Please read the article below and give a summary about what the article is about. I have a hard time understanding long readings. Thank you.

- Language delay is an extremely common childhood concern, and in Western societies, language delay is often the first reason parents seek professional advice about their child's development (De Giacomo & Fombonne, 1998). For many families, the first referral is to a speech and language therapist (SLT); this clinician is responsible for establishing whether there is a clinically significant problem and whether there are any other developmental concerns that would warrant further assessment and perhaps referral to a multidisciplinary team. The clinician is also responsible for developing and delivering treatments designed to increase language capability and communication skills, often in collaboration with other professionals and with the child's family.

In this article, we focus specifically on two common neurodevelopmental disorders, autism spectrum disorders (ASD) and specific language impairment (SLI). Our rationale is that clinicians are often tasked with providing a differential diagnosis between these two conditions, and there is considerable theoretical interest in the degree to which they overlap at biological, cognitive, and behavioral levels (Tager-Flusberg & Joseph, 2003; Tomblin, 2011). Both SLI and ASD are known to have a biological basis and are heavily influenced by multiple genetic risk factors that affect brain development in ways that are nonoptimal for language and communication development (Bishop, 2009). However, presently there are no biological markers or tests for these disorders; instead, diagnosis is based solely on observed behavior and parental report. As with most behaviorally defined disorders, the point at which normal variation converges on disorder is largely an arbitrary decision and is highly likely to be influenced by cultural values and expectations. Culture shapes behavior, and standards of what is and is not acceptable behavior are culturally determined (Bruner, 1990). In addition, the measures used to sample behaviors of interest and the evidence base on which diagnostic decisions are made are also subject to cultural biases. What are the implications of these challenges for understanding developmental disorders?

At a practical level, the challenges are considerable. For instance, in the United Kingdom and the United States, the SLT profession is dominated by White, middle-class, English-speaking women. For example, approximately 97% of qualified SLTs in the United Kingdom report belonging to this social demographic (Royal College of Speech and Language Therapists, 2002), and only 5.3% of American Speech–Language–Hearing Association members identify themselves as bilingual (American Speech–Language–Hearing Association, n.d.). Yet, the populations they serve are often very different. In the London borough of Hackney, for example, only 65% of the residents are monolingual speakers of English, with approximately 100 different language communities in residence. This borough is the second most economically deprived region in England, with 67% of school-age children living in low-income families. For many communities living in Hackney, traditional male–female roles are strictly observed, with women largely responsible for maintaining the home and caring for the children, whereas men more typically work outside the home. When these families present at SLT clinics, cultural issues come to the fore. Some examples from our own clinical practice include:

    Although 100 different languages are spoken in Hackney, standardized tests of language and communication in languages other than English are almost nonexistent.
    Bilingual coworkers (interpreters) are often hard to find: Most are women, few have professional qualifications or extensive knowledge of typical language development in a particular language community, and they are often known to the clients, raising issues of confidentiality.
    Families have reported that assessments involving play with dolls and plastic toys are unusual because they do not have experience with such toys at home.
    Families may feel that it is important for their children to learn English so they can succeed at school. This sometimes means families try not to speak to their children in their home languages and focus on educational English, such as learning numbers and colors, rather than natural conversational discourse.
    Families may reject intervention programs that involve following the children's lead in play, talking about what children were doing, or engaging in pretend play, either because it is culturally unusual for adults to play with children or because the family circumstances are not conducive to free play time (cf. Mbise & Kysela, 1990).
    Some families have opted out of therapy activities because of their cultural/religious beliefs (e.g., males are forbidden from hearing women sing).


Perhaps Hackney is an extreme example, but these experiences raise interesting questions for both researchers and practitioners: How do we determine whether or not a child with a very different developmental experience has a disorder? Are the diagnostic categories and treatment expectations derived from White, middle-class, Western societies appropriate for other cultural communities? What can the cultural experiences of children from other backgrounds tell us about our assumptions about the causal pathways of disorder in our own communities?

These are highly relevant issues for at least two reasons. First, ASD in particular is increasingly recognized as a global challenge and more and more research is directed at identifying ASD in countries throughout the world. At the same time, there is growing recognition in mainstream psychology that much of what is published about human perception, cognition, and behavior is derived from research studies conducted in Western societies and that these cohorts reflect a tiny minority of the world's population that is often not representative of the majority (Henrich, Heine, & Norenzayan, 2010). Thus, the conclusions that psychologists draw about typical human development may be limited to a rather narrow subsample of privileged individuals (i.e., those who are educated, live in democratic societies, and have high socioeconomic status). A similar concern may be raised about research investigating developmental disorders, the vast majority of which comes from Western, English-speaking countries (Dyches, Wilder, Sudweeks, Obiakor, & Algozzine, 2004). Application of diagnostic standards, assessment methods, and treatment approaches from these societies to culturally diverse populations is fraught with difficulties, which are detailed later.

Second, there is growing discussion about how psychologists conceptualize children with neurodevelopmental disorders: Do the communication profiles of these children reflect natural variation in language and communication competence or do they signify a qualitatively different and abnormal developmental trajectory? Should psychologists persist with a very Western medical model of disability, in which the problems faced by individuals are part and parcel of their disease, or should a more social model of disability be adopted, which focuses on society's ability to adapt to the variable language and communication needs of the population as a whole? In recent years, the neurodiversity movement has been influential in bringing these issues to public attention, advocating that ASD in particular is natural human variation and should be viewed as a separate minority culture (rather like the Deaf culture) as opposed to a disorder. The degree to which this view is applicable across the entire autism spectrum is a matter of great debate (Jaarsma & Welin, 2011; Pellicano & Stears, 2011) but does force consideration of what constitutes normal social communication behavior. The dimensional perspective of language delay and SLI (Rescorla, 2009) is similar in viewing language deficits as the lower end of a spectrum of linguistic skills, rather than a qualitatively distinct and aberrant language profile. This perspective leads to discussion of how best to measure optimal outcome in adolescents and young adults and whether this should be measured in relation to scores on standardized tests of language or should be based on the individual's ability to function within society (Tomblin, 2008). Tomblin (2006) considered both of these alternatives with reference to philosophical notions of neutralism and normativism. A neutralist perspective identifies disorder as a deviation from the population mean on some observable characteristic. A normative perspective views deviation from the statistical average as a difference that may attract clinical status only if that difference is disvalued by the individual's society. Both standards require consideration of cultural perspectives: For the neutralist, what aspects of language and communication are to be measured and how can this be achieved in a standardized way, when both most naturally occur in a dyadic milieu? Within a normativist framework, what are the cultural expectations regarding education, employment, and social relationships, and how might a difference in linguistic or social–cognitive ability negatively impact on these expectations?

With these issues and debates in mind, we structure our review as follows: First, we consider the practical challenges to the assessment and diagnosis of children from different linguistic and cultural communities. Here, the limitations of a neutralist perspective are most apparent. Second, we consider the potential of a normativist perspective that focuses on the impact of individual differences in linguistic or social–cognitive ability on a child's cultural acceptance and social well-being. Finally, we explore how cross-cultural research can inform cognitive theories of disorder. How people attend to and perceive the social world is culturally specific (Nisbett & Masuda, 2003); thus, the impact of the social context on language and communication development varies across cultures (Ochs & Schieffelin, 1986). Examining these differences may be particularly informative in helping psychologists to refine theories about the cognitive origins of developmental language disorders. We end with a consideration of how researchers and practitioners might proceed in resolving the question of difference or disorder.
Developmental Difference or Developmental Disorder?

In Western society, conceptualizations of SLI and ASD derive largely from a medical model, in which diagnosis reflects a psychobiological dysfunction within an individual that results in an atypical pattern of behavior or psychological function and confers adverse consequences for everyday functioning (Stein et al., 2010). This may also be considered a deficit model; for instance, the proposed criteria for ASD in the Diagnostic and Statistical Manual of Mental Disorders (DSM–5; American Psychiatric Association, 2011) include (a) persistent deficits in social communication and social interaction across contexts and (b) restricted, repetitive patterns of behavior, interests, and activities. Similarly, SLI may be diagnosed when “language abilities are below [emphasis added] age expectations in one or more language domains” (DSM–5; American Psychiatric Association, 2011). Proposed criteria for SLI stipulate that regional, social, or dialectal variations in language should not be considered disorder; similar cultural variations in social communication and social interaction are not explicitly provided in ASD criteria.

Another assumption of the medical model, inherent in a neutralist perspective of developmental disorder (cf. Tomblin, 2006), is that these deficits represent a biologically based deviation from the typical developmental trajectory. There is certainly overwhelming evidence that many, if not all, complex neurodevelopmental disorders arise from multiple genetic (and environmental) influences that alter neurobiological development (Bishop, 2009). Although in a minority of cases these biological influences are qualitatively different (i.e., a genetic mutation that yields physical and cognitive differences), the majority reflect a confluence of normal genetic variants that together increase risk for developmental delay (Constantino, 2011). In addition, the relationship among biology, environment, and behavior is complex and not well understood at present (Fisher, 2006), meaning that even if biological risk factors could be identified precisely, it would not necessarily be helpful in predicting an individual child's cognitive and behavioral profile.

Thus, the challenge for clinicians and researchers is to determine whether or not a child's behavior deviates significantly from that of peers. In the West, diagnosis for language and communication disorders is largely based on performance on standardized measures, in which normative data are provided and it is possible to calculate the extent to which an individual score deviates from the range of scores expected from other individuals of the same age. Application of standardized assessment in our increasingly multicultural societies is fraught with challenges. Carter et al. (2005) identified five broad categories for consideration in cross-cultural assessment: cultural influences on test performance, familiarity with being tested, effects of formal education, test content, and use of pictures in test materials. We condense these five categories into two major considerations: assessment context and assessment content. Assessment context may include influences on performance, familiarity with testing situations, and educational experience but may also include family perceptions of what is a problem, beliefs about the causes of disorders and the stigma attached to them, family resources and experiences, and the availability of services once a disorder has been identified. These contextual factors may influence both the extent to which families seek or comply with assessment and how psychologists interpret a child's performance on any given test. For instance, families who live in extreme poverty where disease, housing, adequate nutrition, and/or family safety are of primary concern may not view communication difficulties as a high priority or have the resources to participate in assessment and treatment. In other cultural settings, children with developmental disorders may be viewed as representing parental or family misdeeds or the outcome of witchcraft (Cappiello & Gahagan, 2009; Dyches et al., 2004). Such families may be reluctant to seek professional services because of the resulting stigma, which may affect employment or marital prospects for family members.

Once families get to a clinic, it is important to be mindful that the testing situation, in which children are asked to complete unusual tasks or to answer numerous questions posed by an unfamiliar adult who almost certainly knows the answers to these questions, may not be a situation experienced by many children outside mainstream Western cultures (Carter et al., 2005). For instance, Carter et al. (2005) reported their experiences with testing the language competence of children in rural Kenya as part of an investigation into the effects of cerebral malaria on language and cognitive development. They discovered that it was highly unusual for Kenyan children to sit and talk with adults, particularly unfamiliar adults, for extended periods of time. Instead, these children were much more likely to spend their time in polyadic situations with siblings and peers, thus making the testing situation culturally unusual from the most basic assumptions about communication. Similarly, seminal work by Saxe (1988) demonstrated that Brazilian street children demonstrated exceptional computational and problem-solving abilities in their daily market selling activities, yet performed poorly on standard classroom measures of mathematical reasoning. Differences in the amount of formal schooling and literacy a child has may also profoundly affect his or her ability to comply with assessment requirements or his or her motivation to respond quickly, accurately, and logically (Carter et al., 2005).

