Printer Friendly

The expression of evidentiality between lexicon and grammar. A case study from Georgian.

1. Evidentiality-Theoretical Framework

Evidentiality is a linguistic category with reference to the information source, on the basis of which the speaker makes a statement; this may be a direct or an indirect knowledge about an event acquired by inference, hearsay, assumption or perception. Evidentiality is morphologically realized only in a quarter of the languages of the world (Aikhenvald 2004:1), the remaining languages employ several other evidential strategies to convey the information source. Two large areas where evidentiality is encoded in the tense system mainly are the Balkans and the Caucasus (De Haan 2005b: 319). In most Germanic languages as well as in Finnish evidentials are expressed through modals, which acquire secondary meanings (De Haan 2005a: 315). The main source of evidentials is the grammaticalization of lexical categories, for example the grammaticalization of verbs of speech and perception verbs in quotative markers (Aikhenvald 2004). In some languages evidential meanings developed as secondary meanings out of tenses and moods of the verbal paradigm. These extensions are labelled by Aikhenvald "evidentiality strategies". She separates them strictly from the pure evidentials, whose primary meaning is information source (Aikhenvald 2004:105).

The terminology denoting evidentiality is not homogeneous in the linguistic literature; various terms have been adopted in different grammatical traditions. Aronson (1977) employed for Balkan Slavic the term 'status', indicating a verbal category characterized by the speaker's evaluation of the narrated event (Aronson 1977 apud Friedman 1979:339). Shanidze (1973) introduced for Georgian the semantic category of 'act' (Georgian akti) to make a distinction between unmarked (non evidential) and marked (evidential) narrative forms (cf. [section] 2.1.) Evidentials are also labelled as 'mediatives' (Guentcheva 1996), 'confirmatives' (Aronson 1977), 'non-confirmatives' (Friedman 1979:339), 'validationals' or 'verificationals' (Aikhenvald 2004:15) etc.

Regarding the semantic distinction, evidentials are generally divided into direct and indirect evidentials (Givon 1982). Direct evidentials are used when the speaker has some sensory evidence for the action or event he is describing, whilst indirect evidentials are employed when the speaker reports an event which he did not witness but comes to know about it from an actual state. Indirect evidentials are split in two subcategories of inferential and quotative (also known as reportatives, reportive, reported, hearsay or second-hand) evidentials (cf. De Haan 2005a:314,

Aikhenvald 2004:177). Aikhenvald (2004:23) proposes to distinguish evidential systems with two, three, four, five and more choices. Those with only two choices (firsthand vs. non-firsthand) she labels as "small" systems. They are widespread particularly in Eurasia (the Balkans, the Caucasus and the Finno-Ugric area). The distinction between firsthand vs. non- firsthand information is compatible with the subdivision of the Evidentiality domain into direct and indirect evidentiality, introduced by Willet (1988:57):


Evidentiality may acquire mirative (also known as admirative) extensions. Mirativity is frequently related to the emotional factor. It conveys the 'speaker's unprepared mind' (De Lancey 1997), it expresses the speaker's surprise at the moment of discovering unexpected information, which can be positive or negative. Indirect evidentials tend to acquire mirative overtones. However, mirative meanings do not always have to be expressed through evidentials (Aikhenvald 2004).

Aikhenvald makes a clear distinction between epistemic modality, which refers to the subjective evaluation of the speaker with regard to the reliability of the utterance expressed, and evidentiality which marks objectively the information source of the statement. The two notions may co-occur in the same forms or cluster in different ways. Thus, it is useful to keep the domains separate, both on conceptual and functional level.

2. Evidentiality in Georgian

2.1 Previous studies

In the Georgian grammatical tradition the concept of evidentiality was mentioned for the first time by Platon Ioseliani (1863), who observed that the expression of non- witnessed events, in his words 'events which the speaker has not seen with his own eyes' was possible by means of the particle /turme/ 'apparently, evidently'. Shanidze (1930: 124) pointed out that the meaning of /turme/ is encoded within the tense system of the Georgian verb (1930:123). To this regard Shanidze introduced later in 1953 the semantic category of "act": seen vs. unseen i.e. attended vs. non attended paying attention to the capacity of the Georgian verb to express witnessed and non-witnessed events (cf. Shanidze 1953:216, Shanidze 1980: 211): "Act or deed is a verbal form which shows us whether the speaker has seen the action expressed by the verb or not. The Georgian verb can denote not only actions witnessed and seen by the speaker, but also those which have not been witnessed and seen by the speaker. According to that, we have two acts, seen and unseen (i.e. attended and unattended)" (author's translation). Shanidze labelled the perfect and pluperfect tenses, which in contrast to the aorist (unmarked past form) developed evidential extensions as a secondary meaning, with "I turmeobitf (I evidential i.e. perfect) and "II turmeobitf" (II evidential i.e. pluperfect). Furthermore Shanidze detected two possible meanings of the evidential perfect: reportive and inferential: "an unattended action can be recognized by the speaker in two ways: by hearing about it from someone else or by viewing the result of such an action and judging what could have happened" (Shanidze 1980: 211-212).

The morpho-syntactic features of the Georgian perfect have been widely investigated in a large number of linguistic studies, both in diachronic and synchronic perspective (see Chikobava 1950, Arabuli 1980, Harris 1981 and 1985, Natadze 1955, Pkhakadze 1984, Peikrishvili 1974, Shabashvili 2001 among others). In each of them evidentiality is considered one of the meaning of the perfect tense. A short history of evidentiality was drawn up by Kavtaradze (1956), who examined the occurrences of the particle /turme/ in Modern Georgian. He argued that the usage of /turme/ in expressing non-witnessed events is older than the occurrence of the perfect with evidential meaning. In his paper about the typology of 'status' in Georgian and other non-Slavic languages Friedman (1979) considered, along with the evidential perfect, the quotative particles as relevant lexical elements for the category of status (Friedman 1979:343). Furthermore he paid attention to the admirative value of the Georgian perfect, comparing it with admiratives in Balkan Slavic and Turkish. An important contribution has been offered by W. Boeder (2000) who drew up a detailed description of the evidential perfect, as well as evidential-like categories in Georgian.

2.2 Types of evidentials (2)

The evidentiality system in Georgian is subdivided into direct and indirect evidentials. According to Aikhenvald's terminology (Aikhenvald 2004: 24) this is a two term system (3), which makes the distinction between firsthand and non-firsthand information.

Direct evidentiality is generally expressed through unmarked forms (cf. example 1). Non firsthand evidentiality is mainly encoded in the perfect (ex. 2) and sometimes in the pluperfect, since both tenses have evidential extension as a secondary meaning. For the non past tenses or in contexts where the use of the perfect is not possible, lexical means (quotative and evidential particles, adverbial expressions etc.) are employed for expressing the information source.

The inference can be conveyed by the future and the conditional as well. Both lexical and grammatical expressions of evidentiality have to be considered as evidential strategies rather than evidentials proper, since there is no overt morpheme conveying "information source" as its main meaning. Thus, Georgian has no specific grammatical marking for evidentiality, in the sense of Aikhenvald (2004), but a large number of grammatical and lexical strategies for its expression.

2.3 Grammatical strategies

2.4 The perfect

The perfect tense in Georgian developed from the reanalysis of resultative constructions acquiring evidentiality as a secondary meaning. It expresses the unseen corresponding to the Series II (aorist) seen (Harris 1985):
seen:    unseen:

Aorist   Perfect
II       series

(1) man sk'am-i aivan-ze dadga.

s/he:ERG chair-NOM balcony-on put:AOR.3SG

'S/he put the chair on the balcony'.

In contrast to (1), which expresses a witnessed event, the perfect form daudgams in (2a) conveys an unwitnessed event which can have an evidential (inferential, reportive or admirative) or in other contexts (2b) even a resultative or iterative meaning. Thus, the evidential value is one of the meaning of the perfect tense. For the non evidential uses of the perfect see Boeder (2000).

(2a) mas sk'am-i aivan-ze daudgams.

s/he:DAT chair-NOM balcony-on stand:PRF.3SG

'As it seems, s/he put the chair on the balcony'. (Inferential or reportive)

(2b) mas sk'am-i aivan-ze (xsirad) daudgams.

s/he:DAT chair-NOM balcony-on (often) stand:PRF.3SG

'S/he has often put the chair on the balcony' (Resultative, iterative).

In example (2a) only the admirative value can be interpreted unambiguously, indicated by raising intonation in speech and by the exclamation mark in orthography. Inferential and reportive values as well as non-evidential meanings of the perfect can not be distinguished without the context. In (2b) the iterative meaning of the perfect is obvious, since it is reinforced by the temporal adverb /xsirad/. If the information source must be marked in the same sentence, where the perfect has clearly non-evidential meaning, it must be done by lexical strategies of evidentiality, as in (2c):

(2c) mas sk'am-i aivan-ze (xsirad) daudgams turme.

s/he:DAT chair-NOM balcony-on (often) stand:PRF.3SG EV

'Apparently /(as I was told), s/he has often put the chair on the balcony' (Iterative (perfect) + evidential (particle /turme/ 'apparently').

The perfect can be found in contexts where it has a clearly resultative meaning, in particular with 1st person subjects. The perfect is used as stylistic device in the literature and gives more expressivity to the language than the aorist (Peikrishvili 1974:58):

(2d) meckareba, saxl-i bic'-is amara damit'ovebia.

hurry up:PRS:1SG house-NOM boy-GEN with leave:PRF.1SG

'I'm in a hurry, I left the servant alone in the house' (lit. 'I abandoned the house with the servant').

The necessity of avoiding ambiguity caused by the polysemy of the perfect brought about a creation of periphrastic constructions as an alternative to the resultative meaning, which contributed to the consolidation of evidential semantics of the perfect tense. The late 18th--early 19th centuries are considered to be the period of formation of such constructions (Sakhokia 2002).

In example (3) the interpretation of the perfect can be resultative or evidential, but the resultative meaning is minor, it is almost obscured by the evidential one. In (4) the perfect of (3) is replaced by the periphrastic construction (past participle of the main verb + 3rd person present of the auxiliary), such a construction has only resultative meaning, the sentence is unambiguous.

(3) (mas) c'ign-i c'auk'itxavs.

He:DAT book-NOM read:PRF:3SG

'He apparently read the book'./He read the book'. (Evidential or resultative).

(4) (mas) c'ign-i c'ak'itxuli akvs.

he:DAT book-NOM read:P.P have:PRS:3.SG

'He read the book' (Resultative).

The periphrastic constructions can express evidentiality when the auxiliary verb is used in the perfect screeve instead of present as in (5). In this case the only interpretation of the utterance is evidential.

(5) (mas) c'ign-i c'ak'itxuli hkonia.

he:DAT book-NOM read:P.P have: PRF:3.SG

'He apparently read the book'.

The information source can also be marked by adding a lexical marker of evidentiality to the periphrastic construction as in (6).

(6) (mas) c'ign-i turme c'ak'itxuli akvs.

he:DAT book-NOM EV read:P.P have:PRS:3SG

'He apparently read the book'.

For a detailed discussion about the uses of the evidential perfect see Boeder 2000, Giacalone and Topadze 2007, Topadze 2008.

Evidentiality in some contexts can also be expressed by pluperfect (inference or hearsay) conditional or future (only inference) screeves.

2.5 The pluperfect

The Pluperfect expresses an action anterior to another one which has also taken place in the past. The default value of the pluperfect is resultative. It can acquire an evidential extension (inferential or reportive) which can be reinforced by lexical markers:

(7) sulganabul-i bulbul-i potl-eb-si mimaluliq'o.

silent-NOM nightingale-NOM leave-PL-in hide:PPRF:3SG

'The nightingale was silently hidden among the leaves, (as it seemed)'.

(8) turme mama-cem-s k'argi mozrdil-i luk'ma caedva p'ir-si

EV father-my-DAT good-NOM big-NOM bite:NOM put:3.PPRF mouth-in

ise, rom dzer myvdel-s mama-o cven-o ar etkva

so that yet priest-DAT father-VOC our-VOC NEG tell:3PPRF

da supra ar ek'urtxebina.

and table NEG bless:3PPRF

'Apparently my father had put in his mouth a big bite before the priest had said the "Lord's prayer" and had blessed the table'.

2.6 The future

Evidentiality encoded in the future overlaps with epistemic modality as in Romance and Germanic languages (cf. Pietrandrea 2005: 42). It expresses the inference based on internal (reasoning) or external (hearsay or common knowledge) source (cf. Giacalone and Topadze 2007). The inferential future takes on epistemic overtones of uncertainty (Boeder 2000:280), losing its temporal meaning:

(9) v-rek'av, magram aravin p'asuxobs, c'asul-eb-i iknebian.

1SUB-call:PRS but nobody answer:PRS.3SUB gone-PL-NOM to be:FUT:3PL

'I am calling but nobody answers, they might be out'.

The number of verbs which can form the inferential future is quite restricted. These are usually stative verbs as codna 'to know', siq'varuli 'to love', kona 'to have', dzdoma 'to sit', dgoma 'to stand', yireba 'to cost', c 'ola 'to lie', etc.

2.7 The conditional

The conditional is equivalent to the future in the past. It substitutes the inferential future for the past events even for those verbs which do not admit inferential future forms:

(10) ra garudzul-i x-ar! zyva-ze iknebodi.

How tanned-NOM 2SUB-to be:PRS sea-on to be:FUT.2SUB

'How tanned you are! You must have been to the seaside'.

The conditional shares common traits with the perfect, since it refers to past events or their results, but, unlike the perfect, which is limited to stating a fact, the conditional conveys the speaker's commitment towards a statement.

3. Lexical Strategies

The reference to the information source can be expressed by employing lexical items, as mentioned above. They occur autonomously or in combination with grammatical strategies of evidentialty. They can modify the meaning of an utterance or intensify its evidential meaning. Quotative markers, comparative conjunctions, evidential particles, adverbial expressions as well as semigrammaticalized markers derived from verbs of perception can be considered lexical markers of evidentiality (an overview is given in table 2). In traditional grammars of Georgian lexical units with evidential value are usually mentioned along with the invariable words or modal particles (Shanidze 1980:612, Fahnrich 1986:146 among others). Among the markers of lexical evidentiality the quotative particles are undoubtedly the most discussed ones, especially in connection to indirect speech (Shanidze 1980, Hewitt 1981, Boeder 2000, 2002, Harris and Campbell 1995, Kvachadze 1996, etc.). The particle /turme/ 'apparently', lexical marker of evidentiality par excellence, is mentioned in numerous studies in correlation to the evidential meaning of the perfect tense (Shanidze 1980, Boeder 2000, Arabuli 1984, Pkhakadze 1984, Peikrishvili 1973, Fahnrich 1986, Hewitt 1995, among others) or to evidentiality more generally (Kavtaradze 1954, Boeder 2000). In the literature there are also some insights on inferential and reportive markers of evidentiality within investigations about conjunctions (Dzidziguri 1973) and mood (Jorbenadze 1982, Sharashenidze 1995, Harris 1995 etc.), discussing predominantly the epistemic-modal functions of these items.

3.1 Quotative markers

In Georgian there are 3 quotative markers (also known as quotative particles or speech particles): [-metki], [-tko] and [-o]. The markers [-metki] and [-tko] are derived from the verb tkma 'to say'(-metki = me vtkvi, 'I said'. tko<tkva 'he said'). The origin of the particle [-o], which is used mainly for quotations in the third person, is obscure. The quotative markers can generally be cliticized to each constituent of the sentence. Usually they are used as direct speech markers.

The particle [-metki] is used to codify firsthand information. In the first person quotations the speaker uses [-metki] to report a statement made or thought by himself in the past. In this case [metki] serves as a marker of direct evidentiality i.e. firsthand information, since the speaker is personally the information source for the stated utterance. The particle [- metki] generally takes a final position within a sentence:

(11) damagviandeba-metki.

to be late:1OBJ:FUT-1QUOT

'I will be late (I said)'.

In some dialects of Georgian [-metki] occurs as an autonomous lexical unit /metki/ in the headposition preceding the main verb (Dzidziguri 1973:446, Boeder 2002:17, 41 note 34). The original meaning of ' saying' is conserved:

(12) metki gaetrie akedan!

1.QUOT go away:2IMP here:from

'(I said) Out of the way'!

The occurrence of /metki/ as an autonomous unit emerges also in contemporary urban Georgian, where it undergoes morphosyntactic and semantic changes, losing its original meaning of saying and acquiring epistemic overtones of surprise by discovering the opposite information from that expected by the speaker. Such use of /metki/ has not been previously discussed in the literature.

(13) kidev ak xar? metki c'axvedi!

Still here be:PRS:2SUB EV go away:AOR:2SUB

'Are you still here? I thought you went away'.

In colloquial language /metki/ sometimes occurs twice in a sentence as an autonomous lexical unit (at the beginning) and as an enclitic (at the end of the sentence) in order to intensify the marking of reported speech:

(14) metki c'amoiq'vane egec-metki.

1.QUOT to take with-2IMP he-too-1.QUOT

(I said) Take him also with you (I said).

In contrast to [-metki] the quotative markers of the 2nd and 3rd person codify the second hand information. [-tko] is used when the addressee is a mediator between the speaker and a third person. The utterance is addressed to the third person, the information source is the speaker (see also Boeder 2002:15, Giacalone and Topadze 2007):

(15) utxari, dyes cem-i st'umar-i xar-tko.

tell:2IMP today my-NOM guest:NOM to be:2SUB-2QUOT

Tell him/her, you are my guest today (s/he= speaker, 1st person said).

The quotative marker [-o] also reports second hand information. In contrast to [-metki] and [-tko] it does not necessarily specify the information source which can be common knowledge for others. The particle [-o] occurs often in proverbs. For a detailed discussion about quotative markers see (Shanidze 1980, Hewitt 1981, Boeder 2000, 2002, Harris and Campbell 1995, Kvachadze 1996, Giacalone and Topadze 2007).

In formal register the information source is generally marked by adverbial expressions (in the adverbial or instrumental case), such as: mixedvit (from xedva 'vision') 'according to' (lit. with the vision), tanaxmad (from xma 'voice') 'according to' (lit. with the voice), cnobit (from cnoba 'notice') 'according to the notice', gancxadebit (gancxadeba 'declaration') 'according to the declaration', inpormaciit (inpormacia 'information') 'according to the information', etc.

(16) st'at'ist'ik'ur-i monacem-eb-is mixedvit, 2007 c'el-s q'vela-ze

statistic-NOM data-PL-GEN according 2007 year-DAT all-on

met'i p'iradob-is moc'moba dek'ember-si gaica.

most identity-GEN certificate:NOM December-in to issue:AOR.3SG

'According to the statistic data most identity cards have been issued in December 2007'.

(17) arsebul-i inpormaci-is tanaxmad, samxareo p'olicia-m

available-NOM information-GEN according regional policy-ERG

uk've aydzra gamodzieba.

already to open:AOR.3SG inquiry:NOM

'According to the available information, the regional police has already opened the inquiry'.

(18) sinop't'ik'os-eb-is cnob-it xval mtel sakartvelo-si gamoidarebs.

meteorologist-PL-GEN information-INS tomorrow whole Georgia-in to brighten up:3FUT

'According to the meteorologists, tomorrow the weather will brighten up in the whole Georgia'.

3.2 Evidential particles

3.3 Turme

The invariable particle /turme/ 'apparently, seemingly, as it is said' grammaticalized from the union of the conjunction /tu/ 'if, the particle /re/ 'somewhat' and the indefinite particle /me/ (cf. Boeder 2000:281, 313). It does not reflect directly the meaning of the contained elements (Kavtaradze 1956:180). We find the first occurrences of /ture/ (earlier form of /turme/), in texts from the 12th century (Arabuli 1984:142). In modern Georgian /turme/ has the same use as in Old Georgian. It may co-occur with the evidential perfect, but not obligatorily. The Georgian name of evidential screeves (turmeobitis mc 'k'rivebi is derived from the particle /turme/.

/turme/ has both inferential and reportive value. It indicates that the speaker does not have direct knowledge of what he is retelling, it normally does not contain epistemic assessments and does not cast doubt on the reliability of the utterance. The combination [aorist of the main verb + /turme/] can replace the evidential perfect in all declarative phrases of indicative mood:

(19a) c'uxelis uc'vimia.

last night to rain:PRF.3SG

'Apparently it rained last night'.

(19b) c'uxelis ic'vima turme.

last night to rain:AOR.3SG EV

'Apparently it rained last night'.

The particle /turme/ has its counterparts in all Kartvelian languages. In Megrelian it has been borrowed from Georgian, it is attested under the forms of /trume/, /turme/, /rtume/, whereas its Laz counterpart /megerem/ is a Turkic loanword. In Svan there is an autochthonous form equivalent of /turme/ attested under the form /esnar/ and its phonetic variants /ecnar/ /esnarildas/, /esren/, /esrenis/) 'apparently'. The particle /turme/ and its counterparts in Kartvelian Languages can be combined with all (evidential and non evidential) screeves of the indicative mood.

3.4 Mgoni

/mgoni/ 'it seems, it appears, seemingly, I think' (from m-gon-i-a, 1st person singular of the verb /goneba/ 'to think, to appear, to believe') works as a parenthetical element which modifies the meaning of the utterance, it contains a strong epistemic component of the speaker's commitment. /mgoni/ assumes evidential overtones in cases where it indicates the information source which serves as a basis of the statement.

