Language facts

[From Bill Powers (930303.0830)]

Bruce Nevin (930302.1430) --

... I can imagine seeing your glasses in your hand from your
point of view, but unless I have considerable experience with
recording equipment I cannot imagine the acoustic image of my
voice as others hear it.The reference perceptions must be for
the subjective (intracranial) acoustic image, not for the voice
that others hear.

That doesn't matter. EACH person speaks in such a way that the
production is perceived as similar to (or deliberately different
from) what is heard when others speak. The similarity is
perceived only above a certain level (obviously it doesn't feel
the same kinesthetically), and it's only at that level and above
that there can be imitation. There can be great variation at the
lower levels -- as you showed some time ago in remarking that the
kinesthetic aspects of phoneme production vary widely from one
person to another without altering the phoneme that is heard.

The reproduction of heard speech is more analogous to putting the
fork "in" the water glass than to imitating the way the
spectacles are held. The aspects of speech that are reproduced
are those that are independent of point of view, while
reproducing the orientation of the glasses specifically requires
a transformation of the point of view.

So the reference perceptions must be adjusted and established
with respect to factors in others' responses to us.

They must be adjusted and established with respect to WHAT WE
PERCEIVE OF factors in others' responses to us. There's no way
for any individual to know what those factors "actually are."

Do they recognize us as the kind of person we intend to present
ourselves as?

Finding out how others recognize us is a matter of how we
interpret our perceptions of their behavior. I have not noticed
that people are particularly good at grasping the kind of person
I perceive them as, nor am I very good at doing vice versa. Since
effects on other people's perceptions are not actually
observable, one can only guess what they are by comparing others'
behavior with one's own: "If I acted that way, I would be
perceiving the other person in a certain way." There is no way to
verify this guess; one can only look for consistency in the
evidence under one's model of the other person.

Do they accord to us social roles and expectations that are
acceptable or (better, of course) that we desire?

Nobody can "accord us a social role." Another person can behave
as if we are playing a certain social role; that is quite
independent of the social role we see ourselves playing. If these
independent perceptions don't lead to actions and consequences
compatible with both perceptions of social role, there will be
conflict. If they do avoid conflict, there is still no reason to
think that the two perceptions are the same. They are simply not
in conflict.

Do we get a balance of our desire for independence vs. our
desire for acceptance and intimacy that is acceptable to us?
Oops! Do we even know how to talk about independence vs.
intimacy in PCT terms?

To "desire independence" is to misinterpret human nature
according to PCT. The basic problem of human existence is that we
ARE independent, whether we like it or not. Each person is free
every waking movement of the day to set any goal whatsoever,
including the goal of stepping off a cliff or stepping into
traffic. All that prevents selecting such goals is a potential
conflict with other goals -- nothing external has the slightest
say in the matter.

Our actions have consequences outside us; we can't determine what
those consequences are to be. But we can select any goals we
please for those consequences: that they should occur, that they
should have certain forms, that they should not occur.

Independence is not a problem, but intimacy is. All of us, I
think, want to find evidence of a person like ourselves -- at
least of the same basic nature -- behind the exterior appearances
that others present to us. We do not want to be the only sentient
being; we don't want to be alone. So we make great efforts to
contact the person behind the facade, and we set goals that
include doing what is necessary to achieve that contact -- if we
can figure out what is necessary.

This is where a seeming problem with independence arises, because
the other person has goals too, and conflict with them lessens
the contact with the other person. We often find that in order to
entice the other into revealing what we want to experience, we
come into conflict with other goals of our own; to achieve this
intimacy we must do things we don't want to do. But this is still
an internal conflict, because nothing prevents us from giving up
any goal, including the goal of knowing the other, or the goal of
not doing certain things. The basic problem is achieving internal
consistency among our goals.

That [is] what I intended to communicate -- that the illusion
of the glottal stop would appear only in speakers who had to
prevent themselves from extending the vowel into a diphthong
in order to imagine matching the heard vowel.

Now why would a speaker of one language *have* to end a
word-final stressed syllable with a diphthong or a consonant?
Clearly, there is no physiological or neurological constraint,
which would have to apply to speakers of all languages.

Of course not. I said they "had to"

prevent themselves from extending the vowel into a diphthong
in order to imagine matching the heard vowel.

The "in order do" supplies the reason they have to prevent
themselves. The goal is to match the heard vowel. But the vowel
they are used to hearing is extended into a dipthong, so as
speakers of a given language that is their normal reference
perception for the vowel. When they imagine saying "no" they
imagine "no-ooo", but to say "no" in the Spanish way they must
prevent the "ooo" from appearing, which creates an error in the
normal control system that they have learned. At a higher level
they experience this as cutting the vowel short, as in a glottal
stop. People whose reference signal is the Spanish "no" would
have no experience of the illusion. (My own illusion is that the
"o" is nasalized -- sort of the opposite of a glottal stop).

Rather, there is a socially instituted pattern on the basis of
which speaker/hearer's expectations get set.

This is not an explanation, but a description. There is a social
pattern, which means only that most of the people with whom the
test person interacts speak in a similar manner. The existence of
this pattern can't, by itself, have any effect on anyone. If a
person's speech changes when the person interacts extensively
with this group, that is only because selecting the group's way
of speaking suits the person's purposes. There are plenty of
cases to show that there is nothing necessary about the social
influence on pronunciation -- just consider all the British
receptionists who were hired for the neat accent, and make sure
they don't lose it because their job depends on it. Or the movie
and TV actors. Or the taxi drivers from Brooklyn. Or certain ex-
Europeans who consider American phonemes ugly and refuse to adopt
them. You can't base a serious law of speech acquisition on a
generalization with so many exceptions.

Expectations are set, according to HPCT, by higher-level systems,
not by perceptions. All that perceptions can do is provide raw
material in memory, from which higher systems select the
instances that fit their goals. Most people can imitate accents
other than the ones they normally use, at least well enough to be
speaking in an obviously different way. It is not the set of
recorded perceptions of speech that determine the accent we
choose, but our goals for our relationships to those around us.
Some people think (almost always mistakenly) that the way to
entertain people around them is to tell jokes in dialect. They
may not achieve an entertained state in those who hear the joke,
but that is certainly the goal, and the correct explanation for
the way the joke-teller is pronouncing words at the moment. I
have observed that people who tell such jokes also perceive that
the others are entertained, which explains why they persist. They
perceive themselves in a social role that matches what they want
it to be; the others perceive something else, but their self-
images prevent them from being frank.

We feel a need to be consistent with one another, and out of
this does develop impressive consistency in social conventions
or norms associated with subpopulations to which people
identify themselves as belonging.

That I can agree with. Particularly if you add that most people
would identify themselves as belonging to a number of different
subpopulations (e.g., a sports-loving homosexual stockbroker
living with her Vietnamese husband in Bedford-Stuyvesant).

But we don't have to identify subpopulations "objectively."
People--the members of the social groups--do it for us. Or
rather they do it for themselves, and for each other.

Do you mean that they walk up to you and tell you what
subpopulation they belong to, so you don't have to do any
interpreting of your own?

And
social dialect is an excellent index when you consider natives
of a place like NYC, for example. (Don't take my word for it,
have a look at Labov's methods and results in the papers in
_Language in the Inner City_ f'rinstance.)

You can't run that logic both ways at once. If dialect is used to
identify the subpopulation, then you can't prove that there is
any shared characteristic other than the dialect, in which case
all you've shown is that some people speak with that dialect and
some don't. Either you pick the dialect as the discriminator, or
some other set of characteristics like living in the inner city.
To prove an association you'd have to show that not using the
dialect means you don't live in the inner city, or that living in
the inner city means that you have to speak with that dialect,
and the two converse cases. If ALL the cases don't prove out
every time, then all you've got is one of those statistical
generalizations that's true of a population but not of any
particular person. You certainly don't get any explanation for
why one person speaks in a particular dialect on a particular
occasion, and another doesn't.

Sure, there are people that don't fit in, including obviously
tourists and immigrants from other regions, but also effortful
misfits of one sort or another.

So everyone is the same except for those who are different. This
is, if you will excuse me, the usual kind of excuse that is made
for statistical generalizations. The longer you examine the
actual individuals involved, the more excuses are needed. This is
why I have no use for "facts" about individual behavior that are
based on statistics. They're facts, except when they're not.

This is simply true: there are people that don't fit in the
major social strata and groupings of any given community.
Sometimes their seeming idiosyncrasies turn out to be negations
or reversals of the norms of one or more existing groups, by
which they deny membership. Sometimes they take on
characteristics associated with one or more truly "foreign"
groups (the guy from Topeka who talks with a put-on British
accent; Bobby Zinnaman's knock-up of Ramblin' Jack Elliott in
his emerging persona as Bob Dylan). And some people may truly
bridge or alternate membership in more than one local group.

More of the same. Excuses, excuses. Let's face it: this way of
understanding people just doesn't work. So-called social strata
and groupings and norms exist mostly in the imagination of the
statistician. They are perceptual prejudices. If you pick any
individual at random and try to apply your generalization, you
will find that you need a dozen excuses to explain why this
person doesn't fit the category or behave like others whom you
put in the same category. When you finish with the excuses (none
of which is in fact investigated to see if it actually fits), all
you have done is to discard from your sample all the individuals
who don't fit the generalization. If you go through the
population person by person in this way, you may end up with
nobody left.

Just because social psychology and sociology have often been
inept handmaidens of government (and employers), trying to keep
their hands aseptic and "objective", does not discredit social
groupings as something that people identify and identify with
or against.

I have not said that ONLY statisticians commit this sort of
foolishness. People who informally identify with social groupings
don't know any more about what is actually a social grouping than
statisticians do -- if anything, their concepts are even looser
and less consistent, and would lead to confusion even sooner
under any kind of detailed questioning. Everyone imagines living
in a certain kind of society, which doesn't actually exist and
isn't like the society that the next person believes in. When any
individual says "us" you haven't the least idea what he means,
and neither, come to think of it, does he.

Society, in my view, is a continuum in which individuals vary
along a hundred dimensions in ways independent of each other. Yet
the members of the society like to look at superficial secondary
characteristics like where one lives, how one dresses, how one
speaks, what one does for a living, what one says aloud about
beliefs, and so forth, and forms perceptions of groupings that
have almost nothing to do with what makes each person that person
and no other. Worst of all, the individuals then begin to deal
with others whom they see as sharing these apparent patterns of
secondary characteristics as if they were all right in the
middle, and all alike. The use of formal statistics to define
such groupings simply makes the prejudice harder to recognize for
what it is.

Has anyone here read Deborah Tannen's work on manners of
speaking? _You Just Don't Understand_ is the most popular one,
on stereotypically male-female differences in conversational
style and problems that follow from them.

It's really a book about what's wrong with stereotyping -- that
is, with speaking in terms of averages of populations -- isn't
it?

···

----------------------------------------------------------
Best,

Bill P.

[From: Bruce Nevin (Fri 93035 14:20:00)]

[I wrote this in my last hour here Friday, covering the easier half
of my notes scattered through Bill's (930303.0830). When I came in
today 3/8 I found it poised ready to send--but I had never pressed
the carriage return! I'll resist temptation to review it before
sending (that would drag through another week). So, warts and all.]

( Bill Powers (930303.0830) ) --

This must be terse, I hope not too telegraphic.

I think my (930302.1430) was a disturbance because you believe I
am saying that social groups, strata, classes, etc. control
individual people. I am not, I did not, and I don't expect to.

If you now reread what I have said with this in mind, I think
that many of your objections and comments will have no occasion.

It is very easy to mix discussion of matters of different logical
type. To pick a set of distinctions that have often bedevilled
our discussions, are we talking about language learning in the
child, established reference perceptions for language in the
adult, the evolutionary origins of language, the immediate
historical origins of some characteristics of a particular
language such as English? It makes a difference. Or (in this
post) are we talking about acoustic and kinesthetic perception of
phonetic differences between dialects, or about phonemic
contrasts in which these phonetic differences don't make any
difference? Or about the perceptual control by means of which we
arrange that they don't make any difference?

Some individual points:

kinesthetic aspects of phoneme production vary widely from one
person to another without altering the phoneme that is heard.

Some kinesthetic aspects of some phonemes. In certain regions of
the mouth, differences in articulation of the tongue make little
or no acoustic difference. These (acoustic and articulatory)
regions that are relatively disturbance-proof tend to be favored
as "targets" for phonemes in various languages. However, there
are exceptions (usually languages in process of change, I think),
and there are other sorts of phonemes that are not so nicely
favored by the acoustic properties of the vocal tract. I would
be very cautious about overgeneralizing here.

The reproduction of heard speech is more analogous to putting the
fork "in" the water glass than to imitating the way the
spectacles are held. The aspects of speech that are reproduced
are those that are independent of point of view, while
reproducing the orientation of the glasses specifically requires
a transformation of the point of view.

Repetition in my dialect of what I heard you say in yours, yes.
It doesn't matter that I'm using a table fork and a tumbler where
you had a pickle fork and a goblet. Imitation of manner gets
into mimicking detailed differences that don't make any
difference at the higher level of phonemic contrast: my way of
saying the diphthong in "light" vs. yours. For repetition, it
doesn't matter that I realize the contrast (between that
diphthong and, say, the "a" of "barn") at a different point in
the space of acoustic possibilities available to us both. I do
maintain the contrast and each of us is able to map my way onto
your way and vice versa, so we agree that it is the same
contrast, and so indeed it is. But if I am to reproduce your
speaking, I must learn to control the contrast in the same manner
that you do--to speak your dialect.

As a child learning the language, I imitate until I learn the
contrasts. Then I repeat. As an adult, I no longer imitate (or
rarely), I repeat in my own manner. If I were mirroring your
dance movements, comparable differences would be errors; in
repeating what you say, they are not errors, they are differences
that don't make any difference, below the level of phonemic
contrast.

Adapting your words, the aspects of speech that are REPEATED are
those that are independent of point of view, while the aspects of
speech that are IMITATED specifically require a transformation of
the point of view. The terms of repetition (phonemic contrasts,
morphemes, words, etc.) are independent of point of view because
they are socially instituted.

I said they "had to"

prevent themselves from extending the vowel into a diphthong
in order to imagine matching the heard vowel.

The "in order do" supplies the reason they have to prevent
themselves. The goal is to match the heard vowel. But the vowel
they are used to hearing is extended into a dipthong, so as
speakers of a given language that is their normal reference
perception for the vowel.

They "had to" prevent themselves from extending the vowel into a
diphthong in order to imagine matching the vowel they IMAGINED
that they heard. They substitute an imagined perceptual signal,
the reference perception for controlling perception (at a higher
level) that amounts to the patterning in their language.
Before Sapir's students learned about the glottal stop, they
substituted an imagined diphthong, because they could imagine no
consonant there. After they learned about the glottal stop, they
substituted an imagined consonant (the other possibility in the
English pattern for stressed final syllables). This imagined
consonant was the new one they had just learned to write.

(Incidentally, the first part of the o is nasalized in both
English and Spanish "no", but in Spanish the second part, which
is denasalized, is shorter than in the English diphthong.)

Rather, there is a socially instituted pattern on the basis of
which speaker/hearer's expectations get set.

This is not an explanation, but a description. There is a social
pattern, which means only that most of the people with whom the
test person interacts speak in a similar manner. The existence of
this pattern can't, by itself, have any effect on anyone.

The pattern is socially instituted in the individual. The
individual tries very hard to learn and participate in the same
patterns as others, and the child is typically assisted in this
by adults. Look again at Bruner, _Child's Talk_.

If a
person's speech changes when the person interacts extensively
with this group, that is only because selecting the group's way
of speaking suits the person's purposes.

Of course. And that is how and why (a) the individual institutes
the prevalent social patterns in his or her own reference
perceptions and (b) by participating in them contributes to their
ongoing institution as social facts that children learn, etc.

There are plenty of
cases to show that there is nothing necessary about the social
influence on pronunciation -- [...]
You can't base a serious law of speech acquisition on a
generalization with so many exceptions.

I never proposed such a simplictic "law of speech acquisition."
Nor did I say that social patterns, etc., control people. Take
the Martha's Vineyard example. Kids are exposed to a MV dialect
and various mainland dialects (mainly Boston/New England and New
York) all their lives. At puberty, they pick one. THEY do the
picking. The dialect comes with the package when they choose an
image of self for themselves. But they PICK from a range of
possibilities that is socially available. They don't invent
something entirely new. (Even a nonconformist creates a persona
by patching together available materials that are meaningful to
others in intended ways. We all know about the nonconformist
uniform, reinvented with each generation. An individual may
create some unexpected combination or incorporate a new element
in an existing pattern, and is very likely thereby to start a
fashion.) In this sense individuals are constrained in their
choices (even when choosing "against" prevailing norms) by the
social patterns that prevail at the time. But this is not to say
that the social norms control them.

The examples you give as exceptions to the supposed "law of
speech acquisition" are not exceptions, they are each a case in
point. They learned their manner of speaking in childhood and
established an adult persona in their teens which subsequently
they resisted changing.

Character actors, on the other hand, cultivate an ability at
mimicking dialect, body language, etc. to project a persona with
socially identified characteristics. Remember the Oklahoma study
I cited quite a while back? Four grad students in the speech
department recorded multiple readings of a short text, varying
things like speed, pitch variation, nasality, orotundity (not
sure how that was defined), etc. Audiences listening to the
tapes perceived these as many different people, and their
evaluations of the probable personality characteristics of these
different people were virtually unanimous. These correlations of
speech characteristics with perceived personality traits must be
known to the child choosing what to include in a persona, it
seems to me; they apparently were known to the college students
participating in this study, and are known to the audiences of
storytellers and character actors.

Most people can imitate accents
other than the ones they normally use, at least well enough to be
speaking in an obviously different way. It is not the set of
recorded perceptions of speech that determine the accent we
choose, but our goals for our relationships to those around us.

When most people imitate accents, they do so laughably, and would
not be mistaken by anyone as a native speaker of the dialect they
are imitating. But more important here is the question: how,
then, do our goals for relationships with those around us have
any connection with our pronunciations of words? I have proposed
a basis for such a connection. What do you propose as a basis?
Or is my proposal more acceptable to you, now that you know I am
not saying that social structures control individuals?

To "desire independence" is to misinterpret human nature
according to PCT. The basic problem of human existence is that we
ARE independent, whether we like it or not. Each person is free
every waking movement of the day to set any goal whatsoever,
including the goal of stepping off a cliff or stepping into
traffic. All that prevents selecting such goals is a potential
conflict with other goals -- nothing external has the slightest
say in the matter.

So PCT tells us. So we may come to perceive. But so most people
most of the time in fact do NOT perceive. That is one reason PCT
is revolutionary. But many people do perceive their independence
threatened by the actions of others. That is their perceptual
reality.

Do you mean that they walk up to you and tell you what
subpopulation they belong to, so you don't have to do any
interpreting of your own?

Yes. But not by name. Ostensively. If you want to name their
social affiliations or talk about them, you have to do some
interpreting. You may not recognize all that they are "telling"
you (or would be telling, if you had learned the patterns they
are participating in but you are not).

Everyone imagines living
in a certain kind of society, which doesn't actually exist and
isn't like the society that the next person believes in. When any
individual says "us" you haven't the least idea what he means,
and neither, come to think of it, does he.

(A) How do you know? (B) These perceptions, real or imagined,
constitute an important part of his real world.

[Tannen's books]:

It's really a book about what's wrong with stereotyping -- that
is, with speaking in terms of averages of populations -- isn't
it?

Nope. What goes wrong in communication because of being members
of groups whose norms of "conversational style" differ.

Got to run. Snow and blow still in progress. Train to catch.
Until Monday, then,

        Bruce Nevin
        bn@bbn.com