Since you already know the population parameter is normally distributed, then it doesn't matter that the sample is relatively small, so the CLT applies. If the parameter was not normal, then you might need a larger sample to justify its use.
The calculation then is straightforward: the mean of the sample distribution is equal to the mean of the population (2.3), and the standard error of the means is equal to the st. dev. of the population divided by the square root of the sample size, or
0.9 / √20 = 0.20