Thanks Rex and Brian for your feedback. I'm sure that there are many alternatives to the one that I recommended, and they should probably all be tried within certain cost and security parameters. In my opinion, the most important takeaway is the way that the brain processes external data, and what needs to be done to break apart and rebuild our internal models.

The other issue, which I didn't specifically call out in the essay, but perhaps I should, is that of translation. Accurate translation is frequently dependent on cultural context. That's why native speakers are so valuable versus American graduates from the Defense Language Institute. If you can't recruit a native speaker, then perhaps some "cultural context" can be instilled into Americans learning Arabic, Chinese, etc. through various innovative ways, such as we've discussed in this thread.