Practicing ground rules in police interviews with child witnesses

Abstract

In police interviews with child witnesses, ground rules like ‘correct me when I say something wrong’ are established. Establishing these ground rules is required by guidelines, with the aim of enhancing the reliability of children's testimonies. In this article, we use conversation analysis to examine how ground rules are practiced in thirty-eight Dutch police interviews with child witnesses. We focus on the police officers’ use of test questions to practice such ground rules. We found that, often, these questions (at first) only consist of an if-clause. Questions with this format leave open whose turn it is and what the appropriate response should be. If-clause questions allow flexibility in the difficulty of the test question, and a subtle pursuit of a response from the child. Yet, they are also treated as problematic by children, shown by silences and hesitations. Surprisingly, the practicing of ground rules sometimes occasions affiliation. (Police interviews, children, testing, practicing, affiliation, conversation analysis, hypothetical questions)*

INTRODUCTION

In this article, we examine how police officers and children practice ground rules using test questions. We focus on their design in terms of topic and syntactic structure, and on their sequential embeddedness in the wider instruction phase, that is, how test questions are initiated, how children respond to these questions, and how police officers then react and move on.¹

In cases of sexual violence against children, evidence often hinges on one main piece of incriminating evidence: the testimony by the child as a witness (Nationaal Rapporteur Mensenhandel en Seksueel Geweld tegen Kinderen 2014). Hence, it is important that this piece of evidence is reliable and usable in court. It has been noted that children are vulnerable witnesses in the sense that ‘their speech performance is adversely affected by the stressful discourse situation they are in, such as being interrogated by the police’ (Olsson & Luchjenbroers 2014:269). In order to enhance the reliability and legal usability of children's testimonies, guidelines and protocols have been developed (e.g. Lamb, Hershkowitz, Orbach, & Esplin 2008; Ministry of Justice in England and Wales 2011; Walker 2013).

One of the recommendations in these guidelines is that police officers provide the child with ground rules for the interview. The first aim of the ground rules is to avoid over-compliance on the part of the child. That is, the ground rules are meant to eradicate the child's potential misapprehension that they need to guess the ‘right’ answers, that the police officer already knows the answers, or that they need to please the police officer (Dekens & van der Sleen 2013:41, 98). Such misapprehensions threaten the reliability of the resulting testimony because they can lead to children being suspected of having answered what they think the police officer wants to hear, rather than what actually happened. Therefore, it is also important for the police officer to establish as equal a relationship as possible with the child in order for the latter to feel free to intervene when necessary, for example, by correcting the police officer²—this is the second aim. The final aim of the ground rules is that the resulting testimony is considered reliable and usable in court.

In the Netherlands, there are five ground rules that have to be explained to child witnesses before the police officer starts the interview proper (Dekens & van der Sleen 2013:73; emphasis added).

• • Permission to correct the police officer's mistakes in summaries and the permission to complete summaries

• • Permission to say that the witness does not know the answer to a question

• • Permission to ask for clarification when the police officer asks an unclear question or uses difficult words

• • Explanation that the police officer was not there, and does not know the answers to the questions

• • Explanation that when the police officer repeats a question, he [sic] does not do so because the witness has said something wrong, but because the police officer has not remembered or understood the answer. The witness should just give an answer again in that case.

These ground rules are referred to as ‘the instructions’ by Dutch police officers and in the guidebook. This article focuses on the first three instructions, that is, the correction-instruction, the I don't know-instruction, and the I don't understand-instruction. These are the instructions that also have to be practiced with the child according to the Dutch Manual for interviewing vulnerable witnesses, hereafter ‘the Manual’ (Dekens & van der Sleen 2013:37–39). For example, police officers practice the I don't know-instruction by asking the child: “What is your domicile” (see also excerpts (8), (9) and (10)). The Manual mentions two purposes for practicing the instructions. First, it enables the police officer to check whether the child has understood the instruction: ‘The police officer practices [the instructions] with examples so he [sic] can check whether the witness has understood the instruction’ (2013:73). Hence, practicing is described as a test: it is up to the child to demonstrate that they know what to do. When it turns out that the child does not understand the instruction, police officers need to decide if the gain from trying a different explanation or practicing again outweighs the loss of time and limited energy of children (p.c., Jol 2019). Sometimes police officers take a ‘failed’ test as a sign that they should be extra careful not to ask steering questions (p.c., Jol 2019).

The second reason for practicing the instructions is training the child: ‘When children are trained at the beginning of the interview to say “I don't know”, when they don't know the answer to a question, this leads to less incorrect information in their testimony’ (Dekens & van der Sleen 2013:39). The idea is that responding “I don't know”—and probably by extension also “I don't understand” and “that's not correct”—at the beginning of the interview teaches the children to use such responses during the interview-proper (2013:39, 73).

Yet, research has shown that it is questionable whether training or practicing communication is similar enough to real-life communicative events (Stokoe 2013). Moreover, ‘translating’ advice (such as from manuals and protocols) into practice is not straightforward. Attempts to implement advice can have unexpected, and sometimes even undesirable, consequences (e.g. Peräkylä & Vehviläinen 2003; Childs & Walsh 2018; Sikveland, Kevoe-Feldman, & Stokoe 2020; Jol & Stommel 2021). This also applies to the context of police interviews with vulnerable witnesses (see also Richardson, Stokoe, & Antaki 2019). For example, the Achieving Best Evidence guideline in England and Wales encourages police officers to ask vulnerable witnesses whether they have other issues they would like to raise (Childs & Walsh 2018). This is regularly translated into questions like “is there anything else you think I need to know at this time” (Childs & Walsh 2018:364). Although such questions aim to invite the child to bring up additional issues according to the guidelines, they are hearable as starting to close the interview, especially when the police officer has announced being done questioning or is closing their notebook. Children indeed interpret the question as a pre-closing by indicating that they have no further points to discuss (Childs & Walsh 2018). Hence, ‘translations’ of guidelines can sometimes achieve the opposite of the desired effect. Therefore, it is relevant to study how ground rules or instructions are practiced, what practicing establishes in the ongoing interaction, and how it relates to the guidelines. We take a conversation analytical (CA) approach to study how ground rules are practiced, including the sequential embedding and the design of test questions. These findings are then used to reflect on the guidelines.

PRACTICING COMMUNICATION

The practicing of the ground rules resembles role-play communication training. It is often assumed that role play is a good enough approximation of the actual interactional situation (Stokoe 2013). However, this assumption is at least questionable. Stokoe (2013) compared role-played police-suspect interviews with actual police-suspect interviews. She examined parts of the interviews that are regulated by law and that, given the strict regulations, should be similar in real life and the role-play situation. Indeed, Stokoe (2013:183) found similarities in the opening sequences and police officers did not ‘step outside the simulation’. However, she also identified systematic differences. For example, attempts to build rapport (e.g. invitations to use the police officer's first name) were conducted in a more exaggerated way in the role-played interviews. Differences between actual and role-played conversations were even more pronounced in role-played job interviews studied by Linell & Persson-Thunqvist (2003). The authors found that the participants in these training settings often moved from the role-played job interview to advice giving and educating, and vice versa. It was often ambiguous if participants were orienting to the role-played job interview or to the educational setting (Linell & Persson-Thunqvist 2003).

These studies show that role play is an interactional setting in its own right and that it is unlikely that role-played encounters truly resemble the real-life situations they supposedly enact. These studies hence raise the question whether role play is an effective communication training method (Stokoe 2013).

Our analysis centers on particular linguistic and interactional aspects of practicing instructions including aspects of hypothetical questions, test questions, and (designedly) incomplete utterances. In the next sections, we briefly discuss previous research on each of these question types.

HYPOTHETICAL QUESTIONS

Hypothetical questions (henceforth HQs) are questions like: ‘If that sort of thing should occur what do you do then?’ (Speer 2010). They may include three components (Speer & Parsons 2006):

1. (i) an invitation to imagine something

2. (ii) a description of the hypothetical scenario

3. (iii) a question component which asks what the addressee of the HQ would do if the hypothetical scenario would occur (Speer & Parsons 2006:792) or what the hypothetical situation would mean for the addressee (Peräkylä 1993; Noordegraaf, Van Nijnatten, & Elbers 2008)

HQs regularly take the shape of conditional questions, with an if-clause and a then-clause. The if-clause provides a hypothetical scenario X, and often contains markers of hypotheticalness like ‘if’, ‘say’, ‘imagine’, ‘suppose’, ‘assuming’, and so on. The then-clause enquires what would happen if that hypothetical situation occurred, what it means to the addressee (Peräkylä 1993), or how the addressee would cope with the hypothetical situation (Noordegraaf et al. 2008). We would like to comment that ‘questions about hypothetical scenarios’ (see Komter 1987) is a more accurate term, because the questions are usually actual questions and only the scenario is hypothetical. Yet, given the established nature of the term ‘hypothetical questions’ in previous studies (Peräkylä 1993, 1995; Speer & Parsons 2006; Speer 2010, 2012), we use this term when referring to these studies.

HQs have been found in various types of institutional interaction (Komter 1987, 1991; Peräkylä 1993, 1995; Speer & Parsons 2006; Noordegraaf et al. 2008; Speer 2010, 2012). They are used for testing and helping (Noordegraaf et al. 2008), tailored to the specific contexts in which they occur (Speer 2012). In AIDS-counseling, for example, it was found that HQs are therapeutic, or ‘helping’ in nature. The questions are used to first invoke a ‘hostile’ possible future and then invite the person with HIV to consider how they would react (Peräkylä 1993). Counselors attend to the delicate nature by only posing such questions when the patient has already hinted at that problematic future and counselors explicitly do not commit to the accuracy of the hypothetical scenario (Peräkylä 1993).

In contrast, HQs in psychiatric assessments with transsexual patients are more challenging and testing (Speer 2010). In this context, psychiatrists use HQs to test whether the transsexual person is resistant towards pressure to change their stance towards treatment (Speer 2010:153). Similarly, HQs in family assessments of prospective adoption parents challenge and test prospective parents’ pedagogical abilities, psychological capabilities, and sense of reality by presenting them with (usually) increasingly difficult situations and asking how the prospective adoptive parents would respond. At the same time, responses to these questions are regularly treated as an opportunity to provide advice on how to deal with such a situation (Noordegraaf et al. 2008). In job interviews, HQs invite the interviewee to demonstrate, rather than just claim, that they have the skills needed for the job (Komter 1987).

In sum, HQs have been found to regularly test views and capabilities in various settings and various ways, but they also play a part, for example, in providing therapeutic support or in helping prospective adoptive parents to prepare for the adoption. In this article, we add to previous research on HQs by examining a related format that to the best of our knowledge, has not yet been described, namely questions that (at first) only consist of an if-clause. The next section discusses another type of question relevant to learning and practicing, namely test questions or known information questions.

TEST QUESTIONS

‘Test questions’ (Searle 1969) or ‘known information questions’ (Mehan 1979b) do not genuinely seek information. Instead, they elicit a particular answer already known by the questioner, often for teaching purposes. The following examples (Mehan 1979b:285) illustrate how the difference can become clear in sequences:

1. (1)

   

In example (1a), A treats B's answer as something that helps A, and therefore as information that A did not have access to. In (1b), A evaluates B's answer, which implies that A did have access to the time. The uptake by A thus retrospectively displays that the question was eliciting information as a test, rather than informing A. This type of sequence is also known as initiation response evaluation sequence (Mehan 1979a) and has been thoroughly studied in CA and other research disciplines, especially in educational settings (for overviews see Lee 2007; Gardner 2013).

Inviting known information is not always done by questions that are grammatically formatted as interrogatives. Another way to elicit responses from students is using designedly incomplete utterances or DIUs (Koshik 2002; see also Lerner 1991). Koshik (2002) studied DIUs in the context of English language classes. In this context, teachers regularly repeated what students wrote down or said. Yet, they stopped their utterance just before grammatical or other mistakes, thus indicating where the mistake was and inviting the student to complete the teacher's turn and correct the mistake. One of the functions of DIUs, hence, is to provide hints as to what needs correction (Koshik 2002).

Our analysis shows that, in practicing instructions in interviews with child witnesses, police officers’ combine aspects of hypothetical questions, test questions, and DIUs.

DATA AND METHOD

Our dataset consists of video recordings of thirty-eight interviews from two child friendly interview rooms in the Netherlands. The data were pre-recorded by the police in 2011 and 2012 for the criminal investigation. The recordings were obtained with permission of the Public Prosecutor's office. The research proposal has been approved by the faculty's Ethical Review Committee (for a full account of ethical aspects, see Jol & Stommel 2016a). All information in the transcripts that could lead to identification has been replaced, and all publications were checked by the Public Prosecutor's office before publication in order to avoid revealing the identity of the participants.

The youngest child was six and the oldest eleven; there were fourteen boys and twenty-four girls. All of the children were interviewed in Dutch. They are generally victims of alleged sexual violence but are interviewed as witnesses in order to provide evidence for the criminal procedure. The police officers interviewing were female in thirty-two interviews, and male in six interviews. Interviewers were trained and had to pass an exam to interview child witnesses. Interviewers who have not (yet) passed their exam are only allowed to interview children under supervision of an instructor of the police academy.

Interviews usually consisted of roughly the phases indicated in Table 1 (although variations occur). Instructions and practicing generally occurred before the questioning phase.

Table 1. Phases in police interviews with child witnesses.

In line with the methodological underpinnings of CA methods, we became interested in the ways in which instructions are practiced during explorations of the data. We formed a collection of all practicing sequences and transcribed them according to the transcription conventions developed by Jefferson (2004). When relevant, the transcripts were complemented with transcription of embodied behavior, using the conventions developed by Mondada (2018; see an overview of relevant conventions in the appendix) with some adaptations for technical reasons (see n. 4). The transcripts in this article are also complemented with English translations made by the authors that are kept as literal as possible while maintaining the flow of the talk. The attempt to resemble the Dutch wording as much as possible at times resulted in translating awkward Dutch phrases into awkward English phrases. We attempted to maintain as much interactional detail as possible in the translations. Occasionally, an extra line of translation has been added to allow for a more literal translation and a translation that reflects the tone of the utterance.

