Could language analysis tools detect lone wolf terrorists before they act?