How to prove Bossert inequality for the symbol error probability?

In the book : Coding theory-Algorithms, Architectures and Applications p.20, the authors have introduced the following inequality ( and precised that it was taken from Bossert’s Channel Coding for telecommunications ):

$$frac {P_{err}}{k} leq P_{Sym} leq P_{err}$$

Where $P_{err}$ is the word error probability and $P_{Sym}$ is the average symbol error probability.

While the rightmost inequality is intuitive, I cannot fathom the left side.