Hist Ling, a Primer: Part 2 (was Re: Trivial Truths and Genetic "Patterns")

[Robert Whiting]

On Sat, 23 Jun 2001 X99Lynx at aol.com wrote:

>In a message dated 6/22/2001 12:44:17 AM, hstahlke at gw.bsu.edu writes:

><< There's an almost trivial sense in which Steve has to be
>right, but I think he's taken his case beyond that point.  Having
>worked on Niger-Congo languages, especially those of the eastern
>half of West Africa, I've been faced with the question of where
>to begin....  As Bill Welmers used to say, "You get to the point
>where you know that these language can't be unrelated."  Of
>course, he would also add that you then start using the
>comparative method to work out the relationships and make sure
>they're there.  We're dealing with different orders of
>hypothesis.  Using some careful lexicostatistics gives you a
>reasonable hypothesis, but then applying the comparative method
>takes you to a much stronger one.  As I said, this is an almost
>trivially obvious point. >>

[For Herb]:  When Steve is right, it is almost always in a
trivial sense.  Being able to distinguish the trivial from the
significant seems to be one of his major problems.  Of course, it
isn't always easy, and it is exacerbated when one doesn't
understand the discipline or its terminology.  It is almost
impossible to tell what is significant if one does not understand
the parameters.

>Well, of course, my question was, at that point, did you ever
>consider the hypothesis that the language had two ancestral
>language groups to whom it "can't be unrelated."

Not the same thing at all.  Herb is talking about "these
languages" as a putative language family.  "The language" is not
a language family.  A language family is descended from a single
language.  That is the definition of a language family.<-- In
case you don't recognize it, that mark is a "period."  It means
that is the end of the thought.  In other words a language family
is a group of languages that are descended from a single language
PERIOD.

Saying "think of a language family with two parents" is like
offering a mathematical proof and saying "now divide by 0."
It just is not part of the system.  Division by 0 is not defined.
A language family with two parents is not defined.

Of course you could say "let's define a language family as having
two or more parents," but then you aren't doing comparative
linguistics any more.  You're doing a gedankenexperiment much
like saying "assume there is no gravitational attraction at the
earth's surface."  You can only do this as a thought exercise
because such a thing does not exist.  It is like constructing a
point at infinity so you can see what happens when all the
parallel lines in the universe meet (ans:  it gets very crowded)
or like saying "what if three points define a line? -- we only
get our picture of geometry by assuming that two points define a
line but how do we know that that's right?"

>But, in any case, believe me, what you've described is not
>trivial at all when you're dealing with some very 20-20 hindsight
>style explanations.  Larry Trask's explanation of why languages
>can't have more than one genetic ancestor typically offer the
>conclusions as if they were explanations.

On the contrary.  You are making the same mistake again.  Larry
hasn't been talking about individual languages and he hasn't said
that "languages" can't have more than one ancestor.  He has been
talking about language families.  You are the one who has been
talking about language families as if they were languages.  Now a
language can have as many parents as you want it to.  If you
think that every language that has contributed something to
English is a parent of English, then English has at least several
hundred parents.  But this is not the way that comparative
linguists consider language development or the way that they
define genetic links.  A language is not a language family.  A
language can be influenced by any number of other languages,
related or not.  You can think of these influencing languages as
parents of that language if it gives you a warm glow or helps
you sleep through the night.  But a language family has only one
parent because that is the definition of a language family.

>In another post I tried to ask why one "systematic
>correspondence"  - which would alone have been considered enough
>to establish a "genetic' relationship - should be considered
>non-genetic because of the presence of another "more genetic"
>systematic correspondence.

>I was obviously referring to a situation where both "genetic" and
>"non-genetic" elements both create what would ordinarily be
>called "patterns."  (The kind of patterns you might see even if,
>for example, the hypothesis you described in your post about
>relatedness ended up being wrong.)

[Or the kind of patterns you might see if you'd been smoking some
really weird stuff.]

>Here's Larry Trask's reply.  Note that below "miscellaneous
>common elements" aren't patterns.  Only genetic patterns have
>"patterns."

>In a message dated 6/22/2001 10:44:59 PM, larryt at cogs.susx.ac.uk writes:

><<We do not recognize genetic links on the basis of miscellaneous
>elements: instead, as always, we rely entirely on patterns....

>It is patterns that demonstrate common ancestry, not the presence
>of any number of miscellaneous common elements....

>We do not recognize genetic links on the basis of miscellaneous
>elements: instead, as always, we rely entirely on patterns....

>It is patterns that demonstrate common ancestry, not the presence
>of any number of miscellaneous common elements...

>And Albanian is not Greek, Romance, Slavic, Hungarian or Turkic,
>even though elements of these origins greatly outnumber the
>inherited elements....

>It is a distinct branch of IE, because this is the only
>conclusion permitted by the patterns we observe....

>Patterns, patterns, patterns.>>

>There may be an explanation for the one only parent rule.  But
>this cannot be it.

You are so fixed on the word "patterns" that you don't seem to be
able to see the key phrases in what Larry is saying.  These are
"genetic links" and "common ancestry."  These both refer to the
same basic premise:  Genetically related languages were once the
same language.  This is simply the definition of genetic
relationship in comparative linguistics.  "Genetic links" thus
refers to the correspondences between languages that were once
the same language.  "Common ancestry" refers to the same thing.
That two or more forms in two or more languages were once the same
form in a single language.  That's what genetically related (or
"cognate") means.  Let me repeat that in case you didn't get it:
LANGUAGES THAT ARE GENETICALLY RELATED WERE ONCE THE SAME
LANGUAGE.  This is the explanation of what you call the "one
parent rule."

>As you were kind enough to concede, you have to see patterns
>BEFORE you even apply the comparative method.  Heck, borrowing
>creates patterns.  "Patterns" therefore CANNOT explain the
>difference between genetic and non-genetic relationships.

Sure they can.  You just have to be able to tell the difference.
Admittedly, it's not always easy, but the more information you
have, the easier it gets.  And one of the things about doing
linguistic reconstructions is that you use all the information
you can get.

But saying that because inheritance and borrowing both produce
patterns you can't tell the difference between them is just
another of your famous non sequiturs.  It is like saying that
you can't tell the difference between plaids and polka dots
because they are both patterns.  And this is what you call
science.  I can visualize you dressed in plaid trousers, a polka
dot shirt, a paisley tie, and a striped blazer and claiming that
it is a well coordinated outfit because "hey, man, like they're
all patterns and like, y'know, all patterns are the same."
Believe it or not, not all patterns are the same.  It is possible
to tell the difference between different types of patterns.  And
believe it or not, some people can actually tell the difference
between different plaid patterns without a spectrometer.

Quite simply, the patterns that inherited words form are
(usually) different from the patterns that borrowed words form.
Inherited words have always been in the language (that's what
inherited means).  Borrowed words have been in the language for
less time than inherited words.  Thus borrowed words have
followed a different path to get where they are than inherited
words have and hence create different patterns.  For example, PIE
*p comes into Germanic as /f/ but into Greek and Latin as /p/.
Thus in English it is easy to tell inherited Germanic "flat" (ON
'flatr', OS 'flat',  OHG 'flaz') from 'plate', 'platter',
'platen', 'platypus', etc borrowed from Latin or Greek.

Of course, the fact that both Greek and Latin have the same
reflex of PIE *p means that it is harder to tell inherited Latin
words with /p/ from Greek loans, but this doesn't affect the
reconstruction of either Proto-Germanic or PIE, only of
Proto-Italic.  But it does point out that it is not always easy,
or in some instances even possible, to distinguish borrowing from
inheritance.  Sometimes the patterns aren't different enough to
allow a decision to be made.  But this is, in general, an
exception rather than a rule.  And I can't think of another
discipline (that isn't based on "first principles" such as
mathematics is) where there don't arise decisions that are too
close to call one way or the other on the basis of the available
data.

Frequently even borrowings of the same word or root will pattern
in different ways depending on when the borrowing took place.  In
French, Latin <c> ([k]) before [a] developed first into [C]
([tS]) and later into [S].  Because of these changes, it is often
possible, from the patterns which the borrowings form, to tell
when words with the same root from the Latin-French continuum
came into English.  Thus 'cant', 'chant' 'chanteuse' (the
inherited cognate is 'hen', another regular correspondence);
'candle', 'chandler', 'chandelier', etc.

Indeed, the differences in patterning even allow us to recognize
borrowings from closely related languages (or even dialects):

  Native word:    shirt      rear      'em       yard      fox
  borrowing:      skirt      raise     them      garden    vixen

On the other hand, borrowings from unrelated languages often
stand out on the basis of their completely un-native patterns:

  bwana    kangaroo    okapi    quetzalcoatl    kinkajou

So Larry is quite right.  The secret to comparative linguistics
(which you seem to consider some sort of mystical psychic
process) is:  "Patterns, patterns, patterns."  The moral of the
story? -- If you aren't good at pattern recognition, don't take
up comparative linguistics.  But even more important, if you
haven't got the patience to work through all the patterns to see
if they are regular or specious, or whether they result from
inheritance or borrowing, don't even think about comparative
linguistics.

But there is nothing wrong with noticing correspondences or even
patterns before applying the comparative method.  This is known
as a heuristic device.  A heuristic is something that gives you
an indication of where to look for something (sometimes called
a "clue" for those who seem to operate without them and don't
recognize the concept).  The thing about heuristics is that they
don't tell you if what you notice is significant or not.  This is
why one then applies the comparative method -- the comparative
method will tell you whether what you have noticed is just a
flash in the pan or is solid gold.  But the fact that you can
notice patterns before applying the comparative method does not
mean that the comparative method cannot (or CANNOT if your caps
lock is stuck) tell the difference between inherited and borrowed
forms.  That's just one of those little logical hiccups that you
have from time to time that surface as non sequiturs. It may be
true sometimes that you can't tell the difference between
inheritance and borrowing, but not as a general rule.

To illustrate the points that you have raised (about being able
to see patterns without using the comparative method and about
whether it is possible to tell the difference between inherited
and borrowed material) I have a couple of small assignments for
you of the type that I often set for students.  I could just tell
you the answers as examples, but since linguistic arguments just
seem to roll off your back, it might mean more to you if you work
them out for yourself.  I really do think that it is difficult to
see how this stuff works if you don't do some of it yourself.

1)  Greek 'theos' "god" and Latin 'deus' "god" are identical in
    meaning and quite similar in form.  Are these two words in
    fact cognates?  What is the most plausible explanation for
    their similarity?  Finally, how do we know?

2)  There are two words in English that are more or less a fixed
    pair:  'salt' and 'pepper'.  Both of these words appear in a
    similar form in almost all IE languages.  One of them is
    inherited from PIE and one of them is a loanword in
    practically every IE language.  Which one is which and how do
    we know?

Now you don't have to answer these, but then I will know that you
aren't really the self-appointed watchdog of proper scientific
method as you claim, but are just kvetching about the methodology
of comparative linguistics because you don't like its
conclusions.  If you do answer them, then I will give you some
more complex ones.

Incidentally, "are they cognates?" in question 1 means "were they
once the same word (in the same language)?"  I mention this
because at one time you thought that cognates were words that had
the same meaning.  Although you seem not to be using the word
that way at this particular moment, you do have a tendency to
recidivism.

Bob Whiting
whiting at cc.helsinki.fi