[tied] Morphology (9/20)

From: Miguel Carrasquer
Message: 14720
Date: 2002-08-29

9. Assorted imperfects

We saw that the imperfect could be (a) as in Greek, Indo-Iranian or sometimes
Tocharian A, a simple thematic or athematic preterite built on the present stem
(when distinct from the aorist stem), (b) an old optative, as in Tocharian B
(and sometimes A), Armenian and Old Irish (middle optative). There is no
separate imperfect in Hittite or Baltic (the Lith. frequentative past in -davo-
is presumably a recent innovation), but the imperfect formations of Latin,
Slavic and Germanic remain to be discussed.

The Latin imperfect is clearly based on an old a:-preterite of the root *bhuH-
"to be", as follows:

-bam -ba:mus
-ba:s -ba:tis
-bat -bant

From *bhwa:, *bhwa:s, etc. A peculiarity is the addition of an extension -e:-
to the stem of non-a:/e:-stem verbs before the ba:-suffix (em-e:-bam,

An exact parallel is found in Slavic, where verbals stems not ending in -a: or
-e: are similarly extended with -e:- before the imperfect suffix -axU, etc. The
parallel goes even further if we consider that in Slavic too, the imperfect is
built by agglutinating forms of the verb "to be" to this stem ending in -e: (or

-e^-axU -e^-axomU -e^axove^
-e^-as^e -e^-as^ete -e^as^eta
-e^-as^e -e^-axo~ -e^as^ete

Compare this with the imperfect of the verb "to be" (with b- added from the root
*bhuH- as in W.Gmc *bim, *bis(t) for *im, *is(t)):

(b)e:xU < *e:sm (b)e^xomU < *e:smos (b)e^xove^ < *e:sva:
(b)e: < *e:s(s) (b)e^ste < *e:stes (b)e^sta < *e:sta:
(b)e: < *e:s(t) (b)e^s^E < *e:sent (b)e^ste < *e:ste

Apart from the fact that Latin makes it normal past tense of "to be" from the
root *es- (eram, era:s) and uses *bhuH- for the imperfect, while Slavic uses
*es- for the imperfect and has an aorist based on *bhuH- (byxU, by), the
parallel is exact.

The -e:- found in the Latin and Slavic imperfects may be the same e: as the one
of the e:-preterite, a generalized stative conjunctive stem (2/3 sg.) in -e:.
Alternatively, it can be the stative suffix *-ét (> *-eh1) which produces the
Latin category of e:-stems, or the Hittite stative deverbatives in -att- (X-e:
bhwa:m, X-e: e:sm "I was in a state of X").

The Germanic weak dental preterite is not necessarily an old imperfect, but it's
convenient to treat it here. The Gothic forms are:

-da -dêdum *-dêdu
-dês -dêduT -dêduts
-da -dêdun

In most verbs, these endings are attached to the extended verbal stem (ye-stems:
nas-i-da, a:-stems: salb-ô-da, e-stems: hab-ai-da, na:-stems: full-nô-da), but
certain verbs add the endings directly to the root (baúh-ta "bought", brûh-ta
"brought", Tâh-ta "thought", etc.), including the weak preterites of the
preterite-presents (kun-Ta, wis-sa, skul-da, gamôs-ta). The /t/ we see in these
endings, as well as the zero grade of the root, suggest that we are dealing with
past participles in PIE *-tó, to which past tense forms of the verb "to do" were
added, eventually, after some haplological reduction, giving the Germanic weak
preterite as attested. The past form of the verb *dheh1- that was used to make
the weak preterite seems to have been an old imperfect based on the reduplicated
present stem (Grk. tithe:mi, Skt. dadha:mi), as follows (Gothic/OE/OHG):

*-da dede: > -da/-de/-ta *-da de:dum > -dêdum/-don/-tum
*-da dede:s > -dês/-dest/-tôs(t) *-da de:duT > -dêduT/-don/-tut
*-da dede:t > -da/-de/-ta *-da de:dund > -dêdun/-don/-tun

This is compatible with the West-Germanic reflexes of the preterite of *dheh1-,
which are (OE and OHG):

dyde teta
dydes(t) ta:te (< *de:di:)
dyde teta
dydon ta:tum, etc.

Still, a problem remains in the fact that the reduplicated present itself is not
attested in Germanic, where we have instead what are presumably old stative
forms (cf. Hitt. dehhi, dai) in the present of OE and OHG:

dô(m) tuo(m) < *dhóh1-mi
dêst (doe:st) tuos(t) < *dhóh1-si
dêT (doe:T) tuot < *dhóh1-ti
dôT tuom < *dhoh1-mé

Modern Frisian has a paradigm with Verner alternation h ~ g (along the lines of
*sehwan and *slahan): doch, dochst, doch, dogge (inf. dwaan), which, if old,
reminds one of the Latin present tense with k (< *HH?) facio, facis (*dh&1k-).

Miguel Carrasquer Vidal