Limitations of Computers as Translation Tools

[Alexander Gross]

Well, now that the "Wednesday Night Massacre" seems to be over, even
though i'm pretty new here, i suppose i should take some of the
credit (or the blame) for what took place. So far as i can tell, i
sort of functioned as a kind of "wing commander" (to continue Arthur
Murray's brilliant aviation images), swooping down as the first
plane on the site, bouncing a few warning shots off the fuselage,
merely to mark the overall position of the "bogey."

i had no idea all the rest of you would be so quick off the mark
behind me (nor even how many of you there were), but i have to admit
you did a pretty good job (not that there's much honor in shooting
down a sightless target flying in unfamiliar air space without the
necessary tools to defend himself).

in some ways i guess i'm just as much an invader as he was. let me
start by apologizing for that, because i think i have some pretty
good reasons for "invading," if an invasion it truly is.

i've been very impressed by some of the ideas voiced here, and i'm
happy to see that some of the ideas i floated when i organized and
chaired the panel on "Where Do Human Translators Fit Into Machine
Translation?" at the 1991 MT Summit III Conference (which some of you may
even have attended) in Washingon are still around. Not because they came
from me, but because they sprang from the reality of translation and the
very real problems you have all been facing in
trying to port that reality from a human to a digital platform.

some of the ideas i expressed then and am about to express again now may
appear a bit radical to at least some of you. but the reason
why i don't think i really qualify as a "bogey" like the other guy
is that i have always tried to mind my academic p's & q's, citing,
publishing as well and often as i can, seeking permissions, attending
the occasional conference, and almost always submitting these ideas
to the most qualified peers i could find to read them.

Now that we have dealt with that interruption--or at least with what
i think were the causes for that interruption--i'd like to get back
to where i was before and try to present some of the reasons why i
think the quest for truly enhanced MT may be a vain one. as before,
i would appreciate any criticisms you may choose to make, as they
may well influence how i finally deal with the subject. and that
certainly includes negative criticisms as well, as long as the
person making them does not try to claim that the problem has been
totally licked already and he or she is just putting the final
touches on it.

Such claims lie beyond the realm of proof, and i would think they
ought not to belong here, though of course i cannot prevent anyone
from making them. as John Rehling has pointed out, there have been
too many loud cries of bloody triumph in the past, but once all the
bloody clouds had cleared away, little remained of these triumphs.

i've also made it amply clear in a number of published pieces about
MT that two unassailable arguments in its favor remain: 1) no one
opposes MT where it really works, and 2) MT works quite well for
those tasks where it is suitable. I believe MT and its offshoots
have a number of real applications around the edges of translation
(and even in a few specialized cases closer to the center), and that
the number of these applications will continue to grow--gradually
but by no means exponentially. But I have some good reasons for
supposing that the usual engineer's logic--break a problem down into
pieces, solve first 5%, then 10% of piece number one, etc., move on
to piece number 2 and do the same thing, ditto with all the other
pieces until all the pieces are 100% solved--i say, i have some good
reasons for supposing that this process will not work for
translation.

i hope you don't feel too deprived by this. and i really hope you
don't feel too surprised that someone representing translators
(though i have not made my living from translation for some years,
and that branch of translation which i do practice--theatre and song
lyric translation--is never likely to yield to MT) should appear
among you and actually dare to dispute your basic premises with
you.

Put it in ethnic terms--supposing you boasted for fifty years
running that any day now you would soon be able to put the entire
Bleamish people out of work--would you really be surprised to
discover that some rather angry Bleamish citizens had come for a
visit? Would you really dispute their right to discuss your most
basic priorities and methodologies with you? Would you dare to tell
them that they knew nothing about Bleamishness?

Yet this is the tone some of you take whenever an experienced
professional translator begins to speak of linguistics, AI, or MT.

As an example, one contributor recently took me to task for
denigrating the notion of "universal grammar." so sacred was this
notion to him that he even challenged me to prove that "universal
grammar" does not exist.

Can i prove this? i sure as hell can. and in a rigorously logical
way as well.

Please lend me your ears, and all will become clear.

Supposing i had shown you a geometrical figure and told you it was a
triangle.

And supposing you had challenged me to prove that it was a triangle.

I would simply have pointed to the three sides of this figure, asked
you to count them for yourself, and invoked Euclid to the effect
that by definition "all three-sided figures drawn with a
straight-edge on a single plane are triangles."

And if you had objected to this explanation, i would have patiently
explained that these are the rules of geometry, proceeding in a
fixed order including definitions, axioms, corollaries, and theorems.

But already i anticipate your objections: can there possibly be
any definitions as fixed and certain in the nebulous realm of
linguistics, you cry out, as the expressions of Euclid are for
geometry?

Oddly enough, there ARE, and i will now tell you what these
expressions are. Many of you will most probably have never heard of
them before, because linguistics was being taught in a fairly
haphazard way _even before_ the Chomskians came along. Since then
things have simply descended into chaos.

I repeat: There is no universal grammar.

There can be no such thing as "universal grammar" simply because the
statement that "a universal grammar exists" is an example of a
Bloomfieldian secondary response to language.

And by definition, as in Euclid, all Bloomfieldian secondary
responses to language are false.

Worse than false, they are simply irresponsible.

YES, I can hear you quite clearly! Whoever you are, wherever you
are, you are now shouting out with but a single voice:

"Hey, Alex, what the bluddy hell is a 'Bloomfieldian secondary
response to language!?' Where the hell did you dig that up from!?"

I can assure you that I dug it up from a far more responsible source
than the one Chomsky dug up his universal grammar fantasy from.

Once again, the reason you have never heard about this before is
that for two generations now almost no one has been studying
linguistics any more, they've only been studying Chomskianism and
constantly congratulating themselves over their supreme "cognitive"
conquest and how superior this doctrine must be to everything that
went before.

But do the students of today truly know anything about what went before?
For the most part, not a hell of a lot. Almost all they know is what
they have learned from Chomskian sources, often even further corrupted
with the odd dash of meme-mongering and/or francophile pomo-decon-recon
lit-crit titbits. And this truncated, tunnel-vision view of a
remarkably small section of the subject they have the nerve to call
"linguistics."

[yes, i know, even the Chomskians have now given up on "linguistics"
and are trying to get off the hook by calling it "grammar" or
"minimalism," but i'll leave that for another time.]

I hear some nervous rustling in the background. So let me reassure
you about one thing, nothing i am about to tell you is the least bit
secret, much less esoteric. Thank god pre-Chomskian and
non-Chomskian linguists also continue to flourish, and they can bear
me out that what i am about to tell you is more of the truth than
many of you have heard before.

Why just this past Sunday, the summer replacement for Safire in the
NY Times Magazine actually said the following in print:

"Nearly everyone has an opinion about language. In perhaps no
other field, scholars who study social attitudes towards
language say, is unscientific folk belief so consistently
enshrined in high-level policy. It is as if, they say, the
Federal Government seriously entertained a plan to harvest
cheese from the moon."

And make no mistake--Chomskianism itself is a part of that
"unscientific folk belief."

All of this is about to be spelled out in luxuriant detail. So I
hope i have tantalized and/or infuriated some of you sufficiently so
that you read the following piece, which comprises the middle
section of my paper "Limitations of Computers as Translation Tools,"
published in "Computers in Translation: A Practical Appraisal,"
edited by John Newton, Routledge, London & New York, 1992.

As the piece makes clear, (and as i am sure, most of you will agree)
there are many parallels between the theories of Noam Chomsky and
the underlying assumptions of "machine translation" researchers.
I am scarcely alone in making this connection, for as R. H. Robins
points out in his "A Short History of Linguistics," many "critics...
have ascribed his linguistic theory, doubtless because of the
extensive use of mathematical and logical symbols and the overtly
`scientific' style in which discussions between transformationalists
are carried on, to such immediate antecedents as machine translation
and the application of computers to linguistic analysis."

E.F.K. Koerner also clearly makes this connection in the document i
have twice cited by the URL for its on-line abridged version.
And much closer to home, right here on comp.ai.natlang, two MT
developers have already declared their devotion to the Great
Transformationalist.

Professor Chomsky has of course made his pro forma denial of these
connections, partly included here, but the similarities remain, as
do the many limitations of computer translation techniques and
Professor Chomsky's theories alike. The two are self-nourishing and
self-replicating, and I have just finished reviewing two 1998 German
books on MT whose authors look up to Professor Chomsky as an almost
godlike being.

in this piece i have also spelled out in some detail the notion i
have already mentioned once, that mathematics is merely a subset of
language. among my many readers for this paper, i asked a well-
regarded mathematician to review this section before publication:
either he found nothing to criticize or he preferred not to argue with
me. But that's no reason why you can't voice your thoughts about
this or any other aspect of what follows. as i say, i actively seek
out the opinions of my peers, and this is my aim here as well in
preparation for something else along these lines i am now writing.
your comments--however negative--will be most welcome and may form the basis
for changing the views i finally end up expressing.

The section of the paper preceding this one is a nuts-and-bolts
treatment of machine translation problems and techniques, though
punctuated with many insights from human translators. The section
following it presents mutual criticisms leveled by human
translators and MT developers at each other's attainments and
attitudes as well as some predictions for the future. On the whole
these predictions are rather optimistic. If there were an absolute
avalanche of requests, i suppose i could post the other two sections
as well. Those interested in further information should consult the
book itself, which is Copyright 1992 by its editor. Details on
other contributors and on purchasing the book appear at the end of
this piece. Full scholarly notes and bibliography for this section
of the paper are appended.

---------------------------------------

The Limitations of Computers as Translation Tools

Part Two

Deeper Limitations

NOTE: This section explains how changing standards in the
study of linguistics may be related to the limitations in
Machine Translation we see today and perhaps prefigure
certain lines of development in this field. Those only
interested in the practical side should turn immediately to
later sections of this paper [not included here].

Some practical limitations of MT and even of CAT
(Computer Assisted Translation) should already be clear
enough. Less evident are the limitations in some of
the linguistic theories which have sired much of the
work in this field. On the whole Westerners are not
accustomed to believing that problems may be insoluble, and
after four decades of labor, readers might suppose that more
progress had been made in this field than appears to be the
case. To provide several examples at once, I can remember
standing for some time by the display booth of a prominent
European computer translation firm during a science
conference at M.I.T. and listening to the comments of
passers-by. I found it dismaying to overhear the
same attitudes voiced over and over again by quite sane and
reasonable representatives from government, business and
education. Most of what I heard could be summed up as

1) Language can't really be that complex since
we all speak it;

2) Language, like nature, is an alien
environment which must be conquered and
tamed;

3) There has to be some simple way to cut
through all the nonsense about linguistics,
syntax, and semantics and achieve instant
high quality translation; and

4) Why wasn't it all done yesterday?

To understand the reasons behind these comments and why
they were phrased in this particular way--and also to
understand the deeper reasons behind the limitations of
computer translation--it may be helpful to go back to
the year 1944, when the first stirrings of current
activity were little evident and another school of
linguistics ruled all but supreme. In that year Leonard
Bloomfield--one of the three deans of American
Linguistics along with Edward Sapir and Benjamin Lee
Whorf (7)--was struggling to explain a problem that
greatly perturbed him.

Bloomfield was concerned with what he called `Secondary
Responses to Language.' By these he meant the things
people say and seem to believe about language, often in
an uninformed way. He called such opinions about
language `secondary' to differentiate them from the use
of language in communication, which he saw as
`primary.' People delivering such statements, he
observed, are often remarkably alert and enthusiastic:
their eyes grow bright, they tend to repeat these
opinions over and over again to anyone who will hear,
and they simply will not listen--even those who, like
the ones I met at MIT, are highly trained and familiar
with scientific procedures--to informed points of view
differing with their own. They are overcome by how
obvious or interesting their own ideas seem to be. (8)

I would add here that what Bloomfield seems to be
describing is a set of symptoms clinically similar to
some forms of hysteria. As he put it:

`It is only in recent years that I have learned to observe
these secondary ..... responses in anything like a
systematic manner, and I confess that I cannot explain
them--that is, correlate them with anything else. The
explanation will doubtless be a matter of psychology and
sociology.' (9)

If it is indeed hysteria, as Bloomfield seems to
suggest, I wonder if it might not be triggered because
some people, when their ideas about language are
questioned or merely held up for discussion, feel
themselves under attack at the very frontier of their
knowledge about reality. For many people language is so
close to what they believe that they are no longer able
to tell the difference between reality and the language
they use to describe it. It is an unsettling experience
for them, one they cannot totally handle, somewhat like
tottering on the edge of their recognized universe. The
relationship between one's language habits and one's
grasp of reality has not been adequately explored,
perhaps because society does not yet train a sufficient
number of bilingual, multilingual or linguistically
oriented people qualified to undertake such
investigations. (10)

Bloomfield went even further to define `tertiary
responses to language' as innately hostile, angry, or
contemptuous comments from those whose Secondary
Responses are questioned in any serious way. They would
be simply rote answers or rote repetitions of people's
`secondary' statements whenever they were challenged on
them, as though they were not capable of reasoning any
further about them. Here he seemed to be going even
further in identifying these responses with irrational
or quasi-hysterical behavior.

What was it that Bloomfield found so worrisome about
such opinions on language? Essentially he--along with
Whorf and Sapir--had spent all his life building what
most people regarded as the `science of linguistics.'
It was a study which required extended field work and
painstaking analysis of both exotic and familiar
languages before one was permitted to make any large
generalizations even about a single language, much less
about languages in general. Closely allied to the
anthropology of Boas and Malinowski, it insisted on
careful and thoughtful observations and a non-judgmental
view of different cultures and their languages. It was
based on extremely high standards of training and
scholarship and could not immediately be embraced by
society at large. In some ways he and his colleagues
had gone off on their own paths, and not everyone was
able to follow them. Whorf and Sapir had in fact both
died only a few years earlier, and Bloomfield himself
would be gone five years later. Here are a few of
the `secondary' statements that deeply pained Bloomfield
and his generation of linguists:

Language A is more _____ than language B.
(.........`logical,' `profound,' `poetic,' `efficient,'
etc., fill in the blank yourself)

The structure of Language C proves that it is a
universal language, and everyone should learn it as a
basis for studying other languages.

Language D and Language E are so closely related that
all their speakers can always easily understand each
other.

Language F is extremely primitive and can only have a
few hundred words in it.

Language G is demonstrably `better' than Languages H, J,
and L.

The word for `________' (choose almost any word) in
Language M proves scientifically that it is a
worse--better, more `primitive' or `evolved,' etc.--
language than Language N.

Any language is easy to master, once you learn the basic
structure all languages are built on.

Summarized from Bloomfield, 1944, pp. 413-21

All of these statements are almost always demonstrably
false upon closer knowledge of language and linguistics,
yet such opinions are still quite commonly voiced. In
this same piece Bloomfield also voiced his sadness over
continual claims that `pure Elizabethan English' was
spoken in this or that region of the American South (a
social and historical impossibility--at best such
dialects contain a few archaic phrases) or boasts that
the Sequoyan Indian language was so perfect and easy to
learn that all citizens of the State of Oklahoma should
study it in school. (11)

What he found particularly disturbing was that this sort
of urban linguistic folklore never seemed to die out,
never yielded to scientific knowledge, simply went on
and on repropagating itself with a life of its own.
Traces of it could even be found in the work of other
scholars writing about language and linguistics.

Bloomfield's views were very much a reflection of his time.
They stressed a relativistic view of language and culture and
the notion that languages spoken by small indigenous groups
of people had a significance comparable to that of
languages spoken by much larger populations. They
willingly embraced the notion that language, like
reality itself, is a complex matrix of factors and
tended to reject simplistic generalizations of any sort
about either language or culture. Moreover, Bloomfield
certainly saw his approach as being a crucial minimum
stage for building any kind of true linguistic science.

Less than ten years after his death these ideas were
replaced, also in the name of science, by a set of
different notions, which Bloomfield would have almost
certainly have dismissed as `Secondary Responses to
Language.' These new observations, which shared a
certain philosophical groundwork with computational
linguistics, constitute the credo of the Chomskian
approach, now accepted as the dominant scientific view.
They include the following notions:

All languages are related by a `universal grammar.'

It is possible to delineate the meaning of any sentence
in any language through knowledge of its deep structure
and thereby replicate it in another language.

A diagram of any sentence will reveal this deep
structure.

Any surface level sentence in any language can easily be
related to its deep structure, and this in turn can be
related to universal grammar in a relatively
straightforward manner through a set of rules.

These and related statements are sufficient to describe
not only the structure of language but the entire
linguistic process of development and acculturation of
infants and young children everywhere and can thus serve
as a guide to all aspects of human language, including
speech, foreign language training, and translation.

The similarity of these deep and surface level diagrams
to the structure of computer languages, along with the
purported similarity of the human mind to a computer,
may be profoundly significant. (12)

These ideas are clearly not ones Bloomfield could have
approved of. They are not relativistic or cautious but
universalist and all-embracing, they do not emphasize
the study of individual languages and cultures but leap
ahead into stunning generalizations. As such, he would
have considered them examples of `Secondary Responses'
to language.

In many ways these ideas reflect the America of the late
'Fifties, a nation proud of its own new-found dominance
and convinced that its values must be more substantial
than those of `lesser' peoples. Such ideas also
coincide nicely with a seemingly perennial need academia
feels for theories offering a seemingly scientific
approach, suggestive diagrams, learned jargon, and a
grandiose vision.

We all know that science progresses by odd fits and
starts and that the supreme doctrines of one period may
become the abandoned follies of a later one. But the
turnabout we have described is surely among the most
extreme on record. It should also be stressed that the
outlook of Bloomfield, Whorf and Sapir has never truly
been disproved or rejected and still has followers
today. (13) Moreover, there is little viable proof
that these newer ideas, while they may have been useful
in describing the way children learn to speak, have
ever helped a single teacher to teach languages better
or a single translator to translate more effectively.
Nor has anyone ever succeeded in truly defining `deep
structure' or `universal grammar.'

No one can of course place the whole responsibility for
machine translation today on Noam Chomsky's theories
about language--certainly his disciples and followers
(14) have also played a role, as has the overall welcome
this entire complex of ideas has received. Furthermore,
their advent has certainly also coincided with the
re-emergence of many other `Secondary Responses',
including most of the comments I mentioned overhearing
at M.I.T. Much of the literature on Machine Translation
has owed--and continues to owe--a fair amount to this
general approach to linguistic theory.

Overall understanding of language has certainly not
flourished in recent times, and the old wives' tale of a
single magical language providing the key to the
understanding of all other tongues now flourishes again
as a tribute both to Esperanto and the Indian Aymara
language of Peru. (15) Disappointment with computer
translation projects has also been widespread throughout
this time, and at one point even Chomsky seemingly
washed his hands of the matter, stating that `as for
machine translation and related enterprises, they seemed
to me pointless as well as probably quite hopeless.' (16)

Even such lofty notions as those favored by Turing and
Weaver, that removing `language barriers' would
necessarily be a good thing, or that different languages
prevent people from realizing that they are `really all
the same deep down,' could turn out to be `Secondary
Responses.' It may also be that language barriers and
differences have their uses and virtues, and that
enhanced linguistic skills may better promote world
peace than a campaign to destroy such differences. But
popular reseeding of such notions is, as Bloomfield
foresaw, quite insidious, and most of these ideas are
still very much with us, right along with the proof that
they may be unattainable. This is scarcely to claim
that the end is near for computers as translation tools,
though it may mean that further progress along certain
lines of enquiry is unlikely.

There are probably two compelling sets of reasons why
computers can never claim the upper hand over language
in all its complexity, one rooted in the cultural side
of language, the other in considerations related to
mathematics. Even if the computer were suddenly able to
communicate meaning flawlessly, it would still fall
short of what humans do with language in a number of
ways. This is because linguists have long been aware
that communication of meaning is only one among many
functions of language. Others are:

Demonstrating one's class status to the person one is
speaking or writing to.

Simply venting one's emotions, with no real
communication intended.

Establishing non-hostile intent with strangers, or
simply passing time with them.

Telling jokes.

Engaging in non-communication by intentional or
accidental ambiguity, sometimes also called `telling
lies.'

Two or more of the above (including communication) at
once.

Under these circumstances it becomes very difficult to
explain how a computer can be programmed merely to
recognize and distinguish these functions in Language A,
much less make all the adjustments necessary to
translate them into Language B. As we have seen,
computers have problems simply with the communications
side, not to mention all these other undeniable aspects
of language. This would be hard enough with written
texts, but with spoken or `live' language, the problems
become all but insurmountable.

Closely related here is a growing awareness among
writers and editors that it is virtually impossible to
separate the formulation of even the simplest sentence
in any language from the audience to whom it is
addressed. Said another way, when the audience changes,
the sentence changes. Phrased even more extremely,
there is no such thing as a `neutral' or `typical' or
`standard' sentence--even the most seemingly innocuous
examples will be seen on closer examination to be
directed towards one audience or another, whether by
age, education, class, profession, size of vocabulary,
etc. While those within the target audience for any
given sentence will assume its meaning is obvious to
all, those on its fringes must often make a conscious
effort to absorb it, and those outside its bounds may
understand nothing at all. This is such an everyday
occurrence that it is easy to forget how common it
really is. And this too adds a further set of
perplexities for translators to unravel, for they must
duplicate not only the `meaning' but also the
specialized `angling' to an analogous audience in the
new language. Perhaps the most ironic proof of this
phenomenon lies in the nature of the `model' sentences
chosen by transformational and computational linguists
to prove their points. Such sentences rarely reflect
general usage--they are often simply the kinds of
sentences used by such specialists to impress other
specialists in the same field.

Further proof is provided here by those forms of
translation often described as `impossible,' even when
performed by humans--stageplays, song lyrics,
advertising, newspaper headlines, titles of books or
other original works, and poetry. Here it is generally
conceded that some degree of adaptation may be merged
with translation. Theatre dialogue in particular
demands a special level of `fidelity.' Sentences must
be pronounceable by actors as well as literally correct,
and the emotional impact of the play must be recreated
as fully as possible. A joke in Language A must also
become a joke in Language B, even if it isn't. A
constantly maintained dramatic build-up must seek its
relief or `punch-lines' at the right moments. This may
seem far from the concerns of a publication manager
anxious to translate product documentation quickly and
correctly. But in a real sense all use of words is
dependent on building towards specific points and
delivering `punch-lines' about how a product or process
works. The difference is one of degree, not of
quality. It is difficult to imagine how computers can
begin to cope with this aspect of translation.

Cross-cultural concerns add further levels of
complexity, and no miraculous `universal structure' (17)
exists for handling them. Languages are simply not
orderly restructurings of each other's ideas and
processes, and a story I have told elsewhere (18) may
perhaps best illustrate this. It relates to a real
episode in my life when my wife and I were living in
Italy. At that time she did most of the shopping to
help her learn Italian, and she repeatedly came home
complaining that she couldn't find certain cuts of meat
at the butcher's. I told her that if she concentrated
on speaking better Italian, she would certainly find
them. But she still couldn't locate the cuts of meat
she wanted. Finally, I was forced to abandon my male
presumption of bella figura and go with her to the
market place, where I patiently explained in Italian
what it was we were looking for to one butcher after the
next. But even together we were still not successful.
What we wanted actually turned out not to exist.

The Italians cut their meat differently than we do.
There are not only different names for the cuts but
actually different cuts as well. Their whole system is
built around it--they feed and breed their cattle
differently so as to produce these cuts. So one might
argue that the Italian steer itself is
different--technically and anatomically, it might just
qualify as a different subspecies.

This notion of `cutting the animal differently' or of
`slicing reality differently' can turn out to be a
factor in many translation problems. It is altogether
possible for whole sets of distinctions, indeed whole
ranges of psychological or even tangible realities--to
vanish when going from one language to another. Those
which do not vanish may still be mangled beyond
recognition. It is this factor which poses one of the
greatest challenges even for experienced translators.
It may also place an insurmountable stumbling block in
the path of computer translation projects, which are
based on the assumption that simple conversions of
obvious meanings between languages are readily
possible.

