Templatic morphology in the Portuguese verb.

It is argued here that the unmarked forms of the Portuguese verb can be accounted for by assuming a templatic constraint on inflectional morphology. The argument is built on root mid vowel umlaut (e.g., firo / feres, durmo / dormes), and independently supported by stress placement and suffixal morphology. Neither rules nor morphological specifications are required in the present approach, which is shown to be preferable to previous rulebased accounts of the facts in terms of empirical adequacy, formal simplicity and naturalness, and explanatory power. Thus, this paper is intended to provide theoretical and empirical evidence for replacing opaque and idiosyncratic ‘lexical’ rules with structural / phonological and universal explanations. Specifically, Portuguese umlaut implies association of three vocalic melodies with a dissyllabic template, and the templatic basis of lexical umlaut is assumed to be universal. A second point at issue here is that a basic part of Portuguese verbal morphology can be naturally accounted for thanks to a rather unusual concept in Romance linguistics, which is however familiar to Semitists. Comparatively to languages such as Arabic or Hebrew, it is clear that the verbal template has specific characteristics in Portuguese morphology for at least two reasons. Firstly, EP verb system remains concatenative insofar as there are here no consonant-based roots opposed to vowel-based morphemes. Thus, a ‘templatic’ language is not necessarily ‘non-concatenative’. Secondly, while Semitic binyanim form rich paradigms, there is only one template in EP verb system. Thereby, it is both semantically ‘poor’ and morphologically widespread: if there is only one template in EP, then it shows the unmarked CVCV-pattern dictated by stress placement rules; hence, this unique binyan affects the unmarked and most frequent forms of the verb.