A New Zealand correspndent writes to query a point I make at the beginning of By Hook or By Crook, when I meet my supposedly Welsh shepherd and he turns out to be Scottish. I said 'From a phonetic point of view all faces look the same'. My correspondent cites examples like the following:
'I attended a lecture just this week, and as the speaker approached the chair of the meeting she said "hello". I could not hear her, but I could tell from the way her mouth worked she had a strong Scottish accent.'
'In my travels as a boy, in a holiday camp in Devon I met an old man who reckoned to be able to place any person within ten miles of their place of birth just by their accent - and he was stone deaf !'
'When Mike Yarwood impersonated Harold Wilson or Ted Heath, in part his face (and posture) adopted a similar appearance to theirs.'
How does this experience tie in with my statement, he asks?
Because I was talking about anatomy, and my correspondent is talking about physiology. When the mouth is in action, then you can often get a few clues (such as the long pure vowel in the Scottish speaker's hello being visible in the rounded lips being held for longer than in other speakers) - but not when it is static. There is no evidence for anatomical differences in the external mouth, within a racial type, even between languages, let alone dialects. But there is a great deal of difference in physiology (i.e. in the way the mouth and jaw move). And even when there are external differences between racial types (such as Africans and Europeans) this tells us nothing about how the speaker will sound - as is common experience on London buses, when you hear a Cockney voice behind you and then see that the speaker is plainly from Jamaica, or somewhere.
In phonetics there is a conceptual apparatus which has been developed to handle this aspect of speech: it is called 'articulatory setting' - the way we habitually 'set' our vocal organs in preparation to speak. Phonetician John Laver has written a lot about this, as indeed did I, way back in the 60s.
Citing a few anecdotes isn't persuasive. Yes, there are faces where you look and you can guess from the way the face falls that the person very likely speaks in a certain way; but when you take a large enough sample, these instances turn out to be isolated cases, and probably explicable for other reasons. There are always clues in the context which help you reach a conclusion about the linguistic origins of speakers before they start to speak.