Ph.d. 2015 Adam Hyllested

university of copenhagen
University of Copenhagen
Word Exchange at the Gates of Europe

Hyllested, Adam
Publication date:
2014
Document Version
Early version, also known as pre-print
Citation for published version (APA):

Hyllested, A. (2014). Word Exchange at the Gates of Europe: Five Millennia of Language Contact. Kbenhavns
Universitet, Det Humanistiske Fakultet.
Download date: 29. Jun. 2016

FACULTY OF HUMANITIES
UNIVERSITY OF COPENHAGEN
PhD thesis
Adam Hyllested

Five Millennia of Language Contact
3

PhD thesis
Adam Hyllested
University of Copenhagen, 2014
Cover illustrations:
Sarmatie Europenne, a map of The Grand Duchy of Lithuania by the
French cartographer Alain Manesson Mallet, 1685; a Corded-Ware
battle axe and hammer from Nrke, Sweden; a burbot (Lota lota);
European black elderberry (Sambucus nigra); fragment of a Varanasi
silk sari with gold brocade; and s with hek, a grapheme represent-
ing an unvoiced postalveolar sibilant.
4 W ORD E XCHANGE AT THE G ATES OF E UROPE
Table of contents
Preface......................................................................................................................................... 5
Introduction ............................................................................................................................... 9
Fenno-Ugric * as Laryngeal Substitution
in Words of Indo-European Provenance ....................................................................... 11
Stealing the Thunder of alpa:
The Fate of PIE *-bo- in Anatolian ................................................................................ 25
The Precursors of Celtic and Germanic ............................................................................... 43
Again on Pigs in Ancient Europe: The Fennic Connection .............................................. 69
The Other Horse: Germanic Cognates of caballus? ............................................................ 91
Balto-Fennic Loanwords in Proto-Germanic ...................................................................... 99
Gothic mammo meat in the Light of Saami Evidence .................................................... 107
The Mysterious Elder: Common Traits
in European Names for Sambucus nigra and Viburnum opulus ................................ 117
Balto-Fennic *kakra oats, the Etymology of hail
and Another Exception to the Germanic Sound Shift ................................................ 133
Two Issues on Indo-European Substrates in Slavic .......................................................... 153
The Story of time: The Etymology of Finnish aika
(with an Excursus on aita fence and Balto-Slavic v-Prothesis) ............................... 161
Albanian hund nose, and Faroese, SWNorwegian skon,
Finnish kuono snout ..................................................................................................... 167
Estonia and the Aestii: Baltic Etymology as a Key to Fennic Ethnonyms ...................... 179
Word Migration on the Silk Road:
The Etymology of English Silk and its Congeners ...................................................... 187
English mink, Finnish portimo ermine,
and Baltic Fur Trade from Antiquity to the Hanse .................................................... 203
Latin and Slavic Loanwords in Hungarian:
Exceptions to Helimskis Vowel-harmony Adaptation Rules ................................... 213
English abstract ..................................................................................................................... 222
Dansk resum ........................................................................................................................ 224
Preface
The present dissertation is the result of my 3-year Ph.D. scholarship fi-

nanced by the University of Copenhagen as a part of the five-year pro-
gramme of excellence Roots of Europe Language, Culture, and Migra-
tions. It is dedicated to five people who did not live to see the final
product, but whose passion for the study of language directly or indi-
rectly continue to exercise influence on my work: First of all my enthu-
siastic and inspiring teacher, head of centre and initial supervisor Jens
Elmegrd Rasmussen whose warm and welcoming introduction to the
Indo-Europeanist circles in 1994 I will never forget; my good friend,
fellow etymologist and inexhaustible source of knowledge and inspira-
tion Jan Katlev; H.E. Ambassador Paul George Jyrknkallio, who taught
me about the wonders of Hungarian grammar and passed on his per-
sonally dedicated copy of Aulis Jokis Uralier und Indogermanen to
min unge broder i filologi; my grandfather, devoted Romanicist, Lat-
inist and Germanicist, professor Povl Kristian Hyllested, whom I never
got to know in person, but whose note-scribbled library I have inherited
and use every day, probably the main reason I became a linguist; and
my great-grandmother Ragnhild Jensen, born Hst, passionate defender
and recorder of the Bornholmian dialect and penfriend in the early 20th
century with U.S. ambassador to Copenhagen and Belgrade, John
Dyneley Prince, who wrote, among other linguistic works, Tatar Ele-
ments in Old Russian (referred to in this dissertation).
I wish to thank my other teacher and final supervisor, the indefatiga-
ble Birgit Anette Rasmussen and the other colleagues at Roots of Europe
with whom I have spent five wonderful years: Bjarne Simmelkjr Sand-
gaard Hansen, Anders Richardt Jrgensen, Guus Kroonen, Jenny Hele-
na Larsson, Benedicte Nielsen Whitehead, Thomas Olander and Oliver
Simkin; the tried and tested members of the Roots of Europe evaluation
board, professors Douglas Q. Adams, Michael Janda, Joshua T. Katz,
Rosemarie Lhr, Brent Vine and Andreas Willi whose advice and criti-
cal comments at their annual visits have proven invaluably relevant; the
ever-growing and talented Indo-European student community in Co-
penhagen, counting many dear friends; my good colleague, associate
professor Tuula Eskeland who has continuously provided me with sup-
port, advice and tips on recent literature in the Uralic field; fellow Ural-
icist Ilda Hallas-Mller whose competent management of the library
facilities has been instrumental; and a number of other people with
whom I have had fruitful discussions specifically on matters treated in
the final dissertation: Henning Andersen, Lars Brink, Johnny Cheung,
Paul S. Cohen, Michael Fortescue, Bernd Gliwa, Berit Hildebrandt,
Martin Huld, Santeri Junttila, Petri Kallio, Peter Alexander Kerkhof,
Ptr Kocharov, Agnes Korn, Kristian Kristiansen, Martin Kmmel,
Ranko Matasovi, Craig Melchert, Simon Mulder, Robert Orr, Kaspars
Ozoli, Janne Saarikivi, Zsolt Simon, Merlijn de Smit, David Stifter,
Patrick Stiles, Erik Thau-Knudsen, Sen Vrieland and Nicholas Zair.
This dissertation only forms part of the research I have carried out
during the five project years. My employment as a PhD scholar has, in
effect, been interwoven with other projects at the Roots of Europe re-
search centre and is hard to view in isolation from them. The best
known part is probably my work on the Indo-Uralic hypothesis, which
was originally thought to make up the majority of my thesis. The origi-
nal title of my PhD project was The Shared Indo-European and Uralic
Lexicon, deliberately uniting the stratigraphy of loanwords on one
hand and vocabulary which I believe to be inherited from a common
past on the other. Scholars who investigate old loanwords tend to see
them in every case of similarity, arguing vigorously against the possibil-
ity of uncovering inherited material. Adherents of the Indo-Uralic hy-
pothesis, on the other hand, often reject rather obvious intances of bor-
rowing, trying perhaps to maximize the amount of evidence for genetic
affinity. My idea was to introduce an open mind to both approaches,
including both Indo-Uralic material and older loanwords in a lexical
stratigraphy. However, the Indo-Uralic part, comprising also a compar-
ative historical phonology and derivational grammar, grew to such pro-
portions that it will become an independent publication.
During the last half of the project I furthermore became increasingly
involved in studies in Albanian language history, especially phonology,
morphophonology and etymology. Finally in 2013, I devoted much time
to the study of plant-names from alleged European substratum lan-
guages, a subject which ideally ought to have been included in this the-
Preface 7
sis but which proved to complicated for me to able to reach a satisfacto-

ry conclusion within the time-frame given.
Word Exchange at the Gates of Europe:
Introduction
The study of early lexical exchange between Indo-European and Uralic

languages has a proud tradition in Denmark, not least by virtue of the
the pioneering works by Vilhelm Thomsen: Den gotiske Sprogklasses
indflydelse p den finske (1869) and Berringer mellem de finske og de
baltiske (litavisk-lettiske) sprog: En sproghistorisk Undersgelse (1890).
The latter still constitutes the most important reference work on con-
tacts between Baltic and Balto-Fennic in how many other scholarly
fields today can you say that about a work written in Danish? Linguistic
contacts between Uralic and Indo-European and their respective
branches is still today an extensively studied and vibrant field. In some
respects, though, I think that important evidence is consistently over-
looked because of the power of tradition which affects not only how you
carry out your research, but also what you search for, where you look for
evidence, and from which angle.
Traditionally, most lexemes shared by Indo-European and Uralic
language branches are viewed as having been transferred from the for-
mer to the latter. To mention the most obvious example, it is well
known that both Balto-Fennic and Saami languages possess an abun-
dance of ancient terms borrowed from (Pre-)Proto-Indo-Iranian, Pro-
to-Baltic, Proto-Germanic, Proto-Slavic and Proto-Scandinavian. Like-
wise, the vocabularies of more easterly Uralic languages (Mordvin, Ma-
ri, Permian and Ob-Ugrian) have been affected by intense contact with
Iranian languages apart from non-Indo-European languages such as
Turkic while Hungarian has added to its lexical stock hundreds of
(Medieval) Latin, Pannonian Slavic and Alanic (Iassic) as well as Old
and Middle High German loans. Samoyedic languages are even sup-
posed to have loanwords from Proto-Tocharian. Everywhere Indo-
European is automatically assumed to be the provider and Uralic the

target language while the assumed share of Uralic loanwords in older
Indo-European languages is close to absent. Such an asymmetry is
commonly supposed among linguists to be typical for a relationship be-
tween two peoples where one had the upper hand, technically and polit-
ically, at the time of borrowing. While the amount of borrowings rarely
numbers the same on both sides, I find it unlikely that there are any cas-
es where extensive lexical transmission in one direction leaves zero trac-
es in the opposite direction. True, there are famous examples of extreme
asymmetry such as Old Germanic languages versus Old Slavic or
French versus Breton. However, even in these cases at least a small
number of loans in the atypical directions are identifiable. Most such
borrowings have a limited semantic and geographical distribution; they
typically refer to trade objects, important plants and animals, religion,
or other concepts specifically linked to the kind of contact in question.
This thesis aspires to convey to the field of Indo-European and Ural-
ic linguistics a new methodology, where Uralic and Indo-European data
are viewed as equally potential sources for loanwords. Much weight is
put on our ability to reconstruct shared semantics, not least semantic
anomalies, even in cases where the actual lexeme has been replaced. I
also seek to underline the importance of using Uralic material as a key
to unsolved issues in Indo-European. I stress especially that ignoring
variation in the Uralic material (such as dialectal forms, semantic scope,
irregular vocalism, forms from less well-known languages, and older
attestations) can be detrimental or even fatal, leading the etymologist
totally off the scent. Citing one Standard Finnish form is not enough, for
example, when Balto-Fennic languages exhibit a multitude of irregular
forms and deviant meanings. Finally, but equally importantly, I endeav-
or to establish a number of new subfields within the field of IE-Uralic
contact linguistics by showing that hitherto unheeded lexical exchange
took place from Proto-Mari to Proto-Baltic, from Proto-Balto-Slavic to
Proto-Fenno-Permian, and from Proto-Balto-Fennic to Proto-Celtic
and Proto-Germanic.
The dissertation contains 16 articles which are intended to appear in
chronological order, starting with the earliest contacts.
Fenno-Ugric *- as Laryngeal Substitution
in Words of Indo-European Provenance1
Abstract
Jorma Koivulehtos claim that PIE laryngeals in word-internal position

are substituted with *-- in Fenno-Ugric must be reformulated. The de-
velopment is hard to account for phonetically, and a closer look at the
material reveals that there is no uambiguous evidence from outside Bal-
to-Fennic, a subgroup of Fenno-Ugric where *-- regularly yields -h-.
This means that, while word-internal laryngeals may very well be pre-
served in some loanwords (just like initial laryngeals are clearly pre-
served as *k-, as shown by Koivulehto himself), they can simply have
had the manifestation *-h- from the beginning. If this was before the
emergence of *h as a phoneme, it could still have occurred as a loan
phoneme. This loan phoneme would be one of the factors triggering the
development of *-- > -h- in general and thereby the introduction of a
new phoneme proper. One further observation is that only laryngeals
before a dental stop (> BF *t) are represented among the certain exam-
ples.
1 Frozen PIE laryngeals in loanwords in FU
In a number of ground-breaking investigations, Jorma Koivulehto (1988

[1999], 1991, 2001, 2003) has shown that in early Indo-European loan-
words in Fenno-Ugric languages laryngeals are still visible, having been
substituted by a variety of sounds depending on their position in the
word and the time of borrowing. Most convincing is his manifestation
*k- for PIE *H- in an older layer of loanwords, e.g.
1 This paper was presented at the 20th Annual UCLA Indo-European Conference
on 1 November, 2008.
Fi. kes summer PIE *hes-en- harvest (Koivulehto 1991: 36-40)

Fi. kaski burn-beaten land PIE *hez-g()- ashes (Koivulehto
2001: 241)
Fi. kalvas, kalpea pale, N Saami guolbben sandy plain; chalky layer
underneath the top soil PIE *hl-bo-s white, NHG dial. Al-
ven chalky layer underneath the top soil (Koivulehto 2003: 28)
N Saami guovssu < PSaami *kawsoj- dawn PIE *hus-o- (Koi-
vulehto 2003: 29)
Fenno-Volgaic *kua- to weave PIE *heu-d- (Lith. udiu id. )
(Koivulehto 1991: 50)
Koivulehto argues for a conditioning by which the velar stop k- occurs

only in initial position, while any word-internal laryngeal (-h-, -h-,
-h-) is substituted with the postalveolar fricative -- at a certain stage
(but not in the oldest layer of loanwords). Fenno-Ugric *-- remains a
postalveolar sibilant all the way through the Fenno-Permic, Fenno-
Volgaic and Fenno-Saami (Early Proto-Fennic) stages, but develops
regularly into *-h- in Middle and Late Proto-Fennic, merging with the
results of *k, *, * and *z (and even the original *-s- of some Baltic and
Germanic loanwords which may have been pronounced with a more
retracted sibilant than the Fennic dental-alveolar representative). The
consequence of this hypothesis is that any Finnish (or Estonian, Veps
etc. any Balto-Fennic) -h- may be a relic of an Indo-European laryn-
geal.
However, the three PIE laryngeals, although sharing notation, the
term laryngeal, and certain behaviors such as lengthening, vowel col-
oring and a tendency to disappear, were clearly distinct sounds, and
there is no particular reason to believe that they would all behave the
same way in every position. Besides, it makes one a little bit suspicious
that these postalveolar fricatives materialized as laryngeals (more spe-
cifically glottals) again when reaching the Balto-Fennic stage. Koi-
vulehto in his main work on the subject (1991) gives 17 examples:
FU * as Laryngeal Substitution in Words of IE Provenance 13
PFU or PFP form Example alleged NW PIE source

*(j)ekV possibility Fi. ehk, ehki possibly *)h-geh
(Latv. jg ability)
*(j)ete- Fi. dial. ehti be able to *)h-g-)eo- be able to
to have enough power
*ine human being OFi. inhe-minen *enh-
Fi. ihminen beget a child; be born
*keta-ta to dare Fi. kehtaa *geh-d- be ashamed
*kone tool Fi. kone machine *nh-)o- wonder
< magic remedy (ON kyn)
*lete leaf Fi. lehti *bleh-t- leaf; sprout
N Saami lasta
*pewenV sieve Votyak pu-n- *peuH-eno- winnow-
ing
*pota- to winnow Fi. pohta- *pouH-)e- sift (grain)
*p- to fry Lule Saami pass- *beh- bake

(OHG b(j)en)
*putas clean Fi. puhdas *puH-to-s clean
(Skld 1960)
*punV- to winnow Mordvin ponavtoms *pu-ne-H- cleans, win-
nows
*reto line, order Fi. rehto *(h)r-to- line
(Da., Sw. rad)
*roto grass, plant Fi. rohto *groh-to- plant;
growth
*tatas dough Fi. tahdas *th)-s-to-s dough
*tete deed Est. teht *deh-ti- deed
*wita once; at last Fi. vihdoin at last *uiH-to- course, se-
quence
*wo(j)a ramification E Mari (Cheremis) o *uo)H-h branch (OI
branch vay0)
Katz (2003) implicitly supported Koivulehtos thesis regarding the reflex

*-- although he sees them in Indo-Iranian loanwords rather than in
NW Indo-European ones. We may add Mari (Cheremis) r milk
PIE *ksih-r-m, which is in fact the only additional example in his book
for which *-- explicitly substitutes H-.
2 Analyses of the individual proposals
In the following, I will comment on each of Koivulehtos relevant ety-

mologies.
2.1 PFP *( J ) ETE TO HAVE ENOUGH POWER ?

PBF *( J ) EK V POSSIBILITY PIE *) H - G EH POWER ,
*) H - G -)eo- BE ABLE TO ?
PFP *(j)ete- to have enough power (Koivulehto 1991: 77-79; > Fi. ehti
have enough time for a given purpose, dial. ehtii olla can be, N Saami
asta- have enough time, Mari te- do, Komi jeti- be ready; be able
to; be in time for; mature) is traced back to a PIE denominal verb *)h-
g-)eo- be able to (> Lith. jgti, jgi be able to, Latv. jgt, jdzu id.)
but this verb is actually only attested in Baltic. Because of its presence in
the Permian languages it cannot be a Baltic loanword proper, at least
not everywhere, and it may not even be a Balto-Slavic loanword alt-
hough *-- could reflect *-- as a rendering of the palatalized *-g-; in-
stead it could very well instead be a satem reflex of PIE *He)5- to have
in ones power, *Hi5ti- ~ *Ho)5ti- (> e.g. Av. ti- possession ~ PGmc.
*aihti- possession, belongings, property), or more specifically a borro-
wing from Proto-Indo-Iranian into Proto-Fenno-Permic2. Phonologi-
cally, this would make the process more straightforward and account for
the missing *j- in Balto-Fennic.
Koivulehto (1991: 72-74) further derives a Fenno-Saami (Early Proto-
Fennic) *(j)ek possibility (e.g. > Fi. ehk perhaps) from the same
underlying derivative *)h-geh power with reference to typological
parallels like NHG mglich possible ~ vermgen be able to. The rele-
vant Indo-European reflections are Lith. pa-jg ability, Latv. jga id.;
sensibility, Gk. vigor; manliness; young age. Since this word is
only attested as a loanword via Proto-Balto-Fennic (Late Proto-Fennic)
*ehk- (> e.g. Fi. ehk perhaps), one can further object that there is no
direct evidence for a step -- even if the Indo-European etymology
should be correct; it is only reconstructed because BF *-h- is supposed
to have derived from this source.
2
There are related forms in Ugric (Khanty / Ostyak), but they are considered
borrowings from Permian languages (SSA 100).
2.2 PFP * KONE TOOL PIE * G N H -) O - WONDER ?
Fi. kone machine goes back to BF *koneh tool which must mechani-
cally be transposed to Early Proto-Fennic *kone. Koivulehto conjec-
tures a PIE nh-)o- wonder as the source. Personally, I find this one of
the more speculative etymologies one might say wonderful in the
sense full of wonders. More important than my own subjective impres-
sion is that - in *kone is a well-known derivational suffix in Fenno-
Ugric, which may by the way sometimes represent PIE *-s, cf. *omme
fungus (PGmc. *swambaz), *vene boat (Skt. vna- timber); so even
if this is an Indo-European loan, there is no evidence as such that the
laryngeals were not simply lost.
2.3 PFP * INE HUMAN BEING PIE * G E NH - BEGET A CHILD ;

BE BORN ?
Proto-Fenno-Volgaic *ine human being is reconstructed on the basis

of Mordvin (Erzya) ine, (Moksha) indi guest and Balto-Fennic forms
meaning man, human being, not least OFi. inheminen. However, most
Finnish dialects and some of the older attestations point rather to a form
*ineh-minen, as do all of the remaining Balto-Fennic languages, e.g.
Veps iehmoi bitch (pejorative of a woman), Lude inahmoi, Votic
inehmn man, human being. Although it is of course possible that a
lone Old Finnish attestation could be the most archaic one, I find it
more plausible that the majority of languages, dialects and old forms in
this case reflect the original form. Besides, the Mordvin meaning guest
is not immediately compelling. A more probable source would then be
the PGmc. antecedent of Goth. inahs wise (etymologically obscure, but
synchronically analyzable as in- + ah- in aha mind and ahma spirit,
an n-stem ahman-), either a) directly combined with the productive
Balto-Fennic derivational suffix -minen or b) reshaped from a hypothet-
ical *inahman-, consisting of the elements in both inahs and ahman-. In
this case, *inehminen would originally have meant soul. The original -
a- may then have been retained in Lude inahmoi, or the -a- in Goth.
inahs may reflect PGmc. *-e- (as in, e.g. fadar < *faer-, liuha <
*leuhea-). Another possibility is an origin in Baltic, cf. Lith. mons
people, mon woman and -ym famous (Liukkonen 1999)3; Baltic

* is regularly substituted with Balto-Fennic *h.4
2.4 PFS * LETE LEAF PIE ** B LEH -T - LEAF ; SPROUT ?
Proto-Fenno-Saami *lete can be reconstructed on the basis of PBF

*lehte (Fi. lehti, Est. leht) and PSaami *latV (> N Saami luota). Koi-
vulehto (2003: 25), whose own surname contains a derivative of this
word, lehto grove, sees a loan from PIE *bleh-to-(s), a formation re-
flected in PGmc. *bla- blossom, sprout, fruit and *blaa- leaf. A
dedicated and distinguished Germanicist, there is no doubt that Koi-
vulehto is happy to be able to trace a part of his surname (koivu means
birch) back to Proto-Indo-European times, using a Germanic for-
mation as a model for the protoform. However, *lehte it is more likely to
have been borrowed from some Balto-Slavic form related to Lith. lakas
leaf or lktas sheet, OCS list id. These forms are not completely mu-
tually compatible except semantically, but they seem to be linked neatly
by the PFS term. N Saami lasta cannot be from Baltic *lapsta leaf (pace
Liukkonen 1999: 83-84), because *-a- is substituted with N Saami -uo- in
Baltic loans (e.g. N Saami uoldni dew Baltic *aln, cf. Lith. aln).
2.5 PBF * TATAS DOUGH PIE * TH )- S -TO - S DOUGH ?
A good match both formally and semantically is PBF *tatas (> Fi.
tahdas) dough ~ *th)-s-to-s id. (> OCS tsto, OIr. tis, tis and
PGmc. *ais- in OHG deismo sour dough; Koivulehto 2003: 27). Cru-
cially, however, PIE -- reflects the PIE sibilant *-s- rather than the lar-
yngeal. I think this is a quite justified objection, especially since Koi-
vulehto mysteriously does not account for the loss of the PIE *-s- that is
implied. Maybe he thinks that it was assimilated into the sibilant, but
3
I generally do not favor most of Liukkonens etymologies, but this one I would
definitely count as a possibility.
4
Yet another possibility, if *inehminen is the correct reconstruction, is to hypoth-
esize an ancient borrowing from some IE form of *an-mo- spirit, soul; the
forms certainly match semantically and share the same skeleton of consonants,
but since it is not clear precisely what language we would be talking about (we
have no other evidence for a correspondence PIE *a- ~ BF *i-, for example), and
since suffixed forms are only found in Balto-Fennic, it does not seem easy to
come up with credible evidence for such a solution.
surely Ockhams razor dictates that the sibilant we encounter reflects

the sibilant that was there already, while the laryngeal was lost (by as-
similation or not) as laryngeals go. I have no other explanation for Koi-
vulehtos hypothesis other than associative influence from his own na-
tive language in Finnish the word acquires -h-, so when Koivulehto
first saw the PIE reconstruction he may have been reminded of (and
later blinded by) his own word for it. No doubt that the etymon is the
correct one, though.
2.6 PFP * PEWEN V SIEVE PIE * PEU H - ENO - WINNOWING ?,

PFP * POTA - TO WINNOW PIE * POU H - ) E - TO SIFT
( GRAIN )?
PFB * PUTAS CLEAN PIE * PU H-TO - S CLEAN ?
PFV * PUN V- TO WINNOW * PU - NE -H- CLEANS , WIN-
NOWS ?
Skld (1960: 37) was the first to point to an Indo-Iranian protoform of

Ved. ptas the source of BF *puhta- clean (> Fi. puhdas) and *pohta-
to winnow (> Fi. poht-i-a). He did not visualize an intermediate step
*--, and it is implicit that the laryngeal was borrowed directly as *-h-,
which is also the point of this article; however, he did not give additional
evidence to back this claim up. Koivulehto (1988 [1999: 301-302], 2001:
246, 2003: 26) adds PFV *punV- to winnow (> Mordvin ponavtoms)
and *pewenV (UEW 738). These forms can hardly be separated from
each other, nor from the forms mentioned by Skld. While we are defi-
nitely dealing with a laryngeal in the root, this becomes less relevant if
there are extensions with a sibilant since a sibilant in the target language
is more likely to reflect that sibilant than a neighboring laryngeal. I
would asserting Baltic *ptas clean (> Lith. puostas) as the common
source, following Liukkonen (1999: 107-108).
2.7 PFU * PE - OR * P - TO FRY PIE * B EH - BAKE ?
This root (Koivulehto 1981: 355-356, Koivulehto 1988 [1999: 301], Koi-
vulehto 1991: 85, fn. 11.6) is not represented in Balto-Fennic, but is re-
flected in Saami (N Saami bassi- Lule Saami pass-), Permian (Ud-
murt/Votyak pi-, Komi/Zyryan pe-, also be ready/done (of food))
and Ugrian (Mansi/Vogul pt-, Khanty/Ostyak pl-, pat- to fry in fat).
Koivulehto asserts as the source PIE *beh- to bake ( to fry, to

bake, PGmc. baka- to bake, perhaps including the present-tense suffix
*-)eo- (PGmc. *bjan- > OHG b(j)en bake)5.
Again his analysis does not address an inherent and obvious possi-
bility that constitutes a serious weakness for the analysis of -h- as the
reflex of *--: If the present suffix is potentially contained in the substi-
tuted form, where did it go? Why not just say that *-- is the rendering
of both -h- and the -)- together, probably realized as a palatal affricate?
Of course the laryngeal then plays a role in the emergence of *-- in this
case, but how can it then be used to explain other cases of *-- where
there were no present suffixes or other -)-sounds to be articulated with
it. At least the loss of the suffix needs to be addressed if one (like Koi-
vulehto) counts as a possibility that it was contained in the form that
was borrowed. Note how similar this problem is to the lack of explana-
tion of the lost -s- in *tehistos > *tatas above if the PIE -s- played a
role in the emergence of PFU --, how can it be used as evidence that
other laryngeals without sibilant neighbors yielded -- all by themselves?
We cannot rule out that Koivulehto might be on the right track by
postulating an origin in PIE *beh-, but in that case the origin of *-- is
likely to be the PIE cluster *-h-)- and not the laryngeal alone. In my
opinion, however, an origin in an unidentified satem language with sec-
ondary palatalization from PIE *pek- to cook (Satem *pek- ~ *pe-, cf.
Skt. pcati) appears more probable. In that case the attestations do not
even have to reflect a single synchronic transmission, but can be the re-
sult of several independent borrowings which would explain the diffi-
culties of reconstructing the exact vocalism. If there in fact was a PFU
protoform *p- the rounding could be explained by the preceding *p-;
note that PFU *jwa- grain is from PIE *!euo- with the same kind of
secondary rounding affected by the following *-u- (Koivulehto 1981:
355).
My alternative proposal here obviously stands on less firm ground
than those for the other items, seeing as we barely have additional evi-
dence of the relevant correspondences in material from an early Satem
dialect of Indo-European. I do think, though, that the potential exist-
ence on PFU *pe- ~ *p- is enough reason to start searching for
5
According to Koivulehto (1981: 348-356), BF *pai-sta- > *pajsta- (Fi. paista-) to
bake and BF *pejtt (Fi. peitt-) to cover are borrowed from each of the stems
attested in Germanic, the latter via a meaning wrap into something hot or
something that keeps the heat; cf. for a parallel semantic development the rela-
tionship between Olonets Karelian suoju cover and Est. sooe, gen. sooja heat.
such evidence, which could also provide new important information for
the discussion whether Satem isoglosses overlapped with actual PIE dia-
lect divisions.
2.8 PFP * WO ( J ) A BRANCH ; BENDING PIE * U O )H- H ID .?
A form *wo(j)a branch; bending may be reconstructed for the Proto-

Fenno-Permian stage, but in that case it disappeared completely from
Fenno-Saami since it is attested only in Volgaic and Permian. It is re-
constructed on the basis of Mordvin (Erzya) uo, Moksha u corner,
Mari (Cheremis) W a E o branch; bending, point where something
bends, and Komi (Zyryan) vo river arm; fork; branch; (in Permyak
also) sprout. The normal reconstruction is *woa (Itkonen 1953-1954:
165). Koivulehto (1991: 96-99) asserts PIE *uo)H-h branch as the
source (> Ved. vay0, OCS vja id., derivatives of *ue)H- to wind) and
therefore wonders whether the FU protoform had been *woja, which is
also a possible reconstruction, though he prefers *woa after all as Balto-
Fennic has no inherited words of the structure *CVihV. He shows, how-
ever, that Udmurt (Votyak) might have preserved such a structure (cf.
vaji next to Komi vo pole (on a wagon) < PFP *aja)6, thereby
providing evidence that it existed, but incidentally no reflex of the IE
loan in question is attested in Udmurt. Koivulehto also thinks that it
belongs to a younger layer of loanwords and therefore does not consti-
tute an exact parallel to our branch-word.
In any case, *wo(j)a is phonetically closer to PIE *u)-s-o- branch,
yielding e.g. Slavic (LCS) *vcha, than *uo)H-h branch (> OI vay0,
OV), incidentally an extension of the same PIE root. All other things
being equal, there is of course no reason why a PIE form with -H-
should be a more obvious source than a minimal pair with *-s- attested
in the same daughter-languages. On the contrary, Ockhams razor
speaks for the latter. Here Koivulehto makes no methodological error,
however, but simply appears unaware of the alternative IE form. Need-
less to say, there are no semantic problems involved in either case.
6
UEW (825-826) thinks that Mordvin (Erzya) aija also shows the old *-j-, only
by metathesis, but this is rejected by Katz (1983: 118) and Koivulehto (1991: 98)
who believe that the Mordvin -j- in this word comes from palatalization of *--.
2.9 P ROTO -M ARI * R MILK PIE (PIIR.) * KSIH -R -

MILK ?
Now let us turn to Katz (2003: 193-194) lone clear example with *-- as a
reflex of an Indo-Iranian laryngeal, Proto-Mari (Proto-Cheremis) *r
milk > Meadow or Central Mari (Carevokokaisk subdial.) r, Hill
Mari (W Mari) er, East Mari (Malmy subdial.) r, Urum Mari
r id.. He asserts PFP *r Pre-PIIr. *kihrm (in his own unor-
thodox notation; > Ved. kr- milk, Oss.Dig. xsir id.). This does not
work either since *r is not a simplex; the base-word in Mari is r
(Moiso & Saarinen 2008: 715). As shown by Aikio (2014: 131ff.), -- in
Mari almost exclusively occurs before *-r- or between *n and *l, even in
loanwords.
2.10 PFS * KETA - TO DARE PIE * G EH - D - BE ASHAMED ?

PFS * RETO LINE , ORDER PIE *( H ) R -TO - LINE ?
PFS * ROTO GRASS , PLANT PIE* G ROH -TO - PLANT ;
GROWTH ?
PFS * TETE DEED PIE* D EH -TI - DEED ?
I will address the remaining examples together (Koivulehto (1988 [1999:

300-301], 2003: 26). They have probably been assigned the right Indo-
European etymologies although one can always discuss the exact source
language and chronological stage7. But note that they are all attested in
Balto-Fennic only where * yields exactly *h! The only protoforms that
we can reconstruct on the basis of direct evidence is BF *kehta- (> Fi.
kehdata, stem kehta- to not be ashamed, Est. khta- to be able to; BF
*rehto (> Fi. rehto row, line; side; various kinds of layer), BF *rohto (>
Fi. rohto, rohtu medicine, (medicinal) plant; weed; green herb; cattle
feed, Est. roht id.); and BF tehte deed (> Est. teht). Meanwhile, in all
these examples the laryngeal occurs in front of a dental stop, and, corre-
spondingly, the Balto-Fennic *-h- precedes -t-. This means that PIE *-h-
cannot provably have been borrowed into *-- in Fenno-Ugric lan-
guages at all; in fact, we have no certain examples of borrowings with
medial laryngeal substitutions outside Balto-Fennic. What we can say is
that we have a handful of examples showing that a PIE aspirated laryn-
7
Note, however, that Middle Proto-Fennic *-kt- yields Late Proto-Fennic *-ht-.
A word like Est. teht deed could therefore just as well be an inner-Balto-Fennic
formation from teke- to do.
geal i.e. *h or *h were transmitted to Balto-Fennic as *h. This is quite

surprising since the loans are probably older than Balto-Fennic itself
which means that there must have existed some kind of back fricative
(maybe not exactly a glottal *h) at least as a marginal loan phoneme in a
stage before Balto-Fennic. Critics would say that this is not necessary
since that stage had exactly * which yielded BF *h anyway, but there is
no direct evidence for this phonologically quite odd intermediate step.
3 Discussion and conclusions
A more minute critical analysis of Koivulehtos and Katzs entire mate-

rial remains to be carried out, but on the basis of the above considera-
tions, I vow to conclude that, while Koivulehtos analysis that PIE initial
laryngeals were substituted by *k- in early loanwords in Fenno-Ugric
(i.e. at least at the Proto-Fenno-Ugric stage and probably even later) is
virtually unchallengeable and counts as a real discovery, his bid for what
happened to medial laryngeals is subject to serious misinterpretations
and a high degree of uncertainty. First of all, the postulation that these
three different phonemes articulated in the back of the mouth were all
substituted with the postalveolar sibilant in all medial positions is hard
to understand on phonological grounds. It is not so much the cross-
linguistic rarity that a back fricative is replaced by a palatal fricative
we would have to accept this anyway if the material showed credible
constistency but the fact that this is combined with an unconditioned
regular outcome of all three phonemes in all medial positions, inde-
pendent of surroundings, makes one suspicious and demands a high
standard of evidence to back it up.
This leads us to the next issue: Most of Koivulehtos etymologies in-
volving word-internal laryngeals are not as good as those with initial
laryngeals. The problem is not so much the semantics, a point where
Koivulehto is markedly more cautious than many of his fellow scholars
(although not necessarily a desirable approach, it is at least an uncon-
troversial one), but rather the fact that he in several cases overlooks al-
ternative and more obvious candidates for Indo-European source words
where *-- can simply be the reflex of a PIE sibilant.
A handful of Koivulehtos etymologies are serious candidates, but
crucially, in these cases the only evidence comes from Balto-Fennic
where * regularly yields h. This leaves us with no uanmbiguous piece of
direct evidence for *-- as a manifestation of an Indo-European larynge-
al since -h- in Balto-Fennic can simply be direct reflexes of the similar

back fricatives in Indo-European Ockhams razor, I would argue, dic-
tates that there is no reason to go through an undocumented stage *--
especially if this stage is hard to account for on phonological grounds.
Except there might be one reason: It is true that *-h- is normally re-
garded to have arisen in Middle Proto-Fennic when a postalveolar sibi-
lants, sequences of stop + such sibilants, the voiced sibilant (in loan-
words only) and the postalveolar affricate all merged into *-h-. My crit-
ics may then object that there was no *h present with which the IE
laryngeals could be substituted because it did not exist as a phoneme
before the Middle-Proto-Fennic stage and at that stage the laryngeals
are normally supposed to have been long gone from Indo-European.
However, we must remember that such thing exists as a loan phoneme,
widespread in the worlds languages. It is directly observable today that
most languages, when adopting loanwords, preserve some foreign
sounds that do not otherwise occur in the system. One can mention as
an example -r- in English loanwords in continental European languages
like Danish or Dutch which retain their original -r-quality8. In Danish,
voiced sibilants and affricates do not occur, not even in English loan-
words, and postalveolar sibilants are commonly substituted with the
more fronted, almost palatal, variants, found in Danish native words.
Thus, [] in English loanwords will be rendered by Danes typically as
[]. The fact that r keeps its English pronunciation in the target language
sets it apart as a loan phoneme. An example of a loan phoneme which
has fully integrated into the system is // in English itself, originally
from French loanwords. In fact, a commonly accepted (and inherently
uncontroversial) hypothesis in Uralistics is that exactly the adoption of
loanwords with *-- may have triggered the emergence of that phoneme
in Proto-Fenno-Ugric. Perhaps, then, the regular rendering of a laryn-
geal as [h] in stages older than Balto-Fennic became one of the trigger-
ing factors that ultimately made Fenno-Saami * (as well as * and *k,
and in loanwords even *z and *) develop into *h, creating a fully inte-
grated phoneme.
However, I must also conclude that we do not know at this point
what the substitution was other than before a dental stop, neither what it
was in Fenno-Ugric languages other than Balto-Fennic, nor what the
substitution of *-h- was in any of the languages. The quest continues.
8
Of course it is subject to linguists interpretation in every individual case when
these loan phonemes start counting as real phonemes.
Koivulehtos important conclusions that a) PIE *h was an aspirated

fricative (not a glottal stop), and b) that laryngeals were still around
even at the time of NW PIE-PF(P) contacts, remain untouched.
References
Aikio 2014 = Luobbal Smmol Smmol nte (Ante Aikio): On the reconstruction
of Proto-Mari Vocalism. Journal of Language Relationship 11: 125-157.
Itkonen, Erkki, 1953-1954: Zur Geschichte des Vokalismus der ersten Silbe im
Tscheremissischen und in den permischen Sprachen. Finnisch-Ugrische For-
schungen 31: 139-345.
Katz, Hartmut, 1983: Hethitisch ia- und Zubehr. Orient 52: 116-122.
Katz, Hartmut, 2003: Studien zu den lteren Indoiranischen Lehnwrtern in den
Uralischen Sprachen. Heidelberg: Winter.
Koivulehto, Jorma, 1981: Reflexe des germ. /e/ im Finnischen und die Datierung
der germanisch-finnischen Lehnbeziehungen. Beitrge zur Geschichte der
deustchen Sprache und Litteratur 103: 167-203, 333-376.
Koivulehto, Jorma, 1988: Idg. Laryngale und die finnisch-ugrische Evidenz. A.
Bammesberger (ed.): Die Laryngaltheorie und die Rekonstruktion des indoger-
manischen Laut- und Formensystems. Heidelberg. Pp. 281-297 [Reprinted with
postscript in Verba Mutuata, Helsinki: Suomalais-Ugrilainen Seura, 1999, pp.
295-308].
Koivulehto, Jorma, 1991: Uralische Evidenz fr die Laryngaltheorie [= SbAW 566].
Vienn: Verlag der sterreichische Akademie der Wissenschaften.
Koivulehto, Jorma, 2001: The Earliest Contacts between Indo-European and Ural-
ic Speakers in the Light of Lexical Loans. Chr. Carpelan & al.: Early Contacts
between Uralic and Indo-European. Helsinki. Pp. 235-263.
Koivulehto, Jorma, 2003: Frhe Kontakten zwischen Uralisch und Indogerman-
isch im nordwestindogermanischen Raum. A. Bammesberger & T. Venne-
mann: Languages in Prehistoric Europe. Heidelberg. Pp. 279-317.
Liukkonen, Kari, 1999: Baltisches im Finnischen [= Mmoires de la Socit Finno-
Ougrienne 235]. Helsinki: Suomalais-Ugrilainen Seura.
Moiso, Arto & Sirkka Saarinen, 2008: Tscheremissisches Wrterbuch [= Lexica So-
cietatis Fenno-Ugricae XXXII]. Helsinki: Suomalais-Ugrilainen Seura.
Skld, Tryggve, 1960: Drei finnische Wrter und die Laryngaltheorie. Zeit-
schrift fr Vergleichende Sprachforschung LXXVI, 1/2: 27-42.
SSA = Ulla-Maija Kulonen & al. (eds.): Suomen Sanojen Alkuper. Etymologinen
Sanakirja. I-III. Helsinki: Suomalaisen Kirjallisuuden Seura and Kotimaisten
Kielten Tutkimuskeskus. 1995.
Stealing the Thunder of alpa:
The Fate of PIE *-bo- in Anatolian1
Abstract
Finnish kalvas and kalpea pale and N Saami guolbben white layer un-
derneath the top soil must be Indo-European loanwords as shown by
Jorma Koivulehto, indicating by their initial *k- that the PIE word for
white, *helbos, had an initial laryngeal. Kortlandts arguments for
deeming the troublesome Hitt. alpa a loanword is supplemented with
the surprising fact that both the PIE nominal suffix *-bo- and the verbal
root extension *-b- are virtually absent in Anatolian. Even PIE roots of
the structure CVRb-, of which some might at least be candidates for
roots containing original *-b-extensions, turn out to be restricted to
one or two examples. This remarkable state of affairs strengthens the
hypothesis (presented in Hyllested 2010) that nominal *-bo- and verbal
*-b- are ultimately identical. It is clear from the material that the use of
*-bo- was already declining in PIE, gradually becoming replaced by
other suffixes such as *-nt- for the present participle. Since Anatolian
was the first branch to split off the IE core, it is logical if use of *-b- was
weakened further in this branch, paving the way for the multifunctional
Anatolian *-nt- that we know so well. It is only to be expected that a few
lexicalized forms with *-bo- be preserved in Anatolian as relics, but
there are simply no unambiguous examples.
1 This paper was presented at the XII. Arbeitstagung of the Indo-European Socie-
ty in Erlangen, September 2011, and is planned to appear as an article in Mn-
chener Beitrge zur Sprachwissenschaft. However, there is still time for revision
and elaboration.
1 The distribution, function and origin of PIE *-bo-
In an earlier article (Hyllested 2010), I have contested the widely-held

views that the PIE nominal suffix *-bo- was used mainly in the for-
mation of a) animal names, b) color adjectives, and c) abstract nouns;
and that it originated by thematicizations of the verbal roots *beh-
shine and *bueh- be ; grow2.
I pointed out3 that animal names with *-bo- hardly ever denote the
same animal in two branches, and that they are virtually all secondary
formations in Indic and Greek where the suffix has been added to a vo-
calic or nasal stem.4 None of them can be safely reconstructed for PIE,
with the possible lone exception of *hl-n-bo- (and this only if Gk.
deer is in fact related to PGmc. *lambaz/*lambiz- sheep
and/or the Gaulish month-name Elembiu).
As for the color adjectives, there are no more than four safe exam-
ples, some of which even have a limited geographical distribution5. Cor-
respondingly, at least 20 PIE colour adjectives occur without a single
attested *-b- added directly to the root in any language.6 Thus, *-bo-
2
The identification of *-bo- with the root *beh- originates from Brugmann
Grdr. Bammesberger suggested that *beh- forms the basis of *-bo- only in
color adjectives, while verbal abstracts would have *-bo- from *bueh-. For a
modernized version of this view, see Balles 2010.
3
For details and examples, I refer to the original article.
4
In Old Indic, -(a)bh-, in most cases with -a- from PIE *-n-, became productive
in the formation of animal names, e.g. r0sa-bha- donkey, ara-bh- grasshop-
per. In Greek, both the conglomerates -a-- (< *-n-b-), ---, --- and
the diminutives ---, --- became productive and were by no means
restricted to animal names; -a- fox (~ orangy); --
underworld demon, an owl (~ unhoed); -- blackbird (~
SCr. kos id.), --- little animal (~ animal, beast), ---
little place, --- little present; trickery, cheating; gambler;
dice-box; and ground, perhaps from *ued- water.
5
These are *delH-bo- yellow (Arm. deb yellow, blond ~ Arm. dein yellow,
wan, pallid, Lat. fulvus dark yellow, Early Mo.Du. dluw, delluw light yellow,
yellowish pale, sallow, fallow; Driessen 2005 : 58); *ro)-bo- striped, spotted
(Lith. rabas grey-spotted, OPr. roaban striped ~ Lith. ranas grey-spotted,
striped, OE rha, rGge roe-deer); *s(u)or-bo- dark red or black (Lat. sorbum
serviceberry, OIr. sorb stain, dirt, Lith. serbent red currant ~ Latv. srts red
in the face) and *hel-bo- white' (Gk. blister, Lat. albus white,
PGerm. *alba- chalky soil [> NHG Albe(n), Dan. alver], Hitt. alpa cloud ~
Lith. aHvas tin, OHG alunt roach, alant elecampane).
6
Apart from 18 roots mentioned in IEW and M&A (Hyllested 2010: 210-211), this
list also includes PIE *deud- brown (Skt. ddhit- epithet to tmas-, Gk.
Stealing the Thunder of alpa: PIE*-bo- in Anatolian 27
cannot have been used specifically for color adjectives, neither in PIE7,
nor in the history of the individual IE branches8. The use of *-bo- in
preference to other adjectival suffixes was not governed by semantics ;
rather, morphophonotactic restrictions seem to have applied:
a) It occurs almost exclusively with roots ending in a sonorant, and

no roots ending in a stop or -s- form adjectives with *-bo- added
directly to the root. The other adjective suffixes *-ro- and *-u- are,
conversely, nearly always added to obstruents (the internal distri-
bution of *-ro- and *-u- being dependent on the roots syllable
peak; see Rasmussen 2010). Roots ending in a laryngeal can appa-
rently take either ending.9
squid, PGmc. *dura- yellow ; dodder [Cuscuta europaea], Toch. B

tute yellow; Schindler 1967) and PIE *(s)le)h- blue (IEW 965, M&A 246
plum-coloured + Gk. lotus; jujube; black tree , Skt. nIla- blue; sap-
phire, fig; Hyllested 2004a + Lith. liis tufted vetch [Vicia cracca], laiys
dogs mercury [Mercurialis perennis]; Gliwa & Hyllested 2006).
7
such as *5o)H-bo- swift > Ved. bhm adv. fast, OHG heif-tg ~ ghr-
swift, OE hgian strive for ; *lo)h-bo- weak > Lith. libas, lebas thin, lean,
OS lf ~ Lith. lelas, linas thin, ON linr weak, lean, Gk. hunger ; and
and *n)-bo- > OPers. naiba- good, OIr. nib sacred ~ Lat. nite shine, MIr.
na hero).
8
E.g. OI sthla-bh- thick ~ sthl- id. ; Gk. stiff, hard; infertile ~
, ON starr stiff ; and Alb. n-gjel-b-t salty (< *en-sal-bo-). In
Hyllested 2010, I included the example PGmc. *hal-ba- half (< *5ol-bo- ~ Lith.
als side). The connection with Baltic is, however, uncertain ; it is not favored
by Orel 154 and not even mentioned as a possibility in Kroonen forthcoming,
and even if the connection is correct, there are no obvious candidates for co-
gnates outside the Northern European branches. Nonetheless, no alternative
etymologies are generally accepted. I now believe that PGmc. *halba- is a loan
from Fennic, cf. Fennic *halpa- reduced, gen. *halan-, which cannot be a loan
the other way round because the Fennic word comes from *alV cheap. The
motivation for borrowing a word for reduced, cheap would exactly be natural
in a trade context which the Fennic meanings point to. Later, the Indo-
European term for half would have been replaced by a semantically bleached
and functionally strengthened version of the Fennic word.
9
The sole exceptions are inner-Greek formations of which only smooth,
flat (~ id.) seems to be of non-onomatopoeic character; hip
probably does not reflect **host-bhu- (~ *host- bone), but is rather to be seg-
mented *-- and contains the root in ankle and hurry <
*spu-H/d- move rapidly (cf. the connection between Eng. hip and hop). Per-
haps it is even a compound *host-spu-(H/d-)-s of the same type as OI
ahv(nt), Av. ascuua- shin-bone < IIr. ast-(s)iHua- where the second mem-
b) Formations with *-bo- directly added to roots beginning with a

labial stop (*p-, *b-) are avoided. We see no formations such as
ber-bo- or pelh-bo- in the material.
When -bo- occurs in deverbal nouns, these are very often result
nouns10, but may also have retained an action-noun character11, or re-
flect earlier agent nouns12 . In some cases, the distinctions are not clear13.
The multifunctional use of *-bo- constitutes an almost exact parallel to
Mod. Eng. -ing; thus, a form like *skerbo- corresponds to Eng. cutting
which is not only a present participle, but also an adjective meaning ca-
pable of or designed for cutting, an action noun meaning the act of cut-
ting and a result noun meaning a part cut off from a main body; a clip-
ping.
The use of *-bo- for the formation of verbal nouns was rapidly
declining at the time of the dissolution of PIE and remained productive
in the individual branches only when accompanied by other suffixal
elements14. The oldest function of *-bo- was the formation of present
participles, indifferent to voicei.e., both active and passive present
participles.
ber is *(s)kiHuo- shin-bone(Lubotsky 2002). Hitt. wapa shroud (Kloekhorst

2008: *uos-bo-) probably contains -p- and not *-bh(o)-, see below.
10
E.g. PIE *gol-bo- > Goth. kalbo, OE cealf calf, Gaul. galba fat person ~ Lat.
glomus bunch, wad ; PIE *tuH-bo- swelling > OIr. taimm heap, Gk.
pillow covering, tick, Lat. tber tumor ~ Skt. tla- n. (cotton)wad), OHG
griubo crackling, tear strip < tearing off; cf. Eng. a cutting.
11
E.g. Arm. ob lamentation , Gk. id. (~ lament with se-
condary -- from the synonymous ; Olsen 1999 : 37)
12
E.g. PGmc. *wamb- rumen (Goth. wamba paunch, womb, Dan. vom, . vm,
vom rumen, paunch ~ Latv. viMbas pl. vomit, spit ~ *uem(h)- vomit. Com-
pare Lat. rmen ~ Skt. romantha- chewing the cud.
13
E.g. Lith. grba, garbN (an) honour ~ giri (to) honour, PIE *lm-bah fe-
male water spirit > Gr. nymph (> Lat. lumpa id., lympha clear water),
Skt. Rmbh name of a water nymph ~ Lith. LaumN water fairy, older Alb.
lumet the fairies ~ to enchant, bewitch; Hyllested 2004b) may be an agent
noun or (a concretization of) an action noun.
14
For example, Baltic verbal abstracts in *--b- (Lith. darba building ~ darti
build; Latv. medba hunting ~ medt hunt; cerba hope (subst.) ~ cert hope
(vb.) ; Lith. nominal abstracts in -b, often concretised/individualised
(grab beauty; beautiful girl); and Slavic verbal abstracts *-V-ba and result
nouns in *-V-b (OCS zloba evilness ~ zl evil; gost-ba party ~ gost
guest, SCr. stube ladder, steps ~ CS stlati spread, stretch).
The notion that *bo- derives from thematicization of the verbal

roots *beh- shine and *bueh- be ; grow can be easily contested on a
number of grounds15 :
a) The use of the suffix is much broader than initially described by

Brugmann ; thus, *beh- shine makes little sense in most adjec-
tives in question ;
b) The physical similarity between suffix and root is limited to a
single, very frequent consonant;
c) There are real compounds like Gk. enormous, marvel-
lous ~ Lat. superbus and Ved. bhva- n. monster < *-bhuo-
(Kuiper 1962, Meier-Brgger 1991) which retain -u-;
d) The alleged modifying effect of *-bo- (looking like X, of Xs
kind, shining X) lies in the modifying nature of derivation itself
The recurring extension *-b- in verbal roots (*gle)- ~ *gle)-b- smear;

*heu- ~ hu-ebh- weave ; *steh ~ *st-b- stand etc.) is ultimately
identical to nominal *-bo- and reflect either lexicalized participial stems
or simply the use of present participles for the 3rd person finite. Lexica-
lized or parallel formations common to IE and Uralic reveal that PIE *-
bo- must be ultimately related to the PU present participial ending *-
pa, indifferent to voice, which is also used as the marker of the (original-
ly unmarked) 3rd person marker of the verb.16 A parallel development
took place in Indo-European where another participial element, *-t- ~ *-
nt-, came to occupy that function. Formally, then, nothing distinguishes
IE verbal root variants with an extra -b- from the 3rd person of the
Uralic verb.
15
For Slavic -ba, this idea was first conceived by Iljinskij (1902).
16
The verb to cut, PIE *sker- and PU *kere-, provides the largest number of ex-
amples. With *-bho- and *-pa-, respectively, added to the naked root, the verb
to cut acquires shared specialized meanings such as to be sharp or to
scratch, and as a noun it means crust, whereas a suffix PIE *-i- ~ *-ja- added
before it, typically occur in derivatives denoting stripes, lines or pattern and
in verbs meaning to incite (later to write). Gk. sketch, outline;
stylus, Lat. scrb write, Latv. skrpa scratched stripes ~ Est. kirjav striped,
spotted; Fi. kirjava; ~ kirja pattern, figure, script > book). When other suf-
fixes replace *-pa, unpredictable meanings are still shared by Indo-European
and Uralic (see Hyllested 2010 for details).
2 Alleged examples of *-bo- in Anatolian
In any case, the use of *-bo- as a suffix must be old in PIE. It is there-
fore highly surprising that both nominal *-bo- and the verbal extension
*-b- are virtually absent in Anatolian. In six cases, *-bo- has been sug-
gested as the source for what seems to be a derivational ending, but they
can either be refuted right away or remain disputed. Let us have a look
at the candidates:
2.1 L YCIAN XAHBA GRANDCHILD
Lyc. xahba (suggested by Shevoroshkin 1979: 179, fn. 5), is now known
to have meant grandchild and not ruler, king; it goes back to earlier
*aswa- which is in itself a thematicization of *Honsu- > Hitt. au-,
HLuw. asu- (Melchert 1994: 63, 307).
2.2 L YCIAN XTABA RULE
Lyc. xtaba (suggested by Shevoroshkin 1979: 178-179) is recte xntawa-

rule, and its source is not Luw. anda(i)- determine, fix, arrange,
which rather means care for, but the stem ant- ' front', cf. Luwian
antawat(i)- 'king' (Zsolt Simon, p.c.).
2.3 C ARIAN - BA - IN PLACE - NAMES
It was suggested by Neumann (1988: 187 and n. 4) that the recurrent

element -ba- in Carian place-names derive from PIE *-bo-. Most of the
examples mentioned by Neumann are, however, etymologically quite
obscure, and some of them surely continue Anatolian *-wa-, e.g. kib
city of Kindy < *Hinduwa- (cf. Simon 2008). A sound law PA *w >
Carian b/C_ can be established on the basis of kib and ksbo- PN <
*aswa- (HLuw. asu-, Lycian ahba- grandson; grandchild; Simon
2008: 334).
2.4 H ITTITE ALPA - DOG S EXCREMENT
Hitt. alpa- dogs excrement does not for sure reflect *sl-bo- grey;
filthy (> Arm. ab dung) ~ *sal-uo-, *sl-o- dirty, grey; dirt
(Schindler 1978; see also Olsen 1999: 37); an alternative source is still
PIE *solp-o- derived from *selp- grease; greasy (Sahowkyan 1987). If
indeed derived from *sal-, this item would stand alone in the sense that
*-bo- would form a substantive and not an adjective. In that case, it
seems appropriate either to a) reconstruct an intermediate adjective
*sal-bo- dirty which later became substantivised or b) to assert a ver-
bal meaning of the root be dirty, produce dirt (cf. as a parallel the
double meaning of Dan. griset dirty, filthy and messy) which obtained
the meaning dirt as a kind of result noun. In any case, alpa- cannot
count as a safe example of a derivative with PIE *-bo-.
2.5 H ITTITE WAPA CLOTHES ( OF THE DEAD ) ; SHROUD
Hitt. wapa clothes (of the dead) ; shroud (Goetze 1969) is related to
Lat. vespillo undertaker; grave robber (Watkins 1969) and derived from
*ues-p- dress (see also Katz 2000). Kloekhorst (2008) reconstructs
*uos-bo-, morphonotactically illicit according to me; I dont see any
reason not to accept *-p- in this context. Even so, Kloekhorst might be
right that we are ultimately dealing with the same morpheme because
if *-bo- does not occur following -s- and *-p-o- does, we could argue
that the extension *-p- in fact reflects an allomorphic variant of -b-.
2.6 H ITTITE ALPA CLOUD
Hitt. alpa cloud constitutes a problem because the expected reflex of

*h- does not surface (Lubotsky 1989). Kortlandt (2003: 11) argues that it
is a loanword from a non-Anatolian language. He gives five reasons:
a) it is not found in Indo-Iranian or Tocharian

b) it has a variant *elb- in Slavic
c) it has an alternating suffix -it-, -ut- in Slavic and the same suffix
with an infixed nasal in Slavic in the word for swan
d) it plays a role in Germanic mythology (cf. Eng. elf)
e) it is frequent in European geographical names (Alba, Albion, El-

be, the Alps)
We might add
f) the fact that -pa- < *-bo- is virtually non-existent in the older
Anatolian languages
g) The initial stop in Fi. kalvas, kalpea pale and North Saami
guolbben (regularly < *kalpen-; ~ NHG dial. Alven, ODan. alur
[mod. al] chalky sand underneath the top soil; sandy plain) di-
rectly reflects a laryngeal in PIE *hl-bo- (a joint etymology by
Petri Kallio and Jorma Koivulehto, see e.g. Kallio 1998 and Koi-
vulehto 2003: 289, 298);
The seemingly absent initial laryngeal is thus secured by loans in Fennic.

I do not see any semantic problem in connecting alpa with Lat. albus
white and its cognates (as a loanword from a non-Anatolian, but still IE
language), pace Puhvel HED 1/2: 38 and Kloekhorst 2008: 169; it does
not always refer to dark thunder clouds, which can easily be covered by
a generic cloud term anyway. Neither does the unique Hittite meaning
need worry us since the motivation for the borrowing was most likely
mythological, either going via vapour, spirit as in Gmc. *albi- white
creature connected with the fog (an original dichotomy of white ljsal-
far as opposed to the dark dkklfar, cf. also NHG Weie Frauen, Dutch
Witte Wieven) or upper world as in Celtic, cf. also Eng. sky ~ ON sk
cloud.
One might visualize a connection with Hitt. alpant- if this is a variant
of alwanz- being bewitched, affected by sorcery (Kloekhorst 2008: 171).
But note that the similarity with Turkic arba hexen, bezaubern, wahr-
sagen, at first glance of course superficial, constitutes a parallel to Gk.
n. barley-groats, Alb. elp, -bi barley vs. Turkic arpa barley,
Mong. arbaj id. The narrow semantics is in both cases coupled with a
correspondence between -l- in the Indo-European forms and -r- in Alta-
ic. At the same time, a form *arpa also occurs in Uralic word for
withcraft. As is well known, religious and agricultural terms are both
typical loanwords. While Blaek (2012) prefers an Indo-European (Ira-
nian) origin of the Altaic words for barley, several facts do point to a
borrowing in the reverse direction. First of all, the word is unusually
common in the Turkic languages (Stachowski 12); the Turkic word is
already regarded the source of the Mongolian and Tungusic forms;
within Iranian, the word is not found oustide East Iranian (Stachowski);
and its only cognates are found in Greek and Albanian, both Balkan
languages. As Stachowski writes, most previous works have uncritically
quoted previous works about the possible Iranian origin of the Turkic
term. Tatarincev (2000) suggests that the word is an inner-Turkic deriv-
ative, formed by *ar- multiply oneself, be numerous with a suffix de-
noting intensification, cf. Old Turkic arka multitude; collection; crowd,
group, Mong. arbin plentiful. Martin (1987) and Omodaka (2000)
have added OJap. *apa millet as a plausible cognate; I do not see how
Stachowski can conclude that this speaks for Tatarincevs inner-Turkic
derivation, but in any case it strengthens the hypothesis that we are deal-
ing with an Altaic agricultural term of great age.
If the barley-word is indeed of Turkic or even Altaic origin, it seems
justified to hypothesize a similar origin of Hitt. alpant- ~ alwanz-. The
lambdacization in either word does not have to have happened after the
borrowing since confusion between liquids is a common phenomenon
already within older Altaic languages (Granberg 2008). However, there
is also a possibility that *arpa is a Uralic word borrowed into Turkic at
an early stage if -pa could be identified as the participial suffix.
Whatever the exact history of these two words, most signs point to an
extra-Indo-European origin of both of them. Hence, alpant- and al-
wanz-, as well as the designations for barley, should be kept apart from
alpa until stronger evidence for a connection shows up.
3 Verbal *-b- and roots of the structure CVRb- in Anatolian
Let us now have a look at Anatolian verbal roots of the structure CeR(-
)b-, since these are all roots that could possibly contain a verbal exten-
sion *-b- :
3.1 H ITTITE KARP - IYEZZI TO TAKE ( AWAY ), LIFT ( UP ), PLUCK
Kloekhorst (1998: 453) derives Hitt. karp-iyezzi to take (away), lift up,
lift, pluck from PIE *(s)kerp- (Lat. carp pick, pluck etc.), as opposed
to Oettinger (1979: 345) who traces it back to PIE *greb()h- to dig.
Even in the latter case, it seems to have root-final *-h- (see Olsen 1993)
and thus does not count as an original example of *-b-.
3.2 H ITTITE KARAPI , KARE / IPANZI TO DEVOUR
Kloekhorst (2008: 442-444) derives Hitt. karapi, kare/ipanzi to devour

from the root *gerbh- (> Skt. grbhnti to seize and ON grpa id.) But
ON grpa continues < PGmc. *grpP- < PIE *grb-n- rather than
*grb- (Kroonen 2012). Instead this verb should be grouped with Nw.
garva, garpa, gurpa devour, gobble, belch, assuming the doublets reflect
an ablauting iterative doublet *garppi, *garbunani ~ *gurppi,
*gurbunani < *g(o)rb-nh-ti, *gh(o)rb-nh-nti. In either case, the
aspiration of the root-final *-b- seems not to be original, but reflects
PIE *-b()-H- (see again Olsen 1993).
3.3 H ITTITE ARP ( P )- TO CHANGE ALLEGIANCE
Kloekhorst (1998: 311, 2008: 442-444) connects Hitt. arp(p)- to change

allegiance etc. to PIE *herb- (> Gk. orphan etc.). This is the
only certain example of a final *-b- in a triconsonantal root, but it still
bears no signs of having resulted from extension of a shorter root.
3.4 H ITTITE UPPIYA - THROWS , HURLS
In Hyllested & Cohen (2007), our aim was to show that it is phonologi-
cally unproblematic to link Gk. weave to Hitt. uwapp- (alleg-
edly interlace, entangle, Puhvel 1991), despite the lack of prothetic vow-
el in Greek. This is because there are no examples of initial u-diphthongs
before a labial in Greek except for late inner-Gk. formations; both full-
grade *(H)euP- and zero-grade *HuP- regularly yield Proto-Gk. *uP-.
Recently, however, Melchert (2007) has shown that uwapp- rather
means throw, hurl. This obviously does not contradict the Greek rule,
but it does remove one important piece of positive evidence, and, more
importantly, it seems to undermine the evidence for *h in weave (cf.
also van Beek 2011).
Since Neu (1998), another candidate for a cognate of weave etc. has
been the hapax wepu wpta in the fragment KBo. 42.6, 9 (13th c. BC)
whose exact interpretation is still debatable:
(8) [...]-zi-mi-i D-a ar-ru-ma-ar e-ep-t[a ...]

(9) [...]x-ni -e-pu-u -e-ep-ta nu=mu TG-an=mi-i[t ...]
[sby] took the washing of the river [or in the rivers]

[sby] wep-ed wep-s and [ed] my clothing for me
Reconstructing PIE weave as either *hueb- or *ueb- is problematic

since it precludes a -b-extension of PIE *heu- id. > e.g. Skt. vyati,
Lith. udiu (< *heu-d-), whose initial laryngeal is needed to account
for the initial *k- in the Fenno-Permian loan *kua- weave (> Fi. kuto-,
N Saami godde-, Mordv. koda-, Komi kyj-).
Furthermore, the VN arrumar washing is mentioned before the wep-
sequence, suggesting this does not refer to fulling. It could refer to the
washing of wool fibers before the preparation for spinning, but this was
carried out in hot water (i.e. not a river), and the process involved a lot
of intermediate activities drying, beating, cleansing, carding, grading,
bleaching not mentioned in the fragment (Breniquet 2010).
Since the concept of throwing is central to also to ancient weaving,
cf.:
a) Eng. warp ~ OE weorpan to throw and

b) to throw the shuttle
it could be that uwapp- belongs with weave after all, having preserved
an original PIE meaning that was specialized in Core IE after the Anato-
lian split-off; Andrs-Toledo (2010), too, suggests a late semantic nar-
rowing, but from an original meaning bind, interlace, based partly on
the now rejected Hitt. meaning, and partly on Indo-Iranian which also
displays the meaning weave. As Melchert notes, the Hitt. hapax pala-
fish-net does not have to be derived from a verb weave because a net
is something you cast out.
The sumerogram TG clothes represents Hitt. wapa (Goetze
1969) which often specifically means shroud; clothes of the dead, hence
Lat. vespillo undertaker; grave robber (Watkins 1969) < *ues-p- dress
(see also Katz 2000). It occurs elsewhere in Hitt. texts that a dying man
himself is calling for his shroud or his washing. In the Old Hittite-
Akkadian Testament ( 3, Kbo III 64-73, Melchert 1991: 183), the dying
king Hattuili says wash me well; protect me at your bosom from the
earth (Melchert 1986) and the Soldiers Dirge reads
Nea wape Nea wape

tiya=mmu tiya
nu=mu anna=ma katta arnut

tiya=mmu tiya
nu=mu uwa=ma katta arnut
tiya=mmu tiya
shrouds of Nea, shrouds of Nea

wrap me, wrap
put me down for burial with my mother
wrap me, wrap
put me down for burial with my forefathers
wrap me, wrap
It is noteworthy that the passage makes use of alliteration (involving

four words):
pta wepu wpta *wapan
B ecause a figura etymologica of a similar shape, again used with a word

for toga in the acc., occurs in the S Picene epitaph TE 2 from Bellante:
postin : viam : videtas : tetis : tokam : alies : esmen : vepses : vepeten

along the road you see / the toga (or covering) of Titus Allius (?) / bur-
ied (?) in this tomb (?)
This stylistic feature is of PIE age (Watkins 1995: 131-133, Fortson 2002:
73), and the SPic. vep- even occurs in non-etymological alliterations such
as veiat vepet lies in the tomb in MC 1 from Loro Piceno:
apaes : qupat: esmn : ppnis : nr : mefin : veiat : vepet

The elder lies, the Picene chief, in the middle of the tomb
Correspondingly, Ved. vap- strew and its derivative vapu- wonder,

transposed to meanings like color; covering; clothing, forms an allitera-
tive pair with vas- to wear, to dress (Katz 2009, Jackson):
RV 3.55 14 (Heaven and Earth according to Syaa; but perhaps Dawn

and her Sun-god husband, Srya)
pdy vaste pururUp vpsy rdhv0 tasthau tryvi rrih

rtsya sdma v carmi vidv0n mahd dev0nm asuratvm kam
Unten kleidet sich die Vielfarbige in schne Formen; sie richtet sich
empor, das anderthalbjhrige Rind leckend. Ich durchwandere als Wis-
sender die Sttte der Wahrheit. Gro ist die einzige Asuramacht der
Gtter.
RV 1.160, 2 (Heaven and Earth):
uruvycas mahn asact pit0 mt0 ca bhuvanni rakata

sudhrame vapuy n rdas pit0 yt sm abh rpar vsayat
Breitrumig, gromchtig, nie versiegend, behten Vater und Mutter

die Geschpfe. Die sehr kecken (?) Rodas sind wie zwei schne Frauen,
da der Vater sie in Farben kleidete.
MS 3.11.9
srasvati manas pealam vasu n0satyabhy vayati daratm vpu

pealam, cf. RV 1.92.4 (Dawn): psi vapate, but no figura etymologica
(Katz 2009)
It is thus conceivable that KBo. 42.6, 9 describes a burial rite with a dy-
ing or even dead person speaking, and that both the Hittite and South
Picene items represent PIE *uep- to adorn, to make ready by adorning
(pace Meiser 48-49). Katz (2009) adds to this root Gk. marry on
the basis of a new sound-law for Greek that makes *u- disappear in this
context.
I see no reason to leave out ORu. vap color, vapno chalk, OPr.
woapis color and Latv. vpe glaze from this family; cf. the parallel in
OPr. sirmen funeral rite ~ sirmes washing lye made of ashes ~ Lith.
irmas white; grey (Gliwa 2005). On the concept of color in prehistoric
funeral rites in general, see Jones-Bley (2005).
The spelling with single -p- in wepu is problematic, but it is a hapax
preceding wpta which may have influenced it. The plene spelling of
uwapp- is no less problematic, but at least the semantic comparison
between a verb meaning to throw and to weave need not be.
Katz connects them with Hitt. wappu- riverbank (< heaped-up
earth) and Skt. vpra- heap of earth via the meaning heaped up (fin-
ery). I propose an alternative: These belong with *feraz ~ *feran-
bank, shore (MHG uover, NHG Ufer etc.), Gk. the land as op-
posed to the sea and perhaps Lith. up river. This would be the only
case of expected PGmc. *wf- ~ *wb-, and it is conceivable that such a
sequence with two labial fricatives and a rounded vowel in the middle
would be subject to dissimilation.
4 Conclusions
The lack of evidence for both nominal *-bo- and verbal extensions in
*-b- in Anatolian strengthens the hypothesis that these two elements
are ultimately identical. They were not derived from verbal roots in PIE;
rather do they belong to a more distant past where they formed present
participles indifferent to voice (like PU *-pa), and, like Eng. -ing, it end-
ed up synchronically as a derivational suffix for both agent nouns, ab-
stract nouns, result nouns and adjectives. Its occurrence in animal-
names is language-specific, based on substantivizations of color adjec-
tives. PIE *-bo-, thus already declining as a participle marker, gradually
became replaced by *-nt- (which also has a counterpart in Uralic). Since
Anatolian broke off the core first, it is logical if the tendency was weak-
ened further (and *-nt- correspondingly strengthened) in this branch.
As is well-known, the use of *-nt- in Anatolian goes far beyond the for-
mation of participles. We would expect a few lexicalized forms with
*-bo- to be preserved as relics, although not necessarily for us to study
as attestations in the corpus. What is relevant is not whether we can
eliminate the examples altogether, but that we have so few of them in
any case.
References
Adiego, Ignacio J., 2007: The Carian Language. Leiden / Boston.

AiGr = Jacob Wackernagel (Albert Debrunner et al.) (1896)/1930-1957: Altindische
Grammatik. Vols. I-III. Gttingen: Vandenhoeck & Ruprecht.
Andersen, Henning, 1998: A Glimpse of the Homeland of the Slavs: Ecological and
Cultural Change in Prehistory. A. Della Volpe (ed.) 1998, pp. 1-67.
Andersen, Henning, 2003: Slavic and the Indo-European Migrations. H. An-
dersen (ed.) 2003, pp. 45-76.
Andersen, Henning (ed.), 2003: Language Contacts in Prehistory [= Current Issues
in Linguistic Theory 239]. Amsterdam / Philadelphia: John Benjamins.
Andrs-Toledo, Miguel ngel, 2010: Some considerations about Vedic, Avestan
and Indoiranian Textile Terminology. Michel & Nosch pp. 431-444.
Balles, Irene, forthcoming: The Old Indic cvi construction, the Caland system, and
the PIE Adjective. J.E. Rasmussen & T. Olander (eds.): Internal Reconstruc-
tion in Indo-European. Copenhagen: Museum Tusculanum Press.
Bammesberger, Alfred, 1973: Abstraktbildungen in den baltischen Sprachen. Gttin-
gen: Vandenhoeck & Ruprecht.
Bammesberger, Alfred, 1998: Die etymologische Herleitung von lateinisch mor-
bus. Studia Etymologica Cracoviensia (SEC) 3.
van Beek, Lucien, 2011: Vowel Assimilation in Greek: The Evidence Reconsid-
ered. Th. Krisch & Th. Lindner: Indogermanistik und Linguistik im Dialog.
Wiesbaden: Reichert. Pp. 49-58.
Benveniste, mile, 1949: Noms danimaux en indo-europen. Bulletin de la So-
cit Linguistique de Paris 45, pp. 74-103 (III. Noms du mle, pp. 100-103).
Breniquet, Catherine, 2010: Weaving in Mesopotamia during the Bronze Age.
In Michel & Nosch pp. 52-67.
Brugmann Grdr. = Karl Brugmann, 1897-1916: Grundri der vergleichenden Gram-
matik der idg. Sprachen I-III. Strasbourg: Karl J. Trbner.
Campanile, Enrico, 1983: Le Restsprachen e la ricerca indoeuropeistica. Edoardo
Vineis (ed.), Le lingue indoeuropee di frammentaria attestazione Die indo-
germanischen Restsprachen [= Biblioteca della Societ Italiana di Glottologia
4]. Pisa: Giardini, pp. 211-224.
Carpelan, Christian, Asko Parpola & Petteri Koskikallio (eds.), 2001: Early Contacts
between Uralic and Indo-European: Linguistic and Archaeological Considerations
[= Mmoires de la Socit Finno-Ougrienne 242]: Helsinki: Suomalais-Ugrilaisen
Seura.
Chantraine DELG = Pierre Chantraine, 1968-1980: Dictionnaire tymologique de la
langue grecque. Paris: Klinksieck.
op, Bojan, 1973: Prispevek k zgodovini labialnih pripon v indoeuropskih jezikih /
Beitrag zur Geschichte der Labialsuffixe in den indogermanischen Sprachen [=
Razred za filoloke in literarne vede, Classis II: philologia et litterae, dela / opera
29]. Ljubljana: SAZU.
Della Volpe, Angela (ed.), in cooperation with Edgar C. Polom, 1998: Proceedings
of the Seventh Annual UCLA Indo-European Conference 1995 [= JIES Studies and
Monograph Series 28], Washington, D.C.: Institute for the Study of Man.
Driessen, C. Michiel, 2005: On the etymology of Lat. fulvus. Meiser & Hackstein
2005, pp. 39-64.
Fortson, Benjamin W., IV, 2005: Linguistic and cultural notes on Latin Inius and
related topics. Mark R.V. Southern (ed.): Indo-European Perspectives [= Jour-
nal of Indo-European Studies Monograph Series 43]. Pp. 61-77.
Frisk GEW = Hjalmar Frisk, 1960-1972: Griechisches etymologisches Wrterbuch.
Heidelberg: Carl Winter Universittsverlag.
Gliwa, Bernd, 2005: Baltische Bestattungsrituale im Spiegel ihre Bezeichnungen:
litauisch ermens und arvjimas, altpreuisch sirmen. Acta Linguistica
Lithuanica 53: 9-21.
Gliwa, Bernd, 2007: Altpreussisch tubo filcz (E 488). Res Balticae 11.
Gliwa, Bernd & Adam Hyllested, 2006: Baltisches zu Idg. *sle)H- blau (sein).
Baltistica XLI (1), pp. 87-93.
Goetze, Albrecht, 1955: Hittite Dress. Corolla Linguistica [Fs. Sommer], pp. 48-
62.
Granberg, Antoaneta, 2008: The Hunno-Bulgarian language. The DSCA Journal

2008:
Hyllested, Adam, 2004a: Greek lotus and the Indo-European Words for
blue. James Clackson & Birgit Anette Olsen: Indo-European Word For-
mation. Copenhagen: Museum Tusculanum Press, pp. 59-64.
Hyllested, Adam, 2004b: L'esprit des eaux: grec , sanskrit Rmbh, lituanien
LaumN et quelques autres formes semblant apparentes. Per Aspera ad Aster-
iscos [Fs Rasmussen]. Innsbruck: IBS, pp. 219-233.
Hyllested, Adam, 2004c: Review of Andersen (ed.) 2003. Acta Linguistica Haf-
niensia 36: 179-184.
Hyllested, Adam, 2009: PIE *-b- in Verbs and Nouns: Distribution, Function,
Origin. R. Lhr & S. Ziegler (eds.): Protolanguage and Prehistory. Wiesbaden:
Reichert. Pp. 202-214.
Hyllested, Adam, 2010: Internal Derivation vs. External Comparison: The Case of
the Indo-Uralic Laryngeals. Jens Elmegrd Rasmussen & Thomas Olander
(eds.): Internal Reconstruction in Indo-European. Copenhagen: Museum Tuscu-
lanum Press.
Hyllested, Adam & Paul S. Cohen, 2007: Monophthong for Expected -diphthong
in Greek. Coulter George, Matthew McCullagh, Benedicte Nielsen, Antonia
Ruppel & Olga Tribulato (eds.), Greek and Latin from an Indo-European Per-
spective [= Proceedings of the Cambridge Philological Society Supplementary Vol-
ume 32]. Cambridge: Cambridge Philological Society.
IEW = Julius Pokorny, 1959: Indogermanisches Etymologische Wrterbuch. Tbing-
en / Basel: Francke.
Iljinskij, Grigorij, 1902: Zur slavischen Wortbildung. Archiv fr Slavische
Philologie 24: 224-228.
Sahowkyan, G.B., 1987: Hayoc lezvi patmowtyown. Yerevan.
Janda, Michael, 2001: Review of Della Volpe (ed.) 1998. Kratylos 46 : 202-205.
Janhunen, Juha, 1982: On the Structure of Proto-Uralic. Finnisch-Ugrische For-
schungen 44: 23-42.
Jokl, Norbert, 1911: Studien zur albanesischen Etymologie und Wortbildung.
Sitzungbericht der Wiener Akademie der Wissenschaft Philosophisch-historischer
Klasse 168, Teil 1.
Jones-Bley, K.arlene, 2005: Red for the Dead. D. Disterheft, M. Huld & J. Grep-
pin: Studies in Honor of Jaan Puhvel I: 211-2005.
Kallio, Petri, 1998: Kalja. Sananjalka 40, pp. 87-93 (abstract: Kalja beer, p.
94).
Katz, Joshua T., 2000: Evening Dress: The Metaphorical Background of Latin
uesper and Greek . K. Jones-Bley, M.E. Huld & A. Della Volpe (eds.):
Proceedings of the Eleventh Annual UCLA Indo-European Conference. Washing-
ton D.C. Pp. 69-93.
Katz, Joshua T., 2009: Greek and its digamma. Ms. / paper presented at
Indo-European Matters in Copenhagen 9 Oct, 2009.
Kloekhorst, Alwin, 2008: Etymological Dictionary of the Hittite Inherited Lexicon.
Leiden: Brill.
Koivulehto, Jorma, 2001: The Earliest Contacts between Indo-European and Ural-
ic Speakers in the Light of Lexical Loans. Carpelan, Parpola & Koskikallio
(eds.): pp. 235-263.
Koivulehto, Jorma, 2003: Frhe Kontakte zwischen Uralisch und Indogermanisch

im nordwestindogermanischen Raum. Alfred Bammesberger & Theo Ven-
nemann (eds.): Languages in Prehistoric Europe . Heidelberg: Winter. Pp. 279-
317.
Kortlandt, Frederik H.H., 2003. Initial Laryngeals in Anatolian.
http://www.kortlandt.nl/publications/art202e.pdf [Also available in print: Or-
pheus 13/14: 9-12 (2003/2004).]
Kroonen, Guus, 2012: Reflections on the o/zero-ablaut in the Germanic iterative
verbs Craig Melchert (ed.): The Indo-European Verb. Wiesbaden: Reichert.
Pp. 191-200.
Kuiper, Franciscus Bernardus Jacobus, 1962: Atharvavedic abhv-, n. monster.
Lingua 11: 225-230.
Le Feuvre, Claire, 2004: Note sur ladjectif , , ,.
Revue de philologie, de littrature et dhistoire anciennes 2004/2, Tome LXXVIII:
257-264.
Lehtisalo, T., 1936: ber die primren Uralischen Ableitungsuffixe. [= Mmoires de
la Socit Finno-Ougrienne 72]. Helsinki: Suomalais-Ugrilaisen Seura.
Lubotsky, Alexander, 2002: The Indo-Iranian word for shank, shin. Journal of
the American Oriental Society 122: 318-324.
M&A = James P. Mallory & Douglas Q. Adams (eds.), 1997: Encyclopedia of Indo-
European Culture. London / Chicago: Fitzroy Dearborn.
McCone, Kim, 2005: Mgliche nicht-indogermanische Elemente in den keltischen
Sprachen und einige frhe Entlehnungen aus indogermanischen Nachbarspra-
chen. Meiser & Hackstein 2005, pp. 395-435.
Meier-Brgger, Michael, 1991: Griechisch Bratfischchen, ved. bhva-
Unding, myk. Ortsname a-phu-. MSS 52: 123-125.
Meiser, Gerhard, 2003: Veni Vidi Vici. Die Vorgeschichte des lateinischen Per-
fektsystems. Munich.
Meiser & Hackstein 2005 = Gerhard Meiser & Olav Hackstein (eds.): Sprachkontakt
und Sprachwandel. Akten der XI. Fachtagung der Idg. Gesellschaft. Wiesbaden:
Dr. Ludwig Reichert Verlag.
Melchert, Craig, 1991: Death and the Hittite king. Roger Pearson (ed.): Perspec-
tives on Indo-European language, culture, and religion. Studies in honor of Edgar
C. Polom [= Journal of Indo-European Studies monograph 7]. Vol. 1, pp. 182-
188.
Melchert, Craig, 1994: Anatolian Historical Phonology. Leiden: Rodopi.
Melchert, H. Craig, 2007: Hittite huwapp-, huppa(i)- and huppiya-. D. Groddek
& M. Zorman (eds.): Tabula Hethaeorum [Fs. Koak]. Wiesbaden. Pp. 513-519.
Michel, C. & M.-L. Nosch, 2010: Textile Terminologies in the Ancient Near East and
Mediterranean from the first to the third Millennium. Oxford / Oakville.
Neu, Erich, 1998: Hittite i-mu-u. HS 111, 1: 55-60.
Neumann, Gnther, 1988: Beobachtungen an karischen Ortsnamen. F. Imperati
(ed.): Studi di storia e di filologia anatolica dedicati a Giovanni Pugliese Carratel-
li. Firenze
Oettinger, Norbert, 1979: Die Stammbildung des hethitischen Verbums. Nrnberg.
Olsen, Birgit Anette, 1993: Vedic and Laryngeals: irajyati and iradhanta badhnti
and ubhnti. G. Meiser (ed.): Indogermanica et Italica. Innsbruck. Pp. 362-
372.
Olsen, Birgit Anette, 1994: Armenian dalowkn jaundice and the IE suffixes *-
gon-, *-gon- and *-don-. Jens Elmegrd Rasmussen (ed.) in cooperation
with Benedicte Nielsen: In honorem Holger Pedersen. Kolloquium der Indoger-
manischen Gesellschaft vom 26. bis 28. Mrz 1993 in Kopenhagen. Wiesbaden:
Dr. Ludwig Reichert Verlag, pp. 331-347.
Olsen, Birgit Anette, 1999: The Noun in Biblical Armenian [Trends in Linguistics
Studies and Monographs 119] Berlin / New York: Mouton de Gruyter.
Olsen, Birgit Anette, 2000: Verb or Noun? On the Origin of the Third Person in
IE. Martin E. Huld & al.: Proceedings of the Twelfth Annual UCLA Indo-
European Conference [= JIES Monograph Series 40]. Washington D.C: Institute
for the Study of Man. Pp. 65-79.
Olsen, Birgit Anette, 2003: Another Account of the Latin Adjectives in -idus. HS
116 (2): 234-275.
Osthoff, Hermann, 1877: Etymologisches, Lautliches und Grammatisches. ZVS
23: 84-94 [90-94: Die Gotischen Adverbia auf -o und -ba].
Prellwitz, W., 1896: Studien zur indogermanischen etymologie und wortbildung.
Beitr. zur Kunde der indogerm. Sprachen 22: 76-114.
Preveden, Francis R., 1930: Church Slavonic korab and Greek . Lan-
guage 6 (4): 279-296.
Pronk, Tijmen, 2009: Sanskrit (v)a-bh-, Greek , : the spraying bull
of Indo-European?. HS 122: 170-181.
Puhvel, Jaan, 1984: Hittite Etymological Dictionary, I: Words beginning with A [=
Trends in Linguistics; Documentation, 1]. Berlin: Mouton de Gruyter.
Rasmussen, Jens Elmegrd, 1989: Studien zur Morphophonemik der indogermani-
schen Grundsprache. Innsbruck: IBS.
Rasmussen, Jens Elmegrd, 2009: Internal Reconstruction Applied to Indo-
European: Where do we stand?. Jens Elmegrd Rasmussen & Thomas
Olander (eds.): Internal Reconstruction in Indo-European. Copenhagen: Mu-
seum Tusculanum Press.
Sadovski, Velizar, 2012: Roots deep in heaven: Indo-Iranian ritual concepts in a
cross-cultural perspective. Paper given at Etymology and the European Lexicon,
the xth Fachtagung of the Indogermanische Gesellschaft, Copenhagen xx-xx
2012.
Schindler, Jochem, 1967: Idg. *dheudh- in Farbbezeichnungen. ZVS 81, 1/2, 68-
71.
Schindler, Jochem, 1978: Hittite alpa-. Die Sprache 24: 45.
Schmalstieg, William R., 1996: Review of Ambrazas 1993. Lituanus 42: 1.
Shevoroshkin, Vitaly, 1979: On the Hittite-Luwian Numerals. JIES 7 (3-4): 176-
198.
Sturtevant, Edgar Howard, 1911: Studies in Greek Noun-formation: Labial Termi-
nations. Classical Philology 6, 2, pp. 197-215.
Simon, Zsolt, 2008: Review of Adiego. Acta Antiqua Academiae Scientiarum Hun-
garicae 48: 457-463.
Stachowski, Kamil, 2008: Names of Cereals in the Turkic languages [= Studia Turko-
logica Cracoviensia 11]. Krakw: Ksigarna Akademicka.
Watkins, Calvert, 1969: A Latin-Hittite Etymology. Language 45, 2: 235-242.
Watkins, Calvert, 1995: How to kill a dragon. Oxford.
On the Precursors of Celtic and Germanic1
Abstract
A Celto-Germanicism may be defined as a lexeme shared by Celtic

and Germanic only and appearing to be older than the emergence of
Proto-Celtic and Proto-Germanic. Around 100 such items can be found,
some of which are IE archaisms, while others look like morphological or
semantic innovations. There is also a group of isolated lexemes which
appears to have been borrowed from the same third source. Four fifths
fall into two semantic spheres: 1) religion and healing, and 2) warfare
and equestrian terminology. Most remarkable is the occurrence of as
many as ten common words for wound, injury. This situation must
reflect close contacts between speakers of the Indo-European dialects
that later evolved into Proto-Celtic and Proto-Germanic respectively.
We may tentatively fix this cultural unity in time and space in Eastern
Central Europe around 2000 BCE, when the pre-Celtic ntice culture
bordered late varieties of the Corded Ware culture. Some shared
loanwords can be traced back to Proto-Fennic, suggesting a continuum
stretching further to the North. There are even indications that Proto-
Fennic may have been in direct contact with Pre-Proto-Celtic, not al-
ways with Pre-Proto-Germanic as the provider. Words of possible Fen-
nic origin include PCelt. *lub- wort ~ PGmc. *lubja- poisoning or
healing plant, PCelt. *sanesto- secret advice, PCelt. *magos plain,
open field, NIr. ln lunch, PCelt. *klamo- disease, and PGmc. *halj-
abode of the dead.
1 The present article was published as Hyllested 2010. Apart from this footnote
(including the reference just mentioned), the abstract, the comments on Zairs
(2012) review and the inclusion of a new item no. 5) *uitelo- (shifting each sub-
sequent footnote by one), the two articles are identical.
1 Loanword or heritage ?
While Germanic has quite a few Celtic loanwords (see, e.g., de Vries
1960, Birkhan 1970, Mees 1998, Rbekeil 2002, Schumacher 2007), the
share of older Germanic material in Celtic is comparatively small (Lane
1933: 264, and Schumacher 2007: 174-176). However, Celtic and Germa-
nic also share lexical material exclusive to these branches that can be
independently traced back to an identical reconstructed protoform.
Therefore, it is often hard to determine whether a given Celto-
Germanicism is inherited from PIE or borrowed from one branch to the
other at a later age. Karsten (1927: 126) wrote on PGmc. *arbja- vs.
PCelt. *orbios heir and PGmc. *aia- vs. PCelt. *oito- oath: kan likas
vara antingen urbeslktat med eller ln [might just as well be inherited
as borrowed]. Krahe (1954: 142) used the same lexeme as an example:
Die Hauptmasse des gemeinsamen nur keltisch-germanischen
Wortschatzes reicht ohne da vom rein linguistischen Standpunkt
Anhaltspunkte fr eine Entlehnung aus der einen in die andere Sprache
gegeben werden knnten bis vor die Periode der Lautverschiebung
zurck (Typus got. ais air. eth usw.). Olsen (1988: 13) writes on
PGmc. *gslo- hostage vs. PCelt. *geistlo- id.: It is not certain whether
the Gmc. examples are inherited or Celtic loanwords. Casaretto (2004:
318, fn. 1051) on PGmc. *r-n- secret vs. PCelt. *r-n- id.: Ob diese
Parallellitt Lehnbeziehungen oder ein gemeinsames Erbe reflektiert, ist
unsicher. Ringe (2006: 296) states: There are also quite a few words
shared only by Celtic and Germanic, which might or might not be
loanwords .... Matasovi (2009: 227) on Proto-Celtic *krumbo- round,
curved: Germ. krumm, OE crumb round point to PGerm. *krumba-,
which was borrowed either from Celtic, or from the same non-IE source
as the Celtic words. Polom (1983: 284) summed up the problem com-
plex, listing four possible origins of a Celto-Germanicism: a) the terms
represented either a common regional innovation in a marginal area of
the Indo-European territory or the localized survival of an archaic term
lost elsewhere throughout the Indo-European Linguistic area; b) the
terms have both been taken over from a same third source be it a Pre-
Indo-European (substrate) language or less well-documented Indo-
European language in their vicinity; c) the Celtic term was borrowed by
Germanic; d) the Germanic term was borrowed by Celtic. Lane (1933)
and Elston (1934) excluded borrowing, i.e. possibility c) and d), whene-
The Precursors of Celtic and Germanic 45
ver it could not be proved directly.2 In the following, I will use the term
Celto-Germanicism for items believed to be older than the emergence
of Proto-Celtic and Proto-Germanic, but shared by these two branches
only, i.e. Poloms categories a) and b).
2 Semantic spheres
Scholars already noted long ago that such Celto-Germanicims pertain to

certain semantic spheres. Thus, Lane (1933) suggested the following
headings:
a) Political and legal vocabulary

b) Warfare
c) Cultural and technical vocabulary, dwelling
d) Nature, earth, land, plant and animal life
e) Motion, locomotion, transportation
f) The body and bodily functions
g) Mental and emotional activity, vocal utterance
h) Sense perception
i) Family
j) Religion, superstition
k) Miscellaneous
and Krahe (1959: 139-141):

a) Religion und Geistiges Leben
b) Pferdezucht und Reiten
c) Siedlung, Hausbau
d) Landschaft, Natur
e) Metalle
f) Sonstiges
Elston (1934) and Campanile (1970) had still other divisions. On one
hand, it is interesting to observe how an overrepresentation of shared
vocabulary in certain semantic fields hints at the character of the rela-
2
Schmidt (1984, 1986, 1987, 1991) proposed a five-strate model: stratum 1, whose
Celtic origin is proved by their form; stratum 2, Celto-Germanic isoglosses with
the same semantic shift; stratum 3, Celto-Germanic isoglosses without the same
semantic shift; stratum 4, a group with special problems in the semantic field of
craftmanship; and stratum 5, name-doublets.
tionship in question. On the other hand, it seems as if Lanes and Kra-

hes lists cover most parts of the lexicon. If the lexical commonalities
could be combined with shared innovations and archaisms in the pho-
nological or grammatical system, it would be the obvious thing to hypo-
thesize that Celtic and Germanic formed a subgroup within the Indo-
European family. This seems not to be the case3. A closer scrutiny of the
material indeed reveals a much less blurred and much more unambi-
guous picture of the character of the earliest Celtic-Germanic relations.
3 The material
Lists of Celto-Germanicisms have been compiled by Lane (1933), Elston

(1934), Krahe (1956), de Vries (1960), Chemodanov (1962), Campanile
(1969, 1970), and Polom (1983). What follows is a revised and updated
synthesis of their material with the addition of new items. Celto-
Germanicisms include a) lexemes with a specialized meaning or use
shared by Celtic and Germanic, b) formations particular to Celtic and
Germanic (although formed from well-known PIE elements) and c)
lexemes whose very roots or root variants are unknown outside Celtic
and Germanic. The items have been grouped according to meaning (see
section 4 below), but within these semantic groupings they appear in a
more or less random order4.
3
The features mentioned by Schmidt (1991: 146-147) are either too weak or too
common to count as obviously shared innovations.
4
Space does not allow a word-to-word treatment of items that I have refused to
include as true Celto-Germanicisms. A few examples may serve as prototypes:
Craig Melchert (p.c.) kindly points out to me that *tegu- thick in OIr tiug, W
tew ~ OE ie, OHG dicki, ON ykkr thick vs. Lith. tnkus id. < *tenk-(-u-/to)-
is probably also attested in Hittite tagu- thick, swollen < *togu- (Neu 1995);
*luH-s louse > W llau lice (< collective *luu < *luH-eh) Corn. low, Bret. laou
id., OW leu-esicc louse-eaten ~ ON ls, OE, OHG ls louse vs. Toch. B luwo A
lu pl. lw animal probably also forms the basis of Lith. lil louse (where -l is
diminutive, cf. brol brother); PCelt. *korkio- oats (believed by Matasovi
2009 to be of a common substratum origin) corresponds to Shughni sip(i)yak a
kind of millet, sepyak grain of wheat according to Stalmaszyk and Witczak
(1991-1992); Rasmussen (1998) regards PGmc. *landa- (open) land as a bor-
rowing from Celtic proper. Despite the intriguing similarity, OIr. nasc ring;
clasp; bond, tie (~ nascim to bind) is most likely unrelated to OHG nusca, OS
nuscia clasp, buckle which is rather a Balto-Fennic borrowing, cf. Fi. nuska,
nurkka corner, nook, especially since another word for buckle, ON sylgja, is
already known to originate from Balto-Fennic (for the semantics, compare the
Category 1
A. Unique meaning
1) PCelt. *soito- sorcery > MW hut, Bret. hud magic ~ PGmc.

*saia- > ON seir magic; spell, charm, enchantment, sa
work a charm through seir. Probably identical to PIE *so)-
to- string, rope, derived from *seh)- to bind5, cf. Lith.
satas sign, soothsaying, soothsayer, talisman, but in Baltic
also still string, necklace etc., cf. Lith. sitas, Latv. sate id.
2) PCelt. *oitos oath > OIr eth, MW an-udon perjury ~
PGmc. *aiaz oath > OE , OHG eid, ON eir, Goth. ais
oath, OS mn- perjury; vs. Gk. faith, all from PIE
*h)-to-s walk(ing), derived from *he)- to go, cf. ON
ganga ei take the oath (see also Schumacher 2007: 176-
177;)6.
3) PIE *kor-)o-no- in the epithet of a god: OBrit. tribal name
Coriono-ttae people of the army-lord (a god, probably Lu-
gus) ~ PGmc. *harjanaz > ON Herjann lord of the army,
epithet of Odin; vs. Gk. ruler, commander <
* (Meid 1991: 48-49)
4) A personalized form meaning ghost of PIE *drougs: OIr.
air-drech phantom, MIr. aur-fraich ghost ~ ON draugr
ghost vs. Ved. drgha- deceiving, Av. draoga- lie (Mallory
& Adams)7.
5) PCelt. *uitelo- > MIr. fiothal dwarf, hag, goblin; anything
stunted ~ PGmc. *wila- > OE wdl impurity, OHG widil,
widillo homosexual, hermaphrodite, effeminate male; both
personified, vs. Lat. vitilg psoriasis, skin affliction, deri-
double meaning of Da. krog nook; hook and Eng. nook, a Scandinavian loan
with the original meaning clasp, hook, ON hnokki).
5
See Rasmussen 1989: 59-60.
6
Nicholas Zair (p.c.) points out to me that since no derivative of ganga in itself
means oath, ganga ei does not in itself suggest that *H)-to- is derived from
to go, and a connection with the root of Hitt. (i)- believe must also be con-
sidered (Puhvel 1991: 10). However, this does not affect its status as a Celto-
Germanicism.
7
OIr -drech may also be identical to drech vision < PCelt. *drik, derived from
*dr5- to see.
ved from *vitilis affliction (Kerkhof 2012)8. Derivatives from

*ueih- to twist, cf. also Lat. vitium defect, fault.
B. Unique morphology
6) PCelt. *rna- > OIr. rn secret, MW rhin spell, enchant-

ment ~ PGmc. *r-n- secret > OE rn, OS rna whisper;
secret; a rune, ON rn rune; secret9.
7) Gaul. (Chamalires) ande-don udiu-mi I praise a god ~
Goth. in-weitan gu to praise God < *ue)d- see where an-
de- semantically corresponds to in- (de Bernardo Stempel
2001).10
8) PCelt. *nem-eto- sacred grove, sanctuary > Gaul.
(Vaison), OIr. nemed sanctuary ~ PGmc. *nemia- > OS,
OLFr. nimidas sacred grove vs. Lat. nemus sacred grove,
Gk. wooded pasture, glade.
9) PCelt. *daun > MIr. dan poem ~ ON tafn sacrificial
animal < *dap-no- vs. Lat. daps sacrificial meal < PIE *dap-
(Watkins 1970)11.
10) PCelt. *uiro-k()-, gen. -kunos werewolf > Celtib. (Botor-
rita III) uiroku, OIr. Ferch, OW Gurci, OBret. Gurki (na-
me of a) werewolf (McCone 1987; McCone 2005: 401) ~ OE
wer(e)wulf, Dan. varulv, Fr. loup-garou id. where -garou <
Franconian *war-ulf-.
11) PCelt. *nerto- > OIr. nert strength, force, OW, MW nerth,
Bret. nerzh, Gaul. PN Nerto-maros ~ Gmc. *ner-u- > god-
dess Nerthus terra mater, ON Njrr, father of Freyr. Deri-
vatives from PIE *hner- man; strong (Meid 1991: 15).
8
The connection to Skt. vetla- demon (Lehmann 1907) is uncertain. If it is in
fact related, the Celto-Germanic character of the present item should perhaps
rather defined as a combination of morphology (found also in Latin) and se-
mantics (found also in Indic).
9
Rasmussens (1986: 1, 310) judgment that the exact correspondence between
Celtic and Germanic probably reflects an ancient borrowing in one direction or
the other is based on an isolated view of this lexeme. Contra Rasmussens con-
nection with some Greek material, see Vine (2002: 206ff.).
10
As she notes, Goth. inweitan takes the accusative while the Greek original takes
the dative, i.e. chances are that this is not a Greek calque.
11
The Celto-Germanic morphology also differs from Hitt. tappala- person re-
sponsible for court cooking, if it in fact belongs to the same root.
12) PCelt. *ab-anko- water creature > OIr., MIr. abac dwarflike
creature associated with water, W afanc beaver ~ PGmc.
*ab(n) monkey etc. (Schrijver 2004).
C. Isolated lexemes
13) PCelt. *uti- > OIr. fith prophet, Gaul. outeis (pl., Strabo)
and *utu- shamanic wisdom > fth prophesy, MW gwawd
ode ~ PGmc. *w- > ON r poetry; furious, Goth. wos
furious, ON inn, OE Wden, OHG Wuotan Odin (Meid
1991: 25-26; Watkins 1995: 118).12
14) PCelt. *rm > OIr rm, W rhif number ~ PGmc. *rma- >
OE rm number, ON rm computation, OHG rm account,
series, number.
15) PCelt. *sketlo- > OIr scl tale, W chwedl saying, fable ~
PGmc. *skala- > ON skld poet.13
16) PCelt. *gaisto- > OIr. ges speculation, cf. geth insanity;
wind ~ PGmc. *gaista- (supernatural) spirit > OHG geist,
OS gst, OS gst (Meid 1965).
17) PCelt. *klamo- > OIr clam, W claff grave ~ PGmc. *skalm
plague, (cows) disease; evil spirit, crook. Perhaps both
from PIE *s5olm-eh disease, evil spirit, but the Proto-Celtic
vocalism is not entirely clear; syllabic *-l- preceding *-m
would normally yield *-li-.14
18) PCelt. *skx-slo- demon, supernatural being > OIr scl
phantom, MW yscawl young hero, warrior ~ PGmc. skh-
sla- > Got. skohsl evil spirit, demon; both from *skk-slo-.
19) PCelt. *buko- > MIr boccnach goblin, W bwg ghost,
hobgoblin, bwgan bogey, ghost, bwgwn fright ~ Fris.
bkk, Swab. bockelman, NE bogle, bogey.
12
I assume Lat. vts prophet, seer to be a loan from Celtic.
13
It is no longer necessary to reconstruct a labiovelar for this word to account for
-w- in Welsh; cf. Schrijver 1992 and Jrgensen 2010. Zair (2012: 80) in his re-
view objects that a connection with Lat. nsece say still seems very plausible. I
do not quite understand this message, since I indeed do follow Schrijver and
Jrgensen in leaving out the labiovelar, paving the way for both forms to match
the Latin material. Even if Lat. nsece is related, I will maintain that the nominal
formation with *-lo- (and its meaning) constitutes a Celto-Germanicism, albeit
belonging to category II.
14
If Alb. helm poison belongs here, the Celto-Germanic connection is less clear.
Category II
A. Unique morphology (and meaning)
20) PCelt. *kol-ino- > Ir. cuilenn, W celyn holly ~ PGmc. *hul-
isa- > OE holen, OHG hulis, OFr. *huls > Fr. houx holly;
vs. OCS *klas ear of grain, Toch. B klese barley meal, Alb.
kall straw, chaff, Skt. kaamba- arrow, all from PIE *kel-
sharp, prickly. According to Pliny, the plant was a popular
house adornment among Celtic and Germanic peoples. In
Germany and Austria, holly is traditionally placed in stables
to protect horses from evil spirits15.
21) *!uond-neh Angelica > Ir. cuinneg wild angelica, Angeli-
ca silvestris ~ PGmc. *hwann > ON hvnn holy ghost, An-
gelica archangelica vs. *!uend-ro- with other meanings in
Lith. vndras reed, reed-mace; Lat. combrtum a kind of
rush. Angelica is an old medicinal herb and was used
against evil spirits (Birkhan 1999).
B. Isolated lexemes
22) PCelt. *lub or *lub > OIr luib wort, plant ~ OE lybb, OHG
luppi magic remedy; strong plant-juice; poison; magic, ON
lf healing plant, Goth. lubja-leisei magic; poisoning. Per-
haps also in ON epli ellilyfs old-age medicine > epli ellifu
eleven apples (in the Eddic lay Skrnisml; see Polom 1994:
142143 on the similar role of apples in Germanic and Celtic
mythology).
Category III
A. Unique meaning
15
The hollys connection to both horses and evil spirits may be due to the near-
homonymy of PGmc. *marha- m. horse, *marhj- mare and *mar- f. female
incubus, let alone their complete homonymy in Scandinaviancf. the ambigu-
ous names of the holly, Dan. maretorn, mareved, maretidsel, marelok, Nw. ma-
rekvist, Sw. markvist, marlock, martova, Icel. marhrsla, MLG marvlechte, mar-
lock, mahrzopf. Other Germanic names refer to the spirits only: Nw. huldrelime,
NHG Schrattelbaum, Hexenbesen, Eng. dial. witchs besom.
23) PIE *str-beh stiffness in the specialized meaning death;

plague: OIr. us-sarb death16 ~ OHG sterbo, OE steorfa pla-
gue vs. Gk. animal skin, leather, all from *ster-
(be) stiff.
24) *(H)rb-)o- m. leavings and *(H)rb-)o-m n. inheritance
> OIr orbae inheritance, Gaul. Orbio- id., OIr. orb(b)e, or-
pe heir; inheritance ~ Goth. arbja heir, arbi inheritance,
OHG erbi, OE ierfe id. vs. *(H)orb-o- orphaned > Lat. or-
bus deprived, Gk. orphaned, Arm. orb orphan,
Skt. rbha- small; weak; child (McCone 1999).
25) PIE *5re)H- in *5riH-no- > PCelt. *krno- > OIr crn en-
feebled by old age, decrepit; withered, OW crin ~ *!ro!H-
uo-m > PGmc. *hraiwa- n. > Goth. *hraiw in hraiwa-dubo
turtle dove, ON hr dead body, OE hrw id., OHG hro
dead body; grave; funeral; death vs. the unextended root
*5erh- to break (Casaretto 2004: 164).
26) PCelt. *uer-t- > OIr fertae (< *-i) burial mound, W
gwerthyr fort (< *-ero-) gweryd (< *-eto-) earth, soil; grave
~ OE weor yard, weard guarding, ON vara, vari mi-
lestone, vrr warden, watchman, defender; guardian spirit,
house spirit, soul of the dead.
C. Isolated lexemes
27) PCelt. *doueno- > OIr pl. dini, doni men, poetic sg. don,
don man17 ~ PGmc. *dewena- > Goth. diwans mortal, cf.
the verb OHG touwen, OS dian ON deyja die.
28) PCelt. *krito- > OIr. crith trembling; fever, crith-galar il-
lness with fever, W cryd fear ~ PGmc. *hra- > OE hr m.
fever, Nw. ri sudden illness; short period; hard weather
(Bjorvand & Lindeman 2000: 724).
16
OIr -rb- in us-sarb may be from *-ru- instead, cf. marb dead < *mr-uo-s.
17
Historically a suppletive paradigm with the sg. duine from PCelt. *gdonio-
earthling corresponding to Ved. kmya- earthly, mortal, cf. Gaul. TEVO-
XTONION (Vercelli) of god and men. Even if Latin fnus funeral procession
is related, the item still constitutes a Celto-Germanicism in terms of semantics
and word-formation (cf. Rasmussen 1988:923).
29) PCelt. *trusko- > OIr. trosc leprous; leper, W trwsgl rude;
clumsy, Bret. trousk polyps ~ Goth. ruts-fill, OE rst-fell
leprosy.
Category IV
A. Unique meaning
30) PCelt. *gaiso- spear > OIr gi, Gallo-Gk. , Gallo-Lat.

gaesum; OIr fo-gae, W gw-aew javelin ~ PGmc. *gaizo- >
OE gr, OHG gr, ON geirr dart, spear; from PIE *g)sos,
cf. Gk. shepherds crook, Skt. heas- weapon.
31) PCelt. *ri)o- > MW ryd, OCorn. rid free ~ PGmc. *frija- >
OE fro, OHG fr, Goth. freis free vs. Ved. priy- dear
(Schumacher 2007: 177).
32) PIE *kel- strike used in words for battle: PCelt. *kellko- >
MIr cellach contention, strife ~ OE hild war, battle, OHG
hiltia, ON hildr battle (corresponding to the ethnonym
Celtae).
33) PCelt. *trex-so- > OIr. tress battle and *trex-s-no- > OIr trn
brave, strong, comp. sup. tressa, tressam ~ PGmc. *rak-ja-
> OE rece force, oppression, OS wpan-threki ability with
arms, ON rekr strength, bravery.
34) PCelt. *kagro- enclosure, fort > W caer, Mbret. ker; and
*kagio- pen, enclosure > MW cae fence, OBret. caiou pl.
fortification, bulwark ~ PGmc. *hagan- enclosure, fence >
ON hagi pasture with a fence, OE haga, OHG hac hedge
and *hagj- > OE hecg hedge.
35) PCelt. *drungo- > Ir. drong troop, MW dronn multitude,

Gallo-Lat. (Vegetius) drungos groups of enemies ~ PGmc.
*drhta- > OE dryht companion, OHG truht troop ON
drtt company, following, Goth. driugan to serve as a sol-
dier vs. OCS drug friend, other dragas friend.
36) PCelt. *uik- fight > Ir. fichim fight, fecht military expedi-
tion, OW guith front ~ Goth. weihan, OE, OHG whan
fight, ON vega kill, fight vs. Lat. vinc conquer, Lith.

vekti make, work.18
37) PCelt. *kauno- harbor > MIr. can ~ PGmc. *haf-na-; ori-
ginally enclosure, shelter (for vessels).
38) PCelt. *baduo- ~ *boduo- > OIr Badb name of the slaughter
goddess,19 Early Ir. badb crow, demon, witch, W bod kite,
NIr. badhbh vulture; hoodie crow; fairy; scold, Gaulish dei-
ty Catu-bodua ~ PGmc. *bau- ~ *bawa- > OHG Batu-
slaughter; battle (in names), OE beadu battle, ON bs,
bsvar war vs. Lat. fodi, Hitt. padda- ~ padd- dig (the
ground), bury, Toch. A pt- plough, OCS bods to stab,
bed to stick, to dig.
39) PCelt. *ntu- > OIr. nth battle, distress, Gaul. PN Nitio-
broges, Nitio-genna ~ PGmc. *na- > Goth. nei envy, en-
mity, OE n, OHG nd battle-rage, hate, envy, ON n li-
bel.
40) PCelt. *magu- > Ogham magu slave, W meu-dwy hermit
(< servant of God), MBret. m(a)oues girl ~ Goth. magus
boy, ON mgr son; youth, OE magu child; son; man.
C. Isolated lexemes
41) PCelt. *nanti- > OIr. nit battle, combat, Nit god of battle,
husband of the war-goddess Nemain or Badb ~ PGmc.
*nanjana- > OE nan, OHG gi-nenden, ON nenna, Goth.
ana-nanjan to dare.
42) PCelt. *poiko- > OIr ech enemy ~ PGmc. *faiha- ~ *faiga-
> OE fh, fg guilty; outlawed; hostile, NE foe, OHG fhida
hate, enmity, Goth. fih deceit.
18
Brent Vine (p.c.) points out to me that while the nasal present in Lat. vinc ap-
pears (predictably) beside an old root aorist in perf. vc, in theory (despite LIV
670-671) the Celtic and Germanic presents could also be derived from the old
root aorist (e.g.: root aorist subjunctive thematic present is well-attested). In
that case, the Germanic and Celtic material might be closely related, morpho-
logically, to the old aorist (as in Latin), and since the Latin semantics is quite
similar to the one shared by Celtic and Germanic, only the development into a
thematic present would then point to a Celto-Germanicism.
19
Remarkably, Badb is the sister of Macha, married to Nemed, and of Mor-rgain;
of these four names, the first three are all Celto-Germanicisms, while cognates
of Mor- are also found in Slavic.
43) PCelt. *slak- strike > MIr slactha struck (ptc.), slacc
sword, Gael. slachdaim strikes with a hammer ~ PGmc.
*slahana- > Goth., OHG slahan, ON sl, OE slan slay.
44) W llost spear, Bret. lost, Ir. loss end; tail ~ ON ljstr fish-
spear, Dan. lyster eel-spear, ljsta strike.
45) PCelt. *mg- conceal > OIr for-migthe, for-michthai
smothered, concealed ~ PGmc. *mk- > OHG mhhen lie
in ambush for, NHG Meuchler assassin, ME micher thief,
Eng. dial. mitch hide (oneself).
46) OIr bgaid fight, boast, bg battle, W beio blame, Gaul.
Bagaudae, probably the fighters, name of Gallic peasants
who rebelled under Diocletian ~ PGmc. *bg- > OHG bgn
quarrel, fight, ON bgjast quarrel, strive.20
47) PCelt. *gwelti- madman, lunatic > MIr. geilt panic-stricken
fugitive from battle, W gwyllt wild, savage, mad ~ Goth.
wileis, OE wilde, OHG wildi, ON vildr wild.
48) PCelt. *ueidu- wild > OIr. fad wild animals, fian troop of
young warriors, MW gwydd wild, gwyddel a Gael, Irish-
man ~ PGmc. *wajaz > OE w hunt.
49) PCelt. *boudi- victory > OIr. baid victory, W buddig vic-
torious ~ ON bta exchange, divide, MLG bte booty21, all
from *budi-.
50) PCelt. *leid-o- succeed > MW llwyddaw ~ PGmc. *fltana-
> OE fltan, OHG flzan attempt, try hard.
51) PCelt. *geistlo- hostage > Ir. gall, W gwystl hostage, Br.
gouestl vow; promise, Gaul. PN Con-geistlus ~ PGmc.
*gslo- > OHG gsal, NHG Geisel, OE gsel, ON gsl hostage.
52) PCelt. *dno- fortification, rampart > Ir. dn, W din, Gaul.
-dnum in place-names ~ ON and OE tn hedged or fenced
lot, enclosure; OHG zn enclosure, hedge.22
53) Ir. clab *shield (of wicker-work) > basket; wicker frame of
a boat; chest ~ ON hlf shield, protection, OHG lpen, lp-
pen protect, Goth. hleibjan take the part of.
20
Even if Latv. buzties be annoyed belongs here (LIV 68 *beh-), Celtic and
Germanic still share a common semantics.
21
NE booty is a borrowing from Scandinavian.
22
Even if these words are derived from a PIE root *deuh- be finished, come full
circle (Watkins 1991:453), the derivative and its meaning are specific to Celtic
and Germanic.
54) NIr tailm, Bret. talm sling, W telm snare, trap ~ ON jlmi
sort of snare. The ON consonantism seems to indicate
common heritage.
Category 5
A. Unique meaning
55) PCelt. *brus-na- > OIr. bronnaim injure, damage, *brus-o-

> W briw, Corn. brew wound ~ PGmc. *brs- > OE brsan
bruise, OHG brsma crumb vs. Lat. frustum fragment.
56) PCelt. *kaiko- having an eye defect > OIr. cech, OCorn.
cuic one-eyed, W coeg-ddall half-blind ~ Goth. hihs one-
eyed; vs. Lat. caecus blind. The Celtic god Lug closes one
eye in his magic ritual, while in Germanic mythology, Odin
is one-eyed (Polom 1994: 145).
57) PCelt. *knid- > OIr cned a wound ~ PGmc. *hntana- >
ON hnta wound to death, OE, OS hntan thrust, stab; vs.
Gk. to scratch. servant of God),
58) PCelt. *aglo- wound, affliction > OIr il insult, MIr *lad
wound, MW aeled pain; grief ~ PGmc. *agla- > OE egle
disagreeable, loathsome, Goth. agls shameful, aglia, aglo
affliction vs. Av. a bad, evil, Skt. agh- bad, aghr- evil,
distress, aghal- terrible, all from PIE *ag- or possibly
*heg-.
59) PCelt. *gen-i- wound > OIr guin wound, injury ~ PGmc.
*banj- > Goth. banja strike, wound, ON ben, OE ben(n)
id., OS beni-wunda wound vs. PGmc. *ban-an- murder in
OE bana, Da. bane-sr deadly wound < PIE *gen- to kill.
wound to death.
60) PCelt. *koldo- > OIr coll destruction, W ar-choll wound ~
*PGmc. *halta- > Goth. halts, OE healt lame.
61) PCelt. *kre(n)x-tu- > OIr crcht wound, W creithen scar,
MBret. creizenn id. ~ PGmc. *skranh-a- > ON skr scroll.
C. Isolated lexemes
62) W gwanu pierce, thrust, stab, ym-wan joust, tilt, ymwanwr

combatant ~ PGmc. *wunda- > Goth. wunds, OE wund,
OHG wunda, ON und wound.23
63) PCelt. *snad-o- > MIr snaidid to cut; to scratch, W neddyf
axe ~ OHG snatta wound; scar; bruise, ON snata spear.
64) PCelt. *saitro- > OIr saethar work, labour and *saitu- > OIr
saeth trouble, MW hoed pain ~ PGmc. *sai-ra- > OE sri
sorry, OHG ser painfully; in a difficult way, ON srr
painful, sr wound24.
Category 6
A. Unique meaning
65) PIE *5ad- in derivatives with the meaning hatred: PCelt.

*kad-s-i-, *kdo- > MIr caiss, W cawdd, Bret. cas hatred ~
PGmc. *hatiz- > Goth. hatis, OE hete, OHG haz hatred, ON
hatr hatred; persecution vs. Av. sdra-, Gk. sorrow;
pain; misery, Osc. gen.sg. cadeis hostility (cf. also Birkhan
1967, Rbekeil 2001).25
66) A secondary thematic derivative *hup-l-o- evil > OIr fel

evil ~ PGmc. *ubila- evil > Goth. ubils, OE yfel, OHG ubil
vs. Hitt. uwapzi ill-treats, dispoils, Toch. A umpar bad, all
from PIE *huep- treat badly (cf. Cohen & Hyllested
2007:16).
67) PCelt. *kloino- > OIr. cloen crooked; unfair; evil ~ PGmc.
*hlaina- hill > Goth. hlain hill, Nw. dial. hlein steep slope,
both with *-no- from PIE *5le)(H)- to lean, cf. Ved. ryati,
Lith. liti id., lains slanting.
23
Zair (2012: 80) in his review of Hyllested 2010
24
Even if Lat. saevus wild, ferocious and Hittite i- be sullen, angry (see on the
latter Kloekhorst 2008:6923) are related, the Celtic and Germanic items form a
semantic entity.
25
Zair in his review (2012: 80) asks if the Celtic/Germanic meaning hatred is
unique enough to be seen as a shared feature compared to hostility in the Os-
can form. The two meanings differ in an important way, namely that hatred re-
fers to a feeling while hostility refers to a behavior.
C. Isolated lexemes
68) PCelt. *loktu > OIr. locht fault, blame; mistake ~ PGmc.
*lahana- > OE lan, OHG lahan, ON l to blame.
Category 7
A. Unique meaning
69) PCelt. *reid-o- ride; riding; chariot > Gaul. rda travelling-
carriage with four wheels, OIr. radaim ride (in vehicle), Ir.
d-riad team of two horses, W rhwyddau facilitate, speed ~
ON ra, OE rdan, OHG rtan to ride; ON rei riding; hor-
se-riding band; wagon vs. Latv. raidt send quickly; hunt.
70) Compounds with *hekuo- horse and *re)d-: Gaul. PN Epo-

rdo-rx ~ OE eo-red, OS eo-rid-folc cavalry, ON PN J-
reir.
71) i-stem adjectives meaning easy, ready derived from *re)d-
via driving or ready to go: PCelt. *reidi- > OIr. rid simple,
easy, flat, W rhuidd, OBret. ruet easy, quick ~ OHG bi-reiti
ready, Eng. ready.
72) PCelt. *axsil > W echel, MBret. ahel axis ~ PGmc. *ahsulaz
> ON xull axis vs. formations without *-loin Lat. axis,
Lith. as id..
73) PCelt. *uegno- > OIr fn, W gwain, Gaul. co-vinnus wagon ~
PGmc. *wagna- > ON vagn, OHG wagan wagon vs. other
formations in Skt. vhana-, Lat. vehiculum id..
74) PIE *sent- to travel in nominal formations meaning road;
retinue; PCelt. *sentu- path > OIr. st, MW hynt path,
epynt horseroad ~ PGmc. *sina- and *gasinja- retinue >
OHG Gisindi war retinue.
C. Isolated lexemes
75) PCelt. *marko- horse > MIr. marc, W march, Bret. march,
Gaul. (acc.sg.) horse, Marco- in place-names ~
OHG mar(a)h, OE mearh, ON marr id..
76) PCelt. *drux-to- > MIr. drochta tub, vessel ~ PGmc. *trugaz
> OE, ON trog, OHG troc trough.
77) PCelt. *kanx-s-ik- > W caseg, Bret. kazeg mare, not formal-
ly identical to PGmc. *hangista- ~ *hanhista- horse, stallion
etc. (Jrgensen 2006), but their similarity can hardly be
coincidental in the light of other equestrian commonalities;
cf. also that PCelt. *keng-o- to tread, step, walk is irregular
in the first place.
78) PCelt. *mongo- mane > MIr mong, W mwng id. ~ ON
makki upper part of a horses neck, Dan. manke mane; cf.
also ON mn, OE manu mane.
79) PCelt. *doklo- > OIr dal strand, lock (of hair) ~ PGmc.
*tagla- > ON tagl, Dan. tavl hair of a horses tail, OE tgl
tail, Goth. tagl a hair.
Category 8
A. Unique meaning
80) PCelt. *rd- > OIr rdim to say, to speak, MW ad-raud to

tell ~ PGmc. *riana- > Goth. rodjan, ON ra to speak.
B, Unique morphology
81) PCelt. *bana-tlo- > W banadl, MBr. malazn broom ~ OE

bnian polish, OS bnn scrub, polish, both from < *beh-
n- ~ *bn- vs. Gk. shine, Arm. banam open, reveal
(Olsen 1988: 26).
82) PCelt. *gablo-, *gabl- > OIr gabul, NIr gabhal, W gafl fork
~ PGmc. *gabal- > OHG gabala, OE geafol id..
83) PCelt. *lro- floor > OIr lr ground, surface; middle, W
llawr floor, Bret. leur id. ~ PGmc. *flruz > OIr flrr floor
of a cow stall, OE flr floor vs. Lat. plnus.
84) PCelt. *stl- > MIr. sithlad sieving, W hidl, MBret. sizl sie-
ve ~ PGmc. *sla- > ON sld sieve, Fi. (< Gmc.) siekla,
seula id. all from *seh)-tlo-, *sih-tlo- vs. seh)-to-, *sih-to in
Lith. sxtas, CS sito.
85) PCelt. *atim-, *atam- > Gael. aitheamh, W edafedd

yarn; thread ~ PGmc. *famaz > OHG fadum, OE fm,
ON famr spread arms, embrace; thread (Hamp 2008).26
86) PCelt. *iexti- > Ir icht tribe, W ieith language; nation,
MBret. yez language ~ PGmc. *jehti- > OHG jiht utterance
(cf. jehan to speak) vs. Lith. jukas, Lat. iocus jest.
87) PCelt. *rextus > OIr recht law, justice, MW kyf-reith id. ~
PGmc. *rehtuz > ON rttr justice, law < *hre-tu- with a
shared, unpredictable meaning (Schumacher 2007:177).
88) PCelt. *roino- > OIr roen road; mountain range; Bret. run
hill ~ PGmc. *raina- > ON -rein strip of land (in com-
pounds), OHG rein ridge of earth as boundary mark.
C, Isolated lexemes
89) *suek- > W chweg, Bret. chouek sweet, pleasant (of taste),
W chwaeth taste ~ OE swecc, swcc taste, (pleasant) smell,
OHG swehhan to smell (bad).
90) *suem- > OIr to-seinn hunts; follows27 ~ OHG, OE swim-
man, ON svim(m)a to swim, Goth. swum(f)sl lake <
*swum-sla- (Bjorvand and Lindeman 2000:8935, but they
reject the connection; Casaretto 2004:408).
91) *sueng- to bend in PCelt. *swengo- slender > MIr. seng,
Gaul. PN Singi-dnum ~ OE swancor, MHG, MLG swanc
slender, Dan. svang arch of foot vs. *sueg- and suenk- in
other formations and languages (IEW 1047).
26
Hamp includes Alb. p, pl. penj thread, but Celtic and Germanic still agree both
on o-grade and semantics.
27
If Zair in his review (2012: 80), following LIV 532-3, is right that to-seinn is re-
lated to Hitt. sanazi sought < *senh-, this item is of course not a Celto-
Germanicism. Note also Kroonen 2013 forthc.: The verb has no good extra-
Gm. Etymology. The connection with OIr. seinnid is extremely doubtful, both
on the formal and semantic side. In fact, eliminating this item would only
strengthen the hypothesis presented here since it reduces the number of items
belonging neither to category 8. However, the combination of the initial conso-
nant cluster *su-, shared with items 89 and 91, and the derivational suffix *-slo-
of PGmc. *swumsla-, shared with items 19 and 51, perhaps point to an origin in
the Celto-Germanic stratum after all. Indeed, the PGmc. ablauting forms
*swammjan-, with a causative, not denominative, meaning make swim,
*swamn swim and *sunda- sound (beside *swumsla-), indicates that it is
fairly old.
92) PCelt. *grando-, *grendo- beard > MIr. grend; MW grann

beard; chin; cheek, Provencal gren moustache (< Gaul.) ~
PGmc. *gran- f. > OE granu moustache, OHG grana hair
of the beard, ON grn, Goth. grano hair of the beard; spruce
(needle).
93) PCelt. *lind-o/u- drinkable water (cf. Matasovi 2009:240) >
OIr lind liquid, W llyn (m/f) drink, (m) lake ~ Icel. lind
spring, fountain, MHG lnde wave.
94) PCelt. *gluo- > MW glo charcoal and PCelt. *goulo- (< dis-
similated from *glou-lo-?) > MIr. gal28 ~ PGmc. *kula-,
*kulan- charcoal > ON kol (pl.), OE col, OHG kolo.
95) PCelt. *druxtu- (< *drup-tu-) > OIr drcht dew, a drop ~
PGmc. *drupa- > ON dropi, OE dropa, OHG tropfo drop.
96) PCelt. *kaito- wood > OW coit OCorn. cuit, MBret. coat fo-
rest, wood, Gaul. PN , Cto-briga, Eto-ctum ~
PGmc. *haija- > Goth. haii field, heath, NHG heis-ter
small tree or bush, ON heir heath, moor.
97) *gan(d)-no- in MIr gann vessel, jug, pitcher ~ PGmc.
*kann f. > ON kanna, OE canne, OHG channa can, jug.
4 Revision of semantic areas and their implications
Our revised list may be said to fall into the following categories:
(1)-(19) cosmology, spirits, supernatural creatures

(20)-(22) medicinal herbs or plants connected to popular beliefs
(23)-(29) sickness and death
(30)-(54) battle and warfare, fortifications, weaponry
(55)-(64) words for wound, injury, defect
(65)-(68) hostility
(69)-(79) equestrian terminology
(80)-(97) words belonging to other parts of the vocabulary
Or, in a boiled-down version (exlcuding (80)-(97)):
(a) religion and healing ((1)-(29))

(b) warfare and equestrian terminology ((30)-(79))
28
Even if PCelt. *goulo- and the Germanic forms are related to Skt. jvlati burns,
Toch. B oliye hearth, Celtic and Germanic still share a specialized meaning.
Quite a few of the lexemes in question can be placed in either category.

