Contextual Control over Lexical and Sublexical Routines When Reading English Aloud

Article excerpt

Are the processes responsible for reading aloud single well-formed letter strings under contextual control? Despite the widespread contention that the answer to this question is "yes," it has been remarkably difficult to provide a compelling demonstration to that effect. In a speeded naming experiment, skilled readers read aloud exception words (such as PINT) that are atypical in terms of their spelling sound correspondences and nonwords (such as FLAD) that appeared in a predictable sequence. Subjects took longer to name both words and nonwords when the item on the preceding trial was from the other lexical category, relative to when the preceding item was from the same lexical category. This finding is consistent with the relative contributions of lexical and sublexical knowledge being controlled. We note a number of different ways that this control could arise and suggest some directions for future research.

Learning how to translate print into a phonological code is, for most children, critical for learning how to read. Many theorists think that this ability is also functionally important for skilled readers (see Frost's 1998 review). One major account of this skill is provided by the dual-route cascaded (DRC) model (Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001). The DRC model has two ways of converting print to sound without semantic involvement. One relies on sublexical spelling-sound correspondence rules (the nonlexical route), and the other relies on whole-word associations between orthography and phonology (the lexical route).

The dual-route architecture is central to the DRC model's account of reading aloud because neither route alone can lead to the correct reading aloud of all types of stimuli. The nonlexical route generates a correct pronunciation for regular words (HINT) and nonwords (FINT) on the basis of spelling-sound correspondence rules but not for items that violate these rules. Instead, it regularizes them (e.g., exception words like PINT are read so as to rhyme with regular words like MINT). The lexical route generates a correct pronunciation for regular words and exception words on the basis of orthographic and phonological whole-word representations. This route cannot name nonwords correctly, because they are not represented in the lexicons. Critically though, both routes for translating print into sound are active, irrespective of the stimulus being named.

One question that arises naturally in the context of a theory with multiple procedures for converting print into sound is whether any of these processes can be contextually or strategically controlled. There have been a number of attempts to demonstrate such control when single words and nonwords are read aloud under blocked and mixed conditions (e.g., Baluch & Besner, 1991; Rastle& Coltheart, 1999; Tabossi & Laghi, 1992; Zevin & Balota, 2000). All of these authors reported reaction time (RT) data consistent with the interpretation that control can be exerted over whether the lexical or nonlexical routine will dominate performance. However, the same data have also been argued to be consistent with subjects changing their time criterion for responding. Indeed, this time criterion account, which does not appeal to a shift in the relative dominance of one processing routine over the other, may provide a better explanation of these RT data1 (Chateau & Lupker, 2003; Kinoshita & Lupker, 2002, 2003; Lupker, Brown, & Colombo, 1997; Taylor & Lupker, 2001; see also Raman, Baluch, & Besner, 2004). The present work therefore adopts a different approach in an effort to demonstrate contextual control over basic reading processes as reflected in naming time. This approach is a variant of the task-switching procedure developed by Rogers and Monsell (1995) and subsequently adapted by Shafiullah and Monsell (1999) to study switch costs in naming time when the type of script is changed.

Task Switching

Rogers and Monsell (1995) found that subjects took longer to respond on trials where they were switching between two tasks alternating in a predictable AABB sequence despite the stimuli affording both tasks (e. …