Vous êtes sur la page 1sur 8

A New Approach to Understanding the Georgian Verb Dan Humphries

Abstract. Combining an innovative root-based dictionary and a new system of classifying Georgian verbs morphosyntactically may be the key to understanding the notoriously complex Georgian verb. A rootbased dictionary better represents a verb's morphology vis--vis the more opaque traditional formats, and this new classificatory system better preserves the link between a verb's form and argument structure to account for how voice and version interact in the Georgian verb.

1. Introduction A Georgian verb may encode person and number of its subject, direct object, and indirect object; tense, mood, and aspect; voice; direction and orientation; causativity; and version (Harris 1981: 18). Affixes marking these categories are placed into slots before and after the verbal root and compete for these positions based on a multitude of factors (Aronson 1982: 169-174). Thus, the Georgian verb can be extremely complicated and tends to recapitulate the syntactic structure of its clause entirely. Indeed, a single Georgian verb is often translated into another language as a whole sentence, e.g., amaenebinebdnen 'were they to make me build it' (Melikishvili 2001: 5). The difficulty of parsing Georgian verbs is exacerbated by two facts. First, traditional alphabetized dictionaries do not present the verb in a useful format because affixes often obscure the verb's base form. Second, until recently Georgian verbs have been categorized using the semantic feature of voice. This method works well for monopersonal Indo-European verbs, but falls short in describing the polypersonal Georgian verb. Combining a non-traditional lexicon format and a new system of classifying Georgian verbs morphosyntactically (Melikishvili 2001) addresses both these problems: a verbal root-based lexicon presents a logical framework for the system to be portrayed and the classifier diathesis provides an elegant account of how voice and version interact in Georgian verbs. In addition to offering a satisfying linguistic description of the system, this combined approach allows even a beginning student or translator access to a complex problem. 2. The Lexicon Problem The first obstacle a student or translator encounters when parsing a Georgian verb is finding it in a dictionary. Unlike more familiar Indo-European verbs (e.g., French, German, Russian, etc.) that inflect for grammatical categories after the root, the Georgian verb inflects for grammatical categories both before and after the verbal root. So, searching alphabetically from left to right often does not work well in Georgian since both ends of the verb inflect for grammatical categories. This is largely due to the fact that the Georgian verb is polypersonal and requires additional slots to mark person for its non-subject arguments and how they relate to one another in a clause. Agglutination is a major feature of the Georgian verb. The root carries the basic meaning of the verb, but the root itself is not fully lexicalized -- it has meaning only when various affixes are attached to it, e.g., a finite verb form can contain a preverb, a subject/object marker, and a preradical vowel before the root and a thematic suffix, a

desinence, and a number marker after it. For example, the root cer has the core meaning 'write'. By adding the first person marker v- before it and the singular marker - after it, we derive vcer 'I write/am writing'. Add the past tense (imperfect) thematic suffix -di and plural marker -t and we derive vcerdit 'we were writing'. If you add the preverb da=, object marker g-, and causative circumfix a-X-in-eb around the root, we derive dagacerinebdit 'we would make you write'. Because some or all of these elements may be present in a form encountered, the student or translator must know what to ignore (strip away, so to speak) in order to find the verb in a dictionary. Tschenkeli (1965-1974) was the first lexicographer to list Georgian verbs by root in his comprehensive Georgian-German dictionary. This has value because both semantic and morphological relationships among words are clearly indicated. A foreign student or translator of Georgian can increase his vocabulary once he becomes aware of the interrelationships among the various entries. However, since the roots themselves are not fully lexicalized in most cases, the student or translator has difficulty predicting or generating the form a given root would take. Looking up a verb under a non-lexicalized form (root) is not the optimal solution since a root lacks meaning when stripped of its various affixes (cf. Harrell forthcoming and Gurevich 2003). Most other Georgian dictionaries (see bibliography), however, use the verbal noun as the entry form for the verb. This is the equivalent of an English dictionary listing the noun 'pronunciation' and not giving the verb 'pronounce' (Aronson 1982: 49). With regard to the example root given above, the verbal noun cera means '(act of) writing' and it would represent a range of verbs that contain writing as part of their meaning. Native speakers obviously have no problem generating or parsing a verb's other permutations using the verbal noun as a base form, but foreigners usually find this quite difficult. The publication of the Georgian-English Dictionary (Harrell, et al: 2002) was a great leap forward in improving the lexicon for learners and translators because it combined finite base forms of verbs listed alphabetically by their root. Base forms are also listed alphabetically (working left to right by affix) in their own right and crossreferenced to the appropriate root. For example, if the user came across the sentence zurabma aucera besos ninos garegnoba ('Zurab described Nino's appearance to Beso') and wanted to find the verb aucera, he could find it by root under the section (the list has been shortened considerably for the sake of illustration):
cer v. amocers to copy out (e.g., from a text) acers to describe gamocers to subscribe to (e.g., a magazine) gacers to discharge (e.g., from a hospital) daacers to write on something (e.g., an address on an envelope) dacers to write (e.g., a letter) miacers to write (then and there) as an addition; to add to something written moicereba to write and send (a letter or note) emoacers to impose upon / on, to force upon, to foist upon acers to enter (in), to enroll (in) caacers to autograph, to inscribe

