Based on quite substantial evidence, I would actually recommend AGAINST this program. Although I am NOT an expert, I will give these reasons for you to explore (if you wish). Feel free to ask for clarification on any of these reasons:
1. From what I have heard (and you may want to confirm this), Rosetta Stone is great at market capitalization, but has had a net loss of profits throughout every quarter (at least in the last few years). So, even with their Department of Defense and other contracts, they still have not sold enough programs to be profitable without their aggressive investor support. A normal company (such as Berlitz; or even a university course) would need to have more satisfied customers to be able to make ends meet. This net loss of profits indicates that they aren't the best.
2. In their "immersion" method, you use a set of four pictures followed by their words and pronunciation. However, look at the comparison between two or more languages, they have the exact same methodology; in reality, different languages have different declensions and ways of classifying.
For instance, one of the first screens has a man, woman, boy, and girl. In Latin, that would be vir, femina, puer, puella. Spanish would be hombre, mujer, chico, chica. Arabic would be rajyul, immarra, walad, bint (I hope I have that right). However, with pronouncing "the", it would be the man, the woman, the boy, the girl in English; vir, femina, puer, puella in Latin; el hombre, la mujer, el chico, la chica. Il-rajyul, il-immarra, il-walad, il-bint. So, as you can see, the "compartmentalized system" doesn't carry over from language to language.
3. Rosetta Stone will not tell you the "cultural loads" of some of the words. Each culture has so many different mannerisms and modes that a non-native speaker cannot teach you. For instance, both you and I can usually tell when someone is not a native English speaker. Even if their English is generally sound and makes "mathematical sense", it is not close enough (e.g. when someone says "where IS the tables; versus where ARE the tables"). All the little parallelisms, idioms etc. will not be made clear with Rosetta Stone.
4. These computer programs do not teach you how to write out the pronunciation or use word to type in the respective characteristics. To my memory, the Rosetta Stone program never says "Arabic reads right to left", here is an initial, medial, ending, type of letter etc. How frustrating would that be to try and learn?
5. There are free sources of information already that provide comparable knowledge; e.g. http://en.wikipedia.org/wiki/Arabic
(look at the links on the bottom of the page).
6. The foreign language service institute put the following estimate of how long it takes for one of their people (who are significantly motivated, have decent language learning ability, etc.).
Arabic and Asiatic Languages supposedly take 88 weeks/2200 class hours.
Can you imagine having the discipline and patience to spend that time in front of a computer?
For all the above reasons, I would say to invest either in a government program (like the Department of Defense, Army, Navy, or Air Force linguistic program, etc), or an official Berlitz Program; http://www.berlitz.com/.
The most intensive Berlitz program costs around $1000 a week; but, they supposedly guarantee proficiency within 12 weeks (I would check that statistic out; I may not be entirely correct).
Yes, these are big investments. But, translators overseas (as in Iraq and Afghanistan), can make as much as $225,000 a year (I read that in the LA Times).
I would do plenty of research, try to secure any job opportunities or anything you'd gain as a result of taking the language, talk to a few college professors that teach the subject you want to learn, take up an intensive program or possibly study abroad, etc. (I've used Arabic as an example because that's a "hot language" right now). But, I hope you see the illustration I'm trying to make.
Feel free to email me if you have any more questions; I'm always happy to discuss (and, I've seen your previous posts, I'd like to chat about some of those out of curiosity).