INTRODUCTION

Autism spectrum disorder (ASD) and social pragmatic communication disorder (SPCD) are two neurodevelopmental disorders with many similarities in their symptoms. Atypical social communication, and linguistic and pragmatic development are shared obligatory diagnostic features of ASD and SPCD, and indeed defining the boundary between SPCD and ASD with IQ in the normal range (commonly referred to as high-functioning autism—HFA) often presents differential-diagnostic difficulties (Norbury, 2014). Part of the reason for this is that while the literature on ASD is extensive, in comparison relatively little is known about SPCD. Despite the fact that recognition of children who do not show repetitive behaviour but have pervasive social–communication and pragmatic difficulties is about 40 years old, SPCD has only existed as a separate diagnostic category since 2013 (Diagnostic and Statistical Manual of Mental Disorders Fifth Edition—DSM-5; American Psychiatric Association (APA), 2013), and an accurate diagnostic protocol for SPCD is still lacking. The present study explores the comprehension of grammaticalized implicit verbal meanings in order to uncover any divergences between SPCD and ASD children in this understudied domain, and it investigates whether any differences found in comprehension patterns reflect a simple quantitative difference or involve a qualitatively different course of development.

SPCD is classified as a subtype of developmental language and communication disorders in the DSM-5. Children with SPCD struggle with understanding social communication. The disorder affects both verbal and non-verbal communication. Main symptoms are (1) persistent difficulties in adequate language use for social purposes; (2) poor understanding of the speaker's intentions, emotions, desires or behaviour; (3) difficulty with the interpretation of the contextual and environmental clues, and inferencing; (4) ignoring or misunderstanding discourse rules; (5) poor receptive and expressive non-verbal skills; (6) difficulty with reception and expression of pragmatic meanings; and (7) inappropriate use of language (Adams, 2013; APA, 2013). The disorder causes large-scale limitations in almost all fields of social life, such as social participation and understanding social situations. An SPCD diagnosis is indicated if the symptoms are not better explained by ASD, intellectual disability (ID), low language skills, or another medical or neurological condition (APA, 2013).

Previously similar symptoms were predominantly labelled as pervasive developmental disorder not otherwise specified (PDD-NOS) (Diagnostic and Statistical Manual of Mental Disorders Fourth Edition, Text Revision, DSM-IV-TR; APA, 2000). Labelling children who have significant social–communication and pragmatic impairment without exhibiting repetitive and restrictive behaviour, interests or activities has generated a long-running debate laden with terminological inconsistencies. Until the advent of DSM-5, the term pragmatic language impairment (PLI) (Botting & Conti-Ramsden, 1999; Norbury, 2014) was the most commonly used term in clinical practice.

ASD is characterized by the so-called ‘autistic dyad’ as (1) persistent deficits in social communication and social interaction and participation across multiple contexts that affect verbal and non-verbal communication, including social reciprocity, interests and emotions (criterion A); and (2) restrictive, repetitive patterns of behaviour, interests or activities (criterion B). ASD can be diagnosed when autistic symptoms are present in early childhood, limit everyday functioning, and symptoms could not be better explained by ID or global developmental delay (DSM-5; APA, 2013).

According to the current diagnostic protocol, the main differential diagnostic feature that separates SPCD from ASD is the presence or absence of restricted and repetitive patterns of behaviour, interests or activities along the developmental history of the child (APA, 2013). Persons with SPCD are most difficult to distinguish from those with ASD whose IQ falls within the normal range. A common diagnostic dilemma with both disorders is that the compensatory strategies used by the persons may mask the symptoms (Thapar et al., 2015). ASD without ID is commonly referred to as HFA (for a recent critical discussion of the term, see Alvares et al., 2020). A central question pertaining to the differentiation of SPCD from ASD without ID is whether the distinction is valid based on quantitative or qualitative criteria (Norbury, 2014; Gibson et al., 2013). According to the theory of qualitative difference, we should assume (at least partly) distinct cognitive and/or linguistic causes in the background, while according to the theory of quantitative difference, SPCD should be a form of ASD in which restrictive, repetitive behaviour patterns are usually absent. In order to adequately settle issues of differentiation, it is desirable to go beyond the general recognition of social–communication and pragmatic difficulties in both groups and to study aspects of these difficulties with greater specificity.

