The
Ape that talkes
|
The other proposal, was a scheme for
entirely abolishing all Words, since Words are only Names for things.
It therefore would be more convenient to carry about such things as
seem necessary to express the particular business we are to discourse
on. I have often beheld two of those sages almost sinking under the
weight of their packs, like pedlars, who, when they met in the streets,
would lay down their loads, open their sacks, and hold conversation for
an hour together; then put up their implements, help each other to
resume their burdens, and take their leave.
|
Jonathan Swift, Gulliver’s Travels
III:5
|
William
S. Burroughs suggested
that words could be acting similar to viruses, occupying the human
brain in a
kind of symbiosis. Words need us – “the host” – to exist, we need them
to
operate with more efficiency. Words are not just carriers of
information. They
can affect us directly by triggering emotions. And writing amplifys the
semantic potency of a word even further, putting it before the inner
eye as a
pictorial icon. But how is all this happening, how do words link up
with each
other, how do we create a sentence that is making sense?
Since the seventies we
experiment with artificial speech on neural network computers, trying
to make them read a text with proper enunciation, in other words to
read with a measure of “understanding.” Apparently, such neural network
spontaneously forms a syntactic substructure, creating ad hoc so called
“hidden variables.” Such junctions of interrelated words, represent
semantic categories, guiding the emphasis of the delivery. The truly
interesting aspect here is, that not two such machines initially use
the exact same set of hidden variables, but they all gradually improve
their syntax until reaching a common standard of intelligible delivery.
It really happens and helps the machine to downsize the otherwise
unwieldy hierarchies of layered meaning which encumber Noam Chomsky’s
proposition of an innate “meta-language” with his impractical diagrams
of “phrase markers.” On the other hand there is indeed a
“language-gene” – FOXP2 – assisting the process, which leaves the
speaker severely handicapped if it is missing.
Individuals with this
disorder are unable to select and produce the fine movements with
tongue and lips necessary to speak clearly, which makes their speech
unintelligible to the listener, a disorder that goes hand in hand with
dyslexia, difficulties in processing sentences and poor grammar. Yet in
the final analysis, it is the purely logical requirement of producing
an intelligible sentence that could be the factor urging on the child’s
mind to make up language as it moves along. Linguists refer to such
language as a “creole,” a language in which a strict order of words in
the sequence supersedes any synthetic combination of words that uses
particles (suffixes, infixes, and prefixes), connecting the words in a
more “inverted order,” a term that doesn’t make much sense in synthetic
languages. There are the well known cases of children whose immigrant
parents speak merely a pidgin, an ungrammatical cross-breed of several
languages. But their kids develop a brand-new creole from scratch, a
coherent language which shares fewer features with the parents’ pidgin
than it is sharing with other creoles. As Terence W. Deakon has put it:
"Languages must go through the filter of children's
reduced
associative learning and short term memory constraints in order to be
passed on
effectively from one generation to the next with any degree of
fidelity.
Children selectively hear some structures and ignore others, and so
provide a
major selection force for language structure that is 'child-friendly'" (The
Symbolic Species: The Co-Evolution of Language and the Brain).
That's right: children are the engines of linguistic evolution. The
role of the adults is restricted to insisting on 'proper' speech and
prevent this evolution to derail in anarchy. However the intervention
of parents lacking in linguistic competence is not incisive enough to
prevent the development of a full fledged creole. As it seems neural
networks, to some extent, can mimic a child’s capacity to reinvent
language from scratch. “Creoles” are spontaneous creations.
The history of languages
is the evolution from the complicated towards the simpler and easier to
use. The ancient elites used to spend precious years of a short life on
the rigorous training in cumbersome grammar, as everybody can tell who
has conjugated the Latin verb 'ire:'
('eo,' 'is,' 'it,' 'imus,' 'itis,' 'eunt') in search for the imperative
plural ‘ite,’ and then moves
on to figure out the locative for ‘domus’
(‘domum’). “Yankee go home!” is so much easier,
even if you speak not a word of English, than as a foreigner shouting
at Roman legionaries: “Romani”
(vocative
plural) “ite domum.”
It makes me think that
even English is nothing else but a developed creole. Languages with a
synthetic structure like Gaelic, Latin, Anglo-Saxon and later French,
have contributed with their vocabulary to the making of English.
Conquest and trade had knocked together people of different ethnicity,
thus encouraging the development of a pidgin in the first generation,
and of creoles in the second. A generation further down this has
resulted in a derivative, already resembling the language we are using
now. On a personal note: I’ve received a fairly rigorous language
training, but I know that it pales by comparison with the rigor of
training my instructors had received, and they in turn report the same
of their teachers. Have we grown lax these days, or is this actually a
sign of improvement? Are there any statistics for "incorrect" speech
and dubious neologisms in Shakespeare? Nowadays of course nobody will
censor Shakespeare for "incorrect" speech, what we find unforgivable
with a second grader, we accept from Shakespeare as a "creative grace,"
nay even receive it as a new standard of correctness.
Immanuel Kant
took particular pride in his table of cognitive categories. His critics
mocked the idea as a "glittering
palace" (Schopenhauer), but Kant saw it as the
key to his philosophy. The American philosopher Charles Sanders Peirce (1839 – 1914) adopted
this
view with the important correction, that these categories actually lay
at the
bottom of our linguistic toolbox.
Every statement and every
sentence, in order to make sense, can't do without semantic markers
implicit to its structure. No matter how short or elaborate a
statement, whether embedded in context and situation – as in: “Let’s go!” – or standing alone as
a self-reliant pronouncement, it has to be positive on at least three
of
the following four groups of semantic categories. A sentence must
implicitly refer to an element of:
– quantity (do matters belong together; are they
separate
entities; or are we talking about the totality of everything);
– quality (is something real; or contrary to facts; or
is it conditional?);
– relation (is it an accident based on a specific
immutable condition: a tulip imay be red because all the roses in this
particular flowerbed are red; is it an effect following a cause: the
tulips in the flowerbed are red because the gardener planted red
tulips; is it a matter of reciprocity: the gardener would have chosen a
different color, if there wasn't a majority of red tulips in the first
place?);
– modality (is it possible to exist; does it actually
exist; is it necessary to occur?).
To avoid
misunderstandings, I am speaking here of categorizations implicit to
the process of forming a sentence, a mostly subconscious activity
without which we would not be able of understanding or formulating a
statement. In other words, Kant’s categories resemble the function of
“hidden variables” in our neural networks. This process even carries
across the language barrier. Despite of differences in word-structure
and syntax, a translation answers to the same set of categories that is
universal to all languages. Without such universality, translations
could be well-nigh impossible.
© – 5/5/2009 – edited
by michael
sympson, 1,300 words, all rights reserved
