The ability to read words aloud requires the coordination of the cognitive system responsible for visual word recognition and that responsible for speech production. The most prominent theory of reading – the dual-route cascaded (DRC) model (Coltheart et al., 2001) – proposes that this is accomplished by series of processes that start with initial orthographic analysis of the letter identities which are then processed by both the lexical and sublexical route to generate associated pronunciations which then converge on a level of representation that has been argued to correspond to the post-lexical phonology in theories of spoken production (Goldrick & Rapp, 2007). For most words, the lexical and the sublexical routes converge a common phonological representation, but for words with irregular spelling to sound correspondence, the lexical and sublexical routes activate competing phonological representations. For example, according to this theory, when the word BREAD is read, lexical processing activates the sequence of phonemes /brɛd/ because lexical processing is the process of retrieving whole word level phonological representations, while sublexical processing activates the sequence of phonemes /brid/ because sublexical processing requires orthographic mapping where individual graphemes are mapped onto their most frequent phonemic representation. Competition at this level between the phonological representations generated by the lexical and sublexical routes can explain why readers are slower and more error prone with these irregular inconsistent forms. A further assumption of this theory is that activation cascades through the system, meaning that processing at each level is not fully complete before activation is passed from one level of representation to the next. As a result of this assumption, this theory predicts that the competition between the lexical and sublexical routes at the phonological level should cascade down to the articulatory level. This Master’s thesis examines this prediction in a series of reading experiments with both individuals with acquired dyslexia and neurotypical readers. Previous speech production studies have shown that there are phonetic traces of the phonemes activated, but not selected, during production (Goldrick & Blumstein, 2006). Using this logic, we investigated whether there are phonetic traces of the phonological representations generated by the sublexical route, even when participants correctly read words aloud using the pronunciation generated by the lexical route. However, in all three experiments there is no evidence that irregular words generate multiple phonological plans that cascade to the articulatory planning level. The consequence of this null result on the DRC model is discussed.