In the written monuments of Khanmet the existence of Kh-prefix in all three person forms of I-Prefix Passive Voice is a norm (Therefore, in Haemet we come across H-):          

I.პ. ხჳქმენ

II. პ. ხიქმენ

III. პ. ხიქმნა

Although this is a norm but still we have some cases of deviation. The issue of Kh-//H- Prefix in all three person forms of I-Prefix Passive Voice causes dispute among scholars and linguists. Before answering the problem outlined above, we should briefly discuss the conditions of the ancient Khanmet and Haemet manuscripts.

In the extract from the Khanmet on Jeremiah's Prophecy shown in the Oxford and Cambridge Fragments[1] (VI-VII centuries) [Shanidze, 1937:32] forms of I-Prefix Passive Voice (9 in total) are represented by Kh-//H-[2] Prefix:

I Person. In the monument there is an evidence of only one form of subjective person [ხჳშ]ევ [Jeremiah 20:14][3], but according to the publisher "the letters inserted in the staples are ragged parts" [Shanidze, 1937:32], therefore we cannot rely on this form.

II Person. The forms of second person are not depicted in the fragments.

III Person. The forms of third person subjective have Kh- Prefix: ხიყოს [Jeremiah 12:16, 20:14, 20:15], ნუ ხიყოფ(ინ) [Jeremiah 20:14], შეხიმუსრა [Jeremiah 18:4], ხიყო [Jeremiah 18:3]. But there is an evidence of H-Prefix as well: დაჰიდვა [Jeremiah 12:11], განჰირყუნა [Jeremiah 12:11].

As we can see in the Jeremiah fragment of Khanmet forms of I-Prefix Passive Voice are respresented by Kh-//H- Prefix. There is no evidence of the sequence of Kh- with zero, which we come across in the inscription of Mtskheta Cross (inscription of Demetre Upatosi (end of VI century) [Javakhishvili, 1996:161] can be rad the form „მეოხ ხეყავ'', the same verb that is found in the inscription of Adarnase Upatosi (First half of VII century) [Javakhishvili, 1996:160] without Kh-prefix: „მეოხ ეყავ''. If we consider Jeremiah's fragment as the pure epoch of Khanmet, then chronologically Mtskheta Cross and the inscription of Adarnase Upatosi might preceded it. The existence of Haemet forms in the manuscript can be explained by the linguistic environment that influenced the copyist.

In the mansucripts of Khanmet and Haemet that have reached us so far only the Khanmet Lectionare (second half of VII century) [Shanidze, 1944:019] is not a palimpsest.

I Person. The form of first subjective person in I-prefix Passive voice is not shown in the fragments that have reached us.

II Person. The form of second form subjective is represented by the Kh- prefix ხიქმნები [Shanidze, 20:27].[4]

III Person. For the second person Kh-prefix forms are a norm. In the monument to a large extent the norm of Khanmet is seen, as well as its use in all other cases (ხიქმნა [მ. 28:11; ლ. 24:18], გახითქუა [მ. 28:15], ხიყო [მრკ. 16:4; ლ. 24:15; ლ. 24:26; ი. 20:19] და სხვ.). In two cases the paasive voice form of I-prefix is represented by H-prefix: შეჰიძრნენ [მ. 24:29], ჰიყოს [ლ. 12:34]. To ease reading the Kh-prefix forms is erased in nearly half of the cases. If we take into consideration the first condition of the monument there is no evidence of any form of I-prefix Passive Voice, which is depicted without Kh-//(H-) prefix. This fact highlights the ancientness of the text. The Khanmet Lectionaire shows three forms of Haemet: ჰიხილოთ [მრკ. 16:7; მ. 24:33], ჰიცილობთ [ლ. 24:17], ჰეტყებდენ [მ. 24:30], მიჰეხების [ლ. 12:33], შეჰიძრნენ [მ. 24:29], ჰიყოს [ლ. 12:34]. In this case, the existence of Haemet forms can be explained by the influence of the environment on the copier.

The linguistic analysis of the monument clearly shows that the Khanmet Lectionare is one of the most ancient manuscripts of Khanmet. "It is copied in the period, when Khanmet was an important issue of literary Georgian: Indeed, it is not possible that it was only preserved through educational events and did not have support during speaking" [Shanidze, 1944:020]. If we take into account the epigraphic monuments of VII-VIII century, it coul be argue that Khanmet Lectionare preceed the Mtskheta Cross and the the inscription of Adarnase Upatosi and the date of its creation can be moved back to half a century before.

