[From Bruce Nevin (2003.11.17 16:54 EST]
I think maybe we’ve bottomed out on the reality issue that derailed us,
so I’d like to try to get this “fairness and meaning” thread
back on topic.
Recapitulating where we were as of my (2003.10.22 23:00 EDT) post:
What I am saying does not entail a claim of special direct knowledge of
the environment that somehow bypasses perception.
Here’s a scenario. I control a perception of a sentence in English being
audible to you, for example:
Linguistic structure is in our shared
environment.
A sentence is a structured arrangement of words, which are themselves
structured arrangements of morphemes, syllables, and phonemes or
phonological contrasts. So far we are entirely in the realm of my
perceptions as speaker.
We agree, I think, that I perceive structure in this, and that I
control those perceptions – for example, the words must come in that
particular order to be an English sentence, and to be this particular
sentence.
The input functions, references, and output functions for my control of
these perceptions accord with conventions established as common knowledge
among users of the English language. In the course of learning those
common-knowledge conventions I have become so organized as to be able
control these perceptions by means of various behavioral outputs, in this
example by saying the above sentence to you.
This is also true of you. In the processes of having also learned the
common-knowledge conventions of English, you, too, have become so
organized as to construct the same structure out of your
perceptual inputs from your environment while I am speaking.
You recognize the structures that constitute this sentence. These are of
course structure-perceptions which your input functions have constructed,
but you perceive them to be the words, etc. that I have put in the
environment. Remembering common-knowledge conventions as to what
intentions one has when producing such structures, you ‘read’ into them a
perception of my intentions; in particular, you infer meanings that you
perceive (correctly or incorrectly) that I intended you to understand
from it.
Among these meanings that you infer are meanings that I did not intend.
You say:
You’re asserting that you have direct knowledge
of the
environment, bypassing perception.
This is an inference that you have constructed. I did not actually say
“I know this because I have direct knowledge of the environment,
bypassing perception.”
But you could not have made this inference without first understanding
what is directly said in my sentence, namely:
Linguistic structure is in our shared
environment.
I did not say how I arrived at the conclusion stated in this sentence;
you inferred that I must be claiming to have arrived at it by some sort
of direct apprehension of the environment.
What I intended (but did not say) was that I arrived at this conclusion
about the environment in the same way that you arrived at the above
conclusion about my intended meaning, and indeed the same way that we
arrive at any conclusions about reality: by inference.
I infer this conclusion about reality – that there is linguistic
structure in the acoustic effects of my speaking – from the fact that
you construct the same perceptions of language structure that I do –
namely, the above sentence (prior to your associating any meanings with
it or deriving any inferences from it).
(Any fluent speaker of English will perceive the same linguistic
structures; but it is certainly possible, maybe even very probable, that
any two such fluent speakers of English will attribute somewhat different
meanings to it, and derive different inferences from it.)
My inference that there is linguistic structure in our acoustic
environment seems unavoidable because your only relevant inputs for
constructing these perceptions are the physical effects of my control
actions. There is no other channel by which that information can get from
my output functions to your input functions.
This does not mean that every structural feature that is perceived has a
direct physical counterpart in the environment. The information in the
speech signal is virtually always degenerate. But the physical effects of
my speaking include physical transforms or traces of enough aspects or
parts of that linguistic structure for you to construct the rest. Your
(re)construction of the linguistic information that I intended is based
on your control of those physical effects of my speaking, but could not
be accomplished without the redundancies in language. Language is highly
redundant, and we use that redundancy to help us construct perceptions
for which the evidence in the speech signal is incomplete, ambiguous, or
even absent.
There is redundancy at every level of language structure. Speakers may
reduce redundant or low-information forms. I’ve given various simple
examples: pigpen in a subordinate clause vs. in a freestanding
sentence (phonetics of unstressed or de-emphasized phonemes); can
not, cannot, can’t; sgwout for let’s go
out (phonemic shape of morphemes); John plays piano and Alice
violin (repeated words in syntactic context). That hearers undo these
reductions (Alice plays violin? I didn’t know that!) suggests
control of redundancy. However, for the present discussion it does not
matter whether these redundancies are controlled perceptions or effects
of control.
The differences of formant contours for the same consonant before
different vowels provides another case in point. First, the contrast is
with other consonants.
And the transient is the same for all consonants produced at that point
of articulation. The second formant comes up from below for b,
p, or m; it comes down a bit from above for d,
t, or n; and it comes down from much farther above for
ga, ka, or nga.
Even when you compare the various allophones of the same consonant before
different vowels, the direction from which the transient moves to the
location of the vowel formant is constant. For voiceless aspirated
consonants like p, t, and k, less of that transient
is clearly audible. (Compare the formants in the voiceless h at
the end of the spectrogram for ba.) For nasal consonants m,
n etc., which are fully voiced, a formant turns up at the origin
of the transient to the following vowel. In the course of babbling a baby
doubtless learns that those sound squigglies do that when you
close your lips and say mama, or baba, or papa, and
that when you touch your tongue tip just so and say dada,
or nana, or tata, and this when you touch that other
part of your tongue to the top of your mouth and say gaga or
nganga or kaka.
Surely, like any environment variable these can be disturbed. The noise
masking experiment that Rick mentioned is an example of such disturbance.
Bill Powers (2003.10.24.1226 MDT)–
Something is certainly in the environment, but
is itliterally shared? To create structures that we experience as
linquisticstructures, we must do things to the environment, but what is done to
theenvironment need only evoke the proper perceptions; it need not itself
havethe same structure.
Whatever it is in the environment, it evokes the same perceptions
(phonemes, syllables, morphemes, words, in a certain sequence, under
certain intonation contours) in more than one person simultaneously.
These hearers each independently associate meanings with these language
perceptions and draw inferences from those meanings, but they perceive
the same linguistic structure as a common basis for these associations
and inferences. Repetition demonstrates that they control the identical
linguistic structure. That seems to me to be a pretty strong
demonstration that whatever it is in the environment when one of them
produces a repetition, it’s common input to all of them. That’s what I
mean by shared. If you mean something different by the word shared, then
suggest a technical term that won’t be a source of confusion for
you.
Does the environment control behavior in this scenario? Whatever the
common input may be which is shared in the environment, it is itself a
product of behavior. It is the means by which the speaker controls a
perception of having spoken as intended. But it is also the means by
which the speaker controls a perception of being heard as intended. If
your intention as listener is to hear the speaker as the speaker intends,
then yes, you are controlling for the speaker to control your perceptual
input. Does the speaker thereby control your behavior? Well, is
recognition behavior? Is verification that you have recognized correctly
behavior? Then so be it. Communication, understanding, and cooperation do
seem to require that we intentionally put ourselves at one another’s
disposal at least to this extent.
/Bruce
Nevin
(Attachment 4e7d0ea1.jpg is missing)
···
At 01:38 PM 10/24/2003 -0600, Bill Powers wrote: