Implicit Meaning Comprehension in Autism Spectrum Disorders [1 ed.] 9781443876728, 9781443871129

Citation preview

Implicit Meaning Comprehension in Autism Spectrum Disorders

Implicit Meaning Comprehension in Autism Spectrum Disorders By

Yhara Formisano

Implicit Meaning Comprehension in Autism Spectrum Disorders By Yhara Formisano This book first published 2015 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2015 by Yhara Formisano All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-7112-5 ISBN (13): 978-1-4438-7112-9

To V, C, and K My Safety Net

TABLE OF CONTENTS

List of Illustrations ..................................................................................... ix List of Tables .............................................................................................. xi Preface ...................................................................................................... xiii Acknowledgements ................................................................................... xv Introduction ................................................................................................. 1 Chapter One ................................................................................................. 5 Background on Autism Spectrum Disorders 1. 1 Introduction 1.2 Autism Spectrum Disorders 1.3 Clinical Picture, Etiology, and Pathogenesis 1.4 Cognitive Theories of Autism: Executive Functions, Weak Central Coherence, Theory of Mind Chapter Two .............................................................................................. 29 Pragmatics and Autism Spectrum Disorders 2.1 Introduction 2.2 Pragmatic Competence in Neurotypical Individuals 2.3 Relevance Theory 2.4 Pragmatic Development 2.5 Pragmatic Impairment 2.6 Pragmatic Impairment in Autism Spectrum Disorders 2.7 Summary of controversies in previous research Chapter Three ............................................................................................ 63 The Research 3.1 Introduction 3.2 Research Methodology 3.3 Participants and procedures for data collection and analysis 3.3.1 Participants

viii

Table of Contents

Chapter Four .............................................................................................. 91 Results 4.1 Control Group 4.2 ASD group Chapter Five ............................................................................................ 119 Discussion and Conclusions 5.1 Comparison between the ASD group and the control group 5.2 Results of the ASD group sorted by diagnosis 5.3 Conclusions Appendix ................................................................................................. 129 Bibliography ............................................................................................ 135 Notes........................................................................................................ 151 Index ........................................................................................................ 157

LIST OF ILLUSTRATIONS Fig. 1-1 Navon test Fig. 1-2 False Belief Test Fig. 3-1 Slide 2 Fig. 3-2 Slide 4 Fig. 3-3 Slide 7 Fig. 3-4 Slide 9 Fig. 3-5 Slide 12 Fig. 3-6 Slide 14 Fig. 3-7 Slide 16 Fig. 3-8 Slide 18 Fig. 3-9 Slide 20 Fig. 3-10 Slide 22 Fig. 3-11 Slide 24 Fig. 3-12 Slide 26 Fig. 3-13 Slide 28 Fig. 4-1 Percentages of participants and total target answers for the control group Fig. 4-2 Percentages of participants and total target answers for the control group Fig. 4-3 Percentages of people in the ASD group and their total target answers Fig. 4-4 Mean of non-target answers divided in IMs and control items Fig. 4-5 Percentages of people in the ASD group and their IMs target answers Fig. 4-6 Percentages of non-target answers divided by kind of IM Fig. 4-7 Mean of non-target answers on IMs per participant Fig. 5-1 Mean of non-target answers for both groups Fig. 5-2 Mean of non-target answers for both groups divided by kind of IM Fig. 5-3 Mean of non-target productions on GCIs and PCIs for both groups Fig. 5-4 Mean non-target answers on PCIs Fig. 5-5 Non-target answers related to age for both groups Fig.5-6 Mean of non-target answers sorted by diagnostic group

LIST OF TABLES

Table 1-1. Examples of the empathizing-systemizing theory of ASD Table 3-1. Questions typology and quantity Table 3- 2. Examples from the alternative version of the Test Table 3-3. ASD group, ordered by age Table 3-4 Control participants, ordered by age Table 4-1. Results of the control group Table 4-2 Participants that produced a non-target answer on 8b Table 4-3 Results of the ASD Group Table 4-4. Details on IMs occurrences and non-target answers Table 4-5. Generalized and Particularized Conversational Implicatures Table 4-6. Non-target answers on presuppositions Table 4-7. Non-target answers on conventional implicatures Table 4-8. Non-target answers on conversational implicatures

PREFACE

This book is addressed to practitioners from two main fields. First, to clinical linguists, and more specifically clinical pragmatists, conducting research on Autism Spectrum Disorders and, secondly, to neuropsychiatrists, psychologists, and language therapists who both study this condition and work with people on the spectrum. This being a book addressed mainly to these experts from two different fields, I chose to include very general notions from both fields so as to provide a basis to everyone to follow and understand the theoretical notions underlying the research. Thus, if you are a linguist, you probably will not need to go through the first part of Chapter Two, which illustrates some basic pragmatic notions. Similarly, if you are a neuropsychiatrist, a psychologist, and/or a language therapist, you may not need to read the first part of Chapter One, which provides some general background knowledge on ASD. I also have to specify that the research was conducted in 2011. This means that the diagnostic manuals in use at the time were ICD-10 (still in use today), and DSM-IV. The latter currently has a new edition, but I chose to present the fourth edition considering that it was the one in use at the time of the research. Clinical pragmatics is a young discipline and involves the cooperation of people with different expertise. I am a linguist, and I knew I needed the help of medical doctors to conduct this research. Achieving this was not always easy but, once managed, it proved to be very fruitful. Recognizing one’s own limits and understanding the necessity of collaboration for research in boundary-crossing disciplines is fundamental for scientific improvement and progress.

ACKNOWLEDGEMENTS

This research would not have been possible without the precious collaboration of the people tested; thus the first acknowledgement is for the 58 people who patiently sat the test and answered all my questions. I would not have been able to recruit the people in the ASD group without the help of several people who deserve to be mentioned here: Doctor Petrillo from Ospedale San Donato–Arezzo; Doctor Radice from Ospedale Le Scotte–Siena; Doctor Grittani from Centro Autismo e Disturbi dello Sviluppo–Rimini; Doctor Gallo and Speech Therapist Reverberi from Centro Autismo–Reggio Emilia; Doctor Giusti from Istituto Medaglia Miracolosa–Arezzo; Vice-President of the Asperger Group Lazio Mrs. Adami. I owe the possibility of conducting this research to you, your foresight, open-mindedness, and patience. The associations and/or structures that provided the facilities for the test administrations were: Association Gruppo Asperger Lazio–Rome; Centro Autismo, Ward of Child Neuropsychiatry, Azienda Unità Sanitaria Locale–Rimini; Centro Diagnosi, Cura e Ricerca sull’Autismo e i Disturbi Pervasivi dello Sviluppo, Azienda Unità Sanitaria Locale–Reggio Emilia; Istituto di Riabilitazione Medaglia Miracolosa, Viciomaggio–Arezzo; Ward of Child Neuropsychiatry of the Hospital Santa Maria delle Scotte, Siena; Ward of Child Neuropsychiatry of the Hospital San Donato, Arezzo. As for the devising of the test, I am grateful to Professor Vergaro (University of Perugia) and Professor Sbisà (University of Trieste) for having ascertained its pragmatic adequacy, and to Doctor Petrillo (ASL 8 Arezzo) for having ascertained its psycho-neurological adequacy. I also thank Professor Cummings (Nottingham Trent University) for her insightful inputs. My gratitude goes to Professor Vergaro and Professor Sandford (University of Perugia), for having read the manuscript, for their helpful comments and suggestions, and for their support.

INTRODUCTION

This book presents the results of a research project focusing upon implicit meaning comprehension in people along the spectrum of autistic disorders. The rationale for such an investigation is based on the fact that, although research on the topic abounds and a general agreement is emerging about the fact that autistic people are characterized by a ‘pragmatic impairment’, even in cases in which all other levels of linguistic competence are not affected by this medical condition, the patients still have problems in the use of language, not recognizing communication as a social act. Consequently, autistic people are reported to have problems with metaphors, emotional language expression and comprehension, prosody and inferential language. The cause of this impairment of pragmatic skills in toto is still unknown. A possible explanation for such a discrepancy may be due to the fact that researchers appear to be carrying out their research with the aim of demonstrating that the theory on which their work is based is the definitive explanation for such a phenomenon. Taking a critical stand towards this type of approach, the present research is based on a solid theoretical basis, but does not try to demonstrate the superiority of one theoretical hypothesis over another. It starts from the assumptioníbased on previous researchíof an inferential language impairment in Autistic Spectrum Disorder (hereafter ASD) patients and analyzes data with no theoretical bias. Only after the analysis does it attempt to identify which theory best accounts for the data collected. As the First Chapter illustrates, there are three main theories that account for pragmatic impairment in people with ASD: weak central coherence (WCC), executive functions (EEF), and theory of mind (ToM)1. Crucially, they have all been developed by psychologists who, in their attempt to identify the causes of autism, analyze not only the social and psychological dysfunctions of these patients, but also the linguistic ones. This type of approach seems to be flawed in that the study of language functions and dysfunctions, being based on linguistic data, are the privileged object of investigation of linguistics, which, of course, interacts with other disciplines such as psychology or neurology. Nonetheless, when it comes to analyzing linguistic data, the pivotal role of linguistics can be neither ignored nor played down.

2

Introduction

After having dealt with what pragmatic competence actually is and how it develops and functions in neurotypical individuals, Chapter Two presents previous research about pragmatic impairment in ASD patients, pointing out its strong and weak points. Faced with the need to test pragmatic skills in these patients, the existing material was analyzed, and what was found out is that there are only two tests available for Italian professionals to analyze pragmatic competence: Prove per la valutazione della Comunicazione Referenziale2 (PCR) (Camaioni et al., 1995), and Abilità Pragmatiche nel Linguaggio3 (APL) Medea (Lorusso, 2009). They both utilize written material, even if in APL there are some visual aspects like pictures and comics4. It seems difficult to test pragmatic skills using only written materials, given that by definition pragmatic skills refer to the ability to use language in communicative contexts. Even tests devised for English professionals (see §2.6) are characterized by this underestimation of the role of context in communicative exchanges. Presenting people with transcripts of conversations does not give them the necessary tools to interpret the exchange. Thus, this kind of approach appeared to be flawed because it seems unnatural and inauthentic. Based on these observations, I decided to devise a test that would recreate more authentic communicative exchanges. The test and the methodology are illustrated in Chapter Three. The participants tested were required to watch short videos of conversational exchanges containing some kind of implicit meaning, and were then asked questions about it. Videos create a scene which allows for inclusion of all of the contextual clues necessary for interpretation and, also, for analysis of the extent to which the inability to interpret implicit meaning depends on testing modality. The test devised for this research differs from the previous ones mainly because of this feature, in that it gives context its due importance, thereby allowing for interactions to be as natural and realistic as possible. Furthermore, I was present during the test administration and not only did I record the answers and the physical responses to the testing items presented, but I also intervened by asking questions about the choices being made by the participants. As for methodology, the study follows the ‘case-control’ design in that the same test was administered to an experimental group of 29 people along the autistic spectrum and to a control group of neurotypical individuals matched for age and sex. The results of both groups were then compared and analyzed to establish the extent of implicit meaning comprehension in people with ASD and its divergence from the norm.

Implicit Meaning Comprehension in Autism Spectrum Disorders

3

As Chapters Four and Five discuss, providing the appropriate context fostered the inferential process. In other words it allowed these participants to select the most relevant interpretation of an utterance, even if it was not the literal one. The results of the study would seem to go against the predictions and the results of previous research as far as the severity of the impairment is concerned, in that ASD patients demonstrate an ability, of course at different levels, to disambiguate sentences and assign a non-literal meaning to utterances.

CHAPTER ONE BACKGROUND ON AUTISM SPECTRUM DISORDERS

This chapter presents an overview on autism and autistic spectrum disorders. It firstly presents this condition; secondly, it discusses the means available for its classification and diagnosis; then, it describes this condition from a medical perspective presenting its etiology and pathogenesis; lastly, it presents the most accredited cognitive theories that account for it.

1. 1 Introduction When the term autistic (from the Greek Į‫ރ‬IJȩȢ) was coined by the Swiss psychiatrist Eugen Bleuler in 1911, it was used to describe schizophrenic patients who isolated themselves from the outside world and social relationships, withdrawing into themselves. In 1943, Leo Kanner hypothesized that some schizophrenic patients, described until then as autistic, could actually be considered as affected by a syndrome that was different from schizophrenia and from the other mental disorders known at that time. For this reason, he named this syndrome autism. After the observation of 11 children, Kanner (in Rutter 1978:139) described nine symptoms characteristic of this newly identified syndrome: “[…] inability to develop relationships with people, a delay in speech acquisition, the non-communicative use of speech after it developed, delayed echolalia, pronominal reversal, repetitive and stereotyped play activities, an obsessive insistence on the maintenance of sameness, a lack of imagination, a good rote memory and a normal physical appearance”.

Autism is today defined as a pervasive developmental disorder that persists throughout life. The term ‘pervasive’ refers to the fact that it affects several areas of cognitive and behavioral functioning. ‘Developmental’ refers instead to the fact that it usually manifests itself in the first years of life. Even if not all of the symptoms described by Kanner are recognized

Chapter One

6

today as being systematically associated with autism, he had the considerable merit of having identified three main features characteristic of this disorder. According to the World Health Organization (1992) and the American Psychiatric Association (2000), these are: -

Impairment in social interaction Communication deficits Restricted interests and activities, and stereotyped, repetitive behaviors.5

The pediatrician Hans Asperger was conducting research on a syndrome similar to autism in Austria, at the same time as Kanner was working in the United States. The syndrome today bears his name and is characterized by the same core features as autism. Asperger patients, though, do not present mental retardation or delay in language acquisition. However, given that Asperger wrote in German, his work became known only in the 1980s thanks to the British psychiatrist Lorna Wing, who was the first to use the term ‘Asperger’s Syndrome’6. She also conducted research that led her to put forward the idea of quantitative and not qualitative differences among autistic patients. In her 1979 article she coined the term ‘Autism Spectrum Disorders’7 which refers to and includes autistic-like conditions that vary from severe classical autism at one end, to High Functioning Autism (hereafter HFA) and Asperger’s Syndrome (hereafter AS) at the other.

1.2 Autism Spectrum Disorders Autism and Asperger’s Syndrome are classified by two manuals: the International Statistic Classification of Diseases and Related Health Problems (hereafter ICD), and the Diagnostic and Statistical Manual of Mental Disorders (hereafter DSM). The two manuals identify criteria that are substantially the same. Nevertheless, both will be presented for the sake of completeness. It is important to point out, as underlined in Cummings (2009:118), that there are no biological markers for autism yet, and thus clinicians rely on behavioral criteria for the diagnosis of this disorder.

1.2.1 ICD-10 The World Health Organization provides a description of the symptoms that characterize autism spectrum disorders in the classification of diseases

Background on Autism Spectrum Disorders

7

known as ICD. In this classification, the autistic spectrum is assigned the code F84 and includes: classical autism, atypical autism, Rett’s Syndrome, Other Childhood disintegrative disorder, Overactive Disorder Associated with Mental Retardation and Stereotyped Movements, Asperger’s Syndrome, Other Pervasive Developmental Disorder and Pervasive Developmental Disorder Not Otherwise Specified (hereafter PDD-NOS). The ICD is currently in its tenth edition8, which was published for the first time in 1992, and lists the following diagnostic criteria for childhood autism9: A. Abnormal or impaired development is evident before the age of 3 years in at least one of the following areas: 1. 2. 3.

receptive or expressive language as used in social communication; the development of selective social attachments or of reciprocal social interaction; functional or symbolic play.

B. A total of at least six symptoms from (1), (2) and (3) must be present, with at least two from (1) and at least one from each of (2) and (3): 1. Qualitative impairments in social interaction are manifest in at least two of the following areas: a. failure to use adequately eye-to-eye gaze, facial expression, body postures, and gestures to regulate social interaction; b. failure to develop (in a manner appropriate to mental age, and despite ample opportunities) peer relationships that involve a mutual sharing of interests, activities and emotions; c. lack of socio-emotional reciprocity as shown by an impaired or deviant response to other people’s emotions; or lack of modulation of behavior according to social context; or a weak integration of social, emotional, and communicative behaviors; d. lack of spontaneous seeking to share enjoyment, interests, or achievements with other people (e.g., a lack of showing, bringing, or pointing out to other people objects of interest to the individual).

