Friday, December 16, 2022

Language Related, Again

Language Related · Who Can Understand Quenya and Sindarin? Anyone who takes the trouble to learn them · Language Related, Again · Language Related - It's Not Tedious for Me!

Often mentioned
or at least implied, in the following answers, is "Brugmann" - more precisely Karl Brugmann, and the work by him and Berthold Delbrück that is called

Grundriß der vergleichenden Grammatik der indogermanischen Sprachen

How do linguists use the comparative method to reconstruct Proto-Indo-European?

Hans-Georg Lundahl
amateur linguist
Let’s use Romance languages as a test case. In this case we know that Latin was the common mother language. I’ll take “ploughshare” and “water” and “milk” and “honey” in the order French, Italian, Spanish, Portuguese, Catalan, Romanian. What I reconstruct need not be Latin. It could be something from a few centuries after Latin (as we learn it).










prag / fier de plug



le acqua
las agua
a água / agua



le lait
il latte
la leche
o leite



le miel
le miele
la miel
o mel

Now, I’ll start by saying - ploughshare is basically a no-go, except the couple Spanish and Portuguese. For reja / relha we can reconstruct a common ancestor which was presumably something like *relya (I’ve just transscribed the Portuguese), and this became reyya, rezha, resha, rekha in Spanish. Or at least LY between vowels - remains in Portuguese, becomes ach-laut in Spanish (the stages are actually known from rezha on … it was still resha in Cervantes’ day, that’s why Quixote is pronounced kishott in French, but kikhote in Spanish, which went on). To cheat some more, I’d say that this *relya presumably comes from Latin “regula” which means rule or ruler. The object is to ploughing like a ruler is to drawing. This kind of observation can’t be made for Proto-Indo-European. All the other languages have different “root words” (etymons or etyma) for ploughshare.

Next, three arbitrary words in a row that just gave same etymon in all languages. Won’t happen if you take French, Irish, Greek, Russian, Armenian, Kurdish instead of French, Italian, Spanish, Portuguese, Catalan, Romanian. You will get far more examples like ploughshare, and nearly no straight “same etymon for ALL” and the “same etymon for MOST” will also be rare, it is probably more like same etymon for less than half.

So, these words teach us … QU between vowels becomes P in Romanian, and some TT kind of thing becomes PT in Romanian.

QU between vowels is then disappeared in French (or possibly a W sound that led to the U part of EAU), remains or doubles like CQU in Italian, weakens to GU in Spanish, Portuguese and Catalan (where it also adds an Y to previous vowel), becomes P in Portuguese.

Some TT kind of thing (between vowels of course) … T in French (or possibly YT, if the I in AI is from this group), remains as TT or becomes TT in Italian, becomes CH in Spanish and YT in Portuguese, T again in Catalan, and PT in Romanian.

L between vowels is L in French, Italian, Spanish, Portuguese (not quite true, here it is kept because final, because last vowel dropped in time), Catalan, but becomes R in Romanian.

L before a first vowel remains in all but Catalan, where it becomes LL (= LY). M and R before first vowels remain in all.

Final E drops in French, Spanish, Portuguese (that’s why the L remains in “o mel”) and Catalan, but is retained in Italian and Romanian. Wait, final E actually remains in Spanish and Portuguese after the TT group.

Final A drops in French, remains in Italian, Spanish, Portuguese, Catalan, and becomes “ă” in Romanian.

Stressed vowels. A remains everywhere in *AQUA (= Latin “aqua”) except for French where it seems to become E, and it remains in Italian and Romanian for *LAYTTE or possibly *LAKTE, but becomes E in the rest. E remains in Portuguese and Catalan, but becomes IE in the rest.

So, I reconstruct:

ploughshare, common to Spanish and Portuguese = *RELYA
water = *AQUA
milk = *LAKTE
honey = *MELE

How good is this? Well, the Latin words are REGULAM (plough is ARATRUM, which is seen in Catalan, while SOC is a Celtic word); AQUAM; LAC / LACT- with a faulty masculine accusative in LACTEM, and MEL (also neutre, also short accusative identic to nominative, this is why “o mel” is finishing in a remaining L), with forms in MELL-.

When it comes to reconstructing Proto-Indo-European, the disadvantages are:

  • fewer words common to all branches or even most of them
  • more changes in pronunciation in several more steps
  • several more changes of meaning from reconstructed common etyma (we had one example, REGULA becoming REJA / RELHA as in ruler => ploughshare)
  • so many more exceptions to the deduced rules of pronunciation change … why do we have *YEQUR becoming HEPAR in Greek, but *YUGOM becoming (D)ZYGON in Greek?
  • and above all, no written mother language documented to check the results with or even to substantiate the single mother language hypothesis.

What are the features of Indo-European languages?

Hans-Georg Lundahl
amateur linguist
  • Sharing a good portion (not necessarily all or even most for each language) of Pokorny’s vocabulary which he considered as being there in Proto-Indo-European and where each “word” or word family is introduced with the form it was in his day considered to have been correctly reconstructed for Proto-Indo-European. Most widely shared lexical items include numbers from 1 to 10, plus for 100.
  • Sharing some (not necessarily all etc) morphological features like a) 1st and 2nd singular pronouns in m- and t-, with a divergent forms for nominative (like Finnish); b) 1st and 2nd plural have different stems (unlike in Finnish), c) presence of a case system of eight cases or its reconstructably correct reductions, with plural cases originally not given as plural ending plus ending for the case, but things like “genitive plural” given as one specific ending (of Gothic genitive plurals, one declinsion can and one cannot get its genitive plural back to the reconstructed one); d) same case for subject and predicate (unlike Finnish), e) case congruence between attributive adjective and its noun, f) same case for intransitive subjects as for agent-subject in transitive verbs in the active with a different case for patient-objects of the transitive verbs (unlike Basque, but like Finnish), g) verb system with personal endings differing from personal pronouns (unlike Turkish), h) verb systems with different past tenses for imperfective and perfective aspect (exception Germanic and Lithuanian, and newer forms of Slavic).

What language was spoken in Europe before Roman times?

Hans-Georg Lundahl
amateur linguist
Depends on what part of Europe.

Around Rome before it was founded - very old Latin.

In the area of Assisi before Rome arrived after Battle of Sentinum 295 BC - Umbrian, a language related to Latin, but one with lots of P instead of QV.

In Fiesole before Rome arrived in 283 BC - Etruscan.

If we go to the North, to areas like Milan and Turin, it was Celtic among the Insubres, in what’s now Milan, and earlier partially Liguric (perhaps still there among the Taurini). Then a Roman colony was founded in 28 BC among the Taurini, and the Insubres obtained Latin Citizenship in 89 BC.

North of that, you have both Celtic and Rhaetic, the latter related to Etruscan. Rhaetia is like Western Austria and Eastern Switzerland. Noricum is mid Austria, with an East frontier near Vienna, Celtic. Pannonia … “Pannonia was located in the territory that is now western Hungary, western Slovakia, eastern Austria, northern Croatia, north-western Serbia, northern Slovenia, and northern Bosnia and Herzegovina.” - This region had exchanged Pannonian (a language related to Illyrian) for Celtic in the 4th C. BC. All of these three were Romanised under Augustus.

North-West and along the Mediterranean coast, Ligurians are replaced by Celts after some time. This is the case both in Gaul and further North, like in Bavaria - Bohemia, North of which were the Germani.

Way West, you had Iberi, Celts, and Carthaginians on the Iberian Peninsula, and they each had their language (Iberian language is little known, the Celtic language in Spain has lots of QV where the Gaulish one has P (same as between Irish and Welsh), and the Carthaginians spoke Punic - basically Hebrew).

South and East you had people speaking Greek in Naples and in parts of teh Celtic area too … and obviously in Greece and parts of Turkey.

North of these, you had perhaps already Slavs - or perhaps not yet arrived.

In other words, Europe was a very mixed bag before Latin.

What are the differences between Old Norse and other Indo-European languages?

Hans-Georg Lundahl
amateur linguist
What are the differences between Old Norse and Russian and Greek and Latin and Old Irish …

They are different.

When it comes to cognates, if you want to look up how a word that spells *qweqlom in Proto-Indo-European* looks in Old Norse, Greek and Sanskrit, you want to read Brugmann.

He takes branch after branch for Indo-European, on the supposition they developed from one common proto-language and gives the sound changes leading up to each of the languages.

But please note, this is never the entire vocabulary and also never the entire set of grammar endings for any one language, there is always some innovation in morphology and some additions to vocabulary that are not from PIE. So, for instance, English is, like Old Norse, Germanic. The lexical similarity between English and Russian is c. 25 %. Probably calculated from samples of medium difficulty texts that exist in translation. And if we go to older forms of each it will be less, since words that spread with modern culture (like “fantastic”) were absent.

By* the way, *qweqlom in Old Norse, Sanskrit and Greek is hjól, çakra and kuklos.

* Reconstructed forms that are not attested in modern speech or old writing are marked with a previous asterisk.

Why did early Germanic languages diverge so dramatically from Proto-Indo-European?

Hans-Georg Lundahl
amateur linguist
Supposing PIE actually existed, any branch of IE diverged pretty dramatically from it.

For the sound changes, see Brugmann.

There are about as many from PIE to Germanic as there are to Celtic or Italic.

For vocabulary and grammar, NO branch of IE shares all of the reconstructed PIE language. Or even most.

How can we tell if a given word is of Proto-Indo-European origin or not?

Hans-Georg Lundahl
amateur linguist
First, it is possible that no such thing as PIE existed.

Second, for those who presume it did, the criteria:

  • the word occurs in three branches of Indo-European or more;
  • the reconstructed form can derive the forms in the several branches according to the sound laws in Brugmann;
  • if the words don’t mean the same, the changes in meaning should be understandable.

How did Proto-Indo-European phonology change over time?

Hans-Georg Lundahl
amateur linguist
Do you mean in the actual language or in the reconstructions?

For the latter, let me present you with the fastest changing language of the 1870’s - PIE!

Schleicher's fable - Wikipedia

While the next version after Schleicher was made only decades later, by Hirt, it is certain that proposals leading up to (or some of them possibly not) the different versions were, some of them, proposed already in the decade following the publication.

As to the actual language, supposing it existed, it would seem that most reconstructions of it I’ve come across are so to speak “static” and do not take changes over time before the breakup into account.

Decades ago, I saw a proposal involving a kind of explanation why Germanic languages have so very different pronouns for 1st and 2nd plural, compared to the rest. This rested on changes within PIE, but I am far from sure there is a complete model involving them.

Did Proto-Indo-European have nasal vowels?

Hans-Georg Lundahl
amateur linguist
There are several different reconstructions of Proto-Indo-European.

Did it have four laryngals or one … or none? etc.

None of the reconstructions I know of actually includes nasal vowels.

How do we know what sound changes occurred in Proto-Indo-European?

Hans-Georg Lundahl
amateur linguist
  • We do not know that any did, because we do not know if PIE existed.
  • We do have some idea, depending on the reconstruction of PIE, what happened between PIE and given branches.

And if you want to reconstruct a form like qweqwlom:

  • in order to get wheel, you make qw > hw, you drop h after a vowel, you drop the ending and *hwewl already looks a lot like Anglo-Saxon hweol. Attested.
  • in order to get kuklos, you make qw > k before a high rounded back vowel, you also drop the w before a consonant, and you change the neutre nom-acc ending into an accusative only -on, getting a nominative -os from already existing masculines - wait, did I mention you had to presume a side-form without the e, *qwqwlom instead of *qweqwlom? Non-vocalic w automatically becomes a vocalic u between consonants, and this u is then turned into ü in pronunciation;
  • in order to get cakra you first turn qw into ch before e, then change both e and o to a, and drop the final m. Somewhere along the way L becomes R as well.
  • Usually Slavic kolo is not to this noun, but to a verb meaning turn connected to it, but if we assumed *qweqwlom for kolo as well, this is possible. qw becomes k, any k and lots of others drop after vowels before a consonant, final m is also dropped and then in most Slavic languages e becomes in certain conditions o.

Note that each of these sequences of sound changes occurs in one or more of the languages that supposedly derive from PIE, and would be doing that even if PIE didn’t exist, if these words were mutual loans at a very early stage.

How do reputable linguists currently view the theory of a Proto-Indo-European language?

Hans-Georg Lundahl
amateur linguist
Most of them agree with it.

I’m an amateur linguist, not a reputable one.

Nikolai Sergeievitch Trubetskoy was a reputable one (he started Balkan linguistics!) and he disagreed, but he’s dead since 1938. I still think he’s right, though.

He considered the common traits between Indo-European branches are explained, not by a common proto-language (“pater” and “father” both go back to “pehteer” and ”venio” and “I come” both go back to “gwemyo” because Latin and English both go back to PIE), but because they shared an area (“pater” and “father” go back to the same language as they do to the same word, but the one giving ”venio” and “come” could be another one, but they were once situated in a neighbouring area).

I am trying to test this a little.

The vocabulary in Pokorny. How many branches the medium word comes into (5/10 or 3/10) is less important than how the combinations of 3 groups out of 10 line up. Let’s say all possible combinations of 3 groups are roughly speaking equally represented, we can safely say all branches descend from one language having all of the words in Pokorny. But let’s say all the combinations in Pokorny involve two language groups always lining up and two other ones very rarely doing so, one can conclude that there were a series of areal features. And that two language groups sharing many words either were descendent from a proto-language (narrower than PIE, though) or were neighbourly later than the earliest neighborhood mutual influences.

Anyway, I have reputable linguists mostly against me. One man considers Celtic languages come from a Sprachbund Atlantic coast of Europe, that came to include also a contributor from PIE, Cunliff, I think Barry Cunliff. But he has expressed no such thing about the whole Indo-European group.

Did ancient Celtic tribes such as Gauls/Gaels and Britons/Picts share any similarities in culture/language? Did they have any relations with each other?

Hans-Georg Lundahl
Studied Latin (language) at Lund University
Gauls are in fact much closer to Britons than to Gaels.

So, you have Gaels or Goidels on Ireland, and you have basically Britons all over Scotland, England, Wales and most of France.

The differences in language would make mutual understanding without study too difficult, but they were clearly very similar in language. Brit. arguably *mappos, very early Goid. *maqqos, modern Welsh “mab” and Irish “mac”, meaning son. “Pen” and “ceann” mean head. As you can see words in P in British words that have Q, later C in Gaelic. The words for white show another similarity with a difference, “fionn” in Irish and “gwyn” in Welsh. So, original W becomes WH and F in Irish, becomes GW in Welsh. Hundred is respectively “cant” in Welsh and “céad” in Irish. Original EN becomes long E in Irish and AN in Welsh. Of these words, only the word for 100 is shared with most other “branches of Indo-European” while mac, ceann, fionn, / mab, pen, gwyn are unique to Celtic. Given the said, you can guess (as I guess) that Brennus - the Gaul who sacked Rome very early on - has the same name as Bran in Welsh and Brian in Irish.

Is it possible to reconstruct the Proto-Indo-European sound system?

Hans-Georg Lundahl
amateur linguist
Yes. Even multiple different reconstructions are possible.

This remains true whether there was a real PIE language or not.

Any reconstructed PIE is a conlang, like Quenya and Sindarin.

It is reconstructed on a hypothesis, that simplifies, namely that the words and grammatical endings common to several branches come from one single language rather than from all the different “branches” exchanging with each other.

Look at this answer for how reconstruction simplifies other things:

How do linguists use the comparative method to reconstruct Proto-Indo-European?

What's the most scientifically accurate and complete reconstruction of the Proto-Indo-European language?

Hans-Georg Lundahl
amateur linguist
That is a question that scientists differ on.

And, unlike whether it existed, where Trubetskoy’s position is now a clear minority, to which I more or less still belong, what it was is still clearly up for debate.

The standard model with three or four laryngals is being challenged by Jounu Pyysala who accepts one (with - as for many other consonants - a voiced and a voiceless allophone, depending on position, if I got it right), and whether “Indo-European B, D, G’, GW” were voiced stops (like B, D etc) or whether they were glotticised stops or something different is also discussed.

Plus complete and scientifically accurate are not necessarily the same.

Anyone with a laryngal theory will say that Brugmann and Pokorny are outdated as to accuracy, but they will still use these as references, since they are so much more complete than expressions of more recent research.

What is the meaning of the reconstructed Proto-Indo-European word *h₁értus?

Hans-Georg Lundahl
amateur linguist
I do not know.

I tried to look it up in Pokorny, and as Pokorny was before the Laryngal Theory and H1 would preserve the following E as E, I tried ert- sth, but … wait.

J. Pokorny's Indo-European Etymological Dictionary

er-4 with extended versions er-t and er-w

er-4 (er-t-, er-u̯-)
English meaning Earth
German meaning `Erde'


Material Gr. ἔρᾱ `Erde', ἔρα-ζε `zur Erde' (vielleicht davon ἐράω, s. unten S. 336; mit Zusammendehnung wohl πολύηρος · πολυάρουρος, πλούσιος Hes.); ἔνεροι, s. oben S. 312; ἐρεσι-μέτρη · γεωμετρίαν Hes.;

germ. *erþō in got. aírþa, anord. jǫrð, ahd. (usw.) erda `Erde';
germ. *erō in ahd. ero `Erde';

u̯o-Erw. in anord. jǫrvi (*erwan-) `Sand, Sandbank', und

cymr. erw f. `Feld', Pl. erwi, erwydd, corn. erw, ereu ds., abret. mbret. eru, nbret. ero `Furche' (*eru̯i-);

vielleicht arm. erkir `Erde' (Pedersen KZ. 38, 197), wenn für *erg- (idg. *eru̯-) nach erkin `Himmel'.

References WP. I 142, Finzenhagen Terminol. 6, Schwyzer Gr. Gr. I 424.