Masked suffix priming and morpheme positional constraints

Although masked stem priming (e.g., dealer–DEAL) is one of the most established effects in visual word identification (e.g., Grainger et al., 1991), it is less clear whether primes and targets sharing a suffix (e.g., kindness–WILDNESS) also yield facilitation (Giraudo & Grainger, 2003; Duñabeitia et al., 2008). In a new take on this issue, we show that prime nonwords facilitate lexical decisions to target words ending with the same suffix (sheeter¬–TEACHER) compared to a condition where the critical suffix was substituted by another one (sheetal–TEACHER) or by an unrelated non–morphological ending (sheetub– TEACHER). We also show that this effect is genuinely morphological, as no priming emerged in non–complex items with the same orthographic characteristics (sportel–BROTHEL vs. sportic–BROTHEL vs. sportur–BROTHEL). In a further experiment, we took advantage of these results to assess whether suffixes are recognized in a position–specific fashion. Masked suffix priming did not emerge when the relative order of stems and suffixes was reversed in the prime nonwords—ersheet did not yield any time saving in the identification of teacher as compared to either alsheet or obsheet. We take these results to show that –er was not identified as a morpheme in ersheet, thus indicating that suffix identification is position specific. This conclusion is in line with data on interference effects in nonword rejection (Crepaldi, Rastle, & Davis, 2010), and strongly constrains theoretical proposals on how complex words are identified. In particular, because these findings were reported in a masked priming paradigm, they suggest that positional constraints operate early, most likely at a pre–lexical level of morpho–orthographic analysis