Pre-Basque phonology (fwd)

[Larry Trask]

OK; I propose to reply to Roz's long two-part posting in several
instalments, as time permits.

On Wed, 15 Sep 1999, Roslyn M. Frank wrote:

> [LT]

>> First, it is linguistically unusual and unnatural to create new
>> syllables in the middle of a word.  Hence northern <behi> versus
>> southern <bei> points clearly to the conservative nature of <behi>.

> Sorry I don't understand this argument. Particularly I don't follow
> the meaning of the terms "unusual" and "unnatural". The first seems
> to refer to statistically demonstrated probabilities based on
> cross-linguistic/typological studies, while the second's meaning is
> more obscure. In making these comments I have in mind several points
> that Steve Long has made in recent mailings concerning the
> methodology used in comparative reconstructions when several
> dialects/daughter languages are used to recreate the earlier state
> of a linguistic system.

Several things.  First, the loss of syllables is *far* more frequent in
languages than the creation of syllables -- excluding, of course, the
addition of affixes -- though creation of syllables is certainly
attested.  So, given a north/south difference like <behi> / <bei>, we
naturally first suspect syllable loss, not syllable gain.

Second, the loss of the maximally weak consonant /h/ is *extremely*
frequent in languages, and readily understandable.  The history of
English provides some magnificent examples: Jim Milroy has written at
length on the steady loss of /h/ in English during the last thousand
years or so.  But we also find /h/-loss in Latin, French, Spanish,
Greek, Hawaiian, Swedish, and zillions of other languages.  In contrast,
the insertion of /h/ is a rare process.  It is certainly attested,
though the few cases I have seen involve word-initial position, not
word-medial position.  Failing evidence to the contrary, then, we
naturally look for /h/-loss, not for /h/-insertion.

Third, it is demonstrable that the western Basque dialects, Bizkaian and
Araban, formerly had /h/ but have lost it.  Occam's Razor militates
against positing *two* processes, an /h/-insertion and an /h/-loss, in
the same language when the data require only one process.  (By the way,
/h/ has recently been lost also from the coast of Lapurdi, at least in
casual speech, though it remains in formal speech, and the written
records show its former presence there.)

Fourth, we have a dictum in comparative reconstruction.  If variety A
has a contrast which is absent from related variety B, then, unless
there are very good reasons for doing something else, we reconstruct the
contrast for the common ancestor, and conclude that the contrast has
been lost in B.  Since the northern varieties of Basque have an /h/-zero
contrast, absent in the south, we therefore prefer to reconstruct the
contrast for the common ancestor, and to assume that the southern
varieties have lost it.

> [LT]

>> Second, we have minimal pairs in the aspirating dialects, like <sei>
>> `six' and <sehi> `boy, servant'.  If we took *<sei> as the ancestral
>> form in both cases, we would have no principled basis for explaining the
>> modern contrast.

> Unless <sei> is a more recent/ancient loan word, i.e., related
> to items such as <seis> in Spanish.

It is *extremely* unlikely that Basque <sei> is borrowed from Romance:
its form militates against that.  However, there are plenty of other
words available to make the same point: see below.

> In this respect I don't argue
> with your logic, only your particular example. Had it been a
> different one where the loan word status of one of the items was
> less questionable and had the sample in question consisted of a half
> dozen or so such examples of minimal pairs, its power of persuasion
> would have been greater. This is a case where a more statistically
> driven model might give us much better results. But that assumes the
> need to collect data without eliminating one or the other of the
> possibilities. For instance, one would need to collect data for all
> the southern/central dialects in order to see how the problem of
> polysemy is dealt with. In otherwords a stronger argument would be
> to show that in northern dialects there are indeed an extended set
> of minimal pairs in which the presence of /h/ (or [h] ?) is the only
> distinguishing characteristic. The only one that comes to my mind is
> that of /sei/ "six" and /sehi/ "boy, servant." To my knowledge,
> northern dialects do notcontrast /behi/ "cow" with */bei/ meaning
> something else; nor /behe/ "low, below, beneath" with */be/ meaning
> something else. That doesn't mean that there might not be other
> minimal pairs that could be examined.

