Non-agreement on Romance disagreements

Abstract

i. Frans Plank (I984), in the context of discussion of possible universal constraints on linguistic rules, wonders whether phonology can 'interfere' with syntax: he adduces examples of 'Romance disagreement' as an illustration that the operation of syntactic rules can be influenced by phonological factors. Here I shall not challenge his general hypothesis, but merely claim that his examples do not sustain it. He admits that his case rests on the assumption that his examples show DIsagreement, rather than NON-agreement (34I), thus placing them in the realm of syntax, rather than morphology (345-346). So it should be sufficient to demonstrate that the apparently disagreeing elements are phonologically conditioned non-agreeing allomorphs a view Plank himself examines and rejects, because it would involve an implausible degree of' accidental homonymy'. At the risk of being accused of 'ad hoc-ery', I shall seek to show that idiosyncratic features of the languages examined have, plausibly, contributed to the development of such homonymy. Of the three Romance examples Plank discusses, two are to do with apparent gender disagreement of noun determiners (Spanish el before feminine nouns beginning with initial a; French mon, ton, son2 before vowel-initial feminine nouns or preposed adjectives). The other concerns pronominal cross-reference in Barcelona Catalan, where 'inanimate' clitic pronoun hi replaces its 'animate' counterpart li, when conjoined with another pronoun beginning with 1. Here the question of clitic ordering is also involved (Posner, I980: I83), and, in any case, Plank's statement of the facts is not comprehensive enough; as in French and Spanish, the standard language displays anomalies that are absent from more colloquial usage.3 It is