Language as labeling



  • 1. Why modules?
    Evolution _has no particular solution as a goal_ for achieving reproductive success so the question you might ask is why was this modular solution used at every level? It must have been the easiest way to become more complex, in working stages. Stable configurations, isolated from their parts, that could act as wholes and also as parts in a bigger system. And how did it happen without an intelligent designer or programmer to decide what modules to use and how to connect them? JC

Language as labeling

Postby Just Playing » Wed, 09 Nov 2005 22:39:18 GMT

Is it possible to create a system, maybe similar to the periodic table
of elements in chemistry, which would allow us to break words into
elementary components?
IMO such a system would allow to label all the possible combinations of
these elements into words, concepts, etc. and create a kind of meta

Re: Language as labeling

Postby Edgar Svendsen » Thu, 10 Nov 2005 01:19:28 GMT

Sure!  In English the elementary components of words are letters.  In the 
ideographic languages like Chinese the elements are brush strokes.

Perhaps you mean the elements of the *meaning* of words.  If that's what you 
mean then I suspect the answer is no.  Take a word and try to identify the 
elements.  An apt word here would be "philosophy".  I think there are no 
dividing lines between the various shades of meaning that people use when 
using the word "philosophy"; instead of a set of elements there is a shaded, 
blurry map where one meaning blends imperceptably into another.  We talk of 
things like "the main thrust" or "the general gist" or "the general idea" of 
the various things that could be called "philosophy"

There are ways to define "fuzzy" areas but they do no lend themselves well 
to tabular displays.


Re: Language as labeling

Postby Wolf Kirchmeir » Thu, 10 Nov 2005 01:40:06 GMT

Morphology does some of the things JP wants. I think if you take a good 
course in mdoern linguistics, you'll both understand JP's question a bit 

Re: Language as labeling

Postby Immortalist » Thu, 10 Nov 2005 02:06:02 GMT


A word is a unit of language that carries meaning and consists of one
or more morphemes.

 http://www.**--****.com/ (linguistics)

...a morpheme is the smallest language unit that carries a semantic
interpretation. Morphemes are, generally, a distinctive collocation of
phonemes (as the free form pin or the bound form -s of pins) having no
smaller meaningful members.

English example: The word "unbelievable" has three morphemes "un-",
(negatory) a bound morpheme, "-believe-" a free morpheme, and "-able".
"un-" is also a prefix, "-able" is a suffix. Both are affixes.

Types of morphemes

-Free morphemes like town, dog can appear with other lexemes (as in
town-hall or dog-house) or they can stand alone, or "free". Allomorphs
are variants of a morpheme, e.g. the plural marker in English is
sometimes realized as /-z/, /-s/ or /-?z/.

-Bound morphemes like "un-" appear only together with other morphemes
to form a lexeme. Bound morphemes in general tend to be prefixes and

-Inflectional morphemes modify a word's tense, number, aspect, and so
on. (as in the dog morpheme if written with the plural marker morpheme
s becomes dogs).

-Derivational morphemes can be added to a word to create (derive)
another word: the addition of "-ness" to "happy", for example, to give


In human language, a phoneme is the basic theoretical unit that can be
used to distinguish words or morphemes. That is, changing a phoneme in
a word produces either nonsense, or a different word with a different

Phonemes in {*filter*}languages are not physical sounds, but mental
abstractions of speech sounds. A phoneme is a family of speech sounds
(phones) that the speakers of a language think of as being, and usually
hear as, the same sound. A "perfect" alphabet is one that has one
symbol for each phoneme.

In sign languages, a phoneme is a similarly basic theoretical unit of
hand shape, motion, position, or {*filter*} expression. It was formerly
called a chereme, but usage changed to phoneme when it was recognized
that the mental abstractions involved are essentially the same as in

Phonemics, a branch of phonology, is the study of the systems of
phonemes of languages.


In phonetics, a vowel is a sound in spoken language that is
characterized by an open configuration of the vocal tract where there
is no build-up of air pressure above the glottis.

This contrasts with consonants, which are characterized by a
constriction or closure at one or more points along the vocal tract.
The additional requirement is that vowels function as syllabic units:
it is this criterion that distinguishes vowels from semivowels (and
approximants, which in some languages may be slightly more


Re: Language as labeling

Postby Just Playing » Thu, 10 Nov 2005 05:30:12 GMT

I guess that is what I mean and I think or at least I hope that the
answer should be yes.

Your example is far too complex to base your negative answer on it. If
I may use chemistry as analogy, identifying the elements of philosophy
is like trying to find the formula for a very complex molecule before
you know anything about chemical elements.

If it really is imperceptibly there could not be any difference between
them, but if only one or a just a few elements change, as in a complex
formula, then we could have a perceptible change.

Re: Language as labeling

Postby Just Playing » Thu, 10 Nov 2005 05:57:40 GMT

I am not talking at all about morphology.
I will use a simple example to try to clarify what I am asking.
The chemical formula for water is H2O and we can translate the word
water in all the languages without changing the formula.
OTOH we had a word for water a long time before we knew its chemical
formula and its translation in any language was very clear.
IMO the common elements of the meaning of water are related to its
perception by our sensors and not to its chemical formula.
IOW the elements that give meaning to a word might be related to our
perception, to our sensors and not to its morphology.

Re: Language as labeling

Postby Brian Fletcher » Thu, 10 Nov 2005 08:01:27 GMT

This already exists. It is known as numerology.

The equivalent is that of written music as it relates to the actual music.

Having said that, even with music , the beauty (written or actual )is in the 
eye(ear)of the beholder, so , as usual, reality is not what you hear, read 
or have interpreted, but what you make of it at any given time.

You have, of course, to be a trained musician to "hear the  music"-befor it 
is played, as with any discipline, and even the very advanced may hear 
'almost'  the same but be affected differently.

The whole essence here, is that as individuals develop a more advanced sense 
of self, they often attempt to develop an "uncommon" communication level 
with those who are not on the same wavelength.


Re: Language as labeling

Postby Brian Fletcher » Thu, 10 Nov 2005 08:05:31 GMT


Very informative Imm.

Who was the Richard Cranium who decided not to spell phonetic with an 'f'?


Re: Language as labeling

Postby S2 » Thu, 10 Nov 2005 10:38:16 GMT

In the 16th century there was an effort amoung the French Philosphes as
well as the English Royal Society to develop a "Natural Language".
This would be a language conforming to the tennents of the scientific
method.  This lead to the creation of dictionaries and encyclopedias.
A number of them still kicking around today in revised form.  A google
search of Diderot will lead to more on this information.


Re: Language as labeling

Postby Just Playing » Thu, 10 Nov 2005 10:52:53 GMT

Thank you, but this is not what I am talking about. You gave me the
same reference a few months ago.
We can understand each other without dictionaries, it has to be
something that we all have in common. Some type of  model that we
internalize (non verbally) and against which we compare whenever we
communicate verbally.

Re: Language as labeling

Postby Just Playing » Thu, 10 Nov 2005 10:59:34 GMT

I am not talking about numerology but about meaning.
I look at words as "formulas" or "tri dimensional figures", and
these formulas or figures are the equivalent of what we call meaning.
I am trying to find how we create these "formulas" or "tri
dimensional figures", to find what they are made of, their elements
or building blocks.

Interesting stuff but it is not the same as what I am asking.

Re: Language as labeling

Postby forbisgaryg » Thu, 10 Nov 2005 21:23:27 GMT

I've tried reading through the various replies so as to understand
your question and am still not sure what you are asking.

Do you mean something like Roger Shank;s Conceptual Dependency theory
as highlighted at: http://www.**--****.com/ 
or something else?

I haven't checked into Cyc lately.  It appears it's open source now:

Immortalist gave some pretty good information.  You're going to have
some problems with prefixes such as "in" as in "inflammable",
and "inedible".

Most words have several meanings and each person has a slightly
conotations.  I've recently had a discussion about wheather "A
contradicts B"
should be translated into FOPL as A -> ~B or A <-> ~B.  I have yet to
down a authoritative source on the matter.  The Law of
Non-Contradiction as
I understand Aristotle expressing it is ~(A & ~A) and yet this doesn't
the question since LNC is about a single thing being and not being
rather than
about two things where it could be the case that both may not be.  I've
this discussion a rest because I might just be using a non-standard

There are also binding problems.  Consider these sentences:
Mary gave Jim her book.
Mary gave Jim his book.
The dog gave the cat its bone.
The dog gave the cat its scar.

In the last case it's clear what "its" binds to but not so in the prior
Why is that?

Re: Language as labeling

Postby Just Playing » Thu, 10 Nov 2005 22:14:07 GMT


I am looking at words as descriptions, as labels attached to a
The names of the labels have been created mostly accidentally by our
culture and morphology is of little help here.
I will try to use water again as an example.
"Water (from the "/wiki/Old_English_language" word wer; c.f
"/wiki/German_language" "Wasser", from "/wiki/Proto-Indo-European"
*wod-or, "water") is a tasteless, odorless, and nearly colorless (it
has a slight hint of blue) substance in its pure form that is essential
to all known forms of life" and is known also as the most universal
IMO the basic elements of this definition are:
1. taste
2. Odor
3. Color
4. Substance
Now if we look at these elements, they all describe information
perceived  by 4 different sensors, taste, smell, visual, tactile.
The information perceived by all these sensors is connected and a label
is attached, in this case water.
The basic idea is that starting from the information perceived by all
our sensors we create a lot of combinations and we label these
If we would label them the way we do in chemistry it would be clear
that by adding.subtracting or changing elements to a formula we achieve
another formula but it does not happen this way with the language.
For example going back to water if I replace one element in its
description, let's say substance with gas, I could describe air.
OTOH there is no clear connection between the 2 words, water and air,
but according to this way of looking at their perceptive description
they are different only according to the information perceived by one

Re: Language as labeling

Postby makc.the.great » Thu, 10 Nov 2005 22:15:52 GMT

I think it's valid to say that if A contradicts B, it means that A and
B cannot be both true at the same time:
NOT(A AND B) = 1
(NOT A) OR (NOT B) = 1
or a number of other ways.

btw, how is that related to a question at hand?

Re: Language as labeling

Postby makc.the.great » Thu, 10 Nov 2005 22:19:09 GMT what? what's your point?

Similar Threads:

1.Label Autopositioning Algorithm

Situation:  An air traffic control display depicts aircraft using position
symbols and labels.  Position symbols show the two dimensional location of
each aircraft.  Labels contain additional aircraft data and are (usually)
connected to its associated position symbol by a line.  Labels which overlap
are obscured.

Problem:  Labels can be positioned manually to avoid conflict, but this
takes time and effort on the part of the air traffic controller thus
reducing the amount of traffic that can be handled.  An autopositioning
algorithm can reduce overlap of labels but to be effective it must also meet
these requirements:

    * Allow selected labels to be manually positioned
    * Label movement must be smooth (i.e. no sharp jumps)
    * Labels should move relative to the position symbol only when necessary
    * Labels must remain near their position symbol

(There are many other constraints, but this should give you a good feel for
the problem)

Approach:  For each position symbol - label pair, plot the relative paths of
other position symbols and other labels for the next few minutes.  Move the
label in the direction to avoid conflicts with other position symbols and
other labels.

I call this the "dodge ball" model because of its similarity to the
playground game.  The label is constrained to a circle about its position
symbol and moves to avoid incomming objects.

I expect to find rules such as
    * IF no conflict THEN dont move
    * IF conflict on right side THEN move left
    * IF conflict is soon THEN move fast
    and other similar rules

Is this a realistic starting point?
Is fuzzy logic likely to provide a solution to this control problem?
What problems may I encounter since I have not implemented a system like
this before?
How can I test a tentative solution to see if it is realistic?


Bob Oliver

2.standard labels for probability values

Is there a federal standard for labels of probability values?
For example,
  Very Unlikely 0-20%
  Unlikely 21-40%
  Maybe 41%-60%
  Likely 61-80%
  Very Likely 61-100%

Is there any standards such as these?

[ is moderated ... your article may take a while to appear. ]

3.How to compare clustering methods using data without labels


 My problem is I want to compare several unsupervised clustering
algorithms but
 my data has no label.  I don't know on which base should I compare those

 Any insight will be truly appreciated...


 Neset Sozen

4.Multi Class Label Neural Network Implementation

On Oct 17, 4:09m, "Motaz K. Saad" < XXXX@XXXXX.COM > wrote:
> Hi,
> Did anyone come across or use a Neural Network Implementation the
> support Multi Class Label?. I would appreciate if anyone pointed me to
> it.

Any implementation that supports multiple
outputs will suffice. Just use {0,1}
targets for each output.

Hope this helps.


5.The language of aggrevation is Bush language

I hate it, no human should be allowed to be that
aggressive. So agressive that it feels like my
head is going to explode and that the world
will break.

6. Bush language and media language the same brutality (agressive business mentality)

7. [BVE PSA] Logical, natural language based programming language (t3d)

8. CFP: AIPR-07 Special Session on Frontiers of Language Processing and Information Retrieval for Asian Languages

Return to AI


Who is online

Users browsing this forum: No registered users and 50 guest