Abstract
The animate accusative singulars of ky ‘this’, ai ‘that, he’, and kush ‘who’ are këtë, atë, and kë in Tosk dialects, whereas Geg vacillates between final -ë and nasal -ã, of which the latter is limited to argument position. Geg dialects furthermore distinguish kã ‘whom?’ from kãn ‘someone’. This article surveys the dialectal evidence from modern dialects, sketches the Old Albanian situation, and discusses the linguistic history of the relevant forms.
1 Introduction
Tosk and Geg dialects partly present different forms of the animate accusative singular of ky ‘this’ and ai ‘that, he’, and of interrogative and indefinite kush ‘who’, ‘someone’. In Tosk, the only attested forms at all stages of the language are këtë, atë, and kë. In Geg, many modern dialects employ one variant in -ë and another in the nasal vowel -ã for the demonstratives, with a syntactic restriction on the occurrence of the demonstrative accusatives to argument position. Furthermore, many Geg dialects contrast interrogative kã ‘whom?’ with indefinite kãn ‘someone, anyone’. In Old Albanian, vacillation between demonstrative -ë and -ã is found in Bogdani’s Cuneus Prophetarum from 1685, whereas the earlier authors, Buzuku, Budi, and Bardhi, do not show the nasal ending. The aim of this article is to give a survey of the dialectal evidence from modern dialects, to sketch the Old Albanian situation, and to discuss the linguistic history of the relevant forms.
2 Modern dialects
The geographical distribution of the variants of the accusative singular of demonstrative and interrogative pronouns is not treated in the Albanian dialect atlas ADGjSh. I have collected the relevant remarks given in the different dialect descriptions available to me, mostly from the Republic of Albania. Most of the published descriptions have used the same questionnaire and show a similar presentation of the material. Not all of them provide information on the accusative singular forms that interest us, and especially the accusative of the interrogative is only rarely discussed by the authors. As will emerge from the discussion below, the syntactic distinction between adnominal (modifier) and independent (head) usage of the pronouns, as in ‘those children’ vs. ‘I’m talking about those’, is often reflected by two different forms of the pronoun.
It is well known that Albanian dialects are classified along the binary distinction between Tosk (in the south) and Geg dialects (in the north). Tosk dialects lack phonemic nasal vowels.1 The stressed vowel phonemes of Tosk-based Modern Standard Albanian are /i/, /y/, /u/, /e/, /o/, /ə/ and /a/, which may also be found in unstressed syllables. Many of the modern Tosk dialects present similar systems, though part of them lack the front rounded /y/, and another part possesses a length distinction in part of the vowel system. On the contrary, many Geg dialects possess a length distinction in the vowels and feature a series of nasal vowels beside the oral ones. Furthermore, they may show other features, such as the vowels /æ/ or /ø/ (Matzinger 2006: 7–13; Beci 2016; ADGjSh).
2.1 The Tosk evidence
In Arvanitic, /kətə/ ‘this’ and /atə/ ‘that’ occur with two accentual variants, barytone /ˈkətə/ and /ˈatə/ beside oxytone /kəˈtə/ and /aˈtə/ (Sasse 1991: 139–140). Similarly, the Mandrica dialect in Bulgaria (Sokolova 1983: 84–85) distinguishes between /ˈkət/, /ˈat/, which are said to be used in adnominal position and have lost their final schwa there, and /kəˈtə/, /aˈtə/. In this dialect, there is no semantic difference between the two pronouns. In Tomorricë, a northern Tosk variety spoken to the southeast of Mount Tomorr, Xhaferri (1990: 349) observes a tendency—to which some exceptions exist, as the author duly notes—to use /ˈkət/ in adnominal versus /ˈktə/ in independent position, which matches the distribution found in Geg dialects.
The dialect descriptions of other Tosk dialects, such as Gjirokastër (Karagjozi Kore 2014: 146), Kurvelesh (Totoni 1971: 68), Çamic (Haxhihasani 1971: 174), Strelcë (Petriti 1987: 227), and Konispol (Muça 1987: 309), do not provide any data on the accusative singular of the demonstratives.
The accusative ‘whom?’ is attested as /ˈkə/ in all Tosk dialects for which we have data, viz. Arvanitic (Sasse 1991: 145), Konispol (Muça 1987: 309), Çamic (Haxhihasani 1971: 176) and Tomorricë (Xhaferri 1990: 349; here also for the indefinite function).
2.2 Southern Geg
In Weigand’s 1913 grammar, which was based on the Elbasan dialect, the accusatives of the demonstratives are /aˈte/ and /kəˈte/, with an oral, stressed /e/ (Weigand 1913: 56, 67). The carrier sentences provided by Weigand do not show a formal difference according to syntactic usage.
In Shpat (south of Elbasan), the accusative of the near-deictic pronoun in adnominal position is /ˈkət/ for both genders, e.g., ne kët zakon kishim ‘we had this law’, sot m’kët mal ‘today on this mountain’, whereas nasalized /ˈktã/ appears in independent position, also for both genders: e mora ktã ‘I took this one’. The accusatives of ai, ajo ‘that’ are /ˈtã/ and /ˈte/, of which the first is said to be more frequently used; both are given with independent examples only (Çeliku 2020: 159–160).
In Dumre, southwest of Elbasan, a gender difference is reported by Çeliku (2020: 54). In the masculine, /ˈkət/ and /ˈat/ are used for adnominal forms, whereas /ˈktã/ and “rarely” /atə/ (stress not indicated) surface in independent position. In the feminine, the accusative forms are /ˈkte/ and /ate/ (stress not indicated, but surely word-final) both adnominally and independently, e.g., kishim kte pun ‘we had this job’, s’e tha kte ‘he didn’t say this’.
In Krahinë e Peqinit, to the west of Elbasan, the demonstrative accusative is /aˈte/ for both genders. For the interrogative, we find /ˈkət/ in adnominal use versus /ˈkte/ in independent use for both genders, but in the northern, mountainous part of the area, it can also be realized as /ˈktã/ (Çeliku 2020: 235–236).
In Kërrabë, to the northwest of Elbasan, the adnominal variants are /ˈkət/ (more rarely /ˈket/) and /ˈat/, versus independent /ˈktã/, more rarely /ˈkte/, and /ˈtã/, also /aˈte/ (Çeliku 2020: 504).
In Polis, to the east of Elbasan, ‘this’ has the forms adnominal /ˈkət/ versus independent /ˈkta/, whereas ‘that’ has the accusatives /aˈtã/, /aˈtẽ/, apparently (but not explicitly noted) for both animate genders (Haxhihasani 1987: 108–109).
The dialect descriptions for this region give no data on the accusative ‘whom’, except for Krahinë e Peqinit, which has /ˈtʃin/ (Çeliku 2020: 238).
For 19th-century southern Geg, the existence of the nasalized accusatives is confirmed by Kristoforidhi (1872), who prints the accusatives as ⟨ke̥tẽ̆⟩, ⟨ketẽ̆⟩ for /kəˈtẽ/, ⟨atẽ̆⟩ for /aˈtẽ/ (e.g., in 2Petr 3:1, 1 Cor 10:3, 2 Cor 5:14, Rom 6:2), and uses them in adnominal and independent position alike. The spelling ⟨ẽ̆⟩ indicates a short nasalized vowel. The accusative ‘whom’ is attested in ⟨me kẽ̆⟩ /ˈme ˈkẽ/ ‘with whom?’.
Taking stock, Southern Geg has the nasal variants /-ã/ and /-ẽ/, and furthermore a variant in oral /-e/, which is either less frequent than /-ã/ (in Shpat and Kërrabë), used specifically in the feminine (Dumre), or generally used (Elbasan according to Weigand, Krahinë e Peqinit). I interpret the two variants of the nasal vowel as the result of a raising from /-ã/ to /-ẽ/ as is also found in Northwestern Geg (see below). This idea is supported by the fact that the northern and southern fringes of Southern Geg mainly display ã-vocalism. The oral variant /-e/ is difficult to explain by denasalization, which is not otherwise found in these dialects and which would furthermore not explain the gender distribution in Dumre. Therefore /-e/ may have been introduced by analogy with another pronoun, in particular, with the accusative se ‘what?’, ‘which’ of the OAlb. pronoun qish, found in Buzuku and Budi, e.g., Buz. fol. 79v45 [R254.45] /ëndë se ini pagëzuom?/ ‘Wherein are you baptized?’ (lit. ‘In what?’), fol. 81v52 [R262.52] /muor shtratë e tī ëndë se dergjē/ ‘he took his bed, in which he lay ill’, Budi /ndë se ini pshtiellë?/ ‘What are you involved in?’ (RR 344.11), /gishtënitë me se preku shintinë sakrament/ ‘the fingers with which he touched the holy sacrament’ (RR 86.12). The model for the analogy may have been the occurrence of both se and the independent demonstrative accusatives after prepositions (me se, me atã >> me se, me ate).
