[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Research proposal

               State University of New York at Stony Brook
                       Stony Brook, NY 11794-2500

                                            Arthur G. Samuel
                                            516 632-7792
                                            20-Sep-1992 10:24pm EDT
TO:    "Albert Bregman, Tel: 514-398-       (
 _@UBVM.cc.buffalo.edu:AUDITORY@MCGILL1.BITNET )

SUBJECT: RE: Research proposal


I like the general approach you are suggesting -- the "pre-emptiveness" claim of
the speech module theorists is both strong and testable. I think a critical
aspect of the test is one that you referred to, but did not yet have a handle
on: The results are likely to depend on the extent to which the complex is heard
as a "whole". To the extent that your speech bias is successful, you are likely
to induce listeners to hear such good forms, and thus to have less access to the
lowel-level components. Therefore it is critical to induce such a set in the
nonspeech group, and I fear that it may be very difficult to do so, at least to
do so as well as the speech. I have one idea in this regard: Perhaps you could
follow the strategy the Dick Pastore's group, and that Carol Fowler's group,
have taken in the duplex perception phenomenon. You could perhaps construct some
sort of "sine-wave analogs" of a nonspeech sound, such as a particular timbre
contrast (e.g., a cello and a trumpet, filling similar roles to the "ba" and
"da" you might be using in the speech part of the experiment). The idea is that
just as you could induce a speech set for the speech analogs, you could induce a
musical timbre set for the music analogs. Then, you could look for the same sort
of threshold change for the music set as is predicted for the speech set by the
modularists. Of course, if you could really control the nonspeech set to be as
good a form as the speech, this music condition would not be needed. I suggest
it because I am doubtful that such a match will be easy to attain.

This music condition leads me to your question about interpretation. I guess
there are two points here. First, the methodological point raised above can
be thought of as a theoretical point -- if the critical variable is the
"goodness" of the complex percept, then the speech case may be about as good as
it gets. The second point is that from a theoretical perspective of someone like
Fowler, you might get the same effect that All Liberman would predict, but for
rather different theoretical reasons: Fowler would, I think, predict the music
result I've suggested, while Liberman would not.

In any event, I like the direction you are going with this -- good luck.

Arty Samuel