The reality of phonemes

Steve Long Salinas17 at AOL.COM
Thu Apr 3 17:50:54 UTC 2003


In a message dated 4/3/03 9:39:23 AM, geoffnathan at WAYNE.EDU writes:
<< I don't want to start a debate about the reality of phoneme, but I just
couldn't let that one pass... >>

I'm not sure how one would establish the "cognitive" reality of phonemes, but
the physical, physiological and communicative reality of phonemes seems
beyond doubt.
There's little doubt that voice-generated sounds are discriminated in the
listener and that they produce different effects (changes of state) in the
listener.  And there's little doubt that human languages are composed of a
limited number of basic building block "bytes".  And, as Dan Everett pointed
out, basic information theory can readily relate the relative inventory of
phonemes to the complexity on the next level of communication.

Humans could in theory pass consequential information by way of a spoken
binary system of 0's and 1's, but the speed of transfer and tail-end
processing would be cumbersome.  But that does not mean that it would not be
possible.  Sign language and other examples show that its the functional job
of phonemes that must be satified, not a particular structural one.  (Brian
Dickens wrote: "Originally, Morse's telegraph was built such that the letters
of a message had to be typeset by hand before transmission, and the message
would be printed out on a strip of paper on the receiving end (Kline). The
operators, however, realized that they could communicate much faster by
learning to "think" in Morse Code, so they skipped both the encoding and
decoding steps.")

Whatever cognitive grammar means, it cannot mean that humans must employ a
certain number of phonemes or equivalent language "parts".  But the
requirements of tranferring information DOES require phonemes or equivalent
parts to be used or information will not be tranferred.  And it is therefore
a fair bet that information can't be processed without them either.

Conversely, if we try to imagine a language with say 5000 phonemes, it is
possible to see that such a system would burden memory and recognition.  A
language that is able to use 5000 phonemes does not have as intense a need
for word structure (again as Dan Everett pointed out) or perhaps even
grammar, but the costs are high.  Each phoneme would be carrying an
extraordinary amount of information.

Also, phonemes themselves can carry less than a linguistic unit of
information.  One can recognize a unique phoneme in French without
understanding French and immediately be informed that the speaker is speaking
French.  That is information, but it does not tell you what the speaker is
saying.

Steve Long



More information about the Funknet mailing list