Basque <ibili>

[Larry Trask]

[ moderator re-formatted ]

Miguel Carrasquer Vidal writes:

[LT]

>>     <gurpil> 'cartwheel', 'wheel'

>> This is from <gurdi> 'cart' + *<bil>.  The phonology is absolutely regular:
>> *<gurdi-bil> -->  *<gurd-bil> --> *<gurt-bil>  --> *<gurt-pil>  -> <gurpil>.

>>     <opil> 'small round bread roll or pastry'

>> This is from <ogi> 'bread', and again the phonology is perfect:
>> *<ogi-bil>  --> *<og-bil>  -->  *<ot-bil>  -->  *<ot-pil>  --> <opil>.

> Given Mitxelena's reconstruction of "fortis" consonants and your
> interpretation of them as geminates, wouldn't it be preferrable
> to derive:

>  gurdi + bil > gurdbil > gurbbil > gurpil
>  ogi   + bil > ogbil   > obbil   > opil    ?

I formerly favored this view myself, and I would very much like to favor
it now.  Unfortunately, I can't, because the evidence is against it.

One piece in particular.  The Basque word <errepide> 'highway' is a
transparent compound of <errege> 'king' and <bide> 'road'.  The final /e/
of the first element is lost regularly.  The analysis suggested above
would require *<errege-bide> --> *<erreg-bide> --> *<errebbide> -->
<errepide>.  But the word is explicitly recorded in the medieval (early
12th-century) Fuero General of Navarra as <erret bide>:

Libro III, tit. VII, cap. IV, p. 53:

	"...en logares en la cayll, que dize el bascongado erret bide."

This in fact is just one of several attestations of the form <erret(-)>, but
it is the clearest one.  And this, to my mind, is enough to settle the
matter.  Much as I might prefer the other analysis, the facts point clearly
to a change of the first plosive ina plosive cluster to */t/.

> Of course, -b(i) [?], -d(i), -g(i) compounded before initial
> vowel give -t, which then becomes harder to explain.  Maybe an
> initial vowel was formerly preceded by a glottal stop in Basque
> (which isn't the case now), and we might suppose that the fortis
> variant of /?/ became /t/:

>  begi + *?ile > beg?ile > be??ile > betile
>  ardi + *?ile > ard?ile > ar??ile > artile.

But the analysis endorsed above generalizes neatly to handle such cases,
without the positing of any hypothetical sounds.

>>     <ukabil> 'fist'

>> This is from the archaic <uko> 'forearm', recorded in Oihenart in the 17th
>> century, and again the phonology is perfect:

>>     *<uko-bil> --> <ukabil>

> (In old compounds -i and (usually) -u are dropped, while -a/-o/-e
> become -a-).  I wonder about the motivation for that last change.
> Could it have gone through a stage */@/ (schwa), which later
> became Basque /a/?  So, in this case: uko [*uggo] + bil > uk at bil > ukabil.

Precisely this reduction to schwa, followed by the change of svchwa to /a/,
has been cautiously posited.  But we have no evidence.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk