The Research Hub for Language in Forensic Evidence has been publishing up a storm recently. Check some of the new work below, aimed at a variety of different audiences. One of the articles discusses a new experiment using the audio you may have marvelled about on the front page of this website.
If only all forensic audio were as cute as this! For those who like to learn something from every bit of audio – you might like to notice there are several indistinct sections in this recording. Most of them are readily resolved via the internal context (context of other words in the sample). But can you spot any errors or uncertainties in the transcript?
A free online symposium examining the production and use of transcripts in criminal justice.
Thursday 11 March 2021 (morning in the UK, evening in Eastern Australia)
Sign up here
Speaker Individuality in Phonetics and Speech Sciences: Speech Technology and Forensic Applications
Thursday 4th – Friday 5th February 2021
Hosted by University of Zurich Department of Phonetics and Speech Sciences
Kirsty McDougall: Ear-Catching versus Eye-Catching? Some Developments and Current Challenges in Earwitness Identification Evidence
The School of Languages and Linguistics, and the Faculty of Arts, University of Melbourne, have established a new Research Hub for Language in Forensic Evidence.
The Hub aims to coordinate collaborative research involving linguistics, law and law enforcement, along with other disciplines, with the intention of assisting Australian courts to ensure that forensic evidence involving language and speech is used fairly and in the best interests of justice.
by Helen Fraser
‘Enhancing’ has become a routine part of preparing indistinct covert recordings for admission as evidence in criminal trials. Typically, evaluation of its effectiveness is seen as a simple matter of listening to determine whether the enhancement sounds ‘clearer’ than the original. This seems like a straightforward approach, but it brushes over many important issues which can adversely affect the fairness of the trial. This article outlines findings from experiments, case studies and scientific literature to show how enhancing can affect perception in surprising and unpredictable ways, without listeners’ conscious awareness. Discussion demonstrates that enhancing exacerbates, rather than mitigates, known risks of the jury being misled by an unreliable transcript. The conclusion indicates the direction in which to seek better procedures.
I am pleased and honoured to have been awarded the 2019 Rodney Huddleston prize for best article of 2018 in the Australian Journal of Linguistics for my paper Forensic Transcription: How Confident False Beliefs about Language and Speech Threaten the Right to a Fair Trial in Australia.
On 8 December 2017, Australian linguists submitted a ‘call to action‘, asking the judiciary to review and reform legal procedures for handling indistinct covert recordings. The call to action was endorsed by all four major linguistics organisations in Australia, and covered four main areas of concerns: transcription, translation, speaker attribution and enhancing.
Eyewitness evidence is known to be highly compelling in court
Unfortunately, it is also known to be highly unreliable
A very high proportion of wrongful convictions, discovered via DNA testing of historic evidence, have been shown to result from credible but inaccurate eyewitness testimony. (Check the Innocence Project for more detail.)
The third in a series of experimental articles demonstrating some little-known dangers of using ‘enhanced’ versions of indistinct forensic audio used as evidence in criminal trials has appeared. Fraser, H. 2019. Enhancing and priming at a voir dire: can we be sure the judge reached the right conclusion? Australian Journal of Forensic Sciences. The abstract and link are below.
This is the video (53 minutes) of the keynote speech I gave on 4 July at the IAFL conference in Melbourne.
It is intended for a forensic linguistics audience, and, as mentioned in the first minute, it goes rapidly through some established background about forensic transcription and translation, in order to bring together some key points for discussion about what is needed in order to ensure reliable transcription and translation of forensic audio used as evidence in court. If you are not in the target audience, you are of course welcome to new this video, but you might find one of the other videos or readings available on this site gives you a better introduction.
Today I participated (remotely) in a very interesting session on Sociolinguistic and sociotechnical approaches to official transcripts at the 16th Conference of the International Pragmatics Association, in Hong Kong.
The IAFL 2019 Conference will include a thought-provoking and engaging programme of keynote presentations representing academic scholarship, professional legal and language practice, law enforcement and the judiciary on the International Association of Forensic Linguists conference theme of Access to Justice through Language. I’m proud to say I’ll be one of them!
Get a preview of the exciting offerings below
Just $75 for each three-hour workshop, held on Mon 1 July at RMIT in Melbourne CBD.
The reasoning is: though ‘enhancing’ rarely if ever makes unintelligible audio intelligible, it might make it a little clearer, and if it doesn’t, well, there’s no harm done. But that’s not quite right, as a recent article in The Conversation shows via a dramatic demonstration.
This year, the International Congress of Phonetic Sciences (ICPhS) will be held in Melbourne, 5-9 August 2019. As well as many other interesting topics, there is a section on forensic phonetics. I’ll be there! It is too late to submit an abstract, but please consider joining us for one of the biggest phonetics conferences in the world.
I’m pleased and honoured to have been able to collaborate with Prof J Peter French, world renowned expert in forensic phonetics, to write a short and informative article for Australian lawyers about the dangers of providing police transcripts of indistinct covert recordings used as evidence in court.
This year the conference of the International Association of Forensic Linguistics (IAFL) will be held in Melbourne from 1-5 July 2019. The information blurb is below. Note you have a little over a month to submit your abstract (due by 15 Feb).
Forensic Linguistics – Crime and Punishment
I will be running a half-day tutorial on Forensic Transcription on Tuesday 4th December as part of the Tutorial Day preceding the Speech Science and Technology conference hosted by ASSTA and UNSW, and held in Coogee (a beachside suburb of Sydney). It will be possible to register for the tutorial day only – or of course you can stay and enjoy the full conference, which has some great keynote speakers.
Well there has certainly been a lot of publicity for the laurel/yanny clip recently. It is great to have so many people discussing speech and speech perception – but also a little disheartening that so much misinformation gets accepted as valid phonetics.
AAFS in Sydney on Valentines Day
What do problems of forensic transcription reveal about theory and practice of transcription in linguistics?
Departmental seminar – video below (43min; 70mB)
Recent research (e.g. Fraser, 2018) has revealed serious problems in the way transcripts are used to assist in the interpretation of indistinct covert recordings admitted as evidence in criminal trials (see forensictranscription.net.au). It is clear that linguists need to be involved in solving these problems, as shown by the recent Call to Action from ALS and ALAA to the Australian judiciary (see forensictranscription.net.au/news). But exactly what advice should we give? Transcripts of different kinds are used in nearly all branches of linguistic research, with practices varying widely from one context to another (Fraser, 2014). Which kind of transcript is most appropriate to the forensic context? Answering this and other relevant questions requires giving more explicit attention to theoretical concepts. What is a transcript? How can we be sure a transcript gives a reliable representation of the contents of a recording? Why did the legal system get it so wrong?
Legal precedent based on false beliefs proves hard to overturn
Judges consider their profession to be among the most accountable of those whose opinions and actions shape our society – but what happens when they make a mistake in their judgments? After all – judges are only human: mistakes must happen, however rarely.
Two new academic papers give detailed background on the legal and linguistic issues behind the problems with the use of covert recordings as evidence in criminal trials that are canvassed on this site.
I managed to record my recent presentation at the Australian Association for Forensic Sciences and have edited it into a short video for you. At 34 mins, it packs in a lot of new experimental results and case studies.
Problems and solutions from the perspective of forensic phonetics
SPEAKER: Dr Helen Fraser
Summary: Covert recordings are collected during investigation of most major crimes, providing vital intelligence to detectives. Some are also used as forensic evidence in court: it is here that major problems can arise. Covert recordings are often very indistinct, and can easily be misinterpreted. In one colourful example, ‘he died after wank off’ was shown to be a mishearing of ‘he died after one cough’ (French and Stevens 2006). Over the past thirty years, the law has developed practices intended to overcome anomalies like this, and ensure juries reach a reliable interpretation of what is said, and who says it, in forensic audio.
Recent meetings of the Australian Linguistic Society (ALS) and the Applied Linguistics Association of Australia (ALAA) voted unanimously to send a Call to Action to the Australasian Institute of Judicial Administration (AIJA), asking them to facilitate review and reform of current practices for admission and use of covert recordings as evidence in criminal trials.
Many covert recordings contain speech in languages other than English. How should this be presented to the jury? Obviously simply providing a transcript is not enough. The speech needs to be translated into English to enable the court to understand what is going on. The process is called forensic translation.
ANU Press has now published New Directions for Law in Australia: Essays in Contemporary Law Reform. The book is available free as a download, or at a cost as a print-on-demand hard copy.
How Interpretation of Indistinct Covert Recordings Can Lead to Wrongful Conviction: A Case Study and Recommendations for Reform (PDF, 0.1MB)
Find the audio version here to hear all the examples and audience reactions
The Sydney Ideas Post Truth Initiative Series examines fake news, alternative facts, lies, bullshit, and propaganda, from a range of perspective–from the courtroom to cancer research, from philosophy to screenplays, from Orwell to Sean Spicer–with the goal to better understand the post-truth crisis, and to advise on how facts and reason might survive in this climate.
Impressionist Rory Bremner explores the role of the human voice in forensic phonetics.
Forensic phonetics – or voice identification – has long been used in legal proceedings to help determine if the voice on a recording is that of the defendant. But with the electronic age enabling the recording and storage of more data than ever before, its role in criminal investigations is changing rapidly and the race is on to “fingerprint the human voice”.
ABSTRACT Results are reported of a new experiment using an indistinct covert recording from a real murder trial, along with the police transcript admitted to ‘assist’ the court to hear its contents. Previous research using the same material has shown that the police transcript is inaccurate, yet nevertheless highly influential on the perception of listeners ‘primed’ by seeing words it suggests. The current experiment examines the effects of priming participants with a made-up phrase that vaguely fits the acoustics of one section of the recording. Results indicate that a very high proportion of listeners are easily ‘assisted’ to ‘hear’ the made-up phrase. Discussion argues that audio of this quality should only be used as evidence if accompanied by a reliable independent transcript.
ABSTRACT Recent years have seen a great deal of attention given to the reliability of expert evidence admitted in criminal trials. However, almost no attention has been given to the reliability of evidence provided by so-called ‘ad hoc experts’. Indeed, many forensic scientists seem unaware that such a category of witness even exists, much less of the substantial threats they pose to the fairness of our criminal justice system. ‘Ad hoc experts’ are used for a number of evidence types. Here, we concentrate on one type that appears in Australian courts on a weekly basis: interpretation of indistinct covert recordings. The aim is to draw the attention of AJFS readers to serious problems in the handling of this much-used form of evidence, in the hope that the AAFS might develop a position on the issues and support calls for reform of practice.
Thanks so much to those who generously gave their time to participate in the recent find-the-phrase experiment.
The findings (along with discussion) have now been published in the Australian Journal of Forensic Science, along with a guest editorial explaining some of the background and motivation for the experiment.
National Judicial College of Australia Conference 2017
4 to 5 March 2017
The National Judicial College of Australia and the ANU College of Law invite you to attend their annual conference to be held at the John Curtin School of Medical Research at the Australian National University in Canberra on 4 & 5 March 2017.
The difficulty of providing fresh evidence of wrongful conviction that emerges after the trial and associated appeals have been concluded is discussed at this interesting but troubling event hosted by Sydney University: