The Simultaneous Effects of Inflectional Paradigms and Classes on Lexical Recognition: Evidence from Serbian Petar MILIN & Duˇ sica FILIPOVI ´ C DUR DEVI ´ C University of Novi Sad & University of Belgrade, Serbia Ferm´ ın MOSCOSO DEL PRADO MART ´ IN Aix–Marseille Universit´ e & C.N.R.S., Marseilles, France Draft of August 31, 2008 Abstract In this study, we investigate the relevance of inflectional paradigms and inflectional classes for lexical processing. We provide an information- theoretical measure of the divergence in the frequency distributions of two of the paradigms to which a word simultaneously belongs: the paradigm of the stem, and the more general paradigm of the nominal class in which the stem is embedded. We show that after controlling for other variables, this measure is positively correlated with response latencies and error counts in a visual lexical decision experiment in Serbian. We interpret these results as a trace of the simultaneous influence on lexical processing of both, the stem and the inflectional paradigms. Keywords: inflectional paradigm, inflectional class, frequency effect, infor- mation theory, visual lexical decision, Serbian Introduction An inflectional paradigm (IP) can be defined as the set of inflected variants that can be formed for a word by regular or predictable morphological transformations. IPs have long held a prominent role in the study of language as one of the fundamental building blocks of the lexicon (cf., Anderson, 1992; Aronoff, 1994; Blevins, 2003; De Saussure, 1966/1916; Kurylowicz, 1964; Wunderlich & Fabri, 1995; Zwicky, 1985, but for linguistic views that reject the relevance of IPs see also: Halle & Marantz, 1993; Lieber, 1992). A classical example of an IP is presented in Table 1, which shows the inflected variants for the Latin This work was partially supported by the Ministry of Science and Environmental Protection of the Republic of Serbia (grant number: 149039D). The third author was funded by the European Commission through a Marie Curie Individual Fellowship (MC–EIF–010318). We wish to thank R. Harald Baayen, James P. Blevins and Aleksandar Kosti´ c for their generous advice. Also, we thank Cristina Burani, Joseph Stemberger and one anonymous reviewer for their constructive criticism of an earlier version of this paper. Correspondence can be addressed to pmilin@ff.ns.ac.yu