DEPARTMENT OF DUTCH STUDIES UNIVERSITY OF AMSTERDAM
The acquisition of article choice and subject-verb agreement by Dutch speaking
children with Autism Spectrum Disorder
Student: Ana Marija Žagar Student number: 10504796
Thesis supervisor: dr. Jeannette Schaeffer Second reader: prof. dr. Aafke Hulk
Amsterdam, 7-7-2014
Table of Contents
1. Introduction... 7
2. Background... 9
2.1 Pragmatics and grammar in children with ASD... 9
2.2 Acquisition of articles and article choice in Dutch... 10
2.3 Subject-verb agreement in Dutch... 16
3. Hypotheses and predictions... 17
4. Methods... 19
4.1 Participants... 19
4.2 Definiteness production task ... 19
4.3 Definiteness judgement task... 23
4.4 Subject-verb agreement production task... 25
4.5 False-belief task ... 26
5. Results and discussion ... 28
5.1 Results Definiteness production task... 28
5.2 Results False-belief task... 36
5.3 Results Definiteness judgement task ... 37
5.3 Results Subject-verb production task ... 47
6. Conclusion... 49
References ... 52
Abstract
Autism spectrum disorder (ASD) has been known to entail impairments in pragmatics, but not in grammar. Assuming that pragmatics is necessary for the correct article choice and grammar for the acquisition of subject-verb agreement, high-functioning children with ASD are predicted to experience difficulties in the former, but not in the latter. An experimental study with 27 typically developing (TD) Dutch children and 27 Dutch children with ASD was carried out to test these predictions. The findings of the study indicate that high-functioning children with ASD are impaired neither in pragmatics, nor in grammar. However, they do show that younger children with autism are pragmatically immature, which is the reason why they experience some difficulties in the article choice. Since some children seem pragmatically immature, while they experience no trouble with grammar, the results could be considered as an indication that there is dissociation between grammar and pragmatics or that grammar precedes pragmatics.
Keywords: Autism spectrum disorder, pragmatics, grammar, article choice, subject-verb agreement, scalar implicatures, Concept of Non-shared Assumptions, Theory of Mind, Dutch
Acknowledgments
This thesis is one of the results of the project initiated by dr. Jeannette Schaeffer, which I was lucky to be a part of. Hereby, I would like to express my appreciation for all the help and guidance that dr. Schaeffer offered me. I also wish to thank other colleagues that worked on the project, especially Drs. Merel van Witteloostuijn, for collecting and analysing the data and helping me in the analysis of the same.
Writing my thesis would not have been the same without my friends, Lindsey Allen, Mirko Cvetković and Nikolina Pavić, who supported me and were always there to give me tips or advice.
Finally, I would like to thank my family, especially my mother, for their encouragement and support throughout my studies.
1. Introduction
Autism Spectrum Disorder (ASD) has been the subject to much research. Even though there are still questions that need answers, various studies (Astington & Jenkins 1999; Varley, Siegal & Want 2001; Klin 2006; Colle, Baron-Cohen & Hill 2007; De Villiers 2007; Boucher 2012) have helped us understand ASD better. It has become known that autism, as a spectrum disorder, entails degrees and symptoms varying between individuals, but it typically involves impairment in social skills, language and behaviour. Language impairments in children with ASD are usually concentrated in the area of pragmatics, or the social uses of language. For instance, if a mother said to her child “Stop back-talking to me”, the child could say, “I’m sorry mom, I’ll talk to your front”. The child in this case understands it literally and not as a mother’s request for him to stop being rude and answer her questions with respect.
Even though children with autism have impaired pragmatics, their morphology and syntax are relatively spared (Tager-Flusberg 1990). This means that, as opposed to pragmatics, they do not experience problems when it comes to making use of their grammar.
This likely disjunction between grammar and pragmatics inspired many questions, two of them being the acquisition of correct article choice and subject-verb agreement. The acquisition of correct article choice can be argued to be part of pragmatics, because to be able to use definite and indefinite articles appropriately, as in (1) and (2), one needs to be able to distinguish between speaker and hearer beliefs. It is often assumed that children with ASD do not understand that others have beliefs, desires and intentions that are different from their own, the phenomenon known as Theory of Mind (ToM). For example, if somebody picked up a moisturiser instead of toothpaste while brushing their teeth, children with ASD would not understand why the person did that, but would assume that that is something that the person wants to do. If children really do not posses ToM, this could be consequently reflected in pragmatics and have repercussions for the acquisition of correct article choice.
(1) Dit is een lied over een konijntje. Het konijntje woont in een groot huis. This is a song about a rabbit. The rabbit lives in a big house.
(2) De sterren fonkelen. The stars are twinkling.
Dutch indefinite article een is used before a count noun when talking about something that has not yet been defined or is introduced to the conversation or context for the first time, whereby the item introduced is known by the speaker, but not the hearer. Conversely, although definite articles de and het can be used before both count and non-count (mass) nouns, they are used when the noun is defined and both the speaker and the hearer are familiar with the item it denotes and know what it refers to (ans.ruhosting.nl).
If only the speaker believes the referent to exist, an indefinite article should be used (as in een konijntje in (1)). The introduction of een konijntje by the speaker places the referent of konijntje in the so-called common ground between speaker and hearer, prompting the use of the definite article het (as in het konijntje in (1)). Even though it is not previously introduced by the speaker, the referent of sterren in (2) is also part of the common ground. This is so because the referent of sterren is part of long-term shared beliefs between speaker and hearer. The existence of the stars has always been known to the speaker and the hearer and does not have to be explicitly introduced in prior discourse, which makes it a permanent part of the common ground.
As opposed to the pragmatically driven choice of appropriate articles, subject-verb agreement is usually assumed to be a purely grammatical phenomenon. In Dutch, finite verbs agree with the subject in number and person, e.g., hij werk-t (‘he sg’) vs. zij werk-en (‘they work-pl’) and ik werk-0 (‘I work-1st p’) vs. hij werk-t (‘he work-3rd p’).
Since article choice is pragmatically driven and because pragmatics is hypothesized to be impaired in children with ASD, I expect Dutch children with ASD to have difficulties in the choice of articles. However, since their grammar is not affected, they should not experience difficulties with subject-verb agreement.
In section 2.1, more information is given about pragmatics and grammar in children with ASD and some previous studies are presented. Section 2.2 describes the acquisition of articles and article choice in Dutch and section 2.3 describes the subject-verb agreement in Dutch. The hypotheses and predictions are formed in section 3. Section 4 explains the methodology that was used to obtain the data, which are presented and discussed in section 5. Finally, section 6 summarizes and concludes the paper.
2. Background
2.1 Pragmatics and grammar in children with ASD
According to Global Autism Collaboration (autism.org), in severe cases, children with ASD may not interact with others or they may treat people as objects, whereas in milder cases “it involves difficulty in understanding and relating to others and difficulty in understanding other people’s perspectives and emotions”.
People with ASD are often impaired in the use of nonverbal forms of communication and in initiating or sustaining a conversation with others. They often lack behaviours indicating sharing of experiences, shared communication (e.g. joint attention) and spontaneous make-believe or social imitative play. Moreover, they show delays in the development of spoken language and when they do speak their language, “while the syntax and morphology of language are relatively spared, vocabulary and semantic skills may be slow to develop and aspects of the social uses of language (pragmatics) are particularly difficult for individuals with autism” (Klin, 2006, p. 56). As Klin (2006) states, “the person with autism may fail to appreciate the speaker’s communicative intent, resulting in an overly literal interpretation of the utterance” (p. 57).
Several studies (Wang et al. 2006; Pexman et al. 2011) have provided the evidence demonstrating that children with ASD experience difficulties with pragmatics. Wang et al. (2006) did a study on understanding irony in high-functioning children with ASD (HFASD), in which participants listened to short scenarios and decided whether the speaker was sincere or ironic. As Wang et al. (2006) state, “irony […] involves rather complex mental representations, as listener needs to understand not only that the speaker does not mean exactly what she/he said, but also that she/he does not expect to be taken literally” (p. 129). The results of the study showed that even though children with ASD performed above chance levels, they were less accurate than TD children in detecting the communicative intent behind a speaker’s remark, especially in the conditions where the outcome of the event strongly indicated a non-literal interpretation (Wang et al. 2006, p. 938). This suggests, as Wang et al. (2006) explain, “that children with ASD had difficulty in taking advantage of the contextual information available to interpret the speaker’s intent” (p. 938).
Additionally, in their study, Pexman et al. (2011) examined processing of verbal irony in children with HFASD showing that children struggled to appreciate the broader social
functions of irony. Namely, “they did not share the understanding conveyed by the typically-developing children that ironic criticisms were intended to be simultaneously critical and humorous” (Pexman et al. 2011, p. 1108).
As opposed to impairment in pragmatics, in their paper, Tager-Flusberg et al. (1990) show that children with ASD follow the same general developmental path in the acquisition of grammar as TD children. By analysing spontaneous child utterances of higher functioning children with autism in the mean length of utterances measured in morphemes, Index of Productive Syntax, lexical diversity and Form Class Distribution, they show that “autism does not involve a fundamental impairment in grammatical ability” (Tager-Flusberg et al. 1990, p. 15).
2.2 Acquisition of articles and article choice in Dutch
The acquisition of correct article choice can be thought of as part of pragmatics, because to be able to use definite and indefinite articles appropriately, one needs to be able to distinguish between speaker and hearer beliefs. As Rozendaal and Baker (2008) explain, “children must not only learn to use determiners, but also the pragmatic discourse conditions under which these forms can be used” and, as they state, children experience difficulties up until at least six years of age in several aspects of pragmatic language use, such as taking into account the listener’s perspective (p. 774).
When it comes to determiners, or more precisely articles, in Dutch, there are three forms from which a speaker can choose: the definite articles de and het and the indefinite article een. As Van Hout et al. (2008) state, indefinites involve existential quantification and are used when a new referent is introduced, as in (3) (p. 1). However, just as English indefinite article a, the indefinite article een in Dutch can be referential and non-referential. As shown in (4), a referential indefinite article is used when a referent is believed to exist by the speaker, while non-referential indefinite is used when the existence of the referent of the noun phrase is neither known to the speaker nor to the hearer, as in (5) (Schaeffer & Matthewson 2005; Van Hout et al. 2008). Unlike indefinite articles that can be either referential or non-referential, as Van Hout et al. (2008) explain, definite articles are always referential and “carry the presuppositions of existence and uniqueness and typically refer to referents previously introduced in the domain of discourse which are part of the common ground”, as in (3) (p. 1).
(3) A bird and a cat were sitting in a tree. The bird flew away. (Van Hout et al. 2008) (4) (Schaeffer & Matthewson 2005, p. 95)
Situation: picture of Mickey Mouse who just finished painting a car Elmo: Hey, who is this (pointing at Mickey Mouse)?
Child: Mickey Mouse!
Elmo: And what did Mickey Mouse just do? Child: He painted a car.
*He painted the car.
(5) (Schaeffer & Matthewson 2005, p. 96)
Pluto: Oh, I’m so bored, I don’t know what to do. Oh, you know what, I’m going to the book store, and I’m gonna BUY something there.
Elmo: What do you think Pluto is gonna do in the book store? Child: He’s gonna buy a book!
*He’s gonna buy the book!
Some acquisition studies (Schaeffer & Matthewson 2005; Van Hout et al. 2008; Van Hout et al. 2010) show that English-acquiring children younger than 5 overgenerate or overuse the definite article in indefinite referential contexts and several explanations have been offered why this is so.
Schaeffer and Matthewson (2005) attribute the overgeneralization of the English definite article the to the lack of a pragmatic concept, the Concept of Non-Shared Assumptions (CNSA), which states that speaker and hearer assumptions are always independent. The choice between definite and indefinite articles depends on the beliefs of the interlocutors, which means that definite article the is used when the existence of a (unique) referent corresponding to the determiner phrase (DP) is in the common ground of the discourse, i.e. is part of the shared beliefs between speaker and hearer at the time of utterance, making environment A (believed by speaker and hearer), as in (6). However, when indefinite article a is used, the beliefs of the speaker and hearer are not shared and are not part of common ground, as explained in (4) and (5), resulting in environments B (believed by speaker only) and C (believed by neither speaker nor hearer), as presented in (6).
(6) A believed by speaker and hearer part of common ground the B believed by speaker only not part of common ground a
Pragmatically, because of the CNSA, the speaker has to think of the hearer’s assumptions as separate from the ones they themselves hold, even though they can coincide in certain cases. Therefore, if a child lacks the CNSA, in some situations, as a speaker, s/he may attribute his/her own assumptions to the hearer. As Schaeffer and Matthewson (2005) explain, children do not distinguish between speaker and hearer beliefs and attribute their own (speaker-) beliefs to the hearer and therefore do not make a distinction between environment B (believed by speaker only) and environment A (believed by speaker and hearer). Because of that environment B becomes environment A and children use the article appropriate for environment A, which is the definite article the. That is why children would then overgenerate definite article to (adult) B contexts, which are believed by speaker only and require the indefinite article a.
The CNSA is somewhat comparable and could be related to ToM because to have ToM means to know that other people’s beliefs, desires and intentions are not necessarily the same as ours (Schaeffer & Matthewson 2005). For that reason, it would be possible to draw a parallel between children’s overgenerating definite articles in indefinite environments, especially in B contexts, and the same children’s not passing false-belief task, which is commonly used to test children’s ability to distinguish their own beliefs from somebody else’s, i.e. children’s possession of ToM (Colle, Baron-Cohen & Hill 2007). According to Colle et al. (2007), various studies suggest that when children are about 4 years old, they experience a fundamental change in their ToM (p. 716). As Colle et al. (2007) explain, “children’s mastery of the semantics of mental state terms emerges at the same age as their mastery of false belief (FB) tasks”, which implies that false-belief is part of ToM (p. 716). In their study of production and comprehension of definite and indefinite articles, Van Hout et al. (2010) also find the overuse of definite article in production. Unlike Schaeffer & Matthewson (2005), who explain this by the lack of the CNSA, Van Hout et al.’s (2010) reasoning falls within the framework of Optimality Theory. In Optimality Theory, rules of grammar are viewed as an interplay of higher- and lower-ranked constraints (Van Hout et al. 2010). According to Van Hout et al. (2010), there are two constraints at work when it comes to the production and understanding of articles: DETERMINED REFERENCE and AVOID INDEFINITES. As Van Hout et al. (2010) state, “If a language has a definite article, definites are always associated with determined reference”, which makes DETERMINED REFERENCE the higher ranked constraint. This means that it may “overrule lower-ranked constraints, even to the extent that a lower-ranked rule is violated, as long as the
higher-ranked one is satisfied” (Van Hout et al. 2010, p. 1976). AVOID INDEFINITES is a lower-ranked constraint because indefinite articles are to be avoided in general, since definite articles are more informative. According to Van Hout et al.’s (2010) theory, we use the constraints bi-directionally, so “the hearer not only relies on the application of these constraints in the interpretation process from form to meaning”, but also “engages the speaker perspective by considering the other forms that a speaker might have chosen to express her message, but did not” (p. 1976).
In their acquisition of articles, children have to discover the ranking of the constraints from the input. Van Hout et al.’s (2010) results show that children younger than 5 overuse the definite article the, which suggests that children of that age have not yet acquired the correct ranking.
According to Van Hout et al. (2010), while adults use the constraints bi-directionally, children use them in a unidirectional way because they are still pragmatically immature (p. 1976). As presented in the unidirectional tableau 1, when the constraints are unranked in the case of expressing a non-determined reference meaning, both articles are winners since the violations are on the same level. When a is used, AVOID DEFINITES is violated and when the is used, DETERMINED REFERENCE is violated since the referent is non-determined. Because of this, children whose ranking is undetermined produce both articles (Van Hout et al. 2010, p. 1987). However, as Van Hout et al. (2010) explain, “children for whom AVOID INDEFINITES is ranked highest will produce the across board” (p. 1987).
Meaning input Non-determined referent DETERMINED REFERENCE AVOID INDEFINITES a * the *
Tableau 1: The optimal form expressing a non-determined reference meaning with unranked constraints (Van Hout et al. 2010, p. 1987)
However, Van Hout et al.’s (2010) findings are rather problematic when it comes to the overuse of the in indefinite contexts. According to Van Hout et al. (2010), because of their unranked constraints children overuse the definite article the in indefinite contexts. Here, Van Hout et al. (2010) do not make a distinction between indefinite referential and non-referential
contexts. Therefore, according to their theory, the overuse should be found in all indefinite contexts. However, as their experiments include only indefinite referential contexts, we do not find out if their theory proves to be true in indefinite non-referential contexts as well. As opposed to Van Hout et al.’s (2010) explanation that would predict the overuse of the in all indefinite contexts, Schaeffer & Matthewson (2005) predict and find the overgeneralization of the only in indefinite referential contexts, which they explain by the lack of the CNSA. For that reason, I chose to base my predictions regarding the article choice in production on Schaeffer & Matthewson’s (2005), rather than Van Hout et al.’s (2010) theory.
Additionally, in comprehension, Van Hout et al. (2010) find that children interpret the indefinite article a as referring to both unique, mentioned referents and to not mentioned, new referents. The reason for this, as they state, is children’s inability to derive scalar implicatures.
Scalar implicature is a paradigmatic case of conversational or Gricean implicature, which is defined as an inference that consists of attributing an implicit meaning that goes beyond the explicit linguistic, logical meaning of an utterance to a speaker (Noveck 2001, p. 165). As Noveck (2001) states, according to Grice, a hearer searches for implicatures when one of the following submaxims of his Maxim of Quantity is violated:
i. Make your contribution as informative as is required.
ii. Do not make your contribution more informative than is required.
As Chevallier et al. (2010) explain, “scalar implicatures occur when the speaker’s choice of a less informative term is taken to imply that the more informative term is not applicable” (p. 1105). Together, less informative terms and more informative terms form scales and are therefore considered ‘scalar’ terms. As reported by Chevallier et al. (2010), according to Horn (1972), “a scale is a set of alternate linguistic expressions ranked by order of informativeness” (e.g. <might, must>, <some, all>) (p. 1105). For example, in case of the scale <some, all>, All is the stronger term and entails the weaker, Some, but the weaker term Some does not entail All. If the speaker decided to use the word Some, logically meaning at least one, and not a more informative item from the scale, such as All, it means that they chose not to do so because they know that All is not applicable. Therefore, the hearer has to infer that by saying Some the speaker also means Not All. For instance, the statement Some men like cars does not necessarily imply that All men like cars, but it does imply that At least
one man likes cars and that Not all men like cars. However, the statement All men like cars logically implies that Some men like cars (Noveck 2001, p.168).
If this were applied to article choice in Dutch, the scale would consist of the definite articles de/het and the indefinite article een, as follows: <de/het, een>. The definite articles de and het would be the stronger terms since they are more informative and stand for a determined referent, as in het feestje in (7), where it is obvious that the speaker is talking about the party that the hearer is also familiar with. Accordingly de/het entail the indefinite article een because it stands for an undetermined (referential or non-referential) referent. For example, het feestje in (7) is actually een feestje that has been determined at one point. So, while het feestje entails een feestje, een feestje does not necessarily imply het feestje, as it is not determined. Moreover, when we hear the noun phrase het feestje we infer that the speaker also means that it is not een feestje (an undetermined referent), since een feestje has already obviously been determined and speaker refers to that determined referent.
(7) A: Ik ben jarig volgende week en ik ga een feestje geven. Kom je ook?
A: ‘It’s my birthday next week and I’m throwing a party. Would you like to come?’ B: Ja, wat leuk! Kunnen mijn vrienden ook mee naar het feestje?
B: ‘Yes, great! Can my friends come to the party too?’
Therefore, to interpret the articles correctly, children have to acquire the computation of scalar implicatures. In their study, Van Hout et al. (2010) show that children younger than 5 interpret the indefinite article a quite liberally, as referring to both new and old referents and proving that they still cannot compute scalar implicatures. Since children are pragmatically immature and cannot compute scalar implicatures, they cannot disambiguate between the two articles. Upon hearing a, they do not know that by saying a, the speaker also means not the, i.e. not a determined referent and therefore interpret the indefinite article a as referring to both a determined and non-determined referent at chance level.
In their studies, Noveck (2001) and Pouscoulous et al. (2007) show that the routine of carrying out scalar implicatures comes with age. According to Noveck (2001), before children start interpreting weaker scalar terms, such as Some, pragmatically, they do it based on the semantic or as Noveck (2001) states, explicit, logical meaning. In his research, Noveck (2001) shows that even 10-year-olds can still have trouble and are not adult-like in computing
implicatures, possibly because their cognitive resources are not yet as available to them as they are to adults.
In their research, along with similar results to Noveck’s (2001) study, in one of their experiments, Pouscoulous et al. (2007) also find that by facilitating the task and making it relatively free of distractors, 4-year-olds, 5-year-olds and 7-year-olds are generally competent in computing scalar implicatures. Moreover, 7-year-olds performed almost adult-like with only 17% of the children who did poorly. Additionally, they found that 9- to 10-year-olds are much more likely to treat the French term certains semantically or logically (42%) than quelques (0%), both meaning some, which indicates that the computation of implicatures can depend on the scalar term that is being used. As opposed to quelques, which is a simple existential, certains is a partitive implying that something is also a part of a “contextually salient set” and is in that way more complex and consequently more difficult to process (p. 360). Pouscoulous et al. (2007) conclude that “the added processing cost of certains makes the task harder, thus reducing children’s rate of implicature production” (p. 371).
To sum up, Schaeffer & Matthewson’s (2005) and Van Hout et al.’s (2010) findings respectively show that due to their pragmatic immaturity, children younger than 5 overgenerate the definite article the in indefinite referential contexts in production due to the lack of the CNSA, and that in comprehension, because of the inability to compute scalar implicatures, they quite liberally interpret the indefinite article a as referring to both a determined and non-determined referent.
2.3 Subject-verb agreement in Dutch
As opposed to the article choice, agreement is purely a grammatical phenomenon. As Vigliocco, Butterworth and Semenza (2004) state, “agreement consists in copying features (Person, Number and Gender) from a Source or Controller (the Subject) to a Target (the Verb)” (p. 188). That agreement is purely a grammatical phenomenon is supported by the fact that, in English, broken agreements seem to be determined only by the syntactic properties of head and local nouns (Vigliocco, Butterworth and Semenza 2004, p. 188). Thus, “once a lexical head has been selected as ‘Subject’ and its number has been determined, no further semantic information about the head will influence agreement computation” (Vigliocco, Butterworth and Semenza 2004, p. 190).
In Dutch, “finite verbs encode the categories TENSE, NUMBER and PERSON” and “the expression NUMBER and PERSON agrees with the subject of the clause in which finite inflection appears” (Polišenská 2010, p. 44). As Polišenská (2010) explains, “TENSE is usually not considered part of agreement inflection because the choice of the correct tense inflection is not determined by the syntactic structure in which it appears” (p. 44).
In (8), a verbal paradigm in the present tense is illustrated and it includes a regular lexical verb werken “to work”. Three different suffixes are used: -ø for 1SG and 2SG in inversion, -t for 2SG and 3SG and –en for all plural forms, irrespective of person.
(8) Conjugation of Dutch verb werken (to work) in present indicative
Person Singular Plural
1 Ik werk-ø I work Wij werk-en We work 2 Jij/Je werk-t U werk-t You work Jullie werk-en You work 3 Hij/zij werk-t He/she works Zij/Ze werk-en They work
Several studies (De Haan 1996; Polišenská 2010) have been carried out on the acquisition process of the subject-verb agreement in Dutch. After analyzing the elicited production data from Dutch monolingual children between ages three and six and finding high percentages of accuracy in both existing and nonce verbs, Polišenská (2010) concludes that finite verbal inflection is acquired by the age of three (p. 100). In Polišenská’s (2010) words, “Dutch three-year-olds have target-like knowledge of the verbal inflectional paradigm and […] they understand that subject-verb agreement is obligatory” (p. 101).
3. Hypotheses and predictions
Various theories have been proposed and many studies have been carried out on the acquisition of articles and definiteness, about the ASD and about the relation between ASD
and language, some of which I have mentioned above. From these follow my hypotheses, formulated in (9-13).
(9) Dutch children with autism are impaired in pragmatics, but not in grammar. (10) Pragmatics is necessary for the acquisition of correct article choice. (11) Grammar is necessary for the acquisition of subject-verb agreement.
(12) False-belief is part of Theory of Mind (ToM).
(13) The Concept of Non-Shared Assumptions (CNSA) is related to Theory of Mind (ToM).
More precisely, if Dutch children with ASD are impaired in pragmatics, (a) in production, because of the lack of the CNSA, they will experience difficulties in the choice of articles, overproducing Dutch definite articles de and het in indefinite referential contexts, and (b) in comprehension, because of the inability to derive scalar implicatures, they will interpret indefinite noun phrases as referring to both new and old referents.
Furthermore, if Dutch children with ASD are not impaired in grammar, as hypothesized, (c) they will have no trouble in the subject-verb agreement, showing a target-like performance. Finally, if the CNSA is related to ToM, (d) I predict that Dutch children with ASD will not pass or will do badly on a false-belief task.
Conversely, if my prediction in (a) turns out not to be true, my hypothesis that Dutch children with ASD are impaired in pragmatics could be rejected, suggesting that Dutch children with autism do posses the CNSA and show no impairment in that particular aspect of pragmatics. However, this could also mean, even though unlikely, that pragmatics is not necessary for the acquisition of correct article choice.
If my prediction in (b) turns out not to be true, it could suggest that pragmatics is not necessary for the acquisition of correct article choice or that children with ASD are capable of deriving scalar implicatures, thus showing no impairment in pragmatics in that aspect. If my prediction in (c) is not borne out, the results would suggest either that children with ASD are impaired in grammar or that, although unlikely, grammar is not necessary for the acquisition of subject-verb agreement.
Furthermore, if my prediction in (d) turns out not to be true, the data would imply either that the CNSA is not related to ToM, or that false-belief is not part of ToM.
However, if my prediction in (a), (b) and (c) are not borne out, the results might suggest that children with ASD have specific language impairment, as the impairment would seem to be concentrated strictly in grammar and not pragmatics.
4. Methods 4.1 Participants
To test the hypotheses 27 Dutch TD children and 27 fluent monolingual children with ASD were tested on experimental Definiteness production task, Definiteness judgement task, Subject-verb production task and False-belief task. Participants come from different parts of the Netherlands, where the data was collected. The children were from 5 to 14 years old. TD children were individually matched to the children with ASD on age and gender. All children with ASD are high-functioning Dutch speaking children, who were diagnosed with a disorder in the autism spectrum, such as autistic disorder, Asperger syndrome and pervasive developmental disorder not otherwise specified (PDD-NOS). More detailed information about the participants is provided in Table 1.
Group Age (years) Mean age Number of
females
Number of
males Total number
TD 5 – 14 9;9 5 22 27
ASD 6 – 14 9;9 5 22 27
Table 1: Participants
4.2 Definiteness production task
Inspired by Schaeffer & Matthewson’s (2005) study, the Definiteness production task was developed by Schaeffer, Witteloostuijn & De Haan (2014). In the Definiteness production task, a PowerPoint presentation with pictures and video clips was used to elicit definite and indefinite articles. First, children were introduced to a set of animated characters (Dora the Explorer, Miffy, Spongebob, Jip en Janneke…). They saw a short video clip or a picture where a certain character is carrying out an action. Two experimenters were present. The children were told that since the second experimenter was sitting behind the computer,
s/he could not see the screen, but that s/he also wanted to know what was going on. The second experimenter therefore asked the children whom they saw in the picture or video and what the characters were doing, what they had done or what they were going to do. The children then had to answer in full sentences starting with a subject. Moreover, since the second experimenter often did not listen carefully, the children could correct him/her when s/he was wrong. As a filler task, a scrambling production task was used. The children were tested on three conditions: Definite, Indefinite Referential Past, and Indefinite Non-referential Future, as in (14-16). For every condition six items were used, amounting in altogether to 18 items, as presented in Table 2.
Condition Number of items
Definite 6
Indefinite Referential Past 6 Indefinite Non-referential Future 6
Total number of items 18
Table 2: Experimental conditions in Definiteness production task
The enlisted conditions were chosen to test the children on definiteness production in various environments, including definite, referential and non-referential indefinite and both past and future scenarios. In the Definite condition, as presented in (14), the character in the picture or video carried out an action with a previously introduced item and the child had to tell what the character did.
(14) Definite condition
Experimenteerder 2: Hé, wie zie je op het plaatje? Experimenter 2: Hey, whom do you see in the picture? Kind: *Naam*!
Child: *Name*!
Experimenteerder 2: En wat nog meer? Experimenter 2: And what else?
Kind: Een tractor! Child: A tractor!
(*Name* is pushing the tractor)
Experimenteerder 2: Wat deed *naam* daarnet? Experimenter 2: What did *name* just do? Kind: Zij duwde de/het tractor.
Child: She pushed the tractor. * Zij duwde een tractor. * She pushed a tractor.
In the Indefinite Referential Past condition, as in (15), a picture of a character and the result of its action was presented to the child. The child then had to tell what the character did, based on what s/he saw in the picture.
(15) Indefinite referential past
Experimenteerder 2: Hé, wie zie je op het plaatje? Experimenter 2: Hey, whom do you see in the picture? Kind: Jip!
Child: Jip!
Experimenteerder 2: En wat heeft Jip net gedaan? Experimenter 2: What did Jip just do?
Kind: Hij heeft een huis geschilderd. Child: He drew a house.
* Hij heeft het/de huis geschilderd. * He drew the house.
Finally, in the Indefinite Non-referential Future condition, future scenarios, which had to be described by the child, involved an action that will be or is not yet completely carried out by the depicted character, as presented in (16).
(16) Indefinite non-referential future
Experimenteerder 2: Hé, wie zie je op het plaatje? Experimenter 2: Hey, whom do you see in the picture? Kind: Bert!
Experimenteerder 2: Bert zegt: Oh, ik verveel me zo, ik weet niet wat ik moet doen. Oh, weet je wat, ik ga naar de bakker, en dan ga ik daar iets kopen!
Experimenter 2: Bert says: “Oh, I’m bored, I don’t know what to do. Oh, I know! I will go to the bakery and then I’m going to buy something there!”
Experimenteerder 2: Wat denk je dat Bert gaat doen? Experimenter 2: What do you think Bert is going to do? Kind: Hij gaat een brood(je) kopen!
Child: He is going to buy a sandwich. * Hij gaat het/de brood(je) kopen! * He is going to buy the sandwich.
The Definiteness production task was scored as follows: indefinite articles were scored with 0, definite articles with 1, irrelevant answers with 2, such as in (17), één(meaning stressed “one” or one as in “one of”) with 3, like in (18), the change of condition with 4, as in (19), and the omission of articles with 5. Finally, the scores were calculated, put into percentages and used for statistical analysis.
(17) (In the Definite condition)
Kapitein zat te zoenen met die plastic pop. Captain sat to kiss with that plastic doll. ‘The captain was kissing that plastic doll.’ (18) (In the Indefinite Referential Past condition) Nijntje maakt één tart.
Miffy makes one cake. ‘Miffy is baking one cake.’
(19) (In the Indefinite Non-referential Future condition) Pino tekent nog een vogel.
Pino draws again a bird. ‘Pino is drawing another bird.’
4.3 Definiteness judgement task
Also inspired by Schaeffer & Matthewson’s (2005) study, the Definiteness judgement task was developed by Schaeffer, Witteloostuijn & De Haan (2014) as well. The task was set up as a felicity judgement task. The participants saw a PowerPoint presentation with pictures accompanied by a short story. Sometimes, there was a mistake in the last sentence of the story regarding the choice of the article (definite vs. indefinite) and participants were asked if it sounded good or not. While practicing, if a child accepted the items with mistakes as “sounding good”, the experimenter was allowed to play it again and ask the child to judge it one more time. If the child still did not recognize the mistake, the experimenter was allowed to present the child with the alternative correct form and ask the child if that sounded better. Then, if the child still did not recognize the mistake, the experimenter was allowed to hint the child, telling them that they had to pay attention to de/een. However, while performing the task, experimenter was not allowed to give feedback.
Two lists of items were used, list 1 first followed by list 2. Each list consists of four items used for practice and 15 items used for testing, five items per condition, as presented in Table 3.
Condition Number of items
Definite 10
Indefinite Referential Past 10 Indefinite Non-referential Future 10
Total number of items 30
Table 3: Experimental conditions in Definiteness judgement task
The children were tested on three conditions: Definite, Indefinite Referential Past and Indefinite Non-referential Future, as in (20-22).
(20) Definite condition
Liesbeth is in de keuken aan het bedenken wat ze mee moet nemen naar school. Ze vraagt aan haar tante, die in de huiskamer is: Ligt daar nog iets op tafel?
Haar tante antwoordt: Ja, er ligt hier een telefoon en een agenda. Liesbeth vraagt: Mag ik de telefoon?
Liesbeth is in the kitchen thinking what she has to take to school. She asks her aunt, who is in the living room: “Is there anything else on the table?”
Her aunt answers: “Yes, there is a phone and a planner.” Liesbeth asks: “Could you give me the phone?” (21) Indefinite Referential Past condition
Suzanne is in de woonkamer en ruikt iets lekkers. Ze vraagt aan haar oma, die in de keuken is, wat ze aan het doen is.
Oma antwoordt: Kom maar kijken. *Ik heb de appeltaart gemaakt.
Suzanne is in the living room and smells something delicious. She asks her grandma, who is in the kitchen, what she is doing.
Grandma answers: “Come here. *I made the apple pie.” (22) Indefinite Non-referential Future condition
Kirsten komt thuis uit school en vraagt aan haar vader, die in de tuin is, waar hij mee bezig is.
Haar vader antwoordt: Ik ga een boomhut bouwen.
Kirsten comes home from school and asks her father, who is in the garden, what he is doing. Her father answers: “I am going to build a tree house.”
The Definiteness judgement task was scored as follows: incorrect rejection was scored with 01, incorrect acceptance with 02, correct rejection with 11, correct acceptance with 12, irrelevant answers with 2, as in (23), answers such as “the left one/the right one” or één (meaning stressed “one” or one as in “one of”) with 3, as in (24), and Dutch demonstrative pronouns die/dat with 5, as in (25). Finally, the scores were calculated, put into percentages and used for statistical analysis.
(23) (Upon hearing a dialog in the Definite condition where Liesbeth’s aunt tells her niece that there is a phone and a planner on the table and Liesbeth asks for the phone, which she wants to bring to school)
Kind: Fout want je neemt geen telefoon mee naar school. Ik zou eerder agenda naar school meenemen.
Child: Wrong, because you take no phone with to school. I would rather planner to school with take.
Child: ‘Wrong, because you don’t take phones to school. I would rather take a planner.’
(24) (In the Indefinite non-referential future condition) Ik ga één boomhut bouwen.
I go one tree house build.
‘I am going to build one tree house.’
(25) (In the Definite condition) Mag ik dat vliegtuig? May I that airplane?
‘Could you give me that airplane?’
4.4 Subject-verb agreement production task
The Subject-verb agreement production task, developed by Duinmeijer (in progress), was used to investigate whether the children with ASD are indeed not impaired in grammar, but only in pragmatics. Here, subject-verb agreement was elicited for different contexts: 1SG, 2SG, 3SG, 1PL and 2PL. There were seven cards with examples where all verbs were depicted and twelve cards that were used for practicing the task with the verb aaien (to pet). The task was carried out with a doll (P), the researcher (O) and the child (K). Everybody received a separate pile of 30 cards that were turned over at the same time by the researcher and the child. The child was asked to tell what everybody is doing, which was depicted on the cards. Sometimes the pictures were the same so if the child and the doll had the same card and the researcher a different one, the child should tell what s/he and the doll (1PL) are doing compared to what researcher was doing (2SG), as in (26).
(26) Kind: Wij kammen een kat. Jij leest een boek. Child: We pet a cat. You read a book.
Child: ‘We are petting a cat. You are reading a book.’
4.5 False-belief task
The False-belief task is based on Colle, Baron-Cohen & Hill (2007) and it is made of six conditions: a Pre-test, Visible Displacement, Invisible Displacement and Ignore Communicator, which are part of a screening test, a False-belief condition and a Control condition, as described below and presented in more detail in Table 4. The material used consisted of two non-transparent boxes, an eraser and a cardboard screen. There were two experimenters that carried out the task, one being the hider and the other communicator. The hider hid the eraser in one of the boxes and the child was told that s/he had to find it. The child was also told that the communicator would try to help him/her, but that he is not always right. The communicator saw where the eraser was hidden, but the child did not, so he indicated to the child which container he saw the eraser being hidden in.
Condition Number of trials
Pre-test 3 Visible displacement 3 Invisible displacement 3 Ignore communicator 3 False-belief 3 Control 3
Total number of trials 18
Table 4: Experimental conditions in the False-belief task
In the Pre-test, the eraser was hidden in the presence of both communicator and the hider, because the aim was to demonstrate that the communicator was trying to help the child. The hider put the eraser in one of the boxes without the screen being present, so both the communicator and the child could see the entire procedure. The hider then asked the communicator where the eraser was and the communicator pointed to the correct box, which was also seen by the child. The hider then asked the child where the eraser was. The Pre-test ended when the child pointed correctly to the box with the eraser in all three successive trials. In the screening test, the child was tested on three conditions: (1) on Visible Displacement, the ability to follow the eraser as it is moved from one box to another; (2) on Invisible Displacement, the ability to follow the eraser when the box containing it was moved from
one position to another; and (3) on the ability to ignore the sign made by the communicator when this was clearly false.
(1) In the Visible Displacement condition, the communicator left the room and the hider opened the box with the eraser and moved it to the other box within view of the participant. When the communicator came back the child was asked where the eraser was.
(2) The Invisible Displacement condition is the same as the visible condition, except that while the communicator was absent, not only was the eraser moved, but the box containing it was moved as well.
(3) In the Ignore Communicator condition, the hider hid the eraser in the view of communicator, who then left the room. The hider then opened the boxes, showed them to the child and moved the eraser form the first to the second box. When the communicator came back, he was asked where the eraser was. Since he did not know that the eraser had been moved, he pointed to the wrong box. Afterwards, the child was asked to find the eraser and in order to do that s/he had to ignore the answer given by the communicator, as the communicator did not see that the eraser had been replaced.
In the False-belief condition, the hider hid the eraser in front of the communicator but out of the view of the participant, by putting a screen in between the experimenter and the participant. After that, the communicator left the room, the hider took away the screen and switched the closed boxes in view of the participant. When the communicator came back, s/he was asked where the eraser was. The communicator pointed at the box in which s/he had previously seen the eraser. To pass the test, the participant had to point to the other box. Finally, to control that the child does not assume that in the False-belief condition, when the communicator goes away, his indication is always wrong, the Control condition was carried out. In the Control condition, there was no switch of the eraser when the communicator left the room, so when he returned, the communicator pointed to the correct box.
The participants passed the test if they were successful on three successive trials in all parts of every condition.
5. Results and discussion
To test the hypotheses that Dutch children with autism are impaired in pragmatics, but not in grammar, that pragmatics is necessary for the acquisition of correct article choice and that grammar is necessary for the acquisition of subject-verb agreement, 27 TD children and 27 children with ASD were tested on the experimental Definiteness production task, Definiteness judgement task, Subject-verb production task and False-belief task, the results of which are presented in the sections 5.1, 5.2, 5.3 and 5.4.
5.1 Results Definiteness production task
In the Definiteness production task, the children were tested on three conditions: Definite, Indefinite Referential Past and Indefinite Non-referential Future condition. Lack of the CNSA, due to the impairment in pragmatics in children with ASD, predicts the overgeneration of the definite articles de and/or het in the Indefinite Referential Past condition in production. However, the results do not support this prediction since almost no definite articles were used in the indefinite conditions.
Figure 1: Definiteness production – Indefinite Referential Past condition
As seen from Figure 1, none of the children produced the definite articles de and/or het in the Indefinite Referential Past condition. Children in the ASD group performed correctly, producing the indefinite article een 96.91% and the TD children performed correctly as well, producing the indefinite article 98.76%. It can also be seen that the children with ASD
0 20 40 60 80 100 120
Indefinite Referential Past ASD Indefinite Referential Past TD
Indefinite Definite Irrelevant
produced some irrelevant answers (the answers scored with 2, 3, 4 and 5), such as the one repeated in (27), 3.09%, while the TD children did so 1.24%.
(27) (In the Definite condition)
Kapitein zat te zoenen met die plastic pop. Captain sat to kiss with that plastic doll. ‘The captain was kissing that plastic doll.’
I predicted that if the children with ASD are impaired in pragmatics, that they will also lack the CNSA. Consequently, because of their lack of the CNSA the children with ASD will attribute their own assumptions to the hearer and will not make a distinction between environment B (believed by speaker only) and environment A (believed by speaker and hearer). For that reason, the children with ASD were predicted to overgenerate the definite articles de/het to (adult) B contexts, which are believed by speaker only and require the indefinite article een. However, the results do not support my prediction that the children with ASD will overproduce the definite articles de/het in the Indefinite Referential Past condition. Therefore, it can be concluded that the children with ASD do not lack the CNSA. Even though the conclusion that pragmatics is not necessary for the acquisition of articles is also possible, I find it unlikely since many previous studies (Schaeffer & Matthewson 2005; Rozendaal and Baker 2008; Van Hout et al. 2008) show otherwise. For that reason, it is more likely that the children with ASD are simply not impaired in the aspect of pragmatics concerning the CNSA.
Figure 2: Definiteness production – Indefinite Non-referential Future condition
0 10 20 30 40 50 60 70 80 90 100
Indefinite Non-referential Future ASD Indefinite Non-referential Future TD
Indefinite Definite Irrelevant
Lack of the CNSA in the children with ASD predicts the overuse of the definite articles only in the Indefinite Referential condition, so no overuse is expected in the Indefinite Non-referential Future condition. The fact that no overuse was found in the Indefinite Referential Past condition shows that the children with ASD do posses the CNSA, which again implies that there would be no overuse of the definite articles in the Indefinite Non-referential Future condition. From Figure 2, it can be seen that it is indeed so. The children with ASD seem to have no problems in producing the correct article and produce the indefinite article een 84.57% in Indefinite Non-referential Future condition. The children in TD group performed correctly as well, producing een 88.27%. The children did produce a few definite articles, but only 1.23% in the ASD and 0.62% in the TD group. However, as it is noticeable from Figure 2, the percentages of irrelevant answers in the Indefinite Non-referential condition are slightly higher, amounting to 14.2% in the ASD group and 11.11% in the TD group. This result can be attributed to the fact that the questions in the Indefinite Non-referential condition are more open-ended, so they leave more space for variation, and since it is a non-referential condition, it is also cognitively more difficult.
To sum up, against my predictions, children’s performance in the indefinite conditions, especially in the Indefinite Referential Past condition, indicates that children with ASD do posses the CNSA and seem not to be impaired in that aspect of pragmatics.
While no overuse of the definite articles de/het was found in the indefinite conditions, there was some overuse of the indefinite article een in the Definite condition.
Figure 3: Definiteness production – Definite condition
0 10 20 30 40 50 60 70 80 90 100
Definite ASD Definite TD
Indefinite Definite Irrelevant
As it can be observed from Figure 3, in the Definite condition, for which the target were de/het, the ASD group produced the definite articles 85.19% and the TD group 95.06%. However, it is also noticeable that children in both groups produced the indefinite article een, amounting to 10.49% in the ASD group and 3.7% in the TD group. To determine if the difference between the groups is significant, the Mann-Whitney U test was run. The test revealed that the difference between ASD and TD group in the number of incorrectly produced indefinite articles in the Definite condition is not significant (U=280, Z=-1.90, p=.057). However, the results indicate a trend since the p-value is barely over .05.
Upon looking at their distribution in the Definite condition, it was clear that not all the children produced the same amount of indefinite articles. Therefore, I decided to divide the children into two subgroups, failers and passers, failers being the children who produced at least one indefinite article in the Definite condition. The ASD and TD passers and failers subgroups are presented in Table 5.
Group Pass Fail
ASD 17 10
TD 23 4
Table 5: Passers and failers subgroups – Definiteness production
Figure 4: Failers subgroup ASD – ages and mistakes
As seen from Figure 4, the ASD failers subgroup consists of two children at the ages of 6 and 9 that produced an indefinite article in Definite condition in three out of six items, three
0 0.5 1 1.5 2 2.5 3 3.5 6 6 9 9 9 10 10 11 11 11
ASD
ASDchildren at the ages of 6, 9 and 11, who did so in two items and five children, one child at the age of 9, two at the age of 10, and two 11, who produced only one indefinite in the definite condition. From Figure 4, it is noticeable that younger children made more mistakes than older children in the ASD failers subgroup. This is confirmed by the fact that the mean age of children in the ASD failers subgroup that produced more than one indefinite article in the Definite condition is 8;2, whereas the mean age of the children who produced only one indefinite article in the Definite condition is 10;2. Furthermore, the mean age of the passers subgroup of the children with ASD, who made no mistakes in the Definite condition, is 10;5, which implies that there could be development by age.
Figure 5: Failers subgroup TD – ages and mistakes
As presented in Figure 5, the failers subgroup in TD group consists of only one child at the age of 9 that produced an indefinite article in three out of six items in Definite condition and three children at the ages of 8, 11 and 13, who did so in only one item. Again, it is visible from Figure 5 that the older children performed better than the younger children. This is supported by the fact that the mean age of the child who made more than one mistake in the Definite condition is 9, whereas the mean age of the children who produced only one indefinite article in the Definite condition is 10;7. Moreover, the mean age of the passers subgroup of the TD children, who made no mistakes in the Definite condition, is 13;3, which again may suggest development by age.
0 0.5 1 1.5 2 2.5 3 3.5 8 9 11 13
TD
TDFigure 6: Definite condition – ASD failers subgroup and their TD matches
In Figure 6, the percentages of correct and incorrect article choice for the ASD failers subgroup and their TD matches are presented. While the ASD failers subgroup produced indefinite articles 28.33%, their TD matches did so only 6.67%. In Figure 6, it is also noticeable that the TD matches produced more definite articles (91.67%) than the ASD failers subgroup (66.67%). To see whether the differences between the ASD failers subgroup and their TD matches are significant, the results were computed using the Mann-Whitney U test. The test showed that both the difference in the production of definites (U=13.50, Z=-2.89, p=.004) as well as the difference in the production of indefinites (U=11.50, Z=-3.06, p=.002) in the Definite condition are significant.
The overgeneration of indefinite articles in the Definite condition was not predicted for the children with ASD. However, the data show that there is a subgroup of children with ASD that overuse een in the Definite condition. Even though the overuse was found in production, the result could be related to Van Hout et al.’s (2010) explanation concerning the overly liberal interpretation of the indefinite article a in comprehension.
Van Hout et al. (2010) found that TD children younger than 5 accept both old and new referents in their interpretation of indefinites, possible reason for that being their pragmatic immaturity and inability to derive scalar implicatures. When children cannot compute scalar implicatures, upon hearing a, they do not know that by using a the speaker also means not the and are satisfied with the semantic meaning of a (an undetermined referential or non-referential referent), which is also entailed by the. If the children do not know that by a the
0 10 20 30 40 50 60 70 80 90 100 De/inite ASD De/inite TD Inde/inite De/inite Irrelevant
speaker also means not the, they can then interpret a as referring to both a determined and undetermined referent.
The computation of scalar implicatures may not only be necessary for the correct interpretation of article choice, but also for the article choice in production. To use the indefinite article een, the speaker has to know that by doing so, he is implying not de/het, because keeping the hearer in mind, the speaker is making his contribution as informative as required and is avoiding making it more informative than is required by the hearer. However, if the speaker does not know that een also means not de/het and is satisfied by the semantic meaning of een (an undetermined referential or non-referential referent), which is also entailed by de/het, then he will not only use een when referring to an undetermined referent, but also when referring to a determined referent. Consequently, if the speaker chooses to use een to refer to a determined referent, not knowing that he is being less informative than required by the hearer, he will be breaking the Maxim of Quantity. So, for the correct article choice in production, the speaker has to make use of pragmatics and scalar implicatures to know that by saying een he is also implying not de/het in order to convey to the hearer that he is doing so for a reason, namely that he is being as informative and not more informative as required.
The results presented in Figure 6 show that the failers subgroup of the children with ASD overuses the indefinite article een in the Definite condition of the Definiteness production task, which suggests that they cannot compute scalar implicatures and do not know that een also implies not de/het. Additionally, the results presented in Figure 4 show that the younger children in the ASD failers subgroup performed worse than the older children. Moreover, the children in the passers subgroup of the children with ASD are on the average older than the children who do better in the ASD failers subgroup. Taken together, these results suggest development by age, which also seems to be the case in the TD group. When it comes to the TD group, the results are in line with Noveck’ (2001) results that show how even some 10-year-olds still cannot carry out scalar implicatures. Therefore, the reason why the ASD failers subgroup overuses the indefinite article een in the Definite condition of the production task could be pragmatic immaturity and inability to derive scalar implicatures.
Additionally, it seems that computing scalar implicatures for articles is not just pragmatically difficult, but may require more cognitive effort too. Pouscoulous et al.’s (2007) study showed that processing plays a role in computing scalar implicatures, bridging the gap between
semantic and pragmatic meaning of scalar terms. Deriving scalar implicatures for articles seems to be a very subtle and delicate process, because the meanings that articles carry are in itself subtle. Consequently, the process of bridging the gap between semantic and pragmatic meaning of articles may be slightly more difficult even for high-functioning children with ASD.
The TD children and the children with ASD that participated in this research also took part in the Digit Span task. The Digit Span task is a part of the Wechsler Intelligence Scale for Children (WISC) and it tests short-term memory and working memory. It is made of two parts: digit forward and digit backward. The digit span backward test tests the ability to manipulate verbal information and engages working memory. To do the digit span backward test, children listen to a sequence of numbers (e.g. 8, 5, 4) and then they have to repeat the numbers in the reverse order (4, 5, 8). If a child gives a correct response, the length of the sequence increases. Each correct response equals one point and the maximum number of points in one part is 14. The results of the Digit Span backward test are presented in Figure 7.
Figure 7: Digit Span backward
As visible from Figure 7, on average achieving 5.93 points out of 14 on Digit Span backward task and on average being able to repeat 4.52 digits, the TD children outperformed the children with ASD, who achieved a mean score of 4.41 points and on average repeated 3.59 digits in a series. Since the difference between the groups in the mean score (U=217.5, p=.01), and the number of repeated digits is significant (U=207, p=.005), it could be concluded that in terms of working memory, TD children are at a slight advantage compared to the children with ASD. These results suggest that processing might be harder for the children with ASD, which may be the reason why it is more difficult for them to automatize
0 1 2 3 4 5 6 7
Backward correct Backward digits
ASD TD
To sum up, the results show that in the Definite condition of the Definiteness production task, there is a subgroup of children with ASD that overuse the indefinite article een because they are unable to derive scalar implicatures. The reason for that seems to be their pragmatic immaturity. Additionally, the results of the Digit Span also indicate that it could be cognitively more difficult for the children with ASD to derive scalar implicatures, because their working memory is weaker than the working memory of TD children, which, as a consequence, makes it harder for them to make the process of computing scalar implicatures automatic.
5.2 Results False-belief task
If the CNSA is related to ToM, I predicted that since the children with ASD due to their impairment in pragmatics would lack the CNSA, that they would also not be successful on the False-belief task.
As presented in Figure 8, both the TD children and the children with ASD passed the Pre-test and screening with 100% and 97.53%, respectively. The TD group was better than the ASD group in the Control condition, performing at 91.36%, as opposed to the ASD group, which performed at 86.42%. However, in the False-belief condition, the ASD group performed better than the TD group, scoring 75.31%, whereas the TD children scored 61.73%. Subsequently, the Mann-Whitney U test was run, which showed that the children with ASD even performed significantly better than the TD children on the False-belief condition (U=258, Z=-2.02, p=.044).
Figure 8: False-belief task
0 20 40 60 80 100 120 ASD TD Pre‐test Screening FB Control FB
Thus, my prediction that children with ASD would perform badly on False-belief was not borne out. Recall, however, that my prediction that children with ASD would overgenerate de/het in the Indefinite Referential Past condition was not borne out either, which indicated that children with ASD posses the CNSA. The fact that the children with ASD performed so well on the False-belief condition suggests that the children have ToM and know that other people’s beliefs, desires and intentions are not necessarily the same as theirs, which only supports the result of the Definiteness production task in the Indefinite Referential Past condition, because to have the CNSA means to know that the speaker and hearer assumptions are always independent.
Among eleven children in the ASD group that made no mistakes in the False-belief condition, there are ten children who did not make mistakes in the indefinite conditions of the Definiteness production task. This supports the hypothesis that there is a relation between Theory of Mind and the CNSA.
Even though the result is surprising, because the children with ASD performed better than the TD children, it does support the results of the Definiteness production task, where there was no overuse of the definite articles in the indefinite conditions in the ASD group, contrary to my prediction. The results of the Definiteness production task indicated that children with ASD do posses the CNSA, which is only confirmed by the results of the False-belief condition, as the children did so well.
5.3 Results Definiteness judgement task
In the Definiteness judgement task, the children were tested on three conditions: Definite, Indefinite Referential Past and Indefinite Non-referential Future condition. Due to their impairment in pragmatics, children with autism are assumed not to be able to compute scalar implicatures. Based on that, I predicted that in the comprehension of definiteness they would interpret the indefinite article een as referring to both new and old referents at chance level.
