I’ve got a really tough act to follow here. This symposium has been really exciting. Again, I want to thank Christoph for inviting me to this. It’s really been not just exciting, it has really opened my eyes to a number of different ways in which legal communication could be made more effective both written and orally. But I’d like to kind of close this symposium by flipping that around and suggesting that the benefits that we can get by cooperating and collaborating between linguists and lawyers in terms of improving legal communication could also have another positive affect, and that would be to expose us to problems in the law. And more importantly, to point us to possible solutions to those problems.

And so the topic I’m going to talk about today is how is studying the discourse of police interrogation – how does that help us to understand the risks of false confessions and to avoid them and ameliorate them.

So I want to begin by simply saying that confessions are kind of the gold standard of criminal evidence. They’re the absolute most convincing proof of guilt. I mean after all why would you confess unless you were guilty? Sure, we never would. Well, all right, maybe if we were tortured, sure. But absent torture, we would never confess to a serious crime we hadn’t committed. And yet, the evidence is overwhelming that people do falsely confess. In fact, if you look in recent years, post-conviction DNA testing has revealed literally thousands of cases in which the wrong person was convicted of a serious crime.

And amazingly, in about twenty five per cent of those cases, the defendant actually confessed to the crime, that we now know, he didn’t commit. So how does this happen? It turns out that certain methods of police interrogation can lead innocent people to falsely confess to crime. Now, some of those cases involve suspects that are particularly vulnerable like juveniles or people of low cognitive capacity or the mentally ill, but the majority of false confession cases involve people with none of those risk factors.

So I want to illustrate these today through the true story of seven men who were accused of a crime that they didn’t commit. And what I hope to show you by the end of my talk is how an interdisciplinary linguistic analysis of confessions is actually a critical way in which we could prevent and redress these kinds of miscarriages of justice.

So here’s the story. In July of 1997 in Norfolk, Virginia, United States, the body of Michelle Bosko was found in her apartment. She had been stabbed and raped. Police interviewed the neighbours to see who might have a motive for the killing. And one of those neighbours suggested two people – Daniel Williams and Omar Ballard. She told police that she had attended a party where Daniel Williams had danced in what she thought was a sexually provocative manner with Michelle Bosko. And she also reported that Omar Ballard had been known to visit Miss Bosko often while her husband was away at sea. Based on the suspicions of the neighbour, the police began to focus on Daniel Williams as their suspect in the crime. They did not follow up her hunch about Omar Ballard. Had they done so, they would have discovered that Ballard was a suspect in two other sexual assaults that had happened in the days just before and just after the Bosko killing. But by the time of Ballard’s arrest in those cases, the police were already convinced that Daniel Williams was the killer, so Ballard was not investigated further in connection with the Bosko case.

Now, Daniel Williams didn’t exactly fit the profile of a sex killer. He had no police record. He had just gotten married only a few days before the murder and his wife said that he was in bed with her at the time at the killing. Still, the police considered him their prime suspect and they asked him to come down to the police station to help in their investigation. Williams had no idea he was a suspect. He just assumed that he was being asked to help solve his neighbour’s murder. So he arrived at the police station on the evening, and was interrogated thereafter continuously for over twelve straight hours by teams of officers accusing him of the crime. At one point he agreed to take a polygraph, which he passed, but the investigating detective lied to him and told him he had failed it. He was also told that eyewitnesses had seen him leave Bosko’s apartment at the time at the murder. This was another lie. Williams maintained his innocence for hour after grueling hour while the police insisted they knew he had done it and they could prove it.

After about ten hours of questioning, Detective Robert Glenn Ford took over the interrogation. Ford was known as a detective who could get confessions when other officers could not. Although he had been disciplined for his role in inducing a series of false confessions in an earlier case. Ford took an aggressive approach to interrogation. He yelled at Williams, poked him in the chest, told him he knew Williams was a stone-cold killer. And if he didn’t confess, he’d get the death penalty. If he confessed, on the other hand, Ford said he would promise that his life would be spared. Ford even suggested to him they might have amnesia concerning the commission of the crime because remembering what he’d done would be so emotionally painful.

As daylight broke, Williams finally cracked and agreed that he must have committed the murder. For over an hour, Ford prodded the exhausted Williams to change details of his confession to match the facts that Ford knew of the case. Finally satisfied that Williams’ confession matched the police’s version of the facts. Ford turned on a tape recorder and took a tape-recorded confession.

Later on in reviewing the tape, however, the lead detective on the case noticed many discrepancies between Williams’ confession and the autopsy report, which, by the way, Ford didn’t have. For example, Williams said that he had hit Bosko in the head with a blunt object, but the autopsy showed no head injuries. He never mentioned stabbing her but that was her cause of death. He said she was alive and screaming when he fled the apartment, but because of serious throat wounds that would’ve been physically impossible. He said there was no blood at the scene. The actually murder scene was extremely bloody. Clearly, this confession would not do. Detective Evans angrily confronted Williams with the inaccuracies between what he had told Detective Ford and what she knew from the autopsy. And Williams agreed to change his confession to match the facts as recited by Detective Evans. And the second revised confession was recorded. This was not the last time in this case that confession would have to be revised as the police learnt new facts from the case.

Well, the case was closed. The police had their killer. But several weeks later, DNA analysis came back from the crime lab and the semen in the victim’s body didn’t match Williams’ DNA. The police report was not to doubt Williams’ guilt but to look for an accomplice who must have committed the crime along with Williams. They quickly settled his roommate, Joseph Dick, who was hauled in for questioning. Dick too was subjected to an all-night interrogation, including police lies about the evidence and threats about the death penalty. Like Williams, Dick too denied involvement in the crime throughout the night and only crumbled after nine hours of interrogation. Dick agreed to confess to the new police theory that he and Williams had jointly committed the crime. Later on, the crime lab reported Dick wasn’t a match for the DNA either.

Still convinced that they must be guilty, or why would they have confessed? The police concentrated on finding a third accomplice – the one who would match the DNA.

A police informant put into Dick’s cell said that he had mentioned several times someone named Eric Wilson, so Wilson was brought in for interrogation. Ford used the same interrogation technique, so it’d been so successful. And after an all-night interrogation, Wilson cracked and admitting raping Michelle Bosko.

Unfortunately for the police, the DNA analysis excluded Wilson as the rapist. And there wasn’t even any physical evidence – no fingerprints, no hairs, no fibres. Nothing in the apartment linking those men with the crime. Dick’s defense lawyer pleaded with him to be cooperative with the police and amend his confession to add Wilson and a forth guy – the guy whose DNA would finally match. Terrified by the death penalty, Dick agreed to amend his tape confession to include participation by Wilson and a forth guy who he said, ‘I just know his name is George. I don’t know his last name or how to find him.’ So they got a police sketch artist to call in to produce a sketch of George. Armed with George’s sketch, they went to interview the neighbours, and one of them said, ‘That looks a little bit like a guy I know named Derek Tice.’

The police re-interrogated Joseph Dick, this time with a photograph of Derek Tice. And under pressure, Dick caved and said, ‘Yes, Tice is George.’ Again, an all-night interrogation commences by Detective Ford, this time of Derek Tice. Tice too eventually confesses. But this time, Ford doesn’t wait for the DNA evidence – that’s not been his friend. Instead, he said to Tice, ‘Unless you come up right now with the names of everybody else involved in this case, you’ll get the death penalty.’ So Tice offered the first names he could think of – two other sailors who were then arrested on that basis. Joseph Dick was hauled from his jail cell and told he needed to record yet another confession, including the new arrestees as members of the rape-murder plot.

Within two months, the crime lab determined that of the six men jailed in the Bosko case, none matched the DNA of the rapist. Tice was then re-interrogated and threatened with the death penalty if he didn’t name another co-defendant. He offered the name of John Danser, another sailor who was arrested on that basis. Danser actually had physical proof that he was two hundred miles away at the time of the murder and the DNA didn’t match him either. Nevertheless, the prosecutor decided to go forward with the case against all seven men.

About a month after these charges were filed – Omar Ballard, remember Omar Ballard? – Omar Ballard writes a letter to a friend admitting to having killed Michelle Bosko. The letter contained details of the crime that had never been made public. Ballard’s friend took the letter to the police who brought Omar Ballard in from his jail cell for questioning. Within minutes, he freely admitted to having killed Bosko. He told the police he had planned the seduction but Bosko resisted it, and then she started screaming. He tried to quiet her down, and he killed her then, so he wouldn’t be identified. His confession was in fact consistent with his other attacks on women before and after the crime. Oh, and his DNA? It was a match. Sounds like the end of the story, right?

But the police and prosecutor were unwilling to admit that they’d made a mistake in charging the other men. They promised Omar Ballard that he wouldn’t get the death penalty if he would agree to testify that the other men had been involved too. At first, Ballard agreed, although he later retracted his claim that anyone else was involved. It’s like you bring seven other guys along in a seduction, doesn’t really make sense.

Still, the police and prosecutors chose to believe him when he said all those seven men were his co-defendant accomplices and they chose to disbelieve him when he later tried to recant this claims of involvement.

Williams and Dick were represented by appointed defense lawyers who never investigated their claims of innocence. Instead, they were strongly advised to plead guilty and accept the life sentence instead of the death penalty. Feeling as though they had no choice, each did so. Eric Wilson and Derek Tice, however, refused to plead guilty and they went to trial where their confessions were admitted into evidence and they were both found guilty by separate juries despite the failure of their so-called ‘confessions’ to match the physical evidence in the case.

The remaining three sailors incidentally were never tried. Omar Ballard was permitted to plead guilty and avoid the death penalty.

So the question I want to ask is how did everybody in this case – police, prosecutors, defense lawyers and jurors – how did they all become convinced of these sailors’ guilt given the bizarre theory of this case that eight men, most of whom were barely acquaintances, somehow got together, plotted the rape and murder of a woman in which only one guy leaves physical evidence and lots of it, and the others not a trace of fingerprints or hair or fibre or semen or anything.

This case illustrates many of the factors that research has shown can lead to false confessions. Here are some of the hallmarks I want to call your attention to that recur in these cases where innocent people confess. First, from the police side, the phenomenon that psychologists call confirmation bias can distort police’s perceptions of evidence and cause them to doggedly pursue suspects who they ought to appreciate are innocent. Confirmation bias is the tendency that we all have to interpret new evidence to be consistent with what we already know and believe.

Let me give you a good example of an experiment done on confirmation bias. Psychologist Barbara O’Brien prepared a marked case of a burglary and shooting of the homeowner. Subjects looking at the case were divided into two groups and members of each group were shown a case file. That file contained evidence suggesting that a worker who had been fired from his job by the victim was a good suspect in the case. Subjects in one group were asked after reviewing the file, ‘Who’s your prime suspect?’, and they all picked the fired worker. The other group saw the same case file, they were just not asked that question. Then both groups were given a second part of the case file, including some evidence suggesting that the fired worker was guilty, some evidence that strongly suggested that he wasn’t and evidence about two other possible suspects.

What’s interesting is that the prime suspect group read the second half of the file very differently from the people who weren’t asked to name the prime suspect. The prime suspect subjects remembered more facts consistent with the worker’s guilt. They evaluated ambiguous evidence as proof of guilt and they discounted evidence that suggested he probably wasn’t guilty. It’s a classic case of confirmation bias. And remember, these subjects had no professional or personal interest in closing this pretend case.

In the case of real police investigations, confirmation bias has led to disastrous miscarriages of justice. In this case, once the police had focused on Daniel Williams as the murderer, confirmation bias made it easier for them to just kind of rework the details of the case over and over again, rather than reject that theory and start all over again.

Another factor that contributes to coercive interrogations are two related assumptions by police. First, that liars will display certain behavioural cues that they’re lying, and second, that the police can be trained to identify those successfully. Both of these assumptions are very questionable. There’s a lot of research being done right now, but the current state of the research is there is no reliable method to determine when somebody’s lying. And second, when police had been tested in double-blind studies, ‘Can you actually do this?’, they identify liars at about a fifty-fifty clip, that is pretty much a coin flip. But the thing is this: past experience that they have in interrogating suspects who turn out to be guilty, because, after all, most arrestees actually are guilty, convince them that they’re really, really good at lie detecting.

And so, Detective Ford, for example, was really sure that he could always spot liars. He thought all seven sailors were lying when they denied killing Bosko. The objective evidence suggested he was wrong. From the suspect side, being innocent itself is a risk factor for being manipulated in police interrogation. See, guilty people know they are guilty. So when they are being questioned, they are really alert for like, they’re going to try to trap me to make an admission, whereas innocent people are usually very naive. They think this is a big mistake and the more I can just talk to the police, the sooner they’ll figure out it’s a big mistake and I can go home. Even when Daniel Williams was directly accused of the murder, he believed that the truth would come out and set him free.

Innocent people are also susceptible to a common interrogation tactic where the interrogator confronts the suspect with supposedly objective evidence of guilt like a failed polygraph or fingerprints found at the scene. And they are asked, ‘How can you explain this away if you didn’t do it?’ In those cases, the evidence can be faked by the police. Now, US courts have approved the use of fake evidence by the police in interrogation because they reason, ‘Well, that might bluff a guilty person into confessing,’ ‘Oh, I guess you’ve got me,’ but it would never convince an innocent person to confess.

But unfortunately, we know now that the use of false evidence has been implicated in many of these cases where people falsely confessed that were innocent of the crime. Often the use of false evidence is coupled with the suggestion by the police that the suspect might be blocking conscious awareness of having committed the crime and that approach is often coupled with police minimization of the consequences of confession, suggesting that once we’ve all the facts from you, everything will be fine. We can get you psychological help. That’s all what we’re really interested in. Prosecution and prison, they would get admitted as possible consequences of confession.

So if you take fake evidence proving guilt, these tactics can convince an innocent person that they must have committed the crime, even though they have no conscious memory of doing so. And that’s especially powerful if somebody’s been interrogated for many hours and they’re sleep deprived.

Discourse analysis also has a role to play in unmasking false confessions. If you’re doing a pragmatic analysis of police interrogation, you begin by simply recognizing the extreme power asymmetry in this form of question-answer between the officer and the suspect. The interrogator completely controls the subject matter, the tempo, the length of the questioning and is the sole judge of whether a response is satisfactory or not. Police interrogators are actually trained to interrupt denials or explanations by the suspect, conditioning the suspect to accept complete discursive domination. The police often intentionally begin the interrogation late at night and continue it throughout the night to maximize their advantage over sleep-deprived suspects.

Linguistic microanalysis of the discourse in police interrogation can reveal many subtle structural mechanisms of discursive control. Controlling, and I’m not going to go into the details of this, but you know, it’s things like pragmatic discourse markers like that ones to control topics, mark digressions, change the subject, express stance, block attempts by the suspect to say things. Basically, it’s a structure that linguistically has designed to keep the complete narrative of the crime within police control.

I want to end with the conclusion of the story of the Michelle Bosko case at least as of now. Eric Wilson served his entire eleven-year sentence and was released. Daniel Williams, Joseph Dick and Derek Tice were all sentenced to lift without parole. They all filed unsuccessful appellate claims for relief. Upon his relief, Eric Wilson contacted the Innocence Project, the organization that investigates miscarriages of justice. And he told them he’d been convicted of a crime that he hadn’t committed and he wanted to clear his name. And he said there were three guys doing life without parole that he thought were also innocent. So the Innocence Project took on the case and they pressed the governor of the Virginia, a guy named Tim Kaine, to issue exonerations or pardons in the case. And finally, twelve years after Daniel Williams had first been interrogated, the governor issued conditional commutations of sentence to the men and ordered their release from prison, but he refused to declare them innocent. As of today, they still stand convicted of the crime. They have to register as convicted sex offenders.

This case illustrates the way in which standard techniques of police interrogation can literally create false confessions. Is there an alternative? As it turns out, there might be. Britain too has seen a number of high profile convictions overturned when defendants’ confessions turned out to be false. As a result, in England and Wales, they had police interrogation reform implemented to reduce both the coerciveness of interrogation and to try to reduce the adversarial confirmation bias by the police in the process. This method called the PEACE method after its five steps is fundamentally an open-ended investigatory process, one much less susceptible to confirmation bias. Does this solve the problem? There’s been some recent research on the PEACE method and unfortunately, it looks like what happens is all habits are hard to overcome and so they tend to sort of backslide into older interrogation patterns when they don’t get what they want.

But I want to conclude here, finally really conclude, with no extra-curricular activity. This story that I’m telling you today – I tell you because I think it’s a reminder that law and criminology really desperately need the insights and knowledge that social science can bring to bear to prevent miscarriages of justice. And understanding of human communication is incomplete when it doesn’t incorporate the insights of linguistics and psychology. That’s what we’ve been saying all day basically is you need this stuff in order to really make communication work effectively. And it’s exactly this kind of interdisciplinary approach that we’ve been hearing about all day that I’m suggesting can lead to kind of a research grounded public policy in practice and ultimately to a more just and humane world.