Kırkıcı, B. 1 & Clahsen, H. 2
1 Middle East Technical University, Ankara
2 University of Essex
Whilst previous masked priming experiments with native (L1) speakers of English and other Indo-European languages revealed clear priming effects for both regular inflectional and productive derivational processes, studies comparing inflectional and derivational forms in non-native adult second language (L2) learners of English and German (Silva & Clahsen, 2008, Neubauer & Clahsen, 2009, Clahsen & Neubauer 2010) found no priming for inflectional and reduced priming effects for derivational forms. These findings suggest that L2 learners rely more on full-form storage and less on decompositional procedures than native speakers, particularly for inflectional processes. The present study investigates L1 and L2 morphological processing in a non-Indo-European language, Turkish, comparing regularly inflected Aorist verb forms with deadjectival (–lIk) derivational forms, both of which are highly frequent, productive and transparent, in Turkish. A series of masked priming experiments was carried out with groups of monolingual L1 speakers and proficient adult L2 learners of Turkish from different L1 language backgrounds. The SOA for all experiments was set at 50ms. Priming effects were determined by comparing target lexical decision times in the test condition to a control condition with unrelated primes that were neither morphologically, semantically or orthographically related to the targets. Experiment 1 tested Aorist inflection. The critical test primes were 3rd singular verb forms with the regular Aorist marker –Ar (e.g., sorar – ‘s/he asks’), and targets were the corresponding bare verbal stems (e.g., sor – ‘ask’). Experiment 2 had deadjectival nominalizations with the suffix –lIk (e.g., hastalık – ‘illness’) as test primes and corresponding bare adjectives (e.g., hasta – ‘ill’) as targets. Experiment 3 examined the role of purely formal (orthographic/phonological) priming effects. The critical prime-target pairs were morphologically and semantically unrelated, but were similar to those in experiments 1 and 2 in terms of their formal (orthographic/phonological) overlap (e.g., hapis ‘jail’ hap ‘pill’). Experiment 4 tested for potential semantic priming effects. The design, including an SOA of 50ms, was parallel to experiments 1 and 2, but this time the critical condition was made up of orthographically and morphologically unrelated but semantically highly related prime-target pairs (e.g., çabuk ‘quick’ acele ‘fast’), as determined by an offline synonymy pretest. Results. (1) For native speakers of Turkish, experiments 1, 2 and 4 yielded significantly shorter target response times in the test than the unrelated conditions, whereas experiment 3 did not show any priming effects. These results indicate that morphological priming effects can be dissociated from priming effects due to formal overlap. (2) For the L2 group, we obtained priming effects for –lIk derivations in experiment 2, but not for inflected Aorist verb forms in experiment 1. Moreover, experiments 3 and 4 did not reveal any formal or semantic priming effects in the L2 group. The discussion will focus on the L2 results. We argue that morpho-orthographic decomposition (independent of formal and semantic relatedness) can be found in L2 processing, but only for derived word forms. Inflected words, on the other hand, even fully transparent and productive ones, did not show any effects of morphological structure and appear to be processed differently in a non-native language.