Politeness in Task-Oriented Dialogue

Andrew John Merrison


Human communication is a process involving interactive and collaborative effort and its success is dependent on the joint responsibility of all participants involved (Grice, 1975; Clark, 1996). Furthermore, it is often assumed that while part of this effort concerns the management of transactional wants, it is another, entirely separate part which is concerned with the management of face wants (Goffman, 1967b). In short, there is apparently one fundamental organisational principle for talk which is transactional (task-related), and another for that which is interactional (social). What such a view fails to realise, however, is that in certain circumstances part of the collaborative effort required in transactional dialogue can be seen to involve the management of face.

This paper investigates the role of face in task-oriented dialogue. In doing so it suggests a maxim of interactional organisation which recognises the interplay of face wants with transactional wants. Various ‘sensitive’ behaviours are seen to mitigate threats – either threats to the success of the transaction or threats to face. Consequently, and analogous to Brown & Levinson’s (1987) notion of face-threatening act (FTA), this paper introduces the term transaction-threatening act (TTA).

More specifically this paper investigates the role of face in task-oriented dialogues between aphasic and non-aphasic individuals. Results indicate that when engaged in talk with aphasic dialogue partners, non-impaired speakers sensitively manage potential FTAs and TTAs:

(a) by simplifying the interaction by avoiding the generation of unnecessary talk; and

(b) by avoiding highlighting any non-competence on the part of their impaired interlocutor.

Such management is found to be achieved both by active and inactive strategies of restoration. More importantly, much of this restoration involves face-work which obtains despite the very clear transactional aspect to the task-oriented dialogue in which the participants are engaged.

In short, we find that even when engaged in task-oriented dialogue, speakers can be seen to invest a great deal of effort into doing politeness.

Key Words: face, politeness, transactional discourse, sensitivity, aphasia, (perceived) non-competence, Merrison’s maxim


1.0 Introduction

Human communication is a process involving interactive and collaborative effort and its success is dependent on the joint responsibility of all participants involved (Grice, 1975; Clark, 1996). Furthermore, it is often assumed that while part of this effort concerns the management of transactional wants, it is another part which is concerned with the management of face wants (Goffman, 1967b). As Kasper (1990: 205) notes:

Transactional discourse types focus on the optimally efficient transmission of information … Conversational behaviour that is consistent with the requirements of transactional discourse will thus be characterized by close observance of the Cooperative Principle [Grice, 1975]. Interactional discourse, by contrast, has as its primary goal the establishment and maintenance of social relationships. In interactional discourse, therefore, the Cooperative Principle is regularly overridden by the Politeness Principle [Leech, 1983] in order to ensure that participants’ face wants are taken care of.

In short, there is apparently one fundamental organisational principle for talk which is transactional (task-related), and another for that which is interactional (social).

What such a generalised view fails to realise, however, is that despite the fact that the transactional nature of task-oriented dialogue can legitimise bald, on-record ‘politeness’ strategies (Brown & Levinson, 1987), in certain circumstances, part of the collaborative effort required in transactional dialogue can still be seen to involve the management of face.

In this paper I investigate the role of face in task-oriented dialogue between interactants with different levels of knowledge, expertise and linguistic competence, and in so doing I suggest a maxim of interactional organisation which recognises the interplay of face wants with transactional wants.

More specifically this paper investigates the role of face in task-oriented dialogues between aphasic and non-aphasic individuals. However, although the paper deals with aphasia, it is not about the linguistic abilities of aphasic individuals per se. Rather, it is concerned with:

(i) how non-aphasic dialogue partners manage task-oriented interactions with aphasic individuals,

(ii) how their behaviour can be seen to be compensating for the apparent linguistic deficits of their aphasic interlocutor; and, most importantly for this paper

(iii) how that compensation relates to the issue of face management.

However, before addressing the face management issue I first explain what is meant by aphasic discourse and task-oriented interaction. This is done in the following two sections (§2 and §3).

2.0 Aphasic discourse


This paper reports on eight interactions between dyads of previously unacquainted aphasic and non-aphasic individuals (referred to as ‘Aphasic Dialogues’ or ADs). As control data I have analysed eight dialogues between dyads of previously unacquainted non-aphasic individuals (‘Control Dialogues’ or CDs). (See §3.2 for further details.)

Although this is clearly not the place to provide an intricately detailed account of aphasia, it is important for readers to be at least broadly acquainted with this condition.

Aphasia can be defined as the impairment of central language abilities following brain damage. This brain damage may be the result of tumour, trauma, infection or cerebro-vascular accident (the latter more commonly being referred to as ‘stroke’).

Language impairment may be 'more or less complete' (Garman, 1990: 416) and it is possible, in principle at least, to distinguish between partial loss of language (dysphasia) from total loss (aphasia). In practice, however, since total loss of language ability is 'a relatively uncommon and transitory condition' (Garman, 1990: 417), these two alternative terms tend to be used interchangeably. Of the two, though, the more widely used is aphasia as it is this term that has been adopted for the nomenclature of the related research field: aphasiology. Therefore aphasia is this term that I will use in the course of this paper. (For more detailed accounts of the nature of aphasia, see Garman, 1990; Lesser, 1989; and Lesser & Milroy, 1993.)

2.1 Non-competence


Despite having introduced the term aphasia, it is not so much the linguistic nature of the aphasic individuals’ deficits that are pertinent to this paper. Although I would prefer to not use these terms (because of their clear emotive collocations), what this paper is concerned with is the role of face and politeness (see §4) in dialogues between competent and non-competent interlocutors. Many of the points raised here may therefore (to varying degrees) also be relevant to interactions between, say, native & non-native speakers, adults & children, and other ‘experts’ & ‘novices’.

3.0 Methodology

3.1 Task-oriented dialogue

All recordings are of pairs of participants engaged in a task that was designed to elicit natural, yet restricted dialogue. The task in question (developed by Brown, Anderson, Schillcock & Yule, 1984) is known as the Map Task (see also Anderson, Bader, Bard, Boyle, Doherty, Garrod, Isard, Kowtko, McAllister, Miller, Sotillo, Thompson and Weinert, 1991a).

The Map Task has been widely used to support the study of spontaneous speech and communication: it has been used to investigate the language and communication abilities of children (Anderson, Clark & Mullin, 1991b, 1992, 1994; Doherty-Sneddon & Kent, 1996), children with moderate learning difficulties (Lamb, Bibby & Wood, 1997; Lamb, Bibby, Wood & Wood, 1998), non-aphasic adults (Anderson & Boyle, 1994; Boyle, Anderson & Newlands, 1994; Davies, 1997; Kowtko, 1997; Sotillo, 1997), sleep-deprived soldiers (Bard, Sotillo, Anderson, Thompson & Taylor, 1996) and aphasic adults (Merrison, 1992; Merrison, Anderson & Doherty-Sneddon, 1994; Beeke, Dean, Kilborn, Anderson, Robertson & Miller, 1994; Anderson, Robertson, Kilborn, Beeke & Dean, 1997; Merrison, 1998).

Two dialogue partners each have a schematic map drawn on an A3 sheet of paper (see Appendix for examples of maps). The participants sit opposite each other at a table (resembling a double-sloping lecturn) so that neither can see the map of the other participant.

The task involves one participant (designated the Information Giver (IG)) describing a route marked on his map (the IGs in my study were all male) to the other participant (designated the Information Follower (IF)) whose map is lacking the route. The IG’s ultimate aim is to get the IF to successfully (accurately) draw the route onto their map without actually showing them his own map.

Although both IG and IF have copies of the basic map, there exist some differences between the two aside from the fact that only the IG’s map has a route marked on it. Specifically the IG’s map has three landmarks which are absent from the IF’s map, which in turn has three landmarks that are not on the map of the IG. Thus, in total, there are six ‘problem’ points to be discovered en route.[1] The reason for the existence of these landmark mismatches is to set up a genuine information gap between the participants.

The nature of the drawing task was explained to the participants and they were also made aware of potential discrepancies in order to foster collaborative, interactive and genuine communicative strategies. After checking that they had understood what they had to do, the participants were asked to make it clear when they had finished the task by making some signal to the camera. They were also told that there was no time constraint.

All testing was carried out in a soundproofed clinic room in the Department of Speech and Language Sciences at Queen Margaret College, Edinburgh. This was a familiar environment to all the aphasic subjects. Simultaneous recordings were made on video and audio tape. While the audio tapes were made and used for the process of initial transcriptions, the video tapes were used for all analysis.

3.2 Subjects


This study investigates the interactive behaviour of 12 individuals: 4 non-neurologically impaired postgraduate students (2 male, 2 female); 4 male individuals with aphasia and 4 male non-neurologically impaired controls (matched with the aphasic participants for age, handedness and educational background).

The IG in each of the AD dyads is an aphasic participant. The IF in all the dialogues (both ADs and CDs) is always a non-aphasic participant.

3.3 Analytic method: coding task-oriented dialogue


In some ways, the methodology adopted for this paper is not dissimilar to Conversation Analysis (CA). This is because I firmly believe that such an approach can only add to our understanding of talk in interaction. In practical terms this has meant qualitatively rooting around in collections of data and making notes or ‘comments’ (by tagging the extracts with a ‘*C’ code and accompanying commentary) on any points of particular interest (see ten Have, 1999: 108). These codes function to identify particular sequences for further analysis. For this paper the comments that I will be discussing are those which relate to what I have called ‘sensitivity’.[2] (See §6.0.)

My approach is not a ‘pure’ CA one, however— in fact it draws greatly on the game coding tradition developed by Kowtko, Isard & Doherty (1992).[3] Nevertheless, our tendency to use slightly different terminology from Conversation Analysts does not mean that we are not dealing with the same phenomena. For example, the phenomena that Conversation Analysts study under the rubric of ‘repair’ are covered by us (qua Game Coding analysts) mainly by the various types of initiating activity that we call ‘check games’ (though other games can be involved in repair).

4.0 Politeness in task-oriented dialogue

Having briefly explained issues relating to aphasia and task-oriented dialogue, we are almost in a position to turn to the role of face in task-oriented dialogues between aphasic and non-aphasic individuals. First though, because I refer to them in subsequent discussion, I must set out the two main hypotheses concerning the differences between AD and CD interactions that were investigated by Merrison (1998).

4.1 Hypothesis 1

Hypothesis 1

In the Aphasic Dialogues the non-aphasic dialogue partner will do more of the communicative work.

It should be pointed out that this is not an original hypothesis (see, for example, Linebaugh, Kryzer, Oden & Myers, 1982). That said, I am not interested so much in the fact that non-aphasic dialogue partners take on more of the ‘communicative burden’ as much as in how they do so, and in particular, at least for current purposes, how that relates to the management of face.

4.2 Hypothesis 2


Hypothesis 2 (doing being ordinary [4])

In the Aphasic Dialogues the non-aphasic dialogue partner will try to avoid highlighting any linguistic non-competence on the part of their aphasic partner.

This hypothesis warrants a little more detailed explication than Hypothesis 1.

Talk is vulnerable to a variety of troubles relating to speaking, hearing and understanding – from forgetting a word to not hearing something because of a noisy environment or inattention. Because of the abundance of potential trouble sources, we might expect there to be some organised method(s) of dealing with and mitigating those troubles. Such an expectation is indeed correct.

In any next turn, the understanding of the prior turn can be demonstrated. If, however, the prior turn has not been understood sufficiently for current purposes – if something in the talk provides some obstacle to subsequent talk – then work will need to be done to deal with that trouble. This work is known as repair. It is highly organised and has been the focus of much research in Conversation Analysis (see, for example, Schegloff, Jefferson, & Sacks, 1977). For current purposes, however, all we need to note is that in interactions between a Speaker (S) and Hearer (H), there are two possible sources of initiating repair – S (self) or H (other) thereby yielding self-initiated repair and other-initiated repair respectively.

Repair sequences are just one way that talk can become extended. Provision of unnecessary detailed information is another.

In a way, then, Hypothesis 2 relates to such extended sequences. In the Control Dialogues the risk of extended (detailed) sequences of talk generating serious misunderstandings is relatively slight and therefore measures need not be taken to avoid them. With a linguistically impaired partner, however, extended sequences pose a real threat to the smooth running of both transactional and social aspects of the interaction and, wherever possible, they should probably be avoided.

To briefly explain what I mean by the potential disruptive nature of extended sequences, let me take the case of other-initiated repair as an example. Should it occur, other-initiated repair will not only displace the next sequentially implicated turn, it will also make the repair interactional business in its own right, and this will consequently highlight the breakdown (at least to some degree) of the interaction in progress and with that, the non-competence of the producer of the trouble source.

In order to minimise any highlighting of aphasic non-competence in the Aphasic Dialogues, it was therefore hypothesised that the non-impaired dialogue partner might tend to avoid the generation of unnecessary talk, simplifying the interaction by, among other things, avoiding introducing non-essential details and also avoiding other-initiated repair sequences.[5]

4.3 Face


The predominant paradigm in which politeness is investigated tends to be that developed by Brown & Levinson (1987). As their theory is based on Goffman’s (1967b) notion of ‘face’, it is appropriate to explain this term before I briefly sketch Brown & Levinson’s use of it.

Goffman defines face as 'the positive social value a person effectively claims for himself by the line others assume he has taken during a particular contact', where a line is 'a pattern of verbal and nonverbal acts by which [a participant] expresses his view of the situation and through this his evaluation of the participants, especially himself' (1967b: 5). Importantly for us, a person is said to be in wrong face 'when information is brought forth in some way about his social worth which cannot be integrated, even with effort, into the line that is being sustained for him' (1967b: 8). The following quote from Goffman (ibid., emphasis added) highlights the area of particular interest:

When a person senses that he is in face, he typically responds with feelings of confidence and assurance. Firm in the line he is taking, he feels that he can hold his head up and openly present himself to others. He feels some security and some relief – as he also can when the others feel he is in wrong face but successfully hide these feelings from him.

It is this wrong face scenario that I will mainly be discussing in the course of this paper. Specifically, I will be concerned with the work which non-aphasic interactants do to save the face of (or give face to) their aphasic interlocutors. Goffman explains that ‘Following Chinese usage, one can say that "to give face" is to arrange for another to take a better line than he might otherwise have been able to take, the other thereby gets face given him, this being one way in which he can gain face.’ (1967b: 9). ‘Giving face’ is also known as paying face. This work, collectively known as face-work, ‘serves to counteract "incidents" – that is events whose effective symbolic implications threaten face’ (Goffman, 1967b: 12f).

Brown & Levinson (1987: 65) call these incidents which ‘by their nature run contrary to the face wants of the addressee and/or the speaker’ (and hence threaten face) face-threatening acts (FTAs). In Goffman’s words, such face-work prevents a bad moment marring ‘an otherwise euphoric situation’ (1967c: 100) and according to him, it is not merely a matter of social nicety, but rather a sacred and ritualistic undertaking. As Goffman (1967b: 33) notes:

just as there is no occasion of talk in which improper impressions could not intentionally or unintentionally arise, so there is no occasion of talk so trivial as not to require each participant to show serious concern with the way in which he handles himself and the others present

There are essentially two types of face-work: avoidant and corrective. As Goffman points out, ‘The surest way for a person to prevent threats to his face is to avoid contacts in which these threats are likely to occur’ (1967b: 15). But avoidance is not always practical. In such situations Goffman suggests that ‘the participants are likely to give [the threat] accredited status as an incident – to ratify it as a threat that deserves direct official attention – and to proceed to try to correct for its effects’ (1967b: 19).

It is this corrective type of face-work which will be discussed here. It has two manifestations: (i) active mitigation, whereby the degree of face-threat is compensated for by appropriate strategies and (ii) tactful blindness to potentially embarrassing incidents ‘where the person acts as if an event that contains a threatening expression has not occurred at all’ (Goffman, 1967b: 17f).

So to engage in social interaction is to run the risk of losing face and consequently interactants have to jointly cooperate to maintain face of all participants. In short, engaging in social interaction demands proper involvement (Goffman, 1967d: 116).

But involvement in interaction is a delicate process which is also open to misinvolvement and this can lead to what Goffman calls ‘alienation from interaction’ (1967d). Of the various forms of alienation that Goffman discusses, the one which will concern us most is that of interaction-consciousness whereby:

A participant in talk may become consciously concerned to an improper degree with the way in which the interaction, qua interaction, is proceeding, instead of becoming spontaneously involved in the official topic of conversation.
(1967d: 119, original emphasis)

Such interaction-consciousness is more likely in dialogues involving a linguistically-impaired (or otherwise less-competent) interlocutor. It is also potentially more disruptive. Indeed, so important is the smooth running of spoken interaction that Goffman considers uneasiness as a contagious ‘disease’ (1967d: 126). The following quotation summarises this point:

Conversational encounters in which participants feel obliged to maintain spontaneous involvement and yet cannot manage to do so are ones in which they feel uneasy, and ones in which they may generate uneasiness in others.
(1967d: 129)

What concerns us is how such dis-ease may be mitigated through investing effort into doing face-work – or, as Brown & Levinson would say, into doing politeness.

The notion of face has since been developed by Brown & Levinson who state (1987: 61):

We make the following assumptions: that all competent adult members of a society [6] have (and know each other to have)

(i) ‘face’, the public self-image that every member wants to claim for himself, consisting in two related aspects.

(a) negative face: the basic claim to territories, personal preserves, rights to non-distraction – i.e. to freedom of action and freedom from imposition

(b) positive face: the positive consistent self-image or ‘personality’ (crucially including the desire that this self-image be accepted and approved of) claimed by interactants

(ii) certain rational capacities, in particular consistent modes of reasoning from ends to the means that will achieve those ends.

It is Brown & Levinson’s notions of negative and positive face (and associated FTAs) that will be most useful in my own exposition of doing being ordinary (Hypothesis 2 – see §4.2).

Compared to data from the Control Dialogues, results indicate that when engaged in talk with aphasic dialogue partners, non-impaired speakers appear to manage potentially face-threatening situations by attempting to avoid highlighting any non-competence on the part of their interlocutor, and they do so both by active and inactive strategies of face restoration.

Elsewhere (Merrison, 1998) I have termed this restorative face-work ‘sensitivity’ and it constitutes one of a pair of issues that were prevalent in the Aphasic Dialogues, namely sensitivity and mismatch. In many respects both of these could have been considered as sub-types of a higher, superordinate notion, viz. Merrison’s Maxim, and although the thrust of this paper relates only to sensitivity, in order to adequately explain this notion I first need to explain the maxim itself.

5.0 Merrison’s Maxim

Merrison’s Maxim is concerned with paricipants’ monitoring of and participation in talk-in-interaction and I use it to describe regularities in the current data set. Although it might be likened in some respects to the Principle of Parsimony (Shadbolt, 1984) and the Principle of Least Collaborative Effort (Clark & Wilkes-Gibbs, 1986, Clark & Schaefer, 1989), these principles deal only with transactional aspects of discourse: Merrison’s Maxim also incorporates the social.

Merrison’s Maxim operates as a five-clause cyclical decision loop:

Clause 1

Proceed as if everything is okay until you get evidence to the contrary.

Clause 2

With evidence that things are not going smoothly, determine whether the evidence is indicative of a potential threat to:

(i) Social (Face) wants

(ii) Transactional wants

If there is an indication of a potential threat to the wants in Clause 2, proceed to Clause 3. If overall evidence suggests that for current purposes no action is required, take no restorative action and re-enter the loop at Clause 1.

Clause 3

If there is an indication of a potential threat to the wants in Clause 2, determine whether either set of wants is in conflict with the other. If there is conflict proceed to Clause 4. If not, proceed to Clause 5.

Clause 4

Determine which wants should be met to best serve current purposes. Proceed to Clause 5.

Clause 5

Evaluate the cost-effectiveness of restorative action with respect to:

(i) Benefits

(ii) Cost/Effort

(iii) Likely Effectiveness

If Benefits outweigh Cost/Effort, and Likely Effectiveness is sufficiently high, invoke restorative action and then re-enter the loop at Clause 1.

There are two major aspects to the invocation of Merrison’s Maxim. There can be what I call active application of the Maxim at Clause 5 where some trouble is detected and restorative action is taken. Alternatively there is passive (or non-) application of the Maxim, in other words, not going beyond Clause 1 because there is no perceived problem with the interaction.

I can now say why sensitivity and mismatch are considered as two distinct categories, namely because I wish to make a distinction between active and passive behaviour. While ‘mismatch’ involves the passive (or non-) application of Merrison’s Maxim, in other words, not going beyond Clause 1, ‘sensitivity’ involves the active application of Merrison’s Maxim at Clause 5. In the rest of this paper I deal only with this active ‘sensitivity’.

6.0 Sensitivity


Examples of what I have called sensitivity actually cover a range of sensitive interactional behaviour; what the extracts of talk have in common, however, is that they involve (i) the recognition of a potential threat to either transactional or social wants and (ii) some form of restorative action (as per Clause 5 of Merrison’s Maxim). In other words ‘sensitivity’ comments are markers used to tag sequences which involve the mitigation of threats – either threats to the success of the transaction or threats to face. In recognition of these two distinct types of threat I consequently introduce what I believe is a new term to the literature: analogous to Brown & Levinson’s face-threatening act (FTA) I will adopt the term transaction-threatening act (or TTA).

In essence, restorative action can be anything that the initiator of that action believes will act as a remedy to the troubled situation For example, if the exchange of information in transactional discourse is hampered by some concept that your dialogue partner does not share with you, then one form of restorative action might be to invest the effort in explaining that concept to them it. Alternatively it may be to abandon the possibility of talking to that person about that concept.

It all depends on estimates of Benefits, Costs, and Likely Effectiveness and, depending on those estimates, in certain circumstances it may well be deemed more cost-effective to do nothing.

Where restorative action is required, and where it is undertaken by the non-impaired partner, this essentially provides us with additional evidence for Hypothesis 1, namely that the non-aphasic dialogue partner will do more of the communicative work.

Where restorative ‘action’ involves action avoidance – i.e. inactivity – this non-action provides evidence for Hypothesis 2, namely that in order to minimise highlighting (perceived) aphasic non-competence the non-impaired dialogue partner will tend to avoid the generation of unnecessary talk.

Having made these points, I will presently (starting in §6.2) provide extracts of talk that were coded as ‘sensitive’ – a comment that appears exclusively in the ADs.

Because talk can be sensitive to either TTAs or FTAs, and because the restoration to those threats can be made either through action or inaction, I accordingly discuss ‘sensitivity’ comments that cover active transactional restoration (§6.1), inactive transactional restoration (§6.2), active restoration of face (§6.3) and restoration of face by inaction (§6.4).

6.1 Active transactional restoration


If it is discovered that the transaction is not running smoothly, then any active restoration of that transaction will take the form of what is known in the CA literature as conversational repair (see Schegloff, Jefferson and Sacks, 1977). Examples of other-initiated repair are given in Extract 1 in turns GW021 and GW022:

Extract 1: HL (aphasic) & GW(non-aphasic)[7]



Is the bridge on the top right hand corner of your page?



Er (3.8)



The very very top corner (.) do you have a bridge there?



Mhm {nods}



But it’s {headshake} not that bridge



(1.0) Er {nod} yes=



=It {nods} is that bridge.=






˚(·h)˚ OK.


Merrison (1998) reported that in the ADs, check games (repair sequences focussing on checking self-understanding) such as those in Extract 1 are the most prevalent of all the games initiated by the non-impaired IF: of all the games that the IFs initiate, almost 39% are check games (in the CDs, this proportion is just over 28%). Thus, in the ADs there is evidently a great deal of repair activity, which might also be called ‘reparation’ or even transactional restoration. However none of this type of active transactional restoration was coded with *C comments because these comments were used to flag extraordinary instances of talk within the data. Clearly a behaviour (active transactional restoration) that is associated with the most prevalent game type (checks) can in no way be considered ‘sufficiently extraordinary to warrant special mention’.

In short, there is evidence of a great deal of active transactional restoration in the aphasic corpus.

6.2 Inactive transactional restoration


In Extract 2 below, we find two instances of the phenomena I have referred to as ‘sensitivity’. The first is in turn DN014 where it appears that on learning that BA (the aphasic IG) does not have a church in the corner of his map, DN (the non-aphasic IF) decides that the potential transactional benefits associated with explaining the location of the church to BA (namely getting an exact route drawn) are far outweighed by the TTAs (and possibly also FTAs) involved with the generation of additional talk. DN therefore abandons all talk of the church.

In turn DN017 we find that in response to BA’s instruction (BA011) of So (.) {deixis gesture} down first {measures} oh () maybe four inches?, not only does DN not pursue a description of the church’s location, but he also avoids any unnecessary transactional talk by not seeking clarification as to the exact distance that he was meant to go down. We can see this from DN’s resulting route drawing (see Map #2A in the Appendix).

Although this extract is from the beginning of the task (and that DN therefore cannot be blamed for not appreciating the importance of landmark navigation), it should not be assumed that this consequentially impacted on the design of the talk. First, DN’s interaction with BA was actually his fourth aphasic Map Task dialogue. More important, however, is the fact that although BA is communicatively very able, he actually presents as the most linguistically impaired of the four aphasic subjects. Consequently it is plausible to conclude that DN’s inaction is indeed a case of what might be called intelligent communication: DN is making judgements about his partner’s communicative abilities, and, based on the principle of least collabortive effort (Clark & Wilkes-Gibbs, 1986, Clark & Schaefer, 1989), DN is electing to use their combined linguistic resources most effectively by not pursuing a detailed exchange about the location of the church.

Extract 2: BA & DN: IF inaction



{measures} So (.) {deixis gesture} down first {measures} oh () maybe four inches?






=˚(LS)˚ {deixis gesture} Er down.



Is that towards the church? >Have you got a [church ˚in your corner.˚<]



                                                                     [{head shake} No:]=



= {head shake} No okay we- don’t worry.

*C sensitive abandonment



{deixis gesture: down = (.)}=



=But towards the bottom of the page=



={nods} Yeah [˚yeah.˚]



[O]kay >so we’re just going< [straight south.]



                                               [{measures}] ˚Yeah.˚




*C DN takes 3 inches and goes above church but never checks this distance further


Though it may initially seem counterintuitive, what we can see from Extract 2 is an example where apparent transactional inaction on the part of the non-aphasic interactant has the concomitant effect of paying positive (and, to a lesser degree, also negative) face to his aphasic partner. In other words, face-paying is something that can actually be achieved by not doing anything. And although it may be thought that this face-work is merely a fortuitous ‘by-product’ of avoiding TTAs it could instead be argued that it is possibly the avoidance of FTAs that actually drives the avoidance of TTAs. In other words, perhaps what we are seeing in this data are social wants taking precedence of transactional wants despite the very clear task-oriented nature of the dialogue.

Sensitive transactional inaction is not confined to the non-impaired IF, however. We can see evidence of this in Extract 3 where, in turn BA027, the aphasic Information Giver shows that he, too, is an intelligent user of available resources who, by not pursuing matters further, is also able to avoid generating unnecessary extended sequences of talk – a strategy which may also have the added benefit of saving (positive) face (both his own and potentially also his partner’s).

Extract 3: BA & DN: IG inaction



˚(LS)˚ And er <letter box:>



(.) A letter box






No I haven’t got [that.]



                           [˚Oh˚] {iconic gesture: forget it} doesn’t matter [˚˚then.˚˚]

*C this is really very sensitive



6.3 Active restoration of face

We now turn to examples of comments relating to the restoration of social wants, or face (see §4.3). Like active transactional restoration (repair), comments pertaining to the active restoration of face involve talk in which the issue of non-competence is actually addressed, but here, unlike the transactional case, the issue of non-competence is mitigated by active face-saving devices (see Brown & Levinson, 1987: 101ff).

The next extract concerns what conversation analysts would call other-initiated other-repair – the most dispreferred of all repair types (see Schegloff et al., 1977).

In Extract 4, the *C comment line attached to DN049 states that correction comes with opt out clause and is quieter. In other words, it indicates the presence of two types of negative face-saving devices: mitigation in the form of reduced volume (as indicated by the presence of the degree symbols ‘˚’) which serves Brown & Levinson’s negative politeness strategy of non-coercion (give Hearer the option not to act), as well as the presence of ‘Or…?’ which serves to give the IG the legitimate face-saving option of not accepting the correction (Brown & Levinson’s negative politeness strategy of non-presumption). Together these two negative politeness devices serve Brown & Levinson’s positive politeness strategy of disagreement avoidance. (In passing we might also note that it seems that two negatives do indeed make a positive!)

Extract 4: BA & DN



() And then three inches {deixis gesture} west again.



(1.6) Back towards the snail?



(1.2) >No!<=



=˚East then. Or…?˚

*C correction is false: ie, giver is right: here begins Merrison’s Maxim

*C correction comes with opt out clause and is quieter



(2.6) ˚No:˚ er (.) {iconic gesture} there there and there,






(.) {iconic gesture} there (.) and there and there all way

{deixis gesture: left all the way = (2.2)} ˚˚˚two three˚˚˚ {deixis gesture: left all the way}


It is just as well that DN offered this opt-out clause, for BA quite rightly rejects the correction (the reason for the miscommunication being that the IF had been circumnavigating the snail from the wrong direction – see Map #2A in the Appendix).

The next extract also shows active face-saving mitigation on the part of the non-impaired dialogue partner. At the point where we join the interaction, the pair are just negotiating the hills which, on MD’s map, are right next to the sea (see Map #3A in the Appendix). Unfortunately, because of MD’s incorrect distance terminology viz. half an inch in turns MD101 and MD103 and three quarter of a inch in MD104, his non-impaired partner ends up drawing the route quite some distance from the sea.

Extract 5: MD & MB: trouble in them there hills



[So - ] so you've got hills just an inch or two below the axe









Er er a:nd () er go due er [~ ~] er er east



(LS)=East. >So [I go<]


MD101 ctd

                          [And] er so t’s about (1.4) half a: (.) inch



{upward nod} Mm



And then go to the er north






(1.4) Er for half an inch






(2.1) And go to the er east for (0.9) (LS) about half - three quarter of a [of a] inch (·HH)






Right so - (0.9) so come down from the axe to these hills and then we have a little sort of



Er s[ort ˚of˚]


MB134 ctd

       [almost like] a {iconic} square=


MD105 ctd

=Yeah ({draws route shape}= 0.9) er


MB134 ctd

corner. Yeah


MD105 ctd




So sv [=sort of] so I go for the hills and then (.) [east] and then it was north (0.8)








MB135 ctd

for a bit and then east for a bit



And then you carry on





MD108 ctd

=to er south







For: about (1.8) three quarters of an inch






Er the sea directly (1.2) er about half an inch



(2.5) Sea. (2.0) Um so we’re right - at the moment you’re right next to the coast˚line.˚ ˚˚XX˚˚



Well the little bit






er {iconic} further (*hh) in but not much



Oh I think we've gone wrong somewhere 'cos I'm [(.) at the moment] I'm sort





MD141 ctd

[of about] three inches [from] from the sea (ha).



[(*hhhh)]                     [(hhh)]

  MD141 ctd

(·hh) Um (1.0) (LS) (2.4) yeah (1.2) (·hh) I was just wondering actually (1.1)

so - ’cos we went - (1.2) ˚if I˚ {iconic} back-track back to the canoe (1.3) Um



Canoe [er]



[(·hh)] ’Cos I I [think that’s] mi- might be where we went wrong=


MD113 ctd



MB142 ctd

=from the canoe to the axe was l- several inches

*C indirectly saying BA is wrong yet giving him the chance to self-correct!






Like sort of [<four] five inches>



                  [er     ]



Yeah [(.)er]


MB143 ctd

        [to the axe]






The following analytic observations can be made about this extract:

• In turn MB139 MB is displaying some trouble with the comment about preference organisation (see Levinson, 1983: 336f) but the subtlety of the self-initiated self-repair which switches from an inclusive Um so we’re to an exclusive at the moment you’re could also have been flagged. In effect this is an attempt to get MD qua the person responsible for the cause of the trouble to initiate self-repair of that trouble. MB giving MD the opportunity for self-repair is really very face-sensitive. (See Schegloff et al., 1977 for commentary on the preference for self-repair.)

• In turns MB140 and MB141 the discourse marker ‘oh’ (see Schiffrin, 1987; Heritage, 1984; Local 1996), which, together with the explicit (and now inclusive) we’ve gone wrong somewhere (sensitively hedged with I think) also signals trouble and with the admission that he is about three inches from the sea, MB again sensitively offers MD the opportunity for self-repair.

• In the continuation (ctd) of turn MB141 MB provides three opportunities for MD to self-repair the problem:

MB141 ctd

(·hh) Um (1.0) (LS) (2.4) ye:ah. (1.2) (·hh) I was just wondering actually (1.1)

so - ’cos we went - (1.2) ˚if I˚ {iconic} back-track back to the canoe (1.3) Um

The first opportunity for MD to carry out self-repair comes in the (1.0) second pause after (·hh) Um. The second comes in the (2.4) second pause after the lipsmack (LS) and the third opportunity in the (1.2) second pause after the content-free ye:ah.[8]

• And finally in the continuation of turn MB142, we find the *C comment of sensitivity that instigated the discussion of this extract: with with from the canoe to the axe was l- several inches and the Like sort of <four five inches> that follows it in MB143, MB offers yet further opportunities for MD to initiate self-repair.

So every one of MB’s turns from MB139 onwards displays evidence of trouble and almost all of them afford MD the opportunity for self-repair. Yet in none of them does MD take up that offer. Indeed this sensitivity from MB continues over many more turns, only for MD to eventually reaffirm his instructions of half an inch east, half an inch north and then three quarters of an inch east again. At this point MB re-routes his search for the true distances and starts to phrase his concerns in terms of another (for him parochially relevant) landmark, viz. the lighthouse (which, just to make matters even worse, is an IF-specific landmark – i.e. it is not on MD’s map). As we can tell from the final route shape, the problem in them there hills is never actually successfully resolved – but not for lack of trying on MB’s part.

These attempts at sensitive, face-saving behaviour on behalf of his aphasic partner show a great deal of effort from the non-impaired IF – evidence for Hypothesis 1 to be sure. And what is more, all of this discussion has been driven by just one single originally innocuous looking comment. I think that this one example alone is sufficient to prove the analytical worth of the humble *C line and the importance of close observation à la Sacks (1984) and Kelly & Local (1989)!

