Can Parent and Daughter co-exist?

[Larry Trask]

On Fri, 10 Sep 1999 X99Lynx at aol.com wrote:

> In a message dated 9/8/99 4:06:14 AM, kurisuto at unagi.cis.upenn.edu wrote:

> <<Several on the list have brought up cases such as Latin, Sanskrit,
> and Arabic as counterexamples to that last assertion.  Part of the
> problem is in defining what we mean by a "living language". >>

In linguistics, we usually define `living language' as `language
currently existing as a mother tongue'.  Even this doesn't get rid of
all the problems, but it does exclude things like Latin, Sanskrit and
classical Arabic.  In this sense, a language is not living merely
because it is in use.

Of course, this definition of `living language' may not be suitable for
all purposes, but I think it *is* appropriate for some purposes.

> When Larry Trask made the assertion that a parent cannot co-exist with a
> daughter, I tried to avoid it by renaming the narrow PIE language/proto
> languages "Celtic1....Celtic6."   To avoid the terminology issue.

But others have already pointed out the great difficulty with this.

> Somehow this has turned into a dialogue about the eternal changeability of
> language.  Which is irrelevant.

No; I don't think so.  The fact that living languages are constantly
changing is central here, and trying to ignore it leads at once into
absurdities.

> Here's why:

> How long did Old Norse stay Old Norse?

This is purely a matter of definition and convenience, not a matter of
linguistic reality.

> How long has Modern English stayed Modern English?

Same thing.  Labels like `Old English', `Middle English', `Early Modern
English' and `Modern English' are arbitrary conveniences only.  Nothing
happened in 1700 to turn Early Modern English into Modern English.

> What's the longest a "language" can stay the same language?

One day?  ;-)

Since a language is always changing, it must be different today from the
way it was yesterday.  Not very different, of course, but different.

> Can it stay the same "language" for a hundred years?  Two hundred years?

No, it can't stay literally the same, but it can change sufficiently
little in that time that we have no reason to refrain from giving it the
same name.

In fact, it doesn't even have to change so little.  We are happy to
apply the label `Greek' to the language of Pericles and to the language
of an Athenian taxi-driver, but it doesn't follow that Greek has
"remained the same language".

> Does anyone have an answer?  Can a "language" stay the same
> "language" for 400 years?  Let's assume that's the outside margin.

There can be no answer until you explain what you mean by "the same
language".  In the last 400 years, Icelandic has changed rather little,
English has changed much more, and some other languages may have changed
dramatically.

> Now, how long does it take for a dialect to develop in that parent
> and turn into a different language?

There can be no principled answer, since the distinction between a
`dialect' and a `language' is not a principled one, but rather a
linguistically arbitrary one based largely on social, political and
educational factors.

> Anybody who has used "the perpetual changeability of language" buzz
> phrase can put down their hands.  Your answer is obvious - in no
> time flat. Certainly less than 400 years.

It isn't a "buzz phrase": it's a fundamental truth.  You can't wave away
the central fact of ceaseless language change as though it were of no
relevance to the discussion.

> If a language can definitionally stay the same language for 400
> years, and if a daughter can develop in last than 400 years, then
> its obvious.  A daughter can be in existence while the parent is
> alive and well.

No; this doesn't follow.  Let's take a real case.

In the 17th century, Dutch was introduced into South Africa by settlers.
Since then, the Dutch spoken in South Africa has, of course, steadily
diverged from European Dutch -- or, to put it another way, European
Dutch has steadily diverged from African Dutch.

Now, until the 1920s (I think it was), African Dutch, commonly called
`Cape Dutch', was generally regarded as a dialect of Dutch.  But then
perceptions changed: Cape Dutch was officially named `Afrikaans', and it
began to be generally regarded as a distinct language.  Today Afrikaans
has its own distinct standard form, and everybody regards it as a
separate language.  But Dutch and Afrikaans are still largely mutually
intelligible, though each sounds very strange to speakers of the other,
and there are some lexical differences which impede communication.

But the new autonomous status of Afrikaans results from a political
decision, and not from any linguistic events.  Nothing happened in 1925
to convert Afrikaans from a Dutch dialect to a separate language, except
for a political decision that this should become so.

It is probably safe to say that modern Afrikaans is rather more
different from 17th-century Dutch than is modern standard Dutch.
But neither modern variety is identical to 17th-century Dutch.

Moreover, 17th-century Dutch was itself not uniform.  Today the
varieties of Dutch spoken in West Flanders and in northern France are
not at all mutually comprehensible with standard Dutch, and I have no
reason to suppose that things were any different in the 17th century.
For political reasons, though, all these are regarded as varieties of a
single language, Dutch.  But it need not be like that, and, for a while,
it wasn't.  For some time the Dutch-speakers in Belgium took the view
that they spoke something called `Flemish', a different language from
Dutch.  But, some years ago, they abandoned this idea, and they now
agree that they speak Dutch.

> Here's another example: When was the last exact date Latin was a
> "living" language, a "natural" language, a "first language?  Pick
> any date.  January 17, 601 AD.  Let's say the last native speaker
> died that day.

Impossible.  There has never been a "last native speaker" of Latin.
The language has millions of native speakers today.  But, because of
accumulated changes, the modern varieties are so different from the
language of the Romans, and from one another, that we no longer find it
convenient to call them `Latin'.  Of course, if we wanted to, we could
speak of `Paris Latin', `Barcelona Latin', and so on, but no one has
seen any point in this.

> What language was everyone else speaking in the meantime?  Or did
> they all switch to Romance daughter languages the next day?

This is a non-question.  There was no date on which Latin disappeared
and was replaced by Romance.  The linguistic division between `spoken
Latin' and `Romance' is a purely arbitrary one, and any date assigned to
it is no more than a matter of taxonomic convenience.

> (You can't really call those daughters dialects at that point,
> because if a dialect of Latin survived, then Latin survived.)

But "dialects of Latin" survive today!  They just aren't very similar to
the Latin you would have heard in the streets of Rome 2000 years ago.

> Common sense says that Latin as a native language would have had to
> co-exist with vernacular languages that lived on after it died.

No, "common sense" says nothing of the sort.  This view requires that
some varieties of Latin would have had to develop into entirely distinct
languages while other varieties remained entirely unchanged -- an
impossibility.

In fact, written Latin continued for many centuries to coexist with a
large number of Romance varieties, but that's a separate point.

> Or those vernacular languages would have had to come out of nowhere
> the day Latin died.

No.  Such bizarre conclusions arise from the use of an inappropriate
model, one involving an impossible reification of the notion `language'.

> Common sense says parent and daughter can co-exist.

Nope.

> Once again, I believe Larry Trask was referring to a methodological
> assumption.  Though I don't know the basis of it.

If a language splits into two or more regional varieties which become so
distinct that we must count them as distinct languages, then both are
different from their common ancestor.  One may be more conservative than
the other, but both are different, and there is no principled basis for
regarding just one of them as identical to that ancestor.  In the
Dutch/Afrikaans case, we find it convenient to apply the label `Dutch'
both to the 17th-century ancestor and to the modern language of the
Netherlands and Belgium, while we no longer find it convenient to apply
the same label to Afrikaans -- though we did until recently, and we
could do so again if the political, social and educational facts
changed.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk