In Mixtec, wh-words move to the specifier of CP, and wh-possessors move to
the specifier of the possessive DP when they pied-pipe their possessum.
Foci also move to the beginning of the clause, but they cannot front within
a DP. However, both wh-words and foci can subextract out of possessive DPs,
often argued to precede via successive cyclic movement through spec-DP
(Gavruseva 2000). Thus, foci seemingly *can* and *cannot* move to the edge
of DP depending on the derivation. Based on this contrast, I advance a
novel account of syntactic focus movement in the language: movement targets
a class of focus-sensitive particles (that can be overtly realized or
segmentally null), not foci themselves (cf. Cable 2010 on wh-movement).
Movement of phrases headed by these particles will also move any focus that
they c-command. These particles can be attracted to spec-DP (like
wh-words), but semantic foci cannot, as they bear no formal movement
feature. Thus, there is no motivation for foci to front within a pied-piped
constituent.