According to its structure, the Georgian verb is usually regarded as a complex element. Its complexness is stipulated by polypersonalism. Moreover, there are several grammatical categories in a form of a verb (person, number, mood, tense, aspect, voice and others). Some categories are expressed with appropriate morphemes. There is a particular connection between a verb and actants: a verb governs a noun in the case, while a noun governs a verb in the person. Therefore, morphosyntactic constructions are created. They can be regarded as the basis of the Georgian language. The forms of actants which agree with a verb are determined by the particular rules. We are interested in the following questions: What is the determiner of these rules? What regulates these questions?

It's obvious, that the system of the whole language reveals two constructions: inflective and non-inflective (we imply the change of the case of S and direct O according to the series as well as the invariability of cases of S and indirect O according to the series).

In the Georgian linguistics this question is discussed in one direction - mainly, the voice of the verb is concerned.    

Arn. Chikobava makes distinction between dynamic and static verbs. In the first and second schemes of the conjugation of the verb he presents the verbs of active and passive voices (conversion is implied), while static verbs are presented in the third scheme [Chikobava, 1950].

According to the author's point of view, passive is formed from active by means of particular prefixes and suffixes. What forms active? The question is quite complicated. General principles of the conjugation of the verb in the Iberian-Caucasian languages enable us to believe, that an active voice is secondary. It is derived [Chikobava, 1950].

Akaki Shanidze distinguishes four types of conjugation of the Georgian verb. The classification is presented according to the voice.

I type - the conjugation of the verbs of active voice;

II type - the conjugation of the verbs of dynamic passive;

III type - the conjugation of medio-active verbs;

IV type - the conjugation of medio-passive verbs and the verbs of static passive voice [A. Shanidze, 1980: 489].

Damana Melikishvili's work "The system of conjugation of the Georgian verb" presents the classification of the conjugation of the verb according to the diathesis. The author singles out a broader grammatical category on the basis of the relationship of persons and peculiarities of the expression of construction of the Georgian verb. This category comprises a destination system and the comprehension of reflexive and voice (together). It is calls diathesis - a determiner of the construction of the verb. The author singles out three types of diathesis:

I. Diathesis unites the verbs of complete and incomplete construction, which have the structure (model) identical to the present. These verbs are known as actives and the so-called "avtotivebi". Their MS is inflective according to the series.

II. Diathesis unites absolute transitive and relative intransitive (indirect transitive verb) dynamic verbs of the full construction. Their MSnom. and MOind.dat. are non-inflective according to the series.    

III. Diathesis unites secondary verbs having inversive (Dative) construction and different (mixed) structure. This diathesis is created by means of inversion of persons of a verb. It is often formed after losing a person [Melikishvili, 2001:68].

Each diathesis is a complex element, which presents several grammatical categories and shows their interrelation.

Besarion Jorbenadze's monograph ("The questions of the formation and function of the forms of the verb in the Georgian language") gives interesting information about the voice of the Georgian verb. The category of voice is closely related to other categories of a verb and the study of the nature of some of them becomes possible only by taking into consideration the peculiarities of the origin and formation of the voice. It significantly influences the system of conjugation. Therefore, in the Georgian language the types of conjugation are distributed according to the voice (the author means A. Shanidze's types of conjugation, - M. N.)[Jorbenadze, 1975: 3].

The given paper presents the regulations determining the essence of the voice of a verb. Finally, B. Jorbenadze concludes, that voice is a syntactic or a syntactic-morphological category. Its content necessarily implies a particular construction of a sentence - a particular relationship of the subject with the action expressed by the verb. A morphological expression of this syntactic phenomenon is presented only in some languages [Jorbenadze, 1975: 10].

The given definition enables us to distinguish only active and passive verbs, because the dissociation of active and middle voices will be difficult by means of the activity of the Georgian subject. It's known, that A. Shanidze differentiated active and middle voices according to the existence of the direct object ("the action is directed to another person or is not directed") [Shanidze, 1980: 280]. B. Jorbenadze considers, that the indication to the direct object is not expedient in the process of the determination of the voice, because the border between the voice and transitivity "disappears". We share this idea, because we have to deal with two different categories. It's significant, that there are languages, which have a category of transitivity, but did not have a category of voice. Such cases are met in the languages of the mountainous Caucasus.  

A. Shanidze differentiates two types of the middle voice: medio-active and medio-passive (which have the forms of lines of the present circle). Their names are determined by the forms of verbs (their voice) used for the formation of the following lines. According to R. Enukashvili's point of view, it's impossible to place medio-actives and medio-passives on the same flatness. Medio-active verbs, which present the majority of the middle voice express the action of S like the verbs of an active voice [Енукашвили, 1974: 6]. There is a particular contradiction in the definition of the passive voice. The verbs in passive voice are distinguished for their function of passiveness, which is not presented in the Georgian language, for example: the so-called deponents - the verbs, which are considered as passives [Shanidze, 1980: 96], but express an active action: swears, carries, speaks and others. Two forms may have active subjects from the semantic point of view. Hence, besides this similarity, they may represent two different voices. For example: vmalav (I am hiding something or somebody) - S is active and acts on the object (a verb of the active voice), but vimalebi (I am hiding) - S is active again, but it has not got an object (a verb of the passive voice).

A. Shanidze considers conversion as one of the significant factors for the definition of the Georgian verb. The forms of the voice have a conversional relationship. Active and passive voices create a good opposition in the Georgian language. Active represents a "turned version" of passive and vice versa [Shanidze, 1980: 281]. A conversional pair must differ according to the number of persons, which is caused by loosing of an active subject. Active and passive voices are considered as two different forms. At the same time, a conversion exists between them. It indicates that these two forms oppose each other according to the category of voice. A conversion is a semantic phenomenon: after the disappearance of an active subject, a direct object occupies its place. Accordingly, a construction and a form of the verb are changed. In this case two separate processes take place. A transitive verb becomes intransitive, because an active subject disappears and a direct object occupies its place. At the same time the rules of conversion are obeyed and the number of persons is reduced.

For example:

Vano builds a house (builds he (S) it (Odir.) - an active voice);

The house is built (built it (S) - a passive voice).

We think, that in this case transitiveness is essential, because in transitive and intransitive verbs the increase and reduction of persons are connected not only with the conversion, but with different processes as well.

For example: Mother is sewing a dress (she, it - two person verb, an active voice);

Mother is sewing a dress for her child (she, for her, it - three person verb,  an active voice).

The voices of both verbs are active.

For example: Vano is growing up (he - one person verb, a passive voice);

Vano is growing up "for his mother" (he, for her - two person verb, a passive voice).

The voices of both verbs are active.

It's worth mentioning, that the conversion does not exist between active and middle verbs or between passive and middle verbs. What is the relationship between gordeba (is rolling) and goravs (is rolling) concerning the conversion? Obviously, there is no relationship. Therefore, in this case, the separation of the middle voice is doubtful. How can we define the voice of the following verbs: tsukhs (is exited), dughs (is boiling), tiris (is crying), kivis (is shouting), tsekvavs (is dancing) and others?  It's known, that in contrast to A. Shanidze, Arn. Chikobava called this type of verbs "the verbs without the voice". How can be explained the opposition between the following verbs: atsukhebs (troubles) - tsukhs (is troubled), atirebs (makes somebody cry) - tiris (is crying), amgherebs (makes somebody sing) - mgheris (is singing) and others? B. Jorbenadze thinks, that: the verbs of atsukhebs (troubles) - tsukhs (is troubled) type oppose each other according to the reflexivity.  Atsukhebs is nonreflexive, while tsukhs is reflexive. In this case the opposition is only functional. The mark of reflexivity i- will be seen only during the conjugation [Jorbenadze, 1975: 79]. For example: tsukhs (is troubled), itsukhebs, atsukhebs (troubles), amgherebs (makes somebody sing) - imgherebs (will sing), atirebs (makes somebody cry) - itirebs (will cry).

Moreover, the verbs of atsukhebs - tsukhs type oppose each other according to dynamics and statics. Arn. Chikobava believed, that: the static verbs should represent that period of the development of a language, when a verb denoted state and this state was unchangeable in time like a characteristic feature [Chikobava, 1950: 51].  The acquisition of dynamics is semantically accompanied by the indication of time, for instance, tiris (static) - atirebs (dynamic, present tense), tsukhs (static) - atsukhebs (dynamic, present tense). Therefore, B. Jorbenadze thinks, that if we take into consideration the origin and the initial nature of the so-called middle verbs, we can attribute them to the reflexive forms, which are derived from particular transitive verbs. The forms atsukhebs - tsukhs contradict each other according to the reflexivity. Tsukhs is not an intransitive active. It is a reflexive active, where subject is an object of its action. A reflexive form can be static according to its meaning. It can also be transitive. At the same time, it is an opposing unit of the appropriate nonreflexive active dynamic form [Jorbenadze, 1975: 79]. Therefore, it can be noted, that the forms of atirebs - tiris, amgherebs - mgheris type oppose each other according to the voice (the voice of all of them is active). The opposition according to the reflexivity is excluded, for instance, atirebs, amgherebs - are reflexive, while tiris, mgheris - are also reflexive (according to B. Jorbenadze's point of view). We believe, that in these cases we have to deal with an initial causative. 

Therefore, separate lingual phenomena - causative and reflexivity - are independent grammatical categories. This does not imply the category of voice, transitiveness or dynamics and statics. The given categories are different lingual concepts and the necessity of their dissociation emerges. Like other grammatical categories, they are simultaneously presented in the form of a verb, for instance, the form ushenebdes (would build) presents the categories of time, mood, voice, version and others.

B. Jorbenadze believes, that: in the Georgian language the category of voice cannot always be defined as a morphological-syntactic category, which shows the relationship between a subject and a predicate (from the point of view of activeness and passiveness of a subject) by means of a particular form of a verb [Jorbenadze, 1975: 83].

Damana Melikishvili shares the given idea. She believes, that in the Georgian language the category of voice is not vividly separated from the reflexivity. D. Melikishvili also mentions, that A. Shanidze successfully differentiated the categories of voice and version in the system of the contemporary Georgian verb. Hence, the separation of the passive and middle voices (neither from formal, nor from semantic point of view) do not satisfy that meaning of the category of voice, which is presented in the linguistics and therefore, is universal [Melikishvili, 2001: 65]. The author considers, that polypersonalism is the reason of the above mentioned fact.

We share B. Jorbenadze's and D. Melikishvili's ideas about this question and think, that voice in Georgian is a semantic category, which considers relationship between a subject and an action expressed by a verb. This category undergoes the process of formation, which is revealed by the change of forms of static verbs [Сухишвили, 1971: 23].

B. Jorbenadze gives the examples - dzevs (is), khevs (tears) - which prove, that the voice is a semantic category. The forms of the given verbs are identical. The main difference lies in the content, which stipulates the intransitivity of the first and the transitivity of the other. Their forms can be distinguished only in the paradigms of conjugation. There are two kinds of subjects: active and passive. This fact is mainly determined by the context, which causes the appropriate grouping of words. Appropriate grammatical forms appear in this group. They formally differentiate the sentence containing a transitive action from intransitive, which gives the difference between transitive and intransitive verbs.

In the particular language, the groups of transitive and intransitive verbs are created on a special stage, which is mainly stipulated by the semantics. Hence, the second type of verbs is produced from the first type. For example: from the verbs of intransitive semantics (for example: tbeba - warms itself (passive voice)) a transitive verbs (atbobs - warms (active voice)) are produced. Similarly, itsereba - is written (passive voice) is derived from tsers - writes (active voice).

The basis of these derived forms must be searched within the construction.

The difference between transitive and intransitive variants of these verbs leads us to the separation of the forms of the voice of a verb [Менщанинов, 1967:95].

Therefore, it can be concluded, that generally, the formation of the category of transitiveness is preceded by the formation of the category of voice (this is mentioned by a lot of linguists) similarly to the Caucasian languages, namely, the Khundzian language.

A contradiction of the forms of voices became possible after a subject and an object had separated each other and had formed independent categories. This separation is connected with the question of the construction of a sentence, because the formation of a subject and a verb is based on this construction. Moreover, it is connected with the formation of the forms of the case [Jorbenadze, 1975: 19].

It's supposed, that in the Georgian language the voice as the category of a verb was formed comparatively late and the process of its formation is not completed yet [Jordenadze, 1975: 31]. The forms of the voice are developed on the basis of contradictory forms of transitive and intransitive verbs. The language uses special affixes for the production of the forms of the passive voice (there is also a different idea about voiced prefixes, which form the passive voice, but it is not the object of our research) and some forms of the active voice. The forms of transitive and intransitive verbs also differ from each other. It's obvious, that there are two constructions: changeable and unchangeable and these changes are represented in the forms of a verb. Namely, the transitive verbs have endings -ebi/-eba in the future (sometimes, in the present) tense (which is regarded as an ending of the dynamic passive in the Georgian linguistic literature. Several verbs have -evi/eva or -obi/oba endings). No matter whether this form is own or borrowed for the verb. The verbs with -ebi/-eba endings necessarily have an invariable construction. They do not create a variable construction even in case of three person verb. There are several examples of two and three person verbs with the same stem. Hence, the two person verb has a changeable construction, while the three person verb has an unchangeable construction.

For example:

Nick is playing (with) a ball;     Nick is playing (with) a ball with his friend;

Nick played (with) a ball.         Nick played (with) a ball with his friend.

Such examples are not few in number.

According to our point of view, the form of the verb with -ebi/-eba morphemes requires a particular construction, which is connected with the transitivity.

The verbs with a changeable construction have -ebi/-eba ending in the future tense. They are transitive. Among them are all the verbs of active voice and the so-called passive voice. We call them potentially transitive verbs. They mainly produce reflexive forms, for instance: sadilobs - is dining, isadila - dined (dinner), tsekvavs - is dancing, itsekva - danced (dance) and others. 

An unchangeable construction consists of intransitive verbs - according to A. Shanidze's classification all types of passive: dynamic, static and medio-passive.

Dynamic passives produce their own forms, while others are replaced by the forms with  -ebi/-eba ending.

For example: dgeba /is standing up (dynamic passive) - adgeba/will stand up;

apenia/is spread (static passive) - (da)epineba/will be spread;

dgas/is standing (medio-passive) - idgeba/will be standing.

In the Georgian language every person, which is connected with a verb, is represented in its semantics. Each person has its grammatical function, which is presented by the appropriate form. In the European languages only one person is reflected in the semantics of a verb. Therefore, it is always presented with one form (a pronoun, which is combined with a verb). The forms of actants, which are combined with a verb are connected with the polypersonalism of the Georgian verb. They are determined by the transitivity. If the verb is intransitive, the cases of actants are unchangeable, while cases of actants of transitive verbs are changeable according to the series. It's significant, because the difference between transitive and intransitive verbs must be revealed. In the Georgian language this difference is shown by changing S and Odir.  according to the cases. As a result, a changeable construction is formed.   

Therefore, the construction of the Georgian verb is not determined by the activeness and passiveness of S. It is determined by the transitivity. The existence of an object "receiving an action" implies the existence of an active subject. If we state, that a verb in the active voice has an active S, which has an object "receiving an action" and a middle voice is active, when it has not got an object "receiving an action", than the voice and the transitivity will become identical categories. Hence, it is not so.  

We think, that in the Georgian language the voice is a semantic category, which is in the process of development. The constructions of the verb are determined by the transitivity. In the future tense intransitive verbs are characterized by the forms produced by -ebi/-eba morpheme. The formation of the forms of the future tense determined systems of the conjugation of a verb, which were marked off. This phenomenon was expressed in the formation of forms. We connect this fact with -ebi/-eba ending of intransitive verbs in the future tense. Morphemes denoting the third person subjunctive as well as other morphemes (we will not discuss them here) subordinated this separation.

The result is essential for us: we discuss this question as a unity, as a whole construction of forms and categories, which is the basis of the language.

The system of conjugation of the Georgian verb shows two main constructions: changeable and unchangeable. According to other grammatical categories they can be divided into two sub-groups:

Changeable

1. Transitive verbs (khatavs - is painting, grekhs - is spinning, ashenebs - is building, tsers - is writing and others);
2. Potentially transitive verbs (tsekvavs - is dancing, tskhovrobs - is living, tiris - is crying, chivis - is complaining and others).                                                      

Unchangeable

1. Intransitive "direct" verbs (izrdeba - is growing up, tbeba, eperaba - is caressing, emaleba - is hiding and others);
2. Intransitive "inversion" verbs (makvs (I have), mkavs (I have), mikvars (I love), mshia (I am hungry) and others).