Paleo-European languages

[Whimemsz]

It could also just be that intervocalic lenition is extremely common. I'm wary in general of any claims about "substrate" influence to explain changes that are perfectly common and natural, which people seem to make WAY WAY too often (not just in contexts like this). If there's something quite specific and idiosyncratic about the change involved (e.g., the complete restructuring of the vowels and syllable structure of Maghrebi Arabic in a manner that makes it almost exactly like Berber languages), or you can demonstrate that the changes involved are both reasonably specific and happened simultaneously in a localized area (or quickly after the new group moved into the area) among populations that were in significant contact or underwent large-scale language shift, I think you can reasonably claim that there's direct influence, but otherwise the null hypothesis that they're unconnected should be maintained. Especially in instances like this where the proposed similarities, as you say, are quite nebulous, and instantiated very differently in each language group.

[Znex]

It's possible, but the Western Romance and Celtic changes happen way too late to make me think they're related to an older substratum; the shared Western Romance and Celtic changes in fact make most sense in light of the Western Roman Empire covering old Celtic territory. The shift happened as more and more Celtic speakers adopted Latin in place of their native language. If any shift from an older substratum were to have occurred, it seems more likely to me to have happened in pre-(Insular)-Celtic.

That explanation seems tempting, but then you have the problem of the fact that Continental Celtic as attested shows no such sound changes. There are also noticeable mismatches between the two lenition patterns - for one, both branches of Insular Celtic lenite *m as well, whereas I'm pretty sure this is unattested in Romance, rather instead *n is the more likely to be lost, as in Portuguese (this is a change that is shared with Basque - direction of spread is as far as I know unknown). Also no Romance language seems to have included preceding adpositions and articles to create a full-on initial mutation process, which I would expect to have happened if it was the result of imposition.

That Romance doesn't exactly copy what Celtic does is no reason to surmise it didn't at all; in fact it's unlikely it should do completely at all for the same reason that bogolangs don't exist in nature: language systems are still significantly different from each other, whether they have the same speakers or not. I find it likely that Romance and Celtic affected each other not just because they share consonant lenition, but also because they share a number of other sound changes, some more and some less complicated.

No written attestation is hardly evidence for the changes not having existed at all (or that they didn't take place in a shorter space of time than might be imagined). We already have Primitive and Old Irish as an example: Primitive Irish resembles Continental Celtic strongly before it is replaced in inscriptions in only a few hundred years by the greatly modified Old Irish. Breton is another example: up until relatively recently, consonant mutations were never shown in writing like in Welsh and Irish, but we know by testimony of the modern language and by written witness that it has had mutation as long as Welsh has.

And besides, Celtiberian and Gallaecian actually do show lenition up to their last attestations (for voiced stops at least).

[WeepingElf]

Intervocalic lenitions are indeed common enough; and the Western Romance lenition is different enough from the British Celtic one (the biggest difference being that in Western Romance, no initial mutations emerged through it), which in turn is different enough from the Irish lenition (where intervocalic voiceless stops gave voiceless fricatives rather than the British and Romance voiced stops). At least, all three seem to have emerged at about the same time, namely around 400 AD (they preceded apocope in both branches of Insular Celtic). This may be some sort of areal phenomenon, though a common root in an allophonic lenition in early Celtic, which was transferred to Vulgar Latin when the people of Gaul and Hispania shifted to the latter, is possible. In this allophonic lenition, intervocalic consonants would have been spoken with somewhat less force, though they'd still have been the same phonemes and the lenition therefore not indicated in writing. Alas, we don't know.

[Ares Land]

French has liaison which does look a bit like an early stage of the process.
FWIW there are some hints that there might have been something like liaison in late Gaulish -- except it chiefly affects the loss of final nominative -s, so it might have been Vulgar Latin having an influence on a Celtic language rather than the other way around!

https://www.academia.edu/16430640/LA_TA ... n_gaulois_
(The article doesn't quite live up to the English abstract. It turns out there's only so much you can extrapolate from a few words in bad Gaulish.)

As an aside, several Romance language do have initial mutations. Check out syntactic gemination: https://en.wikipedia.org/wiki/Syntactic ... on#Italian, or articles/gender in Neapolitan: https://en.wikipedia.org/wiki/Neapolita ... e_articles
The funny thing, though, is that it occurs in those Italian dialects that are not Western Romance

[Talskubilos]

Basque may have borrowed many words from Iberian. The two languages were neighbours in the Ebro valley, and the Iberians were an urban civilization earlier than the Basques. The Basque word for 'city', hiri in Modern Basque and ili in Proto-Basque, is an obvious candidate, considering the many Iberian city names beginning with Ili-.

The thing is the element Ili- appears in Latinized toponyms and it corresponds to Iberian ildiŕ as in e.g. Ildiŕda ~ Latin Ilerda.
On the other hand, Basque (h)iri 'town' is identical to the compound element -iri 'near', so I don't see any direct relationship with Iberian.

[WeepingElf]

The Vasco-Iberian hypothesis (i.e., the hypothesis that Basque and Iberian are related) is one of the least lunatic ideas about the origin of Basque (and R. L. Trask was unfair in tarring it with the same brush as the many crackpot ideas about Basque in his History of Basque - otherwise an excellent book that anyone with an interest in the pre-Roman languages of the Iberian Peninsula should read). Sure, Basque has so far not been much of a help in understanding Iberian, as opposed to how the classic IE languages were helpful in understanding Hittite once it had been identified as an IE language; but then, the Iberian corpus is much smaller than the Hittite one, and the time depth may be greater. When Hittite flourished around 1500 BC, the time depth of its relationship to the classic IE languages was only about 2,000 years; if Basque and Iberian descended from the language of the Cardial Neolithic farmers (IMHO a plausible speculation), the time depth, reckoned from the date of the Iberian inscriptions, would be up to about 5,000 years. At such time depths, a relationship becomes hard to detect if one of the languages is as fragmentarily known as Iberian and the other one is an internal reconstruction, with all the problems that entails, as is the case with Proto-Basque.

[Talskubilos]

The Vasco-Iberian hypothesis (i.e., the hypothesis that Basque and Iberian are related) is one of the least lunatic ideas about the origin of Basque (and R. L. Trask was unfair in tarring it with the same brush as the many crackpot ideas about Basque in his History of Basque - otherwise an excellent book that anyone with an interest in the pre-Roman languages of the Iberian Peninsula should read). Sure, Basque has so far not been much of a help in understanding Iberian, as opposed to how the classic IE languages were helpful in understanding Hittite once it had been identified as an IE language; but then, the Iberian corpus is much smaller than the Hittite one, and the time depth may be greater. When Hittite flourished around 1500 BC, the time depth of its relationship to the classic IE languages was only about 2,000 years; if Basque and Iberian descended from the language of the Cardial Neolithic farmers (IMHO a plausible speculation), the time depth, reckoned from the date of the Iberian inscriptions, would be up to about 5,000 years. At such time depths, a relationship becomes hard to detect if one of the languages is as fragmentarily known as Iberian and the other one is an internal reconstruction, with all the problems that entails, as is the case with Proto-Basque.

IMHO, there's no way Iberian and Basque would look so similar and yet mutually unintelligible if their common ancestor would date back to the Cardial Neolithic.In fact, the so-called "Aquitanian inscriptions", an epigraphic corpus written in Latin but containing Vasconic theonyms and anthroponyms shows Paleo-Basque/Iberian isoglosses such as halsco- ~ talsco-, hautenn- ~ tautin-.

My guess is Basque was heavily influenced by other languages (mostly but not exclusively Latin-Romance) in the last millenia and it had a divergent evolution from Iberian. For example, Basque buru 'head' is apparently a loanword from an Italoid (i.e. Italic-like) reflex of IE *g^woro- 'mountain' (cfr. the Lusitanian theonyms Boro, Borea). By contrast, its semantic Iberian counterpart bin, found as an anthroponymic element (most Iberian anthroponyms are two-word compounds much alike North American Indians's) would derive from P-Celtic (Gaulish) penno- 'head, end', whose Basque output would be the so-called "superlative suffix" -en.

Another IE loanword in Iberian without Basque counterpart would be baides 'witness(es)', a recurring term in inscriptions which can be readily derived from *weid- 'to see'. On the other hand, Basque has an *e- prefix in non-finitive verb forms, a feature apparently shared with Iberian and which could be related to the augment *h₁e- in Eastern IE languages. This would indicate an Eastern origin for the putative common ancestor of Basque and Iberian, as suggested (although without supporting evidence) by the late Rodríguez Adrados.

[WeepingElf]

Indeed, the relationship between Basque and Iberian may be closer than what I speculated about. I have no problems with that. But are the names on the Ascoli Bronze, which are one of the few items where Basque was helpful, really Iberian? Those soldiers were recruited somewhere near Zaragoza, which lies close to the old border between Proto-Basque and Iberian, so they may actually be Proto-Basque. Surely, Proto-Basque and Iberian were different languages, otherwise we could easily read the Iberian inscriptions. They may have been closely related, though.

[Talskubilos]

Indeed, the relationship between Basque and Iberian may be closer than what I speculated about. I have no problems with that. But are the names on the Ascoli Bronze, which are one of the few items where Basque was helpful, really Iberian? Those soldiers were recruited somewhere near Zaragoza, which lies close to the old border between Proto-Basque and Iberian, so they may actually be Proto-Basque.

Surely you're misinformed, because the names in the Ascoli Bronze are latinized versions of two-word compounds much alike North American Sitting Bull. Thank to this document scholars learned to identify Iberian anthroponyms in inscriptions, but not because they could be translated through Basque.

Surely, Proto-Basque and Iberian were different languages, otherwise we could easily read the Iberian inscriptions. They may have been closely related, though.

Although some scholars tend to identify Aquitanian with Paleo-Basque (the direct ancestor of Basque) on the evidence of lexical isoglosses such as CISON ~ gizon 'man' or ANDERE ~ and(e)re 'lady', it looks like were part of a dialectal continuum which would also include (although more distantly) Iberian. However, I guess Aquitanian/Paleo-Basque (of which we don't possess real texts) would be unintelligible to modern Basque speakers, although probably less than Iberian.

[Talskubilos]

1. Sardinia has its name from the š3rdn, one of the "Sea Peoples" of Egyptian chronicles, settling in Sardinia.

It looks like an IE-satem output of *k´erdh- 'herd' (cfr. Sanskrit śárdha- 'host, troop'). In fact, Basque has the isolated word sarda 'school fish' (Biscayan), in addition to more "metabolized" ones which I won't quote now.

[WeepingElf]

1. Sardinia has its name from the š3rdn, one of the "Sea Peoples" of Egyptian chronicles, settling in Sardinia.

It looks like an IE-satem output of *k´erdh- 'herd' (cfr. Sanskrit śárdha- 'host, troop'). In fact, Basque has the isolated word sarda 'school fish' (Biscayan), in addition to more "metabolized" ones which I won't quote now.

One moment. You snipped a line from an old post of mine such that it seems as if I had asserted that Sardinia had its name from the š3rdn, which I never did. In fact, it was one of three items in a list of what would have to be true in order to assume that Paleo-Sardinian was related to Etruscan, all three items being probably counterfactual. Such editing of quotes which leads to wrong assumptions about the intentions and opinions of the person quoted is illicit!

[Talskubilos]

One moment. You snipped a line from an old post of mine such that it seems as if I had asserted that Sardinia had its name from the š3rdn, which I never did. In fact, it was one of three items in a list of what would have to be true in order to assume that Paleo-Sardinian was related to Etruscan, all three items being probably counterfactual. Such editing of quotes which leads to wrong assumptions about the intentions and opinions of the person quoted is illicit!

OK, my appologies.

I think it's quite obvious the toponym Sardinia derives from š3rdn and the most likely etymology is the one I quoted before.

[WeepingElf]

One moment. You snipped a line from an old post of mine such that it seems as if I had asserted that Sardinia had its name from the š3rdn, which I never did. In fact, it was one of three items in a list of what would have to be true in order to assume that Paleo-Sardinian was related to Etruscan, all three items being probably counterfactual. Such editing of quotes which leads to wrong assumptions about the intentions and opinions of the person quoted is illicit!

OK, my appologies.

I think it's quite obvious the toponym Sardinia derives from š3rdn and the most likely etymology is the one I quoted before.

It may indeed be the case that Sardinia is named for the Š3rdn, but I don't see what a satem IE word does in this quarter of Europe. I wouldn't say it's impossible, but I have my doubts.

[Talskubilos]

It may indeed be the case that Sardinia is named for the Š3rdn, but I don't see what a satem IE word does in this quarter of Europe.

I've already quoted Basque (Biscayan) sarda 'fish school' from the same root, but surely there're more examples of IE-satem loanwords in Western and Central Europe.

[Talskubilos]

An example of an Iberian-Latin bilingual text would be the mortar stamps from Azaila (Teruel):

Latin text: Protemus fecit
Iberian text: Borotenbo denin

From this, we gather the verbal form denin '(he) made'. By contrast, the verb 'to make' in Basque is egin. Quite similar, but by no means identical.

Other interesting inscriptions on mosaics are these ones:

Likine abuloŕaune ekien Bilbiliaŕs (Andelos)
Likinete ekiar Usekeŕteku (Caminreal)

The anthroponym Likine is the rendering of Latin Licinius, while Bilbili and Usekeŕte are names of cities. The segment eki- (or egi- due to the ambiguity of the Iberian Levantine script) has been often interpreted as a form of the verb 'to make' (cfr. Basque egin), but IMHO it should be translated as 'workshop', corresponding to Basque (t)egi. Apparently, this is a loanword from IE *(s)teg- 'to cover', possibly Celtic. As a matter of fact, it seems thematic -o- was adapted as -i in Paleo-Basque and Iberian and the femenine suffix -ā as -e.

[Talskubilos]

The etymology of some pre-Latin loanwords in Romance languages is quite interesting. For example, Spanish barranco 'ravine' is related to Greek pháranks id., purportedly from IE *bʰerH- 'to pierce, to strike'. Also páramo 'moor' would related to Greek palámē 'palm, hand' < IE *plh₂-m-, with rhotacism.

On the other hand, Spanish tonto 'dumb', and secondarily 'cloudy (said of whether)', would a reduplicated output of IE *temH- 'dark'. So it looks like IE languages were spoken in the Iberian Peninsula in Antiquity.

[Znex]

The etymology of some pre-Latin loanwords in Romance languages is quite interesting. For example, Spanish barranco 'ravine' is related to Greek pháranks id., purportedly from IE *bʰerH- 'to pierce, to strike'. Also páramo 'moor' would related to Greek palámē 'palm, hand' < IE *plh₂-m-, with rhotacism.

On the other hand, Spanish tonto 'dumb', and secondarily 'cloudy (said of whether)', would a reduplicated output of IE *temH- 'dark'. So it looks like IE languages were spoken in the Iberian Peninsula in Antiquity.

Not a controversial hypothesis. We already know about Celtiberian and other Celtic languages that were spoken in Iberia, and Lusitanian is likely Para-Italo-Celtic (for one, it would seem to preserve *p).

Intervocalic lenitions are indeed common enough; and the Western Romance lenition is different enough from the British Celtic one (the biggest difference being that in Western Romance, no initial mutations emerged through it), which in turn is different enough from the Irish lenition (where intervocalic voiceless stops gave voiceless fricatives rather than the British and Romance voiced stops).

This is sorta necro-posting, but I must really question the common argument that two different languages affected by a common wave of change must change in exactly the same way. Even languages that we know changed in tandem with each other (eg. in the Balkan sprachbund) did not undergo the exact same sound changes, because the focus was the end product.

eg. for Albanian vs. Romanian

Code: Select all

/e/
Romanian: e > ia / _a; otherwise e > ie
Albanian: e > ia / _i; otherwise e > ie

This would be the exact sound change but for the fact that each language uses completely different environments! But this can be explained if we consider that Albanian underwent early final vowel reduction and eventual loss; the conditioning feature observed would instead be more similar to a schwa /ə/ than any distinct vowel. The closest Latin equivalent to schwa was identified as /a/, so this became the new condition.

We can assume similarly for Latin vs. Insular Celtic. In fact there is evidence that Gaulish stop consonants were already perceived differently to Latin stop consonants or even Greek consonants: voiceless Gaulish stops were often but variably inscribed with aspirate letters (Latin: PH TH CH; Greek: Φ Θ Χ) rather than with tenuis letters, while voiced Gaulish stops were also variably inscribed (evidence for early voicing lenition exists, eg. written omission of /g/ or replacement with /j/).

Why Brittonic and Goidelic emerged differently can be explained by their speakerbase: Brittonic speakers included not only native speakers but also many Romano-British refugees (many of whom likely spoke British Romance) from the Saxon conquest, while Goidelic speakers included mainly native speakers and native Irish adoptees of the language. The different social dynamics certainly pushed the languages to change in unexpected but not unpredictable or random ways.

(the biggest difference being that in Western Romance, no initial mutations emerged through it)

Not identical, but there is actually evidence that initial fortition occurred in tandem with intervocalic lenition in Western Romance, in common with Celtic (especially provable for sonorants). Note here that fortis phonation was early on conflated with palatalisation (since palatalised stops did not undergo voicing lenition):

Code: Select all

m unchanged (note that /mm/ was not nearly as common in Latin as other geminates, or as in Celtic)
n: initial and geminate n > ń (in Ibero-Romance, particularly Astur-Leonese, which we know emerged in Celtiberian-speaking areas)
   intervocalic n > Ṽ~Ø (in Gascon and Portuguese)
   intervocalic n > r (in some Franco-Provençal)
r: initial and geminate r > trilled r (general Western Romance; note how initial trilled r is absent from other Romance branches) > ʁ (Oïl, Portuguese)
   intervocalic r > ɾ (general Western Romance)
   intervocalic r > Ø (Gallo-Italian)
   intervocalic r > z (northern Oïl; otherwise merges with ʁ as in standard Oïl French)
l: initial and geminate l > ĺ (in general Ibero-Romance, particularly Astur-Leonese and Catalan); l > ɖ? > r (in Gascon)
   intervocalic l > Ø (Portuguese)
   intervocalic l > ɫ~w (general Gallo-Romance)
   intervocalic l > r (Gallo-Italian)
w: initial and geminate w (> v) (occurs early generally) > b (occurs notably in Castellano and Astur-Leonese); later loaned w > g(w)
   intervocalic w either forms diphthongs or later disappears altogether
j: initial and geminate j > dʒ (occurs early generally) > ʝ/ć (in Castellano and Astur-Leonese)
   intervocalic j either forms diphthongs or later disappears altogether
s: initial and geminate s > S = s̺ (general Western Romance; distinct from assibilated s)
   later final S > h > Ø (general Western Romance and ongoing in Spanish dialects; obscured by later voicing and merging)
   intervocalic s > z̺ (general Western Romance; distinct from assibilated z; obscured by later devoicing and merging)

in contrast to the Insular Celtic:

Code: Select all

m: initial and geminate m merge > m
   intervocalic m > ṽ (general Celtic; preserved in Breton and Old Irish) > v~w
n: initial and geminate n > N (general Celtic; preserved in Goidelic, and Cornish and Breton)
   intervocalic n remains
r: initial and geminate r > R (general Celtic) > r̥ (Welsh)
   intervocalic r > ɾ (Goidelic; preserved in Scottish Gaelic)
l: initial and geminate l > L (general Celtic) > l̻ = ɬ (Welsh)
   intervocalic l remains
w: initial and geminate w > W (general Celtic); W > gw (Brittonic); W > f (Goidelic)
   intervocalic w > β > Ø (Goidelic)
j unchanged (initial j or /jj/ is not nearly as common as other geminates, or as in Latin)
s: initial and geminate s > S (general Celtic); initial s > h (Brittonic partially; otherwise s merges with st > s)
   intervocalic s > h > Ø (general Celtic)

[Talskubilos]

We already know about Celtiberian and other Celtic languages that were spoken in Iberia, and Lusitanian is likely Para-Italo-Celtic (for one, it would seem to preserve *p).

I'd classify Lusitanian as P-Italic, albeit with a Celtic (Gallaecian) superstrate. On the other hand, I don't think Italo-Celtic existed at all, but rather Proto-Celtic had an Italic substrate/adstrate.

[Nortaneous]

That's a lot of adstrates!

Then again, it's possible for languages to have adstrates that don't show up that much - you could make a case for [separate?] German and Yiddish substrates in American English, but for German you have a handful of constructions and possibly some phonetic detail, and for Yiddish you mostly have lexical material.

If American English had been spoken 2000 years ago, it's unlikely that the Yiddish influence would be at all linguistically recoverable - the Yiddish loanwords in common use are generally in the semantic field of complaints and insults, which don't tend to be recorded well.

(Speaking of gaps, I found recently that Adams's dictionary of Tocharian B doesn't include any non-euphemistic words for "penis", since tso has been reinterpreted to probably have meant "abdomen". They must have had one, but it wasn't recorded.)

[WeepingElf]

The etymology of some pre-Latin loanwords in Romance languages is quite interesting. For example, Spanish barranco 'ravine' is related to Greek pháranks id., purportedly from IE *bʰerH- 'to pierce, to strike'. Also páramo 'moor' would related to Greek palámē 'palm, hand' < IE *plh₂-m-, with rhotacism.

On the other hand, Spanish tonto 'dumb', and secondarily 'cloudy (said of whether)', would a reduplicated output of IE *temH- 'dark'. So it looks like IE languages were spoken in the Iberian Peninsula in Antiquity.

Not a controversial hypothesis. We already know about Celtiberian and other Celtic languages that were spoken in Iberia, and Lusitanian is likely Para-Italo-Celtic (for one, it would seem to preserve *p).

Yep. Celtiberian and Gallaecian were Celtic languages, some say that Tartessian was one too, but that proposal is controversial; Lusitanian also was IE, though the retention of PIE *p speaks against a Celtic affiliation (but perhaps Proto-Celtic *f was secondarily hardened again to p?). The whole western half of the Iberian Peninsula is full of Celtic (or maybe "Para-Celtic") place names. And the cognate of Greek pháranks in Celtic would be *baranks, which could be the source of Spanish barranco.

But there almost certainly was no satem IE language so far in the west, as T. claims as the origin of the name of Sardinia!

Intervocalic lenitions are indeed common enough; and the Western Romance lenition is different enough from the British Celtic one (the biggest difference being that in Western Romance, no initial mutations emerged through it), which in turn is different enough from the Irish lenition (where intervocalic voiceless stops gave voiceless fricatives rather than the British and Romance voiced stops).

This is sorta necro-posting, but I must really question the common argument that two different languages affected by a common wave of change must change in exactly the same way. Even languages that we know changed in tandem with each other (eg. in the Balkan sprachbund) did not undergo the exact same sound changes, because the focus was the end product.

Indeed, they are not exactly the same, only similar, and we don't know whether these have a common origin or not. There are, as you say, lots of differences in the details.