FYI: Forensic Transcription Experiment
Some list-members may remember last year hearing about or participating in
(thanks) an experiment about transcription of hard-to-hear audio used as
evidence in legal cases.
I am now doing a new experiment. It is a bit different form last year's, using more
difficult material, and seeking to investigate a different kind of question - but still
quite interesting I think, to those who like this kind of thing.
I would be most grateful if any list-members were able to do the experiment.
Some of you may wish to give the experiment to students, or to friends and
acquaintances. That is great but please do be aware it is suitable for adults only
(due to coarse language and the murder theme). I recommend you look through
it yourself first before passing it on.
I enclose below the info pages from the experiment, so you will know in advance
what it is about. If you feel able to give it a go, you can access it via the link
Thanks in advance!
Legal cases increasingly use evidence in the form of covert recordings, obtained
by undercover listening device or similar means.
Sometimes these recordings are of very poor quality, to the extent it is hard to
make out the words that are spoken.
This raises the problem of how to be sure the jury hears the evidence accurately.
Current legal practice (in Australia) is to play a hard-to-hear recording once in
court, then make it available for further listening in the jury room. However, there
are no specific directions as to how many times a jury should listen, and indeed
no way of knowing how many times they actually do listen, if at all.
In fact, though it is known that, in general, listening several times can aid
perception, there has been little research on the effect of repeated listening. The
present experiment aims to investigate this.
It asks participants to imagine they are on the jury as they listen to hard-to-hear
audio from a real murder trial (now concluded), then correlates their accuracy in
hearing the audio with the number of times they listened.
The experiment is quite demanding. The audio is 4 minutes in duration, and very
hard to hear. You will probably need to listen several times to make it out, plus
there is other evidence about the murder to read. People who enjoy nutting out
criminal cases generally find it interesting.
If you are willing to participate, the next page has a detailed information and
consent form. Please be sure you know exactly what the experiment involves
before giving your consent.
Dr Helen Fraser (independent researcher)
P.S. if you have any questions before starting please feel free to ask via
Information and Consent:
Thank you for your interest in this experiment.
Ethics requires that we start with this page of information to be sure you know
exactly what the experiment is about and how it will work.
Please read the following carefully, especially underlined parts, and then we can
The experiment is being conducted by Dr Helen Fraser, an independent
researcher in the field of forensic phonetics. You can find more information about
Dr Fraser at helenfraser.com.au
Aim of the Study:
This study is about hard-to-hear audio used as evidence in criminal cases. It
aims to investigate how many times a jury needs to listen to this kind of material
before it becomes fully intelligible.
This study is suitable for those over 18 years only. It involves listening to audio
from a real murder trial, which contains coarse language. You can withdraw at
any time if you find the material unpleasant in any way, but it is better not to begin
if you feel you may be affected.
The experiment should take on average about 30 minutes. However, you might
want to set aside a longer time in case you find you would like to explore the
material in more detail.
You need a good connection to the Internet throughout the experiment, and a
quiet location. It is best if you can use headphones, though this is not essential.
The experiment is designed for native speakers of English. Proficient non-native
speakers of English are welcome to participate but please indicate your language
background in the appropriate place.
Your participation is completely anonymous, but you will be asked to fill out
some demographic questions (about your age, language and education
background, etc) at the end.
Responses to questions will sometimes involve written words, and sometimes
simple yes/no or rating scale responses. Written responses may be quoted in a
future report on the results, but there is no means of associating any response
with any individual, as individuals cannot be identified through their participation.
It is anticipated the research will be completed by July of 2013. The results may
be presented at conferences or written up for publication.
Participation is completely voluntary. If you decide to participate, but then change
your mind, you are free to stop at any time simply by leaving the website.
It is unlikely that this study will raise any personal or upsetting issues, but if it
does you may wish to contact your local Community Health Centre.
If you have any questions or concerns, please email the researcher, Dr Helen
Fraser on firstname.lastname@example.org. Contact details will be given again at the
Request for confidentiality:
As you will see, it is important to start the experiment with no preconceptions, so
please do not discuss this experiment with anyone who has not already done it!
P.S. We will show you the contact information again at the end of the experiment.