Indeed, what you predicted is what actually happened. I think Niomi just made an error. The rule for complete utilization of the channel is for each party to talk for the one-way time (15s), and then listen for the same amount of time. The results may take some getting used to, because what you hear when you listen came from someone who hasn't yet heard what you just said.
Or the cartoonist made an error, and is unsuccessfully trying to present the case where the child starts responding as soon as the child hears the parent speak.