Why can't I just use my voice?

Clicker training involves shaping behavior in small steps, identifying the behavior, as it occurs, with some kind of marker signal. But couldn't you just use a word, like "good," or "yes," as a marker signal?