[language] [Fwd: Re: In Click Languages, an Echo of the Tongues of the Ancients [long]]

[H.M. Hubey]

<><><><><><><><><><><><>--This is the Language List--<><><><><><><><><><><><><>




-------- Original Message --------
Subject: Re: [evol-psych] In Click Languages, an Echo of the Tongues of
the Ancients [long]
Date: Sat, 22 Mar 2003 22:04:22 -0500
From: "H.M. Hubey" <hubeyh at mail.montclair.edu>
To: Larry Trask <larryt at cogs.susx.ac.uk>
CC: Ian Pitchford <ian.pitchford at scientist.com>,
evolutionary-psychology at yahoogroups.com
References: <52518452.3257162602 at wren.crn.cogs.susx.ac.uk>



Larry Trask wrote:

> --On Thursday, March 20, 2003 9:26 am -0500 "H.M. Hubey"
>  wrote:
>
> [on my conclusion that all languages are equally old]
>
>> This is very strange. All languages cannot possibly equally old. ?Modern
>> English is  not mutually comprehensible with Old English (Can an average
>> Englishman read and comprehend Chaucerian English?).
>
>
> Mr Hubey is confused.  The discussion is about languages that are
> spoken today, and about nothing else whatever.  The suggestion of the
> geneticist Dr Knight is that certain languages spoken today in Africa
> are "older" than certain other languages spoken today, such as
> English.  As I hope I have shown, this conclusion is incoherent
> nonsense.  Apart from those few special cases, all languages spoken
> today descend in an unbroken line from the earliest beginnings of
> human speech.  Claiming that your language is older than mine is
> rather like claiming that your line of ancestors extends further back
> into the past than does mine.


Every time I write anything I tremble at the thought of being forced to
engage in yet one more inanity with the master
of confusion and ignorance.  Even the two yokels, alleged quantitative
linguists, who allegedly "reviewed" one of my
books in linguistics "Mathematical Foundations of Linguistics" are
incapable of understanding even the simplest
concepts everyone in the quantitative and mathematical scientists deals
with every day of the week and at least
twice on sundays.

There are so many approaches to this problem and yet I fear you will
never understand any of them. Here is
a short list off the top of my head.

1. Logic and Sorites paradoxes
2. Fuzzy Logic
3. Probability Theory
4. Measurement Theory
5. Statistical techniques - regression,  independent component analysis,
principal component analysis, etc
6. Datamining and Feature Extraction
7. Analogy, Biology, etc

All of these eventually wind up being related to the concept of
"dissimilarity" (usually called distance, metric
or distance metric). Two alleged quantitative linguists who reviewed my
book, as can be seen clearly
in their alleged "review'" have not yet understood this concept. I have
explained this many times to many
people but some never get it. I sent a short version of it to "Language"
and it is archived on the
LinguistList.org pages as a "review/comments" of the book on historical
linguistics written by
Kessler (2001) Stanford.

The first "review" of this book (by Kessler) was totally butchered,
obviously by a linguist who seems
to have learned his linguistics from you or your teachers. He got
everything backwards as usual. Kessler
seems to be psychologist and obviously knows what chi-square tests do.
It is an important remark on the
state of linguistics that a psychologist has to teach them how to do
science. And you never miss an opportunity
to show the state of linguistics as the doyen of linguistics to the
thousands of members of evol-psych.

Let's look at this one at a time.

1. Sorites Paradoxes

A modern of version of the Paradox of the Heap can be found in
Paradoxicon (Faletta).  There is
a video of N frames in which the 1st frame shows a tadpole and the Nth
frame shows a frog.
However there is no frame i, such that the ith frame is a tadpole and
the i+1 frame is a frog.
This is the basic problem of family-tree linguistics. That includes
spatial changes in languages
in which we might find dialects 1,2,3,....N stretching accross space in
which adjacent dialects
are mutually intelligible but others are not.

2. Fuzzy Logic

Can we fuzzify crisp logic that produces the sorites paradoxes and get
some reasonable
results? Can we also extend fuzzy logic to produce a kind of standard
modeling language
which has addition, multiplication and normalized variables? (probably
yes, because
I have done some of this myself). But here too the concept of
normalization rears its
head. What is that? Certainly I do not mean "normal probability density".

3. Probability Theory

Fuzzy Logic(s) can be thought of as special kinds of Probabililty theory
in which
the dependence is built into the specific logic chosen. Indeed, it is
more powerful
than fuzzy logic. In fact, the Cox-Jaynes axioms show that with some simple
assumptions it is easy to show that the reasoning has to be Bayesian
reasoning.

4. Measurement Theory

Here is the crux of the matter from yet another viewpoint. Originally
there were only
three scales: ordinal, difference and ratio (absolute). Then "nominal"
was added. And
nominal scale implies simply that one can make statements such as "X is
Y". Now, a dog
that knows how to classify food and non-food can use the nominal scale
of measurement.
Most historical linguists cannot explain to any other human being what
measurement they
use to create language families. And you head this list of linguists.
Because of the example
set by people like you, most linguists (even the obnoxious students) are
some of the
most narcissistic and ignorant students on the face of the planet. They
work like a herd
of sheep. They overpower by sheer numbers.

The "review" of my book is riddled with inanities, useless nitpicking
and even worse,
total inability to even understand what is clearly written in a ~250
page book.
I will take care of it one of these days, as soon as I get the rascals
in my sights in some
international meeting. Everyone will see how little they know. I will
make sure I videotape
and allow people to enjoy watching it for the next few decades. Nobody
should feel free
to slander and libel in public and get to enjoy it freely. The Laws of
Thermodynamics go
something like this;

i) you cannot win
ii) you cannot break even
iii) you cannot quit playing
iv) you have to pay to play

The "no-free-lunch" axiom in economics is built-up on these concepts.
Those that do not
understand these will be taught to understand them. No free lunch, no
free insults.

5. Statistical techniques - regression,  independent component analysis,
principal component analysis, etc

For a good example of this see Kessler's book. I heartily recommend this
book to everyone on
this mailing list. The book is "Significance of Word Lists", Brett
Kessler, CSLI (CEnter for
Study of Language and Information), Stanford, CA.  I am going to quote a
f ew lines from the
Preface.
-----------------------
Historical linguistics has no generally accepted methodology for
calculating whether the
connections it documents between languages are statistically
significant. This has led to
strident controversies.... This polarization is particularly strong
between those who
accept and those who reject the technique of mass comparison, which
groups languages
into families by collecting lists of words for the same concept, and
noting exceptional
similarities......As currently practiced, the traditional comparativist
method is much
more reliable than mass lexical comparison. But when a rigorous statistical
methodology is introduced, the most effective and reliable type of
evidence turns out
to be something that looks a lot like mass lexical comparison:
collecting lists of
words and comparing at a fairly superficial level those that name the same
concept. The kind of evidence most favored by traditional comparativists
turns out
to be intractable or less powerful in the context of a statistical
experiment.
-------------------------
(Disclaimer: I (HMH) do not claim that chi-square tests are the answer
and I am working
on better ones. Of course, it is orders of magnitude better than what
existed before.)
The back cover of the book says:
   Brett Kessler is a postdoctoral research fellow in the psychology
department of
Wayne State University.
-----------------------------
Everyone who reads this book will be contributing to peace on this
planet and will be
working towards the eradication of ignorance and pompousness parading around
as science. Please read the book. You will never have to take anymore
sh*t from
any member of the horde of semiliterate savages. Historical linguistics
will probably
be split into BK (Before Kessler) and AK (After Kessler). And to make
sure that
this horde of illiterate savages shuts up as quickly as possibly, I am
now developing
a semantic metric for Swadesh-like lists so that they can be used with
phonetic-phonemic
distances to do all sorts of statistical analysis and so that myriad of
other datamining
techniques can be used on the data.

Everyone who reads the book will be richly rewarded, if for no other
reason than
knowing that after they read the book, they will know more about historical
linguistics than Larry Trask. And after they read the "review" of the
book, they
will see how a typical linguist always manages to always get everything
backwards.

6. Datamining and Feature Extraction, (KDD)

AI (artificial intelligence) went thru many stages, and this is the last
stage. It is
about finding patterns in data. But that is what scientists are supposed
to do. Hence
KDD (Knowledge Discovery and Datamining) is really about the creation of
"artificial scientists"!  For the past 30 years or so while the
statisticians fell asleep
great advances were made in data analysis by mathematicians, computer
scientists
and engineers. Statisticians missed everything really big time, and now
they are
playing catch up. Anyone who wants to see what is happening should read this
book: Hastie, Tibshirani and Friedman "Elements of Statistical Learning",
Springer-Verlag, 2001. Those who are interested in undergrad level,
intro books
can look into Han & Kamber and also the book by  Hand, Mannila, and Smyth.

In any case, one of the main goals of KDD is "dimension reduction". Most
data
is/are high-dimensional and humans cannot handle 100,000 dimensions. The
goal
of "feature extraction" is to capture most of the patterns in data in a
few dimensions.
Biologists are familiar with this idea; they call them "characters".
Obviously historical
linguists have represent languages in a few dimensions, and these are
"lists"
e.g. lists of words which are specially chosen e.g. Swadesh-like lists.
Everytime we
throw away data (e.g. dimension reduction, feature extraction etc) we
lose precision.
But so what? That is done so humans can handle it. There is no problem
for a computer
to hold thousands of words in its memory and do number-crunching. But
humans want
things neatly done and in small , bite-size pieces. The data in KDD is
represented
in a concise way using mathematics obviously. And these may be
association rules,
or in statistics, the regression coefficients, or cross-correlations
etc. Or the data
may be represented using the principal components or independent components
etc. Everyone can see that historical linguistics has all the problems
that complex
data sets have. It is not surprising that geneticists (mathematicians,
computer scientists)
who run these programs on genetic data also try them on languages e.g. Tandy
Warnow and co-authors, for example. But people like Larry Trask are
giving answer
to problems that they do not even understand. And not be outdone by
their masters,
even many linguistics students are just as obnoxious.


7. Analogical (Biological etc)
Concept of "archaic" features exists in biology obviously. It would be a
reasonably
safe assumption that the genes we share with rats and frogs would be
"archaic". And those
human groups (e.g. "clusters" as argued some time back by one the
members of this
group), or "races" by another name, that preserve more of these genes
could easily
be considered more "archaic" e.g. older. I will not press this issue too
far. The analysis
as always is too superficial and I do not like being constantly
misquoted by those
who do not understand but whose narcissism and megalomania does not
allow them
to admit that there exist concepts on the planet that they do not
 understand better than
everyone else.

In any case, the rest of Mr. Trask's comments are not even fit to be read by
linguistics students and certainly nothing anyone should be proud of
thinking
let alone writing in public lists.

>
>> That means that a speaker of any language that can comprehend the
>> earlier
>> version of that language that was spoken say, 1000 years ago, is able to
>> converse in a language that is older than English.
>
>
> And now Mr Hubey is introducing a new confusion.  He notes that an
> ancestral form of English which we call "Old English" was spoken
> around a thousand years ago.  And, seemingly enraptured by the name,
> he tells us that Old English is "older than [modern] English".
>
> Think about this a moment.  An earlier form of English is "older" than
> a later form.  Let's apply Mr Hubey's incisive reasoning to another
> entity -- let's say Mr Hubey himself.  Suppose Mr Hubey today is 50
> years old.  That means that, 40 years ago, he was ten years old.  Now,
> by Mr Hubey's reasoning, the ten-year-old Mark Hubey of 40 years ago
> is older than today's 50-year-old Mark Hubey.
>
> Even by Mr Hubey's obviously unusual standards of reasoning, this
> strikes me as an extraordinary conclusion.
>
>
> Larry Trask
> COGS
> University of Sussex
> Brighton BN1 9QH
> UK
>
> larryt at cogs.susx.ac.uk
>

--
Mark Hubey
hubeyh at mail.montclair.edu
http://www.csam.montclair.edu/~hubey





--
Mark Hubey
hubeyh at mail.montclair.edu
http://www.csam.montclair.edu/~hubey



---<><><><><><><><><><><><>----Language----<><><><><><><><><><><><><>
Copyrights/"Fair Use":  http://www.templetons.com/brad/copymyths.html
The "fair use" exemption to copyright law was created to allow things
such as commentary, parody, news reporting, research and education
about copyrighted works without the permission of the author. That's
important so that copyright law doesn't block your freedom to express
your own works -- only the ability to express other people's.
Intent, and damage to the commercial value of the work are
important considerations.

You are currently subscribed to language as: language at listserv.linguistlist.org
To unsubscribe send a blank email to leave-language-4283Y at csam-lists.montclair.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/language/attachments/20030323/17f69d36/attachment.htm>