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

Re: musical tones in speech



Martin Braun wrote:
>Then, please, let's stick to the facts. You reported in no way that my
>results "do not generalize". The techniques of data analysis you applied
>were insufficient in all respects to detect similar effects as I detected.

Indeed.  Something specific to your data and analysis techniques seems to
be necessary for your effects to appear.  I reported in detail conditions
to which your results did not generalize, and also argued why they were
expected to generalize to those conditions if the effect has any
generality.  My methods are certainly "insufficient" in that they failed to
reproduce your nice results.  I'd be glad to improve them, for example if
you provide me a formal account of how speech targets should be assigned
based on F0 contours (saying it's done by hand does not qualify).  If you
don't, I can only continue to be skeptical.

>Once you have extracted the speech-target f0 values with similar precision
>as the researchers at the IPO and then applied similar statistical
>techniques as I did, FIRST THEN can you come out and say that my results
>could not be replicated with your speech material.
>
>And please note, the question if my results generalize or not will not be
>decided by ONE follow-up study, even if this is carried out in a technically
>correct way. We are likely to see variations in the results of future
>studies of this issue, simply because the tested speech material is likely
>to vary.

In essence you are saying that your claims cannot be challenged unless with
the same speech material and the same hand-marking technique (I presume
employing the same hand-markers).  This can of course be done if IPO lend
the original data.  Note however that this is a very restrictive rule for
testing whether an effect generalizes.  Especially one that is claimed to
have implications as far-reaching as pitch perception and absolute pitch.
Might those implications be true only for Dutch speech pronunced in the
laboratory?

TO THE LIST:  We've been seeing a flurry of highly attractive propositions
that often spark cascades of elaborations.  Does it make no difference to
anyone but myself whether any given proposition is actually true? If so I
guess it'll soon be time to follow Ed Burn's footsteps.  If not maybe it's
time to raise the background level of skepticism a tiny wee bit...


Alain

--------------------------------------------------------------
Alain de Cheveigne'
CNRS/IRCAM, 1 place Stravinsky, 75004, Paris.
phone: +33 1 44784846, fax: 44781540, email: cheveign@ircam.fr
http://www.ircam.fr/equipes/pcm/cheveign
--------------------------------------------------------------