>> I'd suggest a different criterion: a human language is one that a >> human being can learn to understand and speak in real time. This still supports my assertion that any speech declared as language 'a' is arbitrarily designated as such. You are arbitrarily defining 'real-time'. In doing so, you are making a subtle but arbitrary decision that humans always have had and always will have the same capacity for language learning diachronically and that all humans synchronically have the same capacity for language learning. The distinction remains arbitrary for me. T.P.