According to dual-route reading models, in the lexical route pronunciation is retrieved as a whole from the lexicon, whereas in the perilexical route, it is constructed via phonological decoding of syllabic segments. The study assessed perilexical reading aloud of 270 isolated non-words by 44 college students. A computer analyzed error frequency and temporal parameters of pronunciation: reaction time (RT), duration, and segmentation frequency, as function of non-word characteristics. Results showed that: 1) the shortest and simplest syllables produced the highest reading speed and segmentation frequency; 2) the longest non-words produced the highest segmentation frequency, but the lowest reading speed; 3) digraphs increased error frequency, as well as duration and RT, but decreased segmentation frequency; 4) digraphs (two letters: one phoneme) increased RT only when they occurred in the initial syllables; 5) digraphs produced greater RT than did consonantal combinations (two letters: two phonemes), which in turn produced greater RT than did simple graphemes. Such findings increase the validity of the dual-process reading model, as well as the understanding of serial and parallel processes involved in perilexical processing.
Reading aloud; non-word reading; segmentation frequency; reaction time; duration