[From: Bruce Nevin (Fri 920327 12:20:31)]
It will take me some time to catch up. Many jottings are scattered over
many days' worth of printouts of CSG mail. I won't try that now,
probably piecemeal and unsystematic over time.
For now, a couple of issues that have particular prominence for me.
···
-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-
Martin, your HCI talk. I wish I could plunge into this a lot more. I
suggest a major contribution of HPCT to HCI might be in the research
prior to design and in usability testing after.
How can variations on the Test be used to determine what the user cares
about? The big prior problem everybody knows about is finding out what
the user's goals are, or rather what they would be if the user already
had the optimally designed product incorporated into the work
environment. But the user cares about things not so obviously connected
to work goals and the tasks required to effect those goals. And we have
to learn how to disturb those cared-about perceptions enough for users
to become conscious of them and complain that the disturbance is
interfering with doing their job.
Usability testing often boils down to the old time-motion studies. How
many keystrokes or mouse movements. Can we get it reframed in terms of
how many disturbances to controlled perceptions, and of these
disturbances how many reflect factors that the designers (and the user's
boss) think of as goals and tasks vs how many are extraneous
interference.
Related to this, Bill, your concern about crib sheets. Yes, you want to
index the commands by keywords for uses of the commands. But you have
to recognize that this is and index, not a list. For each command thee
may be multiple possible uses, for each use there are inevitably
multiple possible ways of verbalizing the use, and in each verbalization
there are probably multiple keywords to be pulled out for index
pointers. This is why your crib sheet for PictureThis will differ from
one I might create. It also explains why you don't see it done that way
as often as you might wish. The expansion into an index conflicts with
the aim of having a compact lookup table.
-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-
Bill (920318.0800) --
Continuing where I left off (920320 8:41):
I need to be clearer what I mean when I say (your quote of my 920317)
orienting one's behavioral outputs to social
conventions or norms
Let me introduce the notion of rule-oriented control. Whether one
controls for conformity to norms or rules, to flout convention, or to
change social conventions, all such control is rule oriented. Even
ignoring convention may be rule oriented unless it is done truly in
ignorance of the relevant conventions. Omitting something called for by
one rule, if the carrying out of that rule is redundant (predictable),
can itself become a further rule or convention.
Now, it seems to me that you are treating the control of the reductions
("expansions") in language in the same way that you treat the variable
means by which a control system attains consistent ends in a
disturbance-prone environment
/ variable \
Stimulus --- pressings --- food pellets
\ of bar /
The difference is that the variable behavioral means by which the
experimental rat controls the appearance of food pellets rat is not rule
oriented, but the variable alternative representations of meanings by
verbal means is rule oriented.
You want to make the link:
Linguist--> Expansion rules
/ \
/ \
/ \
/ \
/ -------------------
/ \
/ \
Heard (reduced) form Expanded form
> >
> >
initial meaning ---> imagination ---------> expanded meaning
But the relation between reduced and unreduced forms is much more
constrained than the relation between reduced forms and whatever
elaborations on them that one might come up with out of the play of
memory and imagination. And those constraints have precisely to do with
rule-oriented control of language behavioral outputs. This
rule-oriented control of language behavioral outputs goes on in
pandaemonium parallel with control of meanings--remembered and imagined
nonverbal perceptions. The two parallel tracks look like this:
rule-
Heard (reduced) form ----> oriented ---------> Expanded form
: control :
: :
: :
initial meaning ---> imagination ---------> expanded meaning
The lines of dots represent whatever means it is by which words are
linked with nonverbal perceptions, be it associative memory and
imagination or the hidden innards of category-level ECSs or whatever.
The essential point is that people control for linguistic form in
parallel for controlling for meanings.
