Kevin Tuite

The geography of Georgian q’e

<version of 18 November 1987>
In this paper I will discuss the geographical and semantic range of the Georgian clitic q’e. While it
has been known for some time that this particle serves as a number agreement marker for plural
objects in some dialects, little has been done to identify the factors which condition its use. I will
present evidence that the NPs controlling agreement in q’e are almost always animate and
presupposed, and attempt to relate this to the syntactic changes that have been underway in
Georgian over the past millenium.

1. Agreement morphology.I will begin with a brief discussion of Kartvelian agreement

morphology. The Kartvelian verb can agree with two — in some dialects three — arguments. The
two sets of person agreement affixes used in early Old Georgian are shown in {1a}. The
correlation between case, person agreement set and semantic role is given in {1b}.1

{1a} Set V (grammatical subject) Set M (grammatical object)

1sg v- -0 1excl m-
1pl v- -t 1incl gw-
2sg x/h/s/0- -0 2 g-
2pl x/h/s/0- -t
3sg -s/a/o/n 3 DAT x/h/s/0-
3pl -n/es/en/ed 3 NOM 0-

{1b} ACTIVE STEM PASSIVE STEM

AGENT PATIENT REC/EXP THEME REC/EXP
Series I
agr. V M M V M
case NOM DAT DAT NOM DAT
Series II
agr. V M M V M
case ERG NOM DAT NOM DAT
Series III
agr. M V —— V M
case DAT NOM [+ postp] NOM DAT

Note that the feature of (formal) number is marked by Set V affixes but not, originally, by Set M .

1
The following abbreviations will be used in the glosses to the Georgian examples: (a) case: NOMinative,
ERGative, DATive, GENitive, INStrumental, ADVerbial; (b) verb-paradigm series: I (present/future, imperfect,
conjunctive); II (aorist, optative, permansive); III (present perfect / evidential; pluperfect); (c) verb-stem class:
active, passive; (d) other: QT (direct-quote marker), PLural. Verb agreement marking is given in the order: Set V /
Set M / pluralizer.
 Geography of q’e (K. Tuite, NSL 5) — page 2

By the 10th century, however, the inclusive/exclusive distinction originally coded by Set M gw-
versus m- was no longer productive, and these same prefixes came to mark a plural/singular 1st
person opposition [Shanidze 1982:74; Met’reveli 1978]. Not shown in {1} is the Old Georgian
suffix -(e)n- which crossreferenced plural NOM case arguments for certain classes of verb stems.
The verb in the following sentence employs one marker from each of the three sets: Set V, Set M
and the pluralizer en:

{2} arca mama-man tkwen-man mo-g-i-t’ev-n-e-s tkwen £ecodebul-n-i tkwen-n-i

[nor father-ERG your(pl) forgive-IIa-3/2pl/pl you(pl)-DAT sin-PL-NOM your(pl)]
“Nor will your father forgive you your sins.” [Matthew 6:15]

Note that the verb in {2} agrees in number with the NOM case direct object (“sins”) but not the
DAT case indirect object (“you(pl)”). With the exception of the 1st person Set M prefixes
mentioned earlier, Old Georgian agreement for number was closely correlated with case: ERG and
NOM case NPs could control number agreement (NA), while DAT NPs could not [Chikobava
1968:162-227]. This constraint applied even to so-called inverse constructions (e.g. active verbs
in series III) where the agent NP is assigned DAT case and controls Set M agreement, while the
patient/theme NP is assigned NOM case and controls Set V agreement. Only the latter could
control NA in Old Georgian. This is illustrated in {3}, in which series I and series III forms of
active verbs are juxtaposed. In the second clause in {3} [cited in Chikobava 1968:171], the DAT
NP denoting the agent (“those who”), although plural and animate, is not crossreferenced for
number by the verb; on the other hand, the NOM patient NP (“hands”) is. (In this example, NA
with the NOM NP is marked twice: through Set V 3pl agreement, and also through en
agreement). Compare this to the agreement pattern in the first clause in {3}, where the NOM
agent controls NA and the DAT patient does not.

{3} romel-n-i-me 0-i-cem-d-es mk’erd-ta mat-ta,

[which-PL-NOM hit-Ia-3pl/3 breast-DATPL their-DATPL
da romel-ta-me aƒ-0-e-p’q’r-n-es qel-n-i mat-n-i zec-ad
and which raise-IIIa-3pl/3/pl hand-NOM-PL their-NOM-PL heaven-ADV]
“...those who<NOM> beat their breasts, and those who<DAT> had raised their
hands<NOM> heavenward.” [c’art’q’. ierus. 27:28]

Although these agreement patterns remained normative for literary Georgian through the 18th
century, verb forms indicative of the norms operative in the various dialects of the spoken
language are attested in texts as early as the 11th century. Sarjveladze [1981, 1984:566-8] has
documented some of these early attempts to code the plurality of prominent DAT arguments in
the verb. Consider the following examples:

{4} amat mo-0-u-g-i-an didebuleba-y sul-isa-y [ms Jer-73 (XI c.) 127r]
[they-DAT win-IIIa-3/3pl majesty-NOM soul-GEN-NOM]
“They have won majesty of soul.” <cp early Old Georgian mo-0-u-g-i-e-s>
 Geography of q’e (K. Tuite, NSL 5) — page 3

{5} g-i-txrob-d-i-t aramed rametu tkwen tana v-i-q’av-0 [ms Jer-32:137r,24]

[tell-Ia-1/2pl rather that you(pl)-DAT with be-IIp-1]
“I was telling you rather that I was with you.” <early OG g-i-txrob-d-i>

{6} £e-0-e-£in-a-t priad [I Kings 7:7]

[be-afraid-IIp-3/3pl greatly]
“They were very much afraid.” <early OG £e-0-e- £in-a>

Two distinct strategies are represented here. Example {4} comes from a document
[psevdomak’ari megwip’t’elis sc’avlani “The teachings of Pseudo-Makarias the Egyptian”]
which seems to be of southwest Georgian origin [Sarjveladze 1984:566]. Agreement with the 3pl
DAT “real subject” is coded by a 3rd plural Set V suffix (-an) in conjunction with the expected
Set M 3rd person prefix (0- prevocalically). In examples {5} and {6} the Set V suffix associated
with 1st and 2nd person plurality (-t) is used to mark plural NA with a 2nd person addressee
{5} and a 3rd person experiencer {6}. This second strategy is now normative in modern literary
Georgian; the first strategy is still used by speakers of several western Georgian dialects as well
as the Kartvelian languages Laz, Mingrelian and Svan [Dzidzishvili 1958, K’iziria 1974,
Chikobava 1936:94-102].

2. Use of q’e. Less than a century after the sentences in {4-6} were written, the particle q’e was
being used — sporadically — in written Georgian for a similar purpose. One of the earliest
documents where q’e is attested is a charter granted by King David the Builder to the monastery
at Shio-Mghvime, dated 1125 [Dzidziguri 1984:57]:

{7} twit adr-it-gan mamaoba-sa 0-e-c’er-a-q’e saebisk’op’oso-d

[even early-INS-from father-(coll.)-DAT write-IIIa-3/3/pl episcopacy-ADV]
“From the very beginning it was ascribed to the fathers as an episcopacy.”

Sarjveladze [1984:562-7] gives several other examples from 12th century documents:

{8} esodeni £i£-i da-g-i-p’q’rob-s-q’e tkwen [ms Jer-22:25r,20b]

[so-much fear-NOM seize-Ia-3/2/pl you(pl)-DAT]
“So much fear will seize you.”

{9} ray-mca ara 0-e-kmn-a-q’e [ms A-52:38r,12]

[what-NOM not do-IIIa-3/3/pl]
“which they would not have done”

{10} egret amat-ca m£widoba-sa mi-s-cem-s brdzol-isa-twis,

[thus them-DAT-also peace-DAT give-Ia-3/3 struggle-GEN-for
romel-i mat-da mimart 0-a-kwn-d-a-q’e [ms A-52:146r,10-11]
which-NOM them-ADV toward have-IIp-3/3/pl]
“Thus he will grant them peace instead of the contention which he had toward them.”
 Geography of q’e (K. Tuite, NSL 5) — page 4

The definition of q’e most often given by Kartvelologists is that it codes the plurality of
“grammatical objects,” i.e. arguments crossreferenced by Set M person markers, which can be
assigned either DAT or — in the case of transitive verbs in the aorist/optative series — NOM
case.2 In sentences {7} and {9} the DAT agent NP of a transitive verb which has undergone
inversion is coded for plurality in the verb. Note that notional rather than formal plurality is
marked in {7}, where q’e crossreferences the collective mamaoba “group of fathers [priests].” In
{8} the 2pl direct object controls NA in q’e. Occasionally one comes across instances where the
motivation underlying the appearance of the clitic is difficult to establish. Sarjveladze [1984:568]
views the q’e in {10} as “functionless.” It may be that the plurality of the oblique argument
matda mimart “toward them” is responsible for the occurrence of q’e. (Another possibility is that
it marks habitual Aktionsart, as in some of the modern dialects which we will discuss further on.)
The early 13th century epic poem “The knight in the panther’s skin” by Shota Rustaveli
contains two tokens of q’e, both, interestingly enough, cliticized to the noun preceding the verb
rather than to the verb itself.

{11} mi-0-xoc-d-es da mi-i-srod-es, mindor-s sisxl-ita mi-0-a-sxmid-es;

[slaughter-Ia-3pl/3 and shoot-Ia-3pl field-DAT blood-INS smear-Ia-3pl/3
ra isar-i da-0-e-lev-i-s, mona-n-i-q’e mi-0-a-rtmid-es. [75:2,3]
what arrow-NOM exhaust-Ip-3/3 servant-PL-NOM-pl proffer-Ia-3pl/3]
“They slew and shot, drenching the field with blood;
Whenever their arrows ran out, servants brought them more.”

In this excerpt, q’e marks the plurality of the recipients of the fresh arrows (Rostevan and
Avtandil), who are coded by zero anaphors throughout. (An overt NP denoting them would be
assigned DAT case by the verb miartmides). Before we begin looking at the data from modern
Georgian dialects, I shall lay out some initial observations concerning q’e:
[a] For the most part — excluding cases like {10} — this clitic codes the notional plurality of
grammatical objects, i.e. arguments controlling Set M (object) agreement in the verb.
[b] In the majority of Old Georgian attestations — and all modern ones — q’e is attached to
the end of the verb. Instances like {11} indicate that at one time this morpheme was more
particle-like and less suffix-like than it is now.
[c] Once the old inclusive/exclusive distinction marked by the Set M prefixes gw- and m- was
lost, a coding asymmetry resulted: for arguments controlling Set M person agreement, number
agreement was obligatory for 1st person, but not possible for 2nd or 3rd person NPs. When, in
literary Georgian, the semantic range of the Set V suffix -t was extended to include the coding of
plural grammatical objects, it was (and still is) employed in complementary distribution to the
prefix gw- giving the following “balanced” Set M paradigm:

2
In his dialect dictionary Ghlont’i [1984:605-6] has collected several such definitions, e.g.: (a) “particle placed after
the verb in the sentence to indicate object plurality [V. Beridze]; (b) particle used to express plurality of an object in
the dative or nominative case, or of a subject in the dative case [GTK:679].
 Geography of q’e (K. Tuite, NSL 5) — page 5

{12} Set M affixes (Modern Standard Georgian)

1sg: m- 1pl: gw-
2sg: g- 2pl: g- -t
3sg: h/0- 3pl: h/0- -t

Usage of q’e shows the same complementarity: it is never used to code the plurality of a 1st
person grammatical object.
[d] Unlike person agreement, number agreement in q’e is not obligatory when a plural
argument in the appropriate formal relation to the verb is present. For example, in “The knight in
the panther’s skin” only two of the hundreds of plural DAT arguments are correlated with q’e.
The same is true for Set M -t in the standard language and in those dialects where it is used: NA
with 2nd and 3rd person arguments controlling Set M person agreement is not obligatory,3 while
agreement with 1st plural arguments in gw- — and for that matter, agreement in -t with 1st and
2nd plural NPs in the Set V paradigm — is.

3. q’e in the modern Georgian dialects. With the above as background, let’s turn to some
data from the modern Georgian dialects. The accompanying map shows the location of the
principle dialects and subdialects, as described in Gigineishvili, Topuria and K’avtaradze 1961 —
henceforth abbreviated GTK.
These can be divided into five main dialect groups:

{13} I. Northeast dialects

Moxevian
Mtiuletian-Gudamaq’rian
Xevsurian
Pshavian
Tushetian
II. Eastern dialects
K’axetian
Ingiloan
Fereidanian
Tianetian
III. Central dialects
Kartlian
Javaxian
Mesxian
3
Of course, the factors conditioning the appearance of this morpheme as a Set M NA marker for 2nd person and for
3rd person arguments are very different. Although NA with 2pl grammatical objects was relatively uncommon in
late Old Georgian, in modern standard Georgian it is for all intents and purposes obligatory; its appearance is only
blocked by a morphological rule permitting no more than one plural suffix [Tschenkeli 1958:354; Shanidze
1953:184-6]. So, instead of the expected *g-i-c’er-en-t “they are writing to you(pl),” the form g-i-c’er-en “they are
writing to you(sg or pl)” must be used. The appearance of Set M 3pl -t is dependent on the same factors that are
relevant for q’e,though the threshold varies from dialect to dialect. In the modern standard language, the basic rule is
that 3rd person “real subjects” have the potential to control number agreement, but other 3rd person arguments do
not [Chikobava 1967; K’vach’adze 1977:96-111; Harris 1978]. In practice, NA with 3pl experiencers, benefactives,
and — rarely — recipients and addressees is widely attested in written Georgian [see Tschenkeli 1958:486-90;
Jorbenadze 1981; K’iziria 1985; Tuite 1987].
 Geography of q’e (K. Tuite, NSL 5) — page 6

IV. Southwest dialects

Gurian
Ach’arian
Imerxevian
V. Northwest dialects
Imeretian
Lechxumian
Rach’an

As a Set M plural marker q’e is not found in the conservative northeast dialects. We will begin
with the eastern dialects.
3.1. Ingiloan. Ingiloan, a Georgian dialect spoken on the Azerbaidjanian side of the Alazani
River, makes extensive use of q’e; some examples are given below:

{14} insn-eb da-g-i-k’lav-q’ [Imnaishvili 1966:187]

[man-PL-NOM kill-IIIa-3/2/pl]
“You(pl) have killed the men.” <NA with 2pl DAT agent of active verb>

{15} danar™en bi™’-eb-s da dad-eb-s 0-u-k’eteb-en plav-s, ™ey-s,

[remaining boy-PL-DAT and bridesmaid-PL-DAT make-Ia-3pl/3 pilaf-DAT tea-DAT
0-a-™’mev-en-q’e, 0-a-levineb-en-q’e. [GTK:244]
feed-Ia-3pl/3/pl cause-to-drink-Ia-3pl/3/pl]
“For the remaining boys and bridesmaids they prepare pilaf and tea, feed them and give
them something to drink.” <NA with 3pl DAT recipients>

The second example is particularly interesting, in that it gives some idea of how q’e is used in
connected speech. The first verb in the series of three — uk’eteben — does not agree in number
with its indirect object, while the second and third verbs do. There is considerable evidence that
discourse-prominence factors play an important part in determing the occurrence of certain types
of NA in Georgian, including the literary language. In this instance, the newly-introduced topic
“boys and bridesmaids” does not control NA in q’e until it becomes presupposed information, as
indicated by the use of null anaphora.
3.2. Fereidanian. The isolated eastern dialect of Fereidan, spoken in Iran by the descendents
of a group of Georgians captured and resettled in the 17th century, is noted for its frequent use of
q’e. As in Ingiloan, discourse topicality is an important factor; this is illustrated in {16}:

{16} tko ro “namaia gamo-(v)-a-ƒidz-e-q’e-o, minac em sxo otageb™i c’vanano”

[say-IIa-3 that good-is wake-IIa-1/3/pl-QT who-NOM the other rooms-in lie-3plQT
— ro keniz-eb-i i-q’v-nen. £a-vid-a da gamo-0-a-ƒidz-a-q’e.
that maid-PL-NOM be-IIp-3pl enter-IIp-3 and wake-IIa-3/3/pl
ema-t h-k’itx-a-q’e ro: “em dedak’ac-tan min ari ro c’ev-s-o?”
them-DAT ask-IIa-3/3/pl that this woman-with who is that lie-Ip-3-QT
emeeb-ma 0-u-txr-es ro: “es dedak’ac-i ari erti tajer-is dedak’ac-i-o,
they-ERG tell-IIa-3pl/3 that this woman-NOM is one merchant-GEN woman-QT
 Geography of q’e (K. Tuite, NSL 5) — page 7

da ans-i-c tavis bi™’-i-a-o....” ema-s balki go-0-u-xard-a,

and this-NOM-also her boy-NOM-is-QT this-DAT very-much rejoice-IIp-3/3
ama-0-i-ƒ-o rakteni pul-i, ™o-0-u-q’ar-a-q’e emeeb-sa:
take-out-IIa-3/3 so-much money-NOM throw-IIa-3/3/pl them-DAT
“£a-d-i-t, gamo-0-a-ƒidz-e-t-o.” em keniz-eb-ma-c, pul-i k’i go-0-u-xard-a-q’e,
enter-IIp-2pl wake-IIa-2pl/3-QT the maids-ERG money indeed rejoice-IIp-3/3/pl
amma 0-e-£inod-a-q’e ro: “em dros min ariese-o, ro ™amo-sul-a-o?!”
but fear-IIp-3/3/pl that this time-DAT who-NOM is this-QT that come-IIIp-3-QT]
He said: “It’s a good idea to wake up the people who are sleeping in the other rooms” —
where the maids were. He went in and woke them up. He asked them: “Who is lying with
this woman?” They told him: “This woman is the wife of a merchant, and that is her
son.” He was happy (to hear this) and took out a large sum of money and tossed it to
them: “Go in and wake them up.” The maids were indeed very happy about the money,
but still frightened: “Who is this person who has come to us at this hour?” [GTK:266]

The interesting thing to note in this passage is the use of q’e to mark NA with a new topic, even
before it is first explicitly mentioned. This use of the particle seems to be more frequent than the
more locally-determined thematicity conditioning exemplified in {15}. Also, note that the
speaker did not use q’e to mark NA with the plural object of gamoaƒ idzeto “wake up <the
woman and the boy>“ in the 7th line, evidently preferring to reserve Set M NA for the primary
topic “maids.” Now consider the short Fereidanian passage given in {17}:

{17} zog dƒe-sa sakme ver ™a-i-gd-i-s xel-™i, sadil-ze £amo-id-i-s-q’e

[some day-DAT affair-NOM cannot grasp-IIa-3/3 hand-in dinner-toenter-IIp-3-q’e
da 0-u-txr-i-s-q’e q’ein-is q’or-sa: dƒe-s sakme ar ™a-m-i-vard-a-o.
and tell-IIa-3/3-q’e king-GEN girl-DAT day-DATaffair-NOM not fall-IIp-3/1-QT
zog dƒe-s ro e-mu£avn-i-s-q’e saƒamo-s em q’ein-is q’or-ma h-k’itx-i-s-q’e:
some day-DAT that work-IIIa-3-q’e eve-DAT the king-GEN girl-ERG ask-IIa-3/3q’e
dƒe-s sad i-q’av-o, mis-tana i-mu£avob-d-i-vo? [GTK:260]
day-DAT where be-IIp-2-QT who-GEN-with work-Ia-2-QT]
“On those days when he can’t find something to do, he comes to dinner and tells the
king’s daughter: I didn’t find a job today. On those days when he had been working, in the
evening the king’s daughter asks him: Where were you, with whom were you working?”

It is clear from the context that no plural grammatical object is being referred to — in fact, some
of the verbs are monopersonal. Examination of Fereidanian texts indicates these “anomalous”
occurrences of q’e are most often with verbs in the permansive/habitual aspect, as in the above
example.4 This use of the particle is by no means limited to Fereidanian. Chikobava [1937:54-5]
reports that an enclitic particle k’e — which he claims is cognate with q’e — is frequently used in
conjunction with the past habitual in the northeast dialect of Mtiuleti.5
4
Don Stilo, who is as far as I know the only American to have done fieldwork on the Fereidanian dialect in Iran,
reports that in the texts he collected [as yet unpublished] q’e was employed only as an aspectual marker.
5
The same particle, in the same contexts, is observed in the neighboring Moxevian dialect as well. The texts in
 Geography of q’e (K. Tuite, NSL 5) — page 8

3.3. K’axetian. In most of the K’axetian dialect region, -t is the Set M plural NA affix of choice,
and it used pretty much the same way q’e is in Ingiloan. In eastern K’axeti (Q’vareli and Gurjaani
Raions) q’e is used, as well as -t, which is said to be supplanting it [Chikobava 1968:277].
3.4. Central and southwest dialects. Number agreement in q’e is not — or longer — found in
the central and southwest dialect area.
3.5. Imeretian. Moving northward into Imereti we observe that of the two major subdialects
spoken in this region, Upper Imeretian prefers -t and Lower Imeretian q’e for NA with
prominent grammatical objects, though both morphemes are attested in both dialects.In many of
the Imeretian texts that I have read, it appears that -t is preferred for grammatical objects that
have “real subject” status, i.e. the DAT case-marked agents of active verbs that have undergone
inversion, and the experiencer arguments of verba sentiendi. For other topical arguments
controlling Set M person agreement, NA in q’e is generally used. Compare the use of the two
plural NA markers in this Upper Imeretian example:

{18} k’itx-es rac 0-u-ndod-a-t; 0-u-txr-a-q’e [GTK:458]

[ask-IIa-3pl/3 what-NOM want-IIp-3/3pl tell-IIa-3/3/pl]
“They told him what they wanted; he said to them...”

For the most part, NA in Imeretian is as in Ingiloan and K’axetian. In western Imereti, most
notably in C’uluk’idze Raion, a surprising extension of the usage range of q’en (a variant form of
q’e) is observed (Dzidziguri [1940:164, 1954:152]; K’ublashvili [1985:140-2]):

{19} ame-0-i-q’van-o-c’-q’en kal-eb-ma q’vel-i-o [Dzidziguri 1940:164]

[bring-up-IIa-3/3/pl woman-PL-ERG cheese-NOM-QT]
“The women should bring up the cheese.” <NA with 3pl ERG agent>

{20} £e-me-xvec’eb-od-e-c’-q’en [K’ublashvili 1985:141]

[plead-Ip-3/1/pl]
“They would be pleading with me.” <NA with 3pl NOM agent>

{21} ver dzlebulob-s-k’en amden sakme-s ertad [ibid]

[cannot manage-Ia-3/3/pl this-many matter-DAT together]
“They cannot manage to do so many things at the same time.”
<NA with 3pl NOM agent>

It may be the case that the range of q’en/k’en has been expanded by analogy with the range of the
3pl Set V agreement marker -en. In the southwest dialects, -en is used to code NA with DAT-
case subjects which control Set M agreement; compare {4} above, which is believed to have
originated in southwest Georgia. The suffix, therefore, codes plural real subjects, regardless of

GTK contain several examples:

{i} gazapxul-is p’ir-ci ker-i da-v-tes-i-t-k’e da stwel-ze mo-v-mk’-i-t-k’e
[spring-GEN mouth-in barley-NOM sow-IIa-1pl/3 and vintage-at mow-IIa-1pl/3]
“In early spring we (habitually) sow barley, and at vintage time we reap it.”
 Geography of q’e (K. Tuite, NSL 5) — page 9

their case. In C’uluk’idze Raion it appears that q’en/k’en has taken on similar characteristics.
3.6. Lechxumian and Rach’an. The semantic range of q’e in Lechxumian and Rach’an is
about the same as in Ingiloan and (most of) Imeretian.6 NA in q’e is found here in essentially the
same contexts as in the other northwest dialects. The following passage was recorded in the
Rach’an village Ch’iori:

{22} ™agele-et 0-q’av-d-a-q’e bat’on-i. katam-i-c rom da-0-e-k’l-a-s-q’e,

[Chagele-DATPL have-IIp-3/3/pl lord-NOM chicken-NOM that kill-IIIa-3/3/pl
naxevar-i bat’on-i-tvin unda mi-0-e-c-e-s-q’e
half-NOM lord-GEN-for must give-IIIa-3/3/pl
mara ertxel mo-0-u-vid-a-q’e iaƒli£i:
but once come-IIp-3/3/pl mishap-NOM
bat’on-eb-i i-txov-d-en met’-s; met’i ama-t ar 0-kon-d-a-q’e.
lord-PL-NOMdemand-Ia-3pl/3 more-DAT more-NOMthey-DAT not have-IIp-3/3/pl
mo-vid-en, a-0-u-t’q’d-en-q’e da c’a-i-q’van-es-q’e tavisi ak’van-eb-ita
come-IIp-3pl attack-IIp-3pl/3/pl and take-IIa-3pl/3/pl their cradle-PL-INS
da tavisi baƒv-eb-it kal-eb-i-ca da k’ac-eb-i-c.
and their child-PL-INS woman-PL-NOM and man-PL-NOM-too] [GTK:524]
“The Chageles had a (feudal) overlord. Whenever they would kill a chicken, they had to
give half to the overlord. One day misfortune came upon them: The overlords demanded
more, but they did not have more (to give). So they [i.e. the overlords] came, attacked
them and carried them off, women and men, with their cradles and their children.”

3.7. Factors relevant to use of q’e. On the basis of the Georgian dialectological
materials available to me, the following general observations can be made concerning the
morpheme q’e:
[1] It almost invariably crossreferences an NP with an animate denotatum. Only one of the
over 200 occurrences of q’e in my sample coded NA with an inanimate. This is to be expected, of
course, given the tendency observed in most of the Georgian-speaking area for all NA processes
to be sensitive to animacy.
[2] Animacy is not the whole story: topicality is also an important criterion for number
agreement. For example, in the corpus I examined, in 86 of 121 instances of q’e NA with a (non-
real-subject) direct or indirect object the latter was represented by a null anaphor.

4. Parallels in Georgian morphosyntax. In the course of its recorded history Georgian has
undergone significant changes. Among the more extensive of these changes has been the
realignment of the number agreement component of the syntax. In every modern dialect, save the
most conservative dialects of northeast Georgia, NA, however it might be marked, is sensitive to

6
According to Dzidziguri [1954:230-1] for NA with plural grammatical objects in Rach’an both -t and q’e are used
when the grammatical subject is 3rd person. With a 1st or 2nd person gram. subj., only q’e is used for this purpose,
to avoid confusion with the primary use of -t as a 1st/2nd Set V plural NA marker: hence, (sen) 0-u-txar-i-q’e (mat)
“you(sg) told it to them,” but not *(sen) 0-u-txar-i-t (mat); the latter verb can only mean “you(pl) told it to
him/her/them.”
 Geography of q’e (K. Tuite, NSL 5) — page 10

the animacy, and in some cases topicality, of the argument concerned. In Old Georgian, formal
number was the primary determinant of NA: if the NP in question was marked with the
appropriate plural suffix [NOM -n-i, ERG -t(a)], NA generally occurred; otherwise 3sg
agreement was marked, even if the argument was notionally plural, or marked with the pluralizer
-eb- [Shanidze 1982:182-3; Harris 1985:210-3; for exceptions see Sarjveladze 1984:543-64]. In
modern standard Georgian, -eb- is the unmarked plural suffix, and NA with plural 3rd person
NPs controlling Set V person agreement is largely dependent on animacy (Chikobava 1968:272-3;
K’vach’adze 1977:99-104).7

{23} mucela-s k’bil-eb-ma k’ac’k’ac’-i da-0-u-c’q’-o [Vazha-Pshavela]

[M.-DAT tooth-PL-ERG chatter-NOMbegin-IIa-3sg/3]
“Mucela’s teeth began chattering.”

{24} jarisk’ac-eb-£i ga-i-sm-i-s xm-eb-i [Chikobava 1968:272]

[soldier-PL-in is-heard-Ip-3sg voice-PL-NOM]
“Voices are heard from among the soldiers.”

Chikobava comments that “ga-i-sm-i-an xm-eb-i [is-heard-Ip-3pl voice-PL-NOM] is

unacceptable in modern Georgian, although in Old Georgian i-sm-i-an qma-n-i [is-heard-Ip-3pl
voice-PL-NOM] would have been normal” [ibid:272]. Likewise, in the Mountain Rach’an village
Glola, the one place outside of the northeast dialect area where NA in -(e)n- with NOM direct
objects is still found, animacy appears to be a necessary condition for this agreement to occur.8
Furthermore, there is evidence that in at least some dialects, notably Lower Imeretian [K’iziria
1974:81] even NA with animate subjects is to some degree dependent on topicality. Consider the
following excerpt from a Lower Imeretian folk tale [in GTK:474] about three brothers — two
smart and one foolish — seeking their fortunes. The foolish brother is spending the night in a
grain hopper.

7
Consider the following near-minimal pair, from two Old Georgian translations of the gospels, which demonstrate
the syntactic properties of the -t(a) and -eb-ma ERG plural forms; only the former governs NA:
{i} k’ac-ta mat vitarca i-xil-es sasc’aul-i igi ... [John 6:14 (Op’iza)]
[man-ERGPL the-ERGPL as see-IIa-3pl/3 miracle-NOM the-NOM]
{ii} k’ac-eb-man man vitarca i-xil-a sasc’aul-i igi ... [John 6:14 (Adish)]
[man-PL-ERG the-ERG as see-IIa-3sg/3 miracle-NOM the-NOM]
“When the men saw the miracle ...”
In modern Georgian, the plural suffixes -n-i [NOM] and -t(a) [ERG, DAT, GEN] are stylistically marked, restricted
for the most part to literary or officialese registers, and to fixed phrases.
8
Animacy was not a relevant factor for determining the occurence of -en- agreemen in Old Georgian. For example,
in the 6th century Xanmet’i texts edited by Molitor [1956], of 45 instances of -en- NA with transitive direct objects,
only half of them [22] refer to animate beings; likewise, in a sample of texts from the northeast dialect area (where
the NA mechanism is basically the same as in Old Georgian) less than half [21 of 44] of the direct objects control-
ling NA in -en- have animate referents. By contrast, in the Glola Rach’an texts that I have examined, almost all
such NPs [24 of 26] denote animate beings.
 Geography of q’e (K. Tuite, NSL 5) — page 11

{25} dila-ze gare-dan pxak’a-pxuk’-it mo-vid-a tagv-eb-i

[morning-at outside-from scratching-INS come-IIp-3sg mouse-PL-NOM
da ga-0-a-ps-en xvimir-i pul-it.
and fill-IIa-3pl/3 hopper-NOM money-INS
game-0-e-k’id-a sulel-i tagv-eb-s da da-0-a-t’ueb-i-a pul-eb-i.
chase-IIp-3/3 foolish-NOM mouse-PL-DAT and leave-IIIa-3/3 money-PL-NOM
ga-0-a-ps-o didi sapule pul-it da c’e-vid-a sax- £i.
fill-IIa-3/3 big money-bag-NOM money-INS and go-IIp-3 home-in
™’k’viani dzman-eb-i sax-£i da-0-xt-a.
smart brother-PL-NOM home-in meet-IIp-3sg/3
veraper-i ver 0-e-£on-a-t da ke da-brunebul-iq’v-en calieri.
nothing-NOM cannot find-IIIa-3/3pl and indeed return-IIIp-3pl empty]
“In the morning the mice came from outside with a scratching sound and filled the hopper
with money. The foolish brother chased away the mice and the money was left behind.
He filled a large bag with the money and went home. The smart brothers met him at home.
They had not found anything, and had returned empty-handed.”

Two formally plural animate NPs appear in this passage: tagv-eb-i “mice” and dzman-eb-i
“brothers.” Although both serve as both grammatical and “real” subjects of their respective verbs
at their first appearance, both control 3rd singular Set V agreement — or perhaps it would be
more accurate to say: 3rd person Set V, not specified for plurality. Verbs in the following clauses,
however, crossreference these same arguments with specifically 3pl agreement markers.
Some aspects of the use of the NA marker -t with arguments controlling Set M agreement in
Modern Standard Georgian also remind us of the discourse-related phenomena mentioned above.
Consider the following excerpt from a recently published short story:

{26} me megobar-i ƒvt-is c’q’alob-it bevr-i m-q’av-s,

[I-DAT friend-NOM god-GEN grace-INS many-NOM have-Ip-3/3
tanac iset-eb-i ar-ian, st’umroba tu gan-i-zrax-es,
besides that-PL-NOM be-Ip-3pl guesthood-NOM if intend-IIp-3
veranairi amind-i ver da-0-a-brk’oleb-t.
no-kind weather-NOM cannot hinder-Ia-3/3pl]
“By the grace of God I have many friends; furthermore, they are the sort of people that,
if paying a visit is on their minds, no kind of weather can hinder them.” [J. Karchxadze
Mnatobi #1:70 (1986)]
My consultants in Tbilisi told me that the topicality and animacy of the patient (“my friends”)
as opposed to the rhematicity and inanimacy of the agent (“weather”) contributed to the
occurrence of NA in -t with the former.9 If the patient NP is less highly presupposed, plural NA
is less acceptable:

9
According to I. K’ik’nadze 1983, the popular Georgian novelist Otar Ch’iladze makes especially frequent use of
the Set M plural suffix -t to code the number of topical 3rd plural arguments, even when these are not serving as
subjects in the usual sense of the term (and therefore number agreement would not be expected according to the
norms of modern literary Georgian).
 Geography of q’e (K. Tuite, NSL 5) — page 12

{27} veranairi amind-i ver da-0-a-brk’ol-eb-s/??-t st’umr-eb-s

[no-kind weather-NOM cannot hinder-Ia-3/3 guest-PL-DAT]
“No kind of weather can hinder the guests.”

5. Conclusion. In evaluating number agreement in the various modern Georgian dialects,

we can list several hierarchies which are relevant to the process:

{28} a. DISCOURSE-FUNCTIONAL: topical > non-topical

b. REFERENTIAL: animate > inanimate
c. FORMAL: 1st/2nd > 0 > pronoun > other NPs

To some extent these hierarchies overlap with each other, of course. Speech-act participants are
almost always animate, and both the topicality and formal NP-type hierarchies reflect degree of
presupposedness [Silverstein 1976,1981]. The Georgian dialects differ in which specific
hierarchies from the list are of importance, and where they draw the line between forms that do
and do not have the potential of controlling number agreement. For Set V number agreement, in
most instances, animacy is the major criterion, though as we saw, topicality can play a role in
Imeretian. For Set M agreement, the number of relevant factors is greater in most dialects, and the
cut-off points are less clear cut. In the case of the topicality criterion, for example, there is
variation — perhaps only idiolectal — concerning the status of newly-introduced arguments that
are destined for a prominent role in the discourse.
Secondly, one can describe the morphosyntactic systems of individual languages or dialects in
terms of the manner of distribution of what one might call “syntactic privileges” within the
clause: which argument classes receive these privileges, and how asymmetric the distribution is.
In early Old Georgian, the privilege of controlling number agreement was distributed according to
formal criteria: the case assigned an NP [ERG or NOM, not DAT] and the set of person
agreement markers it controlled [Set V, not Set M]. Other syntactic privileges, such as ability to
bind reflexive and reciprocal anaphors and likelihood of participating in cross-clausal reference
maintenance, were determined according to semantic criteria, in particular animacy and relative
position on a hierarchy of deep-case roles ranked from most to least agentive 10 [Harris 1981,
Tuite in progress]. One of the important changes in Kartvelian diachronic syntax, as pointed out
by Cole et al 1980 (see also Aronson 1976), has been the reapportionment of these privileges in
the direction of greater convergence onto one type of argument. More precisely, the number
agreement process — that is, a very local syntactic privilege — has realigned to accord more
closely with prominence in the less local, cross-clausal domain. For this purpose, number
agreement morphemes were, in a sense, “recruited” to code prominent arguments which had not
earlier controlled number agreement. The marker -an used in {4} and the -t in {5} and {6} came
from Set V. The origins of q’e are still not known with certainty, though the use of this particle
to code iterative and permansive Aktionsart in Fereidanian and Mtiulian may be an important
clue toward the solution of this mystery.

10
In Harris’ terms, these operations are sensitive to “initial subjecthood,” which is in most cases predictable from
the agentivity hierarchy [Harris 1981:258] though, as in other languages, exceptions occur [cp. Rosen 1984].
 Geography of q’e (K. Tuite, NSL 5) — page 13

Acknowledgements
Much of the investigation for this paper was undertaken in Tbilisi from September 1985 to June
1986, during which time I was a participant in the exchange of American and Soviet researchers
administered by the International Research and Exchanges Board and the Ministry of Higher
Education of the USSR. Among the many scholars with whom I discussed the issues examined
here I especially wish to acknowledge Shukia Apridonidze, Nani Ch’anishvili, Aleksandre
Ghlont’i, Besarion Jorbenadze, Damana Melikishvili and Meri Nik’olaishvili in Tbilisi, and
Howard Aronson in Chicago. Special thanks go to Dee Ann Holisky, whose extensive marginal
comments led to improvements in both the content and its packaging. If any errors remain
despite all of this assistance, they are my fault.

References
abbreviations
EnIMKI = Ak’ad. N. Maris saxelobis enis, ist’oriisa da mat’erialuri k’ult’uris inst’it’ut’is
moambe (Bulletin of the Academician N. Marr Institute of language, history and material culture)
GTK = Gigineishvili, et al. 1961.
IKE = iberiul-k’avk’asiuri enatmecniereba (Ibero-Caucasian linguistics)
IKEC = iberiul-k’avk’asiuri enatmecnierebis c’elic’deuli (Annual of Ibero-Caucasian linguistics)
KESS = kartvelur enata st’rukt’uris sak’itxebi (Issues in the structure of the Kartvelian languages)
TSUG = tbilisis saxelmc’ipo universit’et’is gamomcemloba (Tbilisi State University Press)

Aronson, Howard 1976 “Grammatical subject in Old Georgian” Bedi Kartlisa 34: 220-231
Chikobava, Arnold 1936 ™’anuris gramat’ik’uli analizi t’ekst’ebiturt (A grammatical analysis of
Laz, with texts) [Tbilisi: Mecniereba]
—— 1937 “mtiuluris taviseburebani” (Characteristics of the Mtiuletian dialect) EnIMKI 2 #1: 43-
65
—— 1954 “mravalobitis supiksta genezisatvis kartul£i” (On the genesis of the Georgian plural
suffixes) IKE 6: 67-76
—— 1967 “Gruzinskij jazyk” in Jazyki narodov SSSR 4: 22-61
—— 1968 mart’ivi c’inadadebis p’roblema kartul£i, I: kvemdebare-damat’ebis sak’itxi dzvels
kartul£i (The problem of the simple sentence in Georgian, I: the issue of subject and object
in Old Georgian) [Tbilisi: Mecniereba]
Cole, P., W. Harbert, G. Hermon and S. N. Sridhar 1980 “The acquisition of subjecthood”
Language 56: 719-743
Deeters, Gerhard 1930 Das kharthwelische Verbum: vergleichende Darstellung des Verbalbaus
der sudkaukasischen Sprachen [Leipzig: Markert und Petters]
Dzidziguri, Shota 1937 “kartuli enis mtara™’uli dialekt’is dziritadi taviseburebani” (Chief
 Geography of q’e (K. Tuite, NSL 5) — page 14

characteristic of the Mountain Rach’an dialect of the Georgian language) EnIMKI 2 #1:
69-109
—— 1954 dziebani kartuli dialekt’ologiigan (Topics in Georgian dialectology) [Tbilisi: Ganatleba]
Dzidziguri, Sh. (chief ed.) 1984 kartuli ist’oriuli sabutis k’orp’usi I: IX-XI ss. (Corpus of
Georgian historical documents I: 9th-11th c.) [Tbilisi: Mecniereba]
Dzidzishvili, Meri 1958 “gramat’ik’ul movlenata t’endenciebi gurul £i” (Grammatical tendencies
in Gurian) IKE 9-10: 193-200
Gigineishvili, Ivane, Varlam Topuria and Ivane Kavtaradze 1961 kartuli dialekt’ologia I (Georgian
dialectology, I) [Tbilisi: TSUG]
Ghlont’i, Aleksandre 1984 kartul k’ilo-tkmata sit’q’vis kona (Georgian dialect dictionary) [Tbilisi:
Ganatleba]
Harris, Alice C. 1978 “Number agreement in modern Georgian” in B. Comrie (ed.) Classification
of grammatical categories [Edmonton: Linguistic Research] pp 75-98
—— 1981 Georgian syntax: a study in relational grammar [NY: Cambridge U Press]
—— 1985 Diachronic syntax: the Kartvelian case (Syntax and syntax 18) [NY: Academic Press]
Jorbenadze, Besarion 1981 “Principy stanovlenija inversionnyx glagolov v gruzinskom jazyke”
IKEC 8: 66-77
K’ik’nadze, Ineza 1983 “erti enobrivi t’endenciis £esaxeb tanamedrove kartul£i” (Concerning a
linguistic tendency in modern Georgian) in B. Jorbenadze (ed.) saenatmecniero dziebani
[TSUG] pp 75-87
K’iziria, Ant’on 1974 “kvemdebare-£emasmenlis urtiertoba kartuli enis dasavluri dialekt’ebis
mixedvit” (The subject-predicate relation in the western dialects of Georgian) KESS 4: 75-
91
—— 1985 “obiekt’is mier zmnis £etanxmeba mravlobit ricxv£i tanamedrove kartul£i” (Object-verb
plural number agreement in modern Georgian) IKE 24:100-112
K’ublashvili, K’lara 1985 kartuli enis kvemoimeruli dialekt’i (The Lower Imeretian dialect of the
Georgian language) [Tbilisi: TSUG]
K’vach’adze, Leo 1977 tanamedrove kartuli enis sint’aksi (Contemporary Georgian syntax)
[Tbilisi: Ganatleba]
Met’reveli, Teimuraz 1978 “Nochmals zur Kategorie von Inklusiv und Exklusiv im
Altgeorgischen” Georgica 1: 23-29
Molitor, Joseph (ed.) 1956 Monumenta Iberica antiquiora: Textus chanmeti et haemeti ex
inscriptionibus, Bibliis et patribus Corpus Scriptorum Christianorum Orientalium
166 subsidia 10
Sarjveladze, Zurab 1981 “Zur Geschichte einer syntaktischen Erscheinung” Georgica 4: 86-87
—— 1984 kartuli salit’erat’uro enis ist’oriis £esavali (An introduction to the history of the
 Geography of q’e (K. Tuite, NSL 5) — page 15

Georgian literary language) [Tbilisi: Ganatleba]

Shanidze, Ak’ak’i 1953 kartuli gramat’ik’is sapudzvlebi, I: morpologia (The fundamentals of
Georgian grammar I: morphology) [Tbilisi: TSUG]
—— 1982 Grammatik der altgeorgischen Sprache (H. Fahnrich, tr.) [Tbilisi: TSUG]
Silverstein, Michael 1976 “Hierarchy of features and ergativity” in R. Dixon (ed.) Grammatical
categories in Australian languages [Canberra: Australian Institute for Aboriginal Studies]
pp 112-171
—— 1981 “Case marking and the nature of language” Australian Journal of Linguistics 1:
220-240
Topuria, Varlam 1967 svanuri ena,I: zmna (The Svan language,I: Verb) [Tbilisi: Mecniereba]
(originally published in 1931)
Tschenkeli, Kita 1958 Einfuhrung in die georgische Sprache [Zurich: Amirani Verlag]
—— 1960-1974 Georgisch-deutsches Worterbuch [Zurich: Amirani Verlag]
Tuite, Kevin 1985 “Syntactic subject in Georgian” [in press: to appear in Folia Slavica]
—— 1987 “Indirect transitives in Georgian” [to appear in BLS 13]
—— in progress Number agreement in Georgian (dissertation, U Chicago)