1) the sample mean of the 33 scores
2) the sample standard deviation of the 33 scores.
Then you convert each standard score, 1 by 1: (standard score - sample mean) / (sample standard deviation).
In theory, 15.87% of standard scores (from a normal distribution) would have a z-score greater than 1. So that cumulatively 84.13% have a standard score less than 1. Another 15.87% would have a z-score less than -1. So 68.26% would have a z-score between -1 and 1 (i.e. absolute value less than 1). I got these numbers from looking at a standard normal table.
You have to read the "fine print" on the table, to see what probabilities are given in the table, e.g. some tables give the "area under the curve to the right of z", others give the "area under the curve to the left of z", and there's other possibilities.