The phonological singularity of the French language

French phonology presents two well-known and intensively studied phenomena, viz. schwa (the traditional 'mute e') and latent consonants (LC). Current analyses of these phenomena involve multitiered representations where latent consonants are floating segments and schwa empty (vocalic) positions. It is argued that such an approach runs into both conceptual and empirical problems. Floating segments are supposed to lack a position, but they must nevertheless remain linerarized with respect to other segments. Deficient segments (both floating contents and empty positions) typically remain unrealized as a consequence (so it may seem) of not being syllabified. Such segment could then be considered 'inert' in the sense that they may take part in syllabification but need not do so. This is however hardly a tenable analysis. Interior schwas do tend to remain silent in cases syllabic structure can be built without including them into it. But on the other hand, final morphological schwas (event when finally silent) crucially need to combine with latent consonants in order to awake them. Generally speaking, combinations of supposedly inert segments ('+LC or LC+') do not remain silent altogether. It is suggested that a possible way out of these conceptual and empirical difficulties implies departing from traditional views concerning linearity and constituent recursivity. Phonological information in morpheme, it is argued, is not prelinearized and may be recursive to certain extent, in particular a C' syllable may function as a coda. Under this latter hypothesis it is possible to account for the phenomenology of schwa through the assumption of a plain straighforward participation in the syllabification process.