Presentation at NSW Judicial Commission (40 mins)
Scroll down for answers to some audience questions.
Text of Call to Action
8 December 2017
Dear Professor Reinhardt,
We write on behalf of the Australian Linguistic Society (ALS) and the Applied Linguistics Association of Australia (ALAA).
These organisations wish to call on the Australian judiciary, through you, to review and reform the legal handling of covert recordings used as evidence in Australian courts. As explained in the attached article*, there are substantial problems with the processes developed within the law over the past 30 years that are creating actual and potential injustice in our courts.
These problems relate to four main areas:
- translation of covert recordings in languages other than standard Australian English
- transcription of indistinct English in covert recordings
- attribution of utterances in covert recordings to particular speakers
- use of ‘enhanced’ versions of poor-quality covert recordings.
We would be grateful if you could circulate this document to relevant organisations.
We have established a Reference Group on covert recordings, chaired by Dr Helen Fraser, to facilitate discussion of these matters. Perhaps you might encourage formation of a similar group of appropriate representatives from law and law enforcement, to liaise with our group.
We believe the best starting point is to convene a 1-2 day workshop to discuss the issues raised by the attached article and related materials, to be held in the first quarter of 2018. We would like to hear back from you with a formal response by the end of February. In the meantime we are available for informal discussion or consultation on any aspect.
*Article attached to Call to Action
Fraser, H. (in press). Thirty years is long enough: It’s time to create a process that ensures covert recordings used as evidence in court are interpreted reliably and fairly. Journal of Judicial Administration. May 2018
Other background materials
Find other relevant material on this website, especially under Too Busy to Browse?.
Call to Action document
Download the actual Call to Action from this link: https://forensictranscription.com.au/als-alaa-aija-call-to-action_signed/.
How can you tell what is really said? How can you say what is not said if you don’t know what is said?
For an analogy, consider the following snippet of messy handwriting:
What does the snippet say? It is hard to be sure. Does this snippet contain the word ‘pact’? It is not too hard to be sure it does not. Why? Because there is no sequence of lines capable of being interpreted as representing the relevant sequence of letters.
Roughly, a similar principle applies to speech. The crucial difference, however, is that with handwriting, anyone can make the relevant visual inspection. With speech, the equivalent of the ‘lines’ in handwriting are ‘acoustic cues’ – and recognising the relevant acoustic cues takes specific expertise in phonetic science (for a real life example, see this news item). Plus, most importantly, speech perception is far more susceptible to the effects of priming by misleading contextual information, as described in my presentation.
Shouldn’t you be talking to defence lawyers about this, not judges?
I have talked to many defence lawyers, under various circumstances. Picking up errors in existing transcripts is far harder than expected. The onus should be on prosecution to demonstrate that their evidence is fair and reliable, not on defence lawyers to show it is not. See further discussion here.
Another aspect that has been brought home to me via case work is that defence lawyers are not primarily interested in finding out what is said in an indistinct recording. They are primarily interested in winning their case. That of course is not a criticism of defence lawyers – it is their job. The problem is that the two aims are not always in perfect alignment. Plus: even if defence does come up with their own transcript, it is still only an alternative opinion, with the jury invited to evaluate each and decide which they prefer. The difficulty of evaluating transcripts is vividly highlighted by the judgment brought to my attention by Judge Berman. This is in many ways a more compelling demonstration than the ones I gave in my talk.
Wouldn’t it be best to play all audio without any transcript?
This is a common response, but ignores the possibility of priming by context or suggestion. As seen in the ‘Adelaide bank account’ example in the presentation, it doesn’t take much of a hint for listeners to confidently ‘hear’ words that aren’t there. From a phonetics perspective, the solution to the problems discussed is to ensure that all audio admitted in court is accompanied by a reliable transcript.
A related common misconception is that to be reliable, a transcript should be produced with no contextual knowledge at all. This is not a good solution from the point of view of phonetic science. Withholding all contextual knowledge does remove unhelpful priming – but at the cost of also removing helpful priming. The right way to create a reliable transcript is to start with no priming (so as to be able to distinguish aspects of the transcriber’s perception that come directly from the audio, from aspects that come from contextual knowledge), and then introduce reliable contextual knowledge in a controlled manner. This is what phoneticians naturally do – but the process has more recently been researched by psychologists (notably Itiel Dror) who give it the title linear sequential unmasking.