It's National Grammar Day! Sadly this isn't as fun as it sounds due to the idiots on the internet. However, here is a thing I have learnt. Jeremy Butterfield wrote an article about how linguists and other people have different ideas about what grammar is, and in the comments someone mentioned the famous(?) Dennis Skinner complaint about being misquoted as saying I were. It's mentioned at the bottom of this link.
As he says there, he's not saying I were, because that would be 'grammatically incorrect'. He's saying I wah, as in dropping the 's' from was, and it just happens to sound like I were. Someone from Yorkshire in the comments agreed with him:
This is fine, I suppose, except that I really want it to be I were. It's a nice symmetrical counterpart to the We was found in many other places. Both are examples of levelling of verb forms, and I teach it as an example of how levelling is something that tends to happen, but that it's more or less chance what form is chosen in what dialect. If it's not levelling, then that means the the levelling only takes place in some dialects, and it's always towards was.
On the other hand, if it is a phonological reduction of I was, then there's other interesting questions to answer. Why don't these dialects level? Do other dialects actually have I were? If not, why would there be a preference to level towards was? How can we tell, for sure, that it's I wa and not I were?