Frost's review inspires reading researchers to take seriously the diversity in languages and writing systems in which reading occurs. Appropriately, Frost focuses on adult visual word recognition research, where the debates have been most heated as to the flexibility of letter-order processing. Frost alludes to the developmental implications of his proposal. I review here three relevant dimensions from, and with implications for, development research that will expand this picture – moving us towards a universal model of reading development.

First, recent developmental research provides a direct test of Frost's proposal that, “for a model to produce different behavior as a function of the statistical properties of the language, the model has to be able to pick up the statistical properties of the language” (sect. 5, para. 5, italics in the original). Learning about letter-order regularities would be considered, in child reading research, within the construct of orthographic processing, or the “ability to form, store, and access orthographic representations” (Stanovich & West Reference Stanovich and West1989, pp. 404). Frost's statement suggests that children learn about orthographic processing through their reading. This is also the prediction of several prominent models of reading development (e.g., Ehri Reference Ehri2005; Share Reference Share1999). There is, however, a competing proposal from empirical research: that children's orthographic processing determines their progress in learning to read (e.g., Wagner & Barker Reference Wagner, Barker and Berninger1994; for a review, see Burt Reference Burt2006). Perhaps surprisingly, there has only been one empirical test to date of the direction of the relationship between orthographic processing and reading. In a longitudinal study of English children between grades 1 and 3 including the most common measures of orthographic processing, Deacon et al. (Reference Deacon, Benere and Castles2012) have demonstrated that children's reading determines their growth in orthographic processing rather than the other way around. This is the key advantage of developmental research: It can tell us about which skills drive reading acquisition and which are the product of reading.

This developmental approach provides empirical confirmation of Frost's suggestion that children learn about orthographic regularities through their reading experience. Considering this cross-linguistically, Abu-Rabia (Reference Abu-Rabia1997) found that orthographic processing did not correlate with reading for Hebrew-learning children, even at the same point in time; this finding diverges considerably from similar studies done with children learning English, where such relationships abound (e.g., Cunningham et al. Reference Cunningham, Perry and Stanovich2001). Abu-Rabia's finding makes sense if one considers the relative paucity of the orthographic regularities to be learned in Hebrew, where, as Frost points out, transitional probabilities are far less informative as to word identity. Clearly, cross-linguistic extensions of these developmental studies will be informative in identifying the language-specific and universal mechanisms that underlie learning to read.

Second, the concern that Frost raises about a hyper-focus on orthographic processing needs to be brought to bear on models of reading development. The two most prominent models propose that early reading acquisition is driven by phonological processes with a shift to orthographic processing, either in a later phase (Ehri Reference Ehri2005) or with specific items (Share Reference Share1995). Ehri's (Reference Ehri2005) phase theory of reading development considers children's learning of letter patterns to be of both syllables and morphemes. According to this approach, all frequent letter patterns are processed similarly, regardless of whether these have a basis in meaning (e.g., -ion versus -ight, respectively). Share's (Reference Share1995) self-teaching hypothesis is effectively a model of orthographic learning; through phonological decoding, children learn word-specific representations. Morphological effects are conceptualised as indexing reading efficiency. It is appropriate that both models include orthographic processing; children have a great deal to learn about the orthographic patterns in a writing system and the orthographic representations of individual words. And yet an over-emphasis on orthographic processing has led the vast majority of developmental research to consider only two facets of the linguistic context. The majority focuses on phonological and orthographic processing (e.g., Cunningham et al. Reference Cunningham, Perry and Stanovich2001). Even the extension to morphological processing in more recent research tends to focus on only two dimensions, in this case phonological and morphological processing (e.g., Carlisle Reference Carlisle2000; Deacon & Kirby Reference Deacon and Kirby2004).

Recent developmental studies have begun to take seriously the full context suggested by Frost. Orthographic processing and morphological awareness have been shown to make distinctive contributions to reading outcomes (from each other and from phonological awareness) both in younger (Deacon Reference Deacon, Benere and Castles2012) and older child readers (Roman et al. Reference Roman, Kirby, Parrila, Wade-Woolley and Deacon2009; see also, Garcia et al. Reference Garcia, Abbott and Berninger2010; Tong & McBride-Chang Reference Tong and McBride-Chang2010). Children are clearly attending to more than orthographic dimensions in the process of learning to read, and so should models of reading development.

The final critical step is to bring together a full linguistic and writing system context with rich cross-linguistic comparisons. All discussions, both developmental and adult, need to be framed in the context of the relative weighting that each language places on these linguistic dimensions, a point duly taken from Frost. As we have already reviewed, orthographic processing does not appear to be predictive of reading outcomes in Hebrew (Abu-Rabia Reference Abu-Rabia1997). Clearly, the role of orthographic processing might depend on its place in the writing system in which children are learning to read. Similarly, turning to morphology, McBride-Chang et al. (Reference McBride-Chang, Cho, Liu, Wagner, Shu, Zhou, Cheuk and Muse2005) demonstrated that morphological awareness was more strongly associated with children's reading outcomes in Chinese and Korean than in English. Both Chinese and Korean represent morphology in the writing system to a much greater extent than does English, and so these data demonstrate the clear need to consider the language context in which reading occurs. The next key move is to integrate these developmental and cross-linguistic comparisons with the full linguistic context of learning (e.g., Share Reference Share2008a).

I share Frost's quest for a universal model of reading; I would put forward that it must be a universal model of reading development. It needs to be universal in considering the phonological, orthographic, and morphological dimensions of the context in which the reading occurs. It needs to be developmental in evaluating the factors that underlie children's reading acquisition, as well as those that continue to inform adult word recognition. Reading researchers across all orthographies need to keep this laudable goal in their sights.