A dynamic adjustment model of saccade lengths in reading for word-spaced orthographies: evidence from simulations and invisible boundary experiments

A cross-linguistic study of spatial parameters of eye-movement control during reading.

“…These empirical findings underlie the new center-Based Saccade Length account of a mechanism of saccade target selection by Cutter et al (2018): following Hautala et al (2022), we label this account dynamic adjustment as opposed to discrete control. On this account, "readers do not plan and execute saccades based on a [preferred saccade length] metric established over a relatively extended period (i.e., an experientially based metric), but instead compute saccade metrics moment to moment on the basis of perceptual information about the length of the word under fixation and the length of the word to which a saccade is to be targeted" (Cutter et al, 2018(Cutter et al, , p. 1439.…”

“…For these analyses, we defined initial landing position relative to the word center (in half-letter intervals) and modeled its nonlinear effect as a second-degree polynomial using SPATIAL PARAMETERS OF EYE-MOVEMENT CONTROL poly() function with raw polynomials. In line with (Cutter et al, 2018;Hautala et al, 2022), we also model the effect of the centerbased distance as a second-degree polynomial, using the same function settings. All spatial parameters were estimated in the direction of reading, making Hebrew data comparable with those of other language samples.…”

“…Thus, one approach to a cross-linguistic study would be to harvest the literature and compare estimates reported for the range of languages with mean word lengths or other metrics of word length distributions characterizing those languages. A recent study of saccade lengths by Hautala et al (2022) adopted this approach to eye-movement corpora from English, Finnish, and Russian (Hautala & Loberg, 2015; Laurinavichyute et al, 2019; Plummer & Rayner, 2012; Schilling et al, 1998) and found these languages to be largely comparable in all measurements related to spatial oculomotor control. Yet, such an approach needs to contend with the undesired variability stemming from potential differences in reading proficiency of participants, the type (sentence vs text), genre (fiction vs nonfiction), and complexity of the reading materials.…”

“…Thus, the distance from the launch site in word N to the center of word N þ 1 is an additional variable of importance for this account. The dynamic adjustment account has found recent "proof-of-concept" support in simulations and experiments by Hautala et al (2022).…”

“…This adaptation is rapid too, taking place within seconds from one trial to the next. On this account, even if preferences for a certain saccade length do form through the lifetime of reading experience, they can be effectively overridden by the actual visual characteristics of the text to be read (Cutter et al, 2017(Cutter et al, , 2018Hautala et al, 2022), see also related work by (Liu et al, 2019;Vitu, 1991b;Yang & Vitu, 2007;Yao-N'Dré et al, 2014). In their discussion of the Bayesian account of saccadic range error, Cutter et al (2017) suggest that even if the prior exists, its origin and temporal scope require further specification.…”