Indeterminacy of actions: Working out a relevant next in interaction with people with late-stage dementia

Abstract

This study examines video recordings of activities within an elderly care home, particularly focusing on interactions involving people with dementia. The study presents instances where the relevance of the current conduct—and consequently the generation of a fitting response—appears indeterminate to the co-participant. Despite the uncertainty surrounding the mutually understood ‘constitutive expectancies’ (Garfinkel 1963) and a lack of shared understanding of ‘motivational relevancies’ (Schutz 1970), the participants continue to engage with one another. The analysis reveals that, in the face of an unrecognizable set of conduct and the indeterminacy of subsequent actions before closing down the activity, participants strive to maintain some degree of intersubjectivity by preserving or revisiting the constitutive order of their interaction. This commitment to the ‘co-operative’ nature of human actions (Goodwin 2018) is argued to be central to their interactions. (People with dementia, co-operative actions, conversation analysis, multimodality, ethnomethodology, intersubjectivity)*

INTRODUCTION

Initiating an action, determining its relevance to the current flow of interaction, and recognizing the sense and meaning of a set of conduct are not always straightforward in social interactions. Sometimes participants’ contributions to the interaction, whether through initiation or response, are not immediately recognizable. In other words, the means that participants use to produce their actions may not conform to the recognizable, conventional social actions (see Enfield & Sidnell 2022, for a discussion of social actions). This can be particularly true in interactions with individuals with communicative difficulties, who may have problems initiating interactions or displaying comprehension of their interactional partners’ conduct.

This study deals with initiations and responses in interaction with people with late-stage dementia who are unable to produce speech. The aim of this study is to demonstrate how, despite the indeterminacy of action in interactions with people with dementia, the interactional partners often strive to maintain some level of intersubjectivity through engaging with them in co-operative actions (Goodwin 2018). Although collaboration for the production of shared practices with people with cognitive and communicative difficulties may differ from the machinery of everyday typical social interactions (see Sacks 1992), efforts to sustain intersubjective relations by identifying and reestablishing the constitutive order (Garfinkel 1963, 1967, 2019) are evident even in these types of interactions. As the analysis of this study shows, the interactional partners may co-operatively create opportunities (through intersubjectively recognizable procedural sets of situated conduct) to work out the sense of the conduct and, thus, the constitutive order of the interaction (Garfinkel 1963, 2019).

When referring to the indeterminacy of actions, I specifically mean indeterminacy in terms of constitutive expectancies of producing practical actions (see Garfinkel 1963, 2019) as it is manifest situatedly in social activities (see also the discussion of ‘determining’ in Sharrock & Button 2003:261). Indeterminacy may be shown in how a set of conduct is momentarily treated as not being recognizable and actionable to be responded to according to the constitutive expectancies emerging from ordinary conduct. In such cases, revisiting the progression of action might be in order. This may include temporally suspending the progression of the action, restarting the conduct (or the whole sequence), or resuming the activity with a (delayed) generation of further conduct that is often synchronically adapted to the skills of the interactional partner (see ‘recipient design’ in Sacks 1992:385–90). In other words, what the person with dementia does is not immediately recognized as fitting into a particular category of social action, making it difficult for their interactional partners to determine the next relevant course of action. The treatment of an action as indeterminate is analytically identified in response sequences, for example, by relatively long pauses and perturbations in talk or/and together with repair initiation, such as the repetition of the prior conduct (for the organization of repair, see Schegloff, Jefferson, & Sacks 1977), asking for confirmation, and so on (see the discussion of response mobilizing in Stivers & Rossano 2010b). These responses indicate that the conversational partners of people with late-stage dementia are unsure how to interpret the current conduct and determine the relevant next action (cf. Antaki, Chinn, Walton, Finlay, & Sempik 2020; see also Pilnick, O'Brien, Beeke, Goldberg, & Harwood 2021).

UNDERSTANDING ACTION IN TURNS OF TALK

Studies of interaction (often with neurotypical participants) have pointed out that the understanding of a prior action is located in the subsequent turns (Sacks, Schegloff, & Jefferson 1974; Schegloff 2007; for a discussion of turn designs inviting particular responses, see Heritage 2002a, Stivers & Rossano 2010a, b, among others). This was conceived to bear two assumptions: Actors and actions can be distinguished through turns at talk, and the contribution of each actor can be singled out and understood as equal to an action. This also led to understanding action as an ascription made through ‘the assignment of an action to a turn as revealed by the response of the next speaker’ (Levinson 2013:104). This is also argued to be reflected in the ‘proof criterion’ (Sacks et al. 1974:729) for the analysis of how turns of talk are interpreted.

