A Psycholinguistic Perspective on Child Phonology

1
A Psycholinguistic Perspective on Child Phonology
Laboratoire de Sciences Cognitives et
Psycholinguistique, EHESS-CNRS,
Paris www.lscp.net/persons/peperkamp

• Sharon Peperkamp
• Emmanuel Dupoux

Second International Conference on Contrast in
Phonology. Toronto, May 3-5, 2002
2
Child phonology

• Jakobsonian view of phonological acquisition
• Childrens productions
• reflect their internal grammar
• evolve from universally unmarked structures
  towards the marked structures present in their
  language

3
Outline of talk

• Discuss problems with child phonology
• Propose a model of phonological processing,
  featuring
• a distinction between perception and production
• a distinction between phonology and phonetics
• Show how in this model
• acquisition can take place
• data of child phonology can be accounted for

4
I. Child phonology
5
Problem 1 Initial state

• Childrens earliest productions
• unmarked structures
• - consonant p
• - vowel a
• - syllable structure CV
• Infants perceptual capacities between 0 and 2
  months
• marked structures
• - pa vs. ba (Eimas et al. 1971)
• - a vs. i (Trehub 1973)
• - V vs. CV vs. VC ?

? Are initial representations really unmarked?
6
Problem 2 Preverbal acquisition
Experimental perception data

• 6 months vowels (Kuhl et al. 1992 Polka Werker
  1994)
• 9 months phonotactics
• (Friederici Wessels 1993 Jusczyk et al.
  1993, 1994)
• 10-12 months consonants (Werker Tees 1984)

? Is grammatical knowledge really reflected in
production?
7
Two time paths in acquisition
Perception
universal marked
adult-like
0
12 months
4 years
universal unmarked
adult-like
Production
8
Two time paths in acquisition
Perception
universal marked
adult-like
0
12 months
4 years
universal unmarked
adult-like
Production
? Do children acquire one or two grammars?
9
Problem 3 Speed of acquisition

• Acquisition of syllable structure (Fikkert 1994)
• Elke age onsets rhymes
• 16.4 stops, nasals fricatives
• 16.25 glides stops
• 17.22 nasals
• 19.24 fricatives
• 111.28 liquids
• 20.11 stopglide
• 23.27 stopliquid
• 24.15 sstop

? Why does acquisition take so long?
10
Problem 4 Inter-individual variability

• Production of 3-syllable tokens by twins
• (Savinainen-Makkonen 2000)
• 15 16 17
• Annika 0 100 100
• Antti 0 27 37

? Why do children follow different time paths?
11
Problem 5 Gradual changes

• Consonant harmony by Trevor (Pater Werle 2001)

? Why is learning gradual?
12
Problem 6 Intra-individual variability

• type variability
• frequent words resist correction (Storkel
  Gierut 2001)
• consonant harmony in French child language
• Hélène, 48.7 très k?? very - trou t?u
  hole
• ? Why is there a frequency effect?
• token variability
• non-frequent words show free variation
• consonant harmony in English child language
• Trevor, 21.5 truck tr?k kr?k
• ? Why is there type-specific variability?

13
II. A theory of phonological acquisition
14
Phonological processing
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
/ti/
/ti/
phonological decoding
phonological encoding
thi?
thi?
phonetic encoding
phonetic decoding
thi?
t?h?i??
15
Initial state
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding ?
phonological encoding ?
phonetic encoding ?
phonetic decoding identity
16
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
Step 2
phonetic encoding
phonetic decoding
Step 3
Step 1
17
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
phonetic encoding
phonetic decoding
Step 1
18
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
bottom-up algorithms - segments (Kuhl et al.
1997 Maye, Werker Gerken in press) -
syllable structure (Peperkamp 2001)
phonetic encoding
phonetic decoding
19
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
phonetic encoding
phonetic decoding
0-12 months
20
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
Step 2
phonetic encoding
phonetic decoding
0-12 months
21
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
bottom-up algorithms - allophonic rules
(Peperkamp Dupoux 2002) - neutralizing rules
(Peperkamp Dupoux 2002) - stress (Dupoux
Peperkamp 2002)
phonological decoding
phonological encoding
phonetic encoding
phonetic decoding
0-12 months
22
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
12 months - ??
phonetic encoding
phonetic decoding
0-12 months
23
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
12 months - ??
phonetic encoding
phonetic decoding
Step 3
0-12 months
24
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
12 months - ??
trial and error algorithms (Boersma 1999)
phonetic encoding
phonetic decoding
0-12 months
25
Perception Production underlying
form underlying form Phonology surface
form surface form Phonetics universal
phonetic motor plan representation acoustic
signal acoustic signal
phonological decoding
phonological encoding
12 months - ??
phonetic encoding
phonetic decoding
1-4 years
0-12 months
26
Dual route for phonetic encoding
surface form
retrieval from storage
assembling rules
motor plan
(Levelt 1989)
27
Stage 1
Creation of simplifying assembling rules
surface form
?
t?, k? ? k?
motor plan
very t?? ? k?? assembled hole t?u ?
k?u assembled
28
Stage 2
Storage of patterns
surface form
t?, k? ? k?
t?? ? k??
motor plan
very t?? ? k?? assembled hole t?u ?
k?u assembled
29
Stage 2
Storage of patterns
surface form
t?, k? ? k?
t?? ? k??
motor plan
very t?? ? k?? retrieved from
storage hole t?u ? k?u assembled
30
Stage 3
Correction of assembling rules
surface form
t?, k? ? t?, k?
t?? ? k??
motor plan
very t?? ? k?? retrieved from
storage hole t?u ? k?u assembled
31
Stage 3
Correction of assembling rules
surface form
t?, k? ? t?, k?
t?? ? k??
motor plan
very t?? ? k?? retrieved from
storage hole t?u ? t?u assembled
32
Stage 4
Correction of stored patterns
surface form
t?, k? ? t?, k?
t?? ? t??
motor plan
very t?? ? k?? retrieved from
storage hole t?u ? t?u assembled
33
Stage 4
Correction of stored patterns
surface form
t?, k? ? t?, k?
t?? ? t??
motor plan
very t?? ? t?? retrieved from
storage hole t?u ? t?u assembled
34
Summary phonetic encoding and childrens
productions

• Why does acquisition take so long?
• articulation is a complex motor skill
• Why do children follow different time paths?
• the phonetic encoder is acquired by means of
  trial and error algorithms, which predict
  variation among individuals
• Why is learning gradual?
• motor skills are learned with ups and downs
• Why is there a frequency effect?
• during the acquisition of the phonetic encoder,
  the correction of stored patterns is more
  difficult for more frequent patterns
• Why is there type-specific variability?
• there is competition both between and within the
  two routes, the outcome of which depends on both
  linguistic (e.g. utterance length) and
  extra-linguistic factors (e.g. fatigue)

35
Conclusion

• Models of phonological acquisition should take
  both perception and production data into account.
• Phenomena of child phonology reflect the
  developing phonetic encoder.
• The phonetic encoder is partly non-linguistic it
  should, therefore, be modeled within a
  psycholinguistic framework.