internet so enthusiastically” (Herring, 1999: 2). The fact that they do, suggests that users employ new strategies that ensure satisfactory communication, and/or that they adhere to traditional communication strategies in a way that is not obvious to the amateur observer.
1.1 Aim and scope
By now, quite a few studies have been conducted that look at how users manage communication in technologically restricted settings. However, few of these look at computer-mediated communication (henceforth CMC), that is communication via web-pages, e-mails, chatrooms and so on, from the perspective of Grice’s (1975) cooperative principle.
Grice (1975) identified and defined four maxims of conversation based on what people generally expect from their communication partners in terms of cooperative behaviour. These maxims cover the notion that conversational contributions are expected to be true (the maxim of quality), relevant (the maxim of relation), well-organised (the maxim of manner) and of appropriate length (the maxim of quantity).
This study starts from the assumptions that Grice’s (1975) maxims are being violated in chatrooms, and that the chatrooms’ anonymous environment especially encourages the undermining of the quantity and quality maxims (for further explanation, see Sections 2.1.1 and 2.1.2). The following investigation thus explores the extent to which all four maxims are
violated, and how chatroom conversations compare to real life conversations in this respect.
One might doubt whether written communication can be compared to oral communication.
However, chatroom communication resembles oral communication in many areas. For instance, chatroom conversations, like real life conversations, are time-bound. This means that a contribution is replied to relatively quickly, or not at all. Furthermore, chatroom conversations are spontaneous, loosely structured and socially interactive, all of which are distinctive features of speech (Crystal, 2001: 42). Although Crystal concludes that the language used on the Internet in general is closer to writing than speech, he notes that the language of chatgroups has been pulled towards that of speech (Crystal, 2001: 47).
Correspondingly, Werry (1996) claims that the users’ employment of different strategies to compensate for the lack of paralinguistic cues reflect a “desire to create language that is as
‘speech-like’ as possible”. The heavily abbreviated language adds to this impression (Werry, 1996: 48). Abras (2002), in turn, refers to Korenman & Wyatt (1996: 238) when she calls the language used in chatrooms ‘pseudo-oral’ (Abras, 2002: 43). Consequently, one can assume that users will bring what they know about norms of oral communication to the chatroom.
Whether they adhere to these norms, however, is the topic of this study.
Thurlow & Brown (2003) propose that mobile phone use has been subject to exaggerated worries and misconceptions. The reason for this, they claim, is a combination of
“popular discourses about young people and about new technologies” (Thurlow & Brown, 2003: 2). In line with this perception, Herring (1999: 5) warns against falling prey to
‘technological determinism’. The main aim for this study, then, is to see whether chatroom conversations are as chaotic as the initial impression indicates, or if they too have been subject to exaggerated worries and misconceptions about young people and their use of modern technology. Hopefully this study can aid in giving a broader empirically-based platform for forming opinions on the topic of computers and communication.
1.2 Research questions
With such an aim in mind, a comparison is made between a chatroom session from irc.com and real life conversations. Both samples are evaluated with reference to the Gricean maxims.
The chatroom session is then examined to see whether failures to fulfil these maxims give rise to any compensational strategies.
As mentioned above, Thurlow & Brown (2003) discovered three other sociolinguistic maxims of high importance in their study of text-messages, and claim that they also underpin much CMC. These maxims are brevity and speed, paralinguistic restitution and phonological
approximation. The main drive behind these maxims seems to be that the key purpose of text- messaging is social exchanges. In managing relationships users want to get their messages across fast, but they still do not want to be misunderstood (Thurlow & Brown, 2003: 15-16).
The same can be assumed about chatters.
When examining ways to apply the Gricean maxims to the medium of Internet, Crystal claims in relation to the manner maxim that “brevity is certainly a recognised desideratum (...), in terms of sentence length, the number of sentences in a turn, or the amount of text on a screen” (Crystal, 2001: 57). Today the IRC program has been improved to an extent where it is impossible to crash it by posting long messages1, and bandwidth in general has increased substantially. Still, the ideal of a short chat message seems to prevail. If this is the case, users will still have to make sure that enough information comes across for their messages to be understood. If they do not, it can be assumed that being understood might be less important in a chatroom than in real life situations. Consequently, the analysis addresses the following questions:
• In which ways do the study subjects fail to fulfil the Gricean maxims?
• Do the different conversation samples (chatroom and real life) display any dissimilarity in terms of how the study subjects fail to fulfil these maxims?
• Do failures to fulfil these maxims in the chatroom give rise to the use of any compensational strategies to ensure successful communication?
• Is the notion of what constitutes successful communication different in a chatroom than in the real world?
2. Literature review
This section starts by looking closer at the Gricean maxims in terms of how they were developed, how they can be applied, and how they are expected to relate to some typical features of Internet communication. Section 2.2 gives a brief overview of research on CMC particularly relevant to this study.
2.1 Theoretical background
Grice (1975) describes communication as adhering to what he calls the cooperative principle.
This means that people do not only seek to accommodate understanding when they communicate, they also expect their conversation partners to do the same. This is not to say
1 This has been confirmed by two former IRC-operators.
that we always behave in such a cooperative manner when we communicate. As Crystal (2001: 49) points out, “common experience shows that we do not”. The point is that we all seem to recognise this principle as a norm against which utterances can be judged (Crystal, 2001: 49). Grice observed that people display symptoms of such an unspoken understanding in their communicative behaviour. Furthermore, he argues that not only do people adhere to such a principle, they should do it as it is rational behaviour (Grice, 1975: 48). He assumes that people communicate for a reason, they have a goal, and to reach that goal they have to take into account the four different maxims of communication that underpin the cooperative principle. In line with this, he describes the general purpose of people’s communication as “a maximally effective exchange of information” (Grice, 1975: 47). As a result, the general cooperative principle is formulated as “make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged” (Grice, 1975: 45). The conversational maxims that follow from this are the supermaxims of quantity, relation, manner and quality, and they each have submaxims2 that state how they work.
The idea of a cooperative principle is in turn the fundament for Grice’s concept of implicature. Although this study does not mainly concern implicature, a brief explanation is necessary to fully understand how the cooperative principle works. Although we do not usually notice it, what we say does often not mean or cover exactly what we want to convey.
One of the many examples given by Grice is person B replying “there is a garage around the corner” to person A’s statement “I am out of petrol” (Grice, 1975: 51). Semantically B’s statement is incoherent with A’s statement, and might thus be seen as irrelevant (although in this case it is immediately clear to most people that it is not). However, because A expects B to be cooperative, and thus the statement to be relevant, A interprets B’s statement as implying that the garage has petrol to sell. Implicature, then, relies on the observance of the cooperative principle.
2.1.1 The maxim of quantity
The supermaxim of quantity has to do with the amount of information provided (the number of words is covered by the category of manner), and its submaxims are:
2 Grice uses the terms ‘supermaxim’ and ‘submaxim’ to separate the general idea of a maxim from its assistant descriptions.
• “Make your contribution as informative as is required (for the current purposes of the exchange)”
• “Do not make your contribution more informative than is required” (Grice, 1975: 45)
On the one hand, a failure to fulfil the first submaxim might leave the hearer(s) in limbo, not knowing what to make of the utterance. A failure to fulfil the second submaxim, on the other hand, may make the hearer(s) lose interest, or confused about the main message of the utterance.
An important feature of Internet communication is the anonymity inherent in the medium. It is possible to be both invisible and inaudible. In a chatroom, for instance, the only necessary element signalling someone’s presence is that a nickname appears on the screen when they log in. This anonymity seems to make many people less inhibited in that they disclose more and different information about themselves (Baron, 2001: 232-233), or behave worse (McCormick & McCormick 1992, Turner 1990, Wilson 1993, in Smith et al, 1997: 10), than they would in face-to-face conversations. In other words, anonymity apparently encourages undermining the quantity maxim in a manner presumably not so common in face- to-face conversations, namely excess of personal disclosure and unpleasant behaviour. Then again, it has also been suggested that the extra effort involved in typing will encourage the chatter to be more “selective in what is said” (Crystal, 2001: 57). This implies that less information might be provided in a chatroom than in a real life situation. The expectations of immediate response (Crystal, 2001: 29) should add to this selectivity.
2.1.2 The maxim of relation
The relevance category has only one maxim, namely “be relevant” (Grice, 1975: 46). A further explanation of this maxim might seem unnecessary, as the obvious meaning is ‘stick to the topic’. However, Grice admits that there are a few problems related to this maxim. For one, conversation topics can be changed legitimately. Also, the focus of relevance can shift during a conversation (Grice, 1975: 46). Accordingly, Sperber & Wilson (1995) point to the major flaw of this maxim, namely that judging what should count as a failure to fulfil it is in the end a question of intuition (Sperber & Wilson, 1995: 36). Still, it can be argued that there is a general agreement, at least within a culture, about what can be considered an irrelevant contribution, and what passes as one of the above-mentioned legitimate shifts. Although there will be exceptions, as people occasionally clearly do fail to fulfil this maxim, most people will
be able to recognise the difference between irrelevance and shift of focus or topic, even though it is difficult to describe precisely how.
In a chatroom the situation is slightly more complicated. For instance, in a real life conversation a finishing statement followed by a pause will typically indicate that a change of topic is accepted. In a chatroom, such pauses may simply imply that someone is away from their computer for a while, or engaged in a parallel activity on the computer. When they rejoin the conversation in question, they might want to add something to an exchange that took place while they were gone, even if the current exchange revolves around a new topic.
2.1.3 The maxim of manner
Unlike the other three maxims, the category of manner does not relate to what is being said, but rather how it is said. The general maxim here is “be perspicuous” (Grice, 1975: 46), which means ‘be clear’. The submaxims included are therefore:
• “Avoid obscurity of expression”
• “Avoid ambiguity”
• “Be brief”
• “Be orderly” (Grice, 1975: 46)
As mentioned earlier, there are two manners in which one can say too much. Presenting too much information is covered by the quantity maxim, producing too many words relates to the submaxim of brevity. Although it seems that these two phenomena often occur together, it is definitely possible to fail to fulfil one maxim and not the other.
Sperber & Wilson (1995) note that the fact that these maxims govern conversation explains how incomplete and ambiguous utterances can express complete and unambiguous thoughts. Given an obscure or ambiguous utterance the hearer can usually eliminate all but one interpretation by assuming that the speaker is observing the cooperative principle and the maxims (Sperber & Wilson, 1995: 34).
As already mentioned, Crystal (2001: 57-56) suggests that the strain of typing produces a “selectivity in expression [that] must lead to all kinds of inclarity”. In other words, the extra effort demanded in typing a message rather than saying it should lead to more instances of obscurity, ambiguity and disorder. Also, the time pressure involved in synchronous chatting might inflict on how much care is being taken to organise the messages.
2.1.4 The maxim of quality
The supermaxim of quality is expressed as “try to make your contribution one that is true”
(Grice, 1975: 46). It is specified in two submaxims:
• “Do not say what you believe to be false”
• “Do not say that for which you lack adequate evidence” (Grice, 1975: 46)
Failing to fulfil the first submaxim, one will have delivered a deliberate lie and thus flouted the maxim (for elaboration on different ways to fail to fulfil a maxim, see Section 3.2). In doing so, the speaker has actually exploited the fact that there is such a thing as a cooperative principle. Knowing that people will generally expect the truth, s/he can tell a lie, and still expect to be believed. Grice notes that this first submaxim holds a special position in that
“other maxims come into operation only on the assumption that this maxim of Quality is satisfied” (Grice, 1975: 46).
