• No results found

The linguistic and cognitive abilities in Dutch-speaking children with Autism and a language impairment as compared to children with Autism Spectrum Disorder

N/A
N/A
Protected

Academic year: 2021

Share "The linguistic and cognitive abilities in Dutch-speaking children with Autism and a language impairment as compared to children with Autism Spectrum Disorder"

Copied!
56
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The linguistic and cognitive abilities in Dutch-speaking

children with Autism and a language impairment as compared

to children with Autism Spectrum Disorder

Anne Nagtzaam 10204946 MA General Linguistics Jeannette Schaeffer 04/07/2018 University of Amsterdam

(2)

Abstract

The current study investigated the (morpho)syntactic, pragmatic and extra-linguistic cognitive abilities of Dutch-speaking children with Autism and a language impairment (ALI) in comparison to children with Autism Spectrum Disorder (ASD) and typically developing children (TD). By means of a large test battery, the linguistic and extra-linguistic abilities of 8 children with ALI were compared to 8 children with ASD and 8 TD children who were all matched on age and gender. As children with ALI have both Autism and a language impairment, I hypothesized that they would show (morpho)syntactic deficits as well as pragmatic and cognitive deficits. The data suggest that children with ALI do not have problems with linguistic tasks that targeted morphemes, such as Subject-Verb Agreement, Mass-Count, and Article Choice. They do show significant differences with their TD peers on tasks requiring more complex syntax and longer word-order. The correlations found with extra-linguistic knowledge (memory and executive functions) suggest that the underlying deficit of children with ALI can be found in their memory and EF skills instead of their pragmatic and/or (morpho)syntactic skills. Future research should be conducted to unveil if there is a correlation between phonological memory, (non)-verbal working memory and executive functioning and syntactic performance. By adding up the group size and including more executive functions tests, there may be stronger evidence found that memory and executive functions are the underlying cause of the language deficits.

(3)

Table of contents

1. Introduction 5

2. Theoretical background 7

2.1. (Morpho)syntactic driven phenomena 7

2.1.1. Mass-count distinction 7

2.1.2. Subject-verb agreement 10

2.1.3. Object relative clause production 11

2.1.4. Sentence repetition 13

2.2. Pragmatics driven phenomena 14

2.2.1. Article production 14

2.2.2. Direct object scrambling 16

3. Current study 18 3.1. Research questions 18 3.2. Hypotheses 19 3.3. Predictions 19 4. Method 23 4.1. Participants 23

4.2. Materials and procedure 24

4.2.1. (Morpho)syntactic tests 24

4.2.1.1. Mass-count distinction 24

4.2.1.2. Subject-verb agreement 25

4.2.1.3. Relative clause production 26

4.2.1.4. Sentence repetition 27

4.2.2. Pragmatic tests 27

4.2.2.1. Article choice production 27

4.2.2.2. Direct object scrambling 28

4.2.3. Extra-linguistic cognitive tests 29

4.2.3.1. Non-word repetition 29

4.2.3.2. Digit span forward and backward 29

4.2.3.3. Odd one out 29

4.2.3.4. Luria’s handgame 29

5. Results 30

(4)

5.1.1. Mass-count results 30

5.1.2. Subject-verb agreement results 30

5.1.3. Object relative clause production results 31

5.1.4. Sentence repetition results 31

5.1.5. Summary 32

5.2. Results pragmatics 33

5.2.1. Article choice production results 33

5.2.2. Direct object scrambling results 34

5.2.3. Summary 36

5.3. Results extra-linguistic cognition 36

5.3.1. Non-word repetition results 36

5.3.2. Digit span forward results 36

5.3.3. Digit span backward results 37

5.3.4. Odd-one-out results 37

5.3.5. Luria’s handgame results 38

5.3.6. Summary 38

5.4. Correlations 39

6. Discussion 40

(5)

1. Introduction

Autism Spectrum Disorder (ASD) is a developmental disorder which can be divided into a cluster of disorders for which the characteristics vary significantly in different individuals. The primary impairments of ASD can be found in both language and communication . The characteristics range from extreme aloofness, poor language use and disturbed social interaction to above-average intelligence, verbal fluency and adequate professional and social skills (Mesibov et al., 2001; Wing, 2000). Yet, the main focus of research into autism traditionally focuses on the communicative deficits that can be found in this population. Following the American Psychiatric Association (1994), ASD is diagnosed on the basis of impairments or delayed development in three areas, namely social interaction, communication and repetitive behavior regarding activities and interests. These impairments are generally detected by means of language observations. Delay or absence of spoken language is often the first presenting symptom of ASD (Kjelgaard & Tager-Flusberg, 2001).

Despite the fact that the diagnosis of autism relies heavily on language, for many years researchers only looked at the universal aspects that define the nature of the language and communication deficits in autism (Kjelgaard & Tager-Flusberg, 2001; Tager-Flusberg, 1996). By exploring the pragmatic deficits that may appear in conversations and other discourse contexts, a body of research has led to the consensus that the primary impairment in children with autism can be found in a serious limitation of the communicative abilities. These communicative impairments have been related to difficulties with referentiality, an immature Theory of Mind and to features regarding social functioning (Lord & Paul, 1997; Schaeffer, 2018; Tager-Flusberg, 1996, 1999). However, the language impairment in children with ASD seems to be much more variable and cannot be easily reduced to one main impairment. Where some children show a language profile where their vocabulary, grammatical knowledge and phonological skills are within the normal range in comparison to typically developing children, another significant proportion of the population remains non-verbal (Lord & Paul, 1997). For this study, we focus on the high functioning group within the Autism Spectrum Disorder. Children with High-Functioning Autism (HFA) are characterized by the presence of a qualitative impairment in their social interaction and nonverbal communication, but at the same time are able to speak fluently and have normal intelligence (Noterdaeme et al., 2010; Schaeffer, 2018). The disorders included in this group are Asperger’s syndrome, Classic Autism and PDD-NOS.

(6)

More recent research into the linguistic profile of children with ASD shows that their language problems are not only related to insufficient pragmatics. Several studies suggest that children with ASD also reveal grammatical weaknesses (Eigsti & Bennetto, 2009; Perovic et al., 2013; Tager-Flusberg, 2000). Kjelgaard and Tager-Flusberg (2001) present a subgroup of ASD which shows grammatical difficulties with verb morphology and non-word repetition comparable to the error pattern of children with SLI. Conversely, children diagnosed with SLI who have receptive language impairments seem to show autistic features (Cantwell et al., 1989, Howlin et al., 2000). The overlap in symptoms between ASD and SLI and the fact that ASD and SLI co-occur in more than half of the cases led researchers to argue whether ASD and SLI should be distinguished as different impairments or if they could be considered as “instantiations of the same continuum” that share the same underlying etiology. (Bishop, 2010; Schaeffer, 2018). According to The International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) (World Health Organization (WHO), 1993) and the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) (American Psychiatric Association, 1994), SLI and ASD cannot coincide with each other. As a consequence, a SLI diagnosis cannot be made when a child is already diagnosed with ASD. However, researchers have suggested that SLI and ASD are associated at a genetic and behavioural level. Family studies by Fombonne et al. (1997) and Tomblin et al. (2003) reveal an increasing ASD risk in siblings of children with SLI and vice versa. Genetic linkage studies by Alarcon et al. (2005), Bradford et al. (2001) and Vernes et al. (2008) have found possible loci on chromosomes 7q and 13q which are associated with SLI and ASD. Bishop (2010) therefore argues that ASD and SLI “correspond to points on a continuum of impairment, rather than being all-or-none diseases” (Bishop, 2010: 619). Schaeffer (2018) refutes this assertion by providing data that indicate that children with SLI have profoundly different profiles from children with HFA when looking at grammar and nonverbal cognition. The resemblance in terms of pragmatics is, according to Schaeffer, not enough to consider HFA and SLI as “instantiations of the same continuum” (Schaeffer, 2018: 5).

