Also, the first word in the array won't be directly meaningful, because this is an array that represents a whole keyboard layout (so the first element is "a", the second "s", etc). The first real "word" is actually "and", which is not an interesting word. And it's first-ness is even less interesting because the order in this array is determined by the keys you press to select an item, not the frequency of that word ("a" has a low bit value, so "and" comes before "the", another uninteresting but very frequent word).

But, in later sections of the array (where you are pressing 5+ keys at a time to produce a longer word), you can choose from nearly any english word, and so it makes sense to choose ones that you use frequently, and so it makes sense to analyze a corpus of text to find the frequent words. The words ultimately chosen lead me to believe it was the developer's personal emails that were scanned.