8

Chapter One

2. Qualitative abnormalities in communication are manifest in at least one of the following areas: a. delay in or total lack of development of spoken language that is not accompanied by an attempt to compensate through the use of gestures or mime as an alternative mode of communication (often preceded by a lack of communicative babbling); b. relative failure to initiate or sustain conversational interchange (at whatever level of language skill is present), in which there is reciprocal responsiveness to the communications of the other person; c. stereotyped and repetitive use of language or idiosyncratic use of words or phrases; d. lack of varied spontaneous make-believe play or (when young) social imitative play. 3. Restricted, repetitive, and stereotyped patterns of behavior, interests, and activities are manifested in at least one of the following: a. an encompassing preoccupation with one or more stereotyped and restricted patterns of interest that are abnormal in content or focus; or one or more interests that are abnormal in their intensity and circumscribed nature though not in their content or focus; b. apparently compulsive adherence to specific, non-functional routines or rituals; c. stereotyped and repetitive motor mannerisms that involve either hand or finger flapping or twisting or complex whole-body movements; d. preoccupations with part-objects of non-functional elements of play materials (such as their odor, the feel of their surface, or the noise or vibration they generate). C. The clinical picture is not attributable to the other varieties of pervasive developmental disorders; specific development disorder of receptive language (F80.2) with secondary socio-emotional problems,

Background on Autism Spectrum Disorders

9

reactive attachment disorder (F94.1) or disinhibited attachment disorder (F94.2); mental retardation (F70-F72) with some associated emotional or behavioral disorders; schizophrenia (F20.-) of unusually early onset; and Rett’s Syndrome (F84.12). The diagnostic criteria for Asperger’s Syndrome10 on the other hand, are the following: A. A lack of any clinically significant general delay in spoken or receptive language or cognitive development. Diagnosis requires that single words should have developed by two years of age or earlier, and that communicative phrases be used by three years of age or earlier. Selfhelp skills, adaptive behavior, and curiosity about the environment during the first three years should be at a level consistent with intellectual development. However, motor milestones may be somewhat delayed and motor clumsiness is usual (although not a necessary diagnostic feature). Isolated special skills, often related to abnormal preoccupations, are common, but are not required for diagnosis. B. Qualitative abnormalities in reciprocal social interaction (criteria as for autism). C. An unusually intense circumscribed interest or restrictive, repetitive, and stereotyped patterns of behavior, interests and activities (criteria as for autism; however, it would be less usual for these to include either motor mannerisms or preoccupations with part-objects or non-functional elements of play materials). D. The disorder is not attributable to other varieties of pervasive developmental disorder; schizotypal disorder (F21); simple schizophrenia (F20.6); reactive and disinhibited attachment disorder of childhood (F94.1 and .2); obsessional personality disorder (F60.5); obsessive-compulsive disorder (F42).11

1.2.2 DSM IV12 The American Psychiatric Association (APA), which periodically publishes the Diagnostic and Statistical Manual of Mental Disorders (DSM), provides another classification tool. The edition in use at the time of the research was the fourth. It was published in 1994, and revised in 2000. Its criteria for diagnosing an autistic disorder are:

10

Chapter One

A. A total of six (or more) items from (1), (2), and (3), with at least two from (1), and one each from (2) and (3): 1. Qualitative impairment in social interaction, as manifested by at least two of the following: a. marked impairment in the use of multiple nonverbal behaviors such as eye-to-eye gaze, facial expression, body postures, and gestures to regulate social interaction; b. failure to develop peer relationships appropriate to developmental level; c. a lack of spontaneous seeking to share enjoyment, interests or achievements with other people (e.g., by a lack of showing, bringing or pointing out objects of interest); d. lack of social or emotional reciprocity. 2. Qualitative impairments in communication as manifested by at least one of the following: a. delay in, or total lack of, the development of spoken language (not accompanied by an attempt to compensate through alternative modes of communication such as gesture or mime); b. in individuals with adequate speech, marked impairment in the ability to initiate or sustain a conversation with others; c. stereotyped and repetitive use of language or idiosyncratic language; d. lack of varied, spontaneous, make-believe play or social imitative play appropriate to developmental level. 3. Restricted, repetitive, and stereotyped patterns of behavior, interests, and activities, as manifested by at least one of the following: a. encompassing preoccupation with one or more stereotyped and restricted patterns of interest that is abnormal either in intensity or focus;

Background on Autism Spectrum Disorders

11

b. apparently inflexible adherence to specific non-functional routines or rituals; c. stereotyped and repetitive motor mannerisms (e.g., hand or finger flapping or twisting, or complex whole-body movements); d. persistent preoccupation with parts of objects. B. Delays or abnormal functioning in at least one of the following areas, with onset prior to age three: (1) social interaction, (2) language as used in social communication, or (3) symbolic or imaginative play. C. The disturbance is not better accounted for by Rett’s Disorder or Childhood Disintegrative Disorder. Asperger’s Syndrome diagnostic criteria are: A. Qualitative impairment in social interaction, as manifested by at least two of the following: 1. marked impairments in the use of multiple nonverbal behaviors such as eye-to-eye gaze, facial expression, body postures, and gestures to regulate social interaction; 2. failure to develop peer relationships appropriate to developmental level; 3. lack of spontaneous seeking to share enjoyment, interests, or achievements with other people (e.g., by a lack of showing, bringing, or pointing out objects of interest to other people); 4. lack of social or emotional reciprocity. B. Restricted repetitive and stereotyped patterns of behavior, interests, and activities, as manifested by at least one of the following: 1. encompassing preoccupation with one or more stereotyped and restricted patterns of interest that is abnormal either in intensity or focus; 2. apparently inflexible adherence to specific, non-functional routines or rituals; 3. stereotyped and repetitive motor mannerisms (e.g., hand or finger flapping or twisting, or complex whole-body movements); 4. persistent preoccupation with parts of objects.

12

Chapter One

C. The disturbance causes clinically significant impairment in social, occupational, or other important areas of functioning. D. There is no clinically significant general delay in language (e.g., single words used by age 2 years, communicative phrases used by age 3 years). E. There is no clinically significant delay in cognitive development or in the development of age-appropriate self-help skills, adaptive behavior (other than social interaction), and curiosity about the environment in childhood. F. Criteria are not met for another specific Pervasive Developmental Disorder or Schizophrenia.13

1.2.3 ADI-R, ADOS and M-Chat The Autism Diagnostic Interview-Revised (ADI-R) and the Autism Diagnostic Observation Schedule (ADOS) are the two main tools used to diagnose and assess autism. ADI-R is a structured interview with parents of individuals suspected of having an ASD, and, on the basis of their answers, the clinician can reach the diagnosis. ADOS consists of a series of semi-structured activities in which the clinician and the individual suspected of an ASD are engaged. It allows the observation of social and communicative behaviors related to a possible diagnosis of an autistic condition. It is important to point out that diagnosis of autism spectrum disorder is traditionally accepted to be possible by the age of three; however, early diagnosis has been proven to be fundamental for early intervention, which is key to a more fruitful treatment of these people. The tool currently used for early diagnosis is the Modified Checklist for Autism in Toddlers (MCHAT) that allows clinicians to hypothesize ASD diagnosis around the age of one.

1.3 Clinical Picture, Etiology, and Pathogenesis The clinical picture of a disease is a description of all the features that are observable in a patient. In autism, these features are mainly behavioral, even if there are also a few observable phenotypic manifestations of this condition. On the one hand, the term ‘etiology’ refers to the causes of a disease, while on the other hand ‘pathogenesis’ describes the origins of the

Background on Autism Spectrum Disorders

13

clinical picture. In the case of autism causes are still unknown, but there is no longer any doubt (see §1.3.2) that this condition has a biological origin. Thus, even if the biological marker responsible for autism has not been identified yet, the research in this field has been intensified.

1.3.1 Clinical Picture As anticipated in §1, summarizing the diagnostic criteria of ICD and DSM, the impairments caused by ASD affect three macro-areas of cognitive and behavioral development: social interaction, communication, and interests and behaviors. People on the autistic spectrum are characterized by an inability or extreme difficulty in establishing social relationships; they are reluctant to share experiences or interests, and to express their attachment to people. The difficulty in engaging in social interaction is only partially explainable in terms of a lack of interest in other individuals, and a self-withdrawal. These people find human interaction puzzling and confusing because of their inability to decode the non-verbal expressions and behaviors that regulate and pervade it. They fail to establish eye contact, make scarce use of body language, and ignore culturally established constraints on proxemic distance. Moreover, they do not feel the need to share their emotions or their interests with their peers, and this contributes to their social isolation. In addition, they do not seem to understand others’ emotions. According to Baron-Cohen (2003, 2008) this failure in emotional sharing and understanding is explainable in terms of an extremely low empathy quotient. Following an original intuition by Asperger (1944), Baron-Cohen (2003, 2008) proposes that the brain of autistic patients has the characteristics of an extreme male brain14. He elaborated a theory to account for gender differences in brain functioning according to which two main forces guide cerebral functioning: empathizing and systemizing. ‘Empathizing’ refers to the ability to recognize others’ emotions, while ‘systemizing’ is the ability to analyze and construct systems, to identify the patterns of their functioning and the rules that govern them. In the broad sense of the word, a system is to be understood as anything that is governed by rules. If these two forces are considered the two extremes of a continuum, it appears that women’s brains are on average more towards the empathizing end, and men’s are on average towards the systemizing end. Baron-Cohen explains that this is also plausible from a phylogenetic perspective. Women had to be able to understand others’ intentions in order to protect their offspring and so their empathy levels increased,

14

Chapter One

while men, being the ones who were out hunting, developed the visuospatial skills necessary for their purposes. The autistic brain would be at the extreme systemizing end. People with ASD in fact perform very poorly in communication and social tasks based on empathy, whereas they obtain excellent results in logic, physics, and mechanics. Baron-Cohen et al. (1997) and Rutherford et al. (2002) conducted the same kind of experiment that required autistic people to infer the emotion of a person just by looking at his/her eyes, or listening to his/her voice. Autistic patients performed significantly worse in both tasks compared to typically developing people, corroborating the researchers’ hypothesis of an impairment in people with ASD to “read the mind in the eyes” (1997:813) and “in the voice” (2002:189). This hypothesis is also supported by biological data according to which autism is explainable in terms of testosterone levels during pregnancy. I will return to this point in §1.3.2 dealing with the causes of autism. Another great difficulty that prevents these people from establishing social relationships is in the sphere of communication. As already stated, their impairment affects non-verbal communication, but also, and crucially, verbal communication. People on the autistic spectrum are all characterized by language impairments. The emergence of language is delayed for the majority of these people (though not for AS) and some of them remain completely or selectively mute15. It is important to point out that people with ASD are physically able to articulate language. The fact that they do not speak is sometimes explainable in terms of a hypersensitivity to sensory stimuli, which makes communication harder. This is because auditory stimuli might be perceived as too loud and, thus, exceptionally bothering. Even the touch of the tongue in different parts of the mouth can destabilize these people’s sensibility and render communication painful to them. In other cases, they do not speak because they do not feel the need to do so. Apart from possibly communicating basic needs, they do not recognize the socio-affective component of communicative exchanges, and prefer to withdraw into themselves and their world. Usually people with ASD who do speak are the ones in the highest levels of functioning, and their language can be characterized by extreme verbosity and erudition, delayed echolalia, pronoun reversal, use of non words, atypical prosody, difficulties in initiating and sustaining a conversation, difficulties in turn-taking, and impairment of pragmatic skills, such as failure in the use of politeness and in understanding nonliteral language (see §2.6).

Background on Autism Spectrum Disorders

15

The third area of impairment is the one that concerns interests and activities. People with ASD are reported to have extremely restricted interests, mostly about unusual topics. They are fascinated, for example, by taxonomies, parts of objects, mechanics, or formal systems. This fascination absorbs them to an extent that it sometimes becomes an obsession; they might feel the need to know everything about the topic they are interested in, and might go into tantrums, and become frustrated about it, when they realize that there is something they do not, or cannot, know. They become so involved in what they are doing that they completely neglect anything else from the outside world and from themselves as well, like for example eating or taking care of their personal hygiene. They also become obsessive about adhering to their routines, e.g. always taking the same road to go to work, or doing the same things every day in the same order. The area of interests is also characterized by surprising islands of ability16. These people are sometimes able to make extremely difficult calculations in very little time. For instance, they can calculate square roots of large numbers or the weekday for any calendar date, they have an excellent memory, which they apply to whatever their interest is, they might be able to memorize entire books or listen to a song once, and reproduce it exactly. They are able to notice very little details that neurotypical individuals usually ignore, like patterns in wallpaper, and are sometimes characterized by exceptional visuo-spatial skills that make them great engineers and architects. It should be underlined that these abilities were once overlooked or negatively characterized as being the expression of a disease. The current view (Shah & Frith 1993, Armstrong 2010), on the contrary, is that people with ASD, especially at high levels of functioning, should be encouraged to use their skills for professional goals. Considering an autistic person from this perspective also explains why at the phylogenetic level autistic genes survived natural selection. Early groups of hominids probably needed elements of the tribe who were able to memorize and notice details that others did not (Reser 2011). Apart from behavioral aspects, the clinical picture also includes some phenotypic characteristics. The most evident are a slight macrocephaly due to a generalized overgrowth of the brain, ambulatory clumsiness, and a more pronounced difference in the length of the index finger and the ring finger. I will return to this last feature in the next section dealing with the relationship between hormones and the etiology of autism.

16

Chapter One

1.3.2 Etiology and Pathogenesis The causes of autism are still unknown. Nevertheless, this condition is known today to have a strong genetic basis, contrary to what was believed in the 1950s when investigation on its causes began. Bettelheim (1967), following Kanner (1949), proposed a pure psychogenic origin and coined the term ‘refrigerator mother’ to account for the etiology of this condition. According to him, emotive inadequacy of mothers caused emotional disorders in their children and this led to autism. As Frith et al. (1991:433) point out, “it is now generally accepted that there must be a biological origin to autism”. Several researchers (Nelson & Nelson 1987, Bauman & Kemper 1987, Amaral et al. 2003, Kleinhans et al. 2009, Millis Schumann et al. 2009, Mosconi et al. 2009, among others) report data concerning the differences between the autistic brain and the typically developed brain. These differences concern brain dimensions, structure and functioning. ASD people are characterized by macrocephaly, which is due to a bigger and heavier brain and probably explainable in terms of a higher number of synapses (caused either by a scarce regulation of the neural death process or by the fact that the autistic brain does not discard useless cerebral connections as the typical brain does with the process of ‘pruning’). Moreover, research on brain structures has shown that autistic brains present abnormal dimensions of: - amygdala, responsible for emotional responses; - hippocampus, which is involved in some mnemonic functions; - caudate nucleus, controlling movements; - some parts of the cerebellum, more precisely those controlling attention switching and motor coordination. It is not clear, though, whether these differences, together with macrocephaly, can be considered as part of the causes or of the consequences of this condition. Nevertheless, given that early intervention brings about extremely encouraging improvement for ASD children (Dawson & Zanolli 2003, Ingersoll et al. 2007, Dawson 2008, Wallace & Rogers 2010, among others), it could be assumed that the anatomical differences are secondary. The differences in functioning have been observed through functional Magnetic Resonance Imaging (fMRI) and Electroencephalography (EEG). These techniques allow us to look at the brain while it is performing a task. Studies carried out using these tools (Baron-Cohen et al. 1999, Williams et al. 2001, Müller et al. 2003, Oberman et al. 2005, Dapretto et al. 2006, Just et al. 2007, Bookheimer et al. 2008, Kana et al. 2009,

Background on Autism Spectrum Disorders

17

Lombardo et al. 2011) have shown that the brains of autistic patients present different patterns of activation compared to people with typical development when performing tasks of mental states attribution, emotion inferencing, imitation, and facial processing. Neurotransmitters as well have been found to work differently in the autistic brain (Hussman 2001, Langford 2006, Pardo & Eberhart 2007). The most significant differences are reported to be in the levels of serotonin (involved in the regulation of aggressiveness, anger, and sleep among other things) and Ȗ-aminobutyric acid (or GABA, responsible for inhibition of responses to certain stimuli). There is also evidence for genetic and hormonal factors in autism (Rutter 2000, Muhle et al. 2004, Millis et al. 2007, Sarachana 2011). Even if there is not an agreement about whether autism is entirely ascribable to genetic causes, research shows that there is certainly a genetic component in its development. Studies on twins (Folstein & Rutter 1977, Bailey et al. 1995, Ronald et al. 2006) were used to determine the degree of heritability of this condition. They agree on the fact that if one twin has autism, the chance of the other twin also having it varies depending on whether the twins are monozygotic (60-90 percent) or dizygotic (5-10 percent), something that clearly demonstrates the high heritability of this condition. Further evidence of this is that autism runs in families (Bishop et al. 2006); if one child in the family has a disorder of the spectrum, there are higher chances for his/her siblings to have it as well, or develop a related disorder. In addition, mutations on some chromosomes have been found in autistic patients17. Both brain differences and genetic factors indicate the biological component of autism. However, this evidence does not account for the causes of autism. A possible etiological explanation of this disorder could be the levels of mothers’ testosterone during pregnancy. This hypothesis is mainly based on gender differences in the incidence of autism, which is attested at 4:1 (male: female) for classical autism and 9:1 for Asperger’s Syndrome. Furthermore, studies on the role of hormones in autism (Rutter 2000, Muhle et al. 2004, Millis et al. 2007, Sarachana 2011) have shown that boys with autism are more likely to reach puberty earlier than their typically developing peers, while girls tend to have a delayed menarche, and to develop the polycystic ovary syndrome. These findings clearly demonstrate hormonal imbalance that favors testosterone over estrogen in these people. High fetal testosterone levels have been found to be significantly related to the likelihood of the child developing an ASD and, even in cases of typically developing individuals, to be associated with lower rates of eye contact during infancy and social difficulties during

18

Chapter One

childhood. This theory fits well with what Baron-Cohen (2003) put forward as the extreme male brain theory of autism to account for the different cognitive processing style of these people.

1.3.3 The Mirror Neuron Theory of autism A different etiological account of autism is provided by those researchers who propose the mirror neuron theory of autism (Williams et al. 2001, Oberman et al. 2005, Dapretto et al. 2006, Ramachandran & Oberman 2006). Mirror neurons (MNs hereafter) were discovered by Giacomo Rizzolatti and his équipe in the 1990s. They are a particular kind of neurons which activate themselves both in the performing of an action and when a person sees that same action performed by someone else18. They are crucial for imitation tasks and, more importantly, for the understanding of others’ intentions by means of mentally simulating their actions (hence the name mirror). Considering these characteristics, they have been proposed to be the neural bases of empathy and social skills in humans and higher primates (Rizzolatti et al. 2001). Soon after their discovery, it appeared clear to researchers that MNs perform all those functions that are impaired in people with ASD, namely the ability to read other people’s minds, and understand others’ intentions through empathy. Ramachandran et al. (2005) conducted an experiment that provided evidence of dysfunctions in the mirror system of autistic patients. The participants were asked to move their hand, and then to watch a video of a hand moving in the same way. The patterns of activation, recorded through electroencephalography, were not the same for people with ASD compared to control participants. While in this last group the area connected to the mirror system activated both when the people were observing the action and when they were performing it, this did not happen for people with ASD, whose mirror system only activated when they were performing the action. Other groups of researchers obtained the same kind of results using fMRI. Dapretto et al. (2005), and Williams et al. (2006) showed that during the performance of imitation tasks, children and adolescents with ASD present a reduced or no activation at all of the area corresponding to the mirror system, compared to controls. In the study by Dapretto et al. (2005), people with ASD were asked to imitate faces (expressing emotions) that they saw on pictures. In the imitation task, ASD people performed as well as people with typical development. Nevertheless, the area corresponding to the mirror system did not activate in the clinical population, so they imitated without

Background on Autism Spectrum Disorders

19

experiencing the emotion of the person they were imitating. The authors conclude: “the lack of mirror neurons activity during both the imitation and the observation of emotional expressions […] provides strong support for the hypothesis that early dysfunction in the mirror neuron system may be at the core of the social deficits observed in autism” (2005:30). Ramachandran & Oberman (2006:65) propose the image of broken mirrors to define autistic conditions and summarize the relationship between MNs and autism as follows: People with autism show reduced mirror neuron activity in the inferior frontal gyrus, a part of the brain’s premotor cortex, perhaps explaining their inability to assess the intentions of others. Dysfunctions of mirror neurons in the insula and anterior cingulate cortex may cause related symptoms, such as the absence of empathy, and deficits in the angular gyrus may result in language difficulties.

This account of ASD etiology, has been strongly criticized by proponents of cognitive theories of autism (Baron-Cohen 2008, among others). They point out the fact that some people with ASD have excellent imitation skills in certain areas including in language, as manifested in echolalia, imitation of accent and intonation, and memorization of dialogues, and this goes against the mirror neuron theory of autism. Nonetheless, the fundamental point that these critiques miss is that mirror neurons do not mirror everything; their functioning is crucially restricted to actions expressing intentions. Echolalia, and other kinds of related behaviors, do not represent acts of intentional communication, so cannot be taken as evidence of a functioning mirror system. Probably, the term ‘mirror’ is too broad, leading to a misinterpretation of what the nature of these unique neurons is: they are not merely mirrors reproducing everything indistinctly. They could be better defined as intention mirror neurons, meaning that they activate only in presence of an intention behind an action. Considering this, it can therefore be stated that the MNs theory of autism could be a plausible explanation of the causes of this condition. The point then, is to establish whether the MN system does not develop in people with ASD or whether it is just dormant and, if so, what can be done to “repair the broken mirrors” (Ramachandran & Oberman 2006:68). The possibility of reactivating a dormant mirror system encouraged several studies aimed at enhancing its functionality. As pointed out in Iacoboni (2008), there are several groups of researchers working on the reactivation of the mirror system in people with ASD through imitation-based treatments. Among them, Field, Sanders, Nadel (2001), Escalona et al. (2002), Ingersoll & Schreibman (2006), Ingersoll &

20

Chapter One

Gergans (2007), whose results represent an encouraging therapeutic outcome.

1.4 Cognitive Theories of Autism: Executive Functions, Weak Central Coherence, Theory of Mind ASD etiology has also been dealt with, and still is, from a different perspective than the purely medical one. Over the years, some cognitive theories of autism have been put forward to account for its causes. Studies have mainly revolved around three theories that describe cognitive dysfunctions in autism, and account for them in terms of deficits in executive functions (EEF), central coherence (WCC) or theory of mind (ToM). As Cummings points out (2009:118), the existence of three theories to account for the same problem posits a primacy problem to establish which one has identified the core deficit, but the problem of deciding on which theory has done so is that all three theories are supported by extensive experimental data.

1.4.1 Executive Functions The term ‘executive functions’ refers to a set of cognitive functions performed mainly by the frontal lobes. It includes planning, problem solving, concept formation, working memory, multi-tasking, attention shifting, abstract and inferential reasoning, impulse control, inhibition, and mental flexibility. People with ASD have been found to have deficits in several of these functions through the Wisconsin Card Sorting Test19 and the Tower of Hanoi Test20 that assess flexibility, working memory, and decision making skills (Ozonoff & McEvoy 1994). The first proponents of executive dysfunction as an explanation for autism were Pennington et al. (1997), who put forward the hypothesis of a severe deficit in the working memory of these people due to an abnormal development of the prefrontal cortex, which would cause disruption in the planning of complex behaviors, and a cascade of other impairments. This hypothesis has been supported by several studies. Lopez et al. (2005) found high correlation between executive dysfunctions and restricted repetitive behaviors typical of ASD patients. Similar results were obtained by Ozonoff et al. (2004) and Joseph et al. (2005). In this last study, executive dysfunctions were related to language impairment of people with ASD, and a positive correlation was found. This correlation was interpreted as the effect of a disrupted EEF system on use of language. Kleinhans et al. (2005) investigated the extent

Background on Autism Spectrum Disorders

21

to which poor performance on EEF tests is indicative of an EEF disruption or just evidence of impairments in more basic cognitive processes. Their group did not present an overall depressed EEF profile. Rather, their patients’ executive dysfunctions were mild and limited to the verbal domain. They concluded that these findings confirm the presence of a malfunctioning EEF system in the ASD population. Nevertheless, it does not appear clear whether this is as a consequence of ASD or as its cause.

1.4.2 Weak Central Coherence The expression ‘central coherence’ refers to the style of cognitive functioning of neurotypical people (Frith 1989). It means that typical brains have the tendency to work out meanings considering the ‘big picture’, their context, preferring the whole to the parts that compose information. Frith (1989) proposed that the autistic brain presents the reverse pattern of functioning and thus exhibits weak central coherence. She abandoned the deficit perspective, pointing out that people with ASD have a different, not deficient, cognitive style. They have the tendency to focus on parts of objects rather than wholes, present an extreme sensitivity to small changes in an environment, are superior at recalling and processing information that to typically developing children seems random21, and do not rely on contextual clues for the interpretation of ambiguous sentences (Jolliffe & Baron-Cohen 1999 a, b). Overlooking the context also applies to visual stimuli. Frith & Happé (1994) showed that people with ASD are less sensitive to optical illusions given that they are not influenced by the context. For example, when people with ASD are presented with a figure from the Navon Test22 (in Figure 1) and are asked what letter they see, they answer ‘A’23. Fig. 1-1 Navon Test A A A A AAAA A A A A

These differences in processing style would also account for language problems given that the integration of contextual information is a key factor for language understanding. The WCC theory thus proposes itself as the explanation of the cognitive differences observed in autism.

22

Chapter One

Nevertheless, Martin & McDonald (2003) point out that it is far from being considered the definitive explanation of this condition. Martin & McDonald (2003) take into account another clinical population, that of people with William’s Syndrome. This rare genetic disorder is characterized by the same kind of cognitive processing style as ASD people, namely privileging parts over wholes and attention to details, but it is at its extreme for language and social interaction abilities. People with William’s Syndrome manifest the use of affective prosody, and are able to decode social and emotional clues through empathy. It follows that, if WCC and the associated pattern of cognitive processing cause language and social impairment in autism, as put forward by its proponents, it must also result in an impairment in language and social skills in people with William’s Syndrome. Furthermore Ropar & Mitchell (1999), trying to replicate findings on visual illusions in people with ASD, found that, contrary to expectations from WCC, these people were susceptible to the illusions, showing that neglect of context is not a pervasive processing feature for these people. As Martin & McDonald conclude: There appears to be evidence for low-level central coherence deficits in William’s Syndrome, but only equivocal evidence for low-level WCC in Asperger’s and Autism. On the other hand, there is no evidence for pragmatic disturbance in William’s Syndrome and strong evidence for pragmatic disturbance in Autism. Clearly, the link between WCC and pragmatic ability on this evidence is problematic (2003:458).

1.4.3 Theory of Mind The picture in Figure 2 illustrates what is meant by theory of mind. Typically developing individuals, by the age of four, are able to infer that Sally has a mental state that differs from their own, and thus she will look for the ball in the basket and not in the box, where they know it actually is. They have developed the notion of ‘belief’ and, crucially, they are able to conceptualize that beliefs sometimes can be false. Earlier in a child’s development, some precursors of the mentalizing ability can be identified, such as joint attention, developed around 10-12 months, and the enjoyment of pretense play24, manifested by the age of 18 months. The hypothesis that this ability constitutes a characteristic of human and higher primates’ brains was put forward for the first time in 1978 by Premack & Woodruff: “In saying that an individual has a theory of mind, we mean that the individual imputes mental states to himself and others […] A system of inferences of this kind is properly viewed as a theory,

Background on Autism Spectrum Disorders

23

first because such states are not directly observable, and second, because the system can be used to make predictions, specifically about the behavior of other organisms” (1978:515). Fig. 1-2 False Belief Test25

ToM accounts for all of those tasks that people, and great apes, are able to perform because they attribute mental states to other conspecifics and, in doing so, they can not only make sense of others’ behaviors but also predict them. The capacity to attribute mental states to others is a key skill to develop social interaction and to understand others, their intentions, their beliefs, and their emotions. The failure to develop such a capacity was proposed to be the core cognitive deficit in autism as early as 1985 by Baron-Cohen et al. (1985), and is still one of the most accredited cognitive theories that account for this condition. Research about ToM dysfunctioníalso called mindblindness in autismí(Perner et al. 1989; Frith et al. 1991; Leslie 1991; Charman & Baron-Cohen 1992; Tager-Flusberg 1992; Happé 1994a,b; Baron-Cohen et al. 1997; Tager-Flusberg 1999; Peterson & Siegal 2000; Frith 2001; Rutherford et al. 2002; Kana et al. 2009; Senju et al. 2009; Lombardo et al. 2011) agrees that people with ASD are delayed in the development of theory of mind, and this results in the cognitive and linguistic abnormalities characterizing them.

24

Chapter One

These people have been found to fail the ‘false belief test’ showing that they are unable to conceive that people may entertain beliefs that are different from their own. They are also unable to lie about describing a situation even if instructed to do so, because they cannot allow someone else to have a false belief about it. Autistic children do not engage in pretense play for the same reason. The lack of this kind of play is actually one of the diagnostic criteria for ASD. In addition, they are seriously delayed in the development of joint attention, which is considered one of the earliest signs of the possibility for a child to manifest autism (see §1.2). Baron-Cohen et al. (1996) conducted a large-scale study on 18month-old infants looking at signs of joint attention and enjoyment of pretense play. They found out that a remarkable percentage of those infants who lacked these two abilities by that age were later diagnosed as autistic. ToM theory of autism also accounts for the linguistic deficits of autistic people, because being unable to attribute an intention to their interlocutor means that they are also unable to attribute a communicative intention to him/her. Happé (1993) found a positive correlation between ToM deficit and the ability to understand figurative, ironic, and sarcastic language. The same correlation has been found in traumatic brain injury patients (Levine et al. 1993) and right hemisphere damage patients (Winner et al. 1998), something that points to a very strong relation between ToM skills and pragmatic skills, even if it is not clear which one of the two enhances the other (see § 2.3). Nevertheless, the hypothesis that a ToM impairment is responsible for autism insurgence is not entirely plausible, because it only accounts for the social, emotive, and communication abnormalities, and does not explain the presence of restricted and repetitive behaviors, motor stereotypes, and superior rote memory skills. Baron-Cohen (2003, 2008) deals with this problem by proposing a ToM superset theory of autism: the empathizingsystemizing theory of autism. 1.4.3.1 The Empathizing-Systemizing Theory of Autism A possible account of cognitive and behavioral differences observed in ASD patients is the ‘extreme male brain’ theory of autism, also referred to as the empathizing-systemizing theory (hereafter E-S). According to this theory, autistic individuals have a very low empathy quotient and this deficiency cascades into all their other problems of which ToM impairment is just a component. On the other hand, they have a high systemizing quotient that accounts for all the areas in which they present

Background on Autism Spectrum Disorders

25

superior skills. As already stated, people with ASD perform very poorly on tasks which require inferring others’ emotions and responding adequately to emotive stimuli; but they outperform typically developing individuals in tasks based on systemizing in tests of intuitive physics, logic, and mathematics. Given that empathizing and systemizing are at two extremes of a continuum, this theory well explains the differences in empathizing-systemizing levels along the spectrum. Moreover, not only does it account for the cluster of social impairments, but it also accounts for the non-social ones, namely motor and behavioral stereotypes, narrow interests and resistance to change. As Baron-Cohen puts it: […] when you systemize, it is essential to keep everything constant, and only vary one thing at a time. That way you can see what might be causing what, rendering the world predictable. And to check the pattern or rule you have identified is correct or consistent, it is essential to repeat the sequence over and over again. (2003:65)

This theory therefore accounts for the classic behaviors present in ASD, even at the lowest levels of functioning, in terms of extreme systemizing. The examples in Table 1-1 are from Baron-Cohen (2003:66-68). Table 1-1. Examples of the empathizing-systemizing theory of ASD Systemizing Area Sensory

Motoric

Collectable Numerical

Behavior Classical Autism Asperger’s Syndrome Tapping surfaces Wearing the same clothes Letting sand run through every day one’s finger Insisting on the same Sniffing people foods each day Spinning round and Practicing sports moves round or back and forth Learning knitting Flicking a straw at high patterns speed in peripheral vision Collecting leaves Making lists and Learning the flags of catalogues different countries Obsessions with Rapid calculation of calendars or timetables prime numbers Memorizing birthdays or historical dates Solving math problems

26

Motion

Spatial Environmental

Social

Natural

Mechanical Vocal/Auditory/Verbal Action Sequences

Musical

Chapter One Watching the washing machine spin round and round Spinning the wheels of a toy car Naming shapes Obsessions with routes Insisting on objects being placed always in the same order Organizing videos on a bookshelf in a strict order Insisting on the same timetable at school Asking over and over about the weather Obsessions with taxonomies Learning to operate the VCR Building Lego models Echoing sounds Parroting sentences Watching the same video over and over again Repeating actions Playing a tune on an instrument over and over again Memorizing long sequences of musical notes

Analyzing when a specific event occurs in a repeating cycle Studying maps Developing drawing techniques Insisting that nothing is moved from its usual position in the room

Insisting on playing the same game over and over again Learning the Latin names of every plant and their optimal growing conditions Taking objects apart and reassembling them Imitating accents Watching the same film over and over again Mastering an instrument Analyzing the musical structure of a song

This cognitive theory seems plausible given its explanatory completeness. What it seems to lack, however, is a link to a possible neural substrate. In other words, it does not explain why people with ASD have a low empathy quotient. As Leslie (1987) proposes about ToM, and which is also applicable to E-S theory, there has to be a neural system to support the processing of specific information which does not depend on a particular modality26.

Background on Autism Spectrum Disorders

27

Considering the above discussion of MN system, which is thought to be the neural basis of empathy, the link between these two theoriesíToM and MNíappears quite clear, they seem to complete one another. A dormant MN system would cause the cognitive and behavioral abnormalities that are justified in the terms put forward by the E-S theory. This chapter has dealt with autism spectrum disorders from a medical and cognitive perspective. The next chapter challenges the cognitive theories of autism from a linguistic point of view with the aim of establishing their pragmatic adequacy to account for the communicative impairments of people with ASD.

CHAPTER TWO PRAGMATICS AND AUTISM SPECTRUM DISORDERS

This chapter firstly discusses pragmatic competence and its development in neurotypical people and secondly the topic of pragmatic impairment, focusing on people with ASD. Lastly, it presents an overview of studies about pragmatic impairment in this clinical population taking a critical stand towards research conducted in this field.

2.1 Introduction A constant feature of the language impairment of autistic people is a disruption of pragmatic skills. Even in cases of high functioning individuals who preserve intact phonology, morphology, and syntax they are reported to fail in the pragmatic use of language. As will be discussed in §2.6, ASD patients have been proven to have impairment in several areas of pragmatic competence among which are their understanding of non-literal language, inferential skills, comprehension of irony, and use of deixis. These findings have supported a view according to which people with ASD are characterized by the impairment of pragmatic skills in toto. Nevertheless, this conclusion seems quite premature given that pragmatic competence includes far more skills than those tested in these patients to the present date. In addition, considering that there is no agreement among pragmatists as to the boundaries of this discipline, let alone the boundaries and the working definition of its young branch of clinical pragmatics, it seems simplistic to put forward the hypothesis of a ‘pragmatic impairment’ without having more fully established what pragmatics is, what phenomena it includes and excludes, its possible neuro-cognitive substrates, and its development in neurotypical people.

30

Chapter Two

2.2 Pragmatic Competence in Neurotypical Individuals The fact that Ariel has dedicated two books27í with the last being published in 2010íto the question of defining pragmatics reveals how, after more than seventy years from its first definition (Morris 1938), a firm definition of this discipline remains elusive due to the heterogeneity of the phenomena analyzed by pragmatics. Moreover, this is also evidence of the lively debate around the definition and the object of study of this discipline. Such discussion is beyond the scope of this work; suffice it to say that what emerges from these recent works is the complexity of the field under discussion. Since Morris (1938:6) defined pragmatics as the “relation of signs to interpreters”, pragmatists have endeavored to establish its boundaries and what falls within the pragmatic domain. Levinson (1983)ía milestone in pragmatics studiesídedicates 35 pages to this issue, and suggests observing what practitioners do to define the field. From being the catch-all for semantic overflow (Bar Hillel 1971), to being downgraded to the performance (vs. competence) level by the generativists, pragmatics has affirmed a solid position both in theoretical and experimental linguistics. The scope of pragmatics broadened from the basic topics of presupposition, implicature, speech acts, reference, deixis, and definiteness, to discourse studies, the study of its interfaces with other linguistic levels and disciplines, and the relationship with cognition. As the term assumed more breadth, it brought about different perspectives on the discipline. Some scholars (Kasher 1991, for example) sustain the idea of pragmatics as a module, à la Fodor; others propose a weak version of modularity (Bara 1999; Tirassa 1999), which maintains the domain specificity but does not include the other Fodorian properties (mandatory operation, limited central accessibility, fast processing, informational encapsulation, shallow outputs, fixed neural architecture, characteristic and specific breakdown patterns, characteristic ontogenetic pace and sequencing). Sperber & Wilson (1986, 1995, 2002) propose the Relevance Principle28 to account for every phenomenon taking place during communication, and thus pertinent to pragmatics. According to Ariel (2010), the traditional view of pragmatics is one that identifies the discipline with its objects of study, what she defines as “big-tent pragmatics” (2010:§1.2). Hence, following this perspective, pragmatics could be defined as the discipline that analyzes speech acts, non-literal and implicit meaning, deixis, context influence, politeness phenomena, conversation, and discourse. On the contrary, what the author puts forward as a single criterion to define pragmatic competence is a comparative one.

