Quick Links:
 • Home
 • Essays
 • Teaching
 • Sitemap

 Course Pages:
 • German 101
 • German 102
 • German 203
 • German 204
The Verbal Syntax and Ergativity of Georgian

It is relatively unusual among languages of the world to find examples of what is called mixed ergativity; this phenomenon occurs when some syntactical elements of a language can be ascribed to an ergative-absolutive typology, but others must be considered part of a nominative-accusative structure. Ergativity in and of itself is not exceptionally marked in a language, although traditional linguistics, which has been based upon the work of Indo-European-speaking scholars, for whom ergativity seemed foreign and therefore unnatural, has tended to regard ergativity as something strange or uncommon. In fact, ergative languages are quite numerous in the world today: in Europe, the only ergative languages are Basque, an isolated language unrelated to any other European tongue, and several languages of the Caucasian family, including Georgian, but outside of Europe there are many different examples, including members of the Mayan and other Amerindian families, as well as many of the Australian and Indo-Pacific languages.

The Caucasian language family is one of the most significant in terms of European ergative language groups; most of its members have some form of ergativity inherent to their syntax and structure, and offer interesting cases of study since they have been fairly well-documented and are still readily accessible to field workers. Georgian, a member of the Southern branch of the Caucasian family, is particularly intriguing, because it has a system of mixed ergativity as described above. The Georgian language is spoken today in its traditional homeland, what is now the Georgian Soviet Socialist Republic in the Soviet Union, as well as in certain dialectical variations in parts of northeastern Turkey; the dialect most often studied by linguists, and that which will be discussed here, is the one spoken in Tbilisi, the capital city of Georgia. The history of Georgian dates back to many inscriptions from the Fifth Century A.D., and is strongly associated with the nationalistic sentiments in Georgia today. Its phonology is not exceptionally marked, but it is worth noting; the Georgians use an alphabet unique to their language, which shares many of the Cyrillic characters. The basic Georgian phonology is as follows:

     Bilabial        Dental        Alveolar        Palatal        Velar        Glottal    
Stop    p                t                k            
     p'                t'                k'            
     b                d                g            
Affricate                    c        c  j        q        x  h    
                     c        c'  j'        g'        j    
Fricative            f        s        s'                    
             v        z        z'                    
Liquid                    r        l                    
Nasal    m        n                                    
Glide    w                                            

Georgian morphology is highly inflected, mainly in verbs and adjectives, although nouns are also marked for case and number. According to Harris, a single Georgian verb may encode the following information: [1]

 person of subject
 person of direct object
 person of indirect object
 number of subject
 number of direct object
 number of indirect object
 aspect (complete/incomplete; habitual/non-habitual)
 direction and orientation

Ergativity, at least as seen in Georgian, is defined as the "system of nominal case-marking where the subject of an intransitive verb has the same morphological marker as a direct object, and a different morphological marker from the subject of a transitive verb." [2] The intransitive subject and the transitive object share the case known as the absolutive, while the subject of the transitive verb appears with an ergative case marking. This phenomenon occurs quite predictably in the Georgian sentences below: [3]

 (1)Student-i mivida.
  student-ABS went
  'The student went.'

 (2)Student-ma ceril-i dacera
  student-ERG letter-ABS wrote
  'The student wrote the letter.'

The remarkable aspect of Georgian, though, occurs in its so-called split ergativity. This takes the following form: in some tenses of the verb, notably the subjunctive and the aorist tense (a form of the past tense which implies neither completed nor uncompleted actions), the constituents will be marked with the ergative and absolutive case markings; in most other tenses, on the other hand, the sentence falls into the 'normal' pattern of nominative-accusative declension. Thus, for the present tense of the ergative sentences above, we see the following accusative forms:

 (3)Student-i midis.
  student-NOM goes
  'The student goes.'

 (4)Student-i ceril-s cers.
  student-NOM letter-ACC writes
  'The student writes the letter.'

This system of split ergativity according to the tense system is very rare, and is shared only by Chol, a Mayan language of Mexico, which, however, assigns the ergative and absolutive cases somewhat differently.

