“…When the model is drawn this way, it makes sense, visually, to access the orthographic output buffer through the phonological output buffer, and it is visually displeasing to go directly from phonological input buffer to phonological output buffer (cutting through other arrows or sketching a long bypass). And indeed papers that assume writing non-words via the phonological output buffer portray the model similar to Figure 2a, with phonology on the left and orthography on the right (see Caramazza et al, 1987;Ellis & Young, 1996;Hillis & Caramazza, 1987;Kay et al, 1996;Laiacona et al, 2009;Morton, 1980;Patterson & Shewell, 1987). Models that assume a direct phonological input to orthographic output, on the other hand, use a different style of graphic presentationeither they put the reading route on the right, like in Figure 2b (Rapp, 2002), or they do not represent reading and phonological output at all in the model (Rapp, 2005;.…”