Pragmatics and Autism Spectrum Disorders

31

She accounts for the differences between the domain of pragmatics vs. grammar in terms of a distinction between inference and code, according to which pragmatics is “the set of general language-independent principles which complete the picture, to determine how utterances are understood” (2010:98). Pragmatic skills correspond to inferential skills, namely those that allow speakers to bridge the gap between what is encoded grammatically and the meaning speakers want to communicate. This criterion encompasses the problem of listing the single skills that constitute pragmatic competence in neurotypical people. Thanks to this criterion then, the object of study and the boundaries of this discipline appear clearer29. Nonetheless, pragmatic neuro-cognitive substrates remain a topic for future investigation. Notwithstanding the complexity of the ongoing debate, all too often the stance taken towards pragmatics by researchers in the field of psychology and neurosciences is very simplistic. This is thoroughly illustrated in Cummings (2009:7): The notion of pragmatics that I want to adopt in this book is one that is more deeply rooted in language use than many practitioners and researchers in clinical pragmatics have tended to adopt. […] Even a brief survey of studies that have been conducted in the area of clinical pragmatics reveals a tendency amongst investigators […] to construe pragmatics in such broad terms that it is not clear what it is intended to exclude. In fact the term ‘pragmatics’ has now become coextensive in many (if not most) clinical studies with the notion of communication itself. […] Pragmatics is about the knowledge that allows a speaker to employ a linguistic utterance to achieve a certain communicative effect. The fact that other behaviors may attend the employment of this utterance should not detract from the centrality of the linguistic utterance to pragmatics30.

Going back to the definition proposed by Ariel, among the inferential skills that make up pragmatic competence is the one that allows people to understand presupposed and implicated meanings. The research presented in this book investigates the extent to which autistic people are able to draw inferences deriving from presuppositions and implicatures. Therefore, before presenting and discussing the results of this clinical population, it seems necessary to illustrate these two kinds of implicit meanings and the process by which typically developing individuals interpret them.

Chapter Two

32

2.2.1 Presuppositions The notion of presupposition originated in the field of philosophy and logic with the works of Frege at the end of the 19th century (Frege 1892). Investigating the nature of reference, he stated that there is always a presupposition of existence when certain linguistic expressions are used, as in the case of nouns that presuppose the existence of a referent for that noun. Consider: (1) John is a good student Using the name John, presupposes the existence of a referent for ‘John’. Frege also identified a fundamental property of presuppositions, namely the fact that they resist negation. In other words, if A presupposes B, then also non A presupposes B, as in (2): (2) a. The king of France is bald b. >> There exists a king of France c. The king of France is not bald In this example c., which is the negation of a., maintains the presupposition of existence in b. Indeed, the negation of the utterance containing the presupposition is a means to distinguish presuppositions from other kinds of inferences. The negation test has traditionally been considered a means to identify presuppositions. Apart from surviving negation, presuppositions also survive modal contexts. In (2), for example, the uttering of ‘it is possible that the king of France is bald’ would not have changed the presupposition in b. The example in (2) constitutes an instance of a debated issue. Given that there is actually no referent for the definite description in a., it is not possible to attribute a truth value to the utterance. Nonetheless, Frege pointed out that expressions such as the one in (2) maintain their meaning even if they lack a referent. What Strawson (1950) proposed is that expressions such as (2) cannot be judged as being true or false if the precondition of existence is not true, namely that there exists a king of France, a precondition that he called presupposition. Thus, for an utterance to be true or false, one has to first assume that its presupposition is true. Indeed, this is how presuppositions are defined in the Pragmatics Encyclopedia (2009:349): “propositions whose truth is taken for granted in an utterance and without which the utterance cannot be assigned a truth value”. Consider:

Pragmatics and Autism Spectrum Disorders

33

(3) John’s wife is pretty For (3) to be true (or false), it has to be presupposed that John actually has a wife. Elements that elicit this kind of inference are called presupposition triggers and are: definite descriptions, factive verbs, change of state verbs, iteratives, implicative verbs, possessive constructions, temporal clauses, cleft sentences, comparatives, counterfactual conditionals (in this order, examples 4 through 13): (4) The president of the United States is Afro-American >> There exists a president of the United States (5) John knows that Susan failed her test >> Susan failed her test (6) Mary stopped smoking >> Mary used to smoke (7) John failed his driving test again >> John has already failed his driving test (8) John did not manage to pass his driving test >> John tried to pass his driving test (9) After the birth of his son, John changed a lot >> John’s son was born (10) John’s car is old >> John has a car (11) It was John who got a son >> Someone got a son (12) John is a better driver than Susan >> John can drive (13) If John had passed his driving test, he would have been happy >> John did not pass his driving test

34

Chapter Two

Presuppositions started to receive linguistic interest in the 1970s. As inferences strictly depending upon the linguistic structure of an utterance, they were thought to belong to the domain of semantics. Nonetheless, it was soon noticed that they are sensitive to some extra-linguistic contexts, and as such, should belong to pragmatics. Indeed, presuppositions are cancelled when they are inconsistent with background knowledge, and when what is uttered is inconsistent with what is presupposed, in (14)’ and (15)’ respectively: (14) She cried before writing her testament >> She wrote her testament (14)’ She died before writing her testament ~>> She wrote her testament (15) You finally managed to open that door >> You tried to open that door (15)’ In a context in which a child has been playing with a ball inside the house, and his mother has been insisting for him to stop because he could have broken a window, if the mother utters: You finally managed to break that window! ~>> You tried to break that window31 According to the linguistic context, presuppositions can also be suspended, as in the example in (16): (16) She will regret having phoned John, if she ever does ~>> She phoned John The fact that presuppositions are sensitive to context and to background knowledge renders clear their belonging to the pragmatic domain. Moreover, as Stalnaker (1972:387) highlighted, “to presuppose a proposition in the pragmatic sense is to take its truth for granted, and to presume that others involved in the context do the same”. In other words, presuppositions can be defined as propositions whose truth a speaker takes for granted and, crucially, assumes that his/her listeners do the same. Consider: (17) A: John’s wife is pretty >> John has a wife B: Well, he is a handsome guy C: What?! I didn’t even know John got married!

Pragmatics and Autism Spectrum Disorders

35

In (17) A uses a definite description taking for granted that both B and C share a knowledge with him/her, namely the presupposition ‘John has a wife’. Of course sometimes speakers’ assumptions of the shared knowledge can be incorrect, as in the case of C who did not know John had a wife. What is fundamental to point out though, is the wording of the utterance in (17) that reflects the speaker’s assumptions of his/her hearers’ background knowledge. If A had known that C did not know that John had a wife, he/she would have uttered something like: (18) A: You know, John got married. I met him yesterday, his wife is pretty. The utterance in (18) makes explicit the presupposition that John has a wife because the speaker knows that his/her hearer does not have that notion in his/her knowledge, and thus provides him/her with that information. As pointed out by Cummings (2009), few neurolinguistic studies have explored the cognitive substrates of presupposition processing. Among these, Wetzel & Molfese (1992) conducted an ERPs32 experiment recording brain activity in neurotypical people during a listening task. The utterances contained either factive or non-factive verbs, and the results showed that both hemispheres are involved in the processing of presuppositions. The authors conclude that “pragmatic processing is neither a purely RH nor LH phenomenon, but rather a complex neurolinguistic effect that changes spatially across the brain and over time” (1992:302).

2.2.2 Implicatures Grice (1957, 1975) was the first to recognize the central role of inferential processing during utterance interpretation. In his words, a speaker means something by uttering x if he “intended the utterance of x to produce some effect in an audience by means of the recognition of this intention” (1957:385). In his attempt to provide an explanation as to how a logical term like ‘and’ for example, is interpreted in natural languages as ‘and then’, and wanting to define the difference between a logical inference and a pragmatic inference, he proposed the concept of implicature. The fundamental difference that Grice underlines is between ‘what is said’, the truth-conditional content of an utterance, and ‘what is implicated’, the meaning that the speaker wants to convey, which is processed by the hearer by means of reasoning. As Levinson (2000:171) puts it, “what is said is the input to the pragmatic reasoning responsible for

Chapter Two

36

output of implicatures: what is implicated is calculated on the basis of what is said”. To guide the pragmatic reasoning in the processing of an implicature, Grice proposes the Cooperative Principle as a rational principle aimed at making sense of what is uttered. As Cosenza (2002:172) puts it: “è per ‘mettere ordine’ nella variabilità dei significati occasionali non verocondizionali che le frasi possono avere, che Grice cerca di individuare pochi principi a un tempo semplici, generali e interconnessi nel modo più sistematico possibile che permettano, appunto, spiegazioni generali e sistematiche. Il cosiddetto “Principio di Cooperazione”, le cosiddette “massime conversazionali” e la teoria dell’implicatura conversazionale che ne discende possono essere visti nel loro complesso come frutto del tentativo di Grice di trovare un modo per governare l’indefinita variabilità dei significati occasionali33”.

The Cooperative Principle (Grice 1989:26) states: Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange. The principle is then articulated in Maxims of conversation, as follows (Grice 1989:26-27): Quality: Try to make your contribution one that is true. 1. Do not say what you believe to be false 2. Do not say that for which you lack evidence Quantity: 1. Make your contribution as informative as is required (for the current purposes of the exchange) 2. Do not make your contribution more informative than is required Relation: Be relevant Manner: Be perspicuous 1. Avoid obscurity of expression 2. Avoid ambiguity 3. Be brief 4. Be orderly

Pragmatics and Autism Spectrum Disorders

37

Some scholars have criticized the Cooperative Principle and the Maxims34. It should be highlighted however that Grice proposed the Principle, and the Maxims, as a rational35 way to interpret communicative exchanges, which provides “rationality based constraints on the expenditure of effort” (Horn, 2004:24). The Maxims were not proposed as prescriptive rules that speakers have to learn to follow; on the contrary, they represent the rational way to bridge the gap between the ‘said’ and the ‘implicated’. Seen from this perspective, the modernity of Grice’s proposal appears clear. Before cognitive sciences existed, Grice put forward a ‘rational principle’, what today would be called a ‘cognitive principle’ (in the sense of the way in which the mind works), to account for the interpretation of utterances. What he did was to underline the fact that the application of the Maxims is not restricted to conversation but can be applied more broadly to human behavior, thereby stressing the rationality, or cognitive validity, underlying it. It is difficult at this point not to notice the common points between these ideas and what Sperber & Wilson (1986) proposed as Relevance Theory (see next paragraph) arguing that a relevance-theoretic approach is more cognitively plausible than what Grice put forward. This is not to say that Grice’s notion of implicature is devoid of theoretical weaknesses. Nonetheless, as Horn (2004:25) concludes, “a suitably refined and constrained notion of conversational implicature remains at the heart of linguistic pragmatics”. .

Summing up, and leaving the theoretical debate aside, we could say that implicatures are a part of what is communicated without being part of what is ‘said’ by the speaker. They are pragmatic inferences derived via the tacit assumption, shared by the speaker and his/her audience, that what is communicated has to be true, informative, relevant, and clear36. Grice divided implicatures into conventional and conversational. The first ones are non-truth conditional inferences that do not derive from superordinate pragmatic principles, but are conventionally attached to certain lexical items and thus do not change according to the communicative context. Some elucidative examples are provided in (19) through (23): (19) She studied very hard for her exam, therefore, I think she’ll pass it [therefore engenders the implicature that studying is linked to passing an exam] (20) The wolf is a slim animal but it is very strong [but is used to implicate that what follows it contrasts with what precedes it]

38

Chapter Two

(21) Even John knew that they broke up [even engenders the implicature of ‘contrary to what might be believed’] (22) Susan is a successful businesswoman, moreover she is a very good mother [moreover indicates that the second proposition is something ‘in addition to’ the first one] (23) She looked at her daughter and, so, she calmed down [so introduces a consequence of the proposition that precedes it]. Grice (1989) identified the following properties characterizing conventional implicatures37: - Non calculability: being conventionally attached to lexical items, they are not derived from considerations of cooperation and rationality, but are given by convention; - Detachability: depending on the particular lexical item that triggers them, using a synonym of that item might cause the loss of the implicature; - Non defeasibility: they are not cancellable. Conversational implicatures on the other hand, have to be inferred according to the communicative context, and, depending on their contextdependence, are divided into generalized (henceforth GCIs) and particularized (henceforth PCIs). GCIs are derived assuming that the context of the utterance is an ordinary conversation. On the contrary, PCIs are highly context-dependent and can be interpreted only in reference to a ‘particular’ context. An example of generalized conversational implicature are scalar implicatures. They are generated by quantifiers that, according to the Maxim of Quantity, receive a pragmatic interpretation. Consider: (24) Some elephants have trunks This utterance is true from a logical point of view given that, as a logical operator, some is interpreted as ‘some, possibly all’. Under the Maxim of Quantity however, this interpretation is not possible. If the contribution has to be as informative as required, the choice of a weak term like some pragmatically means that the interpretation of ‘possibly all’ is ruled out. Thus, the pragmatic interpretation of some is ‘some, not all’. This is what was found by Noveck (2001) who conducted an experiment

Pragmatics and Autism Spectrum Disorders

39

with adults, and found that in 59% of the cases the utterance in (24) was considered to be false. Scalar implicatures are generated by any kind of ordered scale. For example uttering ‘they have two children’ +> ‘they have exactly two children’ referring to the scale of numbers or, also, ‘the coffee is warm’ +> ‘the coffee is not hot’ following the scale of ‘warmth’. Another example of a generalized conversational implicature could be: (25) She went home and had dinner Following the fourth Sub-Maxim of Manner, it has to be assumed that the uttering order is not random but conveys a meaning. Thus, this utterance is interpreted as ‘she went home and then had dinner’. On the other hand, the example in (26) is an instance of PCI which, contrary to GCIs, requires the listener to take into account the ‘particular’ context of the conversation in order to be drawn. (26) Context: A is reading, B turns on the radio A: I’m studying +> could you turn off the radio? The utterance ‘I’m studying’ only in this particular context means ‘could you turn off the radio?’. Grice identified implicatures:

four

properties

characterizing

conversational

- Defeasibility: contrary to conventional implicatures which cannot be cancelled given that they are conventionally attached to a lexical item, conversational implicatures can be cancelled depending on background knowledge or contextual clues. Consider the example reported by Huang (2009:206): (27) John and Mary bought a house in London +> John and Mary bought a house in London together, not one each (28) The Americans and the Russians tested an atom bomb in 1962 ~ +> The Americans and the Russians tested an atom bomb in 1962 together, not one each.

Chapter Two

40

Here the implicature in (28) is cancelled through background knowledge about the fact that the Americans and Russians could not have been testing bombs together in that year. This never happens, though, with conventional implicatures given that they are drawn neither because of prior background knowledge, nor because of the particular context of the communicative exchange. - Non detachability: conversational implicatures are attached to the semantic content of the utterance, and not to its linguistic form, and so are not detachable from the utterance; on the other hand conventional implicatures can be detached from the utterance given that they depend on a particular linguistic construction, as (29) and (30) illustrate: (29) In a context in which it is clear to the participants that the uttering of a. is an ironical comment, and, thus, conveys the implicature in b., even the uttering of c. and d. would still trigger the implicature in b. (Levinson 1983:127) a. b. c. d.

He is a genius +> He is an idiot He is so smart He has an outstanding intelligence.

(30) On the other hand, changing the linguistic form of the utterance in a., and substituting the item that triggers the conventional implicature as in c., the implicature is lost: a. b. c.

He is slim, but strong +>> being slim is in contrast with being strong He is slim and strong.

- Calculability: conversational implicatures are derived via the Cooperative Principle and the Maxims, they are not part of the conventional meaning of lexical items and, as such, they have to be calculated, contrary to conventional implicatures which are conventionally attributed a meaning. Grice (1975) proposes a general inferring scheme that involves the followings steps: 1. P said p

Pragmatics and Autism Spectrum Disorders

41

2. There are no reasons to believe that he is not conforming to the Cooperative Principle (CP), or its Maxims 3. Uttering p and respecting the CP and the Maxims, he has to think that q 4. P knows that it is a mutually shared knowledge that being cooperative involves meaning q 5. P has given no reasons to prevent the interlocutor form thinking that q 6. Therefore, P wants the interlocutor to think that q, thus, he implicated q. - Non conventionality: conversational implicatures are not coded, that is, they are not part of the conventional meaning of linguistic expressions, as opposed to conventional implicatures. Consider (31) and (32): (31) A: Would you like some coffee? B: Coffee would keep me awake +> No (31)’ A knows that B does not want to sleep because he has decided to study the entire night. A: Would you like some coffee? B: Coffee would keep me awake +> Yes (32) He is an Englishman, therefore he is brave +>> being an Englishman is connected in some way to being brave While (31) and (31)’ show the non-conventionality of the implicature triggered by the utterance of B, there are no other contexts in which ‘therefore’ in example (32), could mean something different. To these four properties identified by Grice, three others can be added: - Reinforceability: conversational implicatures can be reinforced, that is to say made explicit, without the utterance resulting in being redundant as (34) (Huang 2009:207) shows38. (33) The coffee is warm +> The coffee is not hot (34) The coffee is warm, not hot

42

Chapter Two

- Universality: being motivated, rather than established by a convention, conversational implicatures tend to be universal39, that is they tend to be triggered by the same kinds of content, in different contexts and different languages. - Indeterminacy (Levinson 1983): conversational implicatures are sometimes difficult to determine, given that different implicatures may be triggered by the same expression, as in (35): (35) Jason is a machine +> he is cold +> he works a lot +> he is efficient +> he is not creative. As suggested by Horn, “pragmatic principles and conventions do as much work when they are apparently violatedíwhen speaker S counts on hearer H to recognize the apparent violation and to perform the appropriate contextual adjustmentías when they are observed or ostentatiously violated” (2004:8). Indeed, conversational implicatures can be also generated by an explicit flouting of the Maxim which itself signals that an inference has to be drawn. An example of this might be (36) from Huang (2009:207): (36) John: Wasn’t that a boring lecture? Mary: Did you remember to call Helen yesterday? +> The lecturer is standing behind you and he may hear what you said Even though Mary appears to be flouting the Maxim of Relation, the Cooperative Principle still holds, so Mary’s utterance is interpreted by John as conveying an implicature. The example in (36) is an instance of another distinction that Grice proposed, the one between standard and non-standard implicatures, not to be confused with the distinction between conventional and conversational implicatures. Standard implicatures are those derived by the speaker following the Maximsíno Maxim is violated or at least it is not clear that any Maxim is violated; the violation of a Maxim is to be explained by the supposition of a clash with another Maxim (Group A and B implicatures, in Grice’s terminology)íwhile non-standard implicatures derive from the explicit exploitation of a Maxim for communicative purposes (Group C implicatures, following Grice’s terminology). Examples of non-standard

Pragmatics and Autism Spectrum Disorders

43

implicatures are those derived from ironic expressions, tautologies, rhetorical questions, and some kinds of metaphors. Implicatures have received more attention from an experimental point of view compared to presuppositions. Sanford & Moxey (2004) analyzed interpretation and choice of quantifiers. They showed that both interpreting and choosing a quantifier depend on the perspective of the speakers’ or hearers’ beliefs. For example, the quantifier not many is interpreted as connected to the speakers’ expectations of the hearers’ belief, while few appears connected only to what the speakers believe. The authors also underlined the problems of connecting experimental data to formalisms, and the need to establish “sensible links between the insights of different disciplines” (referring to experimental linguistics and psychology) (2004:134). Scalar implicatures were also investigated by Bezuidenhout & Morris (2004) with an eye-tracking experiment. They confirmed that a quantifier like some is given the pragmatic interpretation of ‘not all’ monitoring the gazing time of sentence pairs like the ones presented here in (37): (37) Some books had color pictures. In fact all of them did, which is why the teachers liked them. The books had color pictures. In fact all of them did, which is why the teachers liked them. The gazing time on all was longer for the sentences containing some compared to the pair without it. This confirmed the hypothesis of a pragmatic interpretation of the quantifier, which was re-analyzed after all was read. Noveck & Cheveaux (2001) analyzed the pragmatic interpretation of and in cases in which it triggers ‘and then’ both in adults and children. They found that 71% of adults interpreted it as ‘and then’, while this did not appear to be true for 7-year-old children who assigned it a pragmatic interpretation only in 15% of cases and for 10-year-olds for whom the percentage was 37%. Similar findings were reported by Noveck (2001) and Chierchia et al. (2004), who point out the preference for a logical over pragmatic interpretation of quantifiers and logical operators in children, pointing at what the last authors call ‘Pragmatic Delay Hypothesis’ (2004:297). This hypothesis accounts for children’s logical interpretation of quantifiers in terms of a lack of the Gricean Maxim of Quantity related to a developmental factor. To test the validity of this proposal, the latter authors conducted an experiment with children ranging in age from 3;2 to

44

Chapter Two

6;0. “After a story in which all farmers decided to clean a horse and a rabbit, among their animals, two puppets each provided a different description of the story; one sentence contains the disjunction operator or and the other the conjunction operator and. - Every farmer cleaned a horse or a rabbit. - Every farmer cleaned a horse and a rabbit. Children were asked to reward the puppet that ‘said it better’” (2004:297). What they found was that, if provided with two alternative descriptions of the same situation, where one was more pragmatically appropriate than the other, children were able to identify the more appropriate one in 93% of the cases. Chierchia et al. (2004:298-299) propose a ‘Processing Limitation Hypothesis’, according to which (…) children have the relevant semantic and pragmatic knowledge needed to interpret sentences including scalar items, that is, they have Grice’s Maxim of Quantity, but they are indeed unable to calculate an implicature because of (memory system) processing limitations.

This hypothesis is confirmed by Feeney et al. (2004), who examined the interpretation of some in 7-8 year-old children, and found that they were able to associate the pragmatic meaning of ‘not all’ in pragmatically rich contexts. They used a storyboard with photographs of a girl interacting with some or all of a range of objects (e.g. sweets), after her mother had told her not to do so. The girl is thus motivated to hide the truth from her mother when asked about her actions. Participants had to judge the girl’s statements describing what she had done, as true or false. Results of this experiment confirm that children are sensitive to scalar implicatures, especially when these can be computed on the basis of pragmatically rich contexts. They were able to identify statements in which the girl was lying (e.g. she had eaten all the sweets, and uttered ‘I ate some of the sweets’). The authors conclude that “children’s sensitivity to the scalar implicature associated with some is largely dependent on context”. Scalar implicature processing in neurotypical individuals has also been analyzed through ERPs investigations. Noveck & Posada (2003) highlighted that the pragmatic interpretation of a scalar term requires more effort, in terms of reaction time, than logical interpretation, and thus claimed that these findings are consistent with a relevance theoretic (see next paragraph) perspective on scalar implicatures, according to which scalar items are first attributed their logical meaning and, then, given that this meaning does not satisfy expectations of relevance, they are given their pragmatic interpretation.

Pragmatics and Autism Spectrum Disorders

45

Noveck et al. (2004), on the other hand, analyzed implicature processing through fMRI40. They tested the comprehension of syllogisms involving conditional reasoning with the two forms of Modus Ponens (the proposition ‘if p then q’ being true, and p being true, results in q being true as well) and Modus Tollens (the proposition ‘if p then q’ being true, and q being false, results in p being false as well). The authors report the activation of the left parietal frontal region during conditional inferencing tasks in both forms, with a greater activation in the task requiring inferences of the Modus Tollens type. The findings illustrate that language areas are not the only ones involved during inference processing.

2.3 Relevance Theory Sperber & Wilson (1986, 1995, 2002, 2004) and Wilson & Sperber (1991) simplify Grice’s framework, proposing a single cognitive principle that unifies the Maxims and the Cooperative Principle, i.e., the Principle of Relevance (1991:382): We assume that a speaker who thinks it worth speaking at all will try to make his utterance as relevant as possible. A hearer should therefore bring to the processing of every utterance the standing assumption that the speaker has tried to be as relevant as possible in the circumstances. It is this assumption that we call the principle of relevance.

They defend the Gricean approach according to which pragmatic interpretation is “ultimately an exercise in mind-reading, involving the inferential attribution of intentions (given that) an essential feature of most human communication […] is the expression and recognition of intentions” (2002:3). Nonetheless, from a relevance theoretic perspective, the inferential processing takes place following the first, or Cognitive, Principle of Relevance, which states that “human cognition tends to be geared to the maximization of relevance” (2004:610). Hence, the expectation about utterances being relevant is raised not because the speaker is expected to obey the Cooperative Principle and its Maxims, but because human cognition is intrinsically relevance-oriented and, thus, points at the maximization of relevance. Therefore, implicit meaning processing does not derive from the recognition of the violation of any principle or Maxim given that, because of its cognitive nature, the principle of relevance cannot be voluntarily violated to achieve a communicative effect. Following this perspective, implicit meanings are just an expression of being relevant according to the circumstances of the communicative exchange, and are processed according to the cognitive

46

Chapter Two

background of the first Principle of Relevance, which allows humans to predict and interpret the mental states of others. Moreover, the authors propose the notion of ostensive inferential communication (1995:50-54) which involves the production, on the part of the speaker (to whom they refer to as ‘communicator’), of an ostensive stimulus aimed at attracting the audience’s attention. The production of an ostensive stimulus itself induces the audience to presume that it is a relevant stimulus and, as such, worth processing. This is illustrated by the second, or Communicative, Principle of Relevance: “every ostensive stimulus conveys a presumption of its own optimal relevance” (2004:612). Therefore, the comprehension process, independently of the explicitness of the message, is explainable in terms of construction of a hypothesis about the speaker’s meaning which satisfies the presumption of relevance. Relevance Theory is undoubtedly one of the most prominent theories in pragmatics. However, following Sbisà (2007), in the present work the Gricean framework will be maintained. I agree with Sbisà that the Principle of Relevance is an overly bare substitute for the Cooperative Principle and its Maxims, because it “reduces all the properties of ideal cooperative conversation to relevance alone, and impoverishes the range of argumentative strategies supporting implicit meanings”41 (2007:113). Furthermore, the Gricean framework allows us to account for comprehension of implicit meaning in ASD people in a more detailed and articulated fashion compared to the one offered by Relevance Theory.

2.4 Pragmatic Development Notwithstanding the abundance of research about pragmatic competence and its development, an agreement about a developmental paradigm is still missing in the field of theoretical pragmatics. While for the other linguistic levels general agreement about the acquisition stages has been achieved42, as Bara et al. (1999:522) point out: Currently, no single theory covers systematically the emergence of pragmatic capacity. We do not have a unitary account of the major phenomena […] Nor do we have a protocol by which to assess the normal stages at which a child is expected to produce and comprehend the different kinds of speech acts. Important as they could be, it is simply not possible to study deficits in communication without a comparable basis in normal development43.

Developmental pragmatics has been the focus of research throughout the years, but as Cummings (2009:40) underlines:

Pragmatics and Autism Spectrum Disorders

47

The less advanced state of our knowledge of developmental pragmatics compared to other aspects of language development has had serious consequences for the study of disordered pragmatics. Clearly defined accounts of the acquisition of syntax and phonology in normally developing children have provided investigators with a framework within which to interpret findings of syntactic and phonological impairment in language-disordered children. In pragmatics no such framework exists. This is as true today as it was nearly thirty years ago when the first clinical studies of pragmatics began to emerge. The result has been a proliferation of clinical pragmatics study, the findings of which are poorly understood both on their own terms and in relation to normal pragmatic development. In the absence of a theory of developmental pragmatics, or even a clearer account of the chronological acquisition of key pragmatic skills, what we are left with is a burgeoning group of studies which is powerless to explain findings of pragmatic disorder and which is cut off from almost everything else in language study44.

This is not to detract from research in the field of pragmatic acquisition and development. It should be kept in mind, though, when conducting investigations in the field of clinical pragmatics.

2.5 Pragmatic Impairment The definition and boundaries of pragmatic impairment are as debated as the definition and boundaries of pragmatics itself. As pointed out by Perkins (2007:11): Quantifiable definitions of pragmatic disability are far more common (compared to those of pragmatic ability), but typically vague. In most accounts, a range of supposedly normal (or abnormal) pragmatic behaviors are simply given, and individuals are described as being pragmatically impaired if either a sufficient number of behaviors are (or are not) observed or their performance of the behaviors meets (or fails to meet) specified criteria of acceptability or of frequency within a given time frame or situation.

To show the heterogeneity of what is analyzed as ‘pragmatic’, Perkins (2007:12) provides a table with three pragmatic impairment checklists and the category groups of analysis for pragmatic competence. The three checklists were devised by Prutting & Kirchner (1983), Penn (1985), and Bishop (1998) and include the following skills:

48

Chapter Two

Prutting & Kirchner (1983): speech acts, topic, turn taking, lexical selection/use across speech acts; stylistic variation; intelligibility and prosodics; kinesics and proxemics; Penn (1985): response to interlocutor; control of semantic content; cohesion; fluency; sociolinguistic sensitivity; Bishop (1998): inappropriate initiation; coherence; stereotyped conversation; use of conventional context; conversational rapport. Research on pragmatic impairment began before these checklists were devised. In the late 1970s a study by Heilman et al. (1975) on patients with Right Hemisphere Damage (hereafter RHD) encouraged research on communication competence as a distinct component compared to the grammatical one. These RHD patients had problems in interpreting emotional prosody, though maintaining unimpaired linguistic functioning. They also had difficulties in modulating their speech intonation thereby failing to transmit affective elements of language. Myers (1979) was the first study conducted on discourse level disorders in RHD patients. She described them as characterized by “inappropriate verbal output” (1979:43). Ross (1981) showed that RHD patients presented a severe impairment in the affective components of prosody and gestures, while the syntactic and semantic components were intact. These patients lamented an inability to insert affective and attitudinal variation into their speech, and gestural behavior, all the while insisting that they could feel and experience emotions inwardly. However, their monotone voices and lack of gestures gave the impression that they had a flat affect, and a lack of emotional responsiveness. These works paved the way to hypothesizing a cerebral component for pragmatics different from the one controlling morpho-syntactic competence and located in the right hemisphere. It was with Rapin & Allen (1983), that pragmatics entered officially into the description of a clinical condition acquiring diagnostic significance. These authors coined the term semantic-pragmatic syndrome to describe a developmental disorder characterized by an impairment in communication that did not derive from any deficit in structural language. This impairment included verbosity, poor conversational skills, atypical word choice, and word finding deficits. Bishop & Rosenbloom (1987) and Bishop & Adams (1989) modified Rapin & Allen’s term to semantic-pragmatic disorder. They described cases of children whose conversational exchanges were considered inappropriate for “a wide range of semantic, syntactic and pragmatic peculiarities […] they frequently misunderstood the literal or implicit meaning of adult utterances and violated normal rules of exchange structure […] they tended to provide the listener with too much or too little

Pragmatics and Autism Spectrum Disorders

49

information” (1989:241). However, after Frith (1989) had put forward the idea of the spectrum to describe autistic conditions, Aarons & Gittens (1993) noted that children diagnosed as affected by a semantic-pragmatic disorder actually fell into the autistic spectrum, and proposed to use the term semantic-pragmatic disorder as a strictly descriptive, rather than diagnostic, label. Since the 1990s, pragmatic impairment and its possible neuro-cognitive substrate has increasingly become the focus of research45 (Kasher 1991; Paradis 1998; Eisele et al. 1998; Kasher et al. 1999; Nicolle & Clark 1999; Stemmer 1999; Bara & Tirassa 2000; McDonald 2000; Stemmer & Schönle 2000; Zaidel et al. 2000; Carston 2002; Carston et al. 2002; Sperber & Wilson 2002; Martin & McDonald 2003; Lohmann & Tomasello 2003; Hatta et al. 2004; Tompkins et al. 2004; Bartels-Tobin & Hinckley 2005; Botting & Adams 2005; Harris et al. 2005; Cummings 2007; Perkins 2007; Cummings 2009, 2014). As for the other levels of linguistic competence, the difference between acquired and developmental impairments has to be highlighted for pragmatics as well. The first category refers to brain damage caused by a traumatic event, ischemic infarction, cerebral hemorrhage, or tumor in brains that were functioning normally before those events. Developmental impairments, on the other hand, refer to cases in which the linguistic system never gets fully formed during the neural and cognitive development of the individual.

2.5.1 Acquired Pragmatic Impairment Since Paradis (1998) put forward the idea of pragmatic competence being located in the right hemisphere, RHD has been the most investigated among the acquired conditions that affect pragmatics46. RHD patients have been reported to be impaired in discourse production and comprehension (Stemmer 1999; Tompkins et al. 2000; Bartels-Tobin & Hinckley 2005; Lehman 2006), non-literal language understanding, including inferencing skills, comprehension of idioms, proverbs, and metaphors (Brundage 1996; Eisele et al. 1998; Kasher & Batori 1999; Happé et al. 1999; Zaidel et al. 2000; Hatta et al.2004; Rinaldi et al. 2004; Tompkins et al. 2004; Papagno et al. 2006), and comprehension of irony and sarcasm (McDonald 2000; Cheang & Pell 2006). Moreover, as Cummings (2009) points out, RHD patients also present difficulties in processing information about the emotional state, intentions, and beliefs of the speaker. They fail to use prosody to alert listeners to changes in discourse structure, and are characterized by a reduced facial expressivity during conversation.

50

Chapter Two

The most relevant studies for the present work are those investigating inferencing (dis)ability in RHD patients. Eisele et al. (1998) analyzed presuppositions and implications of truth related to factive vs. implicative verbs, and to lexical vs. syntactic negation, in children with RHD compared to those with Left Hemisphere Damage (hereafter, LHD), and control individuals. RHD children showed a “[…] more ‘semantic’ deficit; one limited to computing implication, but not presupposition of truth, and one characterized by a greater deficit in processing lexical, as opposed to syntactic, forms of negation. (On the other hand) LHD appears to have resulted in a ‘grammatical’ deficit, characterized by an inability to compute negation scope (lexical or syntactic), and consequently, an inability to interpret presupposition or implication” (1998:391). Kasher et al. (1999) and Zaidel et al. (2000), investigated the comprehension of conversational implicatures in both LHD and RHD patients, relating to basic speech acts (assertion, question, requests, and commands; hereafter, BSAs) as well as comprehension. Both clinical populations showed impairment in the processing of implicatures. LHD patients though, showed a higher correlation between impairment in the processing of both implicatures and BSAs. The authors conclude that “both cerebral hemispheres appear to contribute to the same degree to the processing of implicatures (even though) this does not imply that LH and RH process implicatures in the same way” (1999:587). Happé et al. (1999) showed that RHD patients’ inferential disability is restricted to inferences involving mental states attribution as opposed to those involving the conceptual relation of sentences. LHD patients on the other hand present the reverse pattern. These findings suggest the presence of two different inferencing mechanisms (social vs. conceptual) that are situated in different parts of the brain. One is dedicated to the processing of inferences that require mental state attribution, and the other to nonmental inferences, namely those that relate sentences conceptually. The authors concluded that the difficultly with inferences involving mental state attribution is specific to RHD and that, crucially, this suggests that it is a different mechanism compared to the general non-mental inferencing ability located in the right hemisphere. Hatta et al. (2004) investigated the comprehension of indirect requests and refusals in RHD and LHD patients. They found that both populations were impaired in implicature processing, and also that their performance in the understanding of indirect refusals was poorer compared to that of indirect requests. Moreover, performance of tasks relating to indirect refusal was also comparatively inferior in people with LHD compared to those with RHD. The findings seem to suggest, as argued by the authors,

Pragmatics and Autism Spectrum Disorders

51

that both hemispheres are involved in the processing of implicatures, and that there is not a general system governing implicatures considering the different performances on indirect requests and refusals. Tompkins et al. (2004) analyzed bridging inferences in RHD patients during text comprehension. “A bridging inference links a referring expression to an intended referent that is neither present in the environment nor mentioned in the ongoing discourse but that is nevertheless inferentially identifiable” (Noveck & Sperber, 2004:11). As these last authors suggest, an example of bridging inference for the sentence a. in (38) is the sentence b.: (38) a. John walked into a room. The window was open. b. The room had a window They showed that, even though characterized by discourse comprehension difficulties, which lead to the activation of contextually incompatible interpretations, these people “do activate information that is semantically related to targeted BIs (bridging inferences), in text conditions designed to be less cognitively demanding” (2004:1392). The research presented so far seems to suggest that RHD patients are characterized by an inferencing disability connected to ToM skills47 (implications deriving from factive verbs, conversational implicatures, inferences deriving from mental states attribution, indirect refusals and requests). Indeed, this clinical population has been proven to show ToM impairment as well48. For instance Surian & Siegal (2001) found that RHD patients performed poorly on ToM tasks, significantly, when these were presented verbally and not when presented visually. The authors conclude “their difficulties on tasks devised to test ToM understanding may stem from impaired visuospatial buffers and working memory processes required for pragmatic competence rather than a fundamental representational deficit” (2001:230).

2.5.2 Developmental Pragmatic Impairment Developmental language impairments are connected to developmental disorders. These disorders affect children’s development of cognitive (and sometimes motor) abilities retarding, and usually preventing, the formation of a typically functioning neuro-cognitive system. Those connected with impairments of linguistic functions, and more specifically pragmatic functions, could be divided into four sub-groups: Specific Language Impairment (SLI), Disorders deriving from Mental Retardation (MR),

52

Chapter Two

Emotional and Behavioral Disorders (EBD), Autism Spectrum Disorders (ASD). The diagnosis of SLI applies to children who present a typical cognitive development with the exception of linguistic functions. In SLI children, conditions that may affect normal language development like craniofacial abnormalities, hearing loss, mental retardation, or socialization impairment are absent. SLI children show inferior receptive and expressive language skills without any clinically significant neurological impairment. The prevalence of SLI is higher in males than females, and is most common among kindergarten children. Even though the most frequently observed linguistic impairments in SLI are in morphosyntax and lexical semantics49, it has recently been proposed that SLI children also have problems in the pragmatic use of language, which are not a mere consequence of structural language problems as was once thought, but rather depend on a more general inferencing deficit (Bishop 2000; Bishop et al. 2000; Botting & Adams 2005). Bishop (2000), and Bishop & Norbury (2002) proposed the term ‘pragmatic language impairment’ (PLI) to describe SLI children who only present problems in the pragmatic component of language and, according to them, do not fall within the autistic spectrum. However, the relationship between SLI, PLI, and ASD is not clear yet. Thus, it is important to stress that research is needed to establish the boundaries between these three conditions, and to clarify whether PLI is a subset of SLI, or could be a form of HFA or AS. As underlined by Cummings (2009), even though more research is needed, pragmatic impairment has also been shown to be present in patients affected by: - mental retardation in: Williams Syndrome, Fragile X syndrome, Down’s Syndrome, Prader-Willi Syndrome, Angelman Syndrome; - emotional and behavioral disorders: Attention Deficit Hyperactivity Disorder (ADHD), selective mutism, and conduct disorders. ASD has certainly been the most investigated clinical population with regard to pragmatic impairment from a developmental perspective.

2.6 Pragmatic Impairment in Autism Spectrum Disorders When autism was identified as a disorder in the 1950s, one of the three features considered central to its diagnosis was the presence of communication deficits. Research on communication dysfunction in this clinical population started quite early. Tager-Flusberg (1981:51) affirms

Pragmatics and Autism Spectrum Disorders

53

that “many researchers and clinicians have pointed out that one of the major impairments in autism is in the area of communicative competence”. Since then, a great deal of research about the extent and the characteristics of pragmatic impairment in ASD has been carried out: Clark & Rutter (1981); Loveland & Landry (1986); Bishop (1989); Stone & CaroMartinez (1990); Tager-Flusberg et al. (1990); Loveland & Tunali (1991); Eales (1993); Ozonoff & Miller (1996); Surian et al. (1996); Dobbinson et al. (1998); Joliffe & Baron-Cohen (1999a,b); Tager-Flusberg (1999); Dennis et al. (2001); Kjelgaard & Tager-Flusberg (2001); Adams et al. (2002); Norbury & Bishop (2002); Emerich et al. (2003); Martin & McDonald (2003); Rapin & Dunn (2003); Charman (2004); Martin & McDonald (2004); Tager-Flusberg (2004); Kuhl et al. (2005); Wang et al. (2006); Lokusa et al. (2007); Colle & Baron-Cohen (2008); Surian & Siegal (2008). Even though only a part of the research concerns the comprehension of implicit meaning in people with ASDíwhich is the topic relevant to the present workía more complete picture of the pragmatic impairment in this clinical population will be presented with a twofold aim. First, to give an overview of which pragmatic components are affected and to what extent; and second, to take a critical stance towards some of these studies. The following section focuses therefore on these studies. The order of presentation is strictly chronological.

2.6.1 Discussion of previous research Stone & Caro-Martinez (1990), analyze spontaneous conversations of 30 autistic children aged from 4 to 13 years. They observed these children during their everyday school activities recording their conversations. The authors report both qualitative and quantitative differences characterizing these children’s conversations as compared to typically developing peers, and peers with mental retardation. Children at higher levels of functioning used speech mostly for requesting and giving information. However, as the authors point out, the most common “communicative event consisted of a child directing a motoric form of communication toward the teacher for the purpose of achieving an environmental end (i.e. requesting something) or attracting attention to him or herself” (1990:449). This study appears to be flawed for different reasons. First, the age range is too low to assess communication (dis)abilities. As is well known, pragmatic competence is fully acquired much later, compared to the other linguistic competences, and thus dysfunctions found in the clinical population might be biased by a developmental factor, rather than depending on the disorder itself.

54

Chapter Two

Second, observing an absence in the use of certain skills does not mean that people do not have that skill, or do not comprehend what derives from those skills when used by someone else. Third, crucially, this study considered mostly non-verbal communicationías the authors put it “motoric form of communication” (1990: 449)íwhich, as pointed out in §2.2, is an incorrect methodology in investigations about pragmatic competence, for which the linguistic component is pivotal. Loveland & Tunali (1991) compare Down’s Syndrome (DS) and HFA people’s responses to social scripts. These people, aged from 5 to 27 years, were told a story in which, during a tea party, someone stole one of the examiners’ wallet. DS people gave a significantly higher percentage of sympathetic answers, and relevant suggestions. HFA, on the other hand, appeared more concentrated on the tea party itself, neglecting the examiner’s emotive state. The authors conclude that the lack of relevant comments about the experimenter’s experience seems to be connected to HFA people’s inability to empathize, rather than to a more specific communication impairment. Eales (1993) presents a noteworthy analysis of conversation samples from adults who in their childhood had received a diagnosis either of autism, or receptive language disorder. He points out that individuals diagnosed as autistic present a more severe pragmatic impairment, compared to the other group, which is not attributable to a general impairment of language skills. Furthermore, he draws a distinction between execution and communicative intention. According to the author, impairment of execution corresponds to the production of sentences that violate Grice’s Maxim of Manner (the only Maxim linked to the linguistic form of the utterance), while impairment of communicative intention corresponds to the violation of the Maxims of Quantity, Quality and Relevance. What seems to characterize people with ASD in this study is only an impairment in the second competence, suggesting an “assimilation of pragmatic abnormalities in autism to social-cognitive impairment” (1993:611). Conversational Maxims and their violations were also analyzed by Surian et al. (1996). The authors presented ASD children (mean age 12;11, SD 2.4) with tape-recorded conversational exchanges in which the violation of a Maxim was present; they were then asked to indicate which character said something “funny or silly” (1996:62). The ASD group’s results were compared to a control group of SLI and typically developing children matched for verbal age. The authors conclude “most children with autism are ‘deaf’ to violations of Gricean Maxims. Such a deficit is itself likely to be the result of an impairment in mindreading” (1996:67). It is

Pragmatics and Autism Spectrum Disorders

55

interesting, though, to consider the conversational exchanges used in this study (1996:70). (39) 1. Violation of the first Maxim of Quantity: How would you like your tea? With milk / *In a cup 2. Violation of the second Maxim of Quantity: What pets do you like? I like puppies and kittens / *I like rabbits which are animals 3. Violation of the first Maxim of Quality: Where do you live? I live in London / *I live on the moon 4. Violation of the Maxim of Relation: What did you do at school? We did some writing / *We had a bath 5. Violation of the Maxim of Politeness50: - Do you like my dress? It’s pretty / *I hate it - May I use your pencils? No, I left them at home / *No, you cannot draw The answers to the questions that are supposed to violate the Maxim (marked with an asterisk), would be acceptable in the right context. They actually present some implicit meanings in them. Answering ‘in a cup’ to the question ‘how would you like your tea?’ would be acceptable if the person answering it likes to change his/her drinking habits, let’s say choosing among glasses, mugs, and cups, and the person asking is looking into the cupboard where there are glasses, mugs, and cups. The example violating the second Maxim of Quantity, is actually a very relevant answer, and it also includes an implicit meaning, namely ‘I like rabbits which are animals’ +>> even though they are not pets, as asked. The sentence in 3., which should be an example of violation of the Maxim of Quality, could be plausible for children as they are commonly exposed to imaginary realities through films, animated cartoons, and similar means of entertainment. The violation of the Maxim of Relation is an uncommon answer, but not impossible. The violations of the Maxim of Politeness, in the right context, are perfectly acceptable. Answering ‘I hate it’ to a question about a dress is possible between people with a very close relationship; as for the second example, not only is it possible, but it also contains an implicit meaning: ‘May I use your pencils?’ ‘No, you cannot

56

Chapter Two

draw’ +>> so if you use them you might break/ruin them. There are also some non-ambiguous examples in the study. Nonetheless the ambiguity highlighted above makes a point for reflection about conducting research in pragmatics. When considering what is not adequate, it is important to keep in mind the crucial role that the context plays in the interpretation of utterances. Providing decontextualized examples renders the tasks of interpretation and judging more difficult, because of the lack of environmental clues to rely on. Ozonoff & Miller (1996) tested three pragmatic abilities in ASD adolescents and adults: comprehension of humor, inferential abilities, and understanding of indirect requests. The authors report that people with ASD performed more poorly than controls on all three tasks. In the humor comprehension task, in which participants were asked to choose a humorous ending to a story, they chose either an ending that logically (but not humorously) completed the story, or a humorous ending not connected to the story. The inferential task required participants to disambiguate sentences according to provided information (e.g., Mr. Smith gave the newspaper boy a tip. He told him how to fold the paper for delivery before throwing it on the porch), or to draw a bridging inference (e.g. Jenny missed the wild pitch. The windshield was shattered). People with ASD had more difficulty interpreting sentences in which the ambiguous item appeared at the beginning of the sentence. In the indirect request task, ASD people showed a neglecting of the context for the choice between directness vs. indirectness, not showing though, as hypothesized by the authors, a preference for direct requests. The authors emphasize that similar patterns of behavior are present in RHD patients as well. They are characterized by problems in the understanding of figurative and nonliteral language, misinterpretation of communicative intentions, use of language in an overly literal way, abnormal prosodic patterns, and reduced use of gestures. Therefore, they propose further investigations on the relationship between ASD and the right and left hemisphere contributions to pragmatic communication. Emerich et al. (2003) also report data on impairment of the comprehension of humorous material in 8 adolescents with HFA or AS, ranging in age from 11 to 17 years. After having been presented with the first frames of a cartoon, or having heard or read the body of a joke, they were asked to choose funny endings for the cartoon or the joke. They showed an overall poorer performance on both tasks compared to controls, and a within-group poorer comprehension on the joke task. The authors relate these people’s performance to a difficulty in mastering the surprise

Pragmatics and Autism Spectrum Disorders

57

and coherence aspects of humor, probably linked to a lower cognitive flexibility due to deficits in executive functions. Dobbinson et al. (1998) interestingly point at the pragmatic abilities, instead of impairments, of a woman with autism. The conversations between the autistic woman and a researcher were transcribed and analyzed using Conversation Analysis. Through this linguistic tool, the authors were able to individuate the woman’s conversational competences. They point out that, even though at first glance she might seem to overlap with her interlocutor’s turn, and not respect turn taking, she is actually able to do so; hence, the presence of overlapping is explainable in terms of repairing and returning to previous phases of conversation. They also analyze the differences as regards topic switching and maintenance, latching, and pauses between the person tested and the experimenter’s conversation, pointing at the coherence in the structure of the woman’s discourse. Although this is a case study, and therefore its results are not generalizable, it is worth pointing out because of the use of a linguistic technique to analyze linguistic data. This should be obvious, but instead linguistic methodology is not considered by most researchers in the field of psychology. Dennis et al. (2001) analyzed the comprehension of different implicit meanings in 8 HFA children (mean age 9;9, SD 1.6): presuppositions (deriving from factive mental states verbs, and implicative verbs), script inferencing, metaphor, and speech acts. The authors conclude that, contrary to previous findings, HFA children appear to be able to make some inferences, namely the processing of presuppositions deriving from mental states verbs, suggesting that they can connect a text to fixed knowledge. Nonetheless, this is not the case with implications, suggesting an inability to connect the text to the local sentence context. The authors also compared the performance on ambiguity vs. comprehension of metaphor, and the latter resulted in a poorer performance, suggesting an inability to detect the communicative intention, rather than understanding that an utterance might have multiple meanings. They also failed to make correct inferences about social scripts, and to produce speech acts. The authors conclude “(HFA children) often fail to make inferences that are the basis of successful social communication; for example, those that elaborate meaning for the listener or signal awareness of intentions. They seem unable to make contextual inferences about thoughts” (2001:53). Lokusa et al. (2007) investigated integration of contextual information in HFA and AS children (aged from 7 to 12 years) who were presented with questions with varying levels of dependence on context: reference assignment, enrichment, basic implicature, routine, and feeling questions.

58

Chapter Two

They were assigned scores both for providing the correct answer, and for the correct explanation of why they had given that answer. Reference assignment was unproblematic for both the experimental and the control group; younger ASD people (7 to 9) had some difficulties with enrichment; both younger and older (10-12) ASD children performed poorly on implicature questions which were the most context demanding. The biggest difference between ASD and control children was in providing explanation for their answers, suggesting that “their communication problems in real-life situations may be partly due to difficulties in understanding what kind of thought processes the partner in communication has to go through to give an answer” (2007:1056). Nonetheless, an interesting developmental pattern emerged from this study: younger children (7 to 9) performed significantly worse than the older ones (10-12). Thus, the data were probably influenced by a developmental factor as well. Additionally, the children tested were able to integrate the contextual information to answer the different types of pragmatic questions, even if with different percentages compared to control children. The authors conclude that “the performance of these children indicated an inefficiency, but not an inability in context use in comprehension” (2007:1057).

2.6.2 Cognitive theories of ASD and pragmatic adequacy The research presented shows a strong connection between the impairment of pragmatic and ToM skills. As discussed in §1.4, ToM is not the only cognitive theory that accounts for pragmatic impairment in people with ASD. Scholars proposing impairment of executive functions and weak central coherence put forward arguments supporting a connection with the impairment of this linguistic component. As pointed out by Cummings (2009:127), none of these three theories satisfies her two criteria of pragmatic adequacy (i.e. 1. explain the inferential process by means of which a speaker’s communicative intentions are established; 2. account for the global nature of that process) and is thus a pragmatically adequate theory of autism. The author, however, recognizes that “the central role of ToM skills […] in utterance interpretation means that this is an investigation that workers in pragmatics can ill afford to neglect” (2009:53). ToM has received a substantial amount of experimental corroboration in connection with pragmatics both in neurotypical and clinical populations. Perner et al. (1989); Loveland & Tunali (1991); Eales (1993); Happé (1993); Ozonoff & Miller (1996); Joliffe & Baron-Cohen (1999a, b); Leinonen & Kerbel (1999); Tager-Flusberg (1999); Peterson & Siegal

Pragmatics and Autism Spectrum Disorders

59

(2000); Carston (2002); Rutherford et al. (2002); Martin & McDonald (2003, 2004); Miller (2004); Colle et al. (2008); Surian & Siegal (2008) represent only some examples of the copious research that found a positive correlation between impaired ToM skills and pragmatic impairment in people with ASD. Taking this correlation for granted, what is interesting to try to understand is which way the causal relationship between these two skills goes. In other words, is it an impairment of ToM skills that causes the pragmatic impairment (as maintained by the majority of these works), or could it be the other way around? Or, is there a third, perhaps neurological, dysfunction that causes one (or both) of the cognitive impairments, and consequently the other? Interesting insights about this relationship are provided by Lohmann & Tomasello (2003) and Harris et al. (2005), which underline the central role of language in developing the understanding of false belief, and mental states. This kind of relationship is supported by a noteworthy study by Peterson & Siegal (2000) about deaf children from hearing families, whose limited exposure to language in their first years of life results in poor performance in ToM skills, comparable to that of autistic children. Moreover, de Villiers & Pyers (2002:1056), analyzing ToM skills in SLI children, argue that “the child needs the full syntax of mental verbs plus sentential complements in order to represent in his own mind the belief states of other people, not simply to encode them for reporting about them in speech”. A positive correlation between ToM skills impairment and pragmatic deficits has also been found in schizophrenic patients (Langdon et al. 2002; Brüne & Bodenstein 2005). As Martin & McDonald (2003: 455) cogently suggest: It is, therefore, possible that poor pragmatic communication skills contribute to deficits in ToM, because a person who lacks communicative experience may fail to comprehend that a verbal message represents the subjective mental states of an individual, and is therefore a separate entity from reality. Alternatively, it may be that some core deficit in ‘interpersonal-affective relatedness’ may underpin both the linguistic and the mentalistic deficits observed in children with autism and that such a core deficit is related to right-hemisphere functioning. Indeed, the critical age period for the ability to provide accurate responses to ToM tests coincides with the spurt of RH growth at around 4 years. It may be that delayed exposure to conversation may influence patterns of RH brain development in both deaf and autistic children. Altered RH activity may then in turn lead to deficits in social and emotional reasoning as reflected by both ToM and pragmatic language tasks.

60

Chapter Two

Findings supporting the correlation of deficits in ToM skills, pragmatic impairment, and RH damage, come from research conducted with Traumatic Brain Injury (TBI) patients. Shamay-Tsoory et al. (2005) report that TBI patients with lesions in the right ventromedial area were impaired in several tasks involving ToM skills, like the recognition of irony, sarcasm, and faux pas. The hypothesis of a connection between the right hemisphere and the deficit in ToM and pragmatic skills, though needing more empirical corroboration, provides an interesting point of view about the causal relationship between these last two skills and the role of language in developing a theory of others’ minds.

2.7 Summary of controversies in previous research Notwithstanding the amount and variety of theoretical frameworks in the research presented, a clear, complete, and pragmatically adequate picture of linguistic deficits in autism appears to be lacking. Research agrees on impairment in: inferential abilities, understanding of non-literal language, comprehension of irony and sarcasm, indirect speech acts, contextual integration, deixis, politeness, understanding of social scripts, and so forth, basically covering almost all the skills that make up pragmatic competence. Nevertheless, as already pointed out, some of the results appear to be controversial for the methodology used during the research. In some studies participants were extremely young (e.g., Stone & CaroMartinez (1990) tested children of 4 to 13 years; Dennis et al. (2001)’s participants’ mean age was 9 years; Lokusa et al. (2007) younger group’s age was comprised between 7 and 9), hence the purported pragmatic impairment might actually be influenced by developmental factors. Some other studies concentrated only on the production (e.g. Stone & CaroMartinez 1990), inferring from the absence of a feature in the linguistic production of people with ASD the impairment of the related skill. Other studies (e.g. Surian et al. 1996, Surian & Siegal 2008) show a misinterpretation of the nature of pragmatics by neglecting the fundamental role of context in the disambiguation of utterances, distorting the notion of context (Jolliffe & Baron-Cohen 1999a) or including nonverbal behaviors in pragmatic competence (e.g. Chandler et al. 2002, Caro-Martinez 1990). Future research will need to concentrate more on elaborating testing methodologies not only cognitively but also, and crucially, to make them pragmatically adequate to analyze the pragmatic skills of these individuals.

Pragmatics and Autism Spectrum Disorders

61

This chapter discussed pragmatic competence, development, and impairment. It provided a review of the studies carried out to investigate the extent of pragmatic impairment in ASD people, and presented some of the controversies emerging from these studies. The next chapter presents the research conducted to analyze implicit meanings comprehension in people with ASD, discussing research methods and data collection procedures.

CHAPTER THREE THE RESEARCH

This chapter presents the research I conducted. First it discusses research methodology, including the devising of the test; then it presents details of the people who participated in the study; lastly, it discusses the procedures of data collection and analysis.

3.1 Introduction The research presented in this chapter aimed at investigating the extent of different implicit meanings understanding in adolescents and adults with Autism Spectrum Disorders. As pointed out in Chapter 2, previous experimental findings (Eales 1993; Ozonoff & Miller 1996; Surian et al. 1996; Dennis et al. 2001; Lokusa et al. 2007; among others) agree on the fact that this clinical population is characterized by a general inferential disability, which manifests itself in an inability to comprehend implicated (Grice 1975) and presupposed meanings, among other non-literal uses of language. According to Dennis et al. (2001), people with ASD do present a certain extent of inferential processing. They are able to understand implicit meanings derived from presuppositions triggered by factive mental state verbs (e.g., know); nonetheless, this is not the case for those triggered by implicative verbs, because, as the authors propose, the former type of presuppositions refer to knowledge that requires neither contextual integration, nor the ability to make inferences about thoughts, as is the case with the latter presuppositions. Putting it à la Grice, people with ASD have been found to be impaired in their ability to discern between what is said and what is meant by the speaker. They seem not to adhere to the Cooperative Principle and to the Maxims of conversation (Surian et al. 1996), not recognizing that communicative exchanges are the result of a cooperative effort between interlocutors. Sabbagh (1999) convincingly proposes integrating the wide range of communicative deficits characterizing people with ASD (common to RHD patients as well) into only one, dependent on “a failure to appreciate the role of CIs (Communicative Intentions) in everyday

64

Chapter Three

conversation” (1999:63). Comprehension of communicative intention guides the production and interpretation of language; it is a key skill in every communicative exchange. A disruption of this crucial skill appears to be a plausible explanation for the different linguistic impairments characteristic of autistic conditions. The test administered to people with ASD in the research presented here is designed to analyze the understanding of communicative intentions linked to the processing of implicit meanings. Gricean notions of implicature and conversational Maxims were the starting point for the devising of the test. Taking into consideration previous findings (see §2.6), presuppositions were included as well, in order to investigate whether the difficulty of understanding implicit meanings (hereafter IMs) is proportional to their dependence on context, as shown in the following implicational scale: Presuppositions not the entire paw) in 21b; - 2 participants on the conversational implicature derived by the Maxim of Relation requiring participants to interpret the sentence ‘it’s raining’ as ‘it’s raining outside the house where the people in the video are’ in 29b. For these people a developmental factor could be hypothesized given that their mean age is 13;10. However, before putting forward a developmental explanation, it should be considered that they do not produce any other non-target answer on IMs, and three out of four also produced a non-target answer on a control item. Thus, a developmental explanation could be possible, but should not be proposed as a straightforward conclusion. The explanations that they provided for their non-target answers are presented in (2) and (3): (2) Testing Item: […] it walked leaning the front part of the paws on the ground 21b Plantigrade animals walk leaning the entire paw on the ground (target answer F because ‘the front part of the paws’ +> not the entire paw)

96

Chapter Four

Explanations provided for answering T: - It’s written in the text - Because that’s how they walk.

(3) G: Shall we go play outside? B: It’s raining Testing Item: 29b It is raining out of the house where the two characters are (target answer T because of the Maxim of Relation). Explanations provided for answering F: - The boy did not want to go to play with his sister so he made that up - He said so but we do not know if it is true.

While for the scalar implicature in 21b the neglecting of the scalar item ‘the front part’ seems clear, for 29b, considering the explanations given, the problem seems to be with the boy’s trustworthiness. Considering that these people understood that the character’s utterance ‘it’s raining’ functioned as an indirect refusal (29a), they thought that he was making up an excuse not to go play with her. This interpretation was counted as nontarget, because it differs from the results of the other control individuals; nevertheless, it is certainly not to be considered as ‘pragmatically inadequate’ and, even less, as a failure to adhere to the maxims or to attribute a mental state to other people. Actually, assuming that someone is lying is a very complex mental state attribution process. To conclude this section on the results of the control group, it seems interesting to report some answers provided by the control participants in answer to the question ‘why’ to explain their target choices of T, F, DK. (4) Answers to the ‘why’ question of participants who provided a target answer on IMs: i. W: Will you show me this ball? M (gives her the ball): yes, it’s crystal Testing Item: 10b The woman has to be careful with the ball (target answer T) - Because the man asked her to be careful - He said that it’s made of crystal to tell her to be careful - He warned her that it’s made of crystal.

Results

97

ii. Oriana was looking at her family pictures. She thought about her brothers, her sisters and some of her nephews […] Testing Item: 11a. Oriana has brothers and sisters (target answer T) - If she was thinking about them it means that she does have brothers and sisters - It’s written that she has brothers and sisters. iii. M: What is the dog doing on the couch? B: Get down (pushes the dog away) Testing Item: 15b. The dog shouldn’t be on the couch (target answer T) - T, The man reproached the boy - T, The man says what is he doing to mean that he should not be there - T, It’s understandable from what the man says. Testing Item: 15c67 The boy pushes the dog away (target answer T) - T, The man asked him - T, The man does not want the dog to be on the couch - T, Because of what the man says. iv. G1: I am studying G2 turns off the radio Testing Item: 27b One of the two girls turns off the radio because she was tired of listening to it (target answer F) - Because the other girl asked her to do so - F, The other girl said that it was bothering her - F, Saying that she was studying she indirectly asks the other girl to turn the radio off.

The underlined words reveal how the IMs are processed and understood by these people. In (4) i., iii., and iv., for example, the participants use the verb ‘ask’ when actually not one of the characters of the videos directly asks anything. In 10b in example (4) i. the man just says that the ball is made of crystal; he does not ask the woman to be careful. In 15b, example (4) iii., the man asks what the dog is doing on the couch and not that the dog should be sent away. In 27b, example (4) iv, the girl says that she is studying, she does not ask the other girl to turn the music off. Nonetheless, these participants, and all the others who answered like them, were very sure about what was being ‘asked’ in these communicative exchanges.

Chapter Four

98

4.2 ASD group This paragraph presents the results of the ASD group. Table 4-3 illustrates the number and percentages of total and IMs target answers ordered by age and decreasing percentage of target answers, as well as the quantity and the kind of non-target implicit meanings. The results will first be presented counting the testing item in 8b, for the sake of completeness, and then without that item. Table 4-3 Results of the ASD Group Code AG_M_15 AG_F_21 AG_P_39 AG_C_12 AG_B_14 AG_A_22 AG_E_19 AG_M_23 AG_V_11 AG_F_12 AG_A_12 AG_Y_15 AG_F_16 AG_D_16 AG_C_16 AG_V_20 AG_V_11 AG_D_29 AG_U_20

Total of target Target answers on Answers IMs number percentage number percentage 45 100% 22 100% 45 100% 22 100% 45 100% 22 100% 44 98% 21 95% 44 98% 21 95% 44 98% 21 95% 43 95% 20 91% 43 95% 20 91% 42 93% 20 91% 42 93% 20 91% 42 93% 19 87% 42 93% 20 91% 42 93% 19 87% 42 93% 19 87% 42 93% 20 91% 42 93% 21 95% 42 93% 19 87% 42 93% 20 91% 41 91% 19 87%

AG_N_11

41

91%

18

81%

AG_M_14 AG_V_23 AG_I_16

40 40 39

89% 89% 87%

17 17 18

77% 77% 81%

AG_E_20

37

82%

16

73%

AG_E_18

36

80%

16

73%

AG_G_21

33

73%

16

73%

Non-target IM(s)

1 CI 1 SI 1 CIR 1 CI, 1 CIQ68 2 CI 2 CI 1 CI, 1 SI 1 CI, 1 SI, 1 CIR 1 CI, 1 SI 1 P, 2 CI 2 SI, 1 CIR 2 CIR 1 CI 1 P, 1 CI, 1 CIQ 1 CI, 1 CIR 2 CIR, 1 CI 1 CIR, 1 CIQ, 1 CI, 1P 2 CIQ, 1 CIR, 1 CI 2 CI, 2 SI, 1 CIR 2 CIR, 1 CIQ, 1 CI 2 CIR, 2 CI, 1 SI, 1 P 2 CIR, 2 CI, 1 CIQ, 1P 2 CI, 1 CIQ, 1 CIR, 1 SI, 1 P

Results

99

AG_L_14

30

66%

14

63%

AG_I_15

30

66%

11

50%

AG_G_17

29

64%

13

59%

3 P, 2 CIR, 1CI, 1 CIQ, 1 SI 2 P, 2 CIR, 3 CIQ, 2 CI, 1 SI, 1 LI69 3 P, 1 CIR, 3 CI, 1 SI, 1 LI

As Table 4-3 shows, only 3 people in the ASD group performed at 100% with the item 8b included, compared to the 12 of the control group. Excluding 8b, just one other person’s performance changes to 100%. Considering that 8b problematic results for typically developing individuals, as already stated, it was eliminated in the calculation of percentages, as was done for the control group. The results excluding item 8b are as follows: -

mean of total target answers: 91% (SD 4.3) mean of target answers on implicit meaning: 88% (SD 2.5)

As the values of standard deviation indicate, the results of the people in the ASD group are more heterogeneous. Figure 4-3 illustrates the percentage of people relating to their performance on total target answers. Figure 4-4 shows how the non-target productions are divided according to the mean value for IMs, and control items. Figure 4-3 shows that almost three quarters of the participants (69%) produced • 41 out of 44 target answers on total comprehension. Four participants out of 29 (14%) performed at 100%. The mean of total nontarget answers is 4, divided as shown in Figure 4-4. Figure 4-4 shows that the mean of non-target answers on IMs is slightly more than 2.5 and on control items is almost 1.5. Figure 4-5 illustrates participants’ percentages relating to their production of target answers on IMs.

100

Chapter Four

Figure 4-3 Percentages of people in the ASD group and their total target answers

Figure 4-4 Mean of non-target answers divided in IMs and control items

Results

101

Figure 4-5 Percentages of people in the ASD group and their IMs target answers

For IMs understanding as well, almost three quarters of the participants (73%) produced 3 or less non-target answers, and 6 out 29 (21%) performed at 100% on IMs understanding. The highest percentage (28%) is that of participants that produced only one non-target answer. The next section illustrates these results in detail, discussing the performances on every IM and pointing out the kind of IMs that was more difficult for people in the ASD group.

4.2.1 Details on IMs understanding by ASD group As pointed out in the previous paragraph, participants from the ASD group produced target answers on 88% of questions testing IMs; consequentially, their percentage of non-target answers on IMs is 12%, that is 76 out of 609, with a mean of 2.6 (SD 2.5). In §3.1, an implicational scale was proposed ordering IMs according to their context dependence. Predictions based on previous research (see §2.6) suggest that the difficulty of IMs interpretation is proportional to their context dependence. Thus, following the scale proposed and reported

102

Chapter Four

here, the percentages of non-target answers should increase from left to right on the scale: Presuppositions< Conventional Implicatures< Conversational Implicatures Actually, this is what was found. Figure 4-6 illustrates how the 12% of non-target answers on IMs is divided among the kinds of IMs. The values were calculated considering the different numbers of occurrences of every kind of IM. Fig.4-6 Percentages of non-target answers divided by kind of IM

Figure 4-6 illustrates that the kind of IM that proved to be the most difficult, in percentage, was the category of conversational implicatures, followed by conventional implicatures, and lastly by presuppositions and the logical implication, which had the same percentage of non-target answers. Figure 4-7 shows the mean values of non-target answers per person for every IM category. The mean of non-target answers on IMs is 2.6. Figure 4-7 illustrates how the non-target productions on IMs are divided, according to their mean values. Table 4-4 provides the number and percentage of non-target occurrences for every kind of IM.

Results

103

Fig.4-7 Mean of non-target answers on IMs per participant

Table 4-4. Details on IMs occurrences and non-target answers

Logical Implication Presuppositions Conventional Implicatures Conversational Implicatures

Total Occurrences 29 (1·29) 203 (7·29)

Non-Target Answers # 2 out of 29 14 out of 203

Non-Target Answers % 7% 7%

87 (3·29)

12 out of 87

14%

290 (10·29)

48 out of 290

17%

The results had to be calculated in percentages given the difference in occurrence of the different kinds of IMs. In Table 4-4, the first column lists the kind of IMs; the second the total occurrences, i.e. the quantity of every IM present in the test multiplied by the number of participants; the third column explicitates the calculation to arrive at the percentages in column four. Only the 7% of occurrences of both logical implication and presuppositions were non-target. For conventional implicatures this percentage was 14%, and for conversational implicatures it was 17%. Thus, following the predictions, the least context-dependent IMs, namely logical implications and presuppositions, were the least problematic. As context dependence increases, the IMs prove to be more problematic for

Chapter Four

104

these people. These results would seem to confirm the hypothesis of a direct correlation between context dependence and difficulty of interpretation. Nonetheless, this is true only at a macro level. Dividing the conversational implicatures into generalized (therefore less context-dependent) and particularized (highly context-dependent), the same is not true. Table 4-5 illustrates the results on conversational implicatures divided between generalized and particularized. Table 4-5. Generalized and Particularized Conversational Implicatures Total Occurrences Generalized (Scalar) Particularized

Non-Target Answers #

Non-Target Answers %

58 (2·29)

13 out of 58

22%

232 (8·29)

35 out of 232

15%

What emerges from Table 4-5 is an unexpected result. Scalar items should be easier to interpret compared to particularized conversational implicatures, considering that they are less context dependent than PCIs. That was not the case for the two instances of scalar items proposed in the test. I will return to this point in §4.2.1.3, discussing the results of the single items engendering conversational implicatures. 4.2.1.1 Presuppositions This section presents the results on presuppositions in detail70. There is a total of 7 presuppositions in the test. The non-target responses on them are different, so it was deemed interesting to present the results sorted by testing item. Table 4-6 presents the presupposition trigger and the number of participants who provided non-target answers on them. Table 4-6. Non-target answers on presuppositions Test item 25c 23c 25a 13b 13a

Trigger other times (>> not the first time) how did your test go? (>> she had a test) looking around (>> not blind) before a bolt of lightning strikes us (>> they have not been stricken) get out of this bad situation (>> they are in bad situation)

Number of people 5 3 3 2 1

Results

11a 25b

she thought about her brothers, her sisters (>> she has brothers and sisters) passed through an opening (>> there was an opening)

105

0 0

The most problematic presupposition, albeit with a low percentage (17%) of the total occurrences of this IM, was the one triggered by the iterative construction ‘other times’. In the five cases, the item was ignored by the participants, who provided the following explanations for their non-target answers: (5) […] under which other times they had stopped by and rested Testing item: 25c It was the first time they stopped by and rested under the walnut tree. (target answer F because ‘other times’>> not the first time) - DK, it is not said - DK, I do not know - T, it was the first time - T, the text does not say the contrary - T, maybe they were passing by and they were tired

The triggers for 23c, 11a, and 25b are definite description. As Table 4-6 shows, no participant produced non-target answers for the last two items. As for 23c, the three participants who were not able to interpret it had problems with the entire testing item 23. Not being able to understand the main IM, namely the fact that the girl asks what they are having for lunch because she wants to change the topic of the conversation, they were very puzzled about the existence of a test. Actually, as 23 is the testing item with the highest number of non-target answers (I will return to its discussion in §4.2.1.3), then this is probably what influenced the failure in understanding 23c, also considering the explanations provided: (6) W: How did your test go? G: Hum … what are we having for lunch today? Testing item: 23c The girl had a test at school (target answer T) - DK, it is not clear - F, no one says so - We don’t know

106

Chapter Four

Considering this, what emerges from these results is that among presuppositions, the least problematic kind is the one triggered by definite descriptions. The other kind of presupposition that was not understood by three participants is the one in 25a, triggered by ‘looking around’ which presupposes that she was not blind. In (7) the explanations provided for the non-target answers were: (7) Chiara was riding her horse and looking around […] Testing item: 25a Chiara was blind (target answer F because she was looking around) - T, riding a horse is a therapy for blind people - DK, I don’t know - F, because blind people cannot ride horses

The last explanation provided is an example in which the first answer is correct, but the explanation clarifies the fact that the person did not arrive at the target answer through the processing of the IM; it was thus counted as non-target. The presupposition in 13b, triggered by the temporal item ‘before’, was not understood by two participants. In (8) are their answers and explanations: (8) […] before a bolt of lightning strikes us Testing item: 13b Josef and his friends have been struck by a bolt of lightning (target F because of ‘before’) - DK, it is not clear - T, because they are in a bad situation

The first participant ignores the item, while the second one is influenced by the fact that the character in the paragraph says that they are in a bad situation, and probably thinks that it is so because they have been struck by a bolt of lightning. In the same set of testing items, one participant did not understand the presupposition in 13a, triggered by the change of state verb ‘get out’.

Results

107

(9) […] let’s get out of this bad situation Testing item: 13a Josef and his friend are in a bad situation (target T because of ‘get out’) - DK, I don’t know, it seems so. I don’t know

This participant seems to understand that the characters are in a bad situation, but does not relate that to the verb ‘get out’ which indicates that this is so. Following the predictions, presuppositions did not prove to be difficult for people of the ASD group. Only 7% of all the occurrences were nontarget. Apart from pointing out that the presupposition triggered by the definite descriptions were the least problematic, it is not easy to put forward an explanation for the other kinds considering both the low percentages, which do not allow us to make any generalization, and the heterogeneity of the motivations provided. To conclude this section on presuppositions, it is interesting to provide some examples of target answers (total 93%) and explanations, to show how the tested participants processed this kind of IM. (10) Examples of target answers and explanations i. […] before a bolt of lightning strikes us Testing item: 13b Josef and his friends have been struck by a bolt of lightning - F, because he says before so that means that they haven’t been struck yet - F, because Josef says before a bolt of lightning strikes us. ii. W: How did your test go? G: Hum … what are we having for lunch? Testing item: 23c The girl had a test at school - T, Oh yes she had! And also didn’t do it well! - T, she had the test because her mother asks her about it - T, she had it but it shouldn’t have gone very well. iii. […] Her horse passed through an opening in a wall Testing item: 25b There was an opening in the wall - T, the horse went through it so there had to be one - T, it says an opening so there was an opening.

108

Chapter Four

4.2.1.2 Conventional Implicatures Non-target answers on conventional implicatures represent 31% of all the non-target answers. Following the predictions, they proved to be easier to process than conversational implicatures, and more difficult than presuppositions and the logical implication. A total of 12 out of 87 (14%) occurrences were non-target. Table 4-7 illustrates the details. Table 4-7. Non-target answers on conventional implicatures Test item 17a 21c 11c

Trigger but it has shorter legs a plantigrade, like bears and so she calmed down

Number of people 6 4 2

The most problematic conventional implicature was the one triggered by ‘but’ tested in 17a. Six participants out of 29 (21%), did not draw the expected implicature, and provided the following explanations: (11) The coyote has a robust body, like the wolf, but it has shorter legs Testing item: 17a. The coyote is exactly like the wolf (target answer F, because it says but +>> what follows it contrasts with what precedes it) - T, because it’s like the wolf - T, they are canids - T, they are alike, he is brown though - T, it is difficult to tell them apart - T, it’s written there like the wolf - T, it has a robust body like the wolf

As (11) shows, some of the participants neglected the ‘but’ and remained with the idea of what the first sentence had said about the body of the coyote, some actually echoed the text to provide an explanation. The person of the second example makes reference to background knowledge that he has about wolves and coyotes, the third probably considers the difference between the two pictures present in the test. Four participants out of 29 (14%), produced a non-target answer on the conventional implicature engendered by ‘like’ and tested in 21c. In (12) the explanations are provided.

Results

109

(12) […] it walked leaning the front part of the paws on the ground. It was thus a plantigrade, like bears Testing item: 21c Bears are plantigrade animals (target T, because of ‘like’) - F, I don’t know but I don’t think so - DK, it’s not said in the text - DK, we can’t say that - DK, the text says so about the miacis.

In this case, too, the participants do not process the implicature engendered by ‘like’ at all. The least problematic conventional implicature was the one engendered by ‘so’ and tested in 11c. Only 2 of the 29 participants (7%) provided a non-target answer, as follows in (13). (13) She thought about her brothers, her sisters and some of her nephews, and so she calmed down. Testing item: 11c Oriana calmed down looking at the landscape out of the window (target F, because of ‘so’ +>> what follows it is an explanation of what precedes it) - T, looking at landscapes is relaxing - DK, it would have said so

The two participants here do not recognize the function of ‘so’ as a linker of cause-effect between the two sentences. Summing up, the least problematic conventional implicature was the latter one, engendered by ‘so’, followed by the one engendered by ‘like’, and finally the most difficult was the one connected to ‘but’. The correct responses on conventional implicatures were 75 (86%). In (14) some examples of correct answers are provided. (14) i. She thought about her brothers, her sisters and some of her nephews, and so she calmed down. Testing item: 11c Oriana calmed down looking at the landscape out of the window - F, she calmed down looking at the pictures - F, she was thinking about her family - F, thinking about her brothers, sisters, and nephews

Chapter Four

110

ii. […] it walked leaning the front part of the paws on the ground. It was thus a plantigrade, like bears Testing item: 21c Bears are plantigrade animals - T, it says like bears - T, there it is written plantigrade like bears so it means that bears are plantigrade animals iii. The coyote has a robust body, like the wolf, but it has shorter legs Testing item: 17a The coyote is exactly like the wolf - F, it has shorter legs - F, only for the body - F, it has shorter legs, and for the muzzle and the ears it is more similar to the fox.

4.2.1.3 Conversational Implicatures This section presents non-target answers on conversational implicatures. This is the most copious group. This kind of IMs proved to be the most difficult one for the ASD group, as predicted. It should be highlighted though, that, contrary to predictions, ASD scored a very high percentage of target answers, namely 83%. Forty-eight out of 290 (17%) were nontarget answers, divided as illustrated in Table 4-8. Table 4-8. Non-target answers on conversational implicatures Test item

Kind

23b

CIR

11b

SI

19b

CIQ

5a

CIR

21b

SI

15b

CIQ

27b

CIR

29a

CIR

10b

CIQ

Trigger W: How did your test go? G: Hum … what are we having for lunch? […] some of her nephews B: Excuse me, do you know what time it is? G: (looks at her watch): Yes. (walks away) W: Would you like some coffee? M: I wouldn’t sleep all night the front part of the paws M: What is the dog doing on the couch? B: Get down (pushes the dog away) G1: I am studying G2 turns off the radio G: Shall we go play outside? B: It’s raining W: Will you show me this ball? M (gives her the ball): yes, it’ crystal

Number of people 10 8 6 5 5 4 4 4 2

Results

29b

CIR

G: Shall we go play outside? B: It’s raining

111

0

The particularized conversational implicature engendered by the exchange between a girl and a woman, presumably her mother, of the video in slide 22 was the most difficult IM to understand by the tested ASD participants and, consequently, also the most difficult conversational implicature. Ten out of 29 participants (34%) had trouble drawing the implicature in 23b (+> the girl wants to change the topic of conversation). In (15) the nontarget answers and explanations are reported. (15) W: How did your test go? G: Hum … what are we having for lunch? Testing item: 23b The girl wants to change the topic of the conversation (target T, because she does not want to talk about the test) - F, she didn’t hear what her mother said - F, she did not understand - DK, she didn’t answer because she didn’t understand the question - DK, I don’t know - T, she is hungry - Because she was hungry - Because she is hungry - She didn’t understand the question - She wants to know what they are having for lunch - She didn’t understand the question.

The majority of the participants answered saying that it is true that she wants to change the topic of the conversation but because she is hungry, while the expected answer was that she wanted to do so because she did not want to talk about her test. The next conversational implicature to be discussed is the one engendered by the scalar item ‘some’ and tested in 11b. Eight participants out of 29 (27%) did not give a target answer. Nonetheless, considering the explanations provided, none of them gave the item a logical interpretation; it would seem that they just neglected it, as (16) shows. (16) She thought about her brothers, her sisters and some of her nephews

112

Chapter Four

Testing item: 11b Oriana thought about all her nephews (target F, because some +> not all) - Yes, she was looking at the pictures - Yes, she saw them in the pictures - T, to calm down - T, she was looking at family pictures - T, they were in the pictures - T, because she loves them and so she calmed down - T, it is written that she thought about her family nephews included - T, and also to her brothers and sisters.

These examples show the crucial role of asking the participants to explain their answers. Not considering the explanations, one might have concluded that, because ASD people have been proven to process language mostly at the literal level, they gave the scalar item ‘some’ the logical interpretation of ‘possibly all’ and not the pragmatic one of ‘not all’, which would have lent support to the view of preference for literal interpretation. Nonetheless, none of the tested people justified their answers by saying that ‘some’ means ‘possibly all’, and therefore they answered T to 11b71. Moreover, the 21 participants who correctly answered F (72%), all provided the presence of the term ‘some’ as an explanation for their answer, showing that they did give a pragmatic interpretation to the scalar item. Six participants out 29 (21%), did not recognize the violation of the Maxim of Quantity in the video 18, tested by 19b. Details in (17). (17) B: Excuse me, do you know what time it is? G (looks at her watch): Yes. (walks away) Testing item: 19b The girl gives the expected answer (target F) - She answered the question - T, she answered well - T, I don’t know - T, she answered correctly because she answered literally - T, the boy asked her - T, I don’t know.

The answers provided by these people to explain their choices of T highlight the failure to recognize the inadequacy of the girl’s answer, which violates the Maxim of Quantity. In this case 21% of the tested

Results

113

participants proved unable to recognize the violation of a conversational maxim, as suggested by previous research (Surian et al. 2006). Five participants out of 29 (17%) produced non-target answers on the conversational implicature deriving from the Maxim of Relevance in the video in 4, tested by 5a. (18) W: Would you like some coffee? M: I wouldn’t sleep all night. Testing item: 5a The man accepts the coffee (target F) - DK, it has nothing to do with it - T, because … - DK, we can’t know - T, he wanted it - T, he likes coffee

This case, too, shows the participants’ non-understanding of an IM. Here, they failed to draw the implicature deriving from the Maxim of Relation, namely that the video character’s utterance that coffee would keep him awake functions, in this context, as an indirect refusal. Five participants (17%) did not answer in a target way the item in 21b which tested comprehension of the scalar implicature engendered by ‘the front part of the paw’ +> not the entire paw. (19) details the answers provided. (19) […] it walked leaning the front part of the paws on the ground. It was thus a plantigrade Testing item: 21b Plantigrade animals walk leaning the entire paw on the ground (target F). - Usually, they walk on their four paws -They walk this way - T, they lean their paws on the ground - T, it’s written that that’s how they walk - T, that’s how one walks

114

Chapter Four

This scalar implicature tests the comprehension of a less common scalar item, compared to the other one deriving from ‘some’ in 11b; nonetheless, it was easier for the ASD participants to process in terms of percentages (17% produced non-target answers on this question compared to the 27% on 11b). In this case too, the participants did not attribute a logical meaning to the scalar item, but simply ignored it, showing again that the non-target answers do not derive from a logical vs. pragmatic interpretation of the scalar item. Considering that both the scalar items tested were problematic for some ASD participants and, again, not due to the influence of their logical meaning, it would be interesting in future research to point out to the participants the presence of a scalar item, and ask them what it means, and why it is used. This was beyond the scope of the present research, which aimed at identifying the kind of IMs that are more difficult for ASD people to process; hence no explanation was included. Future research would need to concentrate on this, in order to plan linguistic treatments that are more efficacious and appropriate for this clinical population. Four participants (14%) did not recognize the utterance in the video in 14 as an indirect request under the Maxim of Quantity. The item that tested participants’ understanding was 15b. In (20) are details of the answers provided: (20) M: What is the dog doing on the couch? B: Get down (pushes the dog away) Testing item: 15b The dog shouldn’t be on the couch (target T) - T, Because dogs should not be allowed to stay on couches, they might pee or leave fleas on the couch - T, Because dogs have to lay on the ground - T, Because he wants to sit - Because animals can’t stay on couches.

This case too, shows that the participants provided what seems a target answer, but not because they adequately processed the IM present in the man’s utterance. They provided general rules as to why dogs, or animals in general, should not be on couches.

Results

115

Four participants (14%) produced non-target answers for the conversational implicature derived from the Maxim of Relation tested in 27b. The examples in (21) present details of the answers provided. (21) G1: I am studying G2 turns off the radio Testing item: 27b One of the two girls turns off the radio because she was tired of listening to it (target F) - She (the girl that turns the music off) has to study72 - T, you can tell from her face73 - DK, we don’t know - T, she was bored.

These participants fail to process the girl’s utterance as an indirect request according to the Maxim of Relation. The same percentage of non-target answers was produced for 29a, testing the understanding of the conversational implicature deriving from the Maxim of Relation. (22) describes the non-target answers. (22) G: Shall we go play outside? B: It’s raining Testing item: 29a The boy refuses to go play outside (target T) - F, he does not say that he refuses - F, he does not refuse - F, he only says that it’s raining - DK, it’s not said.

These people, failing to process the IM, do not understand the indirect refusal. The last conversational implicature with which people in the ASD group had difficulties is the one in 10b. Two participants (7%) did not understand the implicature deriving from the Maxim of Quantity and gave the following answers, in (23).

116

Chapter Four

(23) W: Will you show me this ball? M (gives her the ball): yes, it’ crystal Testing item: 10b The woman has to be careful with the ball (target T) - DK, I don’t know - She had not seen that the ball was made of crystal74.

These participants fail to understand that the man’s utterance is an indirect warning. No one in the ASD group had problems with the conversational implicature derived by the Maxim of Relation in 29b, reported here in (24). (24) G: Shall we go play outside? B: It’s raining Testing item: 29b It is raining out of the house where the two characters are (target T)

As was done in the previous sections for the other IMs here, too, examples of correct processing of conversational implicatures (total target 83%) are reported. (25) i. W: Would you like some coffee? M: I wouldn’t sleep all night Testing item: 5a The man accepts the coffee (target F) - Drinking the coffee he would not sleep and he does not want it75 - The coffee has the effect of keeping him awake he does not want it - He refuses the coffee, that’s why he says that he wouldn’t sleep - Because he has to drink the coffee and, drinking it, he wouldn’t sleep76 - F, he wouldn’t sleep - F, he wouldn’t have slept. ii. W: Will you show me this ball? M (gives her the ball): yes, it’s crystal Testing item: 10b The woman has to be careful with the ball (target T)

Results

117

- Because the woman could accidentally drop it and has to be careful not to do it77 - T, because it’s made of crystal - T, that’s why the man said that the ball is made of crystal - T, the man told her that the ball was made of crystal because she did not know and could have broken it - Because it is precious, she does not have to show it around78. iii. G1: I am studying G2 turns off the radio Testing item: 27b One of the two girls turns off the radio because she was tired of listening to it (target F) - F, because the other said that she was studying - F, it was bothering her friend - F, because she had to allow the other girl to study - Because she understood what the other was saying79 - Because the other girl reproached her - The other girl said that she was studying she meant that she couldn’t study - She had been asked to do so.

These are just a few examples that show how people with ASD are actually able to process highly context-dependent implicit meanings when provided with the appropriate context. They proved able to recognize the violation of conversational maxims and the deriving implicatures in 83% of the cases. This chapter presented the results of the study. In the next chapter the results will be discussed further, and the two groups’ performances will be compared.

CHAPTER FIVE DISCUSSION AND CONCLUSIONS

This chapter presents a comparison of the results of the two groups with the aim of illustrating whether, and in what amount, the results of people in the ASD group diverge from the norm represented by the results of the typically developing people of the control group. An inter-group analysis is then presented sorting the results of the people with ASD by diagnostic group, namely HFA, AS, and PDD-NOS.

5.1 Comparison between the ASD group and the control group The results of the ASD group seem to confirm the hypothesis, based on previous research findings (see §2.6), of a difficulty on the part of individuals in this clinical population to interpret implicit meanings deriving from implicatures and presuppositions. Nonetheless, the mean of non-target answers is quite low, both for general understanding, and for IMs comprehension. The mean of total non-target answers was 4, corresponding to 9%. For the control group this mean was 0.2 (0.5%). Figure 5-1 illustrates the means of non-target answers for the two groups, divided into control items and IMs. The Mann-Whitney Test revealed a statistical significance for these results (p