If the user looked for the verb in the traditional manner (alphabetically, working left to right), then he would find: acers See cer. The obvious advantage is that such a format doubles a user's chance to find an item. However, since its focus was on base forms, the complex morphology of the verb remained a problem.

3. The Classification Problem Melikishvili (2001: 66) uses the term diathesis (1951) to replace the semantic term voice in order to describe and categorize the Georgian verb on the morphosyntactic level, thus replacing the traditional division based on four conjugations with three diatheses. Diathesis is based on a verb's form and argument structure (valency), accounting for the expression of both voice (active, passive, and medial) and version (neutral, subjective [reflexive], objective [benefactive and possessive], and locative [superessive] version). Melikishvili's revolutionary argument for this new system echoes that of Shanidze (1973) when he proposed the English term screeve to replace tense to account for verbs at the paradigmatic level. Voice traditionally refers to the relationship between a verb's action or state and its arguments (subject, object, etc.). When the subject is the verb's agent or actor, the verb is in the active voice, e.g., mcerali1 cigns cers2 'the author1 is writing a book2'. When the subject is patient, target, or undergoer of the verb's action, it is in the passive voice, e.g., cigni2 icereba mcerlis1 mier 'the book2 is being written by the author1'. Thus, voice forms are in an inverse relationship with one another and a shift in voice does not alter the semantics of the sentence. However, such a definition for voice does not explain many Georgian verbs expressing a person's emotional and physical condition, e.g., cuxs 'worry, be troubled', kankalebs 'shake, tremble', tiris 'cry, sob', etc. The agent here is the same as the patient: mascavlebeli cuxs 'the teacher is upset'; bavvi kankalebs 'the child is trembling'; tiris gogona 'the girl is crying', etc. Melikishvili (2001: 62-63) calls such verbs autoactive because their subject is self-acting and his action is redirected back onto the subject reflexively. While most of these verbs are used intransitively, e.g., bii tamaobs 'the boy is playing', mocape scavlobs 'the pupil is studying', glexi cuxs 'the peasant is worried', kali tiris 'the woman is crying', etc., some can take a direct object, e.g., bii tamaobs pexburts 'the boy is playing soccer', mocape scavlobs gakvetils 'the pupil is studying the lesson', glexi cuxs jroxas 'the peasant is worried about the cow', kali tiris vils 'the woman is crying about her child', etc. Verbs that can be used both intransitively (e.g., English 'die') and transitively (e.g., English 'kill') are called labile (e.g., English 'walk', in 'I walk every day after work' [intransitive] vs. 'I walk my dog every day after work' [transitive]). Such verbs are provisionally categorized as middle voice or medial (Holisky 1981: 2) in Georgian grammatical literature since they resemble active verbs morphologically and passive verbs syntactically. However, determining which verbs are passive by formal criteria alone is very difficult in Georgian. Furthermore, when active verbs undergo conversion , e.g., nugzarma daimala vali 'Nugzar hid the apple for himself' vs. nugzari daimala (tviton) 'Nugzar hid (himself)', the subject is not passive (the patient) in these relative forms, but is the active element (the agent), cf., nugzar emaleba nanas 'Nugzar hides from Nana', etc. In addition, "passive" monopersonal (absolute) verbs often have an active subject, which also contradicts the accepted semantic criterion to determine which verbs are passive. For this reason, the ad hoc feature deponence has been used to describe them, e.g., ilanjeba 'curse', irineba 'grumble, gripe', ilukmeba 'gorge

oneself with food', etc. Another problem is that many "passives" with the preradical vowel e- (from the semantic realm of oral expression) lack their own regular monopersonal counterpart form with the preradical vowel i- and are paired with autoactive verbs: laparakobs 'talk' vs. elaparakeba 'talk with someone' ; masxrobs 'joke' vs. emasxreba 'joke with someone'; kamatobs 'argue' vs. ekamateba 'argue with someone'; etc. Other "passive" forms having the suffix -eb-a create a voluntary vs. involuntary semantic opposition vis--vis active and autoactive forms: vixseneb 'I reminisce (on purpose)' vs. maxsendeba 'it comes back to my mind (unintentionally)'; vigvianeb 'I am late (on purpose)' vs. magviandeba 'I am late (by mistake)'; vnanob 'I regret (something) (voluntarily) vs. menaneba 'I feel pity (for someone/something) (involuntarily)'; etc. Thus, assigning voice to a particular verb to mark the active-passive relationship between a subject and predicate may work well for Indo-European verbs, but it often proves inadequate for Georgian, as the above examples demonstrate. In Georgian, voice is not sharply delineated from reflexivity and the division passive and middle voice fails to satisfy (both morphologically and syntactically) the concept of voice that is universally recognized and accepted in general linguistics. Since voice is expressed both morphologically and syntactically, determining subject and object, on the one hand, and distinguishing passive and middle voice forms, e.g., saubrobs 'converse' vs. esaubreba 'converse with someone', and active and middle voice forms, e.g., is akankalebs mas 'he makes her shake' vs. akankalebs mas 'she shakes' (involuntarily), on the other, becomes extremely complicated in the context of the Georgian polypersonal verb. Diathesis captures the interrelationship of voice and version in the Georgian verb. Melikishvili (2001: 66) replaces the traditional classificatory system of four conjugations with three diatheses. Her revolutionary system also takes into account a verb's valency, version, conversion, and inversion, as well as its intrinsic stative or dynamic character. It also addresses how syntactically equivalent actants behave within the semantic confines of the all-pervasive animate/inanimate and personal/non-personal opposition in Georgian. Her system is arranged as follows: I Diathesis: Stative autoactive and dynamic active verbs, whose morphological subject changes case depending on series, make up this group, e.g., autoactive: dedaemi cuxs 'my mom is worried', dedaemma icuxa 'my mom was worried' (for a certain period of time), dedaems ucuxia 'my mom (apparently) was worried'; active: mamaemi avijebs laas 'my dad wakes Lasha up', mamaemma gaavija laa 'my dad woke Lasha up', mamaems gauvijebia laa 'my dad (apparently) woke Lasha up'. Most Georgian verbs belong to the I diathesis. An autoactive verb in this group can expand its argument structure or add affixes to participate in other subgroups of this diathesis:
givi cuxs rusikos 'Givi is worried about Rusiko' givi rusikos tavs ar ecuxebinebs 'Givi will not let Rusiko trouble herself [with his problems]' givi acuxebs rusikos 'Givi annoys Rusiko' givi ucuxebs rusikos guls 'Givi nags Rusiko' (lit. 'he vexes her heart') givi acuxebinebs rusikos vanos 'Givi makes Rusiko bother Vano'

This group also includes labile autoactive forms such as gogo tamaobs 'the girl is playing' - gogo tamaobs ogburts 'the girl is playing tennis'; tavmdomare laparakobs 'the chairman is talking' - tavmdomare laparakobs mas 'the chairman

is saying it'; studenti scavlobs 'the student is studying' - studenti scavlobs rusul literaturas 'the student is studying Russian literature'; etc. By the classical understanding of voice, I diathesis verbs with a complete argument structure are active: the subject's action is transferred to the direct object (and may also affect an indirect object). I diathesis (autoactive) verbs lacking a complete argument structure are basically reflexive. Their subject is active, but its action is redirected towards itself. II Diathesis: Stative and dynamic passives, with complete and incomplete argument structures, and absolute intransitive and relative intransitive dynamic verbs belong in this group. Some of these are primary forms, while others are derived via conversion of I diathesis active verbs. It appears that approximately a third of Georgian verbs belong to the II diathesis. II diathesis verbs whose (usually inanimate) subject undergoes or experiences the action of the verb are passive in their morphology, argument structure, and semantics, e.g., i-type: anta imaleba bavvis mier 'the purse is being hidden by the child'; and d-type: axali saxli endeba 'a new house is being built'. While all II diathesis verbs have passive morphology and argument structure, not all of them are semantically passive:
1. As mentioned above, deponent forms such as ilanjeba 'curse', iloceba 'bless', itqobineba 'inform', etc., are objectless autoactives. 2. tbeba-type and d-type verbs whose subject is a self-reliant animate, personal entity (i.e., a human being), e.g., diddeba 'get bigger', dumdeba 'become silent', etc. are reflexive autoactive forms. 3. i-type verbs, whose subject is animate (usually [+personal]), depict an active action: kata imaleba 'the cat is hiding'; kata emaleba jals 'the cat is hiding from the dog', etc. Here, imaleba is autoactive, while emaleba is active with indirect relations (version). 4. Verbs such as elaparakeba 'converse with someone' which lack a monopersonal (absolute) counterpart with the preradical vowel i- are also semantically active. They usually correspond to monopersonal (absolute) reflexive or autoactive verbs, e.g., xubobs 'quarrel' - exubeba 'argue with someone'; tamaobs 'play' vs. etamaeba 'play with someone'; etc. 5. II diathesis passives of state such as sceria sainao davaleba 'the homework is written', sainao davaleba uceria mascavlebels 'the homework is written for the teacher' are neutral from the perspective of voice.

III Diathesis: Secondary inversive verbs (created from the I and II diathesis) such as akankalebs 'tremble' (involuntarily), emereba 'feel like singing', etc. belong to the III diathesis. Such forms have often lost an argument in the inversion of process, e.g., is marolebs me 'it is making me shiver' -> marolebs me 'I'm shivering'. Passive inversive verbs such as ezareba 'be lazy', ecineba 'feel like laughing', agviandeba 'run late', etc., are also part of this group. A significant number of high-frequency stative root verbs expressing emotion and perception also belong to the III diathesis, e.g., uqvars 'love', uirs 'be difficult, have a hard time of it', ia 'be hungry', cquria 'be thirsty', einia 'be afraid', etc. III diathesis forms are historically secondary inversive verbs with their logical subject in the dative case. They have stative meaning (they fundamentally express emotional and physical states). They are neutral regarding voice, but they can express

various nuances of meaning such as lack of volition, consideration, possibility, etc. Very few Georgian verbs (approximately 4%) belong to the III diathesis. Thus, the broad grammatical category of diathesis encompasses a verb's valency, version, conversion, and inversion, its stative or dynamic character, and the class of its actant(s). As a morphosyntactic category, diathesis recognizes the link between a verb's form and argument structure and groups verbs accordingly. The traditional method of organizing verbs by the semantic category of voice fails to classify verbs properly because the interrelationship between a verb's morphology and syntax is not always preserved. Within the hierarchy of Melikishvili's system, voice is a subcategory of diathesis since Georgian verbs of different voices are arranged within the same diathesis. Voice has been traditionally used to describe the interrelationship of the action, an active subject, and a passive object, i.e., the action either emanates or does not emanate from the subject and is either directed at or not directed at the object. The problem is that this interrelationship is expressed in the Georgian verb by various forms and argument structures -- the same form and argument structure does not always correspond to the same voice. 4. Conclusion Work is currently underway to translate Melikishvili's seminal work (Kartuli zmnis ulebis sistema [The Conjugation System of the Georgian Verb]) into English and convert its massive index into a cross-referencing, root-based verbal dictionary with each form tagged for paradigm assignment as per the classificatory system presented in the text. Each model paradigm in the book will also be expanded to the fullest degree possible. This promises to be the key to unraveling the morphosyntactic and semantic complexities of the Georgian verb because when a student or translator finds a particular verb they can find its entry form quickly, then its meaning and paradigm assignment. They can then go to the paradigm and find a model verb (sharing morphological, syntactic, and semantic qualities with the target word) and get a fuller translation for the item in context (taking all of the verb's categories into account). This will also greatly enhance the lexicon of any machine translation system for Georgian because every form of a great many verbs (over 15,000 entries) will be generated using a morphological template for each paradigm. Such an approach may also work well for other languages that have a similar verbal system, e.g., those of the Philippines. Combining an improved lexicon and a better classification of Georgian verbs presents a more logical framework for the system to be portrayed and an elegant account of how voice and version combine in Georgian verbs. On the theoretical plane it provides a comprehensive linguistic description of the system, while on a very practical level it helps even a beginning student or translator deal with the highly complicated Georgian verbal system.

