Compound processing in Finnish L2 speakers

Several studies have shown that compound processing is affected by lexical properties such as word length and morpheme boundary cues. For instance, the length of compound words modifies the extent to which constituents are used in processing (Bertram & Hyönä 2002). There is also evidence that compound parsing differs for native and nonnative speakers, as nonnative speakers are more likely to use orthotactic parsing cues than native speakers (Lemhöfer et al. 2011). In this study, we investigated to what extent relatively short (mean length 9 characters) and high-frequency compounds (merivesi ‘sea water’ or tyttöystävä ‘girlfriend’) are processed via constituents by beginner level L2 speakers in comparison to native speakers. Eye movements were registered during reading of compounds presented in sentence context. Several eye movement measures showed significant whole-word frequency effects for both L1 and L2 speakers. Moreover, the L2 speakers’ processing was affected by constituent frequency, whereas L1 speakers’ processing was not. Taken together, the results suggest that constituents have a bigger role in processing relatively short Finnish compounds for L2 speakers than for natives. This suggests that the developmental trajectory to process complex words moves from decomposition to whole word processing (cf. Portin et al., 2007) rather than from whole word processing to decomposition as suggested by the shallow structure hypothesis (e.g., Clahsen & Felser, 2006) or the declarative/procedural model of Ullman (Ullman, 2001).