Greg Detre
@12 on 17/10/00
dimensions of the speech production matrix
place + the way in which you restrict the airflow
stopped / fricative etc. � force with which you stop the air
�s� allows some through
place labial, dental, alveolar, palatal, velar
lateral � one side is blocking the airflow
approximate � closing off evenly at the back, but not quite
voiced/unvoiced � whether the vocal folds vibrate or not����� �fffff� vs �vvvvv�
nasal � opening the velum at the back, letting air come through the nose as well as the oral cavity
challenges of speech perception
sounds need to be placed in context
not discrete � parallel transmission and coarticulation
vowels are steady state (smooth, uninterrupted air flow)
each frequency/time line = a formant = a frequency domain in a sound
any sound has multiple frequencies associated
fundamental frequency and harmonies
formant transition from consonant � vowel (di/du)
the formants of a phoneme are affected by what comes after (e.g. du/di)
Lieberman � Motor theory of speech perception
don't just use auditory perception
innate?
using the visual cue to identify the sound
coarticulation complicates the signal too much for the auditory perception system alone to use
need to anticipate person�s auditory gestures
circular
i.e. speech + language is represented in terms of motor signals for speech production
spoken language is heard then decoded into the articulatory gestures that we would need to have made to have produced it
because of the complications of the speech signal, we need an innate human linguistic speech perception
categorical perception � infants capable? is it a linguistic (i.e. innate) skill?
non-monotonic � we don't notice equal acoustic variations
unless they cross category boundaries
chinchillas (S American hairy squirrels) � can categorise + discriminate speech sounds + tone (Crunch(???) + Miller) � and they don't have language
alveola consonsant synthesis � 0 vs 18ms
half of a cage electrocuted
humans may have 2 separate speech and non-speechc perception mechanisms
Massaro + Cohen??? � 3 experiments
1.�� McGurke effect � visual input affects auditory input
3.�� categorical perception experiments � wide range of inputs
but only discrete categories
if give many discrete/continuous reponse choice, then people can discriminate
colours as perceptual categories � parallel with voice perceptual categories
we can perceive colours that we don't have words
is it not possible that certain phenomena are more salient (like the focal colours) and these, unsurprisingly, form the phonemic inventory
a very good argument against innately linguistic speech perception
maybe our auditory system is the innate bit, picking out certain parts
4 year old are worse than adults because they lose infants� universal perceptual ability but you are cognitively deficient compared to adults
�bog� vs �dog