Celtic and Pre-Germanic

[Taranis]

A while back, Maciamo posted this thread about a relationship between the Celic and the Italic peoples. What I'm going to do here is a linguistic consideration about the relationship between the Celtic and the Germanic peoples.

- Celtic name evidence in Germany (or, I should say, Germania Magna as it was perceived by the Romans - that is the lands east of the Rhine and north of the Danube - extending up to the Vistula which marked the border to Sarmatia) extends approximately to the Main river and from there to Silesia. Ptolemy in the 2nd century AD lists plenty of town names in these areas with readily identifiable Celtic etymologies - at a time when these areas when the Germanic peoples already lived as far south as the Danube - thereby being the testimony of a strong underlying Celtic presence in southern Germania. A few of these towns actually still exist today: Tarodunon (latinized as "Tarodunum") became Zarten in the Black Forest, Eburodunon (latinized as "Eburodunum") is Brnno/Brünn in the Czech Republic.

However, in the northern areas of Germania (with exception of the Rhine delta), Celtic place names are absent, even if archaeologically Hallstatt/Le-Tene extended into these areas. To find a (possible) answer for this discrepancy, I would like to explore some possibilities here:

- As I already brought up in a different thread a few weeks back, in some of it's vocabulary, Germanic is decisively closer to Balto-Slavic than to Italo-Celtic. For instance it's silver and gold (I've been taking various branches of the various families - as well as outgroups - in order to get a representative image here):

English - Silver - Gold
German - Silber - Gold
Danish - Sølv - Guld
Gothic - Silubr - Gulth
Lithuanian - Sidabras - (Auksas)
Latvian - Sudabra - Zelta
Bulgarian - Srebro - Zlato
Czech - Stribrna - Zlato
Russian - Serebra - Zolota
Thracian - (unattested) - Saldas

Latin - Argentum - Aurum
Irish - Airgead - Or
Welsh - Arian - Aur
Gaulish - Argenton - (*Auron - reconstructed)
Celtiberian - Arkanta - (unattested)
Albanian - Argjend - Ar

(note that I added Albanian and Thracian here just for the sake of completeness - neither are obviously Italo-Celtic nor Balto-Slavic, but words in these languages obviously have cognates, respectively).

(also note that Balto-Slavic changes proto-Indo-European initial *g´h to *z, whereas Germanic changes it to *g).

We might speculate that these Germanic/Balto-Slavic commonalities stem from the Germanic legacy of the Battle Axe Culture, but since that would get off-topic I will leave it at that.

- On the flip side, the Germanic languages have a considerable number of borrowings from the Celtic languages. In particular, they must have occured before in particular the shift of Initial K to H occured in Common Germanic. To pick a few signature examples:

- The Celtic tribal name "Volcae" yields "Walha" ('foreigner' - which we find today in place names like "Wales", "Wallonia" and "Wallachia").

- The word for "steed":
Gaulish - Marcos
Welsh - March
Breton - Marc'h
German - Mähre
English - Mare
Swedish - Märr

- The word for "iron":
English - Iron
German - Eisen
Swedish - Järn
Gothic - (E)isarn
Gaulish - Isarnos
Welsh - Haern
Breton - Houarn
Irish - Iarann

(note the striking similarity between the Gothic and Gaulish words for "Iron")

Since we know from archaeology the rough time of when iron-working from the Celtic Hallstatt Culture reached northern Germany, we can date the timing of this Celtic-Germanic contact to roughly 600-500 BC. We also can infer that the K to H shift must have occured obviously later. But the question is, however, how much later?

Rome's first contact with Germanic peoples is the invasion of the Cimbri and Teutones in the late 2nd century BC. However, to us these have seemingly overtly Celtic names, in particular their leaders: "Boiorix" ("Cattle King") would easily make sense as a Gaulish or Galatian name. However, the tribal names, which are exclusively given as "Cimbri et Teutonibus" or "Cimbri Teutonique". According to Germanic sound laws, Initial *k should yield *h via the intermediate stage of *x. Likewise, initial *t should yield *θ. Therefore, we should expect the spellings "Chimbri" and "Theutones". Since we are not seeing them, that these changes must have occured even later. This means, by the 2nd century BC, we are technically not talking about a "Proto-Germanic" language but a "Pre-Germanic" one. Specifically, "Pre-Germanic" would be defined as the language that was ancestral to "Proto-Germanic" (more appropriate would be "Common Germanic", that is, the ancestor language of all later Germanic languages).

As stated above with the personal names, these names strike us as surprisingly Celtic. Now, if we account for the absence of sound changes, it's possible to reconstruct this "Pre-Germanic" stage, as examplified by the word for "tribe" or "people":

Singular

Case - Gaulish - Pre-Germanic - Gothic
Nominative - Touta - Teuta - Θiuda
Genitive - Toutas - Teutas - Θiudos
Dative - Toutai - Teutai - Θiudai
Accusative - Toutan - Teutan - Θiuda

Plural

Case - Gaulish - Pre-Germanic - Gothic
Nominative - Toutas - Teutas - Θiudos
Genitive - Toutanom - Teuton - Θiudo
Dative - Toutabo - Teutamis - Θiudom
Accusative - Toutas - Teutas - Θiudos

Note that with exception of genitive and dative plural, these forms are near-identical. There's a few uncertainties to be considered however:

Gothic had a vestigial vocative case, and Gaulish had, in addition to a vocative, also an instrumental and a locative case. It stands to reason that Pre-Germanic had (at least) a vocative case, too, and perhaps more cases, but this is speculation. In any case, for the other cases, there is a striking similarity, and as a result, in my opinion, it is hence likely that the Pre-Germanic language and the Celtic languages (ie, Gaulish - and related dialects/languages spoken by the Hallstatt/La-Tene peoples) probably had a considerably degree of mutual intelligibility. This, in a way, mirrors the similarities of Gaulish and Latin, and could explain the archaeological/linguistic discrepancies. It would also account for Caesar's account of the so-called "Germanic" Belgae (which - for the greater part, all bear overtly Celtic - that is, essentially Gaulish - tribal and personal names): if these crossed the Rhine in earlier times, it would have been relatively easy for them to adopt the Gaulish language.

This brings us to the question of when the sound shifts (note that this actually concerns an entire inventory of sound laws, which was first formulated by Jakob Grimm in the early 19th century as "Grimm's Law") actually take place? The answer must be: even later.

Indeed, Tacitus (late 1st century AD) gives us, amongst others, the following tribal names:

- Chatti
- Chamavi
- Cherusci

This means, by this time, the shift already occured. Which means that we can roughly narrow it down to between the 1st century BC and the 1st century AD. The major event that takes place in this time span is the Roman conquest of Gaul, and the expansion of the Roman sphere to the Rhine (and for a brief period, beyond), as well as the effective end of the continental Celtic peoples as a separate culture. One can speculate now that the disruption of trade routes, and the disconnection of trade routes between Gallia and Germania also triggered a linguistic disjunction between the Celtic and Germanic peoples, and that these drastic changes, which brought about the language shifts that brought about the Common Germanic language in the wake of this.

While this may sound a tad unlikely at first glance, one must consider that for instance Irish language made similarly drastic changes in a very short time frame between roughly 7th to 8th century AD), as it attested by the massive differences between Primitive Irish (the language used in the Ogham inscriptions - which in many aspects is much closer to Gaulish than to modern Irish) and the later Old Irish of the early medieval times.

[Taranis]

What I wanted to add, etymologically, is a very ambiguous case of name:

the "Marcomanni". "Marco-" may either derive from Celtic "Markos" (mare, steed) or Germanic "Mark" (frontier/boundary), and it would make perfectly sense for the Marcomanni to be both "horse men" or "frontier men".

This is further complicated as "Marobuduus" (actually spelled as "Marobvdvvs") makes sense as Gaulish "Marobodwos" ("Great Raven").

[Maciamo]

Great post, Taranis. It's very clear and well summarised.

[Dagne]

I just couldn't help noticing that word "people" in modern Lithuanian is very similar, too. Inasmuch as the Lithuanian is not a germanic language, the similarity must be somewhere on Proto - Indoeuropean level :

Singular

Case - Gaulish - Pre-Germanic - Gothic- Lithuanian
Nominative - Touta - Teuta - Θiuda- Tauta
Genitive - Toutas - Teutas - Θiudos - Tautos
Dative - Toutai - Teutai - Θiudai - Tautai
Accusative - Toutan - Teutan - Θiuda - Tautą

Plural

Case - Gaulish - Pre-Germanic - Gothic - Lithuanian
Nominative - Toutas - Teutas - Θiudos - Tautos
Genitive - Toutanom - Teuton - Θiudo - Tautų
Dative - Toutabo - Teutamis - Θiudom- Tautoms
Accusative - Toutas - Teutas - Θiudos - Tautas

[Taranis]

That's quite fascinating! I'm also quite fascinated that modern Lithuanian retains such a complex declension system (which has - across the spectrum - been abandoned in many modern IE languages) - Lithuanian seems to be in line with Gaulish, Latin, Greek or Sanskrit in that respect.

[Dagne]

yes, Lithuanian is rather conservative in changing its grammar and vocabulary. Which leads to a thought that when comparing languages one must take into consideration the overall "speed of change" in a language, and preferably compare the totality of a text. Otherwise it might seem that Lithuanian and, for instance, Gaulish are very much related, which is not realy the case... because they may be similar in some words and grammar because both languages retained a lot of older proto-indo-european features.

[Taranis]

yes, Lithuanian is rather conservative in changing its grammar and vocabulary. Which leads to a thought that when comparing languages one must take into consideration the overall "speed of change" in a language, and preferably compare the totality of a text. Otherwise it might seem that Lithuanian and, for instance, Gaulish are very much related, which is not realy the case... because they may be similar in some words and grammar because both languages retained a lot of older proto-indo-european features.

What is clear is that the Baltic and Slavic languages are closer related with each other than they are with the other IE languages due to a considerable number of common sound laws (a common Balto-Slavic stage, along with a common Indo-Iranic stage, is probably the most undisputed higher-level relationship inside IE), but when Common Balto-Slavic was spoken is hard to say.

One other aspect is that languages do not change at a constant speed: as seen from numerous examples, languages can remain fairly conservative over long stretches, while they also can make tremendous changes over rather short time frames. This is why attempts of so-called glottochronology have been largely futile, and why estimates for various proto-languages can be very wrong. For example, Forster and Toth came up with Proto-Celtic being spoken 10,000 to 8,000 years ago, which is totally unrealistic if you consider the relationship to other, closely-related IE languages (Lusitanian, the Italic and the Germanic families).

[Dagne]

I was also wondering about somewhat paradoxical development in languages from more complex and precise to more simplified ... The fact is not easily explained as our shrinking brains ...

[Taranis]

This is a good question. Generally, what is noticable is that one language which is extremely simplified is English, which has done almost entirely away with declensions except for a few vestigial remnants. One reason why this is so extremely pronounced in English is that it received influence from so many languages. Old Anglo-Saxon was still a proper Germanic language, but it received such an influx of words from Latin and French that keeping up such a fairly complex declension system was probably rather impractical.

So, the degree of simplification varies quite a bit across the spectrum of Indo-European languages - as examplified by your own mother tongue, Dagne. Also, this does not apply to non-IE languages in the way it applies to IE.

[Haganus]

But celtic names of leaders of the tribes of Cimbri and Teutones do not mean
that these tribes were celts. Foreign names for leaders gave them "prestige".
For example in the 18e century many members of the European high society
had French Christian names and spoke French. They remained their original
nationality. When was it not possible that the same thing has happenend
with the Cimbric and Teutonic tribes?

[Taranis]

But celtic names of leaders of the tribes of Cimbri and Teutones do not mean
that these tribes were celts. Foreign names for leaders gave them "prestige".
For example in the 18e century many members of the European high society
had French Christian names and spoke French. They remained their original
nationality. When was it not possible that the same thing has happenend
with the Cimbric and Teutonic tribes?

I didn't imply that these tribes were Celts. Read closely what I wrote. My point is, the Germanic language before the major sound shifts (the "Pre-Germanic" stage, if you will) was significantly closer to the Celtic languages and probably had a fair degree mutual intelligibility with them. In so far, what you are suggesting, that the Germanic tribes picked Celtic names out of prestige would make even more sense.

Germanic sound laws are very well-established, if you will I can type down some examples to better visualize what exactly happened.

[spongetaro]

I wonder if there is something like intermediate languages between the Germanic language and Italo-Celtic languages

[Taranis]

I wonder if there is something like intermediate languages between the Germanic language and Italo-Celtic languages

Hard to say. If there was, it's probably unattested. In a way, you might argue that Germanic in itself is an intermediate language between Italo-Celtic and Pre-Balto-Slavic. I have to say "Pre-" Balto-Slavic because the Baltic and Slavic families are closer to each other than they are to Germanic because they share a considerable number of common innovations absent in Germanic, therefore it's logical that the Germanic words of Balto-Slavic origin must have entered into Common or Pre-Germanic vocabulary before the Balto-Slavic sound shifts. What woud be interesting is to actually quantify the amount of vocabulary there.

[Kentel]

Very interesting !

However, in the northern areas of Germania (with exception of the Rhine delta), Celtic place names are absent, even if archaeologically Hallstatt/Le-Tene extended into these areas.

First, it is very important to emphasize the fact that their is no obligatory connection between an archaeological culture and a language family. It has been said hundreeds of times, but it is still crucial. La Tène and Halstatt are considered Celtic on the basis of a very thin argumentation, and several "laténiennes" areas are obviously not Celtic (eg. in Hungary).

However, a few Celtic ethnonyms and place-names can be found in southern Jutland and Northern Germany : Ampsivarii and Chauci, the river name Amisia, the place name Abalus, the controversed ethnonym Teutones and some others. Their celticity is not 100% certain (see Sims-Williams : Ancient Celtic Place-Names in Europe and Asia Minor, Blackwell 2006).

- On the flip side, the Germanic languages have a considerable number of borrowings from the Celtic languages. In particular, they must have occured before in particular the shift of Initial K to H occured in Common Germanic. To pick a few signature examples:

- The Celtic tribal name "Volcae" yields "Walha" ('foreigner' - which we find today in place names like "Wales", "Wallonia" and "Wallachia").

In your example Volcae/Walha, *k has already undergone the spirantisation of the PIE velars predicted by Grimm's Law. Could you give another example ? There are very few cases of germanic-celtic words attested in both languages dating back to 600 BC. My guess would be that if a word has not been submitted to Grimm's Law, then it is not Germanic.

As stated above with the personal names, these names strike us as surprisingly Celtic.

Maybe because they are !

Now, if we account for the absence of sound changes, it's possible to reconstruct this "Pre-Germanic" stage, as examplified by the word for "tribe" or "people":

Singular

Case - Gaulish - Pre-Germanic - Gothic
Nominative - Touta - Teuta - Θiuda
Genitive - Toutas - Teutas - Θiudos
Dative - Toutai - Teutai - Θiudai
Accusative - Toutan - Teutan - Θiuda

Plural

Case - Gaulish - Pre-Germanic - Gothic
Nominative - Toutas - Teutas - Θiudos
Genitive - Toutanom - Teuton - Θiudo
Dative - Toutabo - Teutamis - Θiudom
Accusative - Toutas - Teutas - Θiudos

Note that with exception of genitive and dative plural, these forms are near-identical. There's a few uncertainties to be considered however:

(For Gaulish I would say respectively : sing: touta - toutias - touti - toutin/im) If you compare with Old Greek, you will see that the endings are very similar too.

The Pre-Germanic language and the Celtic languages (ie, Gaulish - and related dialects/languages spoken by the Hallstatt/La-Tene peoples) probably had a considerably degree of mutual intelligibility. This, in a way, mirrors the similarities of Gaulish and Latin, and could explain the archaeological/linguistic discrepancies. It would also account for Caesar's account of the so-called "Germanic" Belgae (which - for the greater part, all bear overtly Celtic - that is, essentially Gaulish - tribal and personal names): if these crossed the Rhine in earlier times, it would have been relatively easy for them to adopt the Gaulish language.

That's interesting, but the core Germanic and Celtic lexicons have both many pre-IE items that would have made the mutual understanding completely impossible. The same for Celtic and Italic. Moreover, you assume that Gaulish and Pre-Germanic have been spoken at the same time, while it would be more logical to assume that Pre-Germanic was spoken at the same time than pre- or at least proto-Celtic. In fact, the closer you go to PIE, the more the IE languages are similar to each others.

While this may sound a tad unlikely at first glance, one must consider that for instance Irish language made similarly drastic changes in a very short time frame between roughly 7th to 8th century AD), as it attested by the massive differences between Primitive Irish (the language used in the Ogham inscriptions - which in many aspects is much closer to Gaulish than to modern Irish) and the later Old Irish of the early medieval times.

That's right. Now the question is : why ? If, on the contrary, you compared archaic Greek (Mycenean) with modern Greek, you would be surprised by the phonetic stability through a time span of 3500 years.

[Taranis]

Very interesting !

Welcome.

First, it is very important to emphasize the fact that their is no obligatory connection between an archaeological culture and a language family. It has been said hundreeds of times, but it is still crucial. La Tène and Halstatt are considered Celtic on the basis of a very thin argumentation, and several "laténiennes" areas are obviously not Celtic (eg. in Hungary).

While I agree that archaeological cultures do not automatically reflect linguistic homogenity, the case for Hallstatt/La-Tene being associated with predominantly Celtic-speaking peoples can be made. This is not a "very thin" argumentation, but it is clear that there are (I would say, almost exlusively) Celtic place names in the Hallstatt/La-Tene core areas (eastern France, southern Germany, Austria, Bohemia).

Also, if we ask the other way around, the question would be: what else should Hallstatt and La-Tene be linguistically, if not Celtic?

However, a few Celtic ethnonyms and place-names can be found in southern Jutland and Northern Germany : Ampsivarii and Chauci, the river name Amisia, the place name Abalus, the controversed ethnonym Teutones and some others. Their celticity is not 100% certain (see Sims-Williams : Ancient Celtic Place-Names in Europe and Asia Minor, Blackwell 2006).

"Chauci" doesn't sound exactly Celtic to me, it's clearly shifted according to Grimm's Law.

The controversial area I was thinking about primarily was the approximate area of modern-day Hesse and Thuringia: there are a lot of sites from the Hallstatt period from this area, but centuries later, Celtic name evidence does not extend further north than the Main river, and not further east than the Rhine.

Still, Ptolemy (in the mid second century AD) lists nearly 80 town names in Germania Magna, approximately a sixth of which have overtly Celtic etymologies.

In your example Volcae/Walha, *k has already undergone the spirantisation of the PIE velars predicted by Grimm's Law. Could you give another example ? There are very few cases of germanic-celtic words attested in both languages dating back to 600 BC. My guess would be that if a word has not been submitted to Grimm's Law, then it is not Germanic.

I think you're making a wrong assumption here. Grimm's Law probably only occured in the 1st century BC to 1st century AD. This means the "Pre-Germanic" sound stage lasted across a very long time. If a word is not submitted to Grimm's Law, it must obviously have entered the vocabulary at a later point.

The earliest point of Celtic/Germanic language contact that can be agreed on is approximately the 6th century BC (the time that the Jastorf Culture adopted iron-working from the Hallstatt Culture). It is reasonable to assume that Grimm's Law occured after that date (that is the general consensus), and it's absolutely unreasonable to assume Grimm's Law occured before the 6th century BC.

You can further make the case that if "Cimbri" is a genuinely Germanic name, that the Cimbri spoke a late form of Pre-Germanic in the 2nd century BC. If Cimbri is genuinely Germanic, it should be shifted to "Chimbri" or "Himbri" according to Grimm's Law. Likewise, it should be "Theudones" and not "Teutones".

Anyways, another example of a Celtic borrowing into Germanic would be the word "Marcos" (horse):

Gaulish "Marcos"
Welsh "March"
Breton "Marc'h"

Gothic "Marhs"
German "Mähre"
Dutch "Merrie"
English "Mare"
Swedish "Märr"

As you can see, the word is attested in all branches of Germanic (ie East Germanic, West Germanic and Scandinavian - meaning it was present in Proto-Germanic) and applies to Grimm's Law, suggesting it entered Proto-Germanic vocabulary before Grimm's Law.

Maybe because they are !

Some of them would also make sense as Pre-Germanic. On the other hand, it's clear that the Germanic peoples adjacent to Celtic areas were heavily Celtic-influenced.

(For Gaulish I would say respectively : sing: touta - toutias - touti - toutin/im) If you compare with Old Greek, you will see that the endings are very similar too.

This is correct.

That's interesting, but the core Germanic and Celtic lexicons have both many pre-IE items that would have made the mutual understanding completely impossible. The same for Celtic and Italic. Moreover, you assume that Gaulish and Pre-Germanic have been spoken at the same time, while it would be more logical to assume that Pre-Germanic was spoken at the same time than pre- or at least proto-Celtic. In fact, the closer you go to PIE, the more the IE languages are similar to each others.

It is very much logical to assume that early Gaulish and a late form of Pre-Germanic (I must admit the term "Pre-Germanic" is somewhat confusing, in fact "Proto-Germanic before the First Germanic Sound Shift" would be more accurate, but it's obviously unwieldy) were spoken at the same time (because Grimm's Law applied only later, as I elaborated this above). I absolutely agree however that the farther you go back, the more similar the languages are. But, it's absolutely unreasonable to assume that Grimm's Law occured early.

You also have to consider that Gaulish was relatively close to Proto-Celtic itself (much, much closer than any of the modern Celtic languages). It is very clear though, however, that the splitting of the Celtic languages occured earlier than the splitting of the Germanic langauges.

That's right. Now the question is : why ? If, on the contrary, you compared archaic Greek (Mycenean) with modern Greek, you would be surprised by the phonetic stability through a time span of 3500 years.

Well, we obviously don't know. But we know that languages must have done this because this is what is historically transmitted.

[Kentel]

Thank you for your answer; this is very interesting and raises many questions.

Well, we obviously don't know. But we know that languages must have done this because this is what is historically transmitted.

I think that the reason why Greek underwent almost no changes (at least phonetically) during 3 millenaries is the fact that there was no interference of a substratum during 3500 years. I simply don't believe in the structuralist theory which says that languages evolve separately by the pure chance. The reality contradicts this mechanistic point of view.

French is the Romance language which has been the most altered, both phonetically and lexically in comparison with its Latin source, and I would bet that the very cause of the alteration is the Gaulish substratum.

The same for Irish : its deep alterations have probably been caused by a pre-IE substratum (you can read something about that in The Celtic Languages published by Routledge, I don't have the book to hand right now but I'll check tomorrow). And you could find many other examples like that (just think about Hittite, even Mallory does not reject a pre-IE Hatti influence).

I'll answer to the other problems you mentionned within a day or two.

[zanipolo]

I think that the reason why Greek underwent almost no changes (at least phonetically) during 3 millenaries is the fact that there was no interference of a substratum during 3500 years. I simply don't believe in the structuralist theory which says that languages evolve separately by the pure chance. The reality contradicts this mechanistic point of view.

French is the Romance language which has been the most altered, both phonetically and lexically in comparison with its Latin source, and I would bet that the very cause of the alteration is the Gaulish substratum.

not entirely correct, french language currently has a franckish ( germanic) element and while Latin /romance to a degree was not as latin as the original french
http://en.wikipedia.org/wiki/Occitan_language
which had more Latin influence


Greek survived also due to texts saved after the byzantine capture of the Ottomans
Bessarion's legacies was even more far reaching. In 1469 he presented his immense library of 800 volumes, many of them copied by the scribes, to the church of St. Mark in Venice. Bessarion chose Venice partly because he considered it the most politically stable and secure of the Italian city states, and partly because it had offered a refuge for so many of his fellow Greeks. With the advent of printing in the 1490s, the Venetian printer Aldus Manutius established a Greek press in Venice. The choice of location no doubt dictated by the availability of Greek texts in the library of St. Mark's, for Manutius based his editions on these books. Once again, however, obscure individuals played their part. Manutius made use of the large pool of native Greek speakers provided by the Greek community to assist him in the preparation of texts for publication. It was with the help of émigré Greeks like Demetrius Doukas and Mark Mousouros, that Manutius produced printed editions of nearly all the works of the major Greek authors of antiquity before 1515, thus ensuring their survival for posterity.
so, without the printing press, Greek language could have been deminished

[Taranis]

Thank you for your answer; this is very interesting and raises many questions.

I'm glad you appreciate it.

I think that the reason why Greek underwent almost no changes (at least phonetically) during 3 millenaries is the fact that there was no interference of a substratum during 3500 years. I simply don't believe in the structuralist theory which says that languages evolve separately by the pure chance. The reality contradicts this mechanistic point of view.

Actually, there is quite a bit of changes in Greek, both from Mycenean Greek to Classical Greek and from Classical Greek to modern Greek. They are obviously not quite as drastic as in other languages, but they exist. For instance, Mycenean Greek had "kw" where Classical Greek had "p" (interestingly, Greek did make the same shift there as the Osco-Umbrian languages and the P-Celtic languages). Mycenean Greek also had a "w" sound, which existed only vestigially in Classical Greek. From Classical Greek to modern Greek, one of the major changes, phonologically speaking, is the shift from the aspirated plosives (Th, Ph, Kh) to fricative sounds.

My point is, languages will change, no matter what. But, in regard for major shifts occuring in a short frame of time, I agree there must be a different mechanism.

French is the Romance language which has been the most altered, both phonetically and lexically in comparison with its Latin source, and I would bet that the very cause of the alteration is the Gaulish substratum.

Yes, there definitely is a Gaulish substratum in French. There's also a significant amount of vocabulary in French derived from Gaulish, but the Germanic (Frankish) influence on French is probably greater. One example that comes to my mind that illustrates this is the usage of the Uvular Trill in French as an "R" sound, which also exists in Standard German. The question is, is this sound Gaulish origin or Germanic in origin?

The same for Irish : its deep alterations have probably been caused by a pre-IE substratum (you can read something about that in The Celtic Languages published by Routledge, I don't have the book to hand right now but I'll check tomorrow). And you could find many other examples like that (just think about Hittite, even Mallory does not reject a pre-IE Hatti influence).

With Irish, the huge changes to the language all occured after the 6th century AD. If you take a look at the archaic Irish language as preserved in the Ogham inscriptions (circa 4th to 6th century AD), it was much more similar to Gaulish than to modern Irish, with all the sound changes that are typical of the Goidelic languages not having occured yet, as well as bearing a complex declension system (akin to Gaulish, but also Latin, Greek, etc.).

The idea that a pre-IE substratum influenced Irish is admittedly tempting. Especially, people sporadically brought up this idea of an Afro-Asiatic language being spoken in Britain before the arrival of the Celts, due to features such as VSO order (both Celtiberian and Gaulish as generally SVO) and inflected prepositions. However, it stands to reason that the features didn't develop until very late, thereby making the actual likelihood of these changes being due to pre-IE substrate somewhat unlikely.

I'll answer to the other problems you mentionned within a day or two.

Take your time.

[iapetoc]

I'm glad you appreciate it.



Actually, there is quite a bit of changes in Greek, both from Mycenean Greek to Classical Greek and from Classical Greek to modern Greek. They are obviously not quite as drastic as in other languages, but they exist. For instance, Mycenean Greek had "kw" where Classical Greek had "p" (interestingly, Greek did make the same shift there as the Osco-Umbrian languages and the P-Celtic languages). Mycenean Greek also had a "w" sound, which existed only vestigially in Classical Greek. From Classical Greek to modern Greek, one of the major changes, phonologically speaking, is the shift from the aspirated plosives (Th, Ph, Kh) to fricative sounds.

Thank you for your answer; this is very interesting and raises many questions.

I think that the reason why Greek underwent almost no changes (at least phonetically) during 3 millenaries is the fact that there was no interference of a substratum during 3500 years. I simply don't believe in the structuralist theory which says that languages evolve separately by the pure chance. The reality contradicts this mechanistic point of view.

how sure we are about that?
probably you mean about letter F or στ the digama διγαμα which became nkk cause the r ρ was a vowel while in modern is connected with l to λυγρα ορ υγρα,
according what I know 6 changes in Greek are
F digama stop exist and becomes γγ = νκ nk even in archaic
u υψιλον becomes i Hellenistic due to Ionic dialect
and ai oi ei ui drop from long vowel to ε i i i i Hellenistic due to Ionic or Roman
ευ is pronounced ef or ev unknown time (in fact could not be even a change)
η and ι become 1 to i (in south Greece some keep the ι pure and connected with η) Roman
and the koppa ς from ssz σσζ drops to ς s σ archaic times or Attic times (exist even today in mountain isolated areas of Aeolian or Makedonian)

are we sure that Greek had sound τχ πχ κχ in archaic? that puts greek to larrygeal λαρυγγικες glosses (slavic-germanic)
while greek was a lugre and dental language υγρη και οδοντικη had sounds like 'rr 'ρρ hρρ and ll
(even today Makedonians kept the ll)
while sounds like d b w were and are rare in greek (dental, not rino-dental d=δ is pure dental d=ντ is rino-dental (n is a nose sound))

the biggest problem in Greek language was the begining,
the Hellenistic
and the modern 1820-1920 AD

the first was due to create a common sound
the second was cause every one wanted to be teacher of greek for wealthy roman family,
Alexandreians solve it by using aspirations ' and tones '~
the third is even today a problem the kathareuiusa and demotiki
cause although speak same language every corner had borrowed words and sounds, and many words change, like today we name oplon not the shield but the gun, or the chliari and chliaro,
chliari means spoon from kochliario while chliaro means demi warm demi cold, etc

[Taranis]

Iapetoc, one thing you should consider is that Ph, Th and Kh in Greek generally correspond with Bh, Dh, Gh and Ǵh in Proto-Indo-European. Because Greek is a Centum language, it did merge Ǵh with Gh, but unlike the most other Indo-European languages (Centum or Satem), which basically de-aspirated Bh, Dh and Gh (to merge them with B, D and G, respectively), the Greek language instead devoiced these sounds, basically Bh became Ph (φ) , Dh became Th (θ) and Gh became Kh (χ). It therefore is absolutely necessary to assume that Mycenean Greek indeed did have these sounds, even if they are impossible to be transmitted via the Linear B script (on the other hand, Linear B verymuch could represent "Kw" and "W").

[iapetoc]

Iapetoc, one thing you should consider is that Ph, Th and Kh in Greek generally correspond with Bh, Dh, Gh and Ǵh in Proto-Indo-European. Because Greek is a Centum language, it did merge Ǵh with Gh, but unlike the most other Indo-European languages (Centum or Satem), which basically de-aspirated Bh, Dh and Gh (to merge them with B, D and G, respectively), the Greek language instead devoiced these sounds, basically Bh became Ph (φ) , Dh became Th (θ) and Gh became Kh (χ). It therefore is absolutely necessary to assume that Mycenean Greek indeed did have these sounds, even if they are impossible to be transmitted via the Linear B script (on the other hand, Linear B verymuch could represent "Kw" and "W").

correct, it is another story the PIE, and another mycenean, we are not sure if ph or Bh exist in mycenean,
I believe that they were already non working, as we also see in the romano-latin.

[zanipolo]

correct, it is another story the PIE, and another mycenean, we are not sure if ph or Bh exist in mycenean,
I believe that they were already non working, as we also see in the romano-latin.

http://www.ancientscripts.com/greek.html

[Kentel]

Actually, there is quite a bit of changes in Greek, both from Mycenean Greek to Classical Greek and from Classical Greek to modern Greek. They are obviously not quite as drastic as in other languages, but they exist.

This is my point : there are changes (in a time span of 3500 years the contrary would have been surprising), but, as you said, they are not drastic. There is no comparison with the huge amount of sound changes which occurred from Latin to Old French in no more than 5 centuries.

For instance, Mycenean Greek had "kw" where Classical Greek had "p" (interestingly, Greek did make the same shift there as the Osco-Umbrian languages and the P-Celtic languages). Mycenean Greek also had a "w" sound, which existed only vestigially in Classical Greek. From Classical Greek to modern Greek, one of the major changes, phonologically speaking, is the shift from the aspirated plosives (Th, Ph, Kh) to fricative sounds.

This change is probably mechanistic, I agree. As for the question of the PIE labiovelar in Greek, Oscan and P-Celtic, I have no opinion yet, but this is undoubtedly one of the most fascinating problems in the field of historical phonetics.

Yes, there definitely is a Gaulish substratum in French. There's also a significant amount of vocabulary in French derived from Gaulish, but the Germanic (Frankish) influence on French is probably greater.

I wouldn't say that; several words considered by the etymological dictionaries (TLF and FEW) as Germanic may be in fact Celtic, and many words considered as "Dutch" are in fact not. Just one example : the word "chouette" (owl) is supposedly from old Frankish *kawa (which incidentaly designates a completely different bird), while there is an attested Gaulish caouanos (kaouenn in Breton). The same for blet and blesser, suposedly from Old Frankish *bleizza, saule from *sahla, and others. I am currently writing a paper about that.

One example that comes to my mind that illustrates this is the usage of the Uvular Trill in French as an "R" sound, which also exists in Standard German. The question is, is this sound Gaulish origin or Germanic in origin?

Very good question. I don't know . I would say it is originally Celtic (don't believe I am a Celtomane : I am definitely not !) because I find it hard to believe that the superstratum language of an aristocratic elite could influence the phonetics of a language so deeply (the change of articulation is huge in this case). Phonetics is mostly inherited from substrata.

However, this is not a definitive opinion (in fact I have no definitive opinion at all) and it could be originally Germanic. In Breton you have the same uvular R as in French and Germanic. Is it inherited from Gaulish or is it a borrowing from French or Germanic ? I asked to old peoples in Brittany of they could remember anybody using another sound than the uvular one and the answer was no. On the other hand, in Cornish you have an alveolar flap (as in English) which is probably a borrowing from English, thus the Breton R would be a borrowing from French, and that would contradict my first point. But a borrowing from when ? In Old French, you have (supposedly) an alveolar trill, as in Spanish...

Or maybe the English alveolar flap r is substratic, that's why you have it in English and Cornish (not in Welsh though), and maybe the french R is substratic too. I don't exclude any hypothesis. At the moment, I just don't know.

An why do you have a trill in Norwegian (not in Stavanger though... nothing is simple...).
in Swedish and in Icelandic ? Even in Faroese they have a trill...

With Irish, the huge changes to the language all occured after the 6th century AD. If you take a look at the archaic Irish language as preserved in the Ogham inscriptions (circa 4th to 6th century AD), it was much more similar to Gaulish than to modern Irish, with all the sound changes that are typical of the Goidelic languages not having occured yet, as well as bearing a complex declension system (akin to Gaulish, but also Latin, Greek, etc.).

You certainly know this theory according to which the first Celtic language spoken in Ireland was P-Celtic...

The idea that a pre-IE substratum influenced Irish is admittedly tempting. Especially, people sporadically brought up this idea of an Afro-Asiatic language being spoken in Britain before the arrival of the Celts, due to features such as VSO order (both Celtiberian and Gaulish as generally SVO) and inflected prepositions. However, it stands to reason that the features didn't develop until very late, thereby making the actual likelihood of these changes being due to pre-IE substrate somewhat unlikely.

That's Vennemann's theory I guess ? Unfortunately, all the papers and books I've found by him were written in German, and I must confess that I cannot read German. If you know something in English, I would be glad to read it and to know more about his theory.

[Kentel]

not entirely correct, french language currently has a franckish ( germanic) element and while Latin /romance to a degree was not as latin as the original french
http://en.wikipedia.org/wiki/Occitan_language
which had more Latin influence

I wouldn't be so sure about that. The Visigothic Kingdom lasted 3 centuries, and this happened in Southern France.

I don't think you can borrow phonems so easily. As Sapir said "The highly significant thing about such phonetic interinfluencings is the strong tendancy of each language to keep its phonetic pattern intact". I don't think that French borrowed any phonem from German, nor from any superstratum.

But I may be wrong.

[Taranis]

This is my point : there are changes (in a time span of 3500 years the contrary would have been surprising), but, as you said, they are not drastic. There is no comparison with the huge amount of sound changes which occurred from Latin to Old French in no more than 5 centuries.

Yes, you have a point. This is quite a difference in the mode of language evolution. One must add, such sudden changes make any attempts to make an "absolute dating" are completely screwed up (I'm talking about so-called concept of "glottochronology", which you may have heard about, and which has produced bewildering results, most drastically the claim by Forster and Toth 2003 that Proto-Celtic diverged from PIE in 6000 BC! ). What is possible, though, as can be demonstrated, is that it's verymuch possible to do relative chronology in regard for whether words entered a language before or after a sound change.

This change is probably mechanistic, I agree. As for the question of the PIE labiovelar in Greek, Oscan and P-Celtic, I have no opinion yet, but this is undoubtedly one of the most fascinating problems in the field of historical phonetics.

Yes, this treatment in Greek, Osco-Umbrian and P-Celtic is certainly fascinating. Especially the question of why it happened there. Was there a common influence that caused this, or did this really happen independently?

I wouldn't say that; several words considered by the etymological dictionaries (TLF and FEW) as Germanic may be in fact Celtic, and many words considered as "Dutch" are in fact not. Just one example : the word "chouette" (owl) is supposedly from old Frankish *kawa (which incidentaly designates a completely different bird), while there is an attested Gaulish caouanos (kaouenn in Breton). The same for blet and blesser, suposedly from Old Frankish *bleizza, saule from *sahla, and others. I am currently writing a paper about that.

Yeah, you have a point there. The list is probably larger, and in addition I would think we must also not rule out the possibility of Gaulish loanwords in Old Frankish, to add confusion to the whole situation. Regarding Breton, one interesting question is if Breton has borrowings from Gaulish. I mean, I'm not necessarily arguing that Gaulish was still a living language at the time the Bretons arived in Aremorica (that would probably be a stretch to argue), because those words might have been transmitted to Breton via Vulgar Latin, instead.

Very good question. I don't know . I would say it is originally Celtic (don't believe I am a Celtomane : I am definitely not !) because I find it hard to believe that the superstratum language of an aristocratic elite could influence the phonetics of a language so deeply (the change of articulation is huge in this case). Phonetics is mostly inherited from substrata.

Yes, I absolutely agree there! It would seem far more likely for this to be inherited from the Gallo-Romans than from the Frankish aristocracy.

However, this is not a definitive opinion (in fact I have no definitive opinion at all) and it could be originally Germanic. In Breton you have the same uvular R as in French and Germanic. Is it inherited from Gaulish or is it a borrowing from French or Germanic ? I asked to old peoples in Brittany of they could remember anybody using another sound than the uvular one and the answer was no. On the other hand, in Cornish you have an alveolar flap (as in English) which is probably a borrowing from English, thus the Breton R would be a borrowing from French, and that would contradict my first point. But a borrowing from when ? In Old French, you have (supposedly) an alveolar trill, as in Spanish...

Or maybe the English alveolar flap r is substratic, that's why you have it in English and Cornish (not in Welsh though), and maybe the french R is substratic too. I don't exclude any hypothesis. At the moment, I just don't know.

An why do you have a trill in Norwegian (not in Stavanger though... nothing is simple...).
in Swedish and in Icelandic ? Even in Faroese they have a trill...

Well yeah, it gets confusing there. What has to be added is that the Uvular Trill is far from ubiquitous in German, because there's a number of dialects which have a very different "R" sounds: some southern dialects have the alveolar trill, and certain dialects in the west (the region around the town of Siegen, specifically) even have an alveolar approximant!

You certainly know this theory according to which the first Celtic language spoken in Ireland was P-Celtic...

Yes, I heard about that theory, but honestly, it makes much more sense to assume that the British Isles as a whole were originally Q-Celtic. In particular, a cognate of the word "Britain" (recorded as "Pritennike" in the ancient Greek sources, and also the Welsh word "Prydein") also exists in Irish as "Cruithne". Hence, the root word can be reconstructed as something akin to "Kʷritani". In my opinion, Britain was subsequently P-Celticized (or, participated in the P-Celtic sound shift, if you wish to call it that), whereas seemingly, Ireland was left out of this innovation.

However, what definitely is conceivable - even likely - is that there were P-Celtic peoples who arrived later in Ireland. Specifically, Ptolemy mentions a tribe in Ireland called the "Manapi", which sound very similar to the Menapi of Gallia Belgica, which lived in the region of Cassel (Nord-Pas-de-Calais, France). We know that Belgic tribes migrated into Britain (the Atrebates, for instance, are found on both sides of the Channel), so it's conceivable that Belgic tribes also migrated into Ireland.

That's Vennemann's theory I guess ? Unfortunately, all the papers and books I've found by him were written in German, and I must confess that I cannot read German. If you know something in English, I would be glad to read it and to know more about his theory.

Yes, Vennemann, amongst other people. But principally Vennemann, because he's been the loudest advocate of it. In regard for you being not able to read his works, I am kind of afraid to say that you have not missed much. Basically, he started out with a good idea but he interpreted too much into it, and much of his later works (for instance, he argues that the Phoenicians colonized the North Sea!) are outright crazy!

One problem is that Vennemann explicitly argues that there was a Semitic substrate on the British Isles. I mean, I could see how there might have been an Afro-Asiatic language in the Atlantic region, which may have arrived there in the Neolithic. But I genuinely doubt that any Semitic-speaking peoples reached the British Isles before the Phoenician traders who may have arrived there in the early 1st millennium BC.