The phenomenon of second position cliticization remains a mystery in many of its aspects, one of which is the timing of clitic movement. In this talk, I focus on Serbo-Croatian and argue that clitics move to the second position at PF. A new argument for this approach comes from the interaction of second position cliticization and ellipsis: a study has shown that ellipsis bleeds clitic movement to the second position, as in (1).
(1) Mi ih nismo videli, a oni su rekli da (*ih) jesu <videli (ih)>.
we them aren't saw but they are said that them are saw them
“We didn’t see them, but they said that they did.”
The bleeding effect can be easily accounted for in terms of the relative timing of ellipsis and second position cliticization: ellipsis precedes clitic movement. Two questions arise: (i) what exactly blocks clitic movement to the second position and (ii) why there is no violation of the second position requirement if the clitic ihin (1) stays inside the VP? I answer both questions by suggesting that putative clitics inside the ellipsis site never receive their clitic status.
I argue that the choice between a clitic and an accented form of a pronoun or an auxiliary is made no earlier than at the stage of Vocabulary insertion, and the clitic form comes with the requirement to appear in the second position. If ellipsis is the non-insertion of vocabulary items, a pronoun inside the ellipsis side would never become a clitic and would have no motivation to move to the second position.
In sum, there are no second position clitics in syntax, and clitic movement occurs after Vocabulary Insertion, i.e. at PF.