the Wheel and Dating PIE

[Larry Trask]

[ moderator re-formatted ]

Ed Selleslagh writes:

I'm afraid I'm going to have to disagree with most of what Ed has written here.

> After my first response on Dec. 18, 1999, I have been thinking about the PIE
> root *kwekwlo- (and *rotH-) in a wider context, in particular about its
> possible presence in Basque.  It is almost certain that the Basques, who had
> been living in virtual isolation throughout the last Ice Age,

Maybe a little rash?  The Basque language was certainly in place before the
Romans arrived in the first century BC, and very probably before the Celts
arrived in the first millennium BC.  But anything before that is sheer
speculation: there is no evidence.

> learned about the
> wheel from IE peoples or their otherwise spreading culture, since the terrain
> of the Pyrenean peoples offered little incentive for inventing the wheel, as
> opposed to steppe peoples who had to travel long distances over essentially
> flat terrain.

Maybe, but what evidence is there?  There is no evidence for IE speech near the
Basque Country before the first millennium BC.  But I confess I simply don't
know how early the wheel is attested there.

> In Basque, there is a root *bil that has a meaning of 'round'.

Correct. Though our reconstructed *<bil> is nowhere recorded as an independent
word, its presence in a number of transparent formations makes its former
reality certain.

> It appears in compounds like 'ibili' ('walk', originally: 'go around')

No; I can't agree.

First of all, the verb <ibili> does not mean 'walk', though it is often so
glossed, for want of a better equivalent.  It is a verb of undirected motion,
and the best English gloss I can offer is 'be in motion'.  And no such sense as
'go around' is attested or reconstructible.

Perhaps I should explain that Basque, like its Romance neighbors, is a language
in which verbs of motion incorporate path but not manner.  English is just the
opposite.  English is full of motion verbs expressing manner, such as 'walk',
'run', 'sidle', 'trot', 'swim', 'fly', 'drive', 'ride', 'limp', 'scramble',
'toddle', 'stagger', 'sail', 'float', 'crawl' and 'creep'.  But path is not
incorporated into native verbs, and particles like 'up', 'down', 'in', 'out',
'across' and 'through' must be added to express this: 'ride out', 'go in',
'walk across', 'run down', and so on.  (Formal registers of English of course
have the other type of verb, like 'enter', 'exit', 'ascend', 'descend' and
'traverse', but these are borrowed from French or Latin, in which they are
normal.)

But Basque is full of verbs like <igan> 'go up', <jaitsi> 'go down', <sartu>
'go in', <irten> 'go out', and the like.  And manner must be expressed by
adding an adverbial, such as <oinez> 'on foot', <zaldika> 'on horseback',
<igeri> 'swimming', <ttapa-ttapa> 'toddling', <kotxez> 'by car', <trenez> 'by
train', <hegaz> 'on wing', <lauhankaka> 'on all fours', and so on: there are
lots of these.

Unlike its Romance neighbors, Basque has a dedicated verb of undirected motion,
<ibili>, used when no goal or source is stated or implied.  So, for example,
<oinez ibili> is 'walk', 'be walking' (with no goal), and <igeri ibili> is
'swim', 'be swimming' (with no goal).

Furthermore, the verb <ibili> is definitely not a compound, nor does it contain
the element *<bil> 'round'.  It has the normal structure for a native Basque
verb.

The citation form of a Basque verb is its perfective participle, and a native
verb has a participle of this form: *<e-Root-i>.  Here <-i> is the participial
suffix, possibly identical with the ancient adjective-forming suffix <-i>, as
in <gazi> 'salty', from <gatz> 'salt'.

Removal of this suffix yields the stem of the verb, which was formerly a free
form and still is in the east, where it is called the 'radical'.

The prefix *<e-> is of unknown function, but it occurs in all non-finite forms
of native verbs.  In my 1990 paper, I argued that it was probably a nominalizer
which created verbal nouns.

