1 Introduction

The police interview (PI) serves as a context wherein the central activity involves the reconstruction of preceding events, and a significant tool for the police to initiate and engage in this activity is through the use of questions [1, 4, 50]. The distribution of access to prior events under investigation frequently exhibits discrepancies among participants, since victim(s), suspect(s) and witness(es) are likely to have first-hand information to the events under investigation [cf. 26, 33, 46, 47], whereas the police and other institutional representatives often have to rely on concrete evidence for the reconstruction of them. As a result, the police’s understanding of the alleged crime depend upon testimonies provided by witnesses, which in turn make the interview situation a crucial link in a police investigation. Details crucial for the investigation may reside in the interviewee’s memory, making elicitation as well as recollection of memories recurrent and frequent interactional practices in PIs. Asserting an absence of memory or stating other hindrances in recalling preceding events has been reported to be a potential strategy in the interview situation [28], which recurrently has been studied with a primary focus on cognitive aspects [52, 53]. Nevertheless, scarce empirical focus has been directed towards how knowledge of previous events is embedded in talk about memories and remembering, and how this is interactionally managed in high-stake PIs [51]. This is the focus of the current paper.

Significant contributions that explore language use in PIs have emerged from the domain of Forensic Linguistics, which brings together research that focuses on the “investigation and elucidation of language evidence in a legal context” ([6, p. 129], see [7] for an overview). One early study, still frequently referred to, is Loftus & Palmer’s classical study from 1974 [27] in which they show how witnesses reported different speed estimates of a car in a car crash dependent on which verb was used in the question asking about the car’s forward motion. For instance, the verb smash generated a higher speed estimate compared to the verbs collide, bump, contact or hit, in the question ‘About how fast were the cars going when they smashed into each other?’. Over past decades, studies investigating interactional features, employing either a sociolinguistic or conversation analytical framework, have expanded our knowledge on both language use and how social interaction is organized in PIs. For example, research has focused on topic management and turn-taking as tools for pressuring the suspect [29], how suspects’ rights are explained and exercised [32, 51], how questions constitute a linguistic tool for the police in the PI [1, 4, 50], and how the use of guidelines may facilitate or hinder fair questioning and equal treatment [22, 37].

However, studies have primarily been focusing on English speaking contexts (with exceptions, such as e.g., Jol [22] and Pavlenko et al. [32]). As stated by Haworth [17], conditions and results are contingent upon cultural as well as social factors, emphasizing the need for research drawing on data from other social and cultural contexts, such as Swedish high-stake interviews, which still only have been examined in a few studies [3, 23]. To increase our understanding of interactional processes in this specific context, the present paper employs Conversation Analysis (CA) [41] to explore interactional environments in Swedish high-stake police interviews where the interviewee’s ability to recall, as well as the police’s challenge of the interviewee’s claimed incapability to recall is managed in interaction. Within the realm of Conversation Analysis, focus has been directed toward examining how participants handle epistemic dimensions in social interaction, as these have been found to be omnipresent [19, 46]. In conversation, speakers display a sensitivity towards who knows what as well as who is entitled or ought to know what, and their utterances reflect this orientation, as shown in previous research (see Sect. 2 below). This research sheds light on a fundamental aspect of human interaction, at the same time as it contributes to explore how norms are established and renegotiated in different institutions, as patterns of how knowledge is managed vary depending on contextual factors such as participants’ roles and mandate [20]. Increased knowledge of such helps to expand our comprehension of fundamental organisations of our society. Previous conversation analytical research on epistemics form the theoretical foundation for this study, which explores how participants of social interaction handle epistemic elements in the specific environment that the police interview constitute. The overarching objective of this study is to enhance the understanding of interactional procedures utilized in high-stake interviews conducted in Sweden, and in particular, to explore sequential environments in which claims of not recalling are displayed by the interviewee and challenged by the police, as they manage epistemic claims related to the event under investigation.

Data used for this study contains anonymized audio recordings (N = 51) recorded during a preliminary murder investigation (see Sect. 3 below for further description of the data). The investigation is conducted 20 + years after the suspected crime was committed due to new forensic evidence that may link the suspect to the scene of the crime. A majority of the witnesses in the data, as well as the suspect, have been interviewed at different times in the past. Since more than two decades have passed since the event, the time frame is recurrently constructed as an omnipresent factor by both the police and the interviewees in the interviews. The study focuses on how this time frame is incorporated in the police’s questions, and in turn, what interactional space this creates for the interviewees as well as how it becomes a potential resource for the interviewees when constructing an “I don’t remember/recall”–response. An important result from the study is how a shift in focus of attention from the actual events under investigation towards the ability to remember or recall is facilitated by means of the question formulations. To expand our understanding of the described interactional procedure, more studies including data from other cases and contexts are needed, as this study explores one specific case.

2 Access to Previous Events–Dimensions of Epistemics in Social Interaction

In the police interview, the primary institutional goal for the police is to gather information from the person they interview in order to be able to clear up the case under investigation. This means that epistemic dimensions permeate most facets of the police interview, as the police works to identify who knows what and to make them talk about what they know. Depending on the role of the interviewee in relation to the alleged criminal activity, they may align with or diverge from this objective. At times, the interviewee might resist cooperation, possibly owing to their own involvement. In different scenarios, they might intend to provide information; however, factors like trauma, stress, or threats can render this endeavor challenging. In any event, knowledge on the asked about event is central, and in this section, studies targeting how knowledge are managed in interaction are summarized, as they form a backdrop for the reported study.

A prominent study, recurrently referred to in work on epistemics, is Labov & Fanshel’s study [26] on interaction in a psychotherapy session. Their work demonstrates “what is being done” (ibid, p. 27) interactionally in the analysed session, with particular attention paid to how propositions are embedded and launched into interaction. Labov & Fanshel propose a model of how speakers relate to knowledge in terms of events, and these are classified based on the speakers’ knowledge about them, such as A-events: known to A and not to B, B-events: known to B and not to A, AB-events: known to both A and B, and D-events: known to be disputable. Speakers design their utterances, as well as interpret others’, depending on how they perceive speakers’ relation to these events. For instance, Labov & Fanshel suggest that “[i]f A makes a statement about B-events, then it is heard as a request for confirmation” (ibid, p. 100). This rule emphasises how participants are continuously oriented to what type of knowledge their co-speakers (might) have, as a statement may be treated as e.g., either a request for information or a request for confirmation, depending on the perceived access and distribution of speakers’ knowledge. As shown by Cerović [5], B-events can be used as interactional vehicles by the police in police interviews as either confirmation seeking questions, or as accusations, depending on the local interactional environment. If the IE is cooperative, B-events in confirmation seeking question are charged with lower epistemic claims than in sequences labelled as ‘hostile’, where amplified epistemic claims in B-events are treated as accusations.

Pomerantz [33] has also described speakers’ access to events in terms of Type 1 and Type 2 knowables, which distinguish between who has primary and direct access to an event (e.g., based on own experience), or what a speaker has learned second hand. Depending on the degree of access to the event, speakers may be subject to expectations displaying different degrees of obligations and responsibilities. For instance, an eyewitness may be expected to be able to account for an actual event, whereas information provided via a telling constitute a Type 2 knowable, consequently providing reduced degree of obligation to account for the original event. How access to information may be employed for eliciting further information is described by Pomerantz in a later study [35], in which she examines how speakers propose a possible response by offering a candidate answer, embedded in the original question. This is argued to function as an information seeking strategy, and it also reflects epistemic dimensions as it displays the speaker’s own knowledge – and access, of the asked about event. As shown by Pomerantz, this strategy is deployed in interaction recurrently, and as found in Sect. 4.2 below, it is also employed by the police in the PI for information elicitation (cf. [52]).

Stivers, Mondada and Steensig [46] offer a description of the epistemic layers in interaction by proposing three dimensions of knowledge: epistemic access, epistemic primacy and epistemic responsibility. They argue that speakers display orientation towards who knows what (access), as well as the relative rights and authorities to know what (primacy), and lastly who is expected to know what (rights and authorities). Speakers exhibit sensitivity to these dimensions and shape their utterances accordingly, so that information likely to be known by the recipient is not framed as news [16], and claims are reduced if the recipient is expected to have epistemic primacy to certain information [21]. Yet a different terminology is used by Heritage [19], introducing the notion of an epistemic gradient, where speakers are placed based on their degree of access to a certain epistemic content (K). This way, speaker and recipient may be constructed as either K + or K–, and this is displayed in the turn design. Heritage argues that epistemic stance is encoded into the utterance through various linguistic resources, and function to create space for speakers to e.g., elicit information or confirmations, depending on how and what type of relation to certain knowledge is constructed in any given utterance. In contrast to Labov & Fanshel, Heritage states that two speakers rarely ever share the same access to an event (cf. AB-events), as their experience and status always varies.

As this bulk of research shows, speakers monitor and adjust the design of their utterances reflecting their own and their co-participants’ perceived knowledge distribution. However, the majority of examples from the cited research are derived from interactions among individuals acquainted with each other. In contrast, the police interview offers a setting in which much is at stake, and the entire enterprise is about detecting the actual ‘knowable’. In this context, claims of not knowing can be deployed for  serving multiple purposes. As stated by Keevallik [24], a ’no knowledge’-response can be a resource for providing a response in the response slot, while still rejecting the terms set up in the previous turn. In non-institutional settings this can be an item used for resisting the agenda formulated in a first pair part of an adjacency pair. This has been found in research on couples therapy, where claims of not remembering can function as a resistance or avoidance of the therapeutical agenda [31]. However, in an institutional setting where one party has the mandate to strongly control the interactional agenda, it may be more difficult to claim lack of knowledge, which has been found in studies on regulated settings such as courtrooms [8, 25].

Claiming to have no knowledge can also function as an interactive resource when framed as ‘forgetfulness’ [16]. Goodwin describes how this resource provides an opportunity for speakers to rearrange the participation framework, when for instance switching from the discursive identity of listener to a story, to that of a co-producer of the same story. This may also invoke a co-participant in an ‘identity relationship’ (ibid, p. 116) such as a couple, marking shared and common knowledge. This way, the construction of knowing and unknowing recipients provide opportunities for speakers that may not primarily be connected to what is actually known and unknown information. In the analysis, Goodwin also shows how syntactic and intonational resources form pivotal features for shaping different activities, such as requests. For example, a wh-question may construct a recipient as knowing, which in turn may open for the knowing recipient to take a telling in a new direction. This can be pre-empted through the use of a request for confirmation, which instead forms a limited space for the knowing recipients to contribute to. In conclusion, the display of (in)access to knowledge may serve various interactional purposes, as well as creating opportunities for how actions and sequential trajectories are created moment-to-moment in social interaction.

To ‘gather’ or ‘elicit’ information convey the notion of a neutral effort, but in police interviews, there are several factors that come into play and make it a complex one. Cognitive psychologists have predominantly focused on studying the retrieval and management of knowledge stored in memories during police interviews, and although many of the studies target how memories are formulated [e.g., 40, 42, 54, 55], language is often treated as a product rather than a social activity. One such example is how transcripts are often used as the preliminary source of data, instead of the video- or audio recordings, which include dimensions of social behaviours such as gaze, intonation, and prosody [2, p. 1441]. As demonstrated by Goodwin [15, 16] multimodal cues like tracking the gaze of co-participants can clarify instances of self-repair, which in a transcript might resemble a hesitation. On the other hand, studies with discourse analytic foci have revealed that memories, akin to other cognitive processes, must be regarded and comprehended as resources within interaction, shaped by the social as well as the interactional context [10, 30]. Goodman and Walker [14] studied ‘memory talk’ related to instances of alleged intimate partner violence, i.e., how memories related to these events was talked about in interviews with male perpetrators. They reports how the memory talk functions to manage accountability and identity, as claims of not remembering were employed when talking about their own actions, potentially disclosing the exercised violence. A few studies have investigated the use of “I don’t know” in the police interview setting, using transcripts as data. This includes Earhart et al. [9], who studied how ground rules formulated for “I don’t know”-answers are formulated in interviews with allegedly abused children. As children might be inclined to provide responses to adults’ questions, pre-interview instructions regarding this are sometimes used. Their study challenges prior experimental findings, which suggested that children were more inclined to employ “I don’t know” responses when instructions permitted such usage. They also report that children rejected questions with “I don’t know”-answers in almost 30% of the cases, and these rejected questions were often followed by follow-up questionsto which children provided answers.

In conclusion, previous studies highlight that speakers consistently attend to epistemic dimensions, rendering epistemics a fundamental facet of social interaction. Claiming to have no knowledge and/or to not remember have also be shown to operate as interactive resources, serving different purposes for speakers as they may, for instance, (re)configurate the participation framework.

3 Data and Methodology

Data collection was conducted for the project Interactional Patterns in Swedish Police Interviews, which received funding from the Swedish Research Council [2021–03044], spanning from 2022 to 2024. As a precursor to this project, a pilot study involved interviewing eight police officers regarding their roles as interviewers and the training they had received. Findings indicated that training methods varied based on rank and specialization, and it was revealed that not all police interviews are being recorded in Sweden. Typically, the police do not record interviews during investigations of routine crimes; however, more serious offenses like murder or crimes against children are audio- and/or videorecorded. The dataset comprises 51 anonymized audio-recorded interviews involving suspects and witnesses linked to a single case. The dataset underwent an internal confidentiality review by the Police authority and all personal information (such as names, dates, social security number, and addresses) were removed before the interviews were released to the project researchers in a secure digital environment. The police have been offered to take part of the findings.

Police interviews raise ethical concerns due to their potential depiction of traumatic situations and their potential to expose individuals in vulnerable circumstances. However, in order to increase the knowledge on interview techniques within the police profession, real high-stake interviews must be examined. The conversation analytical perspective specifically centers on the patterns of social interaction as significant social achievements, thus shifting the focus away from the participants as individuals. The police and the interviewee jointly shape the interaction, and it is this dynamic interaction—rather than the individuals—that constitutes the analytical focus. How the interaction is shaped and managed is viewed as a reflection of how this type of situation, as well as the particular institution, is “talked into being” [20, p. 290], rather than how specific individuals conduct a specific interview.

The interviews collectively comprise around 27 hours of recorded material, and they include interviews with several different participant categories, such as suspect, witnesses and forensic pathologists. In total, the police interview 37 individuals, and there are four police officers alternating in pairs conducting the interviews. By the time of the interviews, between 20 and 25 years had passed from the crime under investigation. In the transcripts and in the article, 25 years is used as a time reference.

The data was analyzed employing Conversation Analysis (CA), which provides tools for investigating how participants in interaction are socially oriented towards each other and how norms and institutions are managed and maintained through social actions [41]. CA originates from ethnomethodology, which fundamentally seeks to explore how participants achieve intersubjectivity in interaction through social actions. Garfinkel [12] stated that mutual understanding is not just something that happens by itself, it is accomplished through language and other social actions, and it is thus visible, hence, possible to study. Since language is viewed as a tool for socialization and the primary resource for conducting social actions, CA focuses on the fundamental organization of talk. Initially, interaction in mundane settings were the primary target, but during the last decades attention has shifted towards interaction in institutional contexts. For instance, research has focused on explicating the management of interactional practices and examining the constitution of core activities in specific contexts [20]. In the realm of Forensic Linguistics, the use of CA is growing, given its capability to closely analyze how participants establish and negotiate social actions, such as complaining, defending, accusing, explaining. Through close analysis of the micro level of interaction and the participants’ expectations and orientations, it can be revealed how norms are enacted and established, hence uncovering how institutions are ”talked into being” [20, p. 290]. A recurrent theme found in many of these studies is how the participants organize and accomplish questioning and answering. These social actions compose two components of an adjacency pair that form a basic foundation in many interactions conducted in institutional contexts, through which the interaction is shaped and formed. By studying how questions and answers are managed, social and institutional norms become visible, as well as how they form conditions that the participants have to comply with. This does not mean that institutional norms are fixed, rather that they create frames that the participants orient towards, at the same time as they are being renegotiated in and through interaction [20].

The examples featured in this paper have been chosen from a collection of instances extracted from data where the interviewee asserts an inability to remember/recall, and this assertion is questioned by the police. The showcased examples were specifically selected due to their potential to illustrate a recurring characteristic. They serve to demonstrate how this recurring feature is enacted by the participants in this particular case. Further studies that utilize diverse datasets are encouraged in order to comprehensively grasp the phenomenon described.

In the transcripts, ‘P’ refers to the Police officer, ‘S’ to the Suspect and ‘W’ to the Witness. In the analysis, however, ‘IE’ (Interviewee) is predominantly employed when referring to either the suspect or any witness, since the analytic focus is on the participants’ discursive roles and how they mutually manage the interview situation as an interactional project. In the analysis, their institutional roles (suspect, witness) are subordinate to their discursive ones (having to respond to questions, asking questions), but for the sake of transparency the transcripts reflect this information.

3.1 The Case

The case being investigated in the collected data involves a murder, with the alleged incident occurring over two decades prior to the investigation. Due to advancements in forensic technology, the police reopened the case, conducting a new investigation that involved new interviews. Through new analyses of historical evidence and the integration of witness testimonies, the police linked an individual to the crime scene. This person was subsequently acquitted in court.

3.2 Remarks Regarding the Translation

The focus of analysis is instances in which the interviewee states “I don’t recall” or “I don’t remember”. In Swedish, the verbs used are komma ihåg (recall) and minnas (remember). As per to The Contemporary Dictionary of the Swedish Academy, komma ihåg means “being able to retrieve [something] from memory”, and minnas “bringing stored mental concepts to the forefront” (my translations). Similarly to English, these verbs can be used interchangeably. The distinction is akin to the one in English: for instance, you are more inclined to remember rather than recall fond memories from your childhood, while you might both remember and recall where you parked your car in the parking garage. In everyday Swedish, they are interchangeable, and both are employed within the dataset.

4 Analysis

As stated above, the number of years passed since the crime was committed is treated as an aggravating circumstance by both the police and the interviewee throughout the data. In this initial section of analysis, instances of how this is achieved in interaction are provided, demonstrating the construction of time as an omnipresent element within the data, serving as a potential obstacle to recollection. This section establishes a foundation for the subsequent one, wherein instances are examined in which interviewees formulate responses asserting their inability to recall, prompting various forms of challenge from the police.

4.1 Making Time a Relevant Factor When Prompting Recalling

Based on the recorded interactions, it becomes evident that nearly all interviewees have either received a written summons or a telephone call from the police, announcing that they are called for police questioning, which they must attend. Only the suspect and a crucial witness have been apprehended by the police without prior notification before the interview. At the outset of most recordings, the reason to why the interviewee has been summoned is  reiterated for the benefit of the tape [cf. 48], prior to the police commencing the question-and-answer phase of the interview. In this initial phase leading up to the first question, it is common that the interviewing police officer provides a starting point for the interviewee. This establishes the groundwork for the initial question, and frequently, this phase incorporates a mention of the time frame. This is exemplified in the subsequent instances (Examples 1–6):

Example 1

figure a

Example 2

figure b

In Example 1, the police officer produces several statements prior to initiating the first question line 10. In line 5, the police officer introduces the time frame by “it’s been twenty-five years now”. The witness responds promptly with an affirmative “yes”, accompanied by laughter. This is followed by “long time ago, yes”, also produced with laughter in the voice, indicating that this is a fact that is not entirely neutral [cf. 13]. These turns, produced by P and IE, contribute to setting a scene for the events to be discussed. Thus far, the laughter stands as the sole indicator of the time frame being viewed as a potential obstacle. However, the police officer’s second question in line 12 alludes to this perception as they inquire “if you could try to recall what happened that night”. This formulation imbues the question with the presupposition that recollection might present challenges. ‘Try’ implies that an effort is needed, which displays a stance from the police officer that acknowledges potential difficulties with the request. From a conversation analytical perspective, this illustrates how a participant in interaction can formulate a request so that it enables what is referred to as a dispreferred answer [34], reducing the effort it may take to reject the request in the subsequent answer. IE might be unable to provide an informative response, due to difficulties with retrieving the requested information – despite efforts of trying. By employing this question structure, the emphasis shifts to the potential challenge of remembering, thereby placing the recounting of events in a secondary position. This orientation aligns with the established acknowledgement of potential challenges in turns 5–7.

Moving on to Example 2, the initial question addressed to IE is presented in lines 12–13. This question also involves a verb phrase indicating the potential complexities related to the desired action (“could you try”). There is no immediate uptake from IE, instead a lengthy pause of 2.0 s is found in line 14. This causes the police officer to take the turn in line 15 with “I’m aware it’s a long time ago”. This could be interpreted as the police officer treating the silence in line 14 as a sign of the prior request being difficult, hindering IE to launch a response, hence causing the silence in line 14. In contrast to Example 1, the IE is in this example, prompted to (re)present the narrative “as accurately as possible”. Retelling events that lie more than twenty years back in time may be difficult, but to also do it accurately suggests further implications, particularly in this context where the truth is a central, yet often an implicit, factor.

