Depending also how you analyse long vowels
and diphthongs. A biphonemic analysis would greatly reduce the inventory -- it's
the vowels that inflate it. There are languages like !Xu~ with about eighty
click types, but here again it isn't clear to me whether all clicks should be
treated as unitary phonemes (rather than clusters). If there is disagreement
about how to analyse well-known modern languages (including English), perhaps we
shouldn't worry so much about the exact number of phonemes in Nostratic. But
here's the rub: if you posit a very large inventory, phonemes have to share the
comparative evidence and are not likely to be individually well supported by
reliable correspondences.
Piotr
----- Original Message -----
Sent: Sunday, January 28, 2001 4:10 PM
Subject: Re: [tied] IS's "regular roots"
...a phoneme total of 58 is not far above that of modern
English (48 - 51 in my speech, depending how you count). English
admittedly has fewer consonants and more vowels, but 58 is not in itself
impossible, surely?