Reality versus thoughts: the understanding of Theory of Mind and mental language in TD adolescents and adolescents with SLI or ASD

(1)

Reality versus thoughts: the understanding of

Theory of Mind and mental language in TD

adolescents and adolescents with SLI or ASD

Research Master Thesis Linguistics

Language & Cognition

Liset Rouweler (2017733)

April 2016

(2)

Reality versus thoughts: the understanding of

Theory of Mind and mental language in TD

adolescents and adolescents with SLI or ASD

(3)

ACKNOWLEDGMENTS

I am very grateful to my thesis supervisors Dr. Wim Tops and Prof. Petra Hendriks for their great input and valuable feedback. Through their critical look, they supported me in my search for the right ideas, questions and solutions.

I would also like to thank Dr. Bart Hollebrandse and Jessica Overweg for providing me with the bake sale stories task, Dr. Hannah de Mulder and Ellen van Boggelen - Heutink for the memory for complements task and Dr. George Spanoudis for the mental state verb factive and non-factive task. Moreover, I would like to thank the Dr. J. de Graafschool, het Zernike College and De Mast, and more specifically their students, for their great participation in my study.

A special thank you to my dear friends Bob Holland and Zheng Ng for reading my thesis and for their support during the process of writing. They both provided me with very useful feedback, which encouraged me to keep writing.

From the day I started my new life in Groningen as a fresh, 18 year old student linguistics, my parents provided me with continuous support and love. During the 5,5 years of being a student linguistics, and especially during the process of writing this research master thesis, their endless believe in me has never changed. Now, April 2016, it is time to finally graduate and to be an adult. This accomplishment would not have been possible without you. Thank you.

(6)

ABSTRACT

Comprehension of mental state verbs and sentential complements are argued to relate closely to Theory of Mind (ToM) both in typically developing children (TD), and in children with specific language impairment (SLI) or autism spectrum disorder (ASD) (Astington & Baird, 2005; Miller, 2004; Ziatas et al., 1998). However, up until now it is unknown whether both factors still show a relationship with ToM after childhood, more specifically during adolescence (13 until 18 years). Furthermore, literature shows that there are clear differences between TD children and children with ASD or SLI on ToM understanding. In this case, children with ASD or SLI perform significantly worse on ToM tasks. With this study we also aim to find out whether these differences are still present in adolescence.

(7)

SAMENVATTING

Het begrip van mentale werkwoorden en ingebedde zinnen wordt verondersteld een nauwe relatie te hebben met Theory of Mind (ToM) bij zowel typisch ontwikkelende kinderen (TD) als bij kinderen met een specifieke taalstoornis (SLI) of autisme spectrum stoornis (ASS) (Astington & Baird, 2005; Miller, 2004; Ziatas et al., 1998). Het is echter niet bekend of beide factoren nog steeds een nauwe relatie vertonen met ToM in de periode van adolescentie (13 tot 18 jaar). Verder worden in de kindertijd duidelijke verschillen gevonden tussen TD kinderen en kinderen met SLI of ASS bij het begrijpen van ToM. Kinderen met SLI of ASS presteren significant minder goed op ToM taken. Met deze studie willen we onderzoeken of deze verschillen nog steeds aanwezig zijn in adolescentie.

(8)

1. INTRODUCTION

‘A student, aged 13 years and diagnosed with autism spectrum disorders (ASD), was confronted with the sentence ‘I think (that) it is raining outside’, while it was not raining outside. We wanted to explain that someone can think it is raining outside, even though it is not true. This student did not understand this concept; he did not realize that there can be a discrepancy between someone’s thoughts and reality.’

(9)

understanding of ToM (Miller, 2004; Tager-Flusberg & Joseph, 2005).

It is without a doubt that much more is known about ToM understanding and the relationship between language and ToM in typical and a-typical groups of children compared to adolescents. Up until now it is not completely clear how TD adolescents, but also adolescents with SLI or ASD, perform on ToM tasks and whether there is a relation between ToM understanding and language in adolescence. A new study within this direction could extend our knowledge about ToM understanding and the relationship with language.

Therefore, the main focus of our study is twofold. Firstly, we focus on the relationship between mental state verb comprehension, sentential complement comprehension and ToM in adolescence. Both linguistic factors have shown to be related to ToM understanding in childhood, but so far it remains unclear whether one or both relate to ToM understanding in adolescence. The second purpose of our study is examining the potential differences on ToM understanding between TD adolescents and adolescents with ASD or SLI. Previous literature showed clear differences between the groups on ToM understanding in childhood. TD children outperform children with ASD or SLI on ToM tasks. An effective way to measure the ToM understanding of our groups of adolescents is false belief, which is the ability to distinguish between true and false beliefs (Astington & Baird, 2005). Obviously, people’s understanding of ToM is broader than only understanding false beliefs, but the false belief task is most commonly used to measure ToM understanding (De Mulder, 2011).

(10)

2. BACKGROUND

2.1 Language and Theory of Mind

For the last three decades there has been a constant interest in the relationship between language and ToM (Astington & Baird, 2005; De Villiers & Pyers, 2002; Wellman, Cross & Watson, 2001, for a review). It is without a doubt that some kind of relationship between ToM and language exists, but to this day, the precise nature of this relationship remains unclear. Some researchers argue that ToM is crucial in order to develop language, which is the ToM first

hypothesis (Babar, Lang & Ortlieb, 2013; Ziatas, Dunkin & Pratt, 1998). These researchers

propose that if someone has a more advanced understanding of the mental states of other people, it has a positive effect on language development, in particular on those areas that show a relation with ToM understanding (Babar et al., 2013; Ziatas et al., 1998). Other researchers argue that language predicts ToM performance, the language first hypothesis (see Astington & Baird, 2005, for a collection of papers on this topic; De Villiers & Pyers, 2002). These researchers argue that general language ability, as well as specific semantic and syntactic components, need to be acquired in order to develop ToM skills (Astington & Baird, 2005; De Mulder, 2011; De Villiers & Pyers, 2002). Following this idea, some studies have claimed that the relationship between language and ToM only exists for the child during preschool and primary school (Gopnik & Wellman, 1994), while other studies argue for a relation between language and ToM across the lifespan (Astington & Jenkins, 1999; Astington & Baird, 2005; De Villiers & Pyers, 2002). A third group of researchers suggests a two-way relationship between language and ToM, the

bidirectional ToM hypothesis (Slade & Rufmann, 2005; de Mulder, 2011). Slade and Ruffman

(11)

Up to this moment it is unclear how language specifically relates to ToM understanding. There are researchers that argue that general language ability shows the strongest relationship with ToM understanding (Jenkins & Astington, 1996; Milligan, Astington & Dack, 2007). Jenkins and Astington (1997) and Milligan et al. (2007) showed high correlations between the general linguistic competence level of 3 to 5 year old children and ToM scores at a later age. Furthermore, Hollebrandse, Van Hout and Hendriks (2014) conducted a study with 6-7 year old children, 8-9 year old children and a group of control adults. All participants were tested on a second-order non-verbal false belief task and a second-order verbal false belief task. The results revealed that children were able to pass a second-order verbal false belief task at an earlier age than a second-order non-verbal false belief task. The authors argue that language supports verbal reasoning about beliefs, because it helps you to follow different beliefs attributed by people to other people (Hollebrandse et al., 2014).

Other researchers claim a more specific relationship between language and ToM. One group of these researchers argues that the semantics of mental state verbs have to be acquired in order to develop ToM abilities, which is the semantic approach (Bartsch & Wellman, 1995; Spanoudis & Natsopoulos, 2011; Ziatas et al., 1998). However, the second group of researchers believes that the syntactic properties of sentential complements have to be understood in order to understand ToM, which is the syntactic approach (De Villiers & De Villiers, 2002; Tager-Flusberg & Joseph, 2005).

(12)

factual events that are true, such as know, forget and realize. Non-factive mental state verbs denote non-factual events that are not necessarily true, such as thoughts (think, believe) and obligations (agree, promise). The distinction between factive and non-factive events is seen as meaningful, because it helps you to understand subtle differences between mental states and differences between true and hypothetical situations (Spanoudis et al., 2007; Ziatas et al., 1998). Comprehension of mental state verbs enables a person to understand the basics of their own thoughts, beliefs and feelings, as well as those of others (Babar et al., 2013; Shatz, Wellman & Silber, 1983; Spanoudis & Natsopoulos, 2011; Ziatas et al., 1998). Therefore, researchers have claimed the strong relationship between mental state verb comprehension and ToM. Ziatas et al. (1998) studied the relationship between the mental state verbs know, think and guess on ToM development in children. The results of the study demonstrated a significant relation between mental state verbs and ToM understanding.

Other researchers rather search for the relationship between language and ToM in the syntactic properties of complex sentence structures (De Villiers & Pyers, 2002; De Villiers & De Villiers, 2000). It has been claimed by De Villiers and colleagues that there is one particular construction that is extremely important for understanding ToM: the sentential complement construction. They claim that without the understanding of these complex syntactical structures, children will fail to understand fails belief (De Villiers & Pyers, 2002; De Villiers & De Villiers, 2000). Sentential complements are complement clauses that serve as the extending component of the main clause and provide additional information about the main clause (De Villiers & Pyers, 2002). These complex syntactical structures are embedded under the main clause and occur with mental state verbs like think and know and verbs of communication like say and tell. The syntactic properties of both verb types are fairly similar, because they both require a sentence as its complement (Astington & Baird, 2005). Consider the following two examples:

(13)

Keeping the examples in mind: without the sentential complement ‘that the florist shop is closed between 6 pm and 8 am’, the main clause ‘Sophie thinks – Sam said’ only gives us the information that Sam or Sophie thinks/said something. The sentential complement is embedded under the main clause ‘Sophie thinks – Sam said’ and demonstrates the content of Sophie’s or Sam’s mind. Most important about these complex structures is that the whole sentence can still be true even though the sentential complement is false. The florist shop can have for example longer or shorter opening hours, but the sentence is true as long as Sophie thinks or Sam says that (De Mulder, 2011; De Villiers & Pyers, 2002). According to De Mulder, ‘in order to understand sentential complements, a child is forced to consider the exact nature of mental states and has to understand that it is possible to have ‘false’ mental states’. Consequently, the capacity to understand and embed a (possibly) false proposition is what makes the sentential complement construction an ideal mechanism for representing false beliefs.

De Villiers & Pyers (2002) showed that ‘mastery of sentential complements is a precursor and possible prerequisite of successful ToM understanding’. Lohman & Tomasello (2003) and Hale and Tager-Flusberg (2003) demonstrated the importance of knowledge of sentential complements for ToM in a training study. They demonstrated that when children of 3 years old were trained on sentential complementation constructions, their false belief understanding also improved. The children who did not receive this particular training did not improve their false belief skills.

2.2 Development of Theory of Mind in TD children

(14)

2013). Also, somewhere between the ages of 3 and 4 years old, children learn the syntactic properties of complex sentence structures that involve these mental state verbs, such as the sentential complement construction. They begin to understand that these kinds of constructions can take false propositions. The timing of this understanding of sentential complements roughly corresponds with the moment that children are starting to understand standard ToM tasks such as false belief tasks (Astington & Baird, 2005; Newton & De Villiers, 2002).

Around the age of 4 years old, TD children successfully pass a standard first-order false belief task (Hollebrandse et al., 2014; Liddle & Nettle, 2006). There is only one step in first order ToM: understanding the thoughts of someone else - what does person X think/know/believe (Vetter, 2013)? When a child passes a first-order false belief task, it shows arguably sophisticated knowledge of attributing a belief to another person. Obviously, the understanding of other people’s mental states is broader than just the understanding of false beliefs. However, the false belief task is generally considered to be the best way to measure ToM (De Mulder, 2011). Relatively sophisticated understanding of other humans’ mental states is demonstrated when someone successfully passes a false belief task.

Understanding second-order and higher-order ToM is rather complex and is therefore acquired later than first-order ToM (Vetter, 2013). Second-order ToM involves the ability to understand what another person thinks about the thoughts of a third person: what does person X

think/know/believe that person Y thinks/knows/believes? Around the age of 6 or 7 years, TD

(15)

2.3 Theory of Mind in Clinical Groups

Understanding of ToM is not a natural process for all children. For children with specific language impairment (SLI) and children with autism spectrum disorders (ASD), the development of ToM can cause difficulties (Miller, 2004; Tager-Flusberg & Joseph, 2005).

2.3.1 Theory of Mind in SLI

Individuals with SLI provide an interesting group for the study of ToM, because their language development does not proceed in accordance with the development in other cognitive domains (Miller, 2004; Spanoudis et al., 2007). SLI is characterized by severe problems in both production and perception of language without a clear underlying cause (Zink & Breuls, 2012). Symptoms of SLI cannot be traced back to a developmental, neurological, sensory (hearing or vision), or motor disorder. Moreover, research on SLI covers a wide spectrum of language phenomena at different levels. Different studies show that children with SLI experience problems in phonology, semantics and syntax, whereas these problems continue to exist in adolescence (Clair, Pickles, Durkin & Conti-Ramsden, 2011; Norbury & Bishop, 2003; Spanoudis & Natsopoulos, 2011).

