‘Gossiping’ as a social action in family therapy: The pseudo-absence and pseudo-presence of children

Family therapists face a number of challenges in their work. When children are present in family therapy they can and do make fleeting contributions. We draw upon naturally occurring family therapy sessions to explore the ‘pseudo-presence’ and ‘pseudo-absence’ of children and the institutional ‘gossiping’ quality these interactions have. Our findings illustrate that a core characteristic of gossiping is its functional role in building alignments’ which in this institutional context is utilized as a way of managing accountability. Our findings have a number of implications for clinical professionals and highlight the value of discourse and conversation analysis techniques for exploring therapeutic interactions.


Introduction
The social action of gossiping has received attention in social psychology, sociology, anthropology and organizational studies. From this research a number of core features are consistently reported. For conversation to be considered gossip it should be triadic (Michelson et al., 2010), evaluative (DiFonzo and Bordia, 2007), remedial (Guendouzi, 2001) and is typically (but not exclusively) negative (Noon and Delbridge, 1993). It is noted that the talked-about other should be non-present (Foster, 2004) and that there are social sanctions for engaging in gossiping (McDonald et al., 2007).
Gossiping thus creates demarcation between insiders and outsiders and is a useful communication strategy for building (Duncan et al., 2006) and maintaining social relationships (Fiske, 2004). Problematically, however, although gossip can strengthen relationships between the gossiper and recipient, it has potential to damage relationships with the talked-about third party (Michelson et al., 2010).
While gossiping may build alignments with recipients and affirm solidarity between individuals or groups (Benwell, 2001), it can also be a mechanism for elevating one's position within a social hierarchy. Gossiping can be a means of enhancing prospects in social competition (McAndrew et al., 2007). Through gossiping the social position of the talked-about third party becomes downgraded, which by contrast elevates the gossiper to a higher social status (Tholander, 2003). Engaging in gossiping, however, may risk threats to face (Goffman, 1999) as gossiping is a risky social endeavour which is contingent on trust that the recipient will align with the gossiper (Grosser et al., 2010). In practice this means that the person offering the gossip could be perceived by the recipient in a negative way, rather than the object. This has been shown with other social actions, such as complaining, where the complainer risks being judged as a whinger (Edwards, 2005).
Paradoxically, although gossiping is morally sanctionable, it is ubiquitous (Foster, 2004) despite the gossiper's vulnerability to potential threats to face. It is thus incumbent upon the speaker to engage in some interactional effort to increase the persuasive nature of the content of what is said and amplify its coercive power (Kurland and Pelled, 2000). One way of managing gossip whilst maintaining the social relationship, is for the speaker to present the information as a factual description to disguise its gossiping quality. This is because many of the ethical condemnations of gossiping relate to the rules of privacy and therefore people will seek to guard themselves against a charge of indiscretion (Foster, 2004). This clearly indicates not only that the interactional responses of the recipient shape and direct the gossip (Fine, 1986), but also the context and setting in which the gossip occurs (Behnke, 2007).
Gossip can occur in formal informational exchanges and informal conversations within organizations (Mills, 2010), and much of the literature focuses on organizational settings, generally considering business contexts. Conversation analysis recognizes that institutional talk has some different features from mundane conversation (Drew and Heritage, 1992). So while gossip may occur informally within an institutional setting, such as in a waiting room or over the photocopier, the features of talk recognizable as the social action 'gossip' may also occur more formally in the institutional setting. One institutional setting where gossiping has particular distinctive features is in mental health settings. Our interest in this article is not in the gossiping between professionals outside of the formal institutional talk, or between patients as they await therapy; rather, we focus on the social actions occurring during therapeutic interactions.
In the specific context of family therapy, although the term gossip may seem incongruent with the institutional activity, it does share some typical features with gossip in other settings. The common features that it shares are that it is generally negative, evaluative, remedial, triadic, and sanctions are relevant. Gossip in therapy, however, does have some unique features. It differs in three identifiable ways. First, in family therapy children and adults are usually present together. Previous research on gossiping has tended to only explore the qualities of adult-to-adult (e.g. Foster, 2004;Tholander, 2003) or child-to-child gossip (e.g. Fine, 1986;Goodwin, 1982). Second, the talked-about third party, usually absent in the gossiping context, is typically but not exclusively present in the family therapeutic setting. Third, the therapeutic goals of family therapy shape and contextually frame the gossip. While the application of the concept 'gossip' may seem unexpected given the three core contextual differences, we deliberately stretch the meaning of this term in order to illustrate the significance of the social action that is being performed. In therapy there is an intrinsically asymmetric relationship between therapist and clients. Arguably there is also an even greater asymmetry between adults and children owing to children typically being only afforded half-membership status in adult interactions (Hutchby and O'Reilly, 2010;Shakespeare, 1998).