The second submaxim is perhaps not commonly identified with lying. However, it can be assumed that if somebody says something for which they lack evidence, the hearer(s) will expect to be informed about the doubtful truthfulness of the statement. For instance, there is a substantial difference between saying ‘I am not entirely sure that I am right, but I think the bus leaves at nine sharp’ and ‘the bus leaves at nine sharp’. For the person asking, this difference may be the difference between catching and missing the bus. This because the first statement gives him/her the opportunity to find out for him-/herself, or at least show up ten minutes early in case the bus leaves five minutes to, and not sharp.
It can be expected that the anonymity factor will also invite chatters to undermine the quality maxim. As pointed out by Wallace (1999: 51), anonymity makes it “easy to lie and get away with it”. The only thing that might stop us is if we have trouble living with our lies.
Then again, this is probably easier as long as we do not have to face the people we lie to. In contrast, most people will probably feel some level of discomfort when consciously lying to somebody’s face. Not only is it one of the early rules of childhood that one should refrain from lying (and one of the Ten Commandments), in addition many fear that their body language and/or vocal cues might expose them. This is in spite of psychological research having revealed that “most of us are poor judges of truthfulness” (Wallace, 1999: 50).
Similarly, the anonymous environment of the chatroom might lead to more failures to fulfil the second submaxim than in real life. The difference may perhaps not be as substantial
compared to the fist submaxim, as statements for which people lack evidence seem to flourish in real life conversations. For instance, many apparently think of spreading rumours as harmless.
2.2 Previous research
Over the years, countless studies, papers and books have revisited Grice’s cooperative principle, the maxims that accompany it, and the concept of implicature. Some discuss the ideas on their own terms, as for instance Sperber & Wilson’s numerous works on relevance.
Others discuss Grice in relation to a specific area of linguistics, as for instance Abras’ work on relevance in a discussion group (2002). This section presents a few of the works relevant for this study, namely Abras’ (2002) study, Thurlow & Brown’s (2003) study on young people’s text-messaging, and Herring’s (1999) survey study on coherence in CMC.
In a paper called The principles of relevance and metamessages in online discourse (2002), Abras reports from her study on the presence of metamessages in online discourse. In addition she examines whether online discourse adheres to the principle of relevance, which was developed by Sperber & Wilson (1991) on the basis of Grice’s relation maxim. The focus of the study is the online communication of a graduate level course, as conducted in a discussion group. According to Abras, Sperber & Wilson’s principle of relevance “states that the utterance given has to be relevant in order for it to be understood. It presumes that the receiver will have available the contextual information necessary to derive the meaning of the utterance with minimum effort” (Abras, 2002: 41). If the receiver does not have access to this contextual information, the utterance will not be optimally relevant, as the receiver will not be able to figure out what is being implied with the utterance (Blakemore, 1992, in Abras, 2002:
42). Based on this notion of relevance Abras proceeds to explore the discussion group’s irrelevant contributions. Her results, and why they differ from the results of this study, are discussed in Section 5.2.
In their study of young people’s use of text-messages, Thurlow & Brown (2003) examine non-standard linguistic forms and grammar in these messages. They suggest that rather than adhering to Grice’s maxims and the principle of cooperation, the language of text- messages serves “the principle of sociality” (Thurlow & Brown, 2003: 15). The sociolinguistic maxims that complement this principle are “brevity and speed, paralinguistic restitution and phonological approximation” (Thurlow & Brown, 2003: 15). The maxim of brevity and speed manifests itself in “(a) the abbreviation of lexical items (including letter- number homophones) and (b) the minimal use of capitalization and standard, grammatical
punctuation” (Thurlow & Brown, 2003: 15-16). Paralinguistic restitution is a means of replacing extralinguistic features such as stress and intonation, for instance by using all capital letters or multiple punctuation marks. Finally, phonological approximation might serve both the need for brevity and the need to signal identity, as in the spelling of ‘another’ as ‘anuva’
(Thurlow & Brown, 2003: 15-16). Interestingly, Thurlow & Brown claim that although these creative elements flourish, most of them are neither new, nor especially incomprehensible (Thurlow & Brown, 2003: 16).
In Interactional coherence in CMC (1999) Herring surveys research on cross-turn coherence in computer-mediated interaction, both synchronous and asynchronous. She finds that in chatrooms there is “a high degree of disrupted adjacency, overlapping exchanges, and topic decay” (Herring, 1999: 2). Thus, her findings do not run counter to the popular perception of such communication, namely that it is more chaotic than face-to-face interaction. What is interesting is how the chatters handle these apparent obstacles for successful communication. Herring finds that some of them adopt compensational strategies such as new forms of back-channelling, turn-change signalling and referencing (Herring, 1999: 17-19). In other words, the chatters create new means of ensuring successful communication, using the tools provided by the medium. Herring concludes that these solutions, along with the advantages some find in the loosened coherence (such as increased opportunities for interactivity and language play), explains the popularity of the medium in spite of the initial communication problems associated with it (Herring, 1999: 16).
3. Material and method
This section starts by describing and comparing the material used for this study in relation to both medium variables and social variables. The method section (Section 3.2) then describes how the different Gricean maxims are applied to this material, and considers some of the most important problems related to such an application.
3.1 Data description
The material used for this study is a transcript of real life spoken conversations from the Longman Spoken American Corpus and a chat log from irc.com. The samples were chosen on the basis of their comparability in terms of conversation topics (they should be similar), number and assumed age of participants. Another criterion for selection was that the conversations revolved around everyday topics with a low level of conflict. This last criterion was applied on the basis of two assumptions. First, conversations only involving experts on
the conversation topic could result in less or more maxim undermining than can be found in everyday conversations. For instance, if all conversation participants know the topic well, expectations of relevance may be lowered. Second, a high level of conflict might lead to more failures to fulfil particular maxims, as for instance the manner maxim when someone desperately tries to hold the floor. Had ‘expert’ conversations or conversations with a high level of conflict been chosen, it would probably have been more difficult to find comparable data.
The data sample of the real life conversations consists of 2018 words, and the chat log sample comprises 1728 words. In a material of this size, a difference of approximately 300 words seems significant. However, one of the real life conversation participants tells a story over 450 words long. After 150 words his conversation partners interrupt him in a manner that suggests that he is being too talkative, and correspondingly failing to observe the manner and quantity maxims. As a result, the remaining 300 words are merely a continuation of this maxim undermining. The fact that the real life conversation sample is 300 words longer than the chat log sample is therefore negligible. The two data samples are compared below according to different medium and social variables.
To secure the anonymity of the participants of this study all names have been altered in both the chatgroup and the real life group. As the names and nicknames of the participants are of some importance to the analysis, the alteration has been conducted bearing in mind that the new names should give approximately the same associations as the authentic ones.
The medium variables are used to describe the material from a technical point of view.
This implies that they are the basis for a comparison of the data samples with regard to such features as channel of communication, filtering of communication (by a moderator), anonymity, and synchronicity. The results of this comparison are presented in Table 1.
Table 1. Data comparison with regard to medium of communication
Medium variables Chat session Real life (RL) conversation
Communication channel Textual Multi-modal
Filtering By moderator No filtering
Anonymity Full anonymity possible Anonymity impossible
Synchronicity Synchronous Synchronous
As shown in Table 1, there are two main differences between the two data groups in terms of medium. The first is that whereas the participants in the chatroom rely fully on text for their
interaction, the members of the real life group communicate with each other face-to-face. The real life group can thus communicate through body language and face expressions, as well as with actual words. In relation to this study, however, it should be noted that the analysis of the real life conversation is based on a written transcript. As a result, some elements relevant to this study (such as body language and face expressions) may have been lost. Still, this does not change the fact that the people participating in this conversation have had access to all the extralinguistic cues the chatroom participants have not. This can be expected to cause dissimilarities between the two groups in terms of how they communicate.
The second main difference is a result of the above described disparity when it comes to channel of communication. Whereas full anonymity is possible in a chatroom, there are no possibilities for anonymity in real life interaction. Even if the participants did not know each other’s names, they would still be held responsible for their communicative activities in that they literally have to face their communication partners. In contrast, the chatroom participants are free to choose how much they want to reveal about their real life persona. The only exception is of course if an online persona is recognised by someone who knows the person behind it in real life. In that case, personal information may be published without someone’s consent. This might be the case in the chatroom sample used for this study, as can be seen in example (11), Section 4, below.
Another difference between the two samples is the possibility of filtering communication in the chatroom. This means, among other things, that unwanted participants can be denied access to the chatroom by a moderator if their contributions are considered unwanted. Although it is possible to exclude people from real life conversations as well, this is usually more difficult and not at all as common as in CMC. When it comes to synchronicity, both fora are generally synchronous, although the technology involved might mean that there can be a short time lag in the chatroom.
The social variables are used to describe the material in terms of social context.
Accordingly, they cover such areas as participation structure, participant characteristics, purpose, topic, tone and norms of communication, and linguistic code. The results of this comparison are presented in Table 2 below.
As mentioned above, the degree of anonymity is one of the main differences between these two groups. The chatroom offers a possibility for complete anonymity, whereas there is no anonymity in the real life group. However, chatgroup participants may vary in how they choose to approach this possibility, as Table 2 indicates. From the exchange used for this study, one can easily get the impression that some of the participants know each other both
Table 2. Data comparison with regard to social context
Social variables Chat session RL conversation Participation structure Many-to-many
Public channel
Uncertain degree of anonymity 28 participants
188 turns
Even rate of participation Seven dominant participants
Many-to-many Public place No anonymity 6 participants 187 turns
Rate not possible to detect Three dominant participants Participant characteristics Apparent female domination
Aged 20-40
Participants familiar with group Some difference in status
Even gender distribution Aged 20-30
Participants familiar w/group No difference in status Purpose of group
of interaction
Debate Social
Social Social Topic of group
of interaction
Christian discussion
Food, weather, personal past
No given topic
Storytelling, personal past
Tone Relaxed
Casual
Generally friendly
Relaxed Casual
Generally friendly
Norms No explicit norms of conduct No explicit norms of conduct Linguistic code Standard English
Some Netspeak elements
Standard English
offline as well as online, as can be seen in example (1):
(1) <Queen> birdie john comes home today?
Although ‘BirdD’ (‘birdie’s screen name) might have told ‘Queen’ about ‘John’s homecoming in the chatroom, and they thus do not have to know each other outside the chatroom, the level of familiarity in such a question suggests that ‘Queen’ knows ‘John’ in real life. It could of course also be that ‘Queen’ knows ‘John’ from the chatroom, and knows that ‘John’ and ‘BirdD’ know each other in real life.
Another difference between the two groups is that the chatroom has an explicit purpose for its existence, namely the purpose given in its name, ‘Christian discussion’.
However, the exchange used for this study displays no signs of the participants engaging in any typically Christian conversation. The material is therefore comparable with the various topics in the real life group, that sit around a picnic table sharing fictional and true stories from their past.
What may look like a significant difference between the two groups is the number of participants. However, there are only seven chatroom participants who contribute regularly throughout the session, or that take part in any kind of conversation. The others typically log on, say ‘hi’ or deliver a quick remark, and then remain quiet throughout the session (the phenomenon of presumed silent participants is discussed in Section 5.1). This makes the chatroom session comparable to the real life group with its six participants. Although three of these are more active than the others, this does not mean that they stay quiet, as is the case in the chatroom.
The two groups are also similar when it comes to age, although the dominant chatroom participants generally seem slightly older than the participants in the real life group.
The real life group participants’ ages have been collected from the data material, as they have to provide information about where they come from and how old they are. There is no reason to suspect that they lied when providing this information. Additionally, their age can be estimated from what they say, and how they say it. For instance, as will be discussed later on, bragging about one’s past drug abuse is not typical behaviour once one has passed 30, although exceptions do of course occur.
This is how the age of the chatroom participants has been estimated, as there are no other means of obtaining this information. The fact that they get into the details of how to make a chicken and barley casserole, that one of the participants corrects a misspelling of chicken although spelling errors are commonly excused in chatrooms (Crystal, 2001: 88), that one of the participants’ screen name includes the title ‘Mrs’, and finally that one of them creates, and all of them participate in a forum called ‘Christian discussion’, all add up to the impression that at least some of them may have passed the age of thirty-five. It is of course possible to discuss casseroles, correct misspellings, be married, participate in a Christian forum, and still be well under the age of thirty-five, but the impression prevails.
Another impression of the chatroom participants that separates them from the real life group is that the group apparently is dominated by female participants, and three of these seem to enjoy a slightly higher status than the others. Again it has to be stressed that almost
every assumption made about the chatroom participants are just that, assumptions. Still, considering the topics of the forum, it is not unlikely that the participants who use screen names associated with either males or females are of the corresponding gender in real life.
Thus, the two most dominant participants ‘BirdD’ and ‘Queen’ are probably women. That
‘BirdD’ has a high status in the group should not come as a surprise, as she is the creator and moderator of the group. However, it seems that she shares this position to some extent with
‘Queen’. This is manifested in the way ‘Queen’ relates to the other participants, as is evident in the analysis below. Similarly, three people dominate the conversations in the real life group. Yet, this verbal dominance seems to be more a result of who is more talkative, than who enjoys a higher status.
Otherwise the two groups are similar in that they are both examples of a publicly accessible many-to-many conversation. In relation to this study, however, it is worth noting that it might be difficult, or even impossible, to get access to these conversation samples. The chat log taken from irc.com has been removed from the log site, and the real life conversation sample is only available for those who have access to the Longman Spoken American Corpus.
Another similarity is that there are no explicit norms of conduct, although one of the chatroom participants is kicked out for accusing one of the other participants of being a child molester.
Finally, although they apparently speak the same variety of English, i.e. American English, there is an essential language difference between the two groups. Netspeak is one of the many terms coined to acknowledge the agreement on the fact that there are language elements specific to the medium of Internet (Crystal, 2001: 17-18)3. Some of these elements are present in the chatroom sample, along with other language elements commonly found on the Internet, although not necessarily restricted to it. For instance, there is very little use of capitalisation, even in names and first person singular pronouns. Exceptions to this are instances where capitalisation is used as a means of paralinguistic restitution, as described above in Thurlow & Brown’s study on text-messaging (Thurlow & Brown, 2003: 16). A case of such paralinguistic restitution is demonstrated in example (2):
(2) <KittyKitten> geez do you have any GOOD news today?
Here capitalisation is used to signal stress on the word ‘good’. Other examples of paralinguistic restitution are excess use of punctuation marks (to achieve similar effects as
3 See Crystal, 2001: 81-93 for more on Netspeak features.
seen in example (2)), so-called emoticons (such as ;-( and :-/ ), and described actions preceded by an asterisk, so-called emotes. This last phenomenon can be seen in example (3):
(3) *Queen crosses her eyes
All the above examples of paralinguistic restitution compensate for the fact that the chatroom participants cannot see or hear each other, thus this disparity from the real life group is obvious.
Another feature described by Thurlow & Brown (2003: 7), is that of phonological approximations. There are a few of these in the chatroom sample, all of which at the same time serve the purpose of brevity. One example is ‘dunno’, which in addition may be used to signal that the speaker is not really interested. There are also a few examples of abbreviations, such as ‘k’ instead of ‘ok’, and ‘lol’ for ‘laughing out loud’. Still, it cannot be said that there is much Netspeak in the chatroom sample. The language variety should thus be comparable to that of the real life group.
3.2 Method
Grice lists four different ways in which one can fail to fulfil the maxims of quality, quantity, relation and manner (Grice, 1975: 49). They can be violated quietly, by which Grice seems to imply unintentionally, or at least unconsciously. The cooperative principle can be opted out of, either by saying or indicating clearly that one will not adhere to a certain maxim.
Moreover, there may be a clash of maxims, as for instance when someone is not able to fulfil the maxim of quantity without failing to fulfil that of quality. Finally, maxims may be flouted, that is violated on purpose, and thus inherently exploited. In other words, there are many ways of undermining these maxims, and there may be different reasons for doing so.
Both Grice (1975: 53-54) and Wilson & Sperber (2002: 219) note that different tropes, such as metaphors, hyperboles and irony, are exploitations of these maxims. If the hearer does not expect relevance, for instance, s/he will not be able to understand the intended meaning of a metaphor. In this sense, tropes are exploitations of the cooperative principle.
Brown & Levinson (1987) look at politeness as another motive for not observing Grice’s maxims. They claim that in general people do not only cooperate to maintain efficiency and rationality in communication, but to mutually maintain face, i.e. public self- image (Brown & Levinson, 1987: 61). This suggests that in a situation where not threatening someone’s face implies a violation of one or several of Grice’s maxims, politeness may be
preferred over efficiency. Vice versa, concern for a conversation partner’s face needs might be ignored in order to preserve efficiency (Brown & Levinson, 1987: 62). Lakoff (2005) claims that the former is more often the case than the latter. When people have to choose between clarity and politeness, they normally choose politeness (Lakoff, 2005: 8). Grice (1975) also recognises politeness as a force governing conversation. The reason why it is not included in the cooperative principle is that the maxims here are those concerned with what Grice sees as the primary purpose of talk, namely a “maximally effective exchange of information” (Grice, 1975: 47).
One might think that choosing politeness over adhering to Grice’s four maxims should count as a violation of the cooperative principle. However, such choices are generally recognised for what they are. This means that in most situations hearers will not think of what is said as uncooperative behaviour, even though a politeness phrase might clash with the brevity aspect of the manner maxim. Similarly, tropes will be recognised as tropes, and not seen as irrelevant or untruthful. With this in mind, this study is limited to analysing situations where undermining the Gricean maxims lead to conflicts. In other words, the focus of this study are situations where fluency in communication is disrupted because someone fails to observe the cooperative principle.
