Aperçus de morphologie du français [Insights into French morphology] (review)

This book is a collaboration between colleagues who have a dual interest in providing a coherent description of French word formation and developing a theoretical framework in which to present that description. The book is a series of essays, each of which has a theoretical point to make as well as the aim of describing some recalcitrant bit of French morphology. The latter point is important: these papers are deliberately targeting the description of the more awkward bits of French morphology, not the most central areas, and they are making generalizations based on the findings from these more marginal parts of the field.

The first chapter is an introduction by the editors. They position the work of the volume in the framework of later twentieth-century developments of the study of morphology, from Chomsky 1970 to Corbin 1987 and optimality theory (OT). In OT it is the construction of a particular output form that is primary, while in the earlier models it is the use of particular affixation processes according to well-defined rules. The editors provide a brief summary of the papers in the volume, commenting on how they fit into the various frameworks.

In 'Morphologie grammaticale et extragrammaticale' ['Grammatical and extragrammatical morphology'] (21-45), BERNARD FRADIN, FABIO MONTERMINI, and MARC PLÉNAT try to illustrate extragrammatical morphology by two processes in French. The first is the jargon known as verlan. The name verlan comes from the French expression l'anvers 'backwards', but just what is backwards (and how backwards it is) is determined by the syllable structure of the French word that is processed into verlan. In the word verlan, syllables are inverted, but in three-syllable words, any of the original syllables may be placed in first position, and in single-syllable words, the individual phonemes are reversed. The claim made here, but not illustrated in enough detail for it to be clear how far it works, is that once a particular starting element has been chosen for the processed word, the rest of the form falls out naturally because of IDENTITY, MAXIMALITY, CONTIGUITY, and other constraints, and through principles of minimum markedness.

The authors then turn their attention to blends. The data they work with are interesting for the English reader, in that they include several examples where one of the blended elements ends up apparently infixed into the other, something that is rare in English, for examples, ridicoculiser < ridiculiser 'to ridicule' + cocu 'cuckold', s'embellemerder < s'emmerder 'to be bored' + bellemère 'mother-in-law'. There are also other patterns that sound unnatural for the English speaker. The authors appear to want to claim that these are not created by regular morphological processes, while at the same time wanting to claim that the forms are-on the whole at least-totally predictable. Part of the reason they want to make the first claim is that they see these forms as being deliberately created, and they follow Schultink 1961 in believing that conscious creation cannot be productive. As has been pointed out by Bauer 2001 and others, however, there is no way to operationalize 'consciousness' in this context, so that predictability of form would seem to be a better guideline. Again, their discussion is unfortunately not explicit enough to show precisely how they see the constraints applying to blends to create the forms, but they see a constraint of MAXIMALITY, along with another maximizing the amount of deletion of the overlapping forms of the words blended. They note, perceptively, that there seems to be a demand that at least two syllables of both base words must be present to allow identification of the bases (38), and thereby introduce a psycholinguistic notion of recognizability into the study of blends. This factor is usually ignored, though I am convinced of its importance, but in purely formal terms it seems to make...