NOTE: L'Hopital's and Taylor series not allowed!
Taking the log and exponenting the entire thing I get $$\lim_{n\rightarrow\infty}\left(\frac{1+\sqrt[n]{a}}{2}\right)^n=\lim_{n\rightarrow\infty}e^{n\ln\left(\frac{1+\sqrt[n]{a}}{2}\right)}=\lim_{n\rightarrow\infty}e^{n\ln\left(\frac{1+\sqrt[n]{a}}{2}-1+1\right)\cdot\frac{\frac{1+\sqrt[n]{a}}{2}-1}{\frac{1+\sqrt[n]{a}}{2}-1}}.$$
Letting $k=\frac{1+\sqrt[n]{a}}{2}-1$ we see that the RHS can be simplified to
$$\lim_{n\rightarrow\infty}e^{nk\frac{\ln\left(k+1\right)}{k}}.$$
It follows that $k\rightarrow0$ as $n\rightarrow\infty,$ and $\ln(k+1)/k$ then tends to $1$ (standard limit). So we can write
$$\lim_{n\rightarrow\infty}(e^{k})^n=\lim_{n\rightarrow\infty}\left(\frac{e^{k}-1+1}{k}\cdot k\right)^n=\lim_{n\rightarrow\infty}\left( \frac{e^k-1}{k}\cdot k+1\right)^n=(1 \cdot0+1)^n=1.$$
The answer should be $\sqrt{a}.$ Why is my method wrong?