Jan 5
=====

Strand:1999
-----------

From an ethical point of view, can one include stereotypes in
representations and still make sure that no harm is done?  I.e. many
stereotypes are actually harmful to the perspective groups.  Also, as
stereotypes change over time and differ from person to person, how can
one take into account the specific stereotypes a participant in a
study holds to be able to account for them within the representational
value of the study?

I totally see how the Face Gender effect is in conflict with modular
models of perception. In the conlusion, Strand states that "any model
of speech perception that is rigidly modular with a quasi symbolic,
discretely categorized output just above the level of sensation, with
gender stereotypes then acting on this output, will not work" - are
there any sociolinguistics/researcher then who try to explain
interspeaker variability in perception with (adjusted) modularity
models?

There was mentioned the study of Zebrowitz and Montepare (1992)
devoted to face perception. It is interesting, how the situation with
stereotypes can vary across languages (taking into account the
cultural aspect, appearance difference, etc.)?

"longer vocal tracts (i.e., those typical of men) induce boundaries at
lower frequencies, and shorter vocal tracts (i.e., those typical of
women) induce boundaries at higher frequencies" (p.90) Could this
stereotype have developed from physiological differences between men
and women? I could imagine that men produce their S and SH sounds with
lower frequencies than women to match the lower frequencies of their
vowels compared to women's vowels.

Also, especially with Hay & Drager (2007) in mind, both papers seem to
focus on gender differences, where I could well imagine other social
categories to be decisive.

While dealing with the "Face Gender Effect", the author explains that
participants were presented a set of voices which were either
prototypically female, nonprototypically female, prototypically male
or nonprototypically male sounding. Which aspects cause a voice to be
perceived as being prototypical for one gender or not?

- How do we differentiate between sex and gender stereotypes?
 
- Is using stereotypes triggered by visual features of faces in speech
  processing similar to reading subtitles of a film one is watching?
  Both raise expectations to the following speech signal. Or do
  subtitles fall into the same category of general expectations
  regarding how the speaker will continue?
 
- Would a similar effect be triggered if people were given imaginary
  names - which might also trigger sterotypes - to the audio samples
  they're hearing?

Does face perception belong to the category of social expectations
about gender?

Are there studies that investigate the differences between female and
male production of the fricatives in other languages or in
non-American English speaking groups that underline the evidence of
the study mentioned in the paper?

1. From what I understand, normalization is typically viewed as a
phonetic-phonological process of "deciphering" phonemes from a speech
signal as affected by some social and physiological factors. In the
studies covered by Strand (1999), the listeners are claimed to perform
this normalization based on only auditory or visual cues. However, in
real speech it seems to be a rare case that listeners would rely on
phonetic information solely to make sense of an utterance. It appears
to me that the actual normalization happens on a semantic level,
rather than on phonetic one. Otherwise, a listener would have a hard
time trying to understand what a non-stereotypical voice in the
absence of visual cues is saying.

2. The notion of stereotypicality is also an interesting point. In the
field of TTS where visual cues are not accessible and in some cases
context isn't too reliable (e.g. names of stops in public transport),
are there any standards or criteria for speaker selection with respect
to their typicality?

Gender stereotypes: can we suggest a possible relationship between
gender and age in speech perception? Is the role of gender in speech
perception observable in children, and can this role increase or
decrease with age?

Johnson2006a found that stereotypical talkers are processed more
efficiently than non-stereotypical ones - would we e.g. find that
reaction times for nonstereotypical male talkers on token 4 would be
longer than stereotypical? Would priming with a male face reduce that
time (and a female face increase it)?

-  In the first experiment from Strand & Johnson 1996: did they
   synthetically alter the frequency of the onset consonant, and the
   gender of the speaker is evident to the participant based on the
   vowel, and the participant normalizes based on that?
   
- Naslund 1993- I'm assuming that they found that younger children
  produced the fricative in a manner independent of gender?

1. No direct question, but I do have commentary on the face perception
research below.

Pg 16: The Face Perception research reminded me of Japanese Noh masks
which have very distinct facial features and were used in plays, and
those features are associated with types of people in media in Japan
when trying to attribute specific feelings to the person even if they
haven't spoken. For instance the Kumasaka mask is meant to represent a
criminal and they're usually balding with rough facial hair a large
nose and big eyes (I attached an image but if you look up "kumasaka
mask" you would get the same result). So people with these features in
media are generally assumed to have the same type of stereotypes
placed upon them like how the babyface effect was put onto people in
Berry and McArthurs (1985) study. For more info:
http://www.historyofmasks.net/famous-masks/noh-mask/