AustLII Home | Databases | WorldLII | Search | Feedback

Alternative Law Journal

Alternative Law Journals (AltLJ)
You are here:  AustLII >> Databases >> Alternative Law Journal >> 2007 >> [2007] AltLawJl 24

Database Search | Name Search | Recent Articles | Noteup | LawCite | Author Info | Download | Help

Heydon, Georgina --- "When Silence Means Acceptance: Understanding the Right to Silence As a Linguistic Phenomenon" [2007] AltLawJl 24; (2007) 32(3) Alternative Law Journal 149

  • WHEN SILENCE MEANS ACCEPTANCE: Understanding the right to silence as a linguistic phenomenon
    Understanding the right to silence
    as a linguistic phenomenon


    The suspect’s right to silence during a police interview, and the debate about the confidentiality of the suspect’s choice to exercise that right, have thus far been considered primarily a civil rights concern best addressed by the legal fraternity. Scholarly articles and newspaper editorials and opinion pieces have considered the controversy from a number of angles — moral, ethical, legal — but none, to my knowledge, have considered the linguistic implications of a ‘right to silence’.

    This article has emerged from my ongoing research into the discursive behaviour of participants in police interviews with suspects[1], which is based on the analysis of data from police-suspect interviews tape-recorded in Victoria, Australia. The article presents an analysis of language data drawn from those police interviews recorded, with the intention of providing an introduction for non-linguists to the use of linguistic analysis in a legal setting.

    Crucially, the analysis of police interview data demonstrates that people expect a vigorous denial when someone is accused of something, and that the ‘conversational rules’ that generate this expectation will apply regardless of the speech context. As a result, a person who exercises their right to silence will be seen as ‘failing to deny’, and so to be accepting the allegations against them.

    Conversation Analysis (CA)

    Naturally occurring conversations provide a rich source of data for the branch of linguistics known as Conversation Analysis (hereafter CA). The analysis of the data in this study draws on tools used within the framework of CA as it was originally proposed nearly forty years ago by the late sociologist Harvey Sacks, and subsequently developed by his colleagues Gail Jefferson and Emanuel Schegloff.

    While a police interview may seem far from natural it is in fact a perfect example of institutional discourse, recorded in its natural or intended setting without any influence from the researcher. For this reason we are able to analyse segments of the interviews in order to identify those conversational ‘rules’ which are observed by the participants as they construct the discourse.

    As we are dealing with police interview data, it is important to recognise that such talk is conducted within certain organisationally-determined parameters, especially the pre-allocation of the police authoritative ‘voice’. However, the literature has clearly established that these parameters are themselves identifiable as conversational rules, and Conversation Analytic methods have been fruitfully applied to a wide range of institutional talk to reveal precisely how the rules of ordinary conversation are applied or adapted to produce the very characteristics of an institutional interview [2].

    One of the conversational rules which forms the focus of the present study, and which will be discussed in detail below, governs the use of what are known in CA as ‘preferred responses’. It may be surprising for those unfamiliar with language analysis to find that participants in an interaction rigidly adhere to rules of any kind, but in fact there are well-established rules that govern who can speak next and the type of contribution that can be offered in response to an initial turn at talk.[3] These rules can be broken, but they cannot be ignored, by which I mean that it will be noticeable to the participants when a rule has been broken and some redress will often be required.

    When a question remains unanswered or a person speaks out of turn, any competent participant in the conversation will know that a rule of conversation has been broken. In fact, our language is replete with phrases and terminology for just such occasions: ‘I hadn’t finished’, ‘wait your turn’ ‘don’t butt in’ ‘sorry to interrupt’, and perhaps a cliché of the media interviewer, ‘Minister, you haven’t answered the question!’. The careful scrutiny of these and other less-obvious rules provides a unique perspective, and offers a micro-level, linguistic response to a macro-level, legal question.

    Legislative background

    In the state of Victoria, linguistic considerations of interview procedure arise from the Crimes Act (1958), and the police Standing Orders which are derived in part from the Crimes Act[4]. For instance, a useful insight into the nature of ‘voluntariness’ in a police interview is provided by s464A (3) of the Crimes Act (1958) and s8.5 of the Standing Orders. The Crimes Act states that before any questioning commences, a suspect must be told ‘that he or she does not have to say or do anything but that anything the person does say or do may be given in evidence’; the Standing Orders state that a confession is defined as ‘voluntary, not in the sense that it is made spontaneously or that it was volunteered, but in the sense that it was made in the exercise of a free choice to speak or be silent.’ Thus we can see that the use of a caution by police officers to advise suspects of their right to remain silent is a step which is intended in itself to render any subsequent confession or admission voluntary.

