Between 20 and 30 VOT (msec) Is the phonemic boundary, the slope isn’t the same across everywhere.
If the slope was constant, there wouldn’t be a phonemic (category) boundary. How do our two theories
account for this?
Lieberman’s Motor theory: Speech is represented in the mind/brain as articulatory motions (gestural
score), not sounds. Know the gestures required to make speech signal (coarticulation) Understand
speech by determining what gestures produced it. Speech module processes speech and preempts
other auditory processing. There is special speech processing module which allows us to do this. The
reason there is categorical distinction between /ba/ /pa/ is because that’s what it is in the
Auditory Theory of Speech perception: Speech is perceived just like any other sounds. Speech
perception just a reflection of fundamental properties of auditory neurons. E.g., If 2 sounds begin within
20 ms of each other, perceive as simultaneous. If 2 sounds separated by >20 ms, then perceived as
sequential. Hence categorical perception of stop consonants. Categorical distinction occurs outside of
speech as well, and applies to sounds such as beeps. Not specific to sound even, if two lights are flashed
within 10 ms separated one flash will be seen, but if one flashes 30 ms apart, two flashes will be seen.
This is just the way the neurons work. Both theories can account for categorical perception of speech
sounds. Although the motor theory would have trouble understanding why there would be categorical
distinction among lights, the auditory wins because it is more parsimonious and can account for more.
Another finding that Lieberman found is called Duplex perception; take a syllable /da/ /ga/ and slice it at
50 ms. Things in bright orange are the base syllable, the same regardless of whether you got /da/ or /ga/
sound. Extract out that which is common to both sounds; the second part is the part of the sound wave
that corresponds between what distinguishes between the two sounds. In the experiment you play the
base in one ear (the orange part which is the same in both), the other ear will play either the /da/ and
/ga/, what will be heard in the base is da or ga, in the other ear they will hear a chirp. Neither ear got
the whole da or ga. This is a strange phenomenon. Why is this happening, what does the motor theory
and the auditory theory make of this? The motor theory says there is a special module for speech; this
module takes over and integrates the two sounds. If stimuli can be perceived as speech, it will happen.
The auditory system says there is no special speech module, but our perceptual system tries to integrate
information, if our auditory system gets inconsistent information, it tries to integrate it. Duplex
perception for non-speech. Motor theory can account for this but it is positing a speech module which is
Infant Speech Perception: High Amplitude Sucking
Infants can’t use speech to interpret the stimuli they are hearing which goes against the Motor theory.
Acoustic theory if it is the same as the audits, the baby should perceive /ba/ and /pa/ even if they can’t
talk. Researchers have gotten baby to indicate if they hear a stimuli and in a sense treat them as if they
were mice. High amplitude sucking technique VIDEO. First assumption is that babies like new things.
Literally identical stimuli is played (habituation stage; get the baby bored. Eventually the sucking rate
goes down). Then there is dishabituation stage, if baby notices that the second stimulus is different from
the first, then the rate of sucking increases. Then 10 ms cases are played, if the baby notices they are
different than the 0 ms cases, it would be expected the sucking would go up. Do the two sounds cross
the phoneme boundary? Adults would perceive them as different. 1. Background: Novelty preference
2. Habituation Stage: acoustically identical sound repeated.
3. Dis-habituation Stage: New sound presented
4. Does the suck rate change?
There is a difference in stimulus during the second phase. When 20 ms cases are played and then
switched to 40 ms cases, the sucking goes up. What we need to know is if we compared 0 ms and 20 ms
which are on the same categorical boundary, would the baby notice a change? This data suggests that
between 1 and 4 months of age, babies are distinguishing sounds categorically along the same phonemic
boundaries as adults. The Motor theory accounts for the infant data in the following way; gestural
scores are innate. It is a weak theory in this case. Auditory theory says; baby’s nervous system has
developed to the point that just as adult neurons can tell the difference between two stimuli, so do the
baby’s. The motor theory can say that babies are not speaking yet, but gestural scores are innate. Not a
completely crazy idea. What if we now look at a creature that does not, or will never talk. Experiments
with chinchilla; If it hears 0 ms bilabial sound, it can hit a button and it will be rewarded with food, if it
hears a 60 ms bilabial stop, it will not get any food. Wha