Patterns of 1st and 3rd person marking in Oil-Galloromance New insights into an old problem

This article is a contribution to the long-standing discussion of subject marking in Romance. Its originality lies in its systematically considering data from Oil-Galloromance dialects, i.e. non-pro-drop varieties, which had been ignored because they were thought to pattern like French. On the contrary, a detailed survey of the means of 1SG and 3SG.m. Marking in these dialects reveals that the obligatoriness of the subject clitics in all grammatical persons does not guarantee the absence of ambiguous marks, since cases of syncretism between these two persons were found, besides cases of marking even more redundant than in French. I then conclude that it is yet another refutation of the now generally abandoned wisdom according to which the subject pronouns exactly compensate the loss of verb endings. Moreover, the results make the pro-drop parameter and parametric theory hard to maintain, as has been observed from other microvariational studies. I also argue against a functionalist interpretation of the correlation between the different means of subject marking based on the assumption of avoidance and repair strategies underlying language change/dialectal fragmentation. My own analysis then relies on the assumption of a strong and stable typological property of accusative languages like Romance, called here the principle of recovery of the subject. The surface microvariation within (Oil)-(Gallo) romance is simply seen as the result of non-deterministic properties of language change/dialectal fragmentation.