Lexical versus compositional World-Language Relations: Event-Related Brain Potential effects during Second Language Processing State of the art: Overt verification (hearing piano and matching it to its referent) occurs incrementally in native language comprehension and is distinct by type (e.g., lexical verb-action relations are processed distinctly from compositional role relations). For language learning, verification might also be relevant. Learners must (i) identify words, thematic role relations and (ii) relate language to referents; (iii) verify (mis)match, and (iv) convey matches to memory. The argument for verification-in-learning receives support from backpropagation (verifying error between actual and target output to help learning, Elman, 1990; Rumelhart et al., 1986). (In)congruence and contrast are also key in human studies on learning (e.g., Yu & Smith, 2012) and language processing (Koehne & Crocker, 2014). If verification-in-comprehension is key in adult second-language (L2) learners too, then their verification should resemble those of L1 natives. For beginning second-language learners, we predict lexical mismatches (verb-action) should be processed much like in native comprehenders; by contrast, compositional mismatches should be processed differently from natives (see McLaughlin et al., 2010). The present pilot study (N=16) investigated the functional brain responses associated with lexical (verb-action) and compositional (thematic role relation) mismatches in L2 comprehenders (monolingual < age 6, German native, English L2, advanced). In a picture-sentence verification task, participants were asked to inspect a clipart scene with, for instance, a golfer entertaining a butler (Fig. 1a on poster). Subsequently, a sentence about these characters e.g., The golfer entertains the butler was presented in English word by word. The scene either matched the sentence completely (e.g., Fig. 1a on poster), mismatched in action depiction (e.g., Fig. 1b on poster), mismatched in role relations (e.g., Fig. 1c on poster), or mismatched in both action and role relations (e.g., Fig. 1d on poster). Participants verified picture-sentence congruence post-trial via button-press. Results: Significantly larger mean amplitude negativities to role mismatches vs. matches were replicated at noun1, e.g., golfer (0 - 100 ms, 100 - 300 ms and 300 - 500 ms, Fig. 3 on poster), and the first 100 ms of the verb, e.g., entertains. Action mismatches yielded larger mean amplitude negativities than matches to the verb (300 - 500 ms, Fig. 4 on poster). Differences between L1 and L2 comprehenders: For L1 comprehension, prior extant results from one of the authors revealed a main effect of action but not of role congruence to the verb. For L2 comprehenders, a main effect of role congruence emerged to the verb, with larger relative positivities for role mismatches than matches (verb entertains: 300 – 500 ms; noun2 butler: 0 – 100 ms and 100 – 300 ms). Discussion & Conclusion: Advanced L2 comprehenders of English processed role and action (mis)matches in sentence-picture verification studies incrementally, much like L1 comprehenders. Verification of lexical (verb-action) mismatches in L2 comprehenders resembled extant results in L1 comprehenders. By contrast, verifying role relations differed between L1 natives and advanced L2 comprehenders[1] <applewebdata://7FB95C20-938C-4617-9509-906C06169733#_ftn1>. The role congruence effect at the first noun phrase suggests that L2 comprehenders perceived the role-mismatch at the earliest possible time. However, it is possible that the associated revision and/or reconciliation is time-consuming and thus continues into the (mis)matching verb and post-verbal noun. Further cognitive tests and a follow-up study with beginning L2 learners are planned to further assess the observed lexical-compositional differences in L2 comprehension. [1] <applewebdata://7FB95C20-938C-4617-9509-906C06169733#_ftnref1> Investigation of verification in L2 beginners is pending. References Elman, J. (1990). Finding structure in time. Cognitive Science, 14, 179-211. McLaughlin, J. et al. (2010). Brain potentials reveal discrete stages of L2 grammatical learning. Language Learning, 60, 123-150. Rumelhart, D. et al. (1986). Learning representations by back-propagating errors. Nature, 323, 533-536. Yu, C. & L. Smith (2012) Modeling cross-situational word--referent learning: Prior questions. Psychological Revue, 119, 21-39. doi:10.1037/a0026182.