Hi Stephen,
The “core” of this problem is that R2 is learning the 1.1.1.0/24 route through R4, something that should never happen since it’s a route that went from RIP > OSPF > RIP. We call it a metric redistribution issue since RIP doesn’t have any way to differentiate between an internal or external route…the lower metric from R4 wins. This really is a routing loop which is bad 
Once R2 installs the route through R4, it will no longer advertise 1.1.1.0/24 to R4 so the only way for R4 to get to 1.1.1.0/24 is R3…keep in mind that this is not much of an issue…it’s sub-optimal routing but not a routing loop 
The only way to fix this sub-optimal routing problem so that R4 uses R2 as the next hop is to make sure that:
* R2 is able to advertise 1.1.1.0/24 to R4.
* R4 doesn’t install an OSPF route for 1.1.1.0/24.
You can achieve this with redistribution filtering or offset-lists.