References

Aristoteles. (1951) De anima. New Haven: Yale University Press. Aronson, Howard I. (1982) Georgian: A Reading Grammar. Columbus, Ohio: Slavica. Aronson, Howard I. and Dodona Kiziria. (1999) Georgian Language and Culture: A Continuing Course. Bloomington, Indiana: Slavica. Axveldiani, G. S. and V. T. Topuria. (1990) Gruzinsko-russkij slovar' (Georgian-Russian Dictionary). Tbilisi: Ganatleba Publishing House. Cherkesi, E. (1950) Georgian-English Dictionary. Oxford: Oxford University Press. Chikobava, Arnold. (1990) Explanatory Dictionary of the Georgian Language. Tbilisi: Georgian Academy of Sciences. Chubinov, D. I. (1984) Gruzinsko-russkij slovar' (Georgian-Russian Dictionary). Tbilisi: Sobchota Sakartvelo Publishing House. Chuzischwili, O. and T. Chuzischwili. (1977) Georgisch-Deutsches Wrterbuch. Ganatleba Publishing House. Gurevich, Olya. (2003) "The Status of the Morpheme in Georgian Verbal Morphology". Proceedings of the Twenty-Ninth Annual Meeting of the Berkeley Linguistics Society: 161-173. Gvarjaladze, Isidor and Thamar Gvarjaladze. (1974) English-Georgian and GeorgianEnglish Dictionary. Tbilisi: Ganatleba Publishing House. Gvarjaladze, Isidor and Thamar Gvarjaladze. (1979) Georgian-English Dictionary. Tbilisi: Ganatleba Publishing House. Harrell, S.J. (forthcoming) "Dealing with the Georgian Verb: A Lexicographer's Perspective". Dictionaries (journal of the Dictionary Society of North America). Harrell, S.J., Maia Koupounia, Maia Tsitsishvili, and Ketevan Gabounia. (2002) Georgian-English Dictionary. Springfield, Virginia: Dunwoody Press. Harris, Alice C. (1981) Georgian Syntax: A Study in Relational Grammar. Cambridge: Cambridge University Press. Jelden, Michael. (2001) Wrterbuch Deutsch-Georgisch, Georgisch-Deutsch. Hamburg: Helmut Buske.

Kankava, M. V. (1965) Kratkij gruzinsko-russkij slovar' (A Short Georgian-Russian Dictionary). Tbilisi: Ganatleba Publishing House. Meckelein, Richard. (1928) Georgisch-Deutsches Wrterbuch. Gruyter. Berlin: Walter de

Melikishvili, Damana. (2001) Kartuli zmnis ughlebis sistema (The Conjugation System of the Georgian Verb). Tbilisi: Logos Press. Shanidze, Akaki. (1973) Kartuli enis gramatikis sapudzlebi (Fundamentals of Grammar of the Georgian Language). Tbilisi: Tbilisi University Publishing House. Thrax, Dionysius. (1982) The grammar of Dionysios Thrax (translated by Thomas Davidson). St. Louis, Missouri: George Knapp. Torikashvili, John J. (1992) Georgian-English, English-Georgian Dictionary. York: Hippocrene Books. New

Tschenkeli, Kita. (1965-1974) Georgisch-Deutsches Wrterbuch (three volumes). Zurich: Amirani-Verlag. Zycar', Jurij Vladimirovi. (1982) Rekonstrukcii v oblasti baskskogo jazyka. Tbilisi: Mecniereba.

Vous aimerez peut-être aussi