intonational realisation of topic and focus in child dutch aoju chen max planck institute for...
TRANSCRIPT
Intonational realisation of topic and focus in child Dutch
Aoju Chen
Max Planck Institute for Psycholinguistics
NVFW seminar on ProsodyMPI, Nijmegen, 1 June, 2007
2
Outline
1. Topic-focus and their intonational realisation in Germanic languages
2. Prior work on topic-focus in child language
3. The present study
4. Conclusions
5. Topics for future research
3
Topic and focus at the sentence level
The topic: the entity about which information is provided typically realised as a NP or a pronoun at the sentence level
The focus: the information that is provided about the topic e.g. What did the children do next?
[The children] topic [went to school] focus.
Relationally new
Non-contrastive No, the children [watched TV] focus.
Contrastive/corrective
Relationally given
Non-contrastive
Referentially given
Information focus(presentational focus)
Gundel and Fretheim (in press)
Lambrecht (1994)
4
Intonation of focus Speaker A: Which class do you have today?
Speaker B: I have [physics].
contrastiveContrastive focus(corrective focus)
Information focus(presentational
focus)
Speaker C: I have [mathematics].
Speaker D: No, we have [psycholinguistics]. (C’s classmate)
Gussenhoven (2006)
H*L
5
(Possible) Intonation of topic Deaccented
L*H
H*L
De poetsvrouw eet [een biet]topic.‘The cleaning-lady eats a beet.’
Wie eet een biet? ‘Who eats a beet?’
Lambrecht (1994)
Wie eet een biet? ‘Who eats a beet?’
De poetsvrouw eet [een biet]topic.‘The cleaning-lady eats a beet.’
De pestkop pakt [de vaas]topic. ‘The pest picks up the vase.’
Wie pakt de vaas? ‘who picks up the vase?’
Braun (2006)
Vallduví & Engdahl (1996)Braun (2006) Hedberg & Sosa (2007)
6
Topic and focus in child language
Commonality of the topic-focus structure prior to and in the two-word stage Successive two one-word utterances
e.g. Finger. Touch (Scollon 1976)
Two-word utterances e.g. Federica acqua ‘Federica water’ (child pretends to drink) (D’Odorico 2003)
Use of intonation in expressing topic and focus In the one-word stage
Not expressed intonationally (Bloom 1973)
In the two word stage Accenting the focus and deaccenting the topic (Wieman 1976)
Not expressed intonationally (Chen & Fikkert, to appear)
In the grammatical multi-word speech stage The use of accentuation in marking contrast
7
Topic and focus in child language (cont’d) Contrast + newness in English
A picture-description task Picture 1: A boy is riding a bike Picture 2: A GIRL is riding a bike.
Correct use of ‘contrastive stress’ at the age of 3 and 4
Most frequently in sentence-initial position An increase in the use of contrastive stress in
older children
Contrast + focus in German An imitation task
Child repeats the answer to a WH-questionQuestion: Tomorrow is Eva and Peter’s mother’s birthday… Eva wants to bake cookies. What does Peter back?)Answer: Peter bakes a CAKE.
Higher mean pitch in contrastive focus (337 Hz) than in topic (320 Hz) in German 4- and 5-year-olds’ speech. (Müller et al. 2005) Bigger difference in sentence-final position
0
0,2
0,4
0,6
0,8
1
subject verb object
sentence constituent
% o
f con
trast
ive
acce
nt
1st picture
2nd picture
(Hornby and Hass 1970, MacWhinney and Bates 1978)
8
Topic and focus in child language (cont’d)
No clear predictions on the use of accent placement in marking topic and focus
Conflicting effects of sentence position Sentence final: weaker intonational realisation (H & H 1970)
Sentence final: stronger intonational realisation (Müller et al. 2005)
No information on the use of pitch accent type in the marking of topic and focus Accent type matters in adult language Development of inventory of pitch accent types in child language
9
The present study How do Dutch-acquiring children (>3 yrs) use pitch
accent types and deaccentuation to realise full NP non-contrastive topic and focus?
e.g. Speaker A: What did the boy draw?Speaker B: [The boy]topic drew [a castle]focus.
Are topic and focus typically associated with a certain accent type?
Does the position of topic and focus (sentence initial vs. sentence final) matter?
Do children differ from adults?
Is there development over time?
10
An Answer-reconstruction task (in a picture-matching game)
Participants 4- to 5-year-olds
7- to 8-year-olds
10- to 11-year olds
Adults
Method
Step 1 Step 2 Step 3 Step 4
(N=20, age range: 4;5-5;7, mean age: 5;1)
(N=14, age range: 7;2-8;10, mean age: 7;5)
(N=12, age range:10;3-11;10, mean age: 10;4)
(N=25)
Wat bechermt de vos?
Kijk! Een vos!
11
Method (cont’d) Stimuli
36 question-answer pairs Answers have a fixed word order: SVO 18 pairs with questions about the subject
18 pairs with questions about the object
Each subject NP and each object NP occur in both groups of question-answer pairs but in combination with different VPs and SVs.
e.g. Wat bescherm de vos? De vos beschermt het bos. e.g. Wie beschert de fiets? De vos beschermt de fiets.e.g. Wat steelt de pestkop? De pestkop steelt een fiets.
Initial Focus Final Topic
Final FocusInitial Topic
Who protects the bike?The fox protects the bike.
What does the fox protect?The fox protects the forest.
12
Method (cont’d)
Intonation transcription ToDI + a few additional labels (Gussenhoven 2005)
Annotator 1 Annotating all data 2-3 times with a 2-week interval or longer Blind to the experimental conditions
Annotator 2 Annotating a selection of the data independent of Annotator 1 Checking all labels with doubts from Annotator 1
Disagreements were resolved together
13
Within-subject factors Pragmatic condition (topic, focus) Position (sentence initial, sentence final)
Between-subject factor Age
Dependent variables Mean percentage (%) distributions of
Deaccentuation
H*
H*L
L*H
!H*L
Analyses
14
Results: Adultssentence-initial (N =10)
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istri
butio
ns o
f acc
ent t
ypes
focustopic
Focus Sentence-initial: mostly H*L (78%), H* (13%) Sentence-final: most frequently H*L (46%), !H*L (26%), deaccented (17%)
TopicSentence-initial: mostly H*L (65%), H* (20%)Sentence-final: typically deaccented (62%), !H*L (27%)
sentence-final (N =10)
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
Dis
tribu
tions
of a
ccen
t typ
es
focustopic
15
Interim discussion
Deaccentuation as the default intonation of topic Facilitating the processing of relationally given
information (e.g. Terken and Nootboom 1987, Birch and Clifton 1995)
Facilitating the marking of focus
Why is sentence-initial topic mostly accented? Rhythmically motivated: accent the accentable word
preceding the accent that marks focus within the same IP (Horne 1991, Terken & Hirschberg 1994)
Wat beschermt de vos?
[De vos]topic beschermt [het bos]focus.
1. # H*L L%
2. H*L H*L L%
Pattern 2 was judged to be more pleasant sounding than pattern 1 by native speakers of Dutch.
(Chen 2007)
16
Interim discussion (cont’d)
Consequences for language acquisition
Children need to learn to associate H*L with focus and deaccentuation with topic.
Further, they need to know that topic can be accented for rhythmic motivation when preceding an accent assigned to mark focus in the same intonational phrase.
17
Results: 4- to 5-year-olds
H*
!H*LOTHER
deacc
L*H21%
H*L40%
deacc
L*H9%
H*L49%
H*
!H*L
OTHER
OTHER: H*LH, L*, L*HL
OTHER: H*LH, L*, L*HL, H*LHL, L*HLH
deaccented L*H H*L !H*L H*
Mean % distributions of accent types across conditions
Adults (N=10) 4- to 5-yr-olds (N=20)
18
Results: 4- to 5-year-olds (cont’d)
Deaccentuation: topic > focus, independent of sentence position, contra prior work.
Focus: (1) Sentence-initially, mostly H*L, followed by H*
(2) Sentence-finally, mostly L*H and H*L. (why L*H?)
Topic: (1) similar realisation to focus, with only one difference concerning L*H
(2) realised similarly frequently with L*H, H*L, and deaccentuation
4- to 5-yrs-olds (N=20): sentence-initial
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% di
stribu
tions
of ac
cen t
ypes
focustopic
Sentence-initial (N=20) 4- to 5-yrs-olds (N=20): sentence-final
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istri
butio
ns o
f acc
ent t
ypes
focustopic
Sentence-final (N=20)
19
Results: 4- to 5-year-olds (cont’d) Phrasing
In both topic and focus conditions, an IP boundary occurs more frequently after the sentence-initial NP (subject) than after the verb.However,
an IP boundary occurs after the subject more frequently when it is topical; an IP boundary occurs after the verb (or before the object) more frequently when the subject is focal and the object is topical.
0%
20%
40%
60%
80%
100%
subject verb
Initial focus
Initial topic
Position x Pragmatics:
F (1, 17) = 2.64, p=0.12
S V O
20
4- to 5-yr-olds vs. adults
Sentence-finally, children use H*L similarly frequently in topic and focus; adults use H*L typically to realise focus.
Sentence-initially, children show a weaker preference for H*L over H* in sentence-initial topic/focus than adults.
Children accent sentence-final topic frequently (>85%); deaccentuation is the typical intonation of topic in adult Dutch.
4- to 5-year-olds tend to insert an IP after sentence-initial topic and before sentence-final topic; adults rarely do this.
4- to 5-yr-olds (N=20): sentence-initial
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istr
ibut
ions
of a
ccen
type
s
focustopic
4- to 5-yr-olds (N=20): sentence-final
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istr
ibu
tio
ns o
f a
cce
nt ty
pe
s
focustopic
Adults (N=10): sentence-initial
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istr
ibutio
ns o
f a
cce
nt
type
s
focustopic
Adults (N=10): sentence-final
0%
20%
40%
60%
80%
100%
deacc H*L !H*L
% d
istr
ibutions o
f accent
types
focustopic
21
Interim summary 4- to 5-years-olds have adult-like inventory of pitch accent
types. They are sensitive to the accentuation-focus and
deaccentuation-focus associations, like adults The Lack of effect of sentence position on accentuation
suggest that children under 6 are not sensitive to the role of sentence position in marking topic and focus.
4- to 5-years-olds have not acquired H*L as the typical focus pitch accent and deaccentuation as the default intonation of topic.
They appear to use phrasing as another means to realise topic.
22
Results: 7- to 8-yr-olds7- to 8-yr-olds (N=5): sentence-initial
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istri
butio
ns o
f acc
en ty
pes focus
topic
7- to 8-yr-olds (N=5): sentence initial 7- to 8-yr-olds (N=5): sentence-final
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
focus
topic
7- to 8-yrs-olds (N=5): sentence-final
Adults (N=10): sentence-initial
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
% d
istr
ibu
tion
s o
f acc
en
t typ
es
focustopic
Adults (N=10): sentence-final
0%
20%
40%
60%
80%
100%
deacc L*H H*L H* !H*L OTHER
focustopic
23
Focus
Initial final
deacc L*H H*L H* !H*L deacc L*H H*L H* !H*L
4- to 5-yr 54% 26% 41% 33% 13%
7- to 8-yr 72% 10% 19% 56% 13%
10- to 11-yr 65% 28% 1% 78% 13%
adults 78% 13% 7% 46% 26%
Results: all age groups
24
Topic
Initial final
deacc L*H H*L H* !H*L deacc L*H H*L H* !H*L
4- to 5-yr 22% 47% 18% 32% 27% 26% 11%
7- to 8-yr 10% 53% 23% 64% 7% 25%
10- to 11-yr 8% 46% 44% 36% 23% 39%
adults 4% 65% 20% 62% 6% 27%
Results: all age groups (cont’d)
25
Conclusions Children as young as 4 have adult-like inventory of pitch
accent types in Dutch. Children are sensitive to the accentuation-focus and
deaccentuation-topic associations at the age of 4, as suggested in prior work.
However, children under 6 are not adult-like in intonational marking of topic and focus.
They acquire H*L as the typical ‘focus accent’ and deaccentuation as the typical ‘topic intonation’ at the age of 7 or 8.
Possibly, frequent use of H*L in sentence-initial topic in adult Dutch has made it difficult for young children to extract the functions of H*L and deaccentuation from the input.
Only 4- to 5-year-olds appear to use phrasing to realise sentence-initial topic, which forms its own IP. Older children and adults utter the topic-focus structure mostly as one IP.
26
Topics for future research
How do pitch accent types affect the interpretation of topic and focus by children? Will a deaccented NP create a bias towards a topic
interpretation and an NP with H*L a focus interpretation?
How do pitch accent types affect the processing of given vs. new information in 4- to 5-yr-olds? Will their overuse of accentuation mask the facilitating
effect of pragmatically appropriate intonation?
Can listeners tell the topic apart from the focus in the sentence initial position? Any differences in the phonetic realisation of H*L?
27
Special thanks
Information Structure in Language Acquisition
Christine DimrothLaura HerbstWolfgang KleinBhuvana NarasimhanSarah Schimke
Alice Kruisselbrink (former student-assistant)Marieke Hoetjes Steven Rekké
28
Thank you!