🚀 go-pugleaf

misrael@csi.uottawa.ca (Mark Israel) writes:

>>> "The rate of change of a language is a constant"...)
>>
>> Where did you get this from?

>   From Rich Alderson (an apparent expert on historical linguistics 
>who posts frequently to sci.lang and to alt.usage.english).  I had 
>written, "What I recall reading in the book _In Search of the Indo-
>Europeans_ is that among recorded ancient languages, Sanskrit and 
>Latin were closest to the reconstructed original; and that among 
>modern languages, Lithuanian is closest."  He responded, "Point of 
>order.  They have all, *by definition*, diverged to the same 
>extent."  

"Dogma" is not the proper word for that. "Bare-faced, shameless lie"
is.

Amazing. On LINGUIST, last week, there was someone bleating
"I have a very simple question: What are we doing wrong to
be on the receiving end of items like the following?"
The main item had been posted on comp.ai.natlang under the
title "An AI fable: Borne and the Wheel (a bit long)" and
was a wonderfully written fable about the preposterous dogmas
with which linguistics is... yes, *infested*.

Real linguistics at any rate, that is, hands-on, monolingual
fieldwork, has had me unlearn just about everything I had
been taught.

misrael@csi.uottawa.ca (Mark Israel) writes:

>> I believe only the first:
>>
>>> "The relation between words in a language and their meanings is
>>> arbitrary";

>linguistics course that I took, the experiment was not mentioned;
>even the possibility of doing such an experiment was not mentioned;
>we were just told the above and expected to believe it.

In Sociolinguistics, An Introduction (I believe in the USA it would be
called something or other 101), our lecturer DID mention the fact that
the current belief is as above but that there are attempts to disprove
it. She did not know what the success level was, but the attempt WAS
mentioned.

>>> "Speech is more important than writing";

I suspect that the above comment is actually a corruption of the CORRECT
adage that "speech is more ancient and, even now, much more used than
writing." Whether this makes speech more important is a hard question to
answer. I suspect not.

>Did Saussure do experiments to justify this choice of methodology?
>Apparently not.

This is non-cupatory - writing is the only system we may use for analysis of
ancient systems (languages and even partial languages) and some modern
languages (depending on aspect analysed).

>>> "Unconscious processes are more important than conscious decisions";

>   In a post to sci.lang last year, Edmund Grimley-Evans wrote:  "Not
>all linguists are Chomskyites, though the great majority of North
>American linguists are.  The reason Chomskyan linguists ignore
>writing is that they are interested in instinctive linguistic
>behaviour and are worried that written language might be influenced
>by conscious thought."

That is the opinion propunded by the English Linguistics Department of the
Hebrew University. Well, good on them. I, studying in the English Lings and
General (IE) Lings, hold the reverse.

>>> "Complete specification of a language's syntax without reference
>>> to semantics is both possible and worth doing";

>   You omitted this one.  Was that an oversight?  Did you mean to
>include this among the dogmas that you disbelieve?  You're in a very
>small minority among linguists if so.

>>> "The rate of change of a language is a constant"...)
>>
>> Where did you get this from?

This is bloody nonsense. Nowhere did I hear this. Not ONE lecturr ever
spoke thus. Quite the reverse, ALL languages change, that is dogma, if
you wish. The SPEED of change is spurting, at best.

The above is a reaction NOT to Mark Israel but to the poster HE answered.
--
---MAV

Marc A. Volovic                     Fencers do it with rapid thrusts!
(marc@shum.cc.huji.ac.il)          Guess how SABRE fencers do it, eh!?

Marc A. Volovic (marc@shum.cc.huji.ac.il) wrote:
: The above is a reaction NOT to Mark Israel but to the poster HE answered.

ROFLMAO! He is the one who posted the original deliberately 
inflammatory statements. So this IS a reaction to Mark Israel.

Karen
  karenk@netcom.com

In article <2vno1s$eg4@csi0.csi.uottawa.ca> misrael@csi.uottawa.ca
(Mark Israel) writes:

>In article <199407062210.PAA17283@netcom12.netcom.com>, Karen Kay
><karenk@NETCOM.COM> writes:

>>In article <2vf6b3$sqo@csi0.csi.uottawa.ca>, misrael@csi.uottawa.ca
>>(Mark Israel) writes:

>>> "The rate of change of a language is a constant"...)
>>
>> Where did you get this from?
>
>   From Rich Alderson (an apparent expert on historical linguistics 
>who posts frequently to sci.lang and to alt.usage.english).  I had 
>written, "What I recall reading in the book _In Search of the Indo-
>Europeans_ is that among recorded ancient languages, Sanskrit and 
>Latin were closest to the reconstructed original; and that among 
>modern languages, Lithuanian is closest."  He responded, "Point of 
>order.  They have all, *by definition*, diverged to the same 
>extent."  In e-mail, he clarified that he was not nit-picking about
>a definition; by "by definition", he really means "by dogma".

To quote a fairly well known television character:

	Whoa!  Back! up! the! truck!!!!

Your (apparently willful) misunderstanding of two issues, compounded by a
mistaken belief that I support that misunderstanding, must be cleared up right
now.

First, the notion of closeness, or distance, from a reconstructed protolanguage
has nothing to do with the rate of language change.

Second, do not put your notions into my mouth.  I am a linguist, a member of
that group whom you find so distateful, and as a linguist I seek, occasionally,
to clarify misinterpretations of the theoretical bases of linguistics, most
expecially of historical and comparative linguistics, by non-linguists.  When I
say something is true "by definition," I mean that in a rigourous sense; I most
certainly do *not* mean to say "by dogma."  I am perfectly capable of distin-
guishing between the two, and have nothing to gain by intentional obfuscation.

You note that my e-mail clarification to your query about definition; you fail
to note your return e-mail, to which I have not yet replied because I haven't
arrived at a sufficient response, regarding the distinction made by a biologist
between "divergence" and "diversity" and the possibility of this distinction in
linguistics.

You also fail to note that in my e-mail I stated a belief that you had seen
something in Mallory that was not there, that rather you had interpreted state-
ments as you wished to understand them.

Let me then outline the issues of protolanguages and attested languages, and of
determining rates of linguistic change of different types.

We define a _prototypical model of an unattested language based upon the recon-
struction, comparative or internal, of lexical items (words, bound morphemes,
etc.)_ (or, for brevity, a _protolanguage_) to be a timeless construct derived
by the negation of hypothesized changes from a near-"Platonic ideal" state.
Each language in the family contributes to the reconstruction, ideally with the
same weight though this has not always been entirely true in the past.

This is the first sense in which each language in a family can be said, *by
definition*, to have diverged from its reconstructed ancestor to the same
extent as any other.  It is, however, the less interesting of the two.

Let us examine for a moment the things which can change in the history of a
language.  Obviously, phonetics change vary greatly over time; this is even
true in time frames amenable to experimentation, as for example the work done
by Labov and other sociolinguists.  Such phonetic variability leads us on to
phonemic variation, among speakers of different dialects across time; again,
this is amenable to examination by a single generation of linguists.

Next, we can turn to various forms of lexical change.  It is possible to detect
acquisition and deletion of lexical items, reassignment of lexical semantics
(as, for example, specialization or generalization), and so on.  (NB: In this
context, _lexical_ is to be understood as "meaning assignable to individual
morphemes, free or bound.")  This is more difficult to study in a short time
frame, though not impossible; more certain results are probably to be found in
the long term.

Finally, we can turn to change in syntactic structures.  This is the most
difficult area of historical study, more because the synchronic basis has been
in flux for so long than through lack of data or interest.

Having taken note of the things that can change in the history of a language,
let us try to state what it would mean for any member of a language family to
"be closer" to the ancestral state than another:  We somehow have to provide a
metric for weighting changes against one another, even if it is as simple as
counting changes in different (sub)systems and comparing the raw scores.  The
final outcome of any reasonable metric, I would claim, is that any group of
changes which suffice to make one language distinct from another also suffice
to distinguish it from the ancestor.

Now that we have gone over that notion, let us examine briefly the question of
rates of change in language.

First, the question must be raised:  What kind of change?  The lexicostatisti-
cians and glottochronologists work with lexical replacement, change in a list
of supposed lexical items which resist this; they do not take into account
differences in phonology or phonetics as long as a case can be made for two
words somehow being "the same over time."  On this basis, the amount of change
can vary from nil (Old Icelandic to modern Icelandic) to roughly 60% (Inuit);
one must accept either widely varying rates of change or a much longer timeline
for Inuit than is currently held likely.

If we are to talk of phonetic or phonemic variation, we can note that in New
York City social dialects seem to have gone through three changes of rhotic vs.
non-rhotic (with hypercorrections) in the last 70 years, while the canonical
Vermonter has seemingly not changed in 200.

Similarly, to contrast changes in syntax, the Germanic languages moved from the
reconstructed SOV main clause order of Proto-Indo-European to SVO before our
earliest documents; the Romance languages made the same change much later; and
Indo-Aryan languages are still mostly SOV but in many cases have an ergative
rather than an accusative marking (so that we should talk of Agent-Patient-Verb
instead), but this too can be seen in historical times.

So, I invite you to retract the claim that I told you that linguistic change
occurred at a constant rate.
-- 
Rich Alderson   You know the sort of thing that you can find in any dictionary
                of a strange language, and which so excites the amateur philo-
                logists, itching to derive one tongue from another that they
                know better: a word that is nearly the same in form and meaning
                as the corresponding word in English, or Latin, or Hebrew, or
                what not.
                                                --J. R. R. Tolkien,
alderson@netcom.com                               _The Notion Club Papers_

In article <aldersonCssywn.Gsq@netcom.com>, Richard M. Alderson III <alderson@netcom.com> writes:

>>> They have all, *by definition*, diverged to the same extent."
>
> [...] When I say something is true "by definition," I mean that in 
> a rigorous sense; [...]  You note [...] my e-mail clarification 
> to your query about definition; you fail to note your return 
> e-mail [...] regarding the distinction [...] between "divergence" 
> and "diversity" [...]

   O.K., I misunderstood you.  When you said "by definition", I 
thought you meant "by definition of 'diverged'".  From my query 
about "divergence" vs "diversity", I thought you would have gathered 
that this was what I thought you meant.  It now seems that you meant 
"by definition of 'reconstructed protolanguage'", yes?

> First, the notion of closeness, or distance, from a reconstructed 
> protolanguage has nothing to do with the rate of language change.

   *Nothing* to do with it?  If a set of contemporaneously observed 
languages are all equidistant from a reconstructed protolonguage,
then either each of them has changed at the same average rate, or 
the reconstructed protolonguage can't possibly match the real
protolanguage, right?

> I am a linguist,

   By the way, Rich, how can I verify someone's claim to be a 
linguist?  My first thought was to look in _LInguistics and Language 
Behavior Abstracts_, but I couldn't find you there.

> Each language in the family contributes to the reconstruction, 
> ideally with the same weight though this has not always been 
> entirely true in the past.

   Why "ideally with the same weight"?  I can immediately think
of three reasons why equal weighting would not be ideal.

(1) Our earliest records of the various languages are not 
contemporaneous.  Since we have Latin, we can give zero weight
to French and Italian.  Since we don't have a 2000-year-old 
ancestor of Lithuanian, we have to use modern Lithuanian, but
we should give it a lesser weight than Latin, since it makes
sense that older languages would be tend to be closer to the 
original.

(2) We know that geographical displacement correlates negatively 
with language change:  U.S. English and Quebec French are more
archaic than British English and French French.  Hence we should
give greater weight to languages whose speakers must have migrated.

(3) Some changes are intrinsically more likely to have gone in
one direction than in the other (or don't you believe this)?  We
should give greater weight to languages with forms that are 
intrinsically likely to be earlier.

> [...] the Germanic languages moved from the reconstructed SOV main 
> clause order of Proto-Indo-European to SVO before our earliest 
> documents; the Romance languages made the same change much later;

   But doesn't that mean that the reconstruction gives greater 
weight to the Romance languages than the Germanic, and (according to
your "equal weight" dogma) should be shifted?

> So, I invite you to retract the claim that I told you that 
> linguistic change occurred at a constant rate.

   O.K., you don't believe that language changes at a constant rate.
But you do seem to believe that a reconstruction which assumes it 
does is "ideal".  That's a pretty good example of a dogma.

misrael@csi.uottawa.ca			Mark Israel

In article <300ts4$77t@csi0.csi.uottawa.ca>,
Mark Israel <misrael@csi.uottawa.ca> wrote:
>In article <aldersonCssywn.Gsq@netcom.com>, Richard M. Alderson III <alderson@netcom.com> writes:
>>
>> Each language in the family contributes to the reconstruction, 
>> ideally with the same weight though this has not always been 
>> entirely true in the past.
>

Within the context of Indo-European historical linguistics,
I'd consider this a fairly direct allusion to Sanskrit (maybe
Greek and Latin too).  

>   Why "ideally with the same weight"?  I can immediately think
>of three reasons why equal weighting would not be ideal.
>
>(1) Our earliest records of the various languages are not 
>contemporaneous.  Since we have Latin, we can give zero weight
>to French and Italian.  Since we don't have a 2000-year-old 
>ancestor of Lithuanian, we have to use modern Lithuanian, but
>we should give it a lesser weight than Latin, since it makes
>sense that older languages would be tend to be closer to the 
>original.

True, although I would give a little more than zero weight
to Romance.

>(2) We know that geographical displacement correlates negatively 
>with language change:  U.S. English and Quebec French are more
>archaic than British English and French French.  Hence we should
>give greater weight to languages whose speakers must have migrated.

I doubt that is a "constant".  It's true that U.S. English is more 
archaic in some respects. In others, it is clearly not.  South African
Dutch (Afrikaans) has retained some archaic words, but in grammar 
(inflection), Dutch Dutch is far more 'archaic'.
Furthermore, this argument leads to circularity in matters such as
the "Indo-European homeland".  It's usually safer not to assume too
much about "who must have migrated".

>(3) Some changes are intrinsically more likely to have gone in
>one direction than in the other (or don't you believe this)?  We
>should give greater weight to languages with forms that are 
>intrinsically likely to be earlier.

Yes, some changes have that property.  Unfortunately, most do
not.  I'm afraid I don't understand the phrase "languages with
forms ... likely to be earlier".  This generalization from
individual forms to languages does not apply.  If you compare
any two (related) languages, you'll find forms intrinsically
likely to be earlier in *both* languages.  The best you can do
is find a greater proportion of them in one of the two.  If
that's what you meant, you have a point.

-- 
Miguel Carrasquer         ____________________  ~~~
Amsterdam                [                  ||]~  
mcv@inter.NL.net         ce .sig n'est pas une .cig