Relevant research has commonly used the notions of ‘social communication’ and ‘pragmatics’ in rather broad and, not infrequently, undefined senses, and in many cases as synonyms. A great deal of empirical work has examined everyday communication difficulties without a detailed investigation of impairments in specific pragmatic phenomena. In fact, pragmatic competence covers a wide range of distinct phenomena, such as understanding implicit meanings (e.g., implicatures, speech acts, metaphors, irony); observing and applying discourse rules (e.g., turn-taking, topic maintenance and change, politeness); inference from context; using non-verbal contextual information (e.g., supra-segmentals, kinesthetics, gestures, facial expressions); and even metapragmatic awareness. These pragmatic competences are qualitatively different in that they recruit distinct types of cognitive resources, including structural language skills, sociocultural competence, computation of informational relevance, theory of mind (ToM), world knowledge, attention control and working memory, among others. These factors can be impaired in various ways and combinations, and appropriately isolating them in the background of pragmatic difficulties is an important yet challenging task (Adams et al., 2009).

Linguistically systematic pragmatic meanings in ASD and SPCD

In order to gain a better understanding of the differences between the pervasive pragmatic difficulties characterizing SPCD and those characterizing ASD, the goal of the present study is to investigate a specific well-defined aspect of pragmatic impairments in these conditions in childhood: the comprehension of a particular narrow range of linguistically systematic and conventionalized non-literal verbal meanings.1

While non-literal meanings share the property of not being explicit in the verbal message, they are heterogeneous on various counts. Most notably, implicit meanings are of various degrees of systematicity and linguistic conventionality. For instance, particularized conversational implicatures are ad hoc and rely on specific contextual information and world knowledge. Generalized conversational implicatures are less prone to the vagaries of context and knowledge of the world, but importantly they are also not free from their effects. For instance, scalar generalized implicatures associated with lexical items such as the determiner some and the disjunction or crucially depend on their own contextual relevance: the more relevant they are in the context in which they occur, the more likely they are to be computed by the listener (Sperber & Wilson, 2004). Generalized conversational implicatures are not only optional in a context-dependent way but they are also easily cancellable. Another characteristic property of generalized implicatures (except implicatures of manner) is non-detachability: they are independent of the actual linguistic form of the utterance that carries them, thus they are associated with any other utterance that expresses the same sentence meaning. For these reasons, although different types of generalized conversational implicatures are relatively systematic (albeit to different degrees depending on implicature types and triggers; Doran et al., 2009), they cannot be regarded as linguistically fully conventional implicit meanings.

By contrast, highly conventional implicit meanings such as entailments from the truth-conditional content of sentences, presuppositions and conventional implicatures are neither context dependent nor defeasible in this way.2 If She almost won is true then the entailed negative proposition that She did not win must be inferred in any context, without relying on world knowledge, and it cannot be cancelled. The same is true of presuppositions and conventional implicatures. For instance, if John too is at home is true then the existential presupposition that there is some other relevant person who is also at home must hold irrespective of context. Similarly, as a conventional implicature, contrast is an obligatory part of the meaning of the English it-cleft construction: one can only say that It's Bill that came if Bill is set in (explicit or implicit) contrast to some other relevant person(s). As these latter examples illustrate, some entailments, presuppositions and conventional implicatures are integrated into the linguistic system itself through being lexicalized (e.g., the negative entailment is part of the conventional meaning of the lexical item almost) or grammaticalized (e.g., the existential presupposition is part of the conventional meaning of the grammatical particle too, and contrast is part of the conventional meaning of the it-cleft).3

Although these more systematic and relatively conventionalized types of non-literal meanings are ubiquitous in language and hence play a pervasive role in verbal communication, they have not been thoroughly investigated either in ASD or in SPCD. Most of the existing research on children and adolescents in this subdomain concentrates on generalized implicatures in ASD with IQ in the normal range (referred to in this literature as HFA; or Asperger syndrome, where a distinction is made between the latter two).

