The Development of the Ability to Recognize the Meaning of
Iconic Signs
Michael Tomasello
Max Planck Institute for Evolutionary Anthropology, Leipzig
Early developmental psychologists viewed iconic representation as cognitively less complex than other forms of symbolic
thought. It is therefore surprising that iconic signs are not
acquired more easily than arbitrary signs by young language
learners. One explanation is that children younger than
3 years have difficulty interpreting iconicity. The current
study assessed hearing children’s ability to interpret the
meaning of iconic signs. Sixty-six 2.5- to 5-year-olds who
had no previous exposure to signs were required to match
iconic signs to pictures of referents. Whereas few of the
2.5-year-olds recognized the meaning of the iconic signs
consistently, more than half of the 3.0-year-olds and most
of 3.5-year-olds performed above chance. Thus, the ability
to recognize the meaning of iconic signs gradually develops
during the preschool years. Implications of these findings for
sign language development, receptive signed vocabulary
tests, and the development of the ability to interpret iconic
symbols are discussed.
Iconic symbols are a common means of communication. They are prevalent because individuals can
interpret them without requiring a prior explicit
connection between them and their referents. The
features or actions portrayed by iconic symbols can
generate mental representations and associations that
often match or come very close to the idea the presenter of the icon is intending to convey. The classic
example of this aspect of iconicity is the game charades. This is not true of arbitrary symbols, for which
meanings have to be explicitly made either by associNo conflicts of interest were reported. Correspondence should be sent
to Tammy D. Tolar, Department of Educational Psychology & Special
Education, PO Box 3979, Georgia State University, Atlanta, GA 303023979, ([email protected]).
ation or definition. For this reason, iconic representations are considered relatively easy to recognize and
learn as symbols.
Early psychologists identified iconic representation as a precursor to the development of arbitrary
symbols. More specifically, Werner and Kaplan (1963)
viewed iconic gestures as one of the first forms of
representation in the developmental process of ‘‘distancing between person and symbolic vehicle’’ (p.44).
They considered speech symbols the culmination of
this process because the phonic form of representation could be ‘‘clearly differentiated from pragmatic–
conative bodily mobility’’ (p. 45). In this view, the
ability to understand iconic symbols precedes appreciation of arbitrary symbols. Peirce (1894) distinguished
icons from symbols and considered icons a more rudimentary form of communication. He argued that an
iconic representation has no connection to its referent
in the mind, although its presentation would ‘‘excite
analogous sensations’’ in the mind. On the other
hand, a symbol is connected to its referent in the
mind ‘‘by virtue of the idea of the symbol-using mind.’’
The implication of Peirce’s beliefs about iconic and
symbolic representations is that different mental processes are used in their interpretation. However, the
line between icon and symbol and the mental processes associated with them can be blurred because
once an icon is conventionalized it becomes a symbol.
In fact, Peirce argues that symbols often grow from
icons. The view that different mental processes are involved in interpreting iconic and symbolic representations
Ó The Author 2007. Published by Oxford University Press. All rights reserved.
For Permissions, please email: [email protected]
doi:10.1093/deafed/enm045
Advance Access publication on September 7, 2007
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Tammy D. Tolar
Amy R. Lederberg
Sonali Gokhale
Georgia State University, Atlanta
226
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
noniconic signs in short- and long-term memory
(Beykirch, Holcomb, & Harrington, 1990; Lieberth &
Gamble, 1991).
Evidence concerning the role of iconicity in young
children learning to communicate is contradictory. As
Werner and Kaplan (1963) noted, initial gestures that
toddlers acquire are almost all iconic. However, Namy,
Acredolo, and Goodwyn (2000) argue this apparent
preference for iconic gestures is due to parental modeling of iconic rather than arbitrary gestures. Children
learning sign language are exposed to both iconic
and arbitrary gestures (i.e., signs), and in this situation, there is no preference for learning iconic signs
(Folven & Bonvillian, 1991; Orlansky & Bonvillian,
1984). During the first 18 months, only one-third of
signs produced by hearing children of deaf parents are
iconic. Based on this and their own observations,
Newport and Meier (1985) conclude: ‘‘there is an
abundance of evidence that iconicity in general plays
a minor role, if any at all, in the course of ASL acquisition. (though) iconicity does play a role in the acquisition of individual signs by older children and
adults’’ (p. 889). What is the source of this developmental change? One explanation is that young children
who are acquiring American Sign Language (ASL)
through native input may process signs as arbitrary
symbols because they are learning their first language.
Indeed, skilled signers seem to process signs at the
level of their linguistic representations (phonemes)
and ignore pictorial aspects (Siple, Caccamise, &
Brewer, 1982).
However, it also might be that the meaning of
iconic signs are only transparent to older children
because iconic representation is a cognitively advanced
skill not readily available to young language learners.
Research with hearing children suggests the latter may
be true. Hearing children’s appreciation for iconicity
has been explored in experimental studies, which suggest that the ability to interpret iconic gestures undergoes developmental change during early childhood.
Hearing children (26 months old) show a preference
for learning iconic over arbitrary gestures but
18-month-olds do not (Namy, Campbell, & Tomasello,
2004). Tomasello, Striano, and Rochat (1999) found
that 18- and 26-months-old were able to select the
correct referent for iconic gestures but only when
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
and that the processes involved in interpreting icons are
less taxing or complex than those required to understand
arbitrary symbols has influenced how sign language has
been conceptualized.
Both sign and oral languages share many of the
features that by definition make them languages, such
as symbolic representation and syntactic structure.
However, one aspect of symbolic representation, iconicity, may distinguish the two language forms, and
the potential for iconicity in sign languages has been
a source of some debate. The traditional view is that
oral languages have very little iconicity whereas many
of the seemingly mimetic gestures in sign languages
are inherently iconic. A higher degree of iconicity in
sign language suggests relative ease of acquisition.
However, in the past, sign linguists have generally
downplayed the role of iconicity (Schick, 2006;
Wilcox, 2004). This is at least partially due to the
way in which both ‘‘abstract’’ oral and ‘‘iconic’’ gestural symbols have been perceived. Similar to early
philosophers and developmental psychologists who
viewed the meaning in iconic representations as transparent and related more to perceptual than cognitive
abilities, early linguists viewed the iconicity in sign
language as evidence that signs were not worthy of
the term ‘‘language.’’ More recently with the acceptance of sign languages as true languages, such a view
has been challenged. Scholars have recognized that
iconicity is a legitimate, and interesting, aspect of
linguistic systems (Pietrandrea, 2002; Taub, 2001).
In fact, the relative lack of iconicity in oral languages
may be more a function of the limitations of the medium than an indictment against iconicity as an inherent part of language and abstract thought.
Linguistic analysis suggests that iconic associations are a significant part of handshapes and body
locations of many signs (for an analysis, see Pietrandrea,
2002). In addition, some signs are transparently iconic
and represent features or actions of the referent
(Campbell, Martin, & White, 1992). Evidence suggests that nonsigning adults can spontaneously perceive the meaning of these iconic signs (Klima &
Bellugi, 1979; Pizzuto & Volterra, 2000). In addition,
iconicity seems to facilitate learning of sign language
among adults. For example, adults learning signs for
the first time are more likely to retain iconic than
Spontaneous Recognition of Iconic Signs 227
Tischler (1999) conducted such a study with older
children. Using a forced choice picture task similar to
the Peabody Picture Vocabulary Test (PPVT; Dunn &
Dunn, 1997), they asked children to select pictures
that corresponded to the meaning of ASL signs.
First-, fourth-, and ninth-grade hearing students who
had no exposure to sign identified the correct referent
over 70% of the time. Thus, iconic signs were transparent by first grade.
Based on research of other iconic symbol systems,
we hypothesized that this ability would develop some
time between 2.5 and 3.5 years of age. Researchers
have explored hearing children’s ability to use such
iconic symbols as physical replicas and pictures
(Callaghan, 1999; Callaghan & Rankin, 2002;
DeLoache, 1991, 2002; Tomasello et al. 1999). This
research suggests that the ability to interpret iconic
symbols is fragile before 3 years of age, even for such
transparent representations as pictures and physical
replicas.
We tested the development of this ability in
hearing, rather than deaf children because hearing
children’s performance would not be confounded by
exposure to sign. However, in order to generalize the
results to deaf children and understand the underlying
mechanisms involved in iconic representation, we
assessed the role of language in the children’s ability
to interpret iconic sign. First, we examined if the
ability to recognize iconic signs among young children
was a function of vocabulary level more than general
cognitive development. Second, we assessed whether
language served as a mediator in the development of
the ability to interpret iconic symbols. Callaghan
(2000) found that 2.5- and 3.0-year-olds were able to
interpret the meaning of pictures only in conditions
when words could be used as mediators. Attributing
meaning to an iconic symbol using language may tap
different cognitive processes than those involved in
directly mapping an iconic symbol to its referent
(Callaghan, 2000; Campbell et al. 1992). These processes may be additive and together produce better
performance than either process alone. Callaghan
suggested that young children may require both
processes in order to recognize iconic symbols. Therefore, we investigated whether or not producing an
accurate label for a referent influences performance
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
the gesture had been modeled in the context of using
an object (e.g., hammering with a hammer). It was not
until 35 months that children recognized iconic
gestures that were not modeled just prior to the experimental task. Tomasello et al. suggested that the
younger children’s performance may be based on associations between gesture and object made during the
demonstration, rather than iconic recognition. However, Striano, Rochat, and Legerstee (2003) found
26-month-olds (but not 18-month-olds) were able to
spontaneously select a referent based on presentation
of an iconic gesture even if the referent and its associated action had not been modeled prior to the task. It
is not clear, however, if this is done because the children appreciate the iconic representation in the gestures. The iconic gestures used in these studies could
be interpreted through metonymic processes in the
sense that they represent part of an ‘‘activity scheme’’
that includes both the action and the object (Tomasello
et al. 1999). For example, in a familiar routine such as
using a hammer, the motion of hammering may act as
a signal to enact the entire routine including retrieving
the hammer. In this view, iconic gestures are not treated as symbols nor do they tap the same cognitive
processes used by adults in deriving meaning from
iconicity. The objects and gestures in the Striano
et al. study were highly conventional (e.g., hammer,
brush), so by 26 months the children were likely to
have had enough experience to have activity schemes
associated with these objects. Regardless of what is
influencing young children’s ability to respond to
iconic gestures, before 35 months, performance is relatively low (e.g., 50%–60% success rates), suggesting
that this ability is not very robust. In fact, 3-year-olds
in the Tomasello et al. study and 4-year-olds in the
Namy et al. study performed markedly better than
18- and 26-month-olds in taking advantage of iconicity.
The research in iconic gestures supports the notion that the meaning of iconic signs would not be
transparent to young children acquiring sign language,
at least not until 3 years of age. However, iconic signs
differ from iconic gestures in multiple ways (e.g., representation of static as well as action features, linguistic constraints on sign formation). The goal of the
current study was to directly examine the development
of the ability to interpret iconic signs. White and
228
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
Methods
Participants
Sixty-six children (32 girls and 34 boys) were
recruited from three preschools and were selected because they were within one of six predetermined age
ranges. There were fourteen 2.5-year-olds (M 5 31
months, range 5 30–32 months); fifteen 3.0-year-olds
(M 5 37 months, range 5 36–38 months); fifteen
3.5-year-olds (M 5 43 months, range 5 42–44
months); eleven 4.0-year-olds (M 5 49 months,
range 5 48–50 months); five 4.5-year-olds (M 5 55
months, range 5 54–56 months); and six 5.0-year-olds
(M 5 62 months, range 5 61–62 months). During
data collection, it became evident that the children in
the 4.5 and 5.0 groups were able to perceive the iconicity in a majority of the signs, so we focused our
efforts on increasing the number of participants in
the younger groups. Because the 4.5 and 5.0 groups
consisted of fewer children who performed at similar
levels we combined them into one group for all analyses. The children were mostly white and middle to
upper middle class. Nine children (eight 2.5-year-olds
and one 3.0-year-old) were excluded from the study.
Three of the 2.5-year-olds refused to finish the language assessment. Four of the 2.5-year-olds and the
3.0-year-old completed the language assessment but
did not want to return to the testing room for the Sign
Test. One 2.5-year-old did not understand the sign task.
Materials
The PPVT-III (Dunn & Dunn, 1997) is a test of receptive vocabulary normed for children as young as
2.5 years. In addition to providing a language age, it
served as a control task to ensure the children could
do a symbolic picture task. The test has high internal
reliability (lowest alpha 5 .93 for 2.5- and 3.0-year
age groups) and test–retest reliability (corrected
coefficient 5 .92 for 2.6 – 5.9 years). The PPVT consists of picture plates each of which contains four
black-and-white drawings. The examiner asks the
child to select the picture that represents a particular
word (e.g., ‘‘Show me dog. Which one is dog?’’).
The Sign Test was developed for this study and
consisted of iconic signs from ASL. In this task,
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
in identifying iconic signs and whether or not this
influence is similar throughout development among
preschool-aged children.
If language proves to be an important factor in
hearing children’s ability to interpret iconic signs, then
the results would generalize to deaf children who have
comparable levels of language to the hearing children,
regardless of age. In other words, if language helps
mediate interpretation of iconic signs, the ability to
perceive iconicity may relate more to a child’s language abilities than age or cognitive development.
For deaf children who are language delayed, this ability may develop at a later point in development than
what is evident in this study or in deaf children of deaf
parents.
Finally, we investigated if the way in which a sign
depicts a referent influences comprehension of iconic
signs. Iconic signs differ in terms of what aspects of
the referent are represented. For this study, we developed a classification system based on the degree
to which iconic signs involved representations of action versus representations of static features. Pantomimic (Pn) signs primarily depict actions associated
with the referent. For example, one variation of the
ASL sign for TOWEL mimics the action of using
a towel by rubbing the face in circular motions with
flat hands. Perceptual (Pr) signs primarily depict static
features of a referent. The sign for EYEGLASSES
uses the forefingers and thumbs of both hands to
mimic the shape of glasses on the signer’s face. Finally,
some signs represent both (B) perceptual and pantomimic aspects of the referent. The sign for CAMERA
depicts both the shape of a camera and the action of
taking a picture.
