"Related" debate unproductive?

[Larry Trask]

Lloyd Anderson writes:

[snip passages with which I mostly agree]

>  By considering primarily extreme cases where we probably all
>  agree, we make no progress towards handing the difficult ones.

Indeed.  But may I suggest that anyone wanting to discuss difficult cases
should identify some *particular* cases which he sees as "difficult"?

Most of the discussion so far has either centered on completely hypothetical
scenarios or picked on cases which are anything but problematic, such as
Germanic and Indo-European.  Afrikaans and Michif have now been named as
possibly difficult cases.  Would anybody like to name some others?
Or discuss these two further?

>  I would note first that it is probably impossible in practice to
>  avoid a mixture of ordinary language with technical usage.
>  When Whiting says today in response to someone's

>  >>Genetically (in your terms), English is equally related to both
>  >>French and Italian.

>  as follows:

>  >They're not my terms; I didn't invent them -- they are standard
>  >in historical linguistics textbooks.  But I agree with the
>  >premise -- except that I wouldn't say "equally related"; I would
>  >say "related at the same level."

>  I of course agree.

And so do I.

>  Perhaps our common reluctance to use the phrase
>  "equally related" here is that it has a portion of its ordinary-language
>  meaning,

Well, it shouldn't.  In linguistics, 'related' is a technical term, and
in linguistic work it has, or should have, only its technical sense.  We
linguists must put aside all knowledge of everyday senses of the term
when we are working, or we will only get hopelessly confused.

I can cite a good parallel here.  I used to be a physics teacher once,
and, in physics, 'work' is a technical term, with a sense quite different
from its everyday sense.  And I can report that it is surprisingly difficult
for beginning students to forget about the everyday sense when studying
physics: they keep wanting to see work done where no work is done in the
physical sense.  And, as a result, they become confused and get things wrong.

Once upon a time, related languages were sometimes called 'cognate languages'.
This usage is now dead, I think, and we restrict 'cognate' to individual words
and morphemes.  Maybe that's unfortunate, if 'related' as a technical term is
going to cause so much confusion.

>  and we know clearly English *is* especially closely related
>  to French, as *all* linguists recognize.

No.  I disagree flatly.  I believe I have *never* seen a linguist declare
that English is more closely related to French than to Italian, and I would
not expect to see any such thing, because it's not true.

>  (English loans from
>  Hindi or Chinese or Afrikaans or whatever, are also not really
>  to the point, I think, because such extreme cases were not mentioned
>  by those wishing to question an overly narrow sense of "relationship".
>  So bringing *them* up is at least not to the point of what I
>  believe many of us are concerned with, such would also be red herrings.)

Er -- "extreme" cases?  In what sense are they "extreme"?  English has
borrowed a few words from Chinese and from Afrikaans, and rather more from
Hindi.  What's extreme about that?

Anyway, I reiterate that our established sense of 'relationship' is in no
sense "overly narrow".  It is simply the way the term is used in linguistics,
and that's the end of the matter.

Mathematicians recognize classes of numbers they call 'rational' numbers
and 'real' numbers.  But these numbers are no more "rational", and no more
"real", than any other numbers, in the everyday senses of these words --
and any outsider who tried to claim otherwise would be looked at strangely
by mathematicians.  Why should linguistics be different?

>  It is perhaps my personal interest and bias,
>  but the problem that remains as a subtle and sophisticated one,
>  and which has clearly *not* been resolved by previous scholarship,
>  is the handling of trees vs. dialect areas,

Maybe not "resolved", but these issues are far from unfamiliar in
linguistics, and they have been disussed for generations.

>  and the implications of
>  the following paragraph, which I did not write,
>  and which was referred to by Whiting today:

>  >(b) The idea that there must be a single language progenitor of
>  >daughter languages is widely disputed. Some people accept the
>  >idea that a collection of interrelated languages might never have
>  >had a single ancestor, but as far back as you care to go were
>  >simply a collection of inter-related languages.  The
>  >language/dialect issue comes up here.  We talk of IE "dialects"
>  >within PIE, but this is simply terminology.  The point is that
>  >there is no need whatever for there to have been a single unified
>  >PIE language.

We have already discussed this.  The suggestions are fanciful, and the final
sentence is false.

>  The example Whiting mentions of the Altaic "family"
>  (or perhaps not a family, if long contact and mutual influences
>  and massive borrowings were involved)
>  is closer to a propos to the view expressed
>  by the author of the above paragraph, it seems to me.

No; this is a very different issue.

All parties to the Altaic debate agree that there are just two
possibilities: either Turkic, Mongolic and Tungusic are genuinely related
within a single larger Altaic family, or they are not, and the obvious
connections among them are merely the result of prolonged contact among
unrelated families.  The parties differ as to which view they believe is
the true one, but nobody I know of is arguing for any third view.

>  The possibility that PIE was some close complex of languages
>  which exchanged even morphology, but which retained traces
>  of various distinct substrates or whatever one wishes to discuss,
>  is a real possibility.

It might be a hypothetical possibility, a la Dixon, but any such idea is
heavily falsified by the data.  IE is a simply terrible choice of example to
illustrate this hypothetical scenario -- one of the worst examples possible.

If anybody wants to pursue this scenario, how about showing us a *plausible*
candidate?

>  Not necessarily to a great degree,
>  not necessarily as much as Altaic, but to some degree,
>  it is quite possible and an entirely reasonable hypothesis
>  consistent with *traditional* views of historical linguistics.
>  Merely one that is mostly not discussed, rightly or wrongly.

