Frequency and predictability effects in first and second language of different script bilinguals.

Abstract

Reading efficiently in a second language (L2) is a crucial skill, but it is not universally achieved. Here we ask whether L2 reading efficiency is better captured as a language specific skill or whether it is mostly shared across L1 and L2, relying on general language abilities. To this end, we examined word frequency and predictability effects in sentence reading, and tested the same readers in L1 and L2, recording participants' eye-movements. Participants were 57 undergraduate bilingual speakers of Hebrew and English, languages that use different scripts, allowing for a clearer distinction between L1 and L2 processing. Both word frequency and word predictability effects were more pronounced in participants' L2 than in the L1, suggesting that both lower level and higher-order processes in reading are sensitive to language proficiency. Further, frequency effects in the L2 were linked with L2 proficiency but not general language abilities, and L2 predictability effects were not associated with either variable. Finally, readers' frequency and predictability effects in L1 and L2 were not associated with each other. Taken together, these results suggest that for different-script bilinguals, efficient reading in the L2 is a highly specific skill, dependent upon proficiency in that language, and drawing less on L1 and general language ability.

Insights Into the Processing of Collocations During L2 English Reading: Evidence From Eye Movements

“…This pattern of findings is consistent with a wealth of other research showing that readers make use of their knowledge of the prior linguistic context to guide the processing of new linguistic information (for a review, see Staub, 2015 ). Moreover, our findings are in line with the findings from other studies of L2 readers showing that eye movements are affected by both the frequency of usage and contextual predictability of words (e.g., Whitford and Titone, 2017 ; Mor and Prior, 2021 ). The predictability effects we observed were emerged in measures of first-pass processing and were also observed in later measures of processing.…”

Eye Movement Traces of Linguistic Knowledge in Native and Non-Native Reading

“…Despite these advances, the functional form of both frequency and predictability effects in L2 reading has not been characterized to date. Prior work that examined frequency effects in L1 and L2 assumed a linear effect of log frequency on reading times (Cop, Keuleers, et al, 2015 ; Mor & Prior, 2022 ; Whitford & Titone, 2012 ). Whitford and Titone ( 2017 ) and Mor and Prior ( 2022 ) assumed a linear effect of (untransformed) predictability.…”

“…Prior work that examined frequency effects in L1 and L2 assumed a linear effect of log frequency on reading times (Cop, Keuleers, et al, 2015 ; Mor & Prior, 2022 ; Whitford & Titone, 2012 ). Whitford and Titone ( 2017 ) and Mor and Prior ( 2022 ) assumed a linear effect of (untransformed) predictability. Here we relax this linearity assumption, using nonparametric statistical methods to find the functional form best supported by the data.…”

“…These effects have been shown to apply across languages, and their functional form has been studied in L1 (Kliegl et al, 2004 ; Smith & Levy, 2013 ; Wilcox et al, 2020 ). However, only a few studies have examined benchmark word property effects in L2 reading (Cop, Keuleers, Drieghe, & Duyck, 2015 ; Mor & Prior, 2022 ; Whitford & Titone, 2012 , 2017 ), and both their functional form in L2 and their relation to language proficiency are currently unknown.…”

“…We perform three analyses in which we examine the functional form of benchmark word property effects, their magnitude , and how they depend on language proficiency . These analyses build on prior work that estimates frequency and predictability effects in L1 and L2 using linear modeling (Cop, Keuleers, et al, 2015 ; Mor & Prior, 2022 ; Whitford & Titone, 2012 , 2017 ). We go beyond this prior work by characterizing the functional form of these effects, explicitly comparing them to one another, estimating their dependence on language proficiency using standardized test scores, and using a larger and more linguistically diverse dataset.…”