Jan 26
======

Hay/etal:2006b
--------------

What I found especially interesting where the effects concerning
sentence type (if the variant was in the middle of a phrase or
final). I also would have initially agreed with the doubt about
whether Niedzielski's study really only captures speech perception as
I also find the task to be quite difficult (I myself often need to
concentrate well to listen to differences I know exist in speech, and
would assume memory load and attention to the most recent token plays
a role there).

Even if their results show that the "only" effect was that for the
phrase-final variants, all participants respond with a slightly more
Australian variant, this is an interesting finding to me.

Additionally, I still feel like I don't quite get the exemplar theory,
especially how this relates to the quote on p. 372: "Participants are
asked not to identify which distribution is most activated (which is
the usual task in speech perception), but rather which part of a
distribution is most activated."

Have there been any further experiments conducted? How egregious can
the rift in perception and priming be? Examples would be a "rich man"
in very ragged clothing or counting how many times "eh" was uttered by
a supposed Canadian, who is not actually a Canadian.

The priming cutting through the actual knowledge of the fact not being
true (the participants knew the speaker was from New Zealand not
Australia) reminds me of the placebo effect, which also shows effects
even when the participants know they're taking sugar pills which
shouldn't produce any effect.

The experiment conducted by the authors shows that the participants
are influenced by the label at the top of their answer sheets even
though almost everyone of them classified the speaker as being a New
Zealander. How is it possible that the labels have such an influence
if the listeners are apparently all pretty much sure that it does not
correspond to the origin of the speaker?

The authors also mentioned the fact that a speaker adjusts his speech
production to the listener. But what if the speaker was primed by
listening to a recording before? In the context of Jannedy's papers
about Hood German: could it possibly make a difference if the
participants listened to recordings in either High German or Hood
German?

Why participants with lower social class indices that are more likely
to respond as perceiving a higher token number, or more NZ-like /ɛ/
vowel?

How many contradicting clues do the listeners need to discard the
priming information? In the experiments almost all participants
realized the speaker was from NZ. Super fascinating that the results
were still significant. But if the speaker had only a minor accent? Or
if they used an accent that was closer to NZ ?

I thought participants whose response sheet was labeled with
'Australian' and chose more Australian-like variants would have
expected the speaker would be an Australian.
But all bar one indicated that the speaker was New Zealander.
How is this contradiction explainable?

I would expect that just priming the participants with the idea of
Australianism like this study did would activate exemplars less than
an actual voice that they believed to be Australian. Is this true in
later studies?

I'm not sure I understand how exemplar theory explains how
participants might shift to their own speech styles when having to
hold the token longer in memory.

Are female participants better at distinguishing between different
phonetic sounds in general, or are they more aware of their relation
to social groups? i.e. do we also see similar differences when
controlling for social variation?

Speaking on speakers being unaware they produce certain
sounds/associating them with other dialects: My initial reaction is
that this is a product of listeners perceiving these sounds as
prototypical of another accent. I would like to see further studies on
/ae/ and /ε/ (perceived to be less stereotypical) to confirm this.

I think it would also be interesting to see if the same results would
be gathered when studying Australian speakers? Do Australians also
hold these assumed exemplars to the same degree?

Question: It seems that the researchers had too many confounding
variables with the test as they had socioeconomic, regional, and
gender all play major factors in their ability to tell where the
speaker was from. Do you think that they probably tried testing too
many variables at once, what would you have done differently after
reading the paper?

Comment: It would have been interesting to see a group not from AUS or
NZ but from the greater Oceanic region also try this to see if they
can pinpoint the dialect with similar accuracy. If they did test those
groups, if their accuracy wasn't as good then it proves being
neighbors helped hear the difference despite the label, and if their
accuracy was about the same it would be interesting to see how far you
need to be regionally away from a dialect to not recognize the
speakers origin.

In Experimental Methods we have talked about the ratio of fillers to
stimuli being high, which makes me think the number of fillers seems a
bit low?

It seems weird that the authors state that a label would bias the
participants' perception, so they did not explicitly that the label
represented the nationality of the speaker, but as far as I see,
that's the only thing it could represent, so they took this precaution
which to me seems that it doesn't have an effect in the end.

Not a question, but I audibly laughed when I read that the
participants were paid in chocolate.