Degrees of freedom of a factor predictor

With an intercept present, a categorical predictor having $m$ represented levels adds $m-1$ linearly independent indicator columns and therefore $m-1$ model degrees of freedom.