2.3 South-Central Geg
Along the Durrës-Tiranë axis, the same distribution between adnominal and independent forms of the demonstrative is found as in southern Geg. In Kavajë, /ˈkət/ (also /ˈket/) occurs in adnominal position, whereas non-nasalized /ˈkte/ is found in independent position (Çeliku 2020: 299). Exactly the same the opposition is reported for Durrës (Çeliku 2020: 389) and for Ndroq and Pezë, southwest of Tiranë in the Erzen valley (Çeliku 2020: 435). No information is available on the accusative of the interrogative pronoun.
2.4 Central Geg
These dialects also distinguish between adnominal and independent forms, the latter usually with a nasal vowel. Luzni, in the center of the province of Dibër, has nasalized /ˈktõ/, /aˈtõ/ (also /ˈtõ/) when the accusative is in argument position but also when it is a postposed determiner, e.g., me ktõ ‘with this’, n’ livadh atõ ‘in that meadow’. In prenominal position, Luzni has /ˈkət/ (also /ˈkit/) and /ˈat/: kët venen e kishim ‘we had this wine’, m’ kët kohe ‘in this time’, me at kõm ‘with that leg’ (Beci 2007: 276).
In Grykë e Madhe, in the same province, the masculine accusatives singular are /ˈkət/, /ˈat/ beside /ˈkta/, /aˈta/, and the feminine forms /ˈat/ beside /aˈto/. No nasalization is indicated on the final vowels of /ˈkta/, /aˈta/, /aˈto/ (Beci 2007: 312).
In Arrën, in the southwest of the province of Kukës, the reported forms are adnominal /ˈket/, /ˈkət/ and /ˈat/ versus independent /ˈktã/ (Hoxha 1990: 122–123). The same variants are given for Vilë and Kalis, in the far south of the same province (Hoxha 1975: 159).
In Pukë, the demonstrative accusatives are /ˈket/ and /ˈat/ in adnominal position versus /atã/ and the compound demonstratives /ɲetã/, /tɕetã/ (which contain the deictic prefixes nj- /ɲ-/ and ć- /tɕe-/; stress placement not indicated) in independent use, and plain /tã/ after the prepositions me, më, në (Topalli 1974: 308–309, 336–339).2
In Mirditë, adnominal ‘that’ is cited as /ˈat/, as in at dimen ‘that winter’, at vjet ‘in that year’, whereas the independent forms for the two pronouns are /ktɒ̃/ (/ktõ/), /atɒ̃/ (/atõ/) and uncompounded /tõ/, /n tõ/ (Beci 1982: 73, Beci 2007: 83; accent not indicated). In Mat, the distribution is merely said to be “similar to the other Central Geg dialects” (Beci 2007: 155).
The accusative ‘whom’ is only discussed in the descriptions of Vilë-e-Kalisi and Arrën, the relevant forms being /ˈkãn/ in the former (Hoxha 1975: 160) and /ˈkã/ in the latter (Hoxha 1990: 124).
2.5 Northeastern Geg
For this large dialect area, which comprises northeastern Albania and Kosovo, our data is relatively scarce but points in the same direction. In the Has region, north of Kukës, the adnominal forms of ‘that’ are /tɕat/ (ćat shpell ‘that cave’, ćat djal ‘that boy’), with prefixation of an element ća-, whereas the independent forms are /ktã/ ‘this’ and /tɕatã/, /tɕetã/, after prepositions /tã/ ‘that’ (Gosturani 1975: 218, 232; accent not indicated).
In Çerem and Valbonë, in the extreme northwest of Kukës, adnominal /cet/, /cat/ ‘this’, /at/, /ɲat/ ‘that’ (e.g., at tok ‘that earth’, qat dru ‘this tree’) contrast with independent /jetã/, /ɲetã/, /ɲatã/ (Gosturani 1982: 300–301).
In Kosovë, the forms m.f. /atɒ̃/ ‘him, her’, and m. /tʃetɒ̃/, f. /tʃatɒ̃/ ‘this one’ are reported for Opojë regardless of their syntactic position (Pajaziti 2005: 143–144), but in the sample texts, the adnominal accusative of the near-deictic demonstrative is found as /tʃat/.
The accusative ‘whom’ is given as invariant /ˈkã/ in Çerem and Valbonë (Gosturani 1982: 300–301), Opojë (Pajaziti 2005: 153), and Ragam (Gosturani 1990: 251). A semantic distinction between interrogative /kɒ̃/ and indefinite /kɒ̃n/ is reported for Has (Gosturani 1975: 233)
2.6 Northwestern Geg
Turning to Northwestern Geg, in Kelmend (in the north of Shkodër province) the adnominal forms /kət/, /kt/, in younger generations /ˈkyt/, contrast with independent /ktã/ (reinforced /cetã/, /jektã/) and /atã/ (Shkurtaj 1975: 47).
In Hot, situated between Kelmend and Lake Shkodër, the independent forms are /ktã/, reinforced in /cektã/, /jektã/, and /atã/ (Shkurtaj 1974: 379).
In Rranxë, 10 km south of Shkodër, adnominal /kət/, /kt/, and /at/ (e.g., at djalin ‘that boy’, at vajzën ‘that girl’, kët trimin ‘this hero’, kët gruon ‘this woman’) contrast with independent /ktã/, /jetã/, /ɲãtã/, /atã/. The latter forms also occur as /ktɛ̃/, /jetɛ̃/, /jatɛ̃/, /ɲatɛ̃/, /atɛ̃/, with raising of the nasal which Shkurtaj ascribes to the influence of Shkodër speech (Shkurtaj 1982: 210–211). Unfortunately, the dialect description of the city of Shkodër (Beci 1995) does not provide any data on the form of the accusative of the demonstratives.
The dialect of Breg i Bunës, southwest of Shkodër, also shows raising of nasal /ã/ to /ɛ̃/. The pronouns ky and kjo have the accusative /kɛt/ in all positions, but the accusative of ai and ajo is /at/ in adnominal position versus /atɛ̃/ or /tɛ̃/ in independent use, e.g., edhe atę̃ e mora ‘I also took that one’ (Gjinari 1971: 337, 349).
In Dushman, ca. 30 km northeast of Shkodër to the north of the river Drin, Cimochowski (1951: 86–91) finds the prenominal accusatives /kyt/ (sometimes /kət/) and /at/ for both genders. In independent position, the corresponding forms are /ktɛ̃/ and /aˈtɛ̃/, /tɛ̃/, /aˈtã/, /tã/, the variants in /-ã/ being usual only in Malagji.
The accusative ‘whom’ shows the same vowels as the independent demonstratives, but often with a distinction between the interrogative and indefinite forms. In Hot (Shkurtaj 1974: 380) and Kelmend (Shkurtaj 1975: 47–48), the relevant forms are /kã/ ‘who?’ versus /kãn/ ‘someone’, and nearly the same are found in Rranxë e Mbishkodrës (Shkurtaj 1982: 212), viz. /kã/, /kɛ̃/ ‘whom’ vs. /kãn/ ‘someone’. In Dushman /kã/ and /kɛ̃/ co-occur for the interrogative (Cimochowski 1951: 91), Breg i Bunës only has the raised nasal vowel in /kɛ̃/ (Gjinari 1971: 349), whereas for Shkrel only the indefinite form /kurken/, accusative of kurkush ‘someone’ is given (Beci 1971: 293).
2.7 Summary
In Modern Tosk, the situation is similar to that of the standard language: demonstrative /kətə/ and /atə/ are used without regard to their syntactic position, while the accusative of the interrogative is always /kə/.
In Geg and in one northern Tosk dialect, the form of the demonstrative depends on its syntactic status: 1. The adnominal forms are monosyllabic /kət/, /kt/ and /at/ in all Geg dialect areas; 2. The independent forms vary between /ktã/ and /aˈtã/, /tã/ (the vowel being rounded in northeastern Geg), with phonetic raising to /ktę̃/ and /atę̃/ around Shkodër and occasionally in Southern Geg, in particular, in and near Elbasan. The ending -e found in some southern and south-central Geg dialects may have been adopted from the pronoun se ‘what?’, ‘which’ after prepositions governing the accusative.
The accusative of the interrogative can be /kã/ or /kãn/ in Geg, and /kę̃/ in northwestern Geg. As far as information is available about the difference between interrogative and indefinite usage—which is mainly the case for northwestern Geg—, these variants obey a functional distribution: interrogative /kã/ contrasts with indefinite /kãn/.
3 The Old Albanian evidence
The textual evidence3 for Old Tosk and Old Geg is very uneven: we only have a short Old Tosk text from the end of the 16th century before more abundant transmission sets in in the middle of the 18th century, whereas for Old Geg, we are better off with the printed editions of Buzuku’s missal (1555), Budi’s three books (1618–1621), Bardhi’s Latin-Albanian dictionary (1635) and Bogdani’s Cuneus Prophetarum (1685), followed by further texts in the 18th century. Although the Old Geg authors by and large display the same grammatical and lexical system, we must keep in mind that they stem from different parts of the Geg area (northwestern Geg for Buzuku, northeastern Geg for Bogdani, central Geg for Budi), and span a period of 130 years, so that there may have been dialectal differences, especially in their phonology, which may not always show up in writing.
In the discussion of individual word forms, I will cite the evidence from Old Albanian in the spelling of the original editions, followed by a broad phonemic transcription along the lines of Schumacher & Matzinger (2013: 281–282), who use a system based on the orthography of Modern Standard Albanian. I use the following symbols that deviate from IPA usage: /c/ for IPA /ts/, /ç/ for /tʃ/, /dh/ for /ð/, /ë/ for /ə/, /gj/ for /ɟ/, /ll/ for /ɫ/ or /lɣ/, /nj/ for /ɲ/, /q/ for /c/, /r/ for /ɾ/, /rr/ for /r/, /sh/ for /ʃ/, /th/ for /
3.1 Old Tosk
Old Tosk had only oral vowels. However, it did possess the long, stressed schwa phoneme /ə:/, which may correspond etymologically to Old Geg word-final /-ā̃/ and /-ē̃/, as in Old Tosk (Matranga) ⟨zææ⟩ /zə̄/ beside Old Geg (Buzuku) ⟨Ԑaa⟩ /zā̃/ ‘voice’, and Old Tosk (Variboba) ⟨zɒɒ⟩ /zə̄/ beside Old Geg (Budi) ⟨
Matranga’s text has only a single relevant demonstrative form, the accusative ⟨ctæ mbsuame⟩ = /ktë mbsuame/ ‘this doctrine’. Variboba (1762, Life of Saint Mary) has the accusatives /këtë/, /ktë/, and /atë/ for both genders.
The accusative of interrogative ‘who?’ is ⟨cǽ⟩ /kë/ in Matranga (fol. 21r, 22r, 25r; Sciambra 1964: 38, 40, 56). The same form is found in Variboba, viz. in /me kë/ ‘with whom?’, ‘with what?’ (lines 2686, 3540, 4101; Belmonte 2005).
3.2 Old Geg
In order to arrive at a reliable interpretation of the Old Geg data, we must first establish whether the authors distinguished in their orthography between word-final oral and nasal vowels, and between short and long ones. The earliest authors, Buzuku, Budi, and Bardhi, do not explicitly spell the nasal feature of nasal vowels, even though we can be certain that it was present in their speech (Topalli 1996: 26–30). Therefore, when confronted with a word spelled ⟨baa⟩ in Buzuku and Budi, we cannot decide between the interpretations /bā/ and /bā̃/ without additional information of a text-internal or comparative nature. When these authors do spell a word-final -n, we can be sure that the word ended in consonantal n.
Bogdani is the first Old Albanian author who indicates nasal vowels in a systematic, though not always consistent way. Some of his spelling practices have been studied in more detail by Omari (2005: 21–22, 50–55). Word-final nasal vowels are not always marked, but they are in most cases, viz. by means of -n plus an apostrophe after the last vowel, in spellings such as ⟨atan’⟩ for /atã/ and ⟨hijn’⟩ for /hī̃/. It seems likely that Bogdani added the apostrophe to distinguish ⟨atan’⟩ = /atã/ from words ending in an oral vowel + n, such as ⟨ban⟩ = /ban/ ‘does’. It can hardly be a coincidence that 3sg. /ban/ ‘does’ is 67 times spelled ⟨ban⟩ but only twice ⟨ban’⟩.
Table 1
The spelling of different word-final sequences by the main Old Albanian authors
|
Buzuku |
Budi |
Bardhi |
Bogdani |
|
|---|---|---|---|---|
|
/-ā/ |
-aa |
-aa |
-aa |
-aa |
|
/-ā̃/ |
-aa |
-aa |
-aa |
-aan’ |
|
/-an/ |
-an |
-an |
-an |
-an |
|
/-ān/ |
-aan |
-aan |
-aan |
-aan |
|
[-ã] |
n.a. |
n.a. |
n.a. |
-an’ |
Omari consistently interprets final ⟨-an’⟩ as short /-ã/ and final ⟨-aan’⟩ as long /-ā̃/, but it remains to be investigated whether this length variation was phonemic. Compare Table 1, which contains the main Old Geg spellings of the word-final sequences -an, -ān,4 /-ā̃/ and -ã.
Another relevant issue is the spelling of word-final /-ë/ in Old Geg. Buzuku, Budi, and Bardhi do not distinguish final (nor word-internal) /ë/ from final /e/. In Buzuku, the consonant ⟨-h⟩ is often added to word-final ⟨-e⟩, and while the combination ⟨-eh⟩ often renders /-ë/, this is not consistently the case (Fiedler 2004: 16–17), so that we cannot rely on Buzuku’s spelling to establish the distinction. Budi indiscriminately uses ⟨e⟩ to render /ë/ and /e/ (Svane 1980: 127), and so does Bardhi in his dictionary. In his Cuneus Prophetarum, Bogdani introduces a graphic distinction between ⟨e⟩ and ⟨è⟩, with the probable aim of using the second variant mainly for /ë/; but he applies this distinction inconsistently, and there are many cases of ⟨è⟩ for /e/ and of ⟨e⟩ for /ë/. Hence, all we can say is that Bogdani’s spelling ⟨è⟩ raises the chances that we are dealing with /ë/, while conversely ⟨e⟩ may point to /e/, but we cannot rely on the graphic distinction alone (Omari 2005: 38–43).
3.2.1 Buzuku
In Buzuku, the counterparts of MoAlb. këtë and atë are spelled ⟨cheteh⟩, ⟨chete⟩ and ⟨ateh⟩, ⟨ate⟩. There is no difference between adnominal and independent usage of the demonstrative accusatives: compare Buz. ⟨endeh chete puſſ⟩ /ëndë këtë pus/ ‘into this well’ (fol. 49r34, R136.34; Gen 37:22 Lat. in cisternam hanc), ⟨ban chete caffse⟩ /ban këtë kafshë/ ‘do this thing!’ (fol. 38v41, R94.41; Matt 8:9 Lat. fac hoc), and ⟨per enbii cheteh isne cherubinah⟩ /përëmbī këtë ishnë kerubina/ ‘over it were the cherubins’ (fol. 90r19, R296.19; Heb 9:5, Lat. superque eam cherubin). The pronominal accusatives can hardly be interpreted with a word-final nasal vowel, since all other nasal vowels are spelled with a vowel digraph, e.g., ⟨maa⟩, ⟨maah⟩ /mā̃/ ‘more’, ⟨ђaa⟩ /gjā̃/ ‘thing’, ⟨asђaa⟩ /asgjā̃/ ‘nothing’, ⟨liђee⟩ /liqē̃/ ‘den’, ⟨□u⟩ /vū̃/ ‘hunger’ (cf. Hock 2006), and others.
Nevertheless, it must be pointed out that on one occasion Buzuku spells the word for ‘thing’ as ⟨ђe⟩, a reduced variant of ⟨ђaa⟩, which is generally interpreted as /gjë/ but might in theory also have been /gjẽ/ with a nasal vowel, viz. in ⟨A mūde dalle cun ђe teh mireh en Naʒaretit?⟩ /A mundë dalë kun gjë të mirë ën Naxaretit?/ ‘Can anything good come from Nazareth?’ (Latin A Nazareth potest aliquid boni esse?, John 1:45; f. 93r12–13, R308.12–13). A similar uncertainty holds for Buzuku’s two instances of a reduced variant of the 3sg. copula ‘is’, spelled ⟨eh⟩,5 which is usually transcribed as /ë/ but might also be /ẽ/. It must be a reduced variant of /ã/ ‘is’ as found in Budi (spelled ⟨a⟩) and in modern northern Geg dialects, cf. Hock (2005).
The accusative of /kush/ ‘who’ (n=9) is spelled ⟨che⟩ and ⟨cheh⟩, which is usually interpreted as /kë/, though /kẽ/ and /ke/ are also conceivable. With the meaning ‘whom’ it is attested six times, as an interrogative pronoun in /Kë vetëvetëhenë ban/ ‘Whom do you make yourself?’ (fol. 59v75, R178.75; John 8:53 Lat. quem te ipsum facis?), and /Kë lypëni/ ‘Whom do you (pl.) seek?’ (fol. 68r53 and 68r58, R212.53 and R212.58; John 18:4 and 18:7, Lat. quem quaeritis?); and as a headless relative pronoun in /përëmbī kë të shofsh/ ‘he on whom you will see’ (fol. 37v78, R90.78; John 1:33 Lat. super quem videris), /kë të ënthernjënë/ ‘whom they will pierce’, (fol. 69r89, R216.89; John 19:37 Lat. in quem transfixerunt), /kë të mundë përpie/ ‘whom he can devour’ (fol. 83v66, R270.66; 1 Pet 5:8 Lat. quem devoret).
With the indefinite meaning ‘some’, translating Latin alium, kë is found three times in a single passage: /e kë ënvranë, e kë sosnë ën kreje, e kë ënvranë me gurë/ ‘and one of them they killed, another they finished off facing him, and another they killed with stones’ (f. 49r52–53, R136.52–53; Matt 21:35 Lat. alium ceciderunt alium occiderunt alium vero lapidaverunt).
For the indefinite accusative ‘someone’, ‘anyone’, also in the compound /gjithëkush/ ‘everyone; whoever’, Buzuku always has /kana/ (18×), spelled ⟨cana⟩, ⟨canah⟩, ⟨chana⟩ and ⟨chanah⟩.
3.2.2 Budi
In Budi’s works, the accusatives of the demonstratives are spelled ⟨kette⟩, ⟨kete⟩, ⟨atte⟩, and ⟨ate⟩, in both syntactic positions. Svane (1980: 129) interprets the final vowel as /-ë/, among other reasons because of the apocope in SC 308.11 ⟨mbatt cias⟩ /mb at ças/ ‘at that time’. Adnominal examples include /me këtë natyrë tanë/ ‘with this nature of ours’ (DC 10.23), /nktheva këtë shinte Doktrinë/ ‘I translated this holy Doctrine’ (DC 2.15); an independent example is /e këtë e thosh/ ‘and this he said’ (RR 328.8).
The monosyllabic accusative ‘whom’ occurs as ⟨ca⟩ and ⟨ke⟩. While ⟨ca⟩ may in theory render /ka/ or /kã/, the former is unlikely in the absence of any supporting evidence from other varieties of Albanian, so we may assume that Budi intended /kã/; for the spelling, compare Budi’s reduced form of the 3sg. copula ⟨a⟩ /ã/ ‘is’. This variant ⟨ca⟩ occurs as a headless relative in ⟨me ca⟩ ‘with whom’ (twice, in DC 39.11 and SC 2.21) and in indefinite /kã tjetërë/ ‘someone else’ (RR 196.26). In most cases, however, the indefinite accusative ‘someone’, ‘anyone’ has the form ⟨canaay⟩ /kanāj/ in Budi, also in /gjithëkanāj/ ‘everyone’.
The variant ⟨ke⟩ can in theory be read as /ke/, /kë/, or /kẽ/; As a headless relative, ⟨ke⟩ occurs at least five times, viz. in SC 30.6 and 30.7 ⟨gi
3.2.3 Bardhi
In Bardhi’s dictionary, the demonstrative accusatives are spelled ⟨chete⟩ and ⟨ate⟩: këtë ditionār ‘this dictionary’, këtë librëzë ‘this booklet’, and, in the prose of his introduction, këtë shtãsë ‘this animal’, nd atë vend ‘there, in that place’, nd atë ças ‘in that moment’, once in independent position me atë kini me qienë matunë ‘by that [measure] you will be measured’ (Mk 4:24). The accusative of kush is unattested.
3.2.4 Bogdani
Before discussing Bogdani’s accusatives, we must take a closer look at his spelling habits. As discussed above in 3.2, Bogdani spells ⟨-aan’⟩ for word-final /-ā̃/, but he also uses ⟨-an’⟩, probably for word-final [-ã]. In order to assess the value of the various spellings, table 2 compares the spellings for a number of words with a final vowel that is nasalized in at least part of Bogdani’s tokens. I divide the evidence into four categories:
-
Words generally written with a long vowel in the earlier Old Geg authors, where they ended in a long nasal vowel /-ā̃/, viz. the equivalents of MoAlb. u bë ‘it was done’, zë ‘voice’, and gjë ‘thing’.
-
The word gjë ‘thing’ in the compounds asgjë ‘nothing’, asgjëmangu ‘nonetheless’, and gjëkafshë ‘something’.
-
Pronominal accusatives ending in a short vowel (mostly ⟨-e⟩, sometimes ⟨-a⟩) in Buzuku and Budi, viz. the equivalents of MoAlb. ’të and atë ‘that’, këtë ‘this’, and kë ‘whom?’. The ending is usually interpreted as OAlb. /-ë/, but there is room for discussion. A word-final nasal vowel can be reconstructed for Proto-Albanian and is confirmed by the Modern Albanian dialectal forms discussed in § 2.
-
The counterparts of the MoAlb. degree and quantifier adverbs aq ‘so, so much’ and kaq ‘this much’. These are spelled with final ⟨-e(h)⟩ in Buzuku, rendering /-ë/, /-e/ or /-ẽ/, with final ⟨-e⟩ or ⟨-ij⟩ by Budi, and with final ⟨-e⟩ by Bardhi.6
Çabej (1976: 58–59) explained aq from the prefixation of stressed deictic a- (as in ai ‘he’, ashtu ‘thus’) to the first element qye- of OAlb. qyetë ‘so much, so many’ (Budi, Bogdani). This would have caused the reduction of (the preform of) qye to -që in Old Albanian times. Kaq is explained from the prefixation of near-deictic *kë- to aq. Topalli (1995: 409, 2017: 110) has added that the diphthong /ye/ of qye- is best explained from an earlier sequence *-ēn, which leads him to reconstruct *qēn > *qye, to which suffixal -të was added. Leaving aside speculations as to the exact origin of qye (Topalli reconstructs a PIE pronominal form “accusative *kwōm” which has no place in current PIE reconstruction), his explanation of qye- is attractive. It would presuppose the erstwhile presence of a nasal vowel in the second syllable of MoAlb. aq and kaq, which finds some confirmation in Bogdani’s spellings ⟨achian’⟩ and ⟨kachian’⟩, even if these are in the distinct minority (6 % for /aqã/, 3 % for /kaqã/; see Table 2, nr. 4).
Table 2
The spelling of words having ⟨-an’⟩ and corresponding to MoAlb. -ë or zero in Bogdani’s Cuneus Prophetarum. All spellings are diplomatic renderings of those found in the printed edition
|
MoAlb. equivalent: |
Nasal-final spellings |
|||||
|---|---|---|---|---|---|---|
|
Single vowel |
Vowel digraph |
Nasalless variants |
||||
|
1. MoAlb. stressed -ë |
||||||
|
u bë (n=31) |
ban’ |
4 |
baan’ baan |
7 2 |
ba baa |
3 15 |
|
zë (n=35) |
|
6 |
|
15 |
|
14 |
|
gjë (n=26) |
gian’ |
10 |
giaan’ giaan |
3 1 |
gia giaa |
7 5 |
|
2. MoAlb. compounds with gjë |
||||||
|
asgjë (n=10) |
as gian’ asgian’ asgian |
2 1 1 |
as giaan’ |
1 |
asgia asgiaa |
2 3 |
|
asgjëmangu (n=65) |
as gian’ m. as gian m. asgian’ m. |
8 1 1 |
– |
as gia m. as già m. asgia m. as giaa m. asgiaa m. |
35 7 10 2 1 |
|
|
gjëkafshë (n=20) |
gian’ c. |
3 |
giaan c. |
1 |
gia/già giaa |
13 3 |
|
3. MoAlb. accusatives in -ë |
||||||
|
’të (n=80) |
tan’ tàn’ tan |
71 7 2 |
– |
– |
||
|
kë (n=13) |
|
11 |
– |
|
1 1 |
|
|
këtë (n=205) |
|
16 |
– |
|
36 5 147 1 |
|
|
atë (n=250) |
atan’, atàn’ atan |
44 1 |
– |
atè ate |
167 38 |
|
|
4. MoAlb. adverbs aq, kaq |
||||||
|
aq (n=316) |
achian’ |
20 |
– |
achia achi |
294 2 |
|
|
kaq (n=97) |
|
3 |
– |
|
87 1 5 1 |
|
Abbreviations: c. = ⟨cafsc-⟩, m. = ⟨mangu⟩. Numbers represent tokens
The data from table 2 allow for the following observations:
-
In group 1, the spelling ⟨-aan’⟩ accounts for 29 % (u bë), 43 % (zë), and 15 % (gjë) of the tokens for each word. This is in all cases a minority. Sometimes ⟨-an’⟩ outweighs ⟨-aan’⟩, viz. for gjë, and sometimes the nasalless forms are more numerous that the nasal-final ones taken together, viz. for u bë. The reason for the nasalless spellings is unclear: are they due to the tradition of nasalless spelling found in earlier authors, or do they reflect a phonetic change?
-
In group 2, gjë is only once spelled with ⟨-aan’⟩, but 17× with short ⟨-an’⟩ (n=95). The nasalless spellings make up 50 % of the tokens for asgjë, 85 % for asgjëmangu, and 80 % for gjëkafshë. The strong decrease of nasal spellings in these compounds may be due to an actual phonetic change, such as denasalization of /gjā̃/ before the initial nasal of mangu, the (presumed) unstressed position of /gjā̃/ in the syntagm, or the inconsistent spelling of nasal vowels by the author.
-
In groups 3 and 4, the spellings ⟨-aa⟩, ⟨-aan⟩ and ⟨-aan’⟩ are absent. Hence these words certainly do not have long final vowels.
-
In group 3, final ⟨-an⟩ makes up 100 % of the tokens for accusative të and 84 % of those for kë (n=13). Thus the nasal vowel was well preserved in these forms, on which see further below.
-
Nasal-final spellings make up 18 % of the tokens for atë, 8 % of those for këtë, 6 % of those for aq, and 3 % of those for kaq. I have found no syntactic difference between the nasalless and nasalized spellings (which does not exclude the possibility that such a difference existed). The low percentages of final ⟨-n’⟩ suggest that, unlike in të and kë, denasalization had progressed considerably in these words. Since aq and kaq had word-initial stress (whence the loss of the second syllable in MoAlb.), and atë and këtë have initial stress at least in attributive use, the unstressed status of the second syllable may have been the cause for the loss of nasalization.
-
In the accusatives atë and këtë, the nasalless forms always end in ⟨-e, -è⟩. On the contrary, for aq and kaq the nasalless forms mainly end in ⟨-a⟩. If the OAlb. ending of these four words stems from the same Proto-Albanian nasal-final origin (which remains hypothetical), a possible reason for the differential treatment in Bogdani may lie in the (effect of the) palatal consonant /q/ on the following vowel. Note that Budi also treats the endings of aq and kaq differently, with frequent spellings ⟨-ij⟩ (for /-ī/) beside usual ⟨-e⟩, which remain to be explained.
All of the preceding leads me to the following interpretation of the pronominal accusatives in Bogdani. The full forms of the demonstrative accusatives mostly end in ⟨-è⟩, less frequently ⟨-e⟩; probably they render /atë/ and /këtë/. They compete with the minority spellings in ⟨-an’⟩, which I interpret as /atã/ and /këtã/ with a short nasal vowel. The forms in -ë and in -ã both occur as modifiers and in independent use. An anonymous reviewer suggests that there may have been a pragmatic reason for the vacillation between the two variants, the forms in -ã carrying more phrasal emphasis. A first, superficial check did not confirm this intuition, but it would require a more elaborate investigation to gain clarity on this matter.
A different approach is advocated by Omari (2005: 40), who regards Bogdani’s nasal-final spellings of the pronominal accusatives as “nën dikimin e së folmes së vendlindjes”, that is, “under the influence of his native dialect.” This seems to imply that Bogdani’s usual spelling ⟨-e⟩, ⟨-è⟩ is due to conscious imitation of the existing Old Geg spelling tradition, while occasionally his native dialect had the upper hand, causing the nasal-final spellings. This would still leave unexplained why Bogdani without exception spells ⟨-an’⟩ in the truncated accusative të (n=80), where traditionally ⟨te(h)⟩ is used by the earlier authors.
Bogdani is the first author in whose work the syntactic restriction of the nasalized variant surfaces that we know from modern dialects, but only in the truncated equivalent of MoAlb. atë. The spellings ⟨tan’⟩, ⟨tàn’⟩, and ⟨tan⟩ for /tã/ are only found as a prepositional object, e.g., /gjithë sa u kish thanë Filomeni për tã/ ‘everything Philomen had told them about him’ (1.28.41), /kjenë ngjallunë me tã/ ‘they had resurrected with him’ (1.76.16).
For interrogative and headless-relative ‘whom’, most forms (n=11) show ⟨
The indefinite accusative ‘someone, anyone’ is always kand in Bogdani, spelled ⟨
3.2.5 Kuvendi
Kuvendi i Arbënit (1706) has the acc.sg. forms m. and f. at, atë (spelled ⟨ate⟩) and m. këtë, këta, f. këtë, këta (Demiraj 2012: 321). Nasalization is not indicated by the author of the text (Demiraj 2012: 44–45, 52), but we may assume that ⟨keta⟩ stands for /këtã/ in these animate accusatives, otherwise the form would be identical to the neuter pronoun. The form këta is found in adnominal position, e.g., ndë këta kohë ‘in this time’. The accusative of kush is ⟨kand⟩ (4×, always indefinite).
3.2.6 Other authors
Da Lecce’s grammar (1716) provides the acc.sg.m. and f. ⟨kètè⟩ and ⟨atè⟩ = /këtë/, /atë/. In Kazazi’s Doktrina (1743), the acc.sg.m. and f. is only attested as /këtë/, spelled in various ways. All nine instances are in adnominal position (Demiraj 2006: 446).
4 Internal reconstruction
4.1 The accusative of the demonstratives and of the interrogative ‘who’
The Modern Geg dialectal distribution, with adnominal forms in -ë versus independent forms in -ã, represents a further grammaticalization of the distribution in Bogdani, where only tã is obligatory in argument position, whereas there is vacillation between këtã, atã and këtë, atë in the same position. The syntactically governed competition in Geg between nasalized k(ë)tã, (a)tā and non-nasalized k(ë)të, (a)të cannot be explained on the basis of synchronic morphemes or from analogy after a synchronic model. Since the demonstratives carry more syntactic stress in argument position than when used adnominally, their nasalization may be due to the retention of nasalization under stress, whereas the nasal vowels were reduced to schwa in unstressed position. In other words, the Geg opposition between (-)tã and (-)të reflects earlier *tã́ (stressed) versus *tã (unstressed). Just how far back this phonetic differentiation goes is impossible to say. The synchronic cooccurrence in Bogdani of the adverbs /ˈaqã/, /ˈkaqã/ beside /ˈaqa/, /ˈkaqa/ (in Budi /ˈaqë/, /ˈaqī/) suggests that the posttonic reduction may be a recent phenomenon, at least in Bogdani’s dialect.
We have seen a northwestern Geg tendency to raise final -ã to -ẽ. It also existed in southern Geg, where it is attested in the 19th century in Kristophoridi. The presence of this raised variant in and around Shkodër, and in influential works such as the grammar by Pekmezi (1908: 130, 145–146) and Cimochowski’s description of the Dushman dialect, has led a number of scholars to take -ẽ as the more widespread variant of this ending, but this can be dismissed on the basis of the data from other Geg dialects and from Old Geg. Centralization, which may surface by the raising of low vowels and the lowering of high vowels, is a common change in nasal vowels; see, e.g., Beddor, Krakow, & Goldstein (1986) and Carignan (2018). No raising is observed in the stressed lexemes in long -ā̃, such as the indefinite nominative of ‘thing’ and ‘voice’ and the non-active aorist ‘it was done’.
The Old Geg textual situation remains surprising. Why did Buzuku and Budi (and Bardhi, but we have very few independent forms in his book) spell the demonstrative accusatives only with final ⟨-eh⟩ and ⟨-e⟩, whereas most of modern Geg shows a syntactically motivated distribution of the endings? The nasalized endings cannot have arisen out of the blue in the middle of the 17th century since there was no analogical model by which they could have been formed. Nor does it seem plausible that the earliest Old Geg authors, with their different dialectal backgrounds, had all lost the nasalized variant from their idiolect. A possible answer to our question may be that final ⟨-e⟩, ⟨-eh⟩ in Buzuku and Budi sometimes stands for a nasal vowel /-ẽ/. This would be a relatively unproblematic assumption for Buzuku, for whom a northwestern Geg origin is supposed anyway (Çabej 2013 I: 17–22). For Budi’s Central Geg dialect, however, a value /-ẽ/ would be more surprising.
