Received: by alpheratz.cpm.aca.mmu.ac.uk id QAA17460 (8.6.9/5.3[ref pg@gmsl.co.uk] for cpm.aca.mmu.ac.uk from fmb-majordomo@mmu.ac.uk); Mon, 15 Jan 2001 16:42:37 GMT From: <Zylogy@aol.com> Message-ID: <50.100fd116.279481bb@aol.com> Date: Mon, 15 Jan 2001 11:39:23 EST Subject: phonosemantics To: memetics@mmu.ac.uk CC: Zylogy@aol.com Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: AOL 5.0 for Windows sub 129 Sender: fmb-majordomo@mmu.ac.uk Precedence: bulk Reply-To: memetics@mmu.ac.uk
Hi. Answer for Vincent Campbell- essentially since the early part of the 20th
century its been dogma in linguistics that the outer form of a word- its
surface sound structure as well as its underlying systematic representation
in more abstract form (phonology)- was only arbitrarily linked to whatever
meaning it was intended to convey- thus "dog" as a concept could have
whatever sounds attached to it anyone cared to name. Meaning of any sound
string was thus conventional. The only exceptions generally allowed in this
view are onomatopes.
Problem with the above view is that it is way too simple and downright wrong
in many ways. Sure there are huge variations in the ways concepts are carried
as words- but there is also striking uniformity, at many levels, elsewhere.
Depends on where in the vocabulary you are looking. Linguists interested in
the reconstruction of parent languages or looking for deeper genetic
connections have noticed that certain vocabulary (called "core") tends to be
relatively resistant to replacement over long stretches of time, and this
diagnostic set (which can be arbitrarily as large as one chooses it to be) is
never very large by comparison with the size of the total vocabulary.
Historical change is a given- no form once in a lexicon can escape it, so
obviously words that hang around in the vocabulary because often used will
end up mutated much more often than nonce creations. Since this set of
long-used words is also the diagnostic tool, its what linguists see when they
compare languages. And since historical changes go off in all sorts of
different directions relatively independently in any language, after a while
the basic vocabulary will be radically different from language to language.
That said, let me tell you about the other side of the vocabulary- There are
many languages which have very high proportions of phonosemantically
transparent vocabulary. Phonosemantic transparency means that the form of the
word and its meaning are not totally arbitrarily linked, but that the meaning
is in some way predictable from the form. Onomatopes are like this, but they
only convey sound for sound. Transparent vocabulary can convey all sorts of
sensory impressions, movement patterns, etc. above and beyond mere
onomatopoeia. The breakdown of form to meaning goes down through the root,
past phoneme, and down into the individual abstract features of each phoneme
(such as voicing, consonantality, articulatory position, etc.).
Most SubSaharan African languages have huge numbers of what are termed
ideophones- these are transparent forms in the thousands in any language,
expandable by reduplication (repetition of various portions of the word) and
infixation. Since most of these languages don't have much in the way of
adjectives/adverbs as we normally think of them, they are stand-ins. But they
are also much more. There is very strong evidence that these forms actually
feed the normal lexicon- ideophone roots BECOME regular verb roots (its
happening, for instance, right now in Zulu).
Altaic languages (Turkish, Mongolian, Manchu, Korean, maybe Japanese) have
thousands of "expressives" as they are termed in English, again very similar
to the ideophones in African languages, and similarly used. One of the
reasons they are needed is that the numbers of true verb roots in these
languages tend to be very small- and broad in meaning. Expressives and
ideophones help to specify meaning, particularly in the area of manner of
action. Languages such as English, which have more complex roots which
incorporate manner specification, don't need such a supplementary vocabulary,
and so don't have a separate word class of ideophones.
Other languages with large numbers of such forms include Indonesian and many
of its relatives, Tamil and its relatives in India, Khmer, Vietnamese- I
could go on and on.
But this is a typological thing- there are other kinds of languages where
affixal morphology handles the manner specification- so you have neither
complex roots (as in Indoeuropean languages) nor a special word class. Many,
if not most of the languages of the Americas are/were like this- though there
are also examples in the Old World too.
Language type is a cyclic thing- and even the complex roots found in
Indoeuropean languages appear to have originated from a combination of
expressive root plus auxiliary verb. That process happened thousands of years
ago, and in many of the living languages of the family the evidence is being
worn away by historical change. But reconstructed parent forms show it in its
full flower. Interestingly, this process is happening right now in the Altaic
languages (and as I mentioned, also in Zulu and other African languages).
Once a significant portion has been transferred, the independent word class
will die off. In other languages they are on the rise. Big cycle of renewal.
Scholars working on Uralic languages (such as Finnish, Hungarian) estimate
that perhaps 40% of the total root inventory consists of these
phonosemantically transparent forms- if one considers only verbs the
proportion is much higher.
In phonosemantically transparent vocabulary the root acts like an algebraic
formula, but for meaning. Even in English there are still traces- think of
the sequence of verbs ending in -ag: bag, drag, lag, sag, tag, nag etc.
where though the particular application may be different from form to form,
all give a sense of being held back or down by some force, generally from
behind, at least metaphorically. Or consider the set in -am: ram, cram,
slam, dam, etc., giving the very different sense of pressure upon something
causing densification, confinement, etc. There are actually thousands of
such remnant sets in English and other Indoeuropean languages, but the
connections between items are spotty, the network of related forms diffuse.
All because of a combination of historical change and random attrition of
intermediate forms (if you have a Darwinian slant this'll sound familiar).
In languages with rich sets of expressive or ideophonic forms, you get whole
families of related terms which might vary by one or more features. Japanese
and Korean, for instance, utilize voicing contrast on consonants in these
terms to convey bigness or smallness of the same action/actor. Vowel height
conveys vertical height in the semantics. And so on. It is, in effect, really
"one" word with lots of regular variation, a multidimensional matrix covering
all sensorimotor possibilities.
Experiments have been done on many groups of subjects over the last 75 years,
and depending on the type of experiment almost always show very strong
results, even with speakers of unrelated languages. For instance, if I told
you that the consonantal skeleton m-l referred to a body of water, and then
gave you the series mil, mel, mal, mol, mul you could tell me which was a
wide one, which one a deep one, and so on. Results between individuals seem
to vary, but always are systematic, as if the brain was automatically forcing
order on the situation.
The particular values of the phonemes and features in semantic terms vary
from language to language as well, but not wildly, and the relationships of
the phonemes to each other is also quite systematic. The average language
with large numbers of ideophones or expressives has between four and eight
articulatory positions for consonants, and the phonological relationships can
be mapped as a tetrahedron for the former and a cube for the latter (I didn't
make this up). Each vertex of the former is capable of splitting in two, and
studies have shown that neighboring vowels are the most powerful promotor of
this. There is a method to the madness.
What you end up with is a diagrammatical iconic relationship between phonemes
which can be expressed as axial feature specifications and then feature
strings for each of the vertices. Interestingly, this diagram is completely
unconscious, though it has consequences throughout the lexicon and on the way
we categorize our sensory and motor impressions.
Fair enough? The organization I cofounded to bring together people interested
in iconicity in language structure and usage has a web page at
http://www.conknet.com/~mmagnus/LIA/index.html and there are numerous links
to various related web pages. I recommend Margaret Magnus' (our web guru)
bibliography, which is growing all the time. And on her pages and others' you
will find detailed explication of the topic. We also have a discussion list
with archives going back more than a year and a half, linked to the
organization page. Hope you take a look (and anyone else that might be
interested in this vastly underdiscussed area).
Jess Tauber
zylogy@aol.com
===============================================================
This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)
see: http://www.cpm.mmu.ac.uk/jom-emit
This archive was generated by hypermail 2b29 : Mon Jan 15 2001 - 16:44:12 GMT