Rapid automatized naming (RAN; Denckla & Rudel, 1976) tasks are consistent predictors of fluency that also discriminate between dyslexic and nondyslexic reading groups. The component processes of RAN that are responsible for its relationship with reading ability remain underspecified, however. We report a study on dyslexic and nondyslexic adult groups that experimentally manipulated RAN formats to elucidate how different components of RAN differentially influence dyslexic and nondyslexic performance. The dyslexic group showed a pervasive deficit in rapid access of individually presented items. Additionally, they showed a significant impairment when multiple items were presented, whereas nondyslexic readers showed marginal facilitation for this format. We discuss the implications of these findings with respect to reading-group differences in reading fluency.
An established core deficit of developmental dyslexia is difficulty in manipulating phonemes within words (see Snowling, 2000); however, Wolf and Bowers (1999) proposed an additional core fluency deficit that is independent of phonological impairment. Rapid automatized naming (RAN; Denckla & Rudel, 1976) tasks are a well-known independent predictor of reading fluency (see, e.g., Lervåg & Hulme, in press; Manis, Doi, & Bhadha, 2000; Young & Bowers, 1995). RAN involves visually presented arrays of high-frequency items (letters, digits, colors, or objects) that are repeated multiple times in a randomized order, typically five items, each repeated 10 times across five rows. The participant names all the stimuli, from left to right across the page, as quickly as possible.
RAN performance consistently discriminates dyslexic from nondyslexic readers (see, e.g., Denckla & Rudel, 1976; see Wolf & Bowers, 1999, for a review). In orthographically shallow languages, it also contributes more variance than phonological decoding to reading ability (Wimmer, Mayringer, & Landerl, 2000). Naming rates for alphanumeric stimuli in particular remain strong predictors of reading ability (Wolf & Obregón, 1992) that persist into adulthood (Shaywitz & Shaywitz, 2005). As a measure of reading fluency, therefore, RAN has strong potential to explain differences in reading ability, and at least 104 studies published since 1990 have used RAN as a measure of reading skill.1 Despite this body of research, the underlying component processes that determine RAN's relationship with reading ability remain poorly specified (Georgiou, Parrila, & Kirby, 2006).
The present article reports a study that isolated component processes of RAN to elucidate which aspects of the task discriminate between dyslexic and nondyslexic readers' naming speeds. By extension, we examined the low-level (graphemic and phonological) processing that is involved in reading fluency.
Components of RAN
Performance on RAN is often assumed to reflect "retrieval of phonological codes from a long-term store" (Wagner, Torgesen, Laughon, Simmons, & Rashotte, 1993, p. 84), and impaired RAN performance is often assumed to reflect phonological deficits (Clarke, Hulme, & Snowling, 2005). A number of studies have also shown significant shared variance between RAN and phonologicalawareness constructs (e.g., Savage et al., 2005; Torgesen, Wagner, Rashotte, Burgess, & Hecht, 1997). Savage et al., for example, found that RAN, phonological awareness, and motor balance loaded onto one component.
Other evidence, however, suggests that RAN makes a unique contribution to reading fluency when phonological skill is factored out (see, e.g., Powell, Stainthorp, Stuart, Garwood, & Quinlan, 2007). Research has also suggested that multi-item processing is key to understanding RAN's relationship with reading fluency. Studies have demonstrated a stronger relationship between reading fluency and continuous (multiple, matrix presentation of items) versions of RAN than that between reading fluency and discrete (individual letter presentation) versions (Bowers & Swanson, 1991; Walsh, Price, & Gillingham, 1988). …