Psychosocial and Adaptive Deficits Associated With Learning Disability Subtypes

Rationale for Third Method SLD Approaches in Research and Practice

Although the SLD statutory definition (i.e., processing deficit) was codified in the 1970s, policy makers have always struggled with SLD regulatory language. The statutory–regulatory divide was first sanctioned by the U.S. Department of Education in 1975 when they mandated ability–achievement (A-A) discrepancy for SLD identification. Although some still support a nomothetic global (g) factor/IQ interpretation of intellectual test data (Watkins, Glutting, & Youngstrom, 2005) for A-A discrepancy calculation, global scores have been subsequently found to have limited diagnostic and predictive validity for children with SLD and other disabilities (e.g., Hale, Fiorello, Kavanagh, Holdnack, & Aloe, 2007). Not surprisingly, A-A discrepancy has been maligned for its limited reliability and validity, inability to distinguish SLD from low achievement, and overidentification of children from diverse backgrounds (e.g., Dombrowski, Kamphaus, & Reynolds, 2004; Fletcher et al., 1994; Hale, Fiorello, et al., 2008; Kavale, Kaufman, Naglieri, & Hale, 2005; Siegel, 1989). Given these A-A discrepancy limitations, early research on the academic and behavioral characteristics of children with discrepancy-defined SLD may need to be reconsidered.

As a multitiered system that includes regular curriculum-based progress monitoring and interventions for struggling students (Brown-Chidsey & Steege, 2005), the RTI approach has been touted as a viable alternative to A-A discrepancy for SLD identification (e.g., Reschly, 2005). Although chronic poor RTI is a government-sanctioned method for SLD identification under IDEA (Vaughn & Fuchs, 2003), it is not sufficient because it disregards the SLD statutory definition (as is the case with A-A discrepancy), it does not assess children in all areas of suspected disability (e.g., sustained attention, response inhibition, mental flexibility, long-term memory retrieval), and there are many reasons children do not respond (e.g., instructional mismatch, parental divorce, adjustment disorder, poor intervention fidelity) to the best intervention efforts (Dixon, Eusebio, Turton, Wright, & Hale, 2011; Fiorello, Hale, & Wycoff, 2012; Flanagan, Ortiz, Alfonso, & Dynda, 2006; Hale et al., 2010; Hale, Kaufman, Naglieri, & Kavale, 2006; Schrank, Miller, Catering, & Desrochers, 2006; Willis & Dumont, 2006; Wodrich, Spencer, & Daley, 2006).

Part of the problem for using RTI for SLD research and practice is there is no true positive in an RTI model (Hale et al., 2010). A true positive is said to occur when the individual has the condition (e.g., SLD) and test results indicate the condition is present (e.g., the evaluator rejects the null hypothesis that no condition exists; e.g., Anastasi & Urbina, 1997; Strauss, Sherman, & Spreen, 2006). With RTI there is no null hypothesis to be tested, so there cannot be a true positive in an RTI model. Without a true positive, there is no false positive, true negative, or false negative, and the sensitivity and specificity of the measures in identifying the disorder cannot be ascertained. This significant problem could explain why no study to date has successfully used RTI methods to determine SLD (e.g., responder/nonresponder status), with different criteria leading to different children identified as SLD (e.g., Barth et al., 2008; Fuchs, Fuchs, & Compton, 2004; Speece, 2005; Waesche, Schatschneider, Maner, Ahmed, & Wagner, 2011). As is the case with A-A discrepancy, using RTI for SLD identification for research on academic and behavioral functioning for this population would be quite limited. As a result, a third viable alternative is needed to identify SLD and conduct SLD research.