In the preschool and primary school period, children learn to comprehend and produce mental state verbs and sentential complements (Astington & Baird, 2005; Babar, Lang & Ortlieb, 2013; De Villiers & Pyers, 2002). Spanoudis et al. (2007) and Spanoudis and Natsopoulos (2011) studied the mental state verb comprehension of children with SLI, children with a pragmatic language disorder (SPD) and TD children. The results of both studies showed that the comprehension of mental state verbs was significantly impaired for the two groups with a language impairment compared to TD children.

(16)

Andres-Roqueta, Adrian, Clemente and Katsos (2013) reported similar results regarding false belief performance of children with SLI. Both studies examined the false belief and language abilities of two groups of children with SLI and two groups of TD children, one group matched on age and one group matched on language comprehension. When the linguistic complexity of the task was low, children with SLI performed comparably to the age-matched group on the false belief task (Miller, 2004). When the linguistic complexity of the tasks was higher, the children with SLI performed similarly to the younger group matched on language comprehension. Miller (2004) suggested that children with SLI fail to pass complex false belief tasks because they failed to understand the sentential complement construction. Andres-Roqueta et al. (2013) found similar results as Miller (2004), however, they reported that overall language performance was the strongest predictor of false belief performance.

As long as research into ToM understanding and the relationship between language and ToM in children with SLI goes back, not much focus has been on adolescents. Therefore, our focus in this study is to find out whether adolescents with SLI also demonstrate problems with ToM understanding and whether mental state verbs and sentential complements show a relationship with this understanding.

2.3.2 Theory of Mind in ASD

(17)

There are different cognitive accounts that theorize about the underlying key aspects of ASD, such as the weak central coherence account, the executive functions account and the theory of

mind deficit account for autism (Baron-Cohen, Leslie & Frith, 1985; Booth & Happé, 2010;

Happé, Booth, Charlton & Hughes, 2006). The weak central coherence account proposes that people with ASD appear to show weak central coherence, because they have difficulties with processing information (Booth & Happe, 2010). They prefer to focus on the details rather than the global form or meaning of for example a text (Booth & Happé, 2010). The executive function

account for autism proposes that difficulties experienced by people with ASD are a result of a

deficit in executive functions (Happé et al., 2006). These executive functions are complex cognitive processes that help a person to regulate its cognitive control and attention. According to Happé et al. (2006) ‘executive functions are required for flexible and adaptive responses in complex and new situations and involve several cognitive processes, such as inhibition, self-monitoring or planning’. People with ASD experience difficulties in regulating these cognitive processes (Happé et al., 2006). The theory of mind deficit account of autism proposes difficulties with ToM as the most important marker of ASD. 80 % of the children with ASD fail to pass a ToM task (Baron-Cohen et al., 1985; De Mulder, 2011). However, 20 % of these children with ASD successfully pass a ToM task without problems. Researchers argue that their language skills might be the reason why they pass the task (Durrleman & Franck, 2015; Lind & Bowler, 2009; Tager-Flusberg & Joseph, 2005).

(18)

In a more recent study from Durrleman and Franck (2015) studied the relation between sentential complements and false belief. The authors tested children with ASD from 6 to 16 years old and younger TD children from 4 to 9 years old on several linguistic tasks and false belief tasks. They showed that TD children and children with autism perform similarly on sentential complements tasks. However, TD children perform significantly better on false belief tasks. For both ASD and TD, correlations between the verbal false belief task and sentential complements were found. These findings support the view that sentential complements help children with ASD to pass a false belief task.

In an early stage of ToM research, difficulties with ToM were pointed for children with ASD (Baron-Cohen et al., 1985). At a later moment it was discovered that language may help children with ASD to understand ToM (Tager-Flusberg & Joseph, 2005; Ziatas et al., 1998). Also for this group of participants, not much focus has been on adolescents. In this study we will therefore explore whether difficulties with ToM understanding continue to exist in adolescence and how language relates to ToM understanding for adolescents with ASD.

2.4 Current Study

2.4.1 Aim of the Research

(19)

relationship between ToM understanding and the comprehension of mental state verbs and sentential complements in adolescence and (2) we will study the differences between TD adolescents and adolescents with ASD or SLI.

2.4.2 Research Questions and Hypotheses

Based on previous literature concerning the relation between language and ToM and literature about the differences between TD children and children with ASD or SLI on ToM understanding, this study will attempt to give an answer to the following two main research questions:

1. Is there a relation between specific linguistic mechanisms and ToM in adolescence? 2. Is there a difference between TD adolescents, adolescents with ASD and adolescents

with SLI on ToM understanding?

In order to answer these main research questions, four related sub questions, discussing the ToM performance of the groups, the comprehension of mental state verbs and sentential complements and the relation between these components, will be addressed.

I. How do TD adolescents, adolescents with SLI and adolescents with ASD perform on first-order and second-order ToM?

(20)

II. How do TD adolescents, adolescents with SLI and adolescents with ASD perform on the comprehension of mental state verbs?

We predict that the TD adolescents and adolescents with ASD or SLI will perform differently on the comprehension of mental state verbs. For both clinical groups, we predict to find difficulties with the comprehension of mental state verbs (Spanoudis et al., 2007; Ziatas et al., 1998). However, we expect more problems for the ASD group than the SLI group, because children with ASD have shown to have severe problems with understanding subtle differences in mental states and with the semantic properties of mental state verbs (Ziatas et al., 1998).

III. How do TD adolescents, adolescents with SLI and adolescents with ASD perform on the comprehension of sentential complement constructions?

We predict to find differences in performance on the comprehension of sentential complements for the three groups. For both clinical groups, we expect to find difficulties with the comprehension of sentential complements (Miller, 2004; Tager-Flusberg & Joseph, 2005). However, we more severe problems for the adolescents with SLI, because it is extensively shown that children with SLI show problems with complex structural aspects of language (Miller, 2004)

IV. What is the relation between mental state verb comprehension, sentential complement comprehension and ToM understanding in adolescence?

(21)

relationship (Spanoudis et al., 2007; Ziatas et al., 1998).

(22)

3. METHODOLOGY

3.1 Participants

Sixty-six high school students from 13 to 18 years old participated in this study. We selected twenty-three high school students with SLI, twenty-one high school students with ASD and twenty-two TD high school students. One male student from the TD group was excluded after testing, because there was an indication of severe memory problems, resulting in 21 TD students who were included in the analyses. Participants were, as much as possible, matched on age, sex and level of education in order to make a clean comparison. Group details are listed in Table 1. Table 1. Group details participants

N Age Sex IQ TD 22 16;34 9 female > 80* SD = 0,45 12 male SLI 23 15;82 9 female > 80 SD = 1,34 14 male ASD 21 16;57 2 female > 80 SD = 1,61 19 male

Notes: * no IQ information was available, however school level indicates a normal cognitive development

3.1.1 Typically developing adolescents

(23)

students with further cognitive problems were excluded from the selection. Small visual problems were allowed, if they were corrected by means of visual aids, such as contacts or glasses. One male student was excluded after testing, because there was an indication of severe memory problems, resulting in 21 TD students who were included in the analyses.

3.1.2 Adolescents with Autism Spectrum Disorders

Twenty-one high school students diagnosed with ASD participated in this study. All participants with ASD were recruited from De Mast, which is a school for ASD high school students in Merksplas. Merksplas is located in the North of Flanders, the Dutch speaking part of Belgium, close to the border between Belgium and the Netherlands. We were unable to recruit high school students with ASD from the Netherlands on a short notice; we therefore chose to test students in Belgium, since Flemish is the official language of Flanders, and Dutch, which is the official language in the Netherlands, are both varieties of the Dutch language. The lexicon and grammar are fairly similar. Moreover, Flemish and Dutch people can understand each other almost flawlessly. However, there are some small differences between the languages mainly found in phonology and lexicon (Niederländische Philologie, FU Berlin, 2014). These differences mainly manifest themselves in the spoken accent and differences in meaning of words. Small differences are found in the syntax. However, our materials were checked by a native speaker of Flemish. Furthermore, none of the participants has indicated that he or she was affected by the differences in language.

(24)

participants have not been diagnosed officially with either of these disorders. No students diagnosed with dyslexia were found in this group. The participants from this group did not have further cognitive problems or hearing problems. Small visual problems were allowed, if they were fixed by means of visual aids.

3.1.3 Adolescents with Specific Language Impairment

Twenty-three high school students with Specific Language Impairment (SLI) participated in this study. All participants with SLI were recruited from the Dr. J. de Graafschool, which is a cluster 2 school in the South of Groningen. Cluster 2 covers children with a hearing impairment and children with a communicative impairment, like severe speech- or language difficulties. All students were between 13;7 and 18;4 years old, with an average age of 15;8 years old (SD = 1,34). The group consisted of nine girls and fourteen boys. All students have a normal nonverbal IQ of 80 or higher. By school entrance, all students were tested on the presence of specific language disorders by a clinical linguist or speech therapist. Moreover, all scores and measures of the primary school, if available, were also taken into account. Six participants of this group were also diagnosed with dyslexia. Dyslexia was not an exclusion criterion because participants were not required to read words or texts during the different tasks. Participants with other comorbid disorders were excluded from this study. The participants from this group did not have further cognitive problems or hearing problems. Small visual problems were allowed, if they were fixed by means of visual aids.

3.2 Materials

(25)

3.2.1 Mental state verb factive and non-factive task

The mental state verb factive and non-factive task is based on a test developed by Spanoudis, Natsopoulos and Panayiotou (2007). By using this task, the authors were able to measure the ability of participants to infer factual and non-factual implications. Factive mental state verbs entail a true event, and non-factive verbs denote events that are not necessarily true, for example an obligation or an intention. The task of Spanoudis, Natsopolous and Panayitou (2007) consisted of ten different Greek mental state verbs, of which five factive and five non-factive verbs. An example of a factive test item (translated in English) can be found in 3. Participants had to choose which option (A, B or C) was implied by the first sentence. The correct answer in 3 is option C because the verb forget is a factive verb, which implies a true event.

3) John forgot to feed the dog

a. The dog has eaten some food

b. The dog might have eaten some food c. The dog has not eaten any food.

Our task was based on that of Spanoudis et al. (2007) but used 24 test items in total. In addition, six control items were added to the task. The goal of our task was the same as for Spanoudis et al. (2007). In our task, verbs were also checked for frequency. We included both frequent and less frequent verbs, to check whether frequency is of influence on the comprehension of mental state verbs.

We selected nine factive and nine non-factive verbs from the factive and non-factive task of Spanoudis et al. (2007) and the comprehensive list of emotional state words in participants’

narratives of Babar et al. (2013). Frequencies of these verbs were determined by the CELEX list

(26)

frequent verbs. In doing so, nine frequent verbs and nine less frequent verbs were defined. Although these frequent and less frequent verbs were not equally divided among the nine factive and nine non-factive verbs, we nevertheless decided to pilot these eighteen verbs among ten students from the University of Groningen to check whether they give the same answers to the proposed verbs. For fourteen of the eighteen items, the pilot students were consistent in their answers. For the other four verbs there was an inconsistency in non-factive verbs that denoted an obligation (agree, promise), which were not included.

Feedback was received on the first version of the task. On this basis, we designed a new version of the task. Four factive verbs (weten ‘know’, begrijpen ‘understand’, beseffen ‘realize’,

betreuren ‘regret’) and four non-factive verbs (denken ‘think’, verwachten – ‘expect’, assume

‘veronderstellen’, presume ‘vermoeden’) were included in the task. For both the factive and non-factive verbs, two verbs were determined as frequent verbs and two verbs were defined as less frequent verbs according to the CELEX criteria mentioned above. In 4a, 4b, 4c and 4d, examples can be found of a factive, frequent item (4a), a factive, less-frequent item (4b), a non-factive, frequent item (4c) and a non-factive, less-frequent item (4d).

4a) Mila knows that there was a course about animals a. There was a course about animals

b. There might have been a course about animals c. There was no course about animals

4b) Isa realizes that she lost the book a. The book is lost

b. The book is not lost c. The book might be lost

4c) Niek thinks that it has been raining outside a. It is raining outside

(27)

c. It is not raining outside

4d) Mila assumes that the stores were closed a. The stores were not closed

b. The stores might have been closed c. The stores were closed

The complete task consists of 30 items: 24 test items and 6 control items (see Appendix A). All mental state verbs were repeated three times within different sentences, because repetition gives a good indication of the comprehension of the verb. The control items were equally divided over the test items and imposed approximately the same memory load. An example of a control item can be found in 5.

5) Mila eats because she is hungry a. Mila is hungry

b. Mila might be hungry c. Mila is not hungry

The second version of the task was piloted among 13 TD students from the University of Groningen between the age of 19;5 and 26;8 years old. Different students participated in this pilot experiment as compared to the first pilot experiment. The pilot participants received the task and the instructions via e-mail. All 13 participants gave the same answers on the task, which therefore suggested that TD adults understand the implications entailed by factive and non-fative verbs.

3.2.2 Memory for complements in daily situations

(28)

require a participant to understand the concept of false belief instead of only deducing the sentential complement from the story. In the tasks of De Villiers and Pyers (2002) and Astington and Baird (2005) children were told stories in which the character of the story was telling a lie, having a false belief or making a mistake followed by a sentence that displayed reality. An example of a test item of Astington and Baird (2005) can be found in 6.

6) This is a story about Mary and her friend. One day Mary thought: ‘the stores are closed today’. The stores were actually open.

‘What did Mary think?’

De Villiers and Pyers (2002) and Astington and Baird (2005) claimed for these stories that false belief understanding was not necessary in order to answer the sentential complementation question correctly, because the participant only had to repeat the thought of the protagonist (De Mulder, 2011). However, according to Ruffman and colleagues (2003) and De Mulder (2011) ‘the story of Mary and her friend does not make a lot of sense if there is no concept of false belief available, because without false belief knowledge the story cannot be completely understood’. If this story does not make sense to a participant, it is harder to remember and understand the sentential complement question. For this reason, when a participant gives the wrong answer to the question ‘What did Mary think?’ it is not clear whether the participant is not capable of extracting sentential complements from the story or whether the participant fails to understand the concept of false belief (De Mulder, 2011).

(29)

story can be found in 7 (De Mulder, 2011). When a participant correctly responded “a teddy bear” instead of “a teddy bear, a doll and a book” to the question, (s)he was competent of understanding the content of the sentential complement.

7) [Participant sees a picture of Jan and Karin in a living room sitting next to a teddy bear, a doll and a book] ‘It’s Karin’s birthday and Karin is showing Jan the presents she got. Jan says that Karin got a teddy bear for her birthday. Karin then says that she also got a doll and a book.’

Puppet: ‘That went a bit fast. They both said something, but what did Jan say Karin got?’ What was really important for the task of De Mulder (2011) is that all utterances in the stories were true. For this reason, the child did not also have to keep in mind that one of the utterances was false and therefore keep the concept of false belief in mind. In addition to the six test stories De Mulder (2011) also added four control stories to check whether the participants could deal with the imposed memory load and whether they were able to track the different events in the stories. These control stories had the same set up as the test stories. The control stories did not include sentential complement structures.

(30)

complement and not whether a participant is competent to distinguish two sentential complements structures from each other.

The six test stories always involved two protagonists, Sam and Sophie, and two different objects. The two characters and the objects were shown on a picture to the participant. In the story, Sam or Sophie always says something about one of the two objects. This was followed by a proposition that includes the other object. After the story, the leader of the experiment asked the sentential complement question. An example of a test item can be found in 8. The complete Dutch version of the task can be found in Appendix B.

8) [Participant sees a picture of Sam and Sophie playing in a sandbox with a ball and a bucket] ‘Sam and Sophie are playing in the sandbox. Sophie brought some fun things to play with. Sam says that Sophie brought a ball. But look, she also brought a bucket!’ ‘Sophie brought two things, but what did Sam say Sophie brought?’

Correctly responding ‘a ball’ instead of ‘a ball and a bucket’ or ‘a bucket’ shows accurate comprehension of the sentential complementation construction. Just like for De Mulder (2011) no propositions of falsity were presented in the story.

(31)

9) [Participant sees a picture of Sam and Sophie standing next to a starfish and a shell] ‘Sam and Sophie are at the beach. They’re looking for things they can take home with them. Sophie found a starfish. Sam found a shell.’

‘They both found something, but what did Sophie find?’

The same 13 students were also asked for the pilot of this task. The pilot participants received the task via e-mail, which included the PowerPoint presentation with the pictures, the word document with the stories and questions and the response form. Instructions were given in this e-mail. Participants were able to switch between the word document and PowerPoint, but were asked not to during an item. All 13 participants gave the same answers on the task.

3.2.3 Bake Sale Stories (Hollebrandse, Hobbs, de Villiers & Roeper, 2007)

The bake sale stories task, based on the ice cream truck story task from Wimmer and Perner (1985), was developed by Hollebrandse, Hobbs, de Villiers & Roeper (2007). Hollebrandse et al. (2007) made sure that there were no overlapping beliefs for the protagonists in the stories, in contrast to the task of Wimmer and Perner (1985). The bake sale stories task was implemented in the present study, because a false belief task is considered to be the root test of ToM (De Mulder, 2011).

(32)

Each test item contained two probe questions, an ignor question, two first-order questions (Q1FB1 and Q2FB1) and a second-order question (Q1FB2). The purpose of the probe and ignor questions was to check if the participants were capable of keeping track of the different events in the story. The Q1FB1 was asked directly after the second probe question and involved a first-order false belief. The Q2FB1 was the last question provided to the participants and was the same question as Q1FB1. The purpose of this question was to check whether participants had difficulties with the length and complexity of the story. In doing so, Q2FB1 served as a control question in an effective way of how well participants were able to keep track of the different beliefs despite the length and complexity of the story (Hollebrandse et al., 2014). The goal of the most important question of the task, the Q1FB2, was to check the second-order false belief comprehension of the participants. What is important for this question is that it did not make use of any second-order embedding constructions of the type ‘What does Maria think that Sam

thinks they are selling at the bake sales?’ in the story. This is avoided by eliciting a second-order

answer by asking a “double” first-order question (The mailman asks: ‘What does Sam think they are selling at the bake sales?’ – What does Maria answer to the mailman?). In this case, processing a second-order false belief is still required without the commitment of processing a second-order embedding structure. An example of a test item can be found in 10. The complete Dutch version of the task can be found in Appendix C.

10) [Picture 1 showing Maria and Sam playing together, is shown to the participant] Sam and Maria are playing together. They are looking outside and see that there is a bake sale at the church. Maria says to Sam: “I’m going to buy some cookies for us”, and she leaves to the market.

[Picture 2, showing mom and Sam, is shown to the participant] Then mom arrives home and tells Sam that she just drove along the bake sales. “Do they sell cookies at the bake sales?” Sam asks. “No” says mom, “they sell apple pie”. “Then Maria will probably buy apple pie for us”, Sam says

(33)

[Picture 3, showing Maria at the bake sale, is shown to the participant] Maria arrived at the bake sales. “I would like to buy some cookies”, she says. “There are only waffles left”, says the lady behind the stand. Waffles are also delicious, so Maria buys waffles. Probe 2. Does Sam know that Maria buys waffles?

First-order false belief question. What does Sam think they are selling at the bake sales? [Picture 4, showing Maria and the mailman, is shown to the participant] On her way home Maria comes across the mailman. She says to the mailman: “I just bought waffles and I’m going to share those with my brother Sam”. “That is really sweet of you”, the mailman says. “Does Sam know what you bought?” the mailman asks.

Ignor: What does Maria tell the mailman?

And then the mailman asks: “What does Sam think they sell at the bake sales?” Second-order false belief question. What does Maria tell the mailman?

