When a grammatical phenomenon is singled out, the kaleidoscopic degree of variation exhibited by Romance varieties can often be reduced to few systematic patterns. As for clitic placement, for instance, two main patterns are attested :
The western Ibero-Romance system, in which enclisis occurs in finite clauses lacking “proclisis triggers”, i.e. focus/wh- elements, quantifiers, etc. ; this pattern is akin to the one exhibited by medieval Romance languages, ruled by the so-called Tobler-Mussafia law.
The general Romance system, in which all finite clauses exhibit proclisis, while enclisis is confined to (positive) imperatives and non-finite clauses.
Besides these two major patterns, several peculiar systems have emerged. For instance, in the dialect of San Valentino (south-eastern Italy ; Benincà & Pescarini 2014) clitics can be placed before/after the finite verb, see (1) and (2), and, in compound tenses, before/after the past participle, see (3). I will show that the condition ruling anclisis/proclisis alternations are partly reminiscent of those holding in Tobler-Mussafia systems.
(1) a. mə lu ’maɲɲə ŋgə lə ’mejnə. To.myself= it= I.eat with the hands b. ’maɲɲə=me=lu ŋgə lə ’mejnə. I.eat=to.myself=it with the hands ‘I eat it with my hands’
(2) a. mə l ’ajə məɲ’ɲɐtə to.me= it= I.have eaten b. ’ajə mə lu məɲ’ɲɐtə I.have =to.me =it eaten ‘I have eaten it’
(3) a. ’ajə dʤa məɲ’ɲɐtəməlu ji I.have already eaten=to.me=it I b. ’ajə dʤa mə lu məɲ’ɲɐtə I.have already to.me= it= eaten ‘I have already eaten it’
To the best of my knowledge, Sanvalentinese is the only variety in which proclisis to the past participle is allowed. Conversely, enclisis to the past participle is attested in several Romance areas : Franco-Provençal (Chenal 1986), Piedmontese (Parry 2005 a.o.), Dolomitic Ladin (Rasom 2008), Romanian (limited to the accusative feminine clitic o).
(4) L’ an tot portà-lèi vià. (Chenal 1986:340) They= have everything carried=to.him away ‘They have taken everything away from him.’
(5) I an rangiò-la. (Cairo Montenotte, Parry 2005) They= have fixed=it.F ‘They fixed it.’
(6) Am mâncat-o (Romanian, Dragomirescu 2013 : 193) I.have eaten-it.F ‘I ate it’
Parry 1995, 2005 pointed out that, in origin, Piedmontese displayed proclisis in all finite clauses, while enclisis in compound tenses resulted from a process of clitic copying whereby two instances of the object clitic co-occur in the same clause, see (7a). Successively, the proclitic copy eventually disappeared, giving rise to a pattern of generalised enclisis, as shown in (7)b.
(7) a. a l peul di-lo (18th century Piedm., Parry 1998 : 108) (S)he= it= can say=it b. a *(l) peul di-lo (present day Piedm.) (S)he= it= can say=it ‘(S)he can say’
Moreover, in a narrow set of eastern Piedmontese dialects, enclisis is mandatory with simple tenses as well, see (8) (Tortora 2015).
(8) I porti-la I= bring=it.F ‘I am bringing it’
The talk aims at illustrating the above data, by focusing on the conditions ruling the distribution of enclisis and proclisis. Secondly, abstracting away from the intricacies of clitic placement, I aim to discuss how (micro)variation can be modelled within a principled theory of grammatical competence.