Alex Gross ha scritto:

It isn't really - it did that because you accidentally typed http:\\
instead of http:// in your earlier message.

Mary
(remove "nospam" to reply)

---
OJ's Email address:  enter.slash.slash.backslash.escape

  [...]
aTM:
See the end of this post for an *interlinear" repeat of James Lee's
parody -- which had this former German teacher laughing unto tears.

Mind.Forth did not exist five years ago; only Mind.rexx did.

MFMT (Mind.Forth Machine Translation) will at first be slow
and simple.  The coders will put in syntax structures for two
different natlangs (not GNATlangs, Mr. Lee!)  The machine will
know its knowledge base independently of any particular language,
but you will be able to query it in either language.  The use
of a German word will trigger German syntax and vocabulary
not in the "deep thought" of the AI but in the expression
of thought as it floats upwards through the mindlayers.

Not a chance, for at least ten, eleven, twelve years at least.

No such claim is made for Mind.Forth AI.  As parodied below,
the opposite is admitted (although in geekily garbled English):
a cyborg MT worker will cost society million$ more than a human.

http://www.scn.org/~mentifex/aisource.html 28may1998 Mind.Forth
describes the Prosperity Engine of the cyborg economy accruing
to all persons IF the multinational corporations don't hog it.

------- Appendix: Machine Translation Parody of James Lee ------------

MT -- die Maschine Ubersetzung U96 ist keimlich wenn nicht samenlich

ubergeschrieben in die kostenlos oeffentlich Laendereien kuenstliche
Befruchtung des Muecke (?) lange-her Programm

des belegtes Butterbroten die einzeln menschliche Sprachewoerterbuchen
und die Satzlehre

mitten in ein tiefes Geisteskerngehaeuse U91 Pfund pro Quadratzoll U92
und eine Oberflaeche Gehoergedaechtenis U91 Ohr U92.

Weil der Geist.Fort erlaubt der Einbau der Weichwarema_einheiten

fuer ein Mannigfaltigheid der menschlichen Sprachen U96

aber auf Kosten des Millionen Dollars mehr pro Kyborg-gegen-menschlichen-
arbeiter.

P.S. Habe ich der Turingpreis gewonnen?  [aTM: beinahe aber keine Zigarre]

Yes, it is a lot funnier the second time around, now that i know was
es sei.

And thanks to everyone for helping to bring me up to speed.

Gee, i guess i'm just too heavy-handed, i need to lighten up a bit.

Okay, i'll try to be more amusing.  It's real hard though--here are
the only two jokes I know:

                         THE SQUARE PEG RIDDLE

Q.  What do you call someone who tries to fit a square peg in a
round hole?

A:  A fool.

Q.  What do you call someone who goes on trying to fit a square peg
into a round hole for a whole week?

A:  An idiot.

Q.  What do you call someone who goes on for his whole life trying
to fit square pegs into round holes?

A.  A computational linguist.

                         THE CHOMSKIAN RIDDLE

Q.  How many Chomskians does it take to{*filter*}in a light bulb?

A.  One plus three infinities.  A single Chomskian to{*filter*}in the
light bulb itself.  The first infinity to diagram the sentence
describing this act.  The second infinity to explicate the
relationships between this diagram and the ever-shifting Universal
Grammar/Deep Structure fractal matrix.  And the third infinity to
denounce both the light bulb and the act of screwing it in as merely
part of a capitalist/imperialist plot to enrich the electric
companies and deprive us all of our basic right to freedom of
speech and our duty to revolt against electric companies everywhere
and the system they stand for.

(Which could of course explain why most Chomskians are working in
the dark...)

While it may be true that the current direction taken by the ai natlang
crowd to MT is misguided, I think it inevitable that MT will exist, and
will do as good a job as human translation, if not better, although still
many years away, and maybe with as yet unknown technologies.

[snip]

It already is.  It is inevitable the if MT exists, it will be faster than
human translators, especially with the ever increasing speed of computers.

This is the only claim that will need time to meet.  Currently MT is accurate
and cheap enough to be of great assistance in many areas, and those are
typically the areas where translation would have not taken place before due
to the cost.

