Content uploaded by Manabu Oi
Author content
All content in this area was uploaded by Manabu Oi on Mar 08, 2016
Content may be subject to copyright.
1
When Do Japanese Children
with Autism Spectrum Disorder
Comprehend Ambiguous
Language Overliterally
or Overnonliterally?
Manabu Oi
United Graduate School of Child Development, Osaka University,
Kanazawa University,
and Hamamatsu University School of Medicine
Kanazawa, Japan
Sanae Tanaka
Centre for Child Mental Development Research
Kanazawa University
Kanazawa, Japan
The present study aimed to clarify whether comprehension of ambiguous
sentences in Japanese children with high-functioning autism spectrum
disorder (HFASD) varies depending on the degree to which the use of
these sentences is conventional. We examined the relationship between
comprehension by children with HFASD and college freshmen’s assess-
ment of conventionality of usage of these sentences in contexts encourag-
ing literal or nonliteral interpretation. One hundred ninety-four freshmen
participated. Children’s interpretation preferences were correlated with
freshmen’s assessments of 9 ambiguous sentences for which previous da-
Asia Pacific Journal of Speech, Language, and Hearing
Volume 14, Number 1, pp. 1–12
Copyright © 2011 Plural Publishing, Inc.
2 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
Introduction
Literalness has been highlighted repeat-
edly by researchers to be one of the most
prominent characteristics of language in au-
tism since the first clinical descriptions of au-
tism by Kanner (1943) and Asperger (1944).
The notion has been strongly believed
among researchers even in the past decade
(Frith, 2003; Perkins, 2007). Empirical stud-
ies conducted on this issue by researchers
from English-speaking countries have con-
centrated on comprehension of metaphor
and similar constructs (Happé, 1993; MacK-
ay & Shaw, 2004; Rundblad & Annaz, 2010);
however, linguistic ambiguity applies to a
far wider range of language (Kess & Nish-
mitsu, 1989). The results of these studies
have uniformly indicated deficits in individ-
uals with autism in this respect.
Few studies have challenged the as-
sumption that people with autism demon-
strate literalness. As an exception, a study by
Ozonoff and Miller (1996) suggested a kind
of “overnonliteralness” glimpsed in adults
with high-functioning autism spectrum dis-
order (HFASD), despite their pervasive ten-
dency for literalness in understanding hu-
mor, drawing inferences, and appreciating
indirect requests. These adults were more
likely than controls to choose a nonliter-
al response to indirect requests in contexts
where a literal interpretation was encour-
aged. In addition, the autistic group was
significantly more likely to choose indirect
than direct responses, regardless of the di-
rections of contextual encouragement. In
interpreting their findings, Ozonoff and Mill-
er argued that individuals with autism have
overlearned the rule by which questions be-
ginning with ‘‘Can you . . . ’’ should be inter-
preted in a nonliteral way, as in everyday sit-
uations such syntactic forms are more likely
to be polite requests for action than inqui-
ries about ability. Ozonoff and Miller regard
individuals with autism to be less able to use
context to determine when this rule should
not be applied. They argue that such a par-
adoxic difficulty on indirect requests seen
in these adults originates from their inabil-
ity to inhibit an overlearned and proponent
response (Ozonoff, Strayer, McMahon, & Fil-
loux, 1994).
This exceptional glimpse of “overnonlit-
eralness” from individuals with autism was
ta showed a significant intergroup difference in interpretation between 45
2nd to 6th graders with HFASD and 45 typically developing (TD) children
matched for grade and gender. All the HFASD children fulfilled the crite-
ria for pervasive developmental disorder of DSM-IV-TR, and they ranged in
full-scale IQ from 79 to 129 (mean = 97.56, SD = 17.54) and in verbal-IQ
from 80 to 136 (mean = 98.87, SD = 17.81). In 6 of these 9 sentences, the
interpretation preferred more strongly by children with HFASD than TD
children was rated by the freshmen as significantly less strange than the
nonpreferred interpretation, regardless of whether it was literal or non-
literal. These results suggest that children with HFASD comprehend am-
biguous sentences in accordance with their conventionality as assessed
by freshmen. Even when TD children choose a literal interpretation, chil-
dren with HFASD select the nonliteral one judged more conventional than
its literal counterpart. This conformity of children with HFASD to conven-
tional interpretation of ambiguous language seems to account for overlit-
eralness and overnonliteralness.
Key Words: autism, literalness, ambiguous language, comprehension
WHEN DO JAPANESE CHILDREN WITH ASD COMPREHEND AMBIGUOUS LANGUAGE? 3
regarded by Ozonoff and Miller as the result
of a specific underlying impairment in using
context to understand details. They did not
believe it might stem from a generic effect
of an overly concrete communication style
(Dewey & Everard, 1974). We doubt, how-
ever, whether Ozonoff and Miller’s inter-
pretation is valid in this regard. A concrete
communication style might cause an overly
nonliteral interpretation as well as an inap-
propriately literal one. Dewey and Everard,
although intending to explain how words
and sentences are processed literally in in-
dividuals with autism, have tacitly provid-
ed a possible basis for overnonliteralness
in autism by saying that someone with au-
tism “tends to persevere in his first impres-
sion rather than discarding it to test other
meanings” (p. 349). We think this way of
language learning could lead to both over-
literalness and overnonliteralness. A person
with autism would comprehend ambiguous
language such as indirect requests, exactly
in the same way as she or he has learned
it. Nonliteral response to questions begin-
ning with, “Can you . . . ” might be learned
this way when they are asked indirectly as
is done conventionally. In putting forward
this explanation, however, we do not de-
ny the difficulty experienced by individuals
with autism in using contextual information
to respond to indirect requests when the
context encourages a literal response. Rath-
er, an overly concrete communication style
and impairment in using contextual infor-
mation in comprehending language could
be regarded as two sides of the same coin.
In other words, both might be regarded
not as discrete entities but as an emergent
phenomenon (Perkins, 1998). Literal com-
prehension of humor and the overnonliter-
al comprehension of indirect requests ob-
served by Ozonoff and Miller in adults with
autism could have emerged out of some
common underlying mechanism.
Thus, we need to take the first steps
in elucidating such a common underlying
mechanism. The first step is re-examining
findings indicating the co-occurrence of
overliteralness and overnonliteralness in in-
dividuals with autism. Oi and Tanaka (2010)
have coincidentally shown this co-occur-
rence in elementary school children with
HFASD. They asked 53 2nd to 6th graders
with HFASD and 50 without HFASD to rate
ambiguous sentences on a 5-point scale be-
tween two cartoons representing literal and
nonliteral interpretation of the sentences.
Fifty sentences varying in types of ambigu-
ity were given. These were devised by Oi
and Tanaka on the basis of a cross-linguistic
investigation by Kess and Nishimitu (1989)
on ambiguity in Japanese and English. The
types consisted of lexical ambiguity, gram-
matical ambiguity, and ambiguity in dis-
course acts including indirect requests and
irony. Metaphors, idioms, humor, and puns
also were included. The results of Oi and
Tanaka, contrary to the belief about literal-
ness in autism, showed that most (40/50)
of the ambiguous sentences were compre-
hended in a similar way between children
with and without HFASD; those with HFASD
did not demonstrate more literalness. In 4
sentences, on the other hand, children with
HFASD showed a stronger preference for
the literal meaning, and in 6 sentences, con-
versely, they showed stronger preference
for the nonliteral meaning than 50 typically
developing (TD) children. Of the 4 sentenc-
es responded to “overliterally” in children
with HFASD, 2 were grammatically ambig-
uous, one consisted of an indirect speech
act, and the other was a metaphor. Of the 6
responded to “overnonliterally,” 3 involved
ambiguous discourse acts (with the excep-
tion of indirect requests) and 3 were gram-
matically ambiguous. Out of 2 indirect re-
quests given to the children, one showed no
group difference.
These results, first, dismiss the idea that
children with HFASD tend to comprehend
any ambiguous language literally, and sec-
ond, add another piece of evidence that
“overnonliteralness” occurs in individu-
als with autism, replicating the results of
Ozonoff and Miller in terms of co-occur-
rence of “overnonliteralness” and “overliter-
alness” and broadening these results to dif-
ferent types of language ambiguity. These
findings warrant re-examination as a possi-
4 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
ble key to understanding why this co-occur-
rence happens.
In addition, we anticipate that re-ex-
amining “literalness” in autistic language
could explain why pragmatic deficits in au-
tism are so pervasive. These defects are in-
deed incredibly diverse and occur in almost
all pragmatic aspects of language (Oi, 2006;
Ozonoff & Miller, 1996; Perkins, 2002,
2007). Some have attempted to explain this
generality of the disability using the concept
of a single cognitive dysfunction such as
lack of theory of mind (Happe*, 1993, 1995),
impairment of executive function (Bishop
& Nuburry, 2005), and weak central coher-
ence (Joliife & Baron-Cohen, 1999a, 1999b;
Noens & van Berckelaer-Onnes, 2004). Per-
kins (2007) insists any pragmatic disability,
regardless of differences in underlying brain
pathology, such as specific language impair-
ment (SLI), dementia, right hemisphere dys-
function (RHD), aphasia, traumatic brain
injury (TBI), and autism, could not be attrib-
uted solely to a certain single cognitive dys-
function mentioned above. Rather, he sees
many different cognitive processes as being
engaged simultaneously and interactively in
any communicative behavior. For Perkins,
“each cognitive process which merits a sin-
gle individuating label is itself the complex
product of subsidiary interactions” (Perkins
2007, p. 106). While paying respect to his
considerations, we seek “a more pervasive
(and perhaps ultimately even more maladap-
tive) deficit in general cognitive function-
ing” (Bruner & Feldman 1993, p. 286) spe-
cific to autism.
Lawson (2003) proposed a promising
conceptualization of autism that could ex-
plain the generality of the pragmatic disabil-
ity by bringing together all cognitive defi-
cits believed to cause it. He puts forward the
depth accessibility difficulty (DAD) model as
a way of embracing distinct theories of au-
tism. The DAD model proposes that individ-
uals with autism reduce the world “to closed
systems of atomistic (essentially unconnect-
ed) actualities” (Lawson, 2003, p. 197).
Another possibility of a pervasive defi-
cit causing the generality arises (Ohigashi,
2009) when we consider revolutionary neu-
rologic insights into human consciousness
and (Edelman, 2004) together with mod-
ern philosophical concepts about the rela-
tionship between language and experience
(Agamben, 2001). Taken together, these in-
sights suggest that language in autism can be
regarded to emerge just for the individual,
not for others, in fragmental “primary con-
sciousness” without connection to integra-
tive “higher consciousness.” According to
this view, any pragmatic deficit seen in indi-
viduals with autism is due to the lack of lan-
guage for others and a result of having frag-
mentary language by nature. A person with
autism would produce or comprehend lan-
guage fragmentally in a similar manner to
which she or he has learned it. Hence, lan-
guage production and comprehension may
be based on what individuals have memo-
rized fragmentally in the context they per-
ceived it to be directly related to words from
others heard at the time (Oi, 2010).
Thus, we can hypothesize that the co-oc-
currence of overliteralness and overnonliter-
alness in individuals with autism comes from
their fragmental or atomistic language. An
ambiguous sentence would be comprehend-
ed appropriately when used in a context
(presumably conventional) that is almost
identical to that in which the person learned
it. Inappropriate comprehension would arise
when the context (presumably less conven-
tional) is considerably different from the one
where the language was learned. Such learn-
ing seems to be influenced by conventional-
ity of the usage of a sentence, as Ozonoff and
Miller (1996) indicated in referring to “Can
you . . . ” type questions.
Here we have the question of whether
comprehension of an ambiguous sentence
in individuals with autism, regardless of liter-
alness or nonliteralness, varies depending on
the degree of conventionality of the usage
of the sentence in certain context. Another
question that arises is what causes the dif-
ference between the cases where compre-
hension is overliteral and those in which it
is overnonliteral in individuals with autism.
Based on the assumption given by Ozonoff
WHEN DO JAPANESE CHILDREN WITH ASD COMPREHEND AMBIGUOUS LANGUAGE? 5
and Miller (1996), an ambiguous sentence
would be comprehended overliterally when
the context encourages nonliteral interpre-
tation, and overnonliterally when the con-
text encourages literal interpretation. How-
ever, we predict that this theory would not
hold true because the conventionality of a
sentence would play a key role in this differ-
ence in comprehension.
The purpose of the present study was
to answer the above two questions by ex-
amining the relationship between compre-
hension of ambiguous sentences in chil-
dren with HFASD and the opinion of college
freshmen regarding conventionality of us-
age of these sentences in contexts contras-
tive in the direction of encouragement.
Methods
Participants
University freshmen were randomly re-
cruited to rate the degree of strangeness
(unconventionality) of the combination of
a sentence and a cartoon picture represent-
ing the nonliteral or literal interpretation
of the sentence. These sentences and pic-
tures were the same as those used for chil-
dren with HFASD and TD children. Partici-
pants were 98 male freshmen (mean age =
19.50 years, SD = 0.74) and 96 female fresh-
men (mean age = 19.19 years, SD = 0.64).
The group was bisected into two gender-
matched subgroups. One subgroup (fresh-
men A) was asked to rate the strangeness
of 25 sentences in combination with their
literal interpretations and a further 25 sen-
tences in combination with their nonliter-
al interpretations. In rating, the order of the
type of sentence-interpretation combina-
tion was randomized. The other subgroup
(freshmen B) was asked to do the same but
with the opposite interpretation in combi-
nation with the same sentences. Hence,
when a sentence was rated in combination
with its literal interpretation by half of the
freshmen, the other half rated that sentence
in combination with its nonliteral meaning.
Thus, each of the 50 sentences was rated, in
terms of strangeness of interpretation, twice
separately in combination with its literal or
nonliteral interpretation by 97 freshmen as
shown in Table 1.
The freshmen were asked to rate the
strangeness of the combination of these am-
biguous sentences and their literal or non-
literal interpretation on a 5-point scale.
The strangest was assigned 5 and the least
strange 1.
We next examined the relationship be-
tween these freshmen’s ratings of strange-
ness and children’s interpretation of 100
combinations of a sentence and its interpre-
tation. To do this, we used the data of Oi
and Tanaka (2010) concerning 45 2nd to
6th graders with HFASD (40 boys and 5 girls,
mean grade = 4.29, SD = 1.27) and 45 TD
children extracted randomly from 666 2nd-
to-6th graders in an elementary school (ex-
actly matched for grade and gender). The
TD children all attended regular classes, re-
ceived no special educational services, and
had no sensory or motor impairments. All
Table 1. Allocation of Sentence-Literal (Nonliteral) Meaning Combination in
Rating by the Two Groups of Freshmen
Freshmen A (n = 97) Freshmen B (n = 97)
25 sentences rated sentence-literal meaning
combination
rated sentence-nonliteral meaning
combination
The other 25
Sentences
rated sentence-nonliteral
meaning combination
rated sentence-literal meaning
combination
6 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
the HFASD children were assessed by psy-
chiatrists or paediatricians as fulfilling the
criteria for any of the pervasive developmen-
tal disorders of DSM-IV-TR. All were assessed
using the third edition of the Wechsler Intel-
ligence Scale within a year before the data
were collected. They also all attended reg-
ular classes. HFASD children ranged in full-
scale IQ from 79 to 129 (M = 97.56, SD =
17.54), and in verbal-IQ from 80 to 136 (M
= 98.87, SD = 17.81). No standardized intel-
ligence scale was administered to TD chil-
dren, because it is general practice among
typical council elementary schools in Japan
not to assess pupils’ intelligence.
Procedures
First, the correlation r was calculated
between the children’s mean magnitude of
preference between literal interpretation
and nonliteral interpretation of sentenc-
es and the mean strangeness-rating of sen-
tence-interpretation combinations by fresh-
men (described above). The magnitude of
preference between the literal interpreta-
tion and nonliteral interpretation was de-
fined as the distance between the mean val-
ue of children’s rating and either the most
literal or the most nonliteral interpretation
of the sentence on the 5-point scale. The
most literal interpretation was assigned 1
and the most nonliteral 5.
Calculations of r were conducted as fol-
lows: (1) in calculating r between children’s
mean preference magnitude and freshmen’s
mean strangeness rating of the combination
of a sentence and its literal meaning, the dis-
tance between children’s mean rating value
and the most nonliteral interpretation (5 on
the scale) was adopted and (2) in calculating
r between children’s mean preference mag-
nitude and freshmen’s mean strangeness rat-
ing of the combination of a sentence and its
nonliteral meaning, the distance between
children’s mean rating value and the most
literal interpretation (1 on the scale) was ad-
opted. This calculation method is shown by
a figure in the Appendix.
Second, we examined how the strange-
ness ratings by the freshmen of the two
combinations, the sentence-literal and the
sentence-nonliteral interpretations, relate to
the difference seen in the sentence-interpre-
tation preference between the two groups
of children, for the 9 sentences in which sig-
nificant intergroup difference was shown. In
10 sentences Oi and Tanaka (2010) showed
intergroup differences in sentence-interpre-
tation preference between children with
and without HFASD, although matching be-
tween the 2 groups was not perfect in terms
of grade and gender. In the present study we
ensured perfect matching of these variables,
and as a result, one sentence in which an
intergroup difference was seen in the pre-
vious study did not attain statistical signifi-
cance. This sentence was excluded from the
examination. The strangeness ratings by the
freshmen were compared between the sen-
tence-literal interpretation combination and
the sentence-nonliteral interpretation com-
bination using the Mann-Whitney U-test.
Results
The correlation r between children’s
mean preference magnitude and freshmen’s
mean strangeness rating was −.65 (p <.001)
for children with HFASD, and −.67 (p <.001)
for TD children. The two r values did not dif-
fer significantly from each other.
In 6 sentences (#1 to 6 in Table 2) of the
9 in which the intergroup difference was re-
confirmed in terms of sentence interpreta-
tion preference by children, the interpre-
tation preferred more strongly by children
with HFASD than by TD children was rated
by the freshmen as significantly less strange
in its combination with the sentence than
the nonpreferred interpretation, regard-
less of whether the interpretation was lit-
eral or nonliteral. In 5 (#1 to 5) sentences
for which the children with HFASD showed
stronger preference for the nonliteral in-
terpretation than did the TD children, the
freshmen rated the sentence-nonliteral in-
7
Table 2. Relationships Between Children’s Preference of Literal or Nonliteral Meaning and Freshmen’s Rating of Strangeness of These
Meanings (numbers in parentheses indicate SD; tests were two-tailed).
Sentence in
Japanese
Literal
interpretation
Nonliteral
interpretation
Mean preference
toward nonliteral
Mean strangeness
rated
z p HFASD TD Literal Nonliteral
#1 Doa ni te wo
hasanda.
I made my hand get
caught in the door.
I caught my hand in the
door.
4.42
(1.01)
4.04
(1.21)
3.63
(1.53)
1.23
(0.64)
9.71 <.001
#2 Taku-chan ga ishi
ni ataru.
Taku directed himself
into the stone.
A stone hit Taku. 4.24
(1.23)
4.07
(0.99)
4.10
(1.17)
2.56
(1.63)
12.03 <.001
#3 Otou-san ga niwa
de moratta imo wo
yaita.
Dad baked sweet
potato which he was
given in the backyard.
Dad baked sweet
potato in the backyard.
3.91
(1.27)
3.53
(1.32)
3.27
(1.47)
2.36
(1.36)
4.23 <.001
#4 Kore shukudai ni
suru?—Ii desu.
One more bit of
homework?—Yes.
One more bit of
homework?—No.
4.02
(1.32)
3.33
(1.46)
3.86
(1.18)
2.09
(1.34)
7.81 <.001
#5 Shiroi hato to
karasu ga imasu.
There is a white dove
and a white crow.
There is a white dove
and a (black) crow.
4.16
(1.30)
3.78
(1.36)
3.11
(1.64)
1.71
(1.21)
5.74 <.001
#6 Omawari-san ga
otou-san to dorobou
wo tsukamaeta.
The policeman
arrested the thief and
my dad.
The policeman and my
dad arrested the thief.
2.28
(1.61)
3.27
(1.50)
2.94
(1.54)
3.65
(1.44)
3.53 <.001
#7 Ouchi ni chiisana
neko to inu ga imasu.
We have a small cat
and a small dog.
We have a small cat and
a (big) dog.
2.07
(1.34)
2.78
(1.51)
2.32
(1.41)
2.31
(1.29)
.12 n.s.
#8 Okasan ha imasu
ka?—Iru yo.
Is Mom home?—Yes. Is Mom home?—Mom! 2.35
(1.37)
3.33
(1.44)
2.24
(1.33)
2.49
(1.48)
1.04 n.s.
#9 Oisha-san no
tamago.
Doctor’s egg. A resident. 2.93
(1.55)
3.60
(1.40)
4.31
(1.23)
1.37
(0.64)
11.10 <.001
8 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
terpretation combinations as significantly
less strange compared with the literal mean-
ing (Mann-Whitney U-test, z scores were
applied to calculate p and shown in Table
2, all p-values <.001). In one sentence (#6)
where literal interpretation was preferred
significantly more strongly by the children
with HFASD than the TD children, the fresh-
men rated the sentence-literal interpretation
as significantly less strange (Mann-Whitney
U, z score was applied to calculate p and
shown in Table 2, p <.001).
On the other hand, there were 3 sen-
tences (#7 to 9) for which the above-men-
tioned relationship was not seen. The chil-
dren with HFASD preferred the nonliteral
interpretation more strongly than TD chil-
dren for these sentences. In 2 sentences
(#7, 8) of these 3, the freshmen did not rate
the sentence-literal interpretation combina-
tion and the sentence-nonliteral interpreta-
tion combination differently (Mann-Whitney
U-test, z scores were applied to calculate p
and shown in Table 2, both p-values N.S.).
Finally, in 1 sentence (#9), the freshmen rat-
ed the combination of the sentence and its
literal interpretation less strange than that
with its nonliteral interpretation.
Discussion
The present study addressed two ques-
tions. The first asked whether the appropri-
ateness of comprehending an ambiguous
sentence by individuals with autism varies
depending on the degree of convention-
ality of the context in which the sentence
is used. The second asked whether an am-
biguous sentence would be comprehended
overliterally when the context encourages
nonliteral interpretation and vice versa.
The answer to the first question was
“yes.” The more strangely the combination
of an ambiguous sentence and its context
was rated by freshmen, the more literally
the sentence was comprehended by chil-
dren with HFASD. We presupposed that this
would occur only in children with HFASD
or, at least, that it would occur more explic-
itly in these children than in TD children.
