Recent psycholinguistic investigations of ways in which reading can be impaired by brain damage have led to the identification and definition of several varieties of acquired dyslexia. For each variety, certain types of word pose particular problems for the dyslexic reader, while other types of word can be read with success. Studies of acquired dyslexia in English have identified, as significant classes of stimulus, at least the following: stimuli with many letters, nonword letter strings, irregular words, homophones, and stimuli where the mapping of letters to phonemes is not entirely one-to-one. Each of these classes of stimuli causes particular difficulties in particular varieties of acquired dyslexia. Furthermore, each poses problems for theories intended to explain normal reading; normal readers can read all of these classes of stimuli, so any theory of normal reading must include mechanisms for reading each stimulus class. The aim of this paper is to explore some consequences of the fact that these various stimulus classes are not all present in all languages. There are some alphabetically written languages which have no irregular words, some which have no homophones, and some where the mapping of letter to phoneme is invariably one-to-one. In logographically written languages, the concept "containing many letters" is inapplicable, and it appears that the concept "written nonword" is too. I discuss, therefore, the extent to which models of normal English reading, and the distinctions drawn between varieties of acquired dyslexia in English reading, are applicable to other languages.