Operationalizing Processing Strengths and Weaknesses for SLD Research and Practice

With the questionable validity of A-A discrepancy and RTI methods for identifying SLD, and IDEA (2004) Child Find requirements for comprehensive evaluation, children who do not respond to instruction/intervention should receive comprehensive evaluations in all areas of suspected disability (e.g., Dixon et al., 2011), which includes assessment of processing strengths and weaknesses in relation to achievement deficits per IDEA (Hale et al., 2010). The majority of measures available currently are designed to measure multiple cognitive constructs, not a solitary IQ or a single g factor (Elliott, Hale, Fiorello, Dorvil, & Moldovan, 2010; Fiorello et al., 2007; Flanagan, Ortiz, & Alfonso, 2013; Hale et al., 2007; Hale, Fiorello, et al., 2008; Hale, Flanagan, & Naglieri, 2008; McGrew & Wendling, 2010), making them useful in identifying individual processing strengths and weaknesses so both IDEA (2004) SLD statutory and regulatory requirements are met when a child is identified with the disorder (Flanagan et al., 2010; Hale, Flanagan, et al., 2008; Wright et al., 2013).

There are three main processing strengths and weaknesses approaches currently advocated in the field (see Hale, Flanagan, et al., 2008). The present study uses the concordance-discordance model (C-DM; Hale & Fiorello, 2004) of SLD identification because it can be used across cognitive and achievement measures and has a statistical test of the null hypothesis regarding the presence of a SLD. Establishing a cognitive strength, a cognitive weakness, and an achievement deficit associated with that weakness (see Figure 1), the C-DM approach uses the standard error of the difference (SED) formula (Anastasi & Urbina, 1997) to test the null hypothesis that the child does not have a processing weakness(es) or have an achievement deficit(s) relative to their cognitive strength(s), thus making it a viable, statistically sound method for determining SLD. Hale and colleagues (2010) admonish users to not just take the highest and lowest scores or rigidly follow cutoff criteria. Instead, they provide practitioners with an eight-step process for using C-DM to ensure IDEA (2004) SLD statutory and regulatory requirements are met (Hale et al., 2010).

OPEN IN VIEWER

Figure 1.

The Hale and Fiorello (2004) concordance-discordance model of specific learning disability identification.

Through use of C-DM within the cognitive hypothesis testing (CHT; Hale & Fiorello, 2004) approach to comprehensive evaluation, practitioners can generate, examine, and evaluate hypotheses about a child’s processing characteristics, achievement, and behavioral functioning to develop subsequent intervention strategies on the basis of child’s strength and weaknesses (Fiorello, Hale, Decker, & Coleman, 2009). The CHT/C-DM approach has been used successfully in previous single subject and group research with multiple measures and populations to study reading, math, and psychosocial problems associated with SLD (e.g., Elliott et al., 2010; Fiorello et al., 2012; Hain et al., 2009; Hain & Hale, 2010; Hale et al., 2006; Hale et al., 2007; Hale et al., 2011; Hale, Fiorello, et al., 2008; Hale, Hain, Murphy, & Cancelliere, 2012; Mascolo, Kaufman, & Hale, 2009; Reddy & Hale, 2007). Thus, the C-DM model holds promise for current research and practice because it identifies a true positive and SLD subtypes and is consistent with SLD statutory and regulatory language.

Examining SLD Subtype Academic and Psychosocial Functioning

Given a majority of prior SLD research was conducted using A-A discrepancy to define SLD membership, the relationship among cognitive, academic, and psychosocial functioning in this population has not been fully examined. Studies demonstrating the impact of SLD on learning and psychosocial functioning confirm the heterogeneous nature of the population (Forrest, 2004; Fuerst, Fisk, & Rourke, 1989, 1990; Hendriksen et al., 2007; Mattison, Hooper, & Carlson, 2006; Mayes & Calhoun, 2008; Rourke, 2008; Semrud-Clikeman, Walkowiak, Wilkinson, & Christopher, 2010; Spreen, 2011). In addition, various comorbid SLDs have been identified in children with psychiatric disorders such as bipolar disorder, attention-deficit/hyperactivity disorder (ADHD), and autism spectrum disorders (Mayes & Calhoun, 2006). In addition, youth with SLD are overrepresented in the juvenile justice system (Quinn, Rutherford, Leone, Osher, & Poirier, 2005), with more than 57% of incarcerated youth having a diagnosed learning problem (Bullis & Yovanoff, 2005). Clearly there is a need to recognize how different SLDs affect learning and behavior.

For more than three decades, Byron P. Rourke and colleagues examined the neuropsychological, academic, adaptive, and psychosocial functioning of children with SLD (see Rourke, 2000, 2008; Rourke & Fuerst, 1991; Strang & Rourke, 1985). Rourke (1988) argued that the brain’s white matter, required for information integration and processing efficiency, was deficient in children with what he labeled as “nonverbal” SLD. Rourke (1988) indicated the nonverbal SLD subtype had deficits in bilateral tactile-perceptual and psychomotor skills, nonverbal problem solving, visual-spatial-organizational processes, social skills, and mechanical arithmetic, but relative strengths in reading and spelling (Rourke, 1987), and this finding has been replicated in other studies (Forrest, 2004; Hain et al., 2009; Mammarella & Cornoldi, 2013; Rourke, 2000; Semrud-Clikeman et al., 2010). In several studies, Rourke (2000) showed that children with nonverbal SLD had poor nonverbal relative to verbal abilities on the Wechsler Intelligence Scale for Children, Third Edition (WISC-III; Wechsler, 1991), with the assumption being that white matter and right hemisphere dysfunction caused a “nonverbal” problem.

Although the empirical literature supports Rourke’s (2000) contention that white matter problems lead to right hemisphere dysfunction, the notion that this affects “nonverbal” abilities may be an oversimplification of his findings. For instance, the right hemisphere processes global, holistic, spatial, fluid, and implicit language abilities (Bryan & Hale, 2001), not just nonverbal ones. In addition, poor processing speed has also been associated with nonverbal SLD and limited white matter connectivity (Filley, 2005; Forrest, 2004; Hain et al., 2009; Loveland, Fletcher, & Bailey, 1990; Rourke, 2005). Conversely, adequate processing speed has been linked to white matter integrity (Kanai et al., 2012; Turken et al., 2008). Rourke’s (2000) writings reflect the complexity of right hemisphere and white matter dysfunction, but it is the “nonverbal” focus that perhaps limited our understanding of this particular SLD subtype. This is especially plausible given the WISC-III Performance scale tapped both nonverbal and processing speed abilities.

Research Questions

In our prior subtype research we found two SLD subtypes with math deficits and “nonverbal” SLD, but only those with more anterior executive problems such as mental flexibility, working memory, and processing speed were found to have significant psychosocial problems (Hain et al., 2009). Those with more posterior global-holistic-spatial “nonverbal” deficits did not appear to have significant psychosocial problems. To further examine the relationship between cognitive processes and psychosocial functioning for SLD subtypes in the present study, the authors sought to use the C-DM approach to determine SLD subtypes and then compare groups on teacher-reported behavior problems. We predicted that SLD subtypes would have greater teacher-reported clinical and adaptive behavior problems than children who did not meet C-DM criteria for SLD, and sought to determine is some subtypes had more problems than others. Results could aid in understanding the brain–behavior relationships in SLD subtypes, and the impact that cognitive processing deficits have on classroom and adaptive behavior, which could aid in SLD identification and intervention efforts.

Method

Results

Chi-square analysis revealed that when using A-A discrepancy to determine SLD, 51 (36%) students met the criteria for SLD whereas 92 (64%) did not meet the criteria. In contrast to A-A, when using the C-DM method of SLD identification, 100 (70%) were identified as having a SLD whereas 43 (30%) of students were not found to have a SLD. Results of the chi-square analysis indicated a difference between the two SLD identification methods, χ²(1, N = 143) = 18.62, p < .001, φ = .361. A majority of the difference had to do with C-DM criteria identifying children with SLD whom A-A discrepancy excluded (n = 53). For school district criteria to determine SLD eligibility, 83 (58%) met the criteria for SLD whereas 60 (42%) students did not meet the criteria, so these rates are much more similar to the C-DM model. Results of the chi-square analysis indicated a difference between the two methods used to identify SLD, χ²(1, N = 143) = 34.78, p < .001, φ = .493. Although C-DM identified 26 children district SLD criteria did not, there was agreement on 108 children (74 yes, 34 no). Compare this to school district agreement with A-A discrepancy, with only 91 children agreed on (41 yes, 50 no), with 42 children identified with district SLD criteria but not identified using A-A discrepancy, χ²(1, N = 143) = 16.26, p < .001, φ = .337. Overall, results suggest that using the C-DM statistical approach only (which is not advocated; see Hale et al., 2011) will lead to more children identified as SLD, but C-DM results are more consistent with school district SLD criteria than A-A discrepancy criteria. This suggests the C-DM could provide practitioners with actuarial data to validate their clinical team decision making.

When using C-DM and the SED formula to determine SLD subtypes, three distinct C-DM-determined cognitive weakness SLD subtypes were identified. These included Working Memory Index (WMI; n = 20, 16%), Processing Speed Index (PSI; n = 30, 24%), and Executive (WMI+PSI; WMI+PSI+PRI; n = 32; 25%) subtypes. Analyses also yielded a substantial no SLD group (n = 42, 34%). The no SLD group was a clinical sample of children referred for comprehensive evaluation for learning and/or behavior problems. These children likely have strengths and weaknesses cognitively, academically, and behaviorally, but did not exhibit a significant pattern of strengths and weaknesses indicative of SLD according to the C-DM approach. As noted earlier, these cases had another diagnosed disorder, were not found to have SLD, or were not found to have a disability. Other cognitive weakness SLD subtypes identified were too small in number to report in this preliminary study. For instance, the Perceptual Reasoning Index (PRI) subtype, which would be more aligned with “nonverbal” SLD, did not occur frequently enough for group analyses, with only six participants having PRI as their cognitive weakness. There was also four other people who primarily had PRI subtests as the weakness. One child had Gf-fluid reasoning (Picture Concepts and Matrix Reasoning) as a primary weakness, and three children had Gv-analysis/synthesis (Block Design and Picture Concepts) as a primary weakness.

For the SLD subtypes and no SLD group, there were no significant age, F(3, 118) = 0.90, p = .444, or Full Scale IQ, F(3, 118) = 1.31, p = .275, differences for groups, which could affect subsequent results. For descriptive purposes, Figure 2 illustrates the Wechsler Intelligence Scale for Children–Fourth Edition (WISC-IV; Wechsler, 2003) cognitive profiles and the Woodcock–Johnson III Tests of Achievement (WJ-III ACH; Woodcock, McGrew, & Mather, 2001) achievement profiles for each subtype and the no SLD group. Qualitatively, the groups were similar for achievement, with SLD subtype means below the no SLD sample. The PSI SLD subtype scored higher in reading relative to the others; however, math fluency was particularly low in the PSI group. The PSI and WMI groups both had difficulty with reading fluency, but the PSI group had particular difficulty with spelling and writing fluency. The C-DM approach used to create SLD subtypes and no SLD group leads to profile differences, so significance tests are not reported and results should be considered for descriptive purposes only.

OPEN IN VIEWER

Figure 2.

Wechsler Intelligence Scale for Children–Fourth Edition cognitive and Woodcock–Johnson III Tests of Achievement results for no SLD group and SLD subtypes.

With the three subtypes (WMI, PSI, Executive) and no SLD group established with the C-DM model, the research question regarding psychosocial differences among groups was addressed using MANOVA with BASC-2-TRS Internalizing Problems, Externalizing Problems, School Problems, Behavioral Symptoms, Adaptive Skills as dependent variables (DVs). Table 1 shows the descriptive data for the five scales broken down by group. The highest mean scores were found for the no SLD group and the PSI SLD group. This suggests that the C-DM approach could differentiate children referred for comprehensive evaluations whose processing impairments lead to SLD and psychopathology (e.g., PSI SLD), those whose executive impairments likely to lead to SLD but not psychopathology (e.g., Working Memory SLD, Executive SLD), and those with other types of executive impairments and psychopathology who do not meet SLD criteria (e.g., no SLD). Perhaps this latter group has children with disorders such as ADHD, anxiety, or depression without the academic deficits found in SLD.

OPEN IN VIEWER

Table 1.

Behavior Assessment System for Children–Second Edition, Teacher Report Scale Composite Scale Results for No SLD Group and SLD Subtypes.

Scale	No SLD	WMI SLD	PSI SLD	Executive SLD	F	p
Externalizing Problems
M	57.25	52.57	61.45	53.00	1.82	.154
SD	12.99	9.85	11.79	14.68
Internalizing Problems
M	57.45	53.64	64.75 a	51.56	3.20	.030
SD	10.41	12.57	16.35	6.80
School Problems
M	59.25	61.93	64.15	66.33	1.71	.175
SD	9.16	8.12	9.64	7.19
Behavioral Symptoms
M	61.40	58.86	69.80	58.11	2.38	.078
SD	13.48	12.86	16.16	12.46
Adaptive Skills
M	43.55	41.71	35.90a,b	42.22	5.21	.003
SD	7.98	7.62	4.28	4.55

Note. PSI = Processing Speed Index; SLD = specific learning disability; WMI = Working Memory Index.

a

Different from executive SLD group. ^bDifferent from WMI SLD group;.

To examine group differences, two multivariate assumptions had to be met. First, the nonsignificant Box M’s test of equality of covariance matrices (Box M = 66.74), F(45, 3884) = 1.21, p = .159, indicated the null hypothesis that the observed DV covariance matrices are equal across groups was not rejected, so this did not violate multivariate assumptions. Levene’s test of the equality of error variances, which tests the null hypothesis that the DV error variances are equal across groups, was not rejected for the Internalizing Problems, F(3, 59) = 0.83, p = .482, Externalizing Problems, F(3, 59) = 2.34, p = .083, School Problems, F(3, 59) = 0.40, p = .754, or Behavioral Symptoms scales, F(3, 59) = 1.07, p = .367, but it was for Adaptive Skills, F(3, 59) = 3.66, p = .017, which could limit conclusions for this scale.

Using Hotelling’s trace as the multivariate test statistic, there was an effect for group, F(15, 161) = 1.82, p = .036, indicating differences among BASC-2 scales for SLD subtypes and/or the no SLD group. There was an associated ή² of .145, indicating that the measures accounted for approximately 15% of the between groups variance. Power (1 – β), which examines the probability of a Type II error, was sufficient (Observed Power = .923). Not surprisingly, the intercept was also significant, F(5, 55) = 2995.29, p < .001, ή² = .996, 1 – β = 1.00, which would be expected given the large amount of between groups variance not accounted for by the behavior ratings.

The univariate tests for between subject effects and post hoc analyses for DVs are reported in Table 1. Games–Howell tests were used due to unequal variances and are reported in the table, but Sidak–Bonferroni tests were also examined. As can be seen, there were group differences for the Internalizing Problems and Adaptive Skills scales, and the Behavior Symptoms scale approached significance. Post hoc analyses revealed the PSI subtype to have the most internalizing and adaptive behavior problems relative to other subtypes, but only in the Adaptive Skills scale did these differences did not extend to the no SLD group, which had comparable problem behaviors in all other areas.

Exploratory step-down analyses of subscales were then undertaken, with univariate ANOVAs computed for each subscale for which a significant composite differences was found. Significant results were found for the Depression, F(3, 59) = 3.23, p = .029, Withdrawal, F(3, 59) = 3.42, p = .023, Adaptability, F(3, 59) = 4.54, p = .006, Social Skills, F(3, 59) = 3.84, p = .014, and Functional Communication, F(3, 59) = 3.90, p = .013, subscales. The Conduct Problems subscale approached significance, F(3, 59) = 2.71, p = .053. Results for these showed similar patterns, with the PSI SLD subtype showing the greatest number of internalizing problems and the lowest adaptive behavior. These results should be considered preliminary however, given the likelihood of Type I error due to multiple comparisons. Instead, better confidence can be placed in the broadband scale differences reported earlier.

Discussion

Children with SLD represent a diverse population with considerable learning and behavioral needs. Notable for their heterogeneity, both in terms of characteristics and outcomes, some children with SLD experience significant social, emotional, and/or behavioral concerns in addition to achievement deficits, whereas others seem to possess adequate psychosocial functioning (Rourke, 2000). Understanding SLD heterogeneity is hampered by controversy over SLD statutory (definition) and regulatory (identification method) requirements. Most early SLD studies used A-A discrepancy to identify children, but neither A-A discrepancy or RTI successfully defines SLD, so they should not be used for SLD research purposes (Hale et al., 2010). Creating SLD subtypes using a verbal/nonverbal dichotomy (e.g., Verbal IQ/Performance IQ) is ineffective because the brain has bilateral verbal and nonverbal functions, and creating SLD subtypes based on a reading–spelling/mathematics dichotomy is ineffective because of the frequent myriad causes and frequent comorbidity of reading, spelling, and mathematics SLD (Berninger, 2006; Bryan & Hale, 2001; Fiorello et al., 2012; Geary, 2010; Goldberg, 2001; Hale et al., 2010; Koziol, Budding, & Hale, 2013; Riccio, Sullivan, & Cohen, 2010; Semrud-Clikeman, 2005).

To comply with IDEA (2004) SLD statutory and regulatory requirements, and to develop more homogenous SLD subtypes for both clinical and research purposes, a processing strengths and weaknesses model is preferred over A-A discrepancy or RTI identification approaches (Flanagan et al., 2010; Hale et al., 2010; Wright et al., 2013). The present study found the C-DM approach to be more consistent with district team decisions regarding SLD eligibility than a strict A-A discrepancy approach, which missed many children with SLD. However, it is important to note that the C-DM approach used in this study was strictly psychometric and did not follow the eight-step clinical process suggested by Hale et al. (2010). As Hale et al. caution, hypotheses regarding individual processing strengths and weaknesses, and their relationship to achievement deficits, need verification using a CHT evaluation approach. However, this was not feasible for the archival data used in the present study.

Unlike A-A discrepancy or RTI, the C-DM approach allows for examination of specific patterns of academic, psychosocial, and adaptive function among SLD subtypes, revealing important differences that could be used to guide intervention. For instance, all three SLD subtypes demonstrated low mathematics and written language composites relative to those without SLD, and the WMI SLD and Executive SLD subtypes demonstrated low Broad Reading. The PSI and WMI groups both had difficulty with reading fluency, but the PSI group had particular difficulty with spelling, math, and writing fluency as well. As part of a larger problem-solving model, this information can be useful in problem analysis phase for brainstorming targeted interventions. For instance, the C-DM approach can help determine if a child with a reading comprehension problem might need a rapid naming intervention to address processing speed deficits (Norton & Wolf, 2012) or a mnemonics intervention to address working memory deficits (Dehn, 2011) based on C-DM data (e.g., Fiorello, Hale, & Synder, 2006). The achievement data just tell the practitioner both subtypes are low on reading comprehension. In an RTI model, both may receive more intensive reading instruction (e.g., Vaughn, Denton, & Fletcher, 2010); however, psychological processes that tell you what type of intervention to attempt so they get differentiated instruction. With targeted interventions, and subsequent data collection, evaluation, and recycling of the intervention, educators can get truly differentiate instruction so that is sensitive and specific to the individual child’s need (e.g., Fiorello et al., 2009; Fischer, 2009; Hale et al., 2006; Tomlinson, 2012).

The C-DM approach also clarified the relationship between psychological processes and psychosocial functioning in this study. According to teacher BASC-2-TRS report, participants in the PSI SLD group demonstrated higher levels of internalizing symptoms and poorer adaptive functioning than other subtypes and the no SLD group. Clearly, some executive problems lead to psychosocial problems with and without SLD, and processing speed may be the cause of SLD-psychopathology comorbidity. This finding corroborates processing speed to be an important factor in social adjustment and adaptive behavior (Benner, Allor, & Mooney, 2008) and is consistent with the cognitive and psychosocial characteristics of SLD subtypes with processing speed deficits (Filley, 2005; Forrest, 2004). The PSI group in this study has many of the learning and psychosocial characteristics found in Rourke’s (2008) nonverbal SLD. Rourke’s studies used the WISC-III Performance IQ Scale (Wechsler, 1991) to identify the nonverbal SLD subtype, but this scale was heavily dependent on processing speed. As a result, the white matter problem with “nonverbal” SLD may actually have been due to processing speed. Recent arguments by Semrud-Clikeman et al. (2010), Spreen (2011), and Galway and Metsala (2011) highlight the complexity of white matter dysfunction, and how it negatively influences neuropsychological processes and social competence, beyond Rourke’s traditional nonverbal characterization.

To further the argument, the current PSI SLD results should be considered within the context of another disorder affecting white matter integrity and processing speed—traumatic brain injury (TBI). With the shearing/tearing of white matter following acceleration/deceleration closed head TBI, processing and transfer of information is impaired, which in turn reduces brain efficiency and processing speed (Hendriksen et al., 2007; Mathias et al., 2004; Semrud-Clikeman & Bledsoe, 2011), and leads to psychosocial and adaptive behavior problems (Kinnunen et al., 2011; Kraus et al., 2007; Spreen, 2011). Although children with TBI are 18 times more likely to have poor academic performance, with almost 50% experiencing school failure and/or special education placement (Ewing-Cobbs et al., 2004), psychosocial functioning may be the most prominent deficit and greatest need for intervention (DiScala, Osberg, Gans, Chin, & Grand, 1991; Stancin et al., 2002). As white matter damage decreases functional network connectivity in TBI, resulting in poor information integration and diffuse psychological impairment, the findings reported here suggest a parallel among TBI, the PSI subtype, and Rourke’s (2000) contention regarding integration difficulties in nonverbal SLD.

Although Rourke’s (1989, 2000, 2008) position may be accurate regarding white matter dysfunction and psychosocial disturbance, the data presented here suggest his results may have been due to processing speed deficits instead of “nonverbal” deficits. TBI can impair executive functions (e.g., initiating, planning, organization, fluency, and goal setting), memory (e.g., encoding, storage, and retrieval), attention, working memory, and processing speed (Dennis, 1991; Riccio & Wolfe, 2003; Semrud-Clikeman, 2001). Similar to nonverbal SLD, TBI causes difficulty with visual-spatial processes, visual-motor integration, memory, inferential thinking, new learning, discordant-divergent thought problems, and implicit language (Hale & Fiorello, 2004; Koziol & Budding, 2009; Mathias et al., 2004; Semrud-Clikeman, Kutz, & Strassner, 2005), but 10 years postinjury, it is the adaptive functioning and processing speed impairments that remain for children who experience TBI (Anderson, Catroppa, Godfrey, & Rosenfeld, 2012).

As a result, the PSI subtype findings presented are congruent with previous studies evidencing a relationship among TBI, white matter dysfunction, and impairments in social-affective communication (Chapman et al., 2004; Dennis & Barnes, 2001). Poor processing speed may not only be related to the psychosocial and adaptive deficits in TBI and SLD, but it could also explain the social and adaptive problems in Asperger syndrome (AS; McAlonan et al., 2009; Stothers & Oram-Cardy, 2012; Woodbury-Smith & Volkmar, 2009), which has also been described as a form of low functioning “nonverbal” SLD (Klin, Volkmar, Sparrow, Cicchetti, & Rourke, 1995). This distinct pattern of poor processing speed in AS (Noterdaeme, Wriedt, & Höhne, 2010; Spek, Scholte, & van Berckelaer-Onnes, 2008; Williams, Goldstein, Kojkowski, & Minshew, 2008) suggests a common dysfunction among children with TBI, AS, and the PSI SLD subtype found here.

These findings suggest it may be necessary to reconceptualize nonverbal SLD as a processing speed problem, not as a general right hemisphere problem affecting “nonverbal” functioning. White matter dysfunction results in a similar pattern of dysfunction regardless of origin (e.g., AS, TBI, nonverbal SLD) and includes processing speed impairment (Hendriksen et al., 2007; Mathias et al., 2004b), psychosocial (e.g., internalizing depression and withdrawal), and adaptive behavior problems (e.g., adaptability, social skills, functional communication; Kinnunen et al., 2011; Kraus et al., 2007; Spreen, 2011); however, separation of nonverbal from processing speed deficits was not accomplished during this study as was the case in Hain et al. (2009), where it was found that the posterior “nonverbal” SLD subtype had fewer problems than the subtype that had processing speed impairments. By recognizing the similarities among SLD subtypes, professionals can subsequently develop targeted interventions to address common psychosocial and adaptive dysfunction found within subtypes. For instance, children with PSI SLD could benefit from social skills role play interventions designed to increase processing verbal and nonverbal social stimuli in timely fashion during interpersonal dialog and exchange. Targeting processing speed during intervention may facilitate executive management of social exchange and interpersonal relationships, which in turn could lead to better psychosocial and adaptive functioning in affected children (see Hale et al., 2012).

Limitations and Future Research

The study sample size of the study was small, resulting in fewer children in SLD subtypes, and other SLD subtypes not evaluated because of even smaller numbers. The subtypes were compared to other clinical referral children, not neurotypical children, which could have attenuated findings and subtype differences. In addition, the sample was relatively homogenous for demographic characteristics, thereby limiting study generalizability. There was a narrow age range, so the developmental progression of PSI deficits and psychosocial and adaptive functioning could not be explored. Given that archival data were collected and neuropsychological data were limited, neuropsychological measures could not be used to confirm that processing weaknesses were related to achievement deficits, as is suggested in Hale and Fiorello’s (2004) CHT model. It will be valuable to gather a larger sample of participants to make results more generalizable and to differentiate among additional subtypes. Finally, further examination of white matter dysfunction and psychosocial impairment in other disorders should be undertaken to examine externalizing symptoms, internalizing symptoms, and adaptive skills deficits in processing speed SLD and other clinical populations. Future work with functional (e.g., fMRI) or structural (e.g., DTI) neuroimaging may reveal the nature of white matter dysfunction in SLD and allow for comparison with other disorders like TBI and AS. Finally, future research should explore the treatment validity of SLD subtype findings to determine if individualized interventions targeted to address processing characteristics, academic achievement deficits, and psychosocial and adaptive difficulties leads to better functional outcomes for children with subtypes of SLD.

Children may have learning and/or behavioral problems for numerous reasons, only one of which is SLD. When a child does not respond to our best attempts at intervention (e.g., RTI), a comprehensive evaluation in all areas of suspected disability is not only best practice, it is also required by law. Even if the comprehensive evaluation shows the child has a SLD or some other disorder, understanding both the processing and environmental determinants of the SLD can lead to targeted interventions specific to individual needs. Even when achievement (e.g., reading deficit) or behavioral (e.g., attention problem) outcome variables are the same, they may have very different causes, which necessitates different intervention approaches. Recognizing how different processing strengths and weaknesses can lead to SLD and/or other disorders is just the beginning for a child who has a disability; the real utility in comprehensive evaluation is how it can be used to guide interventions that are effective and ecologically valid (Decker, Hale, & Flanagan, 2013). This is the next essential step for processing strengths and weaknesses research, the ability to link assessment to intervention for children with SLD and other disabilities so that all children receive the differentiated instruction they need for academic and psychosocial success.