New linguistics research casts doubt on decades-old murder conviction
- Written by Helen Fraser, Director of the Research Hub for Language in Forensic Evidence, The University of Melbourne
On September 8 1988, 20-year-old Janine Balding was abducted, raped, and brutally murdered[1] in New South Wales. Police quickly arrested four youths, who accused an older man nicknamed “Shorty”.
Two weeks later, police interviewed Stephen “Shorty” Jamieson. Within a few hours, they had a full confession, typed by one of the detectives, and signed by Jamieson as a “record of interview” given of his own free will.
But when Jamieson arrived at his committal hearing, the youths called out to authorities: “you’ve got the wrong Shorty!”. It seems they had been referring to another man[2], also nicknamed Shorty, known to wear a black bandanna similar to the one used to gag the victim.
Nevertheless, Jamieson’s trial continued. In June 1990, he was convicted, along with two of the youths[3], and sentenced to life in prison. There he remains to this day, despite longstanding efforts by solicitor Peter Breen[4] to have his conviction reviewed.
Recent hearings[5] have focused on DNA analysis of the black bandanna. Our new linguistics research[6] casts doubt on the confession that convicted Jamieson.
Confessions as legal gospel
It’s important to be clear that while the other two were convicted on the basis of substantial evidence of guilt (which both later admitted), Jamieson’s conviction depended wholly[7] on the confession transcribed by police.
Nowadays, police interviews must be electronically recorded. At the time of Jamieson’s trial, a verbal confession could be admitted as a typed “record of interview”.
However, the risk of “verballing[8]” (police faking a confession that was never really made) was already well known[9].
Jamieson’s lawyers opposed the transcript vigorously during the trial, but the detectives testified it was accurate:
Defence lawyer: You see there is a very lengthy answer there that goes on for something in excess of half a page?
Detective: Yes.
Lawyer: Are you saying that those words were recorded exactly as Jamieson said it?
Detective: Yes, I am saying that.
Lawyer: You did not need to prompt him in any way?
Detective: No
Lawyer: Didn’t need to remind him about anything?
Detective: No. I did not.
In convicting Jamieson, the jury must have been persuaded by the detectives’ strong testimony.
A 1992 appeal was unsuccessful. The defence had one last hope: official review of the conviction.
Just the ‘gist’
A 2001 application to review Jamieson’s conviction included linguistic analysis by Rod Gardner (one of the authors of this piece), who compared the 1988 police transcript to a professional transcript of another interview with Jamieson, audio-recorded in 1995.
Finding many differences, Gardner concluded:
it is extremely unlikely that [the police transcript] is an accurate record of what would have been said in a police interview with Jamieson.
However, Justice Bruce James rejected the application, dismissing Gardner’s conclusion. He acknowledged the transcript was not Jamieson’s exact words, but thought it captured the gist of a genuine confession. Any inaccuracies simply reflected the detective typist’s “limited proficiency”.
He even excused the detective’s strong testimony, saying it was merely “an emphatic denial” of the defence suggestion, during cross-examination, that police “had concocted the whole interview”.
Confession or construction?
The Research Hub for Language in Forensic Evidence[10] has undertaken a fresh analysis of Jamieson’s case[11].
This asked if the detectives could have transcribed even the gist of an interview in real time, as they claimed.
A new experiment simulating their task suggests not. It used the video of a recent (unrelated) police interview. Participants had to type as much as they could of a three minute clip, without pausing.
All participants were fast typists. Average speed was 68 words per minute – well into the professional typing range[13]. One live-captioning expert managed an astonishing 142 words per minute.
Nevertheless, their average accuracy was only 34%, compared to the reference transcript.
Importantly, those who typed around 40 words per minute (surely the most Jamieson’s transcriber could claim, given his “limited proficiency”) averaged a mere 20% accuracy. That’s hardly the “gist” of an interview.
Read more: The dark side of mondegreens: how a simple mishearing can lead to wrongful conviction[14]
What does this mean for the 1990 trial?
The detectives, under oath, told the jury the transcript captured the confession “exactly as Jamieson said it”. Our research really questions whether that claim can possibly be true.
Many assume Jamieson’s signature proves the confession was genuine, if not exact. However, flaws in this assumption were clear as far back as 1987[15]. According to the Australian Law Reform Commission:
just as oppressive conduct can cause a suspect to make false admissions, so it can cause a suspect to sign a document containing those admissions.
This was one reason behind 1995 legislation[16] introducing compulsory electronic recording of interviews. By then, it had been officially acknowledged[17] at the highest levels that admitting an unverified transcript risks verballing.
Jamieson’s interview was completely unverified. He was alone with the detectives until a Justice of the Peace came to read the record of interview back to him (he couldn’t read at the time), and witness him signing his “voluntary” confession (which he withdrew as soon as the interview was over).
Where to from here?
Of course, none of this proves conclusively that Jamieson was verballed.
What it does do, surely, is strengthen the case for review of his conviction, to be made again in coming weeks.
Jamieson’s 1990 jury reached their verdict on the basis of testimony that has been acknowledged to be inaccurate, under legal procedures that have been acknowledged to be deeply flawed.
As long ago as 1989, a Queensland inquiry[18] made a recommendation[19] that resonates beyond state borders:
special consideration be given for a review of the convictions of any individuals who have raised allegations of “verballing” […] who are still in prison.
More to the story?
It’s comfortable to think the possibility of verballing died with mandatory electronic recording. But it lives on for those convicted under pre-1995 rules.
It also survives in legal procedures[20] that still allow juries to be misled by inaccurately transcribed confessions[21].
Read more: Covert recordings as evidence in court: the return of police ‘verballing’?[22]
The authors would like to acknowledge researchers Jane Goodman-Delahunty[23], Stephen Cordner[24], Robert Turnbull[25] and James Uy Thinh Quang[26] for their contributions to the research on which this article is based.
References
- ^ brutally murdered (en.wikipedia.org)
- ^ referring to another man (www.youtube.com)
- ^ two of the youths (www.tandfonline.com)
- ^ by solicitor Peter Breen (www.smh.com.au)
- ^ Recent hearings (www.smh.com.au)
- ^ new linguistics research (doi.org)
- ^ depended wholly (www.tandfonline.com)
- ^ verballing (www.youtube.com)
- ^ was already well known (www.unsw.edu.au)
- ^ Research Hub for Language in Forensic Evidence (arts.unimelb.edu.au)
- ^ fresh analysis of Jamieson’s case (www.tandfonline.com)
- ^ Wilkinson Publishing (www.wilkinsonpublishing.com.au)
- ^ professional typing range (en.wikipedia.org)
- ^ The dark side of mondegreens: how a simple mishearing can lead to wrongful conviction (theconversation.com)
- ^ as 1987 (www8.austlii.edu.au)
- ^ 1995 legislation (papers.ssrn.com)
- ^ officially acknowledged (www.australianpolice.com.au)
- ^ a Queensland inquiry (en.wikipedia.org)
- ^ a recommendation (www.ccc.qld.gov.au)
- ^ legal procedures (forensictranscription.net.au)
- ^ inaccurately transcribed confessions (forensictranscription.net.au)
- ^ Covert recordings as evidence in court: the return of police ‘verballing’? (theconversation.com)
- ^ Jane Goodman-Delahunty (www.researchgate.net)
- ^ Stephen Cordner (scholarly.info)
- ^ Robert Turnbull (findanexpert.unimelb.edu.au)
- ^ James Uy Thinh Quang (arts.unimelb.edu.au)
Authors: Helen Fraser, Director of the Research Hub for Language in Forensic Evidence, The University of Melbourne
