PSYC3311 Lecture Notes - Lecture 5: Speech Perception, Phoneme, Spoken Word
How to explain speech perception in the light of the variability (not always 1-1 matching) and
imprecision that exists in the acoustic signal?
- and we don't pronounce every single letter, often we merge so imprecision in signal but
we can still understand whats being said
Analysis-by-synthesis (Stevens & House, 1972)
On the basis of the one-to-one mappings, a hypothesis is generated (synthesised) as to what
the phoneme is. ur synthesizing a candidate. This hypothesis can then be confirmed on the
basis of whether it fits with the context in which it occurs. That is, does it make a word, which
fits the sentence context?
Active process, constantly doing it
America^
Similar idea developing in england --> which provides more current
Logogen model (Morton, late 60s early 70s)
A parallel activation account, with each word represented by an "information/activation
collecting device" = a logogen
When signal comes in, water is poured.
Bottom up activation. info extracted from stimulus and fed into the system
Water lvl rises for relevant logogens
And then u understand k sound so then more water poured into kat
Until it overflows- that’s the threshold
Then that word becomes available for recognition
It becomes a hypothesis as to what the word was
Can be metaphorically conceptualised as each logogen as a test-tube, each represents a word
with water (i.e., information extracted from the acoustic signal) being poured into it.
Once a logogen reaches threshold, it becomes a candidate (hypothesis) for recognition.
The frequency effect: The threshold might be reached more quickly for high frequency words
than low frequency words because of a higher resting level of activation.
So mb higher frequency words have smaller length test tube, or water pours in more easily into
high frequency. or every logoen has resting lvl of activation – so there's already water in the test
tube
Mb words leave residue water in there every time so more water left behind cos more frequency
Context can have its effect at one or both of two loci:
Pre-access –prior to u recognising the word eg domestic animals. so u can pre activate all
relevant words so they have higher resting lvl.
- Context is used to pre-activate logogens that are relevant (top-down processing) so
that they reach threshold more quickly than they otherwise would.
find more resources at oneclass.com
find more resources at oneclass.com