Pijnacker et al. (2009) and Chevallier et al. (2010) investigated scalar implicatures in adults and adolescents with ASD (HFA or Asperger syndrome), respectively. They found no overall difference in the rate of derived implicatures between the clinical groups and healthy controls. In the HFA group both studies detected a correlation between the performance in the interpretive task and verbal intelligence. Pijnacker et al. conclude that persons with HFA are able to compensate their pragmatic deficit to some extent with their verbal intelligence. Hochstein et al. (2018) examined adolescents with ASD (mean age = 14;9) in comparison with typically developing (TD) adults, using four different implicature-comprehension tasks. Results showed that despite the presence of difficulties with certain other pragmatic inferences, similar performances were found in scalar and ignorance implicature tasks in ASD adolescents and TD adults.

Findings are less consistent in investigations that include children. Su and Su (2015) examined the comprehension of scalar implicatures in children with ASD (HFA or Asperger syndrome; ages 4;2‒15;2), who performed similarly to TD peers in almost all of their implicature conditions. While the syntactic complexity of target sentences affected performance in ASD, it played no role for TDs. In contrast, Pastor-Cerezuela et al. (2018) found a broader pragmatic deficit in ASD children (ages 6‒13) in the comprehension of different types of generalized implicatures. ASD children performed significantly worse in each type of implicature than both language and chronological age-matched TDs. By contrast, Mazzaggio et al. (2017) arrived at mixed results. While HFA children (mean age = 7;4) were found to perform more poorly than their TD peers on context-based implicatures, they did not differ from TDs in the comprehension of scalar implicatures. Performance in understanding implicatures correlated with the ToM skill and syntactic skill in the ASD group, but not in the TD group.

Schaeken et al. (2018) investigated the comprehension of scalar and non-scalar implicatures in children with ASD, with somewhat equivocal results. In a binary judgment task, the difference between the ASD and the TD group was significant in a mixed effects logistic regression analysis (interestingly, with ASD children deriving more, rather than less, implicatures than TDs), but not in a non-parametric analysis. In a ternary judgment version of the same task, the rate of those responses that indicated disagreement with statements whose scalar implicature was false was about the same in ASD as in TD. The difference between the two groups lay only in the relative rate of judgments indicating total versus partial disagreement, but even this was absent when ASDs were compared with IQ-matched TDs.

As reflected by the foregoing review, most experimental studies on linguistic pragmatic abilities in ASD concentrate on scalar implicatures, and they have yielded somewhat divergent results. Linguistically more highly conventionalized implicit meanings are understudied in ASD. One exception is Cheung et al. (2017, 2019), who examined the comprehension of presuppositions in ASD children aged 7‒14. In the first study, significant, albeit small differences were found between ASD and age-matched TD groups, beyond what could be accounted for by their lower-than-typical non-verbal intelligence and language ability. In the second study, differences were detected between the ASD group and age-matched TD controls, but no differences were found relative to a language-matched TD group.

As compared with ASD, comprehension difficulties with linguistically systematic non-literal verbal meanings in SPCD is barely chartered territory; furthermore, no experimental studies have directly compared these two clinical groups from this empirical perspective. Existing knowledge of pragmatic difficulties in SPCD, primarily concerning their limitations in social–communication and discourse-organization, predominantly comes from clinical practice, rather than from controlled experimentation.

Most of the few existing experimental treatments, employing comprehension tasks based on short stories, involve a comparison of persons with SPCD (formerly labelled PLI) with non-PLI individuals with developmental language disorder (DLD, formerly labelled specific language impairment—SLI). In their studies targeting reference assignment, Ryder et al. (2008) and Ryder and Leinonen (2013) found that the integration of contextual information and the application of world knowledge is more challenging for SLI individuals with PLI than for those without PLI. Holck et al. (2010) established that while PLI children can answer questions about textually explicit information, they have difficulty drawing particularized implicatures and other ad hoc inferences. Botting and Adams (2005) found that children with pragmatic impairments performed worse than age-matched TD children in responding to comprehension questions testing inferences, but their performance did not differ from that of SLI children. In their study, however, the pragmatically impaired group was mixed: it included both PLI and ASD children. One property shared by these studies is their reliance on highly context-dependent inferences that require a varied range of cognitive resources, including world knowledge. We are aware of no study that has been conducted to explore the performance of children with SPCD on context-independent and other linguistically highly systematic implicit meanings.

The role of structural language and ToM ability in pragmatic competence

One key question that emerges both in the literature reviewed above and also more generally is to what extent and in what ways ToM ability and structural language skills are essential for proper pragmatic functioning (Norbury, 2014; Whyte & Nelson, 2015; see also Bosco et al., 2018). Strong evidence is available from TD children on the correlation of ToM and language ability (Milligan et al., 2007). However, results are inconsistent as to whether and how these skills correlate with pragmatic competence, and if they do, whether and in what direction causal relations might hold (Kissine, 2016). Part of the reason for apparently contradictory conclusions is the fact that different studies focus on different aspects of pragmatic competence.

A growing body of research investigating the comprehension of non-literal pragmatic meanings in ASD scrutinizes the relation between pragmatic competence, on the one hand, and language ability and ToM, on the other (Andrés-Roqueta & Katsos, 2017). Baron-Cohen et al.'s (1985) and Happé’s (1993) seminal studies found a strong correlation between ToM ability and pragmatic performance in children with ASD, and the level of ToM ability was a good predictor of understanding implicit meanings. This strand of research has taken such evidence to suggest that the ToM impairment of ASDs inevitably leads to deficits in their pragmatic competence. It is important to add to this conclusion that impaired ToM ability affects some but not all pragmatic skills in ASD (Hochstein et al., 2018), and similar ToM limitations do not necessarily co-occur with the same kind of pragmatic deficits across different types of populations, because even similar ToM limitations may originate from different cognitive sources (Kissine, 2016).

The relation of pragmatic deficits and language ability in ASD is much less clearly understood. This is due in great part to the fact that social–communication and pragmatic deficits show a significantly narrower range of variation in ASD than language skills (Thapar et al., 2015), and to the fact that the reverse holds in subgroups commonly labelled as HFA.

Although research on the topic is scarce, there is evidence suggesting that the language profile in SPCD shows a developmental pathway divergent from that in ASD despite presenting similar pragmatic difficulties. Children with SPCD often have worse grammatical abilities than both TD children and ASD children without ID and/or comorbid DLD (Botting & Conti-Ramsden, 2003). Weaknesses in the grammar skills of SPCD children have not been found to be correlated with their pragmatic comprehension of metaphors (Vance & Wells, 1994) or with ad hoc, once-off implicatures (Leinonen et al., 2003). Consistent with these results, according to Botting and Adams (2005), the poor pragmatic performance of PLI children cannot be ascribed wholly to their structural language skills, since they performed worse on pragmatic tasks than language-age matched TDs.

Aims

As the foregoing review reveals, experimental research on linguistically systematic non-literal verbal meanings has mainly concentrated on the comprehension of context-dependent generalized implicatures by ASDs, and has yielded conflicting results. The receptive abilities of SPCDs in this domain remain largely unexplored. The present study aims to experimentally test the comprehension of a range of highly systematic, linguistically grammaticalized non-literal meanings (grammaticalized implicatures, presuppositions and semantic entailments) in children with SPCD and children with ASD with IQ in the normal range (for brevity, the latter will henceforth be referred to as ASD). To our knowledge, this is the first experimental treatment to directly compare the performance of SPCD and ASD groups in this highly conventionalized domain of implicit verbal meaning.

