I had one yanked today, so I thought I’d post about the Indo-European forms, which mostly all come from the same root and which beautifully illustrate all sorts of sound changes; this is the sort of thing that got me interested in historical linguistics. The Germanic forms — Old English tóþ, Old Saxon tand, Low German tan, Dutch tand, Old High German zan(a) (German Zahn), and Old Norse tǫnn (Swedish, Danish tand, Norwegian tonn) — all come from a reconstructed *tanþuz (Gothic tunþus has a different vowel that must come from the zero grade); French dent, Italian dente, Spanish diente, and Romanian dinte all come from Latin dent– (nominative dens); Greek odont– (nominative odous) shows the o-grade and an initial laryngeal; and all these, plus Sanskrit dant-, Welsh dant, Old Irish dét (i.e., /de:d/), Lithuanian dantìs, and Armenian atamn, come from a Proto-Indo-European root reconstructed as *dent-. The AHD IE Appendix lays it out by root grade, beginning with the suggestion that it was originally a participle:
dent-
Tooth. Originally *h1d-ent‑, “biting,” present participle of ed– in the earlier meaning “to bite.”1. O-grade form *dont‑. tooth, from Old English tōth, tooth, from Germanic *tanthuz.
2. Zero-grade form *dn̥t‑. tusk, from Old English tūsc, tūx, canine tooth, from Germanic *tunth-sk‑.
3. Full-grade form *dent‑. dental, […] from Latin dēns (stem dent‑), tooth.
4. O-grade variant form *ədont‑, ultimately becoming odont‑ in Greek –odon, […] from Greek odōn, odous, tooth.[In Pokorny ed‑ 287.]
The Slavic words (Russian зуб [Vasmer], Polish ząb, etc., all from related to OCS зѫбъ), like Latvian zùobs and Albanian dhëmb, come from a different root, *gembh- ‘tooth, nail,’ which gives English comb among others, and Irish has fiacail, which is just weird.
eDIL says it’s from Latin figere, i.e. a mouth-fixture.
Wow, I thought you had a post yanked! Sounded serious till I realised it was just a tooth!
And about the adventures of the kind of person who is interested in historical linguistics, reread the wonderful false-teeth scene in Nabokov’s Pnin.
In fact, the Russian zub and Polish ząb don’t continue the OCS zǫbŭ but a (reconstructed identically in this particular case) late Proto-Slavic form. It’s a very common error to mention OCS as the source for any modern Slavic cognate word but really OCS despite its archaicity was already a derived Slavic language with certain innovations of its own, which are not shared by Russian or Polish. E.g. the pre-OCS dialect metathesized a Proto-Slavic *CărC to *CrāC → CraC, while the pre-Polish one metathesized it to *CrăC → CroC, and the pre-Russian one employed epenthesis to form *CărăC → CoroC.
The root of tooth, dens does exist in Slavic as a suffixed derivate, namely in the words dziąsło (Old Polish dziąsná), десна ‘gum’, from an EPSl. *dent-snā.
Excellent points all, thanks!
I like all this talk about roots and teeth. A very incisive post.
For h₁ and such, you can use the Unicode subscript figures (U+2081 in this instance).
With *ǵ, of course, and said to mean “grind” or “crush” or something – as in Gomphotherium, which is named for its teeth.
*ǵ
Sorry, I am not familiar with the meaning of this symbol. How different is it from *g ?
PIE is generally thought to have had plain, labialized, and palatalized velar plosives. The labialized ones are usually indicated with ʷ (though superscript u̯ has also been used, strangely enough); the palatalized ones most commonly get an acute accent these days, but a circumflex has also commonly been used.
It used to be common to doubt the distinction between the plain and the palatalized ones, apparently mostly due to the misunderstanding of the palatalized ones (like [kʲ]) as palatal (like [c]). Pokorny, IIRC, was one of the doubters, and the AHD follows him with only the tiniest of amendments.
As Ксёнѕ Фаўст points out the Slavic reflex refers to the gums. In Croatian the word is “desni/desne” (pl.) = gums.
There is a homograph “desni” (adj. masc. nom. sing.), “desne” (adj. fem. gen. sing. and fem. nom. pl.) which means “right” (ie. opposite of left), but it is pronounced differently. The Croatian word for “right” is cognate to Latin dexter. For etymology search for “desni” at http://hjp.novi-liber.hr/index.php?show=search .
Croatian pronunciation:
dȇsni = gums – low falling tone on the E, short I
dèsnī = right – short rising tone on the E, long I
There is a homograph “desni” (adj. masc. nom. sing.), “desne” (adj. fem. gen. sing. and fem. nom. pl.) which means “right” (ie. opposite of left), but it is pronounced differently.
By those who use the pitch system, but they’re homophones for the many Croatians who use dynamic stress instead.
There is a 17th-c. inghean óg is geal dead and more dead-teeth here (Dáta Ghrá):
http://archive.org/stream/dtagrhaanthology00orah/dtagrhaanthology00orah_djvu.txt
Some dental fixtures and deud elephaint in Gaelic:
http://www2.smo.uhi.ac.uk/gaidhlig/faclair/sbg/lorg.php?faclair=sbg&seorsa=Gaidhlig&facal=deud&eis_saor=on&tairg=Lorg
data. danta. My tablet makes me physically dizzy. Sorrty.
dent-
Tooth. Originally *h1d-ent‑, “biting,” present participle of ed- in the earlier meaning “to bite.”
Could this by any chance have given rise to the verb “to eat” in Slavic languages? The infinitive in Russian doesn’t sound like it at all, but in my language (Bulgarian) “да ядe” (that he/she may eat), ядене (eating, food) sounds very similar…
Indeed it did. The Russian infinitive is opaque because of a sound change *-dt- > -st-, so from *ed-ti you get OCS ѣсти; in the first person, *-dm- > -m-, so *ed-mi gives pre-Slavic *ědmь, leading to OCS ѣмь and Russian ем. You can see more forms here.
Dievas dave dantis; Dievas duos duonos —Lithuanian proverb
Deus dedit dentes; deus dabit panem —Latin version thereof
Deity donated dentition; deity’ll donate doughnuts —English version by Muke Tever
God gave gums; God’ll give granary —Version by Mat McVeagh
God gave goons; God’ll give guns.
I just looked up why Swedish would have tand /tand/ when ON had tǫnn — it could be dialectal between East and West Nordic (since most “ON” forms given are in fact Old East Norse), but it seems to have been intrusive in the plural tennr and spread from there. OSw had tan in the singular.
(Danish tand is not diagnostic since /-nn/ and /-nd/ merged early on and both are spelled {-nd}; ODa did have tan in the singular).
@lars
OHG had both tan and tand, I suspect this is a wider reflex of [theta] after n, maybe some kind of sandhi in declined nouns.
Yes, I was only talking about North Germanic. Gothic has tunþus with zero grade and it’s all over West Germanic too.
The point is that there is no stop or fricative in ON, so it would seem to have been lost in North Germanic except that Swedish shows a /d/. But it was tan in OSw too so it’s not the inherited /þ/, another explanation is needed.
While the participial origin of the PIE “tooth” word looks pretty compelling, a tooth is, of course, a “biter” rather than an “eater.” Is there any other evidence that the PIE root had an “earlier meaning” of “bite”, apart from the very word “tooth” itself?
“Eat” and “bite” are both readily reconstructable in Proto-Volta-Congo, and completely distinct: *dɪ and *dum. I can’t reconstruct “tooth”, but it isn’t derived from either verb in Proto-Oti-Volta (*ɲɪn-), nor in Proto-Bantu (*-jíno.) How often is “tooth” derived from “eat” in other language families?
The noun that is derived from “bite” in many “Gur” languages is “snake” which makes sense; the Proto-Oti-Volta word is unfortunately the much less exciting “long creature” etymologically. On the other hand, the Western Oti-Volta words for “mosquito” have the same stem as “bite.” For some reason.
Is a bit of something a small bite of it ? Ein Bisschen is that.
How often is “tooth” derived from “eat” in other language families?
食む
Japanese
Etymology
/pamu/ → /famu/ → /hamu/
From Old Japanese.
Possibly cognate with 歯 (ha, “tooth”). (Can this(+) etymology be sourced?)
Pronunciation
Kun’yomi
(Tokyo) はむ [háꜜmù] (Atamadaka – [1])[1]
IPA(key): [ha̠mɯ̟ᵝ]
Verb
食む • (hamu) transitive godan (stem 食み (hami), past 食んだ (handa))
1. (archaic, dialectal) to chew and ingest food; to eat
Synonyms: 食う (kuu), 食べる (taberu)
草を食む
kusa o hamu
to graze
2. to get paid, receive a stipend (from a duty, work, etc.) (Can we verify(+) this sense?)
3. to harm, injure
Synonyms: 害する (gaisuru), 損なう (sokonau)
/pamu/
That one looks like an “eat” word (possibly) derived from “tooth” (rather than vice versa) via “chew”, which seems relevant but a bit different.
“Chew” -> “eat” seems quite a likely development in itself; but I was wondering if there was any actual evidence for an earlier “bite” (or “chew”) meaning for the PIE root, other than a circular argument from “tooth” itself.
“Chew” in Proto-Western-Oti-Volta is *ŋʊb-, by the way, a bit reminiscent of PIE *gembh-; another phonaesthetic word?
“Chew” -> “eat” seems quite a likely development in itself
French manger and its Romance relatives, of course …
A fun fact is that Swedish has äta as the normal word for “to eat,” with a regular present participle ätande, and at the same time the inherited word for tooth is tand. I don’t know of any other IE language where the phonemes of the two forms have ended up so nicely aligned (even though it involves a small accident with the /d/s as described above).
Well, dens and edens in Latin …
Hum, I don’t know how I missed that.
David Marjanović (May 23, 2015): It used to be common to doubt the distinction between the plain and the palatalized ones, apparently mostly due to the misunderstanding of the palatalized ones (like [kʲ]) as palatal (like [c]). Pokorny, IIRC, was one of the doubters, and the AHD follows him with only the tiniest of amendments.
True, the AHD’s PIE appendix did not recognize the plain/palatalized distinction through the 3rd edition (1992). It was introduced in the revision for the 4th edition (2000), although they still don’t show it on the headwords, only as a note on “oldest form”, which the original post didn’t quote: “Oldest form *ĝembh‑, becoming *gembh‑ in centum languages.” (They use an inverted breve, but I replaced it with a circumflex since the breve often doesn’t align correctly.) That’s also how they deal with laryngeals. Maybe they thought they looked too intimidating on the headwords.
I think the -d in Swedish was preserved in dialects, maybe helped by some morphophonological process. Ref. Old Gothlandic tandr. The partially parallel “mouth” word is ON munnr/muðr* but OSw. munder.
But I’m not at all sure about the derivation from the participle. Da. tang, Sw, tång, Eng. tongs, Ger. Zange etc. go back to PGmc. *tangō- f. , apparently with o-grade from a verbal theme *den-ḱ- “bite” also attested in Indic. The reflexes of this o-grade are parallel with the full grade of the “tooth” word, while the zero-grade forms of tooth are parallel to the “mouth” word < PGmc. *múnþa- m < PIE *mn̩to- < *men- “stick out, protrude”. Gmc. “tooth” could be from *den- “bite” by the same derivational mechanism. The history is complicated by a partial gender switch, which I’ll suggest might be back-formation from a feminine plural with collective meaning.
* Something I really don’t understand is the conditioning factors for ON -nnr ~-ðr.
Funny that Lithuanian and Latvian take their tooth words from different PIE roots. How’d that happen? According to Wiktionary, Lithuanian does have a word žambas that’s cognate with the Latvian and Slavic tooth words, but it means “sharp edge” instead.
And Wikt also says this word was borrowed from Proto-Balto-Slavic into Proto-Finnic, producing e.g. Finnish hammas ‘tooth’.
As far as I can tell, the ‘nail’ meaning of PIE *ĝembh‑ is supported only by Ancient Greek γόμφος ‘bolt, dowel, nail’ (so that’s nail as in hammer and nail, not as in fingernail).
ktschwartz: Lithuanian does have a word žambas that’s cognate with the Latvian and Slavic tooth words, but it means “sharp edge” instead.
The Norw. cognate kam “comb” is synonomous with egg “edge” as a topographic term.
Something I really don’t understand is the conditioning factors for ON -nnr ~-ðr.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/263156817
The Origin of the nn/ð Alternation in Old Icelandic
Thanks! I like that a lot.
Proto-nail. I am not sure if there are any nails in the Stone Age (or Early Bronze)…
Do tenons count?
Don’t know, but it strikes me that ‘nail’ and ‘nave’, < PIE *h₃nogʰ- vs. *h₃nobʰ-, might ultimately be from the same root. The “joint” meaning of ‘nave’ is preserved in Norw. nov(e) “corner of a house”.
@Trond, ON is given as tǫnn; I don’t know that happened with the PG *-z there, but if OG kept it, it might “protect” the *-þ- as well? And yes, I thought -nnr > -ðr was unconditional in ON, so a †tǫnnr would surprise me on that account.
munnr does surprise me. Zoega does say old nom. muðr so maybe we’re just seeing analogical levelling in mid-spread, with conditioning factors changing on the fly? All the descendants seem to have levelled the -ðr words that survive.
Anyway, the issue is that the *-þ- was lost specifically in ON, so why does it reappear in Swedish? That it was probably still there in the common ancestor of ON and OG is interesting, but does not bear directly.
That’s not H.
The latest idea is actually to set up a separate root “bite”, *h₃od-. It was already reconstructed for other reasons.
It was already reconstructed for other reasons
That’s more like it!
I am not sure if there are any nails in the Stone Age (or Early Bronze)…
Once again:
The use of wood as a tenon can be traced back over 7,000 years, as archaeologists have found traces of wood nails in the excavation of early Germanic sites. Trenails are extremely economical and readily available, making them a common early building material.
https://en.wikipedia.org/wiki/Treenail#History_and_general_use
Unexpectedly in Russian (naval) it is nagel.
Unexpectedly, ‘The noun mortise, “a hole or groove in which something is fitted to form a joint”, comes from c. 1400 from Old French ‘mortaise’ (13th century), possibly from Arabic ‘murtazz’, “fastened”, past participle of ‘razza’, “cut a mortise in”.‘
@lars
OS has tänder in the plural (nom., acc.). This might allow *tand by back-formation.
https://spraakbanken.gu.se/karp/#?mode=oldswedish&resources=soederwall&lang
@PP, yes, that was exactly what I got from Hellquist here. Thanks for the link!
That’s expected, because Russian naval vocabulary is essentially Dutch.
Lars M.: Anyway, the issue is that the *-þ- was lost specifically in ON, so why does it reappear in Swedish?
I’ll just note that I’m still working on this, but not hard.
I’ll just note that I’m overusing “I’ll just note”.
@Trond, that was just to give the context, we have had quotes from two different etymological dictionaries to the effect that it spread from an intrusive d in the plural — unlike Old Gutnish where I’m prepared to believe that it did survive from Proto-North-Germanic in the singular (maybe because the word was moved to an inflection that had -R in the nom.sg.) — so I don’t think there’s any mystery left.
(Danish has /tæner/ [tˢɛ̰nʁ̩̤] with two full syllables, so if there was ever a need for an intrusive /d/ it was lost in the palatalization event. By which I mean, it may have followed Swedish up to something like /tænder/, but then /-nn-/ /-nd-/ /-nt-/ all > /-ɲ:-/ [spelled {nd}] > /-n-/ so you can’t tell if it went to /tæn:er/ instead, with the same spelling. That would depend on when the second vowel was inserted).
I am not sure if there are any nails in the Stone Age (or Early Bronze)…
Free preview of the article on “Nails” in Brill’s New Pauly:
However, it does seem to be anachronistic to project the sense of ‘nail’ for γόμφος back to Proto-Indo-European, since according to Blümner’s Technologie und Terminologie der Gewerbe und Künste bei Griechen und Römern (vol. 2, 1879), its original sense was a wooden connector used especially in shipbuilding and similar to “what we call (in German) Zapfen” = English pin, peg, tenon, dowel. The sense of ‘nail’ (small metal spike fastener) was a later extension. So maybe AHD should have glossed *ĝembh‑ as ‘tooth, peg’.
“Eat” and “bite” are both readily reconstructable in Proto-Volta-Congo, and completely distinct: *dɪ and *dum.
How did no one make a Tweedle*dɪ/Tweedle*dum joke?
They may well have done, but the archaeological record is silent.
I meant in the comment thread, but I’d accept archaeological jokes as well.
JC 2015:
eDIL does not endorse this etymology; rather it quotes two early Irish glossaries, Cormac úa Cuilennáin’s from the 9th century and O’Mulconry’s from the 7th/8th.
Wiktionary says
Tatyana Mikhailova :
An ancient etymology derives it from Latin fīgō (“to fix”)
In turn, transparently borrowed from the Moba fìgń “fix.”
(Merritt Ruhlen had the right idea, but was handicapped by ignorance of proto-Oti-Volta, the true Ursprache. Who knows what he might have achieved otherwise? Alas, it was not to be.)
Feck!
this explanation [Latin figa] is impossible on phonological grounds [contra @DE]
So we’re guessing where Early Irish fecked fecc from? Or where they fecked the meaning?
Both yeet and yoink? That’s worse than cleave.
@DM cf. lift