There is ample evidence that word order is a problematic domain in L2 usage. In particular, production of the verb-second (V2) phenomenon, which requires the finite verb in main clauses to appear in second position, (e.g., Ganuza, 2008 for an overview). Surprisingly, we know very little about how these structures are processed and how production relates to comprehension. We therefore examined how advanced German (N=14) and English (N=14) adult learners, matched for proficiency and age of acquisition (AoA; German M=22, English M=23), process word order in Swedish compared to native speakers (N=20) depending on language background (L1 with [German] or without [English] V2), preposed adverb frequency (frequent idag ‘today’ vs. infrequent hemma ‘at home’, ex. 1), and the length of the preposed constituent (short vs. long prefield, ex. 2). (1) Idag/Hemma läste hon tidningen. vs. *Idag/Hemma hon läste tidningen. Today/At home read she paper.def vs. *Today/At home she read paper.def (2) Idag/Hemma hos Maria läste hon tidningen. vs. *Idag/Hemma hos Maria hon läste tidningen. Today/At home at Maria read she paper.def vs. *Today/At home at Maria she read paper.def. We examined responses to word order violations in an acceptability judgement task and an ERP experiment, and probed the production of word order in a sentence completion task.

Preliminary results from the judgment task indicated that native speakers were faster and more accurate on judging sentences than both L2 groups who did not differ. Overall, the more frequent adverb, idag, also affected accuracy and reaction times positively, but there were no interactions with group. The outcome from the sentence completion task showed similar results: native speakers were more accurate than the L2 groups who did not differ, and an overall adverb frequency effect was found, but not difference across groups. In contrast, the ERP data showed different patterns. In native speakers V2 violations elicited a bimodal ERP response, an anterior negativity followed by a posterior P600. These effects were increased in amplitude and the anterior negativity was left lateralized (LAN) when the prefield was long. In the German group a bimodal response was detected only when V2 violations followed a frequent adverb in a long prefield. In other cases only a posterior P600 was evident. The English group, in contrast, showed an early anterior positivity, and a later lateral parietal negativity in the N400 time window that was followed by a posterior P600. These responses were affected only by prefield length and only in amplitude.

Overall, the results indicated that advanced German and English learners, matched on proficiency and AoA, who performed similarly on behavioural measures of comprehension and production of word order, still differed in online processing. More specifically, language background mattered since the German learners whose L1 share V2 with target Swedish, overall showed similar ERP patterns to native speakers. In contrast, the English learners, whose L1 does not share V2, showed more variation in their ERP responses. We discuss the implications of these findings for theories of crosslinguistic influence and theories of nativelike syntactic processing.