Tager-Flusberg and colleagues identified a subgroup of children with Autism Spectrum Disorder who show language impairments comparable to the psychometric profile which is typical of SLI (Riches et al., 2010). This subgroup shows a poor performance on non-word repetition (Kjelgaard & Tager-Flusberg, 2001) and past tense tests (Roberts et al., 2004). They introduced the term ‘Autism plus Language Impairment (ALI)’ to describe this group. However, as pointed out by Riches et al. (2010), the similar performance on two clinical markers does not necessarily extend to all aspects of language. Loucas et al. (2008)

(7)

observed that receptive and expressive language are equally impaired in children with ALI, whereas in children with SLI receptive language skills are stronger than the expressive skills. Therefore, it is important to use sophisticated tasks to identify the subtle differences between the groups. Many studies investigating the SLI/ALI overlap have used standardized tests which lack the psycholinguistic sophistication to determine the differences. Furthermore, Riches et al. (2010) mention that in the current literature, there has been little research on the production of complex syntax in children with ASD or ALI.

The current study focuses on the (morpho)syntactic, pragmatic and extra-linguistic cognitive development of Dutch-speaking children with Autism and a language impairment in comparison to children with ASD and typically developing (TD) children. By means of a large test battery designed by Jeannette Schaeffer and Iris Duinmeijer, I attempt to get more insight in the language phenotype of children with ALI.

An additional focus of this study is the role of executive functioning and phonological memory. Previous research suggests that memory has some predictive power for the grammatical performance of children with SLI (Henry, Messer & Nash, 2012; Meir & Arnom-Lotem, 2014; Schaeffer, 2018). Regarding ASD there are inconsistent results in studies investigating extra-linguistic cognition. Some studies provide evidence for working memory deficits in individuals with ASD (Joseph et al., 1999; Schuh & Eigsti, 2012), whereas other studies report normal performance (Griffith et al., 1999; Koshino et al., 2005; Schaeffer, 2018) or mixed performance (Benetto et al., 1996; Luna et al., 2002). The potential relation between grammatical impairments and extra-linguistic cognitive weaknesses will be further investigated in the current study to see whether executive functioning and phonological memory are of influence on the linguistic performance of children with ALI and ASD.

The next section provides more detailed information and relevant theories regarding the different tasks used for the current study.

2. Theoretical background

2.1. (Morpho)syntactic driven phenomena

2.1.1. Mass-count distinction

In many languages, including English and Dutch, the distinction between mass and count nouns is syntactically expressed. In these so-called number-marking languages, count nouns

(8)

(e.g. dog, table) are the nouns that can appear in both plural and singular form and can be preceded by numerals and determiners. Mass nouns (e.g. water, sand) are the nouns that cannot be preceded by numerals or determiners and cannot be pluralized. Some nouns can behave as either mass noun or count noun and are therefore called “flexible nouns”, as illustrated in (1) for Dutch:

(1)

A. Het huis is gemaakt van steen. ‘The house is made out of stone’. B. Bij de rivier vond ik drie mooie stenen.

At the riverside found I three beautiful stones. ‘At the riverside, I found three beautiful stones.’

The noun steen (‘stone’) behaves as a mass noun when it is quantized by a measure phrase, as in (1a). The same noun behaves like a count noun when it is preceded by a numeral and followed by a plural morpheme, as in (1b) (Barner & Snedeker, 2005; van Witteloostuijn & Schaeffer, 2018) .

Besides syntax, also semantics takes part in distinguishing between mass and count. Over the years, the relationship between syntax and semantics has been extensively discussed regarding the mass-count distinction. Barner and Snedeker (2005) distinguished three different theories about the mass-count semantics and, in particular, the individuation of mass nouns.

The first theory is based on a theory by Quine (1960) and therefore named the “Quinian correspondence hypothesis”. According to Quine (1960), mass nouns refer cumulatively and count nouns refer to individuals or “atoms”. He argues that mass nouns can refer to either a single amount of the substance or to a combination of several amounts of the substance, and still refer to the same substance. However, this cumulativity criterium does not hold for count nouns. This is illustrated by Barner and Snedeker (2005) who provide the example that if a is water and if b is water, than a and b combined are water, but when a is a horse and b is a horse, then a and b taken together are not a horse (p. 44). Wisnieuwski et al. (1996) adopted this view and proposed that “language users should conceptualize the referents of count nouns as distinct, countable, individuated things and those of mass nouns as non-distinct, uncountable and unindividuated things (p. 271).” A second criterium is proposed by Cheng (1973) who argues that mass nouns are divisive, where count nouns are

(9)

not. This notion of divisity suggests that a portion of water divided in two is still referred to as water, whereas a dog divided in two will neither be referred to as two dogs nor one. These two criteria taken together seem to suggest that mass nouns are cumulative, divisive, non-individuated objects or substances, and count nouns are non-cumulative and non-divisive individuated objects.

Gillon (1996) and Chierchia (1998) put forth two counterexamples which suggests that this one-to-one mapping mentioned above does not hold for all mass nouns. Gillon (1996) notes that there are various count nouns which can be divided, such as rope, rock, ash,

cord and tile. Next to that, he also named a couple of mass nouns in English that have

“individuals in their denotations”, such as furniture, silverware, jewelry, clothing, traffic,

infantry, and footwear (p. 45) (see also Chierchia, 1998). This rules out the suggestion that

divisity is exclusive to mass nouns and that count nouns only quantify individuals. As a consequence of his findings he proposed a new theory, namely the “linguistic non-specification hypothesis” in which is argued that mass nouns are unspecified for whether or not they denote individuals. He proposes that knowledge of the world gives us an indication whether the noun denotes individuals or not. Chierchia (1998) extends this proposal by an “inherent plurality theory”, which suggests that “all mass nouns refer to sets of atoms or individuals, and are inherently plural”(Barner & Snedeker, 2005: 45). Chierchia claims that the denotations of the concepts of mass and count are essentially the same, meaning that mass nouns also refer to individuals. It is the plurality value of the noun that provides the distinction between count and mass.

A third theory can be found in multiple studies proposing that nouns become mass and count as a consequence of their syntactic environment (Bale and Barner 2004, Barner & Snedeker, 2005; Borer, 2005; van Witteloostuijn & Schaeffer, 2018). This theory is supported by the study by Barner and Snedeker (2005) who provide experimental evidence that shows that for flexible nouns, like rope and pizza, the syntactic context of nouns provides the cue for a mass or count interpretation (van Witteloostuijn & Schaeffer, 2018). Barner and Snedeker refer to nouns that can behave as either mass and count nouns as so-called “flexible nouns”. They claim that the feature that licenses individuation provides the cue for mass or count interpretation. The feature [+individual] can be expressed syntactically, by means of articles, plural morphology and numerals, or lexically, by means of nouns such as jewelry which carry a lexical denotation . Borer (2005) takes this even further and suggests that every noun is a flexible noun and is determined as either mass or count as a consequence of the syntax they appear in.