Moreover, as Deppermann & Haugh (2022) rightly argue, how produced formats are understood is a member's concern. This is what originally expressed by Schegloff (1996:165) that an interactional conduct (or the main business of turns of talk) is what the recipient must deal with, i.e. how a turn is heard and understood as a relevant response to the prior turn (see also Levinson 2013). Action is therefore understood as an invariably situated meaningful conduct (as in doing something), which is analytically (both by the interactants and the analyst) recognized in its course of production, and evidently so, often by not always, in following (responsive) action(s), or rather, in the next relevant conduct in the trajectory of a sequential development of the whole activity (Sacks et al. 1974; see also Raymond 2003; Thompson, Fox, & Couper-Kuhlen 2015). However, the subsequent (expected) response may not be produced—and its absence may or may not be treated as noticeable or accountable (see the discussion on this by Stivers & Rossano 2010a).

As Charles Goodwin (2018) has shown us through studies of interaction with both neurotypical and non-neurotypical participants, the production of turns of talk and actions is cooperatively constructed, and action is not a production of a single actor but is co-operatively made through transformative and accumulative interactional processes (see also Lerner 1991); what Goodwin (2018) points to is not only the collaborative (though asymmetrical) distribution of interactional labor for building actions, but also the fuzziness of the boundary of action, with an often unclear beginning and end. It is not always a clear-cut case when, for instance, a piece of conduct is understood as a request before it is followed by a grant or decline. The import of actions may be a matter for the interactional partners to work out, and not just a ready-made category of social conduct (such as greeting, request, etc.; cf. Enfield & Sidnell 2017, 2022). This highlights the significance of sequential analysis of action production and its recognition, by emphasizing the necessity of analyzing the conduct of ‘all’ participants, and their contributions, at any point of a sequence of social interaction while an action is produced. This is particularly significant when analyzing interaction with people with cognitive and/or communicative difficulties (see also Goode 1994).

One of the common practices to work out the meaning of conduct is ‘negotiation’ in the form of confirmation seeking and correction practices (Schegloff, Jefferson, & Sacks 1977). This is particularly reflected in the literature on communication with people with various diminished communicative and cognitive abilities (see Damico, Simmons-Mackie, & Wilson 2006; Beeke, Wilkinson, & Maxim 2007; Griffiths, Barnes, Britten, & Wilkinson 2015). However, these practices may not be viable options when the conversational partner cannot respond to the initiation of the negotiation of meaning or repair. For instance, Antaki and colleagues (2020) note that in response to unintelligible talk, communication partners have various options, ranging from ignoring or not attending to unintelligible talk, to using minimal responses or insisting on understanding it by explicitly initiating correction. The authors (Antaki et al. 2020:975) suggest that the choice of response may depend on ‘the immediate job requirements’ and the sensitivity of correcting others to avoid threatening their face (see also Wilkinson 1995; Perkins 2003). Avoiding correction or glossing over misunderstanding has also been observed in other studies (e.g. Lindholm & Wray 2011; Pilnick et al. 2021). Pilnick and colleagues (2021:281) demonstrate how healthcare professionals may ‘shift away from a hard-to-interpret utterance’ onto ‘a different topic or task’ or ‘treating it as related to the task’ at hand. They also show that in doing so, professionals use other embodied conduct rather than talk when picking up clues for understanding the utterances that are hard to interpret in order to provide a response to them. Although these studies rightly emphasize the significance of interactional work, particularly repair work, in conversation where one party may not be able to contribute in a verbally intelligible way, the points highlighted in this study differ slightly in character. This study focuses on people with communicative difficulties who are unable to produce verbal speech and utterances. The main concern of this study is not limited to repair and the question of the production of account or correction, but extends to how the understanding and the formation of action become of collaborative process from the beginning, and how the indeterminacy of an action becomes a problem to solve for the participants.

UNDERSTANDING ACTION AND THE AGENCY OF THE PARTICIPANTS

When it comes to social action and the agency of the actors, there are two significant caveats regarding what the action is and what the relationship is between actions and the turns at talk. From an EMCA perspective, what is being done and the role of participants (identities) are endogenous to the fabric of situated practices (Rawls 2009). Action should therefore not be considered based on a socially objective category of social acts, nor merely a verbal category clearly manifest in speech production (often speech acts, such as complaint, request, offer, etc.; see Schegloff 1996; see also the discussion by Stivers & Rossano 2010a). Action is considered to be an indexically emergent gestalt of recognizable conduct with its boundaries convergent across various resources/modalities including talk, gaze, body (postures, orientations, or movements), and gestures. The relevance of how the production of conduct is understood as a particular action is also a member's matter (see Garfinkel 1967, 2022), by which the participants recognize what an interactant is doing and/or project what they may do next (see also Schutz 1973:218–20).

Action is moreover assumed to bring about some changes in the current social situation, including the course of the ongoing activity, the participants’ understanding of it, and/or also their positions, roles, and identities. To that change, the recipients of actions also contribute (see C. Goodwin 1979). If interactants treat each other's conduct as ‘the document of’ an action (Garfinkel 1967:40 for ‘documentary method of interpretation’), their treatment is reflexively evident in their own conduct, which is procedurally visible and/or hearable to the co-interactants (as well as to the analyst; see also the ‘proof criterion’ for the analysis of how an action is interpreted according to Sacks et al. 1974:729).

In sequential analysis of talk in interaction (e.g. Schegloff 2007), the proof of understanding conduct is often considered in structurally defined ‘slots’. For instance, the display of understanding of a first speaker's turn is sequentially positioned second to it in the turn taking system. The second position is therefore described as conveying the recipient's understanding. The confirmation of that understanding consequently occurs only in the next turn, which is the third position that is considered to be the warrant of ‘intersubjective action ascription’ (Deppermann, & Haugh 2022:4; Schegloff 2007). These structurally predefined positions for particular ‘items’ or ‘events’ are of course beneficial in many ways, for instance, where the co-participants can expect that second-turn-repairs/corrections may be confirmed or rejected in the third turn (Schegloff 1992). However, the organization of turn-taking may not fully capture aspects regarding the production of action that have been stipulated above, for example, action being the co-product of the co-participants (Goodwin 2018), a ‘multimodal complex gestalt’ (Mondada 2014) whose sense and intelligibility may be conveyed in simultaneous conduct instead of subsequent ones (Goodwin 2018; see similar inquiries in Stivers & Rossano 2010a; Deppermann & Haugh 2022).

Considering action beyond turns of talk and single speakers (see also Auer, Bauer, & Hörmeyer 2020) has led us recognize it not merely as a post-hoc ascribing activity, but as an emergent, describable procedure, and accountably so, in the course of the activity (see Garfinkel 1967:vii). Such an approach is not something new, particularly when it comes to studying social situations that are now termed ‘atypical interactions’ (Wilkinson, Rae, & Rasmussen 2020; see e.g. studies on ‘languageless’ deaf-blind children such as Goode 1994; people with aphasia such as Killmer, Beeke, & Svennevig 2020; and people with dementia such as Hydén, Majlesi, & Ekström 2022). In everyday life, especially in social situations marked by asymmetries of competencies due to cognitive and/or communicative difficulties, the process of action ascription is not always straightforward. Interactants are not always required to explicitly ascribe an action to the prior conduct. If understanding of an action means displaying its situated meaning in interaction, it may be displayed during the production of an action, or it may not be displayed at all (see Stivers & Rossano 2010a).

DATA

The video data used in this study were collected over the span of a year in a dementia unit in an elderly care home in a large city in Sweden. The unit housed seven individuals primarily diagnosed with Alzheimer's disease with five in the advanced stages of dementia. Two of the five residents—Soraya in her eighties and Emma in her nineties—are the primary participants in this study. Soraya, originally from Iran, used to speak Farsi as her mother tongue; Emma used to speak Swedish as her mother tongue. Both residents, however, have lost their abilities to verbally communicate. Nonetheless, they are treated by the caregivers as being able to comprehend talk/speech, as they are regularly addressed verbally. While Soraya produced neither speech nor vocalization during the year of fieldwork, Emma occasionally produced speech-like vocalizations, mainly in responsive turns. Both residents had severe memory loss, and were physically too feeble to move on their own, and thus required assistance for mobility and daily activities such as eating and drinking due to deteriorated hand dexterity.

Throughout the fieldwork, various activities in the care home were video recorded, including table talks, eating events, walking tours, and other recreational activities such as watching TV, playing bingo, choral singing, and participating in physical fitness. The study focuses on three interaction sequences: greeting, assisted drinking, and table talk using communication picture cards, also known as cue cards. The examples presented involve the researcher's interactions with Soraya and Emma. These sequences were chosen because they demonstrate interactional situations where the indeterminacy of action is typically illustrated. Additionally, the participants in these interactions had not been acquainted with each other for an extended period compared to the formal caregivers, and family members. If examples were selected from conversations between residents and their loved ones or long-time caregivers, the communication circumstances might have been different. The interactions with the researcher were all spontaneous and emerged naturally. The sequences are strictly approached from an ethnomethodological conversation analytic perspective, drawing evidence only from the immediate context of the interaction, without extrapolating beyond that.

The sequences were transcribed using transcription conventions by Jefferson (2004) and Mondada (2022). Action-illustration through drawings and the descriptions of bodily movements of the participants were also used within the transcription. The analysis employed an ethnomethodological and multimodal conversation analytic approach (e.g. C. Goodwin 2018; Goodwin & Cekaite 2018; Mondada 2022), providing a detailed analysis of the sequential and temporal organization of interaction. For data collection, appropriate ethical approvals were obtained from the Regional Ethical Committee (dnr. 2017/469-31).

RESULTS

This section presents and analyzes three examples. The first example illustrates an occasion of greeting between Soraya (SOR), the resident with late-stage dementia, and the researcher/visitor Ali Reza (AR). The second example shows how Ali Reza uses cue cards to communicate with Emma (EMM), the resident with late-stage dementia. The third example demonstrates how Ali Reza assists Emma in drinking coffee. To indicate the exact timing of nonverbal actions in relation to turns at talk, symbols such as asterisks (*), deltas (Δ), daggers (‡), and others are used in the transcript. The number sign or hash (#) is also used to indicate the exact location of actions depicted in pictures during turns at talk. The duration of pauses, shown in parentheses, is given in seconds (for multimodal transcription, refer to Mondada (2022)).

EXAMPLE 1: GREETING

Prior to the beginning of the sequence, Tina (TIN), a caregiver in the care home, assisted Soraya (SOR) in moving to the sitting room where the residents were watching a documentary film about Swedish nature, landscapes, and rural life. During the transfer, Tina engaged in conversation with Soraya. Upon their arrival, Tina points out the location where Soraya will be seated in the room (line 1). It is worth noting that Tina is multilingual and speaks different languages, including Farsi. The conversation with Soraya was conducted in Farsi.

When Soraya is moved into the sitting room in her wheelchair, Tina, while looking at Soraya, indicates where Soraya will be placed by saying, “here” (line 1). She places Soraya in her wheelchair next to the sofa where she also sits herself (next to Soraya). Tina then refers to the other residents in the room, and says “here. (.) everybody is gathered” (line 2), while, at the same time, securing the brakes on Soraya's wheelchair. At that moment, Ali Reza enters the room and walks past Soraya and Tina. Ali Reza places a recorder on the coffee table, which is located in front of the sofa (see Image 1 above). Soraya looks at Ali Reza, and Tina immediately comments on AR's presence:

Image 1. SOR attends to AR when he puts the recorder on the table.

Soraya's gaze directed toward Ali Reza (line 2) is attended to by Tina as a sign of noticing, a meaningful action that invites joint attention and provides a topic for conversation (see also Sacks 1992:87–97). When Tina mentions Ali Reza's name (line 3), he stops and turns toward Soraya and Tina. This creates a new participation framework in which Ali Reza is being noticed and given the role of an addressee when Tina introduces him to Soraya by saying “here is ar- mr. ali reza (he)'s also here” (line 3). At the exact moment as his name is mentioned, Ali Reza turns back and bends down to face Soraya. Tina draws Soraya's attention to Ali Reza's facing position by saying “look” (line 4), and points at his direction (see Image 2). This leads to a mutual gaze between Soraya and Ali Reza. Tina further continues with accounting for AR's presence.

Image 2. TIN points to AR, drawing SOR's attention to AR who is bending over and looking at SOR.

Tina introduces Ali Reza to Soraya as her guest (“oh you have a guest”), while Tina is still holding her finger toward Ali Reza (line 7). Pointing can be understood as contributing to maintaining the action, and in conjunction with the talk, holding the turn until AR takes over the floor (see also Mondada 2007). At this moment, Ali Reza greets Soraya in Swedish, saying hej (‘hi’) with chuckles, followed by salam (‘hi’) in Farsi, while maintaining a mutual gaze with her. He then expresses his pleasure in seeing her by saying “I'm happy to see you” (line 12). The first pair part of greeting sequences typically mobilizes response for the second pair (Schegloff 2007; Stivers & Rossano 2010b). However, Soraya does not reciprocate his greeting, either verbally or through any facial expressions. After a prolonged pause, both Ali Reza and Tina nod at Soraya (line 13), followed by Tina speaking on behalf of Soraya. She responds to Ali Reza's greeting for Soraya and at the same time provides a directive as to how Soraya could answer the greeting: “say me too” (meaning that Soraya is also happy to see Ali Reza, line 14; cf. Nilsson, Ekström, & Majlesi (2018) for talking on behalf of people with dementia). Ali Reza responds to this immediately, as if it was Soraya who had greeted back. In his response, Ali Reza seeks confirmation with “yeah?” and asks politely if she is alright, using the plural pronoun of “you” in Farsi, a sign of respect for older people: “are you(pl.) we:ll? are you(pl.) we:ll?” while nodding (line 15), encouraging a confirming response (Heritage 2002a). Tina laughs and departs from the conversation to attend to other residents in the room, while Ali Reza remains in the same position. During a long pause of approximately five seconds (line 17), Ali Reza continues to maintain gaze with Soraya, nodding a few times before engaging with her haptically by stroking her arm (see Image 3).

Image 3. AR strokes SOR's arm.

During the time when Ali Reza is bending forward, nodding, and stroking Soraya's arm, another resident and Tina are engaged in a conversation, which is not represented in lines 18–20. Subsequently, Ali Reza points to a nearby chair beside Soraya (see Image 4) and declares his intention to sit there (line 21), while simultaneously seeking confirmation from Soraya (line 23). Despite not receiving a response, Ali Reza waits 1.6 seconds before redirecting his gaze toward the other residents in the room and ultimately taking a seat next to Soraya.

Image 4. AR points to the chair next to SOR where he later sits.

This first example in (1)–(4) points to two significant issues regarding the formation and understanding of action, especially when verbal resources are limited. The example emphasizes the role of nonverbal cues in communication with people with limited verbal repertoire or an inability to use speech such as people with late-stage dementia. Co-participants rely on embodied resources, such as changes in gaze direction to make sense of the person's conduct and respond accordingly. Thus, the person's gaze direction is understood as sustaining attention and as an account-able action (see also Killmer et al. 2020). Moreover, the example above illustrates how the participatory roles which are staffed through the design of actions may be taken up on behalf of the person with dementia by their coparticipants. The contribution made by the person with dementia is mainly through direct gaze distribution as a device for the display of engagement. What the co-participants do takes into account that the conduct of the person with dementia is meaningful and can be responded to (cf. Auer et al. 2020; C. Goodwin 2018; Pilnick et al. 2021; Hydén, Ekström, & Majlesi 2024). In other words, they ascribe an agentive role to the person with dementia. In this sense, the production of action and its accountability can be considered a joint achievement (see also C. Goodwin 2018; Enfield & Sidnell 2022).

EXAMPLE 2: PICTURE CARDS

The second example deals with a table talk in which Emma, a person with late-stage dementia, is seated at a table alongside Ali Reza, the same visitor from the previous example. In order to communicate with Emma, who is unable to verbalize but can vocalize sounds, Ali Reza uses picture cards as communication cues. These cards depict everyday objects on one side and the name or definition of the object on the other. In this example, Ali Reza presents Emma with a card featuring an image of an egg.

At the outset of the excerpt, Ali Reza initiates a new topic by introducing a new card featuring an egg, which he shows to Emma (line 1). Ali Reza tilts his head and directs his gaze toward Emma, closely monitoring her reaction to the card (see Image 5). Emma, in turn, looks at the card during a long pause of 1.4 seconds before she averts her gaze away (line 2). Following this, Ali Reza employs a polar question, offering Emma a choice between ‘yes’ and ‘no’ as an interactional practice (cf. Heritage 2002b; Heritage & Raymond 2012), which gives Emma an opportunity to respond bodily, for example, by nodding or shaking her head if not verbally/vocally; simple options that she might be able to act on (cf. Antaki, William, Finlay, Walton, & Pate (2008) on offering choices). However, after a long pause of 3.7 seconds, when Ali Reza does not receive any reaction from Emma, he answers the question himself on Emma's behalf: “yeah (0.3) maybe” (line 5). The response, while displaying epistemic uncertainty through the use of the word “maybe”, also underscores the sensitivity of producing action on behalf of a co-present party and the potential implications for their role and identity (see also Nilsson et al. 2018). Furthermore, the epistemic uncertainty which is displayed through the use of “maybe” also attenuates the gravity of asymmetry that arises in the talk on behalf of the person with dementia, and moreover, heightens Emma's agency as a co-participant in interaction. Ali Reza's response concurs with his disengagement from eye contact with Emma as he turns his eyes to the cards.

Image 5. AR looking at EMM waits for any possible reaction from her.

Although Ali Reza orients toward the normative order in the production of the question-answer sequence, it is important to note that Emma is neither expected to produce the answer nor is held accountable for not producing the reply. Ali Reza answers his own question (filling the slot of the second pair part in the production of the question-answer sequence), and after a pause of 0.9 seconds (line 6), he moves on to lay down the picture card (illustrating the picture of an egg) on the table (see Image 6 below) and thus signaling the end of the conversation on this topic and the transition to the next card. While the conventional logic of answer-following-the-question is observed, the logic of Emma producing the answer to Ali Reza's question, followed by Ali Reza's third-position action as the warrant of understanding, does not appear to be effective in this particular interaction (cf. Schegloff 1992, 1996).

Image 6. EMM begins to vocalize when AR is placing the card on the table.

In the continuation of the interaction above, once Ali Reza transitions to initiating a new sequence, Emma expands on the sequence by producing a speech-like chain of sounds (line 7). These sounds are not only recognized by Ali Reza as Emma's contribution to the ongoing interaction but also as conduct that requires a response. As Emma begins to vocalize, Ali Reza turns toward her and acknowledges her contribution with two nods (line 7). Following a brief pause (line 8), Ali Reza responds to Emma with a receipt token of “yeah” (line 9), and subsequently offers an interpretation of her vocalization (line 11)

In the following segment of the interaction above, Ali Reza's interpretation of Emma's contribution (line 11) involves attributing a linguistic sense to her vocalization. However, he expresses uncertainty in his interpretation by using the adverb kanske (‘maybe/perhaps’). By this hedged claim as a candidate understanding Emma's vocalization with inbuilt indeterminacy, Ali Reza provides room for the possibility of incorrect interpretation, or room for possible correction. Once again, this highlights how the conduct of the person with late-stage dementia, although not conforming to the use of conventional semiotic fields, may be understood as the production of a response, and the interpretation of the sense of their conduct is accountably done with respect to their agency (see also Kitwood 1997; Kontos 2012; Boyle 2014; Backhaus 2018; Bottner 2018).

In the continuation of the interaction, Emma, after a short pause (line 12), produces another chain of speech-like sounds (line 13), followed by a deep breath, lip smacking (line 14), a clicking sound, and a short vocalization (line 15). Ali Reza demonstrates active and attentive recipiency to Emma's conduct by maintaining eye contact and bodily orientation toward her (see Image 7). He responds to her vocalization with a confirming receipt of just de (‘exactly/of course’), followed by further interpretation of Emma's vocalization in the remaining part of the sequence (lines 19).

Image 7. AR nods back and responds to EMM's vocalization.

Ali Reza's further interpretation of Emma's vocalization points to the treatment of Emma's conduct as an account-able interactional contribution in the post-expansion of the question-answer sequence. Ali Reza moves from answering the question, “do you like egg?” (line 3) with “yeah maybe” (line 5) to an account of “you like eggs perhaps” (line 11). This time, after a heightened display of understanding through the verbal means of “yeah” and a head shake (line 18), Ali Reza interprets Emma's vocalizations (lines 13–15) as a further explanation in the sequence: “yeah you have probably eaten much ((eggs))”. This is followed by Ali Reza's production of a backchannel “mhum” (line 21). Upon the production of a further vocalization (“a:eya”, which could also be heard as an affirmative “yeah” in line 22), Ali Reza seeks mutual gaze with Emma by lowering his head and continuing looking at Emma who turns her gaze first away and then to the cards in Ali Reza's hand (line 22). At the end, a long pause of around 2 seconds passes before Ali Reza produces a chain of triple “yeahs” (line 24) followed by showing another card to Emma and moving on to a new topic (line 26).

This second example in (5)–(8) illustrates how working out the relevance of the contribution of a person with late-stage dementia may look. First, it requires recognizing it as account-able conduct and also treating it as sensible with respect to the overall structure of the activity with an orientation toward the normative order in ordinary conversation. For instance, a response to a question is expected when the question is left unanswered, as demonstrated in the first example, too. Additionally, it is possible to expect the expansion of the sequence for a further talk on the same topic and within the same interactional project (Schegloff 2007; Levinson 2013). In the post-expansion of the sequence, Emma initiates the action by producing vocalizations (line 7 and lines 13–15), and Ali Reza produces a set of accounts to infer the vocalizations as relevant responses. This demonstrates an orientation toward the participatory role of the person with late-stage dementia as an active agent in interaction. It also shows how their conduct is treated as an account-able action that not only acknowledges their contribution but also heightens their co-participation by attributing an account to their conduct. This act of acknowledging and heightening the co-participatory role of a person who may otherwise be at risk of marginalization or even objectification can also be called partner-positioning (see also Majlesi, Ekström, & Hydén 2022; Hydén et al. 2024).

Working out a next relevant action in interaction with a person with late-stage dementia is also exemplified in the next sequence, in which the same participants as in the second example, Ali Reza and Emma, are engaged in an assisted coffee-drinking activity.

EXAMPLE 3: COFFEE DRINKING

The following example is a sequence where Ali Reza notices that Emma, who is sitting at a table in the sitting room, has her fingers positioned in and around the handle of a coffee cup. This is interpreted by Ali Reza as Emma requiring assistance with drinking her coffee (Emma is too feeble in her hands to pick up and hold objects as heavy as cutlery or cups).

At the beginning of the sequence, Ali Reza offers to help Emma (line 1) after she tries to hold a cup. Emma quietly vocalizes a short response (line 2) which Ali Reza interprets as confirmation to receive assistance. Ali Reza walks toward Emma, sits beside her, and stretches his hand toward the cup, which Emma is still trying to hold (line 3; Image 8). Emma notices the hand movement and withdraws her own hand (lines 3–4). Ali Reza's hand stretched toward the cup is followed by an incomplete verbal instruction (“you can”, line 4) since Emma seems to have already acknowledged the request and left the cup to Ali Reza (line 4). Ali Reza then provides an account for the action being underway by saying “yeah I can help you” and takes hold of the cup (line 5).

Image 8. AR stretches hand to take the cup; EMM withdraws her hand.

After holding the cup, Ali Reza brings it up toward Emma's mouth to assist her in drinking the coffee (Image 9). However, as soon as the cup reaches her chest, Emma raises her right hand and jerks Ali Reza's hand (line 8). She then looks at Ali Reza, holds her hand up in the air, and shakes it once (line 9). Ali Reza immediately interprets Emma's conduct as a rejection of the offer. However, he displays uncertainty by using a try-marked TCU (see Sacks & Schegloff 1979) with a slightly rising intonation in the production of the turn: “no,” (the slight rise of intonation is shown by the comma at the end of the TCU, line 10). Ali Reza also uses a head shake to ask for clarification of the meaning-intent of Emma's hand movement. At the same time, Ali Reza withdraws the cup slightly away from Emma. He produces another try-marked “no,” (line 11), showing that he is still uncertain about Emma's response. The uncertainty in recognizing Emma's action continues in the continuation of the sequence.

Image 9. AR raises the cup to EMM's mouth; EMM seems to have other preferences.

In a long stretch of interaction between Ali Reza and Emma, it appears that they engage in a negotiation of sorts to determine whether or not Emma's conduct—holding her hand in the air in front of the cup—should be understood by Ali Reza as a rejection of his offer to assist her (as previously granted by Emma; see (9)). Although Ali Reza is still holding the cup in the air, albeit slightly away from Emma (line 11), Emma moves her right hand up and down twice in the air (line 12). As Ali Reza cannot figure out what Emma's hand movement might mean, he directly asks Emma if she would like to have coffee (a question that he had not asked from the beginning because he had assumed that Emma left the cup to him to assist her in drinking the coffee). Ali Reza raises the cup toward Emma's mouth as he poses the question, and Emma responds by raising her hand again in front of the cup, touching the cup, and holding her hand in the air in front of the cup (line 14; Image 10a), and quietly saying “aa” (line 15). Ali Reza tries to ask again with the refusal “no”, while Emma also touches the cup a few more times, prompting him to withdraw the cup to the table (line 16; Image 10b).

Image 10a. AR raises the cup again; EMM raises and holds her hand in front of the cup.

Image 10b. AR moves back the cup; EMM draws her hand along her throat.

Emma then moves her hand in front of her chest a few times as Ali Reza leaves the cup on the table (line 17). At this moment, Ali Reza poses the question again: “would you like to have coffee?” and raises the cup slightly while nodding (line 18). Emma opens her mouth and makes a sound (this part is not hearable in the recording, line 19), to which Ali Reza smiles and responds with a sharp rising intonation producing both affirmative and negative proposals, “yeah?” (with a nod initial), “no?” (with a head shake initial) (line 20). Emma, in response, produces a vocalization of speech-like sounds, “me-via-af- (0.5) me-oveghagha” (line 21). Ali Reza interprets this as a possibility that Emma would like to take the cup herself, but with some uncertainty. Emma makes another vocalization, “°°aea°°”, combined with a hand gesture as she raises her right hand to her chest and lowers it (lines 25–27). Ali Reza responds with quick nods (lines 25–26) and a receipt token of “yeah” (line 28). When Emma averts her gaze away from the focused interaction, Ali Reza strokes Emma's arm, and after a short pause, takes his leave.

Overall, Ali Reza's use of polar questions both in the form of a single word, using affirming or rejecting tokens of “yeah” and “no”, and also an interrogative utterance (“would you like coffee?”), and repeated attempts to clarify Emma's actions, despite their indeterminacy, demonstrate how he avoids assuming any epistemic authority over her (Heritage & Raymond 2012) even if he treats Emma's hand movements and vocalization as respondable and orientable as potentially meaningful. By responding to Emma and taking a lower epistemic stance toward what Emma may or may not want, Ali Reza creates space for Emma to assert her own agency and participate in the co-construction of action. By addressing Emma directly and asking what she wants, Ali Reza positions her as an active participant in the interaction (partner-positioning). Negotiations over determinacy of action are clear evidence that actions are not always clear-cut products of one actor's conduct but are co-operatively constructed in the course of interaction (which may or may not be settled among the participants).

CONCLUDING DISCUSSION

In this study, three examples were presented. The first example demonstrates how the absence of response in a greeting (the second part of the adjacency pair) as an expected conduct can be produced on behalf of a non-responding actor to complete the sequence of the action, based on normative understandings of ordinary conversation (Sacks et al. 1974). The second example shows how the vocalization of a person with dementia is interpreted as meaningful conduct by her co-conversationalist in relation to the assessment of an object (whether or not she likes eggs). The third example deals with assisting the person with dementia in drinking coffee, in which the meaning-intent of the embodied behavior, including the vocalization, of the person with dementia is oriented to as indeterminate. Unlike the first and second examples, the presumed relevant conduct by the co-conversationalist in the third example is not observed as warranted by the person with dementia (neither by the interactional partner in the course of interaction, nor by the analyst in the course of analysis). Neither action of offering-granting or offering-rejecting can be achieved under the current circumstances of interaction. Thus, the indeterminate conduct is left without a warrant of understanding or intersubjective action ascription (Deppermann & Haugh 2022:4). Nonetheless, in the third example, the person with dementia, by rejecting the presumed relevant next in the production of action, demonstrates her agency and contributes to the interaction by her embodied conduct and vocalizations (to see the significance of approaching the person with cognitive and communicative difficulties as a social agent, please see Kitwood 1997; Kontos 2012; Boyle 2014; Backhaus 2018; Bottner 2018; Killmer et al. 2020; Majlesi et al. 2022; Hydén et al. 2022, 2024; cf. Goode 1994; Goodwin, 2018).

These examples highlight that the understanding of action may not always be straightforward in interaction (either for participants or analysts). Conduct cannot always be recognized and easily classified into generic categories known as social actions (Levinson 2013; Enfield & Sidnell 2017, 2022). There are certain situations, especially involving people with cognitive and communicative difficulties, where participants’ conduct is indeterminate in terms of meaning-intent. That is, their conduct is not easy for co-participants to recognize in terms of what it points to or is ‘the document of’ (Garfinkel 1967:40; cf. Pilnick et al. 2021). This indeterminacy of action leads to, and is recognized by, uncertainty in producing the next relevant move in interaction, according to normative expectations of ordinary talk in interaction, pertaining to turn-taking systems (Sacks et al. 1974). In such situations, action formation and ascription do not fit into turn designs in the typical sequence organization (Levinson 2013), but they become co-constructed projects (over the stretch of various turns and embodied resources). Furthermore, the speaker's and recipient's contributions as discrete actions may not be easily discerned, and recipient's understanding may not always validate the relevance rule in conversation (Schegloff 2007). The indeterminacy of action is, however, reached after observable and describable attempts to identify or reestablish the constitutive order of the interactional activity and thus co-building intersubjective relations among the participants. Such an intersubjective procedural gestalt allows the person with dementia to remain socially engaged and also positioned as an interactional partner, although the relevant next turn may be left unsettled or is produced under presumed relevance without its warrant (if not contested).

As a result, it can be emphasized that social engagement and interaction are not an aggregate of individuals’ actions, but a collective construct of co-operation and collaboration (see e.g. Lerner 1991), however the outcome may not fit into an identifiable category of social action and remains indeterminate. The procedure of exchanging conduct in interaction under the auspices of ‘recipient design’ (Sacks et al. 1974:727) and the rule of relevancies (Schegloff 2007:20–21), however indeterminate they seem (to the participants as well as the analysts), show how participants per axiom—unless proven to be otherwise—treat one another's contributions account-able and thus presumed to be recognizable and understandable initiations or responses in interaction (see also Goode 1994; Webb 2017; Auer et al. 2020; Pilnick et al. 2021). Treating the conduct of the co-participants potentially meaningful and situatedly respondable allows the agency of contributors, no matter how impaired in their functions or abilities, to be maintained as acknowledged (see also ‘partner-positioning’ in Majlesi et al. 2022).