So apparently the syntax of CTL does not allow boolean combinations of path formulas. I'm not even sure I fully understand this sentence. I think it states something like "in CTL, we cannot express a logic statement using the universal or existential quantifier such that the statement holds for every path in a model".
My question is: How can I rewrite a formula which is valid in CTL* in terms of CTL? E.g.: $ A[Fp \rightarrow Xq]$
This expression says: "For all paths in my model, if there is a future state p, then the next state will be q."
In CTL, I would write this as: $EFp \rightarrow AXq$
This translates to "If there is some future state p, then for all states in which p is true, the next states will be q."
Indeed, in CTL every quantifier must be followed by a path formula and every path formula operation must be quantified.
Your second expression is vaild in CTL as $E$ quantifies $F$ and $A$ quantifies $X$.
However, they are not the same. In the CTL* formula only paths that eventually lead to $p$ are required to have have $q$ at the next state, while in the CTL formula you say if $p$ is reachable, all possible following states must be $q$.
You can rephrase the CTL* statement (paths that eventually lead to $p$ are required to have have $q$ at the next state) with negation like this: "For all following states, if the next state is NOT $q$, then from that state on we must never reach $p$. Or as CTL:
$AX(\neg q \rightarrow AG(\neg p))$