Have you considered playing a silence for the duration of a pause, like you do between utterances? Previous answer: https://stackoverflow.com/a/11079302/597849
Since playSilence() is asynchronous you don't want to just loop a bunch of times since it won't wait for the previous utterance to finish. You can use the tts progress listener to know when to start a new clip (or silence).
Another poster has also suggested writing the TTS output to a file, the playback of which can be paused.