Aims of the article
In this article we aim to explore the process of social positioning between parents and children within a family therapy context. We investigate how parents seek to build alignment between themselves and the therapist, simultaneously distancing themselves from their child's behaviour. In this article we also consider more widely how parents talk about their children during therapy and the multiple discursive strategies used to 'do' therapeutic work. We explore how therapists manage these delicate social actions and resist particular alignments in order to maintain the wider therapeutic relationships with the family unit. We also examine the position of the child as the talked-about other in triadic interactions as at times 'pseudo-present', when the child freely interjects with a turn without invitation, and at times 'pseudo-absent', when the child refrains from interjecting a turn without invitation. They are therefore not invisible/ignored as is anecdotally suggested with many vulnerable groups, but neither are they fully present as they become talked about by the adults in the room.

The discursive approach
We utilize the discursive approach for studying family therapy as this version of analysis is methodologically congruent with family therapy theory and practice (Roy-Chowdhury, 2003). For our analysis we follow Edwards and Potter (1992), which has the benefit of using a conversation-analytic framework to elucidate the nuances of interaction. Using this type of analysis allows the researcher to explore the contribution of each party within the therapy from their respective positions (Roy-Chowdhury, 2006). This allows for a rigorous analytically and empirically grounded account of the data.

Setting and context
The data for this research were provided by a UK-based family therapy centre. We were provided with approximately 22 hours of video-taped sessions of naturally occurring family therapy. Data consist of two therapists, Joe and Kim, and four families (see Table 1).
The family therapy team uses a systemic approach and works with families of children who have mental health problems and diagnosed disorders. This team of family therapists routinely video-tape the sessions as part of reflecting clinical practice, and thus the sessions were not primarily recorded for research purposes. Informed consent was obtained for the tapes to be used for research.
The video-taped data were subjected to transcription in accordance with the analytic method, and Jefferson guidelines designed for conversation analysis were followed (Jefferson, 2004).

Ethics
For our research we utilized the principlist approach to ethics incorporating the four core principles of autonomy, justice, beneficence and non-maleficence (Beauchamp and Childress, 2008). In practice this meant that informed consent was collected from managers, therapists and families.

