Are prefixed units processed and represented like suffixed ones? Towards a hybrid model of morphological processing

Hélène Giraudo (1) , Madeleine Voga (2)

1 CLLE-ERSS - Cognition, Langues, Langage, Ergonomie
2 LLACS - Langues, Littératures, Arts et Cultures des Suds

The fact that in most languages affixed words are present in a very high proportion leads to the conclusion that morphology constitutes an important variable in word processing. Thirty years of investigation permitted to confirm that morphology intervenes in automatic processes that operate during the very early stages of lexical access, suggesting that morphemes were independently coded or stored somewhere in the mental lexicon. The masked priming paradigm (Forster & Davis, 1984) is the privileged technique used by the psycholinguists to examine the early processes of word recognition. The principle governing this paradigm lies indeed on the transfer of activation from a first processed stimulus (the prime) on the recognition latency of a second stimulus (the target). This activation transfer is admitted to operate on the basis of the shared representations (orthographic/phonological/morphological/semantic) by prime-target pairs. Moreover, given that the prime is presented very briefly (SOAs under 60 ms) and is generally masked (by a string of hash marks), any effect of the prime is considered to be the result of unconscious processes. In the precise case of morphology, many studies manipulated morphologically related words as well as pseudo-words and found systematically very robust positive priming effects: two morphologically related words prime each other across different languages (e.g., Boudelaa & Marslen-Wilson, 2005 in Arabic; Duntildeabeitia, Laka, Perea, & Carreiras, 2009 in Basque; Drews and Zwitserlood, 1995, in both German and Dutch; Frost, Deutsch & Forster, 1997 in Hebrew; Giraudo & Grainger, 2000 in French; Rastle, Davis, Marslen-Wilson & Tyler, 2000 in English) and in experimental settings that include multiple control priming conditions (unrelated but also orthographic/phonological and semantic controls in order to neutralize any interference effect). This general result being established, the question of the nature of morphemic units represented in longterm memory and their precise role within the lexicon remains unanswered. Two possible hypotheses of representation have been proposed: either

Domaines: Psychologie

Hélène Giraudo, Madeleine Voga. Are prefixed units processed and represented like suffixed ones? Towards a hybrid model of morphological processing. 7èmes Décembrettes, International Morphology Conferenc, 2010, Toulouse, France. ⟨hal-00986176⟩