Example 3 exemplifies a comparable procedure to the one identified in Example 2:

Example 3

figure c

Turning to Example 3, the police officer starts by stating that the witness has undergone previous questioning. This directs attention to a specific time period, dating back over two decades. Subsequently, in line 4–5, the police officer asserts, “I do understand if it might be difficult to recall and so on”, resonating the initial question in Example 1. A similar formulation is produced after the pause in line 14, Example 2, potentially due to the police officer interpreting the lack of uptake in the possible speaker exchange slot as a sign of the witness’ difficulty regarding producing a response.

In examples 1 and 3 the time frame is established as a potential challenge for recollection. This positioning occurs within a preliminary sequence [38] that leads up to the question, forming the first component of an adjacency pair. While not delving into the second components of the adjacency pairs, which harbour insights into the interviewee’s interpretation of the initial queries, one can still deduce that the context surrounding the launch of the first question configures the act of recollection as a potentially intricate endeavor, at the same time as it pushes what happened further into the question. Through the design of the question(s), the police officer situates IE in a position of possible limited access to the relevant event, while also acknowledging the plausibility of being unable to recollect. This situates IE in a position categorized as K– [19], where the prospective knowable [33] remains unspecified in terms of being a Type 1 or Type 2 knowable. Examples 1–3 also showcase how the police imbue the questions with diminished responsibility, thereby projecting low epistemic assertions.

As previously mentioned in connection to the description of the data, the police possess evidence linking the suspect to the crime scene. The police can potentially employ this information strategically across various interviews. However, the methods for such strategic deployment, as well as potential inconsistencies regarding e.g., provided details across different interviewee categories, extend beyond the scope of this study. What is evident in this study, however, is how the police set the frame for the interviewees by means of how the  initial question is designed :

Example 4

figure d

In Example 4, the initiation of the interview’s inaugural question is depicted (“could you describe how you knew [the victim]”) in line 1–2. This sequence contains a clarification of what type of interview is being conducted (line 6–7) and a brief instruction of how to respond to the questions (line 8–9). This contrasts  this interview from the previous instances. The initial query is followed by two more questions (line 5 and 11) before P leaves the floor to IE. This opening phase diverges moderately from those presented in Examples 1–3 in terms of question order and the reference to the elapsed years. Nevertheless, the mentioning of the time frame  occurs here, potentially constructing its relevance. Yet, the way in which it might operate as an influential factor remains unarticulated by the police. All three questions are also formulated with ‘could’, implying a limited level of epistemic access and, consequently, a diminished expectation of response capacity [19].

Examples 1–4 illustrate how the police employ the initial phase of the interview to underscore the significance of the considerable time elapsed since the investigated events occurred. Furthermore, numerous initial questions are formulated with phrases like “if you could try”, coupled with reference to the time frame. This construction contributes to cultivating a diminished level of anticipation concerning the interviewee’s capability to furnish responses pertaining to the subject event. This may constitute a tactic employed by the police to build rapport [see e.g., 53] and foster a sense of ease for the interviewee. However, this introduction also serves to situate the interviewee as someone potentially deficient in, or at least possessing limited access to, the event. In essence, the police create a context where the interviewee is invited to adopt a K–position, given that the knowable(s) are positioned behind a temporal barrier—a challenge that the interviewee is prompted to surmount. The interviewee is summoned to the interview with with the anticipation that they are ‘A-eventers’ [26], possessing Type 1-knowable(s) [33]. This implies expectations associated with the interviewee’s role [cf. 56], and the instances demonstrate how, in the interview’s preliminary phase, the police diminish this expectation.

Additionally, the data unveils an apparent anticipation to remember, as depicted in Example 5, where the police officer commences by specifying the years during which the interviewee had been previously questioned:

Example 5

figure e

In line 9, IE’s statement “I don’t recall that”, offered in response to preceding utterances, conveys their perspective on the information presented in prior turns. That the sotto voce  (i.e., silent voice), indicated by the circles in the transcript, might convey a sense of hesitation is confirmed in uptake by P in line 10–11, as P states that “it’s been many years, it’s twenty-five years since it happened”. Thus, the police officer invokes the time frame as an explanatory factor and a potential mitigating circumstance for IE’s stated inability to recall. While the connection is not overtly stated, the sequential arrangement illustrates how “twenty-five years” is regarded as a rationale for the failure to recall, as it is placed directly adjacent to the turn where non-recollection is expressed. Subsequently, a “yes” is offered in line 12, affirming P’s interpretation. The statement “I don’t recall that” serves to establish a K– position concerning an A-event, a stance validated by P as an admissible perspective.

The final example in this section demonstrates IE’s anticipations concerning their own performance, as they evaluate their own performance in relation to what they were able to recollect:

Example 6

figure f

In Example 6, IE clearly demonstrates an anticipation of furnishing information related to an event queried by the police. The possible deficiency in performance is articulated in connection to memory, fashioning the potential limitation as a trait one might possess, rather than, for instance, being characterized as forgetfulness. Furthermore, as highlighted by Speer [43], self-deprecation creates an opening for co-participants to counter the critique; in line 8–9 + 10 the police officer offers a favorable assessment of IE’s performance.

To this point, the examples reveal that both police officers and interviewees are treating the time frame as a probable obstacle for evoking memories. As discussed earlier, this might function to facilitate rapport-building, however, simultaneously, it establishes a frame that diminishes expectations with regard to the pivotal task at hand: recollecting past events. While doing so, the police underscore the significance of recalling as a high-stakes activity central within the context of the interview, while concurrently projecting it being a task afflicted with potential difficulties.

4.2 Treating “I Don’t Recall/Remember” as an Insufficient Response

This section focuses on interactional contexts where the interviewee’s use of “I don’t recall/remember” (hereinafter referred to as IDR) is treated as an inadequate response by the police. Examples illustrate the initiation and framing of IDRs, its subsequent challenge by the police, and the participants’ departure from the sequence. In the first example, the police are inquiring about a specific car and potential access to its keys. The police have evidence tying the suspect to this car in connection with the time of the crime. In Example 7, an interview with a car owner unfolds, starting with the police officer introducing a question about the car keys in line 1:

Example 7

figure g

The turn in line 1 introduces the subject of car keys, and in lines 3–5, P presents what Pomerantz [35] refers to as a candidate answer. In this case, the police officer proposes a potential circumstance that would explain how someone other than the owner of the car was driving it. As argued by Goodwin [16], this creates a defined slot for the recipient, as the question embeds a candidate answer into the question (“if you were the only one who had the car keys or if someone eh there was someone else who had car keys”). In line 6, IE responds in sotto voce “I don’t remember”. Through recycling IE’s IDR-answer in line 7, P creates an opportunity to halt and direct attention to the given answer. While the police officer doesn’t overtly contest IE’s IDR response, the repetition with an interrogative tone renegotiates the IDR as a final response to the preceding question. Instead, the police officer produces a first pair part, projecting a second pair part that forces IE to reside on the topic, indicating that the provided response is treated as dispreferred [34]. This in turn, forces IE to respond again to the initial question, which is done in line 8 as they respond “no” with what can be referred to as a smiley voice, i.e., with voice qualities bordering laughter, such as respiration. This answer is treated in a similar manner, as P repeats this answer as well (line 9). In line 10, IE briefly develops the response by adding “it’s not possible”, which serves as an assessment of the requested recalling. This resonates with Edwards & Potter’s [11] exploration of subject-side and object-side assessments, seemingly interchangeable but serving different interactional functions as the former formulates “a disposition of the speaker toward that object” (ibid, p. 497), where the latter is used more like a description. The IDR in line 7 displays the speaker’s stance towards the proposed facts in lines 3–4, that they cannot recall if this was the case. The “it’s not possible” forms a more objective viewpoint, and it indicates that IE treats P’s repetition as a request for an account. IE does not remember because this is objectively impossible. This is picked up by the police officer in line 11, who confirms that “it’s a long time ago”. This way P affiliates with IE’s stance, invoking the time frame as a motive for not being able to remember.

However, the police does not fully affiliate with the impossible part, as they add “it’s not easy to remember”. Not easy displays an understanding of the difficulties with recalling circumstances related to the keys, but as seen in what follows, it also projects that the subject is not exhausted, and IE’s effort is not considered complete on this matter. Simultaneous to the talk in lines 13 + 15, the police officer displays an image of the car to IE. During this, they shift the attention away from car keys and emphasize the car itself. From what can be gathered from the audio recording, the witness looks at the picture and responds, “yeah that’s my car”, confirming what has already been established earlier in the interview. To this, the police officer responds “that you do remember”, with a word order that stresses the initial object (that), contributing to emphasizing the police’s orientation towards IE’s ability to recall: they may not remember the keys–but at least they remember the car. This produces a new basis in the interaction, as the police continue to ask questions about the car.

This sequence does not demonstrate an explicit challenge of the IDR, however, by repeating IE’s IDR-answer, the police create an opportunity to halt and engage attention to this specific answer. As becomes visible when examined in close detail, the police treat the IDR as a dispreferred response, which is mildly challenged through the recycling and the repetitions that forces the witness to re-confirm the response. As we will see in the next example, the police have not abandoned the issue of the keys.

The following example, collected from the same interview takes place about nine minutes after Example 7 has ended. Here, the police officer has, after some digressions, returned to the car previously discussed:

Example 8

figure h

In line 1–4, P produces an indirect question in the form of a statement, which is a reformulation of the information given by the suspect. The suspect has said that s/he has borrowed the car, “now and then” (line 2) and “on and off” (line 4). IE responds to this in line 5, with an initial and cut off “jag ve- “ (I know), which is altered through a repair to “jag minns inte det” (I don’t remember that). In Swedish, the negation is placed after the verb in normal word order, if the subject is produced before the verb. The verb swap is an interesting detail, which opens for questions regarding how speakers in the data, as well as more generally, make interactional distinctions between know and remember [cf. 49]. In this example, this turn is followed by a lengthy pause of 3.0 seconds, after which IE returns to the previous statement and develops this response. IE initially produces a non-negated phrase (line 7) “yes cause I rememb-” + “maybe”, followed by “I really don’t remember”. In line 8, IE opens for a yes-response to the police’s question: “maybe he has”, before moving out of the turn with “well I don’t remember much of that time”. In the turn produced in lines 7–8, IE hesitates and admits to the possibility of the car being used by the suspect, at the same time as IE states that they do not remember. As the speaker moves through the turn, focus shifts from the lending of the car, which is treated as the ‘knowable’ by the police, to a position where IE states not being able to recall much at all from that time. This way, the ‘knowable’ is pushed further away, and the distance allows the speaker to gain a K– position regarding the factual events. In line 9, the police officer frames the upcoming disaffiliative turn with a minimal response, a common strategy used by speakers as shown by Stivers [45], before producing a turn that challenges IE:s prior statement as it displays noticeable distrust: “but surely you ought to remember that”. This turn projects a high degree of epistemic responsibility back onto IE, treating what is requested as a Type 1-knowable to IE [33]. This is a rich example of when an IDR is not treated as an acceptable response to the question at hand and is explicitly challenged by the police. The ‘ought to’ does not only project a high degree of responsibility, it also projects obligation [19]. For that reason, the police officer borders blaming IE for stating that they do not recall.

The challenge manifested through the IDR results in IE revising a prior response into a different one (line 11 + 13). This is accomplished by an explicit change of state, described by Heritage [18] as when speakers exhibit a change in stance due to e.g., new information. In this case, the change of state comes as a result of the police officer’s challenge, displayed with an initial “aah” in line 11, followed by an alteration (lines 11 + 13) of the previous response. The initial “aah yes now that you mention it” creates an opportunity for IE to reverse the prior stance without performing a repair [39] of a prior turn. Rather, the change of state, which results in a yes-response to the initial question (line 1–4), is framed as being triggered by P’s preceding turn. The affirmative answer is also somewhat vague, as IE admits to the suspect having lent the car once, whereas the question implied that the suspect repeatedly borrowed the car. This way, IE provides an affiliative response, as is projected by the question formulation [36]. After a brief pause in line 14, IE repeats the statement in line 15: “now that you mention it I’m thinking it’s true”. Again, IE ties the change of state to the information provided by the police, and while doing so, they use thinking it’s true, which still contains a degree of uncertainty, as think is a rather neutral activity exposing a low degree of epistemic responsibility, compared to e.g., believe, which is a closely related verb [cf. 19]. P produces minimal responses in line 16, before IE in line 18–19, further mitigates the response: “but I don’t remember like when he used it and the like no”.

In sum, IE initially produces a negative response to the question regarding lending the suspect the car. When this is being challenged by P, IE immediately changes the response, but the information provided is rather vague, as they only admit to lending the car, not narrowing it down to any specific date or time. In this example the police officer’s  challenge did not lead to a hostile interactional environment [cf. 5], but the data contains such, as will be illustrated in the next example.

In the interviews with the suspect, the police recurrently direct attention towards inconsistencies both regarding information provided by the suspect, as well as in relation to evidence collected elsewhere.

In the following example, which is collected from the second interview with the suspect, prior to line 1, the suspect claims to have been in contact with a man, and the police officer questions when this meeting occured.

Example 9

figure i

Just like in Example 8, IE produces a “I don’t know” response (line 3) before stating that s/he does not recall. The turn in lines 3–4 is produced with hesitations, a cut off, a pause and multiple prolonged sounds. P produces a minimal response in proximal position (line 5) after which IE adds that “it’s very hard”, referring to the action of recalling. This accomplishes a shift in focus, as IE targets the difficulties with recalling. The police officer, however, continues with “you have also the-” which is cut off, but projects that they are in possession of further details related to the question that IE is not responding to. It is likely, that “the” in line 7 projects a time or date in Swedish. By claiming to have more facts (also) in addition to the minimal non-affiliative response, P makes clear that they have not exhausted the subject. IE interrupts P in line 8–9, stating that s/he “can’t afterwards fabricate things”. What this turn accomplishes, is that it introduces a new verb and a potentially new activity: fabricate. This is immediately confronted by P with a “no” (in line 10), which functions as a rejection to IE:s allegation of P trying to elicit a fabricated response, i.e., one that is not truthful. (Cf. Example 4 where S is requested to respond briefly yet without leaving any information out.) This allegation put forward by IE, also implies that s/he is not a person that would fabricate an answer, i.e., implying that s/he leaves truthful and honest responses to the police – when this is possible. In turn, this constructs the police as someone potentially forcing the interviewee to lie. In line 11, S adds that “you are asking me to do something that is impossible” [cf. 11], connecting to the prior turn expressing that the recalling is difficult. What IE ultimately says, is that fabricating would be the only possible way to provide something that would be treated as sufficient answer to the question, since the endeavour demanded by the police cannot be accomplished. This leads to the police referring to information provided by IE in an interview two weeks prior, something that is confirmed by IE. Then, IE takes one step back and admits to the claims made by the police (lines 18–19 + 21), but through introducing the concept of details. This echoes the example above with the car keys, which ended with IE admitting to having lent the car, but not having access to details such as date or time.

The police officer continues to produce minimal silent responses. That these demonstrate non-affiliation with the suspects’ responses becomes evident when the police pursue the topic in line 22. Throughout this sequence, the police show that indicating difficulty in recalling is not regarded as a satisfactory answer to the query. However, the suspect succeeds in shifting the focus from the inquiry about their contact with a particular person to the daunting challenge of recalling—an activity presented as impossible.

The final example of this section is extracted from a witness interview, with a witness who may possess knowledge of the suspects’ whereabouts and potential activities on the day of the crime. This example has been slightly altered in order to reveal minimal details of the witness’ whereabouts. The witness’ driving style, referred to in line 1, has been mentioned in a different interview.

Example 10

figure j

In this last example, we initially find the police referring to a different witness' testimony, in which that witness accounts for this witness' driving style, as they have been going in the same car together the day that the crime was committed. The police officer treats this information as evidence for IE having experienced something traumatic that day, related to the crime under investigation. IE claims to have no knowledge of the crime, and in line 5 IE disputes the assertion provided by the police. The police officer reconfirms the information given to them, and states that the description regarding the witness’ driving style was the experience of the other witness’. From this turn, IE choses to reaffirm that this was indeed the other person’s experience, and this is immediately put in contrast to IE’s memory, as IE states not remembering this (in line 8). This functions as a semi-admittance of the other witness’ observation, however, it also challenges the grade of epistemic access. Experiencing someone else’s driving style may on the one hand provide an A-event, and this experience is something that provides epistemic access to an event. However, the actual driving can only be performed by one person, in this case IE, and by refusing to even remembering this activity, IE disqualifies the other witness. As the potential driver, IE is also an ‘A-eventer’, and maybe even more so, as being the one whoactually drove, knows why they drove in a specific fashion. This challenge functions as an acceleration in the discreditation of the other witness, as the irrationality referred to was based on the driving style— and IE does not even admit to remembering driving a car. To this, the police officer responds with multiple no:s in line 11, before IE engages the police officer personally, by saying that P would not remember having driven a car that day either, implicating that P would give the same answer if roles were reversed. This functions to illustrate that the request made by the police officer is an impossible one, and it resonates with previous examples in this section, in which speakers frame their incapability to recall as an impossible task. This receives no uptake from P, who moves on to the next topic.

This section has examined  sequences in which  the interviewee responds with “I don’t recall/remember” (IDR), and the police subsequently challenge this in a third position. Examples have illustrated two distinct trajectories that unfold after the police officer’s challenge(s):

  1. (1)

    The challenge prompts a pause in the interaction, forcing IE to provide confirmations resulting in partial revision of the previous answer (Example 7–8)

  2. (2)

    The challenge is treated as a confrontation requesting an impossible task: remembering or recalling is claimed to be an impossible achievement  (Examples 9–10)

In summary, the police aim to elicit information concerning specific segments within a broader chain of events. In doing so, they direct their focus toward distinct ‘knowables’, treating IE as an individual with not only epistemic access to these elements but also accountability. However, IE dismisses this position, asserting an inability to recall. The police counter this assertion through diverse methods that effectively halts the interaction. This juncture serves as an opportunity for the interviewee to enter a renegotiation of the requested action in the initial question. IE accomplishes a shifts from the primary event–the subject of the question–and reconsiders the nature of the request: is it a reasonable inquiry? By shifting the question towards details from two decades ago, IE effectively constructs the requested knowable as unattainable. Thus, focus pivots to the activity of recalling, rather than engaging in accounts of events in connection to the committed crime.

5 Concluding Remarks

This study presents findings that explore interactional contexts within police interviews where the interviewee offers the response “I don’t recall/remember” (IDR) to a question, which in turn is challenged by the police in the third position. Results show that recalling/remembering are made central activities throughout, as both the police and the interviewee are oriented towards potential difficulties with recalling/remembering due to the time span of twenty-some years between the crime and the interviews. Consistently, the interviewees are prompted to make efforts in recalling/remembering, while the time frame is explicitly presented as a potential impediment to the requested task. This might establish rapport with the interviewee by mitigating the perceived stakes, while concurrently highlighting the substantial nature of this task. In this process, the focus is subtly shifted from the actual events under investigation to the endeavour of attempting to recall/remember. As shown in the introductory analytical section (4.1), interviewees manifest an awareness of the expectations regarding their capability to recall, along with the anticipation that they can provide accounts of the events in question. The analysis also suggests that the interactional work performed by the police officers while downplaying the stakes contributes to projecting a candidate understanding (“I don’t recall/remember”), framing the circumstances as difficult. This contradicts expectations typically ascribed to the interviewee, who is invited with the premise of occupying a knowledgeable position, endowed with access to the knowable aspects—preferably Type 1 knowledge.

Stating challenges with remembering/recalling is observed across the dataset, however, interactional environments in which IDRs are challenged by the police (section 4.2), are found in close connection to what has been described as ‘hostile environments’ by Cerović [5]. In these environments, IE is ascribed with high degree of epistemic access, and the police’s challenge is treated as accusatory. In these instances, the question does not merely serve as a neutral inquiry for information; instead, it implies a specific sequence of events. In these contexts, the questions appear designed to elicit confirmations regarding particular events, resulting in IDR responses from the interviewees. Upon being challenged by the police, the interviewees shift focus away from the event, to the challenges they encounter when attempting to recall the specific incident. Interactionally, the challenge is employed as a device to redirect attention away from the potential knowable that the police aim to elicit towards the seemingly challenging task of recalling. This operation aligns with Goodwin’s description of forgetfulness as an interactive resource [16], as forgetfulness appears to serve as a vehicle for shifting focus, yet facilitating the interviewees’ with a resource for providing responses in the response slot, resonating with Keevallik’s argument [24].

This study contributes to the understanding of fine-grained details of interactions within high-stakes police interviews conducted in Swedish. The study illustrates a connection between question formulation and response in terms of orientation towards the time frame, which may act as an obstacle to memory retrieval. The findings also contribute to the ongoing exploration of epistemic dimensions, shedding light on how they are realized in  high-stakes interactions. However,  the study also generates  new inquiries, including an exploration of how claims of not being able to remember or recall are systematically deployed in different sequential settings within police interviews. This is particularly relevant in hostile environments where, for example, presented evidence may be resisted by the interviewee, or where the police conduct confrontations with interviewees less inclined to provide responses to the police’s questions  [cf.5, 44, 52].