Second first-order false belief question. What does Sam think they sell at the bake sales? 3.3 Research procedure

(34)

the participants received list one for all three tasks and the other half of the participants received list two for all the tasks. Participants had the opportunity to ask for a break if needed.

(35)

4. RESULTS

The purpose of this study was twofold: (1) examining the relation between the comprehension of mental state verbs and the comprehension of sentential complements and ToM and (2) studying the potential differences between TD adolescents, adolescents with ASD and adolescents with SLI on ToM understanding. We present the results to the reader in four parts (1) descriptive statistics, (2) comparisons between groups, (3) comparisons within groups and (4) correlations between mental state verbs and sentential complements on first-order and second-order false belief.

4.1 Descriptive statistics

(36)

Table 2. Descriptive statistics

SLI ASD TD

Task Measure Min Max M SD % Min Max M SD % Min Max M SD %

Mental state verbs Test items 14 24 19.78 2.98 82 12 24 20.86 3.65 87 20 23 22.29 1.27 93 Factive 6 12 10.26 1.84 86 5 12 10.81 2.25 90 10 12 11.33 0.73 94 Non Factive Frequent Less Frequent 4 8 3 12 12 12 9.43 11.09 8.70 2.21 1.24 2.26 79 92 72 0 6 4 12 12 12 9.95 11.10 9.71 3.15 1.58 2.63 83 92 81 9 9 9 12 12 12 11.00 11.67 10.67 0.95 0.73 1.07 92 97 89 Fac-Freq Fac-Less Freq 3 1 12 6 5.70 4.61 0.77 1.59 95 77 4 0 12 6 5.81 5.10 0.51 1.79 97 85 9 5 6 6 5.86 5.43 0.36 0.68 98 90 NonFac-Freq NonFac-LessFreq 2 1 6 6 5.35 4.17 1.03 1.50 89 70 0 0 6 6 5.14 4.86 1.71 1.53 86 81 4 4 6 6 5.81 5.24 0.51 0.83 97 87 Sentential Complements Test items 0 6 3.39 2.52 56 0 6 4.14 2.27 69 4 6 4.33 2.48 72 Bake Sales Q1FB1 4 8 7.48 1.04 93 3 8 7.52 1.17 94 0 8 7.10 0.94 89 Q1FB2 Q2FB1 1 4 8 8 5.04 6.91 2.23 1.13 63 86 0 6 8 8 3.71 7.43 3.07 0.87 46 91 5 5 8 8 5.24 7.05 0.92 2.05 65 88

Note: Participants were excluded from the analyses when a score of lower than 75 % was received on the control items of a task. For this reason, one participant from the SLI group was excluded from

(37)

4.2 Comparisons between groups

4.2.1 Mental state verb comprehension

We first studied the potential differences between the groups on the test items of the mental state verb task (figure 1). Since the data were not normally distributed, group differences between the SLI group, the ASD group and the TD group were evaluated using the non-parametric equivalent of the One-Way ANOVA test, the Kruskal-Wallis H test. The Kruskal-Wallis H test showed a statistically significant difference between the SLI group, the ASD group and the TD group, χ2 (2) = 8.35, p = 0.02. We added post-hoc analyses using the Mann-Whitney U-test. The TD group performed significantly better than the SLI group, U = 114.00, p = 0.00. No significant difference was found between the TD group and the ASD group, U = 198.50, p = 0.57, nor between the ASD and the SLI group, U = 169.50, p = 0.09.

Figure 1. Proportion of correct answers given by the participants on the mental state verb task. * indicates a

significant difference

(38)

the score on the factive verbs, χ2 (2) = 4.63, p = 0.01. However, a significant difference was found between the three groups on the non-factive verbs, χ2 (2) = 7.84, p = 0.02. Post-hoc analysis showed that the TD group performed significantly better than the SLI group, U = 130.50, p = 0.01. There was no significant difference between the TD group and the ASD group,

U = 204.50, p = 0.66, nor between the ASD and the SLI group, U = 170.50, p = 0.09.

Figure 2. Proportion of correct answers given by the participants on factive and non-factive mental state verbs

Subsequently, the test items are divided in the subcategories frequent versus less frequent. Differences in mean scores between the groups on frequent verbs and less frequent verbs are presented in Figure 3. The Kruskal-Wallis H test showed that there was a statistically significant difference between the SLI group, the ASD group and the TD group on the less frequent verbs,

(39)

Figure 3. Proportion of correct answers given by the participants on frequent and less frequent mental state verbs

Finally, the test items were divided in the subcategories factive - frequent verbs, factive less - frequent verbs, non-factive - frequent verbs and non-factive - less frequent verbs. Mean scores on all subcategories for the SLI group, the ASD group and the TD group are presented in Figure 4. The results were evaluated with a Kruskal-Wallis H test. A statistically significant difference was found between the groups on the score on the non-factive - less frequent verbs, χ2 (2) = 6.56,

p = 0.04. Post-hoc analysis showed that the TD group performed better than the SLI group, U =

139.00, p = 0.01. No other significant differences were found between the groups.

Because of the lack of a non-parametric alternative, we ran a repeated-measures ANOVA to check for interaction effects. For factivity, we found a trend, F(1,64) = 3.49, p = .067, indicating higher scores for factive verbs by the three groups. A main effect was found for frequency,

F(1,64) = 39.69, p < .00, showing significant higher scores for frequent verbs for the

(40)

Figure 4. Proportion of correct answers on factive-frequent, factive- less frequent, non factive-less frequent and non

factive-less frequent mental state verbs

4.2.2 Sentential complement comprehension

The results for the groups for the total score on the test items on the sentential complement comprehension task are presented in Figure 5. Since the data were not normally distributed, group differences were evaluated with the Kruskal-Wallis H test. The analysis showed that there was no statistically significant difference between the groups on the score on the test items, χ2 (2) = 3.72, p = 0.16.

Figure 5. Proportion of correct answers on the sentential complements task

0,00 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 TD SLI ASD P rop or ti on of c or re ct an sw er s Group

Sentential complements task

(41)

4.2.3 False belief understanding

In Figure 6 the results for the first-order false belief question (Q1FB1), the second-order false belief question (Q1FB2) and the second first-order belief question (Q2FB1) are presented. There was no significant difference between groups for the Q1FB1, but a trend was found, χ2 (2) = 5.52, p = 0.06. No difference was found for Q1FB2, χ2 (2) = 3.32, p = 0.19, or for Q2FB1, χ2 (2) = 3.243, p = 0.198. Post-hoc analysis on Q1FB1 showed that the TD group performed significantly worse than the ASD group, U = 146.50, p = 0.035. No other significant differences were found.

Figure 6. Proportion of correct answers given by the participants on Q1FB1, Q1FB2 and Q2FB1

4.3 Comparisons within groups

For the mental state verb task we looked into differences between the scores for frequent verbs versus less frequent verbs and between factive verbs versus non-factive verbs within each group. For the false belief task we checked whether there was a difference between the score on the Q1FB1 and Q1FB2 within each group.

(42)

the SLI group, Z = 3.75, p < 0.01, the ASD group, Z = 2.55, p = 0.01 and the TD group, Z = -2.77, p = 0.01 All three groups performed better on frequent verbs than on less frequent verbs. No significant differences were found between the factive and non-factive verbs for all groups. For the false belief task we checked the differences between the score on Q1FB1 and Q1FB2. A Wilcoxon signed-rank test showed that the participants with SLI performed better on Q1FB1, Z = -3.46, p < 0.01, than on Q1FB2. This was also the case for the ASD group, Z = -3.53, p < 0.01 and the TD group, Z = -3.05, p < 0.01

4.4 Correlations between tasks

Spearmans rho correlation analyses were used to check for correlations. Cohen (1988, 1992 – cited by Field, 2009) proposed a guideline for correlations. According to Cohen, r = (-).10 to (-) .30 is indicated as a small correlation, r = (-) .30 to (-) .50 as an average correlation and r = (-) .50 to (-) 1.00 as a strong correlation. Only marginally significant or significant correlations will be mentioned.

(43)

Table 3. Spearman’s correlations between the mental state verb task, the sentential complements task and the false belief task

SLI ASD TD Q1FB 1 Q1FB2 Q1FB1 Q1FB2 Q1FB 1 Q1FB2 MSV SC Test Items Test items 0.13 (0.57) 0.21 (0.33) 0.41^ (0.05) -0.02 (0.94) 0.48* (0.03) 0.11 (0.65) 0.34 (0.14) 0.37 (0.10) -0.05 (0.82) 0.05 (0.85) 0.14 (0.54) -0.42^ (0.06)

(44)

5. DISCUSSION

There has been a continuous interest in the relationship between language and ToM understanding in children over the last decades. Our study followed up on this idea and examined the relationship between linguistic mechanisms and ToM understanding in adolescence and the potential differences between TD adolescents, adolescents with SLI or ASD on ToM understanding. Our main research questions were as follows:

‘Is there a relation between specific linguistic mechanisms and ToM in adolescence?’

‘Is there a difference between TD adolescents, adolescents with ASD and adolescents with SLI on ToM understanding?’

In order to answer these main research questions, we addressed four related subquestions discussing the ToM performance of the groups, the comprehension of mental state verbs and sentential complements and the relation between mental state verbs or sentential complements and ToM. All four subquestions will be discussed in this particular order.

5.1 ToM understanding in adolescents

Despite the considerable variance between participants in all groups, the general pattern found for ToM understanding was fairly similar for all three groups. However, for first order ToM we found a significant difference between the TD and ASD group. TD adolescents performed significantly worse compared to adolescents with ASD. Explanations regarding this result will be given below. No significant differences were found on second-order ToM understanding.

5.1.1 TD adolescents

(45)

(Hollebrandse et al., 2014). Nevertheless, the TD participants from this group passed first-order ToM in 89 % of the cases and second-order ToM in only 65 % of the cases. For first-order ToM, they performed significantly worse compared to the ASD group, which was not in line with previous literature nor our expectations (Franck, & Durrleman, 2015; Miller, 2004). However, an 89 % score shows sophisticated first-order ToM understanding. The significant difference found between the TD group and the ASD group is considered a shortcoming of the methodology. The circumstances for the TD group were less ideal compared to the two clinical groups. The TD participants received less information about the study and tasks beforehand. Moreover, motivation to participate in this kind of studies is higher for adolescents from clinical groups, because clinical groups experience a more direct benefit.

(46)

5.1.2 Adolescents with SLI

Previous literature demonstrated two important conclusions about the ToM understanding of children with SLI: (1) children with SLI understand ToM at a later age than TD children, and (2) children with SLI especially experience difficulties with ToM when the linguistic complexity of the task is high (Andres-Roqueta et al., 2012; Farmer, 2000; Miller, 2004). Assuming these differences are still present in adolescence, we predicted to find lower scores for adolescents with SLI. First-order ToM was passed in 93 % of the cases and second-order ToM in 63 %, which is a significant difference between the two. This pattern was similar to the pattern shown in the TD group. A possible explanation for this similar pattern is that the developmental difference between SLI and TD that was present in childhood disappeared over the years. Adolescents with SLI were able to catch up with the TD group over the years. This result is quite promising, because it shows that at least some difficulties experienced by children with SLI diminish over time. This found result is not unique for this study. Other studies have also suggested that appearances of developmental disorders, such as SLI, change when children become older (Bishop 1997).

5.1.3 Adolescents with ASD

(47)

the years, because ASD is a developmental disorder.

But how can the low scores for all three groups on second-order ToM be explained? Providing a valid explanation for this question is rather difficult, because at the minimum, sophisticated ToM knowledge would be expected in the TD group. It might be that the linguistic and cognitive complexity of the bake sales task is too high to pass second-order ToM at the ceiling level. In future studies, a different false belief task, such as the commonly used change of location task, or alternatively the strange stories task (Happé, 1994) should be preferred to check whether the task choice influences the understanding of ToM.

By checking the answers of those who failed the second-order questions, it was found that a first-order ToM answer was given to the second-first-order question in most cases. This result demonstrates that some participants may not be completely capable of separating first-order ToM from second-order ToM, but also that the adolescents participating in our study truly try to understand other people’s mental states and do not simply perform at chance.

5.2 Mental state verb comprehension and ToM

(48)

pragmatics, and those children whose problems mainly manifest in syntax and phonology. It might be possible that adolescents in our study experience more difficulties with semantics rather than with syntax. Another possible explanation might be that other linguistic difficulties of adolescents with SLI become less visible during adolescence, while difficulties with semantics continue to exist in adolescence.

Previous literature showed that children with ASD perform worse on mental state verb comprehension compared to TD children (Ziatas et al., 1998). In our study, participants of the ASD group and TD group performed similarly on the comprehension of mental state verbs. This demonstrates that they are able to improve the understanding of subtle differences in mental states while transitioning from childhood to adolescence. Finally, we did not find a difference between factive and non-factive verbs within each group. However, a significant difference was found between frequent and less frequent verbs within each group. This shows that our participants are more sensitive to frequency effects than effects of factivty.

Evidence for a relationship between mental state verb comprehension and ToM understanding was mixed. Only a positive relation was found for the SLI group between mental state verb comprehension and second-order ToM. However, this relation was a marginally significant relation. Mental state verb comprehension did not predict the scores on the ToM task for the other two groups. For this reason, no strong indication for the relation between mental state verb comprehension and ToM in adolescence was found. A possible explanation regarding this result could be that the influence of mental state verbs on ToM is limited to the preschool and primary school period, because children learn to comprehend and produce these mental state verbs in an early stage of life (Ziatas et al., 1998).

5.3 Sentential complement comprehension and ToM

(49)

indicates that differences found in childhood disappear over the years.

It was also predicted that sentential complement comprehension would show a strong relation with ToM understanding (Astington & Baird, 2005; De Villiers & Pyers, 2002). Evidence for this relationship was also mixed: we did not find any correlations between sentential complement comprehension and ToM understanding for the SLI group. However, a significant correlation was found for the ASD group between sentential complements and first-order ToM. This suggests that if a participant has great knowledge of sentential complements he or she is more likely to succeed on first-order ToM. Tager-Flusberg and Joseph (2005) found the same result. In contrast, we found a negative relation between sentential complement comprehension and second-order ToM for the TD group: a low score on the sentential complements task is most likely to result in a positive score on second-order ToM, which is contrary to previous literature (De Villiers & Pyers, 2002). There does not seem to be a straightforward explanation for this result, but the result might be explained as follows: our ToM task did only make use of single embeddings, such as Maria thinks they are selling cookies at the bake sale, and not of double embeddings, for example Maria thinks that Sam thinks they are selling cookies at the bake sale. It is therefore possible that our task does not rely specifically on sentential complements skills but more generally on linguistic skills.

5.4 Conclusions

(50)

period. More research in this direction is necessary in order to give a satisfying answer to this question.

With respect to our second main research question concerning the differences on ToM understanding for the groups, no important differences were found. Despite the considerable variance between participants in all groups, the general pattern for ToM was fairly similar for TD adolescents, adolescents with ASD and adolescents with SLI. However, for first order ToM we found a significant difference between the TD and ASD group, but we considered this a shortcoming of our methodology. No differences between the groups were demonstrated on second-order ToM.

We also suggested that for both the SLI and ASD group, developmental differences in mental state verb comprehension and sentential complement comprehension diminish in the period from childhood to adulthood. This result was also demonstrated in other studies, in which it was suggested that the manifestation of developmental disorders, such as SLI and ASD, change when children become older. Seen from a developmental perspective this is quite promising for ASD and SLI, because it shows that people suffering from either disorder may experience positive changes in adolescence.

Some strong questions remain regarding the performance of the ASD group. Researchers have tried to explain the difficulties that people with ASD experience on several accounts. We focused on the ToM deficit account for autismt, but this account cannot fully explain the problems experienced by adolescents with ASD. Two other proposed accounts are the central coherence

account and the executive functions account (Happé et al., 2006). It might be possible that the central coherence account or the executive functions account, or a combination of accounts,

could clarify the difficulties and behavior of adolescents with ASD. Future studies regarding these directions are necessary in order to find out which problems adolescents with ASD experience.

(51)

of ASD, like classical autism, Asperger and PPD-NOS, are placed under the umbrella term ASD. DSM-5’s predecessor, the DSM-IV, divided ASD into the several disorders stated above. It was noted during testing that some of the participants failed both first-order and second-order ToM. By analyzing this further, it was found that these participants have been diagnosed with classical autism in the past (American Psychiatric Association, 2000). If the criteria of the DSM-IV would have been used for this study, we might have found some differences between adolescents with autism and adolescents with SLI or TD adolescents.

5.5 Limitations

The methods used in this study have certain limitations. The limited amount of items used in the sentential complements task constrained the research. An amount of six test items might not be enough in order to test the comprehension of sentential complements. If more items will be included in future research, perhaps some of the group differences and relations will reach statistical significance.

Another possible limitation of the study was the non-ideal circumstances for the TD group. Participants from this group received less information about the study and tasks beforehand. Moreover, participants in this group had never participated in an experimental study before. Participants of the other groups were more familiar with participating in experimental studies and more motivated to participate in the study, because clinical groups experience a more direct benefit. Furthermore, during testing the environment was noisier for the TD group than for the other two groups, which could have resulted in attention difficulties. In future research, equal circumstances for all participants should be guaranteed.

5.6 Future directions

(52)

research could be to study the production of mental state verbs and sentential complements together with the comprehension of mental state verbs and sentential complements. If participants were also tested on the production of mental state verbs and sentential complements, this could give more insight in the fact whether both skills are acquired properly, because production skills are usually acquired later than comprehension skills.

(53)

6. REFERENCES

American Psychiatric Association (2000). Diagnostic and statistical manual of mental disorders DSM-IV text revision (4th edition). Washington DC: APA Press.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Andres-Roqueta, C., Adrian, J.E., Clemente, R.A., Katsos, N. (2013). Which are the best predictors of theory of mind delay in children with specific language impairment?

International Journal of Communicational Disorders, 48(6), 726–737

Astington, J.W., & Jenkins J.M. (1999). A Longitudinal Study of the Relation Between Language and Theory-of-Mind Development. Developmental Psychology

Astington, J.W. & Baird, J. (Eds.). (2005). Why language matters for theory of mind. Oxford, UK: Oxford University Press. In J. Astington & J. Baird (Eds.), Why language matters for theory of mind. Oxford, UK: Oxford University Press.

Astington, J.W. & Baird J. (2005). How language facilitates the acquisition of false-belief understanding in children with autism.

Babar, A., Baird, S., Lang, B., Ortlieb, A., & Schneider, P. (2013). Children's expression of emotional and cognitive mental states in their story generation from pictures. University of Alberta, American Speech-Language-Hearing Association Convention, Chicago. Baron-Cohen, S., Leslie, A.M., & Frith, U. (1985). Does the autistic child have a ‘‘theory of

mind?’’ Cognition, 21, 37–46.

Baron-Cohen, S. (2000). Theory of mind and autism: a fifteen-year review. Understanding other

minds: perspectives from developmental neuroscience, 3-20

Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind. New York: Oxford

University Press.

Reality versus thoughts: the understanding of Theory of Mind and mental language in TD adolescents and adolescents with SLI or ASD

Reality versus thoughts: the understanding of

Theory of Mind and mental language in TD

adolescents and adolescents with SLI or ASD

Research Master Thesis Linguistics

Language & Cognition

Liset Rouweler (2017733)

April 2016

Reality versus thoughts: the understanding of

Theory of Mind and mental language in TD

adolescents and adolescents with SLI or ASD

TABLE OF CONTENTS

ACKNOWLEDGMENTS

ABSTRACT

SAMENVATTING

1. INTRODUCTION

2. BACKGROUND

3. METHODOLOGY

4. RESULTS

Sentential complements task

5. DISCUSSION

6. REFERENCES