Elsevier

Cognition

Volume 122, Issue 1, January 2012, Pages 110-117
Cognition

Brief article
Chicken or egg? Untangling the relationship between orthographic processing skill and reading accuracy

https://doi.org/10.1016/j.cognition.2011.09.003Get rights and content

Abstract

There is increasing evidence of a relationship between orthographic processing skill, or the ability to form, store and access word representations, and reading ability. Empirical research to date has not, however, clarified the direction of this relationship. We examined this question in a three-year longitudinal study of children from Grades 1 to 3. We included standard measures of orthographic processing skill, at both the lexical and sublexical level, and word reading accuracy, as well as controls of vocabulary, non-verbal reasoning, and phonological awareness. In all analyses, word reading predicted progress in acquiring orthographic processing skill, regardless of grade level or orthographic processing measure. In contrast, orthographic processing skill did not predict progress in word reading. Our results suggest that, between Grades 1 and 3, children acquire orthographic processing skill through their reading and that this ability, as characterized by the most common tasks used to date, does not play an independent role in supporting reading acquisition.

Introduction

Psychological research has contributed greatly to our understanding of the reading acquisition process by identifying factors that appear to determine reading success. The most prominent of these is phonological awareness, or the ability to access and manipulate speech sounds, which has been demonstrated to be a strong predictor of reading outcomes (National Reading Panel, 2000, Snow et al., 1998). However, a second key factor is orthographic processing skill, broadly defined as the “ability to form, store, and access orthographic representations” (Stanovich & West, 1989, p. 404). Despite accumulating evidence of a relationship between measures orthographic processing skill and word reading accuracy, we do not yet have a full picture of the nature and direction of this relationship: there have been no systematic investigations of the extent to which measures of orthographic processing skill determine reading outcomes compared to the extent to which the relationship flows in the reverse direction. The goal of this study was to explore these relationships systematically.

Such an investigation first requires consideration of the ways in which orthographic processing skill is currently measured. The first common measure evaluates orthographic processing skill at the lexical level, and it is often conceptualized in terms of the “specific visual/spelling patterns that identify individual words” (Barker, Torgesen, & Wagner, 1992, pp. 335–336). Accordingly, in the classic pseudohomophone choice task, children are asked to choose between two phonologically plausible alternatives (e.g., tertle and turtle; Olson, Forseberg, Wise, & Rack, 1994). The second measure focuses on regularities in sublexical letter-patterns that occur across a given writing system (e.g., Perfetti, 1984, Treiman and Cassar, 1997). For example, word-final consonant doublets are legal in English (such as mill), but these are typically followed by a vowel in French (as in mille or ‘thousand’). Tasks defined at this level typically request children to display knowledge of these regularities by selecting the most plausible spelling for a nonword (e.g., baff versus bbaf in English; Cassar & Treiman, 1997).

Two recent reviews have suggested that lexical-level orthographic processing tasks are very similar to the outcome measures that they are designed to predict (Burt, 2006, Castles and Nation, 2006, see also Vellutino, Scanlon, & Tanzman, 1994). The similarity in format introduces the possibility that any relationships uncovered essentially reflect the prediction of word reading by word reading. It has been postulated that sublexical measures avoid this circularity, allowing for a more direct evaluation of the process-based knowledge that might potentially be useful in reading acquisition. Including both lexical and sublexical measures of orthographic processing skill in the same study would test this possibility.

Existing cross-sectional studies have demonstrated that orthographic processing skill, as measured with established lexical and sublexical tasks, is correlated with word reading accuracy in young readers, independently of specific control variables (e.g., Barker et al., 1992, Stanovich et al., 1991; for review see Wagner & Barker, 1994). However, correlational studies do not tell us about the direction of the relationship. As Burt (2006) notes, the assumption has been that orthographic processing skill plays an “independent causal role” (p. 401) in the development of reading. This is entirely possible. It is equally possible that children might learn orthographic processing skill, as measured by existing tasks, through their reading. These non-mutually exclusive possibilities need to be tested.

Two longitudinal empirical studies have specifically explored whether orthographic processing skill determines progress in learning to read (Cunningham et al., 2001, Wagner and Barker, 1994). Both have done so using cross-lag analyses with auto-regressive controls. The logic of these analyses is that, should orthographic processing skill assessed at Time 1 predict reading ability at Time 2, after controlling for reading ability at Time 1, this can be taken as evidence that orthographic processing skill determines progress in learning to read. Cross-lag analyses are considered to be conservative and can incorporate multiple controls for spurious third variables (Kenny, 1975). They have been used across developmental research (e.g., McBride-Chang et al., 2008), including in establishing the bi-directional relationship between phonological awareness and reading (Perfetti, Beck, Bell, & Hughes, 1987).

With such analyses, Wagner and Barker (1994) showed that orthographic processing skill assessed in Grade 1 predicted word reading in Grade 2, even after controlling for verbal ability, phonological awareness and the auto-regressive effects of word reading in Grade 1. These findings certainly suggest that orthographic processing skill determines progress in learning to read. However, the measures of orthographic processing skill in this study included children’s letter-name knowledge, which has a demonstrated relationship with reading outcomes (Näslund and Schneider, 1996, Share et al., 1984), and which is typically treated separately from orthographic processing (e.g., McBride-Chang & Ho, 2005). These results, then, are likely to be confounded with the effects of another reading-related variable.

Cunningham et al. (2001) showed that orthographic processing skill in Grade 2, evaluated with a battery of tasks, predicted unique variance in word reading at Grade 3, after controlling for phonological awareness and earlier measures of reading. However, these analyses included the auto-regressive effects of pseudo-word reading, which likely evaluates primarily letter-sound correspondences, and not real word reading. Given that auto-regressive controls are the linchpin of cross-lag analyses, we do not yet know whether established measures of orthographic processing skill determine progress in learning to read words accurately: the hallmark of children’s reading development in English.

A final missing piece of the puzzle is that neither of these studies tested the relationship in the reverse direction: evaluating the effects of early word reading on improvement in performance on measures of orthographic processing. This is precisely the direction of relationship predicted by several models of reading development. Ehri (2005) suggests that children develop orthographic representations as they analyze the properties of print. Similarly, Share (1995) proposes that children’s successful word decoding allows them to develop representations of an individual word’s orthographic pattern. At the sublexical level, statistical learning theories suggest that children develop implicit knowledge of orthographic regularities as they are exposed to print (e.g., Pacton, Perruchet, Fayol, & Cleeremans, 2001). All of these theoretical approaches make the prediction that reading should determine subsequent orthographic processing skill, as measured by existing tasks, rather than the other way round.

The study that we report on here provides the first comprehensive investigation of the direction of the relationships between orthographic processing ability and reading. This study was conducted with the two most common measures of orthographic processing skill: lexical and sublexical tasks. This evaluation was carried out within a longitudinal study between Grades 1 and 3, with the direction of relationships being tested using cross-lag analyses. Specifically, we examined the influence of early measures of orthographic processing skill on later reading accuracy outcomes, after controlling for early reading abilities. We also examined the contribution of early word reading to later orthographic processing measures, after controlling for early orthographic processing ability. Our inclusion of both lexical and sublexical orthographic processing measures allowed us to evaluate whether the contributions of sublexical orthographic processing remain after controlling for lexical level abilities. Multiple control variables were incorporated to reduce the possibility that any effects could be due to a spurious third factor. As such, we obtained data uniquely suited to assessing the question of whether orthographic processing skill, as typically measured, determines progress in learning to read, whether orthographic processing skill is acquired through reading, or both.

Section snippets

Participants

One hundred and twenty-three children were recruited in the first grade. Grade 1 represents the first year of formal reading instruction and the earliest age at which children’s word reading can be reliably measured with standardized tools. Controlling for word reading at this time point provides a strict control for children’s reading and academic experiences to that point. Analyses were conducted with the children who remained in the study across all three years (100 children, including 53

Results

Table 1 displays descriptive statistics for age and all variables organized by grade. All standardized assessment measures are reported as both standard scores and raw scores. Participants scored within the average range on vocabulary, word reading, and non-verbal reasoning. Scores for the orthographic processing tasks were better than chance at each grade level: lexical (ts (99)  6.00, ps < .001) and sublexical (ts(99)  13.00, ps < .001). Reliabilities are all above .7, with the exception of lexical

Discussion

This research evaluated the direction of the relationship between orthographic processing skill, as indexed by previously-used lexical and sublexical measures, and reading ability. We found that early word reading significantly predicted later orthographic processing ability after controlling for age, vocabulary, non-verbal reasoning, phonological awareness, and earlier measures of orthographic processing skill. These results were remarkably consistent, emerging when predicting from Grade 1 to

References (34)

  • D. Kenny

    Cross-lagged panel correlation: A test for spuriousness

    Psychological Bulletin

    (1975)
  • C.J. Lee

    Evidence-based selection of word frequency lists

    Journal of Speech-Language Pathology and Audiology

    (2003)
  • C. McBride-Chang et al.

    Predictors of beginning reading in Chinese and English: A 2-year longitudinal study of Chinese kindergartners

    Scientific Studies of Reading

    (2005)
  • C. McBride-Chang et al.

    What’s in a word? Morphological awareness and vocabulary knowledge in three languages

    Applied Psycholinguistics

    (2008)
  • National Reading Panel (2000). Teaching children to read: An evidence-based assessment of the scientific literature on...
  • R. Olson et al.

    Measurement of word recognition, orthographic, and phonological skills

  • S. Pacton et al.

    Implicit learning out of the lab: The case of orthographic regularities

    Journal of Experimental Psychology: General

    (2001)
  • Cited by (68)

    • Silent word-reading fluency is strongly associated with orthotactic sensitivity among elementary school children

      2021, Journal of Experimental Child Psychology
      Citation Excerpt :

      It has been linked to reading accuracy (e.g., Conrad, Harris, & Williams, 2013; Coyne, Farrington-Flint, Underwood, & Stiller, 2012) as well as oral reading fluency (O’Brien, 2014). Furthermore, evidence suggests that reading facilitates the development of orthotactic sensitivity (Deacon, Benere, & Castles, 2012; Leslie & Shannon, 1981). During formal reading instruction, children’s attention is drawn to words frequently and intensively, which increases orthotactic sensitivity (Cunningham & Stanovich, 1993).

    • “Zooming in” on orthographic knowledge to clarify the relationship between rapid automatised naming (RAN) and word reading

      2019, Learning and Individual Differences
      Citation Excerpt :

      As noted by Apel (2011), some researchers have narrowly defined OK as the stored mental representations of written words saved in long-term memory (e.g., Bowers & Wolf, 1993; Ehri, 1980; Ehri & Wilce, 1982; Masterson, Apel, & Wasowicz, 2006; Moll & Landerl, 2009; Stanovich & West, 1989; Wolter & Apel, 2010). By comparison, an emerging consensus suggests that OK consists not only of this word-specific knowledge (variously referred to as lexical OK or mental graphemic representations [MGR]; Deacon, Benere, & Castles, 2012; Apel, 2011), but also “generic OK” (GOK) consisting of sublexical knowledge about graphemic patterns. This knowledge is not word-specific but generalised across words (e.g., which letters are allowed to follow other letters, legal letter combinations, and the frequency of letter positions [Conrad, Harris, & Williams, 2013; Vellutino, Scanlon, & Tanzman, 1994]).

    View all citing articles on Scopus

    This research was funded by a SSHRC grant to the first author.

    View full text