Contrary to our presupposition, however, it
occurred to the same degree in TD children;
the significant correlation r was almost iden-
tical between the two groups of children.
When considering the notion that the
younger the child the more literal the com-
prehension of ambiguous language such as
indirect requests (Kelly, 2001) and meta-
phor (Happé, 1993), before withdrawing
the abovementioned presupposition, we
have to test the possibility of a high depen-
dence on conventionality in comprehend-
ing these sentences that is specific to old-
er individuals with autism. We need to ask
whether the correlation between strange-
ness rating by freshmen and literalness in
sentence comprehension in children with
HFASD is replicated in adolescents or adults
with autism; specifically, whether r for this
correlation would differ from that in typical-
ly developing individuals.
In regard to the second question, the
present findings seem to favor our predic-
tion rather than the assumption of Ozonoff
and Miller (1996). They assumed that non-
literal interpretations of language such as
responses to “Can you . . . ” questions had
been overlearned by adults with autism to
be comprehended nonliterally even when
the context encourages literal interpreta-
tion. On the contrary, the present results
showed that, regardless of whether they
were literal or nonliteral, interpretations of
ambiguous language preferred by children
with HFASD more strongly than by TD chil-
dren were ones which freshmen rated less
strange. This means the co-occurrence of
“overliteralness” and “overnonliteralness”
in individuals with autism could not be ex-
plained simply as failure to use context or
inability to inhibit an overlearned and pro-
ponent response. Rather, the results suggest
seemingly high conformity of children with
HFASD with conventional language usage
in adults. This reminds us of formulaic lan-
guage often seen in second language learn-
ers (Weinert, 1995) as well as in people with
autism and patients with Alzheimer’s dis-
WHEN DO JAPANESE CHILDREN WITH ASD COMPREHEND AMBIGUOUS LANGUAGE? 9
ease (Wray & Perkins, 2000), although this
was discussed not from the viewpoint of re-
ceptive language but from that of produc-
tive language. Formulaic language in autism
is considered by Prizant (1983) as the prod-
uct of “an inability to segment others’ utter-
ances and realise their internal structure”
(p. 303). This might also apply to recep-
tive language, as production and reception
are closely related to each other. The co-oc-
currence of overliteralness and overnonlit-
eralness in children with HFASD has to be
investigated as a part of language formulai-
city which can be regarded as a “‘Hobson’s
choice’ solution to processing constraints”
(Wray & Perkins, 2000, p. 23).
We cannot shift away immediately, how-
ever, from the assumption of overnonliteral-
ness in autism by Ozonoff and Miller (1996).
This is because we found a relatively small
number of sentences where a stronger sen-
tence-interpretation preference in children
with HFASD than in TD children coincided
with freshmen’s lower rating of strangeness
of the interpretation. Although these sen-
tences accounted for 6 of the 9 sentences
with an intergroup difference, this propor-
tion could not attain statistical significance
with binomial distribution.
Contradictory to our prediction, some
of the present results supported the belief
that autistic language is literal by nature. Lit-
eralness in children with HFASD was seen
in 2 sentences when literal and nonliteral
interpretations were not rated differently
in terms of strangeness. This suggests that
literalness in autism is a potential tenden-
cy that could be activated when interpreta-
tions of an ambiguous sentence are equally
conventional.
In addition, children with HFASD
showed significantly stronger literalness
than TD children in one metaphor for which
the literal interpretation was rated by fresh-
men as significantly stranger than the non-
literal interpretation. This supports the idea
that literalness in autism is the result of fail-
ure to use context, as Ozonoff and Miller
(1996) assumed. This, however, must not
be overemphasized because this was the
case only in 1 of 5 metaphors investigated
by Oi and Tanaka (2010), with the rest not
comprehended differently between the two
groups of children.
For further investigation that moves us
toward an underlying mechanism of the co-
occurrence of overliteralness and overnon-
literalness in autism, we first need to clarify
why the results of the present study varied
in terms of the relationship between chil-
dren’s preference and adults’ rating of the
sentence-meaning combination. A larger
number of sentences is required for accurate
multivariate analysis in these circumstances.
Finally, taking presumable cross-cultur-
al/linguistic differences between Japanese
and English (Hinds, 1987) into account may
enrich further investigation on literalness in
autism seen in comprehending ambiguous
language. Hinds (1987, p. 143) suggested
that in “a typology that is based on speak-
er and/or writer responsibility as opposed to
listener and/or reader responsibility . . . in
English, the person primarily responsible for
effective communication is the speaker . . .
in Japanese, the person primarily responsi-
ble for effective communication is the listen-
er.” Theoretically, a language imposing re-
sponsibility on the listener is considered to
allow a speaker to be more ambiguous than
a speaker-responsible language, and accord-
ingly members of a community with listener-
responsible language are exposed to more
ambiguous language than those whose lan-
guage is speaker-responsible. How autism
interacts with the language type in this re-
gard is worth investigating. Japanese individ-
uals with autism might be more knowledge-
able about nonliteral and literal meanings
of ambiguous language than their English-
speaking counterparts. The study by Oi
and Tanaka (2010) was not the only one to
show that sentences where no difference
between children with and without HFASD
was seen in terms of comprehending ambig-
uous language far outnumbered sentences
with intergroup difference. This finding has
been replicated in other studies on compre-
hension of indirect speech acts in Japanese
(Taguchi, Oi, & Takahashi, 2010; Yata & Oi,
10 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
2009). The difference between findings on
literalness in autism from studies conduct-
ed in English-speaking countries and from
those in Japan might be based on this pre-
sumable cross-cultural/linguistic difference
between the two languages. The former
studies indicate a uniform deficit in individ-
uals with autism, whereas the latter ones de-
pict a different and more complicated pic-
ture in this respect.
Conclusion
The results of the present study, as ex-
pected, suggest that children with HFASD
comprehend ambiguous language in accor-
dance with its conventionality as assessed
by college freshmen, regardless of whether
this involves literal or nonliteral comprehen-
sion. This could be the basis of the co-oc-
currence of overliteralness and overnonlit-
eralness in these children. Even when TD
children choose a literal interpretation of a
sentence, children with HFASD select the
nonliteral one assessed less strange by col-
lege freshmen than the literal one. This con-
formity of children with HFASD to conven-
tional interpretation of ambiguous language
seems to account for the co-occurrence of
overliteralness and overnonliteralness. Re-
peating the present study using a larger
number of sentences may help in the grad-
ual examination of our hypothesis that this
co-occurrence stems from the fragmental or
atomistic nature of language in autism.
The present results have a clinical impli-
cation for assessing comprehension of am-
biguous language such as metaphor, irony,
indirect speech acts, and the like in individ-
uals with autism. In taking into account to
what extent adults with autism regard inter-
pretations of ambiguous sentences as con-
ventional, we can avoid mistaking their
seemingly high conformity to conventional
language use for a lack of knowledge of non-
literal meaning or a failure to use context.
Acknowledgments: This research was
supported by a grant (Grant-in-Aid for
Scientific Research No.14310059) from the
Japan Society for the Promotion of Science.
The authors thank Ms. Harue Ohoka for
her assistance with data collection.
Address Correspondence to: Manabu
Oi, United Graduate School of Child
Development, B-b43, 13-1, Takaramachi,
Kanazawa 920-8640, Japan, Tel: +81-76-
264-5512; Fax: +81-76-264-5510; E-mail:
oimanabu@ed.kanazawa-u.ac.jp
References
Agamben, G. (2001). Infanzia e storia. Distru-
zione dell’esperienza e origine della storia.
Torino, Italy: Einaudi.
Asperger, H. (1944). Die “Autistischen Psychopa-
then” in Kindersalter. Archiv für Psyciatrie
und Nervenkrankheiten, 117, 76–136.
Bishop, D. V. M., & Norbury, C. F. (2005). Ex-
ecutive functions in children with commu-
nication impairments, in relation to autistic
symptomatology I: Generativity. Autism,
9(1), 7–27.
Bruner, J. S., & Feldman, C. (1993). Theories of
mind and the problem of autism. In S. Bar-
on-Cohen, H. Tager-Flusberg, & D. J. Cohen
(Eds.), Understanding other minds: Perspec-
tives from autism (pp. 267–291). Oxford,
UK: Oxford University Press.
Dewey, P., & Everard, P. (1974). The near-nor-
mal autistic adolescent. Journal of Autism
and Child Schizophrenia, 4, 348–356.
Edelman, G. (2004). Wider than the sky: A revo-
lutionary view of consciousness. London,
UK: Penguin Books.
Frith, U. (2003). Autism: Explaining the enigma
(2nd ed.). Oxford, UK: Blackwell.
Happé, F. G. E. (1993). Communicative compe-
tence and theory of mind in autism: A test of
relevance theory. Cognition, 48, 101–119.
Happé, F. G. E. (1995). Understanding minds and
metaphors: Insights from the study of figura-
tive language in autism. Metaphor and Sym-
bolic Activity, 10, 275–295.
Hinds, J. (1987). Reader versus writer respon-
sibility: A new typology. U. Connor & R. B.
Kaplan (Eds.), Writing across languages:
Analysis of L2 text (pp. 141–152). Reading,
MA: Addison-Wesley.
WHEN DO JAPANESE CHILDREN WITH ASD COMPREHEND AMBIGUOUS LANGUAGE? 11
Joliffe, T., & Baron-Cohen, S. (1999a). The
strange stories test: A replication with high-
functioning adults with autism or Asperger
syndrome. Journal of Autism and Develop-
mental Disorders, 29(5), 395–406.
Joliffe, T., & Baron-Cohen, S. (1999b). A test of
central coherence theory: Linguistic process-
ing in high-functioning adults with autism or
Asperger syndrome: Is local coherence im-
paired? Cognition, 71, 149–185.
Kanner, L. (1943). Autistic disturbances of affec-
tive contact. Nervous Child, 2, 217–250.
Kelly, S. D. (2001). Broadening the units of analy-
sis in communication: Speech and nonverbal
behaviors in pragmatic comprehension. Jour-
nal of Child Language, 28, 25–349.
Kess, J. F., & Nishmitsu, Y. (1989). Linguistic
ambiguity in natural language: English and
Japanese. Tokyo, Japan: Kuroshio Shuppann.
Lawson, J. (2003). Depth accessibility difficul-
ties: An alternative conceptualisation of au-
tism spectrum conditions. Journal for the
Theory of Social Behaviour, 33, 189–202.
MacKay, G., & Shaw, A. (2004). A comparative
study of figurative language in children with
autistic spectrum disorders. Child Language
Teaching and Therapy, 20, 13–32.
Noens, I., & van Berckelaer-Onnes, I. (2004).
Making sense in a fragmentary world: Com-
munication in people with autism and learn-
ing disability. Autism, 8, 197–218.
Ohigashi, Y. (2009). Personal communication.
Oi, M. (2006). Pragmatic impairment in indi-
viduals with high-functioning pervasive de-
velopmental disorders: Characteristics, back-
grounds, and intervention. Japanese Journal
of Communication Disorders, 23, 87–104.
Oi, M. (2010). Language and mind in autism:
A perspective to relate the triad of autism to
social brain hypothesis. Paper presented at
International Conference on Social Brain: Au-
tism and Neuroethics, Kanazawa.
Oi, M., & Tanaka, S. (2010). Comprehension of
ambiguous language in children with high-
functioning autism spectrum disorder. Japa-
nese Journal of Communication Disorders,
27, 10–18.
Ozonoff, S., & Miller, J. N. (1996). An explora-
tion of right-hemisphere contributions to the
pragmatic impairments of autism. Brain and
Language, 52, 411–434.
Ozonoff, S., Strayer, D. L., McMahon, W. M., &
Filloux, F. (1994). Executive function abilities
in autism and Tourette syndrome: An informa-
tion processing approach. Journal of Child
Psychology and Psychiatry, 35, 1015–1032.
Perkins, M. (1998). Is pragmatics epiphenome-
nal? Evidence from communication disorders.
Journal of Pragmatics, 29, 291–311.
Perkins, M. (2002). An emergentist approaches
to clinical pragmatics. In F. Windsor, M. L.
Kelly, & N. Hewlett (Eds.), Investigations in
clinical phonetics and linguistics. (pp. 1–14)
New York, NY: Laurence Erlbaum.
Perkins, M. (2007). Pragmatic impairment.
Cambridge, UK: Cambridge University Press.
Prizant, B. M. (1983). Language acquisition and
communicative behavior in autism: Toward
an understanding of the “whole” of it. Jour-
nal of Speech and Hearing Disorders, 48,
286–296.
Rundblad, G., & Annaz, D. (2010). The atypical
development of metaphor and metonymy
comprehension in children with autism. Au-
tism, 14, 29–46.
Taguchi, A., & Oi, M. (2010). Comprehension of
indirect speech in children with high-func-
tioning pervasive developmental disorders
under different task conditions. Japanese
Journal of Communication Disorders, 27,
168–177.
Weinert, R. (1995). The role of formulaic lan-
guage in second language acquisition: A re-
view. Applied Linguistics, 16, 180–205.
Wray, A., & Perkins, M. (2000) .The functions
of formulaic language: An integrated model.
Language and Communication, 20, 1–28.
Yata, A., & Oi, M. (2009). Comprehension of
indirect speech in children with high-func-
tioning pervasive developmental disorders.
Japanese Journal of Learning Disabilities,
18, 128–213.
12 ASIA PACIFIC JOURNAL OF SPEECH, LANGUAGE, AND HEARING, VOL. 14, NO. 1
Appendix