Another cross-cultural example concerns a well-known
wager AI pioneer Marvin Minsky has made with his M.I.T.
students. Minsky has challenged them to create a
program or device that can unfailingly tell the
difference, as humans supposedly can, between a cat and
a dog. Minsky has made many intriguing remarks on the
relation between language and reality, (19) but he shows
in this instance that he has unwittingly been
manipulated by language-imposed categories. The
difference between a cat and a dog is by no means
obvious, and even `scientific' Linnaean taxonomy may not
provide the last word. The Tzeltal Indians of Mexico's
Chiapas State in fact classify some of our `cats' in the
`dog' category, rabbits and squirrels as `monkeys,' and
a more doglike tapir as a `cat,' thus proving in this
case that whole systems of animals can be sliced
differently. Qualified linguistic anthropologists have
concluded that the Tzeltal system of naming
animals--making allowance for the fact that they know
only the creatures of their region--is ultimately just
as useful and informative as Linnaean latinisms and even
includes information that the latter may omit. (20)
Comparable examples from other cultures are on record.
(21)

An especially dramatic cross-cultural example suggests
that at least part of the raging battle as to whether
acupuncture and the several other branches of Chinese
Medicine can qualify as `scientific' springs from the
linguistic shortcomings of Western observers. The
relationships concerning illness the Chinese observe and
measure are not the ones we observe, their measurements
and distinctions are not the same as ours, their
interpretation of such distinctions are quite different
from ours, the diagnosis suggested by these procedures
is not the same, and the treatment and interpretation of
a patient's progress can also radically diverge from our
own. Yet the whole process is perfectly logical and
consistent in its own terms and is grounded in an
empirical procedure. (18) The vocabulary is fiendishly
difficult to explain to non-specialists in this highly
developed branch of the Chinese language. No one knows
how many other such instances of large and small
discontinuities between languages and their meanings may
exist, even among more closely related tongues like
French and English, and no one can judge how great an
effect such discontinuities may have on larger
relationships between the two societies or even on
ordinary conversations between their all too human
representatives.

Just as the idea that the earth might be round went against
the grain for the contemporaries of Columbus, so the notion
that whole ranges of knowledge and experience may be
inexpressible as one moves from one language to another
seems equally outrageous to many today. Such a notion,
that Language A cannot easily and perfectly replicate
what is said in Language B, simply goes against what
most people regard as `common sense.' But is such
insistence truly commonsensical or merely another
instance of Bloomfield's `Secondary Responses?'
Something like this question lies at the root of the
long-continuing and never fully resolved debate among
linguists concerning the so-called Whorf-Sapir
hypothesis. (7)

Mathematical evidence suggesting that computers can
never fully overtake language is quite persuasive. It is
also in part fairly simple and lies in a not terribly
intricate consideration of the theory of sets. No
subset can be larger than the set of which it is a
part. Yet all of mathematics--and in fact all of
science and technology, as members of a Linguistics
school known as Glossematics (22) have argued--can be
satisfactorily identified as a subcategory--and possibly
a subset--of language. According to this reasoning, no
set of its components can ever be great enough to serve
as a representation of the superset they belong to,
namely language. Allowing for the difficulties involved
in determining the members of such sets, this argument
by analogy alone would tend to place language and
translation outside the limits of solvable problems and
consign them to the realm of the intractable and
undecidable. (23)

The theory of sets has further light to shed. Let us
imagine all the words of Language A as comprising a
single set, within which each word is assigned a
number. Now let us imagine all the words of Language B
as comprising a single set, with numbers once again
assigned to each word. We'll call them Set A and Set
B. If each numbered word within Set A meant exactly the
same thing as each word with the same number in Set B,
translation would be no problem at all, and no
professional translators would be needed. Absolutely
anyone able to read would be able to translate any text
between these two languages by looking up the numbers
for the words in the first language and then
substituting the words with the same numbers in the
second language. It would not even be necessary to know
either language. And computer translation in such a
case would be incredibly easy, a mere exercise in
`search and replace,' immediately putting all the people
searching through books of words and numbers out of
business.

But the sad reality of the matter--and the real truth
behind Machine Translation efforts--is that Word # 152
in Language A does not mean exactly what Word # 152 in
Language B means. In fact, you may have to choose
between Words 152, 157, 478, and 1,027 to obtain a valid
translation. It may further turn out that Word 152 in
Language B can be translated back into Language A not
only as 152 but also 149, 462, and 876. In fact, Word #
152 in Language B may turn out to have no relation to
Word # 152 in Language A at all. This is because 47
words with lower numbers in Language B had meanings that
spilled over into further numbered listings. It could
still be argued that all these difficulties could be
sorted out by complex trees of search and goto
commands. But such altogether typical examples are only
the beginning of the problems faced by computational
linguists, since words are rarely used singly or in a
vacuum but are strung together in thick, clammy strings
of beads according to different rules for different
languages. Each bead one uses influences the number,
shape, and size of subsequent beads, so that each new
word in a Language A sentence compounds the problems of
translation into Language B by an extremely non-trivial
factor, with a possible final total exceeding by several
orders of magnitude the problems confronted by those who
program computers for the game of chess.

There are of course some real technical experts, the
linguistic equivalents of Chess Grand Masters, who can
easily determine most of the time what the words mean in
Language A and how to render them most correctly in
Language B. These experts are called translators,
though thus far no one has attributed to them the power
or standing of Chess Masters.

Another large irony: so far the only people who have
proved capable of manipulating the extremely complex
systems originally aimed at replacing translators have
been, in fact.....translators.

----------------------------------------------------

EPILOG: I should perhaps add, writing now in 1999, that machine
translation, which seemed so important less than a decade ago, has
now been largely abandoned--except for a few academic enclaves and
installed industrial bases still devoted to it--in favor of a new
approach, called Translation Memory, based on massive lists of terms
and sentences in various specialties and specific language pairs.
It is still not a fully automatic process and requires considerable
expertise and oversight by a presiding translator. And it is still,
as was MT, applicable only for massive translation projects
involving a single technical vocabulary. And ironically enough,
this latest development owes no more to the theories of Noam Chomsky
or machine translation than a lexicon does to a log table.

--NOTES--

The following notes apply to the excerpt from the original article
and are shown here. A partial bibliography follows this section.

(7) Both Sapir and Whorf carried out extensive
study of American Indian languages and together evolved
what has come to be called the Whorf-Sapir Hypothesis.
Briefly stated, this theory states that what humans see,
do and know is to a greater or lesser extent based on
the structure of their language and the categories of
thought it encourages or excludes. The prolonged and
spirited debate around this hypothesis has largely
centered on the meaning of the phrase "to a greater or
lesser extent." Even the theory's most outright
opponents concede it may have validity in some cases,
though they see something resembling strict determinism
in applying it too broadly and point out that
translation between languages would not be possible if
the Whorf-Sapir Hypothesis were true. Defenders of the
theory charge that its critics may not have learned any
one language thoroughly enough to become fully aware of
how it can hobble and limit human thinking and further
reply that some translation tasks are far more difficult
than others, sometimes bordering on the impossible.

(8) Bloomfield, Secondary and Tertiary Responses
to Language, in Hockett 1970 , pp: 412-29. This piece
originally appeared in Language 20.45-55 and was
reprinted in Hockett 1970 and elsewhere. The author's
major work in the field of linguistics was Bloomfield 1933/1984.

(9) Bloomfield, in Hockett 1970, page 420.

(10) Since so many people in so many countries
speak two or more languages, it might be imagined that
there is a broad, widely-shared body of accurate
knowledge about such people. In point of fact there is
not, and the first reasonably accessible book-length
account of this subject is Grosjean. Some of this
book's major points, still poorly appreciated by society
at large:

Relatively few bilingual people are able to
translate between their two languages with
ease. Some who try complain of headaches,
many cannot do it at all, many others do it
badly but are not aware of this. Thus,
bilingualism and translation skills are two
quite different abilities, perhaps related to
different neurological processes.

No bilinguals possess perfectly equal skills
in both their languages. All favor the one
or the other at least slightly, whether in
reading, writing, or speaking. Thus, the
notion of being brought up perfectly
bilingual is a myth--much of bilingualism
must be actively achieved in both languages..

One does not have to be born bilingual to
qualify as such. Those who learn a second
language later, even as adults, can be
considered bilingual to some extent, provided
they actively or passively use a second
language in some area of their lives.

(11) Bloomfield, in Hockett 1970, pp. 414-16.

(12) Though presented here in summarized form,
these ideas all form part of the well-known Chomskian
process and can be found elaborated in various stages of
complexity in many works by Chomsky and his followers.
See Chomsky, 1957, 1965, & 1975.

(13) The bloodied battlefields of past scholarly
warfare waged over these issues are easily enough
uncovered. In 1968 Charles Hockett, a noted follower of
Bloomfield, launched a full-scale attack on Chomsky
(Hockett, 1968) Those who wish to follow this line of
debate further can use his bibliography as a starting
point. Hostilities even spilled over into a New Yorker
piece and a book of the same name (Mehta). Other
starting points are the works of Chomsky's teacher
(Harris) or a unique point of view related to computer
translation by Lehmann. Throughout this debate, there
have been those who questioned why these
transformational linguists, who claim so much knowledge
of language, should write such dense and unclear
English. When questioned on this, Mehta relates
Chomsky's reply as follows: `"I assume that the writing
in linguistics is no worse than the writing in any other
academic field," Chomsky says. "The ability to use
language well is very different from the ability to
study it. Once the Slavic Department at Harvard was
thinking of offering Vladimir Nabokov an appointment.
Roman Jakobson, the linguist, who was in the department
then, said that he didn't have anything against
elephants but he wouldn't appoint one a professor of
zoology." Chomsky laughs.'

(14) See for example Fodor or Chisholm.

(15) See Note 5 for reference to Esperanto. The
South American Indian language Aymara has been proposed
and partially implemented as a basis for multilingual
Machine Translation by the Bolivian mathematician Ivan
Guzman de Rojas, who claims that its special syntactic
and logical structures make it an ideal vehicle for such
a purpose. On a surface analysis, such a notion sounds
remarkably close to Bloomfieldian secondary responses
about the ideal characteristics of the Sequoyan
language, long before computers entered the picture.
(Guzman de Rojas)

(16) See Chomsky, 1975, p. 40.

(17) The principal work encouraging a search for
`universal' aspects of language is Greenberg. Its
findings are suggestive but inconclusive.

(18) This section first appeared in a different
form as a discussion between Sandra Celt and the author
(Celt & Gross).

(19) Most of Marvin Minsky's thoughts on language
follow a strictly Chomskian framework--thus, we can
perhaps refer to the overall outlook of his school as a
Minksian-Chomskian one. For further details see
Sections 19--26 of Minsky.

(20) See Hunn for a considerably expanded
treatment.

(21) A rich literature expanding on this theme can
be found in the bibliography of the book mentioned in
the preceding note.

(22) Glossematics is in the U.S. a relatively
obscure school of linguistics, founded by two Danes,
Louis Hjelmslev and Hans Joergen Uldall, earlier in the
century. Its basic thesis has much in common with
thinking about computers and their possible
architectures. It starts from the premise that any
theory about language must take into account all
possible languages that have ever existed or can exist,
that this is the absolute minimum requirement for
creating a science of linguistics. To objections that
this is unknowable and impossible, its proponents reply
that mathematicians regularly deal with comparable
unknowables and are still able to make meaningful
generalizations about them. From this foundation
emerges the interesting speculation that linguistics as
a whole may be even larger than mathematics as a whole,
and that `Linguistics' may not be that science which
deals with language but that the various so-called
sciences with their imperfect boundaries and
distinctions may in fact be those branches of
linguistics that deal for the time being with various
domains of linguistics. Out of this emerges the
corollary that taxonomy is the primary science, and that
only by naming things correctly can one hope to
understand them more fully. Concomitant with these
notions also arises an idea that ought to have attracted
computer translation researchers, that a glossematic
approach could lay down the down the basis for creating
culture-independent maps of words and realities through
various languages, assigning precise addresses for each
`word' and `meaning,' though it would require a truly
vast system for its completion and even then would
probably only provide lists of possible translations
rather than final translated versions. The major
theoretical text of Glossematics, somewhat difficult
to follow like many linguistic source books, is
Hjelmslev. One excellent brief summary in English is
Whitfield, another available only in Spanish or Swedish
is Malmberg.

(23) Different strands of this argument may be
pursued in Nagel and Newman, Harel, and Goedel.

-----------------------------------------

Partial Bibliography, comprising mainly the
sources mentioned in this extract from the paper.
For the full bibliography consult the published
edition.

Bloomfield, Leonard (1933) Language, New York, (reprinted in
great part in 1984, University of Chicago).

Bloomfield, Leonard (1944) Secondary and Tertiary Responses
to Language. This piece originally appeared in Language
20.45-55, and has been reprinted in Hockett 1970 and
elsewhere. This particular citation appears on page 420 of
the 1970 reprint.

Celt, Sandra & Gross, Alex (1987) The Challenge of
Translating Chinese Medicine, Language Monthly, April. .

Chisholm, William S., Jr. (1981) Elements of English
Linguistics, Longman.

Chomsky, Noam (1957) Syntactic Structures, Mouton, The Hague.

Chomsky, Noam (1965) Aspects of the Theory of Syntax, MIT
Press.

Chomsky, Noam (1975) The Logical Structure of Linguistic
Theory, p. 40, University of Chicago Press.

Fodor, Jerry A & Katz, Jerrold J. (1964) The Structure of
Language, Prentice-Hall, N.Y.

Goedel, Kurt (1931) Ueber formal unentscheidbare Saetze der
Principia Mathematica und verwandte Systeme I, Monatshefte
fuer Mathematik und Physik, vol. 38, pp. 173-198.

Greenberg, Joseph (1963) Universals of Language,
M.I.T.Press.

Grosjean, Francois (1982) Life With Two Languages: An
Introduction to Bilingualism, Harvard University Press.

Guzman de Rojas, Ivan (1985) Logical and Linguistic Problems
of Social Communication with the Aymara People,
International Development Research Center, Ottawa.

Harel, David (1987) Algorithmics: The Spirit of Computing,
Addison-Wesley.

Harris, Zellig (1951) Structural Linguistics, Univ. of
Chicago Press.

Hjelmslev, Louis (1961) Prolegomena to a Theory of
Language, translated by Francis Whitfield, University of
Wisconsin Press, (Danish title: Omkring sprogteoriens
grundlaeggelse, Copenhagen, 1943)

Hockett, Charles F. (1968) The State of the Art, Mouton, The
Hague.

Hockett, Charles F. (1970) A Leonard Bloomfield Anthology,
Bloomington, [(contains Bloomfield 1944)].

Hunn, Eugene S. (1977) Tzeltal Folk Zoology: The
Classification of Discontinuities in Nature, Academic Press,
New York.

Malmberg, Bertil (1967) Los Nuevos Caminos der la
Linguistica, Siglo Veintiuno, Mexico, , pp. 154-74 (in
Swedish: Nya Vagar inom Sprakforskningen, 1959)

Mehta, Ved (1971) John is Easy to Please, Ferrar, Straus &
Giroux, New York, (originally a New Yorker article,
reprinted in abridged form in Fremantle, Anne (1974) A
Primer of Linguistics, St. Martin's Press, New York.

Minsky, Marvin (1986) The Society of Mind, Simon & Schuster,
New York, especially Sections 19--26.

Nagel, Ernest and Newman, James R. (1989) Goedel's Proof,
New York University Press.

Paulos, John A. (1989) Innumeracy, Mathematical Illiteracy
and its Consequences, Hill & Wang, New York.

Robins, R. H. (1967) A Short History of Linguistics, Longmans,
London.

Sapir, Edward (1921) Language: An Introduction to the Study
of Speech, Harcourt and Brace.

Saussure, Fernand der (1913) Cours der Linguistique
Generale, Paris (translated by W. Baskin as Course in
General Linguistics, 1959, New York).

Vinay, J.-P. and Darbelnet, J. (1963) Stylistique Comparee
du Francais et der l'Anglais, Methode der Traduction,
Didier, Paris.

Weaver, Warren (1955) Translation, in Locke, William N. & Booth,
Albert D.: Machine Translation of Languages, pp. 15-23, Wiley, New
York.

Whitfield Francis (1969) Glossematics, Chapter 23 of
Linguistics, edited by Archibald A. Hill, Voice of America
Forum Lectures.

Whorf, Benjamin Lee (1956) Language, Thought and Reality,
(collected papers) M.I.T. Press.

____________________________

NOTE: The essay excerpted above appeared in "Computers in
Translation: A Practical Appraisal," published by Routledge in
1992. This book also contains papers by Dale A. Bostad, John
Chandioux, Annette Grimaila, Alan Melby, John Newton (also the
editor), Jeanette Pugh, Klaus Schubert, Harold L. Somers, Patrica
Thomas, Muriel Vasconcellos, and Yorick Wilks.

It can be purchased from Routledge on-line at:

http://www.routledge.com

or by mail from Routledge, 29 West 35th Street, New York, NY 10001.

----------------------------

People behave as though they were the moulders
and masters of language, while it remains in fact
*THEIR*
mistress.
--Martin Heidegger

Ontology recapitulates philology.
--James Grier Miller, 1968
(cited by W.V.O. Quine)

[Chaumont Devin]

On Wed, 15 Sep 1999 15:23:13 -0400, "Alexander Gross"
<alex...@sprynet.com> wrote:

>Well, now that the "Wednesday Night Massacre" seems to be over, ...

Mr. Gross,

Why don't you just see if you can't open a new NewsGroup called
comp.trx.impossible, and vent your frustrations there? This really isn't
the appropriate venue for all your negativism, and, NO, you don't sound
all that smart.

--CD.

[Sergio Navega]

Alexander Gross wrote in message <7rreum$l7u$1...@nntp2.atl.mindspring.net>...

>Well, now that the "Wednesday Night Massacre" seems to be over, even
>though i'm pretty new here, i suppose i should take some of the
>credit (or the blame) for what took place. So far as i can tell, i
>sort of functioned as a kind of "wing commander" (to continue Arthur
>Murray's brilliant aviation images), swooping down as the first
>plane on the site, bouncing a few warning shots off the fuselage,
>merely to mark the overall position of the "bogey."

>[really big snip]

I'd like to commend Alexander Gross for the posting of such an
interesting article. It does indeed touches on several interesting
points, and appears to be prone to raise some disagreements,
although not much from me.

I largely buy your conclusions about universal grammars and
machine translation problems, although for different reasons.
Your report on Bloomfield was most interesting and unknown to
me and I'm glad to have invested time in reading this huge text
and having read citations of Sapir and Whorf again.

However, I would be a bit reluctant to accept what you said
about the mathematical problems involved with language and MT.
This is part of what I feel is a common misunderstanding of
associating mathematical reasoning (that essentially deals with
abstractions over arbitrary axioms) and natural languages (that
essentially deals with abstractions drawn in tandem with
natural world or perceived reality).

But I already said that I take your side regarding the difficulties
of MT in computers. My reasons, as I said, are different, and they
are related with the fact that to effectively translate something
one has to *understand* what it means. I'll try to escape from
the trap of trying to define what I mean by "understanding".
Suffice it to say that it is related to what we appear to do when
we transform syntax to our most inner concepts, whatever this
really means.

This process, as much of the cognitive and neuroscientific
literature is able to grant, occurs mostly unconsciously, which
is our greatest problem: it is not easy to comprehend what is
going on, our introspection is unable to follow the "steps"
being performed. This is what fools most of those who use
"inner" methods of investigation: they see only tips of
icebergs and take that this is all there is to it.

The problem of MT is, in my opinion, closely related to the
fundamental goals pursued by Artificial Intelligence, that
of making computers capable of intellectual activities. What is
constantly emerging from recent research is that reasoning
and language and perception and cognition are activities
that demand a body to be accomplished. More than that: they
demand interaction of this body with its environment, and the
learning from such interactions. Much of this interaction is
reflected directly in the organism's cognition and becomes
the foundation upon which language emerges.

Language does not make sense without being related to the
world that surrounds the organism.

So it appears to be futile to think of computerized MT without
computers (or should I say robots?) that can experience their
environments. Even then, there will always be areas where
the translations will fail, notably because of the huge chasm
that separates machines from humans in emotional terms. Human
level translation is only possible (with the same accuracy)
by using humans to do the job.

Sergio Navega.

[Chaumont Devin]

Sergio Navega! My dear companion and comrade at arms, it thrills me to
hear that you are yet among the living!

>I'll try to escape from the trap of trying to define what I mean by
>"understanding".

I object to this escapism. It strikes me as unscholarly. A man who knows
what is talking about should know what he is saying.

>Suffice it to say that it is related to what we appear to do when
>we transform syntax to our most inner concepts, whatever this
>really means.

>This process, as much of the cognitive and neuroscientific
>literature is able to grant, occurs mostly unconsciously, which
>is our greatest problem: it is not easy to comprehend what is
>going on, our introspection is unable to follow the "steps"
>being performed. This is what fools most of those who use
>"inner" methods of investigation: they see only tips of
>icebergs and take that this is all there is to it.

I am sorry you take me for a fool, and I object to this, too.

So now, because of these objections, I am going to close this message and
go to my roost for another out-of-body experience with the fairies of
yesteryear whose out-of-body phantoms keep dancing ever before my eyes!
Oh that through language I might somehow materialize them and then
dematerialize them again. That I might SPEAK them into BEING. This world
might be such a Paradise instead of the Hell that it is! Perhaps someday
in Cyberspace?

--CD.

[Alexander Gross]

Sergio, thanks many times over for having taken the trouble to read
that VERY LONG file and for reacting so positively about it (and so
incredibly soon after i posted it!).

i hope others here will be inspired by your reactions and also
manage to get through it. i tried to write everything in it as
interestingly as possible, especially the first part but even the
"academic" part of it just so people would be able to follow it.
i've learned how to write in a lot of different styles just so i can
choose the right one for a given subject.

let me try the math part again, since i didn't succeed the first
time--and please note, i am not arguing with you in the slightest, i
am merely trying to clarify my own thought for something else i will
be writing, and it sounds like you might be able to help me--let's
see if it makes more sense this way:

is it true or not true that anything that can be said in
mathematical notation, however advanced it may be, must at some
point or another ALSO be sayable in ordinary natlang words? i mean,
not only can the numerical expression

1,579,843,296,843

also be said as:

one trillion five hundred and seventy-nine billion, eight
hundred and forty-three million, two hundred and ninety-six
thousand, eight hundred and forty-three,

but is it or is it not also true that even advanced expressions and
formulas (which i can't draw here because of graphics limitations)

must _at some point in their being taught by teachers or learned by
students_ also be turned either partially or entirely into something
like natlang words or supplemented by explanations in natlang words?

isn't this how the entire pyramid of lower math leading to higher
math leading to sublime math must be built up?

how else would it be possible to begin teaching a concept without
explaining it in natlang words? and isn't it true that we begin
learning it at its lower levels of advanced algebra and trig only
because someone takes the trouble of explaining it to us either
orally or in book form, but in either case by using natlang words in
a specialized way?

granted, any math concept will be a hell of a lot longer expressed
in natlang words, but isn't that true of all valid scientific jargons?

lots of people make the mistake of supposing that when they have
learned something in words, they then know it in another "purer,"
more mental medium and can forget about the words. that's why, as
i was saying earlier here, people suppose they have learned a
language, whether a natlang or a complang, painlessly and easily,
when this was probably not the case at all.

why would this not also be true of mathematicians? Granted, some of
them are able to "sight-read" articles and entire books composed of
almost nothing but formulas, and they certainly believe with some
justice that they are reading math straight with no recourse to
words, just as musicians can read musical scores in their heads and
hear all the orchestral parts internally. But could either do so if
they had not previously built up this ability step by step over a
long period of time, mathematicians by slowly converting words into
concepts and musicians by slowly learning how to pick out notes
until they could hear entire chords and symphonic passages inside
their heads? (And come to think of it, could some verbal element
also have crept into the learning process of musicians?)

This is part of what i mean by math being a subset of language. i
believe there are several other proofs that this may be true:

1) Backward conversion: if a more advanced mathematician must
explain a concept to a less advanced one who cannot grasp it from
written formulas, will he not be forced--even if does not want to do
so--to reconvert his understanding into at least some words of
explanation if he wants to get his point across?

1a) in this sense, does not language function as some kind of
freezing or dehydration process: one has to "freeze" or "dehydrate"
meaningful concepts into "brainstuff" to get them in, but one also
has to unfreeze or rehydrate them back into words to get them out
again...

