The first step of the experiment was to have all French-speaking participants record the
stimuli a first time, so as to have control recordings and allow subsequent comparisons
between the pre-training productions and the post-training productions within each group.
The recordings took place in a quiet room at the University of Lille III. The materials used
were a microphone and a computer, and the recording application was the software Audacity.
The subjects were left alone while they were recording to avoid any background noise or
perturbation by other people. In turn, they sat at the computer, and they could see a fullscreen
slide presentation in which the twenty words and sentences appeared one by one, in a
randomized order – different for each subject. The randomizations were done thanks to a
spreadsheet in OpenOffice Calc. The numbers of the stimuli (W01, W02, …, P01, P02, …)
were typed into a table and randomized twenty times – i.e. two recordings (pre- and posttraining)
by ten subjects – with the RAND function. The resulting orders were thus used for
the creations of the twenty slide presentations (OpenOffice Impress).
When the slide presentation was open and the recording application was on, the
instruction first appeared; what the subject had to do was read out the item in the
microphone, and click on the space bar to go to the next one, until the word “the end”
appeared in French (fin). The participants were allowed to read the item mentally before
saying it out loud in the microphone. Thanks to the software Audacity, it was then possible to
divide the recordings of the twenty items into separate sound files (.wav format) for each
stimulus and each speaker. Once all ten subjects had recorded the twenty items, and the
divisions into separate files were completed, there were two hundred sound files,
corresponding to twenty items recorded by ten speakers.