6.4 Restoration of face by inaction


In the previous section we saw a lot of active face-work from the non-impaired dialogue partner. This fitted in very well with the expectations set out in Hypothesis 1, namely that in the aphasic dialogues the non-aphasic dialogue partner will do more of the communicative work. In this section we will see examples of how inaction from the non-impaired dialogue partner can be utilised to mitigate potential face-threats to the aphasic individual; this inaction is of the special type that has been discussed by Sacks (1975: 75):

In circumstances in which alternative answers to a question are known, and the alternative answers have alternative consequences for that conversation or for other events, then one way in which people are known to attempt to control those alternative consequences is to select answers by reference to their intended selection of a consequence.

In other words, our linguistic choices are driven by our desired outcomes.

In this section we will consider cases where, in an aphasic dialogue, the non-impaired partner is asked a question, the truthful answer to which would violate Hypothesis 2 (which states that the non-aphasic dialogue partner will try to avoid highlighting any linguistic non-competence on the part of their aphasic partner).

We will therefore be concerned here with circumstances where the non-impaired interactant will choose to respond to their partner’s query in such a way that might save their aphasic partner’s face; in other words, cases where the non-impaired partner’s sensitive ‘inaction’ will be to lie. [9]

The first extract to be discussed is taken from the dialogue between GM & MB, and MB’s little white lie is to be found in MB128 as a safer alternative response to GM’s preceding alignment query if you know what I mean.

Extract 6: GM & MB: if you know what I mean



And there’s a win- win[::]till.                                   [˚Wind-˚]



                                  [(LS) {upward nod}]           [Wind]mill.









Well you go round {trace} (.) the back of that



Yeah () [˚that’s˚] (.) [˚˚to your˚˚]



             [˚˚X˚˚] (.)    [So you go] pa- {head turn = deixis gesture} you go past it
{deixis gesture}



˚˚(LS)˚˚ To the: (.) {deixis} le[ft]



                                            [An’] and you’re going () {deixis gesture} right

you’re going round - round it {trace = (.)} round it




Round. So I - (0.5) after the goat (0.6) I go down and then onto the {invisible deixis gesture} right hand side of the {point} windmill (1.2) where it (0.7) um (0.8) (LS)=



=You walk past the door



(0.9) Past the door. Ah.



You walk past the door (.) ˚˚if˚˚ you know what I mean




*C lie! but it saves face and serves to simplify structure!!![10]


MB128 ctd

(.) So to get there >˚˚X-˚˚< I keep (.)



˚(LS)˚ You {deixis gesture} keep going



From the goat I carry {deixis gesture} on down so that I end up on the

left hand side [Whe- where the ‘W’ is]



                     [Yeah. Yeah. Yeah.] where the door is ˚aha˚=



=And then down (.) ˚(LS)˚=then underneath, {looks up}






below the door. () Right.



˚˚Aha. Yeah.˚˚

The following evidence enables us to infer that MB’s response in MB128 is indeed (sensitive though it may be) a lie:

(1) Immediately prior to GM’s query at GM109, MB displays non-understanding through the micro-pause as well as through his utterance of Ah in line MB127.

(2) Not only does MB display his non-understanding, GM orients to this display in his next turn (GM109) by not only emphatically re-clarifying his contribution, but also by providing a hedge on the Gricean maxim of manner (Grice, 1975: 46). This hedge (which GM uses to indicate that he is not being as clear as MB might have hoped) takes the form of the very utterance that prompts the lie, namely if you know what I mean.

(3) Immediately after his response to GM’s hedge, MB initiates two check games (requests for clarification) at MB129 and MB130 to ensure that he has indeed truly understood what GM meant. If he had truly understood what GM had meant, then we would not have expected these checks to be initiated.

Before turning to the final example of this section, I would like to point out that not only is MB’s little white lie sensitive, it is actually doubly sensitive because GM’s hedge is itself orienting to his own non-competence (i.e. it indicates that he knows that he is not being optimally clear). GM has not made light of the trouble that is his (minor) non-competence, unlike the way he did earlier in this dialogue as we see in Extract 7:

Extract 7: GM & MB: It affects me you know



You go[: (.)] you go (.)





GM048 ctd

{trace: south} north then (.) >˚you go˚< {deixis gest: south} north ˚again˚



˚(LS)˚ North? Or (.) do you mean ˚˚s-˚˚ {deixis} down



{deixis gest: south} ˚It’s north˚=



=oh south



˚South˚ - [south. {iconic} Sorry X X]



              [South. Yeah Th(h)at’s] (haha) (·hhhh)=




=It affects me [˚you know˚]





At GM051 we see that the aphasic interactant himself makes light of his own non-competence. We can tell that the participants are in fact treating this meta-linguistic comment as face saving mitigation and not as a troubles-telling (Jefferson, 1984) because of the laughter tokens in MB063 and MB064.

Consequently, at MB128 in Extract 6, if MB had baldly stated that he didn’t know what GM meant, this would have been tantamount to showing a lack of empathy, or in terms that I have been using throughout this section, a lack of sensitivity.

And so we come to the final example of inactive sensitivity which is given in Extract 8. It comes from the same dialogue between MD & MB whence came the example of active face-restoration that involved the "trouble in them there hills".

Extract 8: MD & MB: Oh I see



And then ju[st] directly: south



                  [(LS) >And then<]



Directly (.) {eyebrow flash} south? (.) Not south-east?



No south



Oh south



And then er () south er () er east



>(·hhh)< Oh - so after going south a bit we go south-east. [Mm ˚(LS)˚]



                                                                                        [And] there’s a anchor






er it’s er about () half an inch up it s- stops [it finish]



                                                                 [˚Half an inch˚] Oh I see.


*C liar

MB190 ctd

Is is the anchor- Do you have a fla:g anywhere on your

(1.5) pa[ge - ˚on your map˚]



            [F- f- a fla]g. [Er] as yes







In MB190 MB’s Oh I see is a sensitive lie – as a cumulative effect of the ‘trouble in them there hills’ (see the discussion in §5.1.3), MB doesn’t actually ‘see’ at all, yet he knows that to admit his confusion would only serve to threaten his partner’s positive face.

Consequently we find a face-sensitive lie which we know is a lie because of the ensuing talk about the location of the anchor and the finish point – that and the fact that it takes sixty-four further turns and much elaborate negotiation for MB to finally get the finish point ‘right’, and even then, in MB’s own opinion, only ‘roughly’ so (see Map #3A in Appendix).

In his 1975 paper Sacks examined the validity of the statement Everyone Has to Lie. He concluded with the following words:

the statement is true if the organization of conversation is such that any next conversation can formally produce the problem of having to deal with some such sequentially implicative question as How are you? where the question is asked by one with whom the respondent … is placed in a situation that he sees involves either getting into a sequence in this conversation that he should not get into or lying so as to avoid that sequence. The organization of conversation being such, the statement is true.
(1975: 78, original emphasis)

It appears that my data apparently confirm this organisation of conversation and, consequently, the truth of the statement can be reaffirmed.