Epistemic use (information source is not known):

(20) ramden-i c'l-isa-a? mgoni xut-is.

how many-NOM year-GEN-to be:PRS:3SG INF five-GEN

'How old is s/he? Five, as I believe' (speaker's memory).

Evidential use (information source is indicated):

(21) narcit'as cit'bat'onasac edzaxian mgoni.

kinglet:DAT goldfinch:DAT:too call:3SUB:PRS INF

'The kinglet is also called goldfinch, as it seems/I think (Hearsay).

3.5 Et'q'oba

/et'q'oba/ 'noticeably', 'seemingly', grammaticalized as inferential marker from the 3rd person of the verb /set'q'oba/ 'to notice'/dat'q'oba/ 'to be noticeable' (lit. 'It is noticeable on him/her'). It contains an epistemic element showing the speaker's attitude towards the statement. The use of /et'q'oba/ is characteristic for the colloquial language. It also appears in less formal registers of the written language.

(22) t'elepon-i gamoirto, et'q'oba element'-i daudzda.

phone-NOM to switch off:AOR:3SG:PASS INF battery-NOM to discharge:AOR:3SG

'The phone got switched off. Apparently its battery got discharged'.

With past events /et'q'oba/ occurs sometimes in the imperfect screeve. For this reason it must be considered as a semi-grammaticalized element, since it is not completely fossilized and is collocated on the continuum between lexicon and grammar (Topadze 2008):

(23) zvav-ze dzixv-eb-is naval-i mravlad iq'o.

landslide-on ibex-PL-GEN track-NOM much:ADV to be:AOR.3.SG

et'q'oboda c'q'l-is dasalevad ak evlo-t zevi-dan.

INF water-GEN for drinking here go:PPRF-3PL up-from

'There were many tracks of ibexes (mountain goats) on the landslide. Apparently (lit. as it seemed) they came down here to drink'.

3.6 Cans

/cans/ 'it is visible' is another semi-grammaticalized inferential marker derived from the 3rd person singular of the verb /gamocena/ 'to appear', which loses its original function of marking direct evidence and assumes an inferential value. /cans/ can co-occur with the evidential perfect:

(24) es p'rozauli variant'-i, cans, gauleksav-s vinme mesx-s.

this prose:ADJ version-NOM INF versify:PRF-3SUB some Meskhi-DAT

'This prose version, as it appears, was versified by some Meskhi (=inhabitant of Meskheti, Georgian region)'.

Like /et'q'oba/, the marker /cans/ can occur also in the imperfect under the form /canda/ 'it was visible') when it refers to past events:

(25) rogorc canda, pot'o k'oncert'-ze iq'o gadayebuli.

as be visible:IPF.3SG photo:NOM concert-on to be:AOR.3SG taken

'As it seemed/as it was visible, the photo was taken during the concert'.

4. Inferential and Reportive Markers Derived from Conjunctions

/titkos/ and /vitom/ (5) 'as if, as though' are evidential markers grammaticalized from the homonymous comparative conjunctions. In many contexts they may be used synonymously. /titkos/ (derived from /tu/ 'if + optative of the verb /tkma/ 'to say') developed in modern Georgian an evidential (inferential and reportive) function with the meaning 'as it seems, as it is said'.

/vitom/ 'as if', 'as though' (derived from the Old Georgian comparative conjunction /vit(a)/ 'as' (cf. vit(a) >vitamca>vitomc>vitom), developed a reportive value. In some context of illusory perception (such as dreams, etc ...) it can have an inferential interpretation. Both /titkos/ and /vitom/ contain epistemic overtones. In comparative constructions the evaluation of the statement by the speaker expressed through /titkos/ can be positive or negative, whereas /vitom/ implies necessarily a negative attitude of the speaker towards an utterance.

(26a) ise ikceva, titkos araperi icodes am sak'itx-is sesaxeb.

so behave:PRS:3SG EV nothing know:SUBJ.3SG this:GEN issue-GEN about

'He behaves as if he knew nothing about this issue'.

(26b) ise ikceva, vitom araperi icodes am sak'itx-is sesaxeb.

So behave:PRS:3SG EV nothing know:SUBJ.3SG this:GEN issue-GEN about

'He behaves as if he knew nothing about this issue' (Negative attitude of speaker)'.

/titkos/ and /vitom/ with evidential value occur often as parentheticals, losing in such cases the original function of connectors or conjunctions.

(27a) mariam-i titkos gusin dabrund-a mtidan.

Mary-NOM EV yesterday return:AOR-3SUB mountain:from

'Mary seems to have returned yesterday from the mountains'(Inferential or reportive)'.

/vitom/ is often accompanied by the hearsay marker [-o]:

(27b) mariam-i vitom gusin dabrund-a mtidan-o.

Mary-NOM EV yesterday return:AOR-3SUB mountain:from:QUOT

'Mary seems to have returned yesterday from the mountains, as they said' (Only reportive).

By using the marker /titkos/ the speaker does not assume any responsibility for the validity of the statements and reserves the possibility to change or withdraw his opinion or statement:

(28) gsia? titkos ara.

Be hungry:PRS.2SG INF no

'Are you hungry? Seemingly not' (as I feel)'.

/vitom/ can be additionally used independently as a modal-interrogative particle, as in (29). In such a context it can not be replaced by /titkos/ (Topadze 2008):

(29) -am problem-is gadac'ra advilad seidzleba.

this:GEN problem-GEN solve-MASD easily be possible:PRS:3SG

This problem is easily solvable.