2) Other forms of "clean jargon" do much the same thing, only on a
less complex level. here i use "clean jargon" to denote those forms
of specialized medical, scientific, or technical vocabularies that at
least give a trade-off of precision and concision in return for
saddling us with a term understood only by specialists:

"systemic lupus erythematosus" is certainly a mouthful, but
it describes a precise condition and even presents a bit of
differential diagnosis--and as long-winded as it is, it is
far shorter than having to start from scratch and explain
everything about the disease. but for a doctor to tell you
what it means, he will once again have to "unfreeze" or
"rehydrate" it into words most people can understand. if he
has to explain certain terms to laymen frequently, he will
have committed this "unfreezing" process to memory, but if
he is suddenly forced to explain a rare term, you will see
him going through some "unfreezing pains."

in this sense, could mathematics merely be the "ultimate
jargon?" the briefest way of freezing truly complex
concepts, but so complex and refined a form of "archiving"
that some mathematicians can barely explain themselves in
natlang at all.

(i am of course using the term "clean jargon" to contrast it with
"dirty jargon," which is neither more concise nor more precise than
the concepts it purports to represent but little more than a mess.
The kind of jargon used by Chomskians, tagmemics people,
meme-mongers and derrie`rist Derridans, because they mistakenly
suppose it makes them look "scientific.")

3) In a more purely logical sense, is it or it is not true that
natlang is PRIOR to math, both logically and historically?

In other words, could it be that without language there can be no
math at all. i wonder if there can even be finger counting without
words for the first ten numbers. Or did those numbers in fact arise
from people pointing to their fingers and saying "Duh, Duh, Duh..."
for each number until they realized they had better find a better
way...?

And historically there is not the slightest doubt that language has
been PRIOR to math simply by preceding it in time: many cultures
never seem to have invented math at all beyond the most rudimentary
scale.

4) For many decades we were treated to the claims that "math is
eternal," "math is universal," etc. But quite a few new books on
math have now reneged on this and actually describe our math as
planet- and even as culture-bound. In other words, Whorf-Safir
applies to math as well as to language. Do we need any further
proof that Whorf, Sapir, & Bloomfield were on the RIGHT TRACK all
along, that TGG is a crock, and that the last forty years of
linguistics have been largely wasted?

5) Finally (for now at least) doesn't the notion of language being
prior to math perfectly explain Goedel's Entscheidungsproblem?
If human language is neither complete not entirely consistant, (as i
think most of us would have to agree is the case), why the hell
would math be?

do tell me if i have now succeeded in the eternal linguistic battle a bit
more successfully--does what i have just said about math now make better
sense to you?

i'm also delighted by your insistence on the body as a major
component of speaking & understanding. if you take a look at my
semi-humorous animated text show "Spray It Again, Sam," (which is
one part of my shareware program "Truth About Translation,"
downloadable immediately by clicking on:

ftp://oak.oakland.edu/pub/ad-edu/trutra.zip

you'll see that IMO the biology and evolution of the human body
from lower animals is closely connected with everything having to
do with language. and as someone who has worked in the theatre,
radio, etc., like many actors and public speakers i have personally
experienced the sheer centrality of the body to language.

What a pity that so many others keep presenting language as such a
purely cerebral, intellectual process...!

i'll be posting a few other pieces as we go on, lots more about MT &
language, but also a few items most definitely touching on AI. i
had a lot of fun debating John McCarthy, the grand-daddy of AI, on
another newsgroup, and maybe i'll post some of that too. and if you
haven't already seen it, by all means click on the following too:

http://www.accurapid.com/journal/09review.htm

judging from your choice of words, about the only fault i might find
in your observations is that you seem to have been tarred with a
touch of the cogsci brush. but you may soon outgrow that... :-)

since you're also interested in the "nature of understanding," i'll
leave you with a few thoughts by Wittgenstein on this elusive cutie:

-------------------------------------------

(from a series of questions about MT submitted to MT Summit III in
1991, also available in full in the DL of "Truth About Translation")

Does the following reflection by Wittgenstein apply to MT:

"A sentence is given me in code together with the key. Then of
course in one way everything required for understanding the
sentence has been given me. And yet I should answer the question
`Do you understand this sentence?': No, not yet; I must first
decode it. And only when e.g. I had translated it into English
or German would I say `Now I understand it.'

"If now we raise the question `At what moment of translating
do I understand the sentence? we shall get a glimpse into the
nature of what is called `understanding.'"

To take Wittgenstein's example one step further, if MT is used,
at what moment of translation does what person or entity
understand the sentence? When does the system understand it? How
about the hasty post-editor? And what about the translation's
target audience, the client? Can we be sure that understanding has
taken place at any of these moments? And if understanding has
not taken place, has translation?

be happy!

alex

I am not like that court lady at Versailles who said: "What
a pity that accident with the Tower of Babel should have got
languages all confused--otherwise everyone would have always
spoken French."

--Voltaire, 1767

Every language is an experiment.
(Jede Sprache ist ein Versuch.)

--Wilhelm von Humboldt, ca. 1820

[Alexander Gross]

>I may be a good reader but you're a hell of a good typist!
>
And so are YOU! Thanks many times over again!

At the moment though, i'm not a very good reader, so i'll take some
time reading, mulling over, and trying to absorb what you've written. i'm
glad i asked you those questions, and i suspect i will profit from your
answers!

Thanks again!

best!

alex

[Sergio Navega]

Chaumont Devin wrote in message <7rrrl4$5mi$2...@mochi.lava.net>...

>Sergio Navega! My dear companion and comrade at arms, it thrills me to
>hear that you are yet among the living!

I'm glad you're glad I'm here!

>
>>I'll try to escape from the trap of trying to define what I mean by
>>"understanding".
>
>I object to this escapism. It strikes me as unscholarly. A man who knows
>what is talking about should know what he is saying.
>

Or a man who knows that he doesn't know enough will for ever be
open to the search of the truth.

>>Suffice it to say that it is related to what we appear to do when
>>we transform syntax to our most inner concepts, whatever this
>>really means.
>
>>This process, as much of the cognitive and neuroscientific
>>literature is able to grant, occurs mostly unconsciously, which
>>is our greatest problem: it is not easy to comprehend what is
>>going on, our introspection is unable to follow the "steps"
>>being performed. This is what fools most of those who use
>>"inner" methods of investigation: they see only tips of
>>icebergs and take that this is all there is to it.
>
>I am sorry you take me for a fool, and I object to this, too.
>

Not so fast, Chaumont! If you're in that category I can assure
you that, occasionally, I'm in there too (I guess sometimes
we all do that). The question I raise is that inner methods
of investigation must be complemented with real world checking
techniques.

>So now, because of these objections, I am going to close this message and
>go to my roost for another out-of-body experience with the fairies of
>yesteryear whose out-of-body phantoms keep dancing ever before my eyes!
>Oh that through language I might somehow materialize them and then
>dematerialize them again. That I might SPEAK them into BEING. This world
>might be such a Paradise instead of the Hell that it is! Perhaps someday
>in Cyberspace?
>

I share with you the wonders of that kind of experience and I often
put myself into those states (spontaneously, of course, no need for
booze). There's nothing that a fully prepared scholar could do
without going into those states from time to time. You know what I'm
talking about, it's about creativity. We certainly wouldn't be here
(I mean, talking in the cyberspace) if it wasn't for the creativity
of the human spirit. But... (there's always a but...).

After creating something we should put it to the scrutiny of the
critical thinking. Without doing that, we would live in a magical,
inexistent world that only pays off when transformed into movies
and fiction books. I've learned how to do that, to start creating
without boundaries and then following with a strict, rigid and
critical process of checking.

Best Regards,
Sergio Navega.

[Sergio Navega]

Alexander Gross wrote in message <7rsfp3$3rb$1...@nntp3.atl.mindspring.net>...

>Sergio, thanks many times over for having taken the trouble to read
>that VERY LONG file and for reacting so positively about it (and so
>incredibly soon after i posted it!).

I may be a good reader but you're a hell of a good typist!

>

>i hope others here will be inspired by your reactions and also
>manage to get through it. i tried to write everything in it as
>interestingly as possible, especially the first part but even the
>"academic" part of it just so people would be able to follow it.
>i've learned how to write in a lot of different styles just so i can
>choose the right one for a given subject.
>

I love academic prose that manages to communicate interesting
stuff. On the other hand, academese without that interestingness
is really a pain in the bottom.

>let me try the math part again, since i didn't succeed the first
>time--and please note, i am not arguing with you in the slightest, i
>am merely trying to clarify my own thought for something else i will
>be writing, and it sounds like you might be able to help me--let's
>see if it makes more sense this way:
>
>is it true or not true that anything that can be said in
>mathematical notation, however advanced it may be, must at some
>point or another ALSO be sayable in ordinary natlang words? i mean,
>not only can the numerical expression
>
> 1,579,843,296,843
>
>also be said as:
>
> one trillion five hundred and seventy-nine billion, eight
> hundred and forty-three million, two hundred and ninety-six
> thousand, eight hundred and forty-three,
>
>but is it or is it not also true that even advanced expressions and
>formulas (which i can't draw here because of graphics limitations)
>
>must _at some point in their being taught by teachers or learned by
>students_ also be turned either partially or entirely into something
>like natlang words or supplemented by explanations in natlang words?

In my opinion, yes and no. Yes, you could "read" an equation, mapping
each mathematical symbol to a natlang word or expression. No, it is
not enough to utter the "language equivalent" of an equation in
order to grasp its deep meaning. This deep semantic component is the
one that, in the case of the "reading" of one equation, stops at the
abstract notions that you have from maths. One can't "understand"
it just at that level, it is even deeper. When you're talking with
your friend about your birthday party, you're also using semantic
components that map, ultimately, to the sensory experiences
that you had. Your friend will "understand" what you say because
he can map your words to *his own sensory equivalents*. Someway
math equations must map to similar things.

>
>isn't this how the entire pyramid of lower math leading to higher
>math leading to sublime math must be built up?
>
>how else would it be possible to begin teaching a concept without
>explaining it in natlang words? and isn't it true that we begin
>learning it at its lower levels of advanced algebra and trig only
>because someone takes the trouble of explaining it to us either
>orally or in book form, but in either case by using natlang words in
>a specialized way?
>
>granted, any math concept will be a hell of a lot longer expressed
>in natlang words, but isn't that true of all valid scientific jargons?
>

These are different things. Mathematical "language" is
self-sufficient, when you've got a set of good axioms. Any computer
can do mathematical deductions (as many programs do, proving theorems).
The big question appears when *we* must understand what is behind
a mathematical concept. It is not enough to map them to natural
language. We can't simply "believe" in the axioms and start from
that point on. Our mind does not work in a "logical", deductive way.
It is inductive, it is able to, from a set of generalizations,
induce laws and rules, many times applicable to different domains.
This is the mapping (analogy or metaphor, if you will) that allows
us to derive basic math concepts from sensory experiences.

To teach arithmetic to a child we must start with simple, sensory
concepts like identifying blocks and adding one block to a set
of blocks to suggest what is addition. Then we proceed with
bean counting and the way to understand division is to share
the beans among his/her friends. It's essentially sensory.
When Roger Penrose is doing some complex mathematical thinking
I bet he is (unconsciously) "distributing beans among its friends".

>lots of people make the mistake of supposing that when they have
>learned something in words, they then know it in another "purer,"
>more mental medium and can forget about the words. that's why, as
>i was saying earlier here, people suppose they have learned a
>language, whether a natlang or a complang, painlessly and easily,
>when this was probably not the case at all.
>
>why would this not also be true of mathematicians? Granted, some of
>them are able to "sight-read" articles and entire books composed of
>almost nothing but formulas, and they certainly believe with some
>justice that they are reading math straight with no recourse to
>words, just as musicians can read musical scores in their heads and
>hear all the orchestral parts internally. But could either do so if
>they had not previously built up this ability step by step over a
>long period of time, mathematicians by slowly converting words into
>concepts and musicians by slowly learning how to pick out notes
>until they could hear entire chords and symphonic passages inside
>their heads? (And come to think of it, could some verbal element
>also have crept into the learning process of musicians?)
>

In my opinion, when a mathematician reads those "garbled" papers
full of equations what they are doing is incredibly similar to
the reading of an ordinary text: they are matching patterns with
his/her own set of learned patterns. In both cases (math and natlang)
what is at stake is the identification (by perception) of patterns
that ultimately stand on a huge number of other patterns and so
on until the final level of simple sensory notions.