Analysis
A notable feature of talk in institutional settings that distinguishes them from mundane conversations relates to rights of access to the conversational floor. In family therapy the therapist has primary authority to engage or disengage, invite participation or obstruct turns. Through shifts in category alignment different members take up particular positions and position others in relation to one another. Shifts in alignment between members simultaneously work to collude and exclude. This creates a context whereby elements of the social action 'gossiping' are displayed in the sense that there is a 'talked-about third party'. In family therapy, however, the third party is typically present. In our data corpus, The four families were given the pseudonyms of Clamp, Niles, Bremner and Webber.
The Clamp family consisted of two parents, Daniel and Joanne, one male uncle, Joe and three children, Phillip ('special needs'), Jordan ('handicapped') and Ronald ('learning difficulties').
The Niles family consisted of two parents, Alex and Sally, and four children (one with a pending diagnosis), Steve (undiagnosed -suspected ADHD), Nicola, Lee and Kevin.
The Bremner family consisted of the mother, the grandmother and two children, Bob (Autistic Spectrum Disorder) and Jeff ('mentally handicapped').
The Webber family consisted of two parents, Patrick and Mandy, and four children (one with a diagnosed disability), Adam, Daniel ('special needs'), Patrick and Stuart.
Terms describing the children (e.g. 'handicapped') are the terms used by the families themselves.
the talked-about present third parties are usually the children, who are talked about in a particularly derogatory way. This has implications for the appropriateness of conversational content in family therapy (O'Reilly and Parker, in press). Whilst this is the case for the majority of the data presented, we begin, however, by introducing an adult triad. At this point in the therapy it is only the adult parties present as the child has been removed from the room earlier in the session. There has been a long discussion about their teenage son Daniel's sexual behaviour which led to a conversation about whether Daniel may have had access to pornographic material in the home. The father orients to his potential accountability in relation to this by providing an historical account of his limited pornography usage. This discussion about the use of pornography in their marriage occurs exclusively between the father and the therapist, in front of the 'talkedabout' mother. The therapist makes several attempts to regain the conversational floor from the father. His first attempt is interruptive of the father's turn 'I'm going to' (line 2), which is unsuccessful. His second attempt, despite occurring at a transition relevant place (Sacks et al., 1974), is still unsuccessful as the father interrupts his attempt. Finally, his third attempt, prefaced with the mother's name 'Mandy' (line 10), has a recipient selection function and is thus a more powerful interactional device for usurping the father's turn. This is further emphasized by his exclusion of the father by using the singular pronoun you, 'I'm gonna talk to you' (line 10).
Notably in lines 14-16, where the preceding talk is self-referenced by the therapist as 'gossip', he specifically makes relevant the gender category of 'men' (line 14); this indirectly orients to the mother's exclusion from the conversation as a woman. This is further indicated by the inclusive pronoun use of 'we're' (line 15). It is this excluded present third party status which appears to constitute the therapist's framing of their social action as 'gossipin' (line 16). This orients to the potentially detrimental aspect of the nature of the talk and the function of gossip to exclude the third party (Guendouzi, 2001). This exclusion of the mother is treated by the therapist as troublesome, and this is displayed in his turns which function to re-include her in the conversation. The mother responds to this by acknowledging and dismissing the impact of the trouble by saying 'it's alright' (line 13). Additionally, following the introduction of the description 'gossipin' (line 16) by the therapist, the mother produces a short series of laughter particles. This may be indicative of her orientation to the trouble as being treated lightly, which is further evidenced by the therapist not joining in (Jefferson, 1984). Jefferson notes that if the recipient were to laugh it would display insensitivity to the trouble and by not laughing alignment can be achieved, which is something oriented to by this therapist.
In the system of family therapy it is desirable that all parties within the family are afforded equal status. However, in this context, where blame and accountability are paramount, it raises a number of anxieties, particularly for parents, and through the process of collusion and alignment between two or more parties, others by default become excluded or marginalized. A tension is then created in an environment loaded with accountability, and in the process of helping families share accountability, alignment shifts are inevitable.
This tension is further exacerbated by the presence of children in family therapy. The most prevalent strategy of positioning the child as the problem serves to help parents mitigate their accountability for requiring therapy (O'Reilly, 2005: O'Reilly and Parker, in press). The process of simultaneously positioning the child as accountable and deflecting responsibility from the parents, functions to largely exclude the children from the therapeutic conversation. What this creates is an environment whereby the children are talked about in a derogatory way in front of them, thus having a 'gossiping' quality.

Part 1: Building alignment
A device that parents use to manage the complexity of their accountability is alignment building. In order to mitigate their responsibility for their children's behaviours they work to exclude the child during the therapy session as the 'problem other'. By casting their child as the talked-about other, by default an alignment is sought between the parents and therapist. Gossiping clearly has a third-person focus, substantiates behaviour and contains a pejorative evaluation which in turn creates an 'us' and 'them' context (Eggins and Slade, 1997). Gossiping also tends to deal with information that typically requires the recipient to accept the information as factual (Michelson and Mouly, 2002). In order to achieve this rhetorical function we identify three incremental strategies used to promote this: 'mere telling', actively voicing to authenticate and providing physical evidence.

Telling
In mundane conversation when a speaker 'tells' a recipient about an event, within that telling there are multiple possibilities of social action that transcend mere description (Edwards, 1997). Within the context of family therapy, telling is ostensibly a necessary prerequisite for information gathering. However, the social action of gossip can also be disguised in its presentation as a simple mechanism of information exchange (Foster, 2004). The following extracts are exemplars of these choices in parental reporting of general behaviour, specific actions and using derogatory descriptors related to their children. The therapist opens this extract with a question directed towards all members of the family. The generality of the father's response 'things are getting worse' (line 3) is nonagentive and non-specific, but indicates a general trend. He then upgrades his progressive generalization from 'getting worse' (line 3) to 'a lot worse' (line 6), adding the third person reference to his son, 'Jordan'. This explicates that the general 'things' relate more specifically to Jordan, while still providing a fairly general description of Jordan's behaviour as 'very violent' (line 6). General descriptors of behaviour tend to give way to more explicit examples of specific actions as adult family members work to dis-align themselves from their children and to construct their versions as believable. The specificity which builds plausibility within this extract is constructed by the telling of detail. To provide specific details about an event or situation is to make it more authentic; vivid detail can be used to build the factuality of the account (Potter, 1996). The specific details of the location, 'on holiday' (line 1), 'the stairs' (line 4), the action 'pushed' (line 4) and the numeric detail 'second time' (line 6), add validity to the father's version. The positioning of agency through ascribing intent 'he pushed' (line 4) and the severity of outcome 'nearly killed a child' (line 3) function to imply the dispositional character of Jordan. The use of detail to authenticate accounts is a common discursive strategy in family therapy when parents work to position their child/ren as problematic (O'Reilly, 2005). By positioning the problem 1 2 3 4 5 6 7 1 2 3 4 5 6 7 8 9 within the child, parents work to resist a systemic interpretation of the child's behaviour which affords them the opportunity to attempt to save 'face' (Goffman, 1999). Problematically, while managing this face-saving dis-alignment by diverting blame from themselves to the children, the negatively talked-about children are actually present in the room. Notably, although these vicarious accounts are produced in a manner which signals the child's behaviour as problematic, the child (Jordan) fails to respond as may be expected by offering a rebuttal, in the form of an excuse, justification or apology for that behaviour (Sterponi, 2009). Instead, he indicates his 'pseudo-presence' by vocalizing in overlap what is recognizable as a child's strategy to indicate 'not listening' by repeating the sound particle 'la la la' (lines 2, 8 and 9). The following extracts are further exemplars of negative evaluative comments reported by parents to the therapist within the hearing of the child. Bearing in mind that the children being described by their parents are listening to what is being said about them, in extract 4 the child is described as 'schizo' (line 2) and in extract 5 as a sexual 'predator' (line 4). Both of these descriptors work to locate the child's behaviour as dispositional and internal rather than inter-relational. Again, the pseudo-absence of the child is notable in that although party to the conversation, and clearly the subject of negative evaluative comments, the child does not respond to the allegations in any of the expected ways. The distancing of parental responsibility is further developed by alignment management using discursive techniques such as the selective self-referent pronoun 'we'. By stating 'we've got to sort' (extract 4, line 1), this lexically excludes the child and thus aligns with the therapist in seeking to 'fix' him.
Within the theme of 'telling', we have outlined three discourse strategies which work to authenticate a particular version of events. Thus telling is not merely a neutral activity for the sake of simply informing the therapist about family life, but actively constructs the nature of the family's problems as located specifically within the child/ren.

Active voicing
When building alignment, active voicing (Wooffitt, 1992) is used as an additional element which co-occurs with telling. This functions to authenticate the parental construction of the child by re-enacting the specific details of the described event. In both of these extracts the sequential ordering of the narrative follows the same pattern. Both begin with the parent repeating in the active voice what was said both by themselves and the child at the time and conclude with the subsequent extreme behaviour the child displays. The parental active voice is constructed as reasonable in both content and tone. This is contrasted with the unreasonable response from the child. In extract 6 the child's inappropriateness is highlighted by the use of the swear word 'fuckin' and in extract 7 by the increased emphasis of delivery. Contrast structures (Smith, 1978) are powerful persuasive devices which mark out the differences between the reasonableness of the parent and the unreasonableness of the child. This functions to develop the construction of the child as the problem and distance the parent from blame, building the alignment of the parents with the therapist to 'fix' the child. The use of the 'active voice' makes claims made by the parents more difficult to refute, which is notable given the presence of the talked-about child: an available party to potentially deny or qualify any claims made.

Evidencing
Given that claims need to be qualified in order to validate them, physical evidence is used as a way of substantiating those claims. Problematically, where there is contradictory evidence immediately available, this can be managed in the current context. In these extracts the parents both orient to the potential disbelievability regarding the claims they make about their children's behaviour at home. They do so by pre-empting the therapist's possible scepticism by drawing attention to the contrast between the child's behaviour in the therapy session and at home. In extract 8, Bob is referred to as sitting in the therapy session 'as good as gold' (line 1), compared to at home where he 'overtakes everything' (lines 2-3). This is similar to extract 9 where Steve is constructed as 'answering your questions' and 'listening' (line 3) but 'not paying attention' (line 4).
In therapeutic contexts it is not sufficient to simply tell the therapist about family life events, rather the family members have interactional work to do to support their claims. This is achieved by adding evidencing to telling and active voicing, to build a convincing picture of their version of family life.

Extract 10
Dad: = Show Joe yer arm >where you've s-< cut a:ll yer arm and >says th[e roses< done it Mum: ['e reckons the rose bushes done it but I s->I reckon 'e's done it< with somet Dad: ↑Show Joe yer arm then [7 lines omitted] Steve: there's nothin' th::ere Mum: Don't tell lies Dad: Looks like 'e's tried t' scratch the name o::r somethin' in 'is arm Steve: NO I ain't (Niles family) In this extract the child's version and the parents' version regarding the nature of the scratches on Steve's arm are constructed as competing versions of an event. The 'truthfulness' of the parental version is evidenced through the presence of physical proof. In providing this evidence they simultaneously discredit the child's version and diminish the status and character of the child. By contrast they elevate their own status as plausible, reasonable parents, which in turn serves to foster the alignments between them and the therapist. Interestingly, the child in extract 10 refutes the parental explanation of the scratches 'there's nothing there' (line 13), in response to the invited action from the father 'show Joe yer arm' (line 5). The pronoun 'your' serves to reference Steve as selected to respond. In contrast to this, however, the father goes back to directly addressing the therapist and talking about Steve in the third person: 'looks like he's tried to scratch' (line 15). Notably Steve initiates a denial 'NO I ain't' (line 17), thus using an opportunity as a pseudo-present (actively contributing without invitation) talked-about individual. Not all children in therapy utilize these opportunities and thus become pseudo-absent.
In this section we have demonstrated that parents will talk to the therapist about their children in front of their children. They present a narrative of home-life events and construct a negative and derogatory picture of their child as a way of positioning the child as the reason for therapy. By positioning the child through telling, active voicing and evidencing they are able to manage their own identity as 'good parents' in a context in which they become accountable for their parenting and family system. By gossiping about the third party one can elevate one's social position by downgrading the position of the talked-about other (McAndrew et al., 2007). By talking about the child in a derogatory way and positioning the child as the problem, they are able to deflect parental responsibility and blame to the child. Through these techniques they attempt to build alignments with the therapist and engage him/her in the gossip. The institutional context and role of the therapist, however, mean that this alignment of adult parties is not always successful, and in part two we consider how the alignments can be resisted and how dis-alignments are managed.

Part 2: Resisting alignment and dis-alignment
In gossip sequences, the gossiper may have some difficulty in obtaining the collaboration of the recipient, and resistance to alignment may be encountered (Tholander, 2003). The parents do considerable work to build alignments with the therapist and to exclude the child through actively dis-aligning with them and their versions of events. However, both the therapist and the children manage stake and interest in actively resisting this process (Potter, 1996). Children tend to resist and deny the accountability positioned with them to varying degrees, but their half-membership status is recognized by the therapist, who actively seeks to engage with them to ascertain their versions of events. Notably the therapist does do some work to demonstrate that despite engaging the child in the therapy, this does not mean that he does not believe the version of events presented by the parents.

Denying
Family therapy is an environment of competing versions and in this context the 'talkedabout other' has the opportunity to negate the version presented. Through cumulative narratives parents build a generalized version of the child whereby they position the inherent character of the child in a derogatory and negative way. This locates the 'problem' as dispositional rather than inter-relational. Dis-alignment with the child's problem is created through the repeated telling of instances of problem behaviour, and simultaneously aligns the parents with the therapist as jointly working together to 'fix' the problem child. At any point during gossip the speaker's negative evaluation is open to challenge (Guendouzi, 2001), and given the nature of the talk about them and the therapeutic context there are instances where children resist the versions presented.

Extract 11
Mum: And 'e got 'is hair off with that and >chucked it< on the flo::or >and I says< we[ll once ↓yo-Steve: [NO I HAVEn't I dropped *it on the ↑flo:or Dad: <YOU [threw it> across the livin' ro:om befo:re now Mum: [N-<YOU CHUCKED IT> .hh I was ↑there and seen ya >and I says< once you break that <you ARE NOT 'avin' another one> (Niles family) Commonly denials contain negations and are typically followed by an account or correction (Ford, 2002). In this extract Steve begins his turn by emphatically denying the version presented by his mother: 'NO I HAVEn't' (line 3). The increased volume of this interjection and its interruptive nature serve to enhance the forcefulness of the denial. This is immediately followed by a correction: 'I dropped it' (lines 3-4). Although in this environment the talked-about third party ostensibly has the opportunity to deny or refute versions and offer corrections or alternatives, what is demonstrated here is that on the rare occasions when children take up that prospect their accounts are quickly and forcibly negated; this is also indicated by research which shows that children's interruptions are treated negatively (O'Reilly, 2006). This is achieved here through the parents' use of collaboration, emphasis and upgrades. Problematically, this creates a hostile atmosphere whereby the child's attempts to offer their versions are discredited. The therapist's role is complicated, therefore, as the therapist seeks to balance the dis-alignment sought between family members as well as respect and work with the parents' versions. The therapist manages this through two main strategies: actively engaging the child in the therapeutic conversation and discursively portraying that he believes the parental accounts.

Engaging
The perspectives of children have become more of a concern for family therapists over time (Cooklin, 2001), but despite this, children's half-membership status in interactions means that it is more difficult for them actively to claim the conversational floor. While the parents could provide their children with floor space, it may be counterproductive to their stake in producing a particular version of the child to allow them that opportunity. Thus it falls to the therapist to use his/her authoritative full-membership status to create a more inclusive environment.
(Clamp family) 1 2 3 4 5 6 1 2 3 4 5 6 7 8 9 Extract 13 FT: we did a lot of talkin' abo::ut (0.8)some of the things that you do (.) that your mum and dad aren't too happy about and I guess I jus' wanted t' say that I know that it's re:ally difficult t' sit there and ↑listen

(Clamp family)
Gossiping is considered to be a negative social action (Noon and Delbridge, 1993) and the therapist orients to the possibility that being talked about in a derogatory way may be upsetting 'does that bother you?' (extract 12, line 4) and 'it's really difficult' (extract 13, lines 4-5). The therapist positions himself as an active member of the collusive partnership between himself and the parents and acknowledges the exclusion of the children during those interactions. This is evident through his use of 'we' (extract 13, line 1), and his listing of the three adults (extract 12, line 1) which aligns him with the parents and includes the three adults together in their discussion about Phillip. In extract 1 the therapist acknowledged the shift in alignment that occurred and made attempts to redress that. Here in extracts 12 and 13 a similar correction to the shifts in alignment is demonstrated.

Extract 14
FT: ↑what we're hopin' t' achieve and >I know that< you're looking uneasy already Da(h)niel Mum: Heh he[h heh FT: [I know that this isn't easy stuff for you t' talk about >is it< (0.6) FT: especially with your parents (0.2) present. but but we kindda had an <idea that> (0.6) FT: actually it's re::ally important <for us all> t' be able t' talk about as well

(Webber family)
Not all the therapist's engagement of the child is retrospectively repairing a social breach. There are occasions when the therapist anticipates what is likely to come in the session and prospectively engages the child early on. The therapist interrupts a general opening to the session aimed at the whole family with an insertion sequence aimed specifically at the child. He displays noticing that Daniel looks uneasy -'you're looking uneasy already Daniel' (line 2) -and does some interactional work to validate that feeling before continuing with the original turn. What this displays is recognition that it may be potentially difficult for Daniel to make contributions to the therapy, but that it is important for all parties to contribute: 'important for us all to be able to talk' (lines 10-11). This makes relevant the prospective nature of the talk. We note, however, that although retrospection and prospective engagement are common, it is also possible for the therapist to orient to the pseudo-absence of the child in the immediate temporal space. Children are active participants in the family therapy context and thus normatively the therapist has some responsibility for taking steps to ensure that the child can express their views (Barker, 1998). In this extract the therapist orients to this normative framework by highlighting the relevance of multi-party contributions, including those from Daniel. By using a lexical and relational method of hearer selection, 'mummy' (line 1), the therapist specifically selects the two children as recipients of the question. This works to engage the children and affords them privileged access to the conversational floor space by excluding the other adult parties present. In this extract, version elicitation is primarily directed towards the two children, with the adult members occupying positions as overhearing others. This shifts alignment between the therapist and the children and positions the mother as the talked-about third party.

Believing
When therapists shift alignments to the children as an engagement strategy there are potential risks to the relationships between the therapist and the adult members. A discursive resource to balance the tensions between multi-party versions of events is for the therapist to display not disbelieving. Where there are competing versions of events, the social actions the parents are engaged in is an attempt at polarization, whereby their version is privileged as 'true' and others by default are 'false'. Alignment/dis-alignment strategies can be similarly oppositional, leaving the therapist to reconcile both versions rather than favouring one over the other. The danger of an alignment shift which seeks to engage children therapeutically is that it may provoke an anxiety in parents to display additional evidencing, considered earlier in the article. One way in which the therapist works to avert dis-alignment with the parents is to actively deny disbelief. In both of these extracts the therapist orients to the believability of the parental versions of events. The therapist treats their actions of cumulative evidencing as attempts to convince him of the factuality of their version of the child. This is displayed by his statements 'I have no reason not to believe what you tell me' (extract 16, lines 2-3), 'I do not disbelieve for one minute' (extract 17, line 1) and 'I don't need Steve to do it here to believe you' (lines 6-7). What this achieves is a re-alignment with the parents which averts their disengagement with therapy. Problematically, this has potential to disengage and dis-align with the children. The challenge for therapists therefore is to retain all members within the zone of relational accountability without dis-alignment with any member.

Discussion
This study demonstrates the complexity of teasing out the social actions inherent within the family therapy environment. Managing blame and accountability with multiple parties is a delicate endeavour for the family therapist to work with. Whilst the conversational features analysed in family therapy have many of the recognizable characteristics of gossip in other mundane and institutional settings, the presence of children and the issue of blame is particularly salient in this context.
Our analysis highlights how blame and responsibility are managed, how the presence of the child is dealt with, and how therapy is accomplished. The first part of the analysis focused predominantly on the perspective of the parents who sought to manage their own accountability for their child's behaviour by attempting alignments with the therapist. Parents attempted to strengthen their relationships with the therapist in three incremental ways. They used mere description, active voicing and providing evidence to substantiate claims. The second part of the analysis focused predominantly on how those alignments were challenged and resisted. Children challenged alignments by denying claims, while the therapist worked to engage the children in the therapy and simultaneously demonstrated belief of the parents' versions. The advantage of using a conversation-analytically informed discursive approach for research of this nature affords an opportunity for a detailed systematic analysis of the sequential aspects of interaction within the family therapy setting. It further provides a framework to interrogate the detail of the social actions as they occur in context, allowing an exploration of process and function rather than simply content. Other qualitative methodologies offer some useful insights into family therapy by offering thematic descriptions and interpretative comments, but this approach has additional value as it fully evidences claims in naturally occurring data and is able to explore the nuances of social action that are occasioned in therapy talk.
In this article we have explored the social action of gossip. In family therapy, the talk has some distinctively identifiable features of gossip as it is normatively considered, although it is unusual to consider any talk in therapy to constitute gossip in the conventional sense. Our use of the term 'gossip' in this article has been used as a useful analytic tool, rather than a literal description, by expanding the typical meaning of the concept and translating it to an institutional context. By using discursive techniques we are able to identify instances where talk is produced to be heard as simply factual description, whilst maintaining an underlying alternative performative social action. In therapy there is a necessity for information gathering as the therapist requires access to the family's private lives. Problematically, as our analysis demonstrates, this element of therapy is loaded with negative descriptions and 'gossip' which function to dis-align from children's behaviours in order to maintain face.
The negative descriptions of children, the management of accountability and attempts at alignment make family therapy a complex institutional accomplishment. Parental descriptions in family therapy can perform multiple social actions simultaneously. Therapists can utilize this knowledge to facilitate decisions about how and from whom narratives and information are elicited, with a conscious awareness of the potential performative nature and functions of those descriptions.
The evidence base in family therapy suggests that there is some concern in the field regarding the inclusion of children in family therapy sessions due to the potential harm that may be caused (Miller and McLeod, 2001). Miller and McLeod note, however, that there is a unique advantage to working with parents and children together as it adds an important dimension to the therapy and facilitates child-parent relations. Recently, the UK Department of Health has taken the position that clinical practice should be evidence driven (Macintyre et al., 2001), and family therapy particularly has a limited evidence base (Roy-Chowdhury, 2003). Whilst anecdotally therapists generally promote the value of including children in family therapy sessions, there is limited empirical evidence to support the efficacy of this approach (Miller and McLeod, 2001). This article goes some way to providing clear evidence of how children are included in family therapy and the ways in which they are 'gossiped' about.
From the evidence in this article we raise a number of considerations for the field of family therapy to reflect upon. When therapists make a clinical judgement to offer initial sessions with the parents without the children present, we propose that they may want to consider the following issues. Therapists may benefit from a more explicit awareness of the nature of multiple social actions which may occur within descriptions, such as accountability, mitigation, excusing and managing threats to face. With this in mind it may be advisable for therapists to work towards eliciting descriptions from parents which are as much as possible performing 'information only' functions, in that they are simply factual and free from judgements. During this time therapists may be able to communicate specific therapeutic boundaries about limiting the accountability management and blame resistant elements of ostensibly descriptive talk, which could be translated to later sessions when the children are also present.
When therapists make a clinical judgement to include the children in family therapy sessions we propose they may want to consider the following issues. Although within the NHS it is recommended that professionals work in a child-centred way (Pickering and Busse, 2010), the pressure towards alignment from the parents, as shown in our data, can make achieving this difficult. Parents have a strong stake in the process and outcomes of therapy and this impetus may be a driving force in their desire to dominate sessions with their own versions of events. Because of this the therapist may need to exert additional effort to afford children more opportunities to offer their own perspectives. Additionally, it will be helpful for the therapist to be cognizant of the danger of inadvertent iatrogenic consequences on the child, who may be detrimentally affected by being party to the kind of negative talk from parents that has been exemplified in this article. Gossiping about a family member can damage relationships with that talked-about party (Fine, 1986) and unchecked or unbalanced negative descriptions of children in therapy could have an undesirable or even harmful impact on the child's mental health. It is, however, recognized that this can be particularly difficult for the therapist to be aware of and manage amidst the dynamics of the moment-by-moment interaction, and thus the value of the reflecting team is acknowledged as especially useful in these instances.
The analysis presented in this article highlights that individuals with a 'half-membership' status may be pseudo-absent or pseudo-present in the interaction. This positions them as the 'talked-about' third party with limited access to the conversational floor. Anecdotal narratives and clinical experience indicate that many vulnerable groups are afforded this pseudo-presence in conversations; those in wheelchairs are typically talked over, those with learning disabilities are talked about in front of them, patients lying in hospital beds are often talked over, and in some cultures women are not given privileged access to the conversation. This article has dealt only with the therapy context, but extensive literature searches reveal that there is limited evidence related to any of these 'half-membership' groups and how they are 'gossiped' about in their presence. Future research has a long way to go to understand the nuances of interaction with these particularly vulnerable groups.