(10)

Previous research on the mass-count distinction for Dutch shows that typically developing children become sensitive to the distinction between mass and count nouns from the age of 6 (van Witteloostuijn & Schaeffer, 2014). As for children with ASD or ALI, studies investigating the mass-count distinction are rare. The study by Schaeffer (2018), of which the current study is part, was the first to examine the acquisition of the mass-count distinction in children with autism. Schaeffer investigated 27 children with HFA, 27 children with SLI and 27 TD children (all groups aged 5-14) using a Quantity Judgment Task based on Barner and Snedeker (2005). The results suggest that children with HFA perform with equal accuracy as the TD children on mass-count distinction, whereas the children with SLI score significantly more poorly than both the TD as the ASD group. As studies investigating the linguistic performance of children with ALI are rare, this study is the first to investigate this phenomenon in children with ASD and a language impairment.

2.1.2. Subject-verb agreement

In many languages, including English and Dutch, subject-verb agreement is a key component of grammatical encoding. When formulating a sentence, the grammatical subject has to agree in number with the main verb. This implicates that singular subjects require singular verbs, and plural subjects require plural verbs (Veenstra et al., 2014). In Dutch, subject-verb agreement is expressed by attaching a suffix to the verb as illustrated in the schema in (2): (2) Subject-verb agreement in Dutch for the verb dansen (‘to dance’)

Person Singular Plural

1 (ik) dans (wij) dans-en

2 (jij) dans-t (jullie) dans-en

3 (hij/zij/het) dans-t (zij) dans-en

The suffix expresses the person (in the singular) and number of the verb. In Dutch, both 3SG as 3PL are marked with an overt agreement suffix, which makes Dutch different from English, Spanish and French. By adding a morpheme after the verbal stem, a phonological cue is given indicating the number and person (Verhagen & Blom, 2014).

(11)

Studies into the production of verb agreement reveal that children acquire subject-verb agreement at a relatively young age in spontaneous speech. Dutch-speaking children start to use verbal agreement with a wider variety of verbs from the age of 2;6 (Blom & Wijnen, 2013; Verhagen & Blom, 2014). This early acquisition is surprising in light of the fact that children seem to comprehend subject-verb agreement at a later stage. Therefore, Verhagen and Blom (2014) suggest that there seems to be a an asymmetry between comprehension and production in number agreement acquisition. As for Dutch-speaking children with SLI, subject-verb agreement is an often observed problem as they have difficulties producing correctly inflected verbs (de Jong, 1999; Rispens & Been, 2007). The study by Schaeffer (2018), from which the current study is part, was the first to examine the acquisition subject-verb agreement in children with (high-functioning) autism. By means of an Elicited Production Task designed by Duinmeijer (2012), Schaeffer tested 2 groups of 27 children who were diagnosed with HFA or SLI and 27 TD children. The results suggest that children with HFA perform with equal accuracy as the TD children on verbal agreement between the subject and the verb, whereas the children with SLI score significantly more poorly than both the TD as the ASD group. As studies investigating the linguistic performance of children with ALI are rare, this study is the first to investigate this phenomenon in children with ASD and a language impairment.

2.1.3. Object relative clause production

Relative clauses can be used to modify the head noun of the sentence (Chierchia & McConnell-Ginet, 1990). According to Fox and Thompson (1990), relative clauses provide a characterization of the discourse function which can be observed when investigating the choice of utterance in production. As Dutch has a Subject-Object-Verb (SOV) word-order, the embedded finite verb appears in final position of the relative clause. As a consequence, the difference between subject and object relative clauses is hard to distinguish as it makes them ‘reversible’. This is illustrated in (3) from an example by Schaeffer and Siekman (2016). The subscripts indicate the constraints with the same thematic role.

(3)

Object relative clause:

A. … Jani diei Marie ti belt

John who Mary calls

(12)

Subject relative clause:

B. … Jani diei ti Marie belt

John who Mary calls

‘… John who Mary is calling’

In example (3a), Jan fulfills the role of the object of the relative clause, where in (3b) Jan is the subject of the relative clause. Although the two relative clauses differ grammatically, there is no difference in sound which results in ambiguity. As previous research reveals that children seem acquire subject relative clauses earlier and with higher accuracy than object relative clauses (Rademaker, 2014), this study only focuses on object relative clauses.

Object relative clauses are relative clauses for which the object of the relative clause is the head of the sentence, as illustrated in (3b) from an example by Schaeffer and Siekman (2016):

(4)

a. The girl pushed the boy.

b. This is the boy that the girl pushed.

In (4b), the head of the relative clause [that the girl pushed] is the original object of the clause [the girl pushed __ ], namely the boy. Following Chomsky (1981), object relative clauses thus seem to derive from movement from the object position of the original clause.

Previous research into the production of object relative clauses in Dutch-speaking adults revealed that adults often produce a passive to resolve the ambiguity and convey the meaning of the object relative clause (Rademaker, 2014). Another strategy that can help resolve this ambiguity and comprehend the object relative clause is the semantic feature [+/-animacy]. By observing the semantic feature of animacy between the actors of the relative clause, the interpretation of the relative clause might become straight forward, as illustrated in (5) from an example by Schaeffer and Siekman (2016):

(5)

… het boeki dati het meisje ti leest

the book that the girl reads ‘… the book that the girl is reading’

(13)

This relative clause is not reversible as the actor books ([-animate]) is not able to read, but the actor girls ([+animate]) is. Therefore, this relative clause can only be comprehended as an object relative clause, rather than a subject relative clause.

There are barely any studies investigating the production of object relative clauses younger than the age of 6 for Dutch. However, various studies report that Dutch-speaking typically developing children are able to produce and comprehend object relative clauses at the age of 6, although not with the same accuracy as adults do. These studies also show that TD children often use passives when producing an object relative clause (Schouwenaars et al., 2014; Rademaker, 2014; Duinmeijer, 2017). As for children with HFA, Schaeffer and Siekman (2016) found the same production pattern as in TD children. By means of an Picture Selection Task designed by Duinmeijer (2017), they investigated the production of 25 children with HFA in comparison to 25 age and gender matched TD children. Schaeffer and Siekman also observed an increasing use of passives as a strategy to resolve the ambiguity of Dutch object relative clauses.

2.1.4. Sentence repetition

Sentence repetition is a commonly used tool in diagnosing Specific Language Impairment (SLI) and is part of several clinical assessments, such as the Clinical Evaluation of Language Fundamentals (CELF). Sentence repetition became of interest as a consequence of the studies by Slobin and Welsh (1968) and Clay (1971) who argued that when an individual has to repeat an utterance which is longer than their word span, the maximum number of random words they can repeat, they cannot depend on their short-term phonological memory only, but are obliged to use syntactic knowledge to process the stimulus in their working memory. More recent studies suggest that during the repetition of sentences individuals make a reconstruction of the stimulus using lexical, conceptual and syntactic representations which they retrieve from their long-term memory (Lombardi & Potter, 1992; Potter & Lombardi, 1998).

