Reading and Writing: An Interdisciplinary Journal 16: 179–194, 2003.
© 2003 Kluwer Academic Publishers. Printed in the Netherlands.
Reading aloud: Dissociating the semantic pathway from the
non-semantic pathway of the lexical route
FRANCESCA PERESSOTTI and REMO JOB
DPSS, University of Padova, Padova, Italy
Abstract. According to dual-route models of reading, consistency effects in pseudoword
reading are evidence for the activation of lexical information. We investigated whether this
lexical interference has a semantic or a non-semantic origin. In Experiment 1, participants
named aloud a set of words and pseudowords. The consistency effect in reading pseudowords
co-occurred with associative priming effects in reading words but not with semantic priming
effects. In Experiment 2, only words were presented. Comparable effects of both associative
priming and semantic priming in naming words were found. This pattern provides evidence
for the existence of a lexical non-semantic pathway in reading aloud. It also shows that this
pathway is sensitive to associative relations among words. Finally, it calls into question the
likelihood of a feedback mechanism from the semantic system to the orthographic input
Key words: Associative priming, Reading aloud, Semantic priming
According to the Dual Route Cascaded (DRC) model of reading (Colt-
heart, Curtis, Atkins & Haller, 1993; Coltheart & Rastle, 1994; Coltheart,
Rastle, Perry, Langdon & Ziegler, 2001), a lexical and a non-lexical route
mediate phonological retrieval. The non-lexical route operates on the basis
of grapheme-to-phoneme conversion rules. The lexical route retrieves whole-
word phonological forms and comprises both a semantic and a non-semantic
The issue of two lexical pathways is highly controversial, in spite of its
relevance for claims about the functional architecture of the reading system
(see, e.g., Besner, 1999; Coltheart, 1987; Seidenberg, 1992). Most of the
empirical data on this issue come from the neuropsychological literature.
Speciﬁcally, patients presenting two patterns of selective damage have been
described, which supports the existence of two functionally distinct print-to-
sound lexical pathways. On the one hand, deep dyslexics’ performance (cf.
Coltheart, Patterson & Marshall, 1980) speaks in favor of the existence of a
lexical semantic pathway. In reading words, these patients produce semantic
errors of various types (cf. Barry, 1984), thus signaling access to a damaged