Question about the delayed choice quantum eraser. 2nd order interference possible?

[al onestone]

OK, a quick intro to the delayed choice quantum eraser is at wikipedia. I have attached a figure of the modified DCQE. In this setup there is no delay, there is no choice, and there is recombination of the idlers instead.
In the modified DCQE the signal beams are identically prepared as that in the DCQE. But the idler beams are prepared differently from the DCQE. Whereas in the DCQE the idlers are subject to either of detection or recombination, in the modified DCQE the idlers are definitely combined at an identical optical pathlength from the source, and the pathlength of the idlers is also identical to the pathlength of the signals. So all four optical paths to the two detectors from the source differ by no more than a coherence length(of the signal and idler).
In this preparation do the detectors display interference(simple second order indistinguishability interference with pathlength difference/phase)? Of course you could get a fourth order coincidence effect, but do you also get second order interference at each detector? (much like in the ZWM 1991 expt of Mandel et. all)
If an attenuator is placed in the upper idler pathway, in which case the distinguishing “which-path” information is implied, then is the interference at both detectors is negated?