Dec 8
=====

Hay/etal:2006a
--------------

The authors say that a distance-based metric using F1 and F2 at a
single point in the diphthong in question (SQUARE/NEAR) doesn't
capture the ways in which the phonemes could be distinct - what would
be a better way this to do this/what have other researchers done
(e.g. taking distances at multiple points in the diphthong, taking
into account length of diphthong)? (p. 467)

What was the motivation for transforming F1 and F2 values into the
Bark Scale? (Could we talk about the usefulness/applications of the
Bark scale?)

It wasn't particularly elaborated on in the paper, but I was wondering
why female listeners responded more accurately to the female voice
than male listeners did (p. 476).

Why did the authors use synthesised speech in the first place if they
were interested in seeing whether there was a perceptual difference in
the way listeners identified certain vowels? I get that it makes it
easier to control for formant values and other acoustic parameters,
but doing so wouldn't remove the naturalness of the stimuli?

What is the applicability of the results found by the paper? What
further accounts within the exemplar theory framework could be
explored based on these results, specifically in addressing issues
related to linguistic stereotypes and bias in social and professional
settings?

How do the authors propose addressing the challenges of diverse
organizational cultures during a merger and its impact on speech
perception, and what practical strategies or recommendations do they
suggest for facilitating effective communication in such contexts?

What could be a reason why close vowels are more likely to be
misidentified when the speaker is female?  Interestingly, the
experimenter has an influence on the performance of the task.

How can you control such things in future experiments? Is it more
advantageous to have multiple experimenters or just one?  What do they
mean by the experiment included four voices and they were not blocked?

It is interesting that the presence of a picture significantly reduced
the likelihood of participants making errors, even though the speaker
was not necessarily the same as the person in the picture.  It's also
worth noting that exposure to other social classes is not necessarily
the same for different social classes (e.g. lower social class when
watching TV) and that the accuracy is higher when the participants
believe that the voice corresponds to an older speaker.

What if the participants were children? Would they be able to perceive
this distinction?

What I found pretty interesting was that the identity of the
experimenters was not meant to be included as an independent variable
in their analysis before the experiment was conducted.  The fact that
this small interaction with the experimenter already had a priming
effect shows us our limited control in an experiment and that there
are so many different factors that have to be taken into
consideration.