Current views of the procedures involved in spelling perhaps diverge most from the views of previous generations with respect to the necessity of a phonological component. Few people in 1988 are surprised by the claim that at least some degree of spelling competence can be demonstrated in a subject or patient with virtually no phonological skills; but this would have been an arresting statement to some previous generations whose models often viewed spelling as parasitic upon speech or at least upon some kind of phonological processing. Before one dismisses such a conception as nonsense, it is perhaps worth looking briefly at the factors that motivated it. Irrespective of which population of subjects one interrogates—normal adult spellers, normal children learning to spell, or individuals with impaired spelling—much of the evidence seems to suggest the involvement of phonology. For example, of normal adults’ spelling errors in English, whether these are apparent slips of the pen such as homophone substitutions (e.g. THERE → THEIR) or perhaps genuine misspellings (e.g. COMMITTAL → COMMITAL), the vast majority are phonologically plausible spellings (see, for example, Campbell & Butterworth, 1985; Wing & Baddeley, 1980). As emphasised by Ellis (Chapter 9), errors of 214these types constitute part of the empirical basis for the design of process models of spelling like the one in Fig. 1 (see also Figs. 1 and 4 in Ellis’s Chapter 9). That is, homophone substitutions support the notion that the 215orthographic output lexicon accepts codes or activation from the phonological output lexicon; and the high proportion of phonologically plausible misspellings suggests a contribution of general knowledge about sub-word level phonological-to-orthographic translation in everyday spelling.