Dec 1 ===== Johnson:2006a ------------- "People (perhaps especially men) perform gender." (Johnson, page 3) How can gender as an overall concept be perofmred by only one subgroup of the concept? If only one subgroup performs the concept, and no other subgroup does anything or barely anything related to the concept, does the original subgroup even perform the concept? Or does it present something else that just appears to be related to the concept? One might not directly experience "nonconscious brain states or events", but I believe that what one does experience is highly dependent on it, so that these things are experienced indirectly. The levels of certain brain chemicals and hormones change ones entire view of a situation and ones reaction thereof (being related to the question of the existence of free will), so with this in mind, nonconscious brain states or events are still a part of every experience a person can have. "watching the ‘‘male’’ cartoon character the voice sounds male, but with eyes closed and reminded of the fact that the actor is female the voice sounds female." Does this still have to do with the phonetic characteristics of the voice, when one has to actively remember that the speaker is female to have it sound female? Does this knowledge perhaps come from the additional world knowledge one possesses, instead of any actual stereotypical phonetic characterisitcs of a female voice? How sure can we actually be that the research tested for gender and not sex, or other factors? I have several ideas follow-up studies: showing a photograph of female / male 'speakers' while playing the stimuli (Does the visual cure enforce the auditory? What if they mismatch?), speakers whose sex does not match their gender, record the listeners when they repeat the stimulus and see whether the gender of the stimulus speaker influences their production (especially when the stimulus gender does not match the participant's gender). Why might differences between female and male speakers have evolved in the first place? In the final part of the discussion, the „power of suggestion“ is mentioned relating to the speaker of Bart Simpson. I wonder how much of the effect shown in the paper would still hold if the listeners would see the speaker (and the speaker presents traditionally male or female). Maybe to some degree this would speed up response times (for the repetition task) for those speakers that were previously identified less stereotypical, as it gives further validation - but I could also imagine it would cause further delay or „confusion“ if the speaker sounds very un-stereotypical and the appearance confirms this apparent mismatch. The paper mentions the notion of normalization several times as a probable artifact in speech perception studies. It’s not clear to me what exactly the author understands by normalization. Does it have something to do with allophones/phonemes dichotomy or some gender classification that we perform when processing speech samples? In line with that, I found interesting the findings “ listeners took longer to begin repeating the words produced by the nonstereotypical talkers ” (p.6) and the conclusion drawn that “ Linguistic material produced by voices that do not fit [gender] expectations is not processed as efficiently as is material produced by stereotypical voices. ” (p.6). I’m wondering, whether it could be that it's not a perception difficulty (word recognition) but social task expenses: we have to make an extra effort in identifying gender? The paper refers to an experiment in which several listeners were asked to name words that were produced by stereotypical and nonstereotypical talkers. I was wondering how it is possible that it took the listeners longer to identify the words of the nonstereotypical talkers since I would assume that the expectations of the listensers concerning the voice of the talkers do not have an impact on the comprehensibility of the words. Would the results for gender variability and recognition possibly change over time as well, depending on which year we look at? How is resonance implemented and how is it able to take into account non-auditory aspects of language? And how does it change when running multiple cycles of the model? How do we define auditory distance for exemplar activation? The paper seems to subtract spectrogram representations of speech, but how are we determining how alike one spectrogram is to another? Is it by some sort of edit distance formula? I would like to know what the specific performance of male and female vowel formant frequencies is and how they distinguish the gender. If formants are predictable from height, could weight also be a predictor for some specific formants, as well? Which part of speech organs is responsible for F3 such that it is predictable from height? The vocal folds? on p. 2 we see the table with formant difference between vowels produced by men and women across 17 languages. Could the diversity of the results be somehow connected with cultural background of speakers? Are there studies examining the question at what age people acquire "phonological gender representation"? Does this happen after adolescents' voice breaks, or does it not depend on it? I'm not sure I fully understand exemplar activation pattern. My understanding is that in the model hearing a voice speak a token "activates" recognition memories of other voices of the similar perceived gender saying the same token. Is this correct? I would also like to discuss the normalization algorithm and how it differs from the exemplar activation pattern. The experiment of identifying the gender of stereotypical and nonstereotypical male and female voices resulted in identifying more correct genders for nonstereotypical male voices than for stereotypical ones. I assume this experiment was done by English speakers and listeners only. So my question is, would it not make more sense to make a similar experiment with different langauges to have a bigger overview? Or maybe using more stimuli would make a difference in the result. What cultural factors could cause some languages to have greater gender difference? (Graph p.2)