DLW's Renewed Absence

[Larry Trask]

--On Monday, May 7, 2001 6:54 pm -0500 "David L. White" <dlwhite at texas.net>
wrote:

>         With regard to "mixed languages", I am of course aware of the
> supposed examples in Thomason and Kaufmann.  I must however object to
> their assertion that, in influence between languages, "anything goes".

Hmmm.  May I know where in their book T and K assert that "anything goes"?
My reading of the book reveals only the more cautious claim that we cannot
know what happens in language contact until we look.

[snip]

>         Taking all these things together, plus the fact that finite verbs
> are ordinarily "higher in the tree" than NPs,

OK; I'm afraid I don't understand this.

I don't think it's *generally* true that syntactic theorists put verbs
higher in trees than NPs.  Verb-at-the-top is more a feature of
dependency-based approaches than of constituency-based approaches.  But
some contemporary dependency theorists put verbs and all argument NPs at
the same level in their trees.  Non-Chomskyan constituency theorists
typically put subject NPs higher than verbs.  Chomskyans change their
analysis regularly, but they typically put abstract elements highest in
their trees, not verbs.

> I suggest that the genetic
> descent of a language can always (theoretically) be traced through finite
> verbal morphology, where this exists.

This strikes me as a highly arbitrary proposal, though perhaps an
interesting one.  But it does have the immediate consequence that a
language with no verbal morphology has no ancestor -- unless the intention
is to supplement this proposal with one or more unstated back-up proposals.

> This will never be mixed,

Well, a bold claim.  I confess I can't falsify it off the top of my head.
But I wonder what a survey of, say, native American languages might turn up.

> and its
> affixes will always be found closer to lexical verbs than are foreign
> affixes, if the examples I am aware of are a reliable guide.  Using this
> standard, Mednyj Aleut is Russian, Michif is Cree, and Ma'a is Bantu.
> Each is a rather bizarre and severely influenced version of its putative
> genetic ancestor, but technically there is little reason to think that
> what Thomason and Kaufmann call "normal transmission" of finite verbal
> morphology has in fact been interrupted, for in each case it is there,
> unmixed and un-intruded upon. To some extent it is a matter of what we
> call things rather than what they are,

OK.  Very interesting.  Let me draw attention to two examples.

First, Anglo-Romani.  According to T and K, this consists of wholly Romani
lexis plus wholly English grammar.  According to David White's proposal,
then, Anglo-Romani is English, since it has English verbal morphology.  Is
this a satisfactory conclusion so far?

Now, consider how Anglo-Romani came into existence.  I assume it didn't
spring into being overnight, but must have evolved gradually.  So there are
broadly two possibilities.

First, Anglo-Romani started off as plain English, but was gradually
relexified from Romani until no English lexis was left.  According to
David's proposal, this *must* be what happened, since the alternative below
seems to be impossible.

Second, Anglo-Romani started off as Romani, but it gradually borrowed more
and more grammar from English, until there was no Romani grammar left.
Under David's proposal, this appears to be impossible, since the language,
in this scenario, has shifted from having Romani verbal morphology to
having English verbal morphology -- seemingly in conflict with the proposal
on the table, which sees verbal morphology as inviolate.  It also, of
course, has the peculiar consequence that Anglo-Romani has changed from
being Romani to being English -- an outcome surely more bizarre than
anything contemplated in T and K.

Assuming overnight creation can be excluded, then, David's scenario forces
us to conclude that the first interpretation *must* be right.  Well, this
is an empirical question.

My second case is the Austronesian language Takia.  As it is commonly
described, Takia has borrowed the *entire* grammatical system from the
Papuan language Waskia, so that every Takia sentence is now a
morpheme-by-morpheme calque of the corresponding Waskia sentence -- while
at the same time Takia has borrowed no morphemes at all from Waskia.

So, in Takia, the *patterns* of the verbal morphology are Waskia, while the
*morphemes* are Takia.  In David's scenario, then, which is decisive?  Is
Takia Austronesian, because it exhibits only Austronesian morphemes?  Or is
it Papuan, because it exhibits only Papuan morphological patterns?

A pretty little puzzle, don't you think?

> but in any event I thought it
> worthwhile to point out that the somewhat wild claims of Thomason and
> Kaufmann are arguably exaggerated, as there are restrictions 1) that they
> fail to note, and 2) that can at least possibly be used to determine
> genetic descent, considerably reducing the incidence of "linguo-genesis"
> and/or "mixed languages", perhaps to zero.

Again, I don't see T and K as making any wild claims.  They seem to me to
be doing no more than quoting Hamlet: "There are more things in heaven and
earth, Horatio, than are dreamt of in your philosophy."

> Be all this as it may, I am
> sure Dr. Trask would not assert that any of the extreme developments
> adduced by Thomason and Kaufmann have anything to do with Proto-Celtic,
> so we are at least in agreement on that.

Indeed we are.

[snip]


Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk

Tel: (01273)-678693 (from UK); +44-1273-678693 (from abroad)
Fax: (01273)-671320 (from UK); +44-1273-671320 (from abroad)