1 Cross-language interference during trilingual picture-naming in single and mixed language conditions Julia Festman Abstract In a picture-naming task, the performance of 17 unbalanced trilinguals was examined in three target languages (L1 German, L2 English, L3 French). Due to frequent switching more interference was found in the L3 from the L2, but in the L1 and the L2 more from the L2 and the L1, respectively. In general, results showed that the more a language was activated, the more frequently it was found to cause interference in weaker target languages. An extension of Green’s (1998) bilingual model is put forward to explain trilingual processing, and the direction of cross-language interference. Zusammenfassung In einem Bildbenennungsexperiment wurde die Performanz von 17 unbalancierten Dreisprachlern in drei Zielsprachen (L1 Deutsch, L2 Englisch, L3 Französisch) untersucht. Durch das häufige Wechseln wurde in L3 mehr Interferenz von L2 gefunden, in L1 und L2 jedoch mehr von L2 beziehungsweise von L1. Im Allgemeinen zeigen die Ergebnisse, dass, je aktivierter eine Sprache ist, umso häufiger verursacht sie Interferenzen in schwächeren Zielsprachen. Eine Ausweitung von Greens (1998) zweisprachigem Modell wird als Erklärung für dreisprachige Verarbeitung und Interferenzrichtungen vorgeschlagen. 1. Cross-language interference It is commonly assumed that a speaker who masters several languages has the ability to choose at any point in time which of them she wants to use. This ability, usually referred to as “language choice”, suggests that when a speaker wants to use a certain language, then this is exactly what she will do. Research on cross-language interference (CLI) (e.g., Williams & Hammarberg, 1998) shows that despite the speaker’s language choice, another language might intrude. CLI is here defined as the unintended production of one language during the intended production of another (e.g., “Baum” (German, for tree) instead of “tree”).