Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Oct 13.
Published in final edited form as: Sci Stud Read. 2020 Oct 4;25(5):397–416. doi: 10.1080/10888438.2020.1821035

In Search of Cognitive Promotive and Protective Factors for Word Reading

Rebecca F Slomowitz a, Angela J Narayan a, Bruce F Pennington a, Richard K Olson b, John C DeFries c, Erik G Willcutt b, Lauren M McGrath a,*
PMCID: PMC8514166  NIHMSID: NIHMS1656134  PMID: 34650325

Abstract

This study examined whether strong cognitive skills (i.e. vocabulary, rapid naming, verbal working memory [VWM], and processing speed [PS]) contributed to resilience in single-word reading skills in children at risk for reading difficulties because of low phonological awareness scores (PA). Promotive factors were identified by main effects and protective factors through PA x cognition interactions. This study included 1,807 children ages 8-16. As predicted, all cognitive skills were significantly related to reading, consistent with promotive effects. A significant, but small effect PA x vocabulary interaction (R2 change=.002, p=.00038) was detected but its form was not consistent with a classic protective effect. Rather, the PA x vocabulary interaction was consistent with a “skill-enhancement” pattern, such that children with strong PA and vocabulary skills had better than expected reading. This study provides a framework for reading resilience research and directs attention to promotive mechanisms underlying reading success.

Introduction

While the dyslexia field has made notable progress in identifying risk factors, especially cognitive risks (Pennington, 2006), much less work has focused on resilience mechanisms. The small body of existing research on resilience in children with or at risk for dyslexia has mainly investigated resilience due to socio-emotional and cognitive factors (Eklund, Torppa, & Lyytinen, 2013; Kiuru et al., 2013; van Viersen, de Bree, Kroesbergen, Slot, & de Jong, 2015; van Viersen, de Bree, & de Jong, 2019; for reviews see Haft, Myers, & Hoeft, 2016; Yu, Zuk, & Gaab, 2018) but consensus findings have yet to coalesce. To contribute to this emerging research area, the current study used the established developmental psychopathology (DP) perspective for studying resilience to illustrate its application to reading research. This study specifically examined whether strong cognitive skills (vocabulary, rapid automatized naming, verbal working memory, and processing speed) could contribute to resilience in single-word reading skills in children at risk for reading difficulties because of low phonological awareness scores. While reading is a complex cognitive skill, we specifically focused on single-word reading skills, which is the core skill weakness in children with dyslexia (American Psychiatric Association, 2013; Lyon, Shaywitz, & Shaywitz, 2003).

Developmental psychopathology perspective for identifying resilience in dyslexia

The DP perspective, which emerged from a broader literature that studies risk and resilience mechanisms associated with the development of psychopathology, has made pivotal contributions to the understanding of resilience (Luthar, Cicchetti, & Becker, 2000; Masten, 2001; Masten & Garmezy, 1985; Masten & Reed, 2009; Wright, Masten, & Narayan, 2013). According to the DP literature, the term “resilience” refers to better than expected outcomes despite the presence of risk (Luthar et al., 2000; Masten, 2001), making the context of risk a prerequisite for resilience (Luthar et al., 2000; Masten, 2001). Resilience typically arises due to the presence of promotive and protective factors.

While previous research has considered resilience mechanisms in the reading and dyslexia fields (e.g.,Donahue & Pearl, 2003; Eklund et al., 2013; Haft et al., 2016; Kiuru et al., 2013; Yu et al., 2018), the terminology and analyses are not always consistently applied. We believe that the DP perspective is valuable for reading research, as it offers well-developed and comprehensive suggestions for studying resilience in the context of reading (Cicchetti & Toth, 2009; Masten, 2001). Thus, the present study builds on previous work by providing further guidance on the applications of the DP perspective.

Promotive Factors

Promotive factors are those that are associated with positive outcomes regardless of the presence or degree of risk (Masten & Reed, 2009; Wright et al., 2013). Importantly, promotive factors improve outcomes equally for individuals at both high and low risk and are identified by main effects in linear models (Narayan, 2015; Wright et al., 2013). While promotive factors can be identified in the context of high and low risk, they are only associated with resilience when they occur in a high-risk context. When a promotive factor is identified in a low-risk context, it is associated with positive outcomes butis not considered a resilience factor because there is no risk to overcome (Wright et al., 2013). A classic example of a promotive factor is the way in which supportive parenting is helpful for all children, regardless of risk status (Wright et al., 2013).

Protective factors

Protective factors attenuate risk, thus leading to better than expected outcomes for the high-risk group compared to the low-risk group, although the low-risk group may still outperform the high-risk group in absolute terms (Masten & Reed, 2009). Protective factors can be distinguished from promotive factors because they provide stronger buffering qualities for individuals at high risk and are identified by interactions (Risk x Protective Factor) in a linear model (Masten, 2001; Wright et al., 2013). Figure 1 illustrates a hypothetical protective factor interaction adapted from Luthar et al. (2000). A classic example of a protective factor explained by Wright et al. (2013) is an airbag. While an airbag can lead to a better outcome for someone in a car accident, it does not change the outcome for someone who is not in an accident (Wright et al., 2013). Figure 1 illustrates a hypothetical cognitive protective interaction where children at risk for reading difficulties benefit more from strong cognitive skills than children at low risk.

Figure 1.

Figure 1.

Open in a new tab

Hypothetical Cognitive Protective Effect

Promotive vs. protective factors

It is important to draw distinctions between promotive and protective factors because they operate differentially in the context of risk. One implication of their different effects across risk status is that promotive factors do not diminish the gap between risk groups, while protective factors contribute to gap-closing effects because high-risk groups benefit significantly more than low-risk groups (Masten & Reed, 2009; Wright et al., 2013). Understanding these mechanisms provides important clarity for the reading resilience field, as conflating these terms has led to analytical and definitional problems.

Promotive and protective factors: Application to reading resilience research

When applying the DP perspective to reading research, two problems become apparent. The first problem is that there has been intense (and justifiable) focus on risk within the literature. Thus, when linear associations between cognitive skills and reading are identified, these cognitive skills are described as risk factors, but the high end of the linear association (i.e. promotive factors) is often neglected. The second problem is that when researchers do attempt to examine strong skills that might contribute to resilience, they often do not distinguish between promotive and protective factors.

Expanding on this first issue, if a linear association is found between a cognitive score and a reading score, then by definition lower cognitive scores will be associated with lower reading scores and higher cognitive scores will be associated with higher reading scores. While this description of a main effect may seem straightforward, we note that researchers often treat the high end and the low end of linear associations differently in terms of their dimensionality. For example, risk factors are often understood to be dimensional where more severe weaknesses are associated with more severe outcomes. Promotive factors can have this same dimensionality where stronger skills are associated with better outcomes. However, the dimensionality of the promotive factor is sometimes oversimplified by referring to these stronger skills as the absence of risk. This practice might inadvertently diminish the role of promotive factors in resilience processes, as it is the presence of stronger skills, not just an absence of weaker skills, that is associated with better outcomes (Masten, 2001; Masten & Reed, 2009) We propose drawing attention to the high end of these linear associations by explicitly labeling promotive effects. In many cases, promotive factors are hiding in plain sight at the opposite ends of the same linear distributions as risk factors (Masten, 2001). Thus, promotive factors are not a new empirical finding, but they are a new terminological distinction for the reading field (Masten, 2001).

Currently, the reading literature commonly uses the phrase “risk and protective factors” to acknowledge both vulnerability and resilience, but this phrase should be amended to “risk and promotive factors” when referring specifically to skills that share linear associations with reading. This revision more accurately aligns with DP specifications and reinforces that risk and promotive (not protective) factors should be considered opposites.

The second problem that arises in the reading literature is that researchers infrequently distinguish between promotive and protective factors. Rather, the term “protective factors” has become a catch-all term for both promotive and protective effects. This is more than just a semantic problem, because promotive and protective factors are identified by different analytic methodology (i.e. main effects versus interaction terms) and have different effects on the outcome gap between high and low risk groups (i.e. gap-maintaining versus gap-closing effects). Therefore, to identify a protective effect, the analysis requires both: 1.) a statistical test of the interaction effect and 2.) the full 2 × 2 matrix of presence/absence of a protective factor and high/low risk group to examine gap-closing effects (Masten, 2001). If promotive and protective factors are not properly differentiated, then research designs may be inadequate for distinguishing the two mechanisms (Luthar et al., 2000). These issues have guided the design of the current study.

While previous DP research has commonly used categorical indicators of risk/protective factors and outcomes, within the reading and cognitive literatures, the line between subclinical and clinical level weaknesses is an arbitrary cut-point on a continuous distribution (Peters & Ansari, 2019; Plomin, Haworth, & Davis, 2009). Thus, this study took a dimensional approach by considering continuous scores of single-word reading and cognitive skills across the full range from very weak to very strong skills. To test for protective effects, we tested the dimensional equivalent of the full 2 × 2 matrix using continuous measures of phonological awareness (risk status) and other cognitive skills (promotive/ protective factors) with single-word reading as the outcome.

Defining risk for reading difficulties

Protective factors are defined in relation to their buffering of risk factors. In this study, risk was determined by phonological awareness (PA) skills. PA is one of the most extensively studied and highly predictive cognitive-linguistic risk factors associated with both individual differences in singleword reading skills and dyslexia, with treatment studies supporting its causal links to reading difficulties (Hulme & Snowling, 2013; Snowling & Melby-Lervåg, 2016; Swanson, Trainin, Necoechea, & Hammill, 2003; Vellutino, Fletcher, Snowling, & Scanlon, 2004). Current research shows that approximately 50% of individuals with dyslexia have a weakness in PA (Pennington et al., 2012; Ring & Black, 2018). Thus, while not perfectly predictive, PA is strongly correlated with reading skills. While there are other important cognitive risk factors for dyslexia, we chose to focus on one core risk variable to demonstrate the DP analytic model in a straightforward manner. We turn now to a consideration of which cognitive skills may serve as promotive or protective factors in the context of risk for reading difficulties defined by PA.

What cognitive skills could be promotive or protective?

The current study focuses on whether cognitive skills serve as promotive and protective factors because of the prominence of the cognitive level of analysis in the dyslexia research literature. The leading candidates are: vocabulary, rapid automatized naming (RAN), verbal working memory (VWM) and processing speed (PS). Each of these cognitive skills has shown a linear relationship with reading in previous work (McGrath et al., 2011; Willcutt, Pennington, Olson, Chhabildas, & Hulslander, 2005), which would imply that they are at least promotive factors since high scores are associated with stronger reading skills. What has not been commonly investigated are the interactive effects between PA and these cognitive variables.

Language skills

The most extensively studied resilience mechanism in children at risk for or diagnosed with dyslexia has been early language skills (Muter & Snowling, 2009; Nation & Snowling, 1998; Snowling, Gallagher, & Frith, 2003; van Viersen et al., 2019). For example, in high-risk longitudinal family designs, children at high risk for dyslexia with better early language skills were less likely to be diagnosed with dyslexia compared to children at low risk (Snowling et al., 2003). Similarly, strong language skills were identified at higher rates in gifted children whose dyslexia remitted versus persisted (van Viersen et al., 2019). While this previous work indicates that language skills likely function as promotive factors through main effect associations, it is not currently clear whether language skills also serve as protective factors. This is because no study has examined whether language skills improve the scores of the high-risk group over and above those of the low-risk group.

To explain how strong language skills may contribute to resilience, Snowling et al. (2003) theorized that children with impaired phonological development may rely more on their semantic pathways for reading development. This resilience mechanism has been termed the “semantic bootstrapping” hypothesis (Muter & Snowling, 2009; Snowling et al., 2003). In line with this theory, we hypothesize that vocabulary will be both a promotive and protective factor for single-word reading.

Rapid automatized naming

Rapid automatized naming (RAN) tasks involve speeded naming of letters, numbers, colors, or objects, and has been frequently identified as a predictor of later literacy difficulties (Norton & Wolf, 2012; Puolakanaho et al., 2007; Thompson et al., 2015; Torppa, Eklund, van Bergen, & Lyytinen, 2015). While RAN has been used to identify children at risk for reading difficulties, van Bergen, van der Leij, and de Jong (2014) suggested that stronger RAN performance may serve as a protective factor for children at risk for dyslexia. Empirical results from children at high risk for dyslexia because of a language disorder (Bishop, McDonald, Bird, & Hayiou-thomas, 2009; McBride- Chang et al., 2011; Moll, Loff, & Snowling, 2013; Vandewalle, Boets, Ghesquière, & Zink, 2010) or speech sound disorder (Peterson, Pennington, Shriberg, & Boada, 2009; Raitano, Pennington, Tunick, Boada, & Shriberg, 2004) also support the notion of RAN as a factor that contributes to resilience despite risk for literacy difficulties. While these results suggest that RAN might be promotive or protective for children at risk for dyslexia, no studies have attempted to distinguish promotive from protective mechanisms. Our hypothesis is that RAN will be a promotive factor and may also be a protective factor for single-word reading.

Verbal working memory

Verbal working memory (VWM) is the ability to hold in mind and manipulate verbal input (Peng et al., 2017). VWM has shown consistent correlations with reading skills (Peng et al., 2018; Reiter, Tucha, & Lange, 2005; Rose & Rouhani, 2012; Willcutt et al., 2005), indicating that VWM may serve as a promotive factor. Strong VWM skills were also more often identified in gifted children with remitted versus persistent dyslexia (van Viersen et al., 2019), supporting the idea that strong VWM may be associated with resilience. While the connection between VWM and single-word reading skills remains under-specified, previous research has suggested that VWM helps children to hold and bind auditory information with visual symbols, such as letter sounds with letter symbols (Wang, Allen, Lee, & Hsieh, 2015). Thus, we could hypothesize that if the quality of incoming phonological information is impacted by weaknesses in PA, children with stronger VWM skills may have larger or more efficient capacities to hold and bind visual-verbal information which would be beneficial for learning letter-sound correspondences and decoding words.

In support of VWM conferring protective effects, Rose and Rouhani (2012) found that in adolescents with dyslexia, VWM skills mattered more for those with weaker than stronger vocabularies in predicting reading fluency outcomes – a VWM x vocabulary interaction (Rose & Rouhani, 2012). In other words, an individual could partially compensate for the impact of weak vocabulary on reading fluency if they had a strong VWM. While it cannot be determined whether VWM was protective (due to the lack of a full 2 × 2 matrix), the discovery of a VWM x vocabulary interaction for reading fluency suggests possible cognitive protective effects of VWM for reading outcomes (albeit using different measures than those in the present study). It will be important to test further how VWM may be associated with resilience in the context of PA risk. Taken together, previous studies suggest that VWM will serve as a promotive factor for single-word reading, but the empirical and theoretical rationale to support VWM as a protective factor are not as strong as those for vocabulary and RAN. Thus, we hypothesize that VWM will serve as promotive factor, and plan to conduct exploratory analyses to examine whether VWM is also a protective factor.

Processing speed

Processing Speed (PS) is conceptualized as the efficiency with which a person can complete cognitive tasks (McGrew & Evans, 2004). In studies examining the cognitive correlates of reading, PS has been associated with single-word reading (McGrath et al., 2011; Peterson et al., 2017). Since the association between PS and single-word reading is linear, there is good evidence that PS will be promotive, but no previous research has investigated whether PS is also a protective factor.

To date, theory regarding the relationship between PS and untimed single word reading is the least developed of all the cognitive domains covered in this study. We can speculate that faster processing speed is a domain-general skill that impacts the efficiency of many other cognitive skills. For example, there is evidence that PS is a developmental precursor for both RAN and VWM and that age-related changes in PS precede and promote age-related growth in RAN and VWM (Fry & Hale, 1996; Kail, 2007; Kail & Hall, 1994). Given that reading has multifaceted influences (Pennington, 2006), a child with weaker PA but stronger PS may derive benefit from faster processing in other skills involved in reading, such as VWM and RAN. Thus, we hypothesize that PS will serve as a promotive factor for single-word reading, and we plan to conduct exploratory analyses to examine whether PS is also a protective factor.

Developmental patterns

Previous work has demonstrated that the strength of associations between specific cognitive skills and reading outcomes can change across the age range (Snowling & Melby-Lervåg, 2016). Therefore, in exploratory analyses, we tested whether the cognitive skills (vocabulary, RAN, VWM, PS) had stable main effects and interactions with PA across the age range in our cross-sectional sample (8–16 years).

Current study and hypotheses

Using the prevailing DP perspective, the present study tested whether stronger skills in specific cognitive domains (Vocabulary, RAN, VWM, PS) served as promotive and/or protective factors in children at risk of single-word reading difficulties because of weaker PA skills. Based on prior research, all skills were hypothesized to serve as promotive factors. We hypothesized that vocabulary and RAN would function as protective factors, but we made no a priori predictions about whether VWM and PS would be protective because there is not enough empirical work on these constructs to make strong predictions. The overarching goal of this study was to illustrate the application of conceptual and analytic DP perspectives for the examination of resilience in the reading literature.

Methods

Participants

The participants for this study (N = 1,807) were 8–16 years old. The participants were recruited as part of the Colorado Learning Disabilities Research Center (CLDRC), which is a long-standing twin study (Willcutt et al., 2019). These participants were recruited from 1990–2017. Detailed explanations of this project’s recruitment and procedures have been published previously (DeFries, 1997; Willcutt et al., 2019, 2005). In brief, twins living within 150 miles of metropolitan Denver were identified through 22 local school districts or through the state’s twin registry. For both recruitment sources (schools or twin registry), all twins were invited to participate with subsequent screening for eligibility. Parents who indicated interest provided informed consent and completed a phone interview to screen for history of reading or attention difficulties. Parents also completed a rating scale measure of DSM-IV symptoms of ADHD (Barkley & Murphy, 1998), and permission was requested to send a parallel questionnaire to each twin’s primary classroom teacher. Questions related to history of reading difficulties included whether the child had had difficulty learning to read, had current reading difficulties, or had been diagnosed with a learning disability in reading. Parents were also asked to provide scores on state based standardized tests, when available, and performance below the proficient range on literacy tests was considered indicative of reading difficulties. In addition to the ADHD rating scales, questions related to history of attention difficulties included whether the child had difficulties paying attention, had a history of hyperactivity, had ever been diagnosed with ADHD/ADD, or had been prescribed stimulant medication. If either member of a twin pair had a history of reading or attention difficulties, the pair was invited to participate in the study. A comparison group of twins was also recruited in which neither twin met the screening criteria for learning or attention difficulties. The comparison group was matched to the clinical groups on age, zygosity, and sex as identified by the parent. Additional inclusion/exclusion criteria for this study were that children resided in primarily English speaking homes, had a Verbal IQ or Nonverbal IQ above 85 and a Full Scale IQ above 70 on the WISCR or WISC-III (Wechsler Intelligence Scale for Children, Revised or 3rd Edition; Wechsler, 1991, 1974), had no prior history of neurological conditions, genetic syndromes, or brain injury, had no uncorrected visual impairments, and were not Deaf or hard-of-hearing. Table 1 gives demographic characteristics for the sample.

Table 1.

Participant Demographics (N=1,807)

Participant Demographics Mean SD Range
Age 11.0 2.2 8.0-15.9
Maternal Yrs Ed 15.1 2.5 7-20
Paternal Yrs Ed 15.1 2.7 6-20
Full Scale IQ Standard Score 106.3 12.7 71-148
Verbal IQ Standard Score 107.0 14.0 64-147
Performance IQ Standard Score 104.4 12.6 61-147
PIAT Reading Recognition Standard Score 102.1 12.3 65-135
ADHD Rating Scale, # of Inattention Symptoms 2.5 3.1 0-9
Sex As Identified by Parent 1 Percentage
Female 48.2%
Male 51.8%
Race 2 Wave 1 (1990-2006) 2 Wave 2 (2006-current) 2
Asian 0.3% 0.0%
Black 1.5% 0.6%
Hispanic or Latino 2.7% --
Multiple groups identified3 15.0% 10.3%
Native American/American Indian/Alaska Native/Indigenous 0.2% 0.8%
White 80.0% 87.5%
Prefer to self-describe 0.2% 0.8%
Ethnicity
Hispanic or Latino -- 3.1%
Multiple ethnicities identified3 -- 10.3%
Not Hispanic or Latino -- 86.6%
Open in a new tab
1

We assessed sex as binary and did not include intersex as an option. We did not assess self-reported gender.

2

Our earliest waves of data collection included a single variable that combined race and ethnicity, consistent with federal guidance at the time. Since 2006, we have been collecting race and ethnicity information separately.

3

In earlier phases of data collection, parents self-reported race and ethnicity for themselves but not their children. Here we indicate if parents endorsed multiple identifications. We want to note the limitations of this approach, however, as it does not capture the identification that families and children would choose for the child. We have made revisions to the race and ethnicity data collection to align with current best practices for inclusiveness in research studies (e.g., Wadsworth et al., 2016).

The current sample has minimal overlap (16.6%, N=299 individuals overlapping) with a previous study utilizing similar cognitive constructs to predict single-word reading (McGrath et al., 2011). That study, however, did not investigate cognitive interactions.

Procedures

The study protocol was approved by the Institutional Review Boards at the University of Colorado, Boulder and the University of Denver and conforms to the recognized standards of US Federal Policy for the Protection of Human Subjects. After obtaining informed consent from parents or legal guardians and assent from their children at both institutions, one 6-hour testing session was completed at the University of Colorado, and one 6-hour testing session was completed at the University of Denver. Each day of testing included a lunch break as well as smaller breaks to minimize fatigue. Testing occurred first at the University of Colorado Boulder and about two months later at the University of Denver. The testing sessions at Boulder included the reading, PA, and RAN measures and cognitive measures, such as IQ. The testing sessions at Denver included most of the neuropsychological measures, such as WM and PS. Testing was conducted in research laboratories by trained examiners who were research assistants with bachelor’s degrees or doctoral level graduate students. Examiners were trained to be sensitive to fatigue and to offer frequent small breaks and behavioral support to maintain motivation. Examiners were blind to participants’ diagnostic status. Participants taking psychostimulant medication were asked to withhold medication for 24 hours prior to each testing session.

Measures

Table 2 provides a brief summary of each construct and its corresponding measures. Supplementary Tables 1 and 2 include descriptives and correlations for constructs, respectively. Because participants were recruited from a long-standing study design, outdated versions of measures have been retained to maintain consistency across time. Standard scores were used in a few cases to establish rank ordering of individuals when raw scores were not appropriate, but norms were not used to establish clinical thresholds.

Table 2.

Brief Description of Measures, Reliability, and References

Measure Reliability1 Reference Brief Description
Outcome Measures
Single-Word Reading Outcomes (r’s between measures=.74-88)
PIAT Reading Recognition .85 (Dunn & Markwardt, 1970) Read words with increasing levels of semantic and phonemic difficulty.
PIAT Spelling .64 (Dunn & Markwardt, 1970) Choose the right spelling of a word heard orally from a selection of phonologically similar words. There are a few points that should be noted about some of the measures. Although it is a spelling task, it is more accurately described as a measure of orthographic word recognition skills.
Time-Limited Oral Reading .94 (Olson, Forsberg, Wise, & Rack, 1994; Olson, Wise, Conners, Rack, & Fulker, 1989) Read words presented to them within two seconds. Time of this task demands are minimal because the requirement is that the participant initiates a response within 2 seconds.
Cognitive Predictor Measures
Phonological Awareness (PA) Composite (r’s between measures= .61-.76)
Phoneme Deletion .80 (Olson et al., 1994) Remove phonemes from words and nonwords and said the resulting word or nonword.
The Lindamood Auditory Conceptualization Test .67 (Lindamood & Lindamood, 1971) Use colored blocks to represent and manipulate the phonemes within word and nonwords.
Pig Latin .78* (Olson et al., 1989) Remove the first phoneme from the beginning of a word, add it to the end of the word and add an “ay” after the phoneme.
WISC-R/III Vocabulary .86-.89 (Wechsler, 1974, 1991) Define words with increasingly complex and abstract meanings.
Rapid Automatized Naming (RAN) Composite (r’s between measures= .42-.68)
RAN (color, number, letters, and pictures) .80-86 (Denckla & Rudel, 1974, 1976) Name 5 repeating targets 50 times (in rows of 10) as quickly as possible.
Verbal Working Memory (VWM) Composite (r’s between measures= .35-.40)
WISC-R/III Digit Span Backwards .73-.77 (Wechsler, 1974, 1991) Repeat strings of numbers of increasing lengths in reverse order.
Sentence Span .65-.71 (Kuntsi, Stevenson, Oosterlaan, & Sonuga-Barke, 2001; Siegel & Ryan, 1989) Fill-in the last word of simple sentences, and then repeat these last words in order as the number of sentences increased across trials.
Counting Span .55-.67 (Kuntsi et al., 2001; R., Kurland, & Goldberg, 1982) Count the number of yellow dots they saw on one page, and then recall each number of dots per page at the end of a set.
Processing Speed (PS) Composite (r’s between measures= .56-.57)
WISC-R/III Coding .71-.77 (Wechsler, 1974, 1991) Quickly copy symbols that were associated with numbers in a key.
Colorado Perceptual Speed (CPS) Test Part 1 & 2 .81 (Decker, 1989) Quickly choose matching unpronounceable strings of letters or letters and numbers from a selection of items including three foils.
Identical Pictures Task (IPT) .82* (French, Ekstrom, & Price, 1963) Match a target picture to the correct picture present with four foils.
WISC-III Symbol Search .81 (Wechsler, 1991) Match two non-namable symbols on the left side of the page with options on the right. Mark “yes” if any of the symbol choices match one of the target symbols on the left, or “no” if none of the symbol choices matched either target symbols.
Covariate Measures
ADHD (Inattention) Symptoms Measure
ADHD Rating Scale .59-.89* (Barkley & Murphy, 1998; DuPaul, Power, Anastopoulos, & Reid, 1998) Inattention symptoms from the parent and teacher report version of the ADHD Rating Scale (ADHD RS). The OR rule was used to determine the severity of inattention symptoms (Lahey et al., 1994), where the highest symptom ratings from parents or teacher was averaged across the 9 symptoms used to create the inattention severity score.
Nonverbal/Visual-Spatial IQ Composite (r’s between measures= .51)
WISC-R/III Block Design .77-.81 (Wechsler, 1974, 1991) Given blocks and instructed to look at a picture of increasingly complex designs and recreate them within a specified time limit.
WISC-R/III Object Assembly .76-.81 (Wechsler, 1974, 1991) Given jumbled puzzle pieces depicting objects and were required to put the pieces together within a time limit.
Maternal Education — — Maternal self-reported years of education.
Open in a new tab
1

indicates test-retest reliability from published norms or research papers unless specified otherwise.

*

Cronbach’s alpha

Covariates

IQ.

Participants were administered subtests from the WISC-R (63%) or WISC-III (37%). A nonverbal/visual-spatial IQ composite comprised of WISC-R/III Object Assembly and Block Design tasks (see Table 2; Wechsler, 1974, Wechsler, 1991) was constructed in order to minimize overlap with study constructs. However, this nonverbal IQ measure was still overly conservative with respect to PS because both Block Design and Object Assembly are timed measures.

Inattention.

This sample includes a mild selection for children with reading difficulties and/or ADHD. We chose to include the children with ADHD in our sample because of the strong comorbidity between reading difficulties and ADHD (Chhabildas, Pennington, & Willcutt, 2001; Willcutt & Pennington, 2000). We were concerned that dropping the children with ADHD would bias the sample in a way that is not representative of the continuous associations between reading and ADHD.

To address the impact of ADHD, we included it as a covariate. Because ADHD is correlated with both reading and cognitive skills (Arnett et al., 2012; Brock & Knapp, 1996; McGrath et al., 2011; Okmi & Ann, 2000; Willcutt, Doyle, Nigg, Faraone, & Pennington, 2005), it could serve as a confounding variable. Previous work has demonstrated that inattentive ADHD symptoms significantly account for the relationship between ADHD and reading, and hyperactive/impulsive symptoms do not add any unique variance to this relationship once inattention is accounted for (Chhabildas et al., 2001; Willcutt & Pennington, 2000). Therefore, inattention symptoms were chosen as covariates.

Maternal education.

Previous studies have demonstrated that there is an association between SES and child vocabulary, and that this relationship impacts later reading skills (Whitehurst & Fischel, 2000). As such a measure of maternal self-reported years of education, one indicator of the broader construct of socioeconomic status (SES; Duncan, Magnuson, & Votruba-Drzal, 2015), was included as a covariate in secondary analyses, to test whether vocabulary effects can be attributed to aspects of the home environment.

Data cleaning and analysis

One twin from each twin pair was randomly selected to avoid statistical dependencies due to familial relationships. We generated composites for all constructs except for vocabulary and inattention where only one indicator was available. Composites were used instead of individual measures in order to improve reliability. The composites were informed by previously published confirmatory factor analyses with these measures (McGrath, 2011; Peterson, 2017) and intra-construct correlations within the current sample.

Before creating the composites, each measure was corrected for age, age2 (nonlinear effects of age), and parent-identified sex. Raw scores were used for each measure except in the case of a few variables (Vocabulary, Block Design, and Object Assembly) where we had to collapse across different test versions with different items (WISC-R/WISC-III subtests). In these cases, we used standard scores instead of raw scores. For each composite, we included scores from participants who completed at least two subtests. Outliers were winsorized to 4 standard deviations. Distributions of the composite scores were visually inspected for normality and found to be satisfactory.

All constructs investigated had <4% missingness except for measures of VWM (35% missing) and inattention (35% missing), as these measures were added later in the battery. There was a small cohort effect (missing at random) such that those with data versus without had slightly lower reading, vocabulary, RAN, and PS scores. As a result, for the two variables with high missingness (VWM and inattention), we conducted secondary analyses to ensure comparable results. For VWM, we conducted analyses with one VWM measure (digit span) which was present for nearly all participants and found similar results. We also comprehensively examined the impact of ADHD by covarying for inattention and dropping children with ADHD from the sample and results were comparable. As such, we conclude that missingness patterns have not substantially influenced these results.

Models testing for cognitive promotive and protective factors

Risk (PA), cognitive skills, and single-word reading outcomes were all entered as continuous variables into multiple regression models. Main effects of the cognitive skills were used to test for promotive factors and PA x cognitive interactions were used to test for protective factors. The models were as follows, where Cog is a stand-in for vocabulary, RAN, VWM, or PS in four separate models:

Reading=PA+Cog+(PA×Cog)

Developmental patterns

We conducted an exploratory set of analyses to examine whether age influenced the PA x cognitive interactions. We added 3-way age x PA x cognition interactions including all possible 2-way interactions. The following equation was used to test for developmental effects, where Cog is a stand-in for vocabulary, RAN, VWM, or PS in the four separate models:

Reading=PA+Age+Cog+(PA×Cog)+(PA×Age)+(Cog×Age)+(PA×Cog×Age)

Multiple testing correction

Because this study examined multiple models, significance was set using a Bonferroni-adjusted alpha. In the four main models, there were 9 independent terms and therefore the Bonferroni-adjusted significance value was .0056 (p =.05/9). This correction is conservative because of the method of the correction and because of the known correlations among the cognitive skills investigated.

Results

Promotive and Protective Factors

All cognitive skills (vocabulary, RAN, VWM, PS) were significantly associated with the single-word reading composite while controlling for PA (Table 3). These main effect associations are consistent with previous research, including results published in partially overlapping samples with the current study (e.g., McGrath et al., 2011; Peterson et al., 2017; Willcutt et al., 2005). The effect sizes for the associations between each age- and sex-adjusted cognitive skills and the reading composite were significant and large for PA (R2=.48) and vocabulary (R2=.32), and moderate to large for PS (R2=.23), RAN (R2=.22), and VWM (R2=.21). These results indicate that each cognitive skill could be considered both a promotive and risk factor.

Table 3.

Multiple regressions predicting Single-Word Reading composite.

Model 1: Vocabulary Model 2: RAN Model 3: VWM Model 4: PS
Variable B SE P-value B SE P-value B SE P-value B SE P-value
Constant −0.025 0.017 0.140 −0.004 0.017 0.810 0.062 0.024 0.011 −0.013 0.017 0.440
PA 0.562 0.017 3.00E-180 0.599 0.018 5.00E-186 0.634 0.026 1.00E-103 0.601 0.018 2.00E-194
Vocabulary 0.325 0.017 5.00E-73
PA x Vocabulary 0.055 0.016 3.80E-04
RAN 0.239 0.018 1.00E-39
PA x RAN 0.011 0.015 0.460
VWM 0.143 0.025 2.40E-08
PA x VWM 0.047 0.024 0.049
PS 0.260 0.017 5.00E-48
PA x PS 0.035 0.015 0.022
n 1,797 1,807 1,179 1,807
Adjusted R2 0.568 0.523 0.481 0.535
P-Value <0.001 <0.001 <0.001 <0.001
Open in a new tab

Interaction terms for each of the four hypotheses (PA x vocabulary, PA x RAN, PA x VWM, PA x PS) tested for the presence of potential protective effects (Table 3). Only the PA x vocabulary interaction term was significant at the Bonferroni-corrected alpha (p<.0056), but its effect size was small (R2 change= .001, p=3.80E-04).

In order to understand the directionality of the vocabulary interaction, simple slope analyses were conducted (Aiken & West, 1991). Figure 2 plots the PA x vocabulary interaction. The graph indicates a pattern where individuals with better PA skills benefit more from better vocabulary than individuals with lower PA skills. This pattern of results was not consistent with classical protective effects (compare to Figure 1).

Figure 2.

Figure 2.

Open in a new tab

Simple Slope Plots of the PA x Vocabulary Interaction

Considering relevant covariates

IQ

A nonverbal/visual-spatial IQ composite score was included in a secondary model to determine whether the main effects of vocabulary, PA, and the PA x vocabulary interaction were robust when controlling for IQ. The main effects and interaction remained stable with some modest and expected decreases in B-values due to correlations among the predictors (Supplementary Table 3).

ADHD

Inattention symptoms were included in a secondary model to determine whether the main effects of vocabulary, PA, and the PA x vocabulary interaction were robust when controlling for this covariate. The main effects and interaction remained stable with some modest and expected decreases in B-values due to correlations among the predictors (Supplementary Table 4).

Maternal education

We also tested whether including a covariate that is an indicator for SES (i.e. maternal education level) would reduce the main effects of vocabulary, PA, and the vocabulary x PA interaction, given previous research linking vocabulary, reading, and SES. The vocabulary main effects and interactions remained stable and significant (Supplementary Table 5).

Developmental patterns

To test whether the PA x vocabulary interaction was stable across age, we included a three-way interaction of age x PA x vocabulary along with accompanying two-way interactions and main effects. The three-way interaction of age x PA x vocabulary was not significant (Supplementary Table 6).

Discussion

The dyslexia research literature has historically focused on risk factors with relatively less attention on resilience. The purpose of this study was to illustrate the application of the DP resilience framework to reading research. We investigated cognitive skills (i.e. vocabulary, RAN, VWM, and PS) that may contribute to resilience in single-word reading outcomes in children at risk for reading difficulties because of low PA scores.

We had a large, well-powered sample representing the full range of cognitive and reading skills, including children selected for dyslexia, and a comprehensive battery of relevant neuropsychological domains. These strengths in our design meant that our study was well-positioned to find cognitive protective factors, yet such effects were elusive. While our hypotheses about each cognitive factor serving as a promotive factor were confirmed, none of our hypotheses about cognitive protective factors were supported. Instead, the only significant interaction between PA and vocabulary was not consistent with the form of a classical protective factor interaction. In what follows, we discuss (1) evidence for promotive factors, (2) skill-enhancement interactions, and (3) limitations and future directions.

Promotive factors

Consistent with previous research, and in line with our hypotheses, results showed robust evidence for linear associations between PA, vocabulary, RAN, VWM, and PS with single-word reading skills. While these associations have been well-documented in previous research, explicitly labeling the high end of these linear relationships as promotive is rarely done and yet is helpful for drawing attention to skills that are associated with positive reading outcomes. In other words, for many children, the presence of stronger cognitive skills may not merely represent the absence of risk, but rather the presence of promotive factors for single-word reading. The effect sizes of these promotive effects were much larger than any of the interactions we detected, indicating that promotive effects are the primary cognitive mechanism for resilience in children at risk for reading difficulties in this large sample.

These promotive factor findings mean that Vocabulary, RAN, VWM, and PS skills are associated with better reading skills regardless of whether a child has stronger or weaker PA. A salient clinical question is whether a child with a strong risk factor might compensate using a strong promotive factor to achieve age-appropriate reading skills. Our regression equations do not explain enough variance to be clinically actionable (i.e., 48–57% of the variance) for this kind of question. However, the question itself points to an important next step for this work. These analyses are variable-centered but the field will need to move to person-centered analyses in future work (Masten, 2001; Masten & Obradović, 2007) in order to increase the clinical application of these resilience models.

Future work will also need to develop methods that can measure cumulative and combinatorial profiles of risk, promotive, and protective factors, as this would be consistent with current multifactorial models (Catts & Petscher, submitted; Masten, 2001; Masten & Obradović, 2007; McGrath, Peterson, & Pennington, 2020; Pennington, 2006; Wright et al., 2013). We chose a straight-forward analytic strategy using individual variables, rather than combinations of variables. While there has been some recent theoretical and empirical work examining cumulative mechanisms of risk and resilience associated with reading outcomes (Catts & Petscher, submitted; Kiuru et al., 2013), there is no current consensus on which variables at which levels of analysis should be included and how these variables should be combined and weighted. As such, we judged that a first step was to examine individual cognitive variables and their interactive effects to lay the groundwork for more advanced cumulative models.

Skill-enhancement interactions

We hypothesized that vocabulary and RAN would serve as protective factors and we conducted exploratory analyses with VWM and PS. None of these hypotheses were confirmed. Even though the PA x vocabulary interaction was significant, its form was not consistent with a classic protective effect. Instead, the interaction showed that children at low risk for reading difficulties benefitted more from strong vocabulary skills than children at high risk for reading difficulties. We termed this pattern a “skill-enhancement” interaction because it captured the synergistic combination of strong skills in PA and vocabulary. This interaction form has also been labeled as a “positive synergy” interaction within the gene x environment interaction literature (Pennington et al., 2009).

We predicted that vocabulary may be protective for single-word reading skills based on existing data supporting the semantic bootstrapping hypothesis (Muter & Snowling, 2009; Nation & Snowling, 1998; Snowling et al., 2003). Our findings do not contradict the semantic bootstrapping hypothesis, but rather provide clarification regarding the effectiveness of semantic bootstrapping across the full range of risk. The present study showed that while vocabulary helped children’s reading skills in the face of phonological risk (i.e., main effect), this effect was more robust for children with lower risk (i.e., skill-enhancement interaction). Together these results clarify previous findings and indicate that when PA skills are set as the risk benchmark, stronger vocabulary functions as both a promotive factor for all children and additionally confers a skill-enhancement effect for children with lower risk (i.e. higher PA skills). However, vocabulary skills do not seem to function as a protective factor in the context of weak PA.

The skill-enhancement pattern is reminiscent of the well-known “Matthew Effect” (Stanovich, 1986) where strong reading skills potentiate other cognitive and language skills over time, and vice versa. However, the Matthew effect refers to mediation patterns over time and the present study used moderation in a cross-sectional sample, so the analytic approaches are not parallel. Nevertheless, the idea that strong cognitive skills can enhance other cognitive skills is a theme evident in both the Matthew Effect and the current study.

In interpreting the skill-enhancement interaction in the current study, there are three salient points to emphasize. First, the effect sizes are small. If we consider the effects in standard score points (M = 100, SD = 15), the interaction effect of a one standard deviation increase in PA (15 points) and a one standard deviation increase in vocabulary (15 points) results in less than 1 standard score point increase in single-word reading above the expected additive effects. This interaction effect is vanishingly small compared to the main effects of these variables, meaning that this skill-enhancement effect is not greatly widening the reading gap between children at high and low risk for reading difficulties because of PA skills.

Second, while no classic protective interactions were identified, this does not mean that children at risk for reading difficulties lack other cognitive strengths. Indeed, we observed the full range of cognitive skills in the children at risk for reading difficulties. Our finding is specific to the question of whether cognitive strengths can buffer against a specific risk factor for reading difficulties, PA. It may be that PA skills are so fundamental to reading development that it is difficult to see buffering via the cognitive skills measured in this study. While our results did not find classic protective effects, we did not test an exhaustive list of potential protective factors. Such protective effects may exist for other cognitive skills, such as other executive function domains besides working memory (i.e. inhibition and shifting; Miyake et al., 2000) or other individual differences such as emotion regulation, frustration tolerance, or motivation. Potential protective factors for single-word reading may also exist at other levels of analysis, such as family systems, classrooms, schools, and neighborhoods (see Dvorsky & Langberg (2016) for a DP analysis of ADHD across levels of analysis that could guide further research in reading). While this study did not find cognitive protective interactions, it is entirely possible that protective effects can be found with other cognitive skills or other levels of analysis.

Third, the lack of classic protective factor interactions does not provide any information regarding the potential effectiveness of interventions for children at risk for reading difficulties. Our study investigated individual differences, which is a separate research question from whether interventions could show protective, gap-closing effects (i.e., more effective for children at high versus low risk). In fact, it has been shown that reading interventions can have such gap-closing (protective) effects for children who are dual language learners and children eligible to receive free school meals (Machin, McNally, & Viarengo, 2016).

Limitations and future directions

The current findings should be interpreted in the context of its limitations.

Demographics

This sample has less socioeconomic, racial, and ethnic diversity than the general United States population. Thus, results may not generalize to all socioeconomic, racial, and ethnic groups from different geographic locations. Future research should include more diverse samples and should consider the influence of self-reported gender in addition to sex assigned at birth.

Measurement

Our language measures were restricted to a single measure of vocabulary, but the semantic bootstrapping hypothesis refers to language and oral comprehension skills more generally. Thus, we may have identified protective interactions if we had measured language more comprehensively.

Previous work has shown that cognitive skills are moderately correlated with each other (i.e., cognitive “g”; McGrew & Evans, 2004; Spearman, 1904), which could make it difficult to find discrepant cognitive profiles (i.e. low PA but high cognitive skills). While our cognitive composites were correlated with PA at moderate levels (r’s = .37-.50), these correlations are not so strong that such discrepant profiles do not exist. For example, we had a meaningful subset of children (8%-10%; N = 95-195) who had highly discrepant profiles with weaker PA (1 SD below the mean) but stronger cognitive skills (1 SD above the mean) for each cognitive construct. Still, future research designs could focus explicitly on multivariate outliers and discrepant profiles.

Longitudinal and Developmental Considerations

The cross-sectional nature and age range of our study did not allow us to thoroughly examine the longitudinal processes and developmental nature of resilience mechanisms. While we did not identify any developmental moderation of skill-enhancement interactions within our cross-sectional dataset, a critical next step for future research will be to continue studying cognitive promotive and protective factors using longitudinal samples where the unfolding of developmental processes can be observed. While some previous longitudinal work has examined whether a subset of these cognitive factors (i.e. vocabulary, VWM, RAN) is associated with resilience in samples of children at risk for dyslexia (Eklund et al., 2013; Snowling & Melby-Lervåg, 2016; Torppa et al., 2015), more longitudinal work using DP guidelines is needed to explicitly test how these factors may function as promotive and/or protective factors across development. As suggested by van Viersen and colleagues, it may take time for promotive factors to “build up” to aid in the maintenance and growth in reading skills across development (van Viersen et al., 2015, 2019). This developmental notion of resilience is in line with conceptualizations from DP research, which considers resilience a developmental process that is best captured and understood via longitudinal analyses (Masten, 2006, 2001).Thus, while more longitudinal research will be needed to confirm the causal mechanisms, promotive factors may make important contribution to future mechanisms of resilience for word reading.

When considering development, it is also important to identify the sensitive age range to capture important mechanisms. Our cross-sectional sample was 8–16 years old. By 8 years of age, most of direct reading instruction has already occurred, and students shift from “learning to read” to “reading to learn” (Chall, 1967). Therefore, it is possible that protective factors operate at younger ages outside of this study’s age range. Future work should investigate the possibility of developmentally sensitive protective factors that are associated with resilience in early language and literacy milestones.

Conclusions

In summary, the goal of this study was to reinforce the application of a resilience perspective within the reading literature. This study draws attention to the importance of explicitly identifying and labeling cognitive risk and promotive factors when linear associations are found between cognitive and reading skills. Consistent with previous research, we found that vocabulary, VWM, PS, and RAN were all risk and promotive factors for single-word reading skills. We did not identify classic cognitive protective interactions, but rather a skill-enhancement interaction where PA and vocabulary combined synergistically at the high end of the distribution. Even though protective effects were not identified at the cognitive level of analysis, this study highlights the need for continued resilience research, with the goal of uncovering protective and promotive factors that may reside at other levels of analysis. Identification of protective factors could enhance existing, evidence-based reading interventions with the goal of promoting more efficient gap-closing effects between children at high and low risk for reading difficulties.

Supplementary Material

Supplemental_Tables

Acknowledgements

We gratefully acknowledge Sarah Crennen, M.A. for research coordination, Nina Anderson for editorial consultation, and the families who participated in this study.

Funding

This study was supported by grants P50 HD027802 and R15 HD086662 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD).

Footnotes

Conflict of interest statement

The authors declare that Lauren M. McGrath and Bruce F. Pennington receive royalties from the textbook, Diagnosing Learning Disorders: From Science into Practice, 3rd Edition from Guilford Press. All other authors declare no conflicts of interest. The findings reported in this manuscript are original, have not been published previously, and have not been simultaneously submitted elsewhere.

References

  1. Aiken LS, & West SG (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage Publications, Inc. [Google Scholar]
  2. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: Author. [Google Scholar]
  3. Arnett AB, Pennington BF, Willcutt E, Dmitrieva J, Byrne B, Samuelsson S, & Olson RK (2012). A cross-lagged model of the development of ADHD inattention symptoms and rapid naming speed. Journal of Abnormal Child Psychology, 40(8), 1313–1326. doi: 10.1007/s10802-012-9644-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barkley RA, & Murpphy KR (1998). Attention-deficit hyperactivity disorder: A clinical workbook (2nd ed.). New York, NY: The Guilford Press. [Google Scholar]
  5. Bishop DV, McDonald D, Bird S, & Hayiou-Thomas ME (2009). Children who read words accurately despite language impairment: Who are they and how do they do it? Child Development, 80(2), 593–605. doi: 10.1111/j.1467-8624.2009.01281.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brock SE, & Knapp PK (1996). Reading comprehension abilities of children with Attention-Deficit/Hyperactivity Disorder. Journal of Attention Disorders, 1(3), 173–185. Retrieved from 10.1007/s10802-012-9686-83 doi: [DOI] [Google Scholar]
  7. Catts HW, & Petscher Y (submitted). A cumulative risk and protection model of dyslexia. 1–34. doi: 10.1017/CBO9781107415324.004 [DOI] [PubMed]
  8. Chall JS (1967). Learning to read: The great debate. New York, NY: McGraw-Hill. [Google Scholar]
  9. Chhabildas N, Pennington BF, & Willcutt EG (2001). A comparison of the neuropsychological profiles of the DSMIV subtypes of ADHD. Journal Of Abnormal Child Psychology, 29(6), 529–540. Retrieved from 10.1023/A:10122812260286 doi: [DOI] [PubMed] [Google Scholar]
  10. Cicchetti D, & Toth SL (2009). The past achievements and future promises of developmental psychopathology: The coming of age of a discipline. Journal of Child Psychology and Psychiatry and Allied Disciplines, 50(1–2), 16–25.doi: 10.1111/j.1469-7610.2008.01979.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Decker SN (1989). Cognitive processing rates among disabled and normal reading young adults: A nine year follow-up study. Reading and Writing, 1(2), 123–134. [Google Scholar]
  12. DeFries JC (1997). Colorado learning disabilities research center. Learning Disabilities: A Multidisciplinary Journal, 8 (1), 7–19. [Google Scholar]
  13. Denckla MB, & Rudel R (1974). Rapid “automatized” naming of pictured objects, colors, letters and numbers by normal children. Cortex, 10(2), 186–202. doi: 10.1016/S0010-9452(74)80009-2 [DOI] [PubMed] [Google Scholar]
  14. Denckla MB, & Rudel RG (1976). Rapid “automatized” naming (R.A.N.): Dyslexia differentiated from other learning disabilities. Neuropsychologia, 14(4), 471–479. doi: 10.1016/0028-3932(76)90075-0 [DOI] [PubMed] [Google Scholar]
  15. Donahue ML, & Pearl R (2003). Studying social development and learning disabilities is not for the faint-hearted: Comments on the risk/resilience framework. Learning Disabilities Research & Practice, 18(2), 90–93. doi: 10.1111/1540-5826.00064 [DOI] [Google Scholar]
  16. Duncan GJ, Magnuson K, & Votruba-Drzal E (2015). Children and socioeconomic status. In Lerner RM, Bornstein MH, & Leventhal T (Eds.), Handbook of child psychology and developmental science (7th ed., pp. 1–40). Hoboken, NJ: John Wiley & Sons, Incoperated. doi: 10.1002/9781118963418.childpsy414 [DOI] [Google Scholar]
  17. Dunn LM, & Markwardt FC (1970). Examiner’s manual: Peabody individual achievement test. Circle Pines, MN: American Guidance Service. [Google Scholar]
  18. DuPaul GJ, Power TP, Anastopoulos AD, & Reid R (1998). ADHD rating scale–IV. New York, NY: Guilford. [Google Scholar]
  19. Dvorsky MR, & Langberg JM (2016). A review of factors that promote resilience in youth with ADHD and ADHD symptoms. Clinical Child and Family Psychology Review, 19(4), 368–391. doi: 10.1007/s10567-016-0216-z [DOI] [PubMed] [Google Scholar]
  20. Eklund KM, Torppa M, & Lyytinen H (2013). Predicting reading disability: Early cognitive risk and protective factors. Dyslexia, 19(1), 1–10. doi: 10.1002/dys.1447 [DOI] [PubMed] [Google Scholar]
  21. French JW, Ekstrom RG, & Price LA (1963). Manual for a kit of reference tests for cognitive factors. Princeton, NJ: Educational Testing Service. [Google Scholar]
  22. Fry AF, & Hale S (1996). Processing speed, working memory, and fluid intelligence: Evidence for a developmental cascade. Psychological Science, 7(4), 237–241. [Google Scholar]
  23. Haft SL, Myers CA, & Hoeft F (2016). Socio-emotional and cognitive resilience in children with reading disabilities. Current Opinion in Behavioral Sciences, 10, 133–141. doi: 10.1016/j.cobeha.2016.06.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hulme C, & Snowling MJ (2013). Learning to read: What we know and what we need to understand better. Child Development Perspectives, 7(1), 1–5. doi: 10.1111/cdep.12005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kail RV (2007). Longitudinal evidence that increases in processing speed and working memory enhance children’s reasoning. Psychological Science, 18(4), 312–314. doi: 10.1111/j.1467-9280.2007.01895.x [DOI] [PubMed] [Google Scholar]
  26. Kail RV, & Hall LK (1994). Processing speed, naming speed, and reading. Developmental Psychology, 30(6), 949–954. doi: 10.1037/0012-1649.30.6.949 [DOI] [Google Scholar]
  27. Kiuru N, Lerkkanen M-K, Pekka N, Poskiparta E, Ahonen T, Poikkeus A-M, & Nurmi J-E (2013). The role of reading disability risk and environmental protective factors in students’ reading fluency in grade 4. Reading Research Quarterly, 48(4), 349–368. doi: 10.1002/rrq.53 [DOI] [Google Scholar]
  28. Kuntsi J, Stevenson J, Oosterlaan J, & Sonuga-Barke EJS (2001). Test–retest reliability of a new delay aversion task and executive function measures. British Journal of Developmental Psychology, 19(3), 339–348. doi: 10.1348/026151001166137 [DOI] [Google Scholar]
  29. Kurland DM, & Goldberg J (1982). Operational efficiency and the growth of short-term memory span. Journal of Experimental Child Psychology, 33(3), 386–404. doi: 10.1016/0022-0965(82)90054-6 [DOI] [Google Scholar]
  30. Lahey BB, Applegate B, MCBurnett K, Biederman J, Greenhill L, Hynd GW, … Al E (1994). DSM-IV field trials for attention deficit hyperactivity disorder in children and adolescents. American Journal of Psychiatry, 151(11), 1673–1685. doi: 10.1176/ajp.151.11.1673 [DOI] [PubMed] [Google Scholar]
  31. Lindamood CH, & Lindamood PC (1971). Lindamood auditory conceptualization test. Boston, Massachusetts: Reading Resources Corp. [Google Scholar]
  32. Luthar S, Cicchetti D, & Becker B (2000). The construct of resilience: A critical evaluation and guidelines for future work. Child Development, 71(3), 543–562. doi: 10.1111/1467-8624.00164 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Lyon GR, Shaywitz SE, & Shaywitz BA (2003). Defining dyslexia, comorbidity, teachers’ knowledge of language and reading A definition of dyslexia. Annals of Dyslexia, 53, 1–14. [Google Scholar]
  34. Machin S, McNally S, & Viarengo M (2016). “Teaching to teach” literacy. CEP Discussion Paper No. 1425. Centre for Economic Performance, 1–42. [Google Scholar]
  35. Masten AS, & Garmezy N (1985). Risk, vulnerability, and protective factors in developmental psychopathology. In Lahey E, Benjamin B, & Kazdin A (Eds.), Advances in clinical child psychology (8th ed., pp. 1–52). New York, NY: Plenum Press. [Google Scholar]
  36. Masten AS (2001). Ordinary magic: Resilience processes in development. American Psychologist, 56(3), 227–238. doi: 10.1037//0003066X.56.3.227 [DOI] [PubMed] [Google Scholar]
  37. Masten AS (2006). Developmental psychopathology: Pathways to the future. International Journal of Behavioral Development, 30(1), 47–54. doi: 10.1177/0165025406059974 [DOI] [Google Scholar]
  38. Masten AS, & Reed MJ (2009). Resilience in development. In Snyder CR & Lopez SJ (Eds.), Handbook of Positive Psychology (pp. 74–89). New York, NY: Oxford University Press. [Google Scholar]
  39. Masten AS, & Obradović J (2007). Competence and resilience in development. Annals of the New York Academy of Sciences, 1094(1), 13–27. doi: 10.1196/annals.1376.003 [DOI] [PubMed] [Google Scholar]
  40. McBride-Chang C, Lam F, Lam C, Chan B, Fong CYC, Wong TTY, & Wong SWL (2011). Early predictors of dyslexia in Chinese children: Familial history of dyslexia, language delay, and cognitive profiles. Journal of Child Psychology and Psychiatry and Allied Disciplines, 52(2), 204–211. doi: 10.1111/j.1469-7610.2010.02299.x [DOI] [PubMed] [Google Scholar]
  41. McGrath LM, Pennington BF, Shanahan MA, Barnard HD, Willcutt EG, & Olson RK (2011). A multiple deficit model of reading disability and attention-deficit/hyperactivity disorder: Searching for shared cognitive deficits. Journal of Child Psychology and Psychiatry, 52(5), 547–557. doi: 10.1111/j.1469-7610.2010.02346.x.A [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McGrath LM, Peterson RL, & Pennington BF (2020). The multiple deficit model: Progress, problems, and prospects. Scientific Studies of Reading, 24(1), 7–13. doi: 10.1080/10888438.2019.1706180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. McGrew KS, & Evans JJ (2004). Internal and external factorial extensions to the Cattell-Horn-Carroll (CHC) theory of cognitive abilities: A Review of factor analytic research since Carroll’s seminal 1993 treatise. Reading and Writing, 2, 44. Retrieved from http://www.iapsych.com/HCARR2.pdf/nhttp://scholar.googleusercontent.com/scholar?q=cache:30MkADjiFTAJ:scholar.google.com/+cattell+horn+adhd&hl=en≈sdt=0,11≈vis=1 [Google Scholar]
  44. Miyake A, Friedman NP, Emerson MJ, Witzki AH, Howerter A, & Wager TD (2000). The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41(1), 49–100. doi: 10.1006/cogp.1999.0734 [DOI] [PubMed] [Google Scholar]
  45. Moll K, Loff A, & Snowling MJ (2013). Cognitive endophenotypes of dyslexia. Scientific Studies of Reading, 17(6), 385–397. doi: 10.1080/10888438.2012.736439 [DOI] [Google Scholar]
  46. Muter V, & Snowling MJ (2009). Children at familial risk of dyslexia: Practical implications from an at-risk study. Child and Adolescent Mental Health, 14(1), 37–41. doi: 10.1111/j.1475-3588.2007.00480.x [DOI] [Google Scholar]
  47. Narayan AJ (2015). Personal, dyadic, and contextual resilience in parents experiencing homelessness. Clinical Psychology Review, 36, 56–69. doi: 10.1016/j.cpr.2015.01.005 [DOI] [PubMed] [Google Scholar]
  48. Nation K, & Snowling MJ (1998). Individual differences in contextual facilitation: Evidence from dyslexia and poor reading comprehension. Child Development, 69(4), 996–1011. doi: 10.1111/j.1467-8624.1998.tb06157.x [DOI] [PubMed] [Google Scholar]
  49. Nigg JT, Willcutt EG, Doyle AE, & Sonuga-Barke EJS (2005). Causal heterogeneity in attention-deficit/hyperactivity disorder: Do we need neuropsychologically impaired subtypes? Biological Psychiatry, 57(11), 1224–1230. Retrieved from 10.1016/j.biopsych.2004.08.025 [DOI] [PubMed] [Google Scholar]
  50. Norton ES, & Wolf M (2012). Rapid automatized naming (RAN) and reading fluency: Implications for understanding and treatment of reading disabilities. Annual Review of Psychology, 63(1), 427–452. doi: 10.1146/annurevpsych-120710-100431 [DOI] [PubMed] [Google Scholar]
  51. Okmi HK, & Ann PK (2000). Language characteristics of children with ADHD. Communication Disorders Quarterly, 21(3), 154. Retrieved from http://proquest.umi.com/pqdweb?did=53548919&Fmt=7&clientId=45089&RQT=309&VName=PQD [Google Scholar]
  52. Olson RK, Forsberg H, Wise B, & Rack J (1994). Measurement of word recognition, orthographic, and phonological skills. In Lyon GR (Ed.), Frames of reference for the assessment of learning disabilities: New views on measurement issues (pp. 243–277). Baltimore, MD: Paul H Brookes Publishing Co. [Google Scholar]
  53. Olson RK, Wise B, Conners F, Rack J, & Fulker DW (1989). Specific deficits in component reading and language skills: Genetic and environmental influences. Journal of Learning Disabilities, 22(6), 339–348. doi: 10.1177/002221948902200604 [DOI] [PubMed] [Google Scholar]
  54. Peng P, Barnes M, Wang CC, Wang W, Li S, Swanson HL, … Tao S (2018). Meta-analysis on the relation between reading and working memory. Psychological Bulletin, 144(1), 48–76. doi: 10.1037/bul0000124 [DOI] [PubMed] [Google Scholar]
  55. Pennington BF (2006). From single to multiple deficit models of developmental disorders. Cognition, 101(2), 385–413. doi: 10.1016/j.cognition.2006.04.008 [DOI] [PubMed] [Google Scholar]
  56. Pennington BF, McGrath LM, Rosenberg J, Barnard H, Smith SD, Willcutt EG, … Olson RK (2009). Gene × environment interactions in reading disability and attention-deficit/hyperactivity disorder. Developmental Psychology, 45(1), 77–89. doi: 10.1037/a0014549 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Pennington BF, Santerre-Lemmon L, Rosenberg J, MacDonald B, Boada R, Friend A, … Black JL (2012). Individual prediction of dyslexia by single versus multiple deficit models. Journal of Abnormal Psychology, 121(4), 212–224. doi: 10.1037/a0025823 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Peters L, & Ansari D (2019). Are specific learning disorders truly specific, and are they disorders? Trends in Neuroscience and Education, 17, 1–8. doi: 10.1016/j.tine.2019.100115 [DOI] [PubMed] [Google Scholar]
  59. Peterson RL, Boada R, McGrath LM, Willcutt EG, Olson RK, & Pennington BF (2017). Cognitive prediction of reading, math, and attention: Shared and unique influences. Journal of Learning Disabilities, 50(4), 408–421.doi: 10.1177/0022219415618500 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Peterson RL, Pennington BF, Shriberg LD, & Boada R (2009). What influences literacy outcome in children with speech sound disorder? Journal of Speech Language and Hearing Research, 52(5), 1175. doi: 10.1044/1092-4388(2009/08-0024) [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Plomin R, Haworth CMA, & Davis OSP (2009). Common disorders are quantitative traits. Nature Reviews Genetics, 10(12), 872–878. doi: 10.1038/nrg2670 [DOI] [PubMed] [Google Scholar]
  62. Puolakanaho A, Ahonen T, Aro M, Eklund K, Leppanen PHT, Poikkeus AM, … Lyytinen H (2007). Very early phonological and language skills: Estimating individual risk of reading disability. Journal of Child Psychology and Psychiatry and Allied Disciplines, 48(9), 923–931. doi: 10.1111/j.1469-7610.2007.01763.x [DOI] [PubMed] [Google Scholar]
  63. Raitano NA, Pennington BF, Tunick RA, Boada R, & Shriberg LD (2004). Pre-literacy skills of subgroups of children with speech sound disorders. Journal of Child Psychology and Psychiatry and Allied Disciplines, 45(4), 821–835. doi: 10.1111/j.1469-7610.2004.00275.x [DOI] [PubMed] [Google Scholar]
  64. Reiter A, Tucha O, & Lange KW (2005). Executive functions in children with dyslexia. Dyslexia, 11(2), 116–131. doi: 10.1002/dys.289 [DOI] [PubMed] [Google Scholar]
  65. Ring J, & Black JL (2018). The multiple deficit model of dyslexia: What does it mean for identification and intervention? Annals of Dyslexia, 68, 104–125. doi: 10.1007/s11881-018-0157-y [DOI] [PubMed] [Google Scholar]
  66. Rose LT, & Rouhani P (2012). Influence of verbal working memory depends on vocabulary: Oral reading fluency in adolescents with dyslexia. Mind, Brain, and Education, 6(1), 1–9. doi: 10.1111/j.1751-228X.2011.01135.x [DOI] [Google Scholar]
  67. Siegel LS, & Ryan EB (1989). The development of working memory in normally achieving and subtypes of learning disabled children. Child Development, 60(4), 973–980. doi: 10.2307/1131037 [DOI] [PubMed] [Google Scholar]
  68. Snowling MJ, Gallagher A, & Frith U (2003). Family risk of dyslexia is continuous: Individual differences in the precursors of reading skill. Child Development, 74(2), 358–373. doi: 10.1111/1467-8624.7402003 [DOI] [PubMed] [Google Scholar]
  69. Snowling MJ, & Melby-Lervag M (2016). Oral language deficits in familial dyslexia: A meta-analysis and review. Psychological Bulletin, 142(5), 498–545. doi: 10.1037/bul0000037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Spearman C (1904). “General intelligence,” objectively determined and measured. The American Journal of Psychology, 15(2), 201–292. doi: 10.2307/1412107 [DOI] [Google Scholar]
  71. Stanovich KE (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21, 360–407. doi: 10.1598/RRQ.21.4.1 [DOI] [Google Scholar]
  72. Swanson HL, Trainin G, Necoechea DM, & Hammill DD (2003). Rapid naming, phonological awareness, and reading: A meta analysis of the correlation evidence. Review of Educational Research, 73(4), 407–440. doi: 10.3102/00346543073004407 [DOI] [Google Scholar]
  73. Thompson PA, Hulme C, Nash HM, Gooch D, Hayiou-Thomas E, & Snowling MJ (2015). Developmental dyslexia: Predicting individual risk. Journal of Child Psychology and Psychiatry and Allied Disciplines, 56(9), 976–987. doi: 10.1111/jcpp.12412 [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Torppa M, Eklund K, van Bergen E, & Lyytinen H (2015). Late-emerging and resolving dyslexia: A follow-up study from age 3 to 14. Journal of Abnormal Child Psychology, 43(7), 1389–1401. doi: 10.1007/s10802-015-0003-1 [DOI] [PubMed] [Google Scholar]
  75. van Bergen E, van der Leij A, & de Jong PF (2014). The intergenerational multiple deficit model and the case of dyslexia. Frontiers in Human Neuroscience, 8(1), 1–13. doi: 10.3389/fnhum.2014.00346 [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. van Viersen S, de Bree EH, & de Jong PF (2019). Protective factors and compensation in resolving dyslexia. Scientific Studies of Reading, 23(6), 461–477. doi: 10.1080/10888438.2019.1603543 [DOI] [Google Scholar]
  77. van Viersen S, de Bree EH, Kroesbergen EH, Slot EM, & de Jong PF (2015). Risk and protective factors in gifted children with dyslexia. Annals of Dyslexia, 65(3), 178–198. doi: 10.1007/s11881-015-0106-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Vandewalle E, Boets B, Ghesquiere P, & Zink I (2010). Who is at risk for dyslexia? Phonological processing in five-to seven-year old dutch-speaking children with SLI. Scientific Studies of Reading, 14(1), 58–84. doi: 10.1080/10888430903242035 [DOI] [Google Scholar]
  79. Vellutino FR, Fletcher JM, Snowling MJ, & Scanlon DM (2004). Specific reading disability (dyslexia): What have we learned in the past four decades? Journal of Child Psychology and Psychiatry, 1(50), 2–40. doi: 10.1046/j.0021-9630.2003.00305.x [DOI] [PubMed] [Google Scholar]
  80. Wadsworth LP, Morgan LP, Hayes-Skelton S, Roemer L, & Suyemoto KL (2016). Ways to boost your research rigor through increasing your cultural competence (part 1 of 2). The Behavior Therapist, 39(3), 76–82. [Google Scholar]
  81. Wang S, Allen RJ, Lee JR, & Hsieh CE (2015). Evaluating the developmental trajectory of the episodic buffer component of working memory and its relation to word recognition in children. Journal of Experimental Child Psychology, 133, 16–28. doi: 10.1016/j.jecp.2015.01.002 [DOI] [PubMed] [Google Scholar]
  82. Wechsler D (1974). Manual for the Wechsler intelligence scale for children–revised edition. San Antonio, TX: The Psychological Corporation. [Google Scholar]
  83. Wechsler D (1991). The Wechsler intelligence scale for children–third edition. San Antonio, TX: The Psychological Corporation. [Google Scholar]
  84. Whitehurst GJ, & Fischel JE (2000). Reading and language impairments in conditions of poverty. In Bishop DVM & Leonard LB (Eds.), Speech and language impairments in children: Causes, characteristics, intervention and outcome (pp. 53–71). Hove, England: Psychology Press. [Google Scholar]
  85. Willcutt EG, McGrath LM, Pennington BF, Keenan JM, Defries JC, Olson RK, & Wadsworth SJ (2019). Understanding comorbidity between specific learning disabilities. In Fuchs LS & Compton DL (Eds.), Models for innovation: Advancing approaches to higher-risk and higher-impact learning disabilities science. New directions for child and adolescent development (pp. 91–109). doi: 10.1002/cad [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Willcutt EG, Pennington BF, Olson RK, Chhabildas N, & Hulslander J (2005). Neuropsychological analyses of comorbidity between reading disability and attention deficit hyperactivity disorder: In search of the common deficit. Developmental Neuropsychology, 27(1), 35–78. doi: 10.1207/s15326942dn2701 [DOI] [PubMed] [Google Scholar]
  87. Wright MO, Masten AS, & Narayan AJ (2013). Resilience processes in development: Four waves of research on positive adaptation in the context of adversity. In Goldstein S & Brooks RB (Eds.), Handbook of resilience in children (2nd ed., pp. 1–527). doi: 10.1007/978-1-4614-3661-4 [DOI] [Google Scholar]
  88. Yu X, Zuk J, & Gaab N (2018). What factors facilitate resilience in developmental dyslexia? Examining protective and compensatory mechanisms across the neurodevelopmental trajectory. Child Development Perspectives, 12(4), 240–246. doi: 10.1111/cdep.12293 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental_Tables
NIHMS1656134-supplement-Supplemental_Tables.docx (39.6KB, docx)

RESOURCES