4.2 The accusative of the indefinite pronoun ‘someone’
Whereas the oldest Geg authors have interrogative and relative Buzuku kë or kẽ, Budi kã and ke, kë, or kẽ, the indefinite accusative variant has a-vocalism plus a nasal extension: kana in Buzuku and kanāj < *kanaj in Budi, who lengthens all vowels before word-final j. Budi also uses kujnāj in the genitive-dative. Since the genitive-dative Buzuku kuj (Budi kūj) consists of the stem ku plus what is synchronically a genitive-dative ending -j, suffixal -na+j probably arose in the (genitive and) dative. I conclude that in Budi’s dialect, the accusative *kana was remade as *kanaj under the influence of the genitive-dative form.
As to its own history, Buzuku’s kana must first of all be compared with the dative-ablative kujnaj (found from Buzuku onwards), which shows that the element -na is common to all non-nominative indefinite forms. In modern dialects, suffixal -a and -na may strengthen the deictic character of pronouns in a number of dialects (e.g., in Tosk, Kurvelesh ky ‘this’ beside the near-deixis variant kja, kyja, Totoni 1971: 68; in Geg, Has has the genitive-dative atij, ktij beside (a)tijna and ktijna, Gosturani 1975: 23), but these are not easily connected with Old Albanian material. For the moment, then, I will leave this matter unexplored.
If kana resulted from the suffixation of -(n)a to the original base, we can reconstruct an accusative *kan which regularly yielded the Old Albanian accusative forms. The difference between Old Geg kë and Bogdani’s kã may then be explained from the opposition between Proto-Albanian unstressed *kan and stressed *kán, similar to the syntactically governed variants të and tã in Bogdani and in Modern Geg.
Since Old Albanian tand ‘your’ has become tãn in (part of the) modern Geg dialects, modern Geg indefinite kãn has probably developed from Old Geg kand. The latter form may in theory be analyzed as ka+nd or as kan+d. Whereas there is no likely parallel for the addition of an element -d (see below on the accusative tand ‘your’), the addition of -nd is also found in the indefinite adverb kund ‘perhaps, somewhere, somehow, sometime’, which consists of the interrogative base ku and an element -nd.
Kund is attested from Budi onwards. In Bogdani, besides being an indefinite adverb, kund serves as an indefinitizer in askund ‘anywhere’ and gjithëkund ‘everywhere’. In Tosk, kund is attested starting from Variboba (18th century), who spells kund’ (in the preface, and in the poem in lines 1773 and 4165).
Instead of kund, Buzuku prints the forms kun and kon, the vacillation being due to the lowering of original u to o. The word functions as a dubitative particle ‘perhaps’ and as an indefinitizer ‘some, any’ accompanying a noun in the nominative or accusative. Since Buzuku precedes Budi by two to three generations, we might conjecture that the earlier form in Albanian was kun, which was later remade into kund. Yet this would imply a twofold, independent replacement of kun by kund, viz. in (Northern) Geg between ca. 1555 and 1620, as well as in pre-1750 Tosk. In theory such a coincidence cannot be excluded, but in the present case it seems implausible because both the motive for replacing kun and the model for the addition of -d remain unclear. Topalli (2017 s.v. kund) explains -d from the accusative pronoun -të, but this cannot be correct: kun is not restricted to the accusative, and the final -d in the possessive form tand ‘your’ (acc.), which Topalli invokes as a model, does not stem from deictic *-të but from the unstressed personal pronoun in t- ‘you’ (sg.). For the compositional nature of the possessives, compare Bogdani’s acc.sg.m.f. tãj ‘your’ (pl.) from *tan + *j- ‘you’ (pl.).
I therefore favour the opposite scenario, viz. that the earlier form was kund, which was phonetically reduced to kun in Buzuku’s dialect. Such a view is supported by the variation between kun and kon in Buzuku. As noted by Ashta (1996: 193), u and o frequently interchange in prenasal position in the same word in Buzuku’s text. An exhaustive search yields the following numbers:8
Table 3
Attestations with ⟨u⟩ and with ⟨o⟩ in Buzuku of words containing original -umb-, -und-, -ung-, plus klumështë. The o-variants of /ˈkundər/ ‘against’ are /ˈkondra/, /ˈkondrə/. See Appendix 1 for the exact distribution of the u- and o-spellings in these words
|
Tokens u |
Tokens o |
Meaning |
|
|---|---|---|---|
|
before m |
|||
|
/ˈkluməʃtə/ |
0 |
4 |
‘milk’ |
|
before mb |
|||
|
/bumbuˈɫimə/ |
0 |
2 |
‘thunder’ |
|
/pəˈɫumb/, /poˈɫumb/ |
12 |
0 |
‘pigeon’ |
|
/ˈ(h)umb/ |
1 |
0 |
‘lose’ |
|
before nd |
|||
|
/ˈfund/, /ˈfunt/ |
1 |
22 |
‘end, bottom’ |
|
/funˈdon/ |
0 |
1 |
‘found’ |
|
/ˈkun/ |
25 |
10 |
‘any, somewhere’ |
|
/ˈkundər/ |
0 |
100+ |
‘against’ |
|
/ləˈkund/ |
5 |
3 |
‘move’ |
|
/ˈmundə/, /ˈmundən/ |
200+ |
1 |
‘be able’ |
|
/munˈdon, -im, -uos/ |
53 |
6 |
‘hit, torture’ |
|
/ˈʃkund/ |
0 |
1 |
‘shake’ |
|
/səˈmund-/ |
23 |
0 |
‘illness’ |
|
before ng |
|||
|
/kunˈgon/ |
0 |
3 |
‘participate’ |
Although the evidence seems to be mixed at first sight, we can detect some important tendencies. The only case of vowel lowering in an open syllable is klomështë ‘milk’, which may be contrasted with unchanged u in gjumë ‘sleep’, shumë ‘much’ and shumon ‘to increase’, lumi and plural lumëna ‘river’. I have no explanation for klomështë and can only guess that the lowering rule only sporadically applied before single m.
All other words that lower u to o do so before a cluster of nasal plus obstruent, i.e., mb, nd, ng. The lowering is general or preponderant in the case of bumbullimë, fund, kundër, kungon, and shkund; it is in the minority in kun, lëkund, and mund-, and absent in pëllumb ‘pigeon’ and in the hapax umb ‘lose’.
I have no explanation for pëllumb and umb other than to suggest than the lowering was phonetically less prone to happen before the bilabial nasal.
The nearly pervasive u in mundë ‘be able’ and its derivatives mundën, mundon, and sëmund- may be explained by relative chronology. Schumacher (2005; see also the summary in Schumacher & Matzinger 2013: 983–985) has shown that mundë ‘be able’ derives from earlier 3sg. *mun plus the subjunctive particle *të. Its aorist mũjti also contains a present stem *mun-. Hence the preponderance of ⟨mund-⟩ in Buzuku may be due to an earlier date of the lowering than the univerbation *mun *të > *mundë and/or to analogical restoration of the u in mund- on the model of the aorist mũjt-.
Kun is the only word with lowering of u that ends in n rather than in a nasal cluster, and, moreover, it is the only word ending in /-un/ at all. If kun originally ended in -nd, its lowering to *kond would allow us to simplify the sound change u > o in Buzuku’s dialect, viz. as occurring before clusters of the type nC and (only in klomështë) before m.
There is an additional argument in favour of reconstructing *kund. The reduction kund > kun would make most phonetic sense before obstruents, with d being lost in a consonant cluster. Indeed, of the 35 instances of kun and kon in Buzuku, 14 tokens are followed by gjā and gjë ‘thing, something’; furthermore they are followed three times by të and six times by another obstruent (tuk, tjetërë, këshill, pak, kafshë).9 The combination kun/kon gjā is a tight-knit semantic unity that is lexified with the meaning ‘anything’, as it is also in Budi, where the combination kund gjākafshë (8×) is relatively frequent.10 Therefore, the combination *kund *gjā may have been the decisive environment for the reduction of kund, viz. of ndgj to ngj, yielding kun gjā. In view of the close (semantic, and therefore possibly also phonological) unity of the combination *kund gjā, we cannot tell whether the lowering *kund > *kond preceded the reduction nd > n, or whether the loss of d happened first, followed by lowering of u to o before ngj. The d-less form kun/kon may then have spread to positions where it did not stand before an obstruent.
If this explanation for Buzuku’s kun is correct, kund was originally pan-Albanian and is indirectly attested 130 years before the indefinite accusative kand in Bogdani. Since kund and kand are semantically so close, and since no other source for the d in kand is in sight, it seems reasonable to surmise that the earlier Geg accusative form *kan, *kã, or *kana—we do not know which one was the exact input form—was remade into kand on the model of kund.
5 Comparative reconstruction
Internal reconstruction presupposes the Proto-Albanian preforms *këtan ‘this’, *(a)tan ‘that’, and *kan ‘whom’ for the accusatives in syntactically stressed -ã and unstressed -ë. It is generally accepted that these forms directly reflect the PIE monosyllabic accusatives m. *tom, f. *tām, and interrogative *kwom as per Pedersen (1900: 314); after the change of short *o to Pre-Proto-Albanian *a, these would have become m. *tam, *kwam. Pedersen adduces the masculine accusative singular of possessive ‘your’ (sg.), Tosk tënd, Geg tãnd, from *tam + *t- ‘you’ (sg.), as additional proof for the reconstruction of *tam. Because of the existence of nasalized endings in Bogdani and in Geg dialects, Jokl (1916: 98–99, 114) assumes that nasalization was preserved in stressed word-final syllables in Geg into the historical period.
Pedersen and Jokl are followed by most subsequent scholars, such as Sh. Demiraj (1986: 466, 502), Topalli (1996: 27–28, 59–61), and Schumacher & Matzinger (2013: 223). Some authors, such as Klingenschmitt (2022: 23) and Matzinger (2006: 110), also adopt Jokl’s conclusion that the Geg pronominal forms in -ẽ are older than those in -ã, but this has been dismissed by Sh. Demiraj (1986: 502), who calls Geg kẽ, atẽ, këtẽ “relatively recent.” He leaves them unexplained, but, as I have argued above, the change may be interpreted as phonetic raising or centralization of the nasal vowel.
The length difference between Bogdani’s këtã, atã, tã, kã, on the one hand, and the long vowel in gjā̃, zā̃, u bā̃, on the other, is surprising, since the same Proto-Albanian ending is reconstructed for both groups of forms. Whereas the pronouns presuppose *tán and *kán, final *-an is also reconstructed for monosyllabic nouns ending in a long nasal vowel such as gjā̃ ‘thing’ < PAlb. *ɟan < Pre-PAlb. *źán- (for which different PIE etymologies have been proposed; see Neri [2023] in DPEWA s.v. gjë) and zā̃ ‘voice’ < PAlb. *zan < Pre-PAlb. *dzwána- < PIE *ǵhu̯onos (Neri [2020] in DPEWA s.v. zë, -ri). It seems likely that the resulting shortness or length of the word-final nasal vowel was linked to morphological features, as suggested by Schumacher & Matzinger (2013: 212). Thus, the long vowel of gjā̃, zā̃, u bā̃ may be due to analogy, that is, to the influence of long vowels in the polysyllabic forms of the same paradigms (e.g., the nom.sg.def. *zā̃ni ‘voice’), whereas such supporting forms were absent in the pronominal paradigms.
The exact Proto-Albanian etymology of kund remains uncertain. In view of ku ‘where?’, kur ‘when?’, and kurrë ‘never’, we may analyse kund as consisting of ku and an element -nd. We can only speculate about the origin of -nd. It can go back to Latin or Pre-Proto-Albanian *nt or *nd, Latin nn (as in Alb. pendë ‘feather’), or to a sequence *-nVt/d- with pre-rhotacism syncope of the internal vowel (as in Alb. shëndet ‘health’ < Vulgar Latin sanitā́te). The element *-nV- could then contain PIE *ne(i) ‘not’, since negation often induces indefinite semantics; note that Pedersen (1900: 317) already proposed to explain kurrë ‘never’ from *kur-në from *kur ‘when’ plus *ne ‘not’. For final -d there are so many possible sources (PIE *-dV as in Latin quandō, inde; PIE -dhV; any form of the PIE demonstrative *to-; a truncated form of the verb do ‘want’) that I will not pursue the matter here.
6 Conclusion
Both Old and Modern Tosk dialects use demonstrative këtë and atë without regard to their syntactic position, while the accusative of the interrogative is always kë. In Geg and in one northern Tosk dialect, the adnominal forms of the demonstrative are kët, kt, and at, whereas the independent forms are ktã and atã, tã in many dialects, with raising to ktę̃ and atę̃ around Shkodër and Elbasan. In parts of south and south-central Geg, the ending is -e, which may be due to analogy with se ‘what? which’. Furthermore, in modern Geg, interrogative ‘whom’ has the shape kã (and kę̃ in northwestern Geg), whereas indefinite ‘whom’ is kãn.
The three earliest Old Geg authors have no clearly nasalized forms of the demonstratives under investigation. But Bogdani has këtë/këtã and atë/atã without syntactic motivation, and exclusively nasalized tã after prepositions. The interrogative accusative ‘whom’ surfaces as kë, ke or kẽ in Buzuku, as kã and (probably) ke in Budi, and mainly as kã in Bogdani. The indefinite accusative vacillates between kë, ke or kẽ and kana in Buzuku and kã and kanāj in Budi, whereas it is always kand in Bogdani.
Etymologically, the accusatives of ky ‘this’, ai ‘that’, and kush ‘who?’ go back to Proto-Albanian *këtan, *atan, *kan. When used with syntactic stress in argument position, *këtán, *atán, *kán retained their nasality in Geg, whereas the unstressed variants developed into këtë, atë, and kë in both Tosk and Geg.
The indefinite accusatives, Bogdani kand, whence Modern Geg kãn, and Buzuku kana, the basis for Budi’s kanāj, are due to the addition of one or more phonemes to the earlier accusative *kã or *kan. Old Geg kand probably arose by analogy with the indefinite adverb kund ‘any, anywhere’, which is found in Geg and Tosk but was reduced to kun, kon in Buzuku’s idiolect for phonetic reasons. Old Geg kana displays the addition of deictic -a or -na to the original accusative.
Acknowledgments
For comments on a first version of this paper I am indebted to Sergio Neri (Basel University) and Rexhep Ismajli (Prishtina University). The second version has profited from many improvements suggested by two anonymous reviewers. The usual disclaimers apply.
