Академический Документы
Профессиональный Документы
Культура Документы
1. Introduction
1.1. Cognitive Developments in Quantification
Research has shown that infants can represent both numerical quantity and
quantity along continuous dimensions (see Feigenson, 2007 for a review). For
example, infants dishabituate to a change in number when arrays are controlled
for continuous variables such as contour length and surface area (Xu & Spelke,
2000; Xu & Arriaga, 2007). Conversely, infants can discriminate between
differences in continuous dimensions, noticing changes in area or volume
(Cordes & Brannon, 2008; vanMarle & Wynn, 2011; Hespos, Dora, Rips, &
Christie, 2012). However, the ability to apply discrete quantification to
continuous dimensions may not be available to prelinguistic infants.
Several studies have shown that infants fail to represent portions of noncohesive substances as individuals and to associate number with these portions
(Huntley-Fenner, Carey, & Solimando, 2002; Rosenberg & Carey, 2006;
Hespos, Mikklia, Rips, Chang, & Dora, 2011; Chiang & Wynn, 2000). For
example, infants tested in a violation of expectancy paradigm (Huntley-Fenner,
Solimondo, & Carey, 2002) failed to track two portions of sand poured at
distinct locations behind separated occluding screens. Infants were not surprised
by the magical disappearance of one portion of sand when the occluders were
removed. In contrast, infants succeeded in tracking the same number of
perceptually similar sand-pile-looking objects being lowered behind the
occluding screens. In a follow-up, using a simpler design without occluders,
Rosenberg and Carey (2006) asked whether infants would dishabituate when
one stationary sand pile was changed to two. Prior to habituation, infants were
offered tactile experience with a sand pile; one group was exposed to a regular
sand pile and another group to a sand-pile-looking object. Only the group
exposed to the object dishabituated. This finding again shows that infants do not
pay attention to number for non-cohesive substances, despite the fact that they
Authors: Jinjing Wang, Johns Hopkins University, jwang185@jhu.edu; Peggy Li,
Harvard University, pegs@wjh.harvard.edu; Susan Carey, Harvard University,
scarey@wjh.harvard.edu. We would like to thank Iris Lee, Yujia Li, Xiaodi Chen, Ruthe
Foushee, and Pierina Cheung for assistance in data collection.
Forty-five 3- to 5-year-olds (16 3-year-olds, M = 3;6, SD = 4 months; 12 4year-olds, M = 4;5, SD = 3.8 months; 17 5-year-olds, M = 5;5, SD = 3.1
months) participated. All children were native English-speakers, recruited either
by phone and brought into the laboratory, or recruited and tested onsite at a local
museum in the Boston area.
Children were tested individually. They were first introduced to two
characters (Mr. Elephant and Mr. Lion). On each trial, each character was
paired with a tray containing sand piles or sets of sand-filled cups (see Table 1).
Children were asked to compare who has more on four trial types (see Table 1,
A-D). On some trials they were asked, Who has more sand?, and on others
they were asked, Who has more cups? Trial type A was to establish that
children understood the question Who has more sand? using a ratio (4:1) that
even infants could discriminate (Hespos et al., 2012). The other trial types made
Exp. 1 Ratios
Exp. 3 Ratios
Who has
more sand?
Who has
more sand?
3:2
(3-yr-olds)
4:3
(3-yr-olds)
4:3
(4- & 5-yr-olds)
5:4
(4-yr-olds)
% Correct
D
100
90
80
70
60
50
40
30
20
10
0
Who has
more sand?
Who has
more cups?
A.
sand
vs.
B.
<
p = .02
=
p = .57
sand
vs.
C.
chance
sand
vs.
D.
cups
vs.
Figure 1. Results for the quantity judgment task of Exp. 1. All bars are above chance
(ps < .05), and p-values indicate the results from paired tests.
An ANOVA did not yield effects of age, or gender. The only significant
effect was trial type, displayed in Figure 1. Children performed near ceiling
when asked Who has more sand? using a 4:1 ratio (trial type A). They
performed worse when the piles had a harder discriminable ratio (trial type B),
and performance was not any better when the same amount of sand was put in
equal-sized cups (trial type C). In contrast, children performed better and nearceiling when asked Who has more cups? Experiment 1 therefore replicated
Huntley-Fenner (2001)s findings that preschoolers are not using the number of
containers to quantify non-cohesive substances, despite the fact that they know
how to compare the number of cups.
3.
two; see Table 2), and the trials within each block was randomized. Half of
the participants were tested on the one block before the two block and vice
versa. There were in total 24 test trials.
Four practice trials prior to the test trials familiarized participants to the
idea that when what Polly wanted was not in the open boxes, it must be in the
closed box. The closed box was the correct choice for two of the trials. These
trials asked children to pick on the basis of number, color, and/or shape (e.g.,
two yellow circles). All children performed correctly on these trials. Unlike
the test trials, children were allowed to open the closed box after each trial.
% Correct
Table 2. Choice of boxes containing all or partial match and their respective
utterances for Exp. 2. Cups = upside-down trapezoid, and triangles = piles. Dark colors
= sand, and light colors = rice.
One Block
Two Block
Display
All Partitive
Noun
Display
All
Number Coordination
Utterance
Match mismatch mismatch Utterance Match mismatch mismatch
one cup
two cups
of sand
of sand
one pile
two piles
of sand
of sand
one cup
two cups
of rice
of rice
one pile
two piles
of rice
of rice
100
90
80
70
60
50
40
30
20
10
0
3yr-olds
4 yr-olds
chance
All Match
One
Partitive
Mismatch
Noun
Mismatch
All Match
Two
Number Coordination
Mismatch
Mismatch
e.g.$One$cup$of$sand
e.g.$Two$cups$of$sand
Figure 2. Results for the language comprehension task of Exp. 2. Sample trials
depicted, with check marks indicating the correct box.
For the test trials, the adults performed as expected, averaging 97% correct
across all trials. Figure 2 plots the percentage correct for the children by trial
type. Four-year-olds performed better than 3-year-olds. Unlike 3-year-olds who
were often at or below chance for the partial match trial types, 4-year-olds
performed above chance. Children were best on the all-match trials, followed by
the number-mismatch and noun-mismatch trials, and then the coordinationmismatch and partitive-mismatch trials. For the majority of the trials on which
children erred (i.e., 98%), they were selecting the open box with some attribute
that partially matched the utterance. In sum, Experiment 2 shows that children
are working out measure phrases that coordinate number, partitive, and noun at
the same age that other studies have reported children are working out measure
word meanings (e.g., Srinivasan et al., in press; Huntley-Fenner, 1995).
4.
reference parts and portions (three parts of a table; three piles of tables). In
Mandarin, the surface structure is the same. For example, the distinction
between three tables (san zhang zhuozi) and three sections of a table (san jie
zhuozi) differ only in the classifier (zhang vs. jie), making the difference easy
to miss. In fact, Li, Huang, and Hsiao (2010) showed that three-year-old
Mandarin-speaking children tend to ignore the classifier (e.g., amongst choices
of a strawberry, a box of strawberry, and a pile of strawberry, children tend to
pick the one strawberry regardless of the classifier that the experimenter used).
If learning quantificational language plays a role in treating portions of noncohesive substances as units, such delays in language acquisition might be
accompanied by delays in conceptual development. That is, if Mandarin learners
are delayed relative to English speakers in mastering measure phrases, they
might also be delayed in their ability to coordinate numerical and continuous
quantification in thought. On the other hand, if conceptual development
proceeds unaffected by the acquisition of language, Mandarin learners should
perform similarly to English learners when it comes to spontaneously using
containers to quantify substances, despite not having mastered the language.
4.1.
Method
partitive and sand as the noun. The number in the measure phrase was either
one or three. Like Experiment 2, children had to assess whether the measure
phrase matched a display (Is this one cup of sand? or
?,
see Table 3). The displays consisted of different numbers of cups in which
sometimes all were filled (Full Cups), sometimes all were empty (Empty Cups),
and sometimes some were filled while others were left empty (Mixed Cups).
The Full Cups trials, consisting of trials in which children should be able to
answer correctly, served as the control condition to verify that children did not
have a bias to always respond yes or to always respond no. Compared to
Experiment 2, the expected yes trials were like the all-match trials because the
display matched all the three components of the measure phrase. The no trials
were like the number-mismatch trials, in which even 3-year-old children
performed above chance. The Empty Cups trials were like the noun-mismatch
trials of Experiment 2, where the content of the (empty) cups mismatched the
noun (sand). We therefore expected some children to make the same kind of
error as Experiment 2, by accepting the measure phrase when the number of
cups in the display matches (see Table 3, asterisked trials of Empty Cups). The
Mixed Cups trials were like the coordination-mismatch trials of Experiment 2,
because correct interpretation of the phrase requires coordination of the number,
partitive and noun. Again, we expected some children to go by the mentioned
number of cups in the measure phrase, regardless of the actual number of cups
containing sand (see Table 3, asterisked trials of Mixed Cups). Children were
tested twice on each of the twelve trials depicted in Table 3, with presentation of
the trials randomized.
Table 3. Display and instruction for the language comprehension task of Exp. 3.
Adult responses are in bold. Hypothesized child responses are in parentheses (on the
basis of Exp. 2). Asterisks (*) mark incongruent trials in which the predicted child
responses differ from adult responses. Unmarked trials are congruent trials.
Conditions:
Control
Test
Full Cups
Empty Cups
Mixed Cups
Display
Question
Is this one
Yes
No
No
No
Yes
No
cup of sand (Yes)
(No)
(Yes)*
(No)
(No)*
(No)
Is this three
No
Yes
No
No
No
Yes
cups of sand
(No)
(Yes)
(No)
(Yes)*
(Yes)*
(No)*
4.2.
was also significant (F(3,45)=21.36, p<.0001; see Figure 3). Like Experiment 1,
children were near-ceiling at picking out which side had more cups (trial type
D). Also like Experiment 1, children were also quite good at discriminating
between the sand piles for the easy ratio (trial type A, 4:1 ratio). They were
worse for the smaller, harder to discriminate ratios (trial type C, 4:3 for 3-yearolds and 5:4 for 4-year-olds). However, importantly and in contrast to
Experiment 1, children were better at discriminating amounts of sand when the
sand was placed inside cups than when it was in piles (trial type C vs. B). We
suspect that this difference lies in the method by which cups were presented to
children. In Experiment 1, cups for each character were presented as a group in a
tray. In the current experiment, cups were brought out individually. It is possible
that presenting cups individually highlighted the containers and the notion that
one could use cups in quantifying the amounts of sand. If true, this finding
provides evidence that under some circumstances preschoolers are capable of
applying number to quantify non-cohesive substances.
(b) Mandarin Speakers
%"Correct"
90
% Correct
100
90
80
70
60
50
40
30
20
10
0
3y.o.
4y.o.
sand
A.
vs.
sand
B.
vs.
sand
C.
vs.
cup
D.
vs.
3y.o.
4y.o.
sand
sand
vs.
vs.
A.
B.
sand
C.
vs.
cup
D.
vs.
Language comprehension task: Figure 4 plots the results for the three
types of conditions (Full Cups, Empty Cups, and Mixed Cups). We first
examined childrens performance for the Full Cups control condition. There was
no effect of language or age. As expected, children were on average above 95%
correct on both the yes trials and the no trials (see Figure 4a).
We next turned to the Empty Cups and the Mixed Cups conditions (Figures
4b and 4c). The pattern of performance for the two conditions was similar when
the trials were split into incongruent trials and congruent trials (see Table 3
for the breakdown of incongruent vs. congruent trials). Analysis revealed an
effect of trial type, with children performing better on the congruent than the
incongruent trials (F(1,45)=65.40, p<0.001). There was also an age effect
(F(1,45)=7.27, p=0.01) mediated by trial type (F(1,45)=5.46, p=0.02). Threeyear-olds performed worse than 4-year-olds especially on the incongruent trials.
Replicating the findings of English-speaking children in Experiment 2, fouryear-old English-speaking childrens performance on the measure phrases was
above chance, while 3-year-old English-speaking childrens performance was at
chance. Lastly, English-speaking children performed better than Mandarin-
%"Correct
100
90
80
70
60
50
40
30
20
10
0
chance
3yr>olds.
4yr>olds
English
Mandarin
General discussion
Hespos, Susan J., Dora, Begum, Rips, Lance J., & Christie, Stella (2012). Infants make
quantity discriminations for substances. Child development.
Hespos, Susan J., Mikklia, Maija, Rips, Lance J., Chang, Yin-Juei, Dora, Begum (2011).
Infants make quantity discriminations for multiple piles of substances. Manuscript
under review.
Huntley-Fenner, Gavin (1995). The representation of objects, non-solid substances, and
collections in infancy and early childhood. ProQuest Information & Learning.
Huntley-Fenner, Gavin (2001). Why count stuff? Young preschoolers do not use number
for measurement in continuous dimensions. Developmental Science, 4, 456462.
Huntley-Fenner, Gavin, Carey, Susan, & Solimando, A. (2002). Objects are individuals
but stuff doesnt count: Perceived rigidity and cohesiveness influence infants
representations of small groups of discrete entities. Cognition, 85(3), 203221.
Levin, Iris, & Wilkening, Friedrich (1989). Measuring time via counting: The
development of children's conceptions of time as a quantifiable dimension.
Advances in Psychology, 59, 119-144.
Li, Peggy, Barner, David, & Huang, Becky H. (2008). Classifiers as count syntax:
Individuation and measurement in the acquisition of Mandarin Chinese. Language
Learning and Development, 4(4), 249290.
Li, Peggy, Huang, Becky, & Hsiao, Yaling (2010). Learning that classifiers count:
Mandarin-speaking childrens acquisition of sortal and mensural classifiers. Journal
of East Asian Linguistics, 19, 207230.
Rosenberg, Rebecca D., & Carey, Ssusan (2006). Infants indexing of objects vs. noncohesive substances. Journal of Vision, 6(6), 611611. doi:10.1167/6.6.611
Sabbagh, Mark A., Xu, Fei, Carlson, Stephanie M., Moses, Louis J., & Lee, Kang (2006).
The development of executive functioning and theory of mind: A comparison of
Chinese and U.S. Preschoolers. Psychological Science, 17, 7481.
Srinivasan, Mahesh., Chestnut, Eleanor., Li, Peggy, & Barner, David (in press). Sortal
concepts and pragmatic inference in childrens early quantification of objects. To
appear in Cognitive Psychology.
Spelke, Elizabeth S. (2003). What makes us smart? Core knowledge and natural
language. Language in mind: Advances in the study of language and thought, 277311.
Stickney, Helen (2007). From pseudopartitive to partitive. Paper presented at the
Proceedings of the 2nd Conference on Generative Approaches to Language
Acquisition North America (GALANA), Somerville, MA.
Syrett, Kristen (in press). The role of cardinality in the interpretation of measurement
expressions. Language Acquisition: A Journal of Developmental Linguistics.
vanMarle, Kristy, & Wynn, Karen (2011). Tracking and quantifying objects and noncohesive substances. Developmental Science, 14(3), 502-515.
Wilkening, Friedrich (1980). Development of dimensional integration in children's
perceptual judgment: Experiments with area, volume, and velocity. Information
integration by children, 47-69.
Wynn, Karen (1990). Children's understanding of counting. Cognition, 36(2), 155-193.
Xu, Fei, & Arriaga, R. I. (2007). Number discrimination in 10-month-old infants. British
Journal of Developmental Psychology, 25(1), 103-108.
Xu, Fei, & Spelke, Elizabeth S. (2000). Large number discrimination in 6-month-old
infants. Cognition, 74(1), B1-B11.