Lucia Pozzan, Kristina Woodard, & John C. Trueswell (University of Pennsylvania)
lpozzan@sas.upenn.edu
Sentence Processing; L2 processing; Revision; Garden Path sentences; Executive Functions; Cognitive Training; Visual world paradigm
Real time language comprehension and production require focusing on and rapidly integrating multiple sources of information. Growing evidence shows that this process is supported by domain-general executive function (EF) skills. Here we explore whether the ability to process complex sentences in a second language (a) is supported by EF skills and (b) can benefit from EF training.
Over 2 months, 20 Chinese child learners of English played games (www.lumosity.com) aimed at measuring and training EF-skills. Two pre- and post-training sentence processing tasks were also administered. After correcting for age and English proficiency, at pre-test EF skills did not predict performance on canonical (e.g., actives), non-canonical (e.g., passives) and unambiguous structures (e.g., non-reduced relative clauses), but reliably predicted differences in sentence processing for temporarily ambiguous structures requiring revision of initial interpretations (e.g., reduced relative clauses: β = .68, t = 4.09, p<.01). After 2 months, the group who underwent EF-training showed improved sentence processing performance compared to an active control group; this improvement was selective to temporarily ambiguous sentences, compared with unambiguous ones, but was present for both canonical and non-canonical structures. In addition, training-related EF-improvements selectively predicted sentence processing improvements for non-canonical (β = .63, t = 2.91, p=.01) and temporarily ambiguous structures (β = .77, t = 2.93, p=.01), but not for canonical and unambiguous structures.
These results suggest that (a) domain-general EF skills support language processing and (b) training-related gains in domain-general EF skills can transfer to untrained domains (e.g., sentence processing). Such findings represent an important step in identifying the cognitive processes that underlie language processing and development, and the circumstances under which language learners might benefit from domain-general cognitive training.