Abstract
Evidence against parafoveal processing of morphosyntactic information in linear inflectional morphology languages (English (Kambe, 2004), Finnish (Bertram, 2003), Malay (Winskel et al., 2014)) has been questioned by new cross-linguistic evidence for such processing (Russian (Stoops & Christianson, 2017; 2019); Finnish (Hyöna et al., 2021); English (Dann et al., 2021)). Thus, Stoops et al. (2017; 2019; 2020) found early and late morphological effects on words n+1 and n, using boundary-change techniques (between words: Rayner, 1975; within-word: Hyöna et al., 2004) on Russian nouns. Moreover, the syntactic predictability of the previews (predictable vs. ungrammatical) modulated the results (Stoops et al., 2020): ungrammatical preview did not differ from identical and yielded preview benefit over the nonword only on the word n+1. On the word n both related and nonword previews interfered with the word processing, supporting the idea of increased attention span for word n (Juhasz et al., 2009). The present study examined whether syntactic predictability modulated parafoveal processing of Russian verbs not examined previously by manipulating syntactic expectations for the previews (word n). In experiment 1, the identical preview had the expected case marker (94% Cloze test score), and the morphologically-related preview had a syntactically unacceptable case. In experiment 2 both identical (60% Cloze test score) and related (40% Cloze test score) previews were acceptable. As with previously examined nouns, ungrammatical morphologically-related previews did not differ from identical and induced preview benefit versus nonword in later measures, suggesting word-level facilitation but no early message-level preprocessing for ungrammatical morphology. However, acceptable related previews yielded a preview cost over the identical preview along with the nonword preview in later measures. Our study contributes to the increasing evidence that parafoveal attention allocation is context dependent rather than a rigid mechanism that behaves the same in all sentential contexts.