Thursday, 16 February 2012

a whole lot of words from *dyeu-

Greek has Ζεὺς πατήρ Zeus patēr, Sanskrit has द्यौष्पितृ dyauṣ-pitṛ, and Latin has Jupiter from Iouis + pater. The first element of all three of these can be derived from Proto-Indo-European *dyeu- "to shine and (in many derivatives "sky, heaven, god")". The second element is from *ph₂ter- "father". So these three deities presumably come from a form meaning "sky-father".

*dyeu- is also apparently found in Armenian աստուած astowaç "god", usually written Astwatz. I'm not sure exactly what the etymology is here.

In English *dyeu- became the name of the god Tīw as in Tuesday.

*dyeu- is also the source of Latin deus and dīuus as in deity and divine, and Avestan daēuau- "spirit, demon" as in Asmodeus.

It's also the source of Sanskrit देव deva "heavenly, divine", and Hindi देवदार devdār "divine tree", whence deodar.

It's also the source of Latin Diāna, the moon goddess.

In Celtic it became Gaulish Dēvona and Welsh duw "god".

In Greek it became δηλος "clear" (from earlier *deyalos) - as in psychedelic.

Addendum: as pointed out in the comments, δηλος is from a variant (Watkins says the variant form *deih₂-).


Nelson said...

Not to nitpick too much, but the last word in this entry probably doesn't belong there. *dyew and its vrddhi derivative *deyw-o- are right for most of the forms (respectively Zeús, Iū-, and dyau- for the base and deus, daēuuō, devá-, Celtic *dēwo-, and Germanic *Tīwaz for the vrddhi derivative). This *dyew was probably not verbal - LIV at least doesn't recognize *dyew as a verbal root, just a nominal base, and Ringe similarly glosses it purely nominally as 'sky' (PIE to PGrmc, p. 14). Some people do reconstruct a verbal *dyew, 'to be bright' (e.g. Sihler, Comp. Gramm., p. 282/s.275.2), but this seems more based on theoretical assumptions than data.

But δῆλος probably comes from the different, labial-less root *deyh2, the source of Sanskrit √dī and Greek δέατο. Verbal *deyh2 and nominal *dyew might be somehow connected, but Beekes is (probably rightly) sceptical of direct derivation of one from the other (Etym. Dict. p. 498). He suggests that they might both be extensions of some sort of earlier **dei-.

goofy said...

well, all I can say is that's what Watkins say in the AHD. too bad it's disputed.

Nelson said...

I've had a look at Watkins' entry, and I don't think there's any real discrepancy. He derives δῆλος from a 'variant' *deiə (= *deyh2), and not directly from *dyew (which is what I thought was being implied, and would be phonologically pretty difficult - sorry for not catching on). Calling *deiə/*deyh2 a 'variant' is certainly perfectly fair if you think these *di- words are all ultimately connected.

goofy said...

I see. Sorry for not giving enough details.

Glen Gordon said...

Are there direct cognates of δῆλος in other IE languages? Or is δῆλος simply hoped, based on thin criteria, to be related to our IE root *dei- by way of assuming whatever "extensions" are necessary to link words together haphazardly?

Áhann Áhim said...
This comment has been removed by the author.
Nelson said...

If δῆλος is from *deih2, then it's related to to an Indo-Iranian root, showing up as Sanskrit dī. said...

Do I have this right: nouns and then an adjective are being derived on top of each other through the use of "stem-vowel" suffixes and vowel-grade alternation.

1.DEIE - verb - "to shine"
2.D^IE-U-S - u-stem noun - "heaven (that which shines), God"
3.DEI^-U---O-S - o-stem noun -"a god"
4.D^I^-U-I-O-S - yo-stem adj "divine"

Or do I have it backwards? I.e. Watkins 22: "Noun *deiwos, god, formed by e-insertion to the zero-grade *diw- and suffixation of (accented) -o-." (Like yudh- -> yodhah, right?) But where is *diw- coming from? Is it the Zero-grade of the noun *di̯ēus, itself a -u- stem noun derived from the primitive verbal root dei-1, dei̯ǝ-, dī-?

I've tried to hash it out, but still looking for a derivational-process pattern. Maybe you guys know it or could shine some light on it? (no pun intended)

Nelson said...

The clearer part of the derivation is a nominal(?) root *dyeu-, 'sky, sky god', from which is derived another nominal *deyw-o-. This is done by a process sometimes dubbed 'proto-vrddhi', which basically means taking the zero grade *dwy and inserting an e-grade at the first phonologically available point (even if, as in this case, that's not where the e-grade would usually go with that root).

The Sanskrit adjective diviyá- (Classical divya-), 'heavenly', is from the noun *dyew with some sort of adjectival suffix, either *iHa or *iya (if the latter, the *i might actually be part of the locative ending of the noun). The Greek δῑος, 'divine', is probably from the same source. Latin has a similar adjective, dīvus, that is often reckoned as being from *deywos, having split off from the deus because of a divergence in the paradigm of Old Latin deivos, and taking on an adjectival function from being used in apposition to god names.

So that gives us a pretty clear set for three of the PIE words:

*dyew- 'sky'
*deywos 'skyling, god'
*diwiyo- 'celestial'

How a root *deyh2 might fit into this is less clear, and is not a connection most of the recent handbooks seem to make. It's hard because the root *deyh2 clearly has a laryngeal, but there's no trace of *h2 in the various 'god, sky' words. If there's a connection, it has to be at the pre-PIE level, and not really be a PIE derivation proper.

(The only thing I could think of to connect *deih2 with these other words is that *dyew is formed as a u-stem derivative from this root, with loss of the laryngeal in the position *CHy-: *dyh2w- > *dyw, which then inflected as a hysterodynamic u-stem noun, which in turn formed an *-iyo- adjective. *deywos would have to be a later derivative, once the laryngeal was well and truly disassociated from this family of words, since **deyh2wos should have left traces in Greek and possibly in Sanskrit. I don't know if this derviation holds up to close scrutiny, though. I'd be worried about whether the *CHy > *Cy change occurred early enough to allow for the development of this word family.)

Nelson said...

Actually, ignore the last paragraph entirely - I was being clueless. The sound change in question (even if you accept it - not everyone does) is *CHy > *Cy, but the zero grade of *deih2 would be *dyH. No motivation for the laryngeal to be dropped out, so no PIE-level connection between *deih2 and *dyew.

Like I said, there might be some sort of pre-PIE connection and root extensions, but that is getting into very uncertain waters. I would say it's safer to say that *deyh2 and *dyew are simply unconnected, at least as far as PIE is concerned. said...

So, what you're saying is that these -u- stem nouns are primitive nouns not derived from verbs: *dieus belongs with *medhus, *sunus, and the other few of the like. Very interesting. To quote Szemerenyi (163), "Only a few IE nouns can be designated as root nouns. The great majority are formed by means of various suffixes from simple roots or from already existing derivatives." So, all the reconstructed IE u-stem nouns are "root nouns"?

Without spending days on it, the only other parallel I could find to shine:divine was live:alive - gweie-:gwīwos. But bare gweie- seems unattested, unless Polish żyje > **gweieti (3p sg) is the sole survivor. Would that make this a "root noun" as well? (i.e. a house: dwell, live). Against this, jīva=vīvus=živъ "alive" is -o- stem, and an intermediate u-stem **gwiēus seems to be unattested.

Thanks for the explain of proto-vrddhi. Do we always insert reconstructed *e? Is *dem- > **dm- > *domos/*domus also an example of proto-vrddhi?

@Glen: Pokorny doesn't list δῆλος under dei-. But δῆλος had to come from somewhere, right?

Nelson said...

Saying that *dyeu is a non-derived noun doesn't imply that all u-stems are root nouns. Sihler, for instance, includes only *gṷou-, 'cow', *dyeu-, 'sky', and *neh2u-, 'ship', in a category he labels 'monosyllabic diphthongal stems'. I think these should be sharply differentiated from the true u-stems of PIE.

The root for 'live' is reconstructed *gṷyeh3 (using ṷ to represent the superscript for feature of labialization and w for the glide proper). This seems to have a corresponding *wó verbal adjective, formed so: root *gṷyéh3 > zero-grade *gṷih3 > suffixed *gṷih3-wó- > laryngeal contracted surface form *gṷīwó- (Sk. jīva-, Latin vīvus, etc.). Compare the more common verbal adjectives in *tó and *nó, also formed from the zero-grade of a verbal root.

So this is not much parallel to *diw-iyó-, which is denominal instead of deverbal, and uses a rather different suffix, *iyó (or perhaps *iHó). This must be from the nominal *dyeu, not from a u-less verbal *dei(h2), as it has the labial glide before the adjectival suffix: Sanskrit div-iyá-. So 'living' is a deverbal adjective in *wó-; 'heavenly' a denominal in *iyó-. It looks like the only thing they have in common is that they are adjectives with thematic suffixes.

Proto-vrddhi should have nothing to do with *dem > *domos, for a couple of reasons. For one thing, the PIE noun was probably an acrostatic noun: nom. *dóms, gen. *déms, with thematic *domos a later formation. There were indeed various thematic nominals that could be derived using an o-grade (e.g. *rówdHos, 'wailing' < *rewdH, 'to wail'), but this also has nothing to do with proto-vrddhi, being just ablaut.

Proto-vrddhi, as I understand it, only introduces e-vowels into places they shouldn't occur by normal ablaut, or into otherwise non-ablauting stems. This type is fairly constrained, taking a nominal stem, putting it in the zero-grade, adding an e-grade (even if in an unexpected place relative to normal ablaut), and attaching an accented thematic suffix. Hence *deywós < *diu < *dyéu.

Glen Gordon said...

Hermeneuo: "Pokorny doesn't list δῆλος under dei-. But δῆλος had to come from somewhere, right?"

Yes but then we need to ask ourselves why we have become so convinced that the etymology must be Indo-European (as opposed to a Bronze Age borrowing).

What continues to concern me about the IE field is that there is such a conviction that much of the vocabulary of IE languages derives from Proto-IE itself when Occam's Razor would suggest that it's more likely for a word to be a product of borrowing than an intact word surviving for thousands of years.