6.5 Conflict between transactional and face-wants


In GM & MB’s if you know what I mean Extract 6 and MD & MB’s Oh I see Extract 8, we have seen that the sensitive lies from the non-impaired partner are followed by active restoration of transactional breakdown (with the Oh I see example demonstrating this point particularly well). What we have in these cases is evidence for Clause 3 of Merrison’s Maxim: namely that there is conflict between transactional wants on the one hand and social (face) wants on the other, with both really needing to be addressed.

Conflict forces the non-impaired partner to invoke Clause 4 of Merrison’s Maxim: Determine which set of wants should be met to best serve current purposes, and my theory allows for three possible outcomes of such a determination:

Outcome 1: Transactional wants get addressed

Outcome 2: Face wants get addressed

Outcome 3: Both transactional and face-wants get addressed

I have suggested that the preponderance of checking (repair) activity in the aphasic dialogues counts as evidence of Outcome 1; we have seen evidence of Outcome 2 in the It affects me you know example; and evidence of Outcome 3 is to be found in the if you know what I mean and Oh I see extracts.

There is, however, yet another possible outcome (though one not driven by Clause 4) that has not yet been discussed: it is entirely possible that neither set of wants actually gets addressed. As yet we have not seen any evidence for this possibility and one might be forgiven for thinking that this is because there is no evidence. Not so. The reason that I have not yet discussed evidence of this state of affairs (which, if I am to be true to my theory, should really be called Outcome Zero) is because that will have to be the subject of another paper.

7.0 Conclusion


What has been shown in the course of this paper, then, is that if we carefully consider individual examples of actually occurring talk, we may find ourselves rewarded with details of social interaction that would otherwise have been (at best) missed, or even (at worst) categorically denied.

Our local looking at examples of Map Task talk between aphasic and non-aphasic individuals has done just that. Contra Kasper (1990), these data have shown that even when engaged in task-oriented dialogue (which is clearly ‘transactional discourse’), speakers can be seen to invest a great deal of effort into doing face-work – in other words, into doing politeness, and that when there is apparent conflict between the need to address both transactional wants and face wants, it can even be the case that it is face-work which takes priority.



(1) For example, in Map #3 in the Appendix the three landmarks that appear only on the IG’s map are (in order from Start to Finish): pine, eskimo and hills. The three landmarks that appear only on the IF’s map are: noose, helicopter and lighthouse. (back)

(2) We must make it very clear that *C comment lines have been used to flag extraordinary instances of talk within the data; just because some of the comment types are apparently corpus-specific this does not mean that we wish to imply that any such behaviour is not to be found in the other corpus – all we wish to imply is that if such behaviour does exist then it was not singled out as being sufficiently extraordinary to warrant a *C comment. What we have, therefore, are comments which indicate ways in which our aphasic data is unusually different from the control data. (back)

(3) Game Coding is based on goal-directed exchanges called Conversational Games and essentially involves coding (or tagging) every dialogue utterance in the transcripts for collaborative interaction in terms of initiation-response-feedback patterns (rather than linguistic content). These codes cover such functions as: instruct, check, query, explain and align. (back)

For fuller descriptions of the various game types see Kowtko et al. (1992) or Carletta, Isard, Isard, Kowtko, Doherty-Sneddon and Anderson (1996). For discussion relating to the specificities of aphasic/non-aphasic dialogues, see Merrison (1998), Merrison (in preparation, a).

(4) Although this term originates in Sacks’ Lecture of Spring 1970 (Sacks, 1995: Volume II: 215-221), it is Wilkinson (1995a: Chapter 7, 1995b) who first applied it to aphasic discourse. (back)

(5) Cf. Gravel & LaPointe (1983), Linebaugh, Pryor & Margulies (1983) and Nicholas & Brookshire (1983). (back)

(6) Here, Brown & Levinson add a footnote: "Juvenile, mad, incapacitated persons partially excepted." Although I will not develop this line of discussion in the current work (for it requires some additional analysis), it appears that aphasic individuals are not to be included in this listing. Indeed, it is precisely because they are deemed to be competent adult members of society that effects their non-aphasic interactants ‘doing being ordinary’ on their account. The rest of this paper is an accumulation and discussion of evidence of this interactionally sensitive work. (back)

(7) In all the extract titles the first set of initials are those of the aphasic interactant and the second set are those of the non-aphasic. (back)

Transcription convetions are, in the main, traditional Jeffersonian Conversation Analysis conventions, with some notable exceptions. These are: (i) speakers’ turns are demarcated by initials and turn numbers usually above the line of talk (e.g. in Extract 2 below, BA011 denotes BA’s 11th turn at talk within the dialogue from which the extract is taken; (ii) (LS) represents an audible lip smack or click; (iii) non-verbal gestures are glosed within braces {}; (iv) untimed pauses are represented by (); (v) each unintelligble syllable is marked by a X; (vi) breathiness and laughter are marked within parantheses (hh), (·hh), (ha)

(8) This ye:ah. is a token that is not so very dissimilar to what in the CA literature on conversational closings would be called a passing turn (see Levinson, 1983: 317). By being content-free, it is a signal that the speaker has nothing more to add and therefore it can be seen as an attempt to pass the conversational floor back to the other partner. (back)

(9) Sacks (1975: 75) offers the following example of children’s sensitive (intelligent) behaviour:

If children are asked some question, one of whose alternative answers may occasion a rebuke and another not, then apparently they learn … to produce answers that are directed to avoiding the rebuke, which answer production can involve them in lying.

If we substitute non-impaired interlocutors for children and threat to their partner’s face for rebuke, then we have a formulation which captures nicely the sense of sensitive intelligent behaviour I intend here:

If non-impaired interlocutors are asked some question, one of whose alternative answers may occasion a threat to their partner’s face and another not, then apparently they learn … to produce answers that are directed to avoiding the threat to their partner’s face, [and this sort of] answer production can involve them in lying. (back)

(10) The lie serves to simplify discourse structure (as well as saving face) in that if MB had not responded to GM’s alignment query directly but rather had embarked immediately on the subsequent check sequences, then GM’s align game would have remained open and the checks would have consequently been one structural level deeper (and thus more complex) than they actually are. For further details on the embeddedness of discourse structure see Merrison (in preparation, b). (back)


Appendix - Maps from the Aphasic Dialogue Corpus

Dialogues between

GM & MB Map #1A (Caravans)
BA & DN Map #2A (Van)
MD & MB Map #3A (Mountain)



Anderson, A. H., Bader, M., Bard, E. G., Boyle, E., Doherty, G., Garrod, S., Isard, S., Kowtko, J., McAllister, J., Miller, J., Sotillo, C., Thompson, H. and Weinert, R. (1991) The HCRC Map Task Corpus. Language and Speech 34: 351-366.

Anderson, A. H. and Boyle, E. (1994) Forms of introduction in dialogues: their discourse contexts and communicative consequences, Language and Cognitive Processes 9 (1): 101-122.

Anderson, A. H., Clark, A. and Mullin, J. (1991b) Introducing information in dialogues: forms of introduction chosen by young speakers and the responses elicited from young listeners Journal of Child Language, 18: 663-687.

Anderson, A. H., Clark, A. and Mullin, J. (1992) Communication Skills in Children: Learning how to make language work in dialogue. Edinburgh: HCRC Publications, University of Edinburgh.

Anderson, A. H., Clark, A. and Mullin, J. (1994) Interactive skills in children Journal of Child Language, 21: 1-25.

Anderson, A. H., Robertson, A., Kilborn, K., Beeke, S. and Dean E. (1997) Dialogue despite difficulties: a study of communication between aphasic and unimpaired speakers, in Givón, T., ed., Conversation: Cognitive and communicative perspectives. Amsterdam: John Benjamins, 1-39.

Atkinson, J. M. and Heritage, J. (1984) Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press.

Bard, E. G., Sotillo, C. F., Anderson, A. H., Thompson, H. S. and Taylor, M. M. (1996) The DCIEM Map Task Corpus: Spontaneous dialogue under sleep deprivation and drug treatment Speech Communication, 20 (1-2): 71-84.

Beeke, S., Dean, E. C., Kilborn, K., Anderson, A. H., Robertson, A. and Miller, J. E. (1994) The relationship between syntactic processing and communicative performance in aphasia in Powell, T., ed., Pathologies of Speech and Language: Contributions of Clinical Phonetics and Linguistics. New Orleans: International Clinical Phonetics and Linguistics Association, 75-88.

Boyle, E., Anderson, A. H. and Newlands, A. (1994) The effects of visibility on dialogue and performance in a cooperative problem solving task Language and Speech, 37 (1), 1-20.

Brown, G., Anderson, A., Shillcock, R. and Yule, G. (1984) Teaching Talk: Strategies for production and assessment. Cambridge: Cambridge University Press.

Brown, P. and Levinson, S. C. (1987) Politeness: Some universals in language usage. Cambridge: Cambridge University Press.

Carletta, J., Isard, A., Isard, S., Kowtko, J., Doherty-Sneddon, G. and Anderson, A. H. (1996) HCRC Dialogue Coding Manual. Edinburgh: HCRC Publications, University of Edinburgh.

Clark, H. H. (1996) Using Language. Cambridge: Cambridge University Press.

Davies, B. (1997) An Empirical Examination of Cooperation, Effort and Risk in Task-Oriented Dialogues. Unpublished PhD thesis, University of Edinburgh.

Doherty-Sneddon, G. and Kent, G. (1996) Visual signals and the communication abilities of children Journal of Child Psychology and Psychiatry 37 (8): 949-959.

Garman, M. (1990) Psycholinguistics. Cambridge: Cambridge University Press.

Goffman, E. (1967a) Interaction Ritual: Essays on face-to-face behavior. New York: Pantheon Books.

Goffman, E. (1967b) On face-work, in Goffman (1967a: 5-45).

Goffman, E. (1967c) Embarrassment and social organization, in Goffman (1967a: 97-112).

Goffman, E. (1967d) Alienation from interaction, in Goffman (1967a: 113-136).

Gravel, J. S. and LaPointe, L. L. (1983) Length and redundancy in health care providers’ speech during interactions with aphasic and non aphasic individuals, in Brookshire (1983: 211-217).

Grice, H. P. (1975) Logic and conversation in Cole, P. and Morgan, J. L., eds. Syntax and Semantics 3: Speech Acts. New York: Academic Press, 41-58.

Heritage, J. (1984) A change-of-state token and aspects of its sequential placement, in Atkinson and Heritage, 299-345.

Jefferson, G. (1984) On the organization of laughter in talk about troubles, in Atkinson and Heritage, 346-369.

Kasper, G. (1990) Linguistic Politeness: current research issues Journal of Pragmatics 14: 193-218.

Kelly, J. and Local, J. K. (1989) Doing Phonology. Manchester: Manchester University Press.

Kowtko, J. C. (1997) The Function of Intonation in Task-Oriented Dialogue. Edinburgh: HCRC Publications, University of Edinburgh.

Kowtko, J. C., Isard, S. D. and Doherty, G. M. (1992) Conversational Games within Dialogue. Edinburgh: HCRC Publications, University of Edinburgh.

Lamb, S. J., Bibby, P. A., and Wood, D. J. (1997) Promoting the communication skills of children with moderate learning difficulties Child Language Teaching and Therapy 13 (3): 261-278.

Lamb, S. J., Bibby, P. A., Wood, D. J. and Wood, H. A. (1998) An intervention programme for children with moderate learning difficulties British Journal of Educational Psychology68: 493-503.

Leech, G. (1983) Principles of Pragmatics. London: Longman.

Lesser, R. (1989) Linguistic Investigations of Aphasia [2nd edition]. London: Cole and Whurr.

Lesser, R. and Milroy, L. (1993) Linguistics and Aphasia: Psycholinguistic and pragmatic aspects of intervention. London: Longman.

Linebaugh, C. W., Kryzer, K. M., Oden, S. E. and Myers, P. S. (1982) Reapportionment of communicative burden in aphasia’, in Brookshire (1982: 4-9).

Local, J. K. (1996) Conversational phonetics: some aspects of news receipts in everyday talk, in Couper-Kuhlen, E. and Selting, M., eds.Prosody in Conversation. Cambridge: Cambridge University Press, 177-230.

Merrison, A. J. (1992) How Aphasics get from A to B: Total Communicative Ability and the Map Task. Unpublished MSc dissertation, University of Edinburgh.

Merrison, A. J. (1998) Doing Aphasia: aphasic discourse from a non-aphasic perspective. Unpublished PhD thesis, University of Edinburgh.

Merrison, A. J. (in preparation, a) Game Coding: extending the system (the case for refined categorisation for refined contexts).

Merrison, A. J. (in preparation, b) The nature of embeddedness in the analysis of discourse structure.

Merrison, A. J., Anderson, A. H. and Doherty-Sneddon, G. (1994) An Investigation into the Communicative Abilities of Aphasic Subjects in Task Oriented Dialogue. Edinburgh: HCRC Publications, University of Edinburgh.

Nicholas, L. E. and Brookshire, R. H. (1983) Syntactic simplification and context: effects on sentence comprehension by aphasic adults. In Brookshire’, 166-172.

Sacks, H. (1975) Everyone has to lie, in Sanches, M. and Blount, B. G., Sociocultural Dimensions of Language Use. New York: Academic Press, 57-79.

Sacks, H. (1984) Notes on methodology, in Atkinson and Heritage, 21-27.

Schegloff, E. A., Jefferson, G. and Sacks, H. (1977) The preference for self-correction in the organization of repair in conversation, ,Language 53: 361-382.

Schiffrin, D. (1987) Discourse Markers. Cambridge: Cambridge University Press.

Sotillo, C. (1997) Phonological Reduction and Intelligibility in Task-Oriented Dialogue. Unpublished PhD thesis, University of Edinburgh.

Wilkinson, R. (1995a) The Application of Conversation Analysis to the Assessment of Aphasic Talk-in-Interaction. Unpublished PhD thesis, University of Central England, Birmingham.

Wilkinson, R. (1995b) Doing "being ordinary": aphasia as a problem of interaction, in Kersner, M. and Peppe, S., eds. UCL Department of Human Communication Science Work in Progress, 5, 134-150.