For the repetition of more complex sentences, several cognitive processes are involved. First, phonological short-term memory is activated to temporarily store the received information, after which the central executive system is activated to manage the exchange of information between the phonological memory and the (verbal) working memory. Finally, phonological output processes are involved for the actual repetition of the sentence (Jefferies

(14)

et al., 2004; Riches et al., 2010; Rummer, 2004; Willis and Gathercole, 2001). While children are able to repeat short sentences depending on their phonological short term memory without comprehending the actual content, longer and more complex sentences demand comprehension of the content as the syntactic and semantic representations in the long term memory are more likely to diverge from the stimulus if a sentence is not comprehended successfully (Riches et al., 2010; Vinther, 2002). This was supported by the study by McDade et al. (1982) who showed that comprehension could function as a predictor for performance, when inserting a short three-second pause after the stimulus. Polisenská et al. (2014) confirmed the importance of comprehension as they compared the performance on sentence repetition with semantic plausible and implausible sentences. The results revealed that plausibility has a significant effect on the performance which suggest a reliance on semantics.

Previous research into the performance of children with ASD with normal language and ASD with impaired language (ALI) on a sentence repetition task shows that children with ASD score with the same accuracy as TD children, whereas children with ALI perform more poorly (Riches et al., 2010; Harper-Hill et al., 2012; Taylor et al., 2013).

2.2. Pragmatics driven phenomena

2.2.1. Article choice production

The choice between a definite and an indefinite article depends on the speaker’s knowledge of speaker/hearer assumptions (Heim, 1982; Schaeffer & Matthewson, 2005; Stalnaker, 1974). The use of a definite article requires a common ground in which speaker and hearer share the same beliefs at time of the utterance. When one or both interlocutors are not familiar with the referent of the sentence, the indefinite article has to be used. Schaeffer (2018) proposed a schema for the three possible assumption states in the Dutch adult article system:

(6) The Dutch adult article system

A. Assumed by speaker and hearer Part of common ground de (‘the’)

B. Assumed by speaker only Not part of common ground een (‘a’)

(15)

This schema can be illustrated by the following examples as presented by Schaeffer (2018): (7)

Dit is een verhaal over een (bepaalde) jongen.

De jongen woonde in een groot kasteel.

‘This is a story about a (certain) boy. The boy lived in a big castle’

In the first sentence, the noun jongen (‘boy’) is introduced by the speaker, while the referent is still unknown to the hearer. Therefore, the speaker is obliged to use an indefinite article, as the noun jongen is not part of the common ground. In the second sentence, the referent of

jongen is known to both speaker and hearer, as it is introduced in the first sentence.

Therefore, the speaker can use a definite article as the referent has become part of the common ground between speaker and hearer.

(8)

Ik heb zin om een boek te lezen (wat voor boek dan ook) ‘I feel like reading a book (whatever book it may be)’

In (8), the referent of the noun boek (‘book’) is unknown to both speaker and hearer. Therefore, an indefinite article has to be used as the referent is not part of the common ground.

Furthermore, according to Hawkins (1991) and Horn (2006) the speaker has to adhere to the Maxim of Quantity as proposed by Grice (1975). This Maxim states that you have to be as informative as is required/necessary (not more and not less). They propose a Definiteness Scale between a and the, from which the definite article the is the logically stronger and more informative member of the scale. Therefore, the choice of an indefinite article implies that the definite article could not be used as a result of the consideration of speaker and hearer assumptions. When adults interpret indefinite NP’s, they thus draw a scalar implicature based on Grice’s Maxim of Quantity.

Summarizing, an indefinite article may be interpreted in two different ways: either the literal logical/semantic meaning deriving from the existence of a referent, or with an inference-driven pragmatic reading that excludes the definite reading as a consequence of the calculation of a scalar implicature (Schaeffer et al., 2017).

(16)

Several studies across multiple languages with a two-article language system based on definiteness show that young typically developing children tend to overuse definite articles in indefinite contexts (van Hout, Harrigan, & de Villiers, 2010; Karmiloff-Smith, 1979; Maratsos, 1976; Schaeffer & Matthewson, 2005; Zehler & Brewer, 1982). This definite overgeneration in definite contexts is explained by different accounts varying from cognitive to pragmatic explanations. Maratsos (1976) proposes an “egocentricity” account suggesting that when children have to choose an article, they are tend to choose from an egocentric perspective. Karmiloff-Smith (1979) explains the overproduction of definite articles in indefinite contexts to the fact that children use articles deictically instead of anaphorically. She argues that when children use definite articles, they are drawn directly from representations in the memory, resulting in an overuse of definite articles as they function like demonstratives (Schaeffer & Matthewson, 2005). Schaeffer and Matthewson (2005) explain this overuse by a lack of the pragmatic Concept of Non-Shared Assumptions (CNSA). The CNSA obliges the speaker to consider the hearer’s assumptions as independently from the speaker’s assumptions. If the CNSA is absent, the speaker may attribute his/her own assumptions to the hearer which may result in a miscommunication. Another explanation is set by van Hout et al. (2010) who argue that children fail to draw a scalar implicature resulting in an arbitrary decision between a determined referent meaning and a non-determined referent meaning when they hear an indefinite article.

The overgeneration of indefinite articles in definite contexts has also been found in several studies. Schaefer and de Villiers (2000) reveal that English-speaking TD children aged 3;6-5;5 overuse the indefinite article in contexts requiring a definite article. Keydeniers et al. (2017) show comparable results for Dutch-speaking 2-year olds. Both studies suggest that the overgeneration of indefinite articles may be caused by a failure to draw scalar implicatures.

Regarding children with ASD, Schaeffer (2016) and Schaeffer et al. (2017) show that they overuse the indefinite article in definite contexts. Schaeffer et al. (2017) propose that children with ASD fail to constantly draw the relevant scalar implicature as a consequence of weak verbal memory and phonological memory. As for children with Autism and a language impairment, this study is the first to investigate this phenomenon.

2.2.2. Direct object scrambling

Direct Object Scrambling (DOS) is a complex linguistic phenomenon in Dutch in a direct object is placed right before or after an adverb or negation. Referential objects need to be

(17)

scrambled (placed before), whereas non-referential objects must remain unscrambled (placed after), as illustrated in the following examples by Schaeffer (2016: 173-174):

(9)

Referential object:

A. Jan heeft het boek goed/niet gelezen - scrambled

John has the book well/not read

‘John read the book well’/’John didn’t read the book’

B. *Jan heeft goed/niet het boek gelezen - non-scrambled John has well/not the book read

C. Jan heeft een boek goed/niet gelezen - scrambled

John has a book well/not read

‘John read a (certain)/one book well.’/’John didn’t read a (certain)/one book.’ Non-referential object:

D. Jan heeft goed/niet een boek/geen boek gelezen - non-scrambled John has well/not a book read

‘John read a book (any book) well.’/’John didn’t read a (any) book.’

Referentiality plays a crucial role in direct object scrambling. Schaeffer (2000) defines referentiality as “a nominal expression that is understood to be referential if it has a “fixed referent” in the (model of the) world, meaning that it can be identified by the speaker and/or by one of the people whose propositional attitudes are being reported” (Schaeffer 2017: 174). Similar to definiteness as discussed in the section above, referentiality is based on shared assumptions between speaker and hearer. When speaker and hearer share the same beliefs, the referent is part of the common ground of both speaker and hearer. Referentiality requires both semantic knowledge (uniqueness, existentiality) and pragmatic knowledge (speaker/hearer beliefs). Furthermore, syntax is required for the movement of the referential direct object. Thus, Direct Object Scrambling involves pragmatic, semantic and syntactic knowledge: to establish definiteness and referentiality (semantics) the speaker has to considerate speaker and hearer beliefs (pragmatics), after which definiteness and referentiality “need to be mapped onto the information structural notions of topic and focus”

(18)

(pragmatics), which are placed in a certain position in the sentence (syntax) (Schaeffer 2017, 177).

Previous research into the acquisition of DOS showed that monolingual TD Dutch-acquiring children between the ages of 2 and 3 years old fail to scramble over negation in referential contexts (Schaeffer 1997, 2000). By means of an Elicited Production Task on DOS, Schaeffer tested a group of 49 TD children between the ages of 2 and 7. Schaeffer attributes this failure as a consequence of the difficulties these young children have in distinguishing between discourse-related referentiality, which is established on the basis of the discourse, and non-discourse-related referentiality, which is established on the basis of long-term shared knowledge. Schaeffer claims that this failure is primarily caused by their immature pragmatics, rather than semantics or syntax.

Regarding children with ASD, Schaeffer (2017) tested their performance on an Elicited Production Task on DOS over negation in comparison to children with SLI and TD children. Both the children with ASD as the children with SLI score significantly worse than the TD group. However, Schaeffer proposes that the underlying cause for the error is different for ASD and SLI. As failure on Direct Object Scrambling can be caused by either a grammatical or a pragmatic impairment, she proposes that children with SLI mainly fail because of their weak grammatical skills, whereas children with ASD fail because of their weak pragmatic skills.

3. The current study 3.1. Research questions

Based on previous research and the materials available, I formulate the following research questions:

1) How do Dutch-speaking children with Autism and a language impairment (ALI) perform on the (morpho)syntactic, pragmatic and extra-linguistic cognitive tests in comparison to children with ASD and TD children?

2) Is there a correlation between the linguistic performance and the executive functions and/or phonological memory?

(19)

3.2. Hypotheses

The research into the linguistic and extra-linguistic cognitive profile of children with Autism and a language impairment is limited, but several studies suggest that there is an overlap between ASD and SLI, resulting in a “new” subgroup within the Autism Spectrum Disorder (Eigsti & Bennetto, 2009; Kjelgaard & Tager-Flusberg, 2001; Perovic et al., 2013; Riches et al., 2010; Tager-Flusberg, 2000). As there are not many studies to base my hypothesis on, I predict that children with ALI have a linguistic profile which is a combination of the impairments found for children with ASD and children with SLI.

Previous research revealed that children with ASD struggle primarily with a pragmatic impairment (Mesibov et al., 2001; Wing, 2000) and children with SLI mainly with a (morpho)syntactic impairment (de Jong, 1999; Rispens & Been, 2007) . I therefore hypothesize that children with ALI perform poorly in both language areas, resulting in both (morpho)syntactic and pragmatic impairments. Additionally, I hypothesize that children with ALI have a weak phonological memory and verbal working memory, as previous research revealed that children with SLI often show signs of weak executive functioning and weak phonological memory (Henry, Messer & Nash, 2012; Meir & Arnom-Lotem, 2014; Schaeffer, 2018).

3.3. Predictions

This study involves many different linguistic and extra-linguistic cognitive tests. In the following section, I formulate predictions for all different tests.

3.3.1. (Morpho)syntactic tests

3.3.1.1. Mass-count distinction

The distinction between flexible mass and count nouns is expressed morphosyntactically (Barner & Snedeker, 2005; van Witteloostuijn & Schaeffer, 2018). Previous research by Schaeffer (2018) shows that Dutch-speaking children with High Functioning Autism perform with the same accuracy as their typically developing controls. Children with a Specific Language Impairment have more problems with this task and score significantly more poorly than the HFA and TD groups. Assuming that children with ALI struggle with syntax, I predict that they perform with less accuracy than the ASD and TD group. As the mass-count

(20)

distinction investigated the morphosyntactic abilities of the participants, I predict both the ASD and the TD group to score at ceiling.

3.3.1.2. Subject-verb agreement

In Dutch, agreement between the subject and the verb is expressed by the addition of a suffix on the verb. This agreement inflection indicates the person and number of the verb, which makes subject-verb agreement a grammatical phenomenon (Schaeffer, 2018; Veenstra et al., 2014; Verhagen & Blom, 2014). Previous research by Schaeffer (2018) shows that children with HFA perform with equal accuracy as the TD children on verbal agreement between the subject and the verb, whereas the children with SLI score significantly more poorly than both the TD as the ASD group. Assuming that the ASD group does not have a grammatical impairment, it is predicted that this group will score with equal accuracy as the TD group. The children with Autism and a language impairment are predicted to score with less accuracy than the ASD and TD group, because they are expected to struggle with this grammatical phenomenon, such as children with SLI do (de Jong, 1999; Rispens & Been, 2007).

3.3.1.3. Relative clause production

The production of object relative clauses demands syntactic and semantic skills. As Dutch has a SOV word-order, the difference between a Subject Relative Clause and an Object Relative Clause is rather complicated. The ambiguity of the different clauses can be resolved by using strategies demanding semantic skills. The correct use of a relative clause demands the use of syntactic skills (Fox & Thompson, 1990; Schaeffer & Siekman, 2016). Following the hypothesis that children with ASD are not grammatically impaired, I predict the ASD group to score like the TD children in the production of object relative clauses. Assuming that syntax and semantics are both part of grammar, I predict that the children with Autism and a language impairment score with less accuracy than the ASD and TD group.

3.3.1.4. Sentence repetition

Sentence repetition is considered a clinical marker in diagnosing Specific Language Impairment as the repetition of read-out sentences demands not only the use of phonological memory, but also involves verbal working memory and syntactic skills (Clay, 1971; Lombardi & Potter, 1992; Potter & Lombardi, 1998; Slobin & Welsh, 1968). Previous research into the performance of children with ASD with normal language and ASD with

(21)

impaired language (ALI) on a sentence repetition task shows that children with ASD score with the same accuracy as TD children, whereas children with ALI perform more poorly (Riches et al., 2010; Harper-Hill et al., 2012; Taylor et al., 2013). Assuming that the ALI group struggles with morphosyntax, it is predicted that this participant group scores with less accuracy than the ASD and TD group, who score with the same accuracy.

3.3.2. Pragmatic tests

3.3.2.1. Article choice

Following the study by Schaeffer, Van Witteloostuijn and De Haan (2017), I predict children with Autism and a language impairment to have difficulties drawing a scalar implicature. For the correct production of definite articles, it is necessary to calculate a scalar implicature for definiteness (Van Hout et al., 2010; Schaeffer et al., 2017). The absence of scalar implicature calculation can influence the choice for an indefinite article in a definite context. Assuming that children with ALI and children with ASD have difficulties in calculating the scalar implicature, I predict that they overuse indefinite articles in definite contexts. Previous research by Schaeffer (2016, 2017) shows that children with ASD tend to overuse indefinite articles in definite contexts. Schaeffer et al. (2017) propose that children with ASD fail to constantly draw the relevant scalar implicature as a consequence of weak verbal memory and phonological memory. However, this absence of scalar implicature calculation will not affect cause the use of definite articles in indefinite contexts, as no scalar implicature is drawn if the definite article, the stronger member of the scale, is used. Therefore, I do not predict overgeneration of the definite article in indefinite contexts and expect them to score with the same accuracy as their typically developing peers.

3.3.2.2. Direct Object Scrambling

To be able to correctly scramble the direct object, the speaker has to make use of pragmatic, semantic and syntactic knowledge. First, the speaker has to take into account the beliefs of both speaker and hearer (pragmatics) to be able to establish definiteness and referentiality (semantics). Then, definiteness and referentiality need to be structuralized in the notions of topic and focus (pragmatics), which reflect the position in the sentence (syntax) (Schaeffer, 2000, 2017). Previous research showed suggests that children with ASD primarily struggle with the pragmatic aspect of this scrambling process. In contrast, children with SLI primarily

(22)

struggle with the syntactic part of the scrambling (Schaeffer 2017). I therefore predict that children with both autism and a language impairment perform more poorly on the scrambling task than children with ASD. I predict that both the ASD and the ALI group will perform less accurate than the TD group.

3.3.3. Extra-linguistic cognitive tasks

3.3.3.1. Non-word repetition

The Non-Word Repetition task tests the phonological memory by requesting the participant to repeat non existing words. Previous research showed that this task is often problematic for children with SLI and can therefore be used as indicator for a possible language impairment (Rispens and Baker, 2012). Some studies also show reduced phonological memory performance by children with ASD (Macizo et al., 2016). Therefore, I predict that the ALI group will struggle the most with this task and will perform with less accuracy than the ASD and the TD group. Additionally I predict the ASD group to score with less accuracy than the TD group as well, assuming that the typically developing controls will score at ceiling.

3.3.3.2. Digit Span Forward

The Digit Span Forward task also tests the phonological memory by asking the participants to repeat digits, up to a level of a maximum of eight digits in a row. Because previous research showed that as well children with ASD as well as children with ALI have a weak phonological loop (Loucas et al., 2010; Macizo et al., 2016), I predict the children with ALI to score with less accuracy than the children with ASD and TD children. But I also predict the ASD group to score more poorly than the TD group.

3.3.3.3. Digit Span Backward

In the Digit Span Backward task, the participant is asked to repeat the digits but in the backward order. In this way, the participant cannot rely on the phonological memory only, but has to process the received information by using the verbal working memory. Previous research showed that children with ASD show a reduced phonological memory, but there were no significant differences found in the working memory overall (Loucas et al., 2010; Schaeffer, 2018). Therefore, in contrast to the predictions for the Digit Span Forward task, I

(23)

expect predict the ASD group to score at the TD norm, while I predict that the ALI group will score with less accuracy than the ASD and the TD group.

3.3.3.4. Odd-one-out

By means of the Odd One Out task, non-verbal working memory is tested. As this task does not contain a linguistic element, I predict that the ALI, ASD and TD group will score with equal accuracy.

3.3.3.5. Luria’s handgame

During Luria’s handgame, participants are requested to copy or inhibit certain hand positions to assess the inhibition. Previous studies show that there are differences in the executive functions in behavioral regulation, including inhibition, between children with ASD and TD (Leung et al. 2016, Berenguer et al. 2017). Assuming that both the ASD as the ALI group share this pattern, I predict the ALI and ASD group to score with less accuracy than the TD group. I predict that there will not be a difference in performance between the ALI and ASD group as this task is non-linguistic.

4. Method 4.1. Participants

To investigate the grammatical, pragmatic and cognitive abilities of children who are diagnosed with ASD and a language impairment (ALI), three groups of participants were used that were tested earlier by Jeannette Schaeffer and Iris Duinmeijer as part of their research. All children were matched individually to children with ASD and TD children on the basis of age and gender. The three groups of participants are presented in table 1.

Table 1. Overview of the three participant groups Group N Age (mean/SD) Sex

ALI 8 10.14 (2.90) 8 M

ASD 8 9.94 (3.22) 7 M, 1 F TD 8 10.17 (2.98) 6 M, 2 F

(24)

The children from the ALI and ASD group were officially diagnosed by a psychiatrist with Autism Spectrum Disorder based on the DMV-IV (American Psychiatric Association, 2000). These children were either diagnosed with Classic Autism, Asperger Syndrome or PDD-NOS, which are all considered as subtypes of High-Functioning Autism. Through Dutch organizations for autism, Social Media groups regarding autism and personal contacts, the children were recruited to be part of this research. For the original study by Schaeffer and Duinmeijer, children with SLI were tested as well. These children were recruited from special schools for children with Speech and Language problems. As this current study focusses on the grammatical, pragmatic and cognitive abilities of children with both ASD and a language impairment in comparison to children with ASD and TD children, the SLI group is not taken into account for the present study.

During and after the study by Schaeffer and Duinmeijer, some of the participants turned out to have an additional language impairment or Autism Spectrum Disorder which was not known at the time of the recruitment. These children were therefore excluded from the original study. For the present study, the children with ASD and a language impairment are the main point of interest presented in the ALI group.

4.2. Materials and procedure

The tests used in the current study are part of a large battery of 16 tests designed by Jeannette Schaeffer and Iris Duinmeijer at the University of Amsterdam. The tests are designed to investigate the grammatical, pragmatic and cognitive development of children with SLI and children with HFA. All tests used are suitable for the Dutch language and executed in Dutch. 4.2.1 Morphosyntactic tests

4.2.1.1. Mass-count distinction

For this test, a Quantity Judgement Task was used based on previous research by Barner and Snedeker (2005) about the distinction between mass and count nouns (see paragraph 2.1.1.). During this task, participants were presented with an image in which two characters were pictured. In every image one of the characters had two large objects and the other character had four, five or six smaller objects of the same kind. Importantly, the volume and surface area of the two large objects was bigger than the smaller objects combined.

The task consisted of 12 different conditions for which flexible nouns were used. These nouns can function as either mass noun (without plural marking) or count noun (with plural marking). During the task, the participants were asked to indicate which character has more

(25)

of “X”. The syntax of the noun (X) indicates which answer is the most fitted for that context. Count syntax presents a noun which elicits an answer based on the number of individual items present in the image. Mass syntax presents a noun which elicits a response based on the volume of the item visible in the picture. Next to the 12 experimental items per condition, 8 filler items were used. The filler items consisted of count nouns, which differed only in the total number of items per character, instead of both number and overall volume. All items were presented to the participants in pseudo-randomized order using Microsoft PowerPoint and prerecorded utterances. An example of an experimental item is presented below.

(10) Count Syntax:

Wie heeft er meer pizza’s?

‘Who has more pizza-PL?’

Target response: the horseman (11) Mass syntax:

Wie heeft er meer pizza?

‘Who has more pizza?’

Target response: the cowboy

4.2.1.2. Subject-verb agreement

Subject-verb agreement was investigated using an Elicited Production Task designed by Duinmeijer (2012). During this task, participants were asked to describe the actions which were presented on different cards. The task was designed as a game in which the participant, the experimenter and a doll named Kim all received a pile of 30 cards. All cards presented a person who acts out one of the six transitive verbs (bake, comb, read, clean, film and drink). The different cards were presented upside down and at each round the upper card of the pile

(26)

was turned over. Subsequently, the participant was asked to describe what everyone is doing according to the image on the card. The five conditions (first, second, and third person singular and first and second person plural) are presented by 12 items each, using six transitive verbs twice.

4.2.1.3. Object relative clause production

To elicit the production of relative clauses, the participants listened to a short story from which the participant had to choose which situation described in the story he/she prefers. The task consisted of three different conditions (presented in table 2), from which only the Object Relative Clause reversible (ORC) is included in this study. The other conditions were used as filler items. To elicit an ORC , the participant listened to a story in which the object had to undergo two different actions. Subsequently, the participant was asked which of the two characters he/she would rather be. An example of an experimental item is presented below: (12)

Er zijn twee jongens en een vader. De vader slaat een jongen en de vader knuffelt een jongen. Welke jongen ben je liever?

‘There are two boys and a father. The father hits a boy and the father hugs a boy. Which boy would you rather be?’

Target: Ik ben liever de jongen die de vader knuffelt (ambiguous)

‘I’d rather be the boy who the father hugs’

Table 2. The different conditions of the elicited RC production task

Condition Number of items

1 – subject RC irreversible 6 2 – subject RC reversible (sg & pl)

6 SG, 6 PL 3 – object RC reversible (sg & pl) 6 SG, 6 PL

(27)

4.2.1.4. Sentence repetition

The ability to repeat sentences from different lengths was tested using the Sentence Repetition Task from the Dutch version of the Clinical Evaluation of Language Fundamentals (CELF-4-NL) (Semel et al., 2008). During this test, the participant was asked to repeat the sentence which was read by the experimenter. The experimenter was only allowed to read out the sentence once. During the task, the sentences gain in length and level of difficulty.

4.2.2. Pragmatic tests

4.2.2.1. Article choice production

The production of articles was tested using an Elicited Production Task based on a study by Schaeffer and Matthewson (2005). During this test, the participants got to see a picture or short video clip on a computer screen using Microsoft PowerPoint. Then, they were asked to describe the shown event to an experimenter (A) who could not see the screen, while another experimenter (B) was sitting next to the participant. The 18 experimental items tested three conditions: six items in a definite condition, six items in an indefinite referential condition and six items in an indefinite non-referential condition. The article choice task consisted of 18 fillers which tested direct object scrambling (see section 4.2.2.2). An example of a test item in the definite condition is presented below.

(13) Definite condition

Picture 1 Picture 2 Picture 1:

(28)

Experimenter A (who cannot see the screen): ‘Hey, who do you see in the picture?’ Participant: [Name of het puppet]

Experimenter A: ‘And what else?’ Participant: ‘A bear!’

Picture 2: [the image changes into a short movie clip in which the puppet hugs the bear] Experimenter A: ‘And what did [name] just do?’

Target response: Hij knuffelde de/het beer. ‘He hugged the bear

4.2.2.2. Direct object scrambling

As mentioned in section 4.2.2.1, direct object scrambling and article choice were tested in the same experiment, using as fillers for each other. Direct Object Scrambling was tested using an Elicited Production Task based on methods used by Schaeffer (2000). All experimental items used sentential negation to elicit direct object scrambling. In this way, the participants are provided a clear context as to whether they have to scramble the direct object, where adverbs do not. As in the article choice experiment, three different conditions are distinguished with six test items each, namely a definite condition, an indefinite referential condition and an indefinite non-referential condition. An example of an experimental item in the definite condition is presented below.

(14) Definite condition

B: ‘Patrick is bored and he is looking for something to do. “Hey”, says Patrick, “a book! But I don’t like books”.

(29)

“Dus dat ga ik NIET lezen”.

so that go I not read - ‘So I’m not going to read it.’

A: Ik weet het. Het boek gaat Patrick WEL lezen.

I know it. the book goes Patrick yes read - ‘I know. Patrick is going to read the book.’

Participant: ‘No’.

B: ‘No? What’s really happening?’

Target: Patrick gaat het boek NIET lezen. (scrambled)

Patrick goes the book not read - ‘Patrick is not going to read the book.’

Nontarget: Patrick gaat NIET het boek lezen. (non-scrambled)

Patrick goes not the book read - ‘Patrick is not going to read the book.’

4.2.3. Extra-linguistic cognitive tests

4.2.3.1. Non-word repetition

Phonological memory was tested using the Non-Word Repetition task as designed by Rispens and Baker (2012). In this task, the participants had to repeat nonsense words which varied in syllable length and phonotactic probability. The participants only got to listen to the word once.

4.2.3.2. Digit span forward and backward

Verbal working memory and phonological memory was assessed using the Digit Span task as designed by Wechsler (1974). In the Forward Digit Span task the participant was asked to repeat digits, up to a level of a maximum of eight digits in a row. This task mainly tests the phonological memory level of the participants. In the Backward Digit Span task, the participant had to repeat the digits in reverse order. Reversing the digits requires the participants to use their verbal working memory in order to process the digits heard.

4.2.3.3. Odd one out

Non-verbal working memory was tested using the Odd-one-out test as designed by Henry (2001). During this test, the participants got to see three figures from which they had to point at the odd-one-out figure. Subsequently, they were asked to indicate the blank position where the odd-one-out figure was before.

(30)

By means of the nonverbal task Luria’s handgame (Henry, Messer & Nash 2012) , the inhibition of the participants was tested. During this game, the participant were asked to copy and inhibit certain hand position which were demonstrated by the experimenter.

5. Results

5.1. Results grammar 5.1.1. Mass-count results

Figure 1 presents the collapsed accuracy scores for the flexible mass and count conditions for all groups. Because not all children participated in this task, we had to exclude some results in our statistical analysis. This analysis is therefore done with only 5 participants per group. A linear regression analysis, using the lm() function in R, reveals a difference between accuracy scores of the groups with a mean of 95% for the TD group, 83.35% for the ALI group and 83.35% for the ASD group. There is no evidence found that there is a significant difference between the groups. Although the ALI and ASD group have a lower accuracy than the TD group, this difference is not significant (p= .211).

5.1.2. Subject-verb agreement results

Figure 2 presents the accuracy scores for subject-verb agreement for all groups. A linear regression analysis, using the lm() function in R, reveals a difference between accuracy scores of the groups with a mean of 100% for the TD group, 83% for the ALI group and 96% for the ASD group. Similar to the results on the Mass-count task, there are no significant differences between the groups. Where the TD group and ASD group score at ceiling, the

(31)

ALI group scores with lower accuracy. However, this difference is not significant (p= .098).

Fig. 1. Accuracy (in percentage) on the Mass-Count test

Fig. 2. Accuracy (in percentage) on the Subject-Verb Agreement test

5.1.3. Object relative clause production results

Figure 3 presents the accuracy of the singular and plural responses in the object condition for all groups in the relative clause production task. Because not all children participated in this task, we had to exclude some results in our statistical analysis. This analysis is therefore done with only 6 participants per group. A linear regression analysis, using the lm() function in R, reveals a difference between accuracy scores in the singular condition of the groups with a mean of 97% for the TD group, 56% for the ALI group and 81% for the ASD group. For the plural condition, the TD group scored with an accuracy of 92%, the ALI group with 45% and the ASD group with 64%. The analysis reveals a significant difference between the TD group and ALI group in the singular condition (p= .041) and in the plural condition (p= .032). There were no significant differences found between the ASD and TD group, or the ASD and ALI group.

(32)

Fig. 3. Accuracy (in percentage) on the RCP task in singular and plural condition.

5.1.4. Sentence Repetition Results

Figure 4 presents the scores for all groups in the sentence repetition task, which was part of the larger test battery of the Dutch edition of the Clinical Evaluation of Language Fundamentals (CELF-4-NL). Because not all children participated in this task, we had to exclude some results in our statistical analysis. This analysis is therefore done with only 7 participants per group. A linear regression analysis, using the lm() function in R, reveals a difference in scores with a mean of 14 for the TD group, 7 for the ALI group and 10 for the ASD group. The analysis reveals that the ALI group scored significantly more poorly than the TD group (p= .0017). We cannot conclude that there was a significant difference between the performance of the ASD and TD group, although the analysis reveals that it was near significant (p= .0542).

(33)

fig. 4. Score on the Sentence Repetition Task. 5.1.5. Summary

Summarizing the statistical analysis on the four (morpho)syntactic tests, we can draw the following picture:

(15)

Mass-count distinction: TD=ALI=ASD Subject-verb agreement: TD=ALI=ASD Object relative clause production: TD>ALI, TD=ASD Sentence Repetition: TD>ALI, TD=ASD

The results of the ALI group do not fully match our predictions. As predicted, the children with ALI performed significantly worse than their TD age mates on the Relative Clause Production test and Sentence Repetition test. However, on the Mass-Count Distinction test and the Subject-verb agreement test, there were no significant differences found between the three groups. The ASD group performed according to the predictions and did not show any significant differences with the TD group. As the children with ALI seem to struggle with only part of the (morpho)syntactic tasks, the question raises whether this poor performance is caused by a morphosyntactic impairment, or that there are other factors underlying this performance.

5.2. Results pragmatics

5.2.1.Article choice production results

Figure 5, 6 and 7 present the accuracy, substitution and irrelevant responses for the definite condition (5) and the two indefinite conditions (6 and 7) of the article choice test for all groups. When the children use a definite article in an indefinite condition or an indefinite article in a definite condition, the response is called a substitution response. By means of a linear regression analysis, using the lm() function in R, I now present the results in the different experimental conditions.

As for the definite conditions, the tall bars show that all groups primarily choose to produce the definite article, which was the target answer. The analysis reveals that none of the differences between the participants groups or response types is statistically significant.

(34)

The children with TD and the children with ASD both score at ceiling with a mean accuracy of 92%. Although the mean accuracy of the ALI group was lower (79%), this difference was not significant (p= .265).

Fig. 5: Accuracy (in percentages) on the Article choice test in definite condition

As for the indefinite conditions, analysis shows that all groups primarily choose the target indefinite answer. For both conditions, all three participant groups score at ceiling. Therefore, none of the differences between the participants groups or response types is statistically significant. The children with ALI score better than predicted and do not show any significant differences with their typically developing peers (indefinite referential: 95% c.i.= -0.54 .. 17.19, p=0.064; indefinite non-referential: 95% c.i.=-9.12 .. 11.12, p=0.835). The children with ASD also score better than predicted as no significant differences were found with either of the participants groups.

(35)

Fig. 6. Accuracy (in percentages) on the Article choice test in indefinite referential condition .

Fig. 7. Accuracy (in percentages) on the Article choice test in indefinite non-referential condition.

5.2.2. Direct object scrambling results

Figure 8 and 9 present the proportions of scrambled, non-scrambled and irrelevant responses of the Direct Object Scrambling test for the non-referential and the referential condition. Figure 9 presents the mean of the proportions of responses in the two original referential condition, namely the definite condition and the indefinite referential condition. The referential condition requires the scrambling of the direct object, while non-referential condition does not. Because not all children participated in this task, we had to exclude some results in our statistical analysis. This analysis is therefore done with only 7 participants per group.

As for the referential condition, a linear regression analysis, using the lm() function in R, reveals a significant difference between the ALI group and TD group for the proportion of non-scrambled responses (95% confidence interval: -71.8 .. -6.49, p= .023). Analysis shows no significant differences between the groups in the proportion of scrambled responses, which were the target responses for this condition (ALI - TD: p= .824; ASD-TD: p= .575).

(36)

Fig. 8. Proportions of responses in the Direct Object Scrambling test for the referential condition.

As for the non-referential condition, a linear regression analysis, using the lm() function in R, reveals a difference in the proportion of non-scrambled responses of the different groups with a mean of 93% for the TD group, 88% for the ALI group and 83% for the ASD group. There is no evidence found that there is a significant difference between these groups (p=0.055,). As well for the scrambled, as the irrelevant responses, there is no evidence found that there is a significant difference between the groups.

Fig. 9. Proportions of responses in the Direct Object Scrambling task in the non-referential condition

(37)

Summarizing the statistical analysis on the two pragmatic tests, we can draw the following picture:

(16)

Article choice Definite: TD = ALI = ASD Indefinite: TD= ALI =ASD

Direct Object Scrambling Referential: TD>ALI, TD=ASD Non-referential: TD=ALI=ASD

Except for the non-scrambled responses in the referential condition, all three groups perform with equal accuracy on the pragmatic tests. This contradicts the prediction that the ASD and ALI group perform more poorly than the TD children on pragmatic tests. As these results were other than predicted, the question raises to what extent the pragmatics of children with ASD and children with ALI is impaired or whether there are other factors, such as memory, that can explain this result.

5.3. Results extra-linguistic cognition

5.3.1. Non-word repetition results

Fig. 10 presents the accuracy for the Non-word Repetition task for all groups .A linear regression analysis, using the lm() function in R, reveals a difference between accuracy scores of the groups with a mean of 73% for the TD group, 46% for the ALI group and 60% for the ASD group. Analysis reveals that as well children with ALI score significantly lower than TD children (p= .0003), as children with ASD (p= .033).

5.3.2. Digit span forward results

Fig. 11 presents the score for the Digit Span Forward task for all groups. A linear regression analysis, using the lm() function in R, reveals a difference between scores of the groups with a mean of 9 for the TD group, 4 for the ALI group and 7 for the ASD group. Analysis reveals that as well children with ALI score significantly lower than TD children (p= .0005), as children with ASD (p= .0253). Children wit ALI also score significantly worse than children with ASD (p=.041).

(38)

Fig. 10. Accuracy (in percentages) on the Non-word Repetition task. Fig. 11. The mean score on the Digit-Span Forward task

5.3.3. Digit span backward results

Fig. 12 presents the scores for the Digit Span Backward task for all groups. A linear regression analysis, using the lm() function in R, reveals a difference between the scores of the groups with a mean of 6 for the TD group, 3 for the ALI group and 4 for the ASD group. Analysis reveals that as well children with ALI score significantly lower than TD children (p= .035), as children with ASD (p= .045).

Fig. 12. The mean score on the Digit-Span Backward task.

5.3.4. Odd-one-out results

Fig. 13 presents the score for the Odd-one-out test for all groups. A linear regression analysis, using the lm() function in R, reveals a difference between accuracy scores of the

Referenties

GERELATEERDE DOCUMENTEN

As one expert said concerning features of software for early literacy (RQ2), “What is important to me is that the teacher knows that the combination of images and spoken words is

In case we failed to find slips of action in our behavioural data, we intended to use the automaticity score of not taking the pill during the beginning (i.e. the first three days)

This paper attempts to study how entrepreneurs’ stable psychological attributes such as thinking style influence entrepreneurial decision-making behaviors associated with the

Echter, de definitie van prenatale gehechtheid zoals is omschreven door de ontwikkelaars van het meetinstrument (Van Bakel et al., 2013) als “de liefdevolle sensitieve band die

The algorithm allows to efficiently compute properties of product lines whose feature model consists of a feature tree and a small number of "requires"

However, where children, a husband and family form important reasons for these Nicaraguan women to leave Nicaragua, come to Costa Rica and stay there for a while, there are also

By experimental design, Grosshans and Zeisberger show price paths have the potential to influence investor satisfaction and risk tolerance, also evidence for the presence of

The saturated semicrystalline polymer (P1-H) is water-insoluble but undergoes rapid backbone hydrolysis under neutral, basic, or acidic conditions when polymer films were immersed