The empirical objectives of our study are twofold. First, it is explored whether the comprehension of the grammaticalized implicit meanings (GIM) is impaired in ASD and SPCD children compared with their TD peers, and whether ASD and SPCD children differ from each other in accessing these verbal meanings. Second, it is investigated whether receptive grammatical competence and first-order ToM abilities are associated with children's performance in any way, and if there is a difference in this regard between the ASD and the SPCD group. Addressing this latter question has the potential to shed light on whether any divergence between ASDs and SPCDs found in the comprehension of GIMs reflects a simple quantitative difference or involves a qualitatively different course of development.

Predictions for outcomes of the experiment were formulated based on the nature of the investigated verbal meanings and the impairment profiles commonly associated with ASD and SPCD. First, given that the meanings under scrutiny are not verbally explicit but remain implicit, on account of their general deficits in exploiting implicit verbal meanings it was expected that both ASD and SPCD children would perform at levels below that of TD peers. Second, because the investigated implicit meanings are encoded at the level of the grammatical system of language, it was hypothesized that SPCD children, whose grammatical competence is not infrequently atypical, would perform worse than ASD children without ID, whose grammatical competence is typically unimpaired.

Finally, depending on diverging assumptions concerning the role of ToM in language comprehension, two alternative predictions were formulated with regard to the potential effect of ToM on test performance. One possible hypothesis, in line with the classic Gricean approach to linguistic meaning, holds that while ToM may play a role in the retrieval of non-conventionalized non-literal meanings that crucially depend on reconstructing the interlocutor's intentions and beliefs, it is not required for an understanding of non-intersubjective, conventionalized implicit linguistic meanings. On this assumption, ToM ability was not expected to play a role in predicting performance levels on our experimental task. According to an alternative view, the role of ToM in verbal communication is more pervasive, also extending to the computation of conventional meanings (as in the relevance theoretic approach of Sperber & Wilson, 2004; Carston & Hall, 2012). The prediction of this latter view is that ToM ability should exhibit an association with test performance in our study.

Clinical relevance

Sharpening the picture of the comprehension difficulties in both SPCD children and ASD children without ID, and uncovering differences in the underlying cognitive sources are critical both for clarifying the status of SPCD as a distinct disorder, and also for the improvement of the accuracy of SPCD children's early diagnosis and timely therapeutic intervention in clinical work.

METHODS Participants

A total of 71 children aged between 4;5 and 9;5 participated in the experiment (ASD: n = 19, SPCD: n = 13, TD controls: n = 39) (Table 1). Participants were native, monolingual Hungarian-speaking children without hearing loss or impairment. The three study groups did not differ in age (F(2,68) = 3.049, p > 0.05) and gender (χ2 = 4.531, p > 0.05).

TABLE 1. Sample characteristics Groups N Mean age (SD) (months) Age range (months) Boys N (%) ASD 19 76.21 (15.42) 54‒113 16 (84%) SPCD 13 79.00 (12.08) 59–109 11 (85%) TD 39 70.46 (10.16) 53–113 24 (62%)

Children with an established ASD diagnosis were included into the ASD group. Similar to other guidelines, the Hungarian national guideline for the assessment and treatment of ASD (ESZK, 2017) states that (1) the diagnosis should be based on the DSM-5 (APA, 2013) and ICD-10 (WHO, 1992) criteria; (2) a multidisciplinary team (child psychiatrist, clinical psychologist, special educator) is needed for the assessment; and that (3) multiple methods should be used including objective measures, that is, the Autism Diagnostic Interview—Revised (ADI-R; Le Couteur et al., 2003) and the Autism Diagnostic Observation Schedule (ADOS; Lord et al., 1999). Since we did not reassess ADOS for the purpose of this study, therefore, we do not report the scores. The establishment of a SPCD diagnosis was based on the same process, and children who fulfilled criteria A (severe deficits in social communication and social interaction), but not criteria B (restricted or repetitive patterns of behaviour, interest, or activities) of DSM-5 for ASD, with symptoms being present in the early years and being functionally impairing, were diagnosed with SPCD. Exclusion criteria was an IQ score < 80 in one of the following standardized tests: the Wechsler Intelligence Scale for Children—Fourth Edition (WISC-IV; Wechsler, 2003), the Woodcock–Johnson Test of Cognitive Abilities—Third Edition (Woodcock, 2001), and the Raven Coloured Progressive Matrices Test (Raven, 1958).

