While oral corrective feedback is a principal focus in second language acquisition research, most studies examine feedback once it has been provided. Investigating how instructors make in-class feedback decisions has not been thoroughly explored, despite the fact that classroom feedback occurs at the discretion of the individual language instructor and evidence from case studies that feedback provision varies greatly between instructors (recently reviewed in Lyster, Saito, & Sato, 2013). This study investigates how 32 instructors make their moment-to-moment feedback decisions in response to learner errors during natural, university-level Spanish foreign language lessons. Each instructor had a 50-minute grammar-focused lesson videotaped and participated in a stimulated recall. Results reveal instructor in-class feedback decision making to be a systematic and ordered cognitive process: Instructors who reflect on learner errors report that contextual (e.g., error type), learner (e.g., perceived student ability), and instructor factors (e.g., research background) influenced their decision whether or not to provide feedback, as well what type(s) to provide, and when. Others report having automatized their feedback practices and not reflecting on learner errors. All instructors report that their individual characteristics, specifically their native language, teaching experience, and training in second language acquisition, mediate the corrective feedback they provide.Keywords: teacher cognition; oral corrective feedback; foreign language classrooms; stimulated recall; instructor characteristics META-ANALYSES ON THE EFFECTIVENESS of corrective feedback (CF) have demonstrated that it plays a facilitative role in language development (Li, 2010;Lyster & Saito, 2010;Mackey & Goo, 2007;Russell & Spada, 2006), particularly in foreign language (FL) settings where learners' opportunities with the target language are limited (Li, 2010;Mackey & Goo, 2007). To date, the majority of CF research has focused on specific feedback-related variables-such as explicitness (Adams, Nuevo, & Egi, 2011 (Gurzynski-Weiss & Baralt, 2014, 2016Mackey, Gass, & McDonough, 2000). Additionally, many studies have identified variables that affect the level of effectiveness of CF, including learners' individual differences such as age (Mackey & Oliver, 2002;Mackey & Sachs, 2012), working memory (Egi, Fujii, & Tatsumi, 2002;Goo, 2012), anxiety (Rassaei, 2013(Rassaei, , 2015, and language analytic ability (Li, 2013;Yilmaz, 2013). In other words, most research has investigated CF once it has been provided. Examinations into how instructors make their in-class feedback decisions represent an understudied area in the descriptive CF research domain. This is surprising given
256The Modern Language Journal 100 (2016) how classroom CF occurs at the discretion of the individual language instructor, evidence from case studies (e.g., Basturkmen, Loewen, & Ellis, 2004;Sheen, 2004) that CF varies greatly between instructors (see Lyster et al., 2013, for a recent re...