Basque 'sei'

[Larry Trask]

Roz Frank writes:

> Four items.

> 1) Larry Trask has brought into play what appears to be compelling toponymic
> evidence in order to refute the claim that <sei> might have been copied into
> Euskera from Gascon. Yet Seguy (1951) whose research is discussed in Ruffie
> and Bernard (1976) indicates that the final element of such the place names
> was rendered variously as <-os>, <-osse>, <-ons> <-ost> and <-oz>. If one
> assumes that the original suffixing element was in all cases <-otz> "cold"
> (which is not entirely clear), there seems to be a certain ambivalence
> concerning which sibilant was chosen to represent the original Basque suffix
> in Gascon. The relevant map with the isoglosses is reproduced in Trask 1996:
> 41). It also indicates attestations of corresponding Aragonese endings in
> <ues> and <ueste>.

Yes.  If this toponymic ending is of Aquitanian/Basque origin, then the
presence of the single Romance sibilant in any given area is readily
understandable.  There is indeed reason to suspect an Aquitanian origin, given
the distribution of the ending, but Basque <hotz> 'cold' doesn't look a good
bet, since it's hard to see why this should recur in dozens of toponyms.

> 2) Larry Trask hasn't mentioned (with respect to this item) that in Euskera,
> /z/ regularly undergoes palatalization, e.g., in the case of the pronoun <zu>
> "you" which commonly becomes <xu> (palatalized) in northern dialects, and
> that Azkue and others write or represent that sound as <s> with a "tilde."
> Furthermore, palatalization of sibilants in Basque creates a situation in
> which both the letter <z> and the letter <s> can become conflated and
> represented by the same letter, e.g., as <s> although at times with a "tilde"
> (cf. the dozens of examples of this in Azkue's dictionary).

Yes; palatalization occurs in various circumstances in Basque.  But, in native
words other than expressive formations, it's always secondary.

> Among the commonplace palatalizations in northern dialects we have the case
> of <sei> (Azkue II: 247).  This permutation is so common in other dialects of
> Euskera that when I learned Basque, primarily through contacts with native
> speakers of the language from Goierri in Gipuzkoa (a southern dialect), I
> thought that it always had a palatalized sibilant.

The variant <xei> for <sei> is indeed found in parts of Bizkaia and Gipuzkoa.
But, as usual, this palatalized variant is secondary.  The textual evidence
shows clearly that <sei> is the original form, and hence the form to be
considered.

> 3) Furthermore, in reference to whether <sei> was copied at some time in the
> past from some non-Basque language, one could construct the following
> argument.  In the case of the numbers 5, 7, 8 and 9, if one were to assume
> that they were all indigenous to Euskera (that none of them are "loans" not
> even ancient ones, (cf. Miguel C.'s contribution suggesting the contrary) we
> would find that we could argue that earlier all of them ended in a consonant
> cluster that had /z/ as an element: <bortz> (*<bortzi> 5, <zazpi> (*<zatzi>
> or *<zaptzi>) 7, <zortzi> 8, and <bederatzi> 9. Such a reconstruction would
> make <sei> an anomaly in the series and, hence, suggest that it was copied
> from some non-Basque language, e.g., perhaps an IE language, which was in
> contact with it at some time in the past.

I think this is going too far.  Both <z> and <s> are frequent in Basque, with
the first being more frequent, and either may occur in any arbitrary word.  The
occurrence of numeral-names with both does not strike me as a problem to be
explained, nor does the greater frequency here of the laminal, which is more
frequent generally in the language.

> The past 2500 years would have brought it into close contact with Celtic,
> Latin/Romance, Germanic (Visogoths) speaking peoples. As a result, there
> would have been a number of opportunities for this IE item to have entered
> the language.

But no plausible IE source has been identified.  Latin and Romance, the usual
sources of loans into Basque, appear to provide no plausible sources.  I don't
know much about Celtic here, but I've never seen a proposed Celtic origin.
Germanic, like Romance, retains a final sibilant in the word for 'six', and
anyway it's not even clear that the Visigoths were still speaking Gothic when
they settled in Spain.  You have to come up with a specific proposal involving
a specific language.

Finally, nobody is suggesting a non-native origin for the Basque names of the
numbers above 'six', so a borrowing of 'six' alone would be isolated and
mysterious.

> 4) I believe that Larry Trast argued that in the case of words copied
> (early?)  from Latin and ending in <x> (sorry I can't find the exact email),
> the Basque word never drops the final sibilant complex, i.e., it is copied
> into Euskera as /tz/ (again forgive me if I"m misquoting). However, there
> would seem to be a few words that don't follow this rule, e.g., Basque
> <trebe> which I was once told comes from Latin <trebax>. Are there others?

It is hard not to see a connection between Basque <trebe> 'adroit, skilful,
expert', and also 'cunning', and the more-or-less synonymous Latin <trebax>.
But the Basque word can't possibly derive directly from <trebax>.

First, Latin loans into Basque almost invariably enter in the accusative, not
in the nominative, and the Latin accusative <trebacem> doesn't end in a
sibilant.

Second, Pre-Basque absolutely did not permit plosive-liquid clusters in any
position, and such clusters were invariably eliminated in loans from Latin.
The usual way of resolving a word-initial /tr-/ cluster was to break it up by
inserting an echo of the following vowel.  Hence *<tre-> should have yielded a
Basque *<dere->, or at best *<tere-> -- not attested.  Compare, for example,
Basque <daraturu> (and variants) 'drill', from the Latin accusative <taratrum>.

Actually, Basque <trebe> is a bit of a puzzle.  The Latin word, itself of Greek
origin, was apparently uncommon, and it appears to have left few traces in
Romance.  The unreliable Lhande notes a supposedly synonymous <treba> ~ <treva>
in unspecified varieties of Romance, but the major Romance sources at my
disposal recognize no such word.  Quite possibly the Basque word is borrowed
from an obscure Romance continuation of the Latin word, but it can't be
borrowed from Latin.

> Conclusion: things are somewhat more complicated than they might appear to be
> at first glance.

True.  But that doesn't mean we can't work out a good deal of the truth.

> Although I don't believe we are any closer to a definitive solution to the
> problem, there does seem to be evidence that could be mustered for several
> different scenarios. The four points above are meant only to provide a bit of
> additional data for the analysis, not necessarily to suppport one position or
> another.  I leave it to the experts to figure out whether any of the above
> information helps one cause over the other.

Well, my own view is that there exists absolutely no reason for seeing Basque
<sei> as anything other than native and ancient -- that is, as going back at
least to Pre-Basque.  The word is in no way unusual or problematic within
Basque.  Whether, as Miguel C V suggests, the Basque word itself goes back
still further to a very ancient pan-Afro-Eurasianism, I can't say.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk