Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 May 1.
Published in final edited form as: Dev Sci. 2017 Aug 15;21(3):e12589. doi: 10.1111/desc.12589

Literacy Acquisition Influences Children’s Rapid Automatized Naming

Robin L Peterson 1,2, Anne B Arnett 3, Bruce F Pennington 3, Brian Byrne 4,5,6,7, Stefan Samuelsson 4, Richard K Olson 4,8
PMCID: PMC5997458  NIHMSID: NIHMS969980  PMID: 28812316

Abstract

Previous research has established that learning to read improves children’s performance on reading-related phonological tasks, including phoneme awareness (PA) and nonword repetition. Few studies have investigated whether literacy acquisition also promotes children’s rapid automatized naming (RAN). We tested the hypothesis that literacy acquisition should influence RAN in an international, longitudinal population sample of twins. Cross-lagged path models evaluated the relationships among literacy, PA, and RAN across four time points from prekindergarten through grade 4. Consistent with previous research, literacy showed bidirectional relationships with reading-related oral language skills. We found novel evidence for an effect of earlier literacy on later RAN, which was most evident in children at early phases of literacy development. In contrast, the influence of earlier RAN on later literacy was predominant among older children. These findings imply that the association between these two related skills is moderated by development. Implications for models of reading development and for dyslexia research are discussed.

Keywords: reading, language development, rapid serial naming, longitudinal, developmental dyslexia

Introduction

Becoming literate causes powerful changes in brain function, particularly for the processing of spoken language (Dehaene et al., 2010). Pioneering work (Morais, Cary, Alegria, & Bertelson, 1979) demonstrated that literacy influences performance on meta-phonological tasks previously conceptualized purely as reading predictors. There has recently been mounting interest in understanding how written language impacts a variety of oral language processes (Nation & Hulme, 2011; Pattamadilok, Perre, & Ziegler, 2011; Rastle, McCormick, Bayliss, & Davis, 2011). However, relatively little research has investigated whether literacy exerts an effect on rapid automatized naming (RAN), a well-established reading-related skill. The current study tested the hypothesis that the relationship between literacy and RAN is reciprocal.

Effects of Literacy on Reading “Predictors”

Reading skill is related to a number of oral language processes. The triad of phoneme awareness (PA), verbal short-term memory, and RAN show particularly robust associations with word reading (Ramus & Szenkovits, 2008). PA tasks require explicit manipulation of individual sounds in words (e.g., “Say ‘mist’ without the /s/ sound); verbal short-term memory tasks require brief maintenance of verbal information in conscious awareness; and RAN tasks require speeded naming of a matrix of a small number of familiar objects. Thus, one view is that each task emphasizes a different aspect of phonological processing: awareness of individual speech sounds (PA), phonological loop function (verbal short-term memory), and efficient retrieval of lexical phonology (RAN). As a group, children with dyslexia exhibit deficits in all three skills (Peterson & Pennington, 2012), and longitudinal family risk studies document poor performance on these tasks in pre-dyslexic preschoolers (Pennington & Lefly, 2001; Snowling, Gallagher, & Frith, 2003). These findings have led to the characterization of dyslexia as a phonological disorder. However, mounting evidence indicates that the phonological deficit in dyslexia is partly a consequence, rather than solely a cause, of reading failure.

Early work suggested a causal link from PA to reading acquisition (Bradley & Bryant, 1978) but Morais and colleagues (1979) demonstrated that illiterate adults could not perform simple PA tasks, while a control group who had learned rudimentary reading skills in adulthood could. They argued that explicit awareness of individual speech sounds did not arise spontaneously in language development, and was mostly relevant in the context of reading an alphabetic orthography. The nature of the PA-word reading link has remained vigorously debated in recent years (Castles & Coltheart, 2004; Hulme, Snowling, Caravolas, & Carroll, 2005). Recently, well-controlled, randomized training studies have demonstrated that improvements in PA in conjunction with letter sound knowledge cause better reading (Hulme, Bowyer-Crane, Carroll, Duff, & Snowling, 2012) and that growth in orthographic skills promotes PA, even in preschool (Castles, Wilson, & Coltheart, 2011). Longitudinal studies also support bidirectional PA-reading relations (Burgess & Lonigan, 1998; Hogan, Catts, & Little, 2005). However, it is not clear that PA training in the absence of explicit teaching about print directly causes reading growth.

Nation and Hulme (2011) asked whether learning to read also changes nonword repetition, a verbal short-term memory measure. They reasoned that becoming literate could influence nonword repetition in two ways. If exposure to an alphabetic orthography promotes more finely segmented phonological representations, this would support the ability to repeat novel phonological forms. Alternatively, literate individuals might use an explicit orthographic strategy to encode nonwords. Consistent with their hypothesis, reading ability at age 6 uniquely predicted nonword repetition at age 7, after controlling for age 6 nonword repetition and language skill. Earlier nonword repetition did not uniquely predict later reading, likely because reading was so stable that little variance was left to explain. This study with children echoes earlier research comparing nonword repetition skills in naturally illiterate adults (who have no formal schooling) versus literate controls of otherwise similar social backgrounds. In comparison to literate adults, illiterate adults had more difficulty repeating nonwords but not real words (Castro-Caldas, Petersson, Reis, Stone-Elander, & Ingvar, 1998). These behavioral findings related to functional underactivations in brain regions supporting phonological loop function as well as attentional/executive processes in the illiterate group (Petersson, Reis, Askelöf, Castro-Caldas, & Ingvar, 2000).

A small body of research has investigated potential effects of literacy on RAN. Not surprisingly, early letter knowledge (a robust early reading predictor) influences later alphanumeric RAN (naming of numbers or letters) (Lervåg & Hulme, 2009; Wagner, Torgesen, & Rashotte, 1994) but no such effect has been detected for non-alphanumeric RAN (naming of pictures or color swatches). A handful of longitudinal studies have followed grade school children learning different orthographies and have tested whether the relationship between reading and RAN is bidirectional. Existing studies have included 1st to 5th grade children learning Norwegian (Lervag and Hulme, 2009), 1st to 2nd grade children learning Dutch (Verhagen et al., 2008), and 2nd to 5th grade children learning Chinese (Wei, Georgiou, & Deng, 2015). In each case, earlier RAN was related to later reading, but no influence of earlier reading on later RAN was detected. In contrast to these null findings, some evidence suggests that there may be an influence of earlier reading on later RAN for children with weak decoding skills. Compton (2003) studied first graders at multiple points throughout the year and reported that earlier literacy influenced later RAN numbers for “late decoders” but not for “early decoders”. Wolff (2014) recently reported that the relationship between reading speed and RAN was reciprocal from 3rd to 4th grade in Swedish children with poor decoding skills; the RAN composite used in this study included primarily, but not exclusively, alphanumeric measures. In sum, existing data have yielded mixed results with regard to whether learning to read influences later RAN. Significant effects have been detected only for young children in the very beginning of literacy acquisition or weaker readers.

Across languages, RAN relates to reading fluency more strongly than accuracy (Araújo, Reis, Petersson, & Faísca, 2015; Norton & Wolf, 2012), likely in part because it is speeded. Indeed, RAN and reading fluency both correlate with non-linguistic speeded measures, such as Wechsler Processing Speed (McGrath et al., 2011). Alternatively, RAN may tap linguistic processes that specifically support fluent reading, such as the ability to efficiently convert visual objects into phonological codes. Our view is that there are several sources for the RAN-reading fluency covariance. Regardless of the reasons for the association, a common assumption has been that RAN taps underlying cognitive processes that contribute causally to reading. Understanding whether the effect also flows in the opposite direction is important both theoretically and practically. A demonstration that reading influences RAN would provide further information about the ways in which literacy acquisition impacts cognitive development. From a clinical standpoint, evaluating children at risk for dyslexia for poor RAN would be less meaningful if low performance was largely a consequence, rather than a cause, of weak reading development.

There are several mechanisms through which learning to read might promote superior RAN. For alphanumeric RAN, an obvious candidate is increased automaticity of letter and digit knowledge. In addition, becoming a fluent reader should lead to improved left-to-right visual scanning and redeployment of visual attention. Supporting this notion is the fact that isolated naming (naming one centrally-presented object at a time) and serial naming show different growth patterns over the school years, with serial naming having the steeper learning slope (Logan, Schatschneider, & Wagner, 2011; Protopapas, Altani, & Georgiou, 2013). By approximately 2nd grade, serial naming is more strongly related to reading fluency than isolated naming is (de Jong, 2011; Georgiou, Parrila, Cui, & Papadopoulos, 2013). One interpretation is thus that the process of becoming a fluent reader improves children’s ability to process multiple targets, which in turn supports better serial RAN. Finally, experimental work has demonstrated that literate adults mandatorily activate orthographic codes in picture naming (Rastle et al., 2011). As children learn to read, such a process could lead to higher activations of lexical phonology (and hence faster naming times) for items in the RAN response set.

Overview of the Current Investigation

To evaluate whether learning to read promotes RAN, we performed cross-lagged path analyses in a longitudinal, international, population-based sample of twins (International Longitudinal Twin Sample or ILTS) from end of preschool through end of fourth grade. The ILTS includes twin pairs from the United States (Colorado), Australia, and Scandinavia (Sweden/Norway). Inclusion of participants from these different countries allows for exploration of potential cross-cultural influences including both effects of different orthographic systems (i.e., the more opaque English versus the more transparent Swedish/Norwegian) as well as potential instructional effects. At the time of data collection, children residing in Scandinavia were exposed to formal literacy instruction later than children residing in the English-speaking countries. Although our primary interest was to test for an impact of literacy on later RAN, we also included PA in our models. Because previous research has established effects of literacy on PA, inclusion of this variable provides a test of whether our methodology is sufficiently sensitive to detect an influence of literacy on RAN.

Method

Participants

Participants included a total of 1,147 twin pairs participating in the ILTS from the U.S. (n = 489 twin pairs), Australia (n = 266 twin pairs), or Scandinavia (n = 201 twin pairs from Norway and 191 twin pairs from Sweden). This sample has been described in detail elsewhere (Samuelsson et al., 2005; Willcutt et al., 2007). Briefly, at Time 1, all participants were in their final pre-Kindergarten (Pre-K) year, with ages ranging from 47 to 70 months (Australia), 54 to 71 months (U.S.), and 58 to 70 months (Scandinavia). Time 2 testing was conducted near the end of Kindergarten (K), and Time 3 testing near the end of 1st grade, and Time 4 testing near the end of 4th grade. At the end of 4th grade, only the U.S. and Swedish participants completed testing. Sample demographics are provided in Table 1.

Table 1.

Sample characteristics

Total Australia U.S. Scandinavia
N (twin pairs) 1,148 264 489 395
% Male 49.8 52.7 49.7 48.1
% Monozygotic 49.1 58.3 45.8 47.1
PreK age in months: M (SD) 58.93 (2.79) 57.24 (3.40) 58.75 (2.31) 60.71 (1.61)
K age in months: M (SD) 76.18 (4.87) 72.86 (4.18) 75.19 (3.67) 81.01 (3.47)
1st grade age in months: M (SD) 88.91 (5.05) 84.12 (4.25) 89.06 (3.81) 92.88 (3.79)
4th grade age in months: M (SD) 126.24 (4.24) n/a 125.43 (3.86) 128.45 (4.45)
% Caucasian* -- -- 64% --
Open in a new tab
*

Data on race were not collected in Australia or Scandinavia, where virtually 100% of participants were Caucasian.

Procedure

At study entry, participants completed a battery of cognitive, language, and preliteracy tests over a period of 5 days. Members of a twin pair were tested separately, either in quiet rooms at their preschool or in their homes. Follow-up testing at the ends of Kindergarten and 1st grades included a battery of cognitive, language, and literacy tests. Testing for each follow-up visit took place in a single session lasting approximately one hour at participants’ homes or schools.

Measures

Literacy

In PreK, most children were not yet able to read independently, so literacy was assessed with several measures of print knowledge, including a measure of environmental print recognition (Samuelsson, 2005), Clay’s (1972) Concepts About Print, and letter name and letter sound knowledge (Samuelsson, 2005). At the other three time points, literacy was assessed with the Test of Word Reading Efficiency Sight Word and Phonemic Decoding subtests, which required children to read as many words and pseudowords, respectively, as possible within 45 seconds (Torgesen, Wagner, & Rashotte, 1999). Both forms A and B were administered to maximize reliability at the end of kindergarten and first grade; at the end of 4th grade, Swedish participants completed both Forms A and B, while participants from the USA only completed Form A.

Phonological awareness

In PreK, PA was assessed with pre-reading measures designed specifically for this study (Byrne et al., 2002). These included tests similar to the Blending, Elision, and Sound Matching tests of the Comprehensive Test of Phonological Processing (CTOPP; Wagner, Torgesen, & Rashotte, 1999) as well as a dynamic measure emphasizing learning of phoneme awareness and a measure of rhyme and final sound awareness. PA was assessed with CTOPP Blending and Elision subtests at the end of kindergarten and first grade and with a phoneme deletion task at the end of fourth grade (Olson, Forsberg, Wise, & Rack, 1994).

Rapid automatized naming

RAN was assessed with CTOPP Rapid Naming subtests at each time point. Specific subtests were chosen to be developmentally appropriate and maximally related to literacy attainment and so varied by testing time, as follows: Color and Object Naming at the end of PreK, Color, Letter, and Digit Naming at the end of Kindergarten, and Letter and Digit Naming at the end of first grade and fourth grade.

Data Cleaning and Reduction

All variables were initially examined for normality, and outliers were Winsorized to within three standard deviations of the mean. Following these transformations, skewness and kurtosis values were within acceptable ranges (skewness: −1.1 to 1.5; kurtosis: −1.4 to 2.0).

As described above, Literacy, PA and RAN were assessed using different subtests at each of the time points. In order to study these constructs over time, we employed two-level confirmatory analyses, using maximum likelihood estimation with robust standard errors, to create single composite factor scores at each time point. Two-level models were used in order to account for non-independence of the twin data. The confirmatory factor analysis model included a single factor for each construct (i.e., literacy, PA or RAN) with the specific subtests described earlier as indicators. The resulting factor scores were saved and used for subsequent analyses.

Analyses

We tested for measurement invariance across countries in Mplus 7.3. Metric invariance was defined as a non-significant change in chi-square between a model with free parameters (configural) and one that held factor loadings constant across country. Scalar invariance was defined as a non-significant chi-square change between the configural model and one that held both factor loadings and intercepts constant across countries.

To test our hypotheses, we utilized a cross-lagged, path modeling design. After identifying the best-fitting model in the full sample, secondary exploratory analyses tested if the model fit equally well across different countries. Given the complexity of our proposed three-factor, cross-lagged model with nesting within twin pairs, we opted to create factor scores in order to reduce computational demand, improve model fit, and simplify interpretation of the constructs. Thus, as described above, we computed factor scores from the two-level confirmatory analysis model and used these factor scores, rather than latent variables, to model the family relations in a two-level path model.

Path modeling is well suited for studying neuropsychological development, due to its capacity to simultaneously estimate direct and indirect associations in longitudinal data (Hays, Marshall, Wang, & Sherbourne, 1994). The advantage of path modeling over another multivariate analysis, such as multiple regression, is that multiple factors (i.e. literacy, PA and RAN) can be entered into the equation simultaneously, as opposed to being forced into a temporal sequence. The autoregressive, cross-lagged path model estimates the association between these three constructs over time. Autoregressive path weights account for the stability of each measure across two consecutive time points, while the contemporaneous correlations between the two factors are also estimated.

There has been some concern in the literature that use of factor scores can induce bias in computing path weights, though any bias would likely be small and conservative, resulting in an underestimation of the weights (Skondral & Laake, 2001). To address this concern we ran a follow-up analysis choosing one twin at random from each pair. We then tested a structural equation model (SEM) investigating reciprocal effects of literacy with PA and RAN over time, treating the constructs of interest as latent variables rather than factor scores.

Results

Measurement Models

We first tested measurement models of our hypothesized constructs of literacy, PA, and RAN at each time point, using one twin from each pair selected at random. Model fit statistics were adequate, with χ2/df < 5.00, RMSEA ≤ 0.06, and CFI ≥ 0.96 for all models.

Next, we tested for measurement invariance across country. We found both metric and scalar invariance for the kindergarten, first grade, and fourth grade time points, evidenced by a non-significant χ2 difference for the configural versus metric and configural versus scalar models. However, at PreK, metric invariance was not found (χ2(30) = 57.16, p = .001). Examination of factor loadings suggested that this finding related to Sweden’s loadings on the literacy factor compared to the other countries. Indeed, when we dropped literacy from the model or dropped the Swedish participants, we found both metric and scalar invariance across countries at PreK. It is not surprising that we would fail to find metric invariance for literacy at this age given the differing educational practices across countries. Descriptive statistics for the literacy, PA, and RAN factor scores by country as well as Pearson correlations among the factors are included in Supplemental Tables 1 and 2.

Models Testing Reciprocal Effects of Literacy with PA and RAN

We tested a path model with literacy, PA, and RAN across the four time points. All first-order autoregressive effects were included. In addition, earlier PA and RAN both predicted later literacy, while earlier literacy predicted later PA and RAN. We allowed the PreK predictors to correlate with one another, and also allowed error terms for variables within the three later time points to correlate. To account for lack of full metric invariance, PreK print factor scores were created within country. All participants were included, with non-independence of twin data modeled using cluster analysis in Mplus. This model is shown in Figure 1.

Figure 1.

Figure 1

Open in a new tab

Path model showing reciprocal relations between literacy and oral language skills.

Notes. Values are standardized. Contemporaneous covariances not pictured. Pre-K = end of preschool; K = end of kindergarten; 1st = end of first grade; 4th = end of fourth grade. PA = phonological awareness composite; PRINT = print knowledge composite; READ = reading fluency composite; RAN = rapid automatized naming composite. Bolded paths are statistically significant.

Model fit was adequate (X2(33) = 294.85, p <.001; CFI = 0.97; RMSEA = 0.06). Overall, relationships between literacy and RAN were bidirectional. The influence of earlier RAN on later literacy was statistically significant across each time interval, although it was small in the youngest children (PreK to K: B=0.07, SE=0.02, p<.001; K to 1st: B = 0.31, SE=0.02, p<.001; 1st to 4th: B=0.37, SE=0.05, p<.001). The influence of earlier literacy on later RAN was most evident in younger children. From end of PreK to end of Kindergarten, this effect was significant (unstandardized B=5.72, SE=0.57, p<.001). The effect of earlier literacy on later RAN was small but statistically significant from end of K to end of 1st grade (B=0.02, SE=0.01, p=.020) and was not statistically significant from end of 1st grade to end of 4th grade (B=0.01, SE=0.01, p=.066).

Relationships between literacy and PA were also bidirectional. The influence of earlier PA on later literacy was statistically significant across each time interval, although it was small from end of kindergarten onwards (PreK to K: B=2.86, SE=0.24, p<.001; K to 1st: B=0.35, SE=0.12, p=.002; 1st to 4th: B=0.21, SE=0.10, p=.039). The influence of earlier literacy on later PA was significant and in the predicted direction from end of PreK to end of K (B=1.57, SE=0.15, p<.001) as well as end of 1st grade to end of 4th grade (B=0.17, SE=0.02, p<.001). From end of K to end of 1st grade, the effect of earlier literacy on later PA was negative and statistically significant after accounting for the autoregressive effect of PA (B=−0.02, SE=0.00, p<.001). This counter-intuitive result may reflect the fact that PA was assessed with identical measures in kindergarten and first grade and quite stable over this time period, leaving less variance to be explained by other measures. (Of note, the zero-order correlation between K literacy and first grade PA was moderate in size and in the predicted direction).

To avoid any possible bias in path coefficients, a follow-up structural equation model, using only one twin from each pair, treated constructs of interest as latent variables. We dropped the 4th grade time point from this model because the USA was the only remaining sample at that time point, and 4th grade literacy and PA factors were each informed by only a single indicator. To account for metric invariance identified in the CFA, above, we permitted factor loadings for the PreK Reading measures to vary for the Swedish group. Autoregressive, contemporaneous and cross-lag path coefficients were constrained equal across all groups.

The model had adequate fit (X2(1126) = 3115.31, p <.001; CFI = 0.94; RMSEA = 0.06; Supplementary Figure A). Given our focus on the developmental relationship between literacy and RAN, those results are presented in more detail here. The magnitude of the cross-lagged paths between literacy and RAN was very comparable to the path model across both time points. However, the small effect of earlier literacy on later RAN was non-significant from end of kindergarten to end of first grade. Specifically, from end of PreK to end of kindergarten, we found a significant influence of earlier print knowledge on later RAN (unstandardized B=1.41, SE=0.53, p<.01) and vice versa (unstandardized B=0.09, SE=0.03, p<.01). From end of kindergarten to end of 1st grade, the influence of earlier RAN on later reading was significant (unstandardized B=0.27, SE=0.03, p < .001), while the influence of earlier reading on later RAN was not (unstandardized B=−0.02, SE=0.03, p=.502).

Alphanumeric RAN vs. Non-Alphanumeric RAN

To further understand the RAN-literacy relationship, we divided the Kindergarten RAN composite into alphanumeric RAN (letters and numbers) and nonalphanumeric RAN (colors) and reran the main path model above separately for these two composites. The results for alphanumeric RAN were very similar to those reported above. With kindergarten non-alphanumeric RAN in the model, the effect of Pre-K literacy on Kindergarten RAN remained significant (unstandardized B=1.81, se=0.52, p=0.001), as did the effect of Kindergarten RAN on 1st grade literacy (unstandardized B = 0.19, SE=0.02, p<.001).

Cross-Cultural Effects

To test the consistency of our results across countries with different orthographies and educational practices, we constrained cross-lag coefficients, one at a time, to be equal for the combined U.S. and Australia versus Scandinavian samples. After correction for multiple comparisons (Benjamini-Hochberg), two significant differences emerged regarding the effect of earlier literacy on later PA, with a stronger effect for the Scandinavian group at the Pre-K to Kindergarten interval (Δχ2(1) = 315.58, p<.001) and a stronger effect for the English speaking group at the 1st to 4th grade interval (Δχ2(1) = 8.04, p=.004.

Discussion

This study tested for reciprocal effects of literacy with PA and RAN from end of PreK to end of 4th grade in population-based samples of twins from Australia, the U.S., and Scandinavia. Consistent with previous research, we found evidence for bidirectional relations between PA and literacy. The most important novel contribution concerns the impact of literacy development on RAN. In accord with our hypothesis, prior literacy skill predicted children’s later rapid naming from end of PreK to end of Kindergarten. However, there was not consistent evidence for an impact of earlier literacy on later RAN in older children.

A few previous studies have tested for an influence of literacy skill on later RAN. It has already been demonstrated that early letter knowledge predicts later alphanumeric RAN (Lervag & Hulme, 2009; Wagner et al., 1994). We replicated this intuitive finding. Importantly, in the current study, the influence of PreK literacy on Kindergarten color RAN was moderate in size and statistically significant, so the effect is not driven exclusively by increased automaticity of letter and number knowledge as literacy instruction gets underway. Similar to previous studies of grade school children in various countries including Norway, Holland, and China (Lervag & Hulme, 2009, Verhagen et al., 2008; Wei et al., 2015) we did not detect an effect of earlier reading on later RAN from 1st to 4th grade. In the current study, primary analyses suggested a small but significant influence of earlier reading on later RAN from kindergarten to 1st grade, but different statistical approaches yielded inconsistent results on this question. Thus, an emerging picture based on our results and previous studies is that the relationship between literacy and RAN is bidirectional, but the primary direction of influence may change with literacy development. Of note, we used exclusively serial RAN tasks, and from kindergarten onwards, assessment of literacy relied on timed reading of word and pseudoword lists. Previous literature would suggest developmental changes in the reading-RAN relationship could vary depending upon the ways in which these skills are assessed and in particular whether children are required to handle multiple targets at once (de Jong, 2011; Protopapas et al., 2013).

We found no cross-country differences in literacy-RAN relations in the current study. The pattern of results was similar for children learning more and less transparent orthographies and who had earlier and later starts to formal literacy instruction. This finding agrees with the growing literature showing an impressive degree of universality in the cognitive and brain bases of literacy development across countries and languages (Peterson & Pennington, 2012). We mainly found consistency across countries varying in orthography and educational practice for PA-literacy cross-lagged effects as well. The only significant differences were for the effects of earlier literacy on later PA, with a stronger effect in Scandinavian versus English-speaking countries from pre-K to K, and a stronger effect in the USA versus Swedish sample from 1st grade to 4th grade. The reasons for these differences are not entirely clear, but may relate to variance in the developmental trajectories of PA across countries owing to both educational practices and orthographies. In the relatively transparent Swedish orthography, PA develops very rapidly with formal literacy instruction and so might be less variable by the end of 4th grade (Anthony & Francis, 2005).

In sum, it appears likely that reading has reciprocal relationships with all three types of phonological processing task (PA, verbal short-term memory, and RAN) often conceptualized predominantly as literacy predictors or precursors. These findings question how much of the RAN deficit in dyslexia is a consequence, rather than a cause, of poor reading, at least early in literacy development. While pre-dyslexic preschoolers do show deficits in RAN (Boets, Wouters, van Wieringen, & Ghesquiere, 2006; Pennington & Lefly, 2001), this result could still reflect early differences in literacy knowledge. Most studies have been conducted in 5-year-olds, when there are well-documented differences in letter name and letter sound knowledge between children who will later develop dyslexia and those who will not (and which would clearly promote better alphanumeric RAN). Even for non-alphanumeric RAN, orthographic effects are difficult to exclude. Orthographic effects emerge early in literacy development (Castles, et al., 2011), and orthographic effects on isolated picture naming have been demonstrated in adults (Rastle et al., 2011). Even partial knowledge about the orthography of a pictured item (such as what letter an object name begins with) could plausibly promote greater phonological activation for the to-be-named items and hence faster RAN. Furthermore, while children with dyslexia have sometimes been reported to show a RAN deficit relative to reading level controls, this is less consistently true than for PA (Bruck, 1992; Chiappe, Stringer, Siegel, & Stanovich, 2002; Pennington, et al., 2001).

Current results have important theoretical and practical implications. They add to the growing body of literature showing that becoming literate changes aspects cognitive functioning for processes close to reading itself. Indeed, there may be no single reading-related skill that is purely predictive of reading and not also influenced by literacy development. We also extend previous research highlighting universality of many aspects of reading development (Peterson & Pennington, 2012). Reciprocal relations between literacy and reading-related skills were quite similar across the countries included, which varied in transparency of orthography and in early literacy practices. In other words, learning to read probably changes the brains of children in similar ways, no matter where they are growing up. From a practical standpoint, results highlight that even for young students, PA and RAN are better conceptualized as correlates of literacy development rather than “pure” causal factors. Thus, in the context of an educational or clinical evaluation for dyslexia or related learning difficulties, a finding of low performance in one of these skills does not necessarily pinpoint the cause of the child’s reading failure. While it is probably still valuable to assess reading correlates such as PA, RAN, processing speed, and broad oral language skill in the context of diagnostic evaluations, performance on these reading correlates cannot be used in any categorical fashion to rule a learning disability diagnosis in or out (see also Pennington et al., 2012).

Limitations and Future Directions

The current study has a number of important strengths including inclusion of a large, international, population-based sample of twins with multiple data points over a 5-year period. However, a number of limitations should also be acknowledged. Although the longitudinal results are consistent with a model in which learning to read causes better RAN early in literacy development, cross-lagged correlations cannot prove causality. Our study design does not elucidate the mechanisms that underlie the RAN-literacy link, and we had no measures of non-alphanumeric RAN beyond kindergarten. Results may not generalize to non-alphabetic languages, different developmental periods, or more highly selected samples at risk for literacy problems for a variety of reasons.

Future experimental training studies would most clearly establish causality; a school cut-off design comparing children of very similar ages who differ by a year of schooling could approximate random assignment to literacy training. It is important to understand whether the relationship between literacy and other speeded measures, such as Wechsler processing speed, is also reciprocal. Future experimental studies should include measures of both alphanumeric and non-alphanumeric serial and isolated naming as well as non-linguistic processing speed to further explore the reasons that becoming literate promotes faster RAN. Finally, the developmental research could benefit from an application of methods used predominantly in the adult literature, in which orthographic characteristics of stimuli are precisely manipulated.

Supplementary Material

Supplemental tables
supplemental figure. Supplementary Figure A.

Structural equation model of longitudinal relations among literacy and oral language skills.

Note: Contemporaneous covariances not pictured. Pre-K = end of preschool; K = end of kindergarten; 1st = end of first grade; 4th = end of fourth grade. PA = phonological awareness composite; Print = print knowledge composite; Read = reading fluency composite; RAN = rapid automatized naming composite. Bolded paths are statistically significant.

Values are standardized path coefficients, with the top values reflecting the USA, Australia and Norway samples, and the bottom values reflecting the sample from Sweden. Factor loadings for the observed variables on the pre-Kindergarten Print Knowledge latent factor were permitted to vary for the Swedish sample. Unstandardized path coefficients among the latent factors were constrained to be equal across Sweden and the other three countries, but standardized coefficients vary slightly due to differing latent factor variances.

Research Highlights.

  • The effects of literacy acquisition and reading-related oral language skills (rapid automatized naming and phonological awareness) were reciprocal in an international population-based twin sample.

  • An influence of earlier literacy on later rapid automatized naming was evident for children in the early phases of literacy acquisition (prekindergarten to kindergarten).

  • Effects were similar in the United States, Scandinavian, and Australian samples, despite differences in educational practices and written systems.

Acknowledgments

Funding was provided by the Australian Research Council (DP0663498 and DP0770805), the National Institute for Child Health and Human Development (HD27802, HD38526, and HD49027), and grants from the Research Councils of Norway and Sweden. We appreciate the input Lauren McGrath and Ted Westling provided on an earlier version of the manuscript. We thank the Colorado and Australian Twin Registries, our testers, and the children and parents involved.

References

  1. Anthony JL, Francis DJ. Development of phonological awareness. Current Directions in Psychological Science. 2005;14(5):255–259. [Google Scholar]
  2. Araújo S, Reis A, Petersson KM, Faísca L. Rapid automatized naming and reading performance: A meta-analysis. Journal of Educational Psychology. 2015;107(3):868. [Google Scholar]
  3. Boets B, Wouters J, van Wieringen A, Ghesquiere P. Auditory temporal information processing in preschool children at family risk for dyslexia: Relations with phonological abilities and developing literacy skills. Brain and Language. 2006;97(1):64–79. doi: 10.1016/j.bandl.2005.07.026. [DOI] [PubMed] [Google Scholar]
  4. Bradley L, Bryant PE. Difficulties in auditory organization as a possible cause of reading backwardness. Nature. 1978;271(5647):746–747. doi: 10.1038/271746a0. [DOI] [PubMed] [Google Scholar]
  5. Bradley L, Bryant PE. Categorizing sounds and learning to read: A causal connection. Nature. 1983;301(5899):419–421. [Google Scholar]
  6. Bruck M. Persistence of dyslexics’ phonological deficits. Developmental Psychology. 1992;28(5):874–886. [Google Scholar]
  7. Burgess SR, Lonigan CJ. Bidirectional relations of phonological sensitivity and prereading abilities: Evidence from a preschool sample. Journal of Experimental Child Psychology. 1998;70(2):117–141. doi: 10.1006/jecp.1998.2450. [DOI] [PubMed] [Google Scholar]
  8. Byrne B, Delaland C, Fielding-Barnsley R, Quain P, Samuelsson S, Høien T, … Willcutt E. Longitudinal twin study of early reading development in three countries: Preliminary results. Annals of Dyslexia. 2002;52(1):47–73. [Google Scholar]
  9. Castles A, Coltheart M. Is there a causal link from phonological awareness to success in learning to read? Cognition. 2004;91(1):77–111. doi: 10.1016/s0010-0277(03)00164-1. [DOI] [PubMed] [Google Scholar]
  10. Castles A, Wilson K, Coltheart M. Early orthographic influences on phonemic awareness tasks Evidence from a preschool training study. Journal of Experimental Child Psychology. 2011;108(1):203–210. doi: 10.1016/j.jecp.2010.07.006. [DOI] [PubMed] [Google Scholar]
  11. Castro-Caldas A, Petersson K, Reis A, Stone-Elander S, Ingvar M. The illiterate brain. Brain. 1998;121(6):1053–1063. doi: 10.1093/brain/121.6.1053. [DOI] [PubMed] [Google Scholar]
  12. Cheung H. Nonword span as a unique predictor of second-language vocabulary language. Developmental Psychology. 1996;32(5):867–873. [Google Scholar]
  13. Chiappe P, Stringer R, Siegel LS, Stanovich KE. Why the timing deficit hypothesis does not explain reading disability in adults. Reading and Writing. 2002;15(1–2):73–107. [Google Scholar]
  14. Clay MM. The early detection of reading difficulties: A diagnostic survey with Reading Recovery procedures. Auckland, New Zealand: Heinemann; 1972. [Google Scholar]
  15. Compton DL. Modeling the relationship between growth in rapid naming speed and growth in decoding skill in first-grade children. Journal of Educational Psychology. 2003;95(2):225. [Google Scholar]
  16. de Jong PF. What discrete and serial rapid automatized naming can reveal about reading. Scientific Studies of Reading. 2011;15(4):314–337. [Google Scholar]
  17. Dehaene S, Pegado F, Braga LW, Ventura P, Nunes G, Jobert A, … Cohen L. How learning to read changes the cortical networks for vision and language. Science. 2010;330(6009):1359–1364. doi: 10.1126/science.1194140. [DOI] [PubMed] [Google Scholar]
  18. Gathercole SE, Willis CS, Emslie H, Baddeley AD. Phonological memory and vocabulary development during the early school years: A longitudinal study. Developmental Psychology. 1992;28(5):887–898. [Google Scholar]
  19. Georgiou GK, Parrila R, Cui Y, Papadopoulos TC. Why is rapid automatized naming related to reading? Journal of Experimental Child Psychology. 2013;115(1):218–225. doi: 10.1016/j.jecp.2012.10.015. [DOI] [PubMed] [Google Scholar]
  20. Hays RD, Marshall GN, Wang EYI, Sherbourne CD. Four-year cross-lagged associations between physical and mental health in the medical outcomes study. Journal of Consulting and Clinical Psychology. 1994;62(3):441–449. doi: 10.1037//0022-006x.62.3.441. [DOI] [PubMed] [Google Scholar]
  21. Hogan TP, Catts HW, Little TD. The relationship between phonological awareness and reading: Implications for the assessment of phonological awareness. Language Speech and Hearing Services in Schools. 2005;36(4):285–293. doi: 10.1044/0161-1461(2005/029). [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hulme C, Bowyer-Crane C, Carroll JM, Duff FJ, Snowling MJ. The causal role of phoneme awareness and letter-sound knowledge in learning to read: combining intervention studies with mediation analyses. Psychological Science. 2012;23(6):572–577. doi: 10.1177/0956797611435921. [DOI] [PubMed] [Google Scholar]
  23. Hulme C, Snowling M, Caravolas M, Carroll J. Phonological skills are (probably) one cause of success in learning to read: A comment on Castles and Coltheart. Scientific Studies of Reading. 2005;9(4):351–365. [Google Scholar]
  24. Korkman M, Barron-Linnankoski S, Lahti-Nuuttila P. Effects of age and duration of reading instruction on the development of phonological awareness, rapid naming, and verbal memory span. Developmental Neuropsychology. 1999;16(3):415–431. [Google Scholar]
  25. Lervåg A, Hulme C. Rapid automatized naming (RAN) taps a mechanism that places constraints on the development of early reading fluency. Psychological Science. 2009;20(8):1040–1048. doi: 10.1111/j.1467-9280.2009.02405.x. [DOI] [PubMed] [Google Scholar]
  26. Logan JAR, Schatschneider C, Wagner RK. Rapid serial naming and reading ability: The role of lexical access. Reading and Writing. 2011;24(1):1–25. doi: 10.1007/s11145-009-9199-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Masoura EV, Gathercole SE. Phonological short-term memory and foreign language learning. International Journal of Psychology. 1999;34(5):383–388. [Google Scholar]
  28. McGrath LM, Pennington BF, Shanahan MA, Santerre-Lemmon LE, Barnard HD, Willcutt EG, … Olson RK. A multiple deficit model of reading disability and attention-deficit/hyperactivity disorder: Searching for shared cognitive deficits. Journal of Child Psychology and Psychiatry. 2011;52(5):547–557. doi: 10.1111/j.1469-7610.2010.02346.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Morais J, Cary L, Alegria J, Bertelson P. Does awareness of speech as a sequence of phones arise spontaneously? Cognition. 1979;7(4):323–331. [Google Scholar]
  30. Muthén LK, Muthén BO. Mplus User’s Guide. 7. Los Angeles, CA: Muthén & Muthén; 1998–2012. [Google Scholar]
  31. Nation K, Hulme C. Learning to read changes children’s phonological skills: evidence from a latent variable longitudinal study of reading and nonword repetition. Developmental Science. 2011;14(4):649–659. doi: 10.1111/j.1467-7687.2010.01008.x. [DOI] [PubMed] [Google Scholar]
  32. Norton ES, Wolf M. Rapid automatized naming (RAN) and reading fluency: Implications for understanding and treatment of reading disabilities. In: Fiske ST, Schacter DL, Taylor SE, editors. Annual Review of Psychology. Vol. 63. 2012. pp. 427–452. [DOI] [PubMed] [Google Scholar]
  33. Olson R, Forsberg H, Wise B, Rack J. Measurement of word recognition, orthographic, and phonological skills. In: Lyon GR, editor. Frames of Reference for the Assessment of Learning Disabilities: New Views on Measurement Issues. Baltimore, MD: Paul H. Brookes Publishing; 1994. pp. 243–278. [Google Scholar]
  34. Pattamadilok C, Perre L, Ziegler JC. Beyond rhyme or reason: ERPs reveal task-specific activation of orthography on spoken language. Brain and Language. 2011;116(3):116–124. doi: 10.1016/j.bandl.2010.12.002. [DOI] [PubMed] [Google Scholar]
  35. Pennington BF, Lefly DL. Early reading development in children at family risk for dyslexia. Child Development. 2001;72(3):816–833. doi: 10.1111/1467-8624.00317. [DOI] [PubMed] [Google Scholar]
  36. Pennington BF, Santerre–Lemmon L, Rosenberg J, MacDonald B, Boada R, Friend A, … Willcutt EG. Individual prediction of dyslexia by single versus multiple deficit models. Journal of Abnormal Psychology. 2012;121(1):212. doi: 10.1037/a0025823. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Peterson RL, Pennington BF. Seminar: Developmental dyslexia. The Lancet. 2012;379:1997–2007. doi: 10.1016/S0140-6736(12)60198-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Petersson KM, Reis A, Askelöf S, Castro-Caldas A, Ingvar M. Language processing modulated by literacy: A network analysis of verbal repetition in literate and illiterate subjects. Journal of Cognitive Neuroscience. 2000;12(3):364–382. doi: 10.1162/089892900562147. [DOI] [PubMed] [Google Scholar]
  39. Protopapas A, Altani A, Georgiou GK. Development of serial processing in reading and rapid naming. Journal of Experimental Child Psychology. 2013;116(4):914–929. doi: 10.1016/j.jecp.2013.08.004. [DOI] [PubMed] [Google Scholar]
  40. Ramus F, Szenkovits G. What phonological deficit? Quarterly Journal of Experimental Psychology. 2008;61(1):129–141. doi: 10.1080/17470210701508822. [DOI] [PubMed] [Google Scholar]
  41. Rastle K, McCormick SF, Bayliss L, Davis CJ. Orthography influences the perception and production of speech. Journal of Experimental Psychology-Learning Memory and Cognition. 2011;37(6):1588–1594. doi: 10.1037/a0024833. [DOI] [PubMed] [Google Scholar]
  42. Rutter M, Caspi A, Fergusson D, Horwood LJ, Goodman R, Maughan B, … Carroll J. Sex differences in developmental reading disability: new findings from 4 epidemiological studies. JAMA: Journal of the American Medical Association. 2004;291(16):2007–2012. doi: 10.1001/jama.291.16.2007. [DOI] [PubMed] [Google Scholar]
  43. Samuelsson S, Byrne B, Quain P, Wadsworth S, Corley R, DeFries JC, … Olson R. Environmental and genetic influences on prereading skills in Australia, Scandinavia, and the United States. Journal of Educational Psychology. 2005;97:705–722. [Google Scholar]
  44. Snowling MJ, Gallagher A, Frith U. Family risk of dyslexia is continuous: Individual differences in the precursors of reading skill. Child Development. 2003;74(2):358–373. doi: 10.1111/1467-8624.7402003. [DOI] [PubMed] [Google Scholar]
  45. Torgesen JK, Wagner RK, Rashotte CA. A Test of Word Reading Efficiency (TOWRE) Austin, TX: Pro-ED; 1999. [Google Scholar]
  46. Verhagen W, Aarnoutse C, Van Leeuwe J. Phonological awareness and naming speed in the prediction of Dutch children’s word recognition. Scientific Studies of Reading. 2008;12(4):301–324. [Google Scholar]
  47. Wagner RK, Torgesen JK, Rashotte CA. Development of reading-related phonological processing abilities: New evidence of bidirectional causality from a latent variable longitudinal study. Developmental Psychology. 1994;30(1):73–87. [Google Scholar]
  48. Wagner RK, Torgesen JK, Rashotte CA. Comprehensive Test of Phonological Processing. Austin, TX: Pro-ed; 1999. [Google Scholar]
  49. Wei W, Georgiou GK, Deng C. Examining the cross-lagged relationships between RAN and word reading in Chinese. Scientific Studies of Reading. 2015;19(6):446–455. [Google Scholar]
  50. Willcutt EG, Betjemann RS, Wadsworth SJ, Samuelsson S, Corley R, DeFries JC, … Olson RK. Preschool twin study of the relation between attention-deficit/hyperactivity disorder and prereading skills. Reading and Writing. 2007;20(1–2):103–125. [Google Scholar]
  51. Wolff U. RAN as a predictor of reading skills, and vice versa: results from a randomised reading intervention. Annals of Dyslexia. 2014;64(2):151–165. doi: 10.1007/s11881-014-0091-6. [DOI] [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
NIHMS969980-supplement-Supplemental_tables.docx (22KB, docx)
supplemental figure. Supplementary Figure A.

Structural equation model of longitudinal relations among literacy and oral language skills.

Note: Contemporaneous covariances not pictured. Pre-K = end of preschool; K = end of kindergarten; 1st = end of first grade; 4th = end of fourth grade. PA = phonological awareness composite; Print = print knowledge composite; Read = reading fluency composite; RAN = rapid automatized naming composite. Bolded paths are statistically significant.

Values are standardized path coefficients, with the top values reflecting the USA, Australia and Norway samples, and the bottom values reflecting the sample from Sweden. Factor loadings for the observed variables on the pre-Kindergarten Print Knowledge latent factor were permitted to vary for the Swedish sample. Unstandardized path coefficients among the latent factors were constrained to be equal across Sweden and the other three countries, but standardized coefficients vary slightly due to differing latent factor variances.

NIHMS969980-supplement-supplemental_figure.jpg (1.3MB, jpg)

RESOURCES