Jan 26
======

Pierrehumbert:2006a
-------------------

The paper states that "speech patterns of older people were viewed
[...] as a window into the past history of language".  While I believe
that to be true to some extend, I think it still is a matter of how
much the speech of a person has been adjusted to more modern speech,
based on how much contact a person has with other/younger people etc.

I confused aoubut the the existence of double dissociations between
the perception and production systems, why it leads to the conclusion
that these systems are separate but highly coupled in normal adults.

I found this paper quite dense and difficult to read, especially
without really knowing much about Varbrul's generative model except
for the short section in the beginning.  However, I did end up
recognizing quite a few names and studies we talked about previously
in the seminar, which I found very cool.

The paper deals with exemplar theory and thereby explains several
aspects that need to be considered in order to represent the way in
which human beings process language. Since the processing of language
is such a complex action and since there are so many components
concerning speech processing which are not clarified yet, in how far
is it even possible to build a model that covers all of the relevant
aspects sufficiently?

Not really a question, but I found the paper rather hard to
read/understand in detail. In case it won't be covered in the talk,
could we maybe talk through how a hybrid of exemplar models and
Varbrul works? In particular, I find it hard to imagine the relation
between different levels of representations.

The author states that "Varbrul was never [...] billed as a cognitive
model". Could we call exemplar theory a cognitive model? From what
I've read, I would probably say that yes, it tries to model a
cognitive process.

According to the paper one example for exaggerating the effects of
gender would be if women  lower the formants to sound more like men
and signal more authority. I don't understand how the exaggeration of
older people is meant? Lowering the pitch? And how is this beneficial
in a social context?

How would we integrate exemplar theory and Varbrul? Is the idea that
we need to add thresholds and phonological boundaries to the frequency
data that’s captured by exemplar theory?
 
When discussing how the sample size for social category learning is
much smaller, did they already consider the effects of bias and
stereotype-formation? It would make sense to me that people can still
have strong social categorization beliefs without a huge sample size
due to reinforcing already-held beliefs.

I would guess that perception of social categories for someone else
would be quite categorical, but perception of one’s own social group
belongings might vary a lot more. Is this something that both of these
theories also support?

My question is regarding phonotactics and lexical neighbors. I don't
understand how high frequency phonotactics and words with a high
amount of lexical neighbors can simultaneously aid and hinder in
speech perception.

Are there modern hybrid exemplar models that the authors were
suggesting in the paper, that are superior to both the standard
standard phonological and exemplar models? To what extent?

Comment: This paper, while very interesting, took several re-reads and
some googling to fully understand.

Question: While the Varbrul model does come with limitations relating
to gradience, many of the items brought up with Exemplar models would
arguably be hard to define on a scale that has yet to fully be
explored (e.g. sexual identity on a scale doesn't necessarily have the
entire spectrum researched). While it would be ideal to use Exemplar
model in more places, application of it seems theoretical at
times. How has the process developed since 2006, if you have done
further research into the topic?

At the end the author mentions that social situations cause emotion
which affects attention and memory, but I can't really imagine how to
control for that.  I don't understand what is meant by "the cumulative
force of the exemplars" in the frequency section.