Our analytical approach was conversation analysis (Sidnell & Stivers 2013), including analysis of sequential organization, turn design, turn taking, and so forth. We specifically examined the potential relationship between the manual and actual interaction (cf. Peräkylä & Vehviläinen 2003). In the following, we present the most salient interactional aspects of practicing the instructions as a basis for a reflection on the relation between instructions in the manual and actual interaction.

ANALYSIS

First, we examine the turn design of test questions with which police officers initiate the interview section in which an instruction has to be practiced. This provides a clear view on the multi-layered complexity of these questions. Second, we show how practicing instructions is embedded in the interaction. This provides interactional context that will help to understand the subsequent analysis. Third, we show how children respond to police officers’ initiations of practicing, finding that they treat the initiations as problematic. In the last analytical section, we show how practicing occasions laughter and accounts.

DESIGN FEATURES OF TEST QUESTIONS THAT INITIATE PRACTICING

In this section, we examine some recurring design features of the turns that initiate the practicing. Examples (2)–(4) provide samples of each type of instruction. Speaker indication P refers to the police officer; C refers to the child. Before each example, the name (a pseudonym) and age of the child is indicated.

1. (2) I don't know-instruction test questions

   

2. (3) I don't understand -instruction test questions

   

3. (4) Correction-instruction test questions

   

The examples above have some salient features marking practicing as ‘not the actual interview’ in various ways. First, they are constructed as hypothetical with ‘if’ and with meta-linguistic verbs such as vraag ‘ask’ (examples (2) and (3)) and zeg ‘say’ (example (4)) (Schiffrin 1980), and with the Dutch quotative marker van, translated as ‘like’ (examples (2b) and (4b)) (Mazeland 2006). Some of the examples are also explicitly introduced as not being ‘real’ by announcing ‘practicing’ (example (2a)) or framing the question as ‘examples’ (examples (2b), (3a), (4b)). At the same time, words that typically occur in hypothetical questions, like ‘imagine’, ‘suppose’, ‘assume’, ‘say’, and ‘should occur’ are absent.

Second, the topics used to test children (henceforth ‘test items’) are striking. In case of the I don't know-instruction, police officers enquire about something in their own knowledge domain (e.g. the police officer's bunny's name) that children are unable to know to elicit a no-knowledge claim. This stands in contrast with the interview-proper which mainly concerns children's knowledge domains about what happened, rather than police officers’ knowledge domains (cf. Sacks 1984).³ In case of the I don't understand-instruction, police officers use a jargon word, ‘domicile’, to train the child to say that they do not understand the police officer. In social interaction, speakers orient to the specific recipient in the way they formulate their turns (recipient design; Sacks 1995). Using a jargon word like ‘domicile’ is recipient designed in a very specific way: it makes the question recognizable as a test question, since it is strongly steering towards a non-understanding response. However, such a question would be highly marked (and unlikely) in the interview-proper. In fact, both the test items for the I don't understand and the I don't know-instruction emphasize that there are knowledge asymmetries between the police officer and the child: there are some things the police officer knows that the child does not. Finally, the test item for the correction-instruction is marked because the correctables are highlighted in various ways. In example (4a), the color lilac is emphasized; in (4b), the police officer projects the error by slowing down. In other interviews, police officers use resources like pausing, hesitation markers before the correctables, or they highlight the error with intonation or volume. In the section Laughter in practicing sequences below, we provide possible sequential evidence for the markedness of the test items.

The third and final salient feature of the test questions is their syntactic structure. Practicing is generally initiated with a test question with a hypothetical format, including ‘if’ (see all examples), some kind of introduction (e.g. ‘I say to you’), and the test item (e.g. ‘you were with a friend’). Only in some cases (e.g. example (3a)) does the police officer produce a ‘full’ if-then-question including the question component (the then-clause) (Speer & Parsons 2006). In most interviews, the police officer only provides the if-clause and leaves it open to the child to respond to the test item without an explicit question component (cf. Koshik 2002). The possible relevance of a response is created by a combination of making explicit that practicing (example (2a)) or an example is forthcoming (examples (2a), (3a), and (4b)), by the silence after the if-clause and sometimes also by the use of intonation. Hence, the structure of the test question that initiates practicing can be summarized as follows.

• • if-clause:

  • *introduction

  • *test item

• • (sometimes: then-clause)

Henceforth, we refer to the format of test questions that does not include the then-clause as if-without-then-questions, or IWT-questions. The difference between IWT-questions and ‘full’ if-then-questions only becomes disambiguated in retrospect. Whereas designedly incomplete utterances (Koshik 2002) make clear that it is someone else's turn afterwards, the IWT-question is hearable as either a complete or incomplete test question. Only afterwards, we can see if the child answered the if-clause (an IWT-question occurs), or if the police officer expands the if-clause with a then-clause (a full if-then-question occurs). We further discuss this in the Children's problematic responses below.

So, the initiation of practicing instructions in police interviews with child witnesses is characterized by marking the activity practicing as not the real interview, by challenging children to provide answers they are unable to give and by test questions that often take the format of an IWT-question that ambiguously allocates the child to take a turn. In the section Children's problematic responses, we discuss observable interactional difficulties connected to the ways the instructions are practiced. First, we address the question of how practicing is embedded in the interview.

SEQUENTIAL EMBEDDING OF PRACTICING

Generally, police officers initiate practicing just after providing the instruction. This default order of initiating practicing is illustrated in excerpt (5), a more elaborate version of example (4b). The police officer just announced that she is going to repeat the child's story (data not shown); in lines 1–2, the police officer (P) instructs the child (C).

1. (5) Suzanna, 7; correction-instruction

   

This excerpt shows that the instruction sequence and practicing sequence are tightly connected. The police officer first provides the correction-instruction with an if-clause that indicates when the instruction applies (line 1) and a then-clause that tells the child what to do (line 2). The child confirms her understanding by nodding (line 3). Right after this instruction sequence, the police officer initiates the practicing sequence by asking a test question, specifically, an IWT-question. The child provides a response (lines 8–11), which the police officer treats as correct (line 12). Additionally, the police officer concludes the practicing sequence (line 13), highlighting the educational aim of the activity. Overall, the sequence is structured as follows (optional components are placed between brackets).

Instruction and practicing sequence

1. (i) Explanation of the instruction

2. (ii)  (Child's confirmation)

3. (iii)  Test question

4. (iv)  Child's response

5. (v)  (Uptake of child's response, e.g. confirmation [excerpts (6), (7), (10), (11), (12)], evaluation [excerpts (8), (12)], answer-repetition [excerpt (8)])

6. (vi)  (Conclusion of the instruction, e.g. explanation [excerpt (7)], reiterating the instruction [excerpts (8), (9)], announcements [excerpt (10)])

The connection between instruction and practicing is not just a matter of the order in which they take place. The police officer retrospectively presents the instruction (lines 1–3) as leading to practicing by using want ‘because’ (line 4; see also examples (2a), (3b), and (4a)). Consequently, the instruction is presented as not sufficient: the police officer also requires a demonstration of understanding from the child. At the same time, this tight connection between the instruction sequence and practicing sequence ensures that the police officer makes available the targeted response or response type in the instruction sequence just before requesting that response in the practicing sequence.

In some cases, however, the police officer departs from the default order, by inserting the test question instead of the then-clause of the instruction. The police officer has just announced that she is going to ask questions about the boy's story.

1. (6) Ben, 6; I don't know-instruction

   

The police officer starts the I don't know-instruction with the if-clause (lines 1–5) but abandons the if-then-construction in line 6. Instead of telling the child what to do (‘just tell me that you don't know’), she inserts a test item for the I don't know-instruction, asking something about the police officer's knowledge domain (lines 7–9). The police officer thus prioritizes eliciting the ‘what to do’ or then-clause from the child over providing it herself. As a consequence, the police officer has not yet made the targeted response type (i.e. ‘I don't know’) available.

As in excerpt (5), the test question starts with the format of an IWT-question. After a 1.8 second silence in line 10, the police officer explicitly hands over the floor to Ben by eliciting a response (line 11), transforming the IWT-question into a full if-then-question. After a 0.7 second silence, Ben indeed answers ‘I don't know’ (line 13). He starts with ‘yeah’, treating the question as also making a point. The police officer confirms with ‘no’ (line 14) implying this was the correct answer. This retrospectively constructs the if-then-question as a test question. The strongly rising intonation of ‘no?’ is hearably enthusiastic. In lines 15–21, the police officer formulates the upshot of the sequence, instructing the child to indicate when he cannot answer a question. In this sequential order, the instruction is the conclusion of an exchange between police officer and child, as the collaborative ‘practicing’ sequence was preparatory to it.

In sum, practicing is usually tightly connected to instruction. Initiating practicing treats the instruction (including children's confirmations of understanding) in itself as insufficient. The instruction sequence usually (but not always) makes available the response type targeted in the practicing sequence. This could make practicing an easy exercise, but in the next section, as we shift focus to the children's responses, we find quite the opposite.

CHILDREN'S PROBLEMATIC RESPONSES

In this section we focus on how children respond to test questions. Excerpt (7) is taken from an interview with Siri (a more elaborate version of example (2a)) and shows how the I don't know-instruction is practiced.

1. (7) Siri, 10/11; I don't know-instruction

   

In lines 1–2, the police officer introduces the test question by announcing that they are going to practice and launches the test question for the I don't know-instruction in line 3. There is a 1.9 seconds silence after this test question (line 4). The child gives the right answer, which is produced softly and with a smile (line 5). The police officer also treats it as a correct answer (lines 7–10). The police officer's uptake and explanation (lines 8–10) retrospectively explains the test question as a question the child could not possibly answer, thus treating the test question as responsible for the I don't know-answer.

The excerpt also provides an example of an IWT-question: the test question does not include a question component. Hence, it is ambiguous whose turn it is and who is going to provide the then-clause in this if-then-construction. The following 1.9 second silence may be related to this ambiguity.

Moreover, we observe that the absence of a question component leaves implicit what question the child is supposed to answer, and that the child needs to decipher what type of response is expected. This makes the IWT-question a difficult test question for the child. From the police officer's perspective, this type of question enables the police officer to assess whether the child has understood the instruction well enough to interpret what the question should be, and whether they are able to provide the targeted answer in the interactional situation of practicing.

The IWT-question also provides the police officer with flexibility, because the IWT-question can be complemented with a then-clause when a response does not follow, thus constructing a full if-then-question. This is illustrated in excerpt (8), which shows how a police officer and six-year-old Lisa practice the I don't understand-instruction. The police officer instructs the child what to do (lines 1–3) and enacts how to say that (lines 5–6). The child confirms understanding of this instruction (line 8), upon which the police officer initiates practicing in line 9.

1. (8) Lisa, 6; I don't understand-instruction

   

The police officer's initiation of practicing with an IWT-question leads to silence (line 11). The child is gazing at the police officer, taking a listener's role to a continuation of the police officer's turn.

After a silence of one second, the police officer expands her turn with an explicit enquiry (line 12). This move does delicate work. It resolves the ambiguity with regard to whose turn it is by explicitly asking Lisa for a response (line 12) and thus orients to the silence as caused by this ambiguity. It is designed as the continuation of the if-then-construction and thus retrospectively constructs the silence in line 11 as a turn-internal pause. This minimizes the child's responsibility for not yet responding since the if-clause is retrospectively constructed as an incomplete turn. Moreover, the police officer's question (line 12) makes clear that the child should say something (rather than, for example, do something). The formulation ‘What do you say then’ (line 12) suggests in Dutch that the police officer is targeting a particular answer, while avoiding clues as to what exactly the child should say. This can be understood as maneuvering within institutional constraints: the police officer pursues the child's participation in practicing while avoiding steering her towards a particular answer.

Although the complete if-then-question requires less interpretative work and the child provides the correct answer as evidenced by the police officers approving uptake (lines 15–16), the child produces her answer as hesitant by delaying (line 13), reducing pace and stretching her answer (line 14). She thus treats the question as difficult.

The excerpt thus shows that, in case of silence, the police officer may adapt the IWT-question by adding a question component, thus reducing the required interpretative work from the child. Simultaneously, this allows police officers to pursue a response without emphatically doing so. However, that less demanding version is still regularly treated as difficult.

Excerpt (9) provides another example of a child's response that displays difficulty.

1. (9) Cornelis, 6; I don't understand-instruction

   

The police officer asks the IWT-question in lines 3–4. The child delays his answer with 0.9 seconds (line 5) and “(h)m” (line 6), and then provides a disfluent I don't know-response rather than the targeted I don't understand-response (line 6). This response attributes the lack of an answer to his own lack of knowledge rather than to the unintelligibility of the police officer's question. The child thus avoids holding the police officer responsible for the unanswerable question, in line with the general preference for self-initiated repair over other-initiated repair (Schegloff, Jefferson, & Sacks 1977). He does so even just after the instruction has been given (not shown). Excerpt (9) thus indicates that the test question may put the child in a position in which he has to provide a response that would be dispreferred in many other contexts. The excerpt shows that just providing instructions does not necessarily remove normal interactional preferences with regard to repair. Similarly, ‘I don't know’ is a dispreferred response in other contexts, in the sense that it does not move forward the activity at hand, and accounts for not providing an answer. Finally, corrections, or other initiated (other) repair, are also less preferred than self-initiated repair (Schegloff et al. 1977).

So, IWT-questions in the context of practicing instructions confront children with an interactional challenge. Interactional factors that contribute to this interactional challenge are the ambiguity as to whose turn it is, caused by the IWT-test question, but also the usually face-threatening and dispreferred actions that are elicited. This challenge is observable in hesitations, as well as in the avoidance of asking for clarification (excerpt (9)), or correcting the police officer and instead just indicating that there is a (certain type of) mistake (see excerpt (11) for an example). In the next section we zoom in on how children and police officers may also orient to the initiation of practicing as laughable.

LAUGHTER IN PRACTICING SEQUENCES

In this section we consider how practicing with the test questions described above may not only elicit a smile from the child as we noted in our analysis of excerpt (7) (see also excerpt (9), line 6), it may also occasion affiliation (cf. Jefferson, Sacks, & Schegloff 1987). This is important because this implies that the participants collaboratively treat the test question as marked. Apparently, shared stance taking is a way to deal with some of the challenges identified above, and arguably build rapport (cf. Stokoe & Edwards 2008). Consequently, the difficulties identified so far could be somewhat nuanced. Therefore, we explore the laughter further in this section.

An instance of a child's response including laughter that is reciprocated by the police officer can be found in excerpt (10).

1. (10) Luca, 6; I don't understand-instruction

   

In this case the police officer resolves the ambiguity as to whose turn it is after a 0.3 second silence by adding the question component (line 5) after the hypothetical scenario (lines 1–3). The child provides a response that is treated as correct by the police officer (lines 8, 9). Additionally, the child treats the test question as laughable with laughter that bubbles through the turn (line 7) (Jefferson 1985). The police officer reciprocates the child's laughter (line 9), thus aligning with the child's stance towards the initiation of practicing as laughable. Hence, initiating practicing appears to be an opportunity to accomplish shared stance taking or affiliation.

In other instances, police officers orient to the accountability of the test question in their uptake of the child's response in an unequivocal manner.

1. (11) Cecile, 10/11; correction-instruction

   

The child's smiley voice and bubbling through laughter in line 7 is treating some aspect of the test question as laughable. The police officer's uptake (line 8) is ambiguously hearable as enthusiastically affirming or aligning with this implicit evaluation. The police officer shows that she actually did know the correct name in line 14, thus emphasizing the accountable nature of the ‘mistake’ (line 3) and treating the child's laughter as legitimate. The child aligns by confirming with a smiley voice (line 15). Hence, the participants treat the test questions as accountable and laughable across a series of turns.

The most explicit orientation to the accountable nature of the test question is in the following example that is unique in the data set.

1. (12) Jan, 6; I don't know-instruction

   

The police officer initiates practicing by asking an IWT-question (lines 1–5). Yet, before the actual test item, the police officer provides a disclaimer in line 3: he is not going to ask this in the interview-proper. He thus explicitly orients to the test item as accountable. After a 2.3 second silence (line 6) the child gives the targeted answer while smiling (line 7). In the context of the account in line 3, this constructs the answer as sharing the same stance towards the test item, namely as something accountable to ask, or at least laughable.

Hence, the test items have the potential to invite shared stance taking or affiliation, by merit of their accountability (cf. Stokoe & Edwards 2008). Children repeatedly treat the initiations of practicing as laughable, and police officers’ reciprocal laughter and accounts (embedded in the test question or in the uptake of the children's response) align with the children's stance, thus creating affiliation. Hence, police officers treat the initiation of practicing and practicing itself as something laughable, and sometimes accountable, but inevitable (in this institutional setting). These moments of shared stance taking thus emphasize that the participants are interdependent and that the police interview is a collaborative endeavor (cf. Fogarty 2010).

CONCLUSION AND DISCUSSION

The activities of instructing children and practicing instructions aim at securing the reliability of the testimonies elicited by police officers. We analyzed how police officers design and embed the practicing activity, as well as how children respond to it. Our analysis shows that practicing as it is conducted currently creates a difficult interactional situation, in terms of turn taking as well as in terms of the difficulty and dispreferred nature of the actions expected from children. From a police perspective, the difficult nature of saying ‘I don't know’ and so on, is presumably precisely why these types of responses need to be practiced. Yet, the analysis suggests that giving the instructions does not remove everyday interactional preferences straightaway (cf. van der Houwen & Jol 2016).

The analysis adds to previous research on hypothetical questions, namely by identifying a format that relates to HQs: the IWT-question. This underlines Speer's (2012) claim that HQs have different functions and design features to suit the specific interactional demands of the setting they are used in. This specific IWT-question format is useful in this particular setting because it is easily adaptable and expandable into an easier test question that seeks the child's response without giving cues as to the required answer. This allows the child to apply the instruction with as little scaffolding as possible. This is relevant for the overhearing audience of lawyers and judges because it arguably implicates that the child has understood the instruction, which adds to the reliability of the testimony. When the police officer converts the IWT-question into a full if-then-question, this is at the expense of some of this implication. The IWT-questions thus take the opposite direction of the HQs studied in the context of adoption interviews that can go from easy to increasingly difficult (Noordegraaf et al. 2008). Moreover, the format allows the police officer to pursue a response in case it is not forthcoming straightaway, whilst not overtly treating the child as having failed the test. IWT-questions that are turned into full if-then-questions thus contribute to a safer interactional environment.

From the perspective of the child, however, IWT-questions require much interpretative work on three levels: whose turn is it, what is the question, and what to respond? We have seen that children regularly delay their responses, which also include other characteristics of dispreferred turn design such as disfluency.

Another point of interest is the (un)representativeness of the observed way police officers practice the test questions. First, the test questions work to invite the child to reproduce an answer type that often has just been made available, rather than actually allowing the child to identify a situation in which such a response would be relevant. Second, the test questions are designed as hypothetical and as meta-talk, include questions, jargon, and mistakes that are unlikely to occur in the questioning phase of the interview. It may be that children and police officers recognize the unrealistic nature of the test questions. Yet, overt orientations are relatively rare. Additionally, recognizing the accountable and perhaps unrealistic nature of test questions does not make practicing more representative for the interview proper (cf. Komter 1987). Therefore, the findings cast doubt on how well the test questions serve the educational aim of the activity. This study thus relates to previous studies of how interaction can be practiced and tested (e.g. Linell & Persson-Thunqvist 2003; Stokoe 2013). Specifically, it unpacks some of the situated complexities in a context where practicing and testing is considered highly important.

Next, an unexpected finding was that unrealistic test items seem to have the potential to contribute to affiliative stance taking through mutual laughter (cf. Stokoe & Edwards 2008). It often remains implicit in the interaction what exactly the mutual laughter establishes. While participants do not treat this ambiguity as problematic, doing extra interactional work, such as explicit orientations to the laughability or accountability of the test question (e.g. excerpts (6), (7), (11), (12)), may create affiliation (cf. Steensig & Drew 2008). Therefore, the option of acknowledging the accountable nature of the test question, for example, ‘it sounds silly but what is the name of my dog’ (such as in excerpt (12)), is a potential practical implication of this analysis. However, when we discussed this with instructors from the police academy, they argued that overt orientations to accountability would render the police officer less credible (p.c., 2021), reasoning: ‘because the example (i.e. the test question) should not be taken as fully serious or representative, the police officer may not be taken seriously in general’. This disregards norms in social interaction and the accountable nature of asking ‘what is the name of my dog’ (cf. Jol & Stommel 2016b, 2021). Yet, more explicit orientations to the marked nature of the test question would also add to the unrepresentative and complex nature of the test questions.

The findings pose problems for the aims of instructing and practicing. One of the aims is that police officers should make clear that they don't know the answers. This aim is not supported by asking test questions that have correct answers as evidenced by police officers’ uptakes. The second aim is that police officers should build a relationship that is as equal as possible; the test questions that highlight knowledge asymmetries between the police officer and the child do not seem congruent to this aim.

It is also relevant to consider the fact that the interviews are produced for an overhearing audience in the judicial chain. This leads to the question of what giving and practicing instructions implies for how children are seen: as more or less reliable witnesses? The fact that police officers instruct children to answer ‘I don't know’, ‘I don't understand’, and ‘that is not correct’ treats children as probably not volunteering such responses on their own. Consequently, providing and practicing instructions contributes to a construction of children's statements as potentially untruthful. The instructions discursively reproduce the idea that a child's responses could be the result of not daring to answer ‘I don't know’ and so on.

This adds to the literature on explaining ‘truth’ and ‘lies’ to vulnerable witnesses in, for example, England and Wales (Ministry of Justice in England and Wales 2011). Whereas the explanation of truth and lies officially aims to enhance the reliability of the testimony, inquiring whether the witness understands the difference between truth and lies repeatedly or after the introduction part of the interview, casts the vulnerable witness as unreliable (Aldridge & Luchjenbroers 2011; Richardson et al. 2019). The problem of portrayal may be more salient in legal systems that involve a jury, yet reliability of vulnerable witnesses is often a point of discussion also in Dutch courts that consist of professional judges (e.g. Court Amsterdam 18 January 2012, case number 13/650843-11). To conclude, this analysis of practicing ground rules leaves us with a paradox: although children are instructed with certain ground rules for good reasons, the way in which these rules are practiced is highly complex and artificial. Also, the potential to achieve affiliation through the test items does not seem to be exploited even though this would arguably compensate for some of the difficulties. Yet, exploiting the test items for achieving affiliation may further highlight the artificial nature of the test items. The question remains as to how to encourage children to answer ‘I don't know’ and ‘I don't understand’, and indicate mistakes when necessary.