The target article accuses researchers in orthographic processing of inadvisedly narrowing the scope of investigation in visual word recognition (VWR). However, the article actually reflects the narrowness of the author's own outlook, rather than the existence of any constrictions on VWR research.

The article makes the obvious point that VWR is not limited to orthographic processing, but must include phonological, morphological, and semantic analysis. None of us who investigate orthographic processing would disagree. The current attention being paid to the topic of letter-position encoding simply reflects the fact that this aspect of VWR has been neglected in the past; we have now successfully pointed out the interesting and important questions associated with this topic.

Frost characterizes research in orthographic processing as focusing on transposed-letter effects, and points to the lack of transposed-letter priming for Hebrew roots as evidence that our research does not address universal questions in VWR. However, his article is inaccurate on both these counts. Research on orthographic processing attempts to answer the question of how a feature-based retinotopic representation is converted into abstract representations of letter identity and order that support morphological, lexical, and phonological analysis. It employs behavioral and brain-imaging experiments evaluating the effects of retinotopic position, within-string letter position, word length, and letter insertions, deletions, and transpositions within and across phonological and morphological boundaries.

Such investigations have led some researchers to propose an open-bigram encoding for lexicosemantic access, as noted in the target article. Although the proposal of open bigrams was based on research in European languages, this type of representation happens to be particularly suited for Hebrew roots, because it encodes the order of non-contiguous letters. Under a universal open-bigram encoding, the degree of sensitivity to transposed-letter priming may simply be a function of the relative strength of inhibitory and excitatory connections between open bigrams and morphological units. For example, a strong inhibitory connection from open-bigram BA to root ABC would prevent facilitation by the prime BAC. In fact, evidence for such inhibition comes from an English study that compared the effect on the target ABCD of the reversed prime DCBA versus a control prime containing none of the target's letters (Still & Morris Reference Still and Morris2010). The reversed prime yielded inhibition with respect to the control, suggesting the existence of inhibitory input from bigrams that are reversals of the word's bigrams. The relative influence of such inhibition may vary with morpheme length, language, and reading experience. Research, not ranting, is required to resolve the issue of whether differences in transposition effects across languages reflect quantitative differences in orthographic processing (as suggested here) or qualitative differences (as claimed in the target article).

However, the study of orthographic processing encompasses much more than the question of what type of representation contacts the lexical/morphological level. It addresses lower levels of processing, asking how a retinotopic representation is converted into a location-invariant encoding, including the issue of how information is integrated across hemispheres (Grainger et al. Reference Grainger, Tydgat and Isselé2010; Tydgat & Grainger Reference Tydgat and Grainger2009; Whitney Reference Whitney2001; Reference Whitney2008; Reference Whitney2011). It asks whether orthographic representations are the same on the lexicosemantic and phonological pathways, and, if not, how and when they diverge (Grainger & Ziegler Reference Grainger and Ziegler2011; Whitney & Cornelissen Reference Whitney and Cornelissen2008). It asks how these representations are learned and constrained by the innate visual and auditory systems, and why such learning may fail (Whitney & Cornelissen Reference Whitney and Cornelissen2005).

Hence, a comprehensive model of VWR should characterize how the preliterate neural systems for phonological analysis, visual object recognition, audiovisual integration, and semantic representation adapt to support orthographic, phonological, morphological, and semantic processing in reading. It should characterize the end point of reading acquisition, and how it varies with language and individual differences. It should detail the ways in which reading acquisition can fail, potentially providing insights into how to provide more effective remediation for reading disability. These are ambitious goals; respect and cooperation among researchers in different aspects of VWR will be required to attain them. Unfortunately, the inaccurate account of orthographic research presented in the target article is not conducive to this mission.