Frequent engagement in extramural English (EE) activities (English-language activities that students engage in outside of the classroom) has been shown to positively influence students’ L2 receptive and productive skills (e.g., Sundqvist, 2009, 2019). There are also indications in previous studies that the type of EE input students receive affects their production (Kaatari et al., 2023). Extending this line of research, we test the role of the type of input students receive through EE activities focusing specifically on their effect on lexical complexity. To do so, we look at junior and senior high school student writing in L2 English from the Swedish Learner English Corpus (SLEC; Kaatari et al., forthcoming). SLEC contains information about how many hours per week students engage in five EE activities: reading, watching, conversation, social media, and gaming. We use three types of psycholinguistic lexical sophistication measures (contextual distinctiveness, concreteness, and age of exposure), along with one measure of lexical diversity (moving average type-token ratio). Specifically, we build on previous research and ask the following research questions that also serve as our hypotheses:

Does frequent engagement with spoken input (conversation and watching) result in a higher degree of linguistic diversity than other types of EE exposure?

Does frequent engagement with longer written input (reading) result in a higher degree of linguistic sophistication, than other types of EE?

To test these specific hypotheses, we use Structural Equation Modeling (SEM; Larsson et al., 2021). Competing measured variable path analysis models were fitted, systematically testing our hypotheses. The best-fitting model (χ2: 0.14, df: 20, CFI: 0.99, RMSEA: 0.033[0.00–0.064], SRMR: 0.067) confirmed both of our hypotheses. It thus seems crucial to avoid grouping EE activities together into a single category, and instead consider what type of input students are exposed to.