Categorical Perception Research Paper

* * Categorical Perception
What is it?
Continuum- something that varies in a particular dimension, results in different perceptions of color
The perception of a continuum in terms of categories
Another way to compensate for the lack of invariance in speech
What is it good for?
Stable perception of a variable signal
Good discrimination between categories
Not hindered by variation within a category
We’ll investigate with experiments:
Is it learned? Innate?
Is it unique to speech?
Is it unique to humans?

Voicing and Voice Onset Time (VOT)
VOT (msec)- when a consonant ends and a vowel starts, when voicing starts of the following vowel, queue for voiced and unvoiced

1. pa (30 msec VOT)- mostly pa, some ba
2. ba (0 msec VOT)
3.
pa (40 msec VOT)
4. ba (10 msec VOT)
5. pa (40 msec VOT)
6. pa (30 msec VOT)
7. pa (50 msec VOT)
8. ba (0 msec VOT)
9. pa (60 msec VOT)
10. pa (40 msec VOT)

25-30 msec VOT- “boundary” some will hear ba and some will hear pa
0-25: perceived unambiguously
After 30: perceived ambiguously

Assessing Categorical Perception
1. Forced choice identification
Subjects have to categorize sound as one thing or another (eg., [pa] or [ba])
2. ABX discrimination
Subjects given two stimuli, A and B, which have different values along the continuum in question
Subjects are then given X, which matches either A or B, and they have to indicate which it matches
If subjects perceive A and B as different things, they are very good at matching X

Categorical Perception in Infants
Are we born perceiving speech categorically?
Testing Method:
High Amplitude Sucking (HAS)
Every time infant sucks, it receives a speech stimulus
When infants get bored with what they’re doing
Habituation Phase
Same stimulus whenever infant sucks
Switch at predetermined threshold
Different stimulus from Habituation Phase
Either Dishabituation (increase sucking rate)
Or continued decrease in sucking
rate

Experimental Method
1 month old infants habituated to an adult [pa] or an adult [ba]
Switched stimulus is either:
Acoustic Change
Phonemic Change

Other findings with infants
Cross-linguistically, infants show a categorical boundary around +30 msec VOT
Prevoicing (eg. Thai)
Infants universally seem to be sensitive to prevoiced/voice distinction
Spanish voicing earlier, near 0 VOT
Spanish-exposed infants sensitive to +30 msec boundary, even though not a boundary in their language

Summary
Changes in VOT are perceived categorically
Infants perceive changes in VOT categorically from very early on
There is also some language dependent learning (e.g. Spanish)

“Ganong” Effect
Categorical perception affected by word status
In-between voicing target d/t e.g. +15 msec VOT
Place between [a]
Perceive as [da]
Place before –ype
Perceive as “type”
Category boundary can be shifted if result would be a word

Categorical Perception Summary
The perception of a continuum in terms of categories
VOT
Formant transitions
Another way to compensate for the lack of invariance in speech
Stable perceptions of a variable signal
Good discriminations between categories
Not hindered by variation within a category