I don't believe Ayer can escape this problem. As I understand it (and I am not an expert here), Ayer
himself came to accept this, after some efforts — such as you describe — to escape it. In the end, his
attempts do come down to a definition by fiat of the sort you mention. If you don't accept that the
meaning of "meaningful" is contained in "verifiable", then there is no reason to accept his principle of
verification. I have heard logical positivism described as 'the only philosophical position that has ever
been conclusively debunked', and I agree.