We don't really need minimal pairs.  Take a look at some further
northern forms.

<gai> ~ <gei> `material', <bai> `yes', <sai> `vulture', <goi> `high
place', <gain> `top', <zain> `root', `vein', <soin> `body', <goiz>
`morning', <lai> `a certain agricultural implement', <laur> ~ <lau>
`four', <gau> `night', <oin> `foot', <min> `pain', and many others.

But:

<nahi> `desire', <ohi> `habitually', <lohi> `mud', <behi> `cow', <behin>
`once', <zahi> `bran', <bihi> `grain', <aho> `mouth', <zohi> `turf,
sod', <ahal> `ability', <ahul> `feeble', <ahur> `palm of the hand',
<mihi> `tongue', and many others.

Now, if you want to maintain that the forms without /h/ are original,
and that the forms with /h/ are innovations, then you must provide a
conditioning factor.  That is, you must explain what the rules are for
deciding when /h/ is inserted and when it is not.  Since I can see no
possible basis for doing this, I conclude -- like everybody -- that the
/h/ is original, and that the southern dialects have lost it, producing
a number of new monosyllables there.

> [LT]

>> It appears that we must reconstruct two Pre-Basque
>> forms [/sei/ and /sehi/] with differing numbers of syllables, with the
>> contrast surviving
>> in the north but lost in the south after the loss of the aspiration.

> In other words, you argue that root-stems
> which are monsyllabic (and unaspirated) in southern dialects should be
> considered bisyllabic in Pre-Basque.

Yes.  Absolutely.  All the evidence points that way.

Consider some further evidence.  Latin <anate(m)> `duck' was borrowed
into Pre-Basque as *<anate>.  With the categorical loss of intervocalic
/n/ in medieval Basque, this has become <ahate> in the north, but
variously <aate>, <agate> or <ate> in the south.  In this case, and
likewise with other loans from Latin, it is perfectly clear that the
forms with three syllables are conservative, and that the two-syllable
form <ate> is an innovation.  The southern dialects have lost syllables.

> [LT]

>> Third, aspiration survives today in the north.  It was also very
>> prominent in the west, in Bizkaia and Araba, during the Middle Ages, as
>> our written records show.  For the central dialects, there is no direct
>> attestation of any aspiration.  By far the most parsimonious scenario is
>> a Pre-Basque aspiration in all varieties, followed by early loss in the
>> center, much later loss in the west, and retention down to today in the
>> north.

> Why is this "by far the most parsimonious scenario"?  Do you mean
> that the lack of aspiration in central dialects means they are the
> least stable and hence most deviant (at least for this item)?

I wouldn't say "least stable" or "most deviant".  I would describe them
as "innovating".

> Isn't there another way of looking at data which portrays features
> found in "central dialects" (geographically speaking) as being more
> representative of "core" or "earlier" features"?  I'm not
> necessarily subscribing to this view, simply stating it.

If I may ignore certain issues in the discussion of the homeland
problem, where such ideas have indeed been put forward, then the most
prominent position among dialectologists -- introduced, I believe, by
the Italian Neolinguists -- is that central dialects tend to be the most
innovating, and peripheral dialects the most conservative -- exactly as
concluded in our case.

> [LT]

>> Fourth, we have good evidence that ancient Aquitanian was an ancestral
>> form of Basque -- and the written aspiration is pervasive in the
>> Aquitanian materials.

> Again there are a number of prior assumptions involved in that
> argument, e.g., that Aquitanian was, indeed, an ancestral form of
> Basque and not simply a(nother northern) dialect.

Since most of the surviving Aquitanian texts come from the north, it is
possible that our texts mostly preserved a northern variety of
Aquitanian which has died out, while Basque descends from a southern
variety which is sparsely recorded.  But the letter <H> is just as
prominent in the few southern inscriptions as it is in the northern
ones.  Anyway, we don't need to appeal to Aquitanian to conclude that
/h/ is a conservative feature in Basque, not an innovation: *all* the
evidence points that way.


Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk