Entirely confused though because I do not know what the determinant of elementary matrices are...
You don't even need to know those elementary matrices. You just need to know how elementary row operations affect the determinant. In this case, we need all three types of operations, and I write the effect in the parentheses behind.
- Multiply a row by a non-zero number. (determinant multiplied by this number)
- Interchange two rows. (determinant multiplied by $-1$)
- Add a scalar multiple of one row to another row. (determinant unchanged)
It's obvious that we can multiply the first row of $A$ by $\frac{1}{4}$ and add to the second row (operation type 3) to get $(d\ e\ f)$. Then multiplying this row by $\frac{1}{2}$ and adding to the second row (type 3) gives us a clean row $(a\ b\ c)$. Now we have a matrix that can be related to $A$ by elementary row operations and have the same determinant as $A$:
$$
\begin{pmatrix}
4g & 4h & 4i\\
d & e & f\\
a & b & c
\end{pmatrix}
$$
Divide the first row by 4 (type 1) and interchange the first and the second last row (type 2), we get the original matrix whose determinant is known to be $2$. Since we know consequences of three types of operation, it's easy to conclude that
$$
\det (A) = -4\times 2=-8
$$
P.S.
I personally don't remember those matrices performing elementary row operations as matrix multiplications. And I don't understand the necessity of working out those operation matrces (or the overall matrix $C$ in @Dan's answer). Doing matrix multiplication is always tedious, after all. As I have said, observing that $A$ can be related to the original matrix by row operations suffice to give the answer.