The original <ebili> (well attested) becomes modern <ibili> by a
well-understood phonological change, a vowel-height assimilation:

	e --> i / #Co __ C(C) V[+high] C V

Now, it seems clear that, in early Basque, verbal roots were sharply
distinguished from all other roots.  A verbal root could only be verbal, and
never nominal or adjectival, unless converted to a nominal or adjectival stem
by the addition of a category-changing affix.  There is no known case of a
nominal or adjectival root appearing inside a prefixing verb (a verb taking the
prefix *<e->).  Hence <ibili> surely cannot contain adjectival *<bil>, and the
resemblance in form is a coincidence.

> and 'biribil' (a reduplicated form meaning 'round').

Agreed.

> This could be a phonological adaptation of a derivation of *kwekwlo-,
> especially its Germanic forms (cf. Eng. wheel, Du.  wiel), or -just maybe-
> one of its oldest Celtic forms (But: mod. Welsh: 'olwyn' = wheel, apparently
> with metathesis), because Basque doesn't have /w/, and the closest Basque
> phoneme is /b/ (a tendency that is still alive in Castilian:  Washington =
> Bassinton, at times even on TV! And a WC is often called 'un bater').

I am not sure what the word 'this' at the beginning is meant to refer to.  But
I cannot see how *<bil> can plausibly be derived from the IE word, and still
less its reduplication <biribil>, which itself can be accounted for within
Basque.

> The Basque word 'ibili' needs some further explanation. It is obviously a
> compound, but of what?

No; I can't agree.  It is not a compound at all.  It consists of one root and
two affixes.

> My hypothesis is as follows: *i-b(i)-bil-i, with haplology.  The initial and
> final i's are common features of Basque verbs. -b(i)- would be a root that
> means 'walk, run' (cf. IE wad-), and -bil- '(a)round', of course.

Those vowels are more than "common features": they are affixes, one of which is
fully understood, the other of which is only partly understood.  And this
proposal strikes me as very fanciful: it is supported by no evidence at all,
and it conflicts with the observation that a native Basque verb has the form
*<e-Root-i>, where the root must be strictly verbal.  Moreover, native verbal
roots are usually monosyllabic and very commonly of the form -CVC-.

> The hypothetical root *b(i) is the subject of a long running historical
> controversy. It is supposedly found in words like:

> bide ('way, road'), probably a compound of *bi- and the common suffix (of
> 'extent') -te, meaning something like 'the physical area where one walks'.

No such root as the suggested *<b(i)> can be defended, in my view.  Moreover,
<-te> is not a suffix of extent, but rather a temporal suffix indicating
'duration'.  Examples: <gerrate> 'wartime' (<gerra> 'war'), <gosete> 'famine'
(<gose> 'hunger'), <eurite> 'rainy season' (<euri> 'rain'), <negute>
'wintertime' (<negu> 'winter'), <agorte> 'drought' (<agor> 'dry'), and many
others.

Finally, an original *<bite> should *not* develop into <bide>.  There is no
parallel for such a development.

> ibi, more commonly ubi ('ford, a place where one can wade through the
> water'), according to e.g. Michelena, u-bide ('water-way': u-, uh- or ug- is
> the form of ur, 'water', in compounds),

Correct.  In Basque word-formation, a first element loses its final /r/.  Hence
<ur> 'water' + <bide> 'way' yields *<ubide> regularly, followed by reduction to
<ubi> (such reductions in final elements are sporadic but frequent), and then
by vowel assimilation to <ibi>.  In fact, the form <ibide>, with assimilation
but no reduction, is recorded in 1630, in the writer Etxeberri of Ziburu, so
the etymology is directly confirmed.

> and to Bertoldi, a compound of ibi-bide, with haplology.

Far less likely, I'd say, and in fact unnecessary: why try to derive <ibi> from
a hypothetical *<ibi-bide>?  What does this achieve?  Anyway, the attested form
<ibide> confirms the proposal *<ur-bide>.

> The existence of a root *(i)b(i) has been posited since Hubschmidt,
> because of its wide diffusion, not only in the Basque areas, but also, and
> mainly in the Iberian zones.

But Hubschmid was not a Vasconist, and his ideas about Basque have been widely
dismissed by specialists as fanciful and unsupported.  His problem was that he
wanted to find Basque sources for just about every problematic word and name in
the Romance-speaking area -- though, to be fair, his conclusions are somewhat
more sober than I'm making them appear here.

> Both explanations are not necessarily contradictory: u-ibi-bide > ubi or ibi.
> A related problem is that of the Basque word for 'bridge': zubi, a compound
> of zur ('wood, wooden', possibly a remote relative of a.Grk. xylon) and ubi
> or bide, thus meaning either 'wooden road' or 'wooden ford'.

Michelena proposed <zur> 'wood' + <bide> 'way', and I endorse this, even though
this time we are not so fortunate as to find *<zubide> recorded.

> ibai ('river') and ibar ('river valley bottom, Sp. vega, Du. waard, polder'),
> but this is very controversial.

Not all of it.  The word <ibai> 'river' is pretty clearly a derivative of
<ibar>.  This may look funny, but recall that a final /r/ is regularly lost in
the first element in word-formation.  Compare cases like <izter> 'thigh',
<iztei> 'groin', the second being <izter> plus a suffix of the approximate form
*<-ei> or *<-i> (or quite possibly *<-egi>, in fact, but that's another story).

> It would be explained via 'running [water]'.

Sorry, but I can't follow this.  How on earth can <ibai> be assigned such a
meaning?

> Ibai is often thought of as the origin of Sp. vega (via ibai-ka),

This etymology is popular in some Romanist circles, but I stress that it
remains speculative at best.  See Corominas and Pascual for some discussion.

> while ibar is usually related to Iberia, the Iberians and the river Ebro,
> etc.

Well, I query that "usually".  Modern <Ebro> is clearly from Latin <Iberus>,
but the origin of this name is unknown, and Basque is an implausible place to
look.  The Romans used the name <Iberia> before they encountered the Basques, I
believe.

> Finally, I would not exclude the possibility of *(i)b(i) being related to IE
> wad- (ua-dh-) (Eng. wade, Du. waden, Lat. vade:re), especially via the forms
> ibai and/or ibar, if the initial i is a prefix, as has often been thought.

But there *is no evidence* for this fanciful *<(i)b(i)>.  It looks to me, I'm
afraid, like nothing more than an excuse for dragging in everything under the
sun containing /ib/ or /bi/ or even just /b/.

As for the idea that initial /i/ is a prefix, this goes back to Schuchardt, but
it has proved entirely fanciful, and it is accepted today by no specialist
known to me.

> All this is, of course, rather speculative, even though based upon a body of
> pretty well accepted ideas.

Not all of the ideas put forward above can reasonably be described as "pretty
well accepted".

> Anyway, it looks like the hard core of very ancient and definitely Basque
> words is still shrinking after words like (h)artz (bear, Gr.arktos), gizon
> (man, PIE*ghdonios) and maybe buru (head, Sl. golova) etc.  have been exposed
> as of IE origin - or was it simply a very ancient common substrate origin?

This is far too fanciful for me, and I object to that word "exposed".

Some kind of IE origin for <hartz> 'bear' is considered plausible by many
specialists, but no good IE source is known, and the proposal remains
speculative.

As for the other two, these strike me as beyond belief.  Sorry to be such an
old grouch, but, with comparisons like these, we can derive anything from
anything.

> If you don't believe Basque has any relatives, not even extinct ones, forget
> what I said.

My own view is that no one has ever made an even vaguely plausible case for
linking Basque genetically to any other language at all, living or dead, apart
from its own ancestral form Aquitanian, and that there now remain so few stones
unturned that it is extremely unlikely that any link will ever be found.

> If you do,

Er...what?  If you *do* believe that Basque has relatives?  How can any
reasonable person believe this, when no link has ever been demonstrated?

> I hope it will stimulate you to look into this type of
> problems. Something interesting might come out of it, both for Basque and for
> P...PIE.

Well, with respect, I think we need a lot more than very vague resemblances and
entirely fanciful etymologies.  We need hard evidence.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk