1. Sound Categories

2. Frequency - Tones

3. Frequency - Tones

4. Frequency - Tones

5. Frequency - Tones

6. Frequency - Complex Sounds

7. Frequency - Complex Sounds

8. Frequency - Vowels
Vowels combine acoustic energy at a number of different frequencies
Different vowels ([a], [i], [u] etc.) contain acoustic energy at different frequencies
Listeners must perform a ‘frequency analysis’ of vowels in order to identify them (Fourier Analysis)

9. Time --> Amplitude Frequency
Any function can be decomposed in terms of sinusoidal (= sine wave) functions (‘basis functions’) of different frequencies that can be recombined to obtain the original function. [Wikipedia entry on Fourier Analysis]
Time -->
Joseph Fourier ( )
Amplitude
Frequency

10. Frequency - Male Vowels

11. Frequency - Male Vowels

12. Frequency - Female Vowels

13. Frequency - Female Vowels

14. Schedule Lab #1A – Classic speech perception tasks
individual data: collect by Weds Sept 5th
due Monday Sept 12th
Lab #1B - New speech perception tasks
Task 1: rapid sequence recall (Dupoux et al. 2008)
Task 2: implicit discrimination (Navarra et al. 2005)
collect individual data by Monday Sept 17th– to
group data files available shortly thereafter – team analysis welcome/encouraged
due Monday Sept 24th

15. Timing - Voicing

16. Voice Onset Time (VOT)
60 msec

17. English VOT production
Not uniform
2 categories

18. Perceiving VOT
‘Categorical Perception’

19. Discrimination
Same/Different

20. Discrimination
Same/Different
0ms ms

21. Discrimination
Same/Different
0ms ms
Same/Different

22. Discrimination
Same/Different
0ms ms
Same/Different
0ms ms

23. Discrimination Same/Different 0ms 60ms Same/Different 0ms 10ms

24. Discrimination Same/Different 0ms 60ms Same/Different 0ms 10ms

25. Discrimination Same/Different 0ms 60ms Same/Different
Why is this pair difficult?
0ms ms
Same/Different
40ms 40ms

26. Discrimination Same/Different 0ms 60ms Same/Different
Why is this pair difficult?
0ms ms
(i) Acoustically similar?
(ii) Same Category?
Same/Different
40ms 40ms

27. Discrimination A More Systematic Test Same/Different 0ms 60ms
Why is this pair difficult?
0ms ms
(i) Acoustically similar?
(ii) Same Category?
Same/Different
40ms 40ms

28. Discrimination A More Systematic Test Same/Different 0ms 60ms 0ms 20ms

29. Discrimination A More Systematic Test D D D T T T Same/Different
0ms ms
0ms
20ms D
20ms
40ms T
Same/Different
0ms ms T T
40ms
60ms
Same/Different
Within-Category Discrimination is Hard
40ms 40ms

30. Cross-language Differences

31. Cross-language Differences

32. Cross-Language Differences
English vs. Japanese R-L

33. Cross-Language Differences
English vs. Hindi
alveolar [d]
retroflex [D] ?

34. Russian
-40ms
-30ms
-20ms
-10ms
0ms
10ms

35. Kazanina et al., 2006
Proceedings of the National Academy of Sciences, 103,

36. Discrimination A More Systematic Test D D D T T T Same/Different
0ms ms
0ms
20ms D
20ms
40ms T
Same/Different
0ms ms T T
40ms
60ms
Same/Different
Within-Category Discrimination is Hard
40ms 40ms

37. Quantifying Sensitivity

38. Quantifying Sensitivity
Response bias
Two measures of discrimination
Accuracy: how often is the judge correct?
Sensitivity: how well does the judge distinguish the categories?
Quantifying sensitivity
Hits Misses False Alarms Correct Rejections
Compare p(H) against p(FA)

39. Quantifying Sensitivity
Is one of these more impressive? Harder to obtain by chance?
p(H) = 0.75, p(FA) = 0.25
p(H) = 0.99, p(FA) = 0.49
A measure that amplifies small percentage differences at extremes z-scores
Both yield the same difference between p(H) and p(FA). But the second one is harder to obtain by chance.

40. √( ) Normal Distribution Dispersion around mean Standard Deviation
A measure of dispersion around the mean.
Mean (µ)
√( )
∑(x - µ)2 n
Carl Friederich Gauss ( )

41. The Empirical Rule 1 s.d. from mean: 68% of data

42. Normal Distribution Standard deviation Heights of American
 = 2.5 inches
Heights of American
Females, aged 18-24
Mean (µ)
65.5 inches

43. Quantifying Sensitivity
A z-score is a reexpression of a data point in units of standard deviations. (Sometimes also known as standard score)
In z-score data, µ = 0,  = 1
Sensitivity score d’ = z(H) - z(FA)

44. see sensitivity worksheet sensitivity.xls

45. Quantifying Differences

46. (Näätänen et al. 1997)
(Aoshima et al. 2004)
(Maye et al. 2002)

47. √( ) Normal Distribution Dispersion around mean Standard Deviation
A measure of dispersion around the mean.
Mean (µ)
√( )
∑(x - µ)2 n

48. The Empirical Rule 1 s.d. from mean: 68% of data

49. If we observe 1 individual, how likely is it that his score is at least 2 s.d. from the mean?
Put differently, if we observe somebody whose score is 2 s.d. or more from the population mean, how likely is it that the person is drawn from that population?

50. If we observe 2 people, how likely is it that they both fall 2 s. d
If we observe 2 people, how likely is it that they both fall 2 s.d. or more from the mean?
…and if we observe 10 people, how likely is it that their mean score is 2 s.d. from the group mean?
If we do find such a group, they’re probably from a different population

51. Standard Error is the Standard Deviation of sample means.

52. If we observe a group whose mean differs from the population mean by 2 s.e., how likely is it that this group was drawn from the same population?

53. Development of Speech Perception in Infancy

54. Voice Onset Time (VOT)
60 msec

55. Perceiving VOT
‘Categorical Perception’

56. Discrimination A More Systematic Test D D D T T T Same/Different
0ms ms
0ms
20ms D
20ms
40ms T
Same/Different
0ms ms T T
40ms
60ms
Same/Different
Within-Category Discrimination is Hard
40ms 40ms

57. Abstraction Representations Behaviors
Sound encodings - clearly non-symbolic, but otherwise unclear
Phonetic categories
Memorized symbols: /k/ /æ/ /t/
Behaviors
Successful discrimination
Unsuccessful discrimination
‘Step-like’ identification functions
Grouping different sounds

58. Let’s Learn Inuktitut!
Video: Nunavik: Building on the Knowledge of Ancestors

59. Vowels
Consonants

60. Three Classics

61. Development of Speech Perception
Unusually well described in past 30 years
Learning theories exist, and can be tested…
Jakobson’s suggestion: children add feature contrasts to their phonological inventory during development
Roman Jakobson, Kindersprache, Aphasie und allgemeine Lautgesetze, 1941

62. Developmental Differentiation
Universal Phonetics
Native Lg. Phonetics
Native Lg. Phonology
0 months
6 months
12 months
18 months

63. #1 - Infant Categorical Perception
Eimas, Siqueland, Jusczyk & Vigorito, 1971

64. Discrimination A More Systematic Test D D D T T T Same/Different
0ms ms
0ms
20ms D
20ms
40ms T
Same/Different
0ms ms T T
40ms
60ms
Same/Different
Within-Category Discrimination is Hard
40ms 40ms

65. high amplitude sucking non-nutritive sucking

66. English VOT Perception
To Test 2-month olds
High Amplitude Sucking
Eimas et al. 1971

67. General Infant Abilities
Infants’ show Categorical Perception of speech sounds - at 2 months and earlier
Discriminate a wide range of speech contrasts (voicing, place, manner, etc.)
Discriminate Non-Native speech contrasts e.g., Japanese babies discriminate r-l e.g., Canadian babies discriminate d-D [these findings based mostly on looking/headturn studies w/ 6 month olds]

68. Universal Listeners
Infants may be able to discriminate all speech contrasts from the languages of the world!

69. How can they do this? Innate speech-processing capacity?
General properties of auditory system?

70. What About Non-Humans?
Chinchillas show categorical perception of voicing contrasts!
PK Kuhl & JD Miller, Science, 190, (1975)

71. Suitability of Animal Models
More recent findings…
Auditory perceptual abilities in macaque monkeys and humans differ in various ways
Discrimination sensitivity for b-p continua is more fine-grained in (adult) humans (Sinnott & Adams, JASA, 1987)
Sensitivity to cues to r-l distinctions is different, although trading relations are observed in humans and macaques alike (Sinnott & Brown, JASA, 1997)
Some differences in vowel sensitivity…
Joan Sinnott, U. of S. Alabama

72. #2 - Becoming a Native Listener
Werker & Tees, 1984

73. When does Change Occur? About 10 months Janet Werker
U. of British Columbia
Conditioned Headturn Procedure

74. When does Change Occur?
Hindi and Salish contrasts tested on English kids
Janet Werker
U. of British Columbia
Conditioned Headturn Procedure

75. What do Werker’s results show?
Is this the beginning of efficient memory representations (phonological categories)?
Are the infants learning words?
Or something else?

76. Korean has [l] & [r] [rupi] “ruby” [kiri] “road” [saram] “person”
[irumi] “name”
[ratio] “radio”
[mul] “water”
[pal] “big”
[s\ul] “Seoul”
[ilkop] “seven”
[ipalsa] “barber”

77. #3 - What, no minimal pairs?
Stager & Werker, 1997

78. A Learning Theory…
How do we find out the contrastive phonemes of a language?
Minimal Pairs

80. Word Learning
Stager & Werker ‘bih’ vs. ‘dih’ and ‘lif’ vs. ‘neem’

82. PRETEST

83. HABITUATION
TEST
SAME
SWITCH

84. Word learning results
Exp 2 vs 4

85. Why Yearlings Fail on Minimal Pairs
They fail specifically when the task requires word-learning
They do know the sounds
But they fail to use the detail needed for minimal pairs to store words in memory
!!??

86. One-Year Olds Again
One-year olds know the surface sound patterns of the language
One-year olds do not yet know which sounds are used contrastively in the language…
…and which sounds simply reflect allophonic variation
One-year olds need to learn contrasts

87. Maybe not so bad after all...
Children learn the feature contrasts of their language
Children may learn gradually, adding features over the course of development
Phonetic knowledge does not entail phonological knowledge
Roman Jakobson,

88. Werker et al. 2002
14 months
17 months
20 months 14 17 20 60
300
600

90. Swingley & Aslin, 2002
14-month olds did recognize mispronunciations of familiar words
Dan Swingley, UPenn

91. Alternatives to Reviving Jakobson
Word-learning is very hard for younger children, so detail is initially missed when they first learn words
Many exposures are needed to learn detailed word forms at early stages of word-learning
Success on the Werker/Stager task seems to be related to the vocabulary spurt, rapid growth in vocabulary after ~50 words

92. So how do infants learn…?
Some possibilities: ‘Use it or lose it’ – they stop paying attention to contrasts that they don’t need for the ambient language Minimal pairs (e.g., rock vs. lock) – requires word meanings Acoustic distributions of sounds, requires no word knowledge Seeking contextually conditioned variation, e.g., Korean r/l contrast

95. (Dietrich, Swingley, & Werker 2007)

96. Exp 1: tam - ta:m
Exp 2: tæm - tæ:m
Exp 3: ta/æm - tem
Length factor ~

98. (Dietrich, Swingley, & Werker 2007)

99. Slides: Swingley 2006, ICIS

100. Slides: Swingley 2006, ICIS

101. Slides: Swingley 2006, ICIS

102. 5 hours’ exposure to Mandarin
± human interaction
[2003, Proceedings of the National Academy of Sciences]

104. Alveo-palatals
affricate
fricative

107. Jessica Maye, Northwestern U.

110. Infants at age 6-8 months are still ‘universal listeners’, cf
Infants at age 6-8 months are still ‘universal listeners’, cf. Pegg & Werker (1997)
Infants trained on bi-modal distribution show ‘novelty preference’ for test sequence with fully alternating sequence
How could the proposal scale up?

116. p(a) = p(b)
p(a) = 2 x p(b)

117. 1.0 .5
.25 .1

120. Slides: Swingley 2006, ICIS

121. Slides: Swingley 2006, ICIS

123. Fenson et al. 2000

124. toast
hat
ants
tooth
table
television
blanket
outside
plant
wait
today
fast
hurt
soft
out
stroller
kitty
water
babysitter
pretty
patty cake
bottle
kitchen
don’t
night (night)
bird
dog
duck
doll
bread
candy
head
dish
radio
outside
feed
today
dark
MacArthur Short CDI - 89 items

125. Fei Xu, Berkeley

126. Xu & Carey 1996
10 mo.: no surprise
12 mo.: surprise
--> “10 month olds do not represent basic sortal/kind concepts”
Xu 2002
Add words!
9 mo.:

128. Fulkerson & Waxman 2007 12 months 6 months
Categorization measured by novelty preference score: % looks to novel / total;
categorization should imply novelty preference
12 months 6 months
Words µ = .59, p = .007 µ = .63, p < .001
Tones µ = .53, p = .2 µ = .54, p = .2

130. Yeung & Werker 2009 Naturally produced Hindi syllables
Dental vs. retroflex
Familiarize sound-object links
Test sound discrimination only
Exp1: consistent links
Exp2: inconsistent links
Effect of Type (±alternating)
Exp1: F(1,18) = 5.74, p < .05
Exp2: F(1,18) = 0.53, p = .47

131. (Feldman, Griffiths, & Morgan, 2009)

132. “Simulations demonstrate that using information from segmented words to constrain phonetic category acquisition allows more robust category learning from fewer data points, due to the inter- active learner’s ability to use information about which words contain particular speech sounds to disambiguate overlapping categories.” (Feldman et al. 2009)

134. Analysis Hypothesis testing (null HT vs. Bayesian)
Linking probabilities to hypotheses
Weighted binomial distributions

136. Questions
Combining words and sound distributions: what do learners need to know?
Too-many-colors problem: how to combine across words?
Why does interaction matter (Kuhl et al on Mandarin)?
What does this predict about 1-year olds’ knowledge of phonological contrast?
What changes between 12 & 18 months?

138. Invariance
(Jusczyk 1997)

139. Training on [g-k] or [d-t], generalization across place of articulation. (Dis-)habituation paradigm.
[Maye & Weiss, 2003]

140. So how do infants learn…?
Phoneme categories and alternations
Perhaps more like a phonologist than like a LING101 student - look directly for systematic relations among phones
Gradual articulation of contrastive information encoded in lexical entries
Much remains to be understood

141. Abstraction in Infant Speech Encoding
From a very early age infants show great sensitivity to speech sounds, possibly already with some ‘category-like’ structure
Although native-like sensitivity develops early (< 1 year), this should be distinguished from adult-like knowledge of the sound system of the language
Children still need to learn how to efficiently encode words (phoneme inventory)
Children presumably still need to learn how to map stored word forms onto pronunciations (phonological system of the language)
Popular distributional approaches to learning the sound system address rather non-abstract encodings of sounds, at best

143. More Issues…
Is there distributional evidence for contrasts in the input?
Maye et al.: children can learn
Werker et al.: demonstration from Japanese/English maternal speech
How well does this scale beyond duration (1-dimensional)?
Child needs to store all exemplars
Child needs to know all relevant dimensions
This could yield at most phones, not phonemes
Why do children fail on minimal pair learning?
Inaccurate representations, qualitatively different representations
Hard tasks
Fennell et al.: context helps

144. Questions about Development
Change from 6-12 months
What changes?
Structure changing vs. structure adding
What causes change to occur?
Statistical distributions of sounds
Reliably separable distributions?
Storing and organizing tokens for analysis
Knowing appropriate acoustic dimensions
Allophony, e.g., k-palatalization in English
Why does it take so long?
Change from months
(Skepticism about the effect)

145. 6-12 Months: What Changes?
(clunky diagram, from Phillips, 2001)

146. Structure Changing
Patricia Kuhl U. of Washington

147. Structure Adding
Evidence for Structure Adding (i) Some discrimination retained when sounds presented close together (e.g. Hindi d-D contrast) (ii) Discrimination abilities better when people hear sounds as non-speech (iii) Adults do better than 1-year olds on some sound contrasts
Evidence for Structure Changing (i) No evidence of preserved non-native category boundaries in vowel perception

148. Sources of Evidence Structure-changing: mostly from vowels
Structure-adding: mostly from consonants
Conjecture: structure-adding is correct in domains where there are natural articulatory (or acoustic) boundaries [cf. Phillips 2001, Cogn. Sci., 25, ]