RUNNING HEAD: RAPID NAMING AND DYSLEXIA 1
Dyslexia and fluency:
Parafoveal and foveal influences on rapid automatized naming.
Manon W. Jones
Bangor University
Jane Ashby
Central Michigan University
Holly P. Branigan
University of Edinburgh
Authors’ Note
Correspondence about the paper can be directed to M.W. Jones, School of Psychology, Adeilad Brigantia, Penrallt Road, Gwynedd LL57 2AS, United Kingdom, email: .
This research was funded in part by grants from the Economic and Social Research Council (RA-000-23-3533). We would like to thank Chuck Clifton for his support during several phases of this project and Jeff Kinsey for developing the software used in this study.
KEYWORDS: Dyslexia; fluency; rapid automatized naming; foveal; parafoveal; linear mixed effects; eye movements.
Word count: 9052 (Abstract, main text body and references)
ABSTRACT
Fluent reading requires the ability to coordinate serial processing of multiple items, an ability known to be impaired in dyslexia. Using a serial naming task that has been shown to index reading fluency, we investigated which aspects of rapid serial naming are impaired in dyslexia. In two display change experiments, we recorded eye movements and voice onsets as adult dyslexic and non-dyslexic readers named letters in an array that included letter pairs which were orthographically and phonologically confusable (similar). Confusable information was presented parafoveally (Experiment 1a) and foveally (Experiment 1b) in the second letter of each confusable pair. Linear mixed effects analyses showed that orthographic and phonological similarity slowed the processing of dyslexic readers more than non-dyslexic readers. Orthographic effects arose when orthographically confusable letters were presented in the parafovea, whereas phonological effects arose when phonologically confusable letters appeared in the fovea. We discuss how these findings contribute to our understanding of fluency impairments in high-functioning, dyslexic adults.
TITLE: Dyslexia and fluency: Parafoveal and foveal influences on rapid automatized naming
A key characteristic of skilled reading is the ability to read fluently, and slow, effortful reading is often the only remaining indicator of developmental dyslexia in high-functioning dyslexic adults, who are reading at the college level (Shaywitz & Shaywitz, 2008). Poor reading fluency in adults and children can be predicted by performance on a rapid automatized naming task (RAN; Denckla & Rudel, 1976), which involves the serial naming of letters, digits, objects or colours arranged in a 50 item array. This apparently simple task is nevertheless problematic for dyslexic readers, who have consistently slower naming times than unimpaired, non-dyslexic readers (e.g., Denckla & Rudel, 1976; see Wolf & Bowers, 1999, for a review). Thus RAN-type assessments are used to identify children who are likely to be slow readers.
Researchers have used RAN-type tasks to examine the cognitive processes that support reading fluency in non-dyslexic readers and impair fluency in dyslexic readers (e.g., Jones, Branigan, Hatzidaki, & Obregon, 2010; Jones, Obregon, Kelly, & Branigan, 2008; Lervåg & Hulme, 2009; Parilla, Kirby, & McQuarrie, 2004; Powell, Stainthorp, Stuart, Garwood, & Quinlan, 2007). The present study offers an initial investigation into how parafoveal and foveal processes operate online during rapid serial naming. We monitored the eye movements of dyslexic and non-dyslexic adult readers as they completed a RAN-type task. In order to disentangle the influences of parafoveal and foveal information in serial naming speed, letter arrays were presented using a display change paradigm (Rayner, 1975) that controlled whether potentially confusable information appeared parafoveally or foveally (see Figure 1). As confusable information was available exclusively in either parafoveal view or foveal view, this novel approach allowed an examination of the distinct contributions of each stream of information to serial naming speed, and potentially to reading fluency. Our data offer initial evidence that fundamental parafoveal and foveal processes operate differently in dyslexics than in typical readers.
Rapid serial naming appears to tap a microcosm of the processes underpinning reading fluency, including: attention, feature detection, the activation of orthographic representations, the integration of visual and phonological information, and motor activation leading to articulation (Wolf & Bowers, 1999; Misra, Katzir, Wolf, & Poldrack, 2004). It is therefore important to examine the various processing requirements of rapid serial naming in order to discover why dyslexic readers are poor performers, as a step in understanding basic impairments in reading fluency separated from word and sentence level influences.
One consistent finding emerging from several studies is that RAN-type tasks are most effective at discriminating good and poor readers’ performance when the stimuli (e.g., letters) are presented simultaneously in an array rather than as discrete, individual letters (e.g., Bowers & Swanson, 1991; Jones, Branigan, & Kelly, 2009). Moreover, evidence shows that when stimuli are presented simultaneously, naming speeds for individual stimuli are influenced by information associated with adjacent letters in the array (Jones et al., 2008). These findings suggest that one key determinant of fluency may be the way in which the reader manages to process parafoveal and foveal information in contexts where more than one stimulus is present.
Serial naming and fluency. When we read words aloud we largely concentrate our cognitive resources on the task of identifying the word we are looking at (the ‘target’ item), which is then articulated. However, even before we have initiated articulation of the target item, we have moved our eyes to fixate on the next item to be processed. This is the case in reading (e.g., Laubrock, Engbert, & Kliegl, 2005) and in object naming (Morgan, van Elswijk, & Meyer, 2008).
It is currently unclear to what extent, and how, processing the ‘next’ item to the right in a series of words and objects overlaps with the processing of the target item. In the reading literature, researchers debate whether processing of consecutive words is serial or parallel (see Engbert, Nuthmann, Richter, & Kliegl, 2005, and Reichle, Rayner, & Pollatsek, 2003, for reviews). In the language production literature, Meyer and colleagues have shown in a series of object-naming studies that information from adjacent objects can influence target item viewing and naming times, suggesting parallel phonological processing of object names (e.g., Morgan & Meyer, 2005; Morgan, van Elswijk, & Meyer, 2008; Malpass & Meyer, 2010). In letter naming tasks, when adjacent items in an array are orthographically or phonologically confusable, this confusability lengthens fixation time and naming latency for the target item (Jones et al., 2008; see also Compton, 2003, for further evidence of reading group differences on ‘visual’ and ‘phonological’ versions of RAN). Participant responses are slower when adjacent letters are similar orthographically (e.g., p vs q) or phonologically (e.g., k vs q). Crucially for our current purposes, these studies found that dyslexic readers naming times are slowed significantly more than non-dyslexics when the adjacent items in the array are orthographically and phonologically confusable. These findings and others (e.g., Powell et al., 2007) suggest that dysfluency in RAN reflects a more complex problem than just retrieval of phonological codes (Wagner, Torgesen, Laughon, Simmons, & Rashotte, 1993).
However, we have yet to resolve how the orthographic and phonological information in two adjacent items intersect with each other and influence processing and naming times. Do such effects occur during the initial, parafoveal processing of the neighbouring item, or during later processing when the neighbouring item is fixated? Further, are the influences of orthographic and phonological processing dissociable, such that each exerts an independent influence on fluency but on different aspects of item processing? Or do orthographic and phonological information influence target item processing and naming similarly in each case? Answering these questions will help us ascertain how item processing during serial naming influences reading fluency, and precisely why serial naming is difficult for dyslexic readers.
We conducted two experiments to examine the extent of processing overlap between orthographic and phonological processing of two adjacent letter items. The second item in each pair was manipulated so that it was either confusable or non-confusable with the first item in the pair in a RAN-type serial naming task. In particular, we were interested in pinpointing the factors that result in the impaired performance of dyslexic readers as compared to their non-dyslexic peers. Experiment 1a examined orthographic and phonological influences in parafoveal processing, in which the second item was parafoveally confusable with the first item; whereas Experiment 1b examined orthographic and phonological influences in foveal processing, in which the second item was foveally confusable with the first item. Foveal and parafoveal vision are defined with respect to the focus of attention. Foveal vision extends out approximately 1 degree in either direction (2 degrees in total) from the centre of vision, and is the only region in the visual field allowing 100% acuity. Parafoveal vision extends from about 1 degree out from the center of vision to around 5 degrees, and displays somewhat reduced acuity (Rayner, 1998).
Foveal processing occurs during fixation, and it plays a crucial role in object naming and silent reading. Eye movement studies of silent reading initially identified the importance of foveal processing. For example, readers were much slower to read text in which the foveal information was unavailable than when it was available (Rayner & Bertera, 1979). The foveal information available during fixation supports word recognition and, thereby, affects text comprehension (Rayner, 1998; 2009). Production studies indicate that foveal processing is essential in naming as well. Speakers rarely name an item without first fixating it (Griffin & Bock, 2000; Jones et al., 2008; Meyer, van der Meulen, & Brooks, 2000). Furthermore, single object naming experiments suggest that word selection and phonological encoding processes operate foveally. Speakers gaze longer at objects with relatively inaccessible names (e.g., flute), than they do when producing more accessible object names (e.g., arm) (Griffin & Bock, 2000; Meyer et al., 1998; Meyer & van den Meulen, 2000). Gaze length also reflects assembly of phonological codes in fluent speech that involves multiple referents. Speakers gaze for longer at lower frequency and lower codability items, even when the critical item occurs in the middle of a description (Griffin, 2001).
While the eyes are fixated on one object or word, viewers automatically begin parafoveally processing information at the next location they will fixate. Multiple-object naming studies demonstrate that participants activate phonological information associated with an object that they have not yet fixated but that they will name next (Morgan & Meyer, 2005; Morgan et al., 2008; Malpass & Meyer, 2010). Eye movement data collected during silent reading indicate that readers extract linguistic information from parafoveal stimuli within 140 ms during the pre-target fixation (Inhoff, Eiter, & Radach, 2005; Sereno & Rayner, 2000). Readers use coarse-grained parafoveal information, such as word length, to direct eye movements during reading (e.g., Jones, Kelly, & Corley, 2007; Rayner, 1998). Skilled readers use parafoveal phonological information, such as initial syllable information, to facilitate word recognition during silent reading (Ashby, 2006; Ashby & Rayner, 2004). Other phonological information, such as the number of syllables, helps to control where the eyes will fixate next during reading (Fitzsimmons & Drieghe, 2011; Ashby & Clifton, 2005).
Previous research suggests a relationship between reading skill and parafoveal processing. Whereas skilled readers benefit from phonologically similar parafoveal previews (Pollatsek, Lesch, Morris, & Rayner, 1992; Ashby, Treiman, Kessler & Rayner, 2006), poor readers do not show phonological preview benefits during reading (Chace, Rayner, & Well, 2005). Letter categorisation is also more difficult for dyslexic readers, compared with non-dyslexics, when the item is flanked by other stimuli, suggesting parafoveal processing impairment in a lateral masking task (Pernet, Valdois, Celsis, & Demonet, 2006). Therefore, it may be that dyslexic readers process parafoveal information differently from non-dyslexic readers. Recent research demonstrates that compensated dyslexic readers are sensitive to parafoveal information (Jones et al., 2008), and that it may in fact act as a source of interference in their naming (Jones et al., 2009).
To further examine the influences of parafoveal and foveal information on serial naming in dyslexic readers, we conducted two eye movement experiments that presented 40 letters in a RAN-like array. We used these multi-letter displays to pinpoint the specific processes that contribute to slow serial naming in dyslexic readers, and thus to identify the possible processes that contribute to reading fluency impairments (or slow reading in dyslexics). An eye–contingent, display change paradigm (Rayner, 1975) controlled when the confusable information appeared (parafoveally or foveally). To our knowledge, this is the first serial naming study to utilize display changes in letter arrays in order to examine the parafoveal and foveal processes that occur during rapid serial naming. This methodological development is important for studying a ‘continuous’ naming paradigm, and it allows the generalization of our results to offline assessments, such as the RAN (Denckla & Rudel, 1976) or Rapid Letter Naming (Wagner, Torgesen, & Rashotte, 1999), that have established slow serial naming speed as an indicator of reading difficulties (Schatschneider, Fletcher, Francis, Carlson, & Foorman, 2004; Wimmer & Mayringer, 2002; Wolf & Bowers, 1999).
Experiment 1a: How does parafoveal information influence serial naming speed?
Experiment 1a examined whether information gained exclusively from parafoveal vision (i.e., early processing) influences processing times and/or the onset of articulation times in a RAN task. We compared the performance of age-matched, high-functioning adult groups of non-dyslexic and dyslexic readers on a RAN-letters task that manipulated parafoveal information. Specifically, we manipulated whether the parafoveal information available to readers in position n+1 was orthographically or phonologically confusable with the information presented in position n, and measured whether confusability affected participants’ eye movements compared to conditions in which the parafoveal information was neither orthographically nor phonologically confusable.
The design of this experiment was in many respects similar to the design of Jones et al. (2008): Letters were presented in an array of four lines containing ten items each, and confusability was manipulated by varying the orthographic similarity of adjacent pairs of letter shapes (e.g., p-q; b-d) and the phonological similarity of the onset in the letter names (e.g., q-k; g-j). However, unlike in Jones et al. (2008), we used a contingent change paradigm, so that confusable information was presented only parafoveally: When the participant actually fixated the location where the confusable information had appeared, the confusable letter was replaced by a non-confusable letter. For example, in the orthographically confusable condition, when a participant fixated the target item q (in position n), the orthographically confusable letter p was viewed in the parafovea (in position n+1). When the eyes saccaded across an invisible boundary to the immediate right of the target item in position n, the confusable letter p (or non-confusable letter k) in position n+1 was replaced by a non-confusable target item, such as f (see Figure 1).