I know this is probably a simple fix for you experienced users, but how do you get an echo on the last word of a sung line? Short of splicing and dicing, can the word be isolated so a delay effect can be used and not affect the rest of the phrase?

Well, I have not done it, but I suppose, have you heard of highlighting?

No. Highlighting doesn't work. How could it, if the FX is in the FX bin through which the whole track goes?
You simply cut the clip separate at the desired place and insert a clip FX in it.

Or you can put the FX in a send bus and automate the volume, but I think clip FX is easier.

Right, I've tried the highlighting and it doesn't work. So, Kalle, if I understand you correctly, I should split the clip and then I'll be able to insert the delay effect on one of the split clips? I'll give it a shot. Thank you.

Crossing at the zero-point line, and after cutting the clip separate you can highlight all the clips in that track again and combine it back together. I know combining it back together can be done, but I still fail to see if you are Editing and marking where you want the FX to occur that Sonar would not just do the part that is highlighted, perhaps not, but it should if it does not perhaps.

I draw the send level in track view. 

copy/paste a few times. insert clip envelope. adjust to taste.

Just to add:

After splicing, sometimes you may have to bounce the spliced part first before applying effects, incase it doesn't work the way you expect it to do so.

Just do what Kalle said - SPLIT the last word of the phrase into it's own clip. Insert a Delay in the CLIP FX Bin, adjust to taste.
If you're using the Sonitus Delay (a good idea), fiddle with the mix control to get the dry/wet balance right.

Have a 128 levels of "do" and "undo" until "redo" works.