buber.net > Basque > Euskara > Larry > Note 15: Number-Names
For security reasons, user contributed notes have been disabled.
Note 15: Number-Names
by Larry Trask
The Basque number-names are mostly ancient and mysterious, but we have managed to learn a little about some of them.
The universal <bat> 'one', with its unusual final /t/, has clearly lost a final vowel. We reconstruct *<badV>, and almost certainly *<bade>. This form is probably the source of such words as northern <bedera> 'one each'. This name is also unique in that it always follows the noun it modifies. Perhaps it had a different kind of origin from most other number-names.
For 'two', the conservative form is <biga>. This form is still in everyday use in the north, alongside <bi>. The general rule is this: <biga> is used when the number-name stands alone, or in counting, but the reduced form <bi> is used when the number modifies something. In the south, the reduced form has been generalized to all contexts. We can guess how the reduction came about. Like <bat>, the word <biga> originally followed its noun. This postposed position is found in some early texts in the east, and of course it is still normal in Bizkaian today. And postposed items tend to undergo reduction in Basque, as in, for example, *<zur-bide> 'wood-way' --> *<zubide> --> <zubi> 'bridge'. In most dialects, though, <bi> has been shifted to preposed position, thus falling into line with all the higher number-names. (By the way, northern varieties show variant forms of <biga>; these are <bida> and <bia>.)
For 'three', <(h)irur> is the conservative form, since everybody says <(h)irurak> for 'all three', and common <(h)iru> results from loss of the final (weak) /r/.
Likewise, for 'four', <laur> is the conservative form, and common <lau> results from the same loss.
For 'five', eastern <bortz> is more conservative than western <bost>, which results from another familiar process, as in eastern <ortzegun>, western <ostegun> 'Thursday', and eastern <(h)ertze>, western <este> 'intestine'. The unusual final cluster in <bortz> suggests that an earlier vowel may have gotten lost: possibly original *<bortza> or something similar?
For 'six', <sei> looks a great deal like the neighboring Romance forms, and many writers have suggested that <sei> is borrowed from Romance. I doubt it, though. All the neighboring Romance varieties have a final sibilant in their word for 'six', as in Castilian <seis>, and so, if the Basque word were borrowed, we would expect something like *<seits>, or at best *<seis>, and not <sei>. I think <sei> is native Basque, and I think its resemblance to the Romance words is coincidence, like the unrelated English 'much' and Spanish <mucho>.
For 'seven', <zazpi> is mysterious. Attempts at deriving this from Coptic or Egyptian strike me as preposterous. Michelena once suggested, very diffidently, a source in a hypothetical *<bortzaz-bi> 'two with five'. For this to work, we need three things: (1) <bortz> was originally *<bortza> (possible); (2) the instrumental/adverbial ending <-z> originally meant 'with' (more than possible; in fact, likely); and (3) the original word has lost its entire first syllable (hard to swallow, I think, but perhaps not wholly impossible).
For 'eight', the universal <zortzi> is wholly unintelligible.
Now, <bederatzi> 'nine' is much longer than the other number-names, and it surely consists historically of at least two pieces. Most people think the first element must be *<bade> 'one', and so the whole formation might originally have meant something like 'one left' or 'one more' -- normally enough, when you're counting on your fingers. (Compare our 'eleven', which is historically <endleofon> 'one left'.) But the eastern dialects show a slightly different form: Zuberoan <bederatzü> (that's u-umlaut) and Roncalese <bedratzu> point clearly to earlier *<bederatzu>. Since a change from final /i/ to /u/ would be without parallel in these dialects, Michelena concludes that *<bederatzu> is the original Basque form, and that common <bederatzi> results from contamination from the preceding <zortzi>. In other words, the Basques used to count "<zazpi>, <zortzi>, <bederatzu>", but they got into the habit of saying instead "<zazpi>, <zortzi>, <bederatzi>". This kind of contamination between adjoining number- names is in fact exceedingly common, and it has occurred in a number of European languages.
For 'ten', <(h)amar> is enigmatic, but it probably derives from earlier *<(h)anbar>, just as <seme> 'son' and <ume> 'child' derive from earlier *<senbe> and *<unbe> -- both of these apparently recorded in Aquitanian, as SEMBE- and as OMBE-, VMME, respectively.
Again, <(h)ogei> 'twenty' is opaque. Northern <hogoi> appears to result from an unusual vowel assimilation.
Finally, <ehun> 'one hundred' has been much discussed. Several people have wanted to see this as a loan from Visigothic. The reason is that the Gothic word for '100' is recorded: it is written as <ain hund>, which is thought to represent the pronunciation [en hund]. Now, if this [en hund] had been borrowed into early Basque, the result would have been *<enun>. And this, after the regular early medieval loss of /n/ between vowels, would have turned into the required <ehun>. This is a pretty story, but I doubt it. There is no known case of a word from Gothic, or from any Germanic language, passing directly into Basque. Basque relations with the Visigoths were uniformly hostile. It is not even clear that the Visigoths carried on speaking Gothic for very long after they settled in Spain. Finally, the Basques had already borrowed the larger number-name <mila> '1000' from the Romans, so what motivation could they have had for a later borrowing of a name for '100' from anybody, least of all from the Visigoths? I think <ehun> is probably native.
Larry Trask COGS University of Sussex Brighton BN1 9QH UK
Tel: 01273-678693 (from UK); +44-1273-678693 (from abroad) Fax: 01273-671320 (from UK); +44-1273-671320 (from abroad)