RM: Yes, it’s kind of a version of the Test. I would say it’s a test of what perceptions a speaker could control. If speakers can tell that the repeated sounds are different (when they are – I hope they included some “catch” trials) then they could control them (if they are able to do what’s necessary, articulation-wise, to produce those sounds). I believe such tests have been done to show that speakers of languages that don’t make the “R” vs “L” distinction cannot perceive that difference when it is produced artificially, which is why they can’t reliably control for producing that difference.
RM: I think there are some really interesting observations described in this post. But your explanation of the the failure to completely compensate for the disturbance in the Katseff et al. is not plausible. You say it is a result of conflict between systems controlling articulation and those controlling auditory input. You say:
BN: The error output from control of the intended word (should be head, feels like had) and the error output from control of the intended sound (should be [ 3], sounds like [ɪ]) counter each other, so that the sound in the headphones is somewhere in between hid and head.
RM: So you say that the failure to compensate completely for the disturbance results from a conflict between the system controlling for the “intended word” and the system controlling for the “intended sound”. From the context of the paper it is clear that by “intended word” you mean the intended articulatory feeling. So you explain the speaker’s failure to completely resist the disturbance as a result of conflict between systems controlling two different kinds of variables - sounds and proprioception.
RM: But such a conflict is unlikely, not only because those two variables are different types but because the state of one variable ( sound) depends on the state of the other (articulation). You say the conflict results from the sound controller perceiving hid and the word controller perceiving had so that the resulting sound that is produced is somewhere between these two sounds. But if the “word” controller were perceiving the articulatory pattern for had then that is the sound that would be heard by the sound controller, not hid. If the sound controller hears “hid” it’s because of the changed feedback connection between sound output and sound input.
RM: If the intended sound is head and the perceived sound is something between hid and had then it must be because the speaker is unable to produce the articulatory pattern that will compensate for the change in the feedback function that turns the articulation produced sound output into heard sound. The results of the Katseff et al.study, then, have nothing to do with conflict; they have to do with the fact that the speaker is simply unable to compensate for the disturbing change in the feedback function. I think the speakers in that study are in a situation similar to the one you would be in if you were trying to speak in your normal voice after inhaling helium. Helium changes the feedback connection between the sounds you produce and the sounds you hear. And it’s a change that is impossible to overcome by changing your articulation in a way that would produce a much lower sound if you were speaking using air rather than helium.
RM: An important rule in testing for controlled variables is not to test whether a hypothetical controlled variable is, indeed, controlled using disturbances that could not possibly be resisted if it were.
Best
Rick