A.1 Appendix 1: Register of occurrences belonging to table 3
/ˈkluməʃtə/: 46r28 [R124,28] ⟨clomeſte⟩, 71v51 [R226,51] ⟨ clomeſte⟩, 76r78 [R240,78] ⟨clomeſtite⟩, 82r49 [R264,49] ⟨clomeſte⟩
/bumbuˈɫimə/: 80r12 [R256,12] ⟨Bonbulime⟩, 82r46 [R264,46] ⟨bonbulimeh⟩
/pəˈɫumb/, /poˈɫumb/: 37v75 [R90,75] ⟨polumb⟩, 44r74 [R116,74] ⟨polunbate⟩, 55v78 [R162,78] ⟨polūbateh⟩, 55v83 [R162,83] ⟨polunbateh⟩, 97r48 [R324,48] ⟨polūbi⟩, 102v3 [R346,3] ⟨polunbate⟩; 70v82 [R222,82] ⟨pelumneh⟩, 70v84 [R222,84] ⟨pelunbi⟩, 70v90 [R222,90] ⟨pelumne⟩, ⟨pelunbi⟩, 71r4 [R224,4] ⟨pelumne⟩, 94v9 [R314,9] ⟨pelunbas⟩
/ ˈ(h)umb/: 24r70 [R64,70] ⟨ȣ vnbb⟩
/ˈfund/, /ˈfunt/: 31r46 [R80,46] ⟨fundit⟩, 43r9 [R112,9] ⟨fōteh⟩, 47r88 [R128,88] ⟨font⟩, 47r90 [R128,90] ⟨font⟩, 47v34 [R130,34] ⟨fonte⟩, 75v77 [R238,77] ⟨font⟩, 88r68 [R288,68] ⟨font⟩, 92r39 [R304,39] ⟨fonteh⟩, 92r41 [R304,41] ⟨fonte⟩, 100r31 [R336,31] ⟨font⟩, 108v65 [R370,65] ⟨font⟩, 108v65 [R370.65] ⟨font⟩, 60r25 [R180,25] ⟨fond⟩, f65r47 ⟨per fondih⟩, 69v23 [R218,23] ⟨fondit⟩, 71v7 [R226,7] ⟨fond⟩, 84r74 [R272,74] ⟨fond⟩, 91r8 [R300,8] ⟨fond⟩, 97v11 [R326,11] ⟨fond⟩, 102r8 [R344,8] ⟨fond⟩, 110v21 [R378,21] ⟨fond⟩, 99r64 [332,64] ⟨fondenateh⟩, 103r54 [R348,54] ⟨fondenate⟩
/funˈdon/: 103r52 [R348,52] ⟨fondoȣ⟩
/ˈkun/: ⟨con⟩ 21v34 [R54,34], 42v82 [R110,82], 44v70 [R118,70], 46r11 [R124,11], 55v4 [R162,4], 70v84 [R222,84], 75r7 [R236,7], 75v88 [R238,88], 79r59 [R252,59]; ⟨cō⟩ 75r70 [R236,70]; ⟨cun⟩ 21r45 [R52,45], 22r37 [R56,37], 22v27 [R58,27], 22v45 [R58,45], 23r19 [R60,19], 23v51 [R62,51], 49r28 [R136,28], 51r56 [R144,56], 60v68 [R182,68], 61r38 [R184,38], 68r31 [R212,31], 70v57 [R222,57], 84v30 [R274,30], 84v54 [R274,54], 87v70 [R286,70], 88r63 [R288,63], 93r12 [R308,12], 93v76 [R310,76], 96r76 [R320,76], 106r52 [R360,52], 110v58 [R378,58], ⟨ђiѯe cun⟩ 79r66 [R252,66], 81r90 [R260,90], ⟨cūteh⟩ 93r17 [R308,17], ⟨cū⟩ 095r39 [R316,39]
/ləˈkund/: 30r33 [R76,33] ⟨lecundune⟩, 30r34 [R76,34] ⟨lecūdete⟩, 30v31 [R78,31] ⟨lecondune⟩, 45v75 [R122,75] ⟨lecundee⟩, 65v17 [R202,17] ⟨lecūdune⟩, 72v1 [R230,1] ⟨lecondune⟩, 76v8 [242,8] ⟨lecōdgnene⟩, 92v32 [R306,32] ⟨lecūdeneh⟩
/ˈmundə/, /ˈmundən/ with ⟨o⟩: 64r85 [R196,85] ⟨monde⟩
/munˈdon, -im, -uos/ with ⟨o⟩: 46v89 [R126,89] ⟨mondognene⟩, 85v73 [R278,73] ⟨mondonemih⟩, 93v63 [R310,63] ⟨monduoſah⟩, 103v42 [R350,42] ⟨mondimete⟩, 110v52 [R378,52] ⟨monduoneh⟩ and ⟨ȣ mondogneneh⟩.
/ˈʃkund/: 81r86 [R260,86] ⟨Scondeni⟩
/səˈmund-/ 23× ⟨mund⟩
/kunˈgon/: 22r50 [R56,50] ⟨meȣ conguom⟩, 55r13 [R160,13] ⟨conguō⟩, 86r78 [R280,78] ⟨cōguomite⟩
I refrain from giving the 100+ instances of /ˈkundər/ ‘against’, which always have ⟨o⟩.
A.2 Appendix 2: Data on the words following kon and kun in Buzuku
10× ⟨con⟩:
/kon gjā̃/ 42v82, 55v4, 75r70, and 79r59 (R110.82, 162.4, 236.70, 252.59);
/kon gjā̃kafshë/ 46r12 and 75r7 (R124.12, 236.7),
/kon një/ 21v34 and 44v70 (R54.34, 117.80),
/kon tuk/ 70v84 (R222.84),
/kē shpūm kon/ 75v88 (R238.88).
25× ⟨cun⟩:
/kun e/ + adj. 84v30 (R274.30);
/kun ëndonjë/ + subst. 51r56 (R144.56),
/kun ëndë ëndonjë/ + subst. 84v54, 87v70, and 106r52 (R274.54, 286.70, 360.52),
/kun ëmbë ëndonjë/ + subst. 110v59 (R378.59);
/kun gjā̃/ 61r38 (R184.38),
/kun gjā̃kafshë/ 23v51, 60v68, 68r31, 70v57, 95r39, 96r76 (R62.51, 162.68, 212.31, 222.57, 316.39, 320.76),
/kun gjë/ 93r12 (R308.12);
/kun kafshë/ 49r28 (R136.28);
/kun kshīll/ 22v45 (R58.45);
/kun një/ 22r37 and 88r63 (R56.37, 288.63);
/kun pak/ 23r19 (R60.19);
/kun të/ + adj. 93r17 and 93v76 (R308.19, 310.76);
/kun tjetërë/ + subst. 21r45 and 22v27 (R52.45, 58.27);
/gjithë kun/ 79r66 and 81r90 (R252.66, 260.90).
The dialect descriptions are not uniform in their rendering of nasal vowels. Often, a circumflex accent is used: â, ê, î, û, ŷ. Here I will transcribe all nasal vowels with a tilde, thus ã, ẽ, ĩ, ũ, ỹ.
No nasalization is indicated for /ta/, /ata/, /tçeta/ on p. 309, but the occurrence of nasalized /tã/ and /ktɒ̃/ in the sample texts on pp. 336–339 shows that this must be a printing error.
For Gjon Buzuku’s Missale from 1555, I have used the diplomatic online edition prepared in 2000–2002 by Wolfgang Hock for the TITUS website,
In Buzuku, word-final long vowel plus -n occurs in the 3sg.pres. ind. /bān/ ‘holds’ (to the stem bart-), /fshān/ ‘sighs’, /hīn/ ‘enters’, /pëgān/ ‘soils’, /shān/ ‘despises’, and /vgjān/ ‘finds’.
Viz. fol. 74r20 [R232.20] ⟨seh chetȣ⟩ /s’ë këtu/ ‘he’s not here’ (Mk 16.6 non est hic); fol. 9v90 [R6.09] ⟨Cuſeh cheio⟩ /Kush ë këjo/ ‘who’s she?’ (Catena legionis Quae est ista?).
Viz. Buz. ⟨aђe⟩ and ⟨aђeh⟩ for /aqë/ (n=21) and ⟨caђe⟩, ⟨chaђe⟩, ⟨chaђeh⟩ for /kaqë/ (n=35); Budi ⟨achie⟩ (n=887), ⟨achij⟩ (n=177) and ⟨cachie⟩ (n=56), ⟨kachij⟩ (n=6); Bardhi ⟨achie⟩, ⟨aghie⟩ (n=4) and ⟨kachie⟩, ⟨kaachie⟩ (n=5). Note the long vowel /ā/ in some of Bardhi’s forms of kaq.
The printed text has ⟨
I exclude Buz. portekë ‘stick’ (13× with o, never with u), since it represents a loanword from SCr. prùtak which has undergone r-metathesis in part of Geg, cf. Budi purtekë, but not in Bogdani’s prutekë. It is not certain, therefore, that Buzuku’s portekë is due to exactly the same kind of lowering as before nasals.
More elaborate data are given in Appendix 2.
Kund occurs dozens of times in Budi’s works. The only combinations which are more frequent than kund gjākafshë are kund ndōnjë/a/ī ‘any’ (more than 50×, in subject or object function) and those with the prepositions ndë or mbë, often followed by ndōnjë or një (in prepositional object function).
References
ADGjSh = Jorgji Gjinari, Bahri Beci, Gjovalin Shkurtaj, & Xheladin Gosturani in collaboration with Anastas Dodi & Menella Totoni. 2007–2008. Atlasi Dialektologjik i Gjuhës Shqipe. 2 vols. Tiranë: Akademia e Shkencave e Shqipërisë, Instituti i gjuhësisë dhe i letërsisë / Napoli: Università degli Studi di Napoli l’Orientale, Dipartimento di Studi dell’Europa Orientale.
AShSh = Akademia e Shkencave e RP(S) të Shqipërisë (until 1990) / Akademia e Shkencave e Republikës së Shqipërisë / Akademia e Shkencave e Shqipërisë.
AShAK = Akademia e Shkencave dhe e Arteve e Kosovës.
Ashta, Kolë. 1996. Leksiku historik i gjuhës shqipe I: Tri dokumentet e para të gjuhës shqipe e leksiku përkatës (gjysma e dytë e shekullit XV). Gjon Buzuku e leksiku i plotë, nxjerrë nga vepra e tij “Meshari” (1555). Shkodër: Universiteti i Shkodrës.
Beci, Bahri. 1971. E folmja e Shkrelit. In: Dialektologjia shqiptare I. Tiranë: Universiteti i Tiranës, 268–330.
Beci, Bahri. 1982. E folmja e Mirditës. In: Dialektologjia shqiptare IV. Tiranë: AShSh, 26–143.
Beci, Bahri. 1995. Të folmet veriperëndimore të shqipes dhe sistemi fonetik i së folