[tied] Morphology (2/20)

From: Miguel Carrasquer
Message: 14340
Date: 2002-08-16

2. The present thematic

The active thematic present in Anatolian is distinguished from the athematic
simply by the presence of the thematic vowel between the stem and the endings.
The quality distribution of the thematic vowel is confuse in Hittite as
attested, but this is the result of recent analogical processes in Hittite, and
the original situation must surely be reconstructed as:

-a-mi (Luw. -a-wi) -a-weni (Luw. *-a-mani)
-e-si -e-teni
-e-zzi (Luw. -e-ti) -a-nzi (Luw. -a-nti)

In other words, the thematic vowel is *o (Anat. a) before voiced -m(w) and -n,
*e (Anat. e ~ i) before voiceless -s and -t.

But what is the thematic vowel? It is not simply part of the stem, for in that
case we would expect the thematic vowel to have been reduced to zero in the
present sg. (**CéCe- > *CéC-), and to have attracted the accent (or likewise to
have vanished) in the plural (**CeCé- > CCé or CeCe-' > CC-'). Also, the fact
that some verbs can be conjugated either thematically or athematically (*bhér-ti
~ *bhér-e-ti) does not favour that analysis.

In general, the thematic vowel behaves in a most peculiar way, setting it apart
from every other PIE vowel: (1) It is not subject to zero grade; (2) It does not
normally cause the accent to shift; (3) It is lengthened before voiced segments.

As to the function of the thematic vowel, in the verb it is used to make the
conjunctive stem, while in the thematic indicative no particular function for it
can be recovered. If we compare this to the function of the thematic vowel in
the noun and adjective, we see that there too it is in part a functionless root
extension attached to nouns, and in part it serves a clear function as an
adjectivizing suffix. The phonetic peculiarities mentioned above under (1) and
(2) can be explained if the thematic vowel at the time of the accent rule and
the zero-grade rule was still an independent or semi-independent accentual unit
("word"). We can unify all the different uses of the thematic vowel if we
regard it as in origin an independent pronominal element, identical to the
anaphoric pronoun *i ~ *e ~ *o. When added to nominal roots, it would have
functioned as a marker of definiteness (like a postponed article, or like the
Balto-Slavic definite adjectives built by adding the same anaphoric pronoun *i ~
*e ~ *o to the thematic stem, again). When added to a verbal root, the function
may similarly have been that of creating a verbal adjective (yielding, after the
addition of personal endings, the conjunctive mood), or to mark a third person
definite object (similarly to e.g. the Hungarian verbal system, where transitive
verbs can be conjugated using the normal intransitive endings when there is no
object or when the object is indefinite, but use a special _objective_ set of
endings when there is a definite direct object).

This explains why transitive verbs like *bher- "to carry" could be conjugated
either thematically (with a definite object) or athematically (with indefinite
or unexpressed object). This does not explain why we also have thematic
intransitive verbs. To explain their existence, we can invoke either (a)
intrusion of conjunctive forms in the indicative, or (b) thematization occurring
after the grammatical function of the thematic conjugation as an objective
conjugation had been lost. The athematic type, with its potential for unstable
consonant clusters in the interface between stem and endings, and with its
mobile accent, was clearly the recessive type (all but lost in a number of
daughter languages), and yielded more and more to the thematic type with its
stable accent and clear vocalic interface.

As to the thematic endings, we have seen that in Anatolian there is no
difference between the thematic and athematic type, except for the presence of
the thematic vowel (and the stable accent). However, a number of languages show
aberrant endings in the thematic conjugation, which has led to speculations that
the thematic endings, at least in the present singular, have a different origin
than the athematic endings. The 1st person sg. outside of Anatolian is almost
unanimously *-o: instead of the expected *-omi (only Sanskrit has -a:mi, but in
alternation with unextended -a: < *-o:, suggesting that we are dealing with
analogical -o: + -mi, not inherited *-o-mi). The 2nd and 3rd persons almost
everywhere have the expected -esi, -eti, but we have the following exceptions:

Old Irish *-i:, *-et for expected *-ei, *-eti
Baltic *-ei, *-o[t] ,, *-esi, *-eti
Slavic -es^I, -etU ,, *-esI, *-etI
Greek -eis, -ei ,, *-ei, *-esi

As to the 1st singular ending, I think it can readily be explained if we depart
from original *-o-mw-i. We have -ami in Hittite, but -awi in Luwian, and -eu (>
-au) in Tocharian B, pointing to a development *-o-mwi > *-o-wi, and then (with
Umlaut of *-i to *-u) *-owu. As in the locative sg. of the i- and u-stems,
where the sequences -owi/-ewi > -owu/-ewu and -eyi regularly developed into
-o:u/-e:u and -e:i, in the 1sg. thematic the ending should regularly have become
*-o:u, which is exactly what is preserved in Tocharian B -eu, -au (cf. *gwo:us >
kew "cow"). This was further reduced to *-o: in most languages, yielding Gmc.
-a, Arm -0 (in the aor. conj. -ic' < -isko:), Old Irish -W (bhero: > biur =
/birW/), Latin -o:, Baltic -ù, Slavic -o~ (< *-o: + -m), Greek -o:, Avestan
-a:(mi), Vedic -a:(mi).

The 2nd sg. in Old Irish points to early loss of -s-, whereby -esi > -ei > -i:.
Lithuanian -ì, refl. -ies is from *-ei, by reanalysis of esì < *esei "you are"
as root *es- plus ending *-ei. The athematic ending was then transferred to the
thematic conjugation. Greek -eis for expected *-ei is simply the analogical
restoration of the 2sg. marker -s after the loss of intervocalic -s- in Greek.
Slavic -es^I (OCS -es^i, with *-i: from the athematic paradigm) directly
reflects PIE *-esWi (*sW regularly giving Slavic *x ~ s^). The labialized
sibilant had become a shibilant (cf. French ch, phonetically [s^W]),
subsequently lost its labialization, and thus did not affect the quality of *-i
in Slavic, unlike what happened in the first person sg., where the labialization
of *mW did cause Umlaut of the present tense marker *-i , and unlike the loc.pl.
ending *-sWi, where the Umlaut does occur everywhere but in Greek (*-sWi > -su ~
-si, Slavic -xU).

In the third person, Old Irish, Balto-Slavic and Greek (as in part Latin with
OLat. -d besides -t) perhaps continue the third person secondary marker *-t
rather than expected primary *-ti. In Slavic, *-et should have given *-e, to
which the pronoun -tU "this/he" was added (unless -t was exceptionally preserved
here (sandhi?), and then extended with obligatory -U, as in the cases of azU <
*eg^, izU < *eg^s). In Greek, it's also possible that -e < *-et was extended
with -i on the analogy present -eis / -? :: past -es / -e. But since the 3pl.
regularly continues primary *-onti, perhaps it was rather regular *-esi which
was replaced by analogical -ei.

While most thematic verbs have the accent on the root throughout, there is also
a second type (Vedic Class VII, tudáti), with accent on the thematic vowel, and
the root in zero grade. This type is also followed by the thematic aorists in
Old Indic, for which see below.

A review of the thematic endings:

-ami -aweni
-esi -etteni
-ezzi -anzi

Tocharian B (^ = palatalization of previous consonant):
-ew, -au -em(o)
-^t -^cer
-^(m.) -em. -tem.

Tocharian A:
-am -amäs
-^t -^c
-^(s.) -eñc

-a -am -os
-is -iþ -ts
-iþ -and --

-em -emk`
-es -êk`
-ê -en

Although this looks like generalization of the e-grade of the thematic vowel,
this is not the case (e-grade would have yielded 1sg. *-e-mi > *-im, 1pl.
*-e-mesW > *-imk` and 3pl. *-ent > *-in). In fact, the -e- comes from the root
*es- "to be", and only the forms with <ê> (*-ey- < *-eti / *-ete-) are true
thematic forms.

Old Irish (conjunct):
-W -am
-^ -ith
-^ -ad

-o: -imus
-is -itis
-it -unt

In the 1pl., -i- does not necessarily indicate e-grade of the thematic vowel
(cf. Slavic). Any unstressed vowel would have been reduced to -i- here.

-ù (< *-o:) -ame (< *-o-me:) -ava (< *-o-wa:)
-ì (< *-(s)ei) -ate (< *-o-te:) -ata (< *-o-ta:)
-a (< *-ot) -- --

Baltic has generalized o-grade of the thematic vowel (for 1 and 2 sg., see

-o~ -emU -ovê
-es^I -ete -eta
-etU -o~tU -ete

The 1pl. has thematic vowel -e- for expected -o-. The 3pl. has -o~tU for
expected -o~tI, like the 3 sg.

-o: -omes, -omen --
-eis -ete -eton
-ei -ousi < *-onti -eton

-a:(mi) -a:mas(i) -a:vas
-asi -atha -athas
-ati -anti -atas

The distinction between thematic vowel *o (> /a:/ in open, /a/ in closed
syllables) and *e (always > /a/) is preserved in all cases where an open
syllable occurs. If the 3du. was originally *-h2tés, the *h2 disappeared
without a trace. The variant form -thana of the 2pl. does not occur in the
thematic conjugation (which makes sense if the element -a is etymologically
identical to the thematic vowel -a-).

Miguel Carrasquer Vidal