Thanks for all your interesting answers. However I still feel a bit overwhelmed by the masses of possible orderings of these effects. Maybe it's easier if I break it up into smaller questions...
1) Say I add reverb and delay through a 'send', is it the norm to add the reverb to the post-delayed signal (i.e. the original vox plus added delayed vox) or just the original signal? Surely there's a 'normal procedure' here?
Typically I use reverb(s) n a mix on Aux tracks, fed via sends (assigned to busses in Logic; naming busses and Auxes is highly recommended IMO). Believe it or not, unless an instrument is using the delay as a part of a special effect, I will typically do that on an Aux, and send to it as well. Why? Allows me to isolate ANY effect (delay, reverb, chorus, etc), send THAT effect to another (such as sending the delay Aux to the reverb), and also allows me to mix the effected signal and dry signal to taste, and also allows me to buss several instruments to that one effect, if desired.
Another example is doing the "parallel drum mix" thing, and iinstead of outputting to two Auxes, I will output to one (main drum submix), and then use sends on the drums I want to slam to the second drum sub. I might even do a third sub, high pass it, and add some overdrive. As has been said- SOOOO many choices, and with DAWs, the sky is nearly the limit now.
2) Say I have enhanced the vocal take with a gentle distortion or exciter, should this 'wet' signal be fed to the reverb or is it generally the case that reverb is generated from the original dry signal?
I would go mix by mix. I might create a copy of the vocal track, leave the one dry(ish), and then use the second to beat the pee out of with effects like distortion and stuff inline. This way, I could then control how much of which vocal (dry or effected, or both) goes to which Aux with whatever effect(s) I have on them. I might want the distorted vocal to go to a reverb with a high pass, EQ, and dynamics to give it a far, far away sound. Then send to the main reverb as well (less signal), and where ever else i want. With the number of tracks we can do now, and so long as signals are verified to be in phase (delay compensation, which Logic has), copying tracks is easily handled, and an excellent idea. Freeze em, print em, whatever even. Flexability is king, and we have the facilities for it!
3) On a slightly different note - would you ever use an expander on a vocal? If so why? Would you use it on top of compression? Seems strange as they are working in opposite directions....
Not normally, but ya never know! Experiment, have fun. Isolate something, and slowly play.
4) Does Logic allow me to adjust the order of the inserts once they have been put in place? E.g. I have compression at the top of the list and I need to move it to second place... How?
Yes. Hover the mouse, and then Control or Command (cannot recall which offhand, sorry) and the mouse will change to a hand. Drag the insert up or down, and drop. they will reorder automatically. Also, you can that plus Alt to copy an effect (including the settings) from one insert to another, same track or different track. Cool stuff!