ABSTRACT
Deep dyslexic patients, when confronted with a printed non-word to pronounce, will generally choose to make no response. The logogen model, when confronted with the question of how readers assign pronunciation to unfamiliar words or non-words, has generally chosen to make no response. The logogen model's primary focus was on recognition of familiar written words by fluent readers; and precious little unambiguous evidence suggested that the procedures used to assemble a pronunciation for a non-word might be significantly implicated in skilled word recognition. By non-lexical, we mean that no reference would be made, at the time of assembling the phonological code, to correspondences between orthography and phonology in specific known words. A non-lexical rule-governed routine and a lexical analogy-based routine do not, presumably, exhaust the theoretical possibilities, but these two are the major bids on the table at present.
