Achernar: Um, I believe you’re thinking about the coefficient of determination (R2) which is the square of the correlation coefficient.
In your example a correlation coefficient of .33 implies that one variable can explain about 11% of the variance.
.33^2=.11
Not sqrt(.33) (=.57)
FWIW, .33 might indicate a moderate effect, except that I’m not sure what variables are being studied here.