Faculty of Arts University of Groningen
The acquisition of the weak-strong distinction
and the Dutch quantifier allemaal
Erik-Jan Smits ([email protected])
Bart Hollebrandse
SiN-day; November 25, 2004
SiN-day, November 2004
2
The acquisition of quantification;the classical picture and the yes-answer
• Question:
Is every farmer feeding a donkey?
• Possible answers:
(1) No; pointing at the donkey
(2) Yes; all the farmers are feeding a donkey (adult answer)
(3) Yes; many donkeys are fed by a farmer
Crain et al. (1996)
SiN-day, November 2004
3
The weak-strong distinction and the acquisition of quantification
• Weak-strong distinction (Milsark, 1979):
There are {many, few, *all, *every} doctors in the room
• Geurts (2003): no experimental data, but:
“the grammatical connection between a quantifier and its domain of quantification is less rigid in children than it is in
adults” (footnote 3, p. 10). (cf. Philip (1995), Drozd and many others)
SiN-day, November 2004
4
Analyzing quantified sentences
In order to interpret a quantified sentences, one should:
1. Correctly localize the domain of the relevant quantifier (or: determine its scope)
2. Correctly interpret the domain of the relevant quantifier (or: determine the nature of the quantifier)
SiN-day, November 2004
5
The Dutch quantifier “allemaal” (1)
• With respect to the correct localization of the domain, “allemaal” is able to quantify over subject or object:
1. Een jongen draagt de koffers allemaal
A boy is carrying the suitcases all
“A boy is carrying all the suitcases”
2. De jongens dragen allemaal een koffer
The boys are carrying all a suitcase
“The boys are all carrying a suitcase”
SiN-day, November 2004
6
The Dutch quantifier “allemaal” (2)
• With respect to the correct interpretation of the domain, “allemaal” is unique because its ambiguity between a strong and a weak quantifier:
1. Een jongen draagt de koffers allemaalA boy is carrying the suitcases all“A boy is holding all the suitcases”
allemaalstrong (A,B) is true iff ||A|| ||B||
2. Er fietsen allemaal papegaaienThere are bicycling all parrots“There are bicycling allemaal (many) parrots”
allemaalweak (A, B) is true iff ||A|| ||B|| | 2 |
SiN-day, November 2004
7
Experimental design
Hypothesis: Difficulties with understanding quantified sentences can not only be found in children unable to correctly localize the domain of the quantifier, but also in children unable to correctly interpret the domain of the quantifier (i.e. a consequent strong or weak reading)
Aim: Distinguish children with an adult-like quantifier system from children with a weak quantifier system.
Two experiments:– Scope-experiment:
Is a child able to make a distinction between “allemaal” quantifying over the subject or object?
– Weak-strong experiment:Is a child able to make a distinction between a weak and strong use of “allemaal”?
SiN-day, November 2004
8
General prediction
A child that is always interpreting a quantifier as a weak one in the weak-strong experiment, regardless its syntactic position, will judge a significantly higher amount of sentences as true in the scope-experiment (regardless the fact whether the subject or object is within the domain of the quantifier) than the child always understanding a quantifier as a strong one
SiN-day, November 2004
9
The scope-experiment
• Predictions:– A child is always quantifying the subject
– A child is always quantifying the object
– A child is a spreader: quantifying over both the object and the subject
• Method: Truth Value Judgment Task – also questioning the yes-answer.
• 39 kids (aged 4 – 6)• 3 items per 2 conditions; 3 no-fillers (total 15 sentences)
SiN-day, November 2004
10
The scope-experiment – test items (1)
• Een paard draagt de meisjes allemaal
A horse is carrying the girls all
object Q
• De mannen dragen allemaal een ezel
The men are carrying all a donkey
subject Q
SiN-day, November 2004
11
The scope-experiment – test items (2)
• Een robot houdt de ballonnen allemaal vast
A robot is holding the balloons all PART
object Q
• De mannen tillen allemaal een kist op
The men are lifting all a box up
subject Q
SiN-day, November 2004
12
Results scope-experiment
• Two groups:1. Adult answer (no) (26)
2. Non-adult answer (yes) (13)
Yes-answer scope-experiment
0,00%
20,00%
40,00%
60,00%
80,00%
100,00%
0 5 10 15 20 25 30
Number of subjects
Ye
s-a
ns
we
r
Non-adult
Adult
SiN-day, November 2004
13
Results scope experiment – domains of quantification
Domains of quantification strong quantifying kids
0%
10%
20%
30%
40%
50%
SQ
OQ
Spr
eadi
ng
Wor
ld
Unk
now
m
SQ
OQ
Spr
eadi
ng
Wor
ld
Unk
now
n
OQ-sentences SQ-sentences
Per
cen
tage
yes
-an
swer
SiN-day, November 2004
14
Results scope experiment – domains of quantification
Domains of quantification weak quantifying kids
0%
10%
20%
30%
40%
50%
60%
70%
SQ
OQ
Spr
.
Wor
ld
Unk
now
n
SQ
OQ
Spr
.
Wor
ld
Unk
now
n
OQ-sentences SQ-sentences
Per
cen
tage
yes
-an
swer
SiN-day, November 2004
15
The weak-strong experiment
• Prediction:
Children differ in their interpretation of a quantifier as:– A weak one
– A strong one
– A weak or strong one depending on its syntactic position – the adult analysis
• 39 subjects (aged 4 – 6)• Method: Truth Value Judgment Task – also questioning the
yes-answer• Total of test sentences: 18 (12 test items, 3 no-fillers, 3 yes-
fillers)
SiN-day, November 2004
16
The weak-strong experiment – test items
• De ezels huilen allemaal
The donkeys crying all
(Strong;3 items
3 items with “alle”, all)
• Er dansen allemaal meisjes
There are dancing many girls
(Weak – 6 items)
SiN-day, November 2004
17
Results weak-strong experiment• Prediction I: children with only a strong reading of allemaal
• Prediction II: children with only a weak reading of allemaal
Yes-answer weak and strong condition
0%
20%
40%
60%
80%
100%
0 10 20 30 40 50
Yes-answer weak/strong
SiN-day, November 2004
18
Results weak-strong experiment (2)
• Prediction III: children with an adult reading (expected: yes-answer in the weak-condtion, no-answer in the strong-condittion)
Yes-answers weak and strong condition
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25 30 35
Yes-answer 'weak-condition'Yes-answer 'strong' condition
SiN-day, November 2004
19
Scope and the weak-strong distinction
• General prediction:
A child that is always interpreting a quantifier as a weak one in the weak-strong experiment, regardless its syntactic position, will judge a significantly higher amount of sentences as true in the scope-experiment than the child always understanding a quantifier as a strong one
SiN-day, November 2004
20
Percentages yes-answer scope-experiment vs. weak-strong experiment
0,00%
20,00%
40,00%
60,00%
80,00%
100,00%
120,00%
1 2 3 4 5 6 7 8 9 10 11 12 13
Scope exp.
Weak-strong exp.
General results; scope and the weak-strong-distinction
Exp. I Exp. II
Strong 26 35
Weak 13 4
SiN-day, November 2004
21
Conclusions
Experiment 1:• The data shows that there are children that have a weak reading
for a universal strong quantifier (13 out of 39).
Experiment 2:• Children have a preference to analyze allemaal as a strong
quantifier, in a situation in which not all the subjects are participating (35 out of 39).
In general:• Children that have a weak quantifier system can only be
discriminated from children that have an adult quantifier system by experiments taking the weak-strong distinction into account.
• Problems with quantification are more widespread than previously thought.