Phonetics, day 2 Oct 3, 2008
Jan 15, 2016
Phonetics, day 2
Oct 3, 2008
Phonetics
1. Experimental
a. production
b. perception
2. Surveys/Interviews
b. perception
What kinds of research questions are asked about speech perception?
1. identification and discrimination
2. categorical perception
3. signal manipulation (phonemic restoration, signal lengthening, native vs. non-native speech)
4. sound recognition in/out of context (cocktail party/white noise/McGurk effect)
1. identification and discrimination
All experiments in speech perception come down to doing one of two things (typically)
1. identification: hear a word/sound and decide what it is.
2. Discrimination: hear two (or three) sounds and decide whether they are the same or different sounds.
identification: Where are the speakers from
1
56
4
32
identification: Where are the speakers from
1
56
4
32
1. 1 2. 2 3. 5 4. 6 5. 2 6. 4 7. 1 8. 3
discrimination
2. Categorical Perception
We perceive sounds in terms of categories—can only hear between category differences, not within category differences.
pp
p
p
pp
bb
b
b
bb b
b
Voice Onset Time Categorical Perception
The time interval between the release of a stop consonant and the onset of glottal vibrations in the following vowel.
b b b b p p p p
10 ms 20 ms 30 ms 40 ms 50 ms 60ms 70 ms 80ms
VOT categorical perception
Humans perceive human speech sounds as belonging to one category or another. They do not perceive them as continuous. But they perceive other sounds as continuous
TOT:
VOT:
a. High Amplitude SuckingInfant sucks on pacifier that is linked to a computer
Used for subjects from 1-6 months
b. Conditioned Head Turn Procedure
Children learn that when stimuli changes they can look somewhere and get a “reward”
3. Signal manipulation
a. Phonemic Restoration
(Warren 1970) It was the *eel that was on the table/axle
gapnormalrestoration
b. Lengthening of Speech Signal
Slower: Regular:
3. Signal manipulation
4. Sounds in/out of context
• out of context• cocktail party• white noise• McGurk effect
4. Sounds in/out of context
Guess the following words:
Words out of context were identified only 47% of the time
b. cocktail party effect
Humans can ignore other voices while focusing on a single person’s voice.
Two Sentences :
Can you hear both sentences?
c. White noise
We can process words better in context than in isolation even in white noise
5. White Noise
We can process words better in context than in isolation
Word in isolation:
Word in Context:
d. McGurk Effect
We use visual and auditory cues in order to understand speech
Why would we do all these experiments?
What kinds of experiments can you think to do with your research and experimental phonetics?
Phonetics
1. Experimental
a. production
b. perception
2. Surveys/Interviews
2. Surveys/Interviews
Go to Dialect survey
In pairs, look at 3 of the questions that examine pronunciation variations—decide
1. Is this a question that would elicit a good response—why or why not?
2. Are the multiple choice answers written in a way that would elicit a good response—why or why not?