As Schick (2006) points out, the ability to appreciate
iconicity may not be ‘‘all or none.’’ We hypothesized that
children will understand action-based iconicity before
feature iconicity. Most iconic gestures used by hearing
toddlers and young children come from motor routines
depicting actions by or on referents (e.g., brushing hair,
using a hammer), suggesting action-based gestures may
be primary (Acredolo, Goodwyn, Horobin, & Emmons,
1999; Werner & Kaplan, 1963). In addition, Schick
found that size and shape specifier (SASS) classifiers,
which represent the visual-geometric features of object,
were acquired relatively late in development.
Spontaneous Recognition of Iconic Signs 229
Figure 1 A page from the sign task booklet. The target
picture on this page is BALL.
participants were required to match each sign to one
of four black-and-white drawings. The four drawings
were on an 8.50 3 110 sheet of paper in a booklet
created from the Carolina Picture Vocabulary Test
(CPVT, Layton & Holmes, 1985). The CPVT is a vocabulary test modeled after the PPVT to provide
a normed test of receptive vocabulary for deaf children
acquiring sign. A page from the Sign Test booklet is
shown in Figure 1.
We selected 30 test and 6 practice signs because we
estimated this was the maximum number of signs we
would be able to test with young children. Sign selec-
Table 1 Sign Test signs and percent of adults who matched sign to referent
Pantomime
only signs
n
Percent
match
Perceptual
only signs
n
Percent
match
Pantomime and
perceptual signs
n
Percent
match
BABY
EAT
MIX
PERFUME
SEWING
TISSUE
WRITE
BASEBALL
HAMMER
TOWEL
HAT
COLD
70
29
29
29
29
29
29
70
70
99
99
29
100
100
100
100
100
100
100
99
99
98
97
97
HOUSE
LOOK
NARROW
VASE
GLASSES
TIGER
SHOWER
GLOBE
SMOOTH
TORNADO
BOX
BALL
29
29
29
29
70
99
70
29
29
99
99
70
100
100
100
100
99
97
97
97
97
96
94
94
BIKE
TELEPHONE
CAMERA
CATERPILLAR
NEEDLE
BANANA
ICE CREAM
RING
WINDOW
MIRROR
JAR
BOOK
70
70
29
29
29
70
70
70
70
99
99
70
100
100
100
100
100
99
99
99
97
94
95
93
Note. Adults matched signs to pictured referents in the first (n 5 29), the second (n 5 70), or both (n 5 99) studies.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
tion was based on two criteria: iconicity and sign type.
We initially selected 76 signs that we perceived to be
iconic. Sign iconicity was determined through two
studies with adults who had no knowledge of sign
language. Using plates from the CPVT, adults were
asked to select the correct referent for the signs. We
ranked the signs according to the percentage of adults
who could match them to pictures of their referents.1
To meet the second criteria for selection, sign
type, we categorized signs as perceptual, pantomime,
or both perceptual and pantomime as described in the
Introduction. Signs rarely purely represent the action
or features of the referent. Therefore, our distinctions
between different types of signs were based on the
aspects of the referents that are most emphasized by
the iconic signs. We selected 10 of the most iconic
signs based on the adult performance from each category. Finally, from the remaining signs we chose six
signs for practice so that the children could experience
two of each type of sign in order to understand the
nature of the task. More than half of the test signs
were recognized by 100% of the adults tested. The
least iconic test sign was recognized by 96% of the
adults. Table 1 shows the signs used in this study and
the percentage of adults who correctly matched the
sign to the referent. Twenty-three of the signs are
from the CPVT.
In the final version of the test booklet, the pages
were randomly ordered, and the target pictures were
equally distributed between the four quadrants on the
pages. A group of adults (N 5 43) was tested on the
230
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
30 test signs from the same test booklet used in the
children’s study. The mean performance of this group
was 96.2% (SD 5 3.2%).
Procedure
Coding
Transcripts were made from the videotapes of the 32
children who participated in the labeling condition.
The children’s labels for the target pictures were
coded whether they were correct (defined as using
the predetermined correct label or its variant based
on the sign, e.g., ball or baseball) or incorrect. Incorrect
labels were further coded. In context labels identified
a reasonable referent for the sign (e.g., snake instead of
caterpillar). Mislabels either identified some aspect of
the picture that was not related to the sign (e.g., boy
instead of pizza) or completely misidentified the picture (e.g., truck instead of camera). The final two incorrect label codes included unintelligible and no
response. Two graduate students were trained on the
coding system. To establish and maintain intercoder
agreement, 20% of the transcripts, randomly selected,
were coded independently by both the graduate students. After every five transcripts coded, the students
would compare their results on the transcript they
both coded and resolve any disagreements. Across
these transcripts, the coders agreed on 96% of the
labels coded.
Results
We conducted two sets of analyses. The first addressed
our research goals concerning the development of
iconic representation. These include the effects
of chronological age, language, and sign type on
children’s ability to recognize iconic signs. The second
set of analyses examined potential confounds to these
results.
We initially addressed our research goals as a whole
by conducting a 3 (sign type) by 2 (labeling condition)
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
The children were tested in quiet, well-lit rooms at
their preschools in two 30-min video-taped sessions.
The PPVT was administered during the first session,
and approximately a week later, the sign task was
given.
The children were randomly selected to be in one
of two conditions for the sign task. In the labeling
condition, the person administering the test would
point to each drawing on the picture plate and ask
the child, ‘‘What is this?’’ A correct label was confirmed with a nod and a verbal affirmative. If the child
mislabeled the indicated picture or did not respond,
the researcher would reply with a positive comment
and provide the correct label (e.g., ‘‘Good guess, but
this is a caterpillar.’’). In the nonlabeling condition, the
examiner would point to each picture and say, ‘‘Look
at this one,’’ to make sure the child attended to each of
the four drawings on the page.
In both conditions, the practice signs were presented first. For both the practice and test signs, after
the researcher directed the child’s attention to each of
the four pictures, she would draw the child’s attention
to herself (e.g., ‘‘Look at me.’’), perform the iconic
sign and say, ‘‘Show me this one. Which one is this?’’
If the child did not point to one of the pictures, the
researcher would repeat the sign and encourage the
child to select one of the pictures (e.g., ‘‘You can
guess.’’). A second researcher recorded the child’s response.
During presentation of the practice signs, if the
child correctly identified the target picture, she was
told that she was correct (e.g., ‘‘That’s right!’’). If
the child selected a drawing other than the target picture, she received positive feedback (e.g., ‘‘Good try,
but.’’). In both cases, the researcher would then
identify the correct picture by drawing attention to
herself as she performed the sign, then pointing to
the target picture (e.g., ‘‘See, this,’’ sign, ‘‘is this,’’
point to the picture). In addition, the researcher would
place her hands over the target picture and perform
the sign directly over the drawing to emphasize the
relationship between the sign and the target picture.
All practice signs were presented even if the child
appeared to immediately grasp the nature of the task.