Remarkably, as many as nine words for wound turn out to be Celto-
Germanicims in one way or another.
Only 1/5 of the lexemes fall outside the two main categories,
and, with a couple of exceptions, even these are typical culture-words.
Such a distribution militates against the possibility of a Celto-Germanic
genetic subgroup (pace Mansion 1912) and, obviously, the existence of
Italo-Celtic need not be refuted on this basis; cf. also that the list of NW
IE innovations compiled by Oettinger (2003) comtains not a single Cel-
to-Germanicism. Instead, the situation presented here seems to reflect
contacts between speakers of the IE dialects that later evolved into Pro-
to-Celtic and Proto-Germanic. Religion and warfare seem to have been
of particular concern.
Linking reconstructed prehistoric languages to archaeological fin-
dings is always risky business, but we may tentatively fix this cultural
unity in time and space in Eastern Central Europe around 2000 BCE,
when the pre-Celtic ntice culture in the present-day Czech Republic
bordered late, possibly pre-Germanic, varieties of the Corded Ware cul-
ture29. This scenario is at least partly compatible with conclusions rea-
ched by Kristiansen & Larsson 2005 and Kristiansen 2009: They envisa-
ge contacts between Pre-Germanic peoples and Pre-Celts immigrating
from the South, spreading out over W Europe 2500-2000 B.C., not least
by means of warfare and horses, until more hierarchical societies arise
in the second millennium B.C.
5 A Fennic connection?
Most of the items in question look old and probably represent regional
IE innovations, while others may have been taken over from the same
third source. Interestingly, some of them seem to be shared with Balto-
Fennic languages, suggesting a larger cultural continuum stretching
further to the North. Particularly intriguing are Fi. hepo, hevonen, Est.
hobune horse, Fi. ratsu riding-horse and kavio hoof (dial. kapja) sin-
ce they all look Indo-European, but at the same time do not show the
regular sound substitutions displayed by any attested Indo-European
branch. Fi. luppo lichen is inherited from Proto-Uralic, so if it is con-
29
ntice bodies are typically buried with jugs meaning that *gan(d)-, too,
could justifiably be categorized as belonging to the religious vocabulary.
nected to item no. 22, it must have been borrowed from Fennic into
Celtic and Germanic.
If Balto-Fennic belongs to this cultural continuum, the question ari-
ses whether lexical exchange has taken place directly between Late Pro-
to-Fennic and Pre-Proto-Celtic, or whether Pre-Proto-Germanic was
always the provider: PCelt. *sanesto- secret advice (Matasovi 2009:
322) is suspiciously reminiscent of Fi. sanasto list of words (synchroni-
cally analyzable as sana word + collective -sto), cf. the semantics of
PCelt. *rno- and PGmc. *rna- (item 6 above) which in itself must be
identical to Fi. runo song; poem. The vowel in runo is unexpectedly
short, i.e. it does not behave as loanwords from Proto-Germanic nor-
mally do and may have been borrowed at an earlier stage. Mod. Ir. ln,
pl. linte (> Eng. lunch) could represent Late Proto-Fennic *louna
southwest; noon; lunch (Fi. lounas) which is derived from Proto-Uralic
*luwe south. Note that this word is already known to have been borro-
wed into Baltic (Latv. launags lunch, Lith. lunagas dinner). Fi. maa
land and its Balto-Fennic cognates go back to Proto-Uralic *mae, re-
miniscent both in form and semantics of PCelt. *magos plain, open
field > OIr mag plain, W ma place, Gaul. PN (Arganto-)magus), cf.
Schrijver 2001:423. Fi. tuoni dead < Late Proto-Fennic *tne could for-
mally represent Proto-Celtic *doueno- (item 27). Fi. kalma grave;
disease, Death-goddess, guardian of the abode of the dead could belong
with PCelt. *klamo- grave (item 17)
For the same concept, PGmc. *halj- f. can be reconstructed (cf. e.g.
ON Hel death goddess). It is most often seen as reflecting PIE *5ol-)eh,
derived from *5el- to cover, conceal. However, if Fi. Koljo name of a
giant is a Germanic loan (IEW 553-554), the Finnish vocalism constitu-
tes a problemwhy is PGmc. *-a- substituted with -o-? Moreover, a
Proto-Finno-Ugric form *kolja can be reconstructed also on the basis of
Komi kul water spirit and Mansi (Pelym dial.) ku-njr master of the
netherworld, devil. This word is internally analyzable as a participial
form or agent noun derivative consisting of the Proto-Uralic verbal root
*kole- to die and the agent-marker -ja with root-final -e regularly being
dropped when a suffix is added30. Formally, nothing speaks against this
word being a borrowing in the reverse direction, from Proto-Fennic into
Pre-Proto-Germanic, i.e. at a stage before the Germanic sound shift and
the development of *o > *a.
30
For a slightly different analysis of the Uralic word, see Katz (2003:183).
6 Conclusions
An analysis of the NW Indo-European lexical material shared by Celtic

and Germanic only is suggestive of the following scenario: The precur-
sors of Celtic and Germanic evolved from different Indo-European dia-
lect groups. Shortly after their migrations into Europe they came to
form part of a cultural community, possibly influenced by indigenous
populations or migrators from elsewhere. This had a significant impact
on specific parts of the vocabulary, notably terms for religion and war-
fare. New derivatives were formed on the basis of Indo-European mate-
rial, while some of the old ones were preserved in this area only. Some
shared loanwords can be traced back to Late Proto-Fennic, spoken in
their Northern vicinity. There are even indications that Late Proto-
Fennic may have been in direct contact with Pre-Proto-Celtic, not al-
ways with Pre-Proto-Germanic as the provider. Celtic and Germanic
peoples continued to influence each other, linguistically and in other
respects, as they gradually developed the characteristics by which we
define them.
References
De Bernardo Stempel, Patrizia, 2001: Gotisch inweiti gu und gallisch ande-

don udiu-mi (Chamalires, Z. 1). Historische Sprachforsching 114: 164-170.
Birkhan, Helmut. 1967: Das gallische Namenelement *cassi- und die germanisch-
keltische Kontaktzone. Wolfgang Meid (ed.): Beitrge zur Indogermanistik
und Keltologie. Innsbruck: Institut fr Sprachwissenschaft der Universitt Inns-
bruck. Pp. 115-145.
Birkhan, Helmut, 1970: Germanen und kelten bis zum Ausgang der Rmerzeit. Vien-
na.
Birkhan, Helmut, 1999: Ein Strau nicht durchwegs bekmmlicher Kruter aus
dem keltischen und germanischen Altertum. Peter Anreiter & Erzsbet Je-
rem (eds.): Studia Celtica et Indogermanica. Festschrift fr Wolfgang Meid zum
70. Geburtstag. Budapest: Archaeolingua. Pp. 43-52.
Bjorvand, Harald, & Fredrik Otto Lindeman, 2000: Vre Arveord. Oslo: Novus.
Campanile, Enrico, 1970: Sulle isoglosse lessicali celto-germaniche. AION 9: 13-
39.
Casaretto, Antje, 2004: Nominale Wortbildung der gotischen Sprache. Heidelberg:
Winter.
Chemodanov, Nikolaj Sergeevich, 1962: Germano-keltskie leksieskie paralleli.
Sravnitelnaja grammatika germanskix jazykov 1: 73-105. Moscow.
Cohen, Paul S., & Adam Hyllested, 2007: Monophthong for Expected -diphthong
in Greek. Coulter George, Matthew McCullagh, Benedicte Nielsen, Antonia
Ruppel & Olga Tribulato (eds.): Greek and Latin from an Indo- European Per-
spective [= Cambridge Classical Journal, supplementary volume 32], Cambrid-
ge: Cambridge University Press. Pp. 9-18.
Elston, C.S, 1934: The Earliest Relations between Celts and Germans. London:
Methuen.
Hamp, Eric P, 2008: Germanic *famaz and Gravity in the North. Early and
Pre-historic Language Development in North-Western Europe [= NOWELE
54/55]. Pp. 349-352.
Hyllested, Adam, 2010: The Precursors of Celtic and Germanic. In Jamison,
Melchert & Vine (eds.) 2010: 107-128.
IEW = Julius Pokorny: Indogermanisches Etymologisches Wrterbuch I-II. Bern:
Francke, 1959.
Jamison, Stephanie W., H. Craig Melchert & Brent Vine (eds.): Proceedings of the
21st Annual UCLA Indo-European Conference. Bremen: Hempen.
Jrgensen, Anders Richardt, 2006: Etymologies to Go: Some Further Reflexes of
Celtic *keng-. Keltische Forschungen 1: 59-72.
Jrgensen, Anders Richardt, 2010: Palatalization of *sk in British Celtic. Bene-
dicte Nielsen, Thomas Olander, Birgit Anette Olsen & Jens Elmegrd Rasmus-
sen (eds.): The Sound of Indo-European. Copenhagen: Museum Tusculanum
Press. Pp. ??-??.
Karsten, Torsten Evert, 1927: Germanerna. Stockholm: Natur och Kultur.
Katz, Hartmut, 2003: Studien zu den lteren indogermanischen Lehnwrtern in den
uralischen Sprachen. Heidelberg: Winter.
Kerkhof, Peter Alexander, 2012: Suffix variation in the PGmc. l-suffixes and the ab-
laut of the PIE l-stems. MA dissertation from Leiden University.
Kloekhorst, Alwin, 2008: Etymological Dictionary of the Hittite Inherited Lexicon.
Leiden / Boston: Brill.
Krahe, Hans, 1954: Sprache und Vorzeit. Heidelberg: Quelle and Meyer.
Kristiansen, Kristian & Thomas Larsson, 2005: The Rise of Bronze Age Society. Tra-
vels, Transmissions, and Transformations. Cambridge: Cambridge University
Press.
Kristiansen, Kristian, 2009: Proto-Indo-European Languages and Institutions: an
Archaeological Approach. Marc Vander Linden & Karlene Jones-Bley (eds.):
Departure from the Homeland: Indo-Europeans and Archaeology. Washington
DC: Institute for the Study of Man. Pp. 111-140.
Kroonen, Guus, 2013 forthc.: Etymological Dictionary of Proto-Germanic. Leiden:
Brill.
Lane, George S, 1933: The Germano-Celtic Vocabulary. Language 9: 244-264.
Lehmann, W., 1907: Ahd. Wdillo = ir. fiothal. Zeitschrift fr Deutsche Wortfor-
schung 9: 313-314.
LIV2 = Helmut Rix (ed.): Lexikon der Indogermanischen Verben, 2nd ed., 2001.
Wiesbaden: Reichert.
Mallory, James P., & Douglas Q. Adams, 2006: The Oxford Introduction to Proto-
Indo-European and the Proto-Indo-European World. Oxford: Oxford University
Press.
Mansion, Joseph, 1912: Kelten en Germanen. Verslagen en Meededelingen der

Koninkrijke Vlaamsche Academie 1912: 1292-1308.
Matasovi, Ranko, 2009: Etymological Dictionary of Proto-Celtic [= Leiden Indo-
European Etymological Dictionary Series 9]. Leiden: Brill.
McCone, Kim, 1987: Hund, Wolf und Krieger bein den Indogermanen. Wolf-
gang Meid (ed.): Studien zum indogermanischen Wortschatz. Innsbruck: Institut
fr Sprachwissenschaft der Universitt Innsbruck. Pp. 101-154.
McCone, Kim, 1999: OIr erbaid entrusts, orb heir and orbae inheritance.
Peter Anreiter & Erzsbet Jerem: Studia Celtica et Indogermanica. Festschrift fr
Wolfgang Meid zum 70. Geburtstag. Budapest: Archaeolingua. Pp. 239-242.
McCone, Kim, 2005: Mgliche nicht-indogermanische Elemente in den keltischen
Sprachen. Gerhard Meiser & Olav Hackstein (eds.): Sprachkontakt und
Sprachwandel. Akten der XI. Fachtagung der Indogermanischen Gesellschaft, 17.-
23. September 2000, Halle an der Saale. Wiesbaden: Reichert. Pp. 395-435.
Mees, Bernard, 1998: Celtic Influence in the Vocabulary of Hierarchy during the
Common Germanic Period. Zeitschrift der Savigny-Stiftung der Rechtsge-
schichte. Germanistische Abteilung 115: 361-388.
Meid, Wolfgang, 1965: Zu einigen keltischen und germanischen Bildungen
mit st-suffix. Die Sprache 11: 122-130.
Meid, Wolfgang, 1991: Aspekte der germanischen und keltischen Religion im
Zeugnis der Sprache. Innsbruck: Institut fr Sprachwissenschaft der Univer-
sitt Innsbruck.
Neu, Erich, 1995: Hethitisch tagu- dick, angeschwollen. Historische
Sprachforschung 108: 1-5.
Oettinger, Norbert, 2003: Neuerungen in Lexikon und Wortbildung des Nord-
west-Indogermanischen. Alfred Bammesberger & Theo Vennemann: Langu-
ages in Prehistoric Europe. Heidelberg: Winter. Pp. 183-193.
Olsen, Birgit Anette, 1988: The Proto-Indo-European Noun Suffix *-tlom and Its
Variants. Copenhagen: Munksgaard.
Polom, Edgar C., 1983: Germano-Celtic Isoglosses (Revisited). Journal of Indo-
European Studies 11: 281-298.
Polom, Edgar C., 1994: A Few Notes about Religious Correspondences between
Celtic and Germanic. Roland Bielmeier & Reinhard Stempel (eds.): Indoger-
manica et Caucasica. Festschrift fr Karl Horst Schmidt zum 65. Geburtstag. Ber-
lin: de Gruyter. Pp. 138-148.
Puhvel, Jaan. 1991. Hittite Etymological Dictionary. Volume 3: Words beginning
with H. Berlin / New York: Mouton de Gruyter.
Rasmussen, Jens Elmegrd, 1987 [1999]. Miscellaneous Morphological Problems
in Indo-European Languages IV, 22.: Greek ask, pray with a Note on
the Indo-European Factitive Iteratives. APILKU 6: 145-156 [Reprinted in Jens
Elmegrd Rasmussen, 1999: Selected Papers in Indo-European Linguistics, I: 309-
311].
Rasmussen, Jens Elmegrd, 1989: Studien zur Morphophonemik der indogermani-
schen Grundsprache. Innsbruck: Institut fr Sprachwissenschaft der Universitt
Innsbruck.
Rasmussen, Jens Elmegrd, 1998: Wege der Kulturverbreitung im Lichte der

Wortformen. Wolfgang Meid (ed.): Sprache und Kultur der Indogermanen.
Akten der X. Fachtagung der Idg. Gesellschaft. Innsbruck: Institut fr Sprachwis-
senschaft der Universitt Innsbruck. Pp. 547-556.
Ringe, Don, 2006: From Proto-Indo-European to Proto-Germanic [= A Linguistic
History of English, vol. 1]. Oxford: Oxford University Press.
Rbekeil, Ludwig, 2001: Einige Bemerkungen zur Wortfamilie um germ. *hatis-.
Bla Brogynyi (ed.): Germanisches Altertum und christliches Mittelalter.
Festschrift fr Heinz Klingenberg zum 65. Geburtstag. Hamburg: Kova. Pp. 239-
294.
Rbekeil, Ludwig, 2002: Diachrone Studien zur Kontaktzone zwischen Kelten und
Germanen [= AW, Phil.-hist. Kl., Sitzungsberichte, 699]. Wien: Verlag der s-
terreichischen Akademie der Wissenschaften.
Schmidt, Karl Horst, 1984: Keltisch und Germanisch. Jrgen Untermann & Bla
Brogynyi (eds.): Das Germanische und die Rekonstruktion der indogermani-
schen Grundsprache: Akten des Freiberger Kolloquiums der Indogermanischen
Gesellschaft, Freiburg 26-27 Februar 1981 [= Current Studies in Linguistic Theory
22]. Amsterdam: Benjamins. Pp. 113-153.
Schmidt, Karl, Horst, 1986: Keltisch-germanische Isoglossen und ihre sprachge-
schichtlichen Implikationen. Heinrich Beck (ed.): Germanenprobleme in heu-
tiger Sicht [= Reallexikon der germanischen Altertumskunde, Ergnzungsband
1]. Berlin: de Gruyter. Pp. 231-247.
Schmidt, Karl Horst, 1987: Handwerk und Handwerker im Keltischen und Ger-
manischen: Beitrge zu einem historischen Vergleich. Wolfgang Meid (ed.):
Studien zum Indogermanischen Wortschatz. Innsbruck: Institut fr Sprachwis-
Schrijver, Peter, 2001: Lost Languages in Northern Europe. Christian Carpelan,
Asko Parpola & Petteri Koskikallio (eds.): Early Contacts between Uralic and
Indo-European: Linguistic and Archaeological Considerations [= Mmoires de la
Socit Finno-Ougrienne 242]. Helsinki: Socit Finno-Ougrienne. Pp. 417-425.
Schrijver, Peter, 2004: Apes, Dwarfs, Rivers, and Indo-European Internal Deriva-
tion. Adam Hyllested, Anders Richardt Jrgensen, Jenny Helena Larsson &
Thomas Olander (eds.): Per Aspera ad Ateriscos. Studia Indogermanica in Hono-
rem Jens Elmegrd Rasmussen anno MMIV. Innsbruck: Institut fr Sprachwis-
Schumacher, Stefan. 2007: Die deutschen und die Nachbarstmme: Lexikalische
und strukturelle Sprachkontaktphnomene entlang der keltisch-germanischen
bergangszone. Keltische Forschungen 2: 167-208.
Stalmaszyk, Piotr, & Krzysztof Tomasz Witczak, 1991-1992: The Celtic word for
oats, Avena sativa, and its Indo-European equivalents. Lingua Posnaniensis
34: 83-87.
Stifter, David, 2009: The Proto-Germanic Shift * > * and Early Germanic Lin-
guistic Contacts. Historische Sprachforschung 122: 268-283.
Vine, Brent, 2002: Att. , Ion. ask. Glotta 78: 203-221.
de Vries, Jan, 1960: Kelten und Germanen. Bern: Francke.
Watkins, Calvert, 1976: The Etymology of Irish dan. Celtica 11: 270-277.
Watkins, Calvert, 1991: A Celtic-Latin-Hittite Etymology. Tsvi Abusch, John

Huehnergard & Piotr Steinkeller (eds.): Lingering over Words: Studies in An-
cient Near Eastern Literature in Honor of William Moran. Pp. 451-453.
Watkins, Calvert, 1995: How to Kill a Dragon. Oxford: Oxford University Press-
Zair, Nicholas, 2012: Review of Jamison, Melchert & Vine (eds.) 2010. Kratylos 57:
78-83.
Again on Pigs in Ancient Europe:
The Fennic Connection1
Abstract
Proto-Celtic *mokku- pig, *sukko- sow and *turko- wild boar are
borrowed from Proto-Fennic where they are analyzable as inherited
formations. Other Northern European terms can be shown to have an
Indo-European origin: Welsh cranan wild sow, OIr crin sow belong
to an earlier layer comprising both Germanic (OLFr. chranni-chaltia
pigs den) and Baltic (Lith. ernas wild boar), while Fi. karjas wild
boar is borrowed from an otherwise unattested PGmc. *garjaz corre-
sponding to Gk. , Alb. derr, from PIE *hor-io-s. Latv. cka pig
is not related to Lith. kial pig as usually assumed, but borrowed from
PFc. *tsuka pig (> Fi. sika, Karel. ugu N Saami sokki id.). NW PIE
*por5o- pig(let) is identified as an Altaic newcomer to the NW IE area
on the basis of its widespread irregular variation in both IE and Fenno-
Ugric, and the similarity with European words for badger, an animal
typologically often compared to pigs. The lessons to be drawn are signif-
icant both culturally and linguistically: The great importance that boars
played in Celtic and Germanic mythology must have been preceded by
a centre of cultural gravity further to the North.
1 Hyonyms in Celtic substratum material or Fennic loans?
It is well known that the wild boar played a significant role in ancient
Celtic and Germanic (as well as in ancient Greek) mythology. Hamp
(1987) has argued that the importance of boars and pigs went back to a
1 The greater part of this article will be published in Birgit Anette Olsen & al.
(eds.): Etymology and the European Lexicon, proceedings from Fachtagung der
Indogermanischen Gesellschaft in Copenhagen (September 2012).
or North or Central European pre-Indo-European substratum2. A

number of non-Indo-European pig terms, which we may conveniently
name hyonyms, are of obscure origin, and many look non-Indo-
European. For Celtic alone, Hamp lists the following as unexplained:
1 *mokku- swine (OIr mucc, W moch, Gaul. PN Moccus, a swine
divinity)
2 *sukko- sow (W hwch, OIr socc-)
3 *turko- (wild) boar (W twrch, OIr torc)
4 *banuo- young pig (W banw, OIr banb)
5 W cranan wild sow (White book of Rhydderch), OIr crin f.
sow
6 OIr mat, mata f. pig
7 OIr cribais, cribu(i)s pig
8 OIr fithend m. boar (?)
1.1 PC ELT . * MOKKU - SWINE
Hamp (1987: 187) began his overview with Proto-Celtic *mokku- swine,
which is reconstructed on the basis of OIr mucc f. (originally a u-stem),
W moch (collective; singulative mochyn), Breton moch (collective; sin-
gulative penmoch), Corn. mgh, late mw id., and Gaul. Moccus, the
name of a pig divinity. Hamp defined it as perhaps the most prominent
term notably lacking in IE cognates, a striking fact for the most per-
vasive generic lexeme for the pig. He did not mention MLG and MDu.
mocke f. sow, but as rightly stated by Kroonen (2013, pace Matasovi
2009: 274-275) these Germanic forms are most likely to be loanwords
from Gaulish and not directly from a third, unrelated source, since they
22
Hamp (1987: 187) supposes that the cultural importance of the pig in Ancient
Greece goes back to the same substratum, seeing that Pre-hellenic was noot a
satem language in type [and] is to be classed among IE dialects with the
North European group We must look, therefore, for a pre-Greek movement
of Indo-Europeans into the Aegean from the North, from as least as far North
as Central Europe. It is clear then that the IE Prehellenic speakers could have
brought with them to Greece the North European cultural values and institu-
tions relating to the pig, these later to be incorporated into the Eleusinian mys-
teries. He further ascribes the irregular variant (next to the regular ) to
this Prehellenic IE language. Fascinating as this scenario may sound, the prehis-
tory and shaping of Greek culture goes beyond the scope of this article and will
not be treated here.
Again on pigs in ancient Europe: The Fennic Connection 71
are confined to the Southwestern (Franconian) part of the Germanic

area. Hamp credited the substratum language even for the source of the
u-stem formation, according to him typical of substratum words in
Celtic, and found in one more pig-term, Welsh hob (see further under
*sukko- below).
Remarkably in the light of Hamps scenario, very similar terms spe-
cifically for sow occur elsewhere in Northern Europe, namely among
the Balto-Fennic languages. On the basis of Finnish emakko, Karelian
emakko, emkk, Olonets em, Lude emu, Estonian emak, Votic
emakko, emikko we are able to reconstruct Balto-Fennic *emakko,
*emkk3 while Fi. emokki reflects BF *emokke. Following Hamps line
of reasoning, it would be natural to conclude that the North European
substratum terms for pigs then stretched all the way to the Balto-Fennic
area. Crucially, however, these words are internally analyzable within
Balto-Fennic itself, even partly synchronically in the individual lan-
guages, as perfectly normal derivatives with the denominal suffix -kko
from the noun em womb; mother (also of animal). The Finnish vari-
ant with -okki (which, if regular, would go back to Balto-Fennic *-okke)
has been formed to emo mother of animals, dam, in itself formed with
a frequent denominal and deverbal suffix -o and common in Finnish
compounds (e.g. emo-lehm calver, mother cow, emo-yhti parent
company).
Since derivatives with *-kko to the stem *em must go back to the
Balto-Fennic protolanguage, I will propose that Proto-Celtic *mokku- is
simply borrowed from Early Balto-Fennic.
Prehistoric Celtic is normally attached to the Urnfield Culture of
Central Europe (1300-750 BC), and its descendants, the Hallstatt Cul-
ture (800-500 BC, continuing into the historic La Tne Culture). Can-
didates for Pre-Proto-Celtic cultures are the ntice culture (2300-
1600 BC) and the Tumulus culture (1600-1200 BC). Kristiansen and
Larsson (2005) and Kristiansen (2009) place the emergence of Pre-
Celtic culture in Western and Central Europe 2500-2000 BC. As I
3
BF *emakko has undergone a common, but not entirely regular, morphopho-
nemic development of > a / eC_C(C)o (cf. e.g. Fi. kes summer ~ kesakko
freckle, el- to live ~ elanto livelihood) while the variant *emkk displays
the expected vowel harmony. -ikko (and -ikk) as in Votic emikko mostly occurs
after stems in -e- or -j-, but sometimes even after --stems, cf. Fi. silm eye ~
silmikko bud, hein grass ~ heinikko meadow. These processes have been de-
scribed by Campbell 1980 (257-258 with references) from a purely Finnish per-
spective, but they must be regarded as common Balto-Fennic phenomena be-
cause reflexes of all types occur throughout Balto-Fennic.
stated in Hyllested (2010) there are other lexical indications that Middle
and Late Proto-Fennic4, spoken in the northern vicinity of these
cultures, may have been in direct contact with Pre-Proto-Celtic, not
always with Proto-Germanic as the provider5. Late Proto-Fennic (Proto-
Balto-Fennic) is often assumed to have emerged around 1500 BC (see
e.g. Janhunen 2009), preceded by the Middle and Early Proto-Fennic
(Proto-Fenno-Saami) stages.
The most important argument for regarding *mokku- as a loan from
Fennic6, however, is the big lexicological picture: The fact that other
Proto-Celtic hyonyms also have good candidates for a source in Balto-
Fennic.
1.2 PC ELT . * SUKKO - SOW ?
Another well-known puzzle concerns PCelt. *sukko- (W hwch, OBret.

hoch, Mod.Bret. houch, Corn. hoch, OIr socc- sow) which looks intri-
guingly similar to the most common Indo-European hyonym *suH-s
(Lat. ss, Gk. , Alb. thi, Toch. B suwe), but whose final part -kk- has
been difficult to account for. The classical handbooks reconstructs a
root variant *seu-k- next to *seu-H-, more or less explicitly interpreted
4
I use Kallios (2007, 2012) trichotomy of Early, Middle, and Late Proto-Fennic.
Of these, Early Proto-Fennic, the stage before any distinctively Balto-Fennic in-
novations, is the traditional name for Proto-Fenno-Saami (I hesitate to accept
Kallios inclusion of Mordvin here), and Late Proto-Fennic equals Proto-Balto-
Fennic, the protolanguage of all the Balto-Fennic languages (perhaps except
South Estonian, see Kallio 2007). Important for our discussion, Middle Proto-
Fennic is the stage largely recoverable by internal reconstruction immediately
before the development *ti > *ci (Kallio 2012: 166, fn. 9).
5
Among my proposals for direct loans from Proto-Fennic into Proto-Celtic are
PFc.*sanasto list of words (from *sana word + loc.suff. *-sto)
PCelt.*sanesto- secret advice (compare the semantics of the reverse loan Fi.
runo (traditional) song, poem < PGmc. *rna- or PCelt. *rn- secret);
*louna southwest; noon; lunch (from PU *luwe south + suff. -na) NIr. ln,
pl. linte lunch (further borrowed into Eng. lunch, luncheon; also a Fc. loan in
Baltic, Latv. launags, Lith. lunagas dinner); PFc. *maa land PCelt.
*magos plain, open field; PFc. *kalma disease; grave; Death-goddess PCelt.
*klamo- diseased, leprous. PFc. *tne dead may represent a reverse borrowing
from PCelt. *doueno- (mortal >) man id. into PFc. (cf. PGmc. *dewena mor-
tal, *dawjan- to die), parallel to the hyonyms. Cf. Hyllested 2010: 123-124.
6
A hypothetical hybrid form *emokko would appear even closer to the Celtic
forms. The deletion of -e- seems to presuppose stress shift to the second syllable
-mk-. We will deal with this below.
either as an old morphophonemic alternation or as an extension proper

whereby an old determinative -k- has been added to the root in its
strictest sense, i.e. *seuH-7. The extended version would then make up
the protoform not only of Celtic *sukko-, but also of Germanic -k-
variants like OE sugu, OS suga sow, as well as Latin sucula young sow
and Skt. skra- (male) boar.8
Reconstruction of root variants and root extensions is problematic in
general because it gives etymologists an extra chance to replace or add a
root-final consonant whenever these final elements turn out not to
match. The field of Indo-European studies have now reached a stage
where scholars should try to either identify the source or function of
such irregular and unexpected elements. If they are just conveniently
reconstructed whenever a solution is needed, without an explanation of
their origin or function, it jeopardizes our chances of staying on the
right track and discover the actual conditions of the past. Unexplained
root-final elements simply leave too many options open. Hamp (188)
justifiably states We have no license to drop the laryngeal in IE *suH-,
and in fact the claimed Indo-Iranian comparande conserve the *. I
therefore see here a substratum *suk-, geminated in Celtic.
Even those scholars who accepted *seuk- as a variant to account for
PCelt. *sukko- still have had difficulties explaining the Celtic gemina-
tion. Needless to say, expressive gemination has been among the sug-
gestions (Polom 1953: 541). Testen (1999) analysed the pig-names as
original compounds where -kku- reflects PIE *-p5u- livestock (the ze-
ro-grade of *p5u n.), but an animal name like *brokko- badger can
hardly count as a term from the field of animal husbandry; besides, oth-
er stems than u-stems are found in the Celtic material.
Kroonen (2011) makes a successful case in trying to eliminate the PIE
variant *seu-k- altogether: He shows, first of all, that West Germanic
forms with -g- (OE sugu, Mod.Du. zeug) simply owe this velar to a regu-
lar development of hiatus or -w- between two high vowels if at least one
of them is u. Norwegian sugge, Sw.dial. sgg sow would have arisen by
normal verschrfung of *-ww- to *-ggv- in Nordic. Second, he points out
that Lat. sucula is simply formed with the normal diminutive ending -
7
Testen (1999: 191) states that *sukko- shows phonological problems that com-
plicate any interpretations based upon its obvious similarity to Indo-European
*s-.
8
The latter interpretation is problematic seeing that no trace of length is left in
Latin, Celtic nor Germanic; all forms in these languages begin with *s- and
cannot reflect *suH-.
cula that is added to vocalic stems (cf. auri-cula little ear, avi-cula little
bird, api-cula little bee); a **s-(u)la would be impossible, and the on-
ly way to form a diminutive with -la to ss would be to add -cula to the
stem. Hence, there is no need to believe that -c- belongs to the root in
the Latin case either. Third, as already suggested by Fick, Falk & Torp
(1909), Skt. s-kra- m. wild boar is probably simply a compound,
meaning literally pig-reproducer9. Kroonens conclusion is that PIE
seuk- did not exist, since only Celtic *sukko- cannot be explained away
and needs a source10. His solution is to assert for the Celtic animal
names in -kko a Germanic source where -kk- derives from n-stems to
roots in -k- via Kluges Law. However, of the Celtic animal names, only
PCelt. *bukko- billy-goat has a safe counterpart in Germanic.
Besides, there is one more Indo-European term to take into account.
Interestingly, PCelt. *sukko- is somewhat reminiscent of Latv. cka pig
whose etymology is also disputed: In native Latvian words c- usually
occurs before -e-, -i- because it has developed from late palatalization,
or it is a borrowing from Estonian, cf. Latv. cirele lilac < Est. tsirel, dial.
for standard sirel. Jnis Endzelns simply described c- in cka as irregu-
lar from *ska which he equates with the Germanic, Latin and Indic
forms in the previous paragraph (Kaspars Ozoli, p.c.). Karulis (1992)
prefers to group cka with Lith. kial pig, visualizing a zero-grade of
*keu-/*k-, cf. kakt 'yell; howl, and various toponyms such as Kkas,
presumably place with a lot of wild pigs. This would imply an analogi-
cally mixed root where the original PBalt. distribution *kiau-/*k- was
analogically levelled to *kiau-/*ki-. While not impossible, the Latvian
and Lithuanian words actually do not have that much in common, and
as shown by Hamp (1986), the original meaning pig was probably con-
nected specifically to a stem containing the -l-. Hamp equates kial
with the element Cul- in the Welsh PN Culhwch, referring to a divine
pig, a cousin of Arthur, of the same class as Twrch Twyth. This means
9
Formally corresponding to the Middle Persian proper name Hukar < Proto-
Iranian *hkara- (Blaek 2010: 90)
10
Middle Persian xk, Modern Persian xg (Blaek 2010: 88, 90), not mentioned
by Kroonen, probably derives from a typical secondary formation in Iranian,
*h-ka-, whereby the suffix *-(i/a)ka is added to stems that would otherwise be
very short. Alternatively, it may have been a diminutive formation denoting the
piglet, cf. e.g. Alb. derk piglet < Proto-Alb. *dar-ika next to derr pig (Orel
1998: 61). Other Iranian forms show expected reflexes of the root-noun, e.g.
Young Avestan h- pig, Ossetic (Digor) xu, (Iron) xy id. Laconian Greek
(? < *), likewise absent in Kroonens account, may simply be onomat-
opoeic (Katz 2003: 206-207), cf. also Polish dzik pig.
that Culhwch is basically a tautological formation, simply meaning

(the) pig. Hamp reconstructs *keuli, not as a PIE form, but rather as a
form borrowed from a North Europan substratum language into at least
Celtic and Baltic. He includes in this word-family Lith. kuils boar,
strikingly reminiscent of Culhwchs fathers name, Cilydd < *klios.
Another possibility for Latv. cka would be to assert an onomatopo-
etic origin parallel to that behind Pol. dzik pig, a sound sequence which
seems to be geographically widespread as a word used for attracting the
pigs. However, such words often seem to originate from real nouns, cf.
e.g. Ukr. gus, a word used for calling geese, and Da. hyp, a word used
for making horses move (cf. hoppe mare < *huppan-).
In this particular case, there is no need to go far for a source. I be-
lieve that the obvious source of Latv. cka is the term for the same ani-
mal in the languages spoken to the north of Latvian, namely the Balto-
Fennic languages. The situation within Balto-Fennic itself is unclear,
too, seeing that while Finnish sika reflects *sika, some of its dialects have
tsika, pointing rather to Balto-Fennic *tsika; Karelian ugu meanwhile
points to *tsuka (BF *suka would have yielded Karelian **ugu), and
borrowings into Saami such as N Saami sokki point to a fourth variant,
*suka. This messy situation has led some Fennicists to believe that we
are dealing with different etyma, but this is too hasty a conclusion. First
of all, the irregular vocalism, although its distribution is unaccounted
for, is not unparalleled: for example, it makes little sense to separate BF
lintu bird from FU *lunta bird (cf. N Saami loddi), and this word also
shows up with -i- for expected -u- in Fennic only. The situation is ad-
mittedly not exactly parallel since, unlike the pig-word, the bird-word
does not have alternating vocalism within Balto-Fennic, but in fact this
exactly speaks for uniting the pig-words since in their case we are not
only conjecturing the occurrence of a -u-; we see it attested in front of
our eyes. In other words, if we accept lintu < *lunta in the first place,
there is no reason to doubt *(t)sika ~ *(t)suka < *tuka on the basis of the
vocalic variation. The consonantal variation in the onset is pretty
straightforward: ts- is not phonotactically allowed in Finnish, and alt-
hough the development of Fennic-Saami (Early Proto-Fennic) *ti- > *si-
and *t- > *s- is not regularly extended to the third high vowel, one
might still visualize a few cases where this tendency initiated, especially
if there was a dialectal variation between *tsi- (maybe still *ts- at the
time) and *tu-. Besides, the Fenno-Volgaic reconstruction is actually
*tuka as can be seen from Mordvin tuvo, so the Balto-Fennic onset al-
ready presents an irregularity that needs to be explained, regardless of

the situation within Balto-Fennic.
These observations mutually confirm each other: While the recon-
struction of a BF *tsuka renders it possible to provide a straightforward
source for Latvian cka pig, correspondingly the identification of Lat-
vian cka as a Balto-Fennic loan confirms the BF reconstruction with
*ts-. Incidentally, both elements of the Latv. compound meacka wild
boar therefore have the same etymological source as those in Fi.
metssika badger (since mets forest is a borrowing from Baltic
*med)a- id..
Despite the obvious chronological differences between the situation
for Latvian cka and PCelt. *sukko-, I believe that the most probable
origin for the latter is also Balto-Fennic *tsuka. Note that while Latvian
shows a secondary lengthening in its reflex, Saami sokki has a geminate
consonant. These different kinds of lengthenings probably reflect differ-
ent attempts to render what was heard as a kind of heavy syllable in the
target language. As a parallel, Modern Fennic words of the structure
CVCV (e.g. sika) are frequently perceived by e.g. speakers of Danish as
having an unexpectedly long (or partly stressed) second syllable, since
the vowel of this second syllable is longer than in Danish words of the
same structure (e.g. mokka [caf] mocha, sikahjort sika deer), making
it sound like a compound to Danes. My guess is then that both the Bal-
to-Fennic consonantal alternation in paradigms and the persistent stress
on the first syllable had some effect on the foreign renderings of Fennic
*tsuka, resulting in a geminate in Saami sokki and PCelt. *sukko- but as
vowel length in (the probably younger, but not necessarily very young)
loan in Latvian, *cka. The Balto-Fennic -k- was thus only rendered by
a geminate consonant in the languages where such consonants existed,
whereas Latvian, not possessing geminates, expressed the length in the
preceding vowel instead. One may also note that secondary lengthening
of both vowels and consonants is pretty commonplace in Fennic itself,
even in inherited words.
The u-stem formation in *mokk-u- and *sukk-u-o- (> W hob) that
Hamp (1987) identifies as part of the substratum features do not seem to
be detectable in Fennic; the Karelian -u is regular from -a, and *tsuka
looks like a normal Fennic a-stem.
1.3 PC ELT . * TURKO - WILD BOAR
A third puzzling term is PCelt. *turko- (W [obs.] twrch pig m., pl.
tyrch(-ot), OCorn. torch, OBret. torch, MBret. [Catholicon] tourch ikd.,
OIr torc [masc. o-stem] wild boar). It is remarkable that even this term
ends in -ko. The sequence -Rkk- was non-existent in Proto-Celtic (the
Brythonic development to -ch is regular after resonant) so we can even
define the last part of all three words as identical11. McCone (1992, 1993)
identified the Celtic term with Avestan rs (occurring once in an
Avestan fragment of the Pahlavi Rivyat accompanying the Ddestn-
Dng), reconstructing PIE *tuors with an original meaning cutter
(referring to the boars notorious talent for tearing and uprooting with
his sharp tusks) which has since been generally accepted (with some
reservations Lubotsky 1994; Mallory & Adams 2006: 139). The word
would then rhyme with *por5os (young?) pig and even with ior5os
deer, seemingly revealing a structure for the formation of names for
mammals (cf. *hrt5os bear; beast, PGmc. *elha- elk, *selha- seal, e.g.
Kroonen 2013 under entry *baruga-).12
As already noted by Hamp (1989: 188), McCones reconstruction of
an o-grade in *tuor5s is misleading since the attested Celtic forms
unanimously point to a persistent zero-grade *turkos. This is significant
since, as later shown by Lubotsky (1994), no other ablaut grade is to be
found in derivatives underlying this postulated root to cut, neither in
Indo-Iranian nor in Greek which would be the two other branches to
have allegedly preserved reflexes of it (however, in Greek and Indo-
Iranian the vocalization is different, pointing to *turk- and not *turk-)13.
This similar behavior obviously speaks in favor of a connection between
the Celtic boar-word and the cut-root.
Meanwhile, however, several factors speak against this. Not only is
the attestation of Av. rs restricted to a hapax in a Pahlavi fragment;
11
Matasovi (2009) reconstructs PCelt. *turkko-, but the development of PCelt. *-
k- to Welsh ch is reular in the position after -r-.
12
Some of these examples are equally disputed; Schindler (1966) regarded *selhaz
as an Early Proto-Fennic loanword in Germanic, cf. LPFc. *hleh (Fi. hylke,
gen. hylkeen) seal < EPFc., PFU *lke or *lke.
13
Av. rs- to cut, shape, upa-rsn acc.pl. hole, split, rssca acc.pl.m.
(an) end, split, rtar creator representing *rtar with a common col-
oring of to by a preceding (or following) labial; Ved. Tvar- the god-
creator, where the original r > a has become identical to the following vowel
under dissimilatory influence of a second -r- later in the word; and Gk. ,
Aeol., Dor. flesh.
its very existence is uncertain and it may in fact be an invention by the

scribe as suggested by Hoffmann (1969: 35). As he writes, the passage
yaa v az sacain yaa hu rs, functioning as the subject of a
following phrase in Pahlavi meaning (are) to be killed in the sacrifice
for the gods, corresponds to an almost identical fragment in the Ni-
rangistn that reads yaaa v az scan yaa hu prs; since the con-
text is about sacrificial animals about to be slaughtered, it seems very
likely that in the Rivyat the obsolete prs was reshaped under the in-
fluence of the verb rsaiti to cut (although of course the replace-
ment would be natural if the two words were really synonyms in the
first place). If rs is really a scribal innovation, the semantic bridge
between Indo-Iranian and Greek to cut on one side and Celtic boar
on the other is obviously destabilized.14
Furthermore, the rhyming word *por5os (> Av. prs) as we shall
see later, probably does not go all the way back to PIE and may be of
non-Indo-European origin. Besides, whatever the reason for the all-
dominant zero-grade in Greek and Indo-Iranian reflexes of *turk-, an
isolated view on Proto-Celtic *turkos as a pig-name would result in an
interpretation of the odd zero-grade as signs of a non-Indo-European
origin (cf. Hamp 1989: 188; the scarcely IE sequence *ur).
Balto-Fennic again provides us with an unheeded candidate for a
source which is remarkably close both semantically and formally: Kare-
lian torakko, torikko means nothing less than tusk of wild boar, and
this word, too, is an unproblematic internal formation within Balto-
Fennic, namely as derivative of tora, also tusk of wild boar, with the
denominal suffix -kko, cf. Fi. tora-hammas tusk (hammas tooth). The
noun is widespread within Fenno-Ugric and identical to the verbal stem
FU *torV- (> e.g. Fi. tora-) to struggle, to fight, to battle whose Saami
reflexes such as N Saami doarro- and Lule Saami trr- specifically
mean to fight with the horns; thrust (of mammals), denoting actions
typically carried out by boar tusks. Thus, while the ultimate underlying
semantics may not be very far from what McCone suggested, the words
character as Proto-Fennic loanword seems clear to me especially in the
light of similar scenarios for *sukko- and *mokku-.
14
McCone was of course aware also of this part of Hoffmanns article, but writes
(1992: 99): Convincing though the interpretation of prs as pars < *por5os
is, its corruption to a non-existent rs somewhat reminiscent of the verb to
cut is a less attractive postulate.
2 Indo-European terms
2.1 PIE *5 ER - N - WILD BOAR TUSK ; BRISTLE ?
PFP terms seem to confirm the cultural importance of the pig in prehis-
toric Northern Europe, even in cases where nothing points to a Fennic
origin. Hamp (1987: 189) declares himself incapable of etymologizing
OIr. cran f. sow, but suggests Welsh cranan wild sow (attested in the
White Book of Rhydderch and then once later) as cognate. An exact
Proto-Celtic reconstruction is admittedly not possible on the basis of
these two forms. However, the common Lithuanian word for the wild
boar, ernas, with a variant ernkas, is suggestive of a formation from
the PIE stem *5er-n- horn , referring to the boars tusk; cf. the double
meaning of Skt. r-g- horn; elephants tusk15. It is tempting to include
the first member of OLFr. chranni-chaltia pigs den (Quak 1983)16
whose first member has so far been considered obscure. A term *5er-n-
designating the wild boar, thus seems to unite Celtic, Germanic and
Baltic.
2.3 PIE * G OR ) O - WILD BOAR
The form *5er-n- mentioned immediately above may appear similar to

Fi. karjas, karju wild boar, karja livestock (cf. also Est. karjane shep-
herd) but these Balto-Fennic forms notably lack the nasal element. They
are conventionally connected to the Finnish verb karjua to roar and
the noun karjainen rut, rutting (of male animals), but the question is in
what direction the derivational process originally went. The ending -as
in karjas suggests that we are dealing with an Indo-European loanword.
A PIE *hr-)o-s can indeed be reconstructed on the basis of Gk.
wild boar and Alb. derr (Mallory & Adams 2006: 142). The Proto-
Albanian form was *darja (Orel 1998: 61), corresponding perfectly to the
Greek form; an alternative shape *h)-ro-s (preferred by Demiraj 1997:
131-132) would admittedly also be possible on the basis of Greek and Al-
banian only if the original Alb. sg. **darr was generalized the umlauted
15
Lith. rnas with mtatonie rude also exists. Smoczyski (2003: 10, 92) prefers
an inner-Baltic derivative from eria (also ers) bristle. Cf. also Hyllested &
Gliwa 2009: 50) on the mechanisms behind this derivational process.
16
I thank Guus Kroonen for having drawn my attention to this form.
pl. derra17. However, note that the Fennic forms now confirm a PIE
*hr-io-s if they are borrowings from an otherwise unattested, but per-
fectly matching, PGmc. *garjaz. One might visualize that another ken-
tum language than Germanic, known or unknown, could be the source
of karjas, but it would have to be a language that lost the velar-palatal
distinction since PIE *- would otherwise be attested as a sibilant, cf. Fi.
salko pole, stake PIE *alg-. In forms old enough to have retained
an o-vocalism, palatals are usually substituted with sibilants in FU, cf.
Fi. koipi leg of a bird; (colloq.) human leg < (NW) PIE *5o)po- pole,
stake (Skt. pa- tail, penis, Alb. thep peak, point, cog, tooth, Lat. cip-
pus pole, stake)18, but the a-vocalism in karjas points to Germanic as
the most probable source. This would, conveniently for our reconstruc-
tion of the prehistoric situation, mean that Indo-European borrowings
of hyonyms into Fenno-Ugric took place much later than the borrow-
ings of hyonyms in the reverse direction.
3 Motivation for borrowing
That Celtic and Germanic hyonyms that can be shown to have originat-
ed in Fennic should strike one as unexpected, partly since boars play a
significant role in Celtic and Germanic (as well as in Greek) mythology,
partly since wild boars are generally Southern animals in Europe.
However, there are chronological layers to distinguish: These terms
must go back to a time from before the emergence of these specific traits
in at least the Celtic (and probably also the Germanic) cultures and
religions. The terms themselves and the culturual significance they re-
veal must both emanate from a common non-Indo-European source.
The question then remains if we can trace any extralinguistic evidence
for a special importance of pigs among the Fennic peoples. Tacitus
wrote on the Aestii, a Northeast European tribe in the Baltics:
They worship the mother of the gods: as an emblem of that superstition

they wear the figures of wild boars: this boar takes the place of arms or of
any human protection, and guarantees to the votary of the goddess a mind
at rest even in the midst of foes
17
Hulds (1984: 148) reconctruction of an unparalleled derivative *suo)n-ro- from
the stem occurring in PGmc. *swna- (PIE *suH-ihno-) is thus unnecessary.
18
In fact, the Fennic form confirms IE o-vocalism which is not otherwise directly
attested in this form.
While the name of the Aestii (also Aestiorum gentes) is no doubt the
source of that of Estonia, it has generally been assumed that the Aestii
were in fact speakers of Baltic and not Balto-Fennic languages, whose
name was later transferred to the Balto-Fennic Estonians. However,
Bammesberger & Karaliunas (1998) convincingly show that the Aestii,
and that the original ethnonym, definitely denoting a Fennic people
from the point of view of the Scandinavians, had an extra -r- in stem
(Eistr-)19. As I demonstrate elsewhere (Hyllested forthcoming), the Bal-
tic stem *aistra- had a meaning synonymous to that of PGmc. *finn-20
and is in all probability a loan translation, clearly indicating that the
name denoted Fennic peoples.
Archaeological evidence can be supplied. Sites from the Pitted Ware
culture (3200-2200 BC) on southern Scandinavian coasts from Svealand
and land to the Danish island of Funen contain pig bones in large
quantities emanating from domesticated pigs rather than wild boar. It is
known that they lived side by side with battle-axe peoples, traditionally
attached to Indo-European and Pre-Proto-Germanic expansions. The
people of the pitted ware were not direct ancestors of Northern Scandi-
navians, but more closely related to peoples of the contemporary Baltic
region (Rowley-Conwy & Stor 2007; Malmstrm & al. 2009), and
blending of styles and techniques between pitted ware and battle-axe
peoples took place especially in the later half of the period, 2700-2200
BC (Larsson 2003). It is clear from the datings listed above that this pe-
riod does not fit exactly with the Balto-Fennic protolanguage but rather
with the traditional dating of Fenno-Volgaic or Fenno-Saami. But at
least we have reason to believe that the cultural significance of the pig
contiued into the Balto-Fennic period, and perhaps, in the light of Taci-
tus account, even into historical times.
19
Cf. Eistra dolgi the Estonian enemy (Ynglingatal) and devenit in Eistriam, puer
Olavis Eistriis in servum venumdatur (Historia Norvegiae). The original vocal-
ism *aist- is secured by Old Gutnish utan foru i aina oy vir Aistland, sum haitir
Dagaii [they] travelled to an island off Estonia called Dag (Guta Saga).
20
Pimple or other protuberances from the skin of humans or animals such as
fish scale, fin, larvae under the skin of cattle (causing folly). The Baltic
meaning is reconstructed on the basis of a loanword in Livonian istar pimple
(the dialectal variant vistar can only be explained on the basis of Baltic) and
Lith. aistr intense passion; cf. Gk. intense passion; larvae under the
skin of cattle, ON eistra testicle, OHG eiz larvae under the skin of cattle, Gk.
tumor.
Also seal bones are prominent among the findings in Pitted Ware
sites; note that the dating fits perfectly with Schindlers etymology for
PGmc. *selhaz seal mentioned in fn. 11 above.
4 Badgers and pigs
It is relevant now to introduce PCelt. *brokko- badger (Ir. broc, W

broch)21, into the discussion for two reasons: First, it is an animal name
of obscure origin containing the geminate -kk- that we recognize from
the pig terms. Second, it is typologically common to compare badgers to
pigs and their offspring to piglets, cf. e.g. the aforementioned Fi.
metssika badger (lit. forest pig ~ Latv. meacka wild boar), Nw.
svin-toks badger (1st part svin swine, 2nd part *ahsu- badger), Eng.
sow, Da. so female of pig, but also female of badger, Eng. boar male of
pig, but also male of badger, and Da. gris piglet, but also young of
badger, cub. 22
Hence we should consider the possibility that PCelt. *brokko- could
be derived from word for pig of non-Celtic origin. It is almost for cer-
tain connected to W Germanic *brakka- (scent) hound, dog used for
hunting (OHG bracko, MDu. bracke sleuthhound) since hounds have
been used for hunting game such as (very often specifically) badgers or
boars since ancient times23, cf. also the modern term dachshund < NHG
21
Borrowed into OE as brocc; from Eng. it has further been borrowed into Danish
as brok, mostly used in the definite form brokken (Brokken) as a kind of semi-
personal name, especially common in hunters language.
22
More examples can be found in Ritter (1975). We might add as a modern exam-
ple the name of the hog badger (Arctonyx collaris) from Southeast Asia. In this
light, it might be worth investigating whether the otherwise opaque Eng. word
badger could be a borrowing from Brythonic (cf. W baedd) and perhaps even
confirm the reconstruction *bad)o- rather than the alternative *bas)o-. The root
in Lat. fodi, Hitt. padda- to dig (the ground), bury, Lith. bed dig comes to
mind when you compare e.g. Danish grvling badger ~ grave to dig (and grav
fox or badger earth, burrow), The root is attested in Celtic but with innovative
meanings (cf. Hyllested 2010: 115).
23
I find Kroonens (2013) alternative reason for linking the two words, that the
badger is an animal with a strong sense of smell less relevant. His semantic re-
construction sleuth dog is also somewhat anachronistic since he seems to pre-
fer that the word is Proto-Germanic; sleuthhound is admittedly the meaning in
both Middle Dutch and Old High German, as well as in modern Dutch brak
and modern German Bracke (hence Eng. bracke) but the breed does not go back
to ancient times. An less specified meaning hound, hunting dog or scent dog
Dachshund, lit. badger dog.24 The origin of the latter breed is uncertain,
but probably not bred as early as when Proto-West Germanic was spo-
ken. Cf. also ON Brokkr, the name of the dwarf who creates the mytho-
logical boar Gullinbursti.
Ru. barsuk badger is a loanword from Turkic *borsuk (*borsuq) id.,
derived internally in Turkic from *bor, *boz grey and akin to Written
Mongolian borki old badger (Khalkha Mongolian borx, Buryat burxi,
Kalmyck bork badger) borrowed into Tuvin (a Turkic language spoken
to the North of Mongolian in Central Asia) as murgu male marmot
(Khabtagaeva 2009: 159). The word has also been borrowed into Hun-
garian as borz badger. Fi. myr comes from older *mkr, cf. Karelian
mkr, Est. mger. The Balto-Fennic word can only be connected via
dissimilation from *mrk. Celtic *br- can come from older *mr-, and a
stress shift with vowel loss in the first syllable is reminiscent of the hy-
pothetical process *emokk- > *-mkk- sketched above. That we are
dealing of a wanderwort of Altaic origin thus seems conceivable, but its
routes are not entirely clear.
5 *por"o- and its many variants
The protoform *porko- (> MIr orc [m. o-stem] young pig, Lat. porcus
id., Lith. paras young pig; castrated male hog, OCS pras young pig,
Av. prs id., PGmc. *farha- pig > e.g. OE fearh, OHG farah) is al-
most universally accepted as a major PIE pig-word. Since Benveniste
(1969) the meaning has been reconstructed as young pig or piglet25.
However, as already noted by Hamp, the word is actually geographically
confined to the Northwestern half of the Indo-European area. It is not
found in Albanian, Greek, Armenian, Anatolian, Indo-Aryan, or To-
charian, and within Iranian, it is exclusively attested in the Northern
fringes, geographically speaking. An Indo-Iranian preform is admitted-
ly widely assumed as the basis for Fenno-Permian *poras or *poras
would be better. Kroonen further declares that the relationship of this word
with the rhyming *rakka- is unsolved. Since both Sw. by-racka mongrel dog
and hundracka cur, the loanword in Fi. rakki id., and MDu. rekel bad dog;
male dog (> Du. rekel villain) are clearly pejorative, I wonder if it is not simply
derived from the verb *ragg/kn- to move to and fro, to stroll, cf. Sw. racka to
roam, to wander about (used typically of dogs).
24
Da. gravhund, lit. grave dog, cf. grvling badger, derived from grav grave.
25
As Hamp (189) says The question is not one of domestication, but of cultural
value.
pig (Benveniste 1949), but as some scholars (notably Napolskich 2002)

have pointed out, the Balto-Fennic forms may just as well have been
borrowed from a stage of Balto-Slavic, while the Eastern (Mordvinian
and Permian) forms either suspiciously lack the Indo-Iranian ending or
are aberrant in other ways. Outside Europe, the only attested forms are
therefore Av. prs and Khot. psa (Kurdish purs is a ghostword, cf.
Hoffmann 1967: 35). These are both important culture languages of Cen-
tral Asia, spoken close to Altaic and especially Turkic languages. As we
shall see in a moment, there is a possibility that *porko- is really a Cen-
tral Asiatic culture-word.
First, it is important to note that the main justification for starting to
look for a non-Indo-European source for *porko- is the variation with
which similar words with similar meanings occur in Europe, displaying
a remarkably colorful variation of irregular correspondences: Alongside
*porko- it seems necessary to reconstruct a variant *poro- as the basis
for PGmc. *farkna- pig and CSl. *porz boar; ram; bull.26 Further-
more, PGmc. *barga- (OHG barug, OE bearg, Eng. barrow, OIc. brgr,
MidDu. barg) would have to go back to a PIE form *borko-, * bor5o- or
*borgo-27. If the Germanic protoform was actually *baruga-, the stem
*baru- correspond to Slavic forms like Ru. brov [m.] hog, castrated
boar, SCr. brv [m.] sheep, dial. hog, castrated boar. Finally, OHG
br, OE br boar most likely go back to PGmc. *baira-28. Pre-Proto-
26
It is perhaps conceivable, but hardly provable, that *porz was reshaped from
earlier *pors under the influence of Slavic *kn-orz (domesticated) boar, lit.
with testicle (on this form, see Kretov 1994), and that PGmc. *farkna- can the-
oretically be interpreted as a diminutive *farhkna- of *farha-. All things consid-
ered, I find it more economical and therefore more probable to assume that we
are dealing with a true irregular variation, and that both forms go back to what
would have been PIE *poro-.
27
Not to be reconstructed *baruga-, cf. Ball & Stiles 1983.
28
*baiza- is also possible. Kroonen (2013) regards *baira- as more likely because
the ON hypocoristic form bassi is more likely to have been derived from *brr <
*baira- than *beirr < *baiza-. However, bassi may just as well be a hypocoristic
form of baggi small and thick, compact animal, cf. Da. basse piglet, but also
small, fat male horse, wether, dog, bull etc., thick insect, rather the same
meanings as those of baggi. Polom (1986) and Schrijver (1997), in the light of
the irregular correspondence with MW baed, W baedd, OCorn. bahet boar,
conclude that both the Germanic and the Celtic form are of the same non-Indo-
European origin. However, if the PGmc. form was really *baira- and especially
if at the same time the Brythonic forms are from *bad)o- and not from *bas)o-,
in my opinion they are not even sufficiently alike for us to regard them as the
same word.
Germanic even had a *pr-o- (PGmc. *fra-), but this is clearly derived
from *per- to give birth, Gmc. *farzan-.
therefore think that the irregularities of the Fenno-Permic forms are
due to the fact that some of them were borrowed from R-Turkic, cf. the
Chuvash form por badger (via piglet? Note the exact meaning of
*por5os) with initial p- and without reminiscences of the suffix -uk, into
the Eastern Fenno-Permic languages: Mordvin (Moksha) pu, purc,
Udmurt par, par, pars, pari, Komi por29, with their *-- actually re-
flecting Turkic -s- (= Chuvash -) and not an alleged PIE *-5-. The Bal-
to-Fennic languages (*porsas) and perhaps Erzya Mordvin (purcos,
puis, pursuz) instead borrowed their forms from a stage of Balto-
Slavic, which comprised the still productive ending -as. The Balto-Slavic
form, in turn, like the Germanic and Italo-Celtic forms, would ultimate-
ly have derived from older Turkic *borsuq. Again, the exact directions
are not clear, but a culture-word situation would account for the geo-
graphical distribution, the irregular variations of similar pig- and
badger-words across Europe, aberrancies in Fenno-Ugric and perhaps
even the exact meaning of *por5os if the meaning went via young of
badger (Da. gris, also pig). Finally, the motivation for a borrowing
from Altaic would be straightforward, namely the fact that badger is a
common traditional dish among Turkish and Mongolian peoples.30
6 Conclusions
Etymological analyses that take Uralic material into account confirm

Hamps (1987) claims that 1) most Celtic terms for pigs derive from a
non-Indo-European language in Northern Europe, and that 2) the pig
occupied a special cultural position among the speakers of this language.
Three terms specifically can be traced directly back to the Balto-Fennic
protolanguage (Late or Middle Proto-Fennic), and it seems possible to
link this to extralinguistic data ranging from prehistoric archaeology to
29
From Komi it has been borrowed into Khanty as V pors, DN pur, O por;
and into Mansi as KU prs, P prs, So. pr; Nenets Sj. pors; and from Khanty
it has been borrowed further into Nenets O as pora, Nj. poes.
30
At the conference leading to this publication, I argued that Turkic *borsq
could have been borrowed to Gmc. as *barza- yielding *brrua- that could
then have been transferred to PCelt. brokko-, involving the same stress shift to
the penultimate syllable -bVr-k- as in *mokku- < *e-mk-. However, this is of
course not possible if the PGmc. form was in fact *barga- and not *baruga-. See
above on the role of PFc. *mkr.
historical accounts. These conclusions have important implications both

linguistically and culturally since they strengthen the hypothesis
sketched in Hyllested (2010) that lexical exchange took place directly
between the speakers of Proto-Fennic and Proto-Celtic, and that parts of
the Celtic vocabulary have a Fennic origin. Other Northern European
pig-terms are of Indo-European origin, but their exact reconstruction in
some cases is only possible if Uralic material is analyzed on an equal
footing. Studies in language contact within a given region can benefit
from the the involvement of all languages actually spoken in thatregion.
It seems worthwhile to investigate whether other portions of the IE lexi-
con of Northern Europe that allegedly derive from long-gone substrata,
are actually just loanwords from (different stages of) the adjacent lan-
guage family, Uralic.
References
Ball, C.J.E. & Patrick V. Stiles, 1983: The derivation of Old English geolu yellow,
and the relative chronology of smoothing and back-mutation. Anglia 101. 5
28.
Balles, Irene, 2010: Some new Celtic and other etymologies. Giovanni Belluscio
& Antonio Mendicino (eds.): Scritti in honore di Eric Pratt Hamp per il suo 90
compleanno. Cosenza: Universit di Calabria. Pp. 1520.
Bammesberger, Alfred & Simas Karalinas, 1998: Zur Frage nach der ethnischen
Identitt der Aisten. Alfred Bammesberger (ed.): Baltistik: Aufgaben und
Methoden. Heidelberg: Carl Winter. Pp. 3947.
Bellquist, J.B, 1993: Badger in Indo-European. Journal of Indo-European
Studies 21: 321346.
Benveniste, mile, 1949: Le nom du porc. Bulletin de la Socit Linguistique de
Paris 45: 7491.
Benveniste, mile, 1969: Le vocabulaire des institutions indo-europennes. Paris.
Blaek, Vclav, 2010: Balto-Slavic uepr)o- boar . Baltistica XLV(I): 8594.
Campbell, Lyle, 1980: The psychological and sociological reality of Finnish vowel
harmony. Robert Michael Vago (ed.): Issues in vowel harmony: Proceedings
of the Cuny Linguistics Conference on Vowel Harmony, May 14, 1977 [= Studies
in Language Companion Series 6]. Amsterdam: John Benjamins. Pp. 245270.
Demiraj, Bardhyl, 1997: Albanische Etymologien: Untersuchungen zum Albanischen
Erbwortschatz [= Leiden Studies in Indo-European 7]. Amsterdam & Atlanta
(GA): Rodopi.
Hamp, Eric P., 1986: Culhwch, the Swine. Zeitschrift fr Celtische Philologie 41.
257258.
Hamp, Eric P., 1989: North European Pigs and Phonology. Zeitschrift fr Celt-
ische Philologie 43: 192193.
Hamp, Eric P., 1987: The pig in ancient Northern Europe. S.N. Skomal & E.C.
Polom (eds.): Proto-Indo-European: The archaeology of a linguistic problem
[Fs. Gimbutas]. Washington, D.C.: Institute for the Study of Man. Pp. 185190.
Hoffmann, Karl, 1967: Drei indogermanischen Tiernamen in einem Avesta-
Fragment. Mnchener Studien zur Sprachwissenschaft 22: 2938.
Huld, Martin, 1984: Albanian Etymologies. Bloomington: Slavica.
Hyllested, Adam, 2010: The precursors of Celtic and Germanic. Stephanie
Jamison, Craig Melchert and Brent Vine (eds.): Proceedings of the 21st annual
UCLA Indo-European Conference. Bremen: Ute Hempen. Pp. 107128.
Hyllested, Adam, forthcoming: Baltic etymology as a key to Fennic ethnonyms.
Ms.
Hyllested, Adam & Bernd Gliwa, 2009: Metatony in Lithuanian internal deriva-
tion. Thomas Olander & Jenny Helena Larsson (eds.): Stressing the past: Pa-
pers on Baltic and Slavic Accentology [= Studies in Slavic and General Linguistics
35].Amsterdam: Rodopi. Pp. 4752.
Hkkinen, Jaakko, 2007: Kantauralin murteutuminen vokaalivastaavuuksien
valossa. MA thesis from the University of Helsinki.
Janhunen, Juha, 2009: Proto-Uralic what, where, when?. The Quasquicenten-
nial of the Finno-Ugrian Society [= Mmoires de la Socit Finno-Ougrienne 258].
Helsinki: Suomalis-Ugrilainen Seura. Pp. 5778.
Kallio, Petri, 2007: Kantasuomen konsonanttihistoriaa. Jussi Ylikoski & Ante
Aikio (eds.): Smit, snit, stnehmit: Riepmola Pekka Sammallahtii
miessemnu 21. beaivve 2007 [= Mmoires de la Socit Finno-Ougrienne 253].
Helsinki: Suomalais-Ugrilainen Seura. Pp. 229249.
Kallio, Petri, 2012: The non-initial vowel reductions from Proto-Uralic to Proto-
Finnic. Tiina Hyytiinen, Lotta Jalava, Janne Saarikivi & ErikaSandman
(eds.): Per Urales ad Orientem: Iter polyphonicum multilingue.Festskrift tillgnad
Juha Janhunen p hans sextiorsdag den 12 februari 2012 [= Mmoires de la So-
cit Finno-Ougrienne 264]. Helsinki: Suomalais-Ugrilainen Seura. Pp. 163175.
Karulis, Konstantns, 1992: Latvieu etimoloijas vrdnca. Riga: Avots.
Katz, Hartmut, 2003: Studien zu den lteren indoiranischen Lehnwrtern in den
uralischen Sprachen. Heidelberg: Winter.
Kerkhof, Peter Alexander, forthc.: Who brings home the *bakkn?. Ms.
Khabtagaeva, Bayarma, 2009: Mongolic elements in Tuvan. Wiesbaden: Harrasso-
witz.
Kretov, Aleksej A., 1994: Baltijskie i slavjanskie prodolenija i.-e. *or-. Statja I:
lit. iara i prasl. *knorz. Linguistica Baltica 3: 197206.
Kristiansen, Kristian, 2009: Proto-Indo-European Languages and Institutions: An
Archaeological Approach. Marc Vander Linden & Karlene Jones-Bley (eds.):
Departure from the Homeland: Indo-Europeans and Archaeology. Washington,
DC: Institute for the Study of Man. Pp. 111140.
Kristiansen, Kristian & Thomas Larsson, 2005: The Rise of Bronze Age Society: Trav-
els, Transmissions, and Transformations. Cambridge: Cambridge University
Press.
Kroonen, Guus, 2011: Neuniederlndisch zeug Sau und spugen speien: zwei
Fllen von paradigmatischem Wechsel der Westgermanischen Verschrfung?.
Amsterdamer Beitrge zur lteren Germanistik 67 [Fs. Quak]: 145157.
Kroonen, Guus, 2013: Etymological Dictionary of Proto-Germanic. Leiden: Brill.
Kmmel, Martin, 2004: Ungeklrtes *-u- neben Liquida in Germanischen Nomi-

na. A.Hyllested, A.R. Jrgensen, J.H.Larsson & Th. Olander (eds.): Per
Aspera ad Asteriscos [Fs. Rasmussen]. Innsbruck: IBS. Pp. 291303.
Larsson, sa M., 2003: Uniting Strategies. Material Culture in Eastern Middle
Sweden at the End of the Middle Neolithic. sa M. Larsson & Ludvig Pap-
mehl-Dufay (eds.): Uniting Sea II: Stone Age Societies in the Baltic Sea Region.
Uppsala: Institutionen fr Arkeologi och Historia. Pp. 132149.
Lubotsky, Alexander, 1994: Avestan rtar- and the Indo-European root tur-
. Die Sprache 36/1: 94102.
Mallory, James P. & Douglas Q. Adams, 2006: The Oxford Introduction to Proto-
Indo-European and the Proto-Indo-European World. Oxford: Oxford Universi-
ty Press.
Malmstrm & al., 2009: Ancient DNA reveals Lack of Continuity between Neo-
lithic Hunter-gatherers and contemporary Scandinavians. Current Biology
19,15. doi:10.1016/j.cub.2009.09.017
Matasovi, Ranko, 2009: Etymological Dictionary of Proto-Celtic. Leiden: Brill.
McCone, Kim, 1992: OIr. torc, Av. rs < PIE *twors (cutter,) boar .
Mnchener Studien zur Sprachwissenschaft 53: 99100.
McCone, Kim, 1993. The etymology of Old Irish torc boar . Bela Brogyanyi &
Reiner Lipp (eds.): Comparative-historical linguistics: Indo-European and Finno-
Ugric. Papers in honor of Oswald Szemernyi III. Pp. 291292.
Napolskich, Wladimir, 2002: Zu den ltesten Beziehungen zwischen Finno-
Ugriern und zentraleuropischen Indogermanen. R. Blokland & C. Has-
selblatt (eds.): Finno-ugrians and Indo-Europeans. Linguistic and literary con-
tacts. Maastricht: Shaker. Pp. 265271.
Olsen, Birgit Anette, 2012: Contributions to Armenian etymology. Stephanie
Jamison, Craig Melchert & Brent Vine (eds.): Proceedings of the 23rd annual
UCLA Indo-European Conference. Bremen: Ute Hempen. Pp. 113124.
Orel, Vladimir, 1998: Albanian Etymological Dictionary. Leiden, Boston & Cologne:
Brill.
Polom, Edgar C., 1986: The non-Indo-European component of the Germanic
lexicon. Annemarie Etter (ed.): O-o-pe-ro-si. Festschrift fr Ernst Risch zum
75. Geburtstag. Berlin & New York (NY): Walter de Gruyter. Pp. 661672.
Quak, Arend, 1983: Zu den salfrnkischen Tierbezeichnungen. Amsterdamer
Beitrge zur lteren Germanistik 19: 766.
Ritter, Ralf-Peter, 1975: Slavisch *euo- Dachs . Die Sprache 21: 191194.
Ritter, Ralf-Peter, 2005: Zur Herkunft von fi. myr Dachs . Cornelius Has-
selblatt, E. Koponen & Anna Widmer (eds.): Lihkkun lehkos! Beitrage zur Fen-
nougristik aus Anlass der sechzigsten Geburtstags von Hans-Hermann Bartens
[= Verffentlichungen der Societas Uralo-Altaica 65]. Wiesbaden: Harassowitz.
Rowley-Conwy, Peter & Jan Stor, 1997: Pitted Ware seals and pigs from Ajvide,
Gotland: Methods of study and first results. G. Burenhult (ed.): Remote
Sensing, vol. 1. Teses and Papers in North-European Archaeology. Stockholm:
Stockholm University. Pp. 113125.
Sainan, Lazare, 1905: La cration mtaphorique en franais et en roman: images
tires du monde des animaux domestiques II: Le chien et le porc avec des appen-
dices sur le loup, le renard et les batraciens (Beihefte zur Zeitschrift fr
romanische philologie. 1., 10. hft.). Halle an der Saale: Max Niemeyer.
Schindler, Jochem, 1966: Germanisch *selxaz Seehund . Die Sprache 12: 6566.
Schrijver, Peter, 1997: Animal, vegetable and mineral: some Western European
substratum words. Alexander Lubotsky (ed.): Sound law and analogy: papers
in honour of R.S.P. Beekes. Amsterdam: Rodopi. Pp. 293316.
Smoczyski, Wojciech, 2003: Studia Bato-Sowanskie 2. Krakw: Wydawnictwo
Uniwersytetu Jagielloskiego.
Testen, David. 1999: Stem-final *-kk- in Celtic terms for pig . riu L: 161164.
Wagner, Heinrich, 1954: Varia 4. Ir. Broc, kymr. Broch Dachs . Zeitschrift fr
Celtische Philologie 24: 9293
WGS = August Fick, Hjalmar Falk & Alfred Torp, 1909: Wrterbuch der Indoger-
manischen Sprachen: Dritter Teil: Wortschatz der Germanischen Spracheinheit.
Gttingen.
The Other Horse:
Germanic Cognates of caballus?
Abstract
Simons (2005) reconstruction of a second PIE term for horse as the

source of Lat. cab, caballus, Iranian *kaba-, OCS kobyla and Fi. hepo
(via Germanic) is accepted, but its shape must have been *keb-, not
*keb-, and the PGmc. form was *heb, not *hepa-. The Germanic evi-
dence is not restricted to loanwords in Finnish; in this article, Da. hoppe
mare and its Germanic relatives are interpreted as the old oblique form
of a PGmc. n-stem paradigm *heb ~ *huppaz whose nominative was
transferred into Northern Balto-Fennic as hepo while the Southern lan-
guages borrowed *huppaz (perhaps even in a Pre-Germanic shape with
*-p-n- before Kluges Law operated). This solution provides the first
serious explanation for the irregular variation within Balto-Fennic, nota-
bly the difference between Fi. hepo and Est. hobune, hopene.
1 Reconstructing PIE *keb- the slow horse
Alongside Alongside the famous PIE term *h!uos horse (most re-
cently treated by Huld 2004, de Vaan 2009, Blaek 2010), Simon (2005)
has reconstructed another Indo-European word for this animal, PIE
*keb-, on the basis of the following forms:
a) OCS koby-la mare(< *kob-n; borrowed into MHG dial. as ko-

bel jade, hack, nag, crowbait, and, as a back-formation, Swabian
kob id.)1;
1
Mod.Eng. cob thickset horse might be unrelated if simply identical in origin to
the numerous other meanings of Eng. cob, most of which can be united by the
meaning plump or roundish object, animal or person.
b) late gloss Lat. cab castrated horse (< *keb-n);

c) Iranian *kaba-, *kabala- (borrowed into Greek as
working horse [Hes.] and into Latin as caballus) > Khotan Saka
kab horse (borrowed into Karakhanidic Middle Turkic [11th c.
AD] as kevel (at) quick horse), Mod.Pers. kawal mule; and
d) PGmc. *hepa-, reconstructed solely to account for supposed
loanwords in Balto-Fennic languages, e.g. Fi. hepo, hevonen
horse
Simon gives convincing arguments for leaving out Slavic *kon and
*komon (pace Snoj 2003) and for rejecting alternative loan directions
such as Greek into Persian. On the whole, the proposal looks quite ac-
ceptable, despite Indo-Europeanists usual reluctance to accept inherit-
ed lexemes with the rare unaspirated *b for PIE.
On closer inspection, it turns out that it is unnecessary to reconstruct
this unaspirated *-b- since the only element that precludes its aspirated
counterpart *-b- is exactly the PGmc. *-p- which Simon supposes to
only be reflected in Balto-Fennic loans. Since Balto-Fennic did not pos-
sess voiced stops2, the language would be unable to show evidence of
original voicing, so PGmc. *-b- < PIE *-b- is just as possible in this in-
stance. In fact, the lack of Winters Law in Slavic (not kabyla) unam-
biguously points to an aspirate, if kobyla is inherited at all (see Blaek
2010 on the East Iranian animal-term suffix *-la-; Gob 1985).
2 The problem of Balto-Fennic variation
A serious obstacle not adressed by Simon is that only the forms occur-
ring in the Northern Balto-Fennic languages (Finnish, Karelian, Lude
and Veps) differ from each other in regular ways. Most forms in the
Southern languages (Votic, Estonian, South Estonian, and Livonian)
exhibit variation which is both internally irregular and deviates from
the Northern forms. The biggest issue is the vocalism with *-o-, but the
variation in the suffixes between *-u- and *-e- and between *-p- and *-
pp- is not unproblematic either. Next to hebu, Estonian variants include
2
In Modern Finnish, voiced stops occur in very recent loanwords, and *-d- oc-
curs as the weak form of *-t- in consonant gradation, but in Balto-Fennic times
it is reconstructed as a voiced dental fricative.
The Other Horse: Germanic Cognates of caballus 93
hobune, hobu, hobene (<b> = /p/), even hopen with geminate /p:/ (ren-
dered in writing as <p>) and Votic has opn next to pn:3
Northern Balto-Fennic:
Finnish Karelian Lude Veps

hepo hebo hepo hebo, hebo, hbo
hevonen heboe heboie S dial. hebe
Ingr.dial. Olonets dial.
heppoin heboine
Kukkuzi Votic
hepoina
Southern Balto-Fennic:
Votic Estonian Livonian

- dial. hebu -
- hobu, hobo bbi, ybbi, ibbi, bbi
opn, pn hobune, hobene, hopen -
The variation in both vocalism and consonantism between different Bal-

to-Fennic forms has been close to ignored in virtually all etymological
proposals for hepo. To mention some of the most recent proposals,
Liukkonen (1999) argues for an origin in Baltic *eva- with metathesis,
cf. Lith. aviens working-horse that also has a nasal suffix; and Kort-
landt (1997) surmises a loan directly from (some stage of) PIE *h!uos
with the laryngeal preserved4. LGLOS (I: 95-96) does mention the
3
Est., Vot. and Liv. represents an unrounded mid-high back-vowel which most
often originates from BF *e.
4
It is tempting to suggest as an alternative Iranian etymology of the word, involv-
ing completely different elements, cf. Chin. chb < Middle Chin. *tit puat,
borrowed from Sogdian rp- (yrp) horse used in the valleys, a for-
mation identical to Lat. quadru-ps four-legged (animal) (Yoshida 2009).
Uralic, Fenno-Ugric, Fenno-Permian and Fenno-Saami * become *h- in Balto-
Fennic. However, neither such a proposal would solve the issue of variation in
Balto-Fennic material; even if one disregarded the suffixes and argued that the
alternation between *-pp- and *-p- (~ weak grade *--) could be due to different
renderings of the Iranian cluster *-rp-) the occasional o-vocalism would re-
main enigmatic.
Southern forms, but fails to account for their occurrence. The possibility
of a Germanic origin (from < *ehwa-), again with metathesis, is left
open, but the entry ends with the judgement Kaum ein germ. Lehnw..
3 Danish hoppe and its closest relatives
The earliest attestation of Da. hoppe mare as a common noun is from

1621, but it is known much earlier from a place-name Hoptrup near Ha-
derslev in S Jutland, 1287-1307 Hoptorp, 1289 Hoppetarp, 1421 Hoptorpe.
Nothing points to an original meaning mare: The names correspond
structurally to other towns in S Jutland: Hostrup, attested 1298 as
Horstrop (near Tnder), ca. 1325 Horsthorp (near Esbjerg), 14th. c. Hors-
torp (near Vejle) where the first member is hors horse (obs.) (Jrgensen
1994). We also know that hoppedreng (obs., attested 1653) simply meant
groom, stableboy and did not refer to mares in particular5. Its Swedish
counterpart is a false friend hoppa old meagre horse (dial., att. 1683),
and the Nordic word has been borrowed into OE as hobin > ME hobyn
nag > Eng. hobby. The semantic development horse mare is quite
common, cf. e.g. PGmc. *marha- > Older Da. mar mare (alongside
mr from the original fem. *marhjn-) and PGmc. *hursa- horse >
Bornholm dial. of Da. horse mare.
On the basis of these forms and LG dial. huppe, we can reconstruct a
PGmc. *huppaz. The standard Danish and Swedish handbooks (e.g.
Svenska Akademins Ordbok, SAOB) otherwise assume, often with some
hesitation, that hoppe is an inner-Scandinavian derivative of the word
hoppe to hop, jump < *huppn-. However, a homonymous verb existed
in Germanic with the meaning move backwards, retreat (especially of
horses) and for the latter non-iterative variant *happn-, and judging
by the Swedish and English semantics, this is a more obvious connection
(which however does not work completely, see below). So far, on the
basis of these forms and LG dial. huppe, we can reconstruct a PGmc.
*huppaz slow horse.
5
The surname Hoppe, known from 1410 onwards, according to Gammeldansk
Ordbogs note collection (www.gammeldanskordbog.dk) may instead be from
hoppe in the meaning shrimp, derived from the verb hoppe to hop (cf. also
Mod.Da. grshoppe grasshopper), or from MLG hoppe hop (the crop).
4 A Germanic ablauting n-stem
I suggest that Proto-Germanic possessed an n-stem *heb horse with

gen. *huppaz of the type with vocalism alternating between -e- and *-u-
described by Kroonen (2011). Although none of the items listed by
Kroonen constitute an exact parallel to such a paradigm, i.e. no item
with an alternation between nom. *-eD- ~ gen. *-uTT-6, one example of
*-eT- ~ *-uTT- does occur (*wek ~ *wukkaz wick) as does *-eD- ~ *-
uRTT- as well as numerous other similar types like *-euD- ~ -uTT- and
*-aD- ~ *-uTT-. The system thus leaves the possibility open of the exist-
ence of a subtype *-eD- ~ *-uTT.
If the Germanic form of the horse-word suggested by Simon (2005)
was actually *heb ~ *huppaz originating from *keb-n- ~ *k()b-n-s,
the -u- at first glance would have to have arisen from an analogical
Germanic use of *-u- in zero-grade surroundings because no sonorant
or *-u- was ever present to trigger the Gmc. -u- vocalism in the weak
stems on regular grounds (as in PIE *greb-on- ~ *grb-n-s > PGmc.
*kreb ~ kurppaz basket). However, there is also the possibility that the
ablaut of the iterative verbs was exported to the nominal level as
Kroonen (2011: 211-212) suggests for parallel cases like *dab- ~ *duppaz
puddle next to the verb *duppn-. The vowel *-u- became productive as
a zero-grade marker in Germanic iteratives even derived from strong
verbs (cf. Kroonen 2012: 193). And it seems reasonable to assume that
the iterative verb *hupp/bn-7 and its non-iterative counterpart *happn-
8
to move backwards, retreat (especially of horses) influenced the par-
adigm of *heb, especially if Simon is right that this word originally de-
noted the slow horse.
6
Kroonen (2011: 335-351) rejects an original paradigmatic alternation *kred-
(OHG chreta, MHG krete) ~ *kruttaz (OHG chrota, MLG krode) for toad, ex-
plaining the Upper German forms with <e> as renderings of umlauted forms
with -- corresponding to NHG Krte.
7
Cf. ON hopa fall back, Icel. hopa turn back, retreat, Far. hopa draw back,
recede, retreat, Nw. hop(p)a, hobba retreat, drive backwards (esp. of horses),
Da. dial. hoppe sig move backwards, drive backwards (Kroonen 2013: 257).
8
Cf. Nw. haba, habba retreat, drive backwards (esp. of horses), Sw. dial. habba
to turn back, drive backwards (Kroonen 2013: 257).
5 From Germanic to Balto-Fennic
If the Balto-Fennic words for horse are borrowed from such a Proto-
Germanic -n-stem, the vocalic and consonantal variation among Balto-
Fennic forms can be explained by the Germanic paradigmatic alterna-
tions. The Northern Balto-Fennic form *hepo and Est. dial. hebu are then
simply a rendering of the old Germanic nominative, while Southern Bal-
to-Fennic borrowed oblique forms with new zero-grade, perhaps with
the -n- of the stem still retained and reinterpreted as domestic derivatio-
nal suffixes:
*heb (loan) Fi. hepo, Est. hebu

>
*huppaz < *hup-na- (loan)Est. hobune /hopune/, hopene /hop:ene/ etc.
> Da. hoppe, Sw. Hoppa (Scand. OE hobin, ME hobyn, hobin,
Eng. hobby), LG dial. Huppe
Both forms were easily incorporatable into the Balto-Fennic system be-
cause BF already possessed the nominal suffixes *-o and *-(i)nen. It
happens quite often that originally borrowed strings come to be inter-
preted as native suffixes (cf. e.g. BF *hom-eh fungus < *ome Gmc.
*swambaz).
6 Concluding remarks
It cannot surprise us that yet another term for horse turns out to be a
culture-word as the invention of riding spread with human migrations,
expansions and trade. The exact directions of culture-words are notori-
ously difficult to trace, not least because they often involve transmission
via unattested languages, and their etymologies therefore often remain
disputed. From a purely Indo-European perspective, Simons proposal
might be considered as plausible as any other. It is his inclusion of bor-
rowings into non-Indo-European languages that suddenly made the
evidence for an inherited PIE word worth considering. Corresponding-
ly, in this article, a closer look at the Balto-Fennic material led us to re-
vise his Germanic etymology and ultimately the Indo-European one,
and consulting the latest research in Proto-Germanic morphophonolo-
gy made it possible for us to explain the otherwise enigmatic differences
between Balto-Fennic forms.
Simons observations already shed new light on the motivation of denot-

ing the horse the quick one: because if *keb- was the slow horse then
*h!uos is not simply the swift one understood as swift runners com-
pared to other animals, but rather the swift (kind of) horse, i.e. the kind
used for riding, as opposed to the slower working-horse.9
References
Blaek, Vclav, 2010: On horse in Slavic. Stephanie W. Jamison, H. Craig

Melchert & Brent Vine (eds.): Proceedings of the 21st Annual UCLA Indo-
European Conference. Bremen. Pp. 13-25.
Gob, Zbigniew, 1985: Slavic komon and kon' equus: An Attempt at Etymology
against the Background of the History of Domestication. Journal of Indo-
European Studies 13: 3-4: 415-443.
Huld, Martin, 2004: An Albanian reflex of Proto-Indo-European *E!uos
horse. Karlene Jones-Bley, Martin E. Huld, Angela Della Volpe & Miriam
Robbins Dexter (eds.): Proceedings of the Fifteenth Annual UCLA Indo-
European Conference. Washington D.C. Pp. 186-195.
Jrgensen, Bent, 1994 [1st ed. 1981-1983]. Stednavneordbog. Copenhagen.
Kortlandt, Frederik, 1997: Labials, Velars, and Labiovelars in Germanic.
NOWELE 30: 45-50.
Kroonen, Guus, 2010: The Proto-Germanic n-stems: A study in diachronic morpho-
phonology [= Leiden Studies in Indo-European 18]. Amsterdam.
Kroonen, Guus, 2012: Reflections on the o/zero-Ablaut in the Germanic Iterative
Verbs. H. Craig Melchert (ed.): The Indo-European Verb: Proceedings of the
Conference of the Society for Indo-European Studies, Los Angeles 13-15 September
2010. Wiesbaden: Reichert. Pp. 191-200.
Kroonen, Guus, 2013: Etymological Dictionary of Proto-Germanic. Leiden: Brill.
LGLOS = A.D. Kylstra, Sirkka-Liisa Hahmo, Tette Hofstra & Osmo Nikkil: Lex-
ikon der lteren Germanischen Lehnwrtern in den Ostseefinnischen Sprachen I-
III. Amsterdam / New York 1991-2012.
Liukkonen, Kari, 1999: Baltisches im Finnischen. Helsinki.
Simon, Zsolt, 2005: Die Etymologie von caballus. Gualtiero Calboli (ed.): Latina
Lingua! Proceedings of the Twelfth International Colloquium on Latin Linguistics
(Bologna, 9-14 June 2003). Roma. Pp. 405-416.
Snoj, Marko, 2003: Slovenski etimoloki slovar. Ljubljana.
de Vaan, Michiel, 2009: The derivational history of Greek and .
Journal of Indo-European Studies 37, 1/2: 198-213.
Yoshida, Yutaka, 2009: Sogdian. Gernot Windfuhr (ed.): The Iranian Lan-
guages. London / New York: Routledge. Pp. 279-335.
9
PIE *5opHo- hoof (PGmc. *hfaz), despite the similarity, appears unrelated to
*keb- even though ON hfr means both hoof and horse; this probably just re-
flects a pars pro toto semantic extension of the type ON horn horn, but also
ox.
Balto-Fennic Loanwords in Proto-Germanic1
Abstract
This article consists of four entries treating what is argued to be Proto-

Balto-Fennic (Middle and Late Proto-Fennic) loanwords into Proto-
Germanic. Many of the candidates for loanwords begin with h- this
feature makes them more easily detectable in cases where Balto-Fennic
h- have cognates elsewhere in Uralic beginning with - or -.
1 Four entries
1.1 PGM C . * HAMARA - HAMMER BF * HAMARA BACK OF AN

AXE
PGmc. *hamara- m. (> ON hamarr, OE hamor, hamer, OFris. hamer,

homer) has traditionally been viewed as inherited from PIE *h-mon-
or *-mon- (sharp) stone, yielding Slavic *kamy/*kamen- stone ,
Lith. amu and akmu, Gk. anvil, Skt. man- and Av. asman-
stone, sky (via vault). Just like PGmc. *wajju- wall as a derivative of
PIE *ue)H- to wind, plait tells us that there was once a time when walls
consisted of plaited branches, quite independently of any archaeological
evidence, we also think because of the standard etymology that we are
able to tell that hammers in Pre-Proto-Germanic times were made of
stone.
However, it is by no means certain that PGmc. *hamara- is really
cognate with the other forms mentioned. Slavic *kamy/*kamen- is mys-
teriously distorted. For those who reconstruct a laryngeal in the PIE
word it looks like a late metathesis: *h-mon- > PBSl. *Hk-men- >
1 This paper was presented at the symposion Germanic, Romance and Slavic in
the Early Middle Ages at the University of Leiden, 29 November 2012.
*kH-men. However, the vowel PGmc. *hamara- is short, which rules

out the metathetic explanation that works for Slavic. For those of us who
prefer an initial a-vowel, a metathesis does not even suffice because
Slavic -a- reflects a long vowel.
Furthermore, PGmc. *hamara- is a masculine thematic stem, and an-
imate *-men-stems are normally not heteroclitic. If it is Indo-European,
it could be either a secondary thematicization of an old *-mer/*-men
heteroclite, or the thematic suffix *-ero- added to a root ending in *-m-.
The existence of a neuter *-mr/-m(e)n- underlying the masculine
stone-word is possible but there seems to be no surviving reflexes of
such a form. Indic derivatives such as the adjective amar- stony can
just be interpreted as *-mn- + suffixal *-ro-.
ON hamarr meant: (1) hammer, (2) back of an axe, (3) crag, preci-
pice (rather than just stone); in compounds it could be used of rocks,
e.g. berghamarr rocky precipice, hamarrifa rift in a crag (Zoga 1910).
The latter meaning, however, is more likely figurative than primary.
Correspondingly, in the Da. place-name Hammer Odde, Hammer-
knuden (on Bornholm, 1539 Hammar), Hammer Bakker (1503 Hammer)
the name refers to the hammerhead-shaped crag of granite again, the
meaning hammer(head) is primary, its use of a rock is a figurative de-
scription of a steep rugged mass of rock projecting outwards and up-
wards. It mostly occurs in coastal areas and never seems to be used just
of rocks or stones in general that are not protruding or hammer-shaped.
This use is confirmed by a common noun hammer (steep) crag in
Danish dialects. There is thus no particular need to reconstruct the
meaning stone or rock for PGmc. *hamara-.2
The first and second meaning of the Old Norse word match perfectly
that of Balto-Fennic *hamara (> Fi. hamara back of an axe, Est. hamar,
hammar back of a knife). In my opinion, PGmc. *hamara- has not
been borrowed into Proto-Balto-Fennic a is otherwise assumed, but is
simply a borrowing the other way around3. The most important reason
is that the word must be inherited in Balto-Fennic from at least the Fen-
no-Volgaic stage. It can be connected to Saami *smr back/pole of an
2
Scholars in general seem to regard the meaning rock as primary in Germanic,
but that may simply be circular reasoning i.e. because they suppose from the
outset that it is connected to PIE *he5-men-.
3
Or at least into NW Germanic since no reflexes are known from Gothic. How-
ever, it must have entered the NW Germanic area already in the 2nd or 3rd c. AD
because it occurs in the personal name Chamarus in the Zlpich-Enzen dedica-
tion: Lat.-Gmc. Matronis M(arcus) Chamari f(ilius) et Allo To the mother god-
desses, Marcus, son of Chamarus, and Allo.
Balto-Fennic Loanwords in Proto-Germanic 101
axe, back of knife (> N Saami ibmar, Lule Saami sjimr, Skolt Saami
ammer) and Mordvin *uvV back of a knife (Erzya ov, ovone, Mok-
sha ov). The Saami forms at first glance appears irregular with its *im-
for expected sam-, but Western Saami *i- ( < Proto-Saami *s-) here
must actually be the regular outcome of Fenno-Ugric/-Permic/-Volgaic
*a- in the position before nasal (the only other example being *ama
apparition, shape mentioned below). While Uralic and Fenno-Volgaic
*- indeed normally yields Saami *s-, the actually attested forms, which
are comparatively few, reveal that when followed by the vowel -a- the
sequence yields Saami *i- before *-m- (probably: any nasal, but there
are no examples with *-n- and *--); in fact there seems to be only one
alternative surrounding attested which is *su- before a labial stop (N
Saami suhpi asp ~ Fi. haapa). When followed by other vowels, both
these vowels and *- behave as expected (e.g. N Saami savvi- heal a
wound < FV *e and N Saami soarvi dead pine-tree < PU *orwa- to
dry (out)). *a- in front of a nasal is not affected either, cf. N Saami
uoi membrane, fleshy fibres on the inner side of the skin (*-m- regu-
larly goes to via *-w- in these surroundings, cf. muomi < Pre-PGmc.
*mams-ma-; see elsewhere in this publication)4.
There is a way to get around it all: The Fenno-Volgaic form *amara
might, in turn, originally have been borrowed from some stage of Balto-
Slavic. This would have been a language with regular satem reflex but
the same metathesis or ablaut form as Slavic kamy, Slovincian kamor.
Thus, a possibility is left open that PGmc. *hamara- could be connected
to *a5-men-after all, however not as a direct reflex of its protoform.
1.2 PG MC . * HAMA - SHAPE BF * HAAMO SHAPE
PGmc. *hama- or *haman- shape, physical form (> ON hamr skin of

animal; guardian spirit, tutelary spirit, Da., Nw. ham skin of animal;
apparition, ghost, OS hamo covering, OE hama, homa id., OHG ha-
mo skin; covering; fishing-net)5 can likewise be a borrowing from Pro-
4
Initial *- in Samic is basically regarded as a loanword phoneme, consistently
indicating loans from Balto-Fennic, Baltic and Late Scandinavian. The inherited
postalveolar fricatives turned into dental-alveolars in Proto-Saami (PU *, *
PS *c, *s). The conditioning presented here might indicate that it rather de-
serves to be characterized as a marginal phoneme, most common in loan-
words but also occurring elsewhere.
5
Kroonen (2013) alongside the n-stem adds a ja-stem *hamja- on the basis of
Mod.Icel. hem and Elfdalian iem fish skin. Bjorvand & Lindeman (2000: 342)
to-Balto-Fennic. For BF we must reconstruct *haamo shape; apparition

on the basis of Fi. hahmo, haamu apparition, facial characteristics,
looks; ghost; shadow, Karelian hoamu shape, contour; apparation,
ghost, Votic mo id. The medial -h- in standard Fi. hahmo is second-
ary as indicated by the shape of its cognates. The medial -h- in Finnish
can either haven arisen by influence of the initial /h/, as also in huhmar
~ huumar mortar; via metathesis from a derivative *haamo-h; or less
likely, by transmission via an extinct Sami language, as in kahlata ~
kaalata to wade Proto-Saami *kl- < PU *kl-).
This form, in turn, goes back to Fenno-Volgaic *ama which also
yields Lule Saami sjipm similarity and Mordvin M ama, E ama
face. The Saami form at first glance appears irregular with its *sjim- for
expected suom-, but as mentioned before, Saami *i- here must actually
be the regular Saami outcome of Fenno-Ugric/-Permic/-Volgaic *a- in
the position before nasal (the only other example being *ama-ra men-
tioned above).
The aforementioned uoi membrane, fleshy fibres on the inner side
of the skin is a reflex of a similar Fenno-Ugric root which exhibits al-
most the same semantics as *ama: FU *amV skin (of animal), mem-
brane is reconstructed on the basis of forms from all Saami languages
compared with Khanty (Ostyak) un skin, hide. The reconstruction of
Proto-Saami *-m- in this root is based exclusively on Kildin Sami
tsmts, while all other Saami languages point to a protoform *cnc.
However, that a reflex of a sequence *-mD- (i.e. -m- + dental) is pre-
served only in a single daughter-language and assimilated everywhere
else is typologically common because (also sporadic) dental assimilation
of -m- before a dental is in itself very common6. Crucially, a reconstruc-
tion *cnc would leave the Kildin form unexplained, while *cmc ac-
counts for all of them (by preservation, in the case of Kildin Saami, or
assimilation). Therefore I will reconstruct Proto-Saami *cmc. It should
be noted that even if the Proto-Saami form was *cnc (which I find un-
likely), this form, in turn, could of course just as easily as the individual
Saami forms be the product of sporadic assimilation of an original clus-
ter with *-m-.
reconstruct a PGmc. *hami-, which could formthe basis of *hamja-, next to

*haman-, while Orel (2003: 158) only sees a thematic stem *hama- m.
6
The situation is parallel to Indo-European where usually only Baltic preserves *-
mD- whereas all other language groups assimilate (not necessarily by a rule-
governed development). Note also the assimilation of Uralic *-mt- > Fenno-
Ugric *-nt- in the ordinal suffix (and in lexemes as well).
The Mordvin forms might just as well < *ama. Since Balto-Fennic
*h- is the result of PU *-, too, we might as well assert *ama as the
source. Both *ama and *amV are reconstructed for stages later than
Proto-Uralic, and their variation may be due to their identity as Eura-
sian culture-words, cf. e.g. Written Mongolian ama shirt ~ Persian
jma apparel.
1.3 PG MC . * HALBA - HALF BF * HALPA , HALA - CHEAP , RE-

DUCED
Among the classical, more or less successful, attempts to etymologize

the PGmc. adjective *halba- half is the reconstruction of a derivative
*ol-bo- with the adjectival suffix *-bo- added to the root of Lith. als
side; strip of land; coast; direction7. The original meaning would then
have been belonging to one side, having to do with one side, and such
a development finds a semantic parallel in Slavic (OCS pol means both
side and (a) half. The connection with als is, however, uncertain; it is
not favored by Orel (2003 : 154) who labels it difficult word, rejected
outright by Bjorvand & Lindeman (2000: 341-342)8 and not even men-
tioned as a possibility in Kroonen (2013: 204). Besides, even if the con-
nection should be correct, there are no obvious candidates for cognates
outside the Northern European branches. No alternative etymologies
are generally accepted. Orel assumes that it is in some way related to
Skt. klpate succeed; fit; be partaken by, klpa crossbeam. This is re-
jected by Kroonen as semantically uncompelling; he simply states that
the word has no clear etymology. Bjorvand & Lindeman likewise give up
on most old proposals, vouching that the only possible solution is to
operate with a derivative of the PIE root *skel- to split, share (proba-
bly *skelH-) without s-mobile, an idea going back to Uhlenbeck (1905:
287). A serious issue is then that the root is unattested with s mobile
all attested forms have *s- in the root.
7
In Hyllested (2009: 204) I adhered to this etymology which I now reject. It goes
back to at least Brugmann Grdr. (II, 1: 388-389) although indirectly ascribed to
von Grienberger (1900: 107-108) by Orel (2003: 154) who gives a summary of
the classical proposals.
8
I do not quite understand that their rejection seems to be based on the differ-
ence in semantics; cf. that the derived substantive *halbn- f. means side
(Ofris. halve, OS, OHG halba) and region, quarter (ON hlfr), to a great extent
overlapping with the Lithuanian semantics.
It is important to take into account the semantic scope of the word

which can be inferred from its use in the attested languages. PGmc.
*halba- seems to have three (related) meanings: 1. incomplete, partial,
reduced; 2. a significant portion of; 3. exactly . There is no particu-
lar reason to think that the latter meaning, historically speaking, is the
(only) primary one numerals and terms related to measurement most
often originate from words with broader meanings.
For Proto-Balto-Fennic a word *halpa-, gen. *halan- can be recon-
structed with the meaning reduced, cheap of prices. The reflexes in the
individual Balto-Fennic languages are Fi. halpa, gen. halvan; Est. halb,
gen. halva bad, evil; poor (e.g. of prospects, arrangement); ill; low (of
spirits), SW and SE dial. also cheap (favorable of prices)9; Votic alpa
bad (Adler & Leppik 1990: I, 99; attested only in folk tales). From
Northern Balto-Fennic it has been borrowed into several of the Saami
languages, e.g. N Saami hlbi cheap of price and goods, Lule Saami
(h)alp- become cheaper (of goods).
Koivulehto (1982; via Hofstra 1985: 162) has proposed that it is a bor-
rowing from Proto-Germanic into Fenno-Saami (= Early Proto-Fennic),
namely from the word *salwa- filthy, dirty (> ON slr dark grey)
which would be manifested in Fenno-Saami as *alwa-, later yielding
Late Proto-Fennic (= Balto-Fennic) *halpa- ~ *halva- where *halpa is
one example of -h- corresponding to PGmc. *-s- in the oldest loans. Sa-
mmallahti (1998: 247-248) adds this etymology with a question mark.
However, Balto-Fennic *halba- cannot be a loan from Germanic as it
is cognate with Mari (Cheremis) Nw W ul-, E Ki U ul-o cheap (cf.
UEW 782)10 and thus come from a Fenno-Volgaic *alV- with the adjec-
9
Cf. further Est. halvustama to hold cheap, belittle. Salaca Livonian alu cheap,
favorable (of prices) has been borrowed from from SW Estonian alv (Kettunen
1938: 9).
10
Proto-Mari *-u- is not the most common outcome of Proto-Fenno-Volgaic (or
Proto-Uralic) *-a-, which is normally -a- ~ -o- depending on the dialect. UEW
(782) mentions, however, that there are more examples of Mari *-u- < Fenno-
Volgaic *a. We can at least find examples of Balto-Fennic *-a- corresponding to
Mari *-u-, one of which is mentioned by Bereczki (1988: 338), namely Fi. tammi
oak ~ Mari O tumo W tum; but the question is which language group is con-
servative in this respect since Balto-Fennic seems to have many secondary as
which might be due to a yet undetected conditioning. In other words, the cor-
rect Fenno-Volgaic reconstruction might in fact be *ulV. It is also conceivable
that the Mari word was contaminated with ula-(la-) to cut a little bit (cf.
ula-las one cut across, one single cut), a derivative of ua-, ula- to cut
which, however, itself may derive from a FU root with -a-, *ale- to cut, split
(pace UEW 459-460). Most important for our purposes here is that the Balto-
tival and participial suffix -pa added to it. Accepting Koivulehtos ety-
mology implies a rejection of the Mari cognates and the reconstruction
of a Fenno-Volgaic term.
I therefore propose instead that the word was borrowed from Proto-
Balto-Fennic into Proto-Germanic. Note that *halpa-, even including its
derived verb (Fi. halventa make cheaper) is already known to have
been borrowed into Saami at a later stage. It is also important that there
are no plausible cognates on the Indo-European side. Even if Mari cog-
nates for *halpa are not accepted, this would not hinder a Finnic
Germanic etymology.
Germanic must then have borrowed the oblique stem with weak
grade (*-- before a closed syllable in the paradigm, gen. *halan) which
explains the Germanic voiced fricative; the reason cannot be Verners
Law since the word would have had initial stress all along11. The Fennic
word would have been borrowed first with the primary meaning 1. in-
complete, partial, reduced, while the originally secondary meanings 2.
a significant portion of; 3. exactly (which of course can overlap
completely in practice, but do not have to) would be semantic speciali-
zations within Germanic, eventually replacing the original PIE word for
half, *smi-, *s-tero-. The motivation for the borrowing is clear: re-
duced of prizes and cheeper of goods would surely have been an im-
portant concept in trading contexts.
1.4 PG MC . * PUNGA - PURSE BF * PUNKA - CHEAP , REDUCED
Turning to words with a different initial consonant, another mystery in

Germanic etymology is *punga- purse, pouch, small bag (> Goth. puggs
/pungs/, ON pungr, OE pun, ODu. pung, pong, MLG punge id. [>
Scandinavian], OHG scaz-pfung money-bag). With its initial *p-, it
looks more than anything like a loanword. Most proposals seem to have
just postponed the problem by suggesting as a source languages where it
it likewise has the air of a foreign origin (cf. Orel 2003: 293) In fact,
MLat. punga, Rumanian pung and MGk. are most likely to be
Gothic loanwords (Bjorvand & Lindeman 2000: 707), and then we are
back to square one. Neither they nor Orel (2003: 293) or Kroonen (2013)
Fennic and Mari forms can belong together and be traced back to a common
form which is older than Proto-Germanic.
11
Thus, Gmc. *-b- next to *h- here cannot be used as evidence for the Germanic
sound shift having preceding Verners law.
offer any solutions except that the two latter surmise that there is some
kind of relationship with *pukn-. Bjorvand & Lindeman laconically
state that it may very well be an old Wanderwort, but obviously
wanderwrter have origins in specific languages, too, although they are
harder to trace by definition. There is no reason a priori why some of
the European wanderwrter would not turn out to have an origin in
Uralic or even be pure loanwords, borrowed in a bilateral process be-
tween a Uralic language and the target language.
Balto-Fennic had a word *punka which goes back to Proto-Uralic
*puka (or *poka) swollen or expanded object (UEW 404, SSA 427)
and has cognates in Saami, Permian and Ugrian languages. It is reflect-
ed in Fi. punka Est. pung something chubby or protruding, clod, bump,
swelling; leaf bud; bag, purse. While the latter meaning in Estonian may
well be due to Middle Low German or even later Swedish influence
(Mgiste 1982-1983 [2000] VII: 2230-2231), the semantic development
from something swollen via bag to purse is straightforward, cf.
Welsh balleg purse < PIE *bl-no- from *bel- swell. Besides, Est. also
has a variant pong, corresponding to Fi.dial. ponki, ponka, ponko, a vari-
ation in line with the one found in rest of Uralic.
I suggested already in Hyllested (2008: 136) that PGmc. *punga- is a
borrowing from Balto-Fennic, and, more tentatively, that the forms that
look as if they go back *puh/kkn- (ON poki bundle, purse, OE pocca,
pohha id., Orkney Norn buggy belly, MDu. poke bag for wool, NHG
Pfoch bag) may in fact be borrowings from a Western Saami precursor
of Mod. N Saami buggi bump, lump, hump; swollen or expanded ob-
ject and/or boggi short fat one (person, animal or thing). The medieval
assimilation of nasal plus stop (*-NT- to *-TT-) in Western Saami serves
as terminus post quem since this development took place after 1000 AD
and spread to Central Saami not until the late 1500s (Sammallahti 1998:
29, Hyllested 2008: 134). The transmission can have taken place in sev-
eral steps, most likely with Nordic as the middleman. Note that the
words bag (< ME bagge) and pack itself is already a borrowing from
Western Saami *pakke- (> N Saami baggi) via Old Norse baggi
(Hyllested 2008: 136) package, bundle; plump animal. Medieval Saami
and Balto-Fennic material in Norse and West Germanic relate typically
to hunting and fishing, the fur trade, the production of hide and down,
and trade concepts in general. Terms for purse or bag (made of hide)
with emphasis on either meaning obviously fit in this picture.
References
Adler, Edna & Merle Leppik, 1990: Vadja keele snaraamat. Tallinn: Keele ja
Kirjanduse Instituut.
Bereczki, Gbor, 1988: Geschichte der Wolgafinnischen Sprachen. Denis Sinor
(ed.): The Uralic Languages. Description, History, and Foreign Influences. Leiden
/ New York / Copenhagen / Cologne: E.J. Brill. Pp. 314-350.
Bjorvand, Harald & Fredrik Otto Lindeman, 2000: Vre arveord. Oslo: Novus.
von Grienberger, Theodor, 1900: Untersuchungen zur gotischen Wortkunde. Vienna:
Gerold.
Hofstra, Tette, 1985: Ostseefinnisch und Germanisch. Frhe Lehnbeziehungen im
nrdlichen Ostseeraum im Lichte der Forschung seit 1961. Groningen: Rijksuni-
versiteit Groningen.
Hyllested, Adam, 2008: Saami Loanwords in Old Norse. Hans Frede Nielsen
(ed.), Early and Prehistoric Language Development in North-Western Europe [=
NOWELE 55/56]. Odense: University of Southern Denmark Press. Pp. 131-146.
Hyllested, Adam, 2009: PIE *-b- in Nouns and Verbs: Distribution, Function,
Origin. Rosemarie Lhr & Sabine Ziegler (eds.): Protolanguage and Prehisto-
ry. Akten der XII. Fachtagung der Indogermanischen Gesellschaft, vom 11. bis 15.
Oktober 2004 in Krakau. Wiesbaden: Reichert. Pp. 202-214.
Kettunen 1938: Livisches Wrterbuch [= V]. Helsinki: Suomalais-Ugrilainen Seura.
Koivulehto, Jorma, 1982: Zur Periodisierung germanischer Lehnwrter im Ost-
seefinnischen; wie findet man germanische Lehnwrter und was sagen sie uns?
Lecture given at Rijksuniversiteit Groningen 12 and 13 May, 1982.
Kroonen, Guus, 2013: An Etymological Dictionary of Proto-Germanic. Leiden: Brill.
Mgiste, Julius, 1982-1983 [2000]: Estnisches etymologisches Wrterbuch. Helsinki:
Suomalais-Ugrilainen Seura. 2nd ed. 2000.
Sammallahti, Pekka, 1998: The Saami Languages: An Introduction. Karasjok: Davvi
Girji.
SSA = Ulla-Maija Kulonen & al. (eds.): Suomen Sanojen Alkuper. Etymologinen
Sanakirja. I-III. Helsinki: Suomalaisen Kirjallisuuden Seura and Kotimaisten
Kielten Tutkimuskeskus. 1995.
Uhlenbeck, Christianus Cornelis, 1905: Bemerkungen zum gotischen Wortschatz.
Beitrge zur Geschichte der deutschen Sprache und Literatur 30: 252-327.
Zoga, Geir T., 1910: A Concise dictionary of Old Icelandic. Oxford: Clarendon.
Gothic mammo meat
in the Light of Saami Evidence
Abstract
The Gothic hapax *mammo [f.] meat reflects Pre-PGmc. *mamz-mn

flesh, which goes back to PIE *moms-mo-, a body-part derivative of
*mems- meat. Decisive evidence for this exact formation is delivered by
North Saami muomi meat between the thigh and ribs (of a reindeer)
which is not a Palaeo-Laplandic substratum word, but a regular substi-
tution of the Pre-Proto-Germanic word.
1 Gothic mammo attestation and earlier proposals
Gothic *mammo [f.] meat is a hapax legomenon occurring in the

gen.sg. mammons from the Epistle of Paul to the Colossians, 1, 221:
in leika mammons is airh dauu du atsatjan izwis weihans jah un-

wammans jah usfairinans faura imma
In the body of his flesh through death, to present you holy and un-
blameable and unproveable in his sight
1
The Gothic translation of the relevant passage is handed down to us via Codex
Ambrosianus A and Codex Ambrosianus B, both of which are now kept in Bib-
lioteca Ambrosiana in Milan.
Uhlenbeck (1905) originally suggested that the word had arisen as pure
nursery language via the meaning mothers breast like Lat. mamma
breast, but a year later (cf. Uhlenbeck 1906) he had abandoned that
viewpoint. Nonetheless it was maintained by Scardigli many decades
later (1973: 72).
Today most scholars regard mammo as an o-grade formation based
on PIE *mms-- (? *mems--) > PGmc. *mimza- [n.] > Goth. mimz
meat. This was first suggested by Mikkola (1897) who at the same time
left the possibility open that it could be derived from a PGmc. *maz-
mn based on the root in OHG muos food. His etymology was admit-
tedly received with some hesitation by the standard etymological dic-
tionaries of Gothic, e.g. Lehmann (1984: 243) who earnestly mention the
possibility without directly endorsing it. Lehmann tags the label dis-
puted on it, and Casaretto (2004) loyally maintains that the etymology
is unclear.
Things did not start to develop until Polom (1967). Kroonen (2013)
express similar thoughts in his new dictionary, stating that mammo can
reflect either
a) Pre-PGmc. mamzn- < *moms-- with an uncertain assimilation

of *-mz- > *-mm-, or
b) Pre-PGmc. *mamz-mn < *moms-m- (*moms-mn-) involving
the well-known assimilatory development *-zm- > *-mm-.
He notes that the same problem is relevant for PGmc. *kramma- moist,
humidity (which correspondingly can come from either *kramzn-
eller *kramzmn-), embracing Poloms viewpoint that *-mzm- is the
more probable reconstruction on phonological criteria because it is un-
certain whether an assimilation of *-mz- into *-mm- took place at all;
the most obvious counterexample is PGmc. *mimza- itself2.
2
On this basis, the reconstruction has been characterized as problematic by
Grienberger (1900: 154), Trautmann (1906: 62), Falk/Torp (1909: 310), and Cas-
aretto (2004: 235), but although the sequence must be morphologically seg-
mented *-mz-m-, it of course forms a phonological string *-mzm-, and while it
is true that *-z- generally is not assimilated to a preceding *-m- in Gothic or
Germanic in general, it does assimilate to a following *-m-. Thus, reconstruct-
ing mammo as *mamz-ma- only appears problematic if the formation is sup-
posed to have taken place in Gothic itself (when the assimilation process was no
longer operating); but there is no need to think that it did not happen in Ger-
manic or even PIE already.
Gothic mammo meat in the Light of Saami Evidence 111
From a morphological perspective, however, *mamz-mn is more

difficult to understand: It is the only attestation of *mems- ~ *mms-
with an o-grade anywhere in Indo-European. And one wonders what
the motivation would have been for forming a new *-mo-/*-men-stem
from a nominal (pseudo-)root which already contains two labial nasals.
2 North Saami muomi and its implications
Turning to neighboring languages, new evidence has appeared. The

North Saami word muomi means meat between the thigh and ribs (on
a reindeer)3 and has no generally accepted etymology. Aikio (2012) on
the basis of its phonological structure deems it as a loanword. Moreover,
he includes it in a long list of reindeer terminology inherited from the
original language of the present-day Saami populations, called Palaeo-
Laplandic (he explicitly avoids Saami here in order to distinguish it
from modern Saami language and culture). It is generally acknowledged
that the Saami populations were not speakers of Uralic from the outset
but at some point took over a Uralic language, and that an abundance of
terms from the old sunstratum language survived, accounting for hun-
dreds of words which cannot inherited nor borrowed from any known
language group. Aikio is the forerunner in the identification and analy-
sis of this substratum; another important work in this field is his essay
from 2002 where muomi however is not mentioned.
It has been known for much longer that Saami languages, like Balto-
Fennic, have borrowed hundreds of words from Germanic through dif-
ferent periods in history, and that there are also Pre-Proto-Germanic
loanwords, i.e. items whose present-day Saami shapes reveal that they
were borrowed before the sound changes that define Proto-Germanic
had all taken place. While it is beyond doubt that a large part of the
Saami lexical stock can be ascribed to one or more non-Uralic and non-
Indo-European substrate languages, we should therefore not be sur-
prised if at least a small part of the proposed substratum words turn out
to be Germanic or Pre-Germanic after all.
The Proto-Saami form of muomi would mechanically be recon-
structed as *masmV-. However, *mamsmV- is in fact also possible: The
normal development of Proto-Saami *-amC- is undoubtedly North
3
Nielsen & Nesheim (1932-1965, bd. II 1934 by Nielsen only: 702) gives the more
elaborate definition: layer of fat and muscles between the thigh and ribs of
reindeer and other animals (goes with the side when the carcase is cut up).
Saami -uovC-, cf. ruovda border, edge Pre-PGmc. *rama- (*rama-

) > PGmc. *rana-. However, the sequence -uov- is apparently phono-
tactically illicit before *-- (with or without a consonant following),
meaning there are simply no words in contemporary North Saami con-
taining the structure -uov-. I could find none in the extensive Nielsen
& Nesheim (1932-1962).
Although *muovm- could very well have existed at some stage, Pro-
to-Saami *mamsmV- would regularly yield the North Saami end result
muomi which is also what we find. It therefore simply seems to be a
Pre-Proto-Germanic loan from the very same word that yielded Gothic
mammo. Semantically there are already no problems, but even the spe-
cific Saami meaning may reveal something about the history of the for-
mation.
3 Motivation for formation in *-mo-
One might argue that even if Pre-PGmc. *mamz-mn- or PIE *moms-

mo- was formed on the basis of *memzan- or *mems- respectively, the
motivation for using exactly the *-mo- or *-men- suffix could still be
based on the iconic nature of reduplicative semantics. Danish nursery
terms like mam(-mam) and mamse food, although they clearly look as
if they are formed like nusery words in any other language, are per-
ceived by speakers as derived from the adult term mad, which may even
partly be historically true (or impossible to determine). In other words,
the use of specific morphological elements in word-formation, although
formally allowed, can still be motivated by conditions outside the mor-
phological system such as association to childrens language or even
just a general human delight of playing with sounds and fascination of
phenomena such as alliteration4.
4
One might also speculate that Wulfila associated *mammo with *mammona
mammon, wealth (a masc. n-stem, only attested in the dat.sg. mammonin) in a
religious context where the concept of meat could be contrasted with spiritual
properties and values (cf. the hunger for wealth). However, it is hard to imag-
ine such an association among the heathen speakers of Pre-Proto-Germanic
who would never have heard of any such Greek-Aramaeic word, let alone the
biblical concept it denoted. My guess is that if Wulfila felt an association be-
tween the two, the similarity of mammn- f. to mammnan- m. was simply a
convenient coincidence of the kind that translators sometimes stumble upon
and are happy to exploit for literal and pedagogical purposes (to the extent that
they have a choice at all).
Still, in this case there is in fact a possibility that we might be dealing

with word-formation based on a very clear semantical motivation. Quite
a few Indo-European names for body parts occur with a suffix *-mo- or
*-men-; although a few of them can be shown to have originally meant
something else, e.g. substantivizations of adjectives (PCelt. *skama-
lung, but also the adj. light), they cannot all be explained in this way.
Even when the term looks like productive deverbal nominal formation
(such as PGmc. *barma- bosom < PIE *ber- to carry) one may won-
der why exactly *-mo- was used in these cases. I think that there are
enough examples for us to at least claim that its use was still in (margin-
al) use in PIE, and that it reflects an earlier grammatical marking of in-
alienable possession, using the 1st person verbal and pronominal marker
*-m-:
PIE *gehu-mo- gum (or *gau-mo-; deverb. from *? *geh-

gape)
PIE *pleu-mn lung, Skt. klman-
PIr. *a-man- eye
Gk. , Ru. derm skin
Gk. eye
Gk. eye
Gk. , OIr cnim lower leg, PCelt. *kn-mi- bone
Lat. abdmen belly, abdomen
Lat. rmen (~ Skt. romantha- chewing the cud)
PCelt. *skama- lung (< adj. *skama- light)
Alb. zemr heart (< *-men; root?)
PGmc. *armaz arm~ Lat. armus upper arm (< PIE deverb. from *ar- link)
PGmc. *barma- bosom (< PIE deverb. From *ber- to carry)
PGmc. *arma- gut, intestine (< PIE deverb. drill (?))
PGmc. *famaz bosom, lap (OHG fadum, OE fm; ON famr spread arms,
embrace; thread) ~ PCelt. *atim-, *atam- yarn, thread (> Gael.
aitheamh, W edafedd), cf. Hamp 2008, Hyllested 2010
PGmc. *skama- shame; nakedness < genitals, naked skin, flain skin, hairless
skin (< *sk-mo- from *sekh- cut, cf. sex, Latin secre)
PGmc. *skerma-, *skermi- skin, hide (< *sker- cut (off)
ON hvarmr eye lid, Far., Icel. hvarmur, Nw. Kvarm, Elfd. ogen-walm via Swe-
dish < PGmc. *hwal(b)-maz
German Kieme gill (? ~ OHG kiuwa, kewa, kouwa)
This situation is parallel to Uralic where I believe to have uncovered a

specific use of the denominal suffix for specifically for body-parts. The
suffix seems to have stayed productive in several branches at least it
must have lived longer than in Indo-European, judging by the impres-

sive number of examples:
Proto-Uralic (attested in both Fenno-Ugric and Samoyed):

*ilm eye
*em kidney
*-mV heart
*pisV-mV, *pi(p)se-mV lip
*umV waist
Proto-Saami *kjm marrow, brain ~ Samoyed: Selkup km, Nenets xwa
marrow, Nganasan kojmu id. (~ Selkup k marrow, brain)
Proto-Fenno-Ugric (attested in both Fenno-Permian and Ugric)

*ake-mV tongue, palate (PU ake tongue, palate)
*lm tongue
*VjmV loin
*wiV-mV marrow; brain
*arma groin (formed from *arV hairless skin, FU)
*kelm, *kem skin (formed from *kee skin)
*kulma corner of the eye; eyelid; temple
Proto-Fenno-Permian (attested in both Fenno-Volgaic and Permian)

*kurmV fist; handful
Proto-Fenno-Volgaic (attested in both Fenno-Saami and Mari-Mordvin)

*ilV-mV canine tooth
*sorme finger
*wajmV heart
*pijra crop (of animals) has *-mV in Saami and Mordvin
Individual branches / languages

PU *piV fist has *-mV in Koibal Samoyed pam, Mator ojme
PU *twe lung has *-mV-ktV in Selkup tyymkt, tyymiekto
PU *kne nail has *-mV in Mator Samoyed kadam
PU *konV + 2nd element arm pit has *-mV in Ostyak xonm pat
PU *tkt bone(s) has *-mV in Hung. tetem
PU *kunV abdomen has *-mV in Komi kynm

PU *polwe knee has *-mV in Mordvin pulma-, puma-
PU *kVkV Adams apple has *-mV- + -s- in N Saami guoggom(as)
PU *aje (*ajwe) brain has *-mV in Lule Saami vuojam < PS *vuojm
PFP *VlV small finger has *-mV in West Saami *VlV-V-mV (the base with
*-V- is common Saami)
PU *ikene gums, gills has -ne replaced by -me in Fi.dial., ikemet, Kar. igimet,
Vot iemet, Est. igemed, igem, igim, ikem, ikim, Liv. igmd (in Est. and
Liv., only forms with -m- occur; nom. -n can come from both consonants)
Estonian habe(me) beard

Finnish siera(-im-(inen)) nostril
Finnish helma lap, bosom
Votic sams inner parts of animals body
4 Conclusion
Although Goth. mammo is normally translated as meat, our knowledge

of it is based on a single attestation where it clearly refers to the human
flesh. And N Saami means not meat as a kind of prepared food, but
exactly a kind of flesh, i.e. meat as a certain part of the animals body.
Not all languages have the same distinction as English, but we may
conveniently use English to illustrate a word-formation scenario where
*-mo- transformed the meaning from meat in general into flesh spe-
cifically:
PIE *mems- meat

+ body-part term suffix *-mo-

PIE *moms-mo- (with body-part term suff. *-mo-) flesh5
Since the suffix does not appear to have survived in any of the Indo-
European subgroups, I prefer to assume that the formation took place in
PIE itself. However, most examples of Indo-European body-part terms
with *-mo- are Germanic so we cannot exclude that it was still produc-
tive at least at the Pre-Proto-Germanic stage where *mamz-mn- would
have been formed (perhaps first as *mamz-m). This word was bor-
rowed into Proto-Saami as *mamsmV and developed ultimately into
North Saami muomi (perhaps via *muovmi) but disappeared from the
other Saami subgroups.
It perhaps appears surprising that a word belonging to reindeer ter-
minology was borrowed from (Pre-)Proto-Germanic into Proto-Saami.
Judging by the number of reindeer terms that survived from Palaeo-
Laplandic into Saami, one might at first glance conclude that reindeer
breeding was important already in Palaeo-Laplandic times. Domestica-
5
The transition of a simple thematic feminine *-eh into the productive weak
class of feminines in *-n in Germanic is of course well known, but has not yet
been assigned a satisfactory explanation. Schmidt (1985) mentions the possibil-
ity that the gen.pl. *-n having arisen could have provided a motivation for
analogy; see also Thny 2013: 119.
tion of reindeer may have taken place in the transition period between
the North European Bronze Age and the Iron Age (1200-600 BC).
However, it is worth noticing that most Palaeo-Laplandic terms do not
specifically refer to breeding; it could just as well reflect a way of life
characterized by reindeer hunting.
References
Aikio, Ante, 2004: An essay on substrate studies and the origin of Saami. Irma
Hyvrinen, Petri Kallio & Jarmo Korhonen (eds.): Etymologie, Entlehnungen
und Entwicklungen: Festschrift fr Jorma Koivulehto zum 70. Geburtstag [= M-
moires de la Socit Nophilologique de Helsinki LXIII]. Helsinki: Socit
Nophilologique. Pp. 5-34.
Aikio 2012 = Luobbal Smmol Smmol nte (Ante Aikio): An essay on Saami
ethnolinguistic prehistory. Riho Grnthal & Petri Kallio (eds.) A linguistic
map of prehistoric Northern Europe [= Mmoires de la Socit Finno-Ougrienne
266]. Helsinki: Suomalais-Ugrilainen Seura 2012. Pp. 441451.
Casaretto, Antje, 2004: Nominale Wortbildung der gotischen Sprache. Heidelberg:
Winter.
Feist, Sigmund, 1939: Vergleichende Wrterbuch der gotischen Sprache. 3rd ed. Lei-
den: Brill.
von Grienberger, Theodor, 1900: Untersuchungen zur gotischen Wortkunde. Vienna:
Gerold.
Hamp, Eric, 2008: Germanic *famaz and Gravity in the North. Early and Pre-
historic Language Development in North-Western Europe (NOWELE 54/55),
34952.
Hyllested, Adam, 2010: The Precursors of Celtic and Germanic. Stephanie W.
Jamison, H. Craig Melchert & Brent Vine (eds.) Review of Grnthal & Kallio.
Kratylos. Pp. 63-117.
Kroonen, Guus, 2013: An Etymological Dictionary of Proto-Germanic. Leiden: Brill
Lehmann, Winfred P., 1986: A Gothic Etymological Dictionary. Based on the 3rd ed.
Of Vergleichendes Wrterbuch der Gotischen Sprache bei Sigmund Feist. Lei-
den: Brill.
Mikkola, Joos J., 1897: Baltische Etymologien. Beitrge zur Kunde der indoger-
manischen Sprachen 22: 239-254.
Nielsen, Konrad & Asbjrn Nesheim, 1932-1962 (bd. I-IV by Konrad Nielsen only):
Lapp dictionary based on the dialects of Polmak, Karasjok and Kautokeino /
Lappisk ordbok grunnet p dialektene i Polmak, Karasjok og Kautokeino [=
Instituttet for Sammenlignende Kulturforskning, Serie B: Skrifter XVII]. Oslo:
Aschehoug & co. (W. Nygaard).I-VI.
Polom, Edgar, 1967: Notes on the Reflexes of IE /ms/ in Germanic. Revue Belge
de philologie et histoire 45, 3: 800-826
Scardigli, Piergiuseppe, 1964: Lingua e storia dei Goti. Florence: Sansoni.
Schmidt, Gernot, 1985: Der Genetiv der indogermanischen -Deklination.

Hermann M. lberg & Gernot Schmidt (eds.): Sprachwissenschaftliche For-
schungen. Festschrift fr Johann Knobloch. Innsbruck. Pp. 393402.
Thny, Luzius, 2013: Flexionsklassenbertritte.Innsbruck: IBS.
Trautmann, Reinhold, 1906: Germanische Lautgesetze in ihrem Sprachtgeschicht-
lichen Verhltnis. Kirchhain.
Uhlenbeck, Christianus Cornelis, 1905: Bemerkungen zum gotischen Wortschatz.
Beitrge zur Geschichte der deutschen Sprache und Literatur 30: 252-327.
Uhlenbeck, Christianus Cornelis, 1906: Aantekeningen bij gotische etymologien.
Tijdschrift voor Nederlandse Taal- en Letterkunde 25: 243-306.
The Mysterious Elder:
Common Traits in European Names for
Sambucus nigra and Viburnum opulus
Abstract
No PIE names for the elder species (Sambucus) or the smaller water el-
der (Viburnum opulus) can be reconstructed although they are indige-
nous in all of the possible homeland areas. However, Lat. sambcus can
be traced back to a meaning the dusty tree, referring to powdery mil-
dew on canes, leaves and berries which gives the tree an overall dusty
impression. This meaning also lies behind Lith. eiv-medis (cf. vas
grey; mildew), a relic of the old Balto-Slavic word which was borrowed
into Fenno-Permian as *ewa. In common Slavic it was replaced by the
loan bz, buz < Old Turkic boz, buz grey (cf. Ru. bsel mildew)
while in the NW Slavic fringes the old name only still survived at least
until the 20th c. in the amalgamated and folk-etymologically reshaped
Sorbian form diwi bz and as a loanword in neighboring German dia-
lects (Schibchen, Ziwecken). Berries of the water elder are called *ar in
Mari (Cheremis) which can be traced back to a loan from (East) Baltic
*eras moldy. Germanic words for elder correspondingly can be ana-
lyzed as containing PIE *pelH)- grey.
The frequent renamings of both species all over Europe may be due
to taboo, linked to popular beliefs of the elder being guarded by The
Elder Mother and her abilities to ward off evil. She was associated with
spinning, considered a partly magical activity with links to the other-
world, which triggered a folk-etymological reshaping of the Balto-
Fennic forms after words for thread. The denotation the grey one
could have had a double connotation elder tree and old lady, remark-
ably alike the homonymy covered by Mod. Eng. elder.
1 The elder is no beech: Lat. sambcus
As is well known, the Indo-European word for beech has played a very
important role in scholarly attempts to locate the Proto-Indo-European
homeland, not least because the distribution of Fagus sylvatica at the
period in question was known to have been restricted to areas to the
West of a line from Kaliningrad to Odessa. For decades the classical
reconstruction of the beech-word was PIE *b-o- ~ *bau-o- (or
*beu-o-), although the mechanisms behind such alternations were not
really understood they were just thought to be reflected in the mate-
rial. This outdated reconstruction can still be encountered once in a
while. At the same time, the beech-word served as the eternal parade
example of how the reconstruction of a homeland and names for species
often rely on specific denotations which may have changed as popula-
tions migrated to new areas.
While most scholars nonetheless agreed on a PIE meaning beech
on the basis of Latin fgus, PGmc. *bk-, *bkjn-, and Continental
Celtic *bk- in Gallo-Lat. Bacena silva beech forest, they also had to
consider a) Greek Kermes oak which phonetically corresponds
to the Western European forms, but has acquired a new meaning; b)
Albanian bung which agrees with Greek semantically but contains a
mysterious nasal1; c) Kurdish buz elm and d) the Slavic word for el-
der, *bz (Cz. bez, Bulg. bz etc.)2 or *buz elder which combined
several problems of phonology reflex of both u-vocalism not present
elsewhere and a palatal stop not matching the Albanian word as well
as the fact that they designated other trees (to the extent that the elder
can be called a tree at all). Furthermore, Germanic with its reflex of an
unaspirated voiced stop seemed to be incompatible with the lack of
Winters law in Slavic3.
Scholars should have put more weight on the suspiciously sparse oc-
currence in Eastern languages in general and Satem languages in parti-
cular the term seemed to be absent in Baltic, Armenian, Indic, and
most Iranian, as well as Anatolian and Tocharian). An excellent histori-
1
Probably just a regular development of *bag-n- (cf. my article on Alb. hund
elsewhere in this publication.
2
Blaek 2002: 201-202 mentions the need to operate with an analogous *baz on
the basis of the genitive *baza to account for Bulg. bs, SCr. baz. However, this
is unnecessary since these forms come regularly from *bz.
3
Kroonen (2013) mentions that beyki can just be an irregular continuant of ON
bki, a form directly continued by Icel. bki.
The Mysterious Elder: Common Traits in European Names 121
cal overview as well as detailed and important considerations based on

lots of new evidence, including from satem languages and Anatolian,
can be found in Blaek (2002 with references). His most important con-
clusion for purposes here is that, based on the last 50 years of research,
no evidence remains for any irregular alternations in the word. In fact,
before Blaek himself, Mod. Icel. beyki is the only form reflecting -u-
that is associated with the beech. He argues that the Icelandic word is
not inherited at all, but a 17th century rendering of Dan. bg with pseu-
do-etymological restitution of -ey- for Da. -- (as in inherited words); he
adds the not important factor that no beech forests, or even any kind of
proper forests, are found in Iceland4. Henning (1963): buz elm is not
related at all, but instead a regular Central Kurdish reflex of PIE *ue!-
(corresponding to wiz in other Kurdish dialects; Slavic *vz, Eng.
witch). Blaek shows that there Iranian reflexes showing ecactly -g-
namely Gilaki fa.
Slavic *bz, *buz (also *buzina, *bzina, cf. Ru. buzin) is proba-
bly not related. Quoting Blaek, it is surprising how many excellent
scholars were ready to accept the relation between *bg(o)- and the
Slavic forms indicating *b()eu()- () in spite of the difference in root
vowel and in semantics. One might add that, even even for the excel-
lent scholars mentioned, the Slavic elder-word remained the only clear
indication of a) an irregular variation between -- and -u- and b) the
palatal character of the following stop. Based on typological evidence,
Blaek himself suggests that *bz, *buz is derived from PIE *bu-
he-goat (Eng. buck; this is the case e.g. in Caucasian languages). He
considers Lat. sa(m)bcus elder a loan from Gaulish, formed with an
alleged IE tree-name prefix also found in e.g. Gk. yew ~ Lat.
lex holm oak, kermes oak.
While I agree completely with Blaek that the Slavic word should not
be included in the beech word-family, I have difficulties accepting his
alternative etymological proposal, mainly because of the structure of the
Latin word. It cannot be ignored in this context that sa(m)bcus clearly
appears to contain the well-known plant-name suffix -cus, -ca as in
albcus asphodil, lactca lettuce, deriving from PIE *-u-ho- look-
ing like X5. Moreover, the Latin words displays a variation between
4
Kroonen (2013) mentions that beyki can just be an irregular continuant of ON
bki, a form directly continued by Icel. bki.
5
Thus unrelated to sambcus harp, flute which, like the synonyms ambbia,
Gk. , is a later loanword from popular Aramaeic with sec-
ondary -mb- from -bb-, cf. Literary Aramaeic sabbk, Syr. abbub (see
samb- and sab- as mentioned by Blaek himself: The form sabcus is

also attested. Unless we are dealing with a folk-etymology, it therefore
seems that we are dealing with an inner-Latin formation to a stem
samb- alternating with sab-. We know one word which shows such al-
ternation, and that is sabulum sand, although forms with a nasal are
not otherwise found within Latin. Before we consider possible links be-
tween the elder and the meaning (looking like) sand, let us return to
the Slavic word.
3 The Turkic origin of Slavic *buz, *bz, Mod.Gr.
In my opinion the best candidate for the Slavic word is a borrowing

from an early Turkic language. OTurk. boz grey and its relatives are
already known to be the source of numerous words in East Slavic, not
least Old Russian, displaying various manifestations of the original vow-
el and sibilant and notably a word among them meaning mildew.
One of them also occurs in West Slavic (examples from Vasmer,
Rsnen 1969, Dyneley Prince 1919).
ORu. busyj, bosyj grey, bust, busovt to become grey, blue or

dark, cf. O.Ru. bosym volkom as a grey wolf (in the Tale of Igors
cam paign) and busovi vrani grey crows
Mod.Ru. buzlk Crocus reticulatus
Mod.Ru. buzlk traces of ice on the boots (Tat. bozluk; boz ice)
Mod.Ru. buzn, bsol, Ukr. busel, Belaru. buse white stork
Mod.Ru. buzn salt from salt lakes
Mod.Ru., Ukr. bsel mildew
Pol. busie, buko, bu, busek young stork
The reason for the large representation of this Turkic word as a loan-
word in Russian is no doubt the importance it has played in the history
of Turkic and Mongolian identity from the times of the ancient steppe
cultures to the present day, mainly connected with its function as epi-
thet of the wolf which functioned as a totem animal for the early Turkic
peoples, cf boz kurd, boz br. Correspondingly, in the Secret History of
Schwyzer 238-239 with numerous parallels). Brchs attempt to relate it to Lat.

scab must be rejected, although we cannot exclude his claim that sabca could
have been reshaped to sambca under the influence of sambcus.
the Mongols, the forefather of Temujin is a grey wolf called Brte Chino
(born 758), with brte meaning grey-blue, grey-white.
These are all East Slavic, but Late Common Slavic bordered areas
inhabited by Altaic tribes (Birnbaum 1998, Andersen 2003), and the et-
ymon is also known from the Bulgarian grey drink buz a grey kvas-like
drink, borrowed by Turkish and perhaps the source of Eng. booze via
Romani (cf. also Chagatai, Osmannic Turkic etc. boza drink made of
camels milk and Chuv. pora, its r-Turkic counterpart, which may ulti-
mately the source of the Gmc. beer-word). Quite remarkably, Mod.Gr.
, obviously a late loan, means water elder either it is a South
Slavic loan although South Slavic forms seem to reflect only *bz or
it is directly from Turkish with a meaning not attested there, having
replaced the name for the same plant as in Slavic earlier in history.
The final Slavic - does not historically represent a vowel, but is just
the automatic LCS (and OCS) manifestation of a word-final non-
palatalized consonant in loanwords, cf. LCS *klobuk 'hat' (< Turkic, cf.
Crimean Tatar kalpak cap), OCS koveg box, casket < supposedly
Avar; cf. Mongolian qagurag), and LCS *tlma 'interpreter' (< Turkic
dolma).
4 Lithuanian eiv-medis, West Slavic *iv-
Interestingly, a meaning grey turns out to be compatible with a Baltic

word for the elder. Lith. eiv-medis elder a compound that can be ana-
lyzed synchronically as the spool-tree, because of its hollow branches
that can be used as bobbins on a loom, cf. eiv (cf. also eiv-kaulis ra-
dius (bone) next to . eiv-kaulis and eiv-film spool film). However,
one of the established PIE reconstructions for grey is the formally iden-
tical adjective *5e)H-uo- grey whose zero-grade *5iH-uo- occurs in
Lith. vas white; light grey; moldy (mostly of cows), vis, vis mil-
dew, OPr. sywas white, Slavic *siv dark grey. Lith. eiv- in eiv-
medis simply looks like the feminine counterpart *5e)H-ueh. Gliwa
(2008) derives the name of the tree exactly from eiv spool but he re-
gards the etymology presented here as equally possible (p.c.)6. One im-
6
Admittedly, it must be noted that there are parallels to the development spool
> elder. At least Wichmann, Uotila & Korhonen (1987: 231) note that in the old
handwritten Udmurt-Russian dictionary by Islenev, it is noted that in the Jela-
buga dialect, eri has come to mean Sambucus. It is a borrowing from Chuvash
portant fact is that the name eiv-medis not only denotes the elder
proper (Sambucus ff.) but also the socalled water elder, also called dwarf
elder or danewort (Viburnum opulus) which has red berries.
The designation is probably the old Balto-Slavic one which can be in-
ferred partly from a loanword in Permian languages denoting the water
elder only (see below), partly from what look like relics from the
Northwest Slavic fringes which notoriously retain archaims from even
before the Late Common Slavic period. In the German dialect of Upper
Saxony the elder is called Schib-chen, and the Mansfeld dialect has ziwe-
cken; both forms must belong with Sorbian diwi bz, which is syn-
chronically wild elder (Brch 233, fn. 1), but probably a folk-etymology;
it must be based on the use of the term wilder Flieder in NE German
dialects that arose in order to be able to distinguish the locally preserved
term Flieder elder from that of the lilac, a new meaning thatFlieder has
acquired in High German after the introduction of the lilac into the
country in the 16th c. (Kluge-Seebold 1989: 220). Sorbian dialectal con-
vergence could easily have caused confusion between initial postalveo-
lar sibilants and affricates even without the folk-etymology (Mucke
1891) but we cannot exclude that the similarity is a coincidence if the
Sorbian loans in German dialects are misunderstandings as asserted by
Brch. With less certainty than for the Baltic and Permian attestations,
then, we can reconstruct a West Slavic *iv- elder.7
For PGmc. we can reconstruct *flira- ~ flira- on the basis of defi-
nitely a derivative with the three-name suffix *-ra. It is now feasible to
assert that the first part is the other widespread PIE adjective meaning
grey (probably originally a different nuance) *pelh)-, known from e.g.
Slavic *pln mold, Lith. plkas grey. The zero-grade was originally
*pliH- which should yield PGmc. *fl-, but alternative zero-grades exist
(Lit. pilkas < *pl-) which might either indicate analogy from formations
where the laryngeal had merged with the suffix (e.g. participial *pliH-to-
> *pli-to-) or that the laryngeal and perhaps even the *-i- are exten-
sions of an originally shorter root *pel-, *ple)- 8.
r, r spool into Votyak from where it has been borrowed further into
Komi as uri; here, it reportedly only means spool.
7
It is perhaps also worth noticing that the old o-grade possibly occurring in the
Lith. dialectal form of spole, aiv, and at least in Latv. saiva, does not occur in
the tree-name (although that would be formaly possible). The tree-name also
occurs as eivmedis.
8
Kroonen on the basis of a dialectal Dutch (N Hollandish) form vlaar recon-
structs Old Frisian *flar and thereby PGmc. *fleura-, but considering how
many Dutch forms are reshaped after verbs for to bvre, flagre (cf. vlinder,
5 Why the grey one ?
At first glance it seems odd that a name for several species of edible
tree-like plants, which, depending on the season, are dominated either
by an impression of shining black or red berries or bright white flowers
would be designated by a color adjective grey. We can consider the
possibility of a semantic shift from either:
a) The white flowers of Sambucus nigra and other species

b) The white pith of the elder, a traditional material for angli
c) The black berries of Sambucus nigra (cf. NHG Holunder elder
and Ru. lalna water elder < PIE *kel- black which already re-
veals that this species and not e.g. Viburnum opulus with its red
berries was the first plant to bear this name), or
d) Grey properties of the willow, cf. that Fi. selja elder is borrowed
from PGmc. *salhj- willow, which is normally regarded to be
ultimately derived from PIE *sal- grey, dirty; ? salt-like ; the
willow can grow as a parasite in elders (Tholle 1944)
As mentioned before, the linking of Sambucus species in popular

taxonomy to Viburnum opulus; cf. e.g. Eng. water elder, NHG Trauben-
holunder, Dan. dvrghyld and Ru. kalna water elder is quite re-
markable seeing that the latter has orangy berries but is still derived
from the root *kel- black. This must mean that its name originally desi-
gnated Sambucus nigra; conversely, OIr. ruis also covers Sambucus ni-
gra, although it is derived from the PIE word for red. From a designa-
tion grey with no semantic shift we can imagine the following motiva-
tions:
e) The particularly grey bark, common in botanic descriptions of

Sambucus species (if used as a material)
f) Blue-grey dye can be produced from the berries of Sambucus
nigra
vliender, vlerk after vlinderen, vlerken, cf. fladderen and even vlieden, vlien move
in the wind), almost as if by a process of phonaestetic value of the sequence vl-
and fl-, I so far hesitate to put to much weight on vlaar. Cf. also the parallel in
Fi. heisipuu becoming hrskipuu after hrsky be frayed, sway and similar
forms from Balto-Fennic below. Note that the Gmc. word is not crucial for the
points presented in this paper, and that a root *pleu- could theoretically also be
an extension of *pel-
g) Powdery mildew on berries, canes and leaves can give elder trees
a grey impression (cf. the second meaning of Lith. vis mil-
dew)
I prefer the latter possibility for two reasons. First of all, if the elder was
once denoted the dusty tree, the powdery tree, we can suddenly under-
stand the formation of its Latin name. The root sa(m)b- we were left
with after having subtracted the productive plant-name suffix -cus
from Lat. sa(m)bcus in is identical to the root in Lat. sab-ulum sand <
PIE *bsam-d-o-. I quite agree with Garnier (2006) that the causes for
this alternation goes back to variants of a PIE collocation *bs-m-eh
deh- to reduce to powder, to pulverize vs. *bs--bo- pulverize; cf.
Skt. bhsma- dust and similar meanings in other IE languages. While I
believe that sambcus was formed in Italic rather than all the way back
to a PIE *bs-m-b-u-ho- or PIE *bs-d-u-ho- (plant) looking
dusty, I do find it very likely, having accepted that sand comes from a
word for powder, dust, that this was still the meaning of *sa(m)b- at the
time of formation. This means that sambcus would be exactly the
dusty tree, i.e. the moldy tree, and not as such the sandy tree which
would be an anachronistic interpretation. Thus, the Latin name can cor-
respond exactly in meaning to Lithuanian and almost with the Slavic
and Germanic ones, meaning rather the grey(ish) tree, but supposedly
still referring to mildew.
6 What loans in Fenno-Ugric can tell us
That at least the Balto-Slavic perception of the elder as the grey one
may be much older than the Slavic-Turkic contacts is indicated by the
following facts :
a) The Fenno-Permic name of the water elder is reconstructed as

*ewe or *ewa, attested in Erza Mordvin ev-gel water-elder
berry, and in the Permian languages : Udmurt (Votyak) u id.,
u-pu water elder and Komi (Zyryan) o(v)-pu id.. This alrea-
dy appears to be a loanword (having *- and no relatives within
Uralic), so I suggest that it is borrowed from a hypothesized
Balto-Slavic word for elder, *e!ua- which is then only attested
in Lithuanian and via this loanword. Note that the Fenno-
Permic word only denotes the water elder which eliminates pos-
sibilities b), c), d), e) and f) as naming motivations ;
b) In Cheremis (Mari), the berries of water elder are called ar,
which must likewise be a loan in the first place ; in my eyes the
only obvious source is East Baltic *er(t)as moldy, cf. Lith.
erktas id. with intrusive *-k-. Note once more that we are
dealing with Viburnum opulus, this time leaving g) as the only
motivation (since a) referred to the flowers and *er(t)as must
have meant moldy, not a specific color, at the time of borro-
wing.
Baltic loanwords in Balto-Fennic are quite numerous (around 330); in

Saami there are some too, and Volgaic languages (Cheremis, Mordvin)
have also borrowed directly from Baltic (van Pareren 2005, 2008; Su-
honen 1988). However, no direct lexical exchange has so far been de-
tected between Permic and Baltic, and at the Proto-Fenno-Permic stage
(= Fennic, Saami, Volgaic + Permic), estimated around 2000 BC, Proto-
Baltic proper could hardly yet have evolved as a distinct dialect of the
Balto-Slavic unity.
If we accept the Slavic word as a Turkic loan, Slavic must conse-
quently have had another word for elder in the preceding centuries
perhaps *siv- preserved in *iv- in Northern West Slavic. And since it is
possible to derive the Lithuanian word from IE, it is conceivable that
*ewa is a manifestation of is the old Balto-Slavic term *e!u, borrowed
into neighbouring Fenno-Permian languages during the Bronze Age.
This would then be the first piece of evidence for a Balto-Slavic word
borrowed into a Uralic stage as old as Fenno-Permian9. The etymology
is confirmed by the fact that the Mari name for the berries of the same
plant, ar, can be traced back to another Baltic word for moldy.
7 The Elder Mother and Indo-European beliefs
Since both Sambucus nigra and Viburnum opulus are indigenous to all
possible PIE homeland territories (including Anatolia), the Proto-Indo-
9
Note that this stage will have to be considered even older by many uralicists in
Finland now, since a consensus seems to have arisen that Uralic did not split in-
to Fenno-Ugric and Samoyedic before Fenno-Ugric then split into Ugric and
Fenno-Permian, but rather that Uralic was divided into West Uralic and East
Uralic where the former would equal Fenno-Permic.
Europeans must have had names for them. Elder belong to what we
could justifiably characterize as surprising lacunae in the IE lexicon
(sturgeon is another one).
The frequent renamings of both species all over Europe may be due
to taboo, linked to popular beliefs of the elder being guarded by The
Elder Mother (Mother Hulda, NHG Frau Holle, Dan. Hyldemo(e)r and
its abilities to ward off evil; the Danish tradition is described by Tholle
1944). The Elder Mother was associated with spinning, considered a
partly magical activity with links to the otherworld; this may have trig-
gered a folk-etymological reshaping of the Balto-Fennic forms; cf. Fi.
heisi elder, Est. (h)is bloom Baltic *eidas bloom) after words for
thread (see Mgiste 1970:357-358):
a) Regular: Fi. heisi, heisi-puu elder, Est. (h)is bloom

b) Influenced by Fi. (Lnnrot) hyt, S Est. he thread: Fi. hisi-,
hysi-puu, Olonets Karelian hdi, Lydian hd-i elder
c) Influenced by hrsk(y), hersy etc. be frayed; sway: Fi. hrski-
puu, hersi-puu elder
In Danish folk-tradition it is told that an elder (a water elder?) will grow

from the middle of Viborg Lake when the enemy (often the Turk) ar-
rives, which is perhaps the reason for the English alternative name
Danewort of the water elder; on the connection between Mother Hulda
and Baltic folk-belief regarding lakes and spirits, cf. Gliwa 2005.
If these religious connotations are very old, we may even cautiously
interpret the grey one as having a double meaning, remarkably, but
presumably incidentally, alike the ones covered by Mod. Eng. elder. In
English folk-belief, Mother Elder is also known as The Old Lady. The
Eng. name itself is reshaped from OE ellrn, partly after other tree-
names in -der (cf. MLG elderne), corresponding to MLG alhorn. It may
very well be that this has happened with the meaning of the homonym
in mind.
We may summarize the scenario as follows: The (unknown) PIE term

for elder became replaced in NW PIE by different terms all meaning
the grey one or the dusty one, referring to powdery mildew and per-
haps only later to Mother Elder. In Balto-Slavic, a reflex of *!e!h-uo-
was used and borrowed into neighbouring Fenno-Ugric languages. In

Late Common Slavic, the new indigenous term was replaced with its
synonym in Old Turkic, used by adjacent Turkic or mixed Altaic tribes;
perhaps the original Balto-Slavic word survived in the Northwestern-
most fringes. Several of the Balto-Fennic forms were reshaped under
influence from the word for thread, referring to Mother Huldas magi-
cal spinning techniques.
We may illustrate the course of events in a chart like this:
Stage 1
An unkown PIE name *X is the original common name for two
kinds of plants Sambucus; viburnum opulus, perhaps already per-
sonified as mother elder and colloquially replaced by the grey one,
the dusty one because of taboo
>
Stage 2
NW PIE the term the grey one or the dusty one gets lexicalized
and becomes the normal term for both species
>
Stage 3
a) Italic uses the dusty one (anachronistically *bs-m-b-u-ho-)
b) Balto-Slavic uses the grey one or the moldy one (*5e)H-uo-)
c) Germanic uses the the grey tree (PGmc. *fli-ra-)
d) Germanic and Slavic also has the black one (*kel-n-) referring
originally to Sambucus nigra, but later also to Viburnum opulus
e) Celtic may have used the red one (OIr ruis) of certain species
and later transferred it to Sambucus nigra
>
Stage 4
The Balto-Slavic word is borrowed to Proto-Fenno-Permic (*ewa)
as the designation for Viburnum opulus only
>
Stage 5
Another (East) Baltic expression for moldy (*era-) is borrowed by
Mari (Cheremis) as the name for the berries of Viburnum opulus
>
Stage 6
a) Late Common Slavic replaces the domestic word with a Turkic
synonym *buz > buz; the original word survives in the extreme
Northwest as *iv- (Sorbian, and as a loan in German dialects)
b) Balto-Fennic forms are reshaped after names for thread (heisi

elder > hisi after hyt etc.
c) The Old English form ellrn is reshaped after other tree names
in -der (such as alder), probably not creating a homonym for old
person by shere coincidence.
References
Andersen, Henning, Slavic and the Indo-European Migrations. Henning An-

dersen (ed.): Language Contacts in Prehistory: Studies in Stratigraphy. Amster-
dam/ Philadelphia: John Benjamins. Pp. 45-76.
Birnbaum, Henrik, 1998: Common Slavic in Time and Space. Scando-Slavica
44:131-143.
Blazek, Vclav, 2002: The Beech-argument: State-of-the-Art. Historische Spra-
chforschung 115: 190-217.
Brch, Joseph, 1922: Lateinische Etymologien [4. Lat. sabucus, sambucus]. - In-
dogermanische Forschungen XL: 232-241.
Dyneley Prince, John, 1919: Tatar Material in Old Russian. Proceedings of the
American Philosophical Society 58: 74-88.
Garnier, Romain, 2006 [2008]: Nouvelles reflexions autour de Grec .
Die Sprache 46/1: 81-93.
Gliwa, Bernd, 2005: Einige litauische Ortssagen. Bodenlose Gewsser und Frau
Holle (KHM24, ATU 480). Studia Mythologica Slavica VIII: 187-223.
Gliwa, Bernd, 2008: Litauisch eivamedis, eiv trinti, Lettisch saiva, Finnish
kvi. Baltu Filoloija XVII (1/2): 41-60.
Henning, Walter B., 1963: The Kurdish Elm. Asia Major 10: 68-72.
Kluge-Seebold 1989 = Friedrich Kluge: Etymologisches Wrterbuch der deutschen
Sprache, 22. ed. revised by Elmar Seebold. Berlin / New York: Walter de Gruyter
1989.
Mgiste, Julius, 1959: Gibt es im Tscheremissischen baltische Lehnwrter?.
Ural-Altaische Jahrbcher 31: 169-176.
Mgiste, Julius, 1970: Estn. (h)is Blte und seine Sippe ein Balt. Lehnwort?.
Velta Re-Dravia (ed.), Donum Balticum [Fs. Stang]. Stockholm: Almquist &
Wiksell. Pp. 352-360.
Mucke, Karl Ernst, 1891: Historische und vergleichende Laut- und Formenlehre der
niedersorbischen (niederlausitzisch-wendischen) Sprache mit besonderer
Bercksichtigung der Grenzdialekte und des obersorbischen. Leipzig [reprinted by
Zentral-Antiquariat der DDR 1965].
van Pareren, Remco, 2005: Baltische leenwoorden in het Mordwiens. MA thesis from
the University of Groningen.
van Pareren Remco, 2008: Die direkten baltischen Lehnwrter im Mordwini-
schen. Finnish-Ugrische Mitteilungen 30-31: 69-147.
Rsnen, Martti, 1969: Versuch eines etymologischen Wrterbuchs der Trksprachen
[= Lexica Societatis Fenno-Ugricae XVII]. Helsinki: Suomalais-Ugrilainen Seura.
Schwyzer, Eduard, 1934: Dissimilatorische Geminatenauflsung als Folge von

bersteigerung usw.. Zeitschrift fr Vergleichende Sprachforschung 61, 3/4:
222-252.
Suhonen, Seppo, 1988: Die baltischen Lehnwrter der finnisch-ugrischen Spra-
chen. Denis Sinor (ed.): The Uralic Languages: Description, History, and For-
eign Influences. Leiden / New York / Copenhagen / Cologne: E.J. Brill. Pp. 596-
615.
Tholle, Johannes, 1944: Hyldebusken. Danske Studier 1944: 1-38
Vasmer, Max, 1953-1958: Russisches etymologisches Wrterbuch.Heidelberg: Winter.
Wichmann, Yrj, 1903: Die tschuwassischen Lehnwrter in den permischen Sprachen
[= Mmoires de la Socit Finno-Ougrienne XXI]. Helsinki: Suomalais-
Ugrilainen Seura. 98-99
Wichmann, Yrj (original manuscript), T.E. Uotila & Mikko Korhonen (eds.),
1987: Wotjakischer Wortschatz [= Lexica Societatis Fenno-Ugricae XXI]. Hel-
sinki: Suomalais-Ugrilainen Seura.231.
Balto-Fennic *kakra oats, the Etymology of hail,
and another Exception to the Germanic Sound Shift
Abstract
Labials develop regularly into velars when preceding -l- and -N- in vari-
ous Scandinavian dialects past and present. Since delabialization is not
common before -r-, Nw. dial. hagre, OGutn. hagri, Mod.Gutn. hagre,
Elfd. ager, all oats, have been interpreted as reflecting an original
PGmc. variant *hagran- alongside the more widespread *habran-.
Kroonen (2013) separates the two forms, reconstructing a meaning
broom grass for *hagran-, with reference to especially Da. hejre id.
However, it has been overlooked that Proto-Balto-Fennic *kakra, an
early Germanic loan, points to a meaning oats attached to the velar
variants in Proto-Germanic already. Witczaks (2003, 2004) PIE recon-
struction *!op-r grass; vegetables is accepted, but a semantic and deri-
vational stratigraphy reveals that core IE remnants of the old heteroclitic
stem (such as Alanic zabar) acquired a specific meaning oats, while
purely thematicized forms were attached to less specific, non-
agricultural, meanings. PGmc. *hagla- hail can be related; deriving
terms for hail from grain is typologically common, and a sound-law
whereby PIE *-opl-, *-apl- yields *-agl- or *-ak(u)l- would allow for five
other Germanic words of disputed origin to be assigned quite straight-
forward etymologies: *tagla- tail could go back not to *do5-lo- but to
*dop-lo-, derived from the root in Sw. tafse tuftan alternative also
applicable to OIr. dal, but leaving out otherwise irregular Slavic forms;
OSw. sakla to drool is linked to EFris. sabben to drool and PGmc.
*safta- juice rather than NWGmc. *sakkan- sink slowly, Sl. *sok
juice ; Da. rakle catkin, Sw. dial. rackel long thing; tall, slim person
are equated with Lith. rpls thongs, akin to Sl. *repj burdock, Alb.
rrap plane tree, PGmc. *raftra- long, thin pole; *skakulaz whippletree;
traces for harness horse; schackle is deemed identical to Lat. capulus
halter for horse; towing rope; handle, scapula shoulder, derived from
the root in scpus, PGmc. *skaftaz shaft; and *hakulaz cloak is seen as
a derivative of the root in MLat. cappa, cpa id., which can be shown to
be of IE rather than Semitic origin.
1 Balto-Fennic *kakra and Germanic words for oats
Labials have developed into velars in the position before -n- and -l- in
various North Germanic dialects, cf. for *-pn- ON vkn weapon, OSw.
(Vstgtaland, stgtaland, Hlsingland, Vstmanland, Sderkalix)
vkn, Far. dial. vkn and even Fi. vaakuna, an old Scandinavian loan-
word, next to ON vpn, OSw. vpn, Far. vpn < PGmc. *wpna- id.;
Norrland Sw. gcken, Mod.Gutn. gauken, Far. geykn handful ~ Sw.
gpen, MDa. giben, ON gaupn both hands held together, OHG
goufana handful; empty hand, palm < PGmc. *gaupn-; OSw. (Vstg-
taland) and Norrland Sw strgna suffocate < *stropna, from strypa
strangle; for *-fn- OSw. ughen, Sw. ugn, SW Nw. and Inner Trndelag
Nw. ogn, Icel. nn oven next to ODa. (Haderslev Stadsret 1292) ufen,
Sw. dial., Nw. omn < PGmc. *ufna- id.1; for *-pl- Nw. drygla to secrete
from the uterus (of a cow in rut) ~ Nw. dropla drip quietly, PGmc.
*draupa- a drop; and for *-fl-, *-bl- *-vl- Sw. dregla to drool ~ OE
dreflian to dribble or run at the nose, slobber, Eng. drivel, drool <
PGmc. *drabljan-2, OSw. sughl meat ~ OSw. sufl, Sw. sovel < PGmc.
*subla-; and OSw. swaghel, Sw. dial. svagel sulphur ~ Standard Swe-
dish svavel, borrowed from MLG swavel.
1
For delabialization in this item in Nordic specifically, cf. Bjorvand & Lindeman
(2001: 704) and Kroonen (2013: 446). Traditionally, OSw. ughen, Sw. ugn have
been considered Verner variants of another form *uh(w)na- reflected in Goth.
ahns. However, Gothic in fact seems to have undergone a similar, although
more restricted, development of f > h before nasal (cf. ahuma upper <
*ufuman-), not contested by any obvious counterexamples (cf. also lauhmuni
lightnig in fn. 40; see Hyllested & Cohen 2007: 15-16). Not a single Germanic
attestation thus points unambiguously to a variant with labiovelar *uh(w)na- or
*ug(w)na-; hence, we need not reconstruct any other forms for oven in Ger-
manic than *ufna-, with perfectly regular developments throughout the daugh-
ter-languages (that is, if one disregards late dialectal confusion resulting in vari-
ation in, say, Modern Standard Swedish where umn would be the expected
outcome).
2
These etymologies for Sw. dregla and Nw. drygla respectively are new. The two
words have otherwise been grouped together (Tamm & Noreen 98; Hellquist
1898) but are usually not assigned further certain etymology.
BF *kakra oats, hail and Exceptions to the Gmc. sound shift 135
Mostly, this is described as a sporadic, yet common phenomenon3,

However, judging from the material in e.g. Rydquist (1868: IV, 239, 249),
it appears quite if not altogether regular in some dialects of (Old) Swe-
dish (especially Norrland and Vstgtaland) and Norwegian (confined
to Old W Norwegian, Mod. SW Norwegian and the dialects of Inner
Trndelagen [Lierne]), and it dominates in Faroese4, while, on the other
hand, it is completely absent from (Middle) Danish5 and Old Gutnish6.
3
Thus Wessn 1965 (I: 46) for Swedish in general, Olson (1904: 116) for the old
dialects of stgtaland and Seip (1931: 188) for Old (S)W Norwegian. Howev-
erm they only mention the development of *p > k, g/_n.
4
Also after loss of intervening -t-, cf. akna become evening < ON aftna id.; Old-
er Far. <okn> (= kn) swan < *lptn- (Lockwood 1961: 57).
5
Unfortunately the situation in Modern Danish is often ambiguous because of
the merger between *-g- and *-v- and sometimes *-b- in relevant positions, and
the possibility of spelling the ultimate result [w] both with <g> and <v>, and
sometimes even <b>. This is also the case for forms with <v> in Bokml Norwe-
gian, even pronounced with labiodental [v], that are historically Danish (e.g.
ovn oven).
6
According to Kroonen (2013), OSw. ljung-eld, Mod.Gutn. liaugn lightning
constitute another example and must come from Old East Norse *ljfn (OSw.
lyghna f. id. < *-jn), which would in itself be a secondary labialization of Proto-
Norse *-uhn- into *-ufn-. Kroonen here implies that an old Verner variant PIE
*leuk-n- would not have yielded PGmc. **leugna- but *leukka- with Kluges
Law instead. Nw. ljon, lyn, Da. lyn lightning (and Mod.Icel. ljn hurricane, at-
tested from the 19th.c.) would still point more directly to a Germanic and Norse
protoform with *-hn-. At first glance, the supposition that *ljfn was a general
East Norse form seems to be contradicted by ODan. liughnth lightning since
the developments *-f/v- > -g- /_l/n and *-p- > -k- /_ l/n do not occur in Old
Danish; however, -gh- in this word can simply reflect a difficulty of identifying
the fricative [] following -u-, cf. the orthographic variants frugh ~ fru <<
OSax. fra or MLG frwa and (Brndum-Nielsen). On the other hand, liugh-
nth is the only example where a purely orthographic (unetymological) -gh-
would occur outside hiatus, so we cannot exclude that *-ghn- does reflect a Pro-
to-Norse *-gn-, which, because of Kluges Law, must still be explained as sec-
ondary from a Proto-Germanic point of view. In fact, no Germanic forms even
point unambiguously to its unvoiced counterpart, PGmc. *leuhna-. In Gothic,
first of all, luhmuni f. flash may straightforwardly reflect *laufmunj- (see the
previous footnote). Second, we cannot be completely sure of the regular out-
come of a PGmc. full-grade sequence *-eufN-, *eubN-, *-aufN-, or *-aubN- in
Old West Norse; what we have apart from Nw. ljon lightning, Mod.Icel. ljn
hurricane (attested only from the 19th.c.) is the variation between ON,
Mod.Icel. ofn oven on one hand and Mod.Icel. nn , attested from the
1500s, on the other. There is a possibility that nn is borrowed from Danish
(since from that century onwards has been pronounced as a dipthong close to
Da. -ov-), but O prefers an old variation in OWN. The latter remaining a pos-
Since delabialization is not common before -r- anywhere, Nw. dial.

hagre, OGutn. hagri, Mod.Gutn. hagre, Elfd. ager, all oats, have been
interpreted as old forms reflecting a PGmc. variant *hagran- (~
*hagrn-) m. alongside the more widespread *habran- (~ *habrn-) m.
(> ON hafri, Sw. havre, OS haboro, OHG habaro, NHG Hafer)7. While
the forms reflecting a variant with velar are simply left out in Orel
(2003), Kroonen (2013) tries to separate the two variants historically,
and on the basis of especially Da. hejre brome grass and, with less
weight, Mod.Icel. hellin-hagra mother-of-thyme, Thymus praecox; wild
thyme, Thymus serpyllum8, he tentatively recontructs the meaning
bromus for PGmc. *hagran-9. He further notes Brome grass is closely
related to the wheat-grass lineage, and is known to infest grain fields,
leaving open various possibilities of relationship to PCelt. *korkio- oats,
OIr corc hair and Nw. dial. hagr, harg coarse hair from a horses mane
or tail via dissimilation or metathesis10. In any event, he chooses to sep-
arate *hagran- with a velar from *habran- with a labial, assuming an old
meaning oats only for the latter.
Both Orel and Kroonen derive *habran- from *habra- goat, thus
operating with an inner-Germanic origin of this word. However,
Witczak (2004) makes an excellent case for a broader Indo-European
sibility, we cannot exclude that Icel. ljn, Nw. ljon reflects the same kind of vari-
ation, and that *-eufN- and *-euhN- merged in (variants of) OWN. The remain-
ing form mentioned by Kroonen, ME lven lightning, flash, flame, is traced
back to a form *lauhumna- but may equally well reflect *laufumna-. Now, even
a lone PGmc. *leuf- needs an explanation: From an Indo-European point of
view, we would obviously expect lightning to be derived from *leuk- light,
which occurs in other derivatives as PGmc. *leuh-; however, the word for light-
ning in ON is actually leiptr (Mod.Icel. leiftur) which belongs with Lith. liepsn
id., going back to an altogether different root *le)p-. A Proto-Germanic con-
tamination between *leuh- in leuhman- beam of light (ON ljmi, OE loma)
etc. on one side and *leif- in leiftraz lightning (ON leiptr-) on the other, result-
ing in new PGmc. words for lightning beginning with *leuf-, appears to have
been unavoidable.
7
Thus, Da. havre can theoretically reflect both *habra- and *hagra-.
8
Glossed in Kroonen (2013) as a kind of thyme. The common name in Modern
Icelandic today is blberg.
9
Elsewhere in the entry he supposes a broader meaning than just oats.
10
An equestrian term common to Celtic and Germanic would match the results
reached in Hyllested 2010 quite well; there, I argue that old vocabulary common
to Germanic and Celtic can be grouped into a few semantic categories; one of
the important subcategories is exactly terms pertaining to horsebreeding and
riding. Another word meaning horsehair, shared by Celtic and Germanic, is
mentioned later in this article.
word family involving Hitt. kappar- vegetables; Skt. pa- m. drifting

reed; reed washed onto a riverbank; Alan. zabar oats, Mod.Pers. sabz
grass; green; dark, Pashto sbah grass; vegetables, Shughni sip(i)yak
a kind of millet (< PIr. *spar-ku-), Roshani sabc pod of bean; Lith.
pas straw, grass, pai what remains in the field after a flood; MIr.
corca, coirce oats; W ceirch, MBret. querch (in the Celtic forms, *-p- has
regularly disappeared, and -ch- is seen as reflecting a suffixal *-k-)o-, cf.
the Iranian forms). Although Witczak reconstructs both PIE forms, a
heteroclitic *5op- and a thematic *5op-o-, with very broad meanings
grass; vegetables, especially the Alanic meaning underlines that the
meaning oats can be old. More specifically, oats seems to have be-
come the specialized meaning of the non-Anatolian IE remnants of the
heteroclitic form, while the thematic form is rather attached to mean-
ings like grass, straw, or dead plants floating in the water. In PIE in-
cluding Anatolian, the exact meaning is less precise, but it appears that
PGmc. n-stems are built upon secondary thematicizations to the origi-
nal heteroclitic.
Several other of Kroonens points can be contested. Designations for
brome grass can be transferred from those from oats or vice versa
simply because of the strong physical similarity between the plants; cf.
e.g. Danish hejre-havre bromus. An Icelandicj name like hellin-hagra
does not tell us much about the original meaning of *hagrn- since it
only makes up the last part of a compound of a type common for grass-
like or edible plants or fodder, cf. besides hejre-havre also Dan. drap-
havre oatgrass, Arrhenaterum. Furthermore, hejre could theoretically
be connected to Nw. hagr, harg coarse hair from a horses mane or tail
(and Shetland Norn hegri thin yarn of wool) without having anything
to do with the oats-word, cf. the almost exact parallel source of Nw.
fakse, Sw. faxe bromus < ON faxi horsetail. While I do find it possible
that a contamination or maybe even a reshaping took place involving
the horsetail-word, so that for example a PGmc. *harga- horsehair
merged with *habra- oats into the variant *hagra- oats, this would not
have taken place earlier than in Common Scandinavian for the follow-
ing reasons:
The velar variant is attested in all languages of the Balto-Fennic
group, everywhere with the meaning oats, and unanimously pointing
to a Proto-Balto-Fennic form *kakra, cf. Fi. kaura, E dial. kakra, Karel.
kagra, Veps kagr, Est. kaer, Vot. kagra, Liv. kagr. Since Proto-Balto-
Fennic is usually regarded to be at least as old Proto-Germanic, we
must maintain a form *hagra- with the meaning oats alongside the
synonymous *habra- already in Proto-Germanic (although it is of

course theoretically possible that the various branches of Balto-Fennic
borrowed the word individually at a later stage, say, from a stage close to
Proto-Norse, a possibility not preferred by Ockhams razor). In other
words, Balto-Fennic confirms that *hagra- did not primarily mean
bromus which would perhaps be a less obvious culture-word anyway.11
Summing up, it seems clear that PGmc. *hagran- m. oats12 and
*habran- m. both existed in Proto-Germanic as variants of a word mea-
ning oats, and that the latter variant was borrowed into Proto-Balto-
Fennic as *kakra. It is an n-stem formed from a secondary thematiciza-
tion *5op-r- of an old heteroclitic *5op-r which in core IE acquired a
meaning oats (in some Eastern Iranian languages, it has further come
to signify a local kind of millet). Derivatives outside the heteroclitic sys-
tem retained meanings of a less specific and less agricultural character
such as grass, vegetables, straw or plants flowing in water. This
provides us with a stratification of meanings that neatly match what we
claim to know about the cultural history involved : That the speakers of
PIE were less hard-core agriculturalists than the speakers of later Wes-
tern IE offshoots.
2 PGmc. *hagra- oats and *hagla- hail
Regardless of its precise meaning, the existence of a PIE root *5ap- or

*5ep- used in words for crops, everywhere occurring in what could be
11
Neither do I agree with Hofstra (1995: 95) that Fi. kakra and its Balto-Fennic
cognates can be loanwords from Old Gutnish during the middle ages. As men-
tioned before, the delabialization is not even specifically Gutnish in the first
place, cf. vapn weapon, hamn harbor, stefna, stemna to point out, afla
breed, gafl fork, cafli piece cut out, although at least gauken handful and
probably liaugn lightning show that it occurs (sporadically) in Modern Gut-
nish which is not an uncontaminated descendant of Old Gutnish. What we
need is a the most plausible scenario to explain the presence of velar forms in all
Balto-Fennic languages.
12
Several people have contested my use of the term delabialization in this case
since by delabializing a labial you do not get a velar. I maintain the use of the
term, for which there is no good alternative, because delabialization in this case
refers to phonological rather than phonetic conditions. The velar stop is simply
the stop you get by removing the feature [+lab] from the labials. Typologically,
dentals also become velar when preceding sonorants, so that velars can be said
to make up the unmarked stops in this position.
the o-grade, paves the way for etymologizing the otherwise mysterious
PGmc. *hagla- hail.
According to the etymological standard handbooks, PGmc. *haglaz
m. ~ *haglan- n. hail (> ON hagl, OE h(e)l, haol, OFris. heil, OS,
OHG hagal) derives from a PIE protoform *kag-lo-, whence also Gk.
pebbles. The Germanic-Greek equation indeed appears accep-
table from both a semantical and a formal point of view. However,
Greek also possesses unexplained variants like with divergent
vocalism, and where the first consonant is missing ; such a varia-
tion is normally interpreted as a sign of Pre-Greek substratum origins
(Beekes 2010 : 606 ; Kroonen 2013)13. Bjorvand & Lindeman (2000: 336),
as well as Orel (2003: 150) believe that we are rather dealing with an in-
herited Verner variant of a PIE o-grade noun, *5okl-, whereby *hagla-
corresponds to PGmc. *hahla- slippery with original stress on the first
syllable, *5klo-. This word would have been formed by means of redu-
plication from the root *5elH- freeze (reconstructed with a laryngeal
because of the acute in Lith. lti freeze (of ice)) ; cf. also p-alas
frozen ground, Du. hal id. ; reduplicated formations with -i- in the first
syllable are known from Skt. iira- cold, ON hla hoarfrost < *5i-
5lH-o-. Neither etymology can be excluded, and, so far, the origin of
*hagla- must be regarded as unsolved.
Semantic typology often proves a fruitful starting-point in etymolo-
gy: When designations for hail actually are synchronically analyzable,
what do they then reveal ? In many languages they turn out to constitute
parallels to the classical etymology involving the Greek word for
pebbles; beginning with Modern English, a single piece of hail is called
hailstone, in Dutch correspondingly hagelsteen and in Portuguese pedra
de granizo, literally stone of hail. This is also the most common way of
designating hailstones in the older Germanic languages: ON haglsteinn,
OE haolstn, MLG hagelstn, MHG hagelstein. On the other hand, cal-
ling hailstones grains of hail appears typologically just as common, cf.
e.g. NHG Hagelkorn, Da. haglkorn, It. chicco di grandine and the etymo-
logy of Sp. granizo (a derivative of Lat. grnum grain which is also the
source of the last part of the Portuguese word). Besides, the term for the
hail-like meteorological phenomenon graupel contain words for grain
in Da./Nw. iskorn and Dutch ijskorrel, lit. ice-grain, and the word
graupel itself is related to both MHG s-grpe hail and NHG Graupe
grain, historically a derivative of PGmc. *greupan- ~ *grpan- to cut
13
It should be noted that Beekes represents the Leiden school where a-vocalism in
itself is regarded as a typically non-Indo-European feature.
up, grind (Kroonen 2013). Furthermore, in Danish and Norwegian lexi-

cography, hail itself is typically defined as iskorn, literally grains of
ice14.
The connection between hail and grain goes further: the H-rune in
the runic alphabets, which in the younger futhark looks much like a sty-
lized snowflake, was called hagall in Old Norse and hgl in Old English.
Although these names are not completely identical to the word for hail
(ON hagl < *hagla- n., not hagall < *haglaz m.) there is no doubt that the
name refers to hailstones. In the Anglo-Saxon rune poem from the 8th
or 9th century, it is described in the following way :
Hgl by hwtust corna

hwryft hit of heofnes lyfte
wealco hit windes scra
wore hit t were syan
Hail is the whitest of grains ;

It whirls down from the heavens height (air)
And gusts of wind toss it about;
Then it is transformed to water
The Norwegian rune poem, dating back to the 13th. century, reads:
Hagall er kaldastr korna

Kristr skp himenn forna
Hail is the coldest of grain;

Christ created the world of old
The Icelandic rune poem, from the late 15th century onwards, says:
Hagall er kaldakorn
ok knappa drfa
ok snka stt
Hail : cold grain

14
In the meaning graupel it is less of an everyday word and therefore sufficiently
little lexicalized to be used as a productively formed noun whose two parts are
understood literally.
and driving sleet

and sickness of snakes15
Griffiths (2006) argues that in the famous Old Irish work Auraicept na
n-ces (the scholars primer, originally from 650, edited until around
1150, preserved in the Book of Ballymote from 1390), a word-play invol-
ving the word for grain, OIr. grin, can lay behind the naming of the h-
rune in the Ogham alphabet, (h)ath, since both hath and grin also
mean horror. The word-play would then also involve Latin horror,
meaning both horror and shivering (from cold). In the so-called ar-
boreal tradition, where Ogham runes are named after trees, (h)ath is
glossed as white-thorn (probably referring to hawthorn, but according
to Griffiths deliberately described like hail). Judging from the different
kennings of the rune, it is obvious that the rune-name is connected me-
taphorically and mythologically with its homonym hath horror (tran-
slations by McManus 1988):
1 (Bratharogam Morainn mac Maon)

condl can
assembly of packs of hounds
2 (Bratharogam Maic ind c)

bnad gnise
blanching of faces
3 (Bratharogam Con Culainn)

ansam aidche
most difficult at night
Thus, the typological perception of hail as grain is clear. It is therefore

worth considering if PGmc. *hagla- contains the same root as *hagra-
oats; cf. that corn in Scots and Hiberno-English means both oats and
grain in general. Such conditions could also explain the similarity bet-
ween the most widespread Indo-European words for hail and grain ,
respectively : a) Lat. grand, OCS grad, Arm. karkowt; and b) Lat.
grnum, PGmc. *kurna-. Lat. grand results either from regular meta-
thesis -nd- < *-d-n- following suffixation with *-no- (Rasmussen 1999
[1984]: 152-154) or a contamination of -d- in the hail-word with -n- in
15
Translations from Griffiths 2006: 90.
the grain-word. Cf. that although Spanish and Portuguese derive from
Latin, their words for hail(stone), despite the great similarity, do not
derive regularly from the Latin hail-word, but seem to be derivatives of
the grain-word.
More than anything, *hagla- now simply looks as if it contains the
same root as our oats-word, obviously with a similar delabialization
like the one in *hagran-. This kind of delabialization is already known to
occur sporadically in Germanic preceding liquids, cf. e.g. Dutch heuvel
next to NHG Hgel hill, and Limburgish swegel < *swebla- sulfur
(Kroonen 2013).
Now, one could of course hypothesize that *-g- for expected *-b- not
only in *hagra- but even in *hagla- is due to sporadic delabialization or
word contamination. However, for the latter it seems more likely that a
sound-law is applying. On closer inspection, it indeed turns out that
examples with a preserved labial before -l- and following PGmc. *-a-
are totally absent : There are simply no cases of the expected outcome
PGmc. *-afl- ~ *-abl- from PIE *-apl- or *-opl-. Quite a few instances of
Germanic *-afl-, *-abl- do occur, but they always come from PIE *-abl-,
*-obl- (e.g. PGmc. *kabln m. a piece cut off > ON kafli, OLFr. cavele,
cf. Lith. bas branch)16. This indicates that *hagla- could be the regular
16
Since I first made this claim, Kroonen (2013) has presented two candidates in
his dictionary that seem to contradict it: *afla- hearth (ON afl id., Far. alvur,
alvi fireplace, forge) which he compares to Hitt. appena baking-kiln, fire pit,
broiler (oven), tentatively reconstructing an l/n-stem *hp-()l, gen. *hp-n-s.
He admits that the Hitt. word can altenatively be compared to PGmc. *ufna-
oven, which, according to him looks like an old wanderwort. Even if one is
willing to accept the existence of PIE l/n-heteroclitics different from r/n-
heteroclitics (on this topic, see now Kerkhof 2012), *ufna- has a better match in
Hitt. uppar oven (< PIE *hup-r, *hup-n-), while Hitt. appena rather be-
longs with Gk. baked (< *hep-). Original PIE word-final -l probably al-
ready developed into *-r while the few safe examples seem to reflect a condi-
tioned preservation of *-l in *-ul-/-uen- stems only. In this particular case, one
might admittedly argue that -l- could represent original *-ul- whose labial
consonant would automatically be lost after -p (as would be the case after any
other labial); this demands, however, that the loss of labial is older than the de-
velopment of -l > -r which does not seem likely. Kroonens other example is the
homonymous *afla- strength, power (ON afl n., Far. alv n. id.), reconstructed
as *hep-lo- on the basis of Hitt. app-. The Germanic noun is traditionally re-
constructed as *afalan- (OE afol) where the labial does not immediately preced
the -l-. The PIE reconstruction in -elo- (classical *apelo- before the inclusion of
Anatolian reflexes) is based mainly on Greek which admittedly proves little
since Greeks apparently avoids even original suffixation with -l- directly added
to a root, replacing it with *-elo-. Even from a purely Germanic viewpoint, a
reflex of *5op-lo- or *5ap-lo-, possibly in PIE already, but more likely in

a late stage of (non-Anatolian) IE when more agricultural meanings of
the original heteroclitic had arisen.
3 Other etymologies with delabialization
Regular delabialization of PIE labials into velars between a PGmc. low

vowel on one side and -l- on the other would render it possible to clarify
five additional, otherwise disputed etymologies:
3.1 P GMC . TAGLA - ( HORSE ) TAIL
PGmc. *tagla- horsetail (> ON tagl, Goth. tagl id.; Da. tavl, Zealand
dial. taggel horsetail ; Nw. dial. tagl fibre; OE tgl, Mod.Eng. tail,
OHG zagal tail) corresponds to OIr. dal tail and is normally recons-
tructed as PIE *do5-lo-. Whether -5- here was palatal or not cannot be
decided on the basis of Germanic and Celtic alone; at first glance, a
plain velar seems to be needed to account for the alleged Slavic cognate
*dolka, yielding SCr. dlka a single hair and Cz. dlk branch. Mataso-
vi (2009: 102) reconstructs a palatal to be able to include Skt. da- (f.)
fringe17. However, this word-family rests on a shaky ground; not-
withstanding the semantic developments, in Slavic an irregular metathe-
sis is needed, involving the original suffix (with subsequent depalatali-
zation of the original palatal before a sonorant), and, besides, it is far
from certain that the Common Slavic form was *dolka since there are no
descendants outside Serbo-Croatian and Czech where *-olC- og *-laC-
merge into -lC-. Another possible Slavic reconstruction is thus *dlaka.
From a semantic viewpoint it is just as possible that the Slavic words are
related to ON tlkn baleen ~ MLG tolle branch, from PIE *del-g()-.
OIr. -aR- can come from *-akR-, but is also regular from *-apR-, cf.
can harbor next to PGmc. *hafna-, meaning the Old Irish form dal is
fully compatible also with an original PIE *-p- instead of *-k-.
Reconstructing *-p- i *tagla- furthermore has the advantage that it
renders possible an equation with other Germanic words of similar
vowel seems to have intervened, thus preventing the delabialization from taking
place at the time in question.
17
Matasovi (2009) reconstructs *do5-eh but the lack of Brugmanns Law dic-
tates e-grade.
meaning, but without the -l-suffix: Dan. tave fibre, tavse, tjavs tuft,
Sw. dial. tafse, S Sw. dial. tav(e) tuft and even PGmc. *tappan- tap. A
close connection between these forms and *tagla- may be reflected not
only in the meaning of Nw.dial. tagl fibre, but also in the Scandinavian
sayings Nw. med topp og tagl18, Da. med top og tavl, med tap og tavl
completely (i.e. with all its body). These sayings of course function as
alliterational figures regardless of the etymologies of their elements, but
the Danish expressions with tap instead of top at least show that they
can be hendiadytic (lit. with hair and hair = with every [kind of] hair)
rather than referring to two opposite extremities like the type Eng. from
tip to toe, head over heels 19. An intermediate type, comparable in mea-
ning to med top og tavl is Da. med hud og hr, lit. with skin and hair.
For phonotactic reasons it is already clear that *-la- in *tagla- must
be a suffix and cannot belong to the root. I will therefore tentatively re-
construct *dap-lo- or *dop-lo- as the PIE form behind PGmc. *tagla-
and OIr. dal, with other derivatives *dap- having reflexes in Germanic
only. Relationship with Skt. da- fringe of course remains a possibili-
ty, in which case the reconstruction is *do5-lo-20:
3.2 OD A . SAKL TO DROOL , SW . DIAL . SAKKEL DROOL
A third relevant item is the Scandinavian verb for to drool, Da. savle,
ODa. sakl, Older Mod.Da. sagle, sgle, Sw. dial. sakla, sagla. It is pro-
bably denominal from Da. savl drool, Older Mod. Da. sagel, sagle,
sgle, Sw. dial. sakkel. This noun has traditionally been connected to
NWGmc. *sakkan- sink slowly, sag (> Da. sakke (bagud) lag (behind),
fall behind), Mod.Icel. sagga become moist, Du. zakken drop, sag,
and, outside Germanic, Slavic *sok juice. Semantically, though, it is as
18
Nw. topp, Da. top, from PGmc. *tuppa- (> ON toppr tuft, lock of hair, Far.
toppur crest, OE topp top, OFris. topp tuft, OHG zopf plait of hair) can
most easily be etymologized as a pseudo-etymological zero-grade of the root in
*tappa-, cf. the almost identical semantics of several forms and the lack of obvi-
ous alternative etymologies. The complexity of Germanic reflexes seem to con-
firm an origin in NW European IE and the Germano-Celtic vocabulary
(Hyllested 2010).
19
In lack of an established term for the latter type of idiom, Petr Kocharov (p.c.)
suggests to introduce polarindrome.
20
Kroonen (2013) as an alternative presents an inner-Germanic etymology ac-
cording to which *tagla- would be the diminutive to either *tahjan- to unravel
or to *tgan- ~ *takkan- prickle, branch.
close you can get to E Fris. sabben to drool, LG sabbe drool; spit, and
further Du. sabbelen to suck, which contains PIE *sap- (> PGmc.
*saf/ppan-, *safta- sap, juice ; moist, Lat. sapa must, new wine boiled
thick, Arm. ham juice ; Kroonen 2013: 336). Here, too, I would there-
fore reconstruct an original labial and assert another derivative with *-
lo-, PIE *sap-lo- > PGmc. *sakla-, with no exact equivalents outside
Germanic.
3.3 D A . RAKLE CATKIN , SW . DIAL . RACKEL LONG THING ; TALL ,

SLIM PERSON
Da. rakle catkin, ament (flower cluster on trees) and Sw. dial. rackel
long thing; tall, slim person are etymologically obscure.21 A PIE recons-
truction *rop-lo-, however, would render possible the establishment of a
larger word-family, seeing that the Danish meaning comes close to Sla-
vic *repj burdock, arctium (> Ru. repej id., Ukr. repyk sticklewort,
Agrimonia) and Alb. rrap plane tree (whose fruits, achenes, are remi-
niscent of burdocks or round catkins of e.g. a hazel tree), while the
Swedish meaning matches PGmc. *raftra- long, thin pole; rafter (> ON
raptr rafter, OE rfter small beam); both of these are derived from
PIE *rep- stick to, pick up (> ON rfr roof on rafters ; Lat. rapi to
snatch, Gk. , Alb. rjep tear off, Lith. ap-rxpiu, ap-rxpti to
grasp; LIV 507), and formally they can even be equated with Lith.
rpls, rpls f.pl., OPr. raples f.pl. thongs, albeit with an alternating
ablaut grade. The original meaning of *rop-lo-, *rep-lo- would then be
snatcher and secondarily burdock, preserved best in Slavic, only later
21
The unexpected -k- for -g- in Danish has not yet been explained. Rakle is not
common in the singular; it is my impression that quite a few speakers only have
the plural in active use. It is thus conceivable that rakle is a comparatively recent
back-formation from the pl. rakler (Older Mod.Dan. pl. rackle), in turn from an
older sing. *rakkel (a structure for which the modern pl. would also be rakler),
corresponding to the Swedish form; cf. the parallel change of the sg. skagel into
skagle below. In Danish, -Vkke- and -Vgge- are graphic renderings of the same
pronunciation, so maybe *raggel was simply a dialectal form of the type Zea-
landic saggel for savl, sagl (see the previous entry). If *raggel was interpreted as
the homophonous rakkel, a new plural rakler could arise. It is also possible that
the irregular consonatism is due to contamination with dial./obs. Dan. rakke,
also meaning catkin, ament, which is otherwise unrelated and constitutes a
pair with the non-assimilated ranke floral vine. However, I find it more likely
that the meaning of rakke, in turn, was influenced semantically by rakkel.
transferred to flower clusters of a burdock-like shape. A reconstruction

*rap-lo- would not match the root vocalism of Balto-Slavic, Greek or
Albanian, so the Germanic *-a- must reflect o-grade, and e-vocalism in
the PIE root is ensured.
A problem compared to *hagla-, *tagla- og *sagla- is of course that
*rakla- has PGmc. *-k-, not *-g-, an issue curiously repeating inself in
Danish with rakle for expected ragle (see the previous footnote). We
will return to this variation below. Two further items point to a Ger-
manic reflex *-akul- from PIE *-apl-:
3.4 PG MC . SKAKULA - WHIPPLETREE ; TOWING ROPE ; SCHACKLE
PGmc. *skakulaz m. (> ON skkull car pole, OE sceacel, sceacul

shackle, gyve, hobble, MDu. schakel chain link ; Da. skagle (<
OMod.Da. skagel), Sw. skakel, Nw. skokle have two basic meanings
whippletree, double tree and traces for harness horse, towing rope; as
a loanword in Fi. kak(k)ula, ka(k)kuli it only means whippletree. As
stated in LGLOS (II : 18) the shape of the Finnish material is too un-
marked to determine at what stage the borrowing took place ; the limi-
ted geographical distributionFinnish only, and mainly West Finnish
dialectstips the scale in favor of a late borrowing from Scandina-
vian/Norse rather than Proto-Germanic, maybe even as late as Old
Swedish. Still, LGLOS leaves both possibilities open since theoretically
kak(k)ula can also represent the Proto-Germanic accusative.
In either case, the meaning of the Finnish word is still important for
our etymological purposes since it confirms that the meaning
whippletree is old in Scandinavian and apparently the only meaning
exported eastwards. In combination with the many Danish meanings
only referring to a piece of wood, both meanings must be reconstructed
for common Scandinavian. The meaning shackle is particularly West
Germanic, but may have fallen within the Proto-Germanic range, too.
According to the standard etymology, PGmc. *skakulaz is a diminu-
tive formed with the suffix *-ula- of the root in Nw.dial., Sw.dial. skk,
Sw. skak towing rope; fettle for animals < *(s)kenk-, cf. Lat. cingere to
gird, Lith. kinkti to harness horses22. Semantically it is an acceptable
match, but there is no satisfying explanation as to why the nasal would
22
Orel (2003: 332) instead derives *skakula- from the root in the Sw. and ON verb
skaka shiver, quiver < PGmc. *skakana-, with W Germanic meanings run
away.
have disappeared in Germanic, and, again, it cannot be excluded that a

contamination between two words has taken place.
Latin provides a clue. It is remarkable that Latin capulus displays a
range of meanings similar to those of PGmc. *skakulaz: a) hackamore,
headstall, b) halter for catching and fastening cattle; lasso, c) sword-
hilt, handle and d) coffin. An original *skap- can be found in Lat.
scpus shaft and the otherwise disputed word for shoulder (blade),
scapula. Scpus is already known to be related to PGmc. *skaftaz ~
*skaftan- m./n. shaft; pole (> ON skapt, OE sceaft, OHG scaft id.), so
the root *skap- probably originally referred to a piece of wood con-
nected to another part of a tool, rather than a rope-like device. I there-
fore suggest reconstructing PIE *skap-lo- shaft; double tree,
whippletree, which later, but already in NW IE, acquired the secondary
meaning traces for a harness horse ; towing rope and shackle(s). In
Latin the missing s- in capulus can be explained by contamination with
the near-synonymous capistrum headstall, harness, derived from caput
head; thereby it is perhaps unnecessary to operate with s mobile in this
case23. The meaning coffin can be viewed as an enlargement of handle,
but may simply be a historically different word, cf. the next item and
especially the meanings of MLat. cappellum chapel, Lith. kpas grave,
Port. campa id.:
3.5 PG MC . * HAKULAZ CLOAK , MANTLE
A very parallel formation can be seen in PGmc. *hakulaz ~ *hakuln

m./f. cloak, mantle (> Goth. hakuls, OHG hahhul, ON hkull, OE ha-
cele id., Da. hagel cloak; (dial.) shawl). It is usually regarded as a deriva-
tive of *hakn hook, nook ; buckle (Orel 2003: 154). I have earlier ar-
gued, however (Hyllested 2010), that MLat. cappa, synonymous to
*hakulaz, is not a medieval Semitic loanword as otherwise assumed, but
inherited from PIE *kpo- which underwent the littera-rule whereby a
long vowel plus a single consonant can be replaced by a short vowel fol-
lowed by a geminate. Incidentally, the variant cpa is attested in this
23
Capulus itself can of course also be derived from caput, which, however, makes
less sense when you consider the meanings b), c), and d). It should be noted
that, regardless of etymological origins, Latin speakers probably had a feeling
that both capulus in the meaning headstall and capistrum belonged with caput
synchronically, i.e. a contamination would not only be based on historical states
of affairs.
case, and must represent the original form. *kpo- is not attested in the
meaning cloak outside Latin, but this problem can be solved by assu-
ming that cloak was not the primary meaning of the derivative. *kpos
is already known as the PIE word for piece of land; holy enclosure; gar-
den.It appears in Gk. garden, PGmc. *hfa- (holy) enclosure
(> ON hof hill with holy place) ~ *hb- piece of land, and, with a
PIE suffix *-i-sth- used for nomina loci24, in ORu. kapite holy place,
idol and Alb. kophst garden. This word-family was established by
Witczak 2004; in Hyllested (2010) I added MLat. cappellum chapel;
holy enclosure25; Lat. castrum in the meaning fortification, dim. castel-
lum26; Lat. campus field27; Capitlium, the name of various hills with
holy places scattered around the Roman realms, most famously the one
24
Cf. e.g. Gk. plane-tree grove, OHG ewist sheepfold, ON vzt f.
fishing ground.
25
Said to be named after the small enclosure in the cathedral of Aachen where the
relics of Saint Martin of Tours, including his cloak, are preserved. There can be
no doubt that cappellum is formally secondary to cappa, but on the other hand
there is no evidence that the meaning moved from a primary cloak to a secon-
dary chapel in the middle ages (the French name of Aachen, Aix-la-Chapelle, is
named after Charles the Greats grave).
26
Lat. castrum fortification is most often understood as a result noun piece cut
out, strip of land corresponding to the homonym agent noun or instrument
noun castrum knife < PIE *5as- cut (out). Formally, though, castrum in the
sense fortification can equally well be derived from PIE *kap-i-sth-ro-m whose
first two parts can be identified with the aforementioned nomen loci formation
*kap-i-sth-. It would regularly yield Lat. castru- via syncope (of short vowel be-
fore *-st(r)- as in monstrum evil omen; monster < *monestrum, sstertius two
and a half; sesterce < *semi-s-tertios; this happened in the 6th-5th c. BC, cf.
Meiser 1998: 66) and subsequent assimilation (or loss, of *-T- > /_sT, e.g. Os-
cus 'Oscan' < Opscus [Enn. Ann. 296], asper raw < *ap-sper-; cf. Meiser 1998:
117).
27
Lat. campus field can formally be identified with PIE *kmpos wave (> Gk.
), but semantically it corresponds more closely to the aforementioned
*kpos which has otherwise left no trace in Latin. The most probable scenario is
therefore that a contamination between *kampos and *kpos took place on the
way to Latin, resulting in a single word that retained the shape of the former
and the meaning of the latter. There are signs that the contamination was not
fully completed even in Vulgar Latin, since the original meaning of *kpos en-
closure seems to have been retained in Port. campa grave (cf. also Lith. kpas
grave), although a homorganous nasal infix would be no less different that the
one in the name of the region Campnia ~ Osc. Kap(v)ans, Etr. capevane, Gk.
; or It. Campidoglio, literally the oil-plant fields, but from Lat.
Capitlium.
in Rome itself28; the name of the Thracian fortification

(cf. Duridanov 1987); the name of the temple Secca di Capistello for Di-
ana and Aeolus from the 4th century BC on the Aeolic island of Lipari;
Capistrello, an Italian place-name historically connected to Roman and
Oscan fortifications; and the name of Santiago de Compostela, the fa-
mous catholic pilgrimage city in Galicia which also functioned as a holy
place in Pre-Christian times29.
A double meaning cloak, blanket, cover on one side and piece of
land ; (holy) enclosure on the other is typologically rather common, cf.:
a) Skt. namata-, Av. namta- cloak next to PCelt. *nem-eto- holy

enclosure (> OIr. nemed, Gaul. Nemeto-) and PGmc. *nem-ita-
(OLFr. nimidas and the Swedish farm-name Nymden); cf. also
Latin nemus holy enclosure
b) W caen f. cover, blanket; skin, hide ~ Lat. caulae < *caholae
enclosure, PGmc. *hagan- garden
c) ON kgurr blanket ~ Kgra, Swedish village- and farm name
(Elmevik 1975)
In a parallel fashion, we can now establish:

d) Mlat. cappa, cpa cloak, mantle, PGmc. *hakula- id. ~ PIE
*kp-o-, *kap-ist-o- piece of land; (holy) enclosure'
Semantically this makes sense if the basic verbal meaning is encircle,

surround < get hold of, hold < catch, take, cf. that the examples in a)
above probably are derived from PIE *nem- to take (> PGmc. *neman-
28
Capitlium can also go back to PIE *kap-i-sth-. According to Roman grammar-
ians, an earlier form was Capitdium, which renders possible an etymological
segmentation capi-td-, cf. cus-td- < *cus-to-sd- he, who guards the treasure; a
derivative *kp-ist-o-sd-)o- would mean he who belongs to the *kp-ist-o-sd-,
in turn he who guards the *kp-ist-o-, the holy hill, and a sequence -sd--st-
could easily be subject of dissimilation, i.e. *kpistosdiom > *kpitosdiom. The
guardian referred to can either ne Rx Nemornsis the king of the holy grove (=
Gaul. Rgonemeti), the goddess of the holy grove Diana, or (Loucetios) Mars.
29
In Old Portuguese (the forefather of Modern Portuguese and Modern Galician),
there was a confusion between the spelling variants <am> and <om> (cf. e.g. or-
fom orphan, found in king Duartes Leal Conselheiro, 1428-1438, from Ro-
mance orfanu-) it is possible to assert an original *Campustella or
*Campistella, simply meaning the holy place. Again, the variation between
forms with nasal and forms without point to confusion between *kampos and
*kpos at a late stage.
). Correspondingly, d) can originally be identical to the verbal root

*kap- to catch nown from PGmc. *habna- to have and Lat. capre.
I conclude that PGmc. *hakula- cloak can be derived from PIE
*kap-lo-, a derivative of the same root as in MLat. cappa, cpa cloak.
4 Conclusions
We end up with six potential examples of the postulated development

PIE *-apl-, *-opl- > PGmc. *-agl-, *-akl-, *-akuland, crucially, no obvi-
ous counterexamples:
PGmc. *hagla- hail < PIE *5op-lo- (~ Late PIE *5op-ro- crops,
grain > PGmc. *habra- ~ *hagra- oats)
PGmc. *tagla- (horse)tail < PIE *dop-lo-, *dap-lo- (~ Da. tave, Sw.
tafse tuft, PGmc. *tappan- tap)
PGmc. *sagla- drool (sb.) < PIE *sap-lo- (~ E Fris. sabben to drool,
LG sabbe drool; spit, PGmc. *saf/ppan-, *safta- sap, juice ;
moist; Lat. sapa must)
PGmc. *rakla- catkin < PIE *rop-lo- (~ PGmc. *raftra- rafter, Lith.
rpls f.pl. thongs, Ru. repej burdock, Alb. rrap plane tree)
PGmc. *skakula- whippletree; towing rope; schackle < PIE *skap-lo-
(~ PGmc. *skafta- shaft; pole, Lat. capulus hackamore; lasso;
handle, scapula shoulder, scpus shaft)
PGmc. *hakula- cloak, mantle < PIE *kap-lo- id. (~ MLat. cappa,
cpa cloak, mantle)
The extra -u- in the two final examples has been explained by Kmmel
(2004) as a reflex of partly analogical, partly regular developments with-
in the paradigms of Germanic nouns with a stop followed by a sonant in
the stem.
The division into three items with voiced spirants and three other
items with unvoiced stops also needs an explanation. Such an alterna-
tion is reminiscent both of Verners Law (voiced vs. unvoiced) and parts
of Kluges Law (retention of an orignal unvoiced stop before sonant;
here -l- which is retained, and not a nasal assimilated into the stop) and
could reflect original accent alternation in exactly this position. The de-
velopment itself can perhaps be characterized as a sort of assimilation, if
*-l- at the point in question was of the thicker, velar kind.
Future studies can perhaps reveal whether other kinds of delabializa-

tion processes in Germanic (see e.g. Markey 1979) can be explained as
the result of similar processes. It is important to note that the situation
outlined in this article does not necessarily contradict Kortlandts (1997:
48) hypothesis that labiovelars went in the opposite direction and be-
came real labials next to sonorants.
References
Beekes, Robert S.P., 2010: Etymological Dictionary of Greek. Leiden: Brill.

Bjorvand, Harald & Fredrik Otto Lindeman (eds.), 2001: Vre Arveord. Etymologisk
Ordbok. Oslo: Novus.
Brndum-Nielsen, Johannes, 1935: Gammeldansk Grammatik i Sproghistorisk Frem-
stilling. Copenhagen: J.H. Schultz.
Duridanov, Ivan, 1987: Die geographische Terminologie indogermanischer Her-
kunft im Thrakischen und Dakischen. Wolfgang Meid (ed.): Studien zum in-
dogermanischen Wortschatz. Innsbruck: Innsbrucker Beitrge zur Sprachwis-
senschaft. Pp. 29-34.
Elmevik, Lennart, 1975: Kgurr och Kgra. Namn och Bygd 63: 38-49.
Griffiths, Alan, 2006: Rune-names: The Irish Connexion. In M. Stoklund, B.
Holmberg, M. Lerche Nielsen and G. Fellows-Jensen (eds.): Runes and their se-
crets. Copenhagen: Museum Tusculanum Press. Pp. 93-101.
Hellquist, Elof, 1898: Om nordiska verb p suffixalt -k, -l, -r, -s, -t samt av dem
bildade nomina. Arkiv fr Nordisk Filologi n.f. 10: 1-46 and 136-194.
Hofstra, Tette, 1995. Beziehungen zwischen Germanen und Ostseefinnen vor der
Hansezeit. Tijdschrift voor Skandinavistiek 16, no. 2: 85-103.
Hyllested, Adam, 2010: Saint Martin of Tours, the Capitolin Triad, and Latin Re-
flexes of PIE *kp-o-, *kap-ist-o- piece of land; (holy) enclosure' . Foredrag
afholdt p Greek and Latin in an Indo-European Perspective III i Bratislava, Slo-
vakiet, juli 2010.
Hyllested, Adam, 2012: Hagl, tavl og skagle. Flere undtagelser til den germanske
lydforskydning? in Hans Frede Nielsen (ed.): ldre germansk sproghistorie [=
Mindre Skrifter Blaa Serie 29]. Odense: University Press of Southern Denmark.
Pp. 5-12.
Hyllested, Adam & Paul S. Cohen, 2007: Monophthong for expected -diphthong
in Greek. Matthew McCullagh, Coulter George, Benedicte Nielsen, Antonia
Ruppel & Olga Tribulato (eds.): Greek and Latin from an Indo-European Per-
spective [= Proceedings of the Cambridge Philological Society Supplementary Vol-
ume 32]. Cambridge: Cambridge Philological Society. Pp. 9-18.
Kerkhof, Peter, 2012: Suffix variation in the PGmc. l-suffixes and the ablaut of the
PIE l-stems. PhD dissertation from Leiden University.
Kortlandt, Frederik, 1997: Labials, Velars and Labiovelars in Germanic.
NOWELE 30:45-50.
Kroonen, Guus, 2013: An Etymological Dictionary of Proto-Germanic. Leiden: Brill.
Kmmel, Martin Joachim, 2004: Ungeklrtes u neben Liquida in Germanischen

Nomina. A. Hyllested, A.R. Jrgensen, J.L.Larsson og T. Olander (eds.): Per
aspera ad asteriscos. Studia Indogermanica in honorem Jens Elmegrd Rasmussen
anno MMIV. Innsbruck: Innsbrucker Beitrge zur Sprachwissenchaft. Pp. 291-
303.
Lockwood, W.B., 1961: The Faroese Bird Names. Copenhagen: Munksgaard.
LGLOS = Andries D. Kylstra, Sirkka-Liisa Hahmo, Tette Hofstra & Osmo Nikki-
l: Lexikon der lteren germanischen Lehnwrtern in den indogermanischen Spra-
chen I-III. 1991-2012.
LIV = Helmut Rix (ed.): Lexikon der Indogermaniscen Verben. 2nd ed. 2001. Wies-
baden: Reichert.
McManus, Damian, 1988: Irish letter-names and their Kennings. riu 39: 127-
168.
Markey, T.L., 1979: Delabialization in Germanic. Journal of Indo-European
Studies 7: 199-211.
Meiser, Gerhard, 1998: Historische Laut- und Formenlehre der lateinischen Sprache.
Darmstadt: Wissenschaftlice Buchgesellschaft.
Olson, Johan Emil, 1904: stgtalagens Ljudlra. Lund: Gleerupska Universi-
tetsbokhandeln.
Orel, Vladimir, 2003: A Handbook of Germanic Etymology. Leiden / Boston: Brill.
Rasmussen, Jens Elmegrd, 1984 [1999]: Miscellaneous Problems in Indo-European
Languages III. APILKU 4: 135-149 [reprinted in J.E. Rasmussen: Selected Pa-
pers on Indo-European Linguistics, 1999. Museum Tusculanum Press. Pp. 146-
155].
Rydquist, Johan Erik: 1868: Svenska Sprkets Lagar I-VI. Stockholm: Norstedt.
Seip, Didrik Arup, 1931: Norsk Sprkhistorie till omkring 1370. Oslo: Aschehou & co.
(W. Nygaard).
Tamm, Fredrik August, & Adolf Noreen, 1890-1905: Svensk Etymologisk Ordbok 1-
8. Akademiska boktryckeriet E. Berling.
Wessn, Elias, 1965: Svensk Sprkhistoria I-III. Stockholm / Gothenburg / Uppsala:
Almquist & Wiksell.
Wiklund, K.B., 1917-1920: Die ltesten germanischen Lehnwrter im finnischen.
Indogermanische Forschungen 38: 48-114. (61-62)
Witczak, Krysztof Tomasz, 2003: Indoeuropejskie nazwy zb. d: Wy-
dawnictwo Uniwersytetu dzkiego.
Witczak, Krzysztof Tomasz, 2004: Albanian kophst garden and its cognates.
Studia Etymologica Cracoviensia 9:179-787.
Two Issues on Indo-European Substrates in Slavic1
Abstract
Georg Holzers (1989) identifcation of an Indo-European substrate

language in Slavic Temematian is correct, as also argued by
Kortlandt (2004). However, Kortlandts skepticism regarding the
specific sound development PIE * > Temematian *Ro has proved un-
founded. Holzer did not give any examples of substantivized past passi-
ve participles, an otherwise very common word-type which, if contai-
ning a liquid in the root, would surface as *CRod or *CRon in Teme-
matic words in Slavic. It is suggested that at least LCS *plod fruit and
*grozd ~ *grozn grape; cluster, perhaps also *drozd thrust, are Te-
mematic borrowings. Another possible Temematism is zabar oats in
the Alanic (Iassic) word-list.
Trubachvs hypothesis that Baltic and Slavic *-st- can reflects PIE *-
5- in lexems from another Indo-European substrate is rejected, seeing as
in all of his examples, st can be accounted for in other, more transparent
ways.
1 The problem of identifying related substrate languages
Virtually all specialists would agree that Indo-European languages are

newcomers in Europe which replaced most of the indigenous languages
in several steps from the Bronze Age onwards. To which extent this lin-
1 This article forms part of a larger paper, Indo-European Substrates in Slavic

Revisited that was presented at the 17th Conference of Scandinavian Slavists in
Copenhagen, August 2007; at the 19th Annual UCLA Indo-European Confer-
ence, 2 November, 2007; and at the 14th International Congress of Slavists in
Ohrid, Macedonia, 11 September 2008. Most of the points in section 3 were
printed in Hyllested 2004, while the etymology for plod only was presented at
the XVI Conference of Scandinavian Slavists in Uppsala, August 2004.
guistic replacement reflects actual human migrations, let alone entire

population shifts, is another (albeit related) matter and a matter of
debate but the great changes that took place may in any case be char-
acterized as Indo-Europeanization. The encounter between indige-
nous Palaeo-European cultures and what we now define as Indo-
European culture, however, was most certainly not the only interaction
that took place. By the time of arrival in Europe, the Indo-European
languages and populations were already fragmented and scattered in
many directions some were still immediate neighbours and must have
exchanged lexemes to some extent.
Convergence between closely related languages in the past consti-
tutes one of the greatest challenges for historical linguists, especially
when one of them is neither directly attested nor has any attested de-
scendants. The bulk of the vocabulary is simply either directly inherited
or borrowed from closely related dialects who got their terms ultimately
from the same precursor. The consequence is typically that linguists
overlook the language which disappeared because loanwords from it are
not interpreted as preserved traces of a forgotten language, but merely
irregular varieties of inherited material in the related language that bor-
rowed them.
2 Holzers Temematisch
Holzer (1989) suggested an extinct and otherwise unknown Indo-

European language, which he calls Temematisch (Eng. Temematian)
to account for a variety of irregular consonant correspondences in Slav-
ic. He ascribes 62 lexemes in Slavic to this hypothetic extinct language
basically all lexemes with troublesome or unsatisfying etymologies. The
language name is an acronym based on the asserted developments: PIE
tenues (unvoiced stops) became mediae (voiced stops), and mediae as-
piratae became tenues. Other Temematian features, according to
Holzer, are that zero-grades of liquids surface as *-Ro-, and that long
vowels become shortened before sonant.
PIE Temematic (in Slavic) Normal Balto-Slavic corresp.

*b, *d, *g > *p, *t, *k ~ *b, *d, *g
*p, *t, *k > *b, *d, *g ~ *p, *t, *k
*r, * > *ro, *lo ~ mostly *l, *r
/_R >V ~ distinct reflexes
Two Issues on Indo-European Substrates in Slavic 155
Among Holzers best examples (in my opinion) are:
PIE *suo-poti- ones own lord (cf. Vedic svpati- id.) > LCS *svobod
free (for expected svopot)
PIE *del-(ent-) suck, nurse (cf. Latvian dle sucking calf) > LCS *tel
calf (for expected del)
PIE *brs-o- grain, crops (cf. CSl. brano flour, ON barr barley) >
LCS *proso millet (for expected *brs or *brso). Compare now
Toch. B proksa (pl.) grain, but cf. also LCS *br millet < *puH-ro-
Kortlandt (2004) evaluated Holzers theory positively, also in a historical

context, but had one minor remark: He did not consider Holzers specif-
ic Temematian development * > Ro more than theoretically possible,
pointing to the fact that the evidence is limited to five lexemes, and that
in none of these zero-grade is to be expected:
2. LCS (via Tem.) *proso < PIE *brso ~ Lat. far

6. LCS (via Tem.) *loboda < PIE *l-pod ~ Lat. olor
10. LCS (via Tem.) *krot < PIE *grdo- ~ Goth. garda
27. LCS (via Tem.) *prok < PIE *brgo- ~ Goth. bargan
30. LCS (via Tem.) *slobod < PIE *sl-poti- ~ ON salr
While I agree with Kortlandt that this is one of the weaker Temematian
correspondences judging from Holzers own material, I find it remarka-
ble that no past passive participles in *-t- are represented since they are
common and constitute the zero-grade environment par excellence. Ac-
cording to Holzers rules, such participles if borrowed from Teme-
matian would end in LCS *-d. If the stem ended in a liquid (perhaps
followed by a laryngeal) it would be easy to check if Holzers rule is cor-
rect since in that case they would have the shape *CRod corresponding
to regular Slavic *CRt and PIE *C(H)-t-. Since the participles
formed part of the conjugational system and could be formed produc-
tively as adjectives and substantives even in the daughter-languages, we
would expect at least a couple of those containing liquids in the root to
show up in the Temematian material but of course they can only be
identified if Holzer was right that their zero-grade was markedly differ-
ent from the regular Slavic one.
2.1 LCS * PLOD FRUIT
The Slavic word for fruit, OCS plod, although of disputed origin (see
e.g. Bezlaj 1995), is most often taken for a loanword from PGmc. *blaa-
leaf, ? fruit < PIE ppp. *bl-t-, ultimately from the root *bel- swell.
However, the exact phonological representation remains a problem:
Why is PGmc. *b- replaced by p-, and why is the vowel -o-?
Shevelov (367) equated it with Celtic forms like Ir. loth foal and
Welsh llwdu young person, while Bezlaj (1995) proposes a back-
formation from PIE *pled-men (with voice assimilation) < *plet-men
rope; thread, whence the meaning lineage; offspring (LCS *plem); he
refers to typological parallels with double meanings such as Skt. tantu
thread; offspring; Slov. pasma race; train column; and SCr. lza rope;
fruit tree.
Curiously, a PIE participle *b(H)-t- of this root, where a zero-
grade is obviously expected, would in fact yield Temematic *plod. It
corresponds formally to PGmc. *bul-i/- which is attested e.g. as Dan.
byld abscess. The root could also have contained a laryngeal because it
is the loss of long-short distinction is another Temematic feature (
*bh-t: -RH- > -Ro-; Kortlandt 2004).
The meaning fruit in PGmc. occurs only with full-grade and femi-
nine gender: OE bld f., MLG blt f., but on the other hand we encoun-
ter a similar semantics in zero-grades of other extended root-forms like
Sw. dial. bljon blueberry (< PGmc. *buli- < *b---), Dutch bolster
fruit shell, husk, and Gk. overripe.
2.2 LCS * GROZD GRAPE ; ? CLUSTER
The same semantic sphere fruticulture and viticulture - remarkably

features another good candidate for a substantivized Temematian parti-
ciple. LCS *grozd likewise has an unexpected -Ro- while ending in -d.
The traditional etymology, deriving it from PIE *erzd- barley, does
not explain the synonym grozn. However, nasal participles are pre-
served in Slavic, so the variation *grozd ~ *grozn is yet another sign
that we have encountered a Temematic participle. They can be com-
pletely identical to PGmc. *hursta- shrubbery (< *krs-t-) and OIr
crann (< *krs-n-) tree (meaning originally vine?) respectively. But
there is also the possibility that the Russian meaning cluster represents
an archaism, and that grape is secondary in that case, the word could
be identical to Latv. grste linen bundle and OCS grst handful < PIE
*grt-sti-.
2.3 LCS * DROZD THRUST
A Slavic *drozd (e.g. Ru. drozd) thrust exists alongside the more wide-
spread variant *trozd (OCS trozd). However, this case is less certain
because forms from other Indo-European languages show that a) *-d is
not participial, but comes from an original PIE sequence *-d-o-, and b)
while zero-grade forms do occur, like Lat. turdus (< *trzdos), the more
closely related Lith. strzdas with o-grade (< *strozdo-) shows that the
sequence *-rV- is original in the full-grade, and Slavic *-ro- does there-
fore not have to reflect a Temematian zero-grade. However, *drozd
would be the expected Temematian outcome of the word in any case.
2.4 A LANIC (I ASSIC ) ZABAR OATS
Johnny Cheung (p.c.) has suggested that zabar in the Alanic (Iassic)
word-list is simply a rendering of Hungarian zab which is loanword
from Slavic. This solution, however, does not satisfactorily account for
the second part of the word ar although Cheung suggests that it
might be Hung. r price. Incidentally, zabar is the only word on the list
without a counterpart in Modern Ossetic. However, all other words in
the left column are consistently Iranian, so it seems reasonable to inter-
pret zabar oats as the corresponding Iassic word for oats.
Holzer includes Slavic *zob in his list of Temematian words. The
Iassic word could either be inherited from Proto-Iranian or reflect yet
another Temematian agricultural term in Slavic (subsequently borrowed
into Iassic). In either case, zabar would be semantically and morpholog-
ically identical to the West Indo-European oats-words (PGmc.
*habran-, PCelt. *korkio-) versus the more archaic formations and
meanings in the East (Hitt. kappar vegetables, Skt. pa- drifting
reed), thus reflecting a morphological and semantic innovation point-
ing rather clearly to an agricultural specialization that follows the earli-
est dissolution of the Indo-European dialects and migrations into Eu-
rope. Iassic).
3 Trubachvs substratum reflex -st- < *-"-
Another Indo-European substratum language, more specifically an ex-

tinct Balto-Slavic dialect or Satem dialect, was established by Trubachv
(e.g. 1991) mostly to account for the numerous discrepant dorsals, i.e.
kentum-like reflexes of PIE palatal stops. However, another reflex ascri-
bed to ancient dialect convergence is Baltic or Slavic st for expected s
from PIE *5. Three examples are mentioned by Andersen (2003):
3.1 L ITH . STIRNA , L ATV . STIRNA DEER
Lith. strna, Latv. stirna deer are seen as substratum reflexes of PIE
*5er-n-, cf. PSl. *sirn. However, not only Slavic, but also Old Latvian
has s-, cf. pl. <Ssirnos>. The forms with st- thus appear recent and are
likely to have been influenced by German in medieval times, cf. OHG
stiora (NHG Stier) bull or OHG stirna forehead; skull pad on war-
horse (< blaze, characteristic forehead)2.
3.2 L ITH . T U KSTANTIS , L ATV . TKSTOTIS THOUSAND
Lith. tUkstantis, Latv. tkstotis deer is another of Trubachvs promi-

nent examples, but here -st- stands in medial position and reflects the
regular Baltic development of the Balto-Slavic cluster *-s-ts- with *-ts- as
the intermediate stage between 5 and its final unmarked outcome s.
Pedersen (1942) thought that this only happened in position before
front vowel whereas the regular outcome in front of back vowel is -sk-.
However, note Lith. lakas leaf vs. OCS list id. This Balto-Slavic word
is a more obvious source for Fi. lehti leaf (< *lete) and Mari lita,
lta id. than (NW) PIE *blh-to-/*bleh-to- as suggested by Koi-
vulehto (1995).
2
Irregular intial position seems to be common in names for big mammals. Even
Lith. stuMbras bison has an unexpected st- although from PIE *- Latv. sumbrs
~ CSl. br; cf. parallels like OIr. fearb ~ earb deer; cow; ON jrr ~ stjrr
bull (PGmc. *eura- ~ *steura-).
3.3 O LD P RUSSIAN PARSTIAN PIG
The sequence in OPr. Parstian, rendered <prastian> (Elbinger Vocubu-

lary 686) is to be segmented pars-tia-n where -tia-(n) forms a diminu-
tive, as in e.g. werstian (little bull) calf (EV 674) vs. Lith. veris calf,
wosistian <wolistian> kid(EV 677) vs. Lith. os goat, and eristian
lamb (E 681) vs. Lith. Nras id. Thus, only -s- and not the entire sequence
-st- reflects *-5- in (NW) PIE *por5o-.
According to Leskien (1891: 583) and Skardius (1943: 332) the suffix
shape is -istian and may be connected with Lith. -iia- (see also
Schmalstieg 2003: 271), assuming that forms without -is- have under-
gone dissimilation, e.g. werstian < *versistjan. It remains that -t- is suf-
fixal (whether it belongs with -s- or not) and cannot reflect PIE *-5-.
References
Andersen, Henning, 2003: Slavic and the Indo-European Migrations. Henning

Andersen (ed.): Language Contacts in Prehistory. Amsterdam/Philadelphia:
John Benjamins. Pp 45-76.
Bezlaj, France, 1995: Etimoloki Slovar Slovenskega Jezika III: P-S. Re-edited by M.
Snoj & M. Furlan. Ljubljana.
Holzer, Georg, 1989: Entlehnungen aus einer bisher unbekannten indogermanischen
Sprache im Urslavischen und Urbaltischen. Vienna.
Hyllested, Adam, 2004: Review of Andersen 2003. Acta Linguistica Hafniensia 36:
179-184.
Koivulehto, Jorma, 1995: Zur indogermanisch-germanischen Kontinuitt in der
Nachbarschaft der Finnougrier. Der Ginkgo-Baum. Germanistisches Jahrbuch
fr Nordeuropa 13. Helsinki. Pp. 116-137.
Kortlandt, Frederik, 2004: An Indo-European Substrate in Slavic?. Alfred
Bammesberger & Theo Vennemann (eds.): Languages in Prehistoric Europe.
Heidelberg. Pp.
Leskien, August, 1891: Die Bildung der Nomina im Litauischen. Leipzig: Knigl.
schs. Gesellschaft der Wissenschaften.
Pedersen, Holger, 1942: Et baltoslavisk problem. Rosally Brndal, Viggo Brn-
dal, Christen Mller & Hedvig Olsen: In Memoriam Kr. Sandfeld. Copenhagen:
Gyldendalske Boghandel Nordisk Forlag. Pp. 184-194.
Schmalstieg, William R., 2003: An isogloss uniting Baltic, Slavic, Germanic.
Alfred Bammesberger & Theo Vennemann (eds.): Languages in Prehistoric Eu-
rope. Heidelberg. Pp. 261-278.
Shevelov, George Y., 1964: A Prehistory of Slavic. Heidelberg: Carl Winter.
Skardius, Pranas, 1943: Lietuviu kalbos odi daryba. Vilnius: Lietuvos moksl
akademija.
Trubachv, Oleg, 1991: Etnogenez i kultura drevneiskikh slavjan. Moscow: Nauka.
The Story of time:
The Etymology of Finnish aika
(with an excursus on aita fence and Balto-Slavic v-prothesis)
Abstract
The traditional etymology deriving aika from Germanic *aiwa- time

must be rejected because PGmc. *-w- would not be rendered as Balto-
Fennic *-k-. Instead, it is a borrowing from a different Germanic deriva-
tive from the same PIE stem, namely *ajuka- life, age, an old thematic
noun forming the basis of the adjective *ajuki- eternal, which can be
reconstructed on the basis of Old English ce and Gothic ajuk- in
ajukds eternity. The development of *-u- > *-e- in Middle Proto-
Fennic and subsequent loss in Late Proto-Fennic is regular in the medial
position of such loans.
While BF *aita fence can be reconstructed for Fenno-Ugric, it
might ultimately be a late Indo-European borrowing from *ho)to- de-
limitation, demarkation that yielded the Celto-Germanic word for
oath, Lith. viet place and Slavic vt council; convention; oath. BF
*aita and vt then form a pair systematically corresponding to*aika vs.
Slavic vk time, indicating a development of initial PIE *(H)oi- to Bal-
to-Slavic *vai- in sandhi contexts contexts.
1 Finnish aika time cannot come from PGmc. *aiwa-
Kari Liukkonen (1999: 18-19) justifiably characterizes the traditional et-

ymology deriving aika from Germanic *aiwa- time as phonetically im-
possible: Germanic -w- would not be rendered as Balto-Fennic *-k-.
Instead he proposes that Balto-Fennic *aika time (> Fi. aika) is a ren-
dering of Baltic *eig course (of events) (> Lith. eig), an old derivative
of the verb eti to go, cf. also eigoje during (< PIE *hei- go).
However, this does not work either. Proto-Baltic *ei is regularly re-
tained as Balto-Fennic *ei which only later develops into ai in South
Estonian and Livonian. Larsson (forthc.) states:
It must first and foremost be clarified that PBalt. *ai can indeed yield
EastBaltic ie (e.g. Lith. dievers brother-in-law and ORu. dver, Gk. dr,
Lat. laevir, Arm. taygr [] Another key example is Fi. taivas heaven, sky
which is generally said to be a borrowing from Balt. *deiuas (Lith. divas,
Latv. devs, OPr. EV deiwis god) [] However, this example is better ex-
plained [ as] an early loan from Indo-Iranian, i.e. IIr. *daiuas.
One could add that the study of semantic fields speaks for an Indo-
Iranian origin. Fi. jumala God, attested already in Ottars account as
ON Jmli is a derivative from an old name for the sky, formed with the
productive nomen loci suffix *-la. The base word juma is of Indo-
Iranian origin and identical to Skt.dyman- sky (derived from PIE
*dei-, *dieu-, famous for its occurrence in names of Indo-European
gods. Together with taivas, religion already seems to constitute a visible
semantic field among Indo-Iranian loans.
A late development of initial *e- to *a- (not only when it forms part
of diphthongs) is known from Lithuanian (cf. dial. agis), but not from
Baltic as such, and since cognates of aika are found all over Balto-
Fennic, the word must be a very old borrowing. One might of course
conjecture that *ei- regularly yielded ai- exactly in initial position, but
there are not many examples (Andersen 1996 does not mention any), so
such an argumentation would be circular. We have a motivation for ini-
tiating a search for alternative etymologies.
2 The oath and the fence an origin in Balto-Slavic?
One possibility is that aika could be older than Baltic, going back to Bal-
to-Slavic. Fi. aita hedge, fence, with attested cognates in all Balto-
Fennic languages but Livonian, is of a similar structure and relevant in
this context. Koivulehto (1973) suggested that aita reflects Proto-
Germanic *aia- oath, mentioning the semantic parallel in Greek
fence, hedge next to oath. Unfortunately he withdrew the
etymology in his 1999 version, but the idea was not bad. The common
semantic denominator would have been something like delimitation,
demarkation or restriction.
The Story of time: The Etymology of Fi. aika 163
Germanic *aia- is otherwise regarded as a Celto-Germanicism

(Hyllested 2010), corresponding to Celtic *oito- (> Old Irish eth) and
usually interpreted as a lexicalization of PIE a walk (< *H1oi-to-) be-
cause of the ON expression ganga ei, but this parallel does not really
hold water since the meaning walk is lost when one removes ganga.
Another possibility is that we are dealing with PIE*heit- fetch, estab-
lished by Tichy (2004) on the basis of Gk. to fetch, take along
and Lat. tor to use and supported by Melchert (2007) who supplies
Cuneiform Luvian izza(i)- fetch (in collocations with a motion verb).
As a semantic parallel we could recall PIE *kag- to hold > Gmc.
*hagan- garden.
BF Gmc. Baltic Slavic IE

*aika *ajuka- *vaika- *vk *Hoiu-go-, *Hoiu-ko-
*ai-h-ta-
*aita *aia- *vaita- *vt *Hoi-to- or *Heit-o-
We already know that PIE *(H)oi- can yield Baltic *vai-, e.g. *oiH-no-s
1 > Lith. venas and *Ho)stro- > Lith. aistra- vehement passion ~ Liv.
aistar ~ dial. vistar pimple, a Baltic loan (cf. both meanings of the
Greek cognate elsewhere in this publication). What is the Slavic reflex of
such an initial diphthong? There seems to be only a single relevant, but
contradicting, example: jdro disease mark on tree ~ in Latv. idra, idrs
which happens to be related to the root of the aforementioned Baltic
word < PIE *Hoid-ro- ~ *Ho)d-tro-. PIE *oiH-no- 1 is only attested in
the zero-grade as jed-in 1 and ino-rog unicorn.
Finnish aika and aita form a curious double pair with Slavic *vk
time and *vt council; convention; oath (cf. also Lith. viet place). I
see two possibilities:
1) Slavic *v- could be the regular reflex of some subset of PIE

*(H)oi-, e.g. *Hoi- (*H being a labial consonant)? This solu-
tion, however, would be based on circular argumentation since
none of the etymologies are properly established. More exam-
ples are needed.
2) Slavic *v- < could have arisen in sandhi of PIE *(H)oi- in cases
when a rounded element proceeded since a rounded element
follows as well. Cf. so-vt convention (Ru. soviet). One weak-
ness of this possibility is that jdro is not an obvious 2nd member
of compounds.
It is important to note that Baltic *vait place and *vaika- child can
come from PIE *(H)oi- without a problem. Could Balto-Slavic *(H)oi-
yield Slavic *v- in sandhi for expected *j-, parallel to regular Baltic
*vai-? It is of course theoretically possible that the Slavic words are bor-
rowings from Baltic, but the conclusion would be the same.
3 Rather PGmc. *ajuka- time *ajuki- eternal
As mentioned above, Liukkonen justifiably characterizes the traditional

etymology aika < Germanic *aiwa- time as phonetically impossible:*-
w- would not be rendered as Balto-Fennic *-k-. However, it has been
overlooked that Germanic possesses a related derivative, PGmc. *ajuki-,
albeit an adjective meaning eternal, which can be reconstructed on the
basis of Old English ce and Gothic ajuk- in ajukds eternity. As al-
ready mentioned in a different context by Weiss (1994: 134-135), *ajuki-
could formally be based on an old thematic noun *ajuka- time; eterni-
ty. PGmc. *ajuka- would have been borrowed as Middle Proto-Fennic
*ajeka, as shown by the following well-known parallels:
Baltic *angurias eel Fi. ankerias

Baltic *perknas oak; the thunder god perkele, a swear-word
and subsequent deletion of the *-e- as in:
Middle Proto-Fennic *kojera > Fi. koira /kojra/ dog1
We would expect the Germanic noun not only to have meant eternity,
vitality, but also life, age, a double meaning detectable from compari-
son with other Indo-European languages. Cf. also the meaning (n)ever
of *h)u kid what(ever) time. contexts contexts.
4 Conclusion
The course of events may be summarized as follows:
1
I thank Petri Kallio for pointing out to me the chronology of these develop-
ments.
The Story of time: The Etymology of Fi. aika 165
a) PIE *hi-u-g()-o- became PGmc. *ajuka-

b) PGmc. *ajuka- time was borrowed to Middle Proto-Fennic and
regularly substituted as *ajeka
c) PGmc. *ajuka- forms the basis of the adjective *ajuki- eternal
(may also have happened before step a) )
d) Middle Proto-Fennic *ajeka regularly becomes Late Proto-Fennic
*ajka
The Fennic meaning may in any case have been affected by contamina-
tion with the inherited lexeme ik age; life; lifetime, as has been sug-
gested to me by Michael Fortescue (p.c.).
References
Aikio 2014 = Luobbal Smmol Smmol nte (Ante Aikio): Studies in Uralic Ety-
mology II: Finnic etymologies. Linguistica Uralica L. 2014, 1: 1-19.
Andersen, Henning: Reconstructing Prehistorical Dialects: Initial Vowels in Slavic
and Baltic [= Trends in Linguistics Studies and Monographs 91]. Berlin / New
York 1996.
Junttila, Santeri, 2012: The prehistoric context of the oldest contacts between Baltic
and Finnic languages. Riho Grnthal & Petri Kallio (eds.): A Linguistic Map
of Prehistoric Northern Europe [= Mmoires de la Socit Finno-Ougrienne
266]. Helsinki: Suomalais-Ugrilainen Seura. Pp. 261-296.
Koivulehto, Jorma, 1973: Germanisch-Finnische Lehnbeziehungen III.
Neuphilologische Mitteilungen 74: 561-609 [reprinted with postscript in Verba
Mutuata. Helsinki: Suomalais-Ugrilainen Seura, 1999: 83-120.
Larsson, Jenny Helena, forthc.: The lexicon of Baltic. Brian Joseph, Joseph Klein
& Matthias Fritz (eds.): Comparative Indo-European Linguistics. An Interna-
tional Handbook of Language Comparison and the Reconstruction of Indo-
European. Berlin: Mouton de Gruyter.
Liukkonen, Kari, 1999: Baltisches im Finnischen [= SUST 235]. Helsinki: Suomalais-
Ugrilainen Seura.
Melchert, H. Craig, 2007: Luvian Evidence for PIE *h3eit- take along; fetch.
Indo-European Studies Bulletin, UCLA 12, 1: 1-3.
Tichy, Eva, 2004: Gr. , lat. tor und die Mittelteile der Duenos-Inschrift.
Glotta 78: 179-202.
Weiss, Michael, 1994: Life everlasting: Latin igis everflowing, Greek
healthy, Gothic ajukds eternity and Avestan yauua living forever.
Mnchener Studien zur Sprachwissenschaft 55: 131156.
Albanian hund nose,
and Faroese, SW Norwegian skon
Finnish kuono snout1
Abstract
Alb. hund nose has no accepted etymology, but the stem corresponds
regularly to Far. skon, Nw. dial. skon snout if these go back to PGmc.
*skuna-. Mod.Icel. skon(n)r, a fem. nick-name, seems to rule out the
alternative reconstruction *skan-. The Balto-Fennic root reflected in Fi.
kuono, Est. koon snout is problematic in the light of the vocalism (the
only parallel of * substituting PGmc. *-u- is ruoste, ruosma rust), but
may nonetheless be a Germanic loan. Former proposals deriving hund
from PIE *skeu- spring forward turn out to be correct, but a derivative
*sku-n- projection (also > Alb. hu penis) must have been formed al-
ready before the emergence of Albanian. The suffixal part, PAlb. *-t > -
d, either reflects an inner-Alb. formation or goes back to the PIE root
extension known from PGmc. *skundjan-, *skundn- drive forward.
1 Albanian evidence for PIE reconstruction
Among Indo-Europeanists today, Albanian has acquired, at least unof-

ficially, a discredited reputation as the more or less useless Indo-
European language branch: it has allegedly retained all too little of the
original lexicon, having replaced many everyday words with borrowings
from (especially) Slavic, Greek, Latin, and Romance; many other inno-
vations belong to the notoriously shadowy Ancient Balkan vocabu-
lary; Albanian often exhibits odd phonological irregularities and aber-
1 The present article was published as Hyllested 2012. Apart from this footnote
(including the reference just mentioned), the abstract and the exact title, the two
articles are identical.
rant derivational patterns; and Albanian lexemes in general are so short

that anything goes in etymology, provided your semantic analysis is
creative enough. In many cases, Albanian forms are mentioned merely
to show the geographical representation of a given lexeme.
This reputation, I would argue, is based on a skewed perception of
the actual state of affairs. It is true that we know comparatively little
about the history of Albanian from its split from the Indo-European
core until its earliest attestation, and the internal history of a language
branch is indeed more difficult to uncover when the branch in question
has few members or only a single one Proto-Albanian must be recon-
structed by dialect material and internal reconstruction (disregarding
the poorly attested candidates for close relatives, such as Messapian).
But Albanian has potential: a fair description would be to say that much
of the lexicon can be defined not as obscure loanwords, but rather as
unexplained. In other words, unheeded archaisms might be hiding even
in the basic vocabulary, waiting for us to give it another try, applying
todays broader knowledge of PIE and the individual branches. One way
forward, which is becoming more and more widespread, seems to be the
inclusion, in Indo-European etymology, of evidence from even more
peripheral languages and dialects. Peripheral, to be sure; but as it turns
out, they may still be extremely relevant and provide crucial infor-
mation about details in PIE reconstruction. This is at least the case with
minor Eastern Iranian languages and Nuristani languages, and in recent
years this has proven to be true not least of all in the case of Germanic
dialects (cf. Kroonen forthcoming). This article is an attempt to solve an
obnoxious etymological riddle by straightforward comparison of forms
in peripheral languages.
2 Albanian hund nose; previous proposals
It is indeed a nuisance for comparativists when lexemes from the very

core of the basic vocabulary defy a generally accepted etymology. One
good example is Albanian hund nose. As a term for a body part, it be-
longs to the basic vocabulary par excellence. Hamp (1965: 130) hinted at
the initial h- as a potential clue to discovery: It belongs to a list of very
basic Albanian words in h- which have consistently resisted etymology.
Since a well-known source of Albanian h- is PIE *sk-, scholars have
tried to search for possibilities among established roots in PIE or just
lexemes in other Indo-European languages with this initial sequence.
Albanian hund, Faroese, Norwegian skon, Finnish kuono 169
Meyer (1891: 153-154) was perhaps on the right track by relating it to

PAlb. *skuna- > Alb. hu penis (Gheg h, gen. hni, Tosk huri pole;
limb), Gk. peg, and Skt. skndate spring forth. Schmidt
(1930: 19) accepted Meyers etymology right away, and it is also em-
braced by lberg (1972) and reappraised by Orel (1998: 152), who recon-
structed for hund a prestage *skun-t, derived from *skuna-. However,
this classical proposal has not won general acceptance because it suffers
from the fact that some of the suggested cognates themselves have dis-
puted etymologies. Beekes (2010: 803) regards Gk. as a sub-
stratum word.2
Pedersen (see Tagliavini 1937: 276) equated hu with Lith. skuj pine
needle instead, leaving hund as simply unetymologized. Bari (1919)
related hund to Lat. senti to feel (via the meanings sense and
smell). abej (1953) connected it with Rum. dial. hud, hudr hole,
crack (which would then be an Ancient Balkan loan of IE origin) and
Skt. kuhara- m. cavity, suggesting an original meaning nostril. Hamp
(1965:126) also prefers a development via nostrils, originally entranc-
es, from a Proto-Alb. *hun- meaning (an) insert, the base of the mod-
ern verb hyj, Gheg hj to enter (of disputed origin). He rejects an oth-
erwise formally possible derivative *skud-V-nt- projection (< *skeud-
shoot) because the Arvanitic dialect of Sophik (Greece) has a crucial
short vowel that rules out such a contraction. Huld (1984) envisages an
Old Alb. metathesis of the PIE nose word: *noh- > *hon- + the suffix -
t, ultimately from PIE *nas- nose; but this would involve both suffix-
ation, the as yet shady raising of *o > u before nasal (perhaps in bung f.,
pl. -a, chestnut oak, Quercus sessiliflora, Quercus petraea)3 and a me-
tathesis which may seem ad hoc, although not unprecedented: cf. the
fact that Alb. hobe catapult, sling seems to be a development of an orig-
inally dialectal bahe id., a singularized plural of the archaic sg. bah, a
loanword from Common Slavic *boj fight (Orel 1998: 14, 150).
2
Cf. Kmmel (2010) on the relatively poor occurrence of safe PIE roots contain-
ing the sequence -ND-.
3
According to Orel (1998: 42), Hamps reconstruction bung < *bug-n finds
support in peng security, pledge < Lat. pignus (also Demiraj 1997: 112-113), but
this is not true if Lat. /gn/ was pronounced [n] (see, e.g., Meiser 1998: 52, 121 on
the details). In that case Alb. -ng- is just the rendering of Lat. -ngn-, written
<gn>, with loss of the final nasal (by assimilation) in such a cluster.
3 The background of Albanian initial h-
Let us begin our own analysis by having a closer look at the different
sources for initial h- in Albanian. Original h- is retained in loanwords
from Slavic (e.g., Alb. hitas to hurry < Common Slavic *xytati, Alb.
hukas to shout < Common Slavic *hukati)4 and Latin (e.g., Alb. her
moment of time, hour < Lat. hra ). But in the case of hund, there is
simply no obvious candidate from any neighboring language (disre-
garding the aforementioned proposal by abej).
It cannot be excluded that Alb. initial h- sometimes reflects a PIE ini-
tial laryngeal. Hamp (1965) reconstructed *h4 > Alb. h- (as in herdh
testicle ~ Gk. id.) while all other laryngeals disappeared. This was
heavily criticized by lberg (1972) and has never been widely accepted,
although it is accepted by Mallory and Adams (1997: 10); and according
to Kortlandt (1998), *h2e- and *h3e- yield Alb. ha- (cf. also Demiraj
1997). Alb. h- may indeed show up in the position of an original laryn-
geal, e.g., Alb. (h)ethe fever < PIE *h2eid-s- and Alb. hut empty, desert-
ed < PIE *h2u-tio- (Goth. aueis barren, desolate). However, the very
occurrence of a laryngeal is not assured in the material in question; and
even if one insists on initial consonants in all PIE roots, the picture is
blurred by a notorious tendency to insert a spontaneous h- before initial
vowel in Albanian:
(h)ark bow Lat. arcus id.
(h)arm weapon Lat. arma id.
hikrr sour milk; buckwheat an inner-Albanian derivative
from ikj to run, go (away)
hok joke, jest Lat. iocus id. (note that secondary -j- and j- in
loanwords is regularly substituted by Alb. h-, e.g., krahin
region, district < SCr. krajina: cf. Rasmussen 1985)
(h)urdhe f.pl. ivy < PIE *urdo- root, wort (OE word thorn-
bush)
(h)urdh pond, pool ~ Common Slavic *vir whirlpool
The rare PIE onset cluster *ks- regularly yields Alb. h- (cf. i/e huaj for-
eign, strange ~ Gk. id. and hirr f. whey ~ Skt. kir- milk.).
Again, there is no obvious candidate available. We are left, finally, with
PIE *sk- and *s-, which merge into Alb. h-, at least before a back vowel
(also in inlaut; see, e.g., Huld 1984: 149, Matzinger 2006: 78):
4
See Svane 1992:256-7.
hal f. chaff < *skol-ieh2 (Goth. skalja, Lith. skeli, sklti to hew;
to split)
harb rudeness < *skor-bo- (OHG scarf, Latv. skarbs sharp,
harsh)
hedh throw ~ hudh hurl < *skeu-d- ~ *sku-d- (ON skjta)
shoot; spring forward
helm poison; disputed, but probably connected to OHG scalmo
plague, W claf sick (cf. Hyllested 2010: 111-112)
hn, Gheg hn moon < *skand- (Skt. cndra- moon, cand-
shine, Bret. cann moon; Lat. cande glow, Gk.
ember)
hi, def. hiri, Tosk dial. h, Gheg h, def. hni ash < *(s)keniso-
(Lat. cinis, gen. cineris id., Gk. dust, Toch. B kentse
rust [sic] < *koniso-)5
hije f. shadow < *sh2i-eh2 (Gk. , Toch. B skiyo) ~ *sh2i-
eh2 (Skt. chy)
humb to leave, to lose, to spoil, to miss < *sku-m-b-, nasal pre-
sent to *skeub- (Lith. skumb, skbti to hurry, to hasten,
Goth. af-skiuban to push away, to reject)
hurdh, hudhr garlic ~ Gk. id.
4 A new proposal
It seems to have escaped everyones attention that several West Nordic

languages possess words with almost identical semantics as well as a
strong formal similarity to Meyers reconstruction: Faroese skon f., -ar, -
5
With PIE *e > Alb. i either by umlaut from -i- in the following syllable (Orel
2000: 145, de Vaan 2004: 70-71) or in secondary consonant clusters such as
oblique forms of s-stems (which are later contracted); cf., e.g., vit year < *uetso-
< *ueteso- (Hamp 1971: 121-122). Meyers (1891: 152) reconstruction *sino-, ac-
cepted by Tagliavini (1937: 312), Huld (1994: 74), and Orel (2000: 131, 218), ren-
ders impossible the otherwise almost universally accepted equation with Gk.
(with o-grade); Lat. cinis does not reflect original i-vocalism, but results
from a vowel assimilation *keni- > /kini-/, as in similis (< *semilis) similar, like.
Alternatively, if one prefers to avoid s-mobile on the basis of Albanian only, one
could derive hi from Early Proto-Alb. *skja (~ Gmc. *skeuja cloud, Eng. sky)
with Early Proto-Alb. > i (~ y) preceding -(C)j- (Orel 2000: 11-12, cf. shi rain
< *sja ~ OPr. suge /su:je/ id., and miz, myz a fly, with dimin. suff. -z, ~ ON
m id. < PGmc. *mja-); but then, in return, one would have to accept second-
ary nasalization of the vowel as in Gheg s eye, dr wood.
ir nose; snout; face; (colloqu.) mug has a counterpart in SW Nw. dial.

(Hardanger, Vosseml, Sogn, Nordhordland; cf. Grunnmanuskriptet6)
skon snout, muzzle, but is otherwise a hitherto obscure and isolated
word, not even represented in ON, let alone East Nordic or the rest of
Germanic. There are two possible Germanic proto-forms:7
a) ~ ON *skn < *skan- (like Far. lon f. -ar, -ir longhouse <
ON ln [~ NE lane] or mon m. < ON mn mane, or
b) ~ ON *skon with a-umlaut of PGmc. *-u- < PGmc. *skuna-,
*skun
The latter finds support in the Mod. Icel. fem. nickname skon(n)r, be-
cause the Norwegian dialect of Vik i Sogn has a similar skon hag, poor
woman (Blndal 1989), and the alternative would result in Mod. Icel.
skn(-).
5 Finnish kuono and its relatives
Furthermore, Jorma Koivulehto suggested in 1982 (see Kylstra et al.

1991-2012, II: 125-126) that the Germanic proto-form of skon (or an early
reflex thereof) had been borrowed into Balto-Fennic, cf. Fi. kuono
6
Grunnmanuskriptet (Norwegian Basic manuscript) is a manuscript dictionary
from the 1930s (completed 1940) which was meant to be released as the first
major dictionary of Norwegian in which both lemma and definition were given
in Nynorsk. Entries were taken from other monumental dictionaries of Norwe-
gian from that period by Ivar Aasen, Hans Ross, Steinar Schtt, and others, but
it does not contain all dialectal forms from minor dialectal dictionaries, which
were meant to be added later. For various reasons, the manuscript was never
published as a complete dictionary; but Dokumentationsprojektet (The Docu-
mentation Project) at the University of Oslo has made the entire manuscript
available in searchable electronic form (see the References below). It now serves
as one of the basic sources for the 12-volume Norsk Ordbok (Norwegian Dic-
tionary), edited at the University of Oslo, which is projected to be finished in
2014.
7
PGmc. *skn- is not an option since this would yield Early ON *skn. In Late
ON, merges with , but not if it was nasalized, in which case it (often) merges
with . Cf. for the whole train of events Proto-Norse *nahtu > Early ON ntt >
Late ON, Mod.Icel. ntt. In Faroese too, nasalized would give , cf. ON vn, pl.
vnir > Far. vn, pl. vnir. Thus Far. skon with short -o- and the pl. skonir pre-
cludes a reconstruction *skn-. The same is true for Nw. skon, as
Grunnmanuskriptet cites the word with the vowel , which can go back to either
ON (< PGmc.*a) or o (< PGmc. *u), but not to ON or .
snout, muzzle, Est. koon, Votic kn id. These forms together point to a
proto-form *knV, with secondary lengthening of PGmc. short *u, as in
Fi. ruoste (Est. rooste) and Karel. ruosma rust (< *Balto-Fennic *rsteh
and *rsma respectively) but this is the only other example of such a
lengthening; otherwise PGmc. *u is substituted with a Fennic short
vowel, mostly *u, but sometimes *o (see Kylstra & al. 1991-2012, I: xviii);
examples:
*hurskas pious, devout (only in North Balto-Fennic; Fi., Karel.

hurskas) PGmc. *hurskaz
*kulta gold (Fi. kulta, Est. kuld) PGmc. *gula
*kuningas king (Fi., Est. kuningas) PGmc. *kuningaz
*lukko, *lukku lock (Fi. lukko, lukku, Est. lukk) PGmc.
*lukn-
*multa humus, topsoil (Fi. multa, Est. muld) PGmc. *mul-
*murkina breakfast (Fi. murkina, Est. murgin(a)) PGmc.
*murginaz
*ruis rye (Fi. ruis, Est. rukis) PGmc. *rugiz
*tureh, *turvas peat, turf (Fi. turve, Est. turvas, dial. turv)
PGmc. *turbaz, *turbz
*kotti bag; scrotum; uterus; trough, etc. (Fi. kotti, Est. kott)
PGmc. *kuan-
*poras degree, step, level; (pl.) stairs: staircase(N only; Fi.
porras, Veps pordaz) PGmc. *buraz
*sorta-a to oppress (as a verb only in North Balto-Fennic; Fi.
sortaa to oppress; Est. srd clearing; margin of a field)
PGmc. *sturtjan-
Secondary lengthening, however, does occur in Germanic loanwords

with front vowel, cf. Fi. viikko < PGmc. *wikn- week and liikki ham
PGmc. *flikkija-, and there is no doubt that ruoste and ruosma must
be substitutions for PGmc. *ruste- and *rusma-, respectively. The three
different outcomes are probably dependent on chronology rather than
phonological conditioning. Thus, it cannot be ruled out that *kuono
comes from *skuna-.
Nikkil (see Kylstra & al. 1991-2012, II: 125-126) connects the Fennic
word with PGmc. *gn (Mod. Icel. gna snout, esp. of seal, shark, or
wolffish) instead, and this etymology is perhaps to be preferred, also
since the -o in Finnish often substitutes a Germanic -stem. This -o,
however, can also represent a Balto-Fennic suffix, and is not present in

all of the Balto-Fennic languages in either case. Both etymologies are in
any case possible.8 I find Koivulehtos proposal that PGmc. *skuna- (via
muzzle) also stands behind Karel. kyn trough, Est. kna, Liv. kin
(Kylstra et al. 1991-2012, II: 137) rather far-fetched.9
6 PIE reconstruction: Formal possibilities
Now, a PGmc. *skuna- snout obviously fits well with Proto-Albanian

*skunt nose, even if there are different ways of accounting for the Al-
banian *-t-. Considering the range of meanings of hund, it would make
sense to maintain relationship with the shoot root:
a) nose
b) point, tip, summit
c) projection, overhang
d) promontory, headland, cape
Affinity with Lith. skutn bald head is formally possible since its ex-
pected Albanian counterpart would be exactly *hund (< *skunt <
*skutn; cf. the metathesis in bung() chestnut oak described above in
2.). Since the Lith. forms and the underlying verb sksti to shave, to
peel are most likely connected to MIr. scoth f. point, edge < PCelt.
*skut, it does not appear semantically impossible either. Note also the
Hesychian gloss , of similar shape, which is given the meaning
head.
It is also conceivable, however, that *skunt simply reflects the
original order of nasal and stop. We know Germanic forms like OHG
scunten, OE scyndan, ON skunda Eng. scoon ( schooner), scun fly
forward < PGmc. *skundjan-, *skundn- drive forward, and these ei-
ther go back to PIE *skund- or Verner variants of *skunt-; when com-
pared to No. dial. skut m. projection, overhang, ON skta kind of ship;
8
Paul Kiparsky (p. c.) has reminded me that original mid vowels are lengthened
in open syllables in Fennic languages; but as a regular development this takes
place much earlier, on the way from Uralic to Fenno-Ugric or Fenno-Permian,
and would hardly affect *-u- in Germanic loanwords in Balto-Fennic, let alone a
later Gmc. *-o- after the operation of a-umlaut.
9
As an alternative, Skt. kua- hole in the ground, pit, etc. comes to mind, but
this is a borrowing from Dravidian.
schooner ( OIr. scta, NE scout, MDu. scte), Lith. skudrs fast, Skt.
cdati drive forward, it is clear that we must assert at least two root-
variants with different stops, hence probably old extensions of an origi-
nal root *skeu-. Nasalized forms may represent generalized nasal pre-
sents. Alb. hund cannot reflect a variant with a voiced stop, which
would be lost in the position after a nasal, yielding hun.
I conclude that hund is ultimately related to Alb. hu penis, hedh
to throw, hudh to hurl, hyj to enter and humb to leave; to lose; to
spoil; to miss as originally suggested by Meyer (and followed by
Schmidt, lberg, Orel, and partly Hamp), but this word-family cannot
be safely established by internal reconstruction alone. It is Germanic
*skuna-, reconstructed on the basis of Faroese and Norwegian material,
as well as possible ancient Germanic forms in Balto-Fennic, that have
provided the clue.
Since a primary word for nose is already known from most Indo-
European languages, and since this word is known to be at least of PIE
age (PIE *nas-),10 it seems reasonable to reconstruct the meaning of
*skun-o-, *skun-to- rather as snout (i.e., nose of an animal [as opposed
to the human nose]), preserved in Germanic and having replaced the
original nose word in Albanian only. Thus *skun-o-, *skun-to- would
be of at least Northwest Indo-European age.
References
Bari, Henrik, 1919: Albano-rumnische Studien I [= Zur Kunde der Balkanhalbinsel.

Quellen und Forschungen VII]. Sarajevo: Institut fr Balkanforschung.
Beekes, Robert, 2010: Etymological Dictionary of Greek. Leiden & Boston: Brill.
abej, Eqrem, 1953: Grupet nd, ng n gjuhn shqipe. Buletin pr Shkencat
Shoqrore 4: 30-38. Tirana.
Demiraj, Bardhyl, 1997: Albanische Etymologien. Amsterdam: Rodopi.
Grunnmanuskriptet.
http://www.dokpro.uio.no/nynorsk/gmanus/gmanussoek_fside.html (accessed
11 July, 2012).
Hamp, Eric P., 1965: Evidence in Albanian. Werner Winter (ed.), Evidence for
Laryngeals.
Hamp, Eric P., 1971: Fils et fille en italique: nouvelle contribution. Bulletin de
la Socit Linguistique de Paris 66: 213-27.
Huld, Martin E., 1984: Basic Albanian Etymologies. Columbus, OH: Slavica.
10
Details of the reconstruction vary, but everyone agrees about the existence of
the etymon.
Huld, Martin E., 1998: Meillets Northwest Indo-European Revisited. Karlene

Jones-Bley & Martin E. Huld: (eds.), The Indo-Europeanization of Northern Eu-
rope [= Journal of Indo-European Studies Monograph 17]. Washington: Insti-
tute for the Study of Man. Pp. 109-125.
Hyllested, Adam, 2010: The Precursors of Celtic and Germanic. Stephanie
Jamison, H. Craig Melchert, and Brent Vine (eds.): Proceedings of the 21st Annu-
al UCLA Indo-European Conference. Bremen: Hempen. Pp. 107-128.
Hyllested, Adam, 2012: Albanian skon nose and Faroese, SW Norwegian skon
snout. Stephanie Jamison, H. Craig Melchert, and Brent Vine (eds.): Pro-
ceedings of the 23rd Annual UCLA Indo-European Conference. Bremen: Hempen.
Pp. 73-81.
Koivulehto, Jorma, 1982: Zur Periodisierung germanischer Lehnwrter im Ostseef-
innischen: Wie findet man germanische Lehnwrter und was sagen sie uns?.
Lecture given at Rijksuniversiteit Groningen, 12 May 1982.
Kortlandt, Frederik, 1998: Reflexes of Indo-European Consonants in Albanian.
Orpheus 8: 35-37 [Reprinted in F. K., Studies in Germanic, Indo-European and
Indo-Uralic, 329-332]. Amsterdam: Rodopi, 2010.
Kroonen, Guus, forthcoming: Etymological Dictionary of Proto-Germanic. Leiden:
Brill.
Kmmel, Martin, 2010: The Distribution of Roots Ending in IE *ND. Paper pre-
sented at The Sound of Indo-European 2 (Opava, November 2010). Forthcoming
in the proceedings volume, Roman Suka et al. (eds.).
Kylstra, A. D., Sirkka-Liisa Hahmo, Tette Hofstra, and Osmo Nikkil, 1991-2012.
Lexikon der lteren germanischen Lehnwrter in den ostseefinnischen Sprachen I-
III. Amsterdam / Atlanta: Rodopi.
Magnsson, sgeir Blndal, 1989: slensk Orsifjabk. Reykjavk: Orabk
Hsklans.
Mallory, James P., and Douglas Q. Adams, 1997: Encyclopedia of Indo-European
Culture. London / Chicago: Fitzroy Dearborn.
Matzinger, Joachim, 2006: Der Altalbanische Text Mbsuame e Krshter (Dottrina
cristiana) des Lek Matrnga von 1592: Eine Einfhrung in die albanische Spra-
chwissenschaft. Dettelbach: Rll.
Meiser, Gerhard, 1998: Historische Laut- und Formenlehre der lateinischen Sprache.
Darmstadt: Wissenschaftliche Buchgesellschaft.
Meyer, Gustav, 1891: Etymologisches Wrterbuch der Albanesischen Sprache. Strass-
burg: Trbner.
Oettinger, Norbert, 2003: Neuerungen in Lexikon und Wortbildung des Nord-
westindogermanischen. Alfred Bammesberger & Theo Vennemann (eds.):
Languages in Prehistoric Europe. Heidelberg: Winter. Pp. 183-193.
lberg, Hermann, 1972: Einige berlegungen zur Laryngaltheorie. An Hand des
Albanischen. KZ 86: 121-136.
Orel, Vladimir, 1985: Review of Huld 1984. timologija 179-182.
Orel, Vladimir, 1998: Albanian Etymological Dictionary. Leiden / Boston: Brill.
Orel, Vladimir, 2000. A Concise Historical Grammar of the Albanian Language.
Leiden / Boston: Brill.
Rasmussen, Jens Elmegrd, 1985: Miscellaneous morphological problems in Indo-

European languages I-II. Lingua Posnaniensis 28: 27-62 [Reprinted in J.E.
Rasmussen, 1999: Selected Papers in Indo-European Linguistics. Copenhagen:
Museum Tusculanum. Pp. 100-144.
Schmidt, Manfred Erwin, 1930: Untersuchungen zur albanischen Lautgeschichte.
KZ 57: 1-42.
Svane, Gunnar, 1992: Die slavische Lehnwrter im Albanischen. Aarhus: Aarhus
University Press.
Tagliavini, Carlo, 1937: Lalbanese di Dalmazia: Contributi alla conoscenza del di-
aletto ghegho di Borgo Erizzo presso Zara. Firenze: Olschki.
de Vaan, Michiel, 2004: PIE *e in Albanian. Die Sprache 44, 1: 70-85. Stockholm
/ Gothenburg / Uppsala: Almquist & Wiksell.
Wiklund, K.B., 1917-1920: Die ltesten germanischen Lehnwrter im finnischen.
Indogermanische Forschungen 38: 48-114. (61-62)
Witczak, Krysztof Tomasz, 2003: Indoeuropejskie nazwy zb. d: Wy-
dawnictwo Uniwersytetu dzkiego.
Witczak, Krzysztof Tomasz, 2004: Albanian kophst garden and its cognates.
Studia Etymologica Cracoviensia 9:179-787.
Estonia and the Aestii:
Baltic Etymology as a Key to Fennic Ethnonyms
Abstract
The name of the Estonians, Est. Eesti, goes back to the tribal name
*Aestii first mentioned by Tacitus, denoting inhabitants of the Baltic in
the broadest sense. Several sources reveal that the name originally had
an -r- in the stem. That the original diphthong was *ai- as in the Lat-
inized forms is assured by the Old Gutnish attestation Aistland. The
name must come from Baltic *aistra- pimple (m.); vehement passion
(f.) which can be reconstructed on the basis of a) a loanword in Livoni-
an aistar pimple and its variant vistar; b) Lith. aistr and c) cognates in
other Indo-European languages with the same double meaning. The
name was given to the Estonians as a translation of PGmc. *finn-. Nu-
merous parallelsexist among Fenno-Ugric ethnonyms, perhaps pointing
to an old designation referring to fish scale as money or fish skin as val-
uable garment.
2 Attestations of the name
The name of Estonia, Est. Eesti, goes back to the tribal name *Aestii first
mentioned by Tacitus in 98 BC (gen.pl. Aestiorum gentes Germ. 45, 2)
and later by Cassiodorus 523-526 BC (H(a)estis Theodoricus rex, Variae
5, 2); Jordanes, d. 552 (gen.pl. Aestorum natio, Get. 23, 119); Einhard in
830 (Aisti in Vita Karoli Magni); Wulfstan in 890 (to, mid stum in his
travel account); and Adam of Bremen in 1073 (Haisti and Aestland,
Hamburgische Kirchengeschichte, 12 and 17). On the basis of these
attestations, we can identify a Latinized ethnonym Aestii or Aesti and a
stem Aest- (Aist-).
According to A. Bammesberger & S. Karalinas (1998), rather than
denoting a specifically Baltic tribe, it was probably a cover term for all
inhabitants of the East coast, including those of Balto-Fennic origin.

They point to the fact that, even if the Aestii of the antique and early
medieval sources denoted Southern Baltic tribes (speaking Baltic lan-
guages), the ethnonym has been used by Scandinavians since the Viking
Age to denote the Fenno-Ugric forefathers of the contemporary Estoni-
ans. Since the 10th century the country and their inhabitants were called
Eistr, more rarely Eistir and Eistland.
That the original diphthong was *ai- as in the Latinized forms (the
only option if the name is inherited into Old Norse, but not if it is late
loan) is assured by the Old Gutnish attestation utan foru i aina oy vir
Aistland, sum haitir Dagaii [they] travelled to an island off Estonia
called Dag (Guta Saga, 13th c.)1. Furthermore, -ai- occurs in ethnonym-
ic compound Tafaistr inhabitant of the Hme region in Finland, used
as a personal name on an 11th c. runestone (U 722) from Lts parish,
Trdgs Hrad in the Swedish region of Uppland: tafaistr lit raisa
stain at a[----- b]rour sin Tafeistr raised this stone in memory of
his brother. Otherwise this name occurs with -ei-, although the modern
Swedish term is tavaster (cf. also Tavastehus the Finnish province of
Hme).
2 Etymology
Bammesberger & Karalinas (1998: 47) further establish on a balance of

probabilities that the name originally meant (agricultural) land; acre
and is derived from Proto-Baltic *ist- to burn2 either via a meaning
dry land (as opposed to the sea), or, more plausibly, referring to the use
of slash-and-burn cultivation techniques. They detect this stem also in
Baltic place-names like Astere and Aisternki in Latvia and Eistrai in
Lithuania (with secondary Ei- for *Ai-), the typically Latvian suffix -ere
of the former showing that the name is indeed of Baltic origin; and they
compare e.g. Lit. dogmN glade; acreage, strip of land ~ dgti to burn;
Lat. terr land < *ters- dry. Their main single argument is that other
names of Baltic and Balto-Fennic tribes are etymologized as land, and
that the traditional Estonian self-designation maarahvas still in the
modern language is literally land-people, maakeel Estonian language
1
I thank Sean Vrieland (p.c.) for having pointed this out to me.
2
Originated from *heid-t- as in Lat. aestus heat, aestas summer and PGmc.
*heid- to burn. For an overview of earlier etymological proposals, see Bam-
mesberger & Karalinas 1998: 46, fn. 1.
Estonia: Baltic Etymology as a Key to Fennic Ethnonyms 181
correspondingly lit. land-tongue, maa-sna genuine Estonian expres-

sion, lit. land-word, etc. Since this term is synchronically analyzable it
is of course a late variant which may nonetheless preserve semantics
and connotations from earlier ethnonyms.
What Bammesberger & Karalinas fail to mention is that this term
contrasted specifically with the name of another Balto-Fennic popula-
tion, the closely related Livonians, who were primarily fishermen and
literally called themselves inhabitants of the coast, rndalist, who spoke
rndak Livonian, lit. coast(al) tongue.
As the authors note themselves, there are indications that the origi-
nal stem contained an -r-: The Old Norse Ynglingatal talks about Eistra
dolgi the Estonian enemy, and in Historia Norvegiae, the name shows
up as Eistriam, Eistriis (in the passage devenit in Eistriam, puer Olavis
Eistriis in servum venumdatur came to Eistria and bought the boy Olav
from Eistria as a slave). They seem to interpret this as an indication that
the ethnonym was borrowed from the suffixated form occurring in the
place-name Astere. However, the suffix -ere is almost certainly not
common Baltic; it is typical of Latvian place-names only and specifically
names connected to lakes. Bga (1923: 383) simply considers it a bor-
rowing from Balto-Fennic *jrvi lake (Fi. jrvi; cf. also Balode & Bus
2007: 37).
If the basis for the full sequence *aistr- is common Baltic, there are
other possibilities, however. Asserting -r- as an original part of the stem
provides a formal basis for comparing it to Lith. aistr intense passion
and Livonian istar pimple, blister; maggot or worm in animal skin;
cockchafer grub. Since Thomsen (1890) the latter has been considered a
loan from the Baltic word behind Lith. anktiri, inkstras pimple, Latv.
ankstiras larva under the skin of cattle. However, there are several
structural problems implied in this etymology. The expected form is
rather something like ahtar since Baltic *-(n)k- normally is substituted
with -h- in Balto-Fennic, cf. e.g. ahingas fish-spear, rake ~ Lith. kti-
nas. One further issue is the Livonian dialectal variant vstar which is
even further away from the alleged Baltic point of departure.
3 Alternative Etymology: Baltic *aistra- is a translation of PGmc.

*finn
I suggest instead that the Livonian word is borrowed from a hypothet-

ical Proto-Baltic *aistra- whose feminine-collective counterpart is at-
tested as Lith. aistr. The former would make up a form identical even
in gender to Gk. gadfly; intense passion , otherwise missing
among the IE cognates, and both meanings of the Greek word would
then also be covered by Baltic. In Late (Western) PIE *o)stro- probably
arose as a merger of two originally independent words:
a) *(h)o)d-tro- one who makes (sth) swell from *he)d- or *a)d-

swell (cf. -d- in e.g. Latv. idra disease mark on tree, Gk. odos
<> tumour, to swell, PGmc. *aita- abscess, ulcer,
*aitra- pus, Lat. aemidus).
b) *he)s-tro- (*ho)s-tro-) irritator, one who sets (sby) in vehement
motion from *he)s- set in vehement motion, urge, irritate 3 (cf.
*-s- in Av. aema- m. anger, Lat. ra id. next to Gk. rush,
attack, rage (of animals)).
Such a merger could have been facilitated not only by the sudden ho-
monymy (after loss of laryngeal and development of *TT > sT), but even
by the semantic connection between stinging insects and swellings on
the skin of animals caused by bugs that operate there.
The etymology proposed here also makes it easier to explain the Liv.
variant vstar as an East Baltic variant *viestra- with prothetic *vie- as
reflex of *(H)oi- as also in *vienas 1 < PIE *iH-no-s.
The comparison can be justified semantically, seeing that many other
Fennic ethnonyms bear strong similarities to appellatives with mean-
ings covered by the PGmc.word*finn-. These meanings are:
1) fish fin (Sw. fena, older fina, MLG vinne, OE finn > Eng. fin)
2) pimple; abscess (Sw. finne, NHG Finne, Da. filipens < finne-
pind)
3) kornaks (Sw. Dial. fen(a) etc.)
4) Nw. dial. finn(e) small horn on animal, stiff grass species
5) any protuberance on the skin of humans or animals related to
diseases, including larvae under the skin, notably on fish (MLG
vinne, Dutch vin, NHG Finne)
6) finne > patch, cf. NHG Flosse fin ~ Da. flosset
I reconstruct the common Proto-Germanic meaning as protuberance

on the skin (including fins and fish scale, larvae under the skin). That
3
Beekes (2010: 1062) regards *he)s-tro- as the only source of .
the meaning 5) of *finn- matches Baltic *aistra- so well is in itself

hardly a coincidence. When compared to homonyms of neighbouring
Balto-Fennic tribes, the equivalence becomes even more striking, cf.
that Fi. Suomi Finland is very reminiscent of suomu fish scale. The
two are normally considered unrelated, but has been suggested as one
possible etymology of the name (cf. Grnthal 1997; Kulonen 1998).
Some of the other names have also been viewed as loan translations
(Napolskich 2007): Lapp ~ Danish lap patch, Veps ~ Saami *veaps
fin, Fi. Vatja Vote ~ Balt. *vadja- patch, the ethnonym sambi, sembi
~ Fi. sampi sturgeon, which is known to derive from an earlier mean-
ing horn core on sturgeon < back-fin of a fish (Napolskich 2003).
In South Moravian dialects of Czech, the word maar means both
Hungarian and pimple, but in this case the primary meaning of
maar is clearly Hungarian, coming from Hung. magyar, meaning that
the signification pimple must originate from a wordplay on the double
meaning of uher. This example thus serves merely as a typological
parallel where the semantic shift went in the reverse direction, from
ethnonym to appellative.
This situation probably reflects several erroneous translations and
folk-etymologies through time, cf. further Old Norse seimenn Saami
people, originally sorcerers, but synchronically also understandable as
pollack men; and ON taf-eistr inhabitant of Hme, lit. tap Estonian ~
latinized in as Historia Norvegiae cornuti fenni, Old Norse horn-finnar.
Correspondingly, *aistra- might simply be the Baltic translation of what
had at some point ended up as the default designation for Balto-Fennic
tribes.
4 Other possibilities than mere translations
Although wordplay, erroneous translations and folk-etymology does

seem to have played a role for the emergence of some of the ethnonyms,
it is on the other hand quite hard to imagine a scenario where these
phenomena took place so consistently and repetitively through the ages
in different geographical areas, and that they would have resulted in a
surviving established ethnonym in every single case. It is thus hardly
credible to present the whole story entirely as one of unconscious and
incidental calques and misunderstandings. It seems at least worth inves-
tigating whether an inherited conceptual association between Fenno-
Ugric peoples and at least one of the meanings involved could be held
accountable for some of the homonymies.
Interesting in this context is Widmers (2003) reconstruction of a
Uralic word for fish scale, *kmV, which after the beginning of minting
(at least after the first half of the 7th c. BC) became used as a term for
coin, especially silver coin, copper coin, but also silver in general,
silver jewel and in Ob-Ugrian folklore even established as an image of
wealth, partly as an attribute to the Urmutter (kam naj wealth(y) no-
ble-lady) , partly of any material making up status symbols of heroes
(weapons, booty). In a version of an Mansi legend, the last word in the
formula oapr-n koam-n silver woman, wealth woman has been re-
placed with a word meaning Russian, pr-n kper-n, and in another
version, even the first word has been replaced with Siberian, pr-n
kpr-n. If some of the Fenno-Ugric peoples had once been named or
named themselves after precious metal, it would not be much different
from deriving trk from the word for silk.
It is also conceivable that a term fish-skin could have constituted an
exact parallel to trk, referring to an important garment. As noted else-
where in this dissertation, Fenno-Ugric forest peoples have traditionally
produced clothing coats, boots, and caps from fish-skin, justifying
an analogy with terrestrial animals hunted for their pelt (Armstrong
1997). The burbot along with the sturgeon and sterlet were so important
for the Khanty (Ostyaks) in the time of Russian expansions in the 1600s
that a particularly bad fishing season could threaten the very existence
of a tribe.
References
Anttila, Raimo & Sheila Embleton, 1994 : On the Origin of Suomi Finland.
Scandinavian-Canadian Studies VII : 25-32.
Balode, Laimute & Ojrs Bus, 2007: Latvian place-names of Finno-Ugric origin.
Ritva Liisa Pitknen & Janne Saarikivi (eds.): Borrowing of place names in the
Uralian languages. Debrecen / Helsinki: Onomastica Uralica. Pp. 27-44.
Bammesberger, Alfred & Simas Karalinas, 1998: Zur Frage nach der ethnischen
Identitt derAisten. Alfred Bammesberger (ed.): Baltistik: Aufgaben und
Methoden. Heidelberg: Carl Winter. Pp. 39-51.
Bga, Kazimieras, 1923: Review of Jnis Endzelns & al. (eds.) : Latvijas vietu vrdi,
part 1 (names from Vidzeme). Tauta ir odis 1: 376397, 441444.
Carpelan, Christian, 1998: Suomi, hme, sabme sek finne arkeologian nkkul-
masta. Johanna Laakso & Riho Grnthal (eds.): Oekeeta asijoo. Commenta-
tiones Fenno-Ugricae in honorem Seppo Suhonen sexagenarii 16.V.1998 [= M-
moires de la Socit Fenno-Ougrienne 228]. Helsinki: Suomalais-Ugrilainen Seu-
ra. Pp. 76-88.
Dcsy, Gyula, 1965: Einfhrung in die finnisch-ugrischen Sprachwissenschaft. Wies-
baden: Otto Harassowitz.
Dcsy, Gyula, 1987: Lappisch eadne und das Suomi-problem. Ural-Altaische
Jahrbcher 59 : 45-48.
Grnthal, Riho, 1997: Liivist livviin : Itmerensuomalaiset etnonyymit [= Cas-
trenianumin toimeitteita 51]. Helsinki.
Jms, T., 1989: Finnish Lappi Lapland and lappalainen a lapp. Ural-Altaische
Jahrbcher 61 : 129-132.
Kallio, Petri, 1998: Suomi(ttavia etymologioitta). Virittj 4: 613-620.
Katz, M. & Koch, Ch., 1987: (J)Ugra. Zur ursprnglichen Gestalt des Ugrierna-
mens. Ural-Altaische Jahrbcher, N.F. 7: 99-104.
Koivulehto, Jorma, 1993: Suomi. Virittj 97, 3, 400-407.
Kulonen, Ulla-Maija, 1998: Benmningar p finsk-ugriska folk. Finsk-ugriske
dager 1995 : Foredragssamling om finsk-ugriske emner. Oslo : Institut for steu-
ropeiske og orientalske studier / Unipub, Akademika AS. Pp. 23-32.
Lindroth, H., 1917: Onomatologiska anmrkningar till T.E. Karstens. Germanisch-
finnische Lehnwortstudien. Namn och Bygd 5: 19-52.
Napolskikh, Vladimir, 2003: Uralic fish-names and original home. Ural-
Altaische Jahrbcher N.F. 12: 35-57
Napolskich, Vladimir, 2004: Der Name der Vepsen und frhere Berhrungen
zwischen Germanen, Lappen und Ostseefinnen. Peter Anreiter, Marialuise
Haslinger & Heinz Dieter Pohl (eds.) Artes et Scientiae: Festschrift fr Ralf-Peter
Ritter zum 65. Geburtstag. Wien: Edition Praesens. Pp. 307-313.
Parpola, Asko, 2004: Old Norse seir, Finnish seita and Saami shamanism. I.
Hyvrinen, P. Kallio, and J. Korhonen (eds.), Etymologie, Entlehnungen und
Entwicklungen: Festschrift fr Jorma Koivulehto zum 70. Geburtstag [= Mmoires
de la Socit Nophilologique de Helsinki 63]. Helsinki. 235-273.
Ross, Alan S.C., 1981 : The Terfinnas and Beormas of Ohtere. London : Viking
Society for Northern Research / University College London.
von Sadovzsky, Otto J., 1995: Fish, Symbol and Myth: A historical semantic recons-
truction [= ISTOR books 6]. Budapest / Los Angeles: Akadmiai Kiad / Inter-
national Society for Trans-Oceanic Research.
Stachowski, Marek, 2002 : Zur Frage der ursprnglichen Bedeutung von finnisch
Lappi Lappland. Studia Etymologica Cracoviensia 7: 151-158.
Thomsen, Vilhelm, 1890 : Berringer
UEW = Kroly Rdei 1988: Uralisches Etymologisches Wrterbuch I-III. Budapest :
Akadmiai Kiad.
Widmer, Anna, 2001: Ural. *kmV Fischschuppe im Obugr., Sam. und Perm.: zur
Benennungsmotivation von Mnze, Metall und Schmuck. Heiner Eichner,
Peter-Arnold Mumm, Oswald Panagl & Eberhard Winkler (eds.): Fremd und
Eigen : In memoriam Hartmut Katz. Wien : Edition Praesens. Pp. 279-291.
Word Migration on the Silk Road:
The Etymology of English silk and its Congeners1
Abstract
European terms for silk display exceptional variation for deriving

ultimately from a single Old Chinese source. It is suggested here that
ON Norse silki silk, transmitted via the Varangians trade in
Byzantium through Kievan Rus to Scandinavia in the 9th c., and OE
seoluc, OHG silihho possibly in a separate wave a couple of centuries
earlier, are loanwords from the Iranian language Alanic (Sarmatian) at
the Western end of the Silk Road. It reflects a regular Alanic
development of *ri > l, the original form being *sirika- which leaves
several possibilities open: Either a) it is a productive inner-Iranian
formation with the suffix *-(V)ka-, possibly meaning silk man,
inhabitant of Silis; b) it is a nativization of Gk. ; or c) it has
entered Alanic via Turkic *sir(e)-lek silk garment or Mongolian sirkeg
silk fabric. Modern Ossetic zly, zldag, despite its typical loanword
characteristics, may be inherited directly from Alanic *silika-, only
reshaped in analogy with zld (today young grass; turf) which must
have preserved its original meaning golden; yellow into medieval Iassic
and preserved with that meaning as a loanword in Hungarian zld.
1 This article will be published in Berit Hildebrandt (ed.): Exchange along the Silk
Roads between Rome and China in antiquity: The Silk Trade. Oxford: Oxbow
2014. Apart from the notations (abbreviations of language names and translit-
erations of Greek words) the two versions are almost completely identical.
1 Silk in Europe and Asia2
The exchange of silk along the Silk Road is often examined through
archaeological finds and the interpretation of historical texts. This
contribution seeks to address the question of exchange along the Silk
Road and especially the ways through which the silk reached the West
by discussing the etymology of silk in Western and Northern European
languages from antiquity to modern times. The chronological scope of
this article is vast in order to better understand the different ntional
concepts through which the material was classified, the different routes
the silk trade could take and the ethnic groups that were involved.
Most etymological dictionaries (notably ODEE 827, de Vries 1962:
487, Vasmer 1953-58, III: 387, Falk/Torp 1960: 966-967) agree with and
basically just repeat the standard etymology of Eng. silk, OE seoloc,
seoluc, sioloc, seolc, and its immediate congeners in Germanic (ON silki,
Da., Nw., Sw. silke3, borrowed into Finnish and Karelian as silkki and
into the Western Saami languages, cf. SKES IV: 10254; OHG silehho
toga, selachin cover), Baltic (Lith. ilkas, ilka, Samogitian dial. silka,
borrowed into Latvian in the expression silkuts sewn with silk; OPr.
silkas) and East Slavic (ORu. lk, borrowed into the New Curonian
dialect of Latvian as ilks and continued in Standard Ru. lk, Belaru.
olk, Ukr. ovk and N Ru. ulk, from where it has been transmitted to
North-Eastern Balto-Fennic languages, and, via Karelian, further into
Eastern Saami5). The Baltic forms are regarded as old loans from Slavic
(Fraenkel 1962-65: 983-984). Schrader (1904-1905: 84) states that the
Germanic forms must have been transmitted via Slavic.
2
I owe my sincere gratitude to Peter Kerkhof, Sen Vrieland, Berit Hildebrandt
and two anonymous referees for their insightful comments and invaluable
amendments.
3
Swedish is one of the very few languages that distinguishes two basic words for
the silk thread (silke) and woven silk (siden). The name of the Silk Road is called
Sidenvgen, thus referring to the fabric.
4
S Saami silke, Pite Saami silhke, Lule Saami silhk, N Saami silki, Inari Saami
silkke (SKES IV: 1025).
5
The Balto-Fennic forms borrowed from Northern Russian are Eastern (Karelian
and Ingrian) Fi. sulkku, Karelian proper and Olonets ulkku, Lude ulk(u), Veps
k (cf. Plger 1973: 190). From Karelian proper comes Skolt and Kildin Saami
olkk, while in Votic, the only Southern Balto-Fennic language to possess the
word, olkka, okk is borrowed directly from Standard Russian. Cf. SKES IV:
1103.
Word Migration on the Silk Road: The Etymology of silk 189
All refer to the ultimate source as Old Chinese, the language of the
area where the silk industry began and became important as early as the
3rd millennium BC (cf. Wang 1993: 225). Most scholars further envisage
that the journey of this word towards the West in some way involves
Written Mong.6 sirkeg silk fabric and Manchu sirge, sirhe silk thread,
silk floss from a cocoon; string of a musical instrument, thus pointing
to the important role nomadic tribes played in the distribution of silk.
Hardly surprising, the standard handbooks also agree that the word was
transmitted in antiquity with the trade of silk fabric along the silk road.
It is a typical wanderwort of the later, historical kind, where the ultimate
origin is at least superficially rather obvious but the ways of
transmission less certain (while, for prehistoric culture-words, even the
source is often obscure). In this particular case, to our benefit, the
transcontinental Silk Road constitutes a concrete historical track on
which we can hope to trace the word and its intermediate stations on its
way from East Asia to Europe and catch glimpses of the transmission of
not only the word, but also the material along the Silk Road.
The Chinese source is the precursor of the Mod.Ch. s silk; thread;
string; it is commonly reconstructed as OCh. *s or *sig (thus Wang
1993 with references) and Middle Ch. *si. It is related to other Sino-
Tibetan words denoting thread, string or sinew. Neighbouring
Asiatic languages all reflect a final r-element, which, if not
reconstructable for Old or Middle Chinese, must be explained as
suffixal in one of the lending languages from which it can have been
transferred further: Middle Kor. sr (> Mod. Kor. shil), Manchu sirge,
sirhe, Written Mong. sirkeg. It is however possible that the -r- does go
back to Chinese and reflect a second noun rn people (Genaust 1996:
578) in which case the word borrowed from neighbouring languages
would not be a designation for silk as such, but rather a compound-like
ethnonym already at the time of contact (whose meaning would
correspond exactly to Gk. silk men; see the next paragraph).
Whatever the exact details of this entanglement, there can be no doubt
about a starting-point in East Asia, as far as its identity as a culture-
word is concerned.
6
Written Mongolian (or Literary Mongolian) is the scholarly term for an inde-
pendent Mongolian language variety, attested in the old Mongolian script and
different from both Classical Mongolian and Middle Mongolian. Although
documented from the Middle Mongolian period, it represents an earlier lin-
guistic stage.
3 The Silk Men
Neither is it at all disputed (e.g. Frisk 1960-66: 697) that Gk.

silken, of silk, substantivated silk and transferred into Latin
as sricus, sricum respectively, is formed from the same root as Eng.
silk. It referred however not only directly to the raw silk and silk fabric,
but also to its oriental origins. The 7 silk men were initially of
unknown ethnicity to the Greeks and Romans because they bought the
silk via middlemen who transported it to the west from trade stations in
Central Asia. Consequently, the land of the Srs acquired legendary
associations until geographical knowledge of the Orient increased in the
first centuries AD (Genaust 1996: 578-579; Thau-Knudsen 2000). At
least Ptolemy knew in his Geographia from 150 AD that their land
stretched beyond the Imauni, i.e. the Pamir mountains in Central Asia.
The Chinese sequence s rn, or whatever exact form was the starting-
point, must have been transferred via Central Asiatic peoples to Europe
and European languages; it cannot have been borrowed directly by
European languages. The question then remains: What kind of Central
Asians?
There is clear evidence that not only Europeans used the silk-term
to denote the peoples of Central Asia, and that the term was not only
used about traders, but about entire populations, again matching the
translation silk people. First, on the famous Tonyukuk monument
from 720 AD, found on the right bank of the upper course of the Tuul
(Tola) river in Mongolia, the third and the fourth line of the Western
side reads trk sir bodun yerint / bod qalmad in the land of the Tur-
kic sir people, no group of people were left (any longer) (Ramstedt,
Gran & Aalto 1958: 30-31), and the term is mentioned again later in the
inscription. Furthermore, the very ethnonym Turk in all probability de-
rives from a translation of this name, cf. that similar words meaning
silk are found in all three Altaic branches: OTu. tork silk fabric,
MMong. turge, torkan, Written Mong. torgan silk, satin, Solon Evenki
trga silk8.
7
Transliterated Sres, but the Latin form is Srs.
8
Cf. also Fi. turkku fur and ORu. *torg marketplace, square, borrowed into
ON torg id. I do not necessarily embrace Wangs (1993) idea that Turk and silk
are ultimately etymologically identical, but in the light of a series of interesting
of loanword proposals showing an alleged loan correspondence *sVlC- vs.
*tVrC- the idea seems at least worth pursuing.
While it seems clear to everyone that the root-element sil- of the

word silk and its congeners must be identical to the ser- in Gr.
and the Srs, and that the source of this element is Chinese, it has not
yet been cleared out how an -ilk-form might have arisen next to the -
rik-form, and why the variant with -l- is found exactly in some Ger-
manic languages, in Baltic and in East Slavic (leaving aside the aberrant
South Slavic forms svila)9. In fact, etymological handbooks in general
seem to avoid explaining the variation. Vasmer (1953-58, III: 387) sur-
mises that the Germanic forms are borrowed directly from Latin, which,
however, does not explain the lambdacism, and that the Old Russian
form, because of its initial consonant - which constitutes yet another
difference from the Mediterranean form, must rather have come from
some unknown Eastern source10. While it is very plausible that the first
intermediate stations were Altaic (Turkic, Mongolian, Tungusic) and
perhaps Tocharian (Wang 1993, cf. also Hilmarsson 1984 on Tocharian
orkm string), pointing to the nomadic peoples of North-Western
China and the Tarim region, no language of these groups preserves a
form with -l- and the meaning silk fabric. What we are searching for is
therefore a language at the Western End of the Silk Road that can have
transformed -r- into -l- and which, at least linguistically, played a crucial
role in the transmission of silk from east to west.
4 A Northeast European isogloss
The geographical distribution of the silk-word in Europe provides some

clues to its further migrations. As can be seen from the above list, it oc-
curs in Germanic languages (from where it has been borrowed into
North-Western Balto-Fennic and Western Saami), in Baltic languages,
and in East Slavic (borrowed into North-Eastern Balto-Fennic, Southern
Balto-Fennic and Eastern Saami). In Western Europe, the successor of
Latin saeta stiff hair, bristle has become the dominant term, however
9
Svila, found in all South Slavic languages (Slovenian, Serbo-Croatian, Macedo-
nian and Bulgarian) is technically unrelated since it formally seems to derive
from a preverb *s- and the feminine of a participial form *vila of the verb *viti
to wind, to roll, to twist, to bend, cf. e.g. Slov. zvila bent (fem.), but this is
likely to be due to folk-etymology (see Bezlaj 1995: 351 with references).
10
Russian - is regular from s- in Nordic loanwords; thus, it seems more likely that
the East Slavic terms have been borrowed from Old Norse (Miller 2012: 66-67).
from the expression saeta srica silk thread (lit. silk hair)11. Thus,
MHG Seide, Du. zijde, Sw. siden, Latv. zds, Est. siid (in the two latter via
MLG, from Latv. further into Liv. as zd) and terms in the Romance
languages (Fr. soie, Sp. and Port. seda, It. seta) are all ultimately, the
Romance forms directly, from Lat. saeta and thus unrelated to the word
silk despite superficial similarities. In other words, the l-form seems to
be confined to Northern and Eastern Europe and the full form involv-
ing both -l- and -k- is certainly only found there12.
During the Viking Age, silk was brought to Northern Europe by the
Varangians, Nordic merchants who traveled through Kiev Rus and
reached Constantinople where they traded with local merchants. One
prominent example from the Icelandic sagas is the account of the
Norwegian king Sigurd the Crusaders visit to Miklagard (Byzantium)
1110, recorded in Snorri Sturlusons Heimskringla (III, 238, cf. Blndal
2007: 136), and from Nestors Chronicle we know how Russian envoys
were gifted with impressive silken brocades during a visit to Byzantium
912 (Krag 2013). Given the geographical distribution of the silk-word
and our knowledge, it is likely that the Varangians brought the word
with them through Russia all the way to Scandinavia.
5 An Alanic sound-law
To sum up so far, although it is undisputed that the Chinese silk word

must be the ultimate source of the several similar European terms for
silk, the reason for the irregular variation between them has remained
a mystery, and the exact languages that transferred them into Europe
via Central Asia have resisted discovery. While inherited words
11
Lat. srica survives only via the Llat. form sareca in OFr. sarge, Fr. serge in the
meaning twill worsted or twill silk, and, via MLat. sarcia and OE s(i)erc, syrc,
s(i)erce shirt in ON serkr (Da. srk, Sw. srk, Nw. serk) undergarment of silk
or flax canvas. Today, the Danish word has either historical or pejorative con-
notations, depending on context; in the latter case, the meaning is less specific
and can refer to any kind of loose garment. The directions of transmission in-
volving ME serk, Mod.Eng.dial. sark, OCS sraky, sraka, sraica garment, Ru.
soroka shirt, id., Lith. rkas garment, and Balto-Fennic forms like Fi. sarkki,
Est. srk, Liv. serk, pl. srkid shirt, is not entirely clear (cf. de Vries 1962: 471
with discussion and references).
12
The exact relation of Pashto sl silk veil and Mod.Pers. sirah to silk as well as
to each other is uncertain (Cheung 2002: 254, Vasmer 387), but cf. below on
other Iranian terms.
generally obey to the regularity of sound-laws, culture-words, because of

their tumultuous history as terms for migrating products and
inventions, typically occur in numerous shapes whose internal relations
are hard and sometimes impossible to disembroil. In the case of silk,
the geographical distribution of the varieties may hold the key to a
solution. We have seen how the -l-variants are found primarily in
Northern and Eastern Europe, notably in Old English, Old High
German, Old Norse, Baltic and Old Russian, while the r-variants have
basically spread from Greek. A scenario where this variant of the silk-
word migrated northwards from Byzantium through present-day Russia
and Ukraine to Central and Northern Europe is compatible with the
historical records. Since Chinese (and Manchu, Mongolian etc.) word-
forms start out with -r-, it seems reasonable to assume that the Greek
form of the word would be the more archaic one. What we search for is
then a language near the Western end of the Silk Road that would
regularly have transformed -r- into -l- before the 7th century AD13 (but
possibly much earlier), and which would be sufficiently culturally
important as to provide or disseminate important culture-words.
Crucially, the regular development of older (Proto-Iranian) *-ri- and
*-ri- in the subgroup of Scytho-Sarmatian languages within Iranian,
comprising Scythian, Sarmatian, Alanic, Iassic and Ossetian, is exactly -
l- (Bielmeier 1989: 241). This can be seen, first and foremost, in the name
of the Alans themselves, Gk. , since it derives from the same
(Indo-)Iranian self-designation *arina- that has yielded the very name
of Iran and the infamous politically misused term Aryan (originally, and
among linguists and historians still, denoting Indo-Iranian, but not
other Indo-European peoples). We know that this name does not derive
from a language that turned r to l in general because a subgroup of the
Alans were called which must reflect PIr. *rauxs-alana-
shining Alans (where *rauxs- in itself comes from PIE *leuk-s- with
regular Iranian development of *l- > r), and, more indirectly, by the
amalgamated name , mentioned by Strabo, where the second
member is the name of the other subgroup, the whose name
derives from PIr. *arua- white (Mod.Oss. urs / ors; Cheung 2002: 7).
At the same time, it known to have been an endonym, i.e. the Alans
name for themselves. A widespread alternative name, originally an
exonym but later adopted as an endonym, was , reflected in the
13
When silk first occurs on the British Isles, simultaneous with the Old English
velar umlaut of *siluc to seoluc, cf. Miller 2012: 67.
name of Iassians, the Iazyges and the Ossetes, with no certain etymolo-
gy.
The PIr. form *ryan- is also indirectly attested as an Alanic
loanword in the NE Caucasian Nakh languages: Chechen la prince (of
a principality), chieftain, Ingush la id. and Tsova-Tush (Batsbi) l
lord, gentleman can be reconstructed with a Nakh protoform *al that,
in all likelihood, was borrowed from Alanic *lan- < PIr. *rian-, cf.
Oss. Allon, a mythological tribal name (Bielmeier 1989: 243; Thordarson
2009). Furthermore, the term occurs with the typical Iranian suffix *-ka-
in Oss. Algat (a Nart tribe) < PIr. *riaka- (Thordarson 1989: 478).14.
Another old attestation of the sound-law occurs in Herodots Scythi-
an name since this most likely comes from PIr. *xwaria-kaia
sun-king. Alkmans shows that the development had
taken place at least in the 7th century BC (Hinge 2005).
One further old example of the development *-ri-, *-ri- > Alanic *-l-
comes from the much later bulk of Alanic (Iassic) loanwords in
Hungarian, transferred via the Iassic settlements in the 13th century.
Hung. zld, zeld means green, unripe and is borrowed from Alanic
*zalda- < PIr. *zarita- (corresponding to Ved. hrita- and ultimately the
same PIE formation as Eng. gold, only with different ablaut). Most
examples of the sound-law are from Modern Ossetian, e.g. the preverb
fl- < *pari-; Iron dialect min, Digor mlun to die < *mria- (the
same IE root as in Lat. mortuus dead, and, as a loanword from French,
Eng. mortal); Digor zld young grass; grass; turf < *zarita yellow (~
Av. zairita-, Ved. hrita- yellowish and, with a different ablaut grade,
Eng. gold), and nl male (Digor originally nl) < *naria- (~ Av.
nairiia male, virile).
This means that if a foreign word containing the sequence *-ri- (or *-
ri-) was borrowed into Alanic or another stage of Scytho-Sarmatian
early enough, it would have yielded *-l- in Alanic itself by regular sound
development. It would have to have happened in antiquity already since
the Alanic self-designation with -l- is mentioned by Strabo and Ptolemy.
According to Pliny the Elder (The Natural History, 6,49), both the
Jaxartes river and the Tanais (present-day Syr Darja and Don
respectively) were called Silis by the Scythians, suggesting a
14
The Sarmatian names and , attested in inscriptions from the
Eastern part of the area North of the Black Sea, are perhaps also developments
of *rya-, but they occur in the Western (Scythian) part as the variants
, [, which may reflect a dialect continuum where the change
into *l was not yet completed in the Sarmatian area (Hinge 2005).
development of *-r- to Scythian -l-. Even though it is far from certain

that the name of Syr Darya has any etymological affiliation to Silis, at
least it shows that Scythian already possessed an l-sound.
6 The Iranian suffix *-(i/a/)ka-
Furthermore, while the Iranian suffix *-ika- formally corresponds to

Greek *-iko-, it became productive in the formation of nouns in Iranian
languages15, most often however in the form *-ka, *-aka (originally
formed to stems in *-- and *-a-) or plain *-ka-, and was often added to
words that would otherwise be or end up as monosyllabic. Thus, if a
word *sir or *ser was borrowed into Alanic before the operation of the
sound-law, or while it was still productive, it is very likely to quickly
have formed part of a new derivative *sirika- which would then develop
into *silka-. It would have stayed as *silk(a)- until it was finally picked
up by Varangians and other traders from the North in Byzantium.
Ciancaglini (2012a: 27-28 and 2012b: 95) notes about the Old Persian
use of the suffix: It seems that it occurs especially in toponyms and
ethnonyms designating non-Iranian peoples, or peoples geographically
distant or little known to the Persians; among the examples she
mentions Karka adj. Carian, Kark Caria where the suffix has been
added to the original stem, but also e.g. Katpatuka- m. Cappadocian
where the final syllable in the foreign name seems to have been
interpreted as the Old Persian suffix. It is at least conceivable that the
word could spread via Old Persian areas with the meaning (the
somewhat remote) silk people even if it was formed in Scythian-Alanic
itself directly to *Sil-i-ka- people of Silis, i.e. the area of Tanais and
Jaxartes. Based on what little we know about Scytho-Sarmatian and
Alanic, we would at least expect a word *silika silk to be homonymous
to an ethnonym (which, though hypothetical, is a possible productive
formation). From the Black-Sea inscriptions we now know *-(a)ka- in
Sarmatian names (Hinge 2005), at least in the variants ,
, , and , representing
Scytho-Sarmatian *(F)liyamanak/gos, formed with this suffix to the Ira-
nian appellative *friia-manah- liebgeistig; having a loving mind16,
15
Except in Old Avestan where it is practically absent due to its low sociolinguis-
tic connotation (Ciancaglini 2012a).
16
Corresponding to Av. friia- dear, Ved. priy- id., Eng. free + Av. manah- soul,
spirit, Ved. mnas-, Gk. .
which, in turn, is attested in Sarmatian and Mod.Oss. lymn

friend. It may even be attested via the Scythian name of the Amazons
mentioned by Herotodus, or (with alternating dia-
critica) if Hinge is correct that is simply a corruption of
* < *aiuauaragta one-breasted (cf. Av. auua- 1 and var-
ah breast).
Another possibility is that a Greek loanword reflecting was
already around in Roman-age Alanic and that the Alanic word was
simply formed on the basis of a Greek model where the Greek suffix -
ik- was interpreted as the Iranian -(i)ka- (cf. the parallel in OP
katpatuka- Cappadocian above). Even if we interpret Alanic *silka- as a
Greek loanword lock, stock and barrel, it could still have been borrowed
early enough for the sound-law to operate17. It is also possible that, after
the completion of the sound-law, new sequences of *-ri- and *-ri- were
so slow to reappear that -l- simply was the realization of a phonemic
sequence /ri/, /ri/ for some time meaning that if the language
absorbed a foreign word with *-ri- or *-ri- it would automatically have
this sequence substituted with phonetic [l]. However, we know that this
is at least no longer the case in Byzantine 13th century Alanic where new
cases of -ri- emerged from an old sequence -raCi-, cf. zrin or zirin <
> yellow, golden (Mod.Oss. zrin) < PIr. *zarania- (Cheung
2002: 104, 254; Engberg & Lubotsky 2003: 43).
A third possibility is that the culture-word indeed entered Alanic
from some other language than Greek, but a language which
nonetheless already contained an unrelated suffix with *-k-, cf. that such
forms are already known in words for silk or thread from both
Mongolic and Tocharian. Turkic in fact possesses a widespread suffix *-
lek (*-lak after back vowels), which is already reconstructed in a textile
term (Lubotsky & Starostin 2003: 261), namely PT *kje-lek shirt
(attested both in Modern Turkic languages and in loanwords in both
Tocharian and Mongolian, and occurring in Altaic cognates with the
meaning silk, string and thread). A Turkic form *sirlek is likely to
have turned into Alanic *silka via *sirlika. Also in this case the word
could either have been interpreted by the Alanic borrowers as a
completely foreign element and assigned a local phonetic rendering
17
A loan from Byzantine Greek could account for the -i-vocalism (Miller 2012:
67), but it is uncertain if this is chronologically compatible with the consonantal
development of *ri to l. At least it would rule out a connection with the province
of Silis because in this name the development of *r to l would have taken place
many centuries earlier.
(*silka), or it could have been reanalyzed by the speakers as a local

formation with an inherited suffix (*sirl-ika > *silka), which is a
perfectly normal development in loanword processes.
In any of these cases, the special Alanic sound-law *-ri- > *-l- plays
an instrumental role. Despite the many uncertainties and possible
subscenarios, it seems likely that the reason for the existence of the -ilk-
form and its confinement to Northern and Eastern Europe is that it
spread here via some stage of Alanic. Whether the derivative is formed
with an Iranian suffix or has been borrowed with the suffix into Alanic
from Turkic is less certain, as is its exact relationship with the Greek
derivative, but it is important to note that the cultural and historical
implications of all three scenarios are basically the same.
7 Golden silk a reflex in Modern Ossetian?
An Alanic *silka- is not directly attested, not even in Iassic or Modern

Ossetian. This is no serious problem since the loss and replacement of
lexical items in a given language is a phenomenon that takes place in all
languages over time. However, it is interesting that the Modern Ossetian
word for silk, while not corresponding regularly with the Alanic
reconstruction, is at least superficially similar and has no known
alternative source. It occurs both as zldag, zldag silk (Abaev IV)
and without the *-ka suffix as zly, Digor izly silk; of silk, silken; silk
scarf (Cheung 2002: 254). According to Abaev, Digor i- here is simply
the definite article although prothetic i- elsewhere can have other
sources (a Proto-Iranian syllable *ia- or the prepositions *ui- and *abi-).
Cases of Ossetic z- normally come from PIr. *z- (e.g. zrd heart <
PIr. *zarita-, zad born < PIr. *zta-; zrond old, old man < PIr.
*zaranta-). On the other hand, there are quite a number of Ossetic
words of unexplained origin beginning with z-. e.g. zaz yew, Taxus and
zyng, zing burning, glowing hot coal; fire. Some of these are in fact
typical culture-words that seem to have had an original s-, e.g. zppaz
elevated or submerged grave-chamber of stone, if this is akin to
Mod.Pers. sabad, Arm. sapat basket (cf. Cheung 2002: 254).
However, we would have expected Alanic *silka- to develop regularly
in Ossetic since Ossetic derives directly from Alanic. One alternative
possibility in the particular case of silk is that zly, zldag() can have
been influenced by the precursor of Oss. zld young grass; grass; turf
(Cheung 2002: 253). This implies that the meaning must still have been
golden at the time of reshaping, and we know that as late as Iassic this
was still the case because the word has been borrowed into Hungarian
as zld with that meaning. Golden would then either have referred to a
particular golden type of silk garment, for example Byzantine
embroidery, or to the golden Muga silk type from Assam, which
reached the West alongside Chinese silk, or simply to the value or the
glistening appearance of silk in general18. Such a contamination or folk-
etymological reshaping could clearly have happened at any time in the
history of Alanic from the time of its absorption of the supposed silk-
term *sirika- till the semantic narrowing of zld from golden into
yellow grass, since apart from the supposed Alanic loanword in
surrounding languages, only the Modern Ossetian output zldag(),
(i)zly is known.
Whether Oss. zly, zldag indirectly continue Alanic *silka- or not, a

hypothetical Alanic form remains the most plausible mediator between
Central Asia and ON silki, not least in view of the East Slavic
attestations Historically, this points to medieval trade routes on the
Northern shores of the Black Sea. While the Alanic shape of the word
seems to have reached Western Europe from at least in the 7th century
AD (judging by the phonological shape of the OE forms) it is possible
that the Alans imported it northwards in several waves, and it might not
have reached Old Norse until the 9th century around the birth of Kievan
Rus and the first attestation of the word in Germanic literary sources.
As can be expected, the export of silk terminology from China
westwards comprises not only the word silk but a range of terms from
this domain whose exact meaning are not always entirely clear (see., e.g.
Ching 2011 on Tocharian kaum and other terms, and Lubotsky and
Starostin 2003: 261 on the cognates of PT *kje-lek). Tracking the exact
routes of such Wanderwrter and their chronology requires minute
phonological and morphological analyses compared with facts from
cultural history.
18
In the tenth-century Old English medical work now known as Balds Leechbook,
jaundice is said to cause the body of the patient to turn yellow like good yellow
silk (geolwa sw g geolo seoluc; Biggam 2006: 3).
Postscript
This article arose from discussions I had with the Danish balkanist Erik
Thau-Knudsen who contributed with an essay on the history of silk
terminology to the Danish National Encyclopedia (Thau-Knudsen
2000) of which I was the editor of etymologies at the time. We did not
reach a satisfactory solution, although Thau-Knudsen hinted at another
Iranian language, Parthian, as the provider. Jens Elmegrd Rasmussen
(p.c.) gave us the tip that Ossetian, with its frequent development of
original *r to l, might have played a role. On May 20, 2000, I presented
the idea at a symposion at the University of Copenhagen (Komparativ
Sprogforskning p Vej) that Alanic might have been the provider not
only of the l-variants, but even the derivative itself, seeing that a) *-
(V)ka- is a frequent nominal suffix in Iranian, b) the development of *r
to l was in fact older than Ossetian proper, and c) this sound-law does
normally not work for Iranian *r alone, but involves a following *-i- that
we also find in the Greek silk-word. I reconstructed the same form
*silika- as Thau-Knudsen did, only for Alanic, and was now able to ac-
count for both morphological and phonological developments on the
basis of our knowledge of Iranian in general and Alanic in particular.
Only at the very end of editing the present article, I discovered that C.
Gary Miller in his recent book External Influences of English: From its
Beginnings to the Renaissance presents an almost identical solution (Mil-
ler 2012: 66-67), in fact with additional details; for example, he notes
that is regular in East Slavic borrowings from Old Norse. He suggests,
too, that Oss. zly derives from *slika- (which he reconstructs with
long *)19, however without explaining the irregularities or mentioning
the important variant zldag. It appears to be time for etymological dic-
tionaries to revise their entries on silk.
References
Abaev, Vladimir I., 1958-1989. Istoriko-etymologiceskij slovar osetinkogo jezyka: I-

IV. Moscow / Leningrad.
Bezlaj, France, 1995: Etimoloki slovar slovenskega jezika, vol. 3 (P-S). Ljubljana:
Izdala Slovenska Akademija Znanosti in Umetnosti & al. (edited by Marko Snoj
& Metka Furlan).
19
On the basis of Byzantine Greek /srikon/; its length is indirectly preserved in Old
Irish sric (via Vulgar Latin *srica).
Bielmeier, Roland, 1989: Sarmatisch, Alanisch, Iassisch und Altossetisch.

Rdiger Schmitt (ed.): Compendium Linguarum Iranicarum. Wiesbaden:
Reichert. Pp. 236-245.
Biggam, Carole, 2006: Old English colour lexemes used of textiles in Anglo-Saxon
England. Graham D. Caie, Carole Hough and Irene Wotherspoon (eds.):
The Power of Words: Essays in Lexicography, Lexicology and Semantics In Hon-
our of Christian J. Kay. Amsterdam: Rodopi. Pp. 1-21.
Blndal, Sigfs, 2007: The Varangians of Byzantium. Cambridge: Cambridge Uni-
versity Press.
Cheung, Johnny, 2002: Studies in the Historical Development of the Ossetic Vocal-
ism. Wiesbaden: Ludwig Reichert Verlag.
Ching Chao-Jung, 2011: Silk in Ancient Kucha: On the Tocharian B word kaum*
found in the documents of the Tang period. Tocharian and Indo-European
Studies. 63-82.
Ciancaglini, Claudia, 2012a: Il suffisso Indo-Ir. *-ka nelle lingue iraniche antiche.
Archivio Glottologico Italiano XCVII, fascicle I: 3-33.
Ciancaglini, Claudia, 2012b: Outcomes of the Indo-Iranian suffix *-ka in Old Per-
sian and Avestan. Gian Pietro Basello & Adriano V. Rossi (eds.): Dariosh
Studies II, Persepolis and its Settlements: Territorial system and archaeology in the
Achaemenid state. Napoli: Universit degli studi di Napoli Orientale (= Asso-
ciazione Internazionale di Studi sul Mediterraneo e lOriente, Series Minor
LXXVIII). Pp. 91-100.
Engberg, Sysse & Alexander Lubotsky, 2003: Alanic marginal notes in a Byzantine
manuscript: a preliminary report. Nartamong: The Journal of Alano-Ossetic
Studies II, 1-2: 41-46.
Falk, H.S. & Alf Torp, 1960: Norwegisch-Dnisches etymologisches Wrterbuch. Oslo
/ Bergen: Universitetsforlaget.
Fraenkel, Ernst, 1962-1965: Litauisches etymologisches Wrterbuch I-II. Heidelberg:
Carl Winter.
Frisk, Hjalmar, 1960-1966: Griechisches etymologisches Wrterbuch I-III. Heidel-
berg: Carl Winter.
Genaust, Helmut, 1976 (3rd ed. 1996): Etymologisches Wrterbuch der botanischen
Pflanzennamen. Basel / Boston / Berlin: Birkhuser Verlag.
Hilmarsson, Jrundur, 1984: East Tocharian orkm peg or string? Die Sprache
30: 16-28.
Hinge, George, 2005 [2006]: Herodot zur skythischen Sprache: Arimaspen, Ama-
zonen und die Entdeckung des schwarzen Meeres. Glotta 81: 86-115.
Krag, Anne Hedeager, 2013: Silkens vej til Norden. SFINX 2013,1: 24-29.
Lubotsky, Alexander & Sergei Starostin, 2003: Chinese and Turkic loanwords in
Tocharian. Brigitte L.M. Bauer & Georges-Jean Pinault (eds.): Language in
Time and Space. A Festschrift for Werner Winter on the occasion of his 80th birth-
day. Berlin / New York: Mouton de Gruyter. Pp. 257-269.
Miller, C. Gary, 2012: External Influences of English: From its Beginnings to the Re-
naissance. Oxford: Oxford University Press.
ODEE = Oxford Dictionary of English Etymology, edited by C.T. Onions with the
assistance of G.W.S. Friedrichsen and R.W. Burchfield. Oxford: Oxford Univer-
sity Press, 1966.
Plger, Angela, 1973: Die russischen Lehnwrter der finnischen Schriftsprache. Wies-
baden: Harassowitz.
Ramstedt, G.J., J.G. Gran & Pentti Aalto, 1958: Materialien zu den alttrkischen
Inschriften der Mongolei. Journal de la Socit Finno-Ougrienne 60: 3-91.
Part I. Die Grabinschrift von Tonyukuk.
Schrader, Otto, 1904-1905. ber Bezeichnungen der Heiratsverwandtschaft bei
den indogermanischen Vlkern (11-36). Part Slavische oder durch Slaven
vermittelte Lehnwrter im lteren Deutsch (29-36). Indogermanische For-
schungen 17.
SKES = 1958-1981: Yrj H. Toivonen, Aulis J. Joki, Reino Peltola, Erkki Itkonen,
Satu Tanner & Marita Cronstedt: Suomen kielen etymologinen sanakirja I-VII.
Helsinki 1958-1981 (= Lexica Societatis Fenno-Ugricae XII): Suomalais-
Ugrilainen Seura.
Thau-Knudsen, Erik, 2000: Silke ordets historie. Etymological essay attached to
the entry silke in Den Store Danske Encyklopdi, vol. 17. Copenhagen: Dan-
marks Nationalleksikon.
Thordarson, Fridrik, 1989: Ossetic. Rdiger Schmitt (ed.): Compendium Lin-
guarum Iranicarum. Wiesbaden: Reichert. Pp. 456-479.
Thordarson, Fridrik, 2009: Ossetic grammatical studies [= Verffentlichungen zur
Iranistik 48]. Vienna.
Vasmer, Max, 1953-1958: Russisches etymologisches Wrterbuch I-III. Heidelberg:
Carl Winter.
de Vries, Jan, 1962: Altnordisches etymologisches Wrterbuch. Leiden: E.J. Brill.
Wang, Penglin, 1993: On the Etymology of English silk: A Case Study of IE and
Altaic Contact. Central Asiatic Journal 37: 3-4. Pp. 225-248.
English mink, Finnish portimo ermine,
and Baltic Fur Trade from Antiquity to the Hanse
Abstract
Eng. mink and its Germanic cognates are loanwords from Baltic where
today in Lithuanian menk means cod, while related Slavic forms de-
note the burbot, a closely related species of freshwater fish. This species
shared its name with the weasel in both Latin (mstla) and Greek
(). In Baltic it is itself a borrowing from Proto-Mari men burbot
which is a Uralic name (cf. Hung. mny-hal burbot ~ mnyt weasel).
This explanation finds support in Kalimas (1936) etymology, deriving
Lith. kas polecat from Mari k, ake weasel. The spread of these
terms are likely to be connected to fur trade around the Baltic. It is fur-
ther suggested that Fi. portimo, dial. porttimo stoat, ermine is a deriva-
tive of a medieval loan from MDu. furet, Late ME forette, ultimately
from Lat. fr polecat, brought to the North Baltic via the Hanseatic
trade.
1 Introduction
Strikingly many Northern European terms for the group of predatory

mammals referred to in biology as mustelids (weasel, stoat/ermine, pole-
cat, ferret, mink, marten, otter, sable, wolverine, badger etc.) have no
established etymologies, e.g.: Eng. stoat ermine; Eng., LG mink, Sw.
mnk, menk, originally European mink; Fi. krpp stoat, ermine; Lith.
kas, Latv. sesks polecat, ferret; Fi. portimo, dial. porttimo stoat, er-
mine and its cognates in Karelian, Lude and Veps; and Nw. jerv wolve-
rine, just to mention a few. It is likely that this situation reflects early
language contacts connected to fur trade along the coasts of the Baltic
Sea from the Roman Ages onwards. This article seeks to cast new light
on a few of terms and their etymologies. New proposals are presented
which have implications for our understanding of this part of European

cultural history and its reflections in language.
2 Lithuanian kas polecat, ferret
Lith. kas, Latv. sesks polecat, ferret were first compared to Ved. (YV)
ka- and the Rigvedic hapax kak0- by Fick (1879: 165) who identified
the correspondence Ved. * ~ Balt. *-- as PIE *. These forms were lat-
er analyzed by numerous scholars (see Kalima 1936: 102-103) as a redu-
plicative formation of the root in kti to shit, i.e. *e-(i)ka-s with typi-
cal loss of -i-. The reconstruction of a PIE form *ke5- (*5e5-) on the ba-
sis of Vedic-Baltic comparison, despite the aberrant initial conso-
nantism, has almost become a generally accepted etymology, mentioned
as the only option in standard handbooks. However, the meaning of
neither Vedic word is decidedly certain (Katz 2002: 303). Equally im-
portant, while numerous kentum reflexes occur in Baltic with regular
counterparts in Slavic and Indo-Iranian, the reverse irregular corre-
spondence Lith. - ~ Indo-Iranian and/or Slavic *k- is at best very rare.
This is understandable since the centum-satem isogloss must have di-
vided the Core Indo-European homeland1 approximately in the middle,
so that the languages spoken in the westernmost fringes of the Eastern
parts absorbed more kentum forms than the other satem languages. One
might ask whether examples of Baltic - ~ Indo-Iranian *k- other than
the one discussed exist at all and cases of Baltic - ~ Slavic *k- where
k- does not precede a sonorant (an environment that would probably
depalatalize it) are close to absent. Thus, if LCS *krm fodder (~ Lith.
erti to feed) really has k- because of the following sonorant, the only
example seems to be Lith. eiv spool < *aiua- alongside Slavic cva <
*kaiua- id..
Thomsen (1890: 223) was the first to etymologize Veps hhk otter
and its Balto-Fennic cognates, Fi. dial. hhk, Olonets Karelian
hehku, Lude hehkine European mink, as a loanword from the Baltic
predecessor of Lith. kas.2 Wichmann (Kalima (1936: 102) instead sug-
gested that the Baltic word was the one that had been borrowed, and
1
Like most scholars today, I use Core Indo-European about the protolanguage
being left behind after the departure of Anatolian and Tocharian.
2
Older Baltic loanwords with * show up with h in Balto-Fennic languages in line
with the fact that Early Proto-Fennic * regularly yields h. *Livonian ssk is a
Latvian loanword.
Mink, portimo: Baltic Fur Trade from Antiquity to the Hanse 205
that Balto-Fennic was the provider. He pointed to the fact that Mari
(Cheremis) has a word k, ak, meaning either mink or otter de-
pending on the dialect some dialects have t-k (the 1st element
being t water), meaning otter only. This word corresponds nicely
with the Balto-Fennic protoform *hhk, and Wichmann saw that the
two must be cognates. His hypothesis was then that the borrowing had
taken place at the Early Proto-Fennic stage, when the form would still
have been *k.3
It has now become an established opinion (see the overview of re-
search in van Pareren 2005, 2008) that Baltic exercised at least a modest
influence upon Mordvin lexicon and toponymy. Thomsen (1890) and
Kalima (1936) thought that even (Proto-)Mari had come into some con-
tact with Baltic. Although Mgiste (1959), much to his own chagrin, was
mostly negative, he did not exclude the possibility completely. Likewise,
the inference that Baltic contributed to the Mari lexicon is implied by a
handful of entries in UEW.
Volgaic languages4 historically stretched further to the West, border-
ing areas inhabited by Baltic tribes. Conversely, Baltic languages
reached much further East, not least reflected in the huge area covered
by the Grand Duchy of Lithuania which extended almost all the way to
Moscow.
Here I would like to propose the further possibility that lexical bor-
rowings within a least one semantic field, animals coveted for their
pelts, even took place in the opposite direction, from Mari to Baltic. Ma-
ri k, ak mink; otter is already known to have been borrowed into
neighbouring Turkic languages (Kalima 1936: 102, Rsnen 1969: 105):
Chuvash ak European mink, Bashkir k id., (dial.) aka mar-
ten, Tatar k, k a water animal. It can still be a Fenno-Volgaic
word with remnants in Balto-Fennic and Mari. But note the important
3
Junttila (2012: 268) nonetheless seems to prefer Thomsens etymology, grouping
it in his category A: Relatively clear etymologies.
4
Most Uralicists now reject that Mari and Mordvin once formed a genetic sub-
group of Uralic. I leave the question open, but in any case it can be used about a
geographical group of Fenno-Ugric languages which must have been subject to
some convergence (e.g. a tendency to dissimilate sequences of two nasals), es-
pecially since the ethnic groups are regarded to have constituted a historical en-
tity. From Kievan Rus we know the names of three more westerly Volgaic
groups, the Merya, the Muromians, and the Meshchyora. The latter survived at
least into the 16th c., judging by Russian chronicles. Curiously, although the five
Volgaic ethnonyms are likely to have at least four different etymologies, they all
begin with M- (except that the non-native name of Mari is Cheremis).
fact that Mari has a similar word, Meadow Mari eke daughter-in-law;
young lady (Paasonen/Siro 1948: 121), Hill Mari daughter-in-law
(Ramstedt 1902: 129); in some dialects the meaning is more specific,
such as the wife of ones son, the wife of ones younger brother, the
wife of ones husbands younger brother and the wife of ones wifes
younger brother (Moisio & Saarinen 2008: 678). The use of words for
bride, daughter-in-law, sister-in-law, or (small) young female about
weasels and other mustelids, historically connected to a symbolism for
pristine, virginal and desired young women, is extremely widespread
across the West Eurasian area (see e.g. Falk 1998: 93-94, Witczak 2004,
Martirosyan 2010: 799-800, Olsen 2010: 16, fn. 17), e.g. Da. brud bride;
least weasel, Mustela nivalis, PGmc. *maru- marten ~ Lith. mart
bride, Latvian mra sister-in-law; It. donnola and Port. doninha
weasel, lit. little lady, Sp. comadreja weasel, little god-mother, Basque
satandre weasel, < *sagut- mouse + andere lady; Gr. weasel ~
sister-in-law, Mod.Gr. weasel; little bride; Hung. hlgy
weasel, bride, Ru. kunica (little) marten; bride (in traditional wedding
rituals), Arm. han-owk weasel ~ harsn-owk little bride and Turk.
gelin bride, dim. gelincik little bride, little young woman; weasel just
to mention a few. This is also the case in the cultures of Western Siberia
from where furs were provided, e.g. the Komi (or Zyryans; Laakso
2005).
This makes it probable that Mari eke daughter-in-law; young lady
is related to k, ak otter, mink, at least indirectly. The former is
likely to be formed as a diminutive of the word for sister, which today
is KB ar, U r, B uar. Two protoforms of this word are recon-
structed for Fenno-Volgaic, *sasare (reconstructable also for the Fenno-
Permian stage) and *sisare or *sesare; the former is regarded as Early
Indo-Aryan or Indo-Iranian, the latter as Baltic, and the Mari word can
come from either of them; the rounded vocalism in the first syllable is
secondary from assimilation to the sibilant (UEW 752, 762). Mari has to
velar diminutive suffixes; - and -ka ~ -(i)k, going back to PU *-kV
and *-kkV respectively. The former is realized as the allomorph -k after
a sibilant (Wichmann 1913-1918: 7-9, 11-13). It is therefore likely that
both k, ak mink; otter and eke daughter-in-law; young lady;
sister-in-law are diminutives of the sister-word, the former of *sasare
and the latter of *sesare, at an early time when the vocalism in the Mari
sister-word had not yet become rounded.
In the following, I will discuss another possible loanword from Mari

in Baltic which ended up ultimately in English as the name of a North
American species.
3 English mink and Lithuanian menk, Sorbian mjenk burbot
English mink, attested as ME menks, mynkes from the 15th-16th c., and its
cognates in Germanic, LG mink otter, Sw. mnk, mink id., are of un-
kown origin. Today the English term denotes a North American animal,
but it was originally used only of the European mink (Mustela lutreola,
German Nertz). Etymological dictionaries typically treat it as an ancient
culture-word connected to fur trade around the Baltic sea, without
speculating further about its origins.
Lat. mstla and Gk. , both have a curious double mean-
ing weasel and burbot, a sweetwater fish species (cf. Schaffner 2006)
with which it shares similarities both in visual characteristics and behav-
ior. While the double meaning in Latin of course could have been taken
over from Greek, connections between the two animals also show up in
Eastern Europe. The Hungarian name for the burbot is mny-hal (hal
fish), lit. weasel-fish, cf. mnyt weasel, mny daughter-in-law;
(OHung.) bride; and a West Slavic name for the burbot is Sorb.
mjenk, Cz. mnk, whose Baltic cognate is Lith. mnk, Latv. mnca,
which was transferred to the salt-water cod, a word that obviously
shares great superficial similarities with our mink.
It thus seems reasonable to establish the Germanic word for mink
as a loanword from a Baltic or a West Slavic language. Since the mink is
often designated by its wet habitat (Lith. audn, cf. Young 2001; Fi.
vesikko, cf. vesi water; Da. flodilder, lit. river polecat) the recurrent
element *min- in Baltic river names comes to mind. But the line of
transmission does not begin in Baltic or West Slavic. The formation
mentioned above is also found in Slovenian menk, menk and derived
from LCS *mn, reconstructed on the basis of Ru. men, Ukr. min, Slo-
vak mie and the rarer Czech simplex me The root man- in SCr. mani,
curiously missing from Vasmer (1953-1958), also fits in (cf. pas dog <
*mn), while the Lechitic derivatives Pol. mitus ( Belaru. mjantz),
Kashubian mitus, as well as Rusyn mnuh, Ru.dial. mentuk ( Moksha
Mordvin mentuk, Erzya Mordvin *mntuk), are less well understood.
These Balto-Slavic forms are normally regarded to constitute a word-
family with PGmc. *muniw (> Eng. minnow) and Gr. small fish,
sprat; a root *menH- with the zero-grade of an i-stem form the basis of
the Germanic, Common Slavic and Greek forms while the Baltic and
West Slavic forms extended with a suffx -k- have full-grade. It is note-
worthy, however, that the Germanic and Greek forms designate specifi-
cally small kinds of fish and could just as well have been formed inde-
pendently from (Pre-)PGmc. *min-u- small and (Pre-)PGk. *mei-n-
id. respectively. The Balto-Slavic forms are terms for a completely dif-
ferent kind of species, overlapping almost completely with Fenno-
Volgaic forms: the aforementioned Hung. mny-hal and Hill Mari
(West Cheremis) men, men-gol, also burbot.
At first glance it appears that Mari simply borrowed their terms for
the fish from Russian men, while the Hungarian word, first attested
1395, would have been taken over from Pannonian Slavic. However,
there are several problems with this seemingly straightforward scenario.
First, the Hungarian vocalism is wrong since the manifestation --
(which is not distinguished from -e- in the standard language) normally
points to back vocalism in the stressed syllable of the lending language,
but this Slavic word only had front vowels. Second, it would be an
amazing coincidence, in the light of other European connections be-
tween terms for burbot and weasel; young lady if the Slavic loanword
meaning burbot just happened to have the same shape in Hungarian as
the inherited word for daughter-in-law; weasel (PU *mi daughter-
in-law; young lady, UEW 276).
Moreover, the Hungarian and Mari words actually correspond to
each other as from a common protoform, and Skolt Saami has manij
(big) whitefish, Coregonus (lavaretus), moanji, moanjiga id. (Col-
linder 1977: 115; SKES 347-348). Since *ekas already constitutes a pos-
sible Mari loanword in Baltic, I find it plausible that the stem *men- in
the meaning burbot; weasel was transferred to from Volgaic to the Bal-
tic Sea region as well. Among other possibilities, it could have happened
from Proto-Mari to Late Common Slavic and (East) Baltic. The suffixat-
ed Slavic forms could either be borrowings from Baltic or parallel for-
mations. From burbot skin Fenno-Ugric forest peoples have traditional-
ly produced clothing coats, boots, and caps justifying an analogy
with terrestrial animals hunted for their pelt (Armstrong 1997). The
burbot along with the sturgeon and sterlet were so important for the
Khanty (Ostyaks) in the time of Russian expansions in the 1600s that a
particularly bad fishing season could threaten the very existence of a
tribe. Novgorod and Moscow exploited the fur resources of the North
for its foreign trade, and luxury furs from mustelids like ermine and
sable were called the gold of ancient Rus (Platonov & Andreev 1922).
Whether the Baltic and Slavic forms are Indo-European or borrowed
from Fenno-Volgaic, it seems clear that the Germanic term for the mink
must derive from the Balto-Slavic fish-name. It is further conceivable
that the term mink spread in Western Europe under the influence of
Du. minneken playful term for a female > Eng. minikin (attested from
the 16th c.).
4 Finnish por(t)timo
Leaving the question of Volgaic loans into Baltic, let us turn to the Finn-
ish name of the stoat, ermine, portimo, dial. porttimo, and its cognates
in Karelian, Lude and Veps. Previous proposals are unsatisfying: Koi-
vulehto (1979) derives it from Fi. porras, gen. portaan step; staircase
etc. with reference to the animals habitat in human houses and farms;
and Liukkonen (1999) imagines a rendering of a hypothetical Baltic
sparteiva swift animal, cf. sparts swift5.
That -ti- fails to assibilate into *-si- speaks for a Germanic rather than
a Baltic origin, because most of the Baltic loanwords in Balto-Fennic
were borrowed before the assibilation. Since the distribution is confined
to the Northern branch of Balto-Fennic, a somewhat recent origin in
Germanic is likely. BF *portti- would be the expected substitution of
Late ME forette, MDu. furet ferret, borrowed from OFr. fuiret <
Vulg.Lat. *furittus, lit. little polecat, derived from fr polecat. The
name could have spread Northwards with the Hanseatic trade between
between the 13th and the 17th century, perhaps even from the end of the
12th century (See Bentlin 2008 for a new account on MLG loanwords in
Finnish). Substituting forett- with portt- would be expected, cf. Fi. per-
jantai Friday alongside MLG vrdach, Old Bavarian pferintag id.
(Bentlin 2008: 156).6
The suffix -imo is not excessively common in Fennic but occurs in
many place-names and a few nouns like tuhkimo ashtray from tuhka
5
Junttila (2012: 268) groups this proposal under his category B: Dubious ety-
mologies, rather than in C: Erroneous etymologies.
6
In fact one may wonder if Eng. polecat is not a folk-etymological reshaping of
the word *poleka that yielded e.g. North Saami buoidaga weasel, brought
Southwestwards with the Hanseatic trade or perhaps earlier in connection with
the fur trade.
ash (Koivulehto 1979; Hakulinen)7. However, -mo here may represent a

Karelian dialectism for original *-ma, cf. another name for a mustelid,
Kar. ohmo, ahmo for Fi. ahma wolverine. In that case, the original dis-
tribution of the word would be confined to the Northeastern Balto-
Fennic languages. This at first glance perhaps renders a Middle Dutch
or Middle Low German origin less likely. One could however also argue
the other way around that a limited geographical distribution is an
argument for a relatively late migration connected to urbanism pre-
historic and early historic vocabulary typically occur with a more scat-
tered distribution. Likewise, the very limited morphological variation
speaks for a historically late introduction of the word. When a non-
native term has been present in a given area for a long historical period,
we typically expect to encounter more kinds of domestic derivatives
formed on the basis of this term, with different nominal suffixes.
References
Armstrong, Peter C.B., 1997: Foreigners, Furs and Faith: Muscovys Expansion into
Western Siberia, 1581-1649. MA thesis from Dalhousie University, Halifax, Nova
Scotia.
Bentlin, Mikko, 2008: Niederdeutsch-finnische Sprachkontakte: Der lexikalische Ein-
flu auf die finnische Sprache whrend des Mittelalters und der frhen Neuzeit [=
Mmoires de la Socit Finno-Ougrienne 256]. Helsinki: Suomailais-Ugrilainen
Seura.
Falk, Harry, 1993: Der Zobel im Rigveda. Gerhard Meiser (ed.): Indogermanica
et Italica [Fs. Helmut Rix = Innsbrucker Beitrge zur Sprachwissenschaft 72].
Innsbruck: IBS. Pp. 76-94.
Fick, August, 1879: Schwa indogermanicum. Beitrge zur Kunde der indoger-
manischen Sprachen 3: 157-165.
Junttila, Santeri, 2012: The prehistoric context of the oldest contacts between Baltic
and Finnic languages. Riho Grnthal & Petri Kallio (eds.): A Linguistic Map
of Prehistoric Northern Europe [= Mmoires de la Socit Finno-Ougrienne
266]. Helsinki: Suomalais-Ugrilainen Seura. Pp. 261-296.
Kalima, Jalo, 1936: Itmerensuomalaisten kielten balttilaiset lainasanat. Helsinki:
Suomen Kirjallisuuden Seura.
Koivulehto, Jorma, 1979: Baltisches und Germanisches im Finnischen: die finn.
Stmme auf -rte- und die finn. Sequenz VrtV. E.F. Schiefer (ed.): Explanatio-
nes et tractiones Fenno-Ugricae in honorem Hans Fromm. Mnchen: Wilhelm
Fink. Pp. 129-164.
7
It is conceivable that the use of exactly this quite rare suffix in a late word for
ermine was furthered by association between tuhkimo and the only other word
in Finnish beginning with tuhk-, namely tuhkuri mink.
Kari, Liukkonen, Kari, 1999: Baltisches im Finnischen [= Mmoires de la Socit

Finno-Ougrienne 235]. Helsinki: Suomalais-Ugrilainen Seura.
Katz, Joshua T., 2002: How the Mole and Mongoose got their Names: Sanskrit
kh- and nakul-. Joel P. Brereton & Stephanie W. Jamison (eds.): Indic
and Iranian Studies in Honor of Stanley Insler on his Sixty-fifth Birthday (= Jour-
nal of the American Oriental Society 122, 2 [special issue]). Pp. 296-310.
Laakso, Johanna, 2005: Our otherness: Finno-Ugrian approaches to womens studies,
or vice versa [= Finno-Ugrian Studies in Austria 2]. Mnster: LIT Verlag.
Mgiste, Julius, 1959: Gibt es im Tscheremissischen baltische Lehnwrter?.
Ural-Altaische Jahrbcher 31: 169-176.
Martirosyan, Hrach, 2010: Etymological Dictionary of the Armenian Inherited Lexi-
con. Leiden / Boston: Brill.
Moisio, Arto & Sirkka Saarinen, 2008: Tscheremissisches Wrterbuch [= Lexica So-
cietatis Fenno-Ugricae XXXII]. Helsinki: Suomalais-Ugrilainen Seura.
Olsen, Birgit Anette, 2010: Derivation and composition: Two studies in Indo-
European word formation [= Innsbrucker Beitrge zur Sprachwissenschaft 136].
Innsbruck: IBS.
Paasonen/Siro 1948 = H. Paasonens Ost-tscheremissisches Wrterbuch, edited and
revised by Paavo Siro [= Lexica Societatis Fenno-Ugricae XI]. Helsinki: Suoma-
lais-Ugrilainen Seura.
van Pareren, Remco, 2005: Baltische leenwoorden in het Mordwiens. MA thesis from
the University of Groningen.
van Pareren Remco, 2008: Die direkten baltischen Lehnwrter im Mordwini-
schen. Finnish-Ugrische Mitteilungen 30-31: 69-147.
Platonov, F.S. & S.I. Andreyev, 1922: Oerk novgorodskoi kolonizatsii severa.
Oerki po istorii kolonizatsii severa. Saint Petersburg: Gosudarstvennoe iz-
datelstvo. Pp. 26-37.
Ramstedt, Gustaf John, 1902: Bergtscheremissische Sprachstudien[= Mmoires de la
Socit Finno-Ougrienne XVII. Helsinki: Suomalais-Ugrilainen Seura.
Rsnen, Martti, 1969: Versuch eines etymologischen Wrterbuchs der Trksprachen
[= Lexica Societatis Fenno-Ugricae XVII]. Helsinki: Suomalais-Ugrilainen Seura.
Schaffner, Stefan, 2006: Lateinisch mstella, mstla Wiesel; Quappe und der
Wortbildungstyp vedisch avatar-. International Journal of Diachronic Lin-
guistics and Linguistic Reconstruction 3: 1-50.
Suhonen, Seppo, 1988: Die baltischen Lehnwrter der finnisch-ugrischen Spra-
chen. Denis Sinor (ed.): The Uralic Languages: Description, History, and For-
eign Influences. Leiden / New York / Copenhagen / Cologne: E.J. Brill. Pp. 596-
615.
Thomsen, Vilhelm, 1890: Berringer mellem de finske og de baltiske (litavisk-lettiske)
Sprog. En sproghistorisk Undersgelse. Copenhagen: Det Kongelige Videnska-
bernes Selskab.
Vaba, Lembit, 1983: Baltische Lehnwrter der Wolga-Sprachen im Lichte neuerer
Forschungsergebnisse. Sovetskoie Finnougrovedenie 19: 138-145.
Vasmer, Max, 1953-1958: Russisches etymologisches Wrterbuch.Heidelberg: Winter.
Wichmann, Yrj, 1911: Zur Geschichte der finnisch-ugrischen anlautenden Affri-
katen bes. im Ungarischen und im Finnischen. Finnisch-Ugrische For-
schungen 11: 173-290.
Wichmann, Yrj, 1913-1918: Beitrge zur tscheremissischen Wortbildungslehre.

Journal de la Socit Finno-Ougrienne 30, 6: 1-42.
Witczak, Krzysztof Tomasz, 2004: Old Prussian moazo mothers sister, mosuco
weasel and related words. Baltistica 39 (1): 131-139.
Young, Steven, 2001: East Baltic terms for wolverine and mink. Linguistica
Baltica 9: 163-167.
Latin and Slavic Loanwords in Hungarian:
Exceptions to Helimskis
Vowel-Harmony Adaptation Rule
Abstract
Eugen Helimski (1998) showed that the accented vowel of High Ger-
man, Latin and Slavic words governs whether they acquire front-vowel
or back-vowel harmony as loanwords into Hungarian. While the Ger-
man material appears exceptionless, some words of Slavic and Latin
provenance exhibit unexpected back-vowel harmony. This article shows
that if a labial sound follows the originally accented vowel, front-vowel
harmony is blocked. The rule applies without exception to both Slavic
and Latin loanwords; it is thus an economical solution.
It follows that variation in Slavic loanwords in Hungarian cannot
serve as a testimony of Old Slavic accent shifts, but merely of the place
of the original (pitch) accent; and that the Slavic language that provided
loanwords in early Hungarian must have been fairly uniform. As for
Latin, it likewise renders an appeal to late accent shifts unnecessary.
Helimski also showed that a subset of Latin words containing a me-
dial cluster *-CiV- could trigger front-vowel harmony even if the origi-
nal accent fell on a back vowel. Here it is shown that the distribution of
front and back vocalism in this type is further governed by the vowel of
the initial syllable. This minor rule possibly applies to Slavic as well.
1 Hungarian vowel harmony and Helimskis rule
Hungarian, like many other Uralic languages, possesses vowel harmony,

which is usually defined as a system of progressive assimilation by
which vowels in the first syllable of a word dictates the vowel quality in
suffixal syllables (inflectional endings, derivational endings etc.). From a
diachronic viewpoint, since the harmony does not only affect suffixes, it
is more correct to say that any given word, derived or underived, must
contain vowels from a certain restricted group, and that it is not neces-
sarily a progressive assimilation it may have started out as regressive,
or even both at the same time, if the trigger occurred in the middle of
the word. Vowel harmonic rules vary from language to language. In
Hungarian, all native words must contain either back vowels (a, o, u, ,
, ) or intolerant front vowels (e, , , , ) while they can always
contain tolerant front vowels (i, , and ).
In a brilliant and important article, Helimski (1992) showed that
original stress in the lending language determines whether loanwords in
Hungarian acquire front (palatal) vocalism or back (velar) vocalim.
As he mentions, there has been no issue regarding the many Turkic
loanwords because they were already subject to a similar vowel harmo-
ny from the outset (and in general consist of vowels that form part of
the Hungarian phonemic system). However, Hungarian also absorbed
many loanwords from languages without vowel harmony since the
Magyar migrations into Central Europe in the 9th century, especially
(unspecified) Slavic, Latin and different stages of High German, all of
which have exported numerous polysyllabic words with back-vowel syl-
lables interchanging with front-vowel syllables into Hungarian. Alt-
hough in a few cases Hungarian ended up with both a back-vowel and a
front-vowel variant, most often there is only one form.
Helimski showed that the outcome was determined by the original
vocalism of syllable carrying the main stress in any of the three source
languages. The rule is completely consistent for High German loan-
words. In this article, I will try to explain a number of loanwords from
Latin and Slavic that seem to deviate from the rule.
2 Loanwords from German
In loanwords from Old High German, Middle High German and Early
Modern High German, the stressed syllable was normally also the first
syllable; however, as Helimski pointed out (1998: 46), this was not al-
ways the case. The dialectal Hungarian word ispotly hospital, bor-
rowed from Middle High German, is important because it was itself a
loanword in Middle High German where it had retained the original
stress on the last syllable (as still in Modern High German Spital):
Latin and Slavic Loanwords in Hungarian: Vowel Harmony 215
Hu. bajor Bavarian (11th c. Paiur) < MHG bayer (Mollay 1982 : 150-
151)
Hu. cukor sugar (1587 : cukor) < Early NHG czukcher, zugher (Mol-
lay 1982 : 150-151)
Hu. (obs.) frstk (about 1395 : feletekum) breakfast < Early NHG
Frstukch, fruestukh (Mollay 1982 : 278-289)
Hu. herceg prince (1201: Herceg) < OHG herzog (Mollay 1982: 308)
Hu. dial. ispotly (Jkai Kdex 1372/1448 hypital-, Spital-; 1527 hyt-
aly) < MHG spitl (cf. NHG Spital)
Hu. kalmr (1301: Kalamar-) < MHG krmre tradesman, mer-
chant (Mollay 1982 : 336)
Hu. (obs.) lbstk (1604 Lobotoc) the plant Levisticum < Early
NHG lbestock, lebestock (< Lat. levisticum) (Mollay 1982 : 389)
Hu. polgr citizen (1229/1550 : Pulgar) < MHG burgre city-
dweller (Mollay 1982 : 444-446)
Hu. pspk bishop (1177/about 15oo Pyspek- ; dial. pspk, pzsbk,
pspk, pispk) < Old Bavarian piscof (Mollay 1982 : 459-465)
Hu. tenyr palm < OHG tenar, MHG tener (Mollay 1982 : 528-530)
Hu. vnkos pillow < MHG wangechusse, wangkss
3 Loanwords from Latin
Helimski further showed that in Latin loanwords the same rule applied,
however only in such loanwords where vocalic harmonization was not
blocked because of their bookish character and influence from the litur-
gical tradition. The affected words comprise many proper names. The
following heterovocalic Latin words acquire back-vowel harmony:
Hu. PN Adorjn < Lat. Hadrinus (Fludorovits 1930: 17, 44)

Hu. angyal angel < Lat. ngelus (TESz)
Hu. PN Damjn < Lat. Daminus (Fludorovits 1930: 17, 44 ; see be-
low on the variant Dmjn)
Hu. kamara chamber, kamra storeroom < Lat. cmera (Fludoro-
vits 1937: 41, TESz)1
Hu. kanonok (about 1405: kananok) clergyman < Lat. cannicus
(TESz)
1
This word was probably borrowed via OHG kamara, kamera (Mollay 1982: 337-
339)
Hu. PN ibolya (1340 iwola) violet < Vulg.Lat., LLat. vila <
Class.Lat. vola (TESz)
Heterovocalic Latin words that aquire front-vowel harmony are more

numerous. They comprise words with original stress on a front-vowel:
Hu. PN Elek (1219/1550 Elexius) Lat. Alxius (Fludorovits 1930: 9)

Hu. PN Erzsbet Lat. (?) Elsabeth or Elisabth (Fludorovits 1930:
9)
Hu. PN Ferenc shortened form of Lat. Francscus (Fludorovits
1930: 9)
Hu. flemle, flemile (1395 filemyle) nightingale Lat. philomla
(TESz)
Hu. PN Lrinc (1138/1329 Leurenci) Lat. Laurntius (Fludorovits
1930: 12)
Hu. mise (1372/1448 miet acc.sg.) mass Lat. mssa (TESz)
Hu. petrezselyem parsley Lat. petroslium (added by me; cf. Benk
& Imre 1972: 187)
Hu. sekrstye (1510 ekretye-) sacristy Lat. sacrstia (TESz)
Hu. szerecsn (1138/1329 Scerecn) Saracen Lat. Saracnus (TESz)
Hu. zsllye (about 1510: sellye) armchair (obs.); stall (in a theatre),
orchestra seat, parterre seat Lat. slla seat, chair (TESz)
Latin words with a medial cluster containing a palatal element in it, i.e.
words containing a medial structure -CiV-, occur mostly with front vo-
wels even if the originally stressed vowel (and many other vowels of the
word) is a back vowel. In such cases variation sometimes occurs, and in
the case of Adorjn mentioned above no front-vowel variant is known:
Hu. PN Adorjn Lat. Hadrinus (Fludorovits 1930: 17, 44)

Hu. PN Cerjk, Cirjk Lat. Cyricus (Fludorovits 1930: 7)
Hu. PN Dmjn, variant of Damjn Lat. Daminus (Fludorovits
1930: 17, 44)
Hu. PN Sebestyn Lat. Sebastinus (Fludorovits 1930: 7)
Helimski does not ascribe the variation in these four individual cases to
any kind of regularity. It is noteworthy, however, that the first vowel in
both Hadrinus and Daminus are back-vowels, leading to back voca-
lism (Adorjn, Damjn) with a front-vowel variant in one of the cases
(Dmjn), while the first vowel of Cyricus and Sebastinus are front
vowels, leading to front-vowel forms only (Cerjk/Cirjk, Sebestyn).

Although the material is limited we may tentatively assume that if a La-
tin word contained medial consonantal -i-, it will always end up having
front vocalism in Hungarian (regardless of the quality of the stressed
vowel) unless both the original first vowel and the original stressed vo-
wel are back vowels.
Helimski ran into two real exceptions which exhibit back vocalism
although the stressed vowel in Latin was a back vowel2:
Hu. dzsma tithe Lat. dcima

Hu. tgla brick Lat. tgula
He suggests (p. 50) that the high degree of compatibility of the harmo-
nically tolerant vowel with the back vowels overweighed its palatal
character. While these two words are indeed the only examples con-
taining an open-syllable -e- in Latin, Helimski does not formulate his
suggestion as arule, in fact he exactly counts these examples as excep-
tions; and it is still a bit difficult to account for the fact that they did not
end up as dzsme or tgle respectively. Below we will see how another
rule must have applied which can also explain similar exceptions within
the bulk of Old Slavic loanwords.
4 Loanwords from Slavic
Slavic loanwords into Old Hungarian poses difficulties to a higher ex-

tent. First of all, we do not know all details of the exact language which
was the main or perhaps even the only provider of Slavic loanwords, in
all probability the Pannonian dialect of Late Common Slavic (Helimski
50). It is therefore crucial that we can find clear patterns in the material.
Helimski shows that by using the same indicator, the accented syllable
of the Slavic word, which is generally preserved in Russian (though as
stress rather than pitch accent), we can see that, again, the place of ar-
2
As he notes, other exceptions are actually not direct borrowings from Latin, but
have been taken over from Old High German, as in the case of monostor mon-
astery where the immediate source was OHG munusturi, munsturi and Latin
monastrium only the ultimate source. This is most likely also the case of several
proper names like Mrton (< Mrtin rather than Lat. MartInus) and goston (<
ugustin, not Lat. AugustInus).
ticulation of the stressed vowel in the source language governs the vowel
harmony of the Hungarian word:
Hu. acl steel (not ecl) LCS *cl; Kniesza 1974: 59-60)
Hu. csahol long shirt (not csehl) LCS *exl, cf. Ru. exl co-
ver)
Hu. csata fight; (obs.) military team (not csete) LCS *eta, cf. Ru.
et match, pair
Hu. csnak boat (not csnek) LCS *lnk, cf. Ru. elnk small
boat
Hu. csorda herd (not csrde) LCS *erda > rda, cf. Ru. ered
turn, order
Hu. cstrtk Thursday (not csutortok) LCS etvrtk, cf. ORu.
etvrtk (changed in Mod.Ru. etvertk four pieces)
Hu.dial. deget wheel-grease (not e.g. dagat) LCS *degt, cf. Ru.
dgot tar, pitch
Hu. ebd dinner (not abd) LCS *obd (Kniesza 1974: 166-167)
Hu. kalapl to hammer (not kelepl) LCS *klepa-, cf. Ru. klept
to rivet
Hu. morotva old riverbed (not mrtve) LCS *mrtva, cf. Ru.
mertv dead (fem.)
Hu. sztn stimulus, drive (not oszton) LCS *ostn, cf. Ru. os-
tn id.
Hu. rosta sifter, sieve (not e.g. rest) LCS *reeto, cf. Ru. reet
id.
Hu. szalonna lard, bacon (not e.g. szelnne) LCS *solnn >
*slann (salted) lard (Kniesza 1974: 487)3
Hu. szelnce casket, small box (not e.g. szalonca) LCS *solnica
box for salt, salt-cellar (Kniesza 1974: 495-497)
Hu. szrda Wednesday (not e.g. szarda) LCS *serda > *srda, cf.
Ru. sred id.
Hu. dial. (and obs.) szosztra ~ szesztra junior nun (not e.g. szsztre
or szesztre) < LCS *sestra, cf. Ru. sestr sister
Helimski finds the following exceptions which he notes might be ex-

plained by Pannonian accent shifts shared by Kaikavian Serbo-Croatian
dialects of Slavonia (a suggestion by Sergej Nikolaev and G. Zamjatina):
3
The same Slavic word entered Hungarian later via Romanian slanin, when the
adaptation rules had stopped working, and became szlanina (Bakos 1977: 132).
Hu. jrom yoke LCS *jarm, cf. Ru. jarm id.

Hu. szilva plum LCS *sliva, cf. Ru. slva id.
There are also harmonic doublets of Slavic origin which are tentatively
ascribed to accentual doublets in Slavic:
Hu. csalad family ~ cseled domestic, servant LCS *eljad

Hu. ketrc ~ dial. katroc cage LCS *kotarc
Hu. varsa ~ dial. verse, vrse fishweir LCS *vra
respectively. Below we will see how another rule must have applied
which can also explain similar exceptions within the bulk of Old Slavic
loanwords.
5 Explaining the exceptions
The exceptions all contain a labial sound following the original stress
which could have triggered a strategy for nativization dominated by
back-vocalism. Perhaps Latin -l- also had this effect since Lat. -u- is like-
ly to represent a centralized epenthetic vowel:
Hu. dzsma tithe Lat. dcima

Hu. tgla brick Lat. tgula
Hu. jrom yoke LCS *jarm, cf. Ru. jarm id.
Hu. szilva plum LCS *sliva, cf. Ru. slva id.
At the same time, none of the regular examples contained an original

labial. This may then also provide an alternative explanation for
Hu. PN ibolya (1340 iwola) violet < Vulg.Lat., LLat. vila <
Class.Lat. Vola
where we do not have to appeal to a Late Latin or Vulgar Latin pronun-

ciation, even if it did apply, because our rule works perfectly even with
the Classical Latin pronunciation.
The rule seems to be exceptionless. In Helimskis material I can find
the following other words with a (non-initial) labial, one from Slavic
and one from Latin ; in both cases you have front-vocalism because the
source word had an accented front vowel, and this is allowed because
the labial sequence precedes, not follows, the accented vowel.
Hu. ebd dinner (not abd) LCS *obd

Hu. flemle, flemile (1395 filemyle) nightingale Lat. philomla
Another example might be added, namely
Hu. bazsalikom (not e.g. bezselikem) Lat. baslicum
where front-vowel harmony, which should otherwise occur because of

the Latin stressed front-vowel -i-, is blocked by virtue of the labial se-
quence -um4. Another herb name, not mentioned in Helimskis article
either, ish
Hu. petrezselyem parsley (not patrazsalyam) Lat. petroslium
The latter does have front-vowel harmony throughout the word despite
the labial element because here our other rule of medial palatal se-
quences -CiV- triggers it. The two words are further distinct by their
place of vowel articulation in the first syllable which may have played a
role at least for bazsalikom in the sense that it could have contributed to
add up the factors in favor of a back vowel.
With this rule we can even get solved where the stress was in the La-
tin form of Elisabeth ; it is now clear that it must have been on the last
syllable because it could not have preceded the labial -b- which would
have resulted in a different vowel harmony with at least one back vowel
substituting the -a-. This may even be applicable if the name was bor-
rowed via another language than Latin. Some Slavic languages have
stress on -bt- (e.g. Serbian jelisavta) but the Hungarian form shows no
trace of the feminine ending a that has been added to the name. Thus,
in a revised version:
Hu. PN Erzsbet Lat. (?) Elisabth (not Elsabeth)
4
This example was suggested to me by Sen Vrieland.
6 Conclusion
I have tried to demonstrate that the accented vowel of a word in the

source language governs whether a it ends up in Hungarian with front-
vowel or (predominant) back-vowel harmony to an even more regular
extent than discovered by Helimski. He inferred that exceptions and
variation within the Slavic material may be due to accent shifts reminis-
cent of those known from neighbouring Kajkavian dialects and is a po-
tentially useful source for research in Balto-Slavonic accentology. As for
the Latin material, he ascribed exceptions to the back-vowel tolerance
of the neutral front-vowel in Hungarian itself, and to accent shifts
from Classical to Late and Vulgar Latin.
I hope to have shown that all these exceptions can be explained by a
rather simple conditional rule of adaptation: If the accented vowel of the
source language, whether (Pannonian) Slavic or Latin, was followed by a
labial sound and/or -l-, front-vowel harmonization was blocked, and
original vowels were substituted with back vowels and neutral front-
vowels only. Although it may disappoint Slavicists who could wish for
Hungarian as yet another testimony of old Slavic accentuation devel-
opments, I find it much more economical to embrace a solution that
explains the Slavic and Latin irregularities by the same exceptionless
rule.
With less certainly, we may also set up adaptation rules for a subset
of Latin words (incidentally, all proper names) with a medial palatal
cluster -CiV-. Helimski showed that this type mostly turns out with
Hungarian front-vowel harmony even if the originally stressed vowel
(and many other vowels in the word) was a back vowel. Again, he had to
deal with variation and exceptions, but he simply deduced that vacilla-
tion between front-vowel harmony and back-vowel harmony was the
result of two conflicting tendencies pulling in different directions. I
pointed out that, although based on scarce material, there are at least no
exceptions to a rule by which Latin words with a palatal medial cluster -
CiV- acquires back-vocalism only if both the stressed vowel and the first
vowel of the source word has back vocalism. Without these two trigge-
ring factors working together, front-vowel harmony overrules the de-
fault law.
Finally, I have suggested that harmonic doublets in the Slavic mate-
rial, although potentially reflecting Slavic accentual doublets, may partly
reflect the same tendencies, cf. that e.g. csald family vs. cseld domes-
tic, servant is the only example with a medial structure *-CjV- that is
similar to the Latin structure *-CiV- causing variation above.
The rules can be formalized this way, where VH represents vowel har-
mony and SL stands for source language:
VH[+front] if SL has V[+stress][+front]

VH [+back] if SL has V[+stress][+back]
References
Andersen, Lloyd B., 1980: Using asymmetrical and gradiant data in the study of
vowel harmony. Robert M. Vago: Issues in Vowel Harmony. Proceedings of
the CUNY Linguistics Conference on Vowel Harmony, 14th May 1977 [= Studies
in Language Companion Series (SLCS) 6]. Amsterdam: John Benjamins.
Bakos, Ferenc, 1977: Les lments roumains du lexique hongrois et quelques pro-
blmes de lemprunt linguistique. Acta Linguistica Academiae Scientiarum
Hungaricase 27 (1-2): 111-160.
Benk, Lornd & Samu Imre, 1972: The Hungarian Language [= Janua Linguarum
Series Practica 134]. The Hague / Paris: Mouton.
Fludorovits, Joln, 1930: Latin jvevnyszavaink hangtana [= A Magyar
Nyelvtudomnyi Trsasag kiadvnyai 26]. Budapest.
Greenberg, Marc L., 2004: Sifting the evidence for the reconstruction of Panno-
nian Slavic. Review of Richards 2003. Canadian Slavonic Papers / Revue
Canadienne des Slavistes 43,1/2: 213-220.
Helimski, Eugen, 2003: Slavic/Latin/German Stress and Hungarian vowel Harmo-
ny. Lszl Honti, Sirkka-Liisa Hahmo, Tette Hofstra, Jolanta Jastrzbska &
Osmo Nikkil (ed.) : Finnischugrische Sprachen zwischen dem germanischen und
dem slavischen Sprachraum. Amsterdam / Atlanta: Rodopi. Pp. 45-54.
Kniezsa, Istvn, 1974: A magyar nyelv szlv jvevnyszavai I. Budapest: Akadmiai
Kiad.
Laakso, Johanna. 2006. Review of Richards 2003. Finnisch-Ugrische Forschungen
56: 254.
Mollay, Kroly, 1982: Nmet-Magyar Nyelvi rintkezsek a XVI. Szzad Vgeig. Bu-
dapest: Akadmiai Kiad.
Nuorloto, Juhani, 2010 : Central Slovak and Kajkavian Structural Convergences :
A Tentative Survey. Slovo 50: 37-45.
Richards, Ronald O., 2003 : The Pannonian Slavic Dialect of the Common Slavic
Protolanguage : The View from Old Hungarian [= UCLA Indo-European Studies,
vol. 2]. Los Angeles.
TESz = A Magyar nyelv trtneti-etimolgiai sztra I-III. Budapest 1967-1976.
Vishogradska, Ina: Vowel harmony in the early borrowings from Slavic to Hunga-
rian: Proof of root vowel harmony? Ms.
English Summary
Word Exchange at the Gates of Europe:

Indo-European and Uralic languages dominate present-day Europe,

but both families are newcomers which replaced most of the indigenous
languages step by step from the Bronze Age onwards. The encounter
between indigenous and instrusive cultures, however, was most certain-
ly not the only interaction that took place. By the time of arrival in Eu-
rope, the Indo-European and Uralic populations had already broken up
and constituted a patchwork of languages and cultures that continued
the process of convergence and interchange. Whether contacts were
connected to trade, war, social interaction, or exchange of inventions is
revealed by the character of the loanwords in each individual case
while the shape of the loanwords expose the time depth and the direc-
tion of borrowing.
Traditionally, scholars have thought that basically all loanwords be-
tween Indo-European and Uralic languages went in one direction
from the former to the latter. Such an asymmetry is supposed to reflect a
past relationship between two peoples where one had the upper hand,
technically and politically, at the time of borrowing.
In this dissertation it is shown that cultures of the Northeast played a
surprisingly important role in the shaping of our continent from prehis-
toric to Medieval times; and it is shown how these circumstances are
reflected even in the vocabularies of modern European languages.
The Indo-European tribes, shortly after their migrations into Euro-
pe, came to form part of new cultural communities, influenced by Ura-
lic populations from the North. This had a significant impact on specific
parts of the vocabulary, notably terms for religion and warfare. Many
trade terms (such as Danish pung purse), and words for tools (e.g.
hammer) and religious concepts (e.g. hell) originate from Fenno-Ugric
and other languages spoken in Northeastern Europe at the time. Even

our word half can be shown to derive from an old Fennic trading term
meaning reduced, cheap (of prices).
Some terms denote animals hunted for their pelt (e.g. mink) and
were exchanged in connection to centuries of fur trade along the Baltic
coasts from the Roman Ages to the Hanseatic period. Other words for
animals, among them quite a few used for pigs and boars, are, quite
astonishingly, much older loans going back the pitted-ware culture
around 3000 BC. Some loanwords show, for the first time, that
(Proto-)Celts and Fennic peoples must have been in direct contact with
each other.
Dansk Resum
Ordudveksling ved Europas porte:

Fem rtusinders sprogkontakt
Indoeuropiske og uralske sprog dominerer i dag Europa, men begge

sprogfamilier er historisk set nytilkomne, som skridt for skridt erstatte-
de Europas oprindelige sprog fra bronzealderen og frem. Mdet mellem
de indfdte og indtrngende kulturer var dog helt sikkert ikke den ene-
ste interaktion, der fandt sted. Da indoeuropisk- og uralsktalende
stammer nede til steuropa, var de allerede splittet op og udgjorde et
kludetppe af sprog og kulturer, som vedblev at vre i kontakt og p-
virke hinanden p kryds og tvrs. Hvilke historiske situationer, de en-
kelte sprogkontakter reflekterer om der var tale om handel, krig, social
interaktion eller udveksling af opfindelser afslres af lneordenes ka-
rakter. Samtidig kan deres form afslre, hvornr og i hvilken retning
lnet fandt sted.
Traditionelt har sprogforskerne ment, at nrmest alle lneord mel-
lem indoeuropiske og uralske sprog var lnt netop fra indoeuropisk
til uralsk stort set aldrig den anden vej. Man regner med, at sdanne
asymmetrier afspejler fortidige relationer, hvor folkeslaget med det ln-
givende sprog var teknologisk og politisk overlegent p lnetidspunktet.
I denne afhandling pvises det, at nordstlige kulturer har spillet en
overraskende vigtig rolle for udformningen af vores kontinent fra forhi-
storisk tid til langt ind i middelalderen, og hvordan disse forhold stadig
afspejles i moderne europiske sprogs ordforrd.
Kort tid efter at de indoeuropiske stammer ankom til Europa blev
de hver isr en del af nye kulturelle fllesskaber, pvirket af uralske
folkeslag nordp. Dette fik stor indflydelse p bestemte dele af ordforr-
det, isr termer inden for religion og krigsfrelse. Mange handelstermer
(fx pung), ord for redskaber (fx hammer) og religise termer (fx helvede)
stammer fra finsk-ugrisk og andre sproggrupper fra datidens nordst.
Selv vores ord halv kan pvises at g tilbage til en gammel finsk handels-
term, der betyder reduceret, billig.
Nogle af termerne har med pelsdyr at gre (fx mink) og er udvekslet
i forbindelse med den pelshandel, der foregik langs stersens kyster i
rhundreder fra romertid til hansetid. Andre dyretermer, fx en lang
rkke ord for grise og svin, er overraskende meget ldre ln, der har
forbindelse mske helt tilbage til den grubekeramiske kultur omkring
3000 f.Kr. Nogle lneord viser for frste gang, at (ur)keltere og finske
folk m have vret i direkte kontakt med hinanden.

Ph.d. 2015 Adam Hyllested

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ph.d. 2015 Adam Hyllested

Uploaded by

Copyright:

Available Formats

university of copenhagen

Word Exchange at the Gates of Europe

Citation for published version (APA):

Download date: 29. Jun. 2016

Word Exchange at the Gates of Europe

Word Exchange at the Gates of Europe

The present dissertation is the result of my 3-year Ph.D. scholarship fi-

sis but which proved to complicated for me to able to reach a satisfacto-

The study of early lexical exchange between Indo-European and Uralic

European is automatically assumed to be the provider and Uralic the

Jorma Koivulehtos claim that PIE laryngeals in word-internal position

1 Frozen PIE laryngeals in loanwords in FU

In a number of ground-breaking investigations, Jorma Koivulehto (1988

Fi. kes summer PIE *hes-en- harvest (Koivulehto 1991: 36-40)

Koivulehto argues for a conditioning by which the velar stop k- occurs

PFU or PFP form Example alleged NW PIE source

*p- to fry Lule Saami pass- *beh- bake

Katz (2003) implicitly supported Koivulehtos thesis regarding the reflex

2 Analyses of the individual proposals

In the following, I will comment on each of Koivulehtos relevant ety-

2.1 PFP *( J ) ETE TO HAVE ENOUGH POWER ?

2.2 PFP * KONE TOOL PIE * G N H -) O - WONDER ?

2.3 PFP * INE HUMAN BEING PIE * G E NH - BEGET A CHILD ;

Proto-Fenno-Volgaic *ine human being is reconstructed on the basis

people, mon woman and -ym famous (Liukkonen 1999)3; Baltic

2.4 PFS * LETE LEAF PIE ** B LEH -T - LEAF ; SPROUT ?

Proto-Fenno-Saami *lete can be reconstructed on the basis of PBF

2.5 PBF * TATAS DOUGH PIE * TH )- S -TO - S DOUGH ?

surely Ockhams razor dictates that the sibilant we encounter reflects

2.6 PFP * PEWEN V SIEVE PIE * PEU H - ENO - WINNOWING ?,

Skld (1960: 37) was the first to point to an Indo-Iranian protoform of

2.7 PFU * PE - OR * P - TO FRY PIE * B EH - BAKE ?

Koivulehto asserts as the source PIE *beh- to bake ( to fry, to

2.8 PFP * WO ( J ) A BRANCH ; BENDING PIE * U O )H- H ID .?

A form *wo(j)a branch; bending may be reconstructed for the Proto-

2.9 P ROTO -M ARI * R MILK PIE (PIIR.) * KSIH -R -

2.10 PFS * KETA - TO DARE PIE * G EH - D - BE ASHAMED ?

I will address the remaining examples together (Koivulehto (1988 [1999:

geal i.e. *h or *h were transmitted to Balto-Fennic as *h. This is quite

3 Discussion and conclusions

A more minute critical analysis of Koivulehtos and Katzs entire mate-

al since -h- in Balto-Fennic can simply be direct reflexes of the similar

Koivulehtos important conclusions that a) PIE *h was an aspirated

1 The distribution, function and origin of PIE *-bo-

In an earlier article (Hyllested 2010), I have contested the widely-held

a) It occurs almost exclusively with roots ending in a sonorant, and

squid, PGmc. *dura- yellow ; dodder [Cuscuta europaea], Toch. B

b) Formations with *-bo- directly added to roots beginning with a

ber is *(s)kiHuo- shin-bone(Lubotsky 2002). Hitt. wapa shroud (Kloekhorst

The notion that *bo- derives from thematicization of the verbal

a) The use of the suffix is much broader than initially described by

The recurring extension *-b- in verbal roots (*gle)- ~ *gle)-b- smear;

2 Alleged examples of *-bo- in Anatolian

2.1 L YCIAN XAHBA GRANDCHILD

2.2 L YCIAN XTABA RULE

Lyc. xtaba (suggested by Shevoroshkin 1979: 178-179) is recte xntawa-

2.3 C ARIAN - BA - IN PLACE - NAMES

It was suggested by Neumann (1988: 187 and n. 4) that the recurrent

2.4 H ITTITE ALPA - DOG S EXCREMENT

2.5 H ITTITE WAPA CLOTHES ( OF THE DEAD ) ; SHROUD

2.6 H ITTITE ALPA CLOUD

Hitt. alpa cloud constitutes a problem because the expected reflex of

a) it is not found in Indo-Iranian or Tocharian

e) it is frequent in European geographical names (Alba, Albion, El-

p- to fry Lule Saami pass- beh- bake

geal i.e. h or h were transmitted to Balto-Fennic as *h. This is quite

The recurring extension -b- in verbal roots (gle)- ~ *gle)-b- smear;

Reconstructing PIE weave as either hueb- or ueb- is problematic

55) PCelt. brus-na- > OIr. bronnaim injure, damage, brus-o-

70) Compounds with hekuo- horse and re)d-: Gaul. PN Epo-

85) PCelt. atim-, atam- > Gael. aitheamh, W edafedd

92) PCelt. grando-, grendo- beard > MIr. grend; MW grann