The phonology of -ize derivatives

Introduction

Having outlined the semantic properties of -ize derivatives, we will now focus on the phonological restrictions operating on these forms. It has been observed that verbs derived by the suffixation of -ize feature some peculiar and apparently variable phonological properties. Thus base-final segments are often deleted (as in emphasis - emphasize), base-final consonants may change (as in Celtic - Celticize), vowels may alternate (as in géntìle - géntilìze¹), consonants appear to be inserted between the base and the suffix (as in stigma - stigmatize) and even stress may be shifted (as in catholic - cathólicìze) or reduced (as in (fèrtile - fertilize)). To complicate matters further, the application of these phonological manipulations of the base word seem to be rather unpredictable. Some words are affected, others are not for no obvious reason. Apart from these allomorphy problems, one has to account for the fact that the possibility to attach -ize to a given base word seems to depend on certain phonological properties of the base. In particular, monosyllables and words with final primary stress almost never appear as stems in -ize formations.

Although these phenomena are well-known, it has remained unclear which mechanisms can best account for the phonological alternations and the phonological productivity restrictions of -ize words. Basically, two kinds of models have been proposed, one based on the application of phonological rules (e.g. Gussmann 1987, Kettemann 1988), the other on well-formedness conditions (Raffelsiefen 1996). Hence, on the theoretical level, the debate on the phonology of -ize derivatives reflects the more general controversy about derivational versus representational theories in linguistics. I will argue here that the phonological behavior of derived verbs can best be accounted for if we assume a number of violable output constraints as in Optimality Theory (henceforth OT) instead of input restrictions and/or morpho-phonological rules.

There are cursory remarks on the phonology of -ize in numerous linguistic sources but only very few in-depth studies of the phonological proper ties of -ize derivatives (Gussmann 1987, E. Schneider 1987, Kettemann 1988, and Raffelsiefen 1996).² These studies will serve as a reference point for the account to be presented below. It will be shown that the earlier attempts are flawed in many respects and that they should be replaced by an account according to which the phonological well-formedness of potential -ize formations crucially depends on the prosodic structure of the derivative. I will also argue for violable constraints of the type proposed in OT and show that non-violable well-formedness conditions as put forward, for example, by Neef (1996), are less adequate.

¹ Following established conventions, I use acute accent to mark primary stress, grave accent to mark secondary stress.

² There is one notable account of the stress pattern of -ize derivatives, namely Danielsson (1948). This historically oriented study presents a wealth of data, but leaves the reader with the impression that hardly any generalizations can be drawn. Danielsson lists a whole range of attested stress patterns, but it remains unclear what, if anything, conditions the diverging patterns.