Abstract Heide et al.
How does morphological structure influence the processing of German prefixed verbs?
Though there is a longstanding debate on how polymorphemic words are stored and processed in the Mental Lexicon, data on derivational affixes in German are still rare and there is only one study investigating German polymorphemic verbs prefixed with ver (Schirmeier et al., 2004). However, because of certain morphological and semantic characteristics these verbs are particularly revealing. Our study focusses on the morphological structure of German ver-verbs and how it influences processing. We replicate and extend the results of Schirmeier et al. (2004) by controlling our material for an additional morphological factor: potential decomposability.
Our study investigates the processing of polymorphemic German verbs prefixed with ver. Questions of interest are whether German ver-verbs are decomposed into their constituent morphemes by default or if the decomposition is influenced by the verb's morphological structure, i.e.
a) the lexicality of the root (word or non-word),
b) the part of speech of the root (verb, noun, adjective or neologism) and
c) the potential decomposability into morphemes (flat [ver]+[ROOT]+[en], right-branching [ver]+[ROOT+en] or both).
134 German ver-verbs were selected from the CELEX database and were assigned to one of five categories according to their morphological structure. The categories were matched for whole word frequency, root frequency, word length, number of orthographic neighbours and semantic transparency.
We used a masked priming procedure (prime duration= 63 ms) to measure the reaction times of 60 native speakers of German in a lexical decision task. We presented 134 existing German ver-verbs as targets and 134 non-existing ver-verbs as well as 150 monomorphemic words and non-words as fillers. Targets were primed by either a related or an unrelated ROOT (conditions 1 and 2) or by either a related or an unrelated [ROOT+en] form (conditions 3 and 4).
We analyzed the reaction times for main effects of morphological category and priming condition as well as for the interaction of these two factors (category x condition) by running ANOVA analyses for participants and items.
Results and Conclusions
A preliminary analysis of nine datasets revealed that reaction times depend both on the morphological category of the target and on the nature of the prime (ROOT or [ROOT+en], related or unrelated). Participants responded faster when the target was preceded by a related ROOT or [ROOT+en] form. If the target was decomposable both as flat and right-branching, the priming effect was bigger for [ROOT+en]-forms than for ROOT primes. If, in contrast, the target was decomposable in a flat manner only, the sole ROOT primed better than the [ROOT+en] form. We therefore argue against a decomposition-by-default mechanism as proposed by Taft & Forster (1975) but show that decomposition is guided by the word’s morphological structure.