They could go the BG2/Project Eternity way... have only one or 2 lines VO'ed at the start (say, the welcome) then the rest text. And main NPC's in cutscene's voiced.
Of course, that will be costly, and I totally don't mind. Again; I think some appropriate city ambiance would add a lot to this without costing as much as VO.
Create the initial phrases for each dialogue is difficult and expensive at this stage of the game, and the main thing is not compatible with the intention to expand the game using mods.
For example, a user modes will be deprived of these phrases and it will be immediately strike the eye (or strike the hearing).
One must also not forget that the translation of the first sentence of each dialogue in all languages ​​will be very silly and not justified.
I think the way to go for "The Sims" is much more logical to their ridiculous babble.