TDs had no known mental, behavioural or learning difficulties, no mood disorders or other psychiatric conditions. In the case of TDs, at least average performance was expected on the compulsory pre-school speech–language assessment at age 5. The evaluation of TDs was taken by a qualified speech–language therapist.

All participants had non-verbal IQ within a normal range.

The experiment with TD children was conducted in a quiet room of the kindergarten or primary school; the clinical group was evaluated in the child psychiatry ward. The experiment was introduced after an informal warm-up conversation. We received an ethical permit for the study. All parents gave their informed consent in writing.

Measures

Receptive language ability was measured by the Hungarian adaptation of the Test for Reception of Grammar (Bishop, 1989; Hungarian version: TROG-H; Lukács et al., 2012) and the Peabody Picture Vocabulary Test (PPVT; Hungarian version: Csányi, 2002). Since first-order ToM was sufficient for our experiment (see Pijnacker et al., 2009 for the low ToM load of scalar implicatures), we used a classical false-belief task (Smarties test, Gopnik & Astington, 1988). In order to make sure that the possible low performance in the GIM comprehension task is not the result of the child's limited working memory, an auditory working memory task (non-word repetition test, Hungarian adaptation: Racsmány et al., 2005) was used for assessment.

We also aimed to ensure that TD children match the clinical group not only in their chronological age, but also in their language competence. Language-age matching was performed on the basis of TROG-H age-equivalent, which, in accordance with a caveat in the test manual, did not prove to be accurate enough on its own. Therefore, all evaluation indicators were taken into account in the analysis (age equivalent, standard point, percentile).

Procedure

In order to test the comprehension of sentences associated with GIMs, a modified truth value judgment experiment was conducted. Each item consisted of an auditorily presented, pre-recorded sentence together with a picture that appeared on a computer screen. Participants' task was to judge to what extent the sentence matches the picture: to what degree the sentence gives a true description of the depicted scenario. A ternary scale was employed (Figure 1) because it has been shown to be more adequate than the binary choice to detect sensitivity to false implicit meanings (Katsos & Bishop, 2011) and deviations from complete truth in the case of partially correct descriptions more generally (Jasbi et al., 2019).

Ternary visual judgment scale [Color figure can be viewed at wileyonlinelibrary.com] Material

Each test sentence was designed to gauge the comprehension of one of three types of GIM: (1a) a presupposition, (1b) an entailment or (1c) a grammaticalized implicature (Table 2).

TABLE 2. Types of grammaticalized implicit meanings (GIM) Example Explicit content Implicit meaning Type of implicit meaning Trigger 1a

A királylány is el ájult

the princess too away fainted

‘The princess (has) fainted too’

The princess fainted Someone else fainted Presupposition (of existence) Focus particle ‘too’ 1b

Csak a kisfiú bújt el

only the little boy hid away

‘Only the little boy (has) hid’

The little boy hid No one else hid Entailment (of exhaustivity) Focus particle ‘only’ 1c

A király ült le

the king sat down

‘It is the king who (has) sat down’

The king sat down No one else sat down Implicature (of exhaustivity) Focus word order

All sentences consisted of a singular definite subject and a (perfective) particle verb, in this order. The structural simplicity of stimulus sentences ensured that syntactic complexity does not arise as a confounding factor (cf. Su & Su, 2015). The implicit meanings shared a key property: they were associated with the subject, which uniformly and unambiguously functioned as a focus in the sentence. This focus was marked in sentence types (a) and (b) by a dedicated grammatical element: a focus particle. In sentence type (c) it was marked solely by grammaticalized word order: in Hungarian, if a phrase appears immediately before the verb, which in turn is followed by a verbal particle, then the immediately pre-verbal phrase is unambiguously interpreted as an information focus, strongly associated with an exhaustivity implicature (Káldi & Babarczy, 2018).

