If we assume that the dichotomy between mi- and
hi-conjugation in Hittite was inherited from the
proto-language, that allows an elegant solution to the
distribution of Germanic strong preterites.
Mi-verbs generally had e-grade of the root in the present
singular, zero grade in the plural (mobile stress).
Hi-verbs generally had o-grade in the singular, e-grade in
the plural (acrostatic stress).
In Germanic, we have two categories of strong verbs: verbs
with e-grade of the root, and verbs with a-grade (<
*o-grade). Long vowel verbs pattern with the e-verbs (*i:?,
*u:) or with a-verbs (*e:, *o:). Germanic e-verbs behave as
if heirs to the mi-conjugation, a-verbs behave as if heirs
to the hi-conjugation.
It is customary to split the e- and a-verbs up into five
classes, according to the root vocalism:
e1 *ei (*i:) a1 *ai (*e:)
e2 *eu (*u:) a2 *au (*o:)
e3 *eRC a3 *aRC
e4 *eR a4 *aR
e5 *eC a5 *aC
The imperfect and aorist were lost as preterite tenses in
Germanic, leaving the perfect and the Germanic dental
preterite ("weak preterite").
Strong e-grade verbs make their preterite (prefect) by
changing the generalized (bacause thematic or thematized)
e-vocalism of the present into sg. /o/ > /a/, pl. zero, so
the verbal forms are:
I. *ei *oi *i *i
sti:ga - staig - stigum - stigans
II. *eu *ou *u *u
biuda - bauþ - budum - budans
III. *eNC *oNC *N.C *N.C
binda - band - bundum - bundans
Classes IV and V are special:
IV. *eR *oR *e:R *R.
nima - nam - ne:mum - numans
V. *eC *oC *e:C *eC
giba - gaf - ge:bum - gibans
In Gothic, the a-verbs go as follows:
I. *oi
haita - hehait - hehaitum - haitans
II. *ou
auka - eauk - eaukum - aukans
III. *oNC
falþa - fefalþ - fefalþum - falþans
Again, classes a-IV and a-V (= class VI) are special:
IV. *oR *o:R *o:R *oR
fara - fo:r - fo:rum - farans
V. *oC *o:C *o:C *oC
slaha - slo:h - slo:hum - slahans
The e:- and o:-verbs:
e:(1) sle:pa sesle:p sesle:pum sle:pans
e:(2) le:ta lelo:t lelo:tum le:tans
[The difference is between original *e: (originally
ablauting with zero, as in Angl. leorton < *le-lt-), and
*-eh1-, originally a normal e-verb with Ablaut *e ~ *o,
later incorporated into the reduplicating group.]
o: flo:kan feflo:k feflo:kum flo:kans
As can be seen, Ablaut plays a minor role, and the
difference between present stem and preterite stem is mainly
expressed by reduplication.
In the other Germanic languages, the a-verbs follow a
different pattern:
I. *oi/*e: *ei
haitan - he:2t - he2:tum - gihaitan
slæ:pa - sle:2p - sle:2pum - gislæ:pan
(læ:ta - le:2t - le:2tum - gilæ:tan)
(As van Coetsem has shown, *e:2 reflects PIE *ei with
a-Umlaut: PIE *ei, *eu split into *ee > *e:2, *eo [a-umlaut]
vs. *ii > *i:, *iu [i-umlaut]).
II. *ou/*o: *eu
hlaupan - hleop - hleopum - gihlaupan
flo:kan - fleok - fleokum - giflo:kan
III. *oNC *eNC
falþan - felþ - felþum - gifalþan
IV. *oR *o:R
faran - fo:r - fo:rum - gifaran
V. *oC *o:C
slahan - slo:h - slo:hum - gislahan
If we depart from a PIE paradigm with *o ~ *e Ablaut in the
hi-conjugation past, it is clear that the North-West
Germanic forms in the preterite (A1..A3) have generalized
the hi-conjugation plural (and, in NW Gmc. also the 2sg.)
with *e-grade of the root.
Diachronically, the mi-conjugation verbs have all become
thematic, with e-grade of the root in the present (sg. and
pl.). The preterite is based on the perfect (i.e. the
hi-conjugation present), with Ablaut *o ~ *0 and apparently
no reduplication.
The hi-conjugation verbs split into two groups: the
"praeterito-praesentia", which have perfect endings in the
present (*oi: wait "I know"; *ou: daug "it's OK"; *oRC: kann
"I know"; *OR skal: "I must", mag "I can"; *o:: o:g "I
fear"), with or without Ablaut (*o ~ 0) in the plural
(witum, kunnum, skulum / magum, o:gum), and making a weak
preterite.
The larger group became thematic (presumably because of 3sg.
*-e), and have persistent o-grade (> -a-) in the present.
The hi-preterite had o-grade in the sg., e-grade in the
plural, with the former generalized in Gothic, the latter in
North and West Germanic. Reduplication was apparently
optional, generalized as a preterite marker in Gothic (where
the Ablaut of present and preterite didn't differ), and lost
in N. and W. Germanic (except for isolated cases like ON
sera < *se-so:-).
That leaves the problem of -e:- in the preterite pl. of
e-verbs of the structure *-eC- (*-eR-), and -o:- in the
whole preterite of o-verbs of the same structure.
The -e:- or -o:- do not appear in praeterito-presents
(skulum, magum), so we may assume it's a thing peculiar to
the preterite. As I suggested earlier, roots of the
structure *-eC- (*-eR-) were prone to retain the aoristic
3rd. person marker *-s (Hitt. gane:szi, Vedic verbs in -a:,
-aC), which led to a category of "s-aorists" based on (1)
active aorists (*g^ne:h3-s-) or on (2) hi-conjugation
aorists (the classical s-aorist). If such categories also
existed in pre-Germanic, the lengthened vocalism caused by
*-s might have been retained in the preterite of verbs
ending in a single consonant (although the s-forms
themselves were apparently lost). Mi-conjugation aorists
had generalized e:-grade to the plural (ne:mum, etc.).
Hi-conjugation verbs must have had -o:- in the singular
(*-oC-s > *-o:C-s), -e:- in the plural. Germanic
generalized o: (faran, fo:r, fo:rum), while elsewhere we
generally find -e:- (OCS 1.sg. nêsU).
=======================
Miguel Carrasquer Vidal
mcv@...