CRCC supports research to improve evidence-gathering when reporting domestic and sexual violence to the police
In our latest post, Dr Emma Richardson discusses how research using conversation analysis (CA) helps us understand ‘evidential difficulties’ in crime reporting of gender-based violence.
According to official data, the impact of violence against women and girls in the UK and beyond is pervasive. For example, in the year ending March 2020, 78% of investigations of domestic abuse-related offences were closed as a result of “evidential difficulties”, while a similar trend was documented in the year ending June 2021, with 60% of police-recorded rapes also closed with the same outcome. Evidential difficulties cover a range of ‘outcomes’, and is the term given to cases where the police were unable to build a strong enough case for the offence to continue in the criminal justice system.
Given the high prevalence and low reporting of gender-based violence, it is essential to better understand if evidential difficulties can be identified in the evidence-gathering activities. These are essentially a series of interactions, or conversations, between lay (the public) and professional (the police) parties. Evidence-gathering activities include the initial telephone calls where people are reporting violence, the investigative interview, and the production of records (transcripts of the interviews) which may later be used in court. Linguistic research can assist the police to improve their evidence-gathering activities, by examining them as they occur in real-time and using real examples to feed back into training for those who conduct these activities. In doing so, potential barriers to ‘evidential difficulties’ can be designed out.
My research focuses on applying conversation analysis (CA) to various stages in the collection, or creation, of ‘evidence’. Conversation analysis is a qualitative research method used to show how conversation is systematic (not “messy”), and that there is ‘order at all points’ in our everyday interactions. Therefore, CA can be used to investigate the organisational structure of evidence-gathering activities (e.g. between the opening, free narrative and probing of the investigative interview) and design of social actions (such as different ways of asking questions to elicit information in the emergency calls). CA is useful in this context as it allows us to understand what is happening in real time for the parties who are reporting and gathering evidence. CA uses ‘naturally occurring’ data, meaning the data is captured at the time of the interaction taking place. For the projects I describe below, this means the police collected this data anyway as part of their evidence-gathering processes.
Through the conversation analytic research into the production of evidence I conducted with a team of colleagues, I have identified three stages in the evidence-gathering process where interactional practices could be contributing to evidential difficulties. These are 1) the initial reporting of violence to the police via emergency and non-emergency calls; 2) the investigative interview process, and 3) the transcription of the interview.
The initial reporting of violence to the police
The first stage in seeking justice is to report the offence to the police. There are many barriers to contacting the police in relation to sexual offences which are also equally applicable to incidents of domestic violence abuse: for example, the fear of retribution by the offender or others connected to the offender. Those who do overcome such barriers and make a formal disclosure to the police often describe negative experiences, including feeling disbelieved or blamed, feelings of shame, and embarrassment. These barriers and negative experiences ultimately result in a lack of confidence or trust in justice organisations.
To better understand the problem, alongside colleagues, I collected and analysed a corpus of emergency (999) and non-emergency (101) calls where the callers were reporting domestic violence abuse (DVA). We explored how callers ask for help when they might be in close proximity to, and therefore potentially overheard by, the perpetrator of the violence they are reporting. We found excellent call-taker practices, with call-takers and callers playing on conversation ‘norms’ (such as opening calls with, “hiya, you all right?” rather than, “I need the police please”) and used other available resources (such as breaths) to communicate they needed help.
Additionally, we found that when victims request help to be sent, call-takers ask a series of ‘routine’ questions in advance of dispatch. These include if anyone at the location has any injuries, if there are any children at the property and if anyone has any weapons. This information is vital for the police to understand if any additional services are required (such as an ambulance) and to inform agents arriving on the scene what to expect. However, included in this list was a question about the victims’ or their partners’ alcohol consumption. Given the stereotypical attitudes (e.g., victim-blaming, rape myths) women experience at each stage of the criminal justice system from institutional parties, including the police, judges, and magistrates, it is not surprising to see women pre-empt and account for their drug and alcohol consumption when reporting gender-based crimes to the police.
Our analysis revealed that, if questions about alcohol consumption come before a confirmation that help will be sent (e.g. “We’re getting help out to you straight away”), and probe the caller’s own drinking (but not the perpetrators’), callers respond to these requests as if they were meant to hold them morally accountable. This finding can be used to mitigate women’s negative experiences when reporting violence, simply by ensuring it is asked after a confirmation that help will be sent. We are working with our police partners to present examples of how the different positions in which this question is asked impacts how callers respond to the questions. This means our findings can feed back into call-taker training.
The investigative interview
After the initial reporting of the crime, complications in the interview stage may lead to evidential difficulties. Research suggests that vulnerable individuals, including witnesses of crimes and particularly those with intellectual impairment or disability, are more likely to be sexually assaulted and less likely to report an assault. In this context, the police employ a guidance document, titled Achieving Best Evidence (ABE), for interviewing vulnerable and intimidated witnesses of crimes, including sexual offences based on a victim-centred and trauma-informed approach. Alongside collaborators, we conducted a study that examined a corpus of ABE interviews to understand how ABE guidance was used in practice in police interviews in England and Wales.
We discovered a series of issues that affect both interviewers and interviewees and ultimately may lead to evidential difficulties in a case. For example, we observed that asking interviewees to demonstrate their intellectual competence by stating the difference between a ‘truth’ and a ‘lie’ may have unintended consequences on the credibility of the interviewee. We also considered how interviewers probe inconsistencies in interviewees’ accounts of what happened and how they question the behaviours of vulnerable child and adult complainants of sexual violence; how they in turn resist any implications of fault, (e.g., “How come if he was doing x, you did y?”) and how interviewers deal with complainants becoming distressed during interviews.
Transcription of the investigative interview to produce written records
Another issue that may result in evidential difficulties is linked to the ways evidential records are produced for investigative and trial use. Alongside collaborators at Loughborough University and Aston University, we recently compared audio-visual investigative interview records to scrutinize the current practices of capturing spoken interaction in legal contexts in England and Wales. We found that police transcribers alter the social actions being performed by speakers when they produce written records of the interviews. For example, the interview transcripts contain omissions in comparison to the audio data and inconsistent use of ‘…’ to represent various audible features such as pauses or overlaps in speech; moreover, sections of audio speech are heavily summarised. Thus, the findings of our study challenge the adequacy of these records for use in our legal system. Notably, if a case goes to trial, the written record of an interview is presented in court instead of the audio. Because we found that this practice distorts interview data, we delved deeper into the matter by conducting additional experimental work. Our study showed that when presented with a written transcript of a police interview, members of the public (and therefore jurors) are more likely to make judgements about the credibility of the suspect and their account than if listening to the original audio. Our findings provide a strong motivation for further research into how we capture spoken interaction in legal contexts, and they highlight the need for reform of how police interviews are transcribed in the UK.
Applying conversation analysis to real data at various points in the evidence-gathering process enables training to be developed based on practices that are happening every day across the judicial system. This work is invaluable to understanding how policy and guidance are enacted in practice in these settings and is key to furthering our understanding of how the interactions may be contributing to ‘evidential difficulties’.