The oddities of Basque

[Talskubilos]

It perhaps might be remembered that proto-Basque seems to have been relatively "labial poor", I recall at least from Trask Historical Linguistics discussing an example of how internal reconstruction suggests that proto-Basque had no *m at least (it being largely confined to loanwords, absent from morphology and the relatively few native instances can generally be shown to be products of nasal assimilation of *b, e.g. mihi "tongue" from *bini (the *n being lost by the regular deletion process found elsewhere in Basque)).

Some points are worth mentioning here. First of all, the original labial in this word was retained in the compound arpin 'plantain', from ardi 'sheep' and *bini 'tongue' (see here). Secondly, the lost of intervocalic nasals isn't an exclusive feature of Basque, as it's also found in the neighbouring Gascon as well as in Gallego-Portuguese.

In that case then one might suppose that perhaps *b might not actually have been a stop in proto-Basque and it might have perhaps instead been a glide *w or similar and underwent fortition to a stop in common with the same process in Spanish.

In fact, Early Romance had a fricative labial β corresponding to Classical Latin w. In loanwords into Basque, this consonant became usually m before a and the back vowels o, u, but there're some cases where Basque b originated in a former w, namely the compound element gab-, a combinatory variant of gau 'night'. So IMHO Paleo-Basque had probably a labial phoneme, a stop without voicing contrast.

[Frislander]

It perhaps might be remembered that proto-Basque seems to have been relatively "labial poor", I recall at least from Trask Historical Linguistics discussing an example of how internal reconstruction suggests that proto-Basque had no *m at least (it being largely confined to loanwords, absent from morphology and the relatively few native instances can generally be shown to be products of nasal assimilation of *b, e.g. mihi "tongue" from *bini (the *n being lost by the regular deletion process found elsewhere in Basque)).

The inventory Trask reconstructs is lenis */b d g z s n l r/ vs. fortis */(p) t k tz ts N L R/, plus a suprasegmental aspiration feature.

In that case then one might suppose that perhaps *b might not actually have been a stop in proto-Basque and it might have perhaps instead been a glide *w or similar and underwent fortition to a stop in common with the same process in Spanish.

There could be phonotactic evidence one way or the other. My impression is that there's no reason to suspect this. Unless *d and *l can be unified, in which case the natural thing to do would be to posit /β l ɣ/, but that seems difficult since there was also *L.

Trask says the typical root had the structure C1VC2C3VC4, with heavy restrictions on the permissible consonants in each position:
- C1 could only be /b g z s l n/
- Permissible medial clusters were {r n l}{p t k b d g z s tz ts} and {z s}{p t k}
- C3 could be any consonant if C2 was absent
- C4 could only be /tz ts L N R/ or possibly {n l r}{tz}

Good to know. I agree there's not quite enough evidence to be conclusive, but I will perhaps also point out that if we're looking for explanations for gaps in systems then *b having been originally a glide might perhaps go some way towards explaining the absence of a fortis version of it. But again, as with the glottalic theory, evidence from phonological distribution isn't really evidence of anything at all, so again I'm not saying I'm convinced this would be how it is, but it would at least be an explanation.

Well, Basque appears to be the last survivor of a lost linguistic landscape which existed in Europe before it all was overrun by the Indo-Europeans. It is as if just one indigenous language had survived European colonization of North America.

I'm afraid this is an old clissé repeated ad nauseam, but this doesn't make it true. In fact, in his last book (aimed to the general public and thus less technical than its other works) the late Rodríguez Adrados suggested the ancestor of Basque could have been brought by the Steppe People along with IE.

IMHO, there're two linguistic facts which would support this:
1) the *e- prefix found in non-finite verb forms could be related to the so-called "augment" in Eastern IE languages (Greek, Indo-Iranian).
2) The verb edan 'to drink' could be related to Iranian dānu- 'river', found as a toponymic element.

I'm sorry, but if this is all you've got then I'm calling BS. Firstly, while the *e- prefix proposal seems initially interesting, it becomes somewhat less convincing when it is remembered that Indo-Europeanists these days tend to see the augment as having originally been an adposition that glomped onto the front of the verb as adpositions have a tendency to do in IE. And secondly, connecting edan to *dānu is semantic stretching of the highest order, and without other correspondences I'm disinclined to consider it not a coincidence.

[Talskubilos]

And secondly, connecting edan to *dānu is semantic stretching of the highest order, and without other correspondences I'm disinclined to consider it not a coincidence.

Firstly, the meaning 'to drink' can derive from 'to flow' through specialization, namely 'to flow through one's throat'. Secondly, the Iranian word seems to be linked (although directly derived) to IE *dhen- 'to flow' > Latin fons 'spring'.

Chance resemblances, nothing else. Try harder.

[Talskubilos]

The inventory Trask reconstructs is lenis */b d g z s n l r/ vs. fortis */(p) t k tz ts N L R/, plus a suprasegmental aspiration feature.

Actually, Trask was following Koldo Mitxelena, the pioneer of modern Vascology, so in purity we should call it Mitxelena-Trask's.

IMHO, geminated (fortis) /N L R/ look more like adaptations from Romance (i.e. Vasco-Romance) than genuine Proto-Basque phonemes. As a matter of fact, the liquids /l r/ merged intervocally, thus leaving a contrasting pair /r ~ /L/. I also suspect a Paleo-Basque retroflex was hidden before the latter, as some Romances (namely Asturian, Pyrenaic Aragonese, Sardinian, Sicilian and some South Italian dialects) have traces of it.

[WeepingElf]

The inventory Trask reconstructs is lenis */b d g z s n l r/ vs. fortis */(p) t k tz ts N L R/, plus a suprasegmental aspiration feature.

Actually, Trask was following Koldo Mitxelena, the pioneer of modern Vascology, so in purity we should call it Mitxelena-Trask's.

IMHO, geminated (fortis) /N L R/ look more like adaptations from Romance (i.e. Vasco-Romance) than genuine Proto-Basque phonemes. As a matter of fact, the liquids /l r/ merged intervocally, thus leaving a contrasting pair /r ~ /L/. I also suspect a Paleo-Basque retroflex was hidden before the latter, as some Romances (namely Asturian, Pyrenaic Aragonese, Sardinian, Sicilian and some South Italian dialects) have traces of it.

Similar fortis sonorants occur in Irish, so this may be an Atlantic European areal or substratal thing. Of course, we don't know how far the family Basque is the last survivor of ever extended, but it seems IMHO likelier that we are dealing with a family associated with the Neolithic Cardial culture than one associated with the Bell Beaker culture.

[Vijay]

What the hell even is this thread?!

[Travis B.]

What the hell even is this thread?!

I still think we're in the millinyland range here.

[WeepingElf]

What the hell even is this thread?!

I don't know why he did not just post to the Paleo-European languages thread - perhaps he felt that he was not welcome there because everybody disagreed with him. But that is nothing that could be circumvented by starting a new thread in the same forum where the same people are around and will disagree with him here too.

Yet, I wouldn't say that everything Talskubilos posted was automatically wrong; for instance, I conceded that his "Baltoid" and my "Southern IE" may be the same thing, only viewed from very different angles and thus interpreted in different ways.

But concerning the oddities of Basque, I feel that Proto-Basque, as reconstructed by Mitxelena, was weirder than Modern Basque, the latter having at least a phonology that doesn't really look out of place in Western Europe, while the former had one that is wildly different from anything we now know in Western Europe. Apparently, Basque converged towards the surrounding IE languages (and Iberian looks quite similar to Proto-Basque). It is harder to say how bizarre the Proto-Basque morphosyntax was; at least, I haven't seen a reconstruction of that yet.

One question where Talskubilos and I (and with me, most historical linguists) disagree is whether Basque came from the steppe together with IE (which appears to be T.'s position) or it is the last surviving language of pre-IE Europe, probably descending from the language of the Neolithic Cardial culture. The Yamnaya expansion from the steppe had a northern and a southern wing, as I already said, and one of them, probably the northern one, is the origin of all the IE languages spoken today, but what did the southern Yamnaya speak? My guess, a different branch of IE which Anatolian is the only part of that has left written records; while T. thinks, it seems, that they spoke Proto-Vasconic. But what then to make of Anatolian? Also, I feel as if "Basque from the steppe" was an attempt to save or strengthen Vasco-Caucasian, the notion that Basque was related to NWC and NEC. But just why? Basque has one thing in common with those two families (of which we don't even know whether they are related to each other or not; they actually have very little in common with each other), and that is ergativity - a far too narrow base to build a relationship hypothesis on. Ergative languages are common enough around the world; it is the same kind of canard as with the Semitic substratum in Insular Celtic because both are VSO.

[Talskubilos]

I don't know. This is all stuff for the Paleo-European languages thread, where it has been pretty much discussed to death, and T. has pretty much lost there, so he opened a new thread as if such a move helped anything, which it doesn't.

Not really. Here the focus is on Basque and its relationships. Other topics have better go in the Paleo-European thread.

[Talskubilos]

IMHO, geminated (fortis) /N L R/ look more like adaptations from Romance (i.e. Vasco-Romance) than genuine Proto-Basque phonemes. As a matter of fact, the liquids /l r/ merged intervocally, thus leaving a contrasting pair /r ~ /L/.

Similar fortis sonorants occur in Irish, so this may be an Atlantic European areal or substratal thing.

This would link /L/ > /l/ to a possible or even likely Celtic substrate behind these Romances, including the "Vasco-Romance" embedded in Basque. Hispano-Romance (as well as Catalan, with the exception of some Occitan loans) palatalized both N and L, and this is reflected in loanwords to Basque.

To summarize:
Romance /N/ > Gallego-Portuguese /n/, Hispano-Romance /ɲ/, Vasco-Romance /n/, Gascon /ɳ/. Here VR goes along with GP.
Romance /L/ > Gallego-Portuguese /l/, Hispano-Romance /ʎ/, Vasco-Romance /l/, Gascon /ɾ/ intervocally (lenis) and /t/ts/c/ at word final (fortis). Once again, VR goes along with GP only, but Gascon goes on its own, due to substrate influence.

I also suspect a Paleo-Basque retroflex was hidden before the latter, as some Romances (namely Asturian, Pyrenaic Aragonese, Sardinian, Sicilian and some South Italian dialects) have traces of it.

There's some indirect evidence Iberian had a retroflex transcribed as /ld/ and adapted in Latin as /l(l)/. Thus Paleo-Basque (the stage previous to PB) inheritance might have been masked by Vasco-Romance.

[Travis B.]

as well as Catalan, with the exception of some Occitan loans

Umm sorry but Catalan groups with Occitan, not with Hispano-Romance.

[Talskubilos]

Umm sorry but Catalan groups with Occitan, not with Hispano-Romance.

That's right, but as far as the results of N and L it usually follows Hispano-Romance.

[bradrn]

Also, I feel as if "Basque from the steppe" was an attempt to save or strengthen Vasco-Caucasian, the notion that Basque was related to NWC and NEC. But just why? Basque has one thing in common with those two families (of which we don't even know whether they are related to each other or not; they actually have very little in common with each other), and that is ergativity - a far too narrow base to build a relationship hypothesis on. Ergative languages are common enough around the world; it is the same kind of canard as with the Semitic substratum in Insular Celtic because both are VSO.

And the funny thing is, Basque is closest to Kartvelian in alignment than it is to anything in ‘Vasco-Caucasian’. Indeed, neither Basque nor Georgian are ergative at all. (Linguists tend to get hopelessly excited whenever they see a non-accusative alignment, and call it ‘ergative’ irrespective of what it actually is.)

[Talskubilos]

And the funny thing is, Basque is closest to Kartvelian in alignment than it is to anything in ‘Vasco-Caucasian’. Indeed, neither Basque nor Georgian are ergative at all. (Linguists tend to get hopelessly excited whenever they see a non-accusative alignment, and call it ‘ergative’ irrespective of what it actually is.)

Really? Could you explain it, please?

[WeepingElf]

Similar fortis sonorants occur in Irish, so this may be an Atlantic European areal or substratal thing.

This would link /L/ > /l/ to a possible or even likely Celtic substrate behind these Romances, including the "Vasco-Romance" embedded in Basque. Hispano-Romance (as well as Catalan, with the exception of some Occitan loans) palatalized both N and L, and this is reflected in loanwords to Basque.

AFAIK, the "Celtic" fortis sonorants are an innovation of Goidelic, connected with its lenition system, and not found in Continental Celtic, but I am not a Celticist.

To summarize:
Romance /N/ > Gallego-Portuguese /n/, Hispano-Romance /ɲ/, Vasco-Romance /n/, Gascon /ɳ/. Here VR goes along with GP.
Romance /L/ > Gallego-Portuguese /l/, Hispano-Romance /ʎ/, Vasco-Romance /l/, Gascon /ɾ/ intervocally (lenis) and /t/ts/c/ at word final (fortis). Once again, VR goes along with GP only, but Gascon goes on its own, due to substrate influence.

Romance did not have fortis "/N/" or "/L/", it had geminate /nn/ and /ll/ inherited from Latin, but one could of course easily have turned into the other or vice versa. And the substratum of Gascon is known - it is Aquitanian, i.e. a variety of Proto-Basque.

Also, I feel as if "Basque from the steppe" was an attempt to save or strengthen Vasco-Caucasian, the notion that Basque was related to NWC and NEC. But just why? Basque has one thing in common with those two families (of which we don't even know whether they are related to each other or not; they actually have very little in common with each other), and that is ergativity - a far too narrow base to build a relationship hypothesis on. Ergative languages are common enough around the world; it is the same kind of canard as with the Semitic substratum in Insular Celtic because both are VSO.

And the funny thing is, Basque is closest to Kartvelian in alignment than it is to anything in ‘Vasco-Caucasian’. Indeed, neither Basque nor Georgian are ergative at all. (Linguists tend to get hopelessly excited whenever they see a non-accusative alignment, and call it ‘ergative’ irrespective of what it actually is.)

Yes, WALS classifies Basque as "active-stative", apparently on the ground that some intransitive verbs govern the ergative case. I know too little about this to make a comment. And in my personal and very humble opinion, Early PIE may also have been active-stative.

[Talskubilos]

And the substratum of Gascon is known - it is Aquitanian, i.e. a variety of Proto-Basque.

I'd say Paleo-Basque instead, a stage previous to the reconstucted Proto-Basque (PB). The thing is Aquitanian has no actual texts, so it's less known than Iberian.

[WeepingElf]

And the substratum of Gascon is known - it is Aquitanian, i.e. a variety of Proto-Basque.

I'd say Paleo-Basque instead, a stage previous to the reconstucted Proto-Basque (PB). The thing is Aquitanian has no actual texts, so it's less known than Iberian.

Fair - call it Proto-Basque, call it Paleo-Basque; what's in a name? Fact is that the Aquitanian names (that's all we have) can be interpreted by means of Basque, and the difference between Aquitanian and Proto-Basque is not great. And what regards Iberian, there is IMHO quite some evidence that the language is related to Basque, perhaps the most compelling being the set of Iberian numerals all of which resemble Basque numerals. Yet, Basque has so far been only of limited help in understanding Iberian. The languages are apparently related, but not close - a good deal more distant than Aquitanian.

[Talskubilos]

Fair - call it Proto-Basque, call it Paleo-Basque; what's in a name?

They're different entitites. Proto-Basque is the putative ancestor of the historical Basque varieties, roughly dated to the 6th century AD, while Paleo-Basque would be an earlier stage more or less contemporaneous to the Roman conquest.

Fact is that the Aquitanian names (that's all we have) can be interpreted by means of Basque,

As a matter of fact, some of these names, e.g. CISON (Basque gizon 'man') or ANDERE (Basque and(e)re 'lady') are loanwords from the Gaulish substrate.

And what regards Iberian, there is IMHO quite some evidence that the language is related to Basque, perhaps the most compelling being the set of Iberian numerals all of which resemble Basque numerals.

There're some genuine matches, but other are chance resemblances, very common between both languages.

[...] and the difference between Aquitanian and Proto-Basque is not great. [...] Yet, Basque has so far been only of limited help in understanding Iberian. The languages are apparently related, but not close - a good deal more distant than Aquitanian.

The problem is we haven't got real texts in Aquitanian, but the names found in the inscriptions tells us Aquitanian and Iberian were cousins: eg. Aquitanian HAVTENN- ~ Iberian tautin.

[WeepingElf]

Fair - call it Proto-Basque, call it Paleo-Basque; what's in a name?

They're different entitites. Proto-Basque is the putative ancestor of the historical Basque varieties, roughly dated to the 6th century AD, while Paleo-Basque would be an earlier stage more or less contemporaneous to the Roman conquest.

I understand. The historical Basque varieties are more closely related to each other than to Aquitanian, so Proto-Basque and Paleo-Basque, as you call them, aren't the same.

Fact is that the Aquitanian names (that's all we have) can be interpreted by means of Basque,

As a matter of fact, some of these names, e.g. CISON (Basque gizon 'man') or ANDERE (Basque and(e)re 'lady') are loanwords from the Gaulish substrate.

Do they have IE etymologies? Do they have cognates in Welsh, Breton or Irish? AFAIK, not! This is the first time I see the claim that Basque borrowed these words from Gaulish! Where is Dewrad when we need him? He should know about this; he is our resident expert on Celtic. Also, Gaulish would be a superstratum in relation to Basque, not a substratum. I am pretty certain that Gaulish belongs to a younger stratum than Basque.

And what regards Iberian, there is IMHO quite some evidence that the language is related to Basque, perhaps the most compelling being the set of Iberian numerals all of which resemble Basque numerals.

There're some genuine matches, but other are chance resemblances, very common between both languages.

Maybe.

[...] and the difference between Aquitanian and Proto-Basque is not great. [...] Yet, Basque has so far been only of limited help in understanding Iberian. The languages are apparently related, but not close - a good deal more distant than Aquitanian.

The problem is we haven't got real texts in Aquitanian, but the names found in the inscriptions tells us Aquitanian and Iberian were cousins: eg. Aquitanian HAVTENN- ~ Iberian tautin.

This reminds me of Vennemann's claim that Basque initial /h/ matches everything in other languages, e.g. Basque handi vs. Late Latin grandis.

[Talskubilos]

Do they have IE etymologies? Do they have cognates in Welsh, Breton or Irish? AFAIK, not!

Quoted form Matasović's Etymological Dictionary of Proto-Celtic (EDPC):

*gdonyo- 'human, person' [Noun]
GOID: OIr. duine [io m]; doíni [p]
W: MW dyn [m and f]
BRET: OBret. don, den, MBret.den
CO: OCo. den gl. homo
GAUL: -xtonio (Vercelli)
COGN: Lat. homo, Go. guma
PIE: *d^hģ^hom-yo- 'human, earthling' (IEW:414)
SEE: *gdon- 'earth'
ETYM: The Gaulish form -xtonio (in the compound form teuoxtonio) should be read -gdonio (the alphabet of Vercelli does not distinguish between voiced and voiceless stop, cf. Delamarre 176). OIr. Nom. pl. doíni is from a suppletive stem (attested also as Nom. sg. doín, doén in poetry).

*anderā 'young woman' [Noun]
GOID: MIr. ander [ā f] (DIL ainder)
W: MW anneir [f] 'heifer' (GPC anner, annair)
BRET: MBret. anner 'heifer'
CO: OCo. annoer gl. uitula
GAUL: ? anderon [Ge p.] (Larzac)
ETYM: MIr. ander may have been an o-stem originally (DIL). The change of meaning attested in Brittonic ('young woman' > 'heifer') is based on a common metaphor in cattle-breeding societies. It has been suggested that there is a connection of this Celtic etymon with Basque andere 'lady, woman', but this might only be a chance ressemblance. The meaning of Gaulish anderon is not certain (it might rather be related to Lat. inferus, so this word might not belong here at all.

This is the first time I see the claim that Basque borrowed these words from Gaulish!

Not Basque but Aquitanian/Paleo-Basque. It isn't my fault Vascologists have been unable to identify those and other Celtic loanwords.

Also, Gaulish would be a superstratum in relation to Basque, not a substratum. I am pretty certain that Gaulish belongs to a younger stratum than Basque.

See here.

This reminds me of Vennemann's claim that Basque initial /h/ matches everything in other languages, e.g. Basque handi vs. Late Latin grandis.

I'm sure you didn't know the Aquitanian inscriptions also contain Iberian anthroponyms, although they're minoritary.