During the presentations of the test signs, the
child was praised after each response (e.g. ‘‘Good
job! You are good at this.’’) and received stickers after
every 5 or 6 signs. No feedback was given as to the
accuracy of the child’s response during the presentation of the test signs.
Spontaneous Recognition of Iconic Signs 231
Table 2 Percent correct on sign task by group and condition
Condition
Label (SD)
n
No Label (SD)
n
Total (SD)
n
36
46
54
67
83
55
7
8
6
6
5
32
37
47
50
59
77
53
7
7
9
5
6
34
36
46
52
64
76
54
14
15
15
11
11
66
(8)
(12)
(11)
(9)
(4)
(18)
(12)
(15)
(13)
(15)
(12)
(18)
(10)a
(13)a,b
(12)b,c
(12)c
(12)
(18)
Note. Group total means that do not share subscripts differ at p 5 .05 in Tukey’s honestly significant difference post hoc comparisons. The Tukey
Critical Difference was 12.6 (i.e., groups that differed by approximately 13% in their performance on the Sign Test were significantly different at the .05
level).
Developmental Progression
There was a distinct developmental progression in the
ability to perceive iconicity with the older children
performing better than the younger children. Tukey’s
honestly significant difference post hoc analysis revealed that the 2.5-year-old children recognized significantly fewer signs than all older children except the
3.0-year-olds (see Table 2). In addition, the children in
the 4.5/5.0-age group were able to recognize more of
the iconic signs than the children in any of the other
groups.
The ability of children in each age group to recognize iconic signs was further examined in two ways.
First, using t-tests, group performance was compared
to chance (i.e., 25% because there were four choices
for each sign). All groups performed significantly
above chance. Second, we assessed individual performance by comparing each child’s performance to
chance. Based on the binomial distribution, in order
for performance to be statistically above chance, a child
had to correctly identify at least 13 (43%) of the signs.
Only two out of the 14 2.5-year-old children performed above chance, whereas more than half of the
children in each of the other groups and all the children in the 4.5/5.0 group performed above chance
(see Figure 2). The results of the individual analysis
suggest that few 2.5-year-olds appreciate iconicity;
however, given the number of signs, a binomial test
is a rather stringent test.
Language
We assessed the relationship between language and
sign task performance in three ways. First, we examined the relative effects of chronological and language
age on performance using hierarchical multiple regression. We performed two analyses. In the first analysis,
we entered chronological age first followed by language age (as measured by the PPVT). Chronological
age accounted for 59% of the variance and language
Above Chance
Chance
100%
Percent of Children
by 5 (age group) repeated measures analysis of variance (ANOVA). Sign type was a within-subject variable and labeling condition and age group were
between subject variables. There were main effects
of age, F(4, 56) 5 23.57, p 5 .000, and sign type,
F(2, 112) 5 12.34, p 5 .000, but no effect of labeling
condition. None of the interactions were significant.
These main effects were examined in more detail
through additional analyses reported below. Table 2
shows the performance by age group and labeling
condition.
80%
60%
40%
20%
0%
2.5
3.0
3.5
4.0
4.5 - 5.0
Age Group
Figure 2 Percentage of children above chance. For individual
performance, at least 43% (13 out of 30) correct on Sign
Test is statistically above chance.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Group
2.5
3.0
3.5
4.0
4.5/5.0
Total
232
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
Table 3 Effect of chronological and language age on sign task performance
Model
R2
R2 change
F change
df1
df2
p
Chronological age
Chronological 1 language age
Language age
Language 1 chronological age
0.59
0.65
0.43
0.65
0.59
0.06
0.43
0.21
91.48
10.11
48.52
38.03
1
1
1
1
64
63
64
63
.000
.002
.000
.000
metric statistical analyses (Bakeman & Gottman, 1997;
Rosner, 2000). The odds ratio for this analysis was
calculated by dividing the odds of identifying the sign
if the target picture was correctly labeled by the odds
of identifying the sign if the target picture was not
labeled correctly.
In general, 2.5- and 3.0-year-old children were
more likely to identify the sign if they labeled the
picture (see Table 4). The odds of 2.5-year-olds identifying a sign if they labeled the target picture were
almost four times the odds of them identifying the
sign if they did not label the picture correctly. Children 4.0 years and older performed the same whether
they labeled the picture correctly or not. ANOVA with
the log-transformed odds ratio as the dependent variable and age group as the independent variable yielded
nonsignificant results, F(4, 27) 5 1.179, p 5 .342.
However, this result is probably due to the lack of
power due to the small sample sizes (groups ranged
from five to eight children in the labeling condition).
The relationship between the odds ratio and chronological age, however, approached significance, r 5
2.325, p 5 .069.
Sign Type
The initial ANOVA revealed that there was a main
effect of sign type but no interactions between sign
Table 4 Sign task performance by group and accuracy in labeling target picture
Percent of signs correctly identified
Age group (n)
Percent of pictures
labeled correctly (SD)
Odds ratio
Picture labeled
correctly
Picture labeled
incorrectly
2.5 (7)
3.0 (8)
3.5 (6)
4.0 (6)
4.5/5.0 (5)
47
44
52
57
65
3.8
2.6
2.5
1.4
1.3
45
55
55
66
83
26
39
50
69
81
(4)a
(13)a
(12)a,b
(11)a,b
(9)b
Note. Groups that do not share the same subscript differ at p , .05 in the Tukey post hoc comparison. Odds ratio 5 [(correct label and recognized
sign)/(correct label and did not recognize sign)]/[(incorrect label and recognized sign)/(incorrect label and did not recognize sign)]. Incorrectly labeled
pictures include all types of mislabels (e.g., labeled part of picture not related to sign, no response) except for those coded as unintelligible.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
age accounted for an additional 6% (see Table 3). In
the second analysis, we entered language age first that
accounted for 43% of the variance and chronological
age accounted for an additional 21%. Based on these
results, relative to language age, chronological age
appears to be a stronger predictor of sign task performance in preschool-aged children.
Next, we asked if children’s ability to recognize the
signs was influenced by labeling the pictures used in
the sign task. The initial ANOVA revealed that there
was no effect of labeling condition (see above). In fact,
as Table 2 shows, the performance was nearly (and
surprisingly) identical between the children in the labeling and nonlabeling conditions with a difference of
only 1% among the 2.5 and 3.0-year-old children. The
labeling condition included trials when the child produced the correct label and trials when the researchers
produced the label because the child did not. Because
research suggests recognition of iconicity is influenced
by the child’s production of labels, we examined this
possibility specifically.
We analyzed if children were more likely to identify a sign if they labeled the target picture correctly
themselves. We used an odds ratio to analyze this relationship because unlike probability measures, odds
ratios are not subject to some of the restrictions that
may cause assumption violations or biases in para-
Spontaneous Recognition of Iconic Signs 233
8
4.5 - 5.0
4.0
3.5
3.0
2.5
7
6
5
4
3
2
Pantomime
Both
Perceptual
Sign Type
Figure 3 Group performance on different types of signs.
type and age group or between sign type and labeling
condition (see above). Bonferroni-adjusted pairwise
comparisons showed that children performed significantly better on pantomime (mean 58%) than on either perceptual (mean 48%) or both type signs (mean
47%). There was no difference in performance between signs depicting perceptual features and those
depicting both perceptual and pantomimic aspects
of the referents (Figure 3).2 The percentage of children who correctly identified each sign is shown in
Appendix A.
