Noisy Bingo – Acadeam

Overview

A simple demo on top-down mechanisms in speech perception.

Students will experience the effect of context and prior knowledge on speech perception.

Required (in-class):
- Instructor computer, with visual and auditory presentation available.
- Bing sheets
- Markers/A way to mark the sheets

Print out, cut, and ideally, laminate the bingo sheets to allow them to be reused across student sections.

For this activity, we are using noise-vocoded speech, which sounds like a robotic, noisy, voice.
It is hard to comprehend at first, but given proper context to what is being said, most people can pick up these words quite reliably.
Start by playing them a demo sentence, and asking if they understood what as said.
Then show the transcription of what was said, and play it again. Ask if it “made sense” now. Feel free to ask some questions of what they expect is happening here.
Hand out the bingo sheets and markers to students, one per.
Explain the rules of bingo, so that the whole class is familiar with it:
- If they hear a word that is on their sheet, they mark that square.
- A full row, column, or diagonal, is a bingo! They should yell it out.
- Getting every single square is a house!
You will play each word once, they will be read by a noisy, robotic voice, and students should mark their sheets if they think their word was heard.
After all the words have been played, play them again one by one, revealing what the word actually was, and ask students to check if their bingos were ” real”.

The trick here is that none of the words played were ever present in any of the bingo sheets. Instead, an array of similar-sounding one-syllable words were present in their place.
Debriefing the students will involve revealing this ” trick”, and asking if they were surprised by it.
A good point of discussion here is how our brains are working with context to attempt to maximize inference from these new sounds, using context to resolve ambiguity, but how that may also come with trade offs of increasing confidence on ” wrong” inferences.

It is expected that you will get at least a few “Bingos”, and that students will be surprised when you reveal that no one was supposed to have had one at all

This is a really short demo, and so relies a lot on the “deception” aspect of the activity. However, people can be quite reliably trained to get better at understanding noise vocoded speech. A different way to approach this activity/topic would be to have students gain experience understanding this speech, and then “test” them with a real, non-deceptive Bingo.
- In that context, the discussion would be centered around the learning involved, instead of the context cues.

The words clips from this study were originally recorded by a native Canadian English speaker. If you are working at a setting with a noticeably different accent for the general population, it may be ideal to re-record them, even if still in English.

This activity was originally designed by Matt Davis, who kindly sent his draft of it to me upon my request. I later adapted it and re-created stimuli to be used in a Canadian classroom.

I would like to explore the idea of the ” training” version of this task I proposed during the Variations section.