Upload
valentine-lambert
View
227
Download
1
Tags:
Embed Size (px)
Citation preview
Sound Categories
Frequency - Tones
Frequency - Tones
Frequency - Tones
Frequency - Tones
Frequency - Complex Sounds
Frequency - Complex Sounds
Frequency - Vowels
bull Vowels combine acoustic energy at a number of different frequencies
bull Different vowels ([a] [i] [u] etc) contain acoustic energy at different frequencies
bull Listeners must perform a lsquofrequency analysisrsquo of vowels in order to identify them(Fourier Analysis)
Joseph Fourier (1768-1830)Time --gt
Frequency
Amplitude
Any function can be decomposed in terms of sinusoidal (= sine wave) functions (lsquobasis functionsrsquo) of different frequencies that can be recombined to obtain the original function [Wikipedia entry on Fourier Analysis]
Frequency - Male Vowels
Frequency - Male Vowels
Frequency - Female Vowels
Frequency - Female Vowels
Schedule
bull Lab 1A ndash Classic speech perception tasksndash individual data collect by Weds Sept 10thndash due Monday Sept 21st
bull Lab 1B - New speech perception tasksndash Task 1 rapid sequence recall (Dupoux et al 2008)ndash Task 2 implicit discrimination (Navarra et al 2005)ndash collect individual data by Thursday Sept 24th ndash email to jbuffintumdedundash group data files available shortly thereafter ndash team analysis welcomeencouragedndash due Friday Oct 2nd
Timing - Voicing
Voice Onset Time (VOT)
60 msec
English VOT production
bull Not uniformbull 2 categories
Perceiving VOT
lsquoCategorical Perceptionrsquo
Discrimination
SameDifferent
Discrimination
SameDifferent0ms 60ms
Discrimination
SameDifferent0ms 60ms
SameDifferent
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
Why is this pair difficult
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
Why is this pair difficult
(i) Acoustically similar
(ii) Same Category
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
Why is this pair difficult
(i) Acoustically similar
(ii) Same Category
A More Systematic Test
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
A More Systematic Test
0ms
20ms
40ms
20ms
40ms
60ms
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
A More Systematic Test
0ms
20ms
40ms
20ms
40ms
60ms
D T
D
T T
D
Within-Category Discrimination is Hard
Cross-language Differences
R L
Cross-language Differences
R L
R L
Cross-Language Differences
English vs Japanese R-L
Cross-Language Differences
English vs Hindi
alveolar [d]
retroflex [D]
Russian-40ms
-30ms
-20ms
-10ms
0ms
10ms
Kazanina et al 2006
Proceedings of the NationalAcademy of Sciences 103 11381-6
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
A More Systematic Test
0ms
20ms
40ms
20ms
40ms
60ms
D T
D
T T
D
Within-Category Discrimination is Hard
Quantifying Sensitivity
Quantifying Sensitivity
bull Response bias
bull Two measures of discrimination
ndash Accuracy how often is the judge correctndash Sensitivity how well does the judge distinguish the categories
bull Quantifying sensitivity
ndash Hits MissesFalse Alarms Correct Rejections
ndash Compare p(H) against p(FA)
Quantifying Sensitivity
bull Is one of these more impressive Harder to obtain by chance
ndash p(H) = 075 p(FA) = 025ndash p(H) = 099 p(FA) = 049
bull A measure that amplifies small percentage differences at extremes
z-scores
Normal Distribution
Mean (micro)
Dispersionaround mean
Standard DeviationA measure of dispersionaround the mean
radic( )sum(x - micro)2
nCarl Friederich Gauss (1777-1855)
The Empirical Rule
1 sd from mean 68 of data
2 sd from mean 95 of data
3 sd from mean 997 of data
Normal Distribution
Mean (micro)655 inches
Standard deviation = 25 inches
Heights of AmericanFemales aged 18-24
Quantifying Sensitivity
bull A z-score is a reexpression of a data point in units of standard deviations
(Sometimes also known as standard score)
bull In z-score data micro = 0 = 1
bull Sensitivity score
drsquo = z(H) - z(FA)
See Excel worksheet
sensitivityxls
Quantifying Differences
(Naumlaumltaumlnen et al 1997)
(Aoshima et al 2004)
(Maye et al 2002)
Normal Distribution
Mean (micro)
Dispersionaround mean
Standard DeviationA measure of dispersionaround the mean
radic( )sum(x - micro)2
n
The Empirical Rule
1 sd from mean 68 of data
2 sd from mean 95 of data
3 sd from mean 997 of data
bull If we observe 1 individual how likely is it that his score is at least 2 sd from the mean
bull Put differently if we observe somebody whose score is 2 sd or more from the population mean how likely is it that the person is drawn from that population
bull If we observe 2 people how likely is it that they both fall 2 sd or more from the mean
bull hellipand if we observe 10 people how likely is it that their mean score is 2 sd from the group mean
bull If we do find such a group theyrsquore probably from a different population
bull Standard Error
is the Standard Deviation of sample means
bull If we observe a group whose mean differs from the population mean by 2 se how likely is it that this group was drawn from the same population
Development of Speech Perception in Infancy
Voice Onset Time (VOT)
60 msec
Perceiving VOT
lsquoCategorical Perceptionrsquo
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
A More Systematic Test
0ms
20ms
40ms
20ms
40ms
60ms
D T
D
T T
D
Within-Category Discrimination is Hard
Abstraction
bull Representations ndash Sound encodings - clearly non-symbolic but otherwise unclearndash Phonetic categoriesndash Memorized symbols k aelig t
bull Behaviorsndash Successful discriminationndash Unsuccessful discriminationndash lsquoStep-likersquo identification functionsndash Grouping different sounds
Letrsquos Learn Inuktitut
Video Nunavik Building on the Knowledge of Ancestors
Vowels
Consonants
Three Classics
Development of Speech Perception
bull Unusually well described in past 30 yearsbull Learning theories exist and can be testedhellip
bull Jakobsonrsquos suggestion children add feature contrasts to their phonological inventory during development
Roman Jakobson 1896-1982Kindersprache Aphasie und allgemeine Lautgesetze
1941
Developmental Differentiation
0 months 6 months 12 months 18 months
UniversalPhonetics
Native LgPhonetics
Native LgPhonology
1 - Infant Categorical Perception
Eimas Siqueland Jusczyk amp Vigorito 1971
Discrimination
SameDifferent0ms 60ms
SameDifferent0ms 10ms
SameDifferent40ms 40ms
A More Systematic Test
0ms
20ms
40ms
20ms
40ms
60ms
D T
D
T T
D
Within-Category Discrimination is Hard
high amplitude suckingnon-nutritive sucking
English VOT Perception
To Test 2-month olds
High Amplitude Sucking
Eimas et al 1971
General Infant Abilities
bull Infantsrsquo show Categorical Perception of speech sounds - at 2 months and earlier
bull Discriminate a wide range of speech contrasts (voicing place manner etc)
bull Discriminate Non-Native speech contrastseg Japanese babies discriminate r-leg Canadian babies discriminate d-D[these findings based mostly on lookingheadturn studies w 6 month olds]
Universal Listeners
bull Infants may be able to discriminate all speech contrasts from the languages of the world
How can they do this
bull Innate speech-processing capacitybull General properties of auditory system
What About Non-Humans
bull Chinchillas show categorical perception of voicing contrasts
PK Kuhl amp JD Miller Science 190 69-72 (1975)
Joan Sinnott U of S Alabama
More recent findingshellip
1 Auditory perceptual abilities in macaque monkeys and humans differ in various ways
2 Discrimination sensitivity for b-p continua is more fine-grained in (adult) humans (Sinnott amp Adams JASA 1987)
3 Sensitivity to cues to r-l distinctions is different although trading relations are observed in humans and macaques alike (Sinnott amp Brown JASA 1997)
4 Some differences in vowel sensitivityhellip
Suitability of Animal Models
2 - Becoming a Native Listener
Werker amp Tees 1984
When does Change Occur
bull About 10 months
Janet Werker
U of British ColumbiaConditioned Headturn Procedure
When does Change Occur
bull Hindi and Salishcontrasts testedon English kids
Janet Werker
U of British ColumbiaConditioned Headturn Procedure
What do Werkerrsquos results show
bull Is this the beginning of efficient memory representations (phonological categories)
bull Are the infants learning wordsbull Or something else
Korean has [l] amp [r]
[rupi] ldquorubyrdquo[kiri] ldquoroadrdquo[saram] ldquopersonrdquo[irumi] ldquonamerdquo[ratio] ldquoradiordquo[mul] ldquowaterrdquo[pal] ldquobigrdquo[sul] ldquoSeoulrdquo[ilkop] ldquosevenrdquo[ipalsa] ldquobarberrdquo
3 - What no minimal pairs
Stager amp Werker 1997
A Learning Theoryhellip
bull How do we find out the contrastive phonemes of a language
bull Minimal Pairs
Word Learning
bull Stager ampWerker 1997
lsquobihrsquo vs lsquodihrsquoandlsquolifrsquo vs lsquoneemrsquo
PRETEST
HABITUATION
TEST
SAME SWITCH
Word learning results
bull Exp 2 vs 4
Why Yearlings Fail on Minimal Pairs
bull They fail specifically when the task requires word-learning
bull They do know the soundsbull But they fail to use the detail needed for
minimal pairs to store words in memorybull
One-Year Olds Again
bull One-year olds know the surface sound patterns of the language
bull One-year olds do not yet know which sounds are used contrastively in the languagehellip
bull hellipand which sounds simply reflect allophonic variation
bull One-year olds need to learn contrasts
Maybe not so bad after all
bull Children learn the feature contrasts of their language
bull Children may learn gradually adding features over the course of development
bull Phonetic knowledge does not entailphonological knowledge
Roman Jakobson 1896-1982
Werker et al 2002
14 17 20
14 months 17 months 20 months
0 60 300 600
Swingley amp Aslin 2002bull 14-month olds did recognize mispronunciations of familiar
words
Dan Swingley UPenn