Although Georgian nouns are marked for case and number, there are several different sets of zero morphemes; this often makes it necessary to refer to the encoding markers on the verb to determine the case of the nouns involved. It is thus helpful to set out the markers that may be added on to verbs in a Georgian sentence, as follows. Note that word and morpheme order in Georgian is fairly free, since there is such a large amount of inflection; common usage dictates S-V-DO-IO or S-IO-DO-V, with no specific tendency toward one or the other. (The dashed lines (--) indicate the position of the verb stem, which includes the root and various formants.) [4]

Subject Markers  
         Singular                Plural        
First Person v  --  ---            v  --  t                
Second Person -  --  ---            -  --  t                
Third Person -  --  s/a/o       -  --  en/es/nen   

Direct Object Markers  
         Singular                Plural        
First Person m  --  ---            gv --  ---              
Second Person g  --  ---            g  --  t                
Third Person -  --  ---            -  --  ---              

Indirect Object Markers  
         Singular                Plural        
First Person m  --  ---            gv --  ---              
Second Person g  --  ---            g  --  t                
Third Persons/h/-  --  ---           s/h/-  --  t                

It is difficult at a first examination to discern which tenses in Georgian take the ergative case markings and which the accusative. Perhaps the most helpful method would be to set out the same sentence in several different tenses, and then to examine the results.

 (5)Txa-i c'am-s venax-s
  goat-NOM eat-3sgS vine-DAT
  'The goat eats the vine.' -- Present

 (6)Txa-m se-c'am-a venax-i
  goat-ERG ASP-eat-3sgS vine-NOM
  'The goat ate the vine.' -- Aorist

 (7)Txa-s se-u-c'am-i-a venax-i
  goat-DAT ASP-VER-eat-FORM-3sgS vine-NOM
  'The goat has apparently eaten the vine.' -- Perfect [5]

Although no further examples are provided on which to base the conclusions, Harris has developed a set of generalizations regarding ergativity and tense relationships in Georgian. Her work utilizes the separations of different verbs into classes, a theory which seems to be generally accepted by other Georgian specialists, as well as a second separation of tenses into different classes. Essentially, she divides all Georgian verbs into four classes, based primarily on phonological and morphological distinctions. A summary of her classification criteria can be rendered thus: [6]

 Class 1:
 a.The future and aorist tenses are formed with a preverb.
 b.In the future tense, the suffix -s marks third person singular subjects, the suffix -en third person plural subjects.
 c.In the aorist tense, the third person plural subject is marked by the suffix -es.

 Class 2:
 a.The future and aorist tenses are formed with a preverb or the character vowel e-.
 b.In the future tense, the suffix -a marks third person singular subjects, the suffix -an third person plural subjects.
 c.In the aorist tense, the third person plural subject is marked by the suffix -nen.

 Class 3:
 a.The future and aorist tenses are formed with the circumfix i -- (eb).
 b.In the future tense, the suffix -s marks third person singular subjects, the suffix -en third person plural subjects.
 c.In the aorist tense, the third person plural subject is marked by the suffix -es.

 Class 4:
 a.The future and aorist tenses are formed with the character vowel e-.
 b.In the future tense, the suffix -a marks third person singular subjects.
 c.In the aorist tense, the third person plural final subject is marked as a singular.

An interesting point to note is that although these verbs are classified primarily according to morphological criteria, there is a remarkable tendency for the classes to assume semantic characteristics. The Class 1 verbs for instance, are primarily causative actions, such as the following:

 gaxsnis'he will open it up'
 modreks'he will bend it'
 gaatetrebs'he will make it white'

Class 2 verbs, meanwhile, are passified, and although they may include actions, are often more like state-of-being verbs:

 daigleba'he will tire out'
 ecereba'it will stand written'
 darceba'he will remain'

Action verbs seem to make up the majority of Class 3 verbs, although there are also impersonal and onomatopoeic examples:

 ixitxitebs'he will giggle'
 kartvelobs'he behaves like a Georgian'
 ibebiavebs'she will do midwifery'

The Class 4 verbs are the most varied, but can be generalized to include most state-of-being verbs and some abstract expressions:

 scxela'he is hot'
 uxaria'he finds it pleasing'
 esxvapereba'it appears different to him' [7]

After setting out the different classes of verbs, Harris then proceeds to classify Georgian verb tenses into three different series, which are divided up not on semantic or morphological boundaries, but according to their probability to take the ergative or the accusative case markings. The tenses, or 'screeves,' the term accepted by many scholars (from the Georgian mckrivi, 'row,level'), are divided as follows: [8]

 Series I
    present subjunctive
    future subjunctive

 Series II
    optative (second subjunctive)

 Series III
    perfect (first evidential)
    pluperfect (second evidential)
    third subjunctive (third evidential)

Once she has established these classifications, Harris proceeds to examine a few cases, and then states her conclusions about the interactions of these different groups in the form of a table in which she comes up with a pattern for the resulting case usage, the two patterns being represented here by the letter A or B:

     Series I        Series II        Series III    
Classes 1 and 3    B        A        B    
Classes 2 and 4    B        B        B    

The patterns she then simplifies in terms of which constituents take which cases:

           Final Subject               Final Direct Object         Final Indirect Object    
Pattern A    ERG        NOM        DAT    
Pattern B    NOM        NOM        DAT    

With this framework established, Harris can now explain the irregularities and other problems that occur in Georgian syntax as regards case markings. For example, the sentences below fit perfectly into her framework, since the verb 'to sow' in Georgian is a Class 1 verb. [9]

 (8)glexi tesavs siminds
  peasant-NOM he-sows-it corn-DAT
  'The peasant is sowing corn.'

 (9)glexma datesa simindi
  peasant-ERG he-sowed-it corn-NOM
  'The peasant sowed corn.'

 (10)glexs dautesavs simindi
  peasant-DAT he-sowed-it corn-NOM
  'The peasant has sown corn.'

There are, of course, some minor and irregular exceptions to the rules Harris has set out, as she herself discusses. The verb uqveba 'he tells it to him' may, for example, freely take on characteristics of either Class 1 or Class 2: [10]

Class 1:
 (11)mamam ojaxs motxroba mouqva
  father-ERG family-DAT story-NOM he-told-him-it
  'Father told the story to the family.'

 (12)turme bics saxeli seukitxia gogostvis
  apparently boy-DAT name-NOM he-asked-it girl-for
  'Apparently the boy has asked the girl her name.'

Class 2:
 (13)mama ojaxs motxrobas mouqva
  father-NOM family-DAT story-DAT he-told-them-it
  'Father told the story to the family.'

 (14)turme bici saxels sekitxia gogos
  apparently boy-NOM name-DAT he-asked-her-it girl-DAT
  'Apparently the boy has asked the girl her name.'

The verb arsebobs 'it exists' also does not behave according to the rules set out above. It is morphologically a Class 3 verb, although it can take on characteristics of Class 2 or 3 in basically free variation. More surprising, however, are the restrictions on its case marking system that appear seemingly at random, as in the following example. (15), which uses the Class 3 case markings, is unacceptable to native speakers of Georgian, although (16), which also uses Class 3 markings but adds a locative adverbial phrase, is perfectly fine. Instead of (15), then, (17), using the Class 2 system, must be used. [11]

 (15)*devebs ar uarsebniat
  devils-DAT not they-exist
  'Devils do not exist.'

 (16)devebs ar uarsebniat betaniasi
  devils-DAT not they-exist Betania-in
  'There are no devils in Betania.'

 (17)devebi ar arsebulan
  devils-NOM not they-exist
  'Devils do not exist.'

Despite these relatively few and minor exceptions, Harris' theory works for the great majority of Georgian verbs. There are still many unresolved problems, however, and it must be pointed out that many linguists and Georgian specialists do not accept the validity of her conclusions. On a related note, it is interesting to examine several other aspects of Georgian ergativity and verbal syntax. One of the most interesting areas is the phenomenon known as the 'versions' of a Georgian verb. This allows the creation of an indirect object for a verb that would not normally have one, and is useful in comparing case markings between the two versions. For each (a) sentence below, the verb is in its normal form, with no specified indirect object. For the (b) sentences, however, which are the versions, a formant is added on or in to the verbal phrase to allow an indirect object to be placed in the sentence, marked in the dative case.

 (18)Benefactive Version:
   (a)gelam sekera axali sarvali senivis
  Gela-ERG he-sewed-it new trousers-NOM you-for
  'Gela made new trousers for you.'

   (b)gelam segikera axali sarvali sen
  Gela-ERG he-sewed-you-it new trousers you-DAT
  'Gela made new trousers for you.'

 (19)Possessive Version:
   (a)mzia cmends
  Mzia-NOM she-cleans-it sister-GEN shoes-DAT
  'Mzia is cleaning her sister's shoes.'

   (b)mzia ucmends das pexsacmlebs
  Mzia-NOM she-cleans-her-it sister-DAT shoes-DAT
  'Mzia is cleaning her sister's shoes.'

 (20)Superessive Version:
   (a)gelam surati daxata kedelze
  Gela-ERG picture-NOM he-painted-it wall-on
  'Gela painted a picture on the wall.'

   (b)gelam surati daaxata kedels
  Gela-ERG picture-NOM he-painted-it-it wall-DAT
  'Gela painted a picture on the wall.' [12]

As a final point of interest in regards to Georgian ergativity, note that, although Georgian is an ergative language, it has a common and grammatically-functional passive construction. This is rather unusual in terms of language universals: if a language is ergative, most often it either does not have a passive construction, or the ergative may in some way double as a type of passive. In fact, of the 19 ergative languages examined by Trask, only Basque, Georgian, Hindi, and a few members of the Mayan languages had an independent passive construction as well as ergativity. [13] Georgian, quite significantly, has both, and this has made it particularly interesting for scholars. The passive construction is not terribly complicated, and a few examples should suffice to give a good general picture of its characteristics, which are, in fact, not unlike English:

 (21)Passive with Object:
   (a)jagli ukbens bavsvs
  dog-NOM he-bites-it child-DAT
  'The dog is biting the child.'

   (b)bavsvi dakbenilia jaglis mier
  child-NOM bitten-it-is dog by
  'The child is bitten by a dog.'

 (22)Passive with Location:
   (a)mascavlebels gaugzavnes cemi tavi
  teacher-DAT they-sent-him-him my self-NOM
  'They sent me to the teacher.'

   (b)me gagzavnili var mascavleblistvis
  I-NOM sent I-am teacher-for
  'I am sent to the teacher.' [14]

Thus, the ergativity of the Georgian language, although only partial, can be well-defined and follows reasonable rules of selection. There are many more topics of interest in Georgian verbal syntax, many of which are closely related to the ergativity vs. nominative distinctions; unfortunately, access to the few works that have been written in these areas is difficult to achieve, although the work of Harris and Comrie seems especially approachable. It is quite remarkable, I find, that Georgian verbs in general are so highly inflected and marked, while the nouns and other sentence constituents are hardly inflected at all. The current debates in Georgian scholarship, instead of focusing on more explanations of the actual ergative usage, are concerned primarily with the evolution of this ergativity and the direction it is taking: it seems that the ergative is a fairly recent development, which is, however, already dying out, to be replaced by the purely nominative-accusative case distinctions. This subject is truly a fascinating issue, and will hopefully add to the limited body of literature available at this time.


(1)  Alice C. Harris, Georgian Syntax: A Study in Relational Grammar. Cambridge: Cambridge University Press, 1981, p. xviii. [return to text]
(2)  Bernard Comrie, "Ergativity" in Winfred P. Lehmann, ed., Syntactic Typology: Studies in the Phenomenology of Language. Austin: University of Texas Press, 1978, p. 329. [return to text]
(3)  Comrie, p. 351-352. [return to text]
(4)  Harris, p. 29. [return to text]
(5)  W. Boeder, "Ergative Syntax and Morphology in Language Change: The South Caucasian Languages" in Frans Plank, ed., Ergativity: Towards a Theory of Grammatical Relations. London: Academic Press, 1979, p. 437. [return to text]
(6)  Harris, p. 261. [return to text]
(7)  Harris, p. 262. [return to text]
(8)  Harris, p. 47. [return to text]
(9)  Harris, p. 1. [return to text]
(10)  Harris, p. 270. [return to text]
(11)  Harris, p. 274. [return to text]
(12)  Harris, p. 87. [return to text]
(13)  R. L. Trask, "On the Origins of Ergativity" in Frans Plank, ed., Ergativity: Towards a Theory of Grammatical Relations. London: Academic Press, 1979, p. 404. [return to text]
(14)  Harris, pp. 103, 109. [return to text]

Written and © Nancy Thuleen in 1991 for Linguistics 12 at Pomona College.

If needed, cite using something like the following:
Thuleen, Nancy. "The Verbal Syntax and Ergativity of Georgian." Website Article. 8 May 1991. <http://www.nthuleen.com/papers/L12paper.html>.