Secondary Analyses of Confounding Variables
Developmental progression. The first set of analyses
examined whether the age effects on sign task performance was due to the children’s familiarity with the
referents rather than their ability to recognize iconicity. Although we did not test directly for familiarity,
we considered children correctly labeling the referent
picture as an indication of their familiarity with the
referent.
We first assessed the effect of age on how well the
children were able to label the pictures. If the older
children in this study could correctly label more target
pictures than younger children, their better performance on the sign task may be due to the ability to
use the target picture as a symbol as much as their
ability to recognize the iconic gesture as one. Table 4
shows the percentage of target pictures correctly labeled by children in each age group. A one-way
ANOVA revealed an age difference in the ability to
label the target pictures, F(4, 27) 5 4.00, p 5 .01.
Three-year-olds correctly labeled the lowest percentage of target pictures (44%), whereas the 4.5 and 5.0-
year-olds correctly labeled the highest percentage
(65%). However, based on Tukey post hoc analysis,
the four lowest age groups (2.5–4.0) did not differ
from each other nor did the three oldest (3.5–5.0).
Perhaps, even more important than the statistical analysis, 2.5-year-olds labeled more pictures correctly than
3.0-year-olds. Because the four youngest age groups
were similar in their ability to label the pictures, the
development of the ability to interpret the iconic signs
is unlikely a function of their ability to identify the
pictures.
Another way of addressing the possible confound
of familiarity is to examine performance on referents
to which all children are likely to be equally familiar.
Using labeling as an indication of familiarity, we identified three subsets of signs: those that were identified
by all children in the labeling condition (house,
eyeglasses); those labeled correctly by all 2.5-year-olds
(house, eyeglasses, banana, window); and those labeled
correctly by at least 70% of the 2.5-year-olds (house,
eyeglasses, banana, window, telephone, bike, baby,
tiger, mirror). As Table 5 shows, the pattern of results
for each of these subsets was almost identical to that of
the performance for all 30 signs. These numbers indicate that familiarity was unlikely a factor in the results.
Sign type. This analysis addressed the potential
confound between sign type (e.g., pantomime, perceptual, both) and referent type (e.g., noun, verb,
adjective). Twenty-one of the referents were nouns
(5 pantomime, 6 perceptual, and all 10 of both type
Table 5 Percent of signs identified among subsets of
signs selected according to familiarity with referent as
indicated by labeling of target picture
Sign set
Age group
1
2
3
All signs
2.5 (14)
3.0 (15)
3.5 (15)
4.0 (15)
4.5/5.0 (11)
32
57
63
55
91
32
50
48
57
80
34
50
50
61
81
36
46
52
64
76
Note. The pictures for sign set 1 (HOUSE, EYEGLASSES) were
labeled correctly by all children in the labeling condition; sign set 2
(HOUSE, EYEGLASSES, BANANA, WINDOW) by all 2.5-yearolds; sign set 3 (HOUSE, EYEGLASSES, BANANA, WINDOW,
TELEPHONE, BIKE, BABY, TIGER, MIRROR) by at least 70% of
the 2.5-year-olds.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Mean Number Correct
9
234
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
Table 6 Percent of pantomime signs correctly identified
by referent type
Verbs
Nouns
2.5 (14)
3.0 (15)
3.5 (15)
4.0 (11)
4.5/5.0 (11)
Total (66)
46
52
52
62
76
56
39
49
65
78
78
60
signs); seven of the referents were verbs (five pantomime and two perceptual); and two of the perceptual
signs referred to adjectives. Because a higher proportion of pantomime signs referred to verbs than was the
case for perceptual or both signs, the better performance on pantomime signs may have been because
they depicted actions more than the other two types
of signs. However, a repeated measures ANOVA
among just the pantomime signs (five verbs and five
nouns) showed neither an age by referent type interaction, F(4, 61) 5 1.29, p 5 .284 nor a main effect of
referent type, F(1, 61) 5 1.16, p 5 .287 (see Table 6).
This suggests that the effect of sign type was not due
to referent type; however, this result should be treated
with caution because it is unclear if the way in which
we categorized the referent types was consistent with
the children’s interpretations of the referents.
Discussion
Iconicity is not transparent, nor does appreciation for
iconicity develop at any one time. Instead, significant
developmental changes seem to occur throughout the
preschool years. The development of the ability to
derive meaning from iconic signs is continuous during
these early years. The ability to spontaneously recognize the meaning of iconic signs is fragile by 2.5 years,
relatively established by 3.5–4.0 years, and perhaps
adult like by 4.5–5.0 years of age.
The lack of insight into the relationship between
iconic gestural symbols and their referents may be one
reason why infants and toddlers do not exploit the
possibilities that iconic gestures and signs offer.
Although both deaf and hearing 1- to 2-year-olds easily acquire the conventionalized gestures and signs
used by their parents, they rarely produce novel gestures and those that they do seem to be ‘‘abstracted’’
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Age group (N)
from motor routines (Namy et al. 2000). Indeed, even
though deaf toddlers of deaf parents live in an environment that is rich in visual iconic gestures and
signs they do not gesture more than hearing toddlers
of hearing parents (Spencer, 1993). In addition, young
native language learners may not show a preference
for iconic over arbitrary signs because the meaning
of many iconic signs is not transparent until sometime
after 3 years of age.
Older deaf children, on the other hand, exploit the
iconic potential of gestures. Hearing and deaf preschoolers are able to communicate to each other without language using novel gestures, much as adults do
in the game charades (Lederberg, Ryan, & Robbins,
1986). Research with older children also suggests that
gestures that are pantomimic in origin are easier to
acquire than those that represent features. Deaf preschoolers of hearing parents deprived of input create
spontaneous novel iconic gestures (Goldin-Meadow,
2003). Descriptions of these ‘‘home-sign systems’’ suggest that early ‘‘characterizing’’ gestures function like
action-like verbs in communication, whereas objects
are more likely to be represented by points, even when
children refer to absent objects. As these children
develop, features become more represented in their
systems. Deaf children acquiring ASL also acquire
manipulative handler classifiers system by 3 years of
age, whereas SASS classifiers are later developing
(Schick, 2006).
These results also offer practical advice for assessing deaf children’s sign vocabulary knowledge using
receptive vocabulary tests such as CPVT or PPVT.
Based on the ability of hearing elementary school children to accurately guess the meaning of many of the
signs of the CPVT, White and Tischler (1999)
expressed concerns that deaf children can do the same
and, thus, be credited with knowing more words than
they do. Our results suggest that by 4.5-years-old,
children possess the ability to flexibly use iconicity
to guess many of the signs used in the CPVT. Although even with younger children caution should
be used in interpreting CPVT scores, the younger
children’s low level of performance suggests iconicity
will play less of a role in their performance on this
test. Indeed, Moeller and Johnson (1988) found that
deaf children do not obviously exploit iconicity on
Spontaneous Recognition of Iconic Signs 235
picture. This is similar to Callaghan’s finding of the
importance of labeling for young children’s ability to
recognize iconic pictures (Callaghan, 2000). Interestingly, the researchers’ labeling of the pictures did not
facilitate the children’s understanding of the gestures,
and surprisingly, the children’s general vocabulary
knowledge did not seem very important to the development of this ability. Children 2.5- and 3.0-years-old
were most successful at interpreting iconic signs when
they produced the pertinent label for the referent. This
primary link (i.e., having the relevant aspect of the
referent robustly represented in the mind) appears
to be critical for 2.5-year-olds and important for 3.0year-olds but gradually less important throughout the
later preschool years. This would also suggest young
deaf children just learning language may be less able
to appreciate the iconicity contained in signs than the
hearing children tested in this study. On the other
hand, language-delayed deaf children may be able to
use iconicity to bootstrap their acquisition of signs.
The second factor that influenced interpretation of
iconic signs was the way in which the referents were
represented by the signs. Across ages, signs that only
depicted the action by or on the referent were easier
than signs that included depiction of perceptual features of the referent. These results are consistent with
observations of the gestures used by hearing toddlers
(Acredolo & Goodwyn, 1988). Acredolo and Goodwyn
noted ‘‘the results showed a clear preference for using
an imitation of an action the child did with the object
(45% of the total). An additional 13% of the total were
imitations of actions inherent in the object, and 10%
were depictions of some perceptual quality of the
object’’ (p. 460). The influence of sign type had a longer developmental trajectory than that of verbal mediation. Whereas the effect of labeling declined by 3.5
years and was relatively unimportant by 4.0 years, the
effect of sign type was evident throughout this period.
The relative difficulty that young children seem to
have identifying signs based primarily on perceptual
features is particularly telling when compared to the
performance of adults. For example, adults recognized
the sign for HOUSE (hands shaped to represent the
inverted V of a roof then sliding down to form the
walls) 100% of the time, even when they were required to label the sign without the aid of pictures.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
receptive vocabulary tests until early elementary
school.
This research can also contribute to understanding
of development of the ability to attribute meaning to
iconic symbols. Our results are consistent with other
experimental research in that the way that 3-year-olds
appreciate iconicity is different from 2.5-year-olds.
At 3 years of age, children seem to become capable
of directly deriving meaning from iconic symbols because of the visual relationship between the icon and
its referent. In our study, the majority of 3-year-olds
performed above chance in their ability to recognize
novel iconic signs. Other research also has noted a developmental shift around 3 years of age in children’s
ability to recognize iconic gestures that have not
been previously modeled and to spontaneously derive
meaning from iconic objects and pictures (Callaghan,
1999, 2000; Tomasello et al. 1999). This suggests that
by 3 years, appreciation of iconicity may be more flexible and is not reliant on unique aspects of the mode of
communication (e.g., the metonymic aspects of iconic
gestures), nor is it inhibited by them (e.g., dual representation aspects of iconic objects). On the other
hand, both our results and that of experimental studies
across domains show that for 2.5-year-olds the ability
to derive meaning from iconicity is largely dependent
on context. When the dual representational aspects of
iconic objects (i.e., as symbols and as objects to be
manipulated) are mitigated (e.g., putting objects behind glass so they cannot be manipulated), 2.5-yearolds can act on the iconic aspects of the objects when
they could not before the intervention (DeLoache,
1991, 2002). They can also recognize iconic gestures when those gestures are derived from actions
performed in demonstrating objects just prior to a task
(Striano et al. 2003). However, as our results indicate,
when an iconic sign is ‘‘novel,’’ a 2.5-year-old has
difficulty relating it to its referent. The literature suggests that whereas these types of contextual supports
may improve the performance of older children, it is
critical to the performance of children younger than
3 years. We have added to this literature by providing
evidence that two of these supporting mechanisms are
important in deriving meaning from iconic signs.
First, being able to label the referent (i.e., picture)
seemed to promote the ability to link the sign to the
236
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
include perceptual differentiation and categorization (Callaghan, Rochat, MacGillivray, & MacLellan,
2004) and analogical reasoning (Callaghan et al., 2004;
DeLoache, 2002). Analogical reasoning, though, may
be the ‘umbrella’ process that is key to deriving meaning from iconicity.
In order to map a nonconventional iconic symbol
onto a referent, an individual must appreciate structural similarities between the icon and referent in
the absence of literal similarities (Gentner & Medina,
1998). For example, occasionally, the older children in
our study would spontaneously talk out loud and describe how they were making the connection between
the sign and the referent by making feature matches
between the gesture and certain aspects of the picture.
They seemed to recognize that the ‘‘rule’’ was that the
picture that most closely matched the abstracted features depicted by the gesture was the one being represented by the symbol. Appreciation for this rule is
important across domains. However, there may be domain differences in the way precursors to analogical
reasoning scaffold the ability to appreciate iconicity.
There is one characteristic of iconic gestures that
makes them unique compared to other domains and
possibly gives them an advantage for some types of
analogical mapping. Iconic gestures inherently lend
themselves to action-based representations. Although
preschool-aged children may have some difficulty recognizing gestures based on static features of referents,
matching gestures based on actions or motor routines
seems to develop early. This is likely distinctive of
iconic gestures and does not fit neatly into domaingeneral theories about iconic representation. This
aspect of iconic gestures may tap a different representational system than those accessed by iconic pictures
and objects. The more easily identified, purely pantomimic signs may have the effect of ‘‘exciting analogous
sensations’’ in the mind as described by Peirce (1894)
in the sense that they access some form of procedural
or action-based memory that is relatively robust and in
many cases not consciously represented. This hypothesis is consistent with research on motor neurons that
shows that in monkeys these neurons are activated not
only when objects are manipulated but also when similar gestures are observed (Rizzolatti & Arbib, 1998).
It has been suggested that these neurons function as
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Yet the preschool-aged children were notably less successful at identifying HOUSE especially when compared to their performance on the other signs. Out of
the 30 signs tested on children, it ranked 18th (52%)
in ease of identification in spite the fact that all children required to do so were able to label the house
picture correctly. These results, though, should be
treated with caution because we did not assess the
relative complexity of the signs. In addition, the categorization of sign type is not straightforward as evidenced by the number of signs that were categorized
differently by researchers and naive adults. Future research that constructs gestures that map to the same
referent but differ on this dimension would provide
more definitive answers to some of the questions
raised by this aspect of our study. This would also
address the potential confound between sign type
and referent type present in this article.
Spontaneously recognizing iconic signs is not easy.
Even adults can have difficulty interpreting iconic
signs, especially if they do not have cues like objects
or pictures to prompt recognition. When we were developing the sign task for this study, we required some
adults to produce labels for the signs (i.e., they did not
have pictures from which to choose). Under these
more difficult conditions, adults were able to label only
seven signs over 80% of the time, and they were completely unable to label five others (see Appendix B).
All these signs were easily recognized by adults when
cued by pictures. It appears that interpretation of
iconic signs is not automatic, and context is critical.
One important difference between most iconic gestures and iconic symbols in other domains (e.g.,
objects and pictures) is the degree of iconicity. There
are other differences, some of which may make the
interpretation of iconic gestures easier for young children (e.g., lack of dual representation), and some of
which may influence development in unique ways
(e.g., the inherent motion in gestures). However, in
spite of these differences, the ability to derive meaning
from iconicity in gestures seems to manifest itself at
approximately the same time as it does in other
domains. The implication is that there are domain
general processes that mature to such an extent that
iconic representation becomes functional across symbol systems at about 3 years. These processes likely
Spontaneous Recognition of Iconic Signs 237
Appendix A
In conclusion, contrary to the beliefs of early psychologists, young hearing children do not find iconic
symbols easy to interpret. There may be a major cognitive shift around 3 years of age that enables children
to interpret the meaning of iconic gestures, pictures,
and objects. Although there may be some mechanisms
that are unique to each of these symbolic domains,
many of the processes that enable appreciation for
iconicity are the same across domains. For young deaf
children learning sign language, it seems unlikely that
the acquisition of iconic signs is influenced by appreciation for iconicity per se. However, for deaf children
older than 3 years, bringing attention to the iconic
aspects of some signs may enhance language development. Future research needs to develop models of
the interaction of the mechanisms within and across
symbol domains, as well as examine developmental
changes in the influence of iconicity in sign language
development.
Percent of children who correctly identified each sign
Age group (n)
Sign (type)
2.5 (14)
3.0 (15)
3.5 (15)
4.0 (11)
4.5–5.0 (11)
All (66)
ICE CREAM (B)
BABY (Pn)
MIX (Pn)
EAT (Pn)
CATERPILLAR (B)
SHOWER (Pr)
BIKE (B)
RING (B)
SMOOTH (Pr)
HAMMER (Pn)
TIGER (Pr)
CAMERA (B)
PERFUME (Pn)
WINDOW (B)
HOUSE (Pr)
BASEBALL (Pn)
LOOK (Pr)
WRITE (Pn)
GLASSES (Pr)
TORNADO (Pr)
TOWEL (Pn)
BANANA (B)
MIRROR (B)
TISSUE (Pn)
64
64
64
64
57
50
50
50
50
43
43
43
43
43
36
36
36
36
29
29
29
21
21
21
80
67
60
60
47
67
60
47
33
87
53
47
47
27
60
53
33
13
53
47
33
60
47
47
87
73
73
53
93
67
60
27
67
73
47
73
73
13
53
67
33
53
73
27
60
53
60
53
82
73
73
73
73
73
100
36
82
91
36
91
82
45
27
91
64
45
82
91
73
73
64
73
100
73
64
91
91
91
82
91
82
100
91
100
82
55
82
91
73
73
100
91
73
82
73
73
82
70
67
67
71
68
68
48
61
77
53
68
64
35
52
65
45
42
65
53
52
56
52
52
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
a mechanism for anticipating actions in conspecifics
(Wilson & Knoblich, 2005). Action-based iconic gestures may be developmental precursors to language
although not necessarily in the way suggested by
Werner and Kaplan (1963). Early in the development
of social communication, toddlers begin to extrapolate
from motor or action routines. This extrapolation
allows them to perform actions intended, although
not completed, by adults (Meltzoff, 1995). This may
be why children as young as 18 months will respond to
iconic gestures that have not been explicitly modeled
although demonstrated repeatedly through actions on
target objects (Namy, 2005). Even children older than
2.5 years who begin to incorporate other processes in
the interpretation of iconic gestures perform better
when they can rely on the earlier developed, more
stable processes that allow them to treat seemingly
novel iconic gestures as metonyms, or signals for action schemas (Tomasello et al. 1999).
238
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
Appendix A
Continued
Age group (n)
2.5 (14)
3.0 (15)
3.5 (15)
SEWING (Pn)
GLOBE (Pr)
NARROW (Pr)
NEEDLE (B)
TELEPHONE (B)
VASE (Pr)
21
21
14
14
0
0
40
20
40
20
20
7
7
27
40
27
13
27
4.0 (11)
4.5–5.0 (11)
27
36
36
45
45
27
55
45
73
82
91
45
All (66)
29
29
39
35
30
20
Note. Pn, pantomime; Pr, perceptual; B, both. Signs are listed based on the performance of 2.5-year-olds. The shaded cells indicate the selected target
picture at greater than chance levels.
Appendix B
Pantomime
only signs
Percentage of adults who identified each sign without picture cue
Percent
identified
Perceptual
only signs
Percent
identified
Pantomime
and perceptual
signs
Percent
identified
100
100
68
35
21
19
15
0
0
0
TELEPHONE
RING
BIKE
ICE CREAM
BANANA
CAMERA
MIRRORa
WINDOW
CATERPILLAR
NEEDLE
100
90
55
45
40
32
25
25
21
15
Test signs
BABY
PERFUME
EAT
SEWING
WRITE
MIX
HAMMER
TISSUE
BASEBALL
TOWELa
90
89
84
74
65
28
15
11
0
0
HOUSE
LOOK
SMOOTH
GLASSES
TORNADOa
NARROW
SHOWER
VASE
TIGERa
GLOBE
Practice signs
a
HAT
COLD
15
11
a
BOX
BALL
45
20
BOOK
JARa
20
3
a
Based on two samples of 20 adults each, all other signs tested on one sample of 20 adults.
Notes
1. A more detailed report of these studies is available from
the first author.
2. Classification into sign type was done by consensus
among the researchers. We also asked 73 college students to
code the signs as depicting action (either by a person or by an
object), depicting a feature or shape of an object, both, or
neither. To make the situation analogous to the children’s task,
the adults were given the picture of the sign but not told the
meaning of the sign. The adults’ categorizations were used to
determine which signs clearly depict the three categories. Repeated measures ANOVA of children’s performance on two
subsets of signs continued to show that pantomimic signs were
easier to recognize than perceptual ones. For the first subset,
18 signs (seven pantomime, six both, five perceptual) were
categorized the same way by the researchers and by a majority
of participants. There was only a main effect of sign type, F(2,
122) 5 35.35, p 5 .000. Performance was significantly higher
on pantomime signs (mean 64% correct) than on either both
(mean 40% correct) or perceptual signs (mean 41% correct).
The second analysis included 10 signs for which at least 70%
of adults agreed on their categorization: pantomime (eat, tissue, towel), perceptual (globe, house, vase), and both (camera,
banana, telephone, caterpillar). There was again only a significant main effect of sign type, F(2, 122) 5 17.79, p 5 .000.
Performance on pantomime (mean 57% correct) and both
signs (mean 56% correct) was the same and significantly
higher than performance on perceptual signs (mean 33%
correct).
References
Acredolo, L. P., & Goodwyn, S. W. (1988). Symbolic gesturing
in normal infants. Child Development, 59, 450–466.
Acredolo, L. P., Goodwyn, S. W., Horobin, K. D., &
Emmons, Y. D. (1999). The signs and sounds of
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Sign (type)
Spontaneous Recognition of Iconic Signs 239
Moeller, M. P., & Johnson, D. (1988, November). Longitudinal
performance of deaf students using manually-coded English.
Paper presented at the ASHA National Convention:
Boston, MA.
Namy, L. L. (2005). The changing role of iconicity in nonverbal
symbol learning. In L. L. Namy (Ed.), When good iconicity
goes bad. Atlanta, GA: Society for Research in Child Development Biennial Meeting.
Namy, L. L., Acredolo, L. P., & Goodwyn, S. W. (2000). Verbal
labels and gestural routines in parental communication
with young children. Journal of Nonverbal Behavior, 24,
63–79.
Namy, L. L., Campbell, A. L., & Tomasello, M. (2004).
The changing role of iconicity in non-verbal symbol
learning: A U-shaped trajectory in the acquisition of arbitrary gestures. Journal of Cognition and Development, 5,
37–57.
Newport, E. L., & Meier, R. P. (1985). The acquisition of
American Sign Language. In D. I. Slobin (Ed.), The
crosslinguistic study of language acquisition (pp. 881–938).
Hillsdale, NJ: Lawrence Erlbaum.
Orlansky, M., & Bonvillian, J. D. (1984). The role of iconicity in
early sign language acquisition. Journal of Speech and Hearing Disorders, 49, 287–292.
Peirce, C. S. (1894). What is a sign? Retrieved June 10 2005,
from http://www.iupui.edu/;peirce/ep/ep2/ep2book/
ch02/ch02.htm
Pietrandrea, P. (2002). Iconicity and arbitrariness in odelin sign
language. Sign Language Studies, 2, 296–321.
Pizzuto, E., & Volterra, V. (2000). Iconicity and transparency
in sign language: A cross-linguistic cross-cultural view. In
K. Emmorey & H. Lane (Eds.), In Signs of Language
Revisited: An Anthology to Honor Ursula Bellugi and
Edward Klima (pp. 261–286). Mahwah, NJ: Lawrence
Erlbaum Associates.
Rizzolatti, G., & Arbib, M. A. (1998). Language within our
grasp. Trends in Neuroscience, 21, 188–194.
Rosner, B. (2000). Fundamentals of biostatistics. Pacific Grove,
CA: Duxbury Thomson Learning.
Schick, B. (2006). Acquiring a visually motivated language: Evidence from diverse learners. In B. Schick, M. Marschark, &
P. E. Spencer (Eds.), Advances in the sign language development of deaf children (pp. 102–134). Oxford University
Press.
Siple, P., Caccamise, F., & Brewer, L. (1982). Signs as pictures
and signs as words: Effect of language knowledge
on memory for new vocabulary. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 82, 619–625.
Spencer, P. E. (1993). Communication behaviors of infants with
hearing loss and their hearing mothers. Journal of Speech
and Hearing Research, 36, 311–321.
Striano, T., Rochat, P., & Legerstee, M. (2003). The role of
odeling and request type on symbolic comprehension of
objects and gestures in young children. Journal of Child
Language, 30, 27–45.
Taub, S. F. (2001). Language from the body: Iconicity and metaphor in American Sign Language. Cambridge, UK: Cambridge University Press.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
early language development. In L. Balter, TamisLeMondaC. S. (Eds.), Child Psychology: A Handbook
of Contemporary Issues (pp. 116–139). Philadelphia:
Psychology Press.
Bakeman, R., & Gottman, J. M. (1997). Observing interaction: An
introduction to sequential analysis. New York: Cambridge
University Press.
Beykirch, H. L., Holcomb, T. A., & Harrington, J. F. (1990).
Iconicity and sign vocabulary acquisition. American Annals
of the Deaf, 135, 306–311.
Callaghan, T. C. (1999). Early understanding and production of
graphic symbols. Child Development, 70, 1314–1324.
Callaghan, T. C. (2000). Factors affecting children’s graphic
symbol use in the third year: Language, similarity, and
iconicity. Cognitive Development, 15, 185–214.
Callaghan, T. C., & Rankin, M. P. (2002). Emergence of graphic
symbol functioning and the question of domain specificity:
A longitudinal training study. Child Development, 73,
359–376.
Callaghan, T. C., Rochat, P., MacGillivray, T., & MacLellan, C.
(2004). Modeling referential actions in 6- to 18-month-old
infants: A precursor to symbolic understanding. Child
Development, 75, 1733–1744.
Campbell, R., Martin, P., & White, T. (1992). Forced choice
recognition of sign in novice learners of British sign language. Applied Linguistics, 13, 185–201.
DeLoache, J. S. (1991). Symbolic functioning in very young
children: Understanding of pictures and models. Child Development, 62, 736–752.
DeLoache, J. S. (2002). Early development of the use of symbolic artifacts. In U. Goswami (Ed.), Blackwell handbook of
childhood cognitive development (pp. 206–226). Malden, MA:
Blackwell.
Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test-Third Edition. Circle Pines, MN: American Guidance Service.
Folven, R. J., & Bonvillian, J. D. (1991). The transition from
nonreferential to referential language in children acquiring
American Sign Language. Developmental Psychology, 27,
806–816.
Gentner, D., & Medina, J. (1998). Similarity and the development of rules. Cognition, 65, 263–297.
Goldin-Meadow, S. (2003). The resilience of language. New York:
Psychology Press.
Klima, E. S., & Bellugi, U. (1979). The Signs of Language. Cambridge, MA: Harvard University Press.
Layton, T. I., & Holmes, D. W. (1985). Carolina Picture Vocabulary Test. Tulsa, OK: Modern Education.
Lederberg, A. R., Ryan, H. B., & Robbins, B. L. (1986). Peer
interaction in young deaf children: The effect of partner
hearing status and familiarity. Developmental Psychology, 22,
691–700.
Lieberth, A. K., & Gamble, M. E. B. (1991). The role of iconicity in sign language learning by hearing adults. Journal of
Communication Disorders, 24, 89–99.
Meltzoff, A. N. (1995). Understanding the intentions of others:
Re-enactment of intended acts by 18-month-old children.
Developmental Psychology, 31, 838–850.
240
Journal of Deaf Studies and Deaf Education 13:2 Spring 2008
Wilcox, S. (2004). Cognitive iconicity: Conceptual spaces,
meaning, and gesture in signed languages. Cognitive Linguistics, 15, 119–147.
Wilson, M., & Knoblich, G. (2005). The case for motor involvement in perceiving conspecifics. Psychological Bulletin, 131,
460–473.
Received May 3, 2007; revisions received July 16, 2007;
accepted July 28, 2007.
Downloaded from jdsde.oxfordjournals.org at MAX PLANCK INSTITUT FUER EVOLUTIONAERE ANTHROPOLOGIE on November 25, 2010
Tomasello, M., Striano, T., & Rochat, P. (1999). Do young children use objects as symbols? British Journal of Developmental Psychology, 17, 563–584.
Werner, H., & Kaplan, B. (1963). Symbol formation: An
organismic-developmental approach to language and the
expression of thought. New York: John Wiley & Sons, Inc.
White, A., & Tischler, S. (1999). Receptive sign vocabulary
tests: Tests of single-word vocabulary or iconicity. American
Annals of the Deaf, 144, 334–338.