Hist Ling, a Primer: Part 3 (was Re: "mono-descent is implicit in the comparative method ...")

[Robert Whiting]

On Thu, 28 Jun 2001 X99Lynx at aol.com wrote:

>In a message dated 6/27/2001 3:51:21 PM, dlwhite at texas.net writes:

><< By the way, since "mono-descent" is implicit in the
>comparative method, saying anything along the lines of "I accept
>the comparative method but reject mono-descent" is sort of like
>saying "I'll take the horse but not the legs". >>

>BTW, let's look at that statement: "By the way, since
>"mono-descent" is implicit in the comparative method,..."

>I have four different historical linguistics textbooks in front
>of me, including the admirable one written by Prof Trask.  I
>don't see a single definition that says anything about
>mono-descent.   I do see references to "systematic
>correspondence" between two languages.  I don't see anything that
>logically demands those "systematic correspondences" be only
>related back to only one ancestor.

Then either you are looking in the wrong place or else you lack
the ability to comprehend what you read.  Personally, I am
somewhat astonished to see you admit such a level of ignorance
quite so openly.  Unfortunately, I do not have Larry's book
available, but I find that I have been unable to look at any of
the texts that I do have without finding a clear statement to the
effect that the comparative method is used to trace the forms
found in related languages back to the most likely form in their
common ancestor.  Perhaps you are being confused by the term
"common ancestor."  If there is no clearer statement, this term
by itself implies "mono-descent."  Note that it does not say
"common ancestors," but "common ancestor" -- i.e., the single
language from which the related languages are all descended.
"Common" here does not mean "not rare" but means "shared by all."
And "common ancestor" does not refer to a grandmother who farts
at the dinner table.  It refers to the single language from which
all the members of a language family are descended.

First let's look at David Crystal's _Cambridge Encyclopedia of
Language_ (2nd edition, 1997).  This is a big, glitzy (especially
the 2nd edition with lots of color photos and side-bars),
coffee-table type book that has a wealth of information on many
aspects of language but not going into any great detail in any of
these aspects.  Here is what he says on the subject of language
families and the comparative method (p. 294):

  The first scientific attempts to discover the history of the
  world's languages were made at the end of the 18th century.
  Scholars began to compare groups of languages in a systematic
  and detailed way, to see whether there were correspondences
  between them.  If these could be demonstrated, it could be
  assumed that the languages were related -- in other words, that
  they developed from a common source, even though this might no
  longer exist.

  Evidence of a common origin for groups of languages was readily
  available in Europe, in that French, Spanish, Italian, and
  other Romance languages were clearly descended from Latin --
  which in this case is known to have existed.  The same
  reasoning was applied to larger groups of languages and by the
  beginning of the 19th century, there was convincing evidence to
  support the hypothesis that there was once a language from
  which many of the languages of Eurasia have derived.  This
  proto-language came to be called Proto-Indo-European.  Very
  quickly, other groups of languages were examined using the same
  technique.

  The main metaphor that is used to explain the historical
  relationships is that of the language family, or the family
  tree.  Within the Romance family, Latin the 'parent' language,
  and French, Spanish, etc. are 'daughter' languages; French
  would then be called a 'sister' language to Spanish and the
  others. ...

  This way of talking must not be taken too literally.  A
  'parent' language does live on after a 'daughter' language is
  'born', nor do languages suddenly appear in the way implied by
  the metaphor of birth.  Nor is it true that, once branches of a
  family begin to emerge, they develop quite independently, and
  are never afterwards in contact with each other.  Languages
  converge as well as diverge.  Furthermore, stages of linguistic
  development are not as clear-cut as the labels on a family tree
  suggest, with change operating smoothly and uniformly
  throughout.  ...

  The Comparative Method

  In historical linguistics, the _comparative method_ is a way of
  systematically comparing a series of languages in order to
  prove a historical relationship between them.  Scholars begin
  by identifying a set of formal similarities and differences
  between the languages and try to work out (or 'reconstruct')
  an earlier stage of development from which all the forms could
  have derived. ... When languages have been shown to have a
  common ancestor, they are said to be _cognate_.

  ...

  Genetic Classification

  This is a historical classification, based on the assumption
  that languages have diverged from a common ancestor.  It uses
  early written remains as evidence, and when this is lacking,
  deductions are made using the comparative method to enable the
  form of the parent language to be reconstructed.  The approach
  has been widely used, since its introduction at the end of the
  18th century, and provides the framework within which all
  world-wide linguistic surveys to date have been carried out.
  The success of the approach in Eurasia, where copious written
  remains exist, is not matched in most other parts of the world,
  where a classification into families is usually highly
  tentative.

Crystal tends to used "historical relationship" in place of
"genetic relationship," but this is unremarkable (see the comment
of Arlotto below).  He also says "common ancestor" rather than
"single parent," but this is also unremarkable since "common
ancestor" means the same thing and is the most frequently used
term.

Now let's look at the comments of Bernard Comrie in the
introduction to the compendium that he edited entitled _The
World's Major Languages_ (1987) on pp. 5-6:

  1.2  Language Families and Genetic Classification

  One of the basic organisational principles of this volume, ...,
  is the organisation of languages into language families.  It is
  therefore important that some insight should be provided into
  what it means to say that two languages belong to the same
  language family (or equivalently:  are genetically related).

  It is probably intuitively clear to anyone who knows a few
  languages that some languages are closer to one another than
  are others.  ...  Starting in the late eighteenth century, a
  specific hypothesis was proposed to account for such
  similarities, a hypothesis which still forms the foundation of
  research into the history and relatedness of languages.  This
  hypothesis is that where languages share some set of features
  in common, these features are to be attributed to their common
  ancestor.  Let us take some Examples from English and German.

  <snip of examples>.  Thus English and German belong to the same
  language family, which is the same as saying that they share a
  common ancestor.

Now let's look at the remarks of Philip Baldi in the section
"Indo-European Languages" in the same volume (pp. 33-36):

  Claiming that a language is a member of a linguistic family is
  quite different from establishing such an assertion using
  proven methods and principles of scientific analysis.  During
  the approximately two centuries in which the interrelationships
  among the Indo-European languages have been systematically
  studied, techniques to confirm or deny genetic affiliations
  have been developed with great success.  Chief among these
  methods is the comparative method, which takes shared features
  among languages as its data and provides procedures for
  establishing protoforms.  The comparative method is surely not
  the only available approach, nor is it by any means foolproof.
  Indeed, other methods of reconstruction, especially the method
  of internal reconstruction and the method of typological
  inference, work together with the comparative method to achieve
  reliable results. ...

  When we claim that two or more languages are genetically
  related, we are at the same time claiming that they share
  common ancestry.  And if we make such a claim about common
  ancestry, then our methods should provide us with a means of
  recovering the ancestral system, attested or not.  The initial
  demonstration of relatedness is the easy part; establishing
  well-motivated intermediate and ancestral forms is quite
  another matter.  Among the difficulties are:  which features in
  which of the languages being compared are older?  which are
  innovations?  which are borrowed?  how many shared similarities
  are enough to prove relatedness conclusively, and how are they
  weighted for significance?  what assumptions do we make about
  the relative importance of lexical, morphological, syntactic
  and phonological characteristics, and about direction of
  language change?

  All of these questions come into play in any reconstruction
  effort, leaving us with the following assumption:  if two or
  more languages share a feature which is unlikely to have arisen
  by accident, borrowing or as the result of some typological
  tendency or language universal, then it is assumed to have
  arisen only once and to have been transmitted to the two or
  more languages from a common source.  The more such features
  are discovered and securely identified, the closer the
  relationship.

  In determining genetic relationship and reconstructing
  proto-forms using the comparative method, we usually start with
  vocabulary.  ...  From these and other data we seek to
  establish sets of equations known as correspondences, which are
  statements that in a given environment X phoneme of one
  language will correspond to Y phoneme of another language
  _consistently_ and _systematically_ if the two languages are
  descended from a common ancestor.

Note that both these authors use the term "common ancestor."
This is the same as saying "single parent."

Now we can look at some textbooks on historical/comparative
linguistics.

First we will look at Hans H. Hock, _Principles of Historical
Linguistics_ (1986).  Hock's discussion of genetic relatedness is
intertwined with a general discussion of the nature of language
change which makes it too lengthy to quote in full, but despite
this, and the tendency toward Germanic syntax, the following clear
statements can be noted:

  Moreover, while all natural languages change, they do not
  necessarily change the same things at the same time.  As a
  consequence, as communication between different groups becomes
  more tenuous or stops altogether, linguistic change may
  increasingly operate in different directions.  Given sufficient
  time, then, the dialects spoken by these different groups may
  cease to by mutually intelligible and become completely
  different languages.

  At the same time, this divergent development in many cases does
  not go so far as to completely obscure the fact that these
  languages are descended from a common source.  In such cases we
  speak of Related Languages.
                                                  (p. 8)

  How long such linguistic relationships may remain discernible
  can be seen by looking at the set of vocabulary correspondences
  from the major languages of Europe ....  In fact, not only is
  it possible to recognize the major linguistic groups; within
  the first and largest one, that of the Indo-European languages
  of Europe, further subgroups can be established without great
  difficulties.

  One of these is Romance.  In the case of these languages we are
  lucky, in that their (near-)ancestral language, Latin, is
  attested.  We are therefore able to confirm our suspicion that
  these languages are related, by being descended from a common
  ancestor through independent, divergent developments.

  For the other groups, no such ancestral language is attested.
  And this is true also for the whole Indo-European language
  family to which Latin and the Romance languages belong ....
  However, by applying what we know about how languages change we
  can in many cases 'reverse' the linguistic developments and
  through 'Comparative Reconstruction' establish what the
  ancestral language looked like.
                                                  (p. 9)

Again, note the term "common ancestor."

Next, we can look at Raimo Anttila, _Historical and Comparative
Linguistics_ (1989).  Anttila is often not considered to be
"mainstream" because he does not generally adhere to any
particular doctrine but is eclectic in his explanations.
Personally, however, I find little that is objectionable in his
writings.  Often he presents a balance that is not obvious in
more doctrinaire or "school"-oriented linguists.  Among his
pithier comments on genetic relatedness we find:

  Languages connected by sets of correspondences form a language
  family.  Thus all the Romance languages are sisters, and
  therefore daughters of Latin, the parent or mother language,
  from which they are all descended.  'Related' is a technical
  term, exactly like the equivalent 'cognate', meaning that the
  items were once identical.
                                                  (p. 300)

and

  Those languages that represent outcomes of one and the same
  proto-language are grouped into a family.
                                                  (p. 318)

This is as pithy as you can get.  I don't see how anyone could
confuse this for anything other than a clear statement of "single
parent" and "mono-descent" as a criterion for genetic relatedness
and language families.  If you can explain how multiple languages
can be "one and the same language" then I will be waiting to
hear.  Otherwise, forget it.

Finally, let's look at Anthony Arlotto, _Introduction to
Historical Linguistics_ (1972).  This is a very elementary work,
so elementary that I often recommend it to those who have
absolutely no training in linguistics.  I tend to think of it as
the "Little Golden Book of Historical Linguistics."  I say this
without being derogatory.  It is just very simple and easy to
understand.  It hits the high spots without getting bogged down
in controversial or disputed points or using complex examples
that can only be understood by someone with linguistic training.
Here is what it says on the subject of related languages:

  It has been said that historical linguistics is based on a fact
  and a hypothesis.  The fact is that certain languages show such
  remarkable similarities to each other that these similarities
  could not be due to chance or borrowing.  The hypothesis is
  that these were once the same language.  We call this ancestor
  language a _common language_.
                                                  (pp. 38-39)

and

  When the evidence that two languages were once the same becomes
  conclusive, we then speak of these two languages as being
  _genetically related_.  Another way of phrasing the same fact
  is to say that they belong to the same _language family_.  Note
  that when we group languages genetically, our claim is purely a
  historical one and does not necessarily imply that the attested
  languages resemble each other in any particularly definable
  degree ....  In the course of time, the languages will have
  changed; the amount of change will be dependent on various
  factors, not all of which are clearly understood at the moment.
  ...

  Often, the similarities between two languages which attest
  their common origin will not be at all obvious at first glance.
  Or, on the other side of the coin, large amounts of borrowing
  may get in the way of correct conclusions.  However, by a
  rigorous application of the methods of historical-comparative
  linguistics, it is often possible to say whether two or more
  languages are genetically related.

Here we note the same point raised by Anttila:  Genetically
related languages were once the same language.  Indeed, here we
learn that this is the fundamental hypothesis of historical
linguistics.  We also note that Arlotto explains that genetic
relationship is a historical relationship, thus validating
Crystal's use of the term "historical relationship."  The two are
not interchangeable, however, for while a genetic relationship is
always a historical relationship, not every historical
relationship is necessarily a genetic one (unless one restricts
"relationship" to meaning "genetic relationship").

I am sure  that there are similar or identical statements to be
found in practically any textbook or introductory work that deals
with language families and genetic relatedness of languages.  Or
it that fails, you could just use a dictionary.  Here is what
Webster's Encyclopedic Unabridged (1994) has to say under
"comparative method":

  _Historical Ling._  a body of procedures and criteria used by
  linguists to determine whether and how two or more languages
  are related and to reconstruct forms of their hypothetical
  parent language.

You will notice that this says "parent language," not "parent
languages."  The single parent is just part of the definition of
related languages -- everybody's definition (except yours).

>What Winifred Lehmann writes is that the comparative method
>"contrasts forms of two or more related languages to determine
>the precise relationships between those forms."  Either as a
>matter of phonology or morphology, it seems it is forms, not
>languages, that are being "contrasted."

>If you can describe why or how you think "mono-descent" is
>implicit in the comparative method, that might make me think what
>you are saying is true.

I doubt it.  I have never known you to be convinceed by evidence
and rational argumentation.  But in case you actually mean it,
here is the reason why "mono-descent" is implicit, not just in
the comparative method, but in historical linguistics:
GENETICALLY RELATED LANGUAGES WERE ONCE THE SAME LANGUAGE.  Now,
what part of this don't you understand?

>At this point, you might want to take a closer look at that
>horse you are selling. It seems those legs are not what you would
>call factory options.

Which is exactly what he said.  You can't have the horse without
the legs.  You can't have the comparative method without
"mono-descent" of language families.  It's not an option.

<snip, snip, snip>

>><<I think you are mistaken about who got knocked out inf the
>>first round.>>

>Of course!  After all, theoretically - it just couldn't happen.
>How many fingers do you see?

Only one.  The middle one on your right hand.  You can put it
down now -- we got the message.

Bob Whiting
whiting at cc.helsinki.fi