I will readily grant that the linguist's attempt to describe rule-
oriented control of linguistic form bears no divine cachet guaranteeing
its veracity. No one has seen inside the black box yet. It is an
attempt to model it from the outside on the basis of regularities in
its determinable inputs and outputs. Its relation to the above is as
follows:
+--> Theory of language & information
> & cognition &c
Linguist--+ |
/ \ | V
/ \ +--> Description (theory) of this
/ \ language
/ \
/ -------------------
/ \
/ rule- \
Heard (reduced) form ----> oriented ---------> Expanded form
: control :
: :
: :
initial meaning ---> imagination ---------> expanded meaning
Different linguists guided by different theories come up with different
descriptions. You are a linguist offering your own description and your
own theory of language--a limiting case, in which there is claimed to be
NO rule-oriented control of linguistic form.
To make this stick, you have to account for people's response to
disturbances, as for example when they encounter a typographical error,
a misspelling, a socially marked usage (can't get none), and so on. I
am imagining disturbances that in no way prevent comprehension of
meaning, but are corrected nonetheless.
You have to account for the way people repair errors and false starts
not just anywhere with any word but at the syntactic "joints" of
sentences. (If you don't know what I mean you may just have to hold
your nose and get your hands a little soiled with the details of how
languages are integumented.)
You have to account for the fact that sentences are related to other
sentences not only in meaning but also in a mesh of forms, where each
step from one node to another of the mesh is one of a small set of
elementary sentence differences. You have to account for the fact that
this property is maintained over time and across language varieties,
such that time-stages and varieties differ in respect to which portions
of this mesh are marginal, which are daggered "infra-sentences," and
which are completely normal for the given variety in the given time-
slice of its history.
All of this and much more you are simply ignoring, Bill, apparently
under the supposition that it bears the same relation to control as the
varieties of the rat's responses bear to the stimulus (disturbance) and
attainment (reference perception) in Guthrie's lens. If you need more
persuasion that there is no parallel, let me know. I'll try again.
Bill (920320.1100) quoting my 920320 --
Language in general is a way of representing nonverbal perceptions for
our inspection and study, and possibly for our manipulation. I don't
care whether these sentences expressing background knowledge are
"basically" or "in origin" nonverbal or in language. By the
correlation of nonverbal perceptions with word-perceptions, nonverbal
perceptions become available for control as part of language, and
things in language like assertions, injunctions, maxims, attributions,
prohibitions, and instructions become available for setting nonverbal
reference perceptions.Frame this and hang it on the wall. I think this statement consolidates a
large amount of progress toward a coherent HCT theory of language. If you
keep talking like this, you can expand as many reductions as you like and
I'll remain meekly silent. You're starting to define the problem. The
better you can define it at this level of discourse, the quicker all those
messy details will fall into place.
I'm not sure you realize that construal of "sentences expressing
background knowledge" are not among the expansions I am talking about.
I suggested in the post quoted here that background knowledge of the
sort conveyed in Avery's (Jackendoff/Wierzbicka) formulations is work
done on the nonverbal side, in memory and imagination, and that I take
it constitutes agreement with your diagram above. But the expansions
and reductions that I am talking about involve rule-oriented control of
linguistic form, control that is oriented to socially constituted norms
regarding linguistic forms.
"I'm glad you agree with me . . . but what,
exactly, are you agreeing with?"
-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-=+=-
Avery (23 Mar 12:16 Australian EST) --
Just a quick note for now, I'm late for my train. Your new formula
responds to my question, how do you know these words are more basic than
"betray" by reference to Wierzbicka's attempt to establish a universal
vocabulary of semantic primitives. More on that later. For now, note
the use of X and Y. These exploit the capacity of language to use
arbitrary tokens as names. It rests upon a presumption that once a
name is used, a repetition of the name refers to the same individual.
This is nothing more than a metalanguage assertion of sameness, reduced
to zero under a socially established convention for the sublanguage of
"semantic representation" being used. A full semantic representation
must give it explicitly in the representation for a sentence. No
cop-outs.
Gotta run. Hasty bagles--and have a great weekend.
Bruce
bn@bbn.com