Class VI wet verbs: slag-

As promised (sort of) I will try to show yet another Germanic class VI
verb is a loan from an a-language the speakers of which had to do with
transport on water.


Dirk Boutkan
Lithuanian s^lãkas, Old Norse slag
Some features of North European substrate words exemplified by an
alleged Indo-European etymon

'1. Germanic, Baltic and Slavic show a series of obvious cognates
denoting "wetness, rain, drop vel sim.", cf. the following examples
(after Pokorny 1959: 957, Lühr 1988: 289, Fraenkel 1962-1965: 998,
1002, 1005):
Old Norse slag "Nässe (von Regen)", slagfiðr "durchschlagende Nässe",
slegin "benetzt", Modern Icelandic slagi "Feuchtigkeit", Norwegian
slagen "feucht", Swedish slaga "Sumpf'
Middle Low German slagge(n) (and denominative verb slaggen
"regnerisches, naßkaltes Wetter sein"), Norwegian slagg "Speichel",
Swedish slagg "Mischung von Regen und Schnee", slagga-väder
"schlackiges, regnerisches Wetter"
Modern High German dial. schlack "Schneematsch"
Low German slack "große und schwere Tropfen", schlackern "verschütten,
sudeln", besleckern "beflecken"
Old English slo:h, Middle Low German slo:ch "Morast"
Lithuanian s^lãkas "Tropfen, Fleck, Kleck", s^lakstau~, -ýti "mehrfach
spritzen, sprengen", s^lakù, -´e.ti "tröpfeln", s^laknóti "triefen,
gelinde regnen", cf. further Fraenkel (1962-1965: 998); s^lekiù,
s^le.~kti "spritzen, Wasser ausschütten", s^le.ktùvas "Sprengwedel,
Gerät zum Sprengen"; s^lìkti "tropfen, tröpfeln, sintern", s^likdamas
"triefend", ants^lìkti "bespritzen", s^liknóti "langsam triefen,
gelinde regnen"
Latvian slaka "Anfeuchtung", slacît "naßmachen, besprengen, fein
regnen", beside forms with initial s^-, such as s^laka = slaka,
s^lacît = slacît, etc.; nasalised slan,k´is "feuchter Boden", s^l,anka
"Tropfe, Boden der Wasser durchläßt". Russian sljakotI, Polish
s´la,kwa "nasses Wetter, Regen mit nassem Schnee", s´la,kna,c´ "naß
werden"(Serbo-Croat sl´éka "Flut" may not belong here for semantic and
formal reasons. I thank my colleague Dr. Rick Derksen for helping me
interpret the Baltic and Slavic evidence.).
The reconstruction yields several problems. Pokorny (loc. cit.)
arrives at the proto-forms *sk^le:k-, *sk^l&k- on the basis of the
Gmc. and Baltic evidence, without taking into consideration the Slavic
forms (see also footnote [4]). If the view is correct that the
opposition between velars, labio-velars and palatals was neutralised
after s in Indo-European (cf. Beekes 1995: 111), the Baltic forms can
hardly go back to a PIE form with initial s-, because they require a
real palatal *k^- > Lith. s^-, Latv. s- ( However, the exact reflex of
*sk- in Balto-Slavic is disputed (perhaps > s- ?), cf. also such
obvious cognates as Goth. skeinan, ON skína, OE sci:nan, OS, OHG
ski:nan "gleam" ~ OCS sino,ti, also sijajo, ,-ati "gleam" (cf. Seebold
1970: 410).). If we therefore start from a form without initial *s-,
another complication is that initial palatals were depalatalised
before *l followed by back vowels (Kortlandt 1978: 240, section 13).
These considerations render Proto-Baltic *k^lVk- > *klVbackk ~
*s^lV+frontk as the most likely reconstruction, in which the
depalatalised *k- was analogically replaced by *s^- in all forms (cf.
in this connection Kortlandt 1978: 240, section 14). The initial s^-
and the nasalisation in some of the Latvian forms provide additional
complications. Perhaps the s^-forms are due to German influence (Dr.
R. Derksen, p.c; but see also section 6).
The Gmc. evidence points to PGmc. *sla/o:K-, with variation of
root-final consonants (not commented upon by Pokorny, but see section
1) as well as of root vowels (see section 4). In order to maintain the
comparison with Baltic, we must start from a simplification of initial
*sk^l- to *sl- (This development may have a parallel in OHG sliozzan,
OFris. slu:ta "close" < *slu:-t-< *sluh2- < *sklh2u- (with laryngeal
metathesis), cf. Lat. cla:vis < *klh2u- without s-mobile.) and assume
that Gmc. shows s-mobile, beside taking for granted the Gmc.
variations mentioned.
The nasalisation in Slavic remains unexplained (cf. also Fraenkel loc.
cit., One could discard the Slavic forms, which indeed are restricted
to Polish and Russian, as loans from Baltic. However, the nasalisation
would imply a specific origin, viz. from Latvian, which makes this
possibility less attractive.). It suggests a proto-stage *sl(e)nk-,
possibly from earlier *k^l(e)nk-. If we combine the evidence of
Baltic, Slavic and Gmc, we arrive at a reconstruction *(s-)k^lV(n)K-.
2. It has been suggested that these forms are to be connected with
further Indo-European material, but this envolves even more problems.
The connection with Avestan srask- "triefen, abtriefen von;
triefregnen" and Armenian srskel "besprengen" requires that Armenian
borrowed from Indo-Iranian; a reconstruction on the basis of the
evidence of both languages, -which is most straightforward - would be
*sresk- (Pokorny 1959:1002). This would no longer make possible a
connection with the forms mentioned in section 1. Note in addition
that Pokorny s.vvw. also expresses his doubts as to the connection of
the two etymological complexes.
Fraenkel (1962-: 1002) briefly mentions Petersson's suggestion that
Lith. s^lãkas and s^lãpias "wet, humid, full of water" are cognates,
the latter of which has an IE etymology, cf. Sanskrit s´laksna-
"glatt, weich". However, the alleged correspondence between the
root-final consonants -k- and -p- lacks a satisfactory explanation.
There is serious reason to doubt an IE origin of the etymon under
discussion. Apart from the apparently limited geographical
distribution (only Balt.-Slav. and Gmc), several of the formal
complications mentioned thus far can actually be interpreted as common
features of North European substrate material. I shall discuss these now.
3. Concerning the consonantism, we state two non-IE features. First, I
already pointed to the variation of Gmc. root-final consonants in
section 1. We find the following vacillation:
(1) *slag- in ON slag, etc.;
(2) *slagg- in MLG slagge, Swe. slagga-; deriving the MLG geminate
from WGmc. jod-gemination leaves the NGmc. forms unaccounted for;
(3) *slakk- (*slak-?) in LG slackern; in the verbal forms the geminate
can hardly represent the WGmc gemination of PGmc. *k before r; apart
from the fact that this development is only a sporadic one, this would
require that the r immediately followed the guttural; at any rate, the
input must have been voiceless: *k or *kk;
(4) perhaps *slo:k- in OE slo:h-, etc.; this variant may represent
grammatical change *-h ~ *-g (< pre-Gmc. *k), however.
Such variation as the one mentioned here is well-attested in Gmc.
words of non-IE, North European origin (cf. Kuiper 1995: 69ff), cf.
also such examples as OE knapa (< *knap-) ~ Norw., Swe. dial. knabbe
"young man" (< *knabb-) or English knob (< *knubb-) ~ MDu. cno:vel
"joint, ankle" (< *knu:b-) ~ MLG knoppe "knot, bundle" (< *knupp-). A
possible explanation of this phenomenon is that the borrowed substrate
items displayed consonants that were not available in the PGmc.
phoneme inventory, e.g. plain (i.e. non-glottalised) voiced stops and
geminates, which had to be substituted by phonetically different Gmc.
sounds. This may have led to hesitation and, subsequently, to
variation. In other instances, we may be dealing with different layers
of borrowing, viz. before Grimm's Law (shifted forms) and after it
(unshifted forms). If we take the Baltic and Slavic evidence for a
voiceless root final consonant seriously, this point of departure
yields a satisfactory explanation for part of the Gmc. forms as
follows: North European *-kk was interpreted as *-k (IE did not, Gmc.
hardly showed geminates) > PGmc. *-x/-g (Grimm + Verner) as against
*-k(k) as an unshifted, late borrowing; the Gmc. form in -gg remains
unexplained in this scenario. Lühr (1988: 289) suggests: "Es ist zu
vermuten, daß der Geminate eine lautsymbolische Funktion zukommt",
which is no more than a possibility.
Second, the Slavic and part of the Latvian forms show a nasal in the
root, which is not present in the other languages and is considered as
problematic (see also section 1.). However, optional prenasalisation
in root final position is a well-attested phenomenon in North European
substrate material, see Kuiper (1995: 68ff), who quotes several Gmc.
examples such as ON klífa "climb" ~ English climb, but cf. also Lith.
dubùs "deep"~ dum~blas "marsh" (on p. 72). Perhaps we have another
outer-Gmc. example here.
4. If we compare Lith. s^lãkas and Gmc. *slaK-, we are forced to
reconstruct *o or radical *a (= non-IE, cf. Lubotsky 1989). It cannot
be decided which vowel is original, so we may be dealing with a non-IE
a-word (cf. also Kuiper 1995: 68 sub 1). At any rate, the forms rule
out the reconstruction of a laryngeal in the root as is suggested by
Pokorny's reconstruction of a & (and which would supposedly have been
vocalised to a), because a sequence *k^l,k- would have led to the
vocalisation of the vocalic resonant in both Gmc, e.g. Goth. kaurjos
"heavy" < *kur- < *krH- (Gr. barús), and Balto-Slavic (Lith. **s^ìlk-).
The "ablaut" in the Gmc. noun is especially remarkable. Elsewhere, I
have pointed to the abundance of seeming ablaut variation in Gmc.
nominal roots, which is not to be expected from an IE point of view
(cf. Boutkan 1999, especially section 3.1). All instances concern
(North) European substrate material and are likely to represent a
vowel vacillation that somehow finds its origin in the donor language.
I repeat the examples here (for details, cf. Boutkan loc. cit.):
Gmc. *ho:d- ~ *hatt- in e.g. OHG huot, MDu. hoet "hat" ~ OE hœtt,
OFri. hath "hat", more frequently hôd.
Gmc. *klo:k- ~ *klak- in e.g. MLG klo:k, MDu. cloec "smart" ~ Mod.
Scand. diall. klak(er) "fertile"
Gmc. *kno:st- ~ *knast- in e.g. MDu. cnoest ~ Norw. knast "knot"
Gmc. *ko:k- ~ *kak- in e.g. MLG ko:ke, Du. koek "cake" ~ Du. kaakje
"cookie", Norw. Swe. dial. kaka "cake"
Gmc. *lo:f- ~ *laP(P)- in e.g. Goth. lofa "palm of the hand", ON lófi
"palm" ~ OHG laffa "palm"; lappo "palm", Mod. Icel. löpp "paw", OFris.
lappe "spade"
Gmc. *lo:m- ~ *lam- in e.g. OFris. lo:m, OHG luom "weak" ~ OE loma,
lama, OS lamo, OHG,MDu. lam "lame"
adj. Gmc. *sko:þ- ~ *skaþ- in e.g. ON skœdr "causing harm", cf. also
skóþ "weapon" ~ Goth. skaþuls "harmful", OHG scadal "evil"
Gmc. *wo:s- ~ *was- in e.g. OE wo:s "humidity", MLG wo:s "foam, sap" ~
MHG,MLG,MDu. wase "wet field"
We seem to be dealing with yet another instance of this phenomenon
here, cf. *slaK- ~ *slo:K- (This does not imply that we can consider
ablaut in nominal roots as a heuristic principle for the establishment
of substrate lexicon in Gmc, nor that IE ablaut is nowhere preserved
in nominal roots, cf. *skunk- in OFris. skunk(a) "shank, femur", East
Fris. schunke "leg", Mod. HG dial. schunke "shank, femur" ~ *skank- in
OE scanca, MLG schenke, East Fris. schanke, Norw. skonk, skank
"shank", *skenk- in OS skinka, OHG scinko, scinka, MLG schinke "shank"
from a PIE root *skeng- (Pokorny 1959: 930), cf. Skt. sákthi- "shank"
(with -thi- for *-ti- after ásthí), perhaps also Gr. skázo: "limp".).
One could claim that the Balto-Slavic cognates show the reflexes of
PIE ablaut, and, hence, must have an IE origin. However, the ablaut in
Lith. s^lakù, -´e.ti ~ s^lekiù, sle.~kti can easily be secondary as
certainly do the forms with a root vowel i, e.g. s´lìkti, which
vocalism does not conform to any PIE ablaut grade. The Slavic forms
point to *sl(e)nk- without traces of a root vowel *a (or *o, *i) in
the root. It remains possible that this vocalism has a secondary
origin as in Baltic. At any rate, a similar correspondence seems to be
attested in PGmc. *landa-, cf. Goth., ON land, OE land, lond, OS land,
OHG, MHG, MLG lant, OFris. lond, land, MDu. lant "land" (beside
*lendjo: in Swedish Dial. linda "fallow land", *lund- in ON lundr
"grove"), as against Russian ljadá "overgrown field" (cf. also OPruss.
lindan "valley"). The a-vocalism is also found in Celtic, cf. OIr.
land "open place", Welsh llann "piece of land". In spite of a possible
reconstruction PIE *londh-yo- (Pokorny 1959: 675), this geographical
term only has a limited geographical distribution and is suspect of
being a substratum word (cf. also Polomé 1990: 335). It appears to be
attested in Basque landa "field", Catalan llanda "plain" (cf.
Scardigli apud Lehmann 1986, s.v. land). Returning to the Slavic
vocalism, we may be dealing with a peculiarity of the vocalism in the
substratum language from which "land", and, possibly, our etymon were
borrowed. At any rate, in my view the possibility of "real", i.e. IE
ablaut does not outweigh the preponderance of non-IE features
mentioned thus far.
5. Note in addition that it is also not a counterargument against a
non-IE etymology that the Gmc. forms would require the reconstruction
of s-mobile. Although PIE indeed showed such a prefix (with unclear
origin, meaning and function), it was productive with both IE and
non-IE etyma in Gmc. It may well be that the North European substratum
layer from which the words under discussion were borrowed also had a
mobile s- (cf. in this connection also Cowan 1974: 245).
6. However, if we start from a non-IE loan word, it may not be
necessary to make a reconstruction on the basis of the Balto-Slavic
and Gmc. evidence according to the correspondences these languages
show in IE words, i.e. *(s-)k^lV(n)K-. This would only be valid if the
borrowing took place during the common proto-stage of the languages
involved. It is possible, if not probable that the languages concerned
borrowed independently from the same source. In that case, the
assumption of a source form *s^lV(n)K- becomes an attractive
alternative. The initial sibilant was substituted by s- in Gmc.
(which, on the side, had a palatal realisation). It could be borrowed
as s^- in Lithuanian, where this phoneme was part of the phoneme
inventory. In Latvian and Slavic, it may have been borrowed as *s^-
(i.e. the reflex of PIE *k^), which later yielded s-. This
interpretation has the advantage of accounting for the unexplained
Latvian forms with s^-: these forms may represent (later) borrowings
with retention of the actual palatal of the source form.
7. I conclude that the etymon under discussion probably has no
cognates outside Gmc. and Balto-Slavic. It may also have contained
radical *a. The seeming ablaut and the variation of root-final
consonants in Gmc. as well as the problematic nasalisation in Slavic
and part of Latvian are problematic from an IE point of view, but can
be understood as features of North European substrate material. The
ablaut in Baltic can easily be of secondary origin, that in Slavic is
problematic, but may be secondary as well. The Gmc. s-mobile does not
speak against this interpretation as it occurs in both IE and non-IE
words in Gmc. and may actually have been present in the substratum
language. However, the reconstruction of initial *(s)k^l- becomes
unnecessary if the languages borrowed from the donor language
independently; a source form *s^lV(n)K- with different substitutions
of the initial palatal sibilant could account for all forms and, in
addition, explain the problematic Latvian forms showing unexpected s^-.
8. Excursus. The etymon thus reconstructed also allows an alternative
etymological interpretation of MLG slachregen, MHG schlage-,
slegeregen, Mod. HG Schlagregen, MDu. slachregen, Mod. Dutch slagregen
"heavy rain (rain which makes you very wet)". It has no separate entry
in the etymological dictionaries, and the communis opinio is that this
WGmc. compound originally contains the noun *slagi- "hit, blow" as its
first member (cf. as to Dutch explicitly WNT XIV, 1538-9, s.v.
slagregen, implictly Franck-Van Wijk-Van Haeringen 1976, s.v. slag
["in alle betekenissen"]; also Onze taal [1967,5], 41-2). The meaning
would be: "rain that hits, i.e. falls down heavily". Although this
interpretation remains possible, it seems an alternative to me that
*slag- actually represents our etymon in a meaning "wet".
Interestingly, with respect to the German word, the dictionary of
Grimm (IX, 422, s.v. Schlagregen) distinguishes between the word with
the current meaning "heftiger regen mit dicken, hörbar aufschlagenden
tropfen" and a homonymous schlagregen denoting "nebelartiger, feiner
und durchdringender regen", etymologically connecting the latter with
our etymon. Indeed, the assumption of a form containing *slagi- "hit,
blow" seems less apt with respect to this second meaning. Perhaps, the
association with *slagi- "hit, blow" may be due to secondary
reinterpretation in all instances. The folk-etymological substitution
of the first member can directly be recognised in the MHG form as it
shows i-mutation.'


de Vries says:
'slag 1 n
'schlag, hieb, kampf,
nisl far norw schw da. slag — >
orkn slag 'heftige bewegung des meeres' (Marwick 167)' >
shetl slag, slog 'schlag'.—
mnd slach verschluss, riegel',
mnl slach 'schlag, hieb, wellenschlag, spur, weg' —
air sleg 'speer' —
vgl slá 2
— 2 n 'nasse',
nisl slag 'plötzlicher regenschauer',
vgl schw dial slaga 'sumpf', —
mnd slagge 'regnerisches wetter',
ne dial slaggie, slack 'sumpf' —
lit s^lãkas 'tropfen, fleck',
s^lakstau~, s^lakstýti 'spritzen, sprengen',
IEW 957 zur idg wzl *sk^lēk —
vgl slagna und sleginn
Torp, Wortschatz 534 verbindet mit slá 2, also eig 'mit nasse
beschlagen, durchschlagende nasse', und dann die bed 'sumpf aus 'durch
schlag entstandene vertiefung', vgl ae slōh (ne slough), mnd slo:ch
'morast'. —
Diese bed entw ist höchst unbefriedigend, aber deshalb braucht man die
Verbindung zwischen den beiden stammen *slag nicht aufzugeben,
vielmehr ist auch noch slafa damit zu verbinden. Weil auch sonst
worter für 'pfutze, morast' aus der arbeit des lehmmischens
hervorgehen (vgl. mosi), konnte man eher an das bewerfender hauswand
mit lehm denken; dabei gehen die begriffe 'schmeissen' und 'schmieren'
in einander über (vgl. smeittr). Die wzln *slaf, *slag und *slak (für
diese vgl. slakna), sind teils mit mobilem teils mit wurzelfestem s
und in diesem fall erweiterungen von der in salr vorliegenden idg. wzl
*sel, falls man nicht vielmehr rein germanische neuschopfung annehmen
soll. Für die beziehung auf die lehmwandtechmk vgl. auch sletta.

The link Boutkan needs to get from "drench" to "kill", and from there
to "strike" (not the other way around, as also all his predecessors
have tried it) is this:

de Vries
sløkkva schw. V. 'loschen' (< *slakwian),
nisl slokkva, far sløkkja, nnorw. sløkkva, sløkkja, nschw. släcka,
dial slokka, nda. slukke. — >
ne dial sleck, slick (Thorson 77); >
orkn slock, shetl. slokk. —
ae. sleccan 'schwachen', slacian 'schwach werden',
ne. slake 'loschen', as. sleckian,
vgl. ae. slæc, sleac 'trage, schlaff. —
vgl. slokna.
Über das Verhältnis von sløkkva. und slekkja herrscht uneinigkeit.
Nach Pipping, Fschr. Lidén 1912, 146 soll *slakwian über *slakian
lautgesetzlich zu aschw slækia fuhren, vgl. aber E. Olson ANF 31,
1915, 126'

with a bed.entw.(semantic development) from "extinguish a fire (by
dousing it with water)" to "kill"; cf the development in Engl. 'snuff'.


Torsten