In the Khanmet extract of Jacob's Gospel (VII century) [Birdsall, 1969:108] the three person of I-prefix Passive Voice is represented by Kh- prefix:

I Person. ხჳქმნე [Jacob 12:9][5], აღხჳწერო [Jacob 20:13-14], ხჳპოვო [Jacob 17:19].

II Person. ხიყო [Jacob 12:1], აღხიზარდე [Jacob 18:14].

III Person. ხიქმნა [Jacob 15:23-16:1], ხიპოვა [Jacob 19:13], ხიკურთხა [Jacob 19:5], მოხიქცა [Jacob 20:10], მოხიწიოს [Jacob 13:2-3], ხიყო [Jacob 13:9-10], ხიყოს [Jacob 20:10], აღხიტყუა [Jacob 14:17], ხიშუეს [Jacob 15:1].

In the manuscript H- prefix is not represented. The ancientness of Jacob's proto-evangel is supported by the fact that in I-prefix Passive Voice the sequence of Kh- with zero does not occur. Also, the first person forms are represented by Kh- prefix. Similar is the case of the first person subject forms, when as a rule, third objective person in the dative case correlates with it: ხუ(ხედავ) [Jacob 17:2], მიხუხედე [Jacob 21:21], დახუფარო [Jacob 21:20].

In the Khanmet extract of Jacob's Gospel all the norms of using Kh-prefix are in accordance not only with I-prefix Passive Voice, but in other cases as well, but the manuscript shows that the destruction of Khanmet process has already started. To put it in other words, in one case we have შეაგინა [Jacob 16:19-20], and in the other - (შე)ხაგინა [Jacob 17:27-28] forms [Shanidze, 1977:11]. Taking into account this fact the manuscript can be dated as of the first half of the VII century.

In the Khanmet Gospel (VII century) in the I-prefix Passive Voice forms, genrally the Khanmet norm is preserved, although there are several deviations:

I Person. There are parallel operation concerning the case of first subjective person form of I-prefix Passive Voice: there is evidence of Kh-prefix (მოხჳვლინე [ლ. 4:43][6]), as well ase non-prefix (ვიყვენით [მ. 23:30]) forms. The latter is characteristic and belongs chronologically to the texts of later period, because it reflects the rupture process of Khanmet. I-prefix Passive Voice in the texts do not stand alone, and are supported by the forms of first subjective person . In this case we have simultaneous operations (ხჳქმ [მრკ. 11:29], განხუასხამ [ლ. 11:20], შეხუძინე [მ. 25:20]... compare: განვასხემდით [მ. 7:22], მოვგუარე [მ. 17:16], ვიტყჳთ [ი. 3:11]...). It is clear that in this period during live speech the easy forms of V-prefix are characteristic, which were skipped by the copier. Kh- Prefix forms of Khanmet should reflect the original situation.

II Person. In the Second Subjective person the form has Kh-prefix [მ. 18:3].

III Person. In the third person the Khanmet norm is generally preserved, (განხიკურნებოდეს [მ. 15:31], ხიქმნა [მ. 28:11], ხიქცეოდა [მ. 21:17], დახირღუეს [ლ. 21:6] და სხვ.), but there are several cases of deviation (იყოს [მ. 5:32; მრკ. 9:50], დაიმარილოს [მ. 5:32], განირყუნიან [მ. 6:16], აღივსნეს [ლ. 6:11], შეიმოსა [ლ. 12: 27]).

As we see in the forming of I-prefix Passive Voice the Khanmet norms are destructed. The Kh- prefix in the Khanmet Gospel is not represented in other cases (ნუ კლავ, ნუ იპარავ [მრკ. 10:19], უჩუენე [მ. 8:4], ულხინჱს [მ. 10:15]...) or over-used (მიხიყვანის [მ. 7:13, 14], შეხცთეს [მ. 18:12], ხისადილნეს [ი. 21:15] და სხვ.). As we see, in the Khanmet Gospel, differently from the manuscript discussed above, the norms of usage of Kh-prefix is destructed (Kh-prefix is lost in front of a vowel). These is caused firstly by the "errors" of the copier, which is motivated by the live speech environment, and secondly - deliberately and consciously written, so-called pseudo-grammatical forms, which is related to dim the norms of using Kh-prefix by the copier.

For the History of Literary Language the existence of S-prefix in the Khanmet Gospel is an important fact: „ესრჱთ არს სასუფეველი ღ ~ჲ,      ვ ~ა იგი კაცმან დასთესის" [მრკ. 4:26]. The "error" of the copier makes obvious that during the period of the creation of the manuscript (in at least in the region where the manuscript was created) the S-prefix exsists, the so-called Sannarevi speech (At least in the initial state). Therefore "the existence of s-prefix forms from the IX century is being questioned" [Sarjveladze, 1995:129].  The Khanmet in the monument gives the impression of being used as a norm of an educational way and can be explained by the literary tradition and the existence of Khanmet original.  Although in four cases we have the form of H-prefix instead of Khamet (ჰკლავ [მრკ. 10:19], ჰრქუა [მრკ. 3:5; ლ. 15:31; 16:2], but this is the result of a vandalism of later period, because below H-prefix the graphic image of Kh-prefix can clearly be seen [Kajaia, 1984:301].

At last we want to touch the issue of the dating of the Khanmet Gospel. If we take into account the errors outlined above, dating it to V-VI centuries may seem wrong (Epigraphic monuments of this period did not reveal features and disorders of Kh-prefix), as suggested by L. Kajaia [Kajaia, 1984:303]. We think that Z. Sarjvelize's idea about dating the Khanmet Gospel to VII century should be shared [Sarjvelaze, 1995:128]. The above discussed manuscript of Khanmet is the text copied later. There is one more theme concerning this issue: To what extent is it justifyied to put the Khanmet Lectionaire and Khanmet Gospel on the same flatness? We believe that the difference between them is at least half a century. Taking into account the fact listed above it should be justified to believe that the Khanmet Lectionaire was copied in the beginning of VII century [Utie, 1973:173], and in other case the dating of Khanmet Gospel to the end of VII century might be correct.

Khanmet Mravalthavi shows an interesting situation, which is dated by A.Shanidze to the first half of the VIII centuries [Shanidze, 1927:109].

I Person. The first subjective person form of I-prefix Passive Voice is presented without Kh-prefix: ვიქცეოდი [მრავ. 130:22][7], განვიბანეთ [მრავ. 132:12], ვიყვენით [მრავ. 132:15]. This is the norm strictly followed in this manuscript,without any eviation. The same happens in the first subjective person form of active forms: დავაცადო [მრავ 127:13-14], ვაჩუენოთ [მრავ. 128:4-5], შევწიროთ [მრავ. 131:18],   ვიტყოდით [მრავ. 134:3] and others. These forms, In fact, reflects new Georgian and Sannarevoba and clearly shows the real picture of the speech of that period. Therefore, the continuation of such linguistic data of Haemet, for which the forms აღჰუმართე, შეჰუწირო, დაჰუაცადო is characteristic, is needed. The confirmation of the აღჰუმართე type forms found in the Tskhisi Inscription dating to 616-619 years is a passed stage for the Khanmet Mravaltavi. Some linguistic processes concerning Khaemet and Haemet occurred simultaneously.  Thus, for some time, even if they are in different areas, we have to assume coexistence of these two data. Therefore, we think that it is unlikely that the Haemet was the next chronological step of Khamet as believed by Ivane Javakhishvili  [Javakhishvili, 1996:278] and Akaki Shanidze's attitude seems more appropriate [Shanidze, 1923:461].

In the Khanmet Mravaltavi is an evidence of old forms of Kh-prefix in several cases (ხუმადლობდეთ [მრავ. 127:6-7], შეხუწირვიდეთ [მრავ. 127:11-12], თავყანის ხუცე(მდე)თ [მრავ. 131:9], ხუემსგავსენით [მრავ. 142:10]), which was in the Khanmet original. As a norm V-prefix forms should be regarded. This peculiarity is stated by A. Shanidze as follows "in the first subjective person Khanmet does not flourish" [Shanidze, 1927:115].

We believe that the situation in the first subjective person is one of the main linguistic criteria (in conjunction with other signs) and should be considered for the chronologization of Khanmet and Haemet Texts. The quantitive abudence of V-forms (compared with the formation of Khu-) and moreover, the existence of this type of form must indicate on the lateness of the monument. Among the Khanmet manuscripts reviewed above, indeed the Khanmet Mravaltavi is the oldest.  

II Person. The second subjective person is without Kh-prefix (ხიყვნეთ [მრავ. 143:3], ხიქ(ცეოდით) [145:18-19], ხიქმნნეთ [მრავ. 145:19]). We believe that in this case the morphological factor interviened not letting Kh-prefix to be lost: Apparently, it was thought as a subjective entity.

III Person. The third subjective person is with Kh-prefix, but there are three exceptions: შეიწირვის [მრავ. 125:2], იქცენ [მრავ. 134:1], გარდაიქცეს [მრავ. 149:2] (The ratio is as follows: 32:3).

In general it can be said that in the Khanmet Mravalthavi Second Subjective Person and Thrid Objective person have the Kh-prefix, with a few exceptions, represented everywhere where Khanmet is an expected norm. The monument is noteworthy because there is none of the Haemet forms mentioned which should not be there. The thing is that during the time when the Mravaltavi was being copied Khanmet and Haemet was already a passed stage. From VII century in live speech Sannarevoba was in action (The South - East region of Georgia at least). The Khanmet used in Khanmet Mravaltavi is a literary tradition, an artificial event, a linguistic issue brought through an educational reform and not a fact depicting reality [Shanidze, 1927:116]. The copier of the monument is not just a copier, but a rather educated editor (that is why there are no errors in the monument),who creats orthographic norms suitable to the new era of, which implies a synthesis of old and new. First of all, we mean the linguistic phenomenon characteristic for the legalization of live speech, declaring a norm, which is clearly a violation of the "live" Khanmet and which is spontanousely done in the Khanmet Gospel. This is the stating V-prefix forms of the first subjective person as orthographic norms.  (ხჳყვენით>ვიყვენით, შეხუწიროთ > შევწიროთ...). By this norm the principle of prefix monopersonalization which was natural for Georgian Language was inserted. 

Haemet Lectionaire (VIII century) [Shanidze, 1923:358], which in fact, is believed to be the oldest example of Sannarev Monuments, where the half of I-prefix Passive Voice forms are represented by H-prefix.

I Person. The first subjective person form of I-prefix Passive Voice is presented with Kh-prefix: (ვიყო [ჰაემ. 370:10][8], მოვივლინე [ჰაემ. 382:15]). In this respect it stands beside the Khanmet Mravaltavi. 

Haemet Lectionaire, in general, the first-person subjective forms the Khu- complex, in fact, is not characterized any more: ვიტყჳ [ჰაემ. 387:10], მოვიმკი [ჰაემ. 385:6], მოველოდეთ [ჰაემ. 366:14], ვიდიდებდე [ჰაემ. 370:10]. A few old norms reflecting a more artificial form of impression, Haemet is explained by the original and cannot explain the general tendency (ჰუყოფ [ჰაემ. 388:15], ჰჳცოდე [ჰაემ. 388:13], სათნო ჰჳყავ [ჰაემ. 370:12]). In the monument we come across one form, that shows the new redistribution of H-U- prefix: დავჰთესი [ჰაემ. 385:1].

II Person. The second person Subjective in two cases are presented by H-prefix, the thir time without it: ჰიყოფი [ჰაემ. 387:12], ჰიყავ [ჰაემ. 376:13], იყვენით [ჰაემ. 378: 2-3]. The latter is of the II imperative row, therefore H-prefix can be  regarded as a sign of the Second Subjective person and this resulted in its loss, but this cannot be said on the previous case  (ჰიყავ verb is of the same imperative row).

III Person. There is an evidence of a sequence between H and Zero, in general, the half of third person subjective is represented by the H-prefix, the second half -without it (შდრ. ჰიქმნა [ჰაემ. 376:7], იქმნებიან [ჰაემ. 368:2-3]; ჰიქცეს [ჰაემ. 388:14], მოიქცა [ჰაემ. 386:7]; ჰიყვნეს [ჰაემ. 377:9], იყვნეს [ჰაემ. 383:8] და სხვ.).

The second subjective and third objective personal sign is always present before Consonants, an the loss of H-prefix before vowels is not fully implemented (შდრ. ჰიტყოდა [ჰაემ. 376:15], იტყოდა [ჰაემ. 369:7]; მიჰუგო [ჰაემ. 369:2], მიუგო [ჰაემ. 381:5]; ჰიცოდე [ჰაემ. 385: 3-4], იხილოთ [ჰაემ. 386:5] და სხვ.). There is shown three forms of Hs-contaminatian prefix ჰს- (მიჰსცემდა [ჰაემ. 377:2; 379:7], დაჰსცხრა [ჰაემ.  383:14]), which represent the ondition of both the live speech (where these verbs were pronounce by S-prefix) as well as in the original of Haemet.

The linguistic analysis of the Haemet Lectionare, shows that the copier was unable to handle the grammatical rules of the original and changed much under the influence of live conversational speech, which at the time of the creation should be a Sannarev. Although, in the text, differently from the Khanmet Gospel, the S-prefix adopted on the basis of assimilation is not confirmed, but the forms of Hs-contaminatian prefix show that in that period in live speech of Second Person and Third Objective Person as one of the allomorphs we have S-prefix.

The linguistic analysis of the Khanmet-Haemet Manuscripts shows the simplification of the form of I-prefix Passive Voice and the process of sequence between Kh-//H-prefix, which can be shown schematically as follows:

It is obvious that in the passive voice of I-prefix the sequence with zero should have started ion the I-III persons. This process was more active in I subjective person, by which we got the simplified V-prefix forms (this is the only norm for Khanmet Mravaltavi da Haemet Lectionaire). The loss of Kh-prefix in the III subjective person does not occur. And in II subjective person the loss of it was a result of a morphological factor: Kh-//H- was regarded as the morpheme of II subjective person. The simplification process of the forms was finally completed in the written monuments of Sannarev.

At the end we want to touch the main problem: What function does the Kh-//H- prefix has in the forms of Passive Voice and why does it occur in all three person? On Old Georgian synchronic levels, in some sense, a second person forms Kh-//H- prefix can be regarded as a morphem of second person [Shanidze,1927:114; Oniani, 1978:115; Sarjveladze, 1997:74], while in first person subject forms Khu-prefix can be qualified as an allomorph of one of the [Oniani, 1978:162; Gamkrelidze, 1979:49; Sarjvelaze, 1997:74]. In the I-prefix Passive Voice of thir person no morphological basis for its reflection can be found and by analogue, can be explained by the equivalence to paradigm [Oniani, 1978:165]. Generally, "Why does Kh-prefix occur in these types of verbs is not satisfactionally explained until now" [Sarjveladze, 1995:125].

To our mind, the opinion that Kh-//H- prefix is a morpheme of second person is followed by one main problem: If we consider it as a sign of the second person, would it be reasonable or even morphologically logical to identify it in the forms of other subjective person? By the opposition of person the concrete morpheme can be the marker of a member and in our case Kh-//H- prefix is being realized in all three persons. One important circumstance is that in the Khaemet and Haemet Texts I-prefix Passive Voice imperative row II subjective person is always represented by Kh-//H- prefix. This is a norm and there is no deviation (ხიყვენით [Birdsall, 1971:67], ხიყავ [ქაჯაია, 2006:50], ჰიყავ [ჰაემ. 376:13], but იყვენით [ჰაემ. 378:2-3]). In ancient Georgian Language the lack of sign in II Subject person is a morphological prohibition. Therefore, Kh-//H- prefix in I-prefix Passive Voice does not obey the general rule, which does not apply to the second-person subjective.  Therfore, we cannot regard the Kh-//H-prefix as a morphem of second-person subjective. But to what extant can we regard Khu- as an allomorph of first-person subjective?

We believe that to express the morphological feature of first person subjective Khu-prefix is associated with the second component, Kh-prefix in this case plays the role of Dysfunctional segment, as found in the forms of I-Prefix Passive Voice II-III person subjective. Since the Kh-segment of Khu-prefix does not have a morphological function, it can be regarded as an allomorph of first-person subjective, and its existence can be explained on a phonetically level. Since the expression of first-person subjective had nothing to do with the first component of Khu-prefix the language easily adapted to it and the simplification of "difficult" prefix: Khu-//Hu-> V-.

In scientific literature there is a belief about I-prefix Passive Voice form of behaviour deriving from it [Shanidze, 1953:354-356]. In the texts of Khanmet and Haemet I-prefix Passive Voice as norm is Kh-//H- prefix forms, which is not shown in the prefixes. Therefore, in the passive of this type the existence of Kh-//H- prefix cannot be explained on the basis of behaviour deriving from it. A. Oniani agrees with the opinion and believes that I-allomorph Passive Voice steps up to the common Kartvelian language an in all Kartvelian languages is one-person, and there is no basis to think that it was at some point two-person [Oniani, 1978:164].

As we see, I-prefix forms of Passive Voice does not have a morphological function of Kh-//H- prefix, it is an element without function, not only on ancient Georgian language, but on diachronic as well. Given the fact that the element shows up in all three person forms, it would be better to detect the Kh-//H- prefix, not on a morphological, but phonetical level. In the Khanmet and Haemet texts the Kh-//H- prefix are used with several functions: Markers of S₂ and O₃ (dative); Adjective and adverb comparative quality forms of the name; S₁ as the element of Khu-//Hu- allomorph. We believe that the usage of Kh-//H- prefix developed on the historical and phonetic basis. Its equipment with Morphological features and its understanding as a morpheme seems a secondary phenomenon. The arising of the Kh-//H- element can be related to aspiration, which is characteristic to common-Kartvelian. "The historical evidence of Kartvelian languages shows that charging aspirate is a characteristic feature of phonetic style of Kartvelian languages. This is evidenced, on the one hand, the old Georgian data and, on the other hand, Kartvelian languages, and live data of dialects" [Jgenti, 1965:112]. This should first have occurred in Megrelian and Lazi, then - Georgian and at last in Svanuri. This explains the impersonality of S₂ and O₃ in Megrelian and Lazi Languages.

If asking the issue is right than in Svanuri there should be found the historic aspiration related to the signs of S₂ and O_3.  Truly, in Svanuri the revealing of Kh-prefix (whether it is the segment of Kh-prefix of S₁ or S₂ and O₃ allomorph) it is pholiphonically stipulated: First of all, it is realized with verbs without prefix, anlaut in previous positions of vowel (# - V), with the stems beginning with consonants Kh-prefix is not found (ხუ-არი „ვარ", ხ-ოსყი „ვუკეთებ", ხ-არი „ხარ", ხ-აჰუდი „აძლევს"... შდრ. ტ-უ-იხე „ვაბრუნებ", ტიხე „აბრუნებ", ათ-ბ-ე „მიაბა"...).[9] Generally, Svanuri treats the morphems differently, Kh-prefix easily sequences with zero that cannot be said about other morphems. We can conclude that in this case, Kh-prefix is not lost, and works in accordance with phonological position.

In the further development of Svan aspirations, during the pronunciation of H spirant ჰ > ხ, მაგ., ჰარაყ //ხარაყ, ჰასაკ//ხასაკ and other, it is seen that in the language collective it is possible for forms of H- and Kh-prefix to co-exist together. If we put the issue this way than we can assume that Khanmet-Haemet is nothing else but a different norm of literary pronunciation [Jgenti, 1965:107]. In south-east Georgia H- was pronounced deeply and therefore it was written as Kh-, whereas in the south-west it was weaker and therefore was written with H-.

To conclude, the revealed Kh-//H-prefix of the I-prefix Passive Voice does not have a morphological function on the level of Ancient Georgian Language. It is a functionless element. (Conditionally, Kh-//H-prefix can be regarded as a consisting part of S₁ allomorph: Khu-//Hu-) Historically, this prefix (H- > Kh-) seems to be developed on the basis of aspiration.

[1] The Oxford Fragment was published by Ivane Javakhishvili [Javakhishvili, 1996:304-305], and the Cambridge one -   by A. Shanidze [Shanidze, 1937:29-42].

[2] H-prefix is found in two cases

[3] According to the publication of A. Shanidze [Shanidze, 1937].

[4] According to the publication of A. Shanidze [Shanidze, 1944].

[5] According to the publication of A. Shanidze [Shanidze, 1977].

[6] According to the publication of L. Kajaia [Kajaia, 1984]

[7] According to the publication of A. Shanidze [Shanidze, 1927].

[8] According to the publication of A. Shanidze [Shanidze, 1923].

[9] The information about prefixes can be seen:  [Tsikhelashvili, 2005:275-276].