Again, it already is.  I can buy software to help communicate with friends
and relatives when I don't have a good grasp of the language, and it
certainly works cheaper and faster than a human, although with a certain
loss of accuracy, but not enough that in general the meaning is lost.

This is just a generic claim.  What is "true and lasting human understanding"
anyway.  We don't have that with people who do speak the same language, so
I don't see how everyone understanding everyone will change that much.

John Turnbull

Wow, i've really learned a lot since i've been on here.  And so many
of you are so deep, profound, wise, and stuff.  And many, many thanks
for being so helpful.

Well, i went to take a look at that Mind.forth programming.  All of
those extended rems.  I mean, just Wow!

I scarcely know what to say.  Especially after i messed up before
and failed to recognise a parody when i saw it.

But i think i've got things straight now, at least i hope so.  All
those rems, they're a parody too...right?

That's how you do things here, am i right?  You just keep writing
parodies all the time and see who you can catch with them...

At least i think it's a parody.  But what do i know?  i can't even
get URLs straight...

Here are some excerpts.  You guys tell me.  Do you think this is a
parody...?

<snip>
------------------------------------------------------------

 4  \ a thought being generated in the "monad{}" mindcore is not
 5  \ expressed in a particular natural language until the thought
 6  \ rises to the surface of the mind by percolating through the
 7  \ Chomskyan grammar structures of the particular human language
 8  \ in which the robot mind is currently thinking or communicating
 9  \ its ideas.

 4  \    Having a built-in ontology of concepts does not reflect a
 5  \ belief (a la Jerry Fodor) that concepts are pre-existent in
 6  \ the human mind at birth.  Rather, the best use of Mind.forth
 7  \ as a demonstration AI will result if the robot brain is primed
 8  \ with ready-to-go concepts as a basis for the operation of the
 9  \ Chomskyan linguistic structures at the core of the robot mind,
 10  \ and as a basis for the interaction of a robot with its world.

 7  \ Nevertheless, it is very important to fix the onset-tag of a
 8  \ word stored in auditory memory, so that a Chomskyan language
 9  \ structure may activate that word and cause it to be "heard"
 10  \ within the auditory memory channel during the generation of a
 11  \ sentence of thought.  Here within Mind.forth, we are about to
 12  \ let the "COMPARATOR" function declare secondary or tertiary
 13  \ "b" onset-tags within a stored word so that the AI may parse
 14  \ a compound word down into its smaller subwords -- a function
 15  \ absolutely essential if Mind.forth is going to learn German.

 7  \ In the face of the coming superiority of machine intelligence
 8  \ over human intelligence, all the varying levels of giftedness,
 9  \ which currently (1998) lead to vast economic disparities among
 10  \ humans, will soon count for nothing when NO human is any match
 11  \ for the superintelligent machines.  Gone will be any basis for
 12  \ the enrichment of the few and the impoverishment of the many.
 13
 14  \ A Nietzschean transformation of values will ensue, with those
 15  \ humans being materially rewarded who give more of themselves.

 2  \ The robot will have free will inasmuch as it matches its motor
 3  \ options with its perceived or instinctual values.  The process
 4  \ of thinking about values or goals and how to attain them will
 5  \ initiate the formulation of goal-directed motor strategies.
 6
 7  \ A robotic free will can be designed to pursue any suggested
 8  \ motor option only when all related mental associations have
 9  \ died down below a threshold level.  If the associations cause
 10  \ concern, fear or alarm, they will not acquiesce in the action.

 2  \ We include here a DREAM subroutine merely to show that it is
 3  \ possible and where the DREAM function would fit within the
 4  \ overall scheme of the artificial mind.
 5
 6  \ In accordance with the theory of REENTRY as elucidated by both
 7  \ Gerald Edelman in a Theory of Neuronal Group Selection (TNGS)
 8  \ [ insert appropriate TNGS Web link here, if obtainable ]

 10  \ any memories which are activated during and AS a dream will be
 11  \ remembered by the waking consciousness only if such memories
 12  \ also trigger the REENTRY function that re-records the memories
 13  \ in their hodgepodge pastiche ensemble as a new experience made
 14  \ up of reentrant bits and snippets of previously recorded data.

 2  \ During robot SLEEP, we may be able to consolidate memories and
 3  \ thus boost intelligence by canvassing the sparse memory arrays
 4  \ for associative tags which need not be widely separated, and
 5  \ we may be able to crowd together the associative tags formed
 6  \ during the period of consciousness immediately prior to SLEEP.

 2  \ There is no intention here to make TRANCE a feature of the
 3  \ AI mind of intelligent robots.  Instead, TRANCE is mentioned
 4  \ only in order to show its place within a Theory of Mind.
 5
 6  \ The theory speculates that hypnosis works by suggesting to
 7  \ a subject mind that it is asleep, i.e, in a condition where
 8  \ any experience, such as a dream, is to be interpreted as true
 9  \ WITHIN THE FRAMEWORK OF THE DREAM.  But since the hypnotized
 10  \ mind is actually not asleep, it ill-advisedly trusts the
 11  \ hypnotist and lends full credence to suggestions, resulting
 12  \ in  HALLUCINATION  -- a mental feature
 13  \ which we are not eager to implement in a trustworthy robot.

---------------------------------------------------------------
<snip>

after all, if you're going to write an MT program & you know that's
a pushover, why stop there--why not solve all the other problems of
free will, social injustice, the transformation of values, sleep,
dreams, trance, and total human ineptitude while you're at it?
Doesn't that make sense?

and I mean, all that reliance on Chomskian grammar is positively
awe-inspiring.  i guess you can rely on that stuff to get you a
degree, a teaching job, maybe even tenure, but IMHO depending on
universal grammar to protect you against the full fury of language
is sort of like sitting down in the middle of a field with your arms
wrapped around your head as the hurricane moves in.

i predict that "Universal Grammar" will be remembered mainly--to the
extent that it is remembered at all--as an oxymoron on the order of
military intelligence or giant shrimp.  To wit:

      To the extent that it is universal, it is not a grammar, and
      to the extent that it is a grammar, it is not universal.

Also, one other little detail.  As you gather, i don't think too
much of Chomsky, but to the extent that he commented on MT at all,
he was sensible enough not to have too much hope for it.  For
instance:

        `as for machine translation and related enterprises, they
        seemed to me pointless as well as probably quite hopeless.'
        The Logical Structure of Linguistic Theory, p. 40,
        University of Chicago Press, 1975.

i also found a comparable 1972 quote, but i'm upstate & away from my
books.  Problem is that he didn't really take the trouble to say
this clearly enough (when did he ever say anything clearly enough?)
plus he also kept spouting all the other stuff that keeps the UG
& MT bandwagons going.

i could add more, but i'll try to act humble & shut up for now.

                            With good will to all!

                                    alex

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

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

Weep not for Evita, Argentinians, but for Alex Gross, whose
autodissected mind shows ineradicable infection by the meme:

We got him out of the hot water of his initial arrogance;
now we must "love bomb" him with cultist MT assurances.

Verbum sapienti:  Not "Random Eye Movements" (REM) but "REMarks"
in Forth:  \  This is a remark.  It plays no role in the logic.

The out-of-breath and dumbstruck memeletic (oh no! John Passaniti
will throw a fit at this neologism for "one infected by a meme")
grasps about for a line of attack to deal with the unshruggable.

In the Biblical NT manual, it often goes, "Jesus came upon a
paralytic...," but here Usenet has come upon a memeletic.

Other than with parodies, there is no way to speak to Usenet.

http://www.*-*-*.com/ ;Mind.Forth
for IBM clones" is where you should have gone for the latest code.

Instead, like the pilot Corrigan landing on the British Isles
instead of California, you went to the archival bloatware site
of the long since abandoned Mind.Forth that had to be resurrected
because Dr. Paul Frenger wrote about the heavily commented Mind.-
Forth in the Association for Computing Machinery Sigplan Notices
33(12):25-31 (December 1998) of the Special Interest Group on
http://www.*-*-*.com/ #proglangs
Programming Languages -- on topic here in comp.lang.misc etc.

And let us further snip most of the quoted Mind.Forth material,
available on-line at http://www.*-*-*.com/ ~mentifex/aisource.html
or in the post of Alex Gross q.v.  See also ACM Sigplan Notices.

 4  \ a thought being generated in the "monad{ }" mindcore is not
 5  \ expressed in a particular natural language until the thought
 6  \ rises to the surface of the mind by percolating through the
 7  \ Chomskyan grammar structures of the particular human language
 8  \ in which the robot mind is currently thinking or communicating
 9  \ its ideas.

Machine Translation (MT) is not a direct goal of Mind.Forth (MF).
Since Mind.Forth clearly separates the deep mindcore (now called
"psi" instead of "monad") from the surface auditory memory channel
and sandwiches English or German or Japanese in between the two --
with the option of any number of natural languages ad libitum --
the MF PD AI will have MT as a by-product, not as a thoughtless,
deterministic machine process like the playing of brute-force chess.
Instead, Mind.Forth (or its derivatives Mind.Lisp etc.) will have
to translate judiciously, ponderously, meticulously -- like a human
engaged in HT -- and at a cost of millions of dollars more per cyborg
than per human translator, because it costs much less to birth, berth
and Berkeley a human translator than a nouveau-niche cyborg automaton.

Whether you now first encounter the Mentifex Mind.Forth meme of
public domain artificial intelligence for a {*filter*}netic economy,
or you have been aware of it since it went abroad in ca. 1981,
ask yourself:  What is going to become of this PD AI gambit?

Is it nonsense, a parody, the ravings of a net.kook crank?
The deciding factor in whether enough minds get infected
to create the AI is probably the question of whether or
not Mind.Forth as posted to Usenet on 9-9-99 develops
enough obvious functionality as machine intelligence
to spread out and lead to not an arms race but to a
minds race and to a race of minds -- first on a par
with us humans and then -- enriching a few of us?

 7  \ In the face of the coming superiority of machine intelligence
 8  \ over human intelligence, all the varying levels of giftedness,
 9  \ which currently (1998) lead to vast economic disparities among
10  \ humans, will soon count for nothing when NO human is any match
11  \ for the superintelligent machines.  Gone will be any basis for
12  \ the enrichment of the few and the impoverishment of the many.
13
14  \ A Nietzschean transformation of values will ensue, with those
15  \ humans being materially rewarded who give more of themselves.

Now that you are a memeletic, of course it makes sense.  The meme
of Mind.Forth has blown open your mind and you are reeling under
the burden of having to readjust the trajectory of your career.

We are not in Kansas any more, Mr/Dr/Rev Gross.  The meme is not
a hurricane, but a whisper -- a most dangerous agent of infection.

Aw gee:  In Pravda there is no news, and in Izvestia, no truth.

As the superarrogant superusers say, you are not even wrong.
The creation of MF PD AI +/- MT is not a championship of Chomsky,
but an exploitation of the Beethovenesque gift he gave the world.

If you think that Mind.Forth is a parody, then try to get some laughs
out of the scariest parody of artificial intelligence ever written:
http://www.*-*-*.com/ ~phoenix/vinge/vinge-sing.html Vinge's
"Technological Singularity."

        As perhaps a partial onlooker (only partially well-informed on
all the readings y'all have put out there, I admit), maybe I could
moderate things as follows:

        Chaumont is developing a system that seems to do a lot of nice
things, is not quite "done" yet, but he tastes {*filter*}, and is
completely confident of complete and utter success. Alex is more than
a little skeptical, and his sarcasm has maybe pushed thing to being a
little more antagonistic than is ideal.

        I think the source of Alex's skepticism, and I echo it (if so)
is that many people have "tasted {*filter*}" before in computational
linguistics. And then it turns out that it was only the rustling of
leaves in the woods, and the beast is far more elusive.

