PS: All the documents that I have translated so far are handwritten ones and I think there is not a software in the world that can possibly do the translation of a handwritten Arabic document.
Actually it might be easier than doing it to most European languages. Looking at the documents, it appears that even in handwritten documents, the characters are distinct. More like hand printed English (or other Latin alphabet language) rather than a written (i.e. cursive) text using the Latin alphabet. This would Kiley make optical character recognition easier. Then it's a matter of parsing the characters into words and then translating the words. I don't know how hard it is to translate Arabic words and sentences into English words and sentences. I'm sure the result of such a machine translation would be rough, but it would certainly be good enough to then employ English speakers to screen the documents. Or you could translate to Spanish, which is probably closer in structure to Arabic than other European languages, and use the many Spanish speakers we have in the US to do the screening. Maybe even translate it to Hebrew, which is probably much closer to Arabic in structure, and let a bunch of Jews and other Hebrew readers translate it. There would be a certain poetic justice in that, don't you think?
In regards to translators with basic language skills doing the translation, it is possible that they can translate a large % of the documents but it will take them much more time to do so than a person with native Arabic language and good English language skills.