engleski

Processing of cognates in Croatian as L1 and German as L2

Cognates, or words with similar form and meaning in two languages have already been recognized as co-determinants of cross-linguistic processing. In order to explore the underlying mechanisms, different methods have been used: ERP experiments (Midgley, Holcomb, & Grainger 2011 ; Peeters, Dijkstra, & Grainger 2013 ; Strijkers, Costa, & Thierry 2010), lexical decision tasks (Dijkstra et al., 2010 ; Lemhöfer, Dijkstra, & Michel, 2004 ; Lemhöfer & Dijkstra 2004), standardized tests (Kelley & Kohnert 2012 ; Pérez, Peña, & Bedore, 2010). Common finding is that the cognates are processed differently from non-cognates. Yet, authors disagree over range of characteristics, including the direction and/or the magnitude of the effect. For example, a number of experiments showed that cognates facilitate the comprehension both in L1 and L2 (Van Assche et al. 2009 ; Van Hell & Dijkstra 2002). In other studies facilitation effect emerged only in L2, but not in L1 (e.g., Caramazza & Brones 1979). Recent study by Radanović, Feldman and Milin (2014) showed even more complex pattern: an inversed effect in L1, where cognates were processed slower than non-cognates, and the effect in L2 which was modulated by the word frequency and participant's language proficiency. The authors attribute this unexpected finding to „special properties of the particular L1 and L2 languages“, that is to Serbian as L1 and to English as L2. In order to further explore this hypothesis, we designed a replication of the study and its methodology using another pair languages: Croatian as L1 and German as L2. Late bilinguals of German L2, students of German with Croatian as L1 will participate in a visual lexical decision tasks. Stimuli will cover a broad range of orthographical and phonological similarity between L1 and L2, from completely overlapping cognates (e.g. film ‒ Film), to translations with entirely different forms (such as čaša ‒ Glas). The data will be analysed using Generalized Additive Mixed Modeling (GAMM). Due to the similarity of language pairs Croatian-German and Serbian-English, we expect this experiment to further explain findings from Radanović, Feldman and Milin (2014). Facilitation or inverse effects from cognateness would enable us to take a more detailed look at the observed „special properties of the particular L1 and L2“ and better explain their characters and roles.

Croatian as L1; German as L2; cognates

nije evidentirano