15 October 2009

Proto-Basque and non-standard Basque

Proto-Basque is the reconstructed ancestor of the historical Basque dialects, presumably spoken in the Late Iron Age and whose main features were worked out by the Vascologist Koldo Mitxelena in his magna opus Fonética Histórica Vasca1. The consonantal system of Mitxelena's Proto-Basque didn't had a voiced/voiceless contrast but a fortis (tense)/lenis (lax) one. This was only relevant in medial position (between vowels), because at word-initial only lenis phonemes might appear.

With regard to stops, the French linguist André Martinet (Économie des changements phonétiques, 1955) posited an older system modelled after Danish, in which plosives could be realized either as voiceless aspirated (fortis) or mild voiceless (lenis) at word-initial and as mild voiceless (fortis) and approximants (lenis) between vowels. In Mitxelena's Proto-Basque, the aspirated plosives would have evolved to [h], which had no phonemic value.

I call this process Martinet's Law, and for comparative purposes we should differentiate between Early Proto-Basque (before Martinet's Law) and Late Proto-Basque (after Martinet's Law), sometimes called "canonical" or "Mitxelena's" as opposed to Bengtson's.

In the Aquitanian inscriptions we can found forms with initial h- like HALSCO or HAVTENN2 vs. others with t- like TALSCO (Iberian talsku) or TAVTINN (Iberian tautin) and different geographical distribution. This isogloss is a particular case of Martinet's Law, which separates Proto-Basque from other linguistic varieties like Iberian.

There are also Basque words like k(h)ako 'hook'3 < IE *ko(n)g- 'peg, hook, claw' < Paleo-Eurasian *ɣoŋɣV 'peg, nail' or tara 'young branch' < IE *dhal- 'sprout' which don't follow Martinet's Law and  constitute what I call non-standard Basque4. They are late loanwords from those linguistic varieties, whose speakers were stigmatizated people like highlanders or nomadic shepherds, a fact which lead to their ultimate extinction in the High Middle Ages (there's evidence in NW Catalonia's toponymy of a Bascoid language spoken there until aprox. 1000 AD).

Unlike his original proposal, Martinet's Law also can occur between vowels. For example, Basque zahar 'old' < EPBasq. *sakh5 vs. Iberian sakaŕ (Starostin's PNC *tʃ’HǝqwV 'big').
1 For a summary of Mitxelena's work, see Trask's The history of Basque (1997).

2 Compare Basque hauta 'election, excellent' < Celtic *toutā- 'people'.
3 In the Salazarian dialect there's an isolated form kaiku 'aquiline' found part of the compound sudur-kaiku 'aquiline nose'.
4 It's a pity Bengtson didn't recognize this as a genuine phenomenon.
5 Basque /z/ represents a lamino-alveolar sibilant [s].


  1. 1. What is the evidence that /aho/ < */pakho/? I mean apart from the highly tentative comparison with Celtic and PNC.

    It could be from **/[h]akho/, **/khakho/ (these would go well with PNC */ɦomɢɢwi/ I suppose), **/thakho/ etc. By the way, can you list all the sounds that could result in zero/[h]? Perhaps, it would be more honest to reconstruct **/(CC)a(CC)o/ instead.

    2. Why do you link Basque /k(h)ako/ to "Starostin's Eurasiatic *ɣoŋɣV ‘peg, nail’"? The phonetics and semantics is rather bad. Why not FU */kokka/ "hook", for example? There are plenty of much better matches Eurasia-wide. Have you looked at IE, by the way?

  2. Hi, Petr. I'm affraid this and other entries are outdated.

    1) Basque aho 'mouth' comes from papo, papu 'cheek', attested both in dialectal Basque as well as in Cantabrian Spanish and Asturian Romance. I take it as the Cantabrian output of the root corresponding to PNC *bek'wo. I'm going to deal with this root in a forthcoming post.

    2) The etymology I choose is sound, but I should have quoted IE *ko(n)g- 'peg, hook, claw', possibly from the above Paleo-Eurasian root. From this root, the Salazar dialect has an isolated form kaiku 'aquiline' in the compound sudur-kaiku 'aquiline nose'.

    Notice that 1) the cluster of nasal + velar stop was reduced and 2) o collapsed into a.

  3. Tavi, you don't seem to understand my point.

    1) What is the internal evidence that

    a. the first consonant was what you've claimed it to have been.

    b. the second consonant was what you've claimed it to have been.

    c. there was a nasal-stop cluster rather than a stop alone.

    2) If there is no internal evidence for the above, you should be fair enough to reconstruct...well, something like */(CC)a(CC)o/ (or */(C:)a(C:)o/ if you like). Sure, this will immensely decrease the value of such comparisons, but it will immensely increase their veracity.

    And, needless to say, it might rid you off the pinkish glasses you've been wearing.

  4. From a purely theoretical point of view, it's true that /aho/ should be reconstructed as */CaCo/ or something like that, but from a practical point of view, I chose this particular etymology because it's the most consistent available.

    As I told you before, i'm going to these roots in separate posts.