A5. The perception experiment

Unfortunately, while police, lawyers, juries and others are happy to accept scientific opinion that DNA or fingerprint evidence is inconclusive, when it comes to phonetic science, non-specialists feel they can reach a valid conclusion simply by ‘listening with their own ears’.

However, phonetic science has well-established findings showing that what people’s ‘own ears’ tell them is quite often wrong.

In order to demonstrate this, I used the audio from this case (with permission) in a formal experiment (co-authored with my former colleague Dr Bruce Stevenson), published in International Journal of Evidence and Proof.

The experiment

The experiment allowed participants to play the section of audio transcribed by the detective as ‘at the start we made a pact’, in its surrounding context, repeatedly, under various conditions – all far more conducive to accurate hearing than those available to the lawyers, judge and jury in the case.

Those interested in this topic will find it best to read the article. Here we merely summarise a few key results.

Key results from the experiment

1. Of well over 200 people who have listened to this phrase, in the experiment and elsewhere, none has ever heard anything remotely like ‘at the start we made a pact’, or anything at all to do with a pact, unless that interpretation has first been suggested.

2. When ‘at the start we made a pact’ is suggested, without other context, a certain number of participants believe they hear those words. However, all of these abandon that interpretation when the counter-suggestion is made that the phrase might be ‘It’s fuckin’ payback’. (Please recall this is only a more plausible transcript, not an accurate one.)

3. When participants, before hearing the audio, are given a story about a murder and an allegation there might have been a pact, well over half confidently accept the suggested phrase ‘at the start we made a pact’, with many others hearing something else about a pact.

It can confidently be expected that similar results would obtain for experiments done on other parts of the indistinct audio.

More importantly, it can confidently be expected that the police transcript would have affected the jury’s interpretation of the indistinct recording.

These findings, along with the fact the police transcript is an inaccurate representation of the audio, and observation of the major role given to this audio by the prosecution, suggest to me that this defendant did not have a fair trial.

Sign up here for occasional news about Forensic Transcription (no spam!)