It *has* been discussed.  This idea that the IE family results from
extensive diffusion across the boundaries of unrelated languages, and
that PIE never existed, was put forward by Uhlenbeck, by Trubetzkoy, and
by Tovar (at least).  But this idea has been categorically rejected by
specialists, because *it is in flagrant conflict with the evidence*.
How many times do we have to say this?  The evidence for the reality of PIE
is *overwhelming*, and it is *not* consistent with any kind of "language-
mixing" scenario.

>  This possibility, which may exist for many proto-languages,
>  *does* have practical, as opposed to purely terminological,
>  implications.

In order to persuade anybody of this, you must first put forward some
plausible candidates, and show how the family-tree model fails badly to
account for the evidence.  Until this is done, it is a waste of time to
invent hypothetical but unsubstantiated scenarios and merely mutter darkly
that these might contain some reality.

It is perfectly conceivable that some human languages were introduced by
alien visitors from another planet.  In fact, just such a proposal was once
made for the origin of Basque (naturally).  But no linguist is prepared to
spend time in considering such a scenario.  And why should we?  Nobody has
made a good case that any particular language must have such an origin, and
so what is the point of discussing it?

>  To recognize that this is a possible situation
>  for a proto-language,
>  we must handle vocabulary and morphological
>  distributions across *portions* of the dialect network of any
>  proto-language in *at least* the frameworks of the following

Sorry, but this is putting the cart before the horse.

We cannot identify any possible cases of variation within a proto-language
before we have first identified the proto-language itself.

>  1) simple family tree, innovations on one branch,
>       replacement on one branch, etc. branch then dividing.

Familiar.

>  2) wave spread of items across a part of a dialect network,
>      which may have no relation to the family-tree structure

Familiar.

>  3) persistence as areal dialect-net isoglosses of what were
>       substrate inheritances in only part of the territory of
>       an eventual proto-language, the substrate inheritances
>       in another part of that territory being from a different
>       language or languages.  When substrates are strong,
>       and morphology can spread, the difference between
>       the various kinds of "inheritance" can become quite blurred.

There are not different kinds of inheritance.  Modern English has inherited
from Middle English a number of words of Norman French origin.  But Middle
English did not "inherit" these words at all: it *borrowed* them from Norman
French.  This is just the way the terms are used in linguistics.

Anyway, I can see absolutely no point in invoking hypothetical "substrates".
If we have no hard evidence for the reality, and the nature, of a particular
substrate in a particular case, then there is no point in raising the issue.

>  4)  proto-languages need not (not even by the narrow
>       definition of "genetically related") be completely uniform,
>       they need not be indivisible points with no internal dimensions
>       Living languages do not fit such a simple model,
>       so what business do we have insisting that dead languages did?
>       That would make them theoretical constructs, useful primarily
>       for making it easier for us to think about them,
>       so artificially simplified.

But nobody has ever disagreed with this.  Once again, *nobody* maintains
that ancient languages were devoid of variation.  It is merely that our
methods, on the basis of the evidence available, do not allow us to
reconstruct the full range of that variation, or even (usually) very much
of it at all.  And no more powerful methods exist.

>       In limiting special cases, sure, when a single family or village
>       migrated, and became the nucleus of a new language family.
>       But those are limiting special cases, they do not define a narrow
>       total range of possibilities which historical and comparative
>       linguistics should restrict itself intellectually to being able
>       to deal with.

By what right do you assume that the origin of a language family in a single
rather homogeneous speech variety constitutes a "limiting special case"?
How do you explain the great success of linguists in reconstructing one
proto-language after another?  PIE, Proto-Uralic, Proto-Algonquian,
Proto-Iroquoian, Proto-Oto-Manguean, Proto-Arawak, Proto-Dravidian,
Proto-Austronesian...

Where are the dozens or hundreds of cases in which reconstruction of a
proto-language has proved impossible?

>  5)  Any proto-language need not be pure,

There is no such thing as a "pure" language, and such a notion has no place
in linguistic work.

>  it may share substrate
>       inheritances from quite a number of substrates,
>       substrates which either are known as separate language
>       families, or which have become extinct as independent
>       languages, or in an intermediate situation, which may appear
>       as substrates also for some other language or language family.
>       It may be possible to reconstruct part of the vocabulary even
>       of a language or family which survives *only* as substrates
>       to two or more other languages or language families,
>       if we can determine that the substrates within each of those
>       latter are indeed substrates, rather than being later innovations in
>       some area which crosses language boundaries,
>       or later loans from part of the area of one language or family
>       to part of the area of another language or family.

Substrates again.  Sigh.

OK.  One day when I have a little spare time, I'll post the entry on
substrates from my new dictionary.

>  So let us sharpen our existing tools, develop new tools,
>  avoid oversimplifications, *and* recognize the inherited
>  wisdom of comparative and historical-reconstructive linguistics.
>  There is absolutely no reason we should need to choose
>  between these as if they were mutually exclusive alternatives.

Between what?  Between comparative linguistics and historical-reconstructive
linguistics?  What's the difference?

>  And let us treat the contributions of others by always trying
>  to find the *most* reasonable view of them, or the part of them
>  which we believe we can make the most productive contributions to,
>  rather than spending most of our words trying to defeat them.

Well, again, let's stop introducing hypothetical but unsubstantiated
scenarios, and let's stop illustrating these with dreadful examples
like IE.  Let's see some *genuinely* problematic cases.

>  There is absolutely no reason to throw out *any* of the tools
>  of comparative-historical linguistics.  There *is* reason to
>  sharpen those tools and to add more tools and to add
>  less simplistic formats for recording results of using those tools.

Sounds good, but what exactly is being proposed?

And what on earth are you describing as "simplistic"?

Grumble, grumble, grumble.  Guess I must be in a bad mood today. ;-)

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk