[Bruce Nevin (2003.09.26.2244 EDT)]
Rick Marken (2003.09.26.1010)–
[Bruce Nevin (2003.09.26.2244 EDT)]
Rick Marken (2003.09.26.1010)–
Bruce Nevin
(2003.09.25.2244 EDT)
Nobody invented the word.
How do you know? Someone always has
to be the first to use a particular sound to refer to a particular
perception, don’t they?
[…]
we know that
historically language was not ever invented
Then how
did it start? Someone way back when had to invent some sound to
correspond to a perception. Many people probably contributed to the very
first vocabulary of “ugs” and “ahs” but each one was
an inventor before the each word caught on.
by someone first categorially parcelling up
the universe of perception and then coining words to
correspond.
I think many people contributed to the invention
of the sounds that a community uses to refer to different aspects of
their experience. The point is that someone, long ago, had to say
something like “ferd” when pointing at a horse and eventually
everyone agreed to call it “ferd”. Others may have said
“syts” and “deght” when pointing to their horse but
most people used “ferd” so “ferd” it became, until it
morphed gradually, over 1,000,000 years or so, into
“horse”.
Except, I suppose, in Germany? They say Pferd today.
The common Indo-European ancestor of Latin equus, Greek
hippos, Irish eapall, English horse, Lithuanian
arklys, etc. etc. was something like *ekwo- with various
inflectional endings. (The asterisk signifies that this is a
reconstructed form. It of course is not attested in any documents or
inscriptions.) If they have a common ancester, how did they become so
different? 1. Language changes. 2. People who talk to one another
maintain similar reference values for their way of talking; people who
don’t, don’t. 3. When communities are separated and the people of one
community no longer talk to or hear people in the other speaking, the
ongoing changes in the speech in one community are no longer are kept in
conformity to the ongoing changes in the speech in the other.
The first point, remarkably, seems abstract and unreal to most people. We
do not perceive language changing, we only perceive differences
afterward. The English spoken by children now differs in small ways from
the English you spoke when you were a child. I’m not talking about
obvious teenage jargons. Ten or fifteen years ago I noticed that children
are quite comfortable saying “this is funner!” That was not
something anyone would say when I was growing up, nor is it anything I
ordinarily say now.
In pursuit of the intention of the speaker, listeners overlook small
discrepancies from the way they themselves would say it, and speakers
overlook them in pursuit of the understanding of the listener. (Or, more
exactly, discrepancies from the way they believe they themselves would
say it. People’s actual speech differs from what they say and believe
their reference values for speaking are. This of course makes the process
even yeastier. People don’t notice changes that they themselves are
promulgating!)
What is noticed are differences in the way different kinds of people
speak when they again come in contact after the divergences in their
respective directions of ongoing change have become sufficiently great
that they cannot be so easily ignored. At no time did anyone invent any
of these changes. In the process of two communities coming to have
different words for the same thing, at no time does someone invent e.g.
arklys or horse or Pferd.
So where did it begin? For the Indo-European ancestor language we’re only
looking at maybe 10,000 years ago, certainly less than 20,000 years, and
even at that point the origins of language were unimaginably remote in
earlier time. We have no direct evidence of any human language that is
not a full-fledged language, with all the grammatical, semantic, and
pragmatic complexity of English, Sanskrit, Chinese, Navajo, or any of the
other (dwindling) thousands of languages presently spoken on this planet.
We do have direct evidence of animal calls. These are limited to
ostensive reference (evidently signifying Hawk! Snake! Fruit! and the
like). It is a virtually inescapable inference that the beginnings of
language were in this kind of ostensive reference. If you’re talking
about the invention of words, that’s where you have to look.
So who invented the distinctive sound that a mother hen repeats as she
drops a speck of food in front of her chicks? A rooster repeats the
identical sound as he picks up some choice morsel that he has found and
drops it repeatedly until some hens come over and peck at it. Who
invented the sound that a vervet monkey makes when it sees a snake? There
may be some evidence that these sounds are conventionalized somewhat
differently in the monkeys of widely separated (non-communicating)
groups. That indicates that they are conventional, that is, social
inheritance may play a role, perhaps along with strictly biological
inheritance as with birdcalls. Careful research has not been conducted
for long enough to find out whether these calls change through time in
the way that humans’ pronunciations change. That’s the important
distinction between genetically ‘programmed’ calls and words.
In hominid groups, with expanding cognitive ability (driven by and
simultaneously supporting the need to comment upon their fellows or
‘gossip’ in order to sustain alliances with larger numbers of individuals
than is possible in grooming relationships – Dunbar’s thesis), it is
plausible to reconstruct a process in which more sound-referent
associations were conventionalized (bog for “fruit”, let
us say), far beyond simple calls. After that there emerged words that
could only be said or understood in company with these simple referential
words (from a sound of revulsion perhaps comes a conventionalized
word-shape fah and the combination bog fah equivalent to
“fruit bad-taste”). The beginnings of syntax.
An essential point is that such conventionalizations are social
products, and not inventions by individuals. Here’s another way past
whatever preconceptions may make it difficult to recognize this: By the
time it becomes conventionalized, it is not the same as it was when the
first individual repeated (roughly) what another individual uttered. So
how can you claim that the individual who was imitated was the inventor,
or that the imitator who thereby began to establish the communicative
(i.e. social) value of the repetition was the inventor, when at that
point this is only a precursor to the word in process of being invented?
How can you say that she (it very likely was a she – Dunbar again)
invented something that in fact would not actually exist until some
protracted time after?
And which of course would continue to change thereafter. There is no
reason to suppose that the development of language, from its earliest
roots in animal calls, has proceeded by any different means than those by
which language changes today, and by which it has demonstrably changed
during the relatively shallow period for which we can document and
reconstruct it.
Words, and their
likelihoods of combination with one another, are a domain of perception
that has only a loose coupling to the perceptions that we think of as the
world to which we believe the words refer.
The words I use
are not coupled to perceptions; I use words to evoke in others
experiences and understandings that are (I presume) similar to my
own.
This is a quibble. I’ll use your phrase if you wish. The association (in
your mind and in that of others) of words with referents, by means of
which you hope to evoke experiences and understandings, is a loose and
approximate one. You can achieve something like the evocation you seek
only if the other party cooperates. These two paragraphs are a case in
point.
[The blind use
vision words] in completely appropriate combinations with other
words.
What does appropriate mean? If they don’t know that
these words have referents then they would be as likely to say “The
grass is green” as “The idea is green”. If they use color
words correctly in terms of both semantics and grammar then they
understand that these words point to particular perceptions.
Certainly they understand that the words have referents. (I never
questioned that. Why do you?) What is relevant is that they don’t have
access to those referents, but they nevertheless use the words
indistinguishably from people who do have access to those referents. If
you listened to them talk without knowing that they were blind you would
assume that they look, see colors, see green
objects. This is what appropriate means. Yes, of course, eventually they
usually stumble where some referent is needed (Bill nodding but the blind
person saying “I see you disagree”). But when a person says
“I see what you mean” it usually does not mean that they
literally see the referents of your words in the shared physical
environment.
So one point here
is that you don’t need the perceptions to have the words and their
meanings and to use them with facility and accuracy. That point is one
side of the ‘loose coupling’.
I completely disagree with this
point. You don’t just need the perceptions to have the meanings; the
perceptions are the meanings.
Well, meanings are not outside the perceptual hierarchy. But if by
“the perceptions” you mean the referents of words, you are
mistaken. Meaning is not limited to reference.
The perceptions include specific referents to which the words
refer on this occasion. There are also other perceptions that might be
specific referents of the words on other occasions. With concrete nouns
(such as brick) and the least abstract operators on them (such as
green), the range of differences among referents is relatively
small, but still considerable.
The meaning of a word, on the other hand, is not identical with
its referent on this occasion or that. Meaning is not limited to
referential meaning.
Most talk of meaning is merely a matter of translation. Dictionary
definitions map words onto other words. Logicians map sentences onto
logical propositions which, however unlike language their symbolism may
appear, are still derivative of language, and specifically the meanings
derive from understandings established in a shared vernacular.
These aspects of meaning are learned as a sense of the combinabilities of
words. The evidence is pretty good for statistical learning.
For ‘concrete’ words with specific and (relatively!) consistent
perceptual referents, like green and telephone, it is
possible to talk of the meanings in terms of example referents or in
terms of generalizations about referents. However, the role of perceptual
referents in learning the meanings of verbs (adjectives, prepositions,
etc.) is much weaker than the role of syntactic context.
For example, when adults see a series of videoclips of a mother and child
playing, with the audio turned off, but with a beep sounded just where
the mother says a certain word, observers guess more of the nouns
correctly than they do the verbs. Some details (from Gleitman’s chapter
in the book I edited):
Adults were shown video clips of a mother and small child (18-24 mo.)
talking and playing. The various objects they played with were visible,
but the situation was otherwise unstructured. The investigators
transcribed the mother’s speech and identified the 24 most frequent nouns
and the 24 most frequent verbs. For each of these words, they selected
“more or less at random” 6 videoclips in which the mother was
uttering the word. Each videoclip started about 30 seconds before the
mother uttered the word, and ended about 10 seconds afterwards. For each
word, they then spliced the 6 video-clips together with a brief color-bar
between them. This gives the observer the opportunity to identify what
all six scenes had in common. After viewing all six samples, the subjects
guessed the ‘mystery word’. They did this for all 48 words. The subjects
were told for each word whether it would be a noun or a verb, but they
did not hear what the mother was saying. When the mother uttered the
word, a beep sounded, but otherwise the audio was turned off.
This simulates the situation when a learner can’t infer anything from
other words in the sentence. The only information available to the
learner is a sample of the nonlinguistic perceptions associated with
whatever the mother said. Of course, actual learners may encounter 7 or
50 or 500 such opportunities as the basis for inferring a word meaning.
On the other hand, in real life these opportunities do not occur one
after another in succession, nor is the learner told that it is a noun or
a verb. Conclusions were limited to seeing if there is a difference
between learning nouns and learning verbs. The 84 subjects, observing the
48 items in 6 contexts each, correctly guessed about 45% of the nouns but
only 15% of the verbs.
With words other than nouns, learning from non-linguistic context is
increasingly difficult. We say that the meanings of such words are more
abstract, that is, there are no simple referents, and the various
occasions of use have less in common. In the above experiment, at least
some of the subjects identified each of the 24 nouns, but eight of the
verbs (one third) were never correctly identified by any subject. These
verbs were know, like, love,
say, think, have, make, and pop.
This is easy to understand: several of them describe
invisible mental acts and states, and others are so inspecific that they
can be used in almost any context. But of the 16 remaining verbs
only 23% were correct, vs. 45% of the nouns. Other experiments showed a
correlation with what they called ‘imageability’, that is, consistent
association with specific perceptions input from the video image.
Ball was easy to guess, a kiss was hard; push was
easy, want was hard, as any player of charades can readily
appreciate.
So the contribution of reference to meaning is limited. What of
linguistic context? In a later experiment, they showed a new group of
subjects nonsense frames for the same test verbs. These frames were
constructed from the sentences the mothers were uttering in the
videoclips, preserving the morphology (suffixes, etc.) but converting
both the nouns and the verbs of the six maternal sentences to nonce
words. For example, two of the six sentences with call became
Why dont ver gorp telfa? and Gorp wastorn, gorp
wastorn!. They saw no video, they knew nothing of the contexts in
which these nonsense verbs had been uttered, and they knew none of the
nouns that co-occurred with the verbs, since they, too, were converted to
nonsense. Yet they identified proportionally more verbs (51%) than they
did in an experiment in which both video contexts and accompanying nouns
were presented (29%). This is not surprising to linguists; the literature
documenting relations between syntax and semantics is huge.
Much of the meaning of a word derives not from what it can be said
about (referential meaning) but from what can be said with it.
Constraints on the combinatorial possibilities of words are both
perceptible and learnable. For starters, these constraints partition the
set of words into classes according to what other of these same word
classes must be said (or understood) with them: dependent on no other
word (concrete nouns), requiring one noun (intransitives like
snore), or two (transitives like throw), or three
(ditransitive give), requiring some word which in turn is
dependent on another (as e.g. true requires some verb, adjective,
etc.), in combination with a noun (startle with the noun
following, imagine with the noun preceding), and so on. These word
classes carry, in effect, aspects of the meanings of words. This can be
seen very directly in information-theoretic terms: by perceiving two
words where the arguments of the verb are expected, you eliminate a large
range of possible meanings from consideration: Why dont ver gorp
telfa?.
Consider unknown nouns A, B, and C with some unknown
verb in the following construction: A verb B from C. You
can guess some verbs that might occur here, such as take (A
takes B from C), borrow, carry, etc. But now you hear
the same nouns and verb in A verb C from B, and then in A verb
B and C. These rule out words like borrow, but others remain,
such as divide and distinguish. As more constructions are
added, such as B verb from C, the possibilities are reduced to
just a few candidates, and, finally, only one: separate. You
identify this unique word, not by its referential meaning, but rather by
those aspects of its meaning that inhere in the combinatorial
restrictions of words.
Even referentially specific words like brick or green have
this sort of constructional meaning as well as their specifiable
perceptual referents. The referents and the constructional meanings are
both perceptions, but they are of entirely different sorts. The referents
are nonlinguistic perceptions, outside of language; the constructional
meanings are linguistic perceptions, perceptions of what is sayable in
language. The two are related, of course, and the relationship is
tightest for the more concrete vocabulary with the most specific
perceptual referents.
You can construct grammatical sequences of
words without knowing the perceptions to which the words refer but you
certainly can’t construct meaningful sequences of words without know the
perceptions to which the words refer. Here’s an example. Consider a
language with three words: A, B and C. The rules for constructing
sentences in the language are S → A + B, S → A+C, S →
B, S → C. So there are four grammatical utterances in this language:
AB, AC, B, C. The remaining utterances, an infinite number, are
ungrammatical; for example, A, AA, BB, BCA. Can you tell which of the
grammatical utterances are meaningful? Of course not. That would be
the situation of the blind person, too, if the words had no meaning for
them.
If people in fact constructed grammatical sequences of words according to
such rules this might be relevant.
And we know that
our ancestors’ ways of conceiving of the world and parcelling it up
categorially are not the same as ours today (though perhaps this bears
emphasizing, since it is so common an error anachronistically to presume
that people of the 16th or 14th or 2nd century, or earlier, or even the
19th, perceived the world as we do today, as though they were no more
than ourselves in funny clothes, as in some B movie).
Now we
really have a fundamental disagreement if you are saying that our
ancestors perceived the world differently than we do. If this is what you
are saying, what is your evidence? I believe that our ancestors perceived
exactly as we do, in terms of the same classes of perceptual variables.
They the same nervous system (and, hence, perceptual) architectures as we
do today. They might have used some words slightly differently than we do
now; they may have referred to a hippo as a horse, for example. But I’m
sure that their perception of a hippo was as different from their
perception of a horse as my perception of a hippo is from my perception
of a horse.
What I said was their way of conceiving of the world and parcelling it up
categorially. To take a hackneyed example, in a lightning storm they
perceived the quarreling of the gods. I do not claim that they perceived
flashes of light and the contours of clouds differently than we do. Their
explanation of it, the story they told themselves about it, was
different. This is a matter of language and the use of language to
construct concepts and parcel up the perceptual universe categorially.
But fabrications though these explanations are, and fictional (“made
up”, from Latin faco, facere, “to make or do”),
they constitute (construct) reference perceptions according to which
nonverbal, nonfictional perceptual inputs are controlled. And that is why
it is crucial to understand culture.
What evidence do
you have for a single perception, corresponding to the word ‘fairness’,
other than the use of the word?
The evidence is the perception (of fairness, in this case). The
word has nothing to do with it. When I see Bush saying that there should
be no tax on dividends I see unfairness, whether I say the word (or think
it) or not.
This is incapable of proof or disproof. The only evidence I see is your
use of the word. You have no means to demonstrate that you are not
fooling yourself. You clearly believe that the story you tell about your
perceptions is true.
I’m
just saying that I can perceive fairness so I know it
exists.
To a rather complicated set of perceptions, or rather to a generalization
over an innumerable variety of these, you apply the word
fairness.
The complicated set of lower level
perceptions is experienced by me as a perception of some degree of
fairness. It’s just like my perception of a horse (or hippo); a
complicated set of lower level perceptions (or intensity, color,
configuration, transition, relationship and so on) is a perception of a
horse (or hippo).
The test is
necessary for an observer to know what perceptions another agent is
actually controlling. You certainly don’t need to test to determine
what you yourself are controlling. You’re just controlling
it!
You don’t need the Test in order to control. You don’t need the Test to
determine that you’re controlling. But you do need it to verify
what you’re controlling. You claim that you are controlling a
perception of fairness. This is just like looking at the coin game in
progress and telling a story about your perceptions, “Oh, he’s
controlling a triangle with the quarter adjacent to the longest
side.” You have to then Test to find out if your story is correct.
This is no less true of the stories you tell about your own
behavior.
Why? When I’m catching a fly ball I am
completely unaware of what I am controlling. If I thought about it
(before I did the research on fly ball catching) I would have thought I
was controlling my location relative to the predicted landing site of the
ball. This, of course, is not what I am actually controlling. But
even when I thought that that was what I was controlling I was still
pretty good at catching balls (so I could certainly control without
knowing what I was controlling).
That’s right. You don’t need to know what you are controlling in order to
control, and what you ‘know’ about what you are controlling can easily
have little or nothing to do with what you are controlling. The story you
told yourself about how you catch fly balls was wrong, but fortunately it
was irrelevant. The story you tell yourself about fairness is equally a
fabrication.
As anywhere in
science, it’s just too easy to fool yourself. And in fact people fool
themselves about their CVs all the time. The most common word for it is
probably rationalization.
Rationalization is another kind of
controlling.
I agree. But it is irrelevant to the control actions that you are
rationalizing. The smoker tells a little story about being able to quit
any time he really wants to, cold turkey. Or a story about enjoying the
taste. Whatever. It doesn’t matter. These stories are irrelevant to what
the smoker is controlling by smoking.
You control for perceiving yourself as acting
in terms of principles of rationality. I suppose it is a way of
“fooling oneself”. But it seems to me that this comes up only
in a special situation; when one is in conflict. That is a
situation where it may help to know what (at a higher level) are one’s
goals (the ones creating the conflict). In most everyday behavior it
doesn’t seem to me that one has to know what one is actually controlling.
It doesn’t help to know, anyway. As in the baseball catching example,
whether I know I’m controlling vertical optical velocity or not doesn’t
seem to matter as long as I can catch the ball.
You keep beating on this Pferd. I did not say that you have to perform
the Test and identify your CV before you can control it. I must have said
something very carelessly for you to make that interpretation of it.
But in fact people
argue about what’s fair all the time, don’t they.
The fact
that people do argue about this seems to me to be clear evidence that
they can perceive “fairness” as a perceptual variable. The
differences between people in terms of what they call “fair”
and “unfair” may reflect differences between people in terms of
how they construct the perception of fairness but it could also reflect
differences in where their reference for fairness happens to be.
The people I’ve know who have been active in unions, say, seem to
perceive fairness in the same way I do; they just seem to have a much
higher reference for fairness than I do.
OK, for your kids, “fair” is an equal partition. That’s a
relationship perception to which the word “fair” is
applied.
You may be interested in a story in the local paper, at
http://www.mvgazette.com/news/2003/09/26/agents_seize_brazilians.php
A bit more than half way down, you read
“The immigration officials were very professional and fair,”
said Sheriff McCormack.
I see no equal partition of anything. The perceptual input function for
the Sherriff’s recognition of fairness in this experience is evidently
more complicated than that of your children with the cake. And yet I am
quite confident that if he witnessed or participated in the “I cut,
you choose” scenario, he’d agree, “that’s fair”.
A little bit farther on, you read
"They took my nephew at four o’clock this morning. He was
wearing his pajamas and sandals. He said, Can I put some clothes on?’
and they said no,’ " said Marci Da Silva, a 44-year-old woman from
Oak Bluffs. Her nephew was arrested from a home next to Tony’s Market on
Dukes County avenue.
I don’t think the story she is telling about her nephew is likely ever to
include “fair” or “fairness”. Is that because she and
Sheriff McCormack have different reference values for fairness? She wants
more and he wants less of the same CV? It sounds to me as though there
are entirely different perceptual inputs involved.
A relationship of equality as something like a cake is divided. Something
unspecified about a process of waking a bunch of people up at 4:30 in the
morning and hustling them off to a Coast Guard cutter. At
www.fair.org/ we read
about diversity in the press, marginalizing of public interest, minority,
and dissenting viewpoints, and censorship. At
http://www.workplacefairness.org/
we read about making sure that people are paid extra if they work more than 40 hours in a week. Elsewhere, we can read about the unfairness of police seeking out traffic details, and lobbying to defend the laws that prohibit anyone else from babysitting a construction crew, and racking up 6-figure annual incomes with the overtime. We read about the unfairness of being taxed at a higher rate, or twice, or at all, and we read about the unfairness of wealthy people finding tax loopholes. At http://www.e-fairness.org/ we read how internet businesses must be made to pay sales tax. Elsewhere, we read how the Supreme Court ruled in 1992 that the diverse welter of state tax codes would be unfair to inflict on internet businesses. The 1949 “Fairness Doctrine” of the FCC required all broadcasters to devote a reasonable amount of time (not equal time) to the discussion of controversial matters of public interest. There are other relations of inequality that are commonly considered fair, e.g. that a surgeon should be paid more than a night watchman, that remuneration should be commensurate with or proportional to skill & training, or experience, or what the market will bear, etc. What you call “the proportionality of reward with respect to merit.”
The more you look at examples of “fairness” (or its lack), the more variety you see. What they have in common are stories that people tell about what is going on, using the word “fair” (and its derivates). I don’t see much else that they have in common. In some cases, we see a simple perception of equality. That’s a relationship perception. How can you elevate that to the Principle level? In some cases, we see a “proportionality” to some other variable (skill, training, merit, experience, market demand, gullibility of the mark). That’s a different relationship. Again, how can you elevate that to the Principle level? And do these different relationship perceptions have the same input function? Whatever Sheriff McCormack meant by saying that the Immigration Service and police were fair to the Brazilians, it must have been more complex than a relationship perception. So that’s quite a complicated input function that you’re proposing for control of fairness as a CV.
···
At 10:12 AM 9/26/2003 -0400, Richard Marken wrote:
So here are some possible specifications of a perception of fairness:
-
You say “that’s fair”, referring to some particular situation, or exchange, or relationship, or retort, etc. etc.
-
You and some other person both say, and agree, “that’s fair”, referring to some particular situation, etc.
3a-m. Referring to some particular situation, etc., you predict that you and any person in your {family, community, ethnic group, age group, country … in the world} reliably will agree “that’s fair”.
4a-n. Referring to any situation characterized by perceptions {i-k}, and only to situations characterizable in that way, you predict that you and any person in your {family, community, ethnic group, age group, country … in the world} reliably will agree “that’s fair”.
Which do you mean?
All of them, in some way or another. In (1) I think you are describing the situation where the fairness of what I perceive matches my reference for fairness. In (2), you are talking about a situation where fairness, as perceived by each party, matches their reference for fairness. In (3) you are talking about my guesses about other’s perceptions and goals regarding the fairness of a situation. Let’s take “executive compensation” as an example. I see executive compensation levels are ridiculously unfair and I think I could pretty accurately guess who would and who would not agree with me. But I would know that, when I talked about “fairness” in that situation, we were all talking about the same perception: the proportionality of reward with respect to merit. I think (4) is not very different from (3).
You haven’t disagreed with me yet. You can’t until you understand what I’m saying.
I disagree with what I think I have understood of what you say. But at least you’ve answered my question: you can perceive disagreement because you know when it’s not happening.
Best regards
Rick
–
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org