Once these hurdles have been addressed, it is important to consider the assessment content and the degree to which what is being asked is appropriate in different cultural contexts. Here a number of different challenges arise. The first is that some aspects of language and communication are easier to measure than others; thus, a considerable bank of normative data exists for developmental milestones. For example, the age at which first words and phrases are acquired is well documented and is remarkably consistent across countries and cultures, though a large range of normal variation exists (Caselli et al., 1995). Similarly, vocabulary can be readily measured, and though variable, identification of children with limited vocabularies for age is reasonably straightforward. Nevertheless, even for speakers of the same language, cultural variations in vocabulary may prove problematic. For example, administration of vocabulary tests published in the United States to children in the United Kingdom (who ostensibly speak the same language) usually requires substitution of items such as flashlight, soccer, and vacation. Most U.S. children may have similar difficulties defining torch, football, and holiday in relevant U.K. terms. Picture materials further complicate matters: U.S. items for flag (the Stars and Stripes) and country (an outline of the continental United States) are less familiar to children outside of America. Carter et al. (2005) described more fundamental problems with picture recognition for children from rural Kenya who did not attend school and did not have access to photographs or drawings that represented known objects. These children may confuse visually similar items or may be better able to provide a definition, rather than a label, for a familiar object that they have never seen in a picture before. Even when pictures are familiar objects, differences in expected responses can be biased against children who respond with answers that are culturally appropriate but are not the expected response on a standardized test. For example, in clinical practice we have observed that Puerto Rican children in the United States consistently provide descriptive phrases instead of labels during picture vocabulary assessments; for instance, when looking at a picture of a knife, the child may reply, “for cutting.” This may reflect a cultural preference for describing an object rather than labeling it (Peña & Quinn, 1997), but the response would nevertheless be marked incorrect on a standardized test.

Other aspects of language show less variation within a population but are more language/culture specific. For example, in English, children typically acquire adult levels of tense marking (e.g., past tense –ed and third person singular –s) by the age of 5; children who fail to acquire these markers by this age would likely be diagnosed with SLI. However, in non-English-speaking communities, verb morphology would not necessarily be such a good marker of impaired language development, and measures of English vocabulary would reveal only a portion of their lexical skills (Leonard, 2009). Dialectal differences in spoken language can also blur the lines between diverse expression and disorder. Omission of the article the, as in he went to hospital, would be very unusual in mainstream American English but would be perfectly typical in mainstream British English. Similarly, he were angry may cause concern in the United States but is an acceptable form in some dialects of British English.

It has become increasingly clear that there are few standardized assessments that are valid and reliable measures of linguistic structure for children outside of White, Western, middle-class populations (Hirsh-Pasek, Kochanoff, Newcombe, & de Villiers, 2005). In recent years, considerable efforts have been made to distinguish cultural difference from disorder in varying dialects of American English. For example, a recently developed and innovative test battery was specifically designed to identify children at risk for a developmental language disorder in a population whose native language shared some linguistic features of children with a diagnosis of SLI. The Diagnostic Evaluation of Language Variation (Seymour, Roeper, & de Villiers, 2005) identifies children whose speech and language differ from mainstream American English, such as native speakers of African American English (AAE), and it determines whether their language differences are indicative of a potential language disorder or are in line with the typically observed features of the AAE dialect they have learned to speak at home. A pioneering component of the test design, and one of the crucial elements behind the success of the test instrument, was found in identifying language features in which mainstream American English and AAE were similar (noncontrastive) and features in which they differed (contrastive features; Seymour, 2004). To do this, extensive research was undertaken to chart language acquisition across the early elementary years with the goal of documenting contrastive versus noncontrastive features in all domains of language—syntax, morphosyntax, semantics, pragmatics, and phonology—for speakers of these two varieties of English. This allowed Seymour et al. (2005) to create an assessment instrument in which responses to test items given by children who speak AAE (and do not speak mainstream American English) could be identified as typical or deviating from the linguistic patterns observed in their home language, rather than using contrasting features of mainstream English to make normative judgments (Pearson, 2004). In fact, the diagnostic procedures that emerged from the research were so effective that in pilot testing the Diagnostic Evaluation of Language Variation, the researchers found that it was also a valid assessment for speakers of other dialects, including Cajun English and Appalachian English (Seymour, 2004), as well as mainstream American English. Assessments using a similar philosophy are being developed for children in the United States from Spanish-speaking backgrounds, but standardized assessments that use geographically local normative data are lacking for many world languages.

If there are problems with developing culturally sensitive measures of children's linguistic structure, it is even more of a challenge to develop measures of pragmatic language and social communication skills, yet deficits in pragmatics and social communication may be equally detrimental to social and academic success. This was dramatically demonstrated in one study of 242 seven-year-old children attending specialist language units and schools for SLI in England, in which their teachers and SLTs independently reported whether children displayed deficits in articulation, phonology, syntax/morphology, or semantics and pragmatics. For a substantial minority of students (∼30%), teacher/SLT report was the only indication that the child had a communication impairment severe enough to warrant a specialist educational placement. These children had significant pragmatic language deficits that were not identified by standardized tests of language competence (Botting, Conti-Ramsden, & Crutchley, 1997), highlighting the limitations of extant assessment instruments for measuring children's pragmatic language competence. This is no small problem given that pragmatics is the one pervasive dysfunction that is observed in all children with a diagnosis of ASD (Rollins & Snow, 1998; Tager-Flusberg, Paul, & Lord, 2005).

Pragmatic aspects of language are notoriously difficult to measure in standardized ways because they are a set of contextually dependent human behaviors that occur in dyadic exchanges; the formal structure of a standardized testing procedure makes it difficult to capture pragmatic problems that may arise in everyday situations where the rules of engagement are less explicit and highly dynamic (Adams, 2002; Norbury, Nash, Baird, & Bishop, 2004). Pragmatic language abilities are also highly susceptible to cultural variation: Discourse rules, such as turn taking, interrupting, appropriate topic choices, use of eye contact and other nonverbal strategies for maintaining interaction, use of humor, and the ability to question and challenge communication partners, are largely determined by cultural rules and the child's relationship with his or her interlocutor (Carter et al., 2005). There are also far fewer hard and fast typical norms for such behaviors. This is particularly true for social interactive behaviors; for example, the Autism Diagnostic Observation Schedule (ADOS; Lord, Rutter, DiLavore, & Risi, 1999) weights appropriate eye contact heavily in its diagnostic algorithm, yet there are few established norms for how much eye contact can and should be expected between children and unfamiliar adults. There may also be considerable variation within the typical population and within an individual depending on the task or topic of conversation. For instance, Norbury et al. (2009) recorded eye movements of adolescents as they watched dynamic scenes of social interaction. As expected, there were many adolescents with ASD who spent little viewing time fixating the eye region of the protagonists, but that was also true of typical peers, with viewing times to the eyes ranging from 0% to 70% in both groups. Similarly, Nadig, Lee, Singh, Bosshart, and Ozonoff (2010) reported that both individuals with ASD and their typically developing peers increased eye gaze to a conversational partner when discussing a topic of interest, rather than a generic topic proposed by the examiners. Thus, although the combination of behaviors included in the ADOS algorithm robustly distinguishes children with ASD from peers, normative developmental data regarding the individual behaviors that make up the algorithm are required.

To overcome some of these challenges, parents or teachers are often asked to provide ratings of pragmatic or social communication/interaction behaviors, often with parents and teachers providing different outcomes for the same child (Norbury et al., 2004; Redmond & Rice, 2002). These discrepancies very clearly illustrate the influence of the cultural standards with which children are compared. For example, when asked to rate friendships, parents of children with ASD may often remark, “Well, he has friends, but probably not the same kind of friendships that you or I would have.” Similarly, one parent's report of a child who recites well-known animated cartoon scenes may fit well within the diagnostic category of restricted, repetitive, and stereotyped behavior, whereas another parent may take pride in their child's ability to perform these sequences. This suggests that a sense of what constitutes stereotyped patterns of interest or inflexible adherence to routines may also vary by family or by culture.

Despite these limitations, the few social–pragmatic assessments that exist are being adopted in quite diverse cultural contexts, increasing the need to consider the cultural implications for task performance. For example, there are currently efforts underway to translate the Children's Communication Checklist–2 (Bishop, 2003) into 30 different language and/or cultural contexts (D. Bishop, personal communication, July 28, 2010), though it was standardized on the communication profiles of British schoolchildren. Although great care is taken to translate test items into the relevant language, empirical investigations of the clinical utility of the instrument in different communities are rare (Geurts et al., 2004; Helland, Biringer, Helland, & Heimann, 2009; Ketelaars, Cuperus, van Daal, Jansonius, & Verhoeven, 2009). The need to understand cultural values in interpreting pragmatic language abilities is evident in the different cultural expectations of successful narrative discourse, a skill highly valued in the West and crucial for early literacy learning (de Villiers, 2004). In Western, English-speaking cultures, people tend to value narratives that are semantically and grammatically complex, with a clear temporal and causal structure. In Hispanic culture, narrators appear to be less concerned with the temporal organization of a story and place much more emphasis on evaluative devices within the narrative, stressing the emotional importance of events and making the story interesting for the listener (Silva & McCabe, 1996; Sparks, 2008). Ensuring that pragmatic assessments tap a range of skills for which there are appropriate, culturally specific norms is necessary for discriminating children with pragmatic differences from those children who would be regarded by members of their own cultural community as having pragmatic deficits. The Diagnostic Evaluation of Language Variation is an example of a measure that does just that, with normative data on the narrative and broader pragmatic skills of AAE speakers (de Villiers, 2004).

Similar cultural challenges exist for the ADOS; there are now at least 17 translations of the ADOS underway, though the effects of potential cultural confounds on the validity and reliability of the instrument have not been thoroughly examined (Bernier, Mao, & Yen, 2010). This is surprising given clear differences in cultural expectations surrounding social-communication behavior. For instance, two behaviors, eye contact with adults and pointing with the index finger, were deemed to be inappropriate for children from a Chinese cultural background (J. Zhang, Wheeler, & Richey, 2006), though these behaviors are heavily weighted in the published diagnostic algorithm of the ADOS (Lord et al., 1999). For Chinese children then, lack of eye contact with adults or using the index finger to point would be a sign of difference from Western social expectations, not a deficit. There may be further cultural issues surrounding the extent to which adults engage in pretend play with children, the extent to which they comfort children with public displays of emotion or discuss emotional states and events openly, or the extent to which children (and particularly boys) are familiar with or likely to play with miniature toy dolls (particularly ones with moving eyes that have positively frightened a few of our clients). In addition, Ametepee and Chitiyo (2009) noted that although only a few studies on autism have come from African countries, two of them reported that certain stereotypical behaviors (e.g., hand flapping or rocking) observed in children with an ASD diagnosis in Western countries were uncommon in children with the same diagnosis in Africa. Thus, an important avenue for future research is to start with the indigenous community to develop a measure that is sensitive to individual differences for a given set of communication behaviors within that community.

However, even if psychologists ensure that children and families are familiar with the assessment context and that the content of the assessment is culturally appropriate for the population they are hoping to serve, it is still a major challenge to determine how deviant scores on standardized measures need to be to warrant a diagnosis. There is little clinical or research consensus on this issue. For instance, researchers may consider anyone falling below the 25th percentile to be impaired, or they may consider only those individuals scoring more than two standard deviations below the normative mean (bottom 3%) to be impaired. There is also little consensus on what pattern of behavioral impairment is the most clinically significant. For instance, a diagnosis of SLI may include deficits in different language domains, such as phonology, vocabulary, syntax, and social use of language. However, a child with a circumscribed impairment in phonology has needs very different from a child with impaired development across multiple domains. Likewise, a number of social communication behaviors are included under the umbrella of ASD, but often outcome is determined by associated factors, such as language and cognitive profile, rather than by severity of social-communication impairment.
Taking a Normativist Perspective: What Is the Impact of Individual Differences in Language and Social-Communication Behavior?

Increasingly, diagnostic frameworks stress the importance of evaluating the impact of disorder on everyday well-being, although standard methods of assessing this impact are not well developed in the area of children's language and social communication. One method for systematically considering impact is offered by the International Classification of Functioning, Disability, and Health (World Health Organization, 2002). This framework considers the biological difference experienced by the individual (e.g., the genetic and neuroanatomical risks that are associated with difficulties in learning grammatical rules) and how this difference (which results in expressive language scores significantly below age expectations) interferes with the individual's activity and participation in daily events. It also advocates consideration of contextual factors, such as social attitudes and beliefs about impairment and practical obstacles to well-being. Contextual factors, particularly those related to local beliefs about child development, may be highly influential in determining which children attract clinical attention or how disorders are viewed in different societies. Tomblin (2006, 2008) considered impact from a normativist perspective. Here, individual differences in behavior and cognition may assume clinical importance only if the impact of that difference is disvalued by the individual's culture. The degree to which an outcome is disvalued by a culture may be gauged by the extent to which caregivers feel compelled to intervene and assist a child in developing a particular skill. For instance, literacy is highly valued by Western, industrialized cultures; thus, children with limited reading abilities are at high risk of negative outcomes, realized by reduced academic attainments and poor employment prospects (cf. Tomblin, 2008). Children in these cultures with a biological difference that confers risk for literacy difficulty are therefore likely to be diagnosed with a reading disorder and labeled dyslexic or poor comprehenders in order to receive remedial services. Children with the same biological risk living in cultures in which literacy is not highly valued or where a lack of literacy is not a barrier to social or economic success would not be labeled as having a disorder.

An illustration of the importance of considering the impact of individual differences and the cultural values attached to those differences may be illustrated by the variable prevalence rates of ASD across the globe. Table 1 recounts current rates of ASD (per 10,000) in different countries around the world. The reasons for different prevalence rates have been discussed extensively (Fombonne, 2003) and may include sample size, ascertainment methods (i.e., medical records review vs. direct assessment), case definition, age at diagnosis, and availability of autism services (Kogan et al., 2009), though biological factors and adverse environmental exposure cannot be ruled out. Although methodological factors almost certainly explain a large amount of the variance in prevalence rates between Western societies and developing countries, some differences are not so easy to reconcile. For instance, both the United Kindom and Denmark are wealthy, democratic, Western European societies, and yet a child in the United Kingdom is 1.7 to 9.6 times more likely to be diagnosed with ASD than a child in Denmark. There are also often discrepancies in prevalence rates within the same country; for instance, in the United States, prevalence rates across the states range from 42 to 121 per 10,000, though this may be linked with availability of autism-related services (Autism and Developmental Disabilities Monitoring Network Surveillance Year 2006 Principal Investigators, 2009). Sharp differences are also reported for different ethnic groups within the same country (Dyches et al., 2004; Lord & Bishop, 2010). Although this may be confounded by factors such as socioeconomic status, this does not appear to be the whole story. For instance, prevalence rates for ASD diagnosis in Hispanic communities are significantly lower than those reported for non-Hispanic communities, even when rates are adjusted for socioeconomic factors (Palmer, Walker, Mandell, Bayles, & Miller, 2010). The picture for African American children is more mixed. Earlier studies reported that African American children were 40% more likely than White peers to be receiving special education services for ASD (Dyches et al., 2004). On the other hand, Kogan et al. (2009) reported that the odds of having an ASD diagnosis were 57% lower for African American children than for White children. The discrepancy could reflect methodological differences in case ascertainment but may also be indicative of the fact that African American children are diagnosed on average 18 months later than White peers (Mandell, Listerud, Levy, & Pinto-Martin, 2002) and are more likely to have changing diagnoses (Mandell, Ittenbach, Levy, & Pinto-Martin, 2007).

dev-49-1-45-tbl1a.gifPrevalence Rates of Autism Spectrum Disorder Around the Globe

When low rates are reported, there is an implicit assumption that these rates must somehow be erroneous and that improved access to medical and educational resources, in combination with more robust diagnostic procedures, would increase consistency in identification rates across countries. For instance, a press release from Autism Speaks (2011) reported on recently published prevalence research conducted in South Korea:

    If researchers look carefully, especially in previously understudied, non-clinical populations, they may find more children with ASD. In addition to the South Korean study, Autism Speaks is supporting similar epidemiological research efforts in India, South Africa, Mexico, and Taiwan, including the translation and adaptation of the gold-standard diagnostic instruments into languages spoken by more than 1.7 billion people worldwide.

Alternative explanations may be that differences in prevalence rates reflect protective factors that exist within that community (Palmer et al., 2010) or that in some communities, the symptoms of ASD do not interfere with everyday life. Some communities and cultures appear to be more accepting or better able to accommodate individuals with language, cognitive, or social-communication impairments, with diagnostic labels largely absent from these communities (Dyches et al., 2004). Dyches et al. (2004) discussed certain cultural values, for instance, emphasis on family co-operation and a common concern for the well-being of all individuals within the family, found in Latino, African American, and Native American communities that may inspire a positive appraisal of children with disabilities. It has also been suggested that cultures that define human worth on the basis of holistic or spiritual values rather than physical or cognitive ability or those that depend on family productivity rather than individual success for economic security may view developmental differences as less problematic (Cappiello & Gahagan, 2009). Within Hispanic communities, religious beliefs and extended family networks could reduce the impact of developmental differences, though future research is needed to uncover whether differences in genetic vulnerability or environmental exposure may be responsible for higher diagnostic rates in non-Hispanic White compared with Hispanic communities (Palmer et al., 2010).

Inspection of Table 1 indicates that some cultures appear to have higher than expected prevalence rates, with the highest reported in East Asian communities, such as China, Japan (Sun & Allison, 2010), and more recently South Korea (Kim et al., 2011). This is somewhat surprising given that the stigma surrounding diagnosis of psychiatric disorder in these cultures (Lauber & Rossler, 2007) could be expected to reduce prevalence estimates. On the other hand, these cultures are renowned for their complex social hierarchies and prescribed social role relations, and an appreciation of contextual factors is paramount to effective functioning (Nisbett & Masuda, 2003). Thus, children who are unable to interpret the mental states of others, insensitive to social rules, and fail to observe or integrate contextual information may be more likely to attract attention and have difficulties succeeding in social and educational environments.

The recent investigation by Kim et al. (2011) speaks to these possibilities. At the time of writing, this study reports the highest known prevalence rates of ASD in a total population sample of school children living in South Korea: 260 per 10,000, or 1 in 38 children. The investigators used a rigorous sampling and assessment protocol that included the gold standard instruments for research diagnosis of ASD in Western North American and European societies: the ADOS and the Autism Diagnostic Interview–Revised (Rutter, LeCouteur, & Lord, 2003). These instruments had been translated and adapted for use in South Korea (though the changes made are not reported in the published article), and diagnosis was based on best clinical estimate using clinicians trained in both North America and South Korea in an effort to reduce “potential cultural biases in case identification” (p. 2). The sample included both a high-probability group of children on the disability register or receiving special educational supports, and a general population sample of children in mainstream schools. An intriguing result of this study was that 60% of children who screened positive and were sampled from the mainstream setting, for whom there was no reported history of educational concern or referral for psychological assessment, met diagnostic criteria for ASD. The conclusion is that there are large numbers of undiagnosed children with ASD in the general population, with consequent implications for improving identification and service provision.

However, this particular finding raises some important questions about difference and disorder and the need to incorporate impact into our diagnostic frameworks. If these children are succeeding in mainstream schools and are not causing concern, should we diagnose them with a disorder? What would the advantages and disadvantages be of labeling more children with ASD?

Before we can answer that question, we need to consider whether a cultural variable hinders identification, even if this may not be in the child's best interest. For example, stigma surrounding identification of disorder may prevent some parents from actively seeking assessment even though they may be concerned about their child's development. This does not seem to explain Kim et al.'s (2011) findings; it is likely that families who actively wished to avoid diagnosis would not consent to take part in the study. Indeed, only 13.4% of the families with children who screened positive for an ASD in mainstream settings were seen for in-depth evaluation, either because families withheld consent or did not attend further assessment. An alternative explanation is that rather than having a disorder, these children represent the tail end of the normal variation in social communication behavior within this culture but that they are displaying traits that are regarded as symptoms of ASD on the ADOS. For example, making eye contact with an unfamiliar adult may be socially inappropriate in South Korea (cf. Blais, Jack, Scheepers, Fiset, & Caldara, 2008), but observed low levels of eye contact could tip the balance on the ADOS algorithm toward a diagnosis of ASD. The research team did rigorously attempt to account for potential cultural differences, by interviewing parents and teachers to identify local beliefs and knowledge about ASDs that could influence the way parents report symptoms or consent to participate in the study and complemented rigorous translations with consistent behavioral coding across examiners from both North America and South Korea. Nevertheless, we must acknowledge that such studies are starting from a Western cultural perspective on ASD symptomology, and there is no independent assessment of the sensitivity and specificity of a modified ADOS for distinguishing children with ASDs from other populations in South Korea.

Interestingly, the children identified in mainstream settings had milder symptom profiles and higher IQs than those identified from the disability registrar and appeared to be succeeding in the mainstream school setting. Kim et al. (2011) noted that the school system in South Korea is highly structured and emphasizes behavioral regulation and academic success more than social activity. Thus, able children with mild social-communicative differences may be able to meet societal expectations of behavior and academic attainment. Should a diagnosis of disorder be restricted to those children with multiple developmental concerns who are unable to fully participate in school and home environments? Or does this finding suggest that we in the West could learn from the experiences of the South Koreans about how to include children with diverse social-communication abilities/styles in mainstream settings? Future research will need to address the question of whether the exportation of diagnosis and treatment models for neurodevelopmental disorders from Western countries is helping children and families from other cultural backgrounds or whether researchers are at risk of pathologizing people who may be well adjusted within their own communities.

A similar argument could be made for identification of children with SLI. The most commonly cited prevalence statistic for SLI comes from a population study of children starting school in Iowa (Tomblin et al., 1997). In this study, SLI was assessed with a battery of tests that tapped three language domains (vocabulary, grammar, and narrative) in two modalities (production and comprehension), yielding five composite scores. SLI was diagnosed if at least two of the five composite scores was more than 1.25 SD below the normative mean (bottom 10th percentile), yielding a prevalence estimate of 7.4%. An intriguing finding from this study was that only 29% of children who met these research criteria for SLI had been identified by parents or practitioners as having language difficulties. If more stringent severity criteria were used to include only those children with composite language scores of −2 SD or greater, the overall prevalence rate was reduced, but the percentage of children clinically referred for language difficulties only rose to 39% (X. Zhang & Tomblin, 2000). This suggests that the features that lead to identification of SLI in everyday circumstances may be different from those identified by standardized tests (Bishop & Hayiou-Thomas, 2008). Specifically, children with poor speech articulation and limited verbal output are more likely to be referred to clinical services (Bishop & Hayiou-Thomas, 2008; Zhang & Tomblin, 2000). Perhaps these deficits are more easily observed by adults because they blatantly interfere with the child's ability to be understood and participate fully in social interactions and educational activities.

These studies suggest that screening without taking account of impact has the potential to overidentify children that are otherwise fine. If the benefits of identifying unsuspecting families with clinical disorders outweighed the costs of potentially misdiagnosing as impaired an otherwise healthy child, we might conclude that this was acceptable. However, it is notable that population screening of the kind reported here has not been recommended for either disorder (Al-Qabandi, Gorter, & Rosenbaum, 2011; Nelson, Nygren, Walker, & Panoscha, 2006). The primary reasons are that even in Western societies, we do not have the capacity to treat all children identified on screening measures as at risk, nor is there a sufficient evidence base with which to recommend effective treatments to improve outcomes for children with ASD or SLI. This is even more likely to be the case in countries outside of North America, Western Europe, and Australasia in which clinical and educational services are in their infancy. Thus, the cost of diagnosis could be increased family concern and distress with little support from clinical or educational services. Others have argued that although it is important to measure impact, this should not be taken into account in the diagnostic process (Rutter, 2011), the reason being that in other areas of medicine, it is useful to know that an individual has a condition even if it has little impact at the present time (i.e., high blood pressure or diabetes). This assertion highlights the fact that impact is not static but changes with developmental time. Tomblin (2008) noted that although many kindergartners with low language scores did not attract clinical attention at school entry, they were more likely to experience later difficulties with reading comprehension, and these difficulties did impact on their educational outcomes. He therefore argued that it is important to identify and monitor early vulnerabilities to circumvent later adverse outcomes. Longitudinal studies charting the developmental outcomes of school-age children with social-communicative vulnerabilities are lacking, but studies of infant siblings at risk of ASD show remarkably varied developmental outcomes in early childhood (cf. Elsabbagh & Johnson, 2010). It may be that the children identified by Kim et al. (2011) succeed in a highly structured educational setting but that they experience more adverse impacts when they leave school and must find employment or fulfilling social relationships. Longitudinal studies following these population cohorts are urgently needed to inform decisions about difference or disorder.

In sum, developmental disorders of language and communication bring issues surrounding diversity into sharp focus. Although we know that these disorders have a biological basis, the impact of these biological differences on the individual's ability to participate fully in society is largely culturally determined. Prevalence rates for disorder vary dramatically across country and cultural boundaries. This is no doubt due in large part to differences in sampling procedures, diagnostic definitions, assessment, and availability of resources. However, we should not discount the possibility that cultural values may in some circumstances magnify behavioral differences and increase the impression of disorder. Shifting from a neutralist model to a more normativist model of disability (as suggested by Tomblin, 2006) may encourage researchers to think of ways that biological differences in language and social-communication behavior may be more readily accommodated in mainstream society.
Why Are Cross-Cultural Comparisons Crucial to Informing Theories About the Cognitive Origins of Developmental Language Disorders?

Despite these challenges, cross-cultural studies are vital in allowing researchers to identify universal features of disorder and to refine hypotheses about the cognitive bases of developmental disorders. Two examples are illustrative of this point and are discussed next in relation to SLI and ASD.
Cognitive Theories of SLI

SLI has often been seen as a test case for the notion that language development is supported by specific, innate neural architectures and that SLI arises from selective impairment to this language module (van der Lely, 2005). The very label specific language impairment denotes that language may be selectively impaired in the face of otherwise normal cognitive development. In English, this is a very appealing hypothesis; obligatory morphosyntactic structures, such as third person singular –s, are omitted from expressive language long after it is developmentally appropriate to do so, and these grammatical errors are not related to vocabulary or nonverbal reasoning (Rice, Tomblin, Hoffman, Richman, & Marquis, 2004). Even more puzzling is that phonologically identical suffixes, such as plural –s, which do not mark grammatical relationships, do not challenge English-speaking children with SLI (Oetting & Rice, 1993; Rice & Oetting, 1993). A parsimonious explanation is that children with SLI experience a delayed maturation of the grammatical system that supports tense marking (Rice, Wexler, & Hershberger, 1998). However, cross-cultural studies have queried this assumption. Most notably, Italian- and Spanish-speaking children with SLI do not omit affixes marking tense and agreement in their spontaneous speech at greater rates than peers matched for mean length of utterance (Bedore & Leonard, 2005; Bortolini, Caselli, & Leonard, 1997). In these languages, tense and agreement inflections are syllabic and appear in stressed or word final positions, increasing their perceptual salience. In contrast, corresponding inflections in English are not syllabic, tend to be unstressed and unvoiced, and are therefore not perceptually salient. These observations led to development of the surface hypothesis (Leonard, 1998), which postulated that children with SLI had a limited capacity for cognitive processing in combination with perceptual weaknesses. This domain-general impairment could be particularly deleterious for learning morphosyntactic forms in English because of the phonological properties of these verb inflections in connected English speech:

    If inflected words were typically heard in one-word sentences separated by pauses, there would be no problem. However, fast on the heels of the inflected word is the next word in the utterance that must be held in working memory and processed, and so on. Thus, processing is pressed from two directions; processing of a first item must be completed before the item fades from memory, and it must be processed in time for the next item. Given the reduced speed of processing assumed for children with SLI, sufficient processing of one item can't be completed before the next item appears. Consequently, some material is processed incompletely or not at all. In a language like English, it is reasonable to expect that if an inflected word is incompletely processed, only the bare stem will be retained. (Leonard, 1998, p. 251)

Further delineation of the SLI phenotype across cultures and across different developmental disorders (e.g., moderate sensorineural hearing loss) has led to refinement of the surface hypothesis (Leonard, 2009) and has begun to suggest some universals in SLI that occur across languages. These may include protracted rates of acquisition of first words and phrases and deficits in implicitly learning and/or remembering novel motor or phonological sequences and complex rules (Ullman & Pierpont, 2005). Assessment of these underlying cognitive skills has the advantage of not requiring explicit linguistic content, which heralds the possibility of an instrument that may be more universally applied (Campbell, Dollaghan, Needleman, & Janowsky, 1997).
Cognitive Theories of ASD

Over the past 30 years, there has been considerable research effort to specify the aspects of autistic cognition that could fully explain the behavioral phenotype. Many investigators have focused on the core social deficit and have attempted to explain how atypical social interactions may adversely affect development of social cognition (Baron-Cohen, 1995; Klin, Jones, Schultz, & Volkmar, 2003). There are varying manifestations of this theory, but substantial credence has been given to the importance of attending to the eye region of faces. Specifically, it has been suggested that individuals with ASD fail to orient to relevant social cues, particularly the eyes. Empirical research supports this observation; numerous eye-tracking studies of Western North American and European individuals with ASD have demonstrated reduced fixation time to the eyes, with some reporting a corresponding increase in fixation time to the mouth, relative to neurotypical peers (Klin, Jones, Schultz, Volkmar, & Cohen, 2002). If eyes are hypothesized to be important conveyors of mental and emotional states of others, a failure to engage in prolonged eye contact will result in reduced opportunities to learn about the internal states of others and a protracted rate or qualitatively different development of a theory of mind. Reduced attention to the eyes may further contribute to abnormalities in social interaction, social learning, and social knowledge.

However, cross-cultural investigations suggest that how people look at faces is shaped by cultural experiences. Comparison of Western Caucasian and East Asian children (Kelly et al., 2011) and adults (Blais, Jack, Scheepers, Fiset, & Caldara, 2008) has demonstrated that individuals from Asian cultures do not preferentially fixate the eye region of the face, instead directing fixations centrally to the nose. This likely reflects cultural expectations that direct eye contact is considered rude in many Asian cultures. On the other hand, one could argue that these findings are irrelevant for a social deficit hypothesis of ASD; regardless of face scanning patterns of older children and adults, it is possible that infants universally demonstrate a preference for fixating the eye regions of caregiver faces and that this is a developmental prerequisite for social cognitive development. Cross-cultural data to support this hypothesis are lacking, but there are data demonstrating that the preference for central (nose) fixations is evident in 9-month-old Asian infants (S. Liu et al., 2011) and that Western infants demonstrate considerable variation in fixation patterns, with fixation to eyes in infancy not generally predicting diagnostic status or social competence in toddlerhood (Young, Merin, Rogers, & Ozonoff, 2009).

We do not wish to suggest that eye contact and visual scanning of faces is irrelevant to social development or that differences in face scanning will not help researchers to understand the developing cognitive phenotype of ASD. However, we suggest that these cross-cultural findings of typically developing individuals indicate that reduced visual fixation to the eye region of human faces is not necessarily maladaptive, nor need it lead to disrupted social understanding. Instead, there may be different routes to deriving social information from people and situations (Akhtar & Gernsbacher, 2008); these may be accentuated by cultural differences but may also reflect typical variation within a culture, rather than disordered development.

Cross-cultural comparisons of neurotypical individuals have also been key in unraveling the causal connections between observed behavior and autistic cognition in another influential theory, the weak central coherence theory (Frith, 2003). Again, there are different manifestations of this theory, but it essentially argues that individuals with ASD lack the typical drive to integrate information holistically at a global level and instead preferentially focus on local details, resulting in a fragmented perceptual and cognitive experience of the world (Happé & Frith, 2006). Cross-cultural investigations, however, have demonstrated that strong local biases do not necessarily go hand in hand with disrupted global percepts. Davidoff, Fonteneau, and Fagot (2008) studied local and global biases in the remote Himba culture of northern Namibia. Adult members of the tribe showed exceptionally strong local biases when making similarity matches to the Navon hierarchical images, in which a shape (large circle made up of small xs) can be matched to another shape at either a global level (large circle made up of small circles) or a local level (large square made up of small xs). The Himba had stronger local preferences than other typical adult populations, and their preferences were in line with results reported for Western individuals with ASD (Happé, 1999). Yet, these same individuals displayed similar patterns of global processing on measures of face recognition, arguing against a direct causal relationship between local processing biases and the behavioral manifestations of ASD.

Unfortunately, we know of no studies that directly compare the cognitive phenotypes of children with ASD from different cultural backgrounds. Such studies could be particularly informative; for example, none of the eye-tracking studies of Western individuals with ASD has explicitly calculated fixation time to nose regions, nor do we know if East Asian individuals with ASD would show face-scanning patterns more typical of peers with the same cultural identity. Such findings would further elucidate the importance of direct eye gaze for understanding the social deficits that characterize ASD, particularly if data were obtained in longitudinal studies that delineated the developmental trajectories of gaze patterns, social behavior, and social cognition.

Cross-cultural studies could also be hugely informative in identifying potential protective factors and their influence on the developmental course of the disorder. For instance, as noted earlier, cultures differ in their preferences for attending to local detail or global context. In contrast to the extreme local biases in Himba culture, East Asian cultures may be at the opposite extreme; they are characterized as holistic cultures, in which attention to the surrounding context is emphasized over attention to focal objects within the context (Nisbett & Masuda, 2003). It would be beneficial to discover whether a child with ASD raised in a more holistic culture would show the same local biases reported for Western children with ASD and/or whether, like the Himba, global processing would be indistinguishable from that of peers because of the pervasive and explicit focus on contextual factors. In a similar vein, many Asian languages explicitly mark the speaker's certainty about belief statements, which may facilitate performance on false-belief tasks for young, typically developing children (D. Liu, Wellman, Tardif, & Sabbagh, 2008; Matsui, Rakoczy, Miura, & Tomasello, 2009). It would therefore be theoretically interesting to discover if these salient linguistic markers support development of theory of mind in Asian children with ASD, just as salient linguistic structures in languages such as Italian and Spanish support production of those structures for children with SLI growing up in that linguistic culture. Thus, cross-cultural studies not only can help researchers to delineate universal features of disorder as well as culturally specific phenotypes but can also help them to refine cognitive theories and identify environmental supports for cognitive development.
Implications for Practice and Research

We began this review with a stark illustration of the challenges faced by clinicians and researchers working in an increasingly global and multicultural society, which prompted us to question whether what we see is a difference or a disorder. There is little doubt that conditions such as SLI and ASD are universal, biologically influenced, and cross country and cultural boundaries. However, we have argued in this article that there is far less consensus on where the boundary between difference and disorder lies. We have also argued that decisions about where to draw the line and the materials used to inform those decisions are heavily influenced by cultural experiences and expectations. How clinicians intervene and support children with developmental disorders and the conclusions they come to about the cognitive underpinnings of disorder are also subject to cultural biases.

This means that clinical practitioners must be mindful of the cultural environment of individual children and their families. The need for SLTs and psychologists from diverse cultural backgrounds is great, as is the need for more culturally appropriate assessment practices (cf. Carter et al.,