When studying people engaging in a face-to-face conversation, it is usually not very difficult to spot when they fail to fulfil Grice’s maxims in the manner described above. The social context will give the observer clues about what is and what is not considered appropriate conversational behaviour in the particular conversations studied. If not, the participants should reveal it through their responses to what is being said, either verbally, physically, or both.
Needless to say, the situation becomes a bit cloudier when working with written material. In a transcript of spoken conversation there are few, if any, extralinguistic cues given4. In the cases where someone’s failure to fulfil a maxim is signalled either by the speaker’s or the other participants’ body language, the observer might not notice it. However, this does not mean that it is impossible for an outsider to judge such situations. People will often signal verbally that a maxim has been undermined. For instance, someone who gives too much information might be interrupted. When someone gives insufficient, or unclear, information they will be asked to elaborate. Someone who makes an irrelevant contribution will usually be asked to explain what s/he means, or may even be ignored. Lying, however,
4 The extralinguistic cues given in the transcript used for this study are ‘clears throat’, ‘sighs’, ‘laugh’ and
‘laughing’.
holds a special position in this context. First, it can be difficult to detect, as one would have to know that the speaker knows that what s/he is saying is, or could be, untrue. Second, because people tend to give each other the benefit of the doubt, they rarely accuse others of lying. This suggests that it can be difficult to detect when a violation of the quality maxim has occurred.
There are also two other problems related to the real life conversation transcript. The first is that some words are missing because they are unclear on the tape. This was not a problem in the material used for this study, as none of the instances where a maxim is undermined are connected to utterances that include such unclear elements. Conversely, utterances that do include unclear elements do not elicit any signals from the hearers that a maxim has been undermined. The second problem is more relevant to this study. As mentioned earlier, there is little additional information in the transcript as regards tone of voice, body language, and such. As a result, it is sometimes impossible to understand what is meant by certain utterances. In addition, the lack of response from the other participants to such utterances can be ambiguous. They might not understand what was meant either, and ignore the utterance, or they understand and respond using signals not detectable to the reader of the transcript. The solution to this problem was to choose data material where the intended meaning of an utterance is evident from the text. This means that whole conversations including such unclear utterances were excluded from the analysis. This should not skew the results and the following analysis, since the main purpose of this study is not to compare how the two groups differ when it comes to signalling maxim undermining, but rather at which rate and how it occurs.
When observing a chatroom conversation, the situation is a little less cloudy as the participants there operate on more or less the same terms as the observer. There are no possibilities for signalling to each other through body language or face expressions, although some remedies such as smileys have been invented to compensate for this. However, if one wants to signal to participant X that participant Y is talking too much, one cannot simply use an adequate smiley as participant Y will see it too. The only opportunity here is to wait until Y has left the room, or go into whisper mode (i.e. exchange messages in a parallel chatroom with restricted access). Then of course the observer will not know what is being said, as the
‘whisper’ can only be seen by the participants in it.
4. Results
This section presents a frequency count of maxim undermining in the chat session and the real life conversation sample. This count is accompanied by an examination of how the two
groups differ, both with regard to the amount of undermining, and the manner in which it occurred. The results of the frequency count are presented in Table 3.
Table 3. Distribution of maxim undermining in real life and chatroom samples Supermaxim undermined Chat
session
RL
conversation
Total %
Quantity - not enough information - too much information
3 5
5 2
8 7
30 26
Relation 7 - 7 26
Manner - 4 4 15
Quality - 1 1 4
Total 15 12 27
% 55 44 ≈ 100 %
As Table 3 shows, more maxims are undermined in the chat session than in the real life conversation. However, the difference is not as significant as one could have expected, a fact that will be reverted to.
The maxim most often undermined in the real life conversation sample is the maxim of quantity. This is chiefly a result of someone not providing enough information to be understood. It is usually signalled by a request for more information, as can be seen in example (4):
(4) <Jim> Totally given up all of my illegal pursuits.
<Steve> What illegal pursuits, you have none do you?
<Jim> I used to do drugs.
This example also illustrates a phenomenon common in both the real life conversation and the chatroom, namely that someone appears to deliberately hold back information with the purpose of attracting attention from the others. As such, it is a typical example of a maxim being flouted and exploited. In the exchange example (4) is taken from, the conversation originally revolves around a ranger having asked them to put out the candles because of fire hazard. ‘Jim’ exploits the situation to make himself more interesting, because the others obviously want to know what other kinds of illegal business than lighting candles he has been
involved in. He could have been interpreted as trying to be funny, but the piece of information he wants to attract the others’ attention to is obviously not a joke.
Failing to fulfil the quantity maxim is also the most common maxim undermining in the chatroom, although here the provision of too much information is slightly more common than not providing enough information. A case of such excess information can be seen in example (5):
(5) <Queen> she's not renewing?
<piffy> not for a year
<piffy> and her fiancee is moving in
<piffy> 3 people in a one-bedroom is too much
<piffy> and i dont even get a bedroom
<piffy> so by not doing an actual lease, she has to pay more per month.
<piffy> its not worth it to me
<piffy> esp. since i have no money
<Queen> oo smells good
This exchange is taken from a short conversation about ‘piffy’ having to move out. It seems that ‘piffy’ mistakes ‘Queen’s polite interest for heartfelt care. As a result, the elaborate messaging about his/her housing situation is rewarded with ‘Queen’ giving ‘piffy’ a cold shoulder in the shape of an eight minute long silence, only to return to the previous topic of her making a casserole. The interpretation that ‘Queen’ merely expresses polite interest is supported by the fact that she ignores ‘piffy’ throughout the rest of the chat session. ‘Queen’s response could have been an innocent attempt at topic change, but as she so suddenly and clearly ignores ‘piffy’s messages, one could say that she is flouting the maxim of relation.
This situation results in two other instances of flouting this maxim, as these two participants engage in a short verbal battle over who gets to control the conversation, and make irrelevant contributions on the way.
