During comprehension, listeners can draw upon different sources of non-linguistic information for reference resolution and structural disambiguation (e.g., Tanenhaus et al., 1995). While non-linguistic information plays an active role in extant accounts of sentence processing, social-context effects (e.g., formality of a situation and register) have received little attention (but see van Berkum, et al., 2008). The present eye-tracking pilot study (n=9, 34 critical items, 64 fillers) examines whether (i) formality conveyed by a linguistic context can rapidly affect the visual interrogation of object photographs and associated comprehension processes; (ii) congruence of verb-argument meaning rapidly affects comprehension; and (iii) congruence in context formality and register interacts with semantic verb-argument congruence (testing to what extent we can assume a single conceptual store and closely-linked mental representation that encompasses register information).

Method and Design: Using the Visual World Paradigm (Figure 1), we examine native German speakers' comprehension of register variants (Latschencolloquial vs. Schuhestandard transl:.‘shoes’) in a formality (mis)matching context sentence, given the object argument either matches or mismatches verb meaning constraints (e.g. Ich binde schnell meine Schuhe/Latschen/*Kleidung/*Klamotten, transt.: ‘I’m about to tie my shoesstandard/ shoescoll/ *clothesstandard/ *clothescoll’). (Mis)matches are assessed two-fold: context formality-target register (Table 1) and semantic verb-argument congruence (1a,1b vs. 2a,2b).

Hypotheses: At the verb region, if participants are sensitive to formality-register and verb- argument congruence we expect more looks to the formality-register and verb-argument matching object when it matches the formality-register and verb-meaning constraints than when it mismatches either or both. If they are sensitive to verb-argument congruence irrespective of formality-register congruence, we predict greater fixations to the object referent in the verb-argument congruent condition irrespective of whether it matches or mismatches the formality of the context sentence. In the post-verbal object noun region, if formality-register congruence rapidly modulates visual attention to the objects and associated comprehension, we expect more looks to the object referent in the formality- register matching (Full match; Verb-Argument mismatch) than mismatching conditions. If fixations to the object referent are modulated by verb-argument congruence (main effects of verb-argument congruence) independent of formality-register congruence, we should likewise see more looks to the object referent in the verb-argument matching than mismatching conditions (Full match; Register mismatch). Moreover, if formality-register and verb-argument congruence interact during the post-verbal object noun region, how verb- argument congruence is processed may depend on formality-register congruence. The latter might indicate that standard language processing mechanisms are in close interaction with register representations, supporting the interpretation of a single conceptual store and set of mechanisms.

Preliminary Analyses and Results: Linear mixed-effects models were fitted to the empirical-logit transformed fixation proportions to the verb-argument (mis)matching and formality-register (mis)matching objects at the verb region, and to the object noun referent for each condition at the post-verbal object noun region. Preliminary results of the pilot study show a main effect of verb-argument congruency, indicating that verb-argument relations are computed and used immediately in on-line language comprehension, yet no main effects of formality-register congruency were found in the verb and post-verbal object noun region. Our analysis also revealed an interaction of formality-register and verb-argument congruency in the verb-region and post-verbal object noun region (Figure 2). These findings taken together imply that situation-formality might indeed modulate the processing of verb-argument congruence, possibly indicating that standard language processing mechanisms are in close interaction with register representations. Given the low power of our pilot, these results need to be interpreted with caution. The main experiment with 32 participants is underway.