80
viewed not as a MI truncation but in terms of
ON onomastics, it can be seen as another
example of the order inversion of the
compound stems that we discuss here (and
which
is common in classical dróttkvætt in the
case of kennings, see next and previous
sections). If we restore the order, we get
Eyjólfs-Þorkell, which is probably not perfect
ON, as such compounds were mostly formed
with place names as the first element, but it at
least matches the regular, recognizable pattern
of the first element of the personal name
compound being the distinguishing feature for
the carrier of the name – so it is employed in
the same function as regular patronymics.
Kennings
Kennings are rightfully considered to be a core
element of the vocabulary and poetics of
classical
dróttkvætt. They form a tightly-knit
system with a rich network of internal links (on
these see Fidjestøl 1974 and particularly
Sverdlov 2015 which presents the network
graphically), the functioning of which relies on
heiti creation (Gurevich 1984; Gurevich &
Matyushina 2000: 17‒30; Sverdlov 2003b: 42‒
62, 103‒120), on patterns of variation and
extension (Marold 1983; Sverdlov 2003b:
107‒109; 2015) and metrical packaging rules
(Sverdlov 2012; Frog 2016). Our poet
demonstrates both his considerable knowledge
of the facts of this system and his
understanding of the fundamental principles of
variation and semantic rules of heiti creation,
which have necessarily been derived from his
own working analysis of the contents of
Laufás
Edda, as well as from
rímur tradition. At the
same time, he bends these rules beyond what is
factually attested in the corpus of classical
skaldic poetry. Some of his innovations could
have arguably been developed within the oral
tradition, had it survived up to his day (or, if
one assumes that it did survive, were
developed this way). Others are not supported
by the skaldic system and remain in the
category of nonce coinages, sometimes at odds
with the principles of the classical system.
A good example of this is provided by the
drakons díkja dreifir sequence from lines 4.7‒
8. On the surface, it is a straightforward
extended kenning:
dreifir, derived from the
verb
dreifa [‘to scatter’] (itself a causative of
drífa [‘to drive’]), is a classic
nomen agentis
baseword for a kenning for
MAN
/
WARRIOR
, and
the kenning reads ‘the scatterer of bed/road of
the dragon’ → ‘scatterer of
GOLD
’ →
WARRIOR
. However, the baseword for the
embedded kenning for
GOLD
is exceptional.
The word
dík [‘dike, ditch’]
, of which here we
have a genitive plural form, is never used in
classical kennings for
GOLD
of the model ‘bed
of the dragon’. Skaldic heitis are created out of
regular words by semantically ‘downgrading’
them (semantic ‘deterioration’ in Smirnitskaya
1992; ‘levelling’ in Frog 2015) to what is best
called, somewhat in the spirit of the famous
work of Anna Wierzbicka (1972), a ‘prime’. In
the case of
dík, the normal ‘prime’ extracted
from it when it is turned into a heiti is
WATER
,
thus its normal use is that of a
determinant in
kennings for
GOLD
(e.g. ‘sun of the dike’ →
GOLD
). Extracting the meaning of
BED
, is,
however, a theoretical possibility: the kenning-
model ‘bed/abode of dragon’ uses heitis
created from ON words denoting ‘road’ and
‘path’, and because snakes (and dragons are
snakes) do live in dikes, dikes might be
interpreted as their homes/beds/roads. The
situation when
different primes are extracted
from the same word when the skaldic system,
or an individual skald, creates a heiti out of it,
is far from unknown in classical
dróttkvætt: for
example, the word
máni [‘moon’] can function
as a baseword for ‘fire of battle’ →
SWORD
kennings if the prime of
FIRE
is extracted,
because the moon is a source of light, and it can
also function as a baseword for ‘something
round or flat of the ship’ →
SHIELD
kennings if
the prime of
ROUNDNESS
is extracted, as
shields are round (Sverdlov 2003b: 54). It also
follows that the same word-turned-heiti is able
to function as both baseword and determinant
in
different kenning-models: e.g. ON
hauðr
[‘earth’] can act as a baseword in kennings of
SEA
such as
knarrar hauðr [‘earth of ship’],
and as determinant in kennings of
SEA
such as
hauðr-men [‘necklace of the earth’] (Sverdlov
2003b: 201‒202). Thus, in theory, the same
thing could have happened with
dík too, if it
were done at the right
time and then picked up
by tradition; the extant classical corpus,
however, does not attest such usage beyond a
single occurrence of a nonce kenning for
HELL
díki djǫfla [‘dike (i.e. abode) of devils’] (
LP:
81).
81
The
vápnageyfa sequence from line 4.8 is a
peculiar
case.
Were
the
word
geyfa
[‘particularly dark snowstorm’] to have existed
in ON times, it could have indeed served as a
good baseword for a kenning of
BATTLE
(
vápnageyfa [‘snowstorm of weapons’ →
BATTLE
]). The classic kenning-model ‘bad
weather of weapon(s)’ uses all kinds of words
denoting bad weather as basewords (e.g. ‘hail
of sword’ etc.). The ON word ‘weapon’ (
vápn)
however, is rarely used as the determinant
(Meissner 1921: 176, 190; Sverdlov 2003b:
189–195; Frog 2016: 189), as classical
dróttkvætt likes to
use words with less abstract
semantics as source material for heiti-making
(Sverdlov 2003b: 50). Nevertheless, the use of
vápnageyfa in the final four positions of the
verse conforms to one of the two metrical
templates in which
vápn was regularly and
even formulaically used for
BATTLE
kennings
in classical
dróttkvætt,
frequently, as here, in a
prepositional phrase (Frog 2014: 63–64). It
seems unlikely our poet had internalized such
subtleties of word usage, leaving it unclear
whether he had an exemplar verse in mind.
More substantial deviations are observable
in the
fyllir dáða sequence from line 2.3. This
is unmistakably a kenning for
WARRIOR
, yet it
looks somewhat awkward. The word
fyllir is
derived from the verb
fylla [‘to fill’] (a
causative derived from the adjective
fullr
[‘full’]), and thus means ‘filler’; it is thus a
classic
nomen agentis baseword for kenning
for
WARRIOR
. The trick is that normally
fyllir is
paired with very specific determinants: those
that are heiti for
BEASTS OF BATTLE
, adding up
to the kenning-model ‘feeder of the wolf’ →
WARRIOR
(i.e. one who, literally, fills the wolf
with food, on which see Sverdlov 2003a).
Here, however,
we have the word dáð [‘deed’]
as a determinant, adding up to a ‘filler of
deeds’. This coinage is a good indication that
the poet’s grasp of principles of skaldic
composition was
good enough, even too good:
its existence depends on understanding that
nomina agentis basewords for kennings for
MAN
/
WARRIOR
are rather devoid of any actual
sense, their semantics having been reduced to
their function of a baseword and vague
meaning of ‘doer’, and that
ultimately each
such baseword is
interchangeable with any
other of its ilk, without a reference to the
source-verb’s semantics. The key word here,
however, is ‘ultimately’: not all opportunities
present in the skaldic system are, in fact, fully
realized and entextualized (Sverdlov 2015: 29‒
31). The tenor of the list of basewords for any
kenning-model is, indeed, to make
any
member of the list interchangeable with any
other, i.e. to allow them to co-occur with any
determinant; in actuality, this ideal is not
(always) achieved. In this particular case, we
see that the knowledge of the actual skaldic
tradition on the part of our poet was somewhat
patchy: in the extant classical skaldic corpus,
dáð is normally only used as a bare stem in
adjectives such as
dáðreyndr [lit. ‘deed-tried’,
i.e. ‘experienced’], with only two kenning-like
coinages like
dáð-geymir (for Christ) and
dáð-
hittir (referring to Bishop Páll) attested.
Further, in stanza 7 we encounter a curious
semantic problem that is retained from
classical
dróttkvætt. Namely, in line 7.2, we
have
hringþoll, accusative singular of
hring-
þollr, normally a kenning for
WARRIOR
, with
þollr [n.masc. ‘fir-tree’] a typical baseword
acting as a heiti for
TREE
, and
hringr being
(here, as in other kennings) a heiti for
SWORD
,
resulting in a classic kenning-model ‘tree of
weapon’. Yet we have a problem here: the
stanza at this point clearly refers to an object,
not to a human actor, Helgi Harðbeinsson, who
is mentioned in the preceding line. The solution
seems to be to read the heitis differently: the
basic meaning of
hringr in kenning vocabulary
is the sword’s guard (literally ‘ring’), and if this
basic meaning is the active one here, then also
þollr might be read not as a heiti for
TREE
but
as a heiti for
ELONGATED OBJECT
, resulting in a
kenning that reads ‘the pole of the guard’, which
matches a well-attested model for a kenning
for
SWORD
, ‘elongated object of sword-part”,
such as
hjalta vǫndr [‘wand of the hilt’]
(Meissner 1921: 162). This reading would make
grammatical sense and would result in the ON
phrase being translated as ‘Helgi Harðbeinsson
ran a sword through Bolli’. However, this would
be
factually wrong, because the saga explicitly
tells us Helgi kills Bolli with a spear (
Laxdæla
55).
It is unnecessary to assume that the poet is
misremembering the saga text here, which he
follows rather closely in most cases, particularly
in stanza 6. This more likely reflects a very real