Another instance of flouting the relation maxim occurs when ‘Tiffany’ in the middle of a conversation about one of the other participants’ illness posts the message rendered in example (6):
(6) * `Tiffany stares at her roses
Similarly she posts the message seen in example (7) in the middle of a conversation about a treasure map:
(7) <`Tiffany> haha
<`Tiffany> i gave my dog a carrot to play with
<`Tiffany> and he's rubbing his neck on it
<`Tiffany> lol
What she might hope to achieve by posting these messages will be discussed in Section 5.2.
As Table 3 above shows, there are no irrelevant contributions in the real life conversation sample.
There are no failures to fulfil the manner maxim in the chatroom sample, whereas there are four instances of someone failing to observe this maxim in the real life conversation sample. One occurs when ‘Steve’ is telling a story. Half way through it, two of the people he is telling the story to engage in a short conversational exchange, as can be seen in example (8):
(8) <Steve> (...) the guy that was interviewing him didn't want to admit that he didn't know what a kluge maker was so he takes the test, passes them with flying colors, I mean just blows the test away so they put him on a ship you know, he goes to OCS first of all, becomes an officer, all the way through people say well you know what's your, what's your training at
<Christa> Are you signing these?
<Steve> I'm a kluge maker, so they stick him on a battle ship right, he's in the middle of the ocean
<Jim> [Look at how I filled mine out]
<Steve> [And of course he's registered as]5, he's registered as kluge maker, first class (...)
As mentioned earlier, general cooperative behaviour does commonly not only include efficient communication, but also concern for the conversation partner’s face. Brown &
Levinson (1987) list interrupting and showing non-attention as two of many face-threatening acts, and claim that they are blatant acts of non-cooperation (Brown & Levinson, 1987: 65- 67). In this light it is clear that the people interrupting ‘Steve’ are sending a strong signal, namely ‘if you do not cooperate and wrap it up soon, we will not cooperate either’. Had it been necessary for ‘Christa’ to get the information she is requesting immediately she could
5 The square brackets indicate which words overlap.
have been excused, but in this case it is not. This signal from ‘Christa’ is repeated at the end of the story, when she interrupts again to finish the story herself, as example (9) shows:
(9) <Steve> (...) now he knows the jig is up, he's walking up to the Commanding Officer and he's scared to death and he's shaking so bad the thing drops out of his hand and goes [overboard and goes kluge]
<Christa> [Into the ocean] <nv_laugh> that's bad.
Although ‘Steve’ is clearly trying to create suspense by carefully adding layers of detail to his story, ‘Christa’ will not allow him to reach the climax of the story himself. This is actually an even stronger signal, as she ruins the moment for him. Also, by replying “that’s bad”,
‘Christa’ reveals that she has probably not listened to the story, since the main character in the story was saved from humiliation and losing his job because the ‘thing’ fell overboard.
Therefore a more natural response from ‘Christa’ would have been something along the lines of ‘good for him!’. All these signals, then, suggest that the storyteller could have used fewer words in telling the story. This impression is supported by the way the story is told, as the storyteller repeats himself several times, and also adds information that is not really necessary to get his point through. This way he also partly violates the quantity maxim.
Another failure to fulfil the manner maxim in the real life conversation occurs when someone is being obscure, as seen in example (10):
(10) <Eve> Are you going to call me splatter any time this weekend?
<Steve> Sclatter?
<Christa> What's sclatter?
‘Eve’ aims this question at her fiancé ‘Jim’, but as the question itself really has no point it can safely be assumed that what she really wants to accomplish is attracting attention from the others, something which she also does. In other words, ‘Eve’ is deliberately being obscure, and in doing so she provides another example of a maxim violation with the purpose of gaining attention.
Finally, despite the earlier-mentioned problems related to detecting someone failing to fulfil the quality maxim, there seems to be two such instances in the material used for this study, one in each sample. The possible maxim undermining in the real life conversation sample concerns the illegal business conducted by ‘Jim’ in example (4) above. As is probably his intention, he captures the others’ attention whereupon he reveals that he “used to do
drugs”. The apparent failure to observe the quality maxim occurs when ‘Jim’ claims that he smoked marihuana for several years without ever managing to “get high”. He then suggests that this was because he had so little body fat that he could not reach the level of THC (the main psychoactive ingredient in marihuana) necessary to feel any effect. The problem, and source of possible maxim undermining, is that it is impossible to smoke marihuana and not feel the effect of the intoxication.
There are also elements to suggest that an undermining of the quality maxim takes place in the chat room. Here one of the participants accuses another participant of being a child molester, as example (11) shows:
(11) <qpd> hey all. ...
<qpd> übertech is a known child molester <übertech> thats right
<qpd> [first name] "tiny cock" [last name] is registered in the state of florida
<qpd> as a level 1 sex rapists
<qpd> look him up on the FL registry if you don't believe me
<BirdD> ok
<BirdD> enough
<übertech> can you ban him please?
* BirdD sets mode: +b *!*@71-35-54-72.phnx.qwest.net
* qpd was kicked by BirdD (qpd`)
What makes this appear as a lie is the use of foul language (which makes it seem like the main purpose is to verbally abuse rather than accuse), and the fact that the group moderator kicks the accuser out of the chatroom. How these two instances compare in terms of undermining the quality maxim is discussed in Section 5.4.
5. Discussion
The following discussion is divided into sections according to the different maxims. This is to facilitate the reader’s navigation between the different parts of the paper, and to ease the reading of the actual discussion. Each section discusses probable causes for why the maxim in question has been undermined in the chat session and real life conversations respectively, and suggests reasons as to why the two groups differ, if so is the case. Where there are signs of compensational strategies being employed to secure successful communication, these are also explored.
5.1. Undermining the quantity maxim
Although there are three instances of someone not being sufficiently informative in the chatroom sample, there is little evidence to support the assumption that typically short chat messages create more situations than real life conversations where an utterance does not contain enough information to be understood. As example (12) demonstrates, there is not much in this exchange that separates it from a typical real life exchange:
(12) <piffy> i have 2 months to find a new home bleh
<Queen> why??
Of course, the lack of capitalisation, and the use of a numeral instead of spelling out the word
‘two’, is evidence that chatters seek to reduce the strain of typing. However, neither of these elements cause the failure to fulfil the quantity maxim. The lack of sufficient information does not follow from the Netspeak elements involved, but rather from ‘piffy’s apparent attempt at capturing someone’s attention. Consequently, this situation is theoretically no different from a real life situation.
The second instance of someone failing to provide enough information in the chatroom sample is also a result of a probable attempt at attracting attention. This participant actually writes five separate posts to attain this effect, as illustrated below in example (17), Section 5.2. She is awarded with no less than four responses to her cryptic message.
The third instance is a short message resulting in a request for more information, as shown in example (13):
(13) <MrsAMD> ineed suggestions
<Queen> for?
<MrsAMD> a password for a game
However, it is highly questionable to count this as a failure to observe the quantity maxim.
Chances are ‘MrsAMD’ was already typing her second post when ‘Queen’ asks for elaboration. This is suggested by several facts. First, who would ask for suggestions without saying what for? Of course, that could happen in a chatroom if the intent was to spoof, that is to send an odd message anonymously for instance to irritate others (Crystal, 2001: 52). This is obviously not the case here, as ‘MrsAMD’ follows up this initially odd-looking message.
Second, judging from this chat session it is quite common to break up longer messages into
several posts. The reason why an otherwise seemingly experienced chatter like ‘Queen’
interrupts this sequence is probably that it takes too long before ‘MrsAMD’ explains what she needs the suggestions for. Third, this last assumption is partly supported by the fact that
‘MrsAMD’s second post appears only seven seconds after ‘Queen’s question, and not everyone can type that fast. Consequently, none of these three instances of someone failing to provide sufficient information is a result of the message, or the post, itself being too short.
Crystal (2001) suggests that a phenomenon present in most chatrooms called lurking should count as a form of undermining of the quantity maxim. Lurking is accessing a chatroom without contributing to the conversation, and Crystal likens it to “a refusal to communicate” (Crystal, 2001: 53). Admittedly, there is evidence of such behaviour in the chatroom sample. The program automatically posts a message whenever someone enters the chatroom, and in this sample a couple of participants enter without posting any messages during the six hours the sample session lasts (although they might be communicating in whisper mode). Still, this was not counted as instances of undermining the quantity maxim for the following reason. As cited earlier, one of the two quantity submaxims says “make your contribution as informative as is required (for the current purposes of the exchange)” (Grice, 1975: 45). Crystal seems to interpret ‘the current purposes of the exchange’ as meaning
‘participation in a chatroom conversation’; thus anyone who is present in a chatroom without contributing is failing to be ‘as informative as is required’. However, all of Grice’s examples (1975: 51-56) display a much more limited view of what constitutes a ‘current purpose of exchange’. These examples concern short conversations dealing with one topic, typically between two or three people. In other words, they are communicative situations limited both with regards to topic, time, and number of participants. It therefore seems that Grice would not consider a chatroom conversation to be one exchange, as Crystal does, but rather see it as several exchanges taking place in one forum. Admittedly, one can view the absence of contributions to a chatroom as uncooperative behaviour, as one would if someone in a real life social gathering failed to say anything throughout a whole day. Yet, there are many kinds of uncooperative behaviour that are not covered by Grice’s maxims, as for instance the earlier- mentioned impoliteness. As stated in Section 2.1, Grice’s cooperative principle is based on the notion that the main purpose of communication is “a maximally effective exchange of information” (Grice, 1975: 47). The lurkers in this chat session are apparently not involved in any information exchange at all (unless they communicate in whisper mode), they are merely present where such information exchange takes place. Therefore, their anti-social and uncooperative behaviour is not uncooperative in the strict Gricean sense.
As established above, the fact that chat messages or posts are typically short does not necessarily affect negatively the amount of information they contain. Thus, it should be no surprise that there is little difference between the real life conversation and the chatroom conversation on the matter of not providing enough information. Similar to the chatroom sample, the real life conversation sample contains two instances of someone more or less accidentally not being sufficiently informative for their conversation partner to understand why they are saying what they are saying. This is demonstrated in example (14):
(14) <Jim> Uh bugs <nv_clears throat> I put down a document that you can look at for influence.
<Christa> For what?
<Jim> On the back of the blue form. You ought to check that book out. It's a, it's a dictionary of words that are used in my sort of
Here ‘Jim’ is not providing enough information for ‘Christa’ to understand that he is still talking about the forms they are filling out, even though they are in the middle of a conversation about them. One might think that ‘Jim’ is being unclear, and thus failing to fulfil the manner maxim. Nevertheless, if one looks at the submaxims of the manner maxim it becomes evident that such is not the case. Given the context, ‘Jim’ presents an orderly, unambiguous, and otherwise clear utterance. It is a piece of information, namely ‘when you fill out your form’, that is missing.
The other three instances of someone failing to provide enough information in the real life conversation sample are similar to two such instances from the chat session in that someone apparently withholds information for the purpose of creating interest. One of these instances from the real life conversation sample is illustrated above in example (4), Section 4.
This means that in more than half of the eight instances (i.e. chatroom sample and real life conversation sample counted together) where someone fails to provide sufficient information, this submaxim is being flouted and exploited. Knowing that hearers expect cooperative behaviour, and accordingly messages that can be understood, the speakers seems to deliberately withhold information to make their hearers curious. Of course, one can never say for sure what someone’s communicative purpose is. Nevertheless, the fact that there are no other equally likely interpretations of these utterances strongly suggest that they are aimed at attracting attention. Why people feel the need to create such extra interest will not be dealt with here, but it should be noted that such behaviour is evidence that expectations of cooperative behaviour exist both offline and online.
