Perception of the same acoustic pattern in different contexts


A spectrogram of a token of the word "say."

Click on spectrograms below to hear them.

Note that "say" consists of a "noise" section, followed by formant transitions, followed by vowel formants.

The formant transitions reflect the release of the fricative gesture. Perceptually, they are integrated with the noise into a percept of a fricative consonant.

If the noise is removed, however, we do not hear "ay", but "day":

Why does this sound like "day"?

The formant transitons are still there, only now they cannot be attributed by the perceptual system to a fricative consonant, so the system assumes that they reflect the release of a stop consonant, which would not have any noise.

In fact, if we add back the noise, but add a silent gap between the noise and the transitions, we hear a fricative followed by a stop ("stay"):

Thus, the same acoustic information (formant transitions) is sometimes perceived as a cue that a stop consonant has taken place, and sometimes it is not perceived that way.

This can be explained if the perceptual system is attempting to interpret these patterns in terms of the gestures (fricative, stop, or fricative+stop) that could have given rise to them.