'Can it be really true? Is it credible' (Scepticism of the speaker).

The derivation of evidential markers from comparative constructions is a typologically widespread phenomenon. (Wiemer 2008a:20, Wiemer 2008:348). Wiemer observes for Slavic languages that evidential elements developed from comparative constructions are inclined to oscillations between inferential and reportive domains. This claim seems to be valid also for the Georgian evidentials /titkos/ and /vitom/ which can be defined as 'undifferentiated indicators of indirect evidentiality, the designation proposed by Wiemer (Wiemer 2008:348).

4.1 Inferential markers derived from modals. The case of unda

The invariable marker /unda/ (3rd person of the verb /ndoma/ 'to wish, desire') developed as a marker of deontic modality, which in combination with the optative of the main verb can acquire an inferential value:

(30) pankar-s ver v-p'oulob! magida-ze unda

pencil-DAT NEG 1SUB-find:PRS table-on must

idos, axlaxans movk'ari tval-i.

lie:OPT.3SG recently glimpse:AOR.1SUB eye-NOM

'I can not find the pencil! It must be on the table, I just glimpsed it'.

4.2 Inferential markers derived from a nominal affix: -savit

The suffix [-savit] is a clitic composed from the dative marker [-s] + epenthetic vowel [a-] and postposition [-vit] 'like', 'as' (Amiridze 2006). Its primary function is comparative:

(31) cem-s-a-vit


'As I'.

The clitic [-savit] grammaticalized as evidential suffix which can be cliticized to the finite verbal forms (Jorbenadze, Kobaidze, Beridze 1988:170, Amiridze 2006). In contemporary urban Georgian some forms can be found, where [-savit] functionally corresponds to the marker /titkos/ ' it seems, it appears' in its inferential uses:

(32a) acivda-savit.

start to get cold:AOR:3SG-INF

'It is starting to get cold, as it appears'.

(32b) titkos acivda.

INF start to get cold:AOR:3SG

'It is starting to get cold, as it appears'.

5. Conclusions

Summarizing results of the present discussion we can conclude that evidentiality in Georgian is a functional, semantic category. Marking the information source is not obligatory and it depends on the choices of the speaker. Evidentiality can be expressed by various grammatical and lexical means: direct evidentiality is codified by unmarked forms of TAM and by the first person quotative marker [-metki], whereas indirect evidentiality and its sub-domains of inference and hearsay are encoded in the TAM-paradigms, which have developed evidentiality as a secondary meaning, or can be conveyed by lexical means. There are no overt morphological i.e. verbal markers of evidentiality, with only the meaning of "information source".

As reflected by the analyzed data, the encoding of evidentiality in Georgian is not restricted to the indirect evidentiality, as sometimes claimed in the literature. The claim according to which Georgian marks only indirect evidentiality (second hand information) is valid for some grammatical strategies (e.g. the perfect), but not for all the means of expression of evidentiality.

In contrast to the grammatical strategies which developed evidentiality as a secondary meaning, like the perfect, some lexical strategies may have evidentiality as unique meaning.

The investigated data confirm that evidentiality in Georgian is a domain in movement, which admits several intermediate stages, as manifested by the presence of both grammaticalization processes involving lexical items, and degrammaticalization processes involving affixes.

The evidential value may overlap with the epistemic one within a marker, as demonstrated by a large number of lexical means. However, such overlap is not obligatory, as for instance in the case of the perfect, which is not compatible with the epistemic elements. The Georgian perfect, which has developed an evidential value in addition to other meanings, corroborates the claim made by some authors (Aikhenvald 2004, De Haan 1999, Wiemer 2010) according to which evidentiality and epistemicity are two different notions, both at the conceptual and functional level.


ADV-Adverb, ADJ-Adjective, AOR-Aorist, DAT-Dative, EPV-epenthetic vowel, ERG Ergative, EV-Evidential, FUT-Future, GEN-Genitive, IMP-Imperative, INF- Inferential, INS-Instrumental, IPF-Imperfect, MASD-Masdar, NEG-Negation, NOM-Nominative, OBJ-Object, PASS-Passive, PL-Plural, P.P-Perfect Participle, PPRF-Pluperfect, PRF-Perfect, PRS-Present, QUOT-Quotative, SG-Singular, SUB-Subject, TAM-Tense/Aspect/Mood, VOC-Vocative.


Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press.

Amiridze, Nino. 2006. Originally Nominal Affixation on Finite Verb Forms in Modern Georgian and Laz. Presentation at the International Workshop on Morphosyntax of Caucasian Languages. Inalco and College de France, Paris (Manuscript).

Arabuli, Avtandil. 1980. Mesame seriis nak'vteulta c'armoeba da mnisvneloba dzvel kartulsi [The formation and meaning of the third series of tenses in Old Georgian]. Tbilisi.

Aronson, Howard I. 1977. The Interrelationships between Aspect and Mood in Bulgarian. Folia Slavica 1, 1.9-32.

--. 1982. Georgian: A Reading Grammar. Columbus: Slavica Publishers. Boeder, Winfried. 2000. Evidentiality in Georgian. In: Evidentials: Turkic, Iranian and

Neighbouring Languages, ed. by Lars Johanson & Bo Utas, Berlin-New York, Mouton de Gruyter, 275-328.

--. 2002. Speech and thought representation in the Kartvelian (South Caucasian) Languages. In: Reported discourse: a meeting ground for different linguistic domains (Typological Studies in Language 52), ed. by. Tom Guldemann & Manfred von Roncador. Amsterdam: John Benjamins, 3-48.

Chikobava, Arnold. 1950. Kartuli enis zogadi daxasiateba [A general characteristisation of the Georgian Language]. Kartuli enis ganmart'ebiti leksik'oni, ed. by A. Chikobava, Tbilisi.

De Haan, Ferdinand. 1999. "Evidentiality and epistemic modality: Setting boundaries", Southwest Journal of Linguistics, 18, 83-101.

--. 2005a. Semantic Distinctions of Evidentiality. In: The World Atlas of Language Structures-WALS, ed. by Martin Haspelmath, Matthew S. Dryer, David Gil, Bernard. Comrie. Oxford: Oxford University Press, 314-317.

--. 2005b. Coding of Evidentiality. In: The World Atlas of Language Structures- WALS, ed. by Martin Haspelmath, Matthew S. Dryer, David Gil, Bernard. Comrie. Oxford: Oxford University Press, 318-321.

De Lancey, Scott. 1997. Mirativity: The Grammatical Marking of Unexpected Information. Linguistic Typology 1, 33-52.

Dzidziguri, Shota. 1973. K'avsirebi kartul enasi ('Conjunctions in Georgian'). Tbilisi.

Fahnrich, Heinz. 1986. Kurze Grammatik der georgischen Sprache. Leipzig, Verlag Enzyklopadie.

Friedman, Victor. 1979. Toward a typology of status: Georgian and other non- Slavic languages of the Soviet Union. In: The elements: A parasession on linguistic units and levels, ed. by Paul R. Clyne, William Hanks & Carol Hofbauer. Chicago Linguistic Society, 339- 350.

Giacalone Ramat Anna & Manana Topadze. 2007. The coding of evidentiality: a comparative look at Georgian and Italian. In: Evidentiality between lexicon and grammar, Italian Journal of Linguistics, n. 19,1, ed. by Mario Squartini, Pisa.

Givon, Talmy. 1982. Evidentiality and epistemic spaces. Studies in Language, 6, 23-49.

Guentcheva, Zlatka. 1996. L'Enonciation mediatisee. Bibliotheque de l'information grammaticale. Louvain: Editions Peeters.

Harris, Alice C. 1981. Georgian Syntax. A study in relational grammar. Cambridge Studies in Linguistics, vol. 33. Cambridge, Cambridge University Press.

--. 1985. Diachronic Syntax: The Kartvelian Case. Syntax and Semantics, vol. 18. Academic Press, Orlando.

--. 1995. "Modal Auxiliaries in Georgian". In: Pilologiuri Dziebani (Festschrift for Guram Kartozia, ed. Aleksandre Gvakharia), Tbilisi, Mecniereba, 195-207.

Harris Alice and Lyle Campbell. 1995. Historical Syntax in Cross-Linguistic Perspective. Cambridge Studies in Linguistics, vol. 74. Cambridge University Press, Cambridge.

Hewitt, George B. 1981. "Eine weitere Betrachtung der georgischen Redepartikel '-tko//-tkva'. Georgica, n. 4.

Ioseliani, Platon. 1863. P'iruel-dac'q'ebitni k'anonni kartulisa yrammat'ik'isa/sedgenilni p'lat'on egnat'es dzis ioselianisagan--aybec'dilni mesamed [First Canons of Georgian Grammar, gathered by Platon E. Ioseliani, printed for the third time], Tbilisi.

Jorbenadze, Besarion. 1982. Nazmnari modaluri element'ebi kartulsi. [Deverbal modal elements in Georgian], Kartuli sit'q'vis k'ult'uris sak'itxebi, 10. Tbilisi.

Jorbenadze Besarion, Manana Kobaidze & Marine Beridze. 1988. Kartuli enis morpemebisa da modaluri element'ebis leksik' oni [Dictionary of Morphemes and Modal Elements of Georgian]. Tbilisi, Mecniereba.

Kavtaradze, Ivane. 1956. Unaxaobis k'at'egoriis ist'oriisatvis kartulsi [On the History of the Category of Non-witnessedness in Georgian]. In: IKE (Iberiul-k'avk'asiuri enatmecniereba), 8, 179-193.

Kvachadze, Leo. 1996. Kartuli enis sint'aksi (Syntax of Georgian language). Tbilisi.

Natadze, Nodar. 1955. Mesame seriis dro-k'ilota c'armoebisatvis kartulsi [On the formation on series III in Georgian]. In: IKE (iberiul-k'avk'asiuri enatmecniereba) 7, 81-98.

Peikrishvili, Zhuzhuna. 1974. Turmeobitebis mnisvneloba da gamoq'eneba axal kartulsi [The meaning and use of evidentials in Modern Georgian]. Kartuli enis st' rukt' uris sak' itxebi 4, 53-70.

Pietrandrea, Paola. 2005. Epistemic Modality. Functional Properties and the Italian System. Amsterdam-Philadelphia, John Benjamins Publishing Company.

Pkhakadze, Darejan. 1984. P'irveli turmeobitispunkciebi kartulsi [Functions of the 1st evidential (present perfect) in Georgian]. Tbilisi: Tbilisis universit'et'is gamomcemloba.

Sakhokia, Maia. 2002. New analytical perfects in Modern Georgian. Proceedings of the 3rd and 4th International Tbilisi Symposium on Language, Logic and Computation, ILLC/ICLC, ed. by H. Zeevat D. de Jongh and M. Nilsenova. Amsterdam/Tbilisi, 1-10.

Shabashvili, Giuli. 2001. P'erpekt'isa da unaxaobis k'at'egoria dzvel kartulsi [The Category of perfect and non-witnessedness in Old Georgian], Doctoral Dissertation, University of Tbilisi.

Shanidze, Akaki. 1930. Kartuligramat'ik'a, I. Morpologia [Georgian Grammar, I. Morphology], Tbilisi.

--. 1953. Kartuli enisgramat'ik'is sapudzvlebi [Foundations of Georgian Grammar]. Tbilisi

--. 1973. (Reprint 1980). Kartuli enis gramat'ik'is sapudzvlebi [Foundations of Georgian Grammar]. Tbilisi: Sakartvelos ssr mecnierebata ak' ademia.

Sharashenidze, Nino. 1995. Unda modaluri pormis ist'oriisa da morposint'aksuri analizisatvis [On the History and Morphosyntactic Analysis of the Modal Form unda] Enatmecnierebis sak'itxebi, 1999/4, Tbilisi, 49-56.

Topadze, Manana. 2008. Mezzi di espressione dell'evidenzialita in Georgiano. Pavia: Universita degli Studi di Pavia (Doctoral Dissertation).

Willett, Thomas (1988) "A Cross-Linguistic Survey of the Grammaticization of Evidentiality", Studies in Language 12, (1988: 51-97).

Wiemer, Bjorn. 2008. Pokazateli s citativnoj i inferentivnoj funkcijami v russkom i pol'skom jazykach--kommunikativnye mechanizmy semanticeskogo sdviga [Indicators with reportive and inferential functions in Russian and Polish languages--Communicative mechanisms of the semantic shift]. In: Lexikalische Evidenzialitatsmarker im Slavischen (=Wiener Slawistischer Almanach, Sonderband 72), ed. by Bjorn Wiemer and Vladimir A. Plungjan, 337-378.

--. 2008a. Lexikalische Markierungen evidenzieller Funktionen: zur Theoriebildung und empirischen Erforschung im Slavischen. In: Lexikalische Evidenzialitatsmarker im Slavischen (=Wiener Slawistischer Almanach, Sonderband 72), ed. by Bjorn Wiemer and Vladimir A. Plungjan, 5-50.

--. 2010. Hearsay in European languages: toward an integrative account of grammatical and lexical marking. In: Linguistic Realization of Evidentiality in European Languages (Empirical Approaches to Language Typology [EALT]), volume 49. ed. by Gabriele Diewald & Elena Smirnova. Berlin, New York, Mouton de Gruyter, 59-130.

Manana Topadze Gaumann

University of Bern

Author's Contact Information:

Manana Topadze Gaumann

Universita di Pavia

(1) The material presented in this article is a brief summary of a part of the PhD thesis completed by the author at the University of Pavia, Italy (Topadze 2008).

(2) The analysed data on Modern Georgian has been drawn from contemporary Georgian literature, magazines and newspapers. The TITUS database (University of Frankfurt) was exploited as the source for Old Georgian data. Data from the colloquial language was taken from different internet forums and collected by means of questionnaires and interviews.

(3) Term used by Aikhenvald (2004) who discusses evidential systems with two, three, four, five and more choices.

(4) Screeve is a paradigmatic set (tense-aspect-mood) of verbal forms which inflects only for person and number. The term screeve was coined by Aronson (1982), it is a phonetic adaptation of the Georgian term mc'k'rivi 'row', introduced by Shanidze in 1930.

(5) The conjunctions titkos and vitom have following phonetic variants: titkos: titkosda; vitom: vitomc, vitomda, vitomdac, vitomcda.
Table 1: TAM paradigm of the Georgian Verb

SERIES          SCREEVE (4)   TENSE      MOOD          ASPECT

I Present       Present       Present/   Indicative    Imperfective
  (Future       Imperfect     Past       Indicative
                Present       Present/   Subjunctive
                subjunctive   Future

                Future        Future     Indicative    Perfective

                Conditional   Past       Indicative

                Future        Future     Subjunctive

II Aorist       Aorist        Past       Indicative    Perfective

                Optative      Present/   Subjunctive

III Perfect     Perfect       Past       Indicative

                Pluperfect    Past       Indicative

                Perfect       Past/      Subjunctive
                Subjunctive   Future

Table 2: Lexical markers of evidentiality in Georgian

Types of       Quotative Markers      Adverbial
Evidentials                           Expressions


Reportive      -metki 1st pers. sg    mixedvit tanaxmad
                                      cnobit gancxadebit
               -tko 2nd pers sg/pl,   'according to'
               1st pers. pl.

               -o 3rd pers. Sg/Pl.


Types of       Evidential     Comparative     Semigrammaticalized
Evidentials    Particles      Conjunctions    Markers derived from
                                              perception verbs

Inferential                                   et'q'oba 'it seems'
                                              cans 'it is visible',
                                              'it seems' mgoni 'it
                                              appears, I think'

Inferential/   turme          titkos 'as if
Reportive      'apparently'   vitom 'as if

Types of       Modals   Degrammaticalized
Evidentials             Affixes

Inferential    unda     -savit 'it seems,
               'must'   like, as'


COPYRIGHT 2011 Dartmouth College Library
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2011 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Gaumann, Manana Topadze
Publication:Linguistic Discovery
Date:Jul 1, 2011
Previous Article:On the expression of spatial relations in Ardesen-Laz.
Next Article:A corpus study of Mexican Spanish three-participant constructions with and without clitic doubling.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters