29 September 2014

Forgotten Derivatives and Their Sexual Implications

What kind of noun is čët? What is its relationship to our hypothetical verb root? One cannot avoid asking such questions when proposing an etymology. A word is more than a root; it has a derivational history. If you add an affix to a word, you may alter its lexical category and its meaning of the base. We already know a good deal about morphological processes in the Indo-European languages, which means that we can tell plausible relationships between possibly related words from unlikely ones.

Let R be a root morpheme. In Proto-Indo-European (and in many of the languages descended from it), a root consists of a consonantal skeleton with a slot where a vowel can be inserted. For example, the verb root *{w_rǵ} ‘make, work’ is normally quoted in the form *werǵ-, called its e-grade, symbolised as R(e). Here, the slot is occupied by the vowel *e. The same root also forms an o-grade, R(o), realised as *worǵ-, and a zero grade, R(z), in which the vowel slot remains empty. In that case, the liquid *r, sandwiched between two other consonants, has to play the role of a syllable nucleus, and the root becomes phonetically *wr̥ǵ- (in the traditional Indo-Europeanist notation, a tiny subscipt ring marks a syllabic consonant).

One of the largest and most productive classes of PIE nominals (nouns/adjectives) were the so-called thematic nouns (also known as o-stems). Their stem ended in the vowel *-o-, to which inflectional endings were attached. In the simplest case, the vowel was added directly to the root; in more complex cases it was part of a suffix (such as *-to-, *-no-, *-tero-, *-tlo-, etc.). Somewhat surprisingly, “simple thematic”  nouns of the shape R(e)-o- were pretty rare in the protolanguage. The neuter action noun *wérǵ-o-m ‘work, activity’ is well supported by the agreement between Germanic *werka- (Old English weorc, German Werk) and Greek érgon; we also have Iranian (Avestan) varəza-, with the same stem (and meaning) but with masculine inflections. Very few such nouns, however, are truly old. More typically, the suffix *-o- was added to R(o), as in *wóiḱ-o- ‘house, dwelling’ (root *weiḱ- ‘enter, occupy’) and sometimes to R(z), as in *jug-ó- ‘yoke’ (root *jeug-, already mentioned in earlier posts).

Marc Greenber (2001) doesn’t define the morphological status of his reconstruction *kʷet- (‘two’ > ‘pair, partner’). In some places in the article he treats it as if it were a root noun (with no suffixes), but the simplest form we actually find in Slavic is represented by Russ. čët (cf. dialectal Polish cot), which appears to reflect a thematic masculine noun *kʷet-o-s ‘even number’. How could it have originated? If *kʷet- was once a verb root (with the approximate meaning of ‘arrange in pairs, pair up’), *kʷet-o- makes sense as a kind of action noun that has acquired a resultative interpretation: by pairing objects together, you end up with an even number of them. (By the way, the verb root is not entirely conjectural: we can see it in Russian četáť ‘form pairs’.) The problem  with *kʷet-o- is that it represents a rare type of stem, at least in terms of PIE morphology. Is it legitimate to posit it just like that?

On the other hand, *kʷet-o- needn’t go all the way back to PIE. The deverbal formation R(e)-o- has enjoyed increased productivity in Slavic. We even have doublets like R(o)-o- and R(e)-o-, where the o-grade variant is more conservative (and has more external cognates), while the e-grade seems to be a younger innovation (with a more restricted distribution).  Thus, the root *tekʷ- ‘run, flow’ has produced Slavic *tekъ (as if from *tekʷ-o-s) ‘waterflow, leak, source’, which coexists with *tokъ (< *tokʷ-o-s) ‘stream, current, flux; (figuratively) course, sequence of events’. The former is an innovation directly connected with the Slavic verb *tekti ‘leak, flow’ (3sg. *tečetь > *tékʷ-e-ti), whereas the latter is a relict form which has drifted away from its etymological base, also semantically. Therefore, if *četъ is a relatively recent derivative of a Proto-Slavic verb, it wouldn’t be surprising if it had an o-grade cousin (possibly with a more “evolved” meaning).

As a matter of fact, Greenberg mentions *kotъ ‘offspring (of animals), litter’ and *kotiti (sę) ‘have young’ as possible members of the same word-family. A connection with the homophonous noun *kotъ ‘domestic cat’ (a European Wanderwort which spread with the introduction of cats) is folk-etymological: the verb may be used of cats, but also of mice, sheep, goats, roe deer, and a variety of other animals. It is used even in those Slavic languages that have a different word for ‘cat’ (e.g. Serbo-Croatian mačka). The verb *kotiti could be an “iterative/causative” built to the root *kʷet-. The structure of such secondary verbs is R(o)-éje/o- (the final vowel of the stem alternates depending on which conjugational ending is added). For example, the Slavic verb *gъnati (3sg. *ženetь) ‘drive on, drive away, rush’ has a corresponding o-grade iterative, *goniti (3sg. *gonitь) ‘chase, run after’. These forms ultimately reflect PIE *gʷʰén-/*gʷʰn- ‘slay, kill with blows’ (a root verb, somewhat  restructured in Slavic) and its PIE iterative *gʷʰon-éje/o-. The verb *tekti (< *tékʷ-e/o-), mentioned above, forms a pair with the causative *točiti ‘cause to flow, (cause to) roll’  (< *tokʷ-éje/o-). Note also such English pairs as lie vs. lay, or sit vs. set, where the first member is a primary verb and the second is its causative (e.g. ‘lay’ = ‘cause to lie’).

The consequences of forming a pair.
[source; © gerald reiner]
The stem *kʷot-éje/o-, originally with middle-voice inflections (whose function was taken over by the reflexive/reciprocal pronoun * in Slavic), would mean ‘form a couple (together)’, hence ‘mate, have sex’, and eventually ‘reproduce, have young’. If so, *kotъ ‘litter’ is not a senior synonym of *četъ (with a hard-to-explain change of meaning), but more likely a separate verbal noun back-formed from *kotiti sę (the consequence of mating), on the analogy of formally similar denominal verbs: *agniti sę ‘yean’, *teliti sę ‘calve’, *žerbiti sę ‘foal’.

The feminine *četa can hardly be a collective (at any rate in the meaning ‘pair’). Not only because it refers to just two things, but also because collectives in *-ah₂ to o-stem masculines are an archaic formation in Indo-European (as opposed to neuter collectives, co-opted as ordinary plurals of neuter nouns and adjectives), and *četъ is unlikely to be sufficiently ancient. But Indo-European *-(a)h₂ was not only a collective suffix and a marker of femininity; it was also employed to coin (formally feminine) abstracts, including action nouns. Quite a few deverbal masculines in Slavic (and more generally in Balto-Slavic) have feminine synonyms like *čarъ ~ *čara ‘sorcery, enchantment’ or *-tokъ ~ *-toka ‘flow, course’, *-sěkъ ~  *-sěka ‘cutting’ (in compounds). Note the familiar morphological formations represented by Greek tómos ‘slice’ (result of cutting) versus tomḗ ‘cut’ (an instance of cutting) – a nice parallel to *četъ (resultative) vs. *četa (an individual instance of pairing).

In the first post of this series I suggested that the stem *kʷet-w(o)r- was originally a deverbal neuter of a familiar type. Before I develop this idea, let me briefly suggest one other possible trace of the root *kʷet-: the second member of the Latin compound triquetrus ‘triangular’. The next post will be about it.

[back to the table of contents]


  1. In Basque we say erdi 'half' and erdi(tu) 'to give birth to', as well as erditu 'to split into two' (being more common the derivatives erdiratu and erdibitu in the sense 'to split into two'); cf. Latin pars, pario...?
    In this case (PIE *kʷet-), could the solution have more to do with "halving"/"giving birth" than with "coupling"?

  2. It probably works for Basque, but not for Latin. Pariō comes from *perh₃- 'procure, provide', which formed a *-je/o- present, *pr̥h₃-jé/ó-. It means 'bear, bring forth', but also 'produce, bring about, provide', etc. Pār, paris is a mysterious word. At first blush it looks as if it contained a fundamental *ā/a (rather than a vowel coloured by a laryngeal).

    1. Latin _pa:r_, _paris_ recalls _La:r_, _Laris_ (Carm. Arv. pl. LASES); _sa:l_, _salis_; and _ma:s_, _maris_. One semantic issue is the superlative in the epitaph of Scipio Barbatus, QVOIVS FORMA VIRTVTEI PARISVMA FVIT. One might expect 'most equal' in satire, but not in a solemn epitaph. More likely PARISVMA is to be taken as 'most fitting', that is, Barbatus was not only a great statesman, but looked the part.

      The impersonal _pa:r est_ may also be understood as 'it is fitting', e.g. Cic. de Orat. 4: _pa:r est omni:s omnia experi:ri:_. Umbrian has _parsest_ in apparently the same sense (Tab. Iguv. 7B:2), parallel to _mersest_ 'the law is' (ib. 6B:55), where _mers_, earlier _mer^s_ (ib. 1B:18) continues the neuter *medos, *medes-. If Umb. _pars_ is a neuter /s/-stem, its /r/ must be rhotacized */s/, for original */r/ syncopated against final */s/ yields -r in nom. pl. _frate(e)r_ 'brothers' < *fra:te:rs < *fra:teres. This suggests that Lat. _pa:r_ in _pa:r est_ is not the nt. adj. but a nt. noun *pa:ros, earlier *pa:sos, syncopated like _vir_ < *virs < *viros.

      The ablaut in 'salt' has been explained by a PIE paradigm with nom. *sáh2ls, acc. *sh2álm, gen. *sh2lós, and secondary derivation from the middle stem *sh2al- (e.g. Skt. _sal-ilá-_ 'salty'). A similar paradigm with nom. *páh2s(s), acc. *ph2ásm., gen. *ph2sós could handle _pa:r_, _paris_ with nom. sg. rhotacism after the nom. pl. due to the latter's high frequency (as with _La:r_ but not _ma:s_); *páh2s-os 'that which is fitted, fitting' would underlie Umb. _pars(est)_ and Lat. _pa:r (est)_. Outside Italic, *ph2s-tó- (*ph2as-tíjo-) '(pertaining to the) well-fitted' would give Gmc. *fast(j)a-, Arm. _hast_ 'strong' and Skt. _pastyàm_ 'dwelling-place, house'.

      With breeding stock, the male is brought in to service the female, turning her into a nursing mother. Thus Lat. _ma:s_, _maris_ (*máh2s(s), *mh2sós) plausibly involves *mah2-s, extended from *mah2- 'to suckle' in *máh2-tr. 'mother'. Likewise from *pah2- 'to feed and protect, take care of' comes *pah2-s- 'to facilitate protection' vel sim., 'to fit together (items to build a shelter)', 'to be fitting, match'.

      Lat. _La:r_, _Laris_ (*láh2s(s), *lh2sós) is more difficult. If it first meant 'apparition (of an ancestor)', it might go with Skt. _lásati_ 'shines, glitters, appears' (from the middle stem *lh2as-). Otherwise if _lásati_ merely reflects *les-, _La:r_ could involve *lah2-s-, extended from the root of _lateo:_, Greek _lantháno:_ 'I lie hidden, escape notice', though it is hard to see how the force of the /s/-extension would square with that assumed in *mah2-s- and *pah2-s-.

  3. Hello Piotr. On behalf of PIE *kwetwóres I would like to pull your attention on my morphological and etymological analysis of this word in S.Neri/S.Ziegler, "Horde Nöss". Etymologische Studien zu den Thüringer Dialekten, Bremen 2012, pp. 65f. Best, Sergio

    1. Hello, Sergio,

      How nice of you to pay a visit to this blog! Knowing the quality of your etymological work, I'm most intrigued. I wonder if you could share an electronic copy of that chapter if you have one. My university library hasn't got the Münchner Forschungen series.

  4. The modern Russian word for "to mate" is 'sparitsa', derived from 'para' -- "a pair, couple".

    I'm going crazy from this amount of evidence

    1. Looks like a calque of German sich paaren to me, itself built on a loan from Latin.