Last week's post on perceptual restoration in toddlers brought a lot of speculation from commenters. To answer some of the questions, I thought I'd elaborate a bit here on the phenomenon and how I created the demo.
First, here's the original recording again, with me saying "dinosaur" three times:
In the first case, I edited out the "s" sound, and everyone with normal hearing can hear that. The last "dinosaur" is complete. Did I edit out the "s" in the middle dinosaur?
Most adults believe they hear the "s" sound in cases like this, even if the sound has been edited out: the perceptual system adds in a sound where it doesn't exist.
Indeed, more than half of the respondents to our poll said they had heard an "s" sound, even though in fact the sound was edited out in the second "dinosaur."
But some commenters speculated that an "s" sound was embedded in the sneeze sound effect I created, thus nullifying the effect. The "sneeze" was actually a composite of a fake cough and a fake sneeze (neither sounded realistic enough on its own). Do you hear an "s" in either of these sounds?
Personally I'm not hearing it, but I agree that it's closer to an "s" sound than other sounds I could have inserted. In fact the authors of the study do point out that perceptual restoration doesn't always occur; it's less likely to occur when the inserted sound is less like the sound it replaces. As a demonstration, I've redone the demo below, using a horn sound instead of a sneeze:
Now do you hear the "s" in the second "dinosaur"? I certainly don't. So our commenters who argued that the sneeze sounded a little like an "s" do have some grounds for their belief, but I still think that some of what they thought was an "s" sound was in fact the perceptual restoration effect.
You could run a spectrograph on the cough/sneeze sound and see if it contains the profile of "s" (high-frequency fricative). Humans do seem to map language sounds onto non-language sounds based on shared acoustic principles. You might disambiguate the interference sound from the linguistic "s" by using something with a lower frequency.
Yes, I did notice the 's' that seems to be actually missing in third clip. Looks like you should have chosen a horn rather than a fake sneeze/cough (or maybe even a different word, like infor_ation).