>This is part of what i mean by math being a subset of language. i
>believe there are several other proofs that this may be true:
>
>1) Backward conversion: if a more advanced mathematician must
>explain a concept to a less advanced one who cannot grasp it from
>written formulas, will he not be forced--even if does not want to do
>so--to reconvert his understanding into at least some words of
>explanation if he wants to get his point across?
>

I'm not sure I agree with you here. Take, for instance, a
mathematician explaining to a lay person what an integral is. He/she
will draw a graph (which is a visual, symbolic construct that can
be mapped to sensory experiences of visually raising objects) and
then say that integral is like the area of the figure below the
graph.

Now if the same mathematician explains triple integrals to a
student who is already knowledgeable in simple integrals, he will
use words yes, but only to map what he is trying to explain to the
*previous* notion, which is that of a simple integral. It is
usually not necessary to go all the way down to areas, the student
can understand this because he already "knows" what a
simple integral is. I believe that the most sophisticated
mathematical reasoning can be traced back to only two kinds of
things: a) patterns of abstraction without any sensory correspondent,
and that stand only over other patterns; b) snippets of patterns
that are directly mappeable to sensory concepts. Obviously, most of
what constitutes the mind of a mathematician are a) structures.

In that regard, mathematical reasoning can really have a "life
of its own" in the mind of a mathematician. Most of the time
they may reason only handling a) type patterns. Natural languages
only participate in the coordination of the movement of these
patterns, in order to assemble new ones.

What is prior to math (and also language) is the understanding of
what to do with the bunch of patterns involved. Think, for instance,
about the formula for the centripetal force: F = (m v^2) / R

To "understand" what this means, it is not enough to map it to
natural language, it is necessary to use language in the process
of mapping this formula to "inner" concepts of linear variation,
variation with the square, inverse variation, etc. All this can
only be understood by a person who had previous notions of what
variation is, sensory notions that help in differing between a
linear and a quadratic growth. One may have the formula in its
head, but one can only say it "understands" it once he maps this
formula to intuitive notions.

This does not mean that all people do that kind of reasoning:
most of the time, novices don't do that, and this is why they
are said to have only rote material, failing to "understand"
what is behind it. They will hardly be able to create new
things, because in order to do that (in a efficacious manner)
you should "think" with those inner concepts, not with the
surface appearance of them.

That's one of the greatest problems in education we have today:
some educators think that formal training is just the rote
learning of formulas in such a way as to pass the examinations.
But when the student faces a real world problem, he/she must
think intuitively about it. To discover new things you must
have more than the syntactical notions about math, you must
"feel" how the things work in order to propose a reasonable
solution.

This is the part of the "thinking" process that an experienced
mathematician or physician or chemist or doctor does when he/she
is doing its job. For instance, ask to an experimented physician
the rationale behind his diagnostic for a mysterious disease. He
may have problems turning his suspicions into words, but he
usually is right in his intuitions.

>In other words, could it be that without language there can be no
>math at all. i wonder if there can even be finger counting without
>words for the first ten numbers. Or did those numbers in fact arise
>from people pointing to their fingers and saying "Duh, Duh, Duh..."
>for each number until they realized they had better find a better
>way...?
>
>And historically there is not the slightest doubt that language has
>been PRIOR to math simply by preceding it in time: many cultures
>never seem to have invented math at all beyond the most rudimentary
>scale.
>
>4) For many decades we were treated to the claims that "math is
>eternal," "math is universal," etc. But quite a few new books on
>math have now reneged on this and actually describe our math as
>planet- and even as culture-bound. In other words, Whorf-Safir
>applies to math as well as to language. Do we need any further
>proof that Whorf, Sapir, & Bloomfield were on the RIGHT TRACK all
>along, that TGG is a crock, and that the last forty years of
>linguistics have been largely wasted?

I may accept that some things in math are culturally-bound, but
nothing we have created so far is closer to universal (meaning
"alien") acceptance than math. The question is that this is
not enough to map our real-world concepts to other civilizations.

Take the SETI project for instance. If they manage to receive
any signs of alien intelligence (I doubt they will), what this
signal will inform us? It will inform only that an intelligent
life produced it. We'll not know much about these guys by the
mathematical signal that we capture. Probably the differences in
sensory apparatus will be so astonishing that we'll not be able
to "grasp" what they are thinking. We already have that kind of
problem when trying to interpret "utterances" of dolphins and
whales.

>
>5) Finally (for now at least) doesn't the notion of language being
>prior to math perfectly explain Goedel's Entscheidungsproblem?
>If human language is neither complete not entirely consistant, (as i
>think most of us would have to agree is the case), why the hell
>would math be?
>
>do tell me if i have now succeeded in the eternal linguistic battle a bit
>more successfully--does what i have just said about math now make better
>sense to you?
>

As I said on my earlier post, I accept your conclusions but for
different reasons. I don't think that language is very different from
math, I think that both are surface representations of deep things
that happen inside our mind. Both can be, often, self-sufficient
(just try to see the typical talk of philosophers and mathematicians).
But both demand solid sensory grounds in order to be fully meaningful.

>i'm also delighted by your insistence on the body as a major
>component of speaking & understanding. if you take a look at my
>semi-humorous animated text show "Spray It Again, Sam," (which is
>one part of my shareware program "Truth About Translation,"
>downloadable immediately by clicking on:
>
> ftp://oak.oakland.edu/pub/ad-edu/trutra.zip
>
>you'll see that IMO the biology and evolution of the human body
>from lower animals is closely connected with everything having to
>do with language. and as someone who has worked in the theatre,
>radio, etc., like many actors and public speakers i have personally
>experienced the sheer centrality of the body to language.
>

I'll take a look at that.

>What a pity that so many others keep presenting language as such a
>purely cerebral, intellectual process...!
>
>i'll be posting a few other pieces as we go on, lots more about MT &
>language, but also a few items most definitely touching on AI. i
>had a lot of fun debating John McCarthy, the grand-daddy of AI, on
>another newsgroup, and maybe i'll post some of that too. and if you
>haven't already seen it, by all means click on the following too:
>
> http://www.accurapid.com/journal/09review.htm
>
>judging from your choice of words, about the only fault i might find
>in your observations is that you seem to have been tarred with a
>touch of the cogsci brush. but you may soon outgrow that... :-)
>

How observant you are! I am indeed touched (or should I say scratched)
by the cogsci brush, but I also happen to be seriously involved
with neuroscience, where much of the answers are laying, awaiting
to be discovered. Much of what puzzled Wittgenstein will be settled
by the advances these two disciplines, sooner or later, will make.

Regards,
Sergio Navega.

[patrik bagge]

Hi Sergio

I asked you before (maybe you missed that one?)
and i am still curious,
you mentioned that you was going to publish
a 'working paper' on your AI-work
at your website. I have studied your page
and could not find such a paper,
when do you expect the first version to be made public?

Regards
Patrik Bagge

[Sergio Navega]

patrik bagge wrote in message <7%tE3.1888$uV5....@news1.online.no>...

Hi, Patrik,

I've been a bit busy with my schedule of seminars, so the
completion of my research statement is delayed. Before the
end of this year I should put some more material I'm working
now and by the beginning of next year I should have some
initial "down to earth" material also put there. Keep checking.

Regards,
Sergio Navega.

[Alexander Gross]

Glad to see that some of you are beginning to comment on my "Limitations"
piece. i think there's a lot there to chew on, plus which there's a whole
lot more i can post later on about specific aspects of MT, AI, human
translation, problems of Chinese, and just language in general.

when i first went on the Internet back when you needed UNIX to do so, i used
to get all roiled up inside over newsgroup messages & felt i had to shoot
back the first thing that came into my mind about every single point anyone
made, particularly if it touched on something _I_ had said.

i don't feel that way any more, so i think i'll just let things roll for a
while. anyway, i've already gone through this process on
alt.fan.noam-chomsky, where people were about four times as ill-mannered as
here.

please remember, i'm really listening and will one way or another take
everything into account you wish to say.

really glad to encounter Sergio here--fear he may be losing his Chomskian
religion, but i think he's strong enough to stand up on his own two feet
without it, unlike one or two others i won't name. Sergio, i'll also get
back to what you have to say both about math and about putting emotions in a
system, but only after a while. i think i'd rather
post a few possibly useful things over a period of time than get involved in
too much detailed in-fighting.

though i promise i won't stand totally above the fray either and look
forward to an interesting match of conceptual capoeira.

also posted the piece on sci.lang.translation, where i just got back the
following message:

From Peter Wells:

> On Wed, 15 Sep 1999 17:19:47 -0400, in sci.lang.translation you wrote:

>> For James Lee, & no doubt others as well, from comp.ai.nat-lang. It
>> looks like we've ended up on the same side of this after all. ;-)
>
>Thank you for posting this highly-informative and interesting piece. As a
translator with
> little or no knowledge of linguistics, I am unable to issue a reasoned
critique, but you
> laid out the issues very clearly in many areas. I particularly appreciated
the points
> about secondary responses to language, as I live in France where "pure"
language is an
> obsession, and asymetrical bilingualisms, something that is difficult to
explain to
> customers.

And on the matter of understanding in order to translate, here's what
someone else
had to say almost two thousand years ago:

At the same time, what they may miss in reading, they
cannot
avoid in translating. From this process, intelligence and
judgement are acquired.

--Pliny the Younger, 105 A.D.

ciao for now!

alex

[John Turnbull]

In article <37e1...@news3.prserv.net>, Sergio Navega <sna...@ibm.net> wrote:

>I'd like to commend Alexander Gross for the posting of such an
>interesting article. It does indeed touches on several interesting
>points, and appears to be prone to raise some disagreements,
>although not much from me.

I missed Alexander's post, however I did read his "Truth about
Translation" file.

>But I already said that I take your side regarding the difficulties
>of MT in computers. My reasons, as I said, are different, and they
>are related with the fact that to effectively translate something

>one has to *understand* what it means. I'll try to escape from

>the trap of trying to define what I mean by "understanding".

>Suffice it to say that it is related to what we appear to do when
>we transform syntax to our most inner concepts, whatever this
>really means.

Certainly "understanding" is crucial, and is the first stage of
translation. Expressing that understanding being the second stage.
That understanding has to be built on the foundation of what we already
understand. Much is written that even though one may "understand" the
words individually, one does not know all the subtelties, and can
complete miss the actual knowledge therein for scientific works. For
literary works one can read the same work very differently than another
person, or maybe than the author intended, because of different
experiences.

>The problem of MT is, in my opinion, closely related to the
>fundamental goals pursued by Artificial Intelligence, that
>of making computers capable of intellectual activities. What is
>constantly emerging from recent research is that reasoning
>and language and perception and cognition are activities
>that demand a body to be accomplished. More than that: they
>demand interaction of this body with its environment, and the
>learning from such interactions. Much of this interaction is
>reflected directly in the organism's cognition and becomes
>the foundation upon which language emerges.
>
>Language does not make sense without being related to the
>world that surrounds the organism.

That is certainly true. The purpose of language is to communicate ideas
from one person to another. Even within one language one might express
the same idea differently to different audiences if you know that they
have differing assumptions and ways of thinking. As a quick example, if
you were describing a traffic accident you would probably describe it
differently to the police than to your friends or family, since your
family already knows the color, make model year of your car, and the
police aren't too interested in your opinions of the bozoness of the
other driver.

>So it appears to be futile to think of computerized MT without
>computers (or should I say robots?) that can experience their
>environments. Even then, there will always be areas where
>the translations will fail, notably because of the huge chasm
>that separates machines from humans in emotional terms. Human
>level translation is only possible (with the same accuracy)
>by using humans to do the job.

I don't think that emotions are impossible for machines, or maybe when
machines have emotions they are no longer machines? While I think the
lack of shared experience may be an impediment, I don't think it is
insurmountable. Human translators have similar problems, in that
learning and internalizing enough about two cultures takes enough time
that it's difficult to fully absorb anyone. People always make all
sorts of interesting guesses as to where I'm from, though nowhere is the
guess that I'm a local.

I think my main argument for MT would be that it seems the brain is
basically composed of lots of little pieces, that are individually
straight forward, but when connected together they think. If this is
true I see no reason we will not eventually build a better version of
the brain as we can exceed the limits of our anatomy.

John Turnbull

[patrik bagge]

>I've been a bit busy with my schedule of seminars, so the
>completion of my research statement is delayed. Before the
>end of this year I should put some more material I'm working
>now and by the beginning of next year I should have some
>initial "down to earth" material also put there. Keep checking.

I will !
one humble request though, incorporate some examples
and 'not-so-general-terms' in your paper.
This will enable us 'amatours' to 'understand'
and 'return fire' :-)

Regards
Patrik Bagge

[Chaumont Devin]

Computer illiteracy and stupidity do not belong on this NewsGroup, so why
don't you just pack up your little limitations and take them elsewhere?

[Sergio Navega]

patrik bagge wrote in message ...

Well, as we are on the subject of MT in this thread, it is
interesting to note that often hard to understand terms or
jargon are usually employed to categorize ideas and concepts
that occur very often in the discipline one is talking. This
means that there is something "relevant" behind those specialized
words (and it is not the word itself). I'd suggest for you
to ask shamelessly what the words mean, people on newsgroups
usually answer that requests with pleasure. The only problem
is that those requests often trigger an even greater flood of
jargon and garbled sentences.

Regards,
Sergio Navega.

[Sergio Navega]

John Turnbull wrote in message ...

I don't have much to disagree, except perhaps for the position of
understanding. I wouldn't say it is the first stage, but the central
stage. For instance, suppose that we're reading a phrase about
someone who was walking and stepped on a stone loosing his balance.
The semantic content of this phrase can be understood by an american
and by a chinese, even if their surface representations differ
substantially. It is this "core" level of processing, that bears
strict relation to the organism's sensory experiences, that I
ascribe to the understanding and it is the second step, of the
three I envision (the first would be knowledge of surface aspects of
the specific language, third would be generation of syntax in the
target language).

>>The problem of MT is, in my opinion, closely related to the
>>fundamental goals pursued by Artificial Intelligence, that
>>of making computers capable of intellectual activities. What is
>>constantly emerging from recent research is that reasoning
>>and language and perception and cognition are activities
>>that demand a body to be accomplished. More than that: they
>>demand interaction of this body with its environment, and the
>>learning from such interactions. Much of this interaction is
>>reflected directly in the organism's cognition and becomes
>>the foundation upon which language emerges.
>>
>>Language does not make sense without being related to the
>>world that surrounds the organism.
>
>That is certainly true. The purpose of language is to communicate ideas
>from one person to another. Even within one language one might express
>the same idea differently to different audiences if you know that they
>have differing assumptions and ways of thinking. As a quick example, if
>you were describing a traffic accident you would probably describe it
>differently to the police than to your friends or family, since your
>family already knows the color, make model year of your car, and the
>police aren't too interested in your opinions of the bozoness of the
>other driver.

This is an important point, I agree in full. This is another example
of how hard it is to translate something, with the content and form
of that translation varying so much in relation to the "receiving"
audience. How's a machine supposed to do in this case without those
notions?

>
>>So it appears to be futile to think of computerized MT without
>>computers (or should I say robots?) that can experience their
>>environments. Even then, there will always be areas where
>>the translations will fail, notably because of the huge chasm
>>that separates machines from humans in emotional terms. Human
>>level translation is only possible (with the same accuracy)
>>by using humans to do the job.
>
>I don't think that emotions are impossible for machines, or maybe when
>machines have emotions they are no longer machines? While I think the
>lack of shared experience may be an impediment, I don't think it is
>insurmountable. Human translators have similar problems, in that
>learning and internalizing enough about two cultures takes enough time
>that it's difficult to fully absorb anyone. People always make all
>sorts of interesting guesses as to where I'm from, though nowhere is the
>guess that I'm a local.
>

Ah! This is a hell of a polemic question, I'm not far from your viewpoint.
I also think that we can put some emotions in machines, but this will
be more "artificial" than "artificial intelligence". This will drift
a bit into philosophical aspects, I hope readers of comp.ai.nat-lang
will forgive me for this.

The way I understand "intelligence", taken in isolation, allows me to
think about mechanisms who are able to perform almost to the point
of dropping the word "artificial". I put intelligence to be the kind
of ability that, once properly implemented, will be as natural as
our own. We could say that the "real" intelligence will have to
be "natural" anyway.

However, when it comes to emotions, and given the kind of machines
we're constructing, I'd say that we will stop on the "artificial".
Emotions are things that appear to be closely related to evolutionary
aspects. It is essentially intertwined with aspects of our biological
nature, to the point of being difficult (if not impossible) to
separate one from another. A machine will, under this hypothesis,
never be able to grasp what we "feel", because what we feel is
closely related to our "elemental biological hardware".

Aspects of emotions can be simulated in machines with reasonable
results, there are some laboratories around the world that do that
for quite some time (Aaron Sloman's is one example). This will be
enough to give us an opportunity to interact with the machines and
empathize with them (and vice versa), enough for us to develop a
personal "relationship". Empathizing (on the side of the machine)
will allow it to, often, translate the right feeling from one
language to another. But don't ask a "personal opinion" to a
machine, it will probably just utter a composite idea of what it
had "listened" before. But we can imagine the development of a
specific kind of "emotion" in machines related to their bodily
conditions. That would be the weirdest kind of thing we could
ever think of, what a silicon-based machine could "feel".

>I think my main argument for MT would be that it seems the brain is
>basically composed of lots of little pieces, that are individually
>straight forward, but when connected together they think. If this is
>true I see no reason we will not eventually build a better version of
>the brain as we can exceed the limits of our anatomy.
>

I surely hope so. It would be nice to have such kind of thing within
my lifetime. There are so many interesting things to do and so little
time...

Regards,
Sergio Navega.

[patrik bagge]

>Well, as we are on the subject of MT in this thread, it is
>interesting to note that often hard to understand terms or
>jargon are usually employed to categorize ideas and concepts
>that occur very often in the discipline one is talking. This
>means that there is something "relevant" behind those specialized
>words (and it is not the word itself).

absolutely, might be due to the fact that complicated
ideas/notions can't be explained with the current vocabulary,
i notice that new words are often invented (do it myself too)
Anyway, as long as people understand there is no problemo.

language seems to evolve all the time, new words become
accepted and other disappear. This should be remembered
when creating MT solutions.
hmm, come to think of it, in (the beginning of) love relations small,unique
'love-words'
are often invented between two people. Even though one has
never heard these words it's possible to 'understand' them,
due to the fact that they often are spoken with affection & cilly faces ;-o.
The way a word is spoken may determine the meaning of it.
the swedish equivalent of 'bitch' can be used in many,many ways
by varying tone.

> I'd suggest for you
>to ask shamelessly what the words mean, people on newsgroups
>usually answer that requests with pleasure.

sure, i have done it a couple of times, it's no shame in that.
the 'shame' in my opinion occurs when people hide behind
jargon due to the fact that they are not so sure what they are talking
about.
Why not say: 'i do not know'. I notice a surprising lack of those words
in this forum. It certainly is no shame in that either.
One could always add 'but i will tomorrow'

The only problem is that those requests often trigger an even greater flood
of
>jargon and garbled sentences.

certainly, here ís an example:
´a general interpolation of knowledge based communication with humanoid
symbols could possibly be transformed into a lower level of abstraction'
...
I am currently probing into Chaumont Devins work and i will get back with
comments/questions when i am done. I think i will order a copy and
evaluate some scandinavian languages on the software.

Regards
Patrik Bagge

[Chaumont Devin]

Dear Sergeo,

I hate to have to cut you down like this after being friends for so many
years, but now I find myself constrained to betray you for science, in
other words to offer you up upon the chopping block of dissection and
analysis, the alter of progress, etc.

You say:

>What is
>constantly emerging from recent research is that reasoning
>and language and perception and cognition are activities
>that demand a body to be accomplished. More than that: they
>demand interaction of this body with its environment, and the
>learning from such interactions.

People have said things like this before, as in Jesus, after having
actually created the world through linguistic means, still having to
descend into it to really understand what he had created through physical
interaction. So I be wondering whether you may not have gotten some of
this from the confessional, and if you have, I think you should confess.

>For instance, suppose that we're reading a phrase about
>someone who was walking and stepped on a stone loosing his balance.
>The semantic content of this phrase can be understood by an american
>and by a chinese, even if their surface representations differ
>substantially. It is this "core" level of processing, that bears

>strict relation to the organism's sensory experiences, ...

Complete nonsense. You could be a cripple and completely bedridden all of
your life and still understand all of this perfectly. Your claim that a
physical experience is necessary for understanding is 100% bogus, and is a
construction of your own imagination put there for purposes of keeping
others from getting at the truth you yourself were unable to apprehend.
At one point in my memory, you were an excited discoverer of new lands.
Now you have shut down and wish to make everybody else shut down as well
because seeing other people move out ahead of you is intolerable. So you
have joined the club of naysayers just like Alex Gross. This is very sad
for us who knew you the way you were before, when you were still
interested in life and looking for breakthroughs. You have chosen evil
over good, and so now your fate is to perish in the gainsayings of Korah,
just like in the days of Moses.

My advice? Repent now and get back on track while there is time, because
you ain't gonna get any younger, and your negativism is gonna make an ass
out of you, just like it has these others, who have no real business
posting on this NewsGroup because this is where we try to figure out what
we CAN do, and not where we are supposed to be forever trying to prove
that computers really can't do what humans can after all.

--CD.

[Sergio Navega]

Chaumont Devin wrote in message <7s0u3n$eb5$1...@mochi.lava.net>...

>Dear Sergeo,
>
>I hate to have to cut you down like this after being friends for so many
>years, but now I find myself constrained to betray you for science, in
>other words to offer you up upon the chopping block of dissection and
>analysis, the alter of progress, etc.
>
>You say:
>
>>What is
>>constantly emerging from recent research is that reasoning
>>and language and perception and cognition are activities
>>that demand a body to be accomplished. More than that: they
>>demand interaction of this body with its environment, and the
>>learning from such interactions.
>
>People have said things like this before, as in Jesus, after having
>actually created the world through linguistic means, still having to
>descend into it to really understand what he had created through physical
>interaction. So I be wondering whether you may not have gotten some of
>this from the confessional, and if you have, I think you should confess.
>

Well, I confess that I didn't understand quite well what you said.
The need of a body to support cognition is something that's being
studied seriously since, at least, Jean Piaget, during the 50's.
Cognitive scientists and neuroscientists are finding increasing
evidences in support for this position. Several cognitive studies
verify, for instance, that the somatosensory cortex (a belt of
neurons responsible for sensory and motor coordination) are
activated during planning tasks involving movement or *thinking*
about movement, as when one is saying some phrase that involves
that kind of action. It's all intertwined.

>>For instance, suppose that we're reading a phrase about
>>someone who was walking and stepped on a stone loosing his balance.
>>The semantic content of this phrase can be understood by an american
>>and by a chinese, even if their surface representations differ
>>substantially. It is this "core" level of processing, that bears
>>strict relation to the organism's sensory experiences, ...
>
>Complete nonsense. You could be a cripple and completely bedridden all of
>your life and still understand all of this perfectly. Your claim that a
>physical experience is necessary for understanding is 100% bogus, and is a
>construction of your own imagination put there for purposes of keeping
>others from getting at the truth you yourself were unable to apprehend.

One of the hardest cases are the unfortunate who were born without
normal sight *and* hearing. These people develop impressive language
abilities, but at the cost of extensive touch exploration of the world.
No matter which sense is left, it is used to construct a representation
of the world that is the foundation for cognition. Without that
foundation, cognition is not reliable.

Several studies were made of sensory deprivation. Normal people, after
a period of just 2 days of complete sensory deprivation, becomes
completely out of balance, without further notions of reality and
incapable of sequenced rational trains of thought. Once their
interaction with the world is resumed, they rapidly return to
normality.

>At one point in my memory, you were an excited discoverer of new lands.

I keep being a discoverer (perhaps more than before), only that the
land I'm after is way ahead of what I used to think.

>Now you have shut down and wish to make everybody else shut down as well
>because seeing other people move out ahead of you is intolerable.

My greatest complaint is the *lack* of people saying things in this
and other forums. If I'm to commend you it is for pursuing your way,
following what you think is correct. I think there's few others who
are doing that and I have my own road to go.

> So you
>have joined the club of naysayers just like Alex Gross. This is very sad
>for us who knew you the way you were before, when you were still
>interested in life and looking for breakthroughs.

The breakthroughs I'm looking for now are different. I'm accompanying
very excitedly some findings of the rim of cognitive neuroscience.
The subjects of the day for me are such things as cognitive binding,
mismatch negativity, encoding in the olfactory bulb, stochastic
models of neurons, etc. When one obtains a sufficiently wide view
of traditional problems faced by AI (symbolicism, connectionism, etc),
one is able to direct his/her attention to the questions that need
insightful answers. The answers to AI are not inside this discipline,
they are outside. This is what I'm doing.

> You have chosen evil
>over good, and so now your fate is to perish in the gainsayings of Korah,
>just like in the days of Moses.
>

Geez, what a twisted version you have here. We're just in different
tracks, Chaumont. Just because my road is different than yours doesn't
mean I'm "evil" and you're "good".

>My advice? Repent now and get back on track while there is time, because
>you ain't gonna get any younger, and your negativism is gonna make an ass
>out of you, just like it has these others, who have no real business
>posting on this NewsGroup because this is where we try to figure out what
>we CAN do, and not where we are supposed to be forever trying to prove
>that computers really can't do what humans can after all.
>

The only thing I must repent is not perceiving these things before.
I'd like I had drifted to cognitive neuroscience way before.
But even that would be a loss, I would have missed the act of discovery,
which is a pleasure all by itself.

By the way, you may think that I'm one of those who think that computers
cannot be intelligent. After all, I've puncturing some balloons recently
and this could be interpreted as a sign that I don't believe it is
possible. Nothing could be far to the truth: I believe computers can
be intelligent, I just have a lot of reasons to think that conventional
approaches aren't close to the solution. We can't forget that AI have
been the primary field of research of thousands of extremely brilliant
minds for the past 4 decades. One must think out of the box if one is
willing to get ahead of that pack.

Regards,
Sergio Navega.

[rob]

Let me condense and reiterate to confirm that I understand you. What your
saying is that people hear what they want to hear? Ok.
As an aside, I do believe that someday there will be machine translation.
However, Mr. Gross is absolutely correct in placing humanity on it's proper
peg in the development timescale. We don't have a fundemental understanding
of the human brain, let alone it's mind. Without these prerequisites, how
can one hope to rewrite a very impressive piece of biological programming
onto a new substrate, such as silicon.

RaNdOm PaTtRn

nite....@usa.net

[Chaumont Devin]

Dear Sergeo,

Please do not evade the point of what I said by returning to Piaget. Just
explain why you believe automated systems need bodies to understand things
when cripples can understand slipping and falling on rocks without ever
slipping and falling themselves. Otherwise you are just wasting my time
and your time.

Thanks, and best regards from Honolulu.

On Mon, 20 Sep 1999 15:02:54 -0300, "Sergio Navega" <sna...@ibm.net>
wrote:

[Sergio Navega]

Chaumont Devin wrote in message <7s8iv9$2ue$2...@mochi.lava.net>...

>Dear Sergeo,
>
>Please do not evade the point of what I said by returning to Piaget. Just
>explain why you believe automated systems need bodies to understand things
>when cripples can understand slipping and falling on rocks without ever
>slipping and falling themselves. Otherwise you are just wasting my time
>and your time.
>

The understanding that cripples have of slipping and falling, or the
understanding that you and I have of floating in space like an
astronaut is very similar: we map some of our sensory experiences
into others, we *imagine* what it is like to have that experience.

This allows us to use those concepts in natural language, but
it does not give us the "real" feeling of those experiences. If
one have never tasted a papaya, one can be aware that it is a
sweet fruit, but he/she will never have "the real feeling".

What I'm saying is that if someone doesn't goes directly through
one experience, *neither* have any similar experience to map from,
then one will not be able to figure out what is behind the idea.
This is what happens with a computer that does not have sensory
inputs: it can't map the concepts into the real world, it "lives"
in a virtual one.

An interesting report by Landau and Gleitman (1985) shows how this
work in practice. They found that congenitally blind children were
able to learn the meaning of the word "look". One may be puzzled
by how this happened, after all they didn't see anything. What
they found is that all references that the children had of "look"
were associated with haptic (touch) exploration of the world,
suggesting that they captured one way to ground the meaning
of that word with activities involving touch, their primary sense.

This adds to several other studies such as ones done by Damasio
in which verbs were found to activate motor areas of the brain
while nouns did not. There is a clear link between sensorimotor
activities of the organism and the support of linguistic
constructs.

Regards,
Sergio Navega.

[Chaumont Devin]

On Tue, 21 Sep 1999 11:41:39 GMT, "rob" <now...@somewhere.com> wrote:

>We don't have a fundemental understanding
>of the human brain, let alone it's mind. Without these prerequisites, how
>can one hope to rewrite a very impressive piece of biological programming
>onto a new substrate, such as silicon.

By elementary deduction, Dear Watson. By elementary deduction.

[Chaumont Devin]

On Tue, 21 Sep 1999 17:38:48 -0300, "Sergio Navega" <sna...@ibm.net> wrote:

>The understanding that cripples have of slipping and falling, or the
>understanding that you and I have of floating in space like an
>astronaut is very similar: we map some of our sensory experiences
>into others, we *imagine* what it is like to have that experience.

QED. Understanding is possible without experiencing.

--CD.

[Sergio Navega]

Chaumont Devin wrote in message <7sb50m$1bo$2...@mochi.lava.net>...

You got that wrong: partial understanding is possible when you use
other experiences to map to new (or to compare with). But you never
have the exact notion of that experience without really going through
it. Just think about the sensation of a new fruit, unknown to you,
that somebody defines as tasting "sweet and dry like an apple".
Now think about that new fruit if you *never* tasted an apple.
The new fruit continues to be unknown to you. Now think about
a whole bunch of links saying "tastes like that, is red like
that one, etc". This mary-go-round never ends and understanding
is never achieved.

In order to make sense of that, you must have some kind of prior
reference(s) to the real world, references that keep some sort
of correspondence with the causal nature of that world, and that's
what cripples do. But nobody can make this mapping without having
something in which to map from: a computer doesn't have *any*
thing to do that.

Regards,
Sergio Navega.

[Chaumont Devin]

On Wed, 22 Sep 1999 17:09:32 -0300, "Sergio Navega" <sna...@ibm.net> wrote:

>In order to make sense of that (slipping and falling), you must have some kind of prior

>reference(s) to the real world, references that keep some sort
>of correspondence with the causal nature of that world, and that's
>what cripples do. But nobody can make this mapping without having
>something in which to map from: a computer doesn't have *any*
>thing to do that.

Well then why not just GIVE it that something?

--CD.

[Sergio Navega]

Chaumont Devin wrote in message <7sdkol$9i3$7...@mochi.lava.net>...

Because that's not "giveable". It is only "apprehendable".
Worse than that, each "detail" of those sensations usually
comes in large groups that cannot be fully defined in conscious
terms, besides having no meaning when taken in isolation. Our
conscious thought (the one we're using to talk right now) cannot
grasp the full content of the huge amount of things that's
going on in the realm of our lower perceptual levels. This
level is essential for cognition and it is elusive enough
to deceive us into thinking that it is not necessary. But
it is the foundation.

Have you ever jumped in those "bungee-jumps", where you're
tied in a huge elastic string and you fall (or is pushed :-)
from bridges or towers? That's a guaranteed way to raise
the adrenaline level of anyone. I've never been subjected to
such an experience (and if depends on my wish, it appears I'll
never will), so I can confidently say that my "idea" of
what it is to bungee-jump is very limited, no matter how
much textual descriptions I may receive. Sure, I can imagine
what it could be, the feeling of blood suddenly raising to
my brain, the movement of my internal viscera, the sudden
loss of consciousness, the dizzyness. All this is imagined,
I have no way to actually feel these things without
experiencing it directly.

That's the same thing with *all* our senses: a deaf fellow
cannot figure out what it is to listen to a good tenor singing
the main aria in La Boheme. He/she may "imagine" what it could
be, when making parallels between sounds and vibrating light,
which he/she can see. But it is less than a rough approximation.

Now, without any sensory reference, of any sort, an entity will
not be able to *analogize*, it will not have any kind of reference
in which to support its cognition.

Regards,
Sergio Navega.

[Chaumont Devin]

[Sergio Navega]

John Burch wrote in message <37ee9...@feed1.realtime.net>...
>Sergio,
>
>I've seen this issue (giving versus apprehesion) come up between you and
>Chaumont many times and I hope this 3rd perspective may say things in a
>different way.
>
>I believe that brains and computers can be considered equilavent to the
extent
>that they implement similar functions, irrespective of how those functions
are
>implemented in hardware(circuits and neurons) and software(programs and bio
>neural net weights).
>
>You, Sergio, appear to be saying that there exists some fundamental
>information in a successful brain that is derived from direct input from
the
>envirionment via the CNS, and that we can not duplicate such information in
a
>program, either because we can not model that input process or we can not
>create the information, in place, that would result from such a process.
>
>It 'feels' to me like the brain would use the same technology to store and
>process this fundamental experience information, the same way it processes
the
>higher level information we call language. Because experience information
is
>mainly unconscious, this is harder to point at and talk about. But I trust
>form follows function and we can see the fuction. Hopefully we can then
deduce
>a working information structure with the same functionality, although it
may
>differ from the information structure in a real brain.
>
>If it does store experience info the same way it stores language info, then
we
>just need to come up with structures that contain the same meaning that is
>inherent in the experiencially derived structures and validate it by trial
and
>error.
>
>If it does not store it the same way, then we have a more difficult
problem,
>but one that still should be solvable in terms of simple cause and effect.
>
>For example, I drive a car and am constantly, without thinking about it,
>calculating the closing speed of different objects and feeling OK or
worried
>depending on those internal predictions. This is an experience thing in
that
>it took years of childhood experiences to calibrate and train that model to
>simulate reality accurately and quickly. I got this simulator set up and
>calibrated by having real world experiences, however, I see nothing except
>ignorance and primative technology that keeps me from going inside and
>reconfiguring it for more accuracy, increasing the number of simultaneous
>degrees of freedom, and increasing the speed of simulations. I can much
more
>easily do the same thing to a software program today. But I don't see the
>difference between configuring a program today, and configuring my own
brain
>30 years from now. In both cases, I would be 'giving' the program new
>information or setting it up for the first time.
>
>I think Chaumont, and I, believe we can figure out structures that mimic
the
>naturally derived structures in our brains to handle non-verbal information
>like kinesthethics without using real world input to 'grow' that
information
>as we do now in childhood.
>
>Does this make any sense?
>John Burch
>

Yes, it makes sense, but I don't think this is correct ;-)

One of the fundamental questions is that what makes sense in terms
of "introspection" does not always make sense when implemented. Suffice
it to read about the more than 4 decades of frustrating experiences
of AI to know that something else lurks there. It is not the
obvious, intuitive solutions, the problem is more complex.

All of us can be said to be expert linguists in one way or another,
we have at least one language in which we're able to communicate
quite well and we "know" what words mean. We do that effortlessly,
even though this is the result of a *huge* amount of processing by
our neurons. This processing is not restricted to linguistic areas
of our brain, it involves motor and sensory areas, even when we're
thinking about "investing in the stock exchange".

The structures and strategies that we naturally envision to solve
this problem are the product of a misled vision of our cognition.
The greatest part of our sensory and perceptual activities are
hidden from "us" (I mean, our consciousness). However, we take
those experiences so "granted" that we often fail to notice
how hard it is (in computational terms) to come up with them.

What I usually say to approaches that try to model only this
"high level", is that one can't reliably put these models to
work without a sound and supportive model of how *our* world
works, and this is a task that occupies most of our childhood.
It involves capturing the essential operations of our body
in a physical world. This is the task that must have a
counterpart in our computers in order to make "them" understand
what we say. How else could a computer have a notion of verbs
such as "give", "push", "break"?

One cannot "understand" what language conveys without referring
to these lower level structures. They are not only the support
of our abstractions, they are also the source of material for
analogical and metaphorical forms of reasoning that are so
fundamental for the reasoning that we do under several
circumstances. There's no hope for AI, in my opinion, unless
one understands seriously what is happening during this
transition from low-level, perceptual tasks and the high-level,
symbolic (and visible) area. The secret lies in there.

Regards,
Sergio Navega.