Basque butterflies (and phonemes) again

Larry Trask larryt at cogs.susx.ac.uk
Mon Jan 24 17:12:32 UTC 2000


[ moderator re-formatted ]

Roz Frank writes:

> [LT]
>> And, of course, native Basque lexical items with the slightest
>> claim to antiquity never begin with /p/ (the ancient Basques apparently
>> couldn't pronounce initial /p-/)

> [RF]
> Just for the record, I'd mention that this is Larry Trask's position, one
> that he has discussed on other occasions. However it is not the only
> position. Indeed, for a rather different rendition of phonological events I
> would suggest taking a look at Jose Ignacio Hualde's article "Pre-Basque
> Plosives" forthcoming in a volume edited by Jon Franco, Alazne Landa and
> Juan Martin, _Grammatical Analyses in Basque and Romance Linguistics_ John
> Benjamins, Vol. 187 of the series Current Issue in Linguistic Theory. Since
> I understand that Larry read an earlier version of that article, I'm
> certain that he's familiar with the evidence. Given the complexity of the
> issue, I would only state that based on my reading of Hualde's data, the
> fact a word appears in modern Basque with an initial /p/ would not in and
> of itself argue for or against its "antiquity".  And by saying this I don't
> mean to imply that I'm taking any particular position on the antiquity of
> the examples found in the current discussion.

OK.  First, let me stress that I have the very greatest respect for Hualde's
work.
In my view, Hualde is the finest descriptive linguist working on Basque today.
And it is not often that I disagree with him, but on this issue I certainly do.

The standard reconstruction of the phonology of the Pre-Basque of some 2000
years ago is that by Michelena.  Michelena reconstructs two contrasting sets of
plosives: the "fortis" */p t k/, of which he says */p/ was rare and may
possibly not have existed at all, and the "lenis" */b d g/.  These symbols are
chosen to reflect the most usual reflexes in the modern language, and they
should not be regarded as necessarily representing phonetic reality in
Pre-Basque.

These two sets, M concludes, contrasted only word-medially and mostly only
intervocalically.  Word-initially, only */b d g/ could occur, and moreover
initial */d/ never occurred in native words, though it was used freely to
render both Latin /d-/ and Latin /t-/ in early borrowings.

This asymmetric distribution, or better neutralization, is not confined to the
plosives.  On the contrary, M reconstructs just such a neutralization for *all*
his proto-consonants, all of which come in fortis-lenis pairs: free contrast
between vowels, only lenes initially (and not all of them), and only fortes
finally (and not all of them).  This reconstruction is powerfully supported by
the evidence, including the modern forms of Basque words, the treatment of
borrowed words, and the medieval graphies.  It cannot lightly be waved away.

Now, I have heard and read several versions of Hualde's proposal here, though I
have not yet read his final published version, and so I must direct my remarks
to the earlier versions, which I hope are not substantially different from the
final version.

Important point: Hualde does *not* reject Michelena's reconstructed
phonological system.  Instead, he proposes to reinterpret its phonetics, and
its later development.

In M's analysis, word-initial */b d g/ in Pre-Basque, which were always
unaspirated and usually voiced, develop regularly into /b d g/ in the modern
language, with certain complications, notably for /b/, which are irrelevant
here and which I shall ignore.

In Hualde's alternative view, initial */b d g/ in Pre-Basque had facultative
voicing: that is, they could be realized, indifferently, either as [b d g] or
as [p t k] -- "indifferently", because there was no contrast of voicing in
word-initial plosives in Pre-Basque.  So far, this view is not significantly
different from M's view, but now comes the difference.  Hualde proposes that,
because of this facultative voicing, Pre-Basque word-initial */b d g/ sometimes
develop into modern /b d g/ but sometimes into modern /p t k/.  In other words,
he reckons, one or the other voicing possibility was selected arbitrarily for
each word, with some words receiving both treatments in different parts of the
country.

This is a perfectly plausible and respectable analysis, of course, but now it
must be evaluated against the evidence.  And here, I think, it falls down.

The problem, in my view, is that this analysis requires that both native words
and borrowed words should turn up in modern Basque, arbitrarily and
unpredictably, with both voiced and voiceless plosives, or in many cases with
both.  Hualde has tried hard to argue that precisely this is indeed the case.
But I can't agree.

To me, it seems overwhelmingly clear that obviously native words with initial
*/b g/ (no native initial */d/, recall) turn up in modern Basque with /b g/
virtually without exception.  I am thinking of words like <buru> 'head',
<bi(ga)> 'two', <bide> 'road', <berri> 'new', <bihotz> 'heart', <bero> 'hot',
<gurdi> 'cart', <gero> 'after, later', <gerri> 'waist', <gorri> 'red', <gaitz>
'bad', <gatz> 'salt', and many dozens of others.  These native words *just do
not* turn up with initial /p k/.  Exceptions to this statement are few,
confined to small areas, and often explicable by the sporadic but familiar
right-to-left voicing assimilation we find in Basque.  For example, <gurdi>
'cart' appears nowhere as *<kurdi>, but its phonologically regular derivative
<gurpil> 'cartwheel' is recorded as <kurpil> in the Gipuzkoan dialect -- by
anticipatory voicing assimilation.

In great contrast, obvious loan words overwhelmingly show both initial /p t k/
and, more particularly, enormous regional variation in initial voicing.
Examples:
<bake> ~ <pake> 'peace'; <bago> ~ <pago> 'beech'; <biper> ~ <piper> 'pepper';
<pinu> 'pine' (no *<binu>); <dekuma> ~ <tekuma> 'tithe'; <golko> ~ <kolko>
'breast'; <gutun> ~ <kutun> 'letter'; <dipula> ~ <tipula> 'onion'; <dorre> ~
<torre> 'tower'; <boltsa> ~ <poltsa> 'bag'; <dolare> ~ <tolare> 'wine-press';
<gontza> ~ <kontza> 'hinge'; <gereta> ~ <keleta> 'rustic gate'; <gapelu> ~
<kapelu> 'cap'; <garden> ~ <karden> 'transparent'; and countless others, with
both variants typically being found in many parts of the country.  Such voicing
fluctuation is so pervasive among loanwords that we may, with considerable
confidence, take it as the signature of a loan word.  When we encounter a
doubtful case like <gorotz(a)> ~ <korotz(a)> 'dung', with both variants found
right throughout the country, we may be almost certain that we are looking at a
loanword.

Accordingly, I find myself obliged to reject Hualde's interpretation.  It
requires native words to behave just like loanwords in their initial voicing,
but they plainly do not.

Finally, on the particular question of initial /p/, which, recall, Michelena
regarded as rare or nonexistent in Pre-Basque and as absent from initial
position.  I challenge anybody on this list to produce even a single Basque
word with initial /p/ which meets my usual criteria: it is recorded before
1600; it occurs in all or in nearly all dialects; it does not appear to be
polymorphemic; and it cannot reasonably be suspected of being borrowed from a
neighboring language.

Any takers?

Given the dozens of words in initial /b/ which satisfy my criteria, Hualde's
thesis requires us to find at least something like a comparable number with
initial /p/ or with /b/ ~ /p/ variation throughout the country.  And I can't
think of one.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk



More information about the Indo-european mailing list