21 September 2014

The Name of the Game: Jakobson Reads Vasmer

With the vast and reliable etymological material put into circulation by Vasmer, a number of new questions naturally arises. I should like to dwell on some particulars.
Roman Jakobson (1955) *) 

The Slavs played at “even and odd” too. In Polish the game used to be called cetno licho (or cetno i licho). The noun licho is still used as a mild euphemism for ‘devil’. Czego chcesz, do licha? means “What the heck do you want?” Polish also has the adjective lichy ‘poor, inferior, in bad shape’. Historically, licho is a neuter form of lichy, substantivised centuries ago, when the adjective had a wider range of meaning, including  ‘mean, evil’; licho was therefore ‘something wicked’. The phrase cetno i licho lingers on on the fringes of literary Polish (people are at best vaguely aware that it refers to some old game of chance), but cetno no longer occurs on its own, and has no obvious relatives  in the modern Polish lexicon.

The man who read Vasmer's dictionary
A few hundred years ago (most examples come from 16th-century texts) cetno and licho could mean, respectively, ‘even number’ and ‘odd number’. Though often contrasted with each other, they were not yet harnessed together into a fixed phrase. Cetnem (instr.sg.) or w cetnie (loc.sg.) meant ‘(occurring) in even numbers’; likewise lichem and w lichu ‘in odd numbers’. This usage has been completely forgotten.

Licho and lichy go back to Proto-Slavic *lixъ ‘strange, irregular, rogue’. In the modern Slavic languages it usually has pejorative conotations (‘bad, lacking, defective, lonely’, etc.); it can also mean ‘excessive, superfluous’. The meaning of Russian lixój, however, ranges – somewhat schizophrenically – from ‘bad, sinister, hard’ to ‘daring, valiant’ (the common ancestor was ‘extraordinary’, whether in a positive or a negative sense)’. Like semantically similar words in other languages (Greek perittós, English odd), *lixъ developed the arithmetical meaning of ‘odd’, which survives here and there in the Slavic branch. For example, in Czech liché číslo means ‘odd number’. As for its origin, *lixъ < *leikʷ-so-, from the widespread Proto-Indo-European root *leikʷ- ‘leave, abandon’.

So much for licho. Where does cetno come from? The Russian term for “even and odd” is čët i néčet. Čët means ‘even number’ (= čëtnoe čisló); néčet is its antonym. The adjective čëtnyj ‘even’ (of a number) is closely related to Polish cetno. Russian č normally corresponds to Polish cz, but some regional varieties of Polish have merged the affricate cz /tʂ/ with c /ts/ for centuries, and the standard language has borrowed a number of dialectal pronunciations of this kind.

On the combined evidence of Polish and East Slavic forms we can reconstruct Proto-Slavic *četъ (n.) and *četьnъ (adj.). Russian also has the noun četá ‘pair, couple’, which is formally and semantically close to them. There are several other Slavic words that might or might not be related to *četъ, but it’s wiser at this stage to exclude more difficult material so as to avoid the risk of contaminating a reliable set of cognates with spurious ones.

Back in the 1950s, as successive volumes of Max Vasmer’s monumental Russisches etymologisches Wörterbuch were published in Heidelberg, the great linguist Roman Jakobson (then at Harvard University) read the entire dictionary (I mean, actually read it like a novel, page by page), jotting down comments on entries that attracted his attention. Those marginalia were published as a journal article (see the reference below) and reprinted in Jakobson’s Selected Writings (Volume II: Word and Language). With regard to čët and its relatives, Jakobson remarked that they “seem to be archaic relics of the same word family as četýre” (the Russian reflex of the Indo-European numeral ‘four’). Having devoted one sentence to the matter, he moved on to the next entry that had caught his eye, čex ‘Czech’. The idea that čët and četýre are somehow related has been picked up by several other authors, but hitherto published attempts to analyse *kʷetwor- in this light have the usual flaws of “root etymologies”: too little attention to morphological details, and too much imaginative semantics. Nevertheless, I think Jakobson’s idea is worth salvaging, so I’ll review those previous attempts and try to see if I can do any better.

*) Roman Jakobson. 1955. “While reading Vasmer’s dictionary”. Word 11: 611-617.

[link to a digitalised Russian translation of Vasmer's dictionary]

[to be continued]

[back to the table of contents]


  1. Concerning Polish cetno - are there any Slavic cognates attested outside of Eastern Slavic? As you know, a correspondence of Polish "c" and Russian "č" could also go back to a Proto-Slavic *tjet- or *ktet-, in which case the Polish form would be regular and not dialectal, but we would lose the possible connection with "4".

  2. As I will show, the putative South Slavic cognates are doubtful and had better be excluded from comparison. Nevertheless, PSl. *ket- seems to be the only plausible analysis for the following reasons:

    Slavic *kt/__e does not generally merge with *tj. The following vowel has to be *i or *ī (Slavic *ь, *i). There is no such development in forms like *plekt-en- > *pletenь 'braid'. In the 'daughter' word (Balto-Slavic *dukter-) medial *-tj- was analogically generalised from the nom.sg. *dъtji < *duktī < *duktẹ̄r.

    PIE lexical roots don't normally begin and end with the same stop, so *tjet- is a priori an unlikely form (it could be morphologically complex, but in what way? *tj-et-? what kind of derivative would that be?).

  3. Thanks for the correction on *kt, I wasn't sure of the exact conditions any more (I thought it could be any front vowel). What are those putative South Slavic cognates? In any case, IMO, if the cognates would indicate a Proto-Slavic *tjet-, the point would be that the word probaly wouldn't be inherited from PIE, would it?

    1. I will discuss those cognates in the next post. If they are cognate, they point to *četъ. By the way, word-initial *tj was vanishingly rare in early Slavic. The only good example I know of is *tjudjь 'foreign, alien, not one's own', an unquestionable Germanic loan. On the other hand, dialectal c for expected cz is well attested in Polish, cf. cudo < *čudo 'miracle, prodigy' (< *keud-es-), ceber < *čьbьrъ 'bucket'.

    2. *tjudjь was the example I was thinking of. In any case, if there are no Slavic cognates indicating *tj-, then č certainly is more probable.

  4. an unquestionable Germanic loan

    How delicious, then, that it was (much more recently) borrowed back into Viennese as, well, /tʃuʃ/ (with short /ʃ/), a xenophobic term.

  5. the adjustment on *kt, I wasn't certain of the specific conditions any more (I figured it could be any front vowel). What are those putative South Slavic cognates? Regardless, IMO, if the cognates would show a Proto-Slavic *tjet-, the point would be that the word probaly wouldn't be acquired from PIE, would it?