        There's some question of where the burden of proof lies. IMO,
it lies on both sides. Chaumont has far from convinced me that his day
of success is inevitable. Meanwhile, Alex makes some rather absolute
claims on the difficulty of MT. In fairness, I think the empirical
evidence is that:
        1) MT is very, very hard.
        2) People saying they are very close to having acheived
comprehensive NL tools has always been an empty claim in the past.
        3) Chomsky-ites' claims about a good many things have proven
empty in the past.
        Based (not exclusively) on such observations, I think Alex
drifts into opinion put forth as fact.

        I think "real" MT (puts virtually all human translaters out of
work, in text->text translation, anyway) is decades away, but not
centuries away. I'm only putting that forth as my opinion, and I don't
have time in the day to do justice to that claim by argument; anyone
is free to feel differently.
        I think that *better* MT than what exists now, that is an
actual convenience, and not just an ugly hack appreciated only by
those with super-low standards is PERHAPS not so far off; somewhere in
the vast, vast gulf between Babelfish and the work of professional
human translators. That is a very wide gulf, and progress out into it
seems not impossible.

        To Chaumont, because I am curious, how would your system
handle things like this?

1)
Context: A policeman accepts a bribe from a criminal. Later, he
complains about the meal his wife has cooked. She says...
Utterance: "I'm just a bad cook. You're a bad cop."

How would it derive that "bad" means ethically for "cop", but "in
skill" for cook?

2)
Bob says...
Utterance: "Years of {*filter*} have stunted my intellectual development."

Would it infer that Bob means his STARRING in {*filter*}, or his WATCHING
{*filter*}?  Some languages might require the verb. How might it integrate
prior context to resolve this (alone, it is ambiguous)?

3)
Utterance 1: "I was walking to school, when I saw a black cat."
Utterance 2: "I was walking to school, when I was a boy."

In some languages, the first clause of each would be in the same tense
(English, obviously; German and Spanish, also). In other languages
(Russian, Navajo; correct me if I'm wrong), the respective second
clauses indicates different verb inflections in the respective first
clauses. Again, how is this handled?

4)
Sentences at the beginning of jokes, or other narratives:
Utterance 1: "A guy walks into a bar."
Utterance 2: "A guy walked into a bar."

IN ENGLISH
In narrative, the choice of present or past can be largely arbitrary,
with no clear difference in denotation. In other contexts, past and
present are obviously not interchangeable!
IN GENERAL
In many languages, this is not permissable. Some languages have
specific inflections for narrative.

Does your system handle this? The arbitrariness (or is it?!) of
past/present in English narrative; narrative as a morpheme in other
languages.

5)
Utterance 1: "Jack was talking to Suzanne."
Utterance 2: "Jack talked to Suzanne."

Some contexts could establish that Utterance 2 denotes a situation
that would be expressed with the perfect in languages with such a
construct; some contexts would establish Utterance 2 as an imperfect
(and some, neither). For contexts that make Utterance 2 imperfect, is
it semantically equal to Utterance 1? I don't think so, but
personally, I'll be damned if I can express the difference.

6)
Context: Speakers A just threw a water balloon at Speaker B.
Speaker A: "Boy, did I get you! Ka-spa-lash!"
Speaker B: "More like pa-woop."

Onamatopeoia... Good luck!

7)
Context: A sidewalk in Germany. A and B walk one way, speaking
English. C walks the other.

C: "Hello."
A: "Buenos dias."
C: "Ich mochte freundlich sein."

This *almost* happened to me once (I was B!). The point, if it is
obscure, is that C tries to use the language of A and B; A feels this
is unnecessary (because we were in Germany) and uses Spanish to poke
fun at C. C wants to defend his position and point out A's rudeness,
but switches back to German to do so. In any event, the difficulty
here is not only in understanding all the utterances, but to assess
the significance of choice of language, which is different in all
three utterances. This is a spin-off on many Gricean examples
regarding Manner (a parent spelling a word so a child doesn't
understand is a more classic one). I've never seen a system that
wouldn't get it's ass utterly kicked, if I may say so, by things like
this.

***

        Mainly, I tried to get at two kinds of objection/question:

        1) In general, I think there are some extraordinarily subtle
points in semantics; most of my examples key on verb inflections. I
don't think ANYONE has a good theory of all of these yet. Experts
disagree on what the time on Arabic verbs mean, and how they should be
translated. How can you claim to have a general solution that
underlies all languages, when even some specific points have so defied
complete understanding? Maybe I am just ingorant, but I'm a smart
person who has thought about English semantics, and I am very unclear
on the semantics of past progressive vs. simple past (for the
imperfect) in English (Example 5 above). Have you solved this, or have
you bypassed it?
        1b) Sarcasm, humor, Gricean norms and violations thereof. Also
hard to represent semantically.

        2) Some utterances are inherently ambiguous when taken alone,
but context can disambiguate which of multiple meanings was
intended. This requires full-blown reasoning, which I'm quite sure is
not a solved problem. I apologize if you have not claimed anything of
the sort; then my objection rests primarily on (1).

        My strong hunch is that these things are beyond your theory,
and that what you have is maybe a couple of steps beyond Conceptual
Dependencies, but a few dozen steps shy of solving NL. And I think
this is generally what Alex was trying to get at; as-is, this is hunch
and not proof, but if you don't see any way to deal with most or all
of the above examples, it is pretty much proof.

        Cheers to all,
        -JAR
--
Poverty is an anomaly to rich people: it is very difficult to make out
why people who want dinner do not ring the bell.
        -Walter Bagehot

What do you know? Lots, but not enough. For a start, biting sarcasm
isn't the way things are usually done here. Most of the time, we watch
what someone is trying to do, holler "{*filter*} emperor!" on occasion, and
suspend judgement about worth that's based on the way concepts are
expressed. You are permitted to interpret "REM" (in it's Forth
embodiment as "\") as "REMOVE". If a hack works, great. If it might
work, encougage it. If relevant comments are obscure, ask for better
ones. As for creeds, all are tolerated, and public sneers are few. We
have a resident astrologer; why not a resident Chomskian? The Phlogiston
theory carried chemistry a long way.

One of the things you do know is that you need to lighten up. We'll help
you practice.

Jerry

P.S. We also argue a lot about POSTPONE and such. I think that's
different.
--
Engineering is the art of making what you want from things you can get.
-----------------------------------------------------------------------

'I'm sure everyone here will understand that.'

No, is that really a valid future tense in English? I don't think so. It
is a stylistic embellishment of a 'present tense' idea: 'I'm sure
everyone here understands that.'. Yet I see it translated by real
translators into (what I assume is) a real future tense in French and
German. How could a poor, benighted machine cope?
--
Regards, John Woodgate, OOO - Own Opinions Only.
Phone +44 (0)1268 747839 Fax +44 (0)1268 777124.
http://www/jmwa.demon.co.uk Did you hear about
the hungry genetic engineer who made a pig of himself?
PLEASE DO ****NOT**** MAIL COPIES OF NEWSGROUP POSTS TO ME!!!!

Can you prepone something?
--
Regards, John Woodgate, OOO - Own Opinions Only.
Phone +44 (0)1268 747839 Fax +44 (0)1268 777124.
http://www/jmwa.demon.co.uk Did you hear about
the hungry genetic engineer who made a pig of himself?
PLEASE DO ****NOT**** MAIL COPIES OF NEWSGROUP POSTS TO ME!!!!

Constantly -- that's one of the bedrocks of Forth development.  There's no
standard word PREPONE, though; we tend to write application-specific
versions.

Recently someone posted an example of (improperly done) preponement; they
had preponed a table-generating function by running the function and
inserting the table into the source.  Unfortunately, they forgot to
document where the resulting table came from and what it was used for.

There are five stages of development; preponing in its strongest form
refers to doing something in an earlier stage.

0) non-existance.
1) concept.
2) design.
3) compilation.
4) runtime.

The ultimate way to solve a problem is to not write a program for it --
just let the problem solve itself.  :)  Using someone else's program to do
the job also fits in here.  Hence the mention of 'non-existance'.  A program
which does not exist contains no bugs.

--
-William "Billy" Tanksley

Thanks, John.  Very fair, very thorough, very observant.

Let's see what Chaumont has to say.  i was toying with the idea of asking
some similarly probing questions, but i held back.   i feared that in the
current mood it could only turn into a flame war, which is the last thing i
want.

Let me make this clear: i have absolutely nothing against Chaumont or his
system, and it may just be bad luck that he is busy promoting it here at the
same time that i am trying to develop some linguistic theories i have been
playing with.

i really wonder if there may be some limits to the completeness and
consistency of human language, similar to the Goedel Entscheidungsproblem
for math.  Frightfully esoteric, i know, but if this were to prove true,
there might be some real-world consequences for MT, AI, and even how people
carry on normal, day-to-day communication with each other.

                                                    very best to all!
                                                             alex

Thanks for your thoughts, Arthur, though i don't think i quite caught your
meming.

Really, if you can be cute, so can i, but i don't think either of us ought
to be doing so.

Truly enjoyed your aviation image--indeed, your style puts me in mind of the
old RAF pilots' tale of the poor blighter who lost control of his plane and
was doomed to fly in ever diminishing spirals until he finally disappeared
up his own {*filter*}aperture.

Do you really want to keep doing this?  i don't.

Why don't we both stop?

   Gruss von Gross!
                alex

Ah, now I see the context. I'm not sure that this is on-topic at all in
sci.lang.translation. 'Preponing' was quite common in the days of slow
8-bit computers like the BBC Micro, especially the generation and
storage of look-up tables at the beginning of a program run, to avoid
having to do complex calculations several times during the run.
--
Regards, John Woodgate, OOO - Own Opinions Only.
Phone +44 (0)1268 747839 Fax +44 (0)1268 777124.
http://www/jmwa.demon.co.uk Did you hear about
the hungry genetic engineer who made a pig of himself?
PLEASE DO ****NOT**** MAIL COPIES OF NEWSGROUP POSTS TO ME!!!!

Well, this is a real challenge, but only as far as these words have
different translations. If you can find a word in the target language
that covers both meanings, you're safe. But in general, ouch.

Chaumont responds "I am not convinced that there is any real
difference, because in both cases it simply means "unsatisfactory"."
This is a bad response, because it only applies to this example. There
are many cases where one word requires different translations. E.g.
"bad food". In my Dutch, German and Spanish you cannot translate it
with the same word you would use for "bad cook" or "bad policeman". And
there are other examples: bad child ({*filter*}), bad cheque (worthless),
bad feeling (ill, injured). This is just one word. There are at least
5000 highly ambiguous words...

Sorry, but the solution to this is well-known.

Translation of these things is very complex too. You need specific
knowledge about the construction of jokes (just any plain narrative
will not suffice). So, first you will need to be able to identify this
as a joke and then to have rules for translating it properly. This is
simply too much work for the moment.

Chaumont seems to think that this can be encoded into some kind of
"Interlinguish atom", but this will take us nowhere. The sentence "a
guy walks into a bar" needs to receive different treatment in this joke
context than in other narrative contexts, and just introducing an atom
doesn't really strike me as the perfect solution for this. Sounds like
alchemy.

And don't get me started about the use of subjunctives in Roman
languages...

Hey, don't make it too hard for the poor guy. Well, once the problem of
general translation is solved, this will follow automatically (this is
not a joke; I am completely serious; I think general translation
requires implementation of such a high level of intelligence that this
will be implemented as a consequence).

In my opinion, there are no subtle points in semantics other than
pragmatics, i.e. the enormous amount of knowledge language users
acquire over the years of the way expressions are used in non-literal
ways. This is nearly impossible to mimic. For this, a model of language
acquisition is needed and years of training.

I don't even think anyone has a good theory of syntax yet. Let alone
the rest.

Much too hard.

Not only that, it requires "reasoning" about illogical things (like
pragmatics and common world knowledge) and huge amounts of them too.
This is not the sort of reasoning computers are particularly good at.

No, MT still has a long way to go...

   Theo