    However, subsequent sections of the police Standing Orders recognise that a confession which follows a caution is not voluntary as a matter of course, and police officers are instructed to avoid certain strategies which may jeopardise the voluntariness of any confession or admission.[5] Although a confession may have been offered which is deemed voluntary by virtue of having been made by a suspect who is aware of his or her right to remain silent, the approach taken by the police officers in the elicitation of such a confession may still render the confession involuntary. Both the legislation and the Standing Orders recognise that, for suspects faced with coercive police behaviour in an interview, merely knowing that they can remain silent is not considered sufficient protection against forced confessions.

    Preference in Conversation Analysis

    One of the key structural features of conversation identified by Sacks and his colleagues is that almost all the speakers’ contributions, or ‘turns’ as they are called in CA, come in pairs: an initiation turn (eg a question or an invitation) and a response turn (eg an answer or an acceptance/rejection). In CA, such pairings are referred to as ‘adjacency pairs’ because they are bound to each other in an adjacent position in the conversation. The initiation turn is referred to as the ‘first pair part’ and the response turn is referred to as the ‘second pair part’, even when the second part of the adjacency pair does not immediately follow the first. Within this conversation structure the phenomenon of ‘preference’ arises.

    In CA the term ‘preference’ is used as a technical notion, not in its everyday sense;[6] CA provides a set of conversational rules which are ‘conventional reference points that actors orient to and that give behaviour its particular intelligibility’, and ‘by which actors understand one another’s behaviour’[7]. Participants in an interaction can draw inferences based on these conventional reference points; this is fundamental to the concept of ‘preference’ in CA.

    The technical notion of preference is that, following an utterance which is a first pair part of an adjacent pair (eg an invitation, request, accusation etc), certain responses, or second pair parts, are ‘preferred’ over others by virtue of the fact that if there is no response, those ‘preferred’ responses will be noticeably absent. For example, following an invitation it is possible for the recipient to accept or reject the invitation. However, if the recipient remains silent, or otherwise fails to offer either an acceptance or a rejection, it is the acceptance which is lacking, and a rejection is assumed to have been offered in its absence. Consider the following exchange:

    John Do you want to see a movie with me next Friday? First pair part (invitation)

    Jane … er … Second pair part (non-response filler)

    John Well, maybe Saturday? First pair part (invitation)

    Jane’s non-response is interpreted by John as a rejection, since an acceptance was not offered, and John offers a modified invitation next on the basis of this interpretation. Although this is an invented conversation, any competent speaker of the language would accept as reasonable the way in which John uses the rules of preference to make inferences about Jane’s second pair part — his modified invitation does not come as any great surprise.

    Of interest here is Bilmes’ discussion of a particular case of preference which concerns accusations. Bilmes agrees that following accusations, denials are preferred,[8] and says: ‘[i]f one fails to deny an accusation, a denial is noticeably absent and is a cause for inference, the most common inference being that the accusation is true’[9].

    Bilmes goes on to demonstrate that in fact this preference for denials following accusations is part of a broader type of preference — ‘when A attributes some action or thought or attitude to B, in B’s presence, there is a preference for B to contradict A interruptively or immediately following the turn in which the attribution was produced… When such attribution occurs without contradiction, a contradiction is relevantly absent’[10]. Bilmes demonstrates the strength of this argument using a number of examples of both contradicted and

    non-contradicted attribution[11].

    Some general features of police interview interactions

    The use of language by speakers in police interviews contributes to the construction of the interview as a ‘chain’ of adjacency pairs, most of which can be loosely classified as question and answer (Q-A) pairs. This does not mean simply that the discourse consists of interviewers asking questions to which interviewees offer answers, which would seem rather an obvious feature of an interview. Rather, it describes the specific conversational rules that are applied by participants to ensure that this is the only structure which is allowed in the interview. Thus, suspect-initiated utterances are produced only within exchanges or turn types that facilitate the return of the floor to the police participant at their conclusion. For example, if a suspect initiates a question, it is always a clarification question, which, together with the police clarification response, is inserted between a prior police question and a subsequent suspect answer such that there is no disruption to the overall chain of Q-A adjacency pairs. There is an inflexible ‘chain rule’[12] governing turn allocation which operates in police interviews, so that recurring sets of adjacency pairs oblige the suspect to respond to first-pair parts, such as questions, and return control of the interview to the police interviewer.

    If we consider the institutional requirements which produce the interview turn structure, we see that it is the role of the police officer as ‘elicitor’ which is crucial in establishing the recurring chain rule. This is made clear in the allocation of ‘topic management strategies’.

    One of the results of the chain rule is that the role of interviewer affords the police officer a far greater range of topic initiation devices than it does the interviewee. Whereas the interviewee is only able to introduce new topics in ways which do not oblige the police interviewer to take up a respondent role, an interviewer can introduce a new topic within any first pair part. The suspect is constrained to topic initiations which are minimally obligating and can be easily ignored, while the police interviewer is able to introduce new topics within highly obligating adjacency pair structures. For instance, the interviewer is able to ask questions which oblige the interviewee to produce a topically-relevant answer, even if the interviewee’s previous turn related to a completely different topic. The following extract demonstrates this:[13]

    Extract 1

    The effort made by the suspect to complete his turn in line 391, when he has been interrupted by a topically disjunctive question put to him by the police interviewer, is subsequently ignored by the interviewer. This underlines the weakness of the obligation on the police interviewer to take up new information provided by the suspect despite the overt display by the suspect that he considers the information important.

    The application of a Q-A chain rule in interviews provides police officers with recurrent ability to produce highly obligating topic initiation devices in any sequential position. The structure of the turn-taking mechanism ensures that police interviewers are endowed with an authoritative voice by virtue of their institutional role, while suspects are heavily constrained in their allowable contributions. We need to recognise that such an authoritative voice can provide the means by which a police officer may use inappropriate pressure to elicit a confession or admission.

    Accusations and attributions

    Accusations and attributions are a vital resource for police interviewers who are trying to establish a police version of events as they present the police narrative in a form that obligates the suspect to respond. The following data illustrate how accusations and attributions are produced and responded to in police interviews. Denials usually occur ‘interruptively or immediately following the turn in which the attribution was produced’ (emphasis added) ,[14] and in the analysis of these extracts it will become clear that the timing of the accusation is critical to its effectiveness.

    In Extract 2 the suspect interrupts the police interviewer in line 316 to deny the allegation made against him:

    Extract 2

    Extract 3 demonstrates again that denials are placed interruptively by the respondent; in lines 334 and 337, the suspect interrupts the interviewer to deny the accusation made in line 333:

    Extract 3

    Extract 4, a different case, provides a contrasting example: the suspect does not directly deny the accusation, implicit in the interviewer’s utterance in line 159, that the suspect ‘forcibly dragged’ his girlfriend out of the house:

    Extract 4

    This is particularly interesting because a little later, in line 181, the interviewer indicates that he has heard a lack of a denial by the suspect as agreement, describing the action as still draggin’ ’er:

    Extract 5

    By failing to immediately and directly deny the accusation made in Extract 4, the suspect leaves open the possibility that he has accepted the accusation, even though a denial is made later in the data.

    Similarly in line 184 and following, the interviewer makes an accusation that the suspect has dragged his girlfriend by the hair. The suspect does not expressly deny this in his immediately following turn but says it happened at a different time:

    Extract 6

    Later, in lines 193–197, the suspect explains that he did not drag the victim outside by the hair, but rather that he had hold of her hair as she was sitting inside and tried to pull her to her feet:

    Extract 7

    Clearly, a denial made at a time removed from the accusation does not have much impact. It is apparent that the police officer is inferring from this lack of an adjacent denial that the suspect accepts the accusation: he restates that supposition in line 203. In line 243 the interviewer again says ‘dragging her outside’, despite the fact that the suspect has never directly admitted that he undertook this action, and has offered various forms of denial:

    Extract 8

    There is evidence in Extracts 4–8 that a suspect’s lack of a contradiction immediately following an attribution, is treated as an implicit acceptance by the suspect of that attribution.

    This analysis supports Bilmes’ findings that denials are routinely treated as ‘preferred’ by interviewers and, importantly, that the timing and placement of the denial is key to its recognition in the interview. This raises a number of concerns about the practicality of invoking of one’s right to silence — concerns that a traditional legal analysis may not identify because they are apparent only from the micro-level management of conversation.

    The implications of ‘preference’ for right to silence

    We have seen that the police interview provides a constrained speaking environment for suspects which leaves them vulnerable to acts of discursive coercion by interviewing officers. We are aware that this vulnerability is addressed in part by legislation requiring all suspects to be informed that they have the choice to remain silent at any time. However, the CA notion of ‘preference’ informs us that when a suspect actually invokes their right to silence in response to an accusation or attribution made by police interviewers, no denial or contradiction will be inferred, and ‘[g]enerally, the conclusion drawn is that the [suspect] is acknowledging the truth of the attribution [or accusation]’[15].

    The implications are extremely serious. If the suspect decides not to respond to an accusation or attribution made by an interviewing officer, it may prove difficult for the suspect to later address any further assertions that the interviewer makes on the basis of inferences drawn from the suspect’s ‘absent denial’. Clearly this will be exacerbated if the suspect wishes to continue to invoke his or her right to silence.

    A complicating factor is the suspect’s lack of access to turn types in the Q-A chain — such as topic initiations, questions and other first pair parts — which may be needed to address police interviewer assumptions. Suspects, because of their role as ‘respondent’, may never feel able to take up a ‘leading’ role in the conversation in order to backtrack and address a police assumption made some time previously in the interview. It is common enough in ordinary conversation for participants to simply ‘let it go’ when some prior comment has been misinterpreted — how much more likely that a suspect will avoid the difficult and confronting task of stopping the interview to take up a conversational role that is not allocated to them, and dispute a police interpretation of some earlier response, or, worse still, lack of response? This is of course assuming that the suspect is even aware of the conversational rules of preference that have led to the misinterpretation.

    If, in a subsequent trial, the defendant were known to have invoked his/her right to silence at any time during the police interview, a judge or jury would be able to draw the same inferences of ‘absent denial’ and acknowledgement of truth from the defendant’s silence in the interview. In other words, a judge or jury, on hearing that the defendant had offered no response to a particular police accusation or attribution, could unconsciously apply the rules of preference and infer that the defendant had agreed with the police utterance.

    The possibility of a court drawing an adverse inference from a suspect’s refusal to respond to police questions has continued to be a cause for concern articulated by the legal fraternity.[16] While they consider the legal arguments surrounding the issue, this research makes it clear that there are, as well, important linguistic considerations operating at a level often undetectable to a non-linguist.

    It is hoped that these linguistic considerations will enhance the legal understanding of silence as a response to questions, and contribute usefully to the current debate over the right to silence in police interviews.

    [*] GEORGINA HEYDON is a forensic linguist and post-doctoral research fellow in the School of Humanities, Communications and Social Sciences at Monash University.

    © 2007 Georgina Heydon

    I am very grateful to Bronwyn Naylor of the Law Faculty, Monash University, for her valuable comments on an early draft of this paper and to the editors and reviewers at the Alternative Law Journal for their assistance in making this research more accessible to a non-linguistic audience.

    [1] Georgina Heydon, The Language of Police Interviewing : a critical analysis (2005).

    [2] Paul Drew and John Heritage (eds), Talk at Work (1992); this edited collection is a seminal text in applying CA to institutional settings.

    [3] The rules of turn-taking were first identified in Harvey Sacks, Emanuel Schegloff and Gail Jefferson, ‘A Simplest Systematics for the Organisation of Turn-taking for Conversation’ (1974) 50(4) Language 696.

    [4] The Standing Orders were superseded in the early 1990s by the Operating Procedures of the Victoria Police Manual, which present the same information in a more succinct format. However, the Standing Orders provide a more interpretative insight to the Crimes Act and in any case, were still current when the data analysed here were recorded. For further discussion of this issue, see Heydon, above n 1, 6.

    [5] Victoria Police Standing Orders Section 8.8.

    [6] Jack Bilmes, ‘The Concept of Preference in Conversation Analysis’ (1988) 17 Language in Society 161.

    [7] Ibid 162.

    [8] J. Maxwell Atkinson and Paul Drew, Order in Court: The organisation of verbal interaction in judicial settings (1979).

    [9] Bilmes, above n 6, 167.

    [10] Ibid.

    [11] Ibid, 167–9.

    [12] Richard Frankel, ‘Talking in interviews: a dispreference for patient-initiated questions in physician-patient encounters’ in George Psathas (ed), Interaction Competence (1990) 231.

    [13] Detailed, or ‘narrow’, transcriptions, using symbols to illustrate as many linguistic features as possible, are an essential part of the CA approach to micro-level language analysis, as the variation in pitch, emphasis and timing can be critical to the interpretation of a speaker’s turn. Symbols show features such as overlapping speech, ‘latching – when the next utterance starts immediately after the previous turn, and pauses measured in seconds and fractions of second. Transcription conventions have, however, been removed from these extracts but are available from the author.

    [14] Bilmes, above n 6, 167.

    [15] Ibid.

    [16] See for instance David Hamer, ‘Does Silence Imply Guilt?’ (2006) 72 (Jan–Feb) Precedent 4; Katherine Biber, ‘On Not Speaking: the right to silence, the gagged trial judge and the spectre of child sexual abuse’ [2005] AltLawJl 5; (2005) 30(1) Alternative Law Journal 19.

    AustLII: Copyright Policy | Disclaimers | Privacy Policy | Feedback