Language and Theory of Mind: A Study of Deaf Children

Abstract

Theory-of-mind (ToM) abilities were studied in 176 deaf children aged 3 years 11 months to 8 years 3 months who use either American Sign Language (ASL) or oral English, with hearing parents or deaf parents. A battery of tasks tapping understanding of false belief and knowledge state and language skills, ASL or English, was given to each child. There was a significant delay on ToM tasks in deaf children of hearing parents, who typically demonstrate language delays, regardless of whether they used spoken English or ASL. In contrast, deaf children from deaf families performed identically to same-aged hearing controls (N=42). Both vocabulary and understanding syntactic complements were significant independent predictors of success on verbal and low-verbal ToM tasks.

Full text

Language and Theory of Mind: A Study of Deaf Children
Brenda Schick
University of Colorado
Peter de Villiers
Smith College
Jill de Villiers
Smith College
Robert Hoffmeister
Boston University
Theory-of-mind (ToM) abilities were studied in 176 deaf children aged 3 years 11 months to 8 years 3 months who
use either American Sign Language (ASL) or oral English, with hearing parents or deaf parents. A battery of tasks
tapping understanding of false belief and knowledge state and language skills, ASL or English, was given to each
child. There was a significant delay on ToM tasks in deaf children of hearing parents, who typically demonstrate
language delays, regardless of whether they used spoken English or ASL. In contrast, deaf children from deaf
families performed identically to same-aged hearing controls (N542). Both vocabulary and understanding
syntactic complements were significant independent predictors of success on verbal and low-verbal ToM tasks.
The purpose of this paper is to examine the possible
role of language in theory-of-mind (ToM) develop-
ment by examining the performance of a large sample
of deaf children. If the development of understanding
others’ mental states requires only normal cognitive
development coupled with acute observation of
human behavior, then as long as a deaf child lives in
a socially and cognitively nourishing environment,
ToM should develop on a normal timetable even if
language were deficient. However, this view that
social observation and cognitive development suf-
fices seems untenable given the increasing number of
studies that show the importance of the child’s lan-
guage environment, and the child’s own language
development, to the development of a ToM. Yet the
precise mechanism by which language has its effects
remains highly controversial. Does language have its
effects only because the tasks demand it, or because it
affects other cognitive processes, or provides a means
of knowledge acquisition? Or is language a more
direct tool of thought in this domain?
Language and ToM Tasks
Initially, attention focused on the fact that the
standard tests of false belief reasoning require rather
sophisticated language skills. For example, the
widely used unexpected contents task (Perner, Le-
ekam, & Wimmer, 1987) involves showing a child a
familiar container, such as an M&M’s (or Smarties)
candy box. Predictably, children believe the con-
tainer holds its familiar contents. However, when
they look inside, the box turns out to contain some-
thing unexpectedFfor example, a plastic fork in the
crayon box. Children are then asked what they
thought (or ‘‘first thought’’) was in the box before
they looked inside. The false belief questions in this
task contain mental state verbs, embedded clauses,
and if/then statements. Another common task, the
unseen change-in-location task (Wimmer & Perner,
1983), involves a character putting an object in a lo-
cation and then leaving the scene. While he is away, a
second character moves the object to a new location.
The first character returns to the scene to get the
object. The child is asked where the character will
look for the object. Even though the question is
simpler, the child needs some linguistic and narra-
tive sophistication in order to follow the story in the
first place. Therefore, clearly a modicum of language
is needed even to participate in the standard tasks.
Language and False Belief Reasoning
There are stronger theoretical claims about why
language might be facilitative, or even necessary, for
reasoning about false beliefs (for an excellent recent
r2007 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2007/7802-0002
This research was funded by NIH Grant DC02872 ‘‘Language
and Theory of Mind in Deaf Children’’ to Smith College, the
University of Colorado, Boulder, and Boston University. We are
grateful to all the consultants and research assistants who made
the study possible: Ben Bahan from Boston University and Gal-
laudet University; Elaine Gale, Jenny Lin, Sarah Hafer, and Gene
Mirus from the University of Colorado in Boulder; Marcia Unger,
Lana Cook, Patrick Costello, and Marie Philip from Boston Uni-
versity; and Jennifer Friedman, Roberta Giordano, and Jennie
Pyers from Smith College. We would like to thank the schools,
families, and children for participating in this study, especially
given the amount of time and testing we needed.
Correspondence concerning this article should be addressed to
Brenda Schick, Speech, Language, & Hearing Sciences, University
of Colorado, 2501 Kittredge Loop Road, SLHS, Boulder, CO 80309-
0409. Electronic mail may be sent to Brenda.Schick@colorado.edu.
Child Development, March/April 2007, Volume 78, Number 2, Pages 376 – 396

survey, see Astington & Baird, 2005). For example,
researchers who emphasize how executive function
skills or working memory capacity contribute to
false-belief reasoning have argued that language fa-
cilitates those cognitive processes (Zelazo, 1999).
Having verbal labels allows the child to keep two
things in mind at once, and being able to remember
verbal instructions enables staying on track in the
task (Gordon & Olson, 1998). Other theorists stress
that language provides evidence about minds to chil-
dren. For instance, Astington (2001) and Dunn and
her colleagues (Brown, Donelan-McCall, & Dunn,
1996; Dunn, Brown, Slomkowski, Tesla, & Young-
blade, 1991; Dunn, 1994) argue that talking about the
mind focuses the child’s attention on explicit mental
explanations of behavior, introducing them to a vo-
cabulary of terms for unseen and abstract concepts
such as thoughts, feeling, ideas, memories, and so
forth that are inaccessible to direct observation (see
also Bartsch & Wellman, 1995; Olson, 1988). Essen-
tially, language provides access to this obscure do-
main. The act of conversation itself has also been
held as an important source of evidence (Harris,
1996; Peterson & Siegal, 2000). As people converse,
the very language they use conveys their intentions,
presuppositions, beliefs, and knowledge, even when
it contains no particular explicit reference to mental
terms. Hence, on that view, language conveys
evidence for minds in action, more so than ordinary
behavior.
Quite apart from the role of language as either a
facilitator or a source of evidence for minds, other
theorists have proposed a different role for language
as enabling representations about mental states. Kar-
miloff-Smith (1992) and Ruffman, Slade, Rowland-
son, Rumsey, and Garnham (2003) argue that the
reflective or offline thinking that language permits
opens up a new level of representation of complex
phenomena (see also Bickerton, 1995). In Karmiloff-
Smith’s terms, the child’s ToM becomes more expli-
citly representational because of the generalized
ability to meta-represent that language enables.
Nelson (1996) supports a similar view, that focuses
on language as a representational tool.
Each of these positions stresses different aspects of
language. The vocabulary of mental state terms is
important for views such as Olson’s (1988) in which
the use of mental state vocabulary is critical. Some
researchers stress the access to interactions, while
others focus on language as a representational tool.
However, in these theories, grammar plays little, if
any, role. Other researchers have emphasized the
contribution that grammar, specifically syntactic
complements, makes to mastering mind-talk (J. de
Villiers, 1995, 2005; de Villiers & de Villiers, 2000,
2003). Mental state verbs fall into a special class of
verbs in that they take a kind of grammatical argu-
ment structure called a complement. Complements
are linguistic structures where one sentence is
embedded within another (Hauser, Chomsky, &
Fitch, 2002). A complement is embedded under
the verb and, importantly, dependent on it. Relevant
to this investigation, the set of mental verbs
(e.g., think, believe, know, forget, pretend, see) and
communication verbs (e.g., say, tell, ask, report,
promise) take complements. Complements can be
irrealisFreferring to states not yet achieved or
hypothetical, as in:
Bill said he would come tomorrow.
Frieda wanted to see the carnival.
In English, irrealis complements are usually en-
coded by modal verbs (should, will) or infinitival
clauses (to see). The complements cannot be judged
as true or false, although promises or desires may or
may not be fulfilled. However, complements can also
be realis (usually clauses with an overt present or
past tense) and refer to states of affairs that are true
or false.
Marge said that her chair collapsed.
Marge said that her chair giggled.
It is a crucial property of these latter forms that a
false propositionF‘‘her chair giggled’’Fcan none-
theless be included in a true sentence. Such a struc-
ture permits the representation of a state of the world
as seen through someone else’s eyes, and the con-
trast of one person’s mental world with another ’s.
These complements are to be distinguished from
other complex clauses with respect to this embed-
ding of truth. Consider the sentence:
Marge screamed after her chair giggled.
Setting aside the possibility of Halloween pranks,
if we know the second clause to be a false propo-
sition, then the whole sentence becomes a falsehood.
An adjunct clause (e.g., after, because, so that) may
be syntactically complex, but it does not have the
right properties for representing contrasting truths.
A relative clause, another form of embedding under
a noun, also has the requirement that all its clauses
be true; therefore, the following statement is false if
the chair did not giggle:
Marge sat on a chair that giggled.
Language and Theory of Mind 377

However, many mental and communication verbs
allow false propositions in their complements.
Marge thought/believed/told him/announced
that the chair giggled.
Learning to distinguish such structures from ad-
juncts, and knowing the verbs with which they can
occur, is a major achievement of preschool syntax
(Roeper & J. de Villiers, 1994). Tager-Flusberg (1997,
2000) and J. de Villiers (1995; de Villiers & de Villiers,
2000; J. de Villiers & Pyers, 2002) proposed that ac-
quisition of this particular form, specifically realis
complement clauses, then provides the child the
representation for false beliefs.
There has been some debate on the role of syntax
in the development of ToM. Perner, Sprung, Zauner,
and Haider (2003) argue that the breakthrough can-
not be in the syntax because in German, desire and
belief verbs have the same surface complement but
children can answer desire questions before belief.
However, J. de Villiers (2005) argues that the com-
plements are not grammatically equivalent, and that
the verb1complement combination is what is crucial.
She claims that the lexical semantics of the verb alone
are insufficient, and the relation of the verb to the
embedded proposition is critical. The range of pos-
sible options in the grammars of languages is only
now being explored.
In principle, then, complement clauses allow the
representation of multiple possible worlds, the
child’s own mind, and the minds of others. Even if a
belief is false, it can be represented in a person’s
mind. Complements therefore provide a way to
discuss lies, mistakes, and other social cognitions
involving false beliefs. This view challenges the
usual thinking that language merely maps concep-
tual understanding. The argument is that the repre-
sentational capacity of the child is enhanced by
having language of this degree of complexity, and
the strongest version of the claim would be that only
a child with language of this degree of complexity
could reason explicitly about the truth and falsity of
the contents of others’ minds (Jackendoff, 1996; Se-
gal, 1998). This is a strong claim regarding the role of
language in ToM development. Language on this
view is more than facilitative; it is necessary.
Evidence that Language Facilitates ToM
Evidence from typically developing preschool
children shows that language plays at least a facili-
tative role in ToM development, but the mechanism
remains ambiguous. For example, many studies
have shown how the linguistic environment of the
child is related to understanding false beliefs (e.g.,
Cutting & Dunn, 1999; Dunn, 1994; Hughes et al.,
2005; Meins et al., 2002; Ruffman, Slade, & Crowe,
2002), but none of the studies can tease apart which
aspect of the language input is facilitative. Second,
several studies have documented that the child’s
own language makes a difference in their late ToM
skills (Farrar & Maag, 2002; Watson, Painter, &
Bornstein, 2001), but few studies pit one kind of
linguistic skill against another to see which is re-
sponsible. Some researchers have tried to distinguish
one kind of influence, such as vocabulary, from an-
other, such as syntax, on a standard test, and found
mixed results, which could be a function of differ-
ential variance in the tasks (Astington & Jenkins,
1999; Ruffman et al., 2002, 2003). The problem with
small cross-sectional studies and even small longi-
tudinal studies in this narrow age range of 3 – 5 years
is that so much is changing at once: grammar is
maturing, vocabulary is expanding, executive func-
tion is being mastered, working memory is increasing,
meta-cognitive skills are emerging, and children are
engaged in rich conversations about social cognitive
topics. Any small-scale study that samples from
these many plausibly relevant abilities alongside
false-belief tasks may find a convincing story to tell
about the interrelationships that another study might
contradict, simply because of the wide range of
variables not considered given the small sample size.
Children in whom development of these skills
may be asynchronous, for whatever reason, help us
see how the relationships among them might vary. In
particular, the performance of deaf children has been
explored because of the opportunity to study a
group of children who have normal social, emo-
tional, and cognitive skills (unlike the case of au-
tism), adequate ability to observe other people’s
behavior, and yet, all too frequently, have delayed
language skills due to lack of access to speech or to
skilled signing (Clark, Marschark, & Karchmer, 2001;
P. de Villiers, 2003; Schick, 2003). By studying deaf
children, one can perhaps assess more directly, over
a longer time span, the role that language plays in
the development of false belief reasoning.
ToM in Deaf Children
For the majority of deaf children, namely those
with hearing parents, language deficits often exist
for those learning American Sign Language (ASL;
Schick & Hoffmeister, 2001; Strong & Prinz, 1997),
English-based signing (Geers, Moog, & Schick, 1984;
Schick & Moeller, 1992), or spoken English (P. de
378 Schick, de Villiers, and de Villiers

Villiers, 2003; Geers et al., 1984). Nevertheless, deaf
children with hearing parents (DoH) are actively
sociable, even with language delays. Studying DoH
children’s ToM reasoning can therefore tease out the
effects of language acquisition from those of cogni-
tive maturation and engagement in social interaction,
at least to the extent that the latter do not themselves
depend on language acquisition. In contrast with
DoH children, deaf children who have deaf parents
(DoD), who provide natural access and exposure to
ASL, demonstrate developmental benchmarks in
language acquisition similar to typically developing
hearing children (Newport & Meier, 1985; Schick,
2003). DoD children provide a natural control for any
effects of deafness per se. If language acquisition
plays a central causal role in ToM development, then
DoH children with delayed language will experience
corresponding delays in their understanding and
reasoning about mental states, but DoD children will
show no delays in ToM development.
Many studies of deaf children suggest some caus-
al role for language acquisition in the development
of an understanding of false beliefs (see P. de Villiers,
2005). Studies that have investigated ToM under-
standing in DoH children have found that these
children often have skills that are quite significantly
delayed compared with their hearing peers (Courtin
& Melot, 1998; Courtin, 2000; Gale, de Villiers, de
Villiers, & Pyers, 1996; Jackson, 2001; Moeller &
Schick, 2006; Peterson & Siegal, 1995, 2000; Peterson,
Wellman & Liu, 2005; Russell et al., 1998; Steeds,
Rowe, & Dowker, 1997), in some cases not reliably
understanding false belief until early adolescence.
Results have been fairly consistent across a variety of
tasks, such as standard tasks involving false belief as
well as appearance-reality tasks. For example, in a
recent study by Peterson et al. (2005), only a third of
the late-signing deaf children ag ed 5.5 – 13.2 years
could pass a false belief task, although the ranking of
five different ToM tasks was similar to that of hear-
ing children, albeit at a much later age (Wellman &
Liu, 2004). Although substantial deficits are reported
in the research literature on deaf children, some
studies find less delay. Moeller and Schick (2006)
showed that deaf children reliably pass false belief
tasks at younger ages (6- and 7-year-olds: 63% pas-
sed; 8- and 9-year-olds: 75%). As would be expected,
there is also evidence that DoD children, who de-
velop language typically, perform significantly better
on ToM tasks than their DoH peers. However, many
of these studies tested DoD children at ages much
older than the age where hearing children typically
pass false belief tasks, with mean ages of participants
ranging from 8 years 11 months to 11 years (Peterson
& Siegal, 1997, 1999) Given the older ages, it is dif-
ficult to determine how DoD children compare with
hearing children in terms of development of false
belief understanding. In the only study so far to
compare DoD with hearing children at the same age,
Courtin (2000) studied ToM skills in a group of 79
French deaf children, including 37 children who
were from signing deaf families (mean age 55 years
4 months). Courtin found that the DoD children
outperformed the 7-year-old DoH children, regard-
less of whether the latter were learning sign lan-
guage or spoken French. The DoD children were
significantly better than a control group of hearing
children on the false belief tasks. As an aside,
Courtin argues that the results show an enhanced
performance in deaf children learning sign language
because most sign languages (if not all) have gram-
matical structures to indicate that point of view and
visual perspective taking underlie aspects of sign
language grammar. Thus, deaf children learning sign
language may be particularly advantaged when
learning concepts about mental states.
Several methodological issues in the studies of
deaf children are problematic. Many of the previous
studies testing signing deaf children have used in-
terpreters or hearing individuals who are reportedly
fluent in sign language. It is also possible that some
results may be due to less-than-fluent sign skills on
the part of the examiner or interpreter. Even when
the interpreter is highly fluent, deaf adults routinely
report that they greatly prefer direct communication,
and information is often lost. Deaf children’s per-
formance may also be hindered by being required
to look at materials presented by the examiner and
simultaneously watch an interpreter, requiring vis-
ual coordination.
An additional problem with many of the existing
studies with deaf children is that only verbal tasks
were used to assess ToM skills. Because many deaf
children have difficulties with language compre-
hension, it is possible that performance on the verbal
tasks is limited by the children’s language skills, and
not by their understanding of ToM, especially given
that the verbal false belief tasks entail language that
is grammatically complex. Because of this, a deaf
child’s failure at a task may reflect general language
limitations rather than ToM issues. However, other
work suggests that DoH children also show delayed
levels of performance on much less verbal or
nonverbal tests of reasoning about cognitive states
(de Villiers & de Villiers, 2000; P. de Villiers & Pyers,
2001; Figueras-Costas & Harris, 2001; Gale et al.,
1996; Woolfe, Want, & Siegal, 2002). These results
suggest that their delayed performance on standard
Language and Theory of Mind 379

tests of false belief reasoning does not just result
from the language demands of the tasks themselves.
According to Peterson and Siegal (1995, 2000), the
primary cause of a delayed development of ToM is
the lack of access to conversations. Because hearing
families with a deaf child have difficulty communi-
cating about everyday routines, they have extreme
difficulty talking about thoughts, beliefs, and inten-
tions. As a result, language-delayed deaf children
miss out on references to abstract, unseen entities
such as mental states, and have less raw material to
develop ToM concepts. Deaf children do not have
any special problems with social interaction other
than that imposed by delayed language skills, unlike
children with autism, who have also been shown to
be significantly delayed in ToM development (Pe-
terson & Siegal, 2000; Tager-Flusberg, 2000). But even
if language-delayed deaf children participate in rich
social interaction, that may not be enough for normal
ToM development.
Very little work has looked at deaf children’s
conversational input directly. Hearing mothers of
deaf children often must also learn the sign vo-
cabulary for mental state concepts and we know that
not all mothers know these signs. Moeller and Schick
(2006) found that hearing mothers varied in their
ability to use signs for mental state terms and that
the mothers’ ability and willingness to talk about the
mind was correlated with their own child’s ability to
pass false belief tasks. There is also evidence that
teachers of the deaf, as well as hearing teachers of
hearing children, vary a great deal in how much they
talk about the mind (Caldwell, Schick, & Hoffme-
ister, 2002). Given that many deaf children often have
a limited range of social partners who can commu-
nicate freely with them, restricted input is a serious
issue for many children.
There is mixed evidence as to whether a deaf
child’s own language skills are related to ToM skills,
and unfortunately, existing studies have not used
very sophisticated language measures. For example,
Lundy (2002) found that the language skills of 34
deaf and hard of hearing children, ages 5 – 10, were
only modestly related to false-belief reasoning, and
age was the most significant predictor of perform-
ance. Unfortunately, the only index of their language
skills was a 56-item rating scale, the Language Pro-
ficiency Profile, filled out by their teachers, and only
verbal false belief tasks were administered. Stronger
evidence was provided by Moeller and Schick (2006),
who found that grammatical skills were highly cor-
related with false belief measures. Both maternal
language and the children’s own language were
highly correlated with the children’s ToM scores, but
in a regression that first took out both age and ma-
ternal language, the children’s own language scores
predicted a small but significant 14% of the variance
in their ToM scores. Similarly, P. de Villiers and Pyers
(2001) found that both vocabulary comprehension
(on the Peabody Picture Vocabulary Test) and syntax
production (the Index of Productive Syntax; Scar-
borough, 1990) were significant predictors of 23 oral
deaf children’s performance on both verbal and
nonverbal tests of false belief reasoning, even when
the effects of age, nonverbal IQ, and hearing loss were
factored out. Other research has shown that even
when DoH children are matched with DoD children in
terms of language age, they are still significantly de-
layed in their false belief reasoning (Woolfe et al.,
2002). Clearly, it is difficult to draw conclusions about
the role of the child’s own language without more
sophisticated measures of that language.
No studies to date have examined the possible
connection of specific complement syntax in ASL to
children’s performance on false belief tasks. There
were no such measures in the Test of British Sign
Language used by Woolfe et al. (2002). Just as in
English, verbs of communication and mental state in
ASL can take embedded propositions as their com-
plements, and these can be false. There is no overt
complementizer, such as that, in ASL, but there are
complements with a wide range of verbs (e.g., want,
think, tell, feel, inform, persuade; Padden, 1987;
Wilbur, 1979). Padden reports examples such as the
following:
MY FATHER PERSUADE ME BUY HIS CAR.
‘‘My father persuaded me to buy his car.’’
MARY HOPE BILL COME VISIT WILL
‘‘Mary hopes Bill will come to visit’’
ASL also distinguishes between irrealis and realis
forms, using explicit lexical tense markers, a full
range of lexical modals, as well nonverbal markers of
modality (Shaffer, 2004, 2005). There is also a rich
system of subject – verb agreement that in many
cases can differentiate the subject of the two verbs
(Padden, 1987). Therefore, like all spoken languages,
ASL has the machinery to carry exactly the meanings
conveyed by English mental verb1complement and it
has the representational power for propositional at-
titudes and possible worlds.
In sum, many studies of both signing and oral
DoH children agree that deaf children with language
delay are significantly delayed in their mastery of
false-belief reasoning. However, existing studies
of ToM in deaf children often suffer from insuffi-
ciencies in design from the perspective of answering
380 Schick, de Villiers, and de Villiers

questions about the role of language (see P. de
Villiers, 2005, for a more detailed review). The extent
to which the native-signing children were delayed or
not in their ToM development could not be assessed
accuratelyFall that could be said was that their
performance was significantly better than that of the
nonnative signers.
In the present study, our goal is to investigate
more thoroughly the ToM skills of deaf children with
and without language delay by investigating the
following predictions:
1. If language skills do not contribute to the un-
derstanding of cognitive states, then language-
delayed deaf children should not be delayed in
ToM as long as the language demands of the
tasks are made unimportant.
2. If general language skills do matter, then either
vocabulary or general grammar should predict
how well children do on ToM reasoning, wheth-
er the task is low or high verbal.
3. If grammar contributes as a representational
tool, then complement mastery with commu-
nication verbs should predict reasoning about
others’ beliefs and knowledge states, whether
the task is verbal or low-verbal.
Method
Participants
The 176 deaf participants ranged in age from four
years to 8 years 3 months, the mean age was similar
for each group (ASL-DoD 56.07; ASL-DoH 56.11;
Oral 56.06), and there were approximately equal
numbers of 4-, 5-, 6-, and 7-year-olds (see Table 1).
An analysis of variance (ANOVA) with post hoc
pair-wise least significant (LSD) tests revealed that
there were no differences in average or distribution
in age between the four groups (Oral-DoH, ASL-
DoD, or ASL-DoH, and Hearing) of 4- through
6-year-old participants, F(3, 164) 51.39, p5.247. We
did not test a group of 7-year-old hearing children
because of ceiling effects on most ToM measures,
even though we did test the 7-year-old DoD children
because previous research has not demonstrated a
ceiling effect in young DoD children and research is
scarce on this population.
Orally taught participants. Eighty-six deaf children
with hearing parents from oral-only educational
settings with hearing teachers participated in the
study (Oral-DoH). These programs focus on the use
of spoken English and do not use any sign language.
The programs were in different geographical regions
of the US: East Coast, West Coast, and Midwest, and
contained ethnically diverse populations (African
American 59.3%; Hispanic 511.6%; Asian 58.1%).
Forty-nine (56.9%) of the children were profoundly
deaf; their average better-ear unaided-hearing loss
was 92 dB (range 47 dB – 120 dB). Fifty-three of the
children wore hearing aids and 33 had cochlear im-
plants. Aided hearing loss from pure tone audiological
testing was available for each of the oral deaf children.
It ranged from 18 to 62dB, with a mean of 35.7 dB. All
of the children experienced their hearing loss before 18
months of age, and they all had nonverbal IQs and
nonverbal sequence memory (tested by Knox’s Cubes
Test) within the normal range. The children repre-
sented a wide range of socioeconomic status ranging
from working class to upper-middle class.
ASL signing participants. Ninety deaf children
were from educational settings that used ASL. These
programs were also in different geographical regions
of the United State: East Coast, West Coast, and
Midwest, and served ethnically diverse populations
(African American 512.1%; Hispanic 514.3%; Asian 5
16.8%). These programs used ASL and written English
as the languages of instruction; no instruction was with
spoken English. Just as important, the children were
surrounded by adults and peers who were fluent in
ASL. Forty-nine children had deaf parents (ASL-DoD);
41 had hearing parents (ASL-DoH). Fifty-seven (63.3%)
of the children were profoundly deaf; average unaid-
ed-hearing loss was 90 dB (range 45 – 120 dB), and all of
the children experienced their hearing loss before the
age of 18 months. As with the oral deaf children, the
signing group all had nonverbal IQs and nonverbal
sequence memory within the normal range.
Hearing control participants. As a comparison
group to determine whether the deaf children in
the different groups were delayed in their mastery
of false belief reasoning, 42 control children with
normal hearing (Hearing) were tested. Although
no exact measures of parental education or income
were obtained, the hearing children were sampled
Table1
Numbers of Children in Each Experimental Group by Age Group
4 years 5 years 6 years 7 years
Hearing 20 11 11
ASL-DoD 11 14 11 13
ASL-DoH 8 10 10 13
Oral-DoH 18 27 17 24
Note. ASL 5American Sign Language, DoD 5deaf children who
have deaf parents, DoH 5deaf children with hearing parents.
Language and Theory of Mind 381

primarily from preschools and elementary schools in
working-class districts to roughly match the socio-
economic background of the ASL children, particu-
larly the DoD children who tend to be from families
in the lower range of income and type of employ-
ment. They ranged in age from 4 years to 6 years 8
months, with an average age of 5 years 4 months,
and all were reported by their schools to have IQs
within the normal range. There was also a range of
race/ethnicity (African American 519.1%; Hispan-
ic 57.1%; Asian 58.1%).
Procedure
Each of the deaf children received a battery of
tasks that included measures of nonverbal intelli-
gence, false-belief reasoning, and language. Testing
occurred individually across four to six testing ses-
sions, each session lasting approximately 30 min. The
hearing control children were tested on all of the
false belief tasks in two individual testing sessions,
but they did not receive the language and nonverbal
IQ tests due to limitations in available testing per-
sonnel when the control data were collected.
For the oral deaf children, the testing was carried
out by examiners who were highly familiar with the
speech of deaf children. All of the children were
using individualized amplification systems (typical-
ly some combination of FM systems, hearing aids, or
cochlear implants). The ASL-signing children were
all tested by deaf examiners with native skills in ASL.
All but one of the examiners had deaf parents. None
of the testing was performed through interpreters.
All ASL examiners had bachelor degrees, most had
master’s degrees, and were certified deaf educators.
Measures of Nonverbal IQ and Memory
The general nonverbal cognitive skills of the deaf
children were assessed by two tests. The Pattern
Construction subtest of the Differential Ability Scales
(DAS; Elliott, 1990) requires the child to manipulate
colored blocks to match pictured patterns of increas-
ing complexity and is standardized for ages 3 years 6
months to adults. Knox’s Cube Test of nonverbal
sequence memory (Stone & Wright, 1979) requires
that the child imitate the examiner in tapping a set of
four identical cubes in an increasingly complex se-
quence. It has been widely used with deaf children.
ToM Measures
Standard verbal false belief reasoning assessments.
There were three unseen change-in-location narratives
(Wimmer & Perner, 1983) presented in a picture se-
quence format. The story was told in simple lan-
guage, using the pictures for clarification. Gale et al.
(1996) argued that the picture sequence story format
was better for deaf children because it was a familiar
educational activity and it is easier to sign and point
to a picture than it is to sign and manipulate puppets
or figures. P. de Villiers and Pyers (2001) reported that
this format was equivalent for hearing preschoolers
to the more typical acted-out story with puppets or
dolls. This was therefore a more suitable format for
the deaf participants because the signer had hands
free to sign, and the oral deaf children could glance
between the book and the speaker to read lips when
the book was held against the speaker’s chest.
In each story, a character put an object in one of
two or three possible locations and then left the
scene. While he or she was away, a second character
moved the object to a new location. Control ques-
tions were asked to ensure that the child remem-
bered where the object was first placed and where it
was moved to. If the child answered one or more of
the control questions incorrectly on the first asking,
the tester turned back in the picture book and re-
viewed the facts of the event again. The control
questions were then asked for a second time. When
the child answered both control questions, he or she
was asked where the character would ‘‘first look’’ for
the object (Siegel & Beattie, 1991).
The children were also tested using two unexpected
contents tasks with two familiar containers: a Crayola
crayon box that turned out to contain a plastic spoon,
and a milk carton that contained a small car (Perner
et al., 1987). The children were asked both about their
own previous false belief (what they ‘‘thought was in
the box before they looked inside it’’) and about what
a friend of theirs would think was in the box before
he or she looked inside it.
Low-verbal ToM tasks. Two low-verbal games that
required minimal language to convey the task or
respond were used to test the children’s ToM rea-
soning about another person’s cognitive states. These
tapped into the children’s understanding of the re-
lationship between seeing and knowing and what
someone might expect based on what they had
seen or not seen. Although these are not strictly an-
alogues of the standard verbal false belief tasks, they
were designed to assess the children’s understand-
ing of a component of ToM (the seeing/knowing
relationship) that is necessary for reasoning about
the characters’ false beliefs in the unseen change-in-
location narratives. Wellman and Liu (2004) and
Peterson et al. (2005) have recently shown that
with equivalent wording of the test questions,
382 Schick, de Villiers, and de Villiers

seeing-knowing is understood before false beliefs in
a verbal test by both hearing and deaf children.
A sticker-finding hide-and-seek task, or hidden
sticker game, was adapted from research by Povinelli
and deBlois (1992) with preschoolers and with
chimpanzees. The children played a game with the
experimenters in which they had to locate stickers
that were hidden in one of four identical white
boxes. A training phase involved only the examiner
and the child, who sat across from each other with a
moveable opaque screen between them. First the
examiner pulled down the screen, hid a sticker in
one of the four boxes, and then opened the screen.
Then the examiner pointed to the box with the
sticker and motioned to the child to take the box, and
keep the sticker. At least two more trials were com-
pleted until the child understood that it was ad-
vantageous always to pick the box that was pointed
to. Then two adult confederates (‘‘helpers’’) were
introduced for the test phase of the game. One of the
confederates (the knower) sat next to the examiner
watching where the sticker was hidden; the other
(the guesser) sat next to the child, screened from
seeing the boxes like the child and wearing a blind-
fold. After the sticker was hidden, the screen was
removed and the confederates moved to the side of
the table opposite the child. Each confederate
pointed to a different box and hence the child re-
ceived ambiguous advice about which box to choose.
The knower pointed to the correct box, and the
guesser pointed to an empty box according to a
preordained rotation. The two adult confederates
took turns in a random sequence at being the knower
or the guesser, so the child could not solve the
problem by just sticking with the same person’s ad-
vice throughout the game. Before each trial the child
was reminded about who was watching and who
was wearing the blindfold. In addition, to reduce the
load on the children’s memory, the guesser lifted the
blindfold, but kept it up on top of her forehead when
she came round to the examiner’s side of the table to
point at a box. Ten test trials with the knower and
guesser were run. In essence the child had to deter-
mine which helper knew where the sticker was
hidden based on their visual access to the event. In
previous studies this game was mastered by hearing
children at approximately the same age as they
passed the standard unseen change-in-location task
(P. de Villiers & Pyers, 2001; Povinelli & deBlois,
1992) and it was significantly correlated with per-
formance on the standard unseen change-in-location
task (P. de Villiers & Pyers, 2001; Gale et al., 1996).
In the surprise face game, the child had to complete
a picture at the end of a story sequence by putting a
surprised or not surprised (neutral) face on a character
depending on the character’s knowledge state (see
de Villiers & de Villiers, 2000; P. de Villiers & Pyers,
2001). P. de Villiers and Pyers reported that per-
formance on the surprise face game was significantly
correlated with scores on the standard false-belief
reasoning tasks in hearing preschoolers and oral deaf
children. Stories involved a familiar container that
would lead someone to expect it to contain certain
items, but in which an unusual object had been
substituted. In three of the stories, the character was
not surprised to discover the substituted object and
in three of the stories, the character was surprised. In
the first type, the target character clearly watched the
unusual item being substituted for the familiar con-
tents of the box, and hence would not be surprised
when he or she tipped that item out of the box at the
end of the picture sequence. In the second type, the
target character had not observed the substitution
and saw only the closed box until the strange con-
tents were revealed, and so would be surprised. The
child had to indicate whether the character would be
surprised or not surprised by the unusual contents of
the container at the end of the picture sequence. The
child did this by choosing between two pictures of
the character’s face: one with the facial expression
showing surprised and the other showing not surprised.
The child placed transparencies of these expressions
over the character’s blank face in the final picture of
the story, the picture in which the unusual contents
of the box were revealed. In the warm-up to the task,
the general idea of the procedure was communicated
to the child through two stories in which characters
were surprised or not surprised depending on what
they had seen before, as well as the vocabulary
(‘‘surprised,’’ ‘‘not surprised,’’ ‘‘see,’’ and ‘‘did not
see’’). The stories were communicated through ges-
tures and minimal language, ascertaining that the
child could correctly identify which face represented
surprised and not surprised. On these trials the chil-
dren were corrected if they chose the wrong face.
The testing phase consisted of six picture se-
quences, each consisting of five colored pictures. As
in the warm-up, minimal language was used in
playing the game. Pointing was used to direct the
child’s attention to the crucial objects and characters
in each picture. The examiner pointed to the char-
acter in the final picture and then back to that char-
acter in each of the pictures in which he or she
appeared and said (or signed): ‘‘This boy/girl . . .
Which face? Surprised or not surprised? (pointing to
the two faces on the transparencies).’’ The order of
the two faces and emotion words varied across
sequences and across children for each sequence.
Language and Theory of Mind 383

Language Measures: Oral Deaf Children
Spoken one-word vocabulary comprehension and pro-
duction was assessed by two picture-based tests of
word knowledge, widely used with both hearing
and deaf children. The Peabody Picture Vocabulary
Test – Revised (PPVT – R; Dunn & Dunn, 1981) re-
quires the child to choose which of four pictures
match the experimenter’s spoken word, and
the Elicited One-Word Picture Vocabulary Test
(EOWPVT; Gardner, 1990) asks the child to name a
single picture and tests primarily simple nouns and
category labels such as fruit.
General spoken English syntax comprehension was
tested on the Sentence Structure subtest of the Clin-
ical Evaluation of Language Function for preschool-
ers (CELF – Preschool; Wiig, Secord, & Semel, 1992).
This subtest consists of 22 sentences varying in
complexity including prepositional phrases, coor-
dinated clauses, passive voice, and relative clauses,
but it does not include any complement clauses. The
stimulus sentences were presented orally and the
child had to choose which of three pictures best de-
picted the meaning of the sentence.
Comprehension of false complement clauses with
communication verbs (tell) was tested on a memory
for complement clauses task (P. de Villiers & Pyers,
2002). The child was shown two colored photo-
graphs depicting the events of a brief two-sentence
anecdote in which a character was described as
making a mistake or telling a lie. The four items in-
volved what the character told someone, (e.g., ‘‘She
told the girl there was a bug in her hair (Picture 1).
But it was only a leaf (Picture 2, close up).’’ After
each sequence the child was asked ‘‘What did s/he
tell X?’’
To answer correctly, the child had to process the
syntax and remember the embedded false comple-
ment with the verb of communication in the story.
J. de Villiers and Pyers (2002) argue that the com-
munication verb does not require any understanding
of false beliefs. In their longitudinal study of hearing
preschoolers, J. de Villiers and Pyers showed that
performance on communication verbs around age
3:4 was the primary predictor of later levels of false-
belief reasoning (around age 3:8). Indeed, memory
for false complements was a separate and stronger
predictor than other more general measures of the
children’s language level.
Language Measures: ASL-Signing Deaf Children
Because there are no existing standardized tests
for ASL, all of the tests were designed specifically for
this study, or were adapted from tests for spoken
English. The design of the tests and the items were
based on the research literature on which aspects of
ASL show developmental timetables across the age
range included in this study, mostly focusing on the
use of complex syntax, morphology, and vocabulary
(Schick, 2003). Three tests of ASL were devised that
parallel the English language assessments carried
out with the oral deaf children: a test of ASL sign
vocabulary based on the structure and format of the
PPVT – R; a test of general ASL syntax comprehen-
sion that did not include complement clauses; and an
ASL version of the false complement comprehension
test. All of the tests were reviewed by linguistically
trained native signers of ASL and were pilot tested
on signing deaf children, with deaf parents and with
hearing parents.
The Receptive ASL Vocabulary Test (ASLVT) was
created to assess the children’s receptive vocabulary
in ASL (Schick, 1997), modeled on the PPVT – R
(Dunn & Dunn, 1981). The PPVT – R itself was
not used for two reasons. First, in translating the
test from English into ASL, there are not always
equivalent signs in ASL. Consequently, some words
must be represented using fingerspelling, which
uses a manual alphabet to spell a word. Children’s
ability to recognize English words in fingerspelling
is not a reflection of their vocabulary skills in ASL,
although the two appear to be related (Padden &
Ramsey, 2000; Sedey, 1995). Second, in the PPVT – R,
the foils were not designed to prevent a child from
using iconicity in a sign to determine which picture
is a plausible response. For example, in the PPVT – R,
in the plate for the target word wheel, the correct
picture is the only one with something round. The
ASL sign for wheel easily communicates the fact that
you are talking about something round, and hence
the child might correctly guess without knowing the
word. In contrast, in the ASLVT, foils were selected
to prevent a correct guess due to iconicity. For ex-
ample, the ASL sign for compass has an element that
iconically represents that the object has a long thin
component. Because of this, the foils consisted of
pictures of a weathervane, a syringe, and a telescope
and hence the iconic clue was ambiguous on its own.
Vocabulary was selected using the following cri-
teria. The sign had to be a commonly accepted ASL
sign, not a word borrowed from English, with min-
imal regional variation. No sign had an obligatory
lexical nonmanual component, such as facial ex-
pression. Each item on the test was reviewed by a
team of hearing and deaf native signers, all with
extensive experience with deaf children who use
ASL. Items were sorted according to perceived
384 Schick, de Villiers, and de Villiers

difficulty, with easier signs first. Pilot testing was
conducted with signing deaf children in the age
range of the larger study, with deaf parents and with
hearing parents. Items were revised, and the order of
the items was changed based on the pilot testing. The
final test consists of 61 plates, each with four pictures.
During testing, the examiner signed a single sign and
the child selected the matching picture. Testing was
stopped when the child missed 10 items in a row. The
percent of the signs scored as correct was used as the
child’s vocabulary comprehension score.
It was not suitable to simply translate an existing
standardized test such as the CELF – Preschool Sen-
tence Structure subtest into ASL because its gram-
matical structure is very different from English;
therefore, it was necessary to devise a new test of
comprehension of ASL syntax. The new test consisted
of a series of 16 plates, each with three or four pic-
tures, which were sufficiently similar that a child
would need to understand some aspect of complex
syntax or morphology in order to choose the right
picture. For this test, the examiner signed a sentence
and the child selected the appropriate picture. The
test included sentences in which a syntactic object
was topicalized and moved to a sentence-initial
position. It also included complex forms of verb
agreement, a morphological marker for person, with
several participants in a scene. It did not include
complements; therefore, in this sense it served the
same purpose as the CELF – Preschool for the oral
deaf children, namely, as an index of how much
general grammar and morphosyntax of ASL the
child understood. The selection of the final 16 items
was based on pilot testing with signing deaf children
in the age range of the larger study, with deaf parents
and with hearing parents. An ASL general syntax
score was calculated as percent correct.
Comprehension of false complement clauses was as-
sessed by translating the communication verb items
from the memory for complements task into ASL
and administering the test in the same way as for the
oral deaf and hearing children. The translations were
made by native signers in the research team and the
task was pilot tested with signing deaf children in
the age range of the present study.
Results
Background Matching of Groups
Before comparing the different groups of deaf
children from different backgrounds, it is necessary
to ensure that they are matched in background
measures. The three groups of deaf children were
closely matched on almost all of the background
measures. Scores for the DAS, Test of Nonverbal
Intelligence – 2 (TONI – 2), and the Knox Cubes test
are shown in Table 2. ANOVAs revealed no statistical
differences between the groups in age, F(2, 173) 5
0.02, p5.98, hearing loss, F(2, 167) 50.41, p5.66, or
action sequence memory on Knox’s Cubes Test,
F(2, 173) 50.75, p5.47. On the DAS Pattern Con-
struction subtest, there was a significant group effect,
F(2, 173) 56.24, po.01, and a post hoc LSD compar-
ison showed that the ASL-DoD children had signif-
icantly higher scores on that test than the Oral-DoH
children (LSD, po.05). The ASL-DoH children
scored midway between the ASL-DoD and Oral-
DoH children, but were not statistically different
than either of those groups.
The groups of signing deaf children were also
compared in terms of the three ASL measures. There
were significant differences between the groups on
the ASLVT, F(1, 89) 58.752, p5.004, but not for the
Syntax Comprehension, F(1, 89) 5.621, p5.433. On
Complement Processing the difference between the
two groups approached significance, F(1, 89) 53.444,
p5.067FASL-DoD Means: 4-year-olds 51.9 (1.6);
5-year-olds 51.8 (1.4); 6-year-olds 53.6 (0.7); 7-year-
olds 53.5 (1.2); ASL-DoH Means: 4-year-olds 51.5
(1.4); 5-year-olds 52.0 (1.3); 6-year-olds 51.9 (1.9);
7-year-olds 52.6 (1.6).
Table2
Mean Scores and Standard Deviations for Each Age Group of Deaf
Children on the DAS, TONI-2, Knox Cubes Test, and the Complement
Processing Test
4 5 6 7 Total
M SD M SD M SD M SD M SD
DAS
ASL-DoD

53.8 5.6 49.8 5.5 56.5 9.4 53.6 9.6 53.2 7.9
ASL-DoH 50.3 6.8 49.5 8.3 51.7 4.6 52.8 13.2 51.2 9.1
Oral

50.3 5.0 47.3 6.2 47.1 7.3 48.3 11.6 48.1 8.1
TONI – 2
ASL-DoD 94.7 15.8 116.5 14.0 102.5 17.2 103.7 17.8
ASL-DoH 104.4 12.3 98.6 13.9 102.4 18.2 101.8 15.0
Oral 101.6 13.8 93.9 17.1 98.3 14.9 98.5 15.1
Knox Cubes
ASL-DoD 5.9 1.9 6.6 3.2 10.9 2.2 10.2 2.7 8.4 3.3
ASL-DoH 4.6 2.4 5.8 1.7 7.2 2.8 11.5 2.4 7.8 3.6
Oral 5.7 2.6 6.6 2.5 8.3 2.9 10.0 3.0 7.7 3.2
Note. ASL 5American Sign Language, DAS 5Differential Ability
Scales, DoD 5deaf children who have deaf parents, DoH 5deaf
children with hearing parents, TONI – 2 5Test of Nonverbal In-
telligence – 2.

ASL-DoD4Oral, po.05
Language and Theory of Mind 385

The normally hearing control group was also
closely matched in age to the three groups of 4- to 6-
year-old deaf children, with no significant differen-
ces between any of the groups in average age,
F(3, 164) 51.39, p5.247.
Scoring of ToM Tasks
For the standard verbal false belief tasks, the chil-
dren were given one point for each of the false belief
questions for which they gave the correct answer, the
three ‘‘where will X first look?’’ questions on the
unseen change-in-location narratives, and two own
belief and two other belief questions on the unexpected
contents containers. For children in all groups, scores
on the unseen change-in-location and the unexpected
contents tasks were significantly correlated (ASL:
r5.37, df 588, po.001; Oral: r5.31, df 584, po.01;
Hearing: r5.61, df 540, po.001). Therefore, a com-
bined verbal false belief score (out of seven false belief
questions) was calculated for each child by combin-
ing the scores from the standard tasks. Chance in the
verbal false belief tasks is considered to be zero be-
cause the default is that the child responds according
to his own belief, not randomly.
On the low-verbal tasks, however, chance guess-
ing produces 50% correct responses and hence the
raw scores for those games are not particularly use-
ful. Instead, the two low-verbal games were scored
on a pass-fail basis with a statistical criterion for
passing (a5.05). Children passed the Sticker Find-
ing game if they located the hidden sticker, that is,
they took the advice of the knower, on nine or more
of the ten test trials (po.05 on a Sign test). On the
Surprise Face game, children passed if they chose the
correct face (surprised or not surprised) on all six of the
six picture sequences (po.05 on a sign test). Pass –
fail performance on the two low-verbal games was
significantly correlated for each of the three groups
of children (ASL: r5.36, df 588, po.001; Oral:
r5.27, df 584, po.01; Hearing: r5.51, df 536,
po.01). Hence, again, a combined Low-verbal ToM
Tasks Score was calculated as the number of tasks
(out of two) that each child passed.
Performance of the Different Groups of Participants
In the next analysis, we compare the performance
of the different groups on both the verbal and the
low-verbal ToM tasks across age. For each group,
Figure 1 shows the average total score on the
standard verbal false belief tasks across the different
age groups. Figure 2 shows the average number of
low-verbal tasks passed by the children in the dif-
ferent groups by age. Table 3 shows the percent of
children who scored perfectly on the verbal and low-
verbal tasks.
ANOVAs with age in years and group as fixed
independent factors compared the average per-
formance of the different groups on the standard
verbal false belief tasks and the low-verbal tasks.
Only the data from the children aged 4 through 6
were considered for this analysis so that the per-
formance of the deaf children could be compared
with that of the hearing controls matched for age.
For the children’s performance on the verbal false
belief questions, there was a significant main effect
for age in years, F(2, 156) 57.26, po.01, and for group,
F(3, 156) 510.07, po.001. Post hoc LSD tests demon-
strated that the hearing control children and the DoD-
ASL children were indistinguishable on the verbal
ToM tasks (LSD, p5.863), but both of those groups
were significantly better than the two groups of deaf
children with hearing parents: Hearing versus ASL-
DoH (LSD, po.01), Hearing versus Oral-DoH deaf
(LSD, po.001), ASL-DoD versus ASL-DoH (LSD
po.01), ASL-DoD versus Oral-DoH (LSD po.01).
The performance of the two groups of deaf children
with hearing parents was equivalent on the verbal
tasks (LSD, p5.785); hence, at these ages there was
no significant effect of the predominant language of
school instruction, ASL versus oral English.
On the low-verbal games, there was again a sig-
nificant main effect for age in years, F(2, 156) 511.72,
po.001, and for group, F(3, 156) 55.15, po.01. The
two groups with normal language acquisition
(hearing controls and ASL-DoD children) were
equivalent in their performance (LSD, p5.511), and
they were significantly better than the children in the
two groups with hearing parents who were delayed
in their language acquisition: Hearing versus ASL-
DoH (LSD, po.05), Hearing versus Oral-DoH deaf
(LSD po.01), ASL-DoD versus ASL-DoH (LSD,
po.01), ASL-DoD versus Oral-DoH deaf (LSD,
po.01). Again no significant difference was observed
in this age range between the ASL-DoH and Oral-
DoH deaf groups (LSD, p5.941). The LSD post hoc
analysis of group effects did not reveal any earlier
ToM understanding in the ASL-DoD children versus
the Hearing control children on either the verbal
false belief tasks or the low-verbal seeing/knowing
games. Further analyses using ttests to compare the
ASL-DoD and Hearing children also revealed no
significant differences for either measure of ToM at
any of the three age groupingsF4-year-olds, 5-year-
olds, or 6-year-olds.
Because there were group differences for the
4- to 6-year-old DoD and DoH children on DAS
386 Schick, de Villiers, and de Villiers

performance, a separate analysis of covariance
ANCOVA was calculated using the DAS Tscore
as the covariate. Results were similar for the ver-
bal tasks, showing a significant group effect,
F(2, 126) 55.97, po.01, as well as an age effect,
F(2, 126) 54.14, po.05, and for the low-verbal tasks;
group: F(2, 126) 53.93, po.01; age: F(2, 126) 5
12.38, po.001. The three groups of children also
differed in their performance on the complement
processing tasks, showing a significant group effect,
F(2, 175) 57.12, po.001, as well as an age effect,
F(2, 175) 53.11, po.05.
An ANOVA was also carried out on the three
groups of 7-year-old deaf children (ASL-DoD, ASL-
DoH, and Oral; see Figures 1 and 2). There were
main effects for group for the verbal false belief
score, F(2, 47) 53.13, po.05, and the number of low-
verbal tasks passed, F(2, 47) 54.97, po.01. Post hoc
LSD analyses revealed that at age 7, the native-
signing ASL-DoD children were still significantly
better than the Oral children on both the standard
verbal and the low-verbal tasks (verbal false belief:
LSD po.05; low-verbal tasks: LSD, po.01). The ASL-
DoH children were now much more similar to the
ASL-DoD children in their performance on both the
verbal and low-verbal ToM tasks. Post hoc LSD tests
revealed no significant differences at age 7 between
the ASL-DoD and ASL-DoH groups (verbal false
belief: LSD, p5.473; low-verbal tasks: LSD, p5.590),
but the ASL-DoH children were now significantly
better than the Oral-DoH children on the low-verbal
tasks (LSD po.05), although not on the verbal false
belief tasks (LSD, p5.127).
To summarize, there were no differences between
the Hearing children and the age-matched native-
signing ASL children who had deaf parents, that is,
the two groups of children expected to have typical
language acquisition. However, both of these groups
were significantly better than the two groups of deaf
children with hearing parents, the ASL-DoH and
0
1
2
3
4
5
6
7
4567
Age Group
ASL DoD
ASL DoH
Oral
Hearing
Figure 1. Mean total score of the children in each group on the standard verbal false belief tasks (out of 7 false belief questions).
0.0
0.5
1.0
1.5
2.0
45 76
Age Group
ASL DoD
ASL DoH
Oral
Hearing
Figure 2. Average number of low-verbal seeing/knowing tasks passed by the children in each group.
Language and Theory of Mind 387

Oral children, on both the standard verbal false-
belief reasoning tasks and the low-verbal ToM
games. By age 7 the ASL-DoH children who had now
had several years of exposure to intensive ASL ap-
peared to be catching up with the native signers,
falling between the ASL-DoD and the Oral-DoH
group in their performance on the low-verbal ToM
tasks but not on the verbal tasks.
Predictors of ToM Performance in the Deaf Children
To begin to address the prediction about factors
contributing to the development of a mature ToM,
the relationships among background measures, per-
formance on the ToM tasks, and language skills in
the deaf children were analyzed in several ways,
using the total verbal false belief reasoning score (out
of 7) and the number of low language games that the
children passed (out of 2) as the dependent variables.
First, bivariate correlations were calculated (Table
4) between the two ToM scores and the background
measures of age, hearing loss, nonverbal IQ on the
DAS Pattern Construction subtest, and action se-
quence memory on Knox’s Cubes Test. This analysis
was carried out separately for the ASL and Oral deaf
groups. Age and sequence memory were signifi-
cantly correlated with both verbal and low-verbal
tasks for both groups of deaf students. DAS non-
verbal IQ was significantly correlated with both ToM
scores for the ASL children, but not for the Oral
children. There were no significant relationships
with degree of hearing loss but there was a limited
range of hearing loss in the sample, which included
mostly children with profound losses.
Partial correlations were calculated to explore the
relationship between the language measures and the
children’s ability to pass the ToM tasks, controlling
for the effects of the three most significant back-
ground variables (Age, DAS nonverbal IQ, and se-
quence memory; see Table 5). The analysis was
conducted separately for the ASL and Oral children
because the language measures were not identical.
For both the ASL and Oral children, all of the lan-
guage measures were significantly correlated with
the verbal false belief score. For the low-verbal tasks
(which were scored differently than the verbal tasks),
however, the general syntax comprehension measure
was not a significant predictor of the ASL children’s
performance. Only ASL vocabulary comprehension
and processing of false communication complements
on the complement comprehension task were sig-
nificantly related to the ToM score.
The most effective procedure for teasing out the
contribution of the different variables to the per-
formance of the children on the false belief tasks is a
Table3
Percent of Children Who Scored 7/7 for the Verbal False Belief Tasks and
2/2 for the Low-Verbal Tasks in each Experimental Group by Age Group
Tasks 4 years 5 years 6 years 7 years
Hearing Verbal 10 64 64
Low-verbal 4 18 27
ASL-DoD Verbal 0 14 36 54
Low-verbal 9 14 36 62
ASL-DoH Verbal 0 0 0 46
Low-verbal 0 0 10 46
Oral-DoH Verbal 0 13 18 25
Low-verbal 0 7 6 12
Note. ASL 5American Sign Language, DoD 5deaf children who
have deaf parents, DoH 5deaf children with hearing parents.
Table4
Bivariate Correlations of Background Measures and ToM Scores for ASL
and Oral Deaf Children
Verbal false belief
score
Low-verbal tasks
passed
ASL Oral ASL Oral
Age .46

.41

.61

.33

DAS nonverbal IQ .22

.04 .25

.16
Sequence memory
(Knox’s Cubes)
.38

.27

.51

.32

Hearing loss .08 .05 .09 .02
Note. ASL 5American Sign Language, DAS 5Differential Ability
Scales.

po.05,

po.01,

po.001.
Table5
Partial Correlations for the ASL and Oral Children between Language
Measures and ToM Scores, Controlling for Age, Nonverbal IQ (DAS),
Sequence Memory (Knox’s Cubes), and Hearing Loss
Verbal false belief
score
Low-verbal
tasks passed
ASL Oral ASL Oral
ASL vocabulary
comprehension
.51

.57

.32

.26

ASL general syntax
comprehension
.41

.47

.18 .28

False communication
complement processing
.42

.46

.37

.37

Note. ASL 5American Sign Language.

po.05,

po.01,

po.001.
388 Schick, de Villiers, and de Villiers

regression analysis that separates out the indepen-
dent effects of different background and language
measures on each of the dependent ToM measures
(total verbal false belief score and number of low-
verbal games passed). Separate linear regressions
were therefore calculated for the ASL and Oral
children. In each regression the background mea-
sures that had shown significant correlations with
the ToM measures were first entered as a block. Once
any shared variance from the background variables
was removed, the language measures were entered
as a second block of predictors to determine what
additional effect language skills seemed to have on
the children’s ToM.
For the verbal false-belief score (Table 6), the same
results were observed for both the ASL children and
the Oral children. The set of background measures
was a significant predictor of false belief reasoning
for each groupFASL: F(3, 84) 58.76, po.001; Oral:
F(3, 76) 53.94, po.01Fwith age being the strongest
and only significant independent predictor in the set.
For both groups, language skills accounted for a
substantial additional percentage of the variance in
ToM. Vocabulary comprehension and the processing
of false complement clauses were independent sig-
nificant predictors for both ASL and Oral children.
However, general syntax measures were not signifi-
cant independent predictors.
For the low-verbal ToM tasks (Table 7), back-
ground measures were again significant predictors
of the children’s performanceFASL: F(3, 84) 519.85,
po.001; Oral: F(3, 77) 53.54, po.05. Age was the
strongest predictor among the background variables
for both ASL and Oral children. The language
measures again accounted for a significant addi-
tional percentage of the variance in ToM. Processing
of false complement clauses with verbs of commu-
nication was the only reliable significant inde-
pendent predictor among the language variables,
although vocabulary comprehension narrowly mis-
sed significance for the ASL children (p5.076). As in
the case of the verbal tasks, general syntax measures
were not significant independent predictors of per-
formance on the low-verbal games.
To summarize the regression analyses, age and
processing of false complement clauses with verbs of
communication were independent predictors of the
deaf children’s reasoning about false beliefs or states
of knowledge, whether the reasoning was tested in the
standard verbal tasks or in games with much lower
Table6
Regression Analysis Predicting Verbal False Belief Reasoning ScoreFASL and Oral Children
bTP
ASL group
Step 1: Background control variables entered
Age .40 3.02 .003

Nonverbal IQ (DASFPattern Construction) .15 1.44 .155
Action sequence memory (Knox’s Cubes test) .06 0.47 .638
Percentage of variance (R
2
) accounted for by background variables 523.8%
Step 2: Language variables then entered
ASL vocabulary .41 2.91 .005

Complement processing .22 2.19 .031

ASL general syntax .12 1.01 .317
Percentage of additional variance (DR
2
) accounted for by language measures 524.5%
Oral group
Step 1: Background control variables entered
Age .30 2.24 .028

Nonverbal IQ (DASFPattern Construction) .06 0.45 .654
Action sequence memory (Knox’s Cubes Test) .10 0.67 .505
Percentage of variance (R
2
) accounted for by control variables 513.5%
Step 2: Language Variables then Entered
PPVT – R vocabulary comprehension .48 3.48 .001

Complement processing .24 2.31 .024

CELF – Preschool sentence structure subtest .06 0.48 .622
Percentage of additional variance (DR
2
) accounted for by language measures 533.6%
Note. ASL 5American Sign Language, CELF 5Clinical Evaluation of Language Function, DAS 5Differential Ability Scales, PPVT –
R5Peabody Picture Vocabulary Test – Revised.

po.05,

po.01,

po.001.
Language and Theory of Mind 389

language requirements for the tasks. Vocabulary
comprehension emerged as a strong predictor in the
more verbal tasks, but not for the low-verbal games.
Very similar results were obtained in these analyses
for both the Oral deaf children and the ASL signing
children, despite the difference in language modality
and syntax between ASL and spoken English.
Discussion
Prediction 1 stated that deaf children at risk for
language delay should show good performance on a
regular timetable if the language requirements of the
tasks are made unimportant. This prediction receives
clear disconfirmation in the present study. Regard-
less of the language demands of the task, the deaf
children from hearing families, who were delayed in
language compared with their DoD peers (Schick &
Hoffmeister, 2001), are also significantly delayed in
their reasoning about cognitive statesFboth false
beliefs and knowledge versus ignorance. This study
included both traditional verbal tasks to assess false
belief understanding as well as tasks that required
minimal language skills, even though these tasks
assess slightly different aspects of performance (and
they are scored differently). As predicted, the DoH
children were significantly delayed on the low-ver-
bal tasks as well as the tasks requiring either the
comprehension or production of complex language
to participate. These results show that it is not simply
the language of the task that causes deaf children
with language delay to demonstrate a delay in such
tasks. Rather, the reasoning about the mental states is
the problem for the child. Even when the language
demands are minimized, deaf children who have
hearing families are significantly delayed compared
with deaf children who are acquiring ASL from birth
from their parents, or hearing children acquiring
English. The nonverbal tasks were not passed easily
by the DoH children, even though they can be argued
to tap slightly earlier achievements such as seeing
and knowing rather than false belief. These data con-
firm results by obtained Figueras-Costas and Harris
(2001), Gale et al. (1996), and Woolfe et al. (2002).
As in other studies, the deaf children with hearing
parents were delayed in their ability to reason about
false beliefs and knowledge states. This was true for
deaf children being educated using spoken English
(Oral-DoH) as well as DoH children being educated
using ASL (ASL-DoH). In contrast, deaf children
Table7
Regression Analysis Predicting Number of Low-Verbal Tasks PassedFASL and Oral Children
btp
ASL group
Step 1: Background control variables entered
Age .53 4.62 .001

Nonverbal IQ (DASFPattern Construction) .15 1.68 .097
Action sequence memory (Knox’s Cubes test) .099 0.81 .421
Percentage of variance accounted for by background variables 541.5%
Step 2: Language variables then entered
Complement processing .26 2.63 .010

ASL vocabulary .25 1.80 .076
ASL general syntax .067 0.58 .564
Percentage of additional variance (DR
2
) accounted for by language measures 510.0%
Oral group
Step 1: Background control variables entered
Age .27 1.98 .050

Nonverbal IQ (DASFPattern Construction) .12 0.89 .375
Action sequence memory (Knox’s Cubes test) .09 0.61 .543
Percentage of variance (R
2
) accounted for by control variables 512.1%
Step 2: Language variables then entered
Complement processing .26 2.13 .036

PPVT – R vocabulary comprehension .11 0.70 .489
CELF – Preschool sentence structure subtest .08 0.50 .616
Percentage of additional variance (DR
2
) accounted for by language measures 512.6%
Note. ASL 5American Sign Language, CELF 5Clinical Evaluation of Language Function, DAS 5Differential Ability Scales, PPVT –
R5Peabody Picture Vocabulary Test – Revised.

po.05,

po.01,

po.001.
390 Schick, de Villiers, and de Villiers

who have deaf parents, and who learn sign language
naturally from birth, performed much like the hear-
ing children, with no significant differences between
the two groups on any of the ToM tasks, either verbal
or low verbal. Contrasts at each ageF4, 5, and 6F
confirmed no differences. This means that the deaf
children with hearing parents are not delayed in
ToM because of their deafness per se. Early access to
an equivalent language in another medium, namely
ASL, is just as effective for communicating ToM, as
demonstrated by the DoD performance. However,
the current study does not show that the DoD children
are advantaged in their ToM development compared
with their hearing peers, providing no evidence that
sign language was facilitative over spoken language,
in contrast to a finding by Courtin (2000).
In sum, any theories that propose that ToM, even
false belief reasoning, is not affected by language
skills, would have difficulty with the results found
here. These deaf children are well above the usual
age at which false belief reasoning is mastered, and
yet they are often failing the tasks of all varieties. It is
clear that language is a critical factor in children’s
ability to reason about the mind.
It is also important to note that in many aspects of
ToM, deaf children reveal a rich understanding of
other people’s mental states. For example, deaf
children are good at predicting simple emotional
reactions from stereotypical situations (P. de Villiers,
Hosler, Miller, Whalen, & Wong, 1997; Pyers & P. de
Villiers, 2003) and readily appeal to desires and other
mental states as explanations of emotion (Rieffe &
Terwogt, 2000). In contrast, their performance on
tasks involving judgments about emotions based on
characters’ false beliefs is delayed to the same degree
as their reasoning about false beliefs on the classic
tasks (P. de Villiers et al., 1997; Pyers & P. de Villiers,
2003). P. de Villiers et al. (1997) found that oral deaf
children’s judgments about characters’ emotional
reactions in different causal situations were pre-
dicted by different background and language mea-
sures, specifically mastery of cognitive state verbs
and the syntax of complementation. The children’s
understanding of simple emotions based on stereo-
typical situations was predicted by nonverbal IQ and
age, not by language measures or by their reasoning
on standard false belief tests.
Furthermore, deaf children’s delay in ToM rea-
soning seems to be specific to the representation of
cognitive states that do or do not correspond to
perceived reality. Other research has shown that they
do not have any problem in judging the contents of a
physical representation (a photograph) that no
longer reflects the scene that is in front of them (P. de
Villiers & Pyers, 2001; Peterson & Siegal, 1997; Pe-
terson, 2002). In fact their performance on the false
photographs test (Zaitchik, 1990) matched that of
hearing peers of the same age (P. de Villiers & Pyers,
2001). Thus, their failure on traditional false belief
reasoning tasks is not the result of general meta-
representational problems or being unable to detach
themselves from the tendency to give reality answers
to the test questions.
Another possible explanation of deaf children’s
difficulties with ToM reasoning relates to potential
delays in the inhibitory or control features of lan-
guage in language-delayed deaf children. For ex-
ample, having language of a certain degree can be
shown to assist working memory (Gathercole & Ba-
ddeley, 1993), and may also allow the child to inhibit
prepotent responses of the sort that could undermine
performance on a false belief task (Jacques & Zelazo,
2005). The present study cannot rule these out as
influences on performance. However, there was no
discernible effect of deafness on Knox’s cube task, as
compared with reported standards for hearing chil-
dren, which can be considered to have some loading
on at least working memory. Furthermore, other
evidence is accumulating that deaf children are not
delayed on a battery of executive function tasks, or
not sufficiently delayed to explain the delay in false
belief reasoning (P. de Villiers, 2005; Jackson, 2002;
Woolfe et al., 2001).
Role of Language in ToM Development
What impact do the results of this study have for
the different theories about the role of language for
false belief reasoning in the literature? We discuss
our data in the light of these theories and what still
needs to be investigated.
Prediction 2 proposed that if general language
matters then measures of the deaf child’s vocabulary
and general grammar may predict ToM measures. In
contrast, Prediction 3 stated that if language plays a
role as a more specific representational tool, then
complement mastery should predict reasoning about
others’ mental states.
General grammar skills were not found to be
predictive of ToM performance in the current study,
despite the fact that the language measures included
other types of complex grammar in ASL and English.
This finding is evidence against the argument that
language is a proxy (like age) for maturation and
might simply indicate the amount of linguistic ex-
posure the child may have had. The specific ability to
process syntactic complements was predictive,
which was consistent with Prediction 3. That is,
Language and Theory of Mind 391

complements may have a role in the ability to talk
about and represent mental state concepts. These
data are in contrast to those of Ruffman et al. (2003),
who argued that general language skills were better
than embedded clauses at predicting false belief
performance in his sample of hearing preschool
children studied longitudinally.
Ruffman et al. (2003) did not use complement
clauses as a measure of complex grammar in their
study because they argued that such clauses entail
false belief reasoning. J. de Villiers (2005) has argued
that using verbs of communication avoids this con-
found with mental state verbs. Furthermore, the task
used in the present study never requires the child to
represent anything about the content of anyone else’s
mind; the task is simply answering a question about
what was actually said. Furthermore, if the comple-
ment task required false belief reasoning as well as
syntactic skill, then one would expect that language-
delayed children would pass nonverbal ToM tasks
before being able to answer questions involving
complements However, both the present study and
that of P. de Villiers and Pyers (2001) found that the
nonverbal ToM tasks were difficult for the children,
deaf or hearing, unless they had mastered comple-
ments (see also Woolfe et al., 2002).
The only other linguistic measure that was an
independent predictor of ToM in the current study
was vocabulary skills, a fact reported before for
verbal false belief tasks (Happe
´, 1995; Peterson et al.,
2005). This can be construed as evidence for the
theory that children learn about minds through
conversation, even when the topic is not mind talk.
So why does general grammatical ability fail to
predict as well? Some studies suggest that vocabu-
lary development is more dependent on conversa-
tional experience than is general grammatical
development (Arriaga, Fenson, Cronan, & Pethick,
1998; Hoff-Ginsberg, 1998). It is very likely that the
vocabulary measure is a proxy for how much rich
conversation the children have been exposed to, and
hence general vocabulary will always predict ToM.
However, two distinct representational alterna-
tives are theoretically possible: one, that learning the
labels for mental events might help set the concepts,
and two, that the formal structure involved in com-
munication and belief sentences would allow the
representation of false beliefs. Clearly, more precise
work needs to be carried out to discover the lexical
precursors for ToM. For instance, no training study
to date has shown that teaching mental state terms in
isolation from syntax could enhance reasoning, but it
might be an important test. We can achieve some
separation in the present study as the complement
comprehension test included verbs of communication
with the necessary structures. Both the complement
comprehension test and the general vocabulary
measures are predictors of false belief reasoning for
the more verbal tasks; therefore, it is possible that
some kind of lexical semantics and the particular
complement syntax contribute independently as
representational bootstraps for a mature ToM.
Similar results have been found with children
with autism. Tager-Flusberg and Joseph (2005) showed
that comprehension of communication verbs with
complements was more predictive than general lan-
guage measures of false belief reasoning in autistic
subjects in a longitudinal study. They suggested that
for some high-functioning children with autism,
cracking the linguistic code for mental events may
supply them with access to a kind of explicit reasoning
about false belief situations. Deaf children with lan-
guage delay might be considered the opposite of
children with autism in this respect. They are richly
endowed with interest in human social behavior, but
they can only get so far in developing a mature ToM
without access to representational structures for
reasoning about other’s cognitive states.
It is informative to turn these arguments around
and ask, what would be the optimal kind of input for
a child to learn about others’ false beliefs? First, the
evidence suggests that time alone with exposure to
ordinary life events and behavioral observation,
even acute behavioral observation, is insufficient, but
it may well be necessary. As Lohmann and Toma-
sello (2003) state, it appears to be ‘‘difficult for
children to construct an understanding of the rep-
resentational nature of mental states purely from
visual scenes alone’’ (p. 1139). They conclude that
both perspective-shifting discourse and the avail-
ability of sentential complement syntax as a repre-
sentational format make independent contributions
to the development of mental reasoning. Second, the
data suggest that language modality does not matter;
signing children do as well as English speakers, as
long as their input is early and complete. Third,
having enough access to the language to learn a
sizeable vocabulary is likely to help, although which
vocabulary, and why, is as yet unclear. Fourth, being
able to understand the syntax of complement-taking
verbs is not just useful but possibly essential to rea-
soning about mental states. Whether this is because
having such comprehension gives the child access to
the evidence about minds, or because it gives the
child a representational bootstrap for reasoning, is
far from settled.
The present data cannot discriminate between
these possibilities, although in conjunction with
392 Schick, de Villiers, and de Villiers

training study results, the latter possibility becomes
at least plausible. In two training studies (Hale &
Tager-Flusberg, 2003; Lohmann & Tomasello, 2003),
preschool children who had not passed false belief
tasks received training on the false complement
structures but not in conjunction with mental verbs,
and yet improved in their posttest false belief rea-
soning. However, both training studies chose chil-
dren on the cusp of both grammatical and ToM
developments. No one would claim that a 2-year old
could be so trained. The intervention was occurring
on top of whatever ordinary life and language ex-
posure had provided the child to that date, and to
that point both processes could have been at work
simultaneously.
It would seem that hearing full sentences or
mental verbs and complements, together with some
claim about their truth value, in the presence of a
behavioral discrepancy, is the optimum condition for
learning. For a variety of reasons this optimum may
not be achieved. If children are deaf, their access to a
range of speakers and range of complex structures
may be limited. Some parents may not engage in rich
talk with children that includes mental state ex-
planations, perhaps because they doubt their chil-
dren’s capacity to understand, or for reasons of time,
exigencies of other responsibilities, or cultural
norms. Parents may not have the sign language skills
to engage in elaborate mental state talk (Moeller &
Schick, 2006). Without a way to connect the sentence
structures to truth values, much of the linguistic in-
put might be uninformative.
In sum, our data suggest a significant role for
language in the development of false belief reason-
ing in deaf children. There is still room for dis-
agreement about exactly what this role is. We believe
that the problem does not lie with the verbal de-
mands of the task. Nevertheless, the specific facili-
tative or enabling role of language for false belief
reasoning is still not clear. It is very hard to tease out
whether a child could glean the evidence for ToM
from conversations without simultaneously learning
the vocabulary and complement structures. Using
complements with communication verbs represents
our best effort to separate the roles played by lan-
guage-as-evidence and language-as-representation.
In this study, complement comprehension plays a
significant role independent of the deaf children’s
general language skills.
References
Arriaga, R., Fenson, L., Cronan, T., & Pethick, S. (1998).
Scores on the MacArthur Communicative Development
Inventories of children from low and middle income
levels. Applied Psycholinguistics,19, 209 – 223.
Astington, J. (2001). The future of theory-of-mind research:
Understanding motivational states, the role of language,
and real-world consequences. Child Development,72,
685 – 687.
Astington, J., & Baird, J. A. (2005). Why language matters for
a theory of mind. New York: Oxford University Press.
Astington, J., & Jenkins, J. (1999). A longitudinal study of
the relation between language and theory-of-mind de-
velopment. Developmental Psychology,35, 1311 – 1320.
Bartsch, K., & Wellman, H. (1995). Children talk about the
mind. New York: Oxford University Press.
Bickerton, D. (1995). Language and human behavior. Seattle,
WA: University of Washington Press.
Brown, J. R., Donelan-McCall, N., & Dunn, J. (1996). Why
talk about mental states? The significance of children’s
conversations with friends, siblings, and mothers. Child
Development,67, 836 – 849.
Caldwell, E., Schick, B., & Hoffmeister, R. (2002, July).
Theory of Mind explanation of action tasks as performed by
teachers. Paper presented at the Joint Conference of the
IX International Congress for the Study of Child Lan-
guage and the Symposium on Research in Child Lan-
guage Disorders, Madison, WI.
Clark, M. D., Marschark, M., & Karchmer, M. (2001). Con-
text, cognition, and deafness. Washington, DC: Gallaudet
University Press.
Courtin, C. (2000). The impact of sign language on the
cognitive development of deaf children: The case of
theories of mind. Journal of Deaf Studies and Deaf Educa-
tion,5, 266 – 276.
Courtin, C., & Melot, A. M. (1998). Development of theo-
ries of mind in deaf children. In M. Marschark & D.
Clark (Eds.), Psychological perspectives on deafness (Vol. 2,
pp. 79 – 102). Mahwah, NJ: Lawrence Erlbaum Associ-
ates.
Cutting, A., & Dunn, J. (1999). Theory of mind, emotion
understanding, language and family background: Indi-
vidual differences and inter-relations. Child Development,
70, 853 – 865.
de Villiers, J. (1995, April). Steps in the mastery of sentence
complements. Paper presented at the biennial meeting of
the Society for Research in Child Development, Indian-
apolis, IN.
de Villiers, J. (2005). Can language acquisition give chil-
dren a point of view? In J. Astington & J. Baird (Eds.),
Why language matters for theory of mind (pp. 186 – 219).
Oxford, UK: Oxford University Press.
de Villiers, J., & de Villiers, P. (2000). Linguistic determin-
ism and the understanding of false beliefs. In P. Mitchell
& K. Riggs (Eds.), Children’s reasoning and the mind (pp.
189 – 226). Hove, UK: Psychology Press.
de Villiers, J., & de Villiers, P. (2003). Language for thought:
Coming to understand false beliefs. In D. Gentner & S.
Goldin-Meadow (Eds.), Language in mind: Advances in the
study of language and thought (pp. 335 – 384). Cambridge,
MA: MIT Press.
Language and Theory of Mind 393

de Villiers, J., & Pyers, J. (2002). Complements to cognition:
A longitudinal study of the relationship between com-
plex syntax and false-belief-understanding. Cognitive
Development,17, 1037 – 1060.
de Villiers, P. (2003). Language of the deaf – acquisition of
English. In R. Kent (Ed.), The MIT encyclopedia of com-
munication disorders (pp. 336 – 338). Cambridge, MA:
MIT Press.
de Villiers, P. (2005). The role of language in theory of
mind development: What deaf children tell us. In J.
Astington & J. Baird (Eds.), Why language matters for
theory of mind (pp. 266 – 297). Oxford, UK: Oxford
University Press.
de Villiers, P., Hosler, E., Miller, K., Whalen, M., & Wong, J.
(1997, April). Language, theory of mind, and reading other
people’s emotions: A study of oral deaf children. Poster pre-
sented at the biennial meeting of the Society for Re-
search in Child Development, Washington, DC.
de Villiers, P., & Pyers, J. (2001). Complementation and
false-belief representation. In M. Almgren, A. Barrena,
M. J. Ezeizabarrena, I. Idiazabal, & B. MacWhinney
(Eds.), Research on child language acquisition: Proceedings of
the 8th Conference of the International Association for the
Study of Child Language (pp. 984 – 1005). Somerville, MA:
Cascadilla Press.
Dunn, J. (1994). Changing minds and changing relation-
ships. In C. Lewis & P. Mitchell (Eds.), Children’s early
understanding of mind (pp. 297 – 310). Hillsdale, NJ: Er-
lbaum.
Dunn, J., Brown, J., Slomkowski, C., Tesla, C., & Young-
blade, L. M. (1991). Young children’s understanding of
other people’s feelings and beliefs: Individual differen-
ces and their antecedents. Child Development,62, 1352 –
1366.
Dunn, L., & Dunn, L. (1981). Peabody picture vocabulary test-
revised. Circle Pines, MN: American Guidance Service.
Elliott, C. (1990). Differential Ability Scales. San Antonio, TX:
Psychological Corporation.
Farrar, M. J., & Maag, L. (2002). Early language develop-
ment and the emergence of a theory of mind. First
Language,22, 107 – 213.
Figueras-Costas, B., & Harris, P. (2001). Theory of mind
development in deaf children: A nonverbal test of false-
belief understanding. Journal of Deaf Studies and Deaf
Education,6, 92 – 102.
Gale, E., de Villiers, P., de Villiers, J., & Pyers, J. (1996).
Language and theory of mind in oral deaf children. In A.
Stringfellow, D. Cahama-Amitay, E. Hughes, & A.
Zukowski (Eds.), Proceedings of the 20th annual Boston
University Conference on Language Development (Vol. 1,
pp. 213 – 224). Somerville, MA: Cascadilla Press.
Gardner, M. (1990). Expressive one-word picture vocabulary
test. Austin, TX: Pro-Ed.
Gathercole, S. E., & Baddeley, A. D. (1993). Working memory
and language. Hove: Erlbaum.
Geers, A., Moog, J., & Schick, B. (1984). Acquisition of
spoken and signed English by profoundly deaf children.
Journal of Speech and Hearing Disorders,49, 378 – 388.
Gordon, A. C., & Olson, D. R. (1998). The relations between
acquisition of a theory of mind and the capacity to
hold in mind. Journal of Experimental Child Psychology,68,
70 – 83.
Hale, C. M., & Tager-Flusberg, H. (2003). The influence of
language on theory of mind: A training study. Develop-
mental Science,6, 346 – 359.
Happe
´, F. (1995). The role of age and verbal ability in
theory of mind task performance of subjects with au-
tism. Child Development,66, 843 – 855.
Harris, P. (1996). Desires, beliefs and language. In P. Car-
ruthers & P. Smith (Eds.), Theories of theories of mind (pp.
200 – 220). Cambridge, UK: Cambridge University Press.
Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The
faculty of language: What is it, who has it, and how did
it evolve? Science,298, 1569 – 1579.
Hoff-Ginsberg, E. (1998). The relation of birth order and
socioeconomic status to children’s language experience
and language development. Applied Psycholinguistics,19,
603.
Hughes, C., Jaffee, S. R., Happe, F., Taylor, A., Caspi, A., &
Moffitt, T. E. (2005). Origins of individual differences in
theory of mind: From nature to nurture. Child Develop-
ment,76, 356 – 370.
Jackendoff, R. (1996). How language helps us think. Prag-
matics and Cognition,4, 1 – 34.
Jackson, A. L. (2001). Language facility and theory of mind
development in deaf children. Journal of Deaf Studies and
Deaf Education,6, 161 – 176.
Jacques, S., & Zelazo, P. D. (2005). Language and the de-
velopment of cognitive flexibility: Implications for the-
ory of mind. In J. Astington & J. Baird (Eds.), Why
language matters for theory of mind (pp. 144 – 162). Oxford,
UK: Oxford University Press.
Karmiloff-Smith, A. (1992). Beyond modularity: A develop-
mental perspective on cognitive science. Cambridge, MA:
MIT Press.
Lohmann, H., & Tomasello, M. (2003). The role of language
in the development of False Belief understanding: A
training study. Child Development,74, 1130 – 1144.
Lundy, J. E. B. (2002). Age and language skills of deaf
children in relation to Theory of Mind development.
Journal of Deaf Studies and Deaf Education,7, 41 – 56.
Meins, E., Fernyhough, C., Wainwright, R., Das Gupta, M.,
Fradley, E., & Tuckey, M. (2002). Maternal mind-mind-
edness and attachment security as predictors of theory
of mind understanding. Child Development,73, 1715 –
1726.
Moeller, M. P., & Schick, B. (2006). Relations between
maternal input and theory of mind understanding in
deaf children. Child Development,77, 751 – 766.
Nelson, K. (1996). Language in cognitive development: Emer-
gence of the mediated mind. New York: Cambridge Uni-
versity Press.
Newport, E. L., & Meier, R. (1985). The acquisition of
American Sign Language. In D. Slobin (Ed.), The cross-
linguistic study of language acquisition: The data (Vol. 1, pp.
881 – 938). Hillside, NJ: Lawrence Erlbaum Associates.
394 Schick, de Villiers, and de Villiers

Olson, D. R. (1988). On the origins of beliefs and other
intentional states in children. In J. W. Astington, P. L.
Harris, & D. R. Olson (Eds.), Developing theories of mind
(pp. 414 – 426). New York: Cambridge University Press.
Padden, C. (1987). Interaction of morphology and syntax in
American Sign Language. Unpublished doctoral disserta-
tion, University of California, San Diego.
Padden, C., & Ramsey, C. (2000). American Sign Language
and reading ability in deaf children. In R. I. Mayberry
(Ed.), Language acquisition by eye (pp. 165 – 189). Mah-
wah, NJ: Lawrence Erlbaum Associates.
Perner, J., Leekam, S., & Wimmer, H. (1987). 3-year-olds’
difficulty with false belief: The case for a conceptual
deficit. British Journal of Developmental Psychology,5,
125 – 137.
Perner, J., Sprung, M., Zauner, P., & Haider, H. (2003). Want
that is understood well before say that, think that, and
false belief: A test of de Villiers’ linguistic determinism
on German speaking children. Child Development,74,
179 – 188.
Peterson, C. (2002). Drawing insight from pictures: The
development of concepts of false drawing and false be-
lief in children with deafness, normal hearing, and
autism. Child Development,73, 1442 – 1459.
Peterson, C., & Siegal, M. (1995). Deafness, conversation
and Theory of Mind. Journal of Child Psychology & Psy-
chiatry,36, 459 – 474.
Peterson, C., & Siegal, M. (1997). Domain specificity and
everyday biological, physical, and psychological think-
ing in normal, autistic, and deaf children. New Directions
for Child Development,75, 55 – 70.
Peterson, C., & Siegal, M. (1999). Representing inner
worlds: Theory of mind in autistic, deaf, and normal
hearing children. Psychological Science,10, 126 – 129.
Peterson, C., & Siegal, M. (2000). Insights into a theory of
mind from deafness and autism. Mind and Language,15,
123 – 145.
Peterson, C., Wellman, H., & Liu, D. (2005). Steps in Theory
of Mind development for children with deafness or
autism. Child Development,76, 502 – 571.
Povinelli, D. J., & DeBlois, S. (1992). Young children’s
(Homo Sapiens) understanding of knowledge formation
in themselves and others. Journal of Comparative Psy-
chology,106, 228 – 238.
Pyers, J., & de Villiers, P. (2003, April). Theory of mind and
understanding emotions: what is the role of complex
language? Poster session presented at the biennial
meeting of the Society for Research in Child Develop-
ment, Tampa, FL.
Rieffe, C., & Terwogt, M. M. (2000). Deaf children’s
understanding of emotions: Desires take prece-
dence. Journal of Child Psychology and Psychiatry,41,
601 – 608.
Roeper, T., & de Villiers, J. (1994). Lexical links in the Wh-
chain. In B. Lust, G. Hermon, & J. Kornfilt (Eds.), Syn-
tactic theory and first language acquisition: Cross linguistic
perspectives. Vol. II: Binding, dependencies and learnability.
Mahwah, NJ: Erlbaum.
Ruffman, T., Slade, L., & Crowe, E. (2002). The relation
between children’s and mothers’ mental state language
and theory of mind understanding. Child Development,
73, 734 – 751.
Ruffman, T., Slade, L., Rowlandson, K., Rumsey, C., &
Garnham, A. (2003). How language relates to belief,
desire, and emotion understanding. Cognitive Develop-
ment,18, 139 – 158.
Russell, P. A., Hosie, J. A., Gray, C. D., Scott, C.,
Hunter, N., Banks, J. S., et al. (1998). The develop-
ment of theory of mind in deaf children. Journal of
Child Psychology and Psychiatry and Allied Disciplines,39,
903 – 910.
Scarborough, H. S. (1990). Index of Productive Syntax.
Applied Psycholinguistics,11, 1 – 22.
Schick, B. (1997). The American Sign Language vo-
cabulary test. Boulder, CO: University of Colorado at
Boulder.
Schick, B. (2003). The development of American Sign
Language and Manually-Coded English Systems. In M.
Marschark & P. Spencer (Eds.), Oxford handbook of deaf
studies, language, and education (pp. 219 – 231). New York:
Oxford University Press.
Schick, B., & Hoffmeister, R. (2001). ASL skills in deaf chil-
dren of deaf parents and of hearing parents. Paper presented
at the Society for Research in Child Development In-
ternational Conference, Minneapolis, MN.
Schick, B., & Moeller, M. P. (1992). What is learnable in
manually coded English sign systems? Applied Psycho-
linguistics,13, 313 – 340.
Sedey, A. (1995). Fast mapping of novel fingerspelled words by
profoundly deaf students. Doctoral dissertation, University
of Wisconsin-Madison, Madison, WI.
Segal, G. (1998). Representing representations. In P. Car-
ruthers & J. Boucher (Eds.), Language and thought
(pp. 146 – 161). Cambridge, UK: Cambridge University
Press.
Shaffer, B. (2004). Information ordering and speaker sub-
jectivity: Modality in ASL. Journal of Cognitive Linguistics,
15, 175 – 195.
Shaffer, B. (2005). Deaf children’s acquisition of modal
terms. In B. Schick, M. Marschark, & P. Spencer (Eds.),
Advances in the sign language development of deaf children
(pp. 291 – 313). New York: Oxford University Press.
Siegel, M., & Beattie, D. (1991). Where to look first
for children’s knowledge of false belief. Cognition,38,
1 – 12.
Steeds, L., Rowe, K., & Dowker, A. (1997). Deaf children’s
understanding of beliefs and desires. Journal of Deaf
Studies and Deaf Education,2, 185 – 195.
Stone, M., & Wright, B. (1979). Knox’s cube test. Wood Dale,
IL: Stoelting.
Strong, M., & Prinz, P. (1997). A study of the relation-
ship between American Sign Language and English
literacy. Journal of Deaf Studies and Deaf Education,2,
37 – 46.
Tager-Flusberg, H. (1997). Language acquisition and
theory of mind: Contributions from the study of autism.
Language and Theory of Mind 395

In L. B. Adamson & M. A. Ronski (Eds.), Research on
communication and language disorders: Contributions to
theories of language development. Baltimore, MD: Paul
Brookes Publishing.
Tager-Flusberg, H. (2000). Language and understanding
minds: Connections in autism. In S. Baron-Cohen, H.
Tager-Flusberg, & D. J. Cohen (Eds.), Understanding other
minds: Perspectives from developmental neuroscience (2nd
ed., pp. 124 – 149). Oxford, UK: Oxford University Press.
Tager-Flusberg, H., & Joseph, R. (2005). How language
facilitates the acquisition of false belief understanding in
children with autism. In J. Astington & J. Baird (Eds.),
Why language matters for theory of mind (pp. 298 – 318).
Oxford, UK: Oxford University Press.
Watson, A. C., Painter, K., & Bornstein, M. (2001). Longi-
tudinal relations between 2-year-olds’ language and
4-year-olds’ theory of mind. Journal of Cognition and
Development,2, 449 – 457.
Wellman, H. M., & Liu, D. (2004). Scaling of theory of mind
tasks. Child Development,75, 523 – 541.
Wiig, E., Secord, W., & Semel, E. (1992). Clinical evaluation of
language fundamentalsFpreschool. San Antonio, TX: Psy-
chological Corporation.
Wilbur, R. (1979). American Sign Language and sign systems.
Baltimore, MD: University Park Press.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs:
Representation and constraining function of wrong be-
liefs in young children’s understanding of deception.
Cognition,13, 103 – 128.
Woolfe, T., Want, S. C., & Siegal, M. (2002). Signposts to
development: Theory of Mind in deaf children. Child
Development,73, 768 – 778.
Zaitchik, D. (1990). When representations conflict with
reality: The preschooler’s problem with false beliefs and
‘‘false photographs’’. Cognition,35, 41 – 68.
Zelazo, P. D. (1999). Language, levels of consciousness, and
the development of intentional action. In P. D. Zelazo, J.
W. Astington, & D. R. Olson (Eds.), Developing theories of
intention: Social understanding and self-control (pp. 95 –
117). Mahwah, NJ: Erlbaum.
396 Schick, de Villiers, and de Villiers