Academic journal article Attention, Perception and Psychophysics

Is Attention Confined to One Word at a Time? the Spatial Distribution of Parafoveal Preview Benefits during Reading

Academic journal article Attention, Perception and Psychophysics

Is Attention Confined to One Word at a Time? the Spatial Distribution of Parafoveal Preview Benefits during Reading

Article excerpt

Eye movements were recorded while participants read declarative sentences. Each sentence contained a critical three-word sequence with a three-letter target word (n), a spatially adjacent post-target word (n+1), and a subsequent nonadjacent post-target word (n+2). The parafoveal previews of words n and n+2 were manipulated so that they were either fully visible or masked until they were fixated. The results revealed longer word n and word n+1 viewing durations when word n had been masked in the parafovea, and this occurred irrespective of whether the target was skipped or fixated. Furthermore, masking of word n diminished the usefulness of the preview of word n+2. These results indicate that the effect of a parafoveally available target preview was not strictly localized. Instead, it influenced target viewing and the viewing of the two subsequent words in the text. These results are difficult to reconcile with the assumption that attention is confined to one word at a time until that word is recognized and that attention is then shifted from the recognized word to the next.

Readers must select to-be-processed linguistic information from an ordered sequence of concurrently available visual language symbols, and the selection of tobe- recognized words for processing is accompanied by the execution of saccadic eye-movements and by fixations during which useful linguistic information is obtained (see Rayner, 1998, for a detailed review). There is a general consensus that readers regularly obtain useful linguistic information from more than one word during a fixation, typically from the fixated word and from the next parafoveally visible word(s) in the text. Furthermore, linguistic information that is obtained from the parafovea is integrated with information that is acquired during the following fixation, thereby contributing to the fluency of skilled reading. These findings have contributed to the recent development of several types of computational models that provide formal accounts of the coordination of eye-movement programming with text processing (Engbert, Nuthmann, Richter, & Kliegl, 2005; Feng, 2006; McDonald, Carpenter, & Shillcock, 2005; Reichle, Pollatsek, Fisher, & Rayner, 1998; Reilly & Radach, 2006; Yang, 2006). At the core of these models are assumptions concerning the selection of text segments for linguistic processing and the integration of linguistic information across fixations.

Morrison (1984) proposed a sequential-attention-shift (SAS) assumption according to which readers select a single word for processing and then, after lexical processing has been completed (or has reached a stage near completion), shift attention to the next word in the text. Selection is thus focused at an individual word at each point in time. Saccade programming is assumed to be a relatively timeconsuming process that ensues when the word recognition process has reached threshold, whereas attention shifting is assumed to be instantaneous; therefore, useful information can be acquired from the next word before it is fixated. Morrison's model could also account for the skipping of some words in the text. This is assumed to occur when lexical processing of the next word reaches a critical threshold before the saccade to this word is committed to action. The saccade is then cancelled and is replaced with a saccade to the following word.

Successor models, notably the family of E-Z Reader models (Pollatsek, Juhasz, Reichle, Machacek, & Rayner, 2008; Pollatsek, Reichle, & Rayner, 2006; Reichle et al., 1998; Reichle, Pollatsek, & Rayner, 2006; Reichle, Rayner, & Pollatsek, 2003) and EMMA (Salvucci, 2001), have refined SAS assumptions. Specifically, the E-Z Reader model postulates two stages of word recognition (L1 and L2) rather than a single stage, and it assumes that saccade programming and attention shifting are functionally decoupled; that is, the programming of a saccade to the next word is initiated after an attended word's L1 processing is completed, and a corresponding shift of attention occurs when its L2 processing has been completed. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.