From a cognitive psychologist’s perspective, reading constitutes a high-level, complex cognitive skill that involves the coordination of linguistic, oculomotor, and working memory processes and their mutual interaction (Kliegl and Laubrock, 2017, Pan et al., 2017, Rayner et al., 2012). Understanding the interdependencies of the underlying processes forms the basis for practitioners to effectively support reading acquisition, in particular for disadvantaged children. One such process that is tightly linked to early reading success and to efficient reading in general is the preview of upcoming text during reading. An eye-movement based measure of preview is the perceptual span (Rayner, 1986, Rayner, 2009, Sperlich et al., 2015, Sperlich et al., 2016). Because preview needs to be integrated with what has already been identified, the perceptual span can be interpreted as an indicator of how attentional resources are flexibly scheduled during preprocessing of upcoming information (Meixner et al., 2022).

Here, we present new data from the fourth and fifth wave of our longitudinal study of perceptual-span and reading development (Sperlich et al., 2016). We examine the effects of linguistic, oculomotor, and executive control processes over ten years. This allows us to answer questions about perceptual-span development in general, about its importance for reading development, and about inter-individual variation in perceptual-span development.

The main focus of the present report is on how well long-term success in reading development can be predicted by inter-individual differences between beginning readers. While taking into account individual differences in initial domain-specific linguistic competencies, we are particularly interested in the predictive value of initial domain-general executive skills. In addition to guiding educational policies and intervention strategies, revealing such differences might be informative about the relationship of executive processes and the perceptual span, which has been neglected so far.

The effective visual field around the current fixation has been termed the perceptual span during reading and is measured using the moving-window technique (McConkie and Rayner, 1975, McConkie and Rayner, 1976, Rayner, 2014). Within a software-controlled window that moves with the gaze, only a pre-determined region around the current fixation is rendered visible while the remaining text is masked with unrelated characters. The moving-window paradigm impairs efficient reading to the degree that the masked information is useful in normal reading. When varying window size, the size of the perceptual span can be estimated as the minimum window size at which reading efficiency is comparable to normal, undisturbed reading. Defined this way, the size of the perceptual span in English adult readers is about 14–15 characters to the right for low-level word spacing information (McConkie & Rayner, 1976), about 8–9 characters to the right for letter identity information (Rayner, 1986; for Finnish see Häikiö et al., 2009), and only about 3–5 characters to the left of fixation for English readers (Pollatsek et al., 1981). Preview of upcoming text is crucial for saccade planning and targeting, which in turn are key elements in common models of eye-movement control during reading (e.g. Engbert et al., 2005, Reichle et al., 2013). Unsurprisingly perceptual-span size is a strong predictor of reading fluency in adults (Choi et al., 2015, Rayner et al., 2010, Veldre and Andrews, 2014) and children alike (Sperlich et al., 2016, Yan et al., 2020).

However, relatively little research has focused on the development of the perceptual span and its developmental causes. Previous studies have demonstrated that a) second-grade children already obtain limited information beyond the foveal region, b) perceptual-span size is larger for children in fourth than in second grade, and c) sixth graders have virtually the same span size as adults (Häikiö et al., 2009, Rayner, 1986, Sperlich et al., 2016). While basic oculomotor measures such as fixation duration and saccade length already vastly improve from first to second year of reading instruction (Meixner et al., 2022), perceptual span shows a major developmental step from second to third year (Sperlich et al., 2015, Sperlich et al., 2016), suggesting that basic oculomotor and linguistic skills need to be well developed before parafoveal preprocessing sets in (Sperlich et al., 2016). In particular, linguistic skills such as sight word reading (Sperlich et al., 2016) and reading fluency (Yan et al., 2020) have been suggested as developmental drivers of perceptual-span size.

Still, many issues in perceptual-span development remain unresolved. First, the developmental course of the perceptual span in children has not been examined beyond grade six. However, reading fluency improves until the end of college (Spichtig et al., 2016, Taylor, 1965), and it is conceivable that enhanced parafoveal processing contributes to these improvements. Even if the first cross-sectional analyses suggest that sixth-graders and adults employ similar perceptual-span sizes during reading (Häikiö et al., 2009, Rayner, 1986), this has not yet been confirmed longitudinally, and deeper levels of parafoveal preprocessing might be achievable later in development with growing cognitive capacity. Second, intra-individual developmental trajectories of perceptual-span development are hitherto poorly understood (Sperlich et al., 2016). However, such longitudinal research is needed to justify, develop, evaluate and establish adequate interventions for disadvantaged children (Francis et al., 1996, Lennon and Slesinski, 1999). A prime advantage of the current longitudinal design is the possibility to follow children’s intra-individual developmental trajectories and evaluate if initial reading performance is predictive for reading performance at the end of middle school. Third, whereas inter-individual differences in reading speed, reading skill, reading age, schizophrenia and deafness have already been taken into account (for a review, see Rayner, 2014), inter-individual differences in domain-general cognitive abilities and their impact on the perceptual span and its development have rarely been discussed. However, it is already established that early advantages in executive functioning have a positive influence on reading development (e.g. Kieffer et al., 2013, Meixner et al., 2019, Peng et al., 2022), and it is highly plausible to assume that executive functioning also shapes developmental trajectories of perceptual-span size as we will argue below. In addition to revealing a potentially strong predictor of individual differences in reading success, demonstrating such an association will help to clarify the theoretical underpinnings of the perceptual-span measure.

A major topic in educational psychology is whether early differences in reading skill compensate with time (developmental lag model); persist into later school years (deficit model); or even increase with time (increasing differences model). The latter is an example of the Matthew effect of reading (Stanovich, 1986), which has been demonstrated for various reading skills, such as word recognition (Bast & Reitsma, 1998), reading speed (Juel, 1988, Klicpera and Schabmann, 1993) and vocabulary (Cain & Oakhill, 2011), although it has been recognized that most reading-related skills tend to stabilize or even compensate at later grades (Pfost et al., 2014, Scarborough and Parker, 2003).

Developmental patterns in eye-movement based reading measures, and in particular in the perceptual span, have only recently been investigated for the first time (Sperlich et al., 2016, Spichtig et al., 2017). In Sperlich et al. (2016) we demonstrated a transient Matthew effect in reading rate and in the perceptual span from second to third grade, with stabilizing differences at a then larger separation from third to fourth grade. A minor aim of the current study was to re-examine this effect up to grade ten by adding data from three new measurement waves – now covering grades one to ten. We further refined the analysis by (1) using continuous rather than dichotomous predictors, (2) combining data from all three cohorts rather than analyzing them separately, and (3) applying nonlinear, asymptotic growth-curve models rather than linear models to better approximate the shape of the developmental course (cf. Meixner et al., 2022).

The Matthew effects discussed in the previous section mean that initial differences in reading skill become inflated. What causes this inflation beyond initial differences in reading skill? Candidate domain-general processes that are among the strongest determinants of individual differences across cognitive domains are executive functions. Executive functions are defined as a set of basic, domain-general cognitive abilities related to the control of working memory that support goal-directed behavior, such as reading (Jurado & Rosselli, 2007).

At the core of executive functions is the ability to regulate attention in order to focus on task goals, refrain from responding immediately, resist distraction, and plan for the future (Zelazo et al., 2016). In particular, three executive functions have been repeatedly identified using factor analyses in adults (e.g. Fisk and Sharp, 2004, Miyake et al., 2000) and children (Huizinga et al., 2006, Lee et al., 2013, Lehto et al., 2003): First, inhibition describes the ability to ignore task-irrelevant information, staying focused on the task, and suppress prepotent responses. Second, updating describes the ability to store and process new, relevant information in working memory and to operate on the contents stored in working memory. Third, shifting of attention or cognitive flexibility describes the ability to efficiently switch the focus between operations or tasks. The current study investigated initial executive functioning as a unitary concept, as developmental research suggests that executive functioning differentiates only later in development into separable subcomponents (Best and Miller, 2010, Connor et al., 2016, Holl et al., 2021, Welsh et al., 2010).

Executive functioning supports the reading process directly, e.g. by disintegrating ambivalent word meanings, integrating decoded words into mental models, switching between reading and comprehension monitoring, or inhibiting inappropriate eye-movement behavior. Executive functioning further supports reading acquisition through perpetuating self-regulation, e.g. setting learning goals, staying focused on task demands, listening attentively to teacher instructions, or actively searching for unknown information (e.g. McClelland and Cameron, 2011, Warner et al., 2017).

Consistent with these findings, a recent meta-analysis found that individual differences in executive functions cross-sectionally predict reading difficulties (Peng et al., 2022). In contrast to working memory or short-term memory, executive functions still retained some predictive value after controlling for phonological processing and language comprehension, although of course language skills explained most of the variance in reading performance.

The current study extends previous research by examining the relationship between executive functioning and the perceptual span. The perceptual span is not purely determined by visual limitations such as reduced parafoveal acuity, but massively shaped by attentional factors. For example, the span’s asymmetry changes with reading direction for bilingual readers of Hebrew and English (Pollatsek et al., 1981); enlarging the font size in the periphery to compensate for degraded visual acuity does not enlarge the perceptual span (Miellet et al., 2009); the perceptual span is smaller in oral compared to silent reading partly due to increased executive demands on coordinating perceptual input sampling and vocal output (Ashby et al., 2012; J. Pan et al., 2016, Pan et al., 2017); the more challenging the text material and the higher the information density of the script, the smaller the perceptual span (Inhoff and Liu, 1998, Rayner, 1986); and the perceptual span dynamically adjusts to current attentional processing demands (Henderson and Ferreira, 1990, Meixner et al., 2022).

The strong contribution of attention to the perceptual span makes executive functioning a prime candidate in explaining inter-individual differences therein. However, the available evidence on the relationship between executive functioning and the perceptual span is scarce, mixed, and indirect at best. The most affirmative results come from studies employing Rapid Automatized Naming (RAN), a reading-related task in which symbols on a screen need to be named as quickly as possible. Yan et al., 2013, Pan et al., 2013 found that dyslexics extract less parafoveal information in RAN than controls. They argued that because dyslexics have less automatized translation routines of visual symbols into phonological output, they need more attentional resources for foveal processing, causing a reduction of the perceptual span. Pan et al., 2013, Yan et al., 2013 reported much larger group differences in the eye-voice span between dyslexics and controls with highly practiced than with novel material. This suggests that the degree of automaticity in translation from visual symbols to phonological codes is a critical factor in determining reading efficiency. Again, because automatized routines need fewer executive resources, this argues for a critical involvement of executive functions.

With reading proper, participants scoring high in working-memory span tasks exhibit shorter fixation durations (Osaka & Osaka, 2002), longer saccades (Luke, 2018; cf. Osaka & Osaka, 2002) and skip more words during reading (Kennison & Clifton, 1995), which might be indirectly linkable to a larger perceptual span for high-span individuals. In contrast, individuals with schizophrenia, typically suffering from inhibitory control deficiencies, have a much smaller perceptual span and make shorter forward saccades, (e.g. Luke, 2018, Whitford et al., 2013). In the current study, we examine (1) if executive functioning is positively associated with the perceptual span and with reading skill initially, and (2) if initial executive functioning longitudinally predicts the development of a larger perceptual span and of reading efficiency. Besides domain-general capabilities, it is well supported that efficiency in domain-specific linguistic processes predicts inter-individual differences in perceptual-span size (Rayner, 2014, Sperlich et al., 2016). Arguably, an increased automatization of linguistic processes puts less demand on the central executive during reading, which will in turn allow for a wider perceptual span (Henderson and Ferreira, 1990, Meixner et al., 2022). Therefore, both executive and linguistic functioning might show a joint impact on perceptual-span development.

To summarize, the current report presents new data from our longitudinal study on eye-movement development during silent reading in general, and perceptual-span development in particular. Three cohorts starting at first, second, and third grade of primary school took part in five measurement waves in total, spanning the time from grades 1 to 10. This is the first study that allows characterizing the longitudinal development of the perceptual span from the beginning of primary school to the end of middle school. Our main focus is on the identification of longitudinal predictors, in particular executive functioning. To this end, we relate reading rate and perceptual span in the final year and their developmental growth to initial reading skill and to initial executive functioning. We hypothesized that executive functioning is an important driver of perceptual-span development.

We compare the influence of initial executive functioning on initial reading skill and on longitudinal development of reading skill and of the perceptual span. We further test whether a longitudinal effect still holds when executive functioning effects on initial reading skill are controlled for, which would indicate that additional executive control functions that are not captured by initial reading later contribute to reading performance. In particular, we test for interactive effects of initial reading skill and initial executive functioning on development, which would suggest that they exert a multiplicative effect on reading and perceptual-span development. Such potential interactive effects can best be conceived of as dual benefits: if controlling for their initial reading skill students with better executive control might develop their reading skills faster than what is expected from their initial executive or reading skills alone.