Auditory Stroop Task Tutorial
This tutorial assumes you’ve already followed the simple Stroop using the PsychoPy builder tutorial. It will guide you through modifying the simple Stroop task to use audio as described in, for example, Most et al. (2007) & Kestens et al. (2021)
In this tutorial you will learn to:
create high quality audio stimuli (voice recordings or tones) for use in perception experiments
load lossless audio recordings into PsychoPy for presentation to listeners
collect data either in-person on the phonetics lab computers or online using pavlovia.org
Open source audio manipulation tools
We use a range of audio tools in the phonetics lab for different tasks. The ones we recommend everyone in the lab learn to use are:
- Praat: https://www.fon.hum.uva.nl/praat/
- Audacity: https://www.audacityteam.org/download/
- SoX: (commmand line audio manipulation) https://sourceforge.net/projects/sox/
- Cecilia 5: (scriptable audio processing with customizable UI) https://github.com/belangeo/cecilia5
Background: Auditory Stroop Task & Emotion
Development and use of Auditory Stroop Task
Creating a PsychoPy experiment is only the end result of a lot of reading, thinking, careful analysis of previous work, and synthesis with previous results. As Mary Beckman likes to say, linguists should spend less time writing and more time reading. Here, for this tutorial, is an example of what a page of my research notes might look like when thinking about a task. I started collecting these references because I was thinking about exemplar models and gender identity, but accidentally discovered a connection to emotion along the way.
- Hamers and Lambert (1972) : French (basse et haute) and English (low and high) words are presented in high and low frequency tones. Complicated result, worth reading carefully, but responses to incongruent stimuli are slower. See also: (Cohen and Martin 1975; Spapé and Hommel 2008)
This is the same Lambert as the classic first Matched Guise study: Lambert et al. (1960)!
Green and Barber (1981) : girl1 and man spoken by male and female talkers show an auditory stroop effect. confusing result alongside the explanation in, e.g. Walker and Hay (2011) or Palmeri et al. (1993). Why would these words have gendered expectations?
Morgan and Brandt (1989) : Auditory stroop effect for pitch and loudness but not duration. Compare with memory results in (Nygaard et al. 1995; Bradlow et al. 1999)
Most et al. (2007) : Another gender study, but kids and adults suggest an opposite pattern for gendered words (e.g. bracelet, lipstick, pirate, football) vs gendered names (e.g. Amy, Jenny, Brian, George) with kids having more interference for words and adults having more interference for names. Again, not obvious that exemplars predict this. See also Christensen et al. (2011) for neural investigation with gender.
Knight and Heinrich (2017) & Kestens et al. (2021) : Methodological insights, implementation details, and relevance to speech in noise. The implementation in this tutorial is largely inspired by these two.
Emotion & Speech Perception
Clone and open your Stroop task
Make a copy of your stroop.psyexp file and rename it auditory-strop.psyexp. Then open it in PsychoPy Builder (File → Open → Choose auditory-stroop.psyexp)
Designing the audio stimuli
Most et al. (2007) assume that gender is a simple binary (Campbell-Kibler and miles-hercules 2021)
The instructions provided for setting up and playing-back audio in PsychoPy
Before we write the actual experiment, let’s change some default settings:
In the Experiment settings window (Basic tab), enter stroop next to Experiment name and remove the session option in the Experiment info box. In the Data tab, make sure the data is saved as sub-{nr}. Finally, in the Screen tab, make sure the experiment uses your own monitor (which you created in an earlier tutorial) and set the Units to “norm”.
Testing
Upload to Pavlovia (optional)
Next Steps
References
Footnotes
chosen because it’s monosyllabic↩︎