I hope I can clarify things a bit. The way I have come to see it is that speaking outloud creates a feedback loop, which is good. Wenger talks about this in his book. But the feedback that is important with a partner, or with a taperecorder, is that you can either talk to your partner, or play back your tape to find out what you said. If you are really in a stream of consciousness, you probably won't remember what you were describing, and so you would miss any answers to your questions that you are using your image stream to answer.
Now if you are just doing it to improve your general intellegence, or work on visualization skills, I'm not sure about the need for a taperecorder or partner, but it still is helpful to be able to review your descriptions and seeing where you might want to improve the way you do it. Like giving more descriptions of non-visual senses, like taste or smell that would complete the experience.
I do not have extensive experience with image streaming, so if anyone that has been doing this for awhile would like to comment, please do.