This study investigates early stages of third language acquisition (L3A). At present, most L3A research focuses on cross-linguistic influence (CLI), i.e. whether properties are transferred based on linguistic proximity (e.g., Westergaard et al. 2017), typological primacy (e.g., Rothman 2015) or further factors. In our study, while also investigating CLI at initial stages, we focus on L3 development. According to the Cumulative Input Threshold Hypothesis (CITH) (Cabrelli & Iverson submitted), CLI from the L2 is overcome faster than CLI from the L1 (Cabrelli et al. 2018). Our research question is whether order of acquisition is a determining factor in overcoming CLI.

The L3 German learners (N=55) who participated after 7, 26 and 42 weeks of exposure are 15-17-year-old L1 Norwegian speakers with high proficiency in L2 English. We conducted an acceptability judgment task with five conditions, two of which are structurally similar to Norwegian (adverb placement in subject-initial declaratives, V2 in non-subject initial declaratives), one to English (obligatory articles in generic contexts that allow article omission in Norwegian), one to both English and Norwegian (prenominal placement of possessive determiners), and one to none of the two languages (object-verb word order). We included twelve (un)grammatical items per condition.

Our findings partially support the CITH. We found a significant main effect of time and condition. Accuracy is increasing significantly over time for three conditions: Possessive condition with facilitation from L1 AND L2, object-verb condition with non-facilitation from L1 AND L2, adverb condition with non-facilitation from the L2 and facilitation from the L1. The findings in the adverb and object-verb conditions suggest that overcoming non-facilitation is possible. Supporting the CITH, non-facilitation from the L2 is overcome faster in contrast to the L1: There is a significant increase of accuracy in the adverb condition but not in the generic condition, nor in the V2 condition.