PhDniX's blog

One hunderd

Lately I've had some questions from several friends what the Centum & Satem split is about. Mostly because I have fairly controversial theories on this. So whenever I talk about Centum vs. Satem language split, and my theories about it, people generally end up asking me what Centum and Satem languages are exactly. Therefore I thought I'd write a definitive post to once and for all end these questions.
I also figured it would be nice to have some slightly easier entries on this blog about Proto Indo-European (PIE) as well. It's not unlikely that I'll write more non-Indo-Europeanist friendly entries in the future.

This blog entry will be about the sound shifts that take place in Proto Indo-European, and the languages that sprout from these. I will especially focus on the so called Centum & Satem split. Which has to do with the `velar series' in PIE. First I'll explain what these `velar series are', after that I'll explain the definition of the Centum & Satem split, and to conclude I'll give practical examples of words in Centum and Satem languages.

In PIE you have the so called `velar series'. Groups of three types of consonants with a velar articulation.
In a row they are the:

• Palato-Velars ḱ ǵ ǵʰ
• Plain Velars k g gʰ
• Labio-Velars kʷ gʷ gʷʰ

What's special about these Velar series, is that every Indo-European language only retains two of them. The so called Centum languages retain the Plain Velars and Labio-velars, and the Satem languages retain the Palato-Velars and the Plain velars.

So in Satem languages Plain Velars and Labio-Velars merge to k sounds, while Palato-Velars stay distinct. And in Centum language Plain Velars and Palato-Velars merge while Labio-Velars stay distinct.

Centum is the Latin word for `Hundred' and `Satem' the Avestan word for hundred. These words  come from the Indo-European word *ḱmtom `Hundred'. This is especially illustrative, because Latin has the "K" sound here (although these days we would pronounce it as s, originally it was K) while Avestan retains the palatal articulation, and actually exaggerates it so much it becomes an s (which is the general tendency of Satem languages).

This split in how these velars are treated are generally thought to be a big split in two dialects of PIE, and this would mean a Centum language and a Satem language would never be able to form a subgroup in the Indo-European family tree together.

In the next few paragraphs I'll illustrate the difference between Centum and Satem. I'll do this by taking three words, with the three different types of velars. Then I'll show how they are reflected in a Centum (English) language and a Satem (Sanskrit) language respectively.

The three words I'll compare are the words for 'Hundred', 'Light' and 'What'.

Centum languages:
English (a Centum language) has hundred where the h comes from palatovelar ḱ (root is also *ḱmtom).
The English word light  comes from the PIE root *leuk- with a plain velar which also becomes h (and later gh, but Old English spelling still has leoht.)

Thus the Palatovelar and and Plain velar have merged into h.

The English word what comes from the Indo-European word *kʷod `what'. Here you can see how  kʷ  (a Labio-velar) became wh and not h as with the plain velars and palatovelars.

So clearly Labio-velars are distinguished from the other velars.

Satem languages:
Sanskrit (a Satem language) has śatám for hundred also the root is *ḱmtom  so the ś comes from ḱ.

The Sanskrit word for light though is roka from the root *leuk-, here you can see the ḱ stays clearly distinct from the k. as k stays k while  ḱ becomes ś.

The Sanskrit word for `what' is kád which comes from the Indo-europed word *kʷod. As you can see here the labiovelar became k, and through that merged with the plain velar.

So to sum it up. English has h for Indo-European ḱ and k. And it as wh for Indo-European kʷ.

Sanskrit has ś for ḱ. And it has k for k and kʷ.

I hope this illustrates the concept Centum and Satem languages. And how the split works. Interestingly enough, a few Anatolian language (not Hittite though, that's a Centum language) retain the full three series, so they can't be defined as Centum or Satem.

As said before the general consensus is that Centum vs Satem split is something dialectal in Indo-European itself, I personally am not very convinced of this theory, which I'll try to explain in later updates.

I want to do this now!

I've been working on a theory with one of my classmates on the Indo-European present stem suffix *neu/nu.
As seen in the Greek δείκνυμι for example. I'm not sure if anyone else ever proposed this, but I think it's very likely that it is actually derived from the root for `now'  *neu/nu.

Not only Greek retains this suffix. Also Tocharian B has this suffix. *nu regularly gives nä. Which isn't attested along, but does appear with the suffix *-ske/o (which is -ṣṣä/ske in Tocharian B).

And the *nu version is then the suffix -nä-ṣṣä/ske. As seen in the word:
lännaṣṣä/ske- (here <a> is phonemically /ä́/) `to leave'
Interstingly, just like in Greek, besides the present there isn't another place where this *neu/nu suffix appears. Not in the subjunctive, nor the preterite or the preterite participle.

So the present class would mean something like `I VERB now', something very appropriate for a present tense I would say. There's the problem though, that you can never really prove that this is indeed the original.

Beekes though, proposes that this present suffix is actually derived from  verb roots ending in *-u with an n-infix. Which resulted in an ending in *-nu/neu which was then reinterpreted as a new suffix. Similarly he proposes that the *-nh2/-neh2 suffix is formed in a similar way but then for roots ending in *-h2.

I personally find the present marker *neu/nu as derived from the Indo-European word for `now' a far more appealing concept. This leaves us with being unable to explain where the *-nh2 is from though.

But we have another problem. Just like that the *nu-ske/o suffix appears in Tocharian B, there's also the *nh2-ske/o suffix. Which functions in same way as the other suffix. An example of the *nh2-ske/o is this:

kärnāṣṣä/ske- `to buy' (<ā> is phonemically /á/).

But unlike the *nu-ske/o suffix, this suffix also appears as a present marker without the *ske/o suffix. In fact it's one of the most productive present classes. The so called na-presents.

Example of a na-present verb is:

tärkana- `to let go' which has a preterite tarka-.

There's no such thing as a  nä-present though.  This distribution is  strange. If they would have the same sort of origin (as Beekes proposes) you'd expect them to have the same sort of distribution. This  could be seen as argument for our hypothesis that *-neu/nu comes from `now' rather than a clipping from n-infixed roots ending in *-u. Not a very strong one though.

I have not put any time into figuring out the origin of the *nh2/neh2 suffix, I will write a new post about that soon. Feel free to add any thought you might have on this subject, or questions for that matter.