Importantly, the explicit content and the implicit meanings were uniform across the sentence types. The explicit content was that the individual denoted by the subject performed the action denoted by the verb. The implicit content concerned whether or not someone else (in the picture) performed the action too. According to test sentences of type (a) they did, while according to test sentences of types (b) and (c) they did not. Half of the test items included sentences of type (a), while the other half included type (b) or (c).

The picture in each sentence–picture pair depicted exactly two persons each of which either performed or did not perform the action denoted by the verb in the corresponding sentence. In each experimental item the explicit content concerned one of the two persons in the picture presented together with the sentence, while the implicit content concerned the remaining other person.

Items varied with respect to the type of the implicit meaning (a function of sentence types 1a–1c), with respect to whether the implicit meaning was true or false, and finally, with respect to the truth/falsity of the explicit content (Table 3). Since no comparisons across different conditions were planned (instead, comparisons were planned across groups), the study did not require a full factorial design. Of a full factorial design with 12 conditions, six conditions were included. These were selected with careful consideration of the information that responses in the different conditions would carry, and with an eye to balancing the expected overall set of responses. Each of the six conditions was represented by three different sentence–picture pairs, yielding a total of 18 items. The items were presented in a pseudo-randomized order.

TABLE 3. Overview of experimental conditions Condition Sentence type Example sentence Explicit content Implicit meaning Picture example I. Presupposition—false (1a) The princess fainted too True False II. Entailment—false (1b) Only the boy hid True False III. Implicature—false (1c) It is the king who sat down True False IV. Presupposition—true (1a) The cook fell too True True V. Entailment—true (1b) Only the clown tripped True True VI. Explicit—false (1a) The binman squatted too False True

The above distribution of false and true content was balanced in view of idealized expected judgments, which we established independently on the basis of the frequency distribution of the judgment types collected from 23 adult speakers. The patterns of these adult speakers’ judgments will be summarized in the next section.

RESULTS Measures

Table 4 presents the performance of participants in receptive grammar, receptive vocabulary, ToM and working memory tasks. As already discussed above, different measurement methods of TROG-H lead to slightly different results. Group differences were explored by means of one-way analyses of variance (ANOVAs) with Tukey post-hoc tests. When the assumption of homogeneity of variance was not met (Non-word repetition task, TROG-H standard points), the Welch test was used. Gender differences were not found in the study variables.

TABLE 4. Measures ASD (N = 19) SPCD (N = 13) TD (N = 39) Age (years) 6.4 6.7 5.10 Receptive grammar (TROG-H) Age equivalent (in years) 8.42 (3.16) 6.66 (2.98) 7.5 (2.32) F(2,68) = 1.708 Standard point 104.79 (15.35) 90.38 (18.02) 105.77 (10.76) 4.136* Percentile 59.58 (31.24) 33.54 (33.80) 61.82 (24.01) 5.194** First-order ToM (Smarties test) Passed 10 (62.5) 9 (69.2) 35 (92.1) χ2 = 7.643* Failed 6 (37.5) 4 (30.8) 3 (7.9) Auditory working memory (non-word repetition task) Percentile 54.74 (30.02) 58.85 (28.22) 51.03 (21.71) F(2,68) = 0.456 Receptive vocabulary (PPVT) (N) Under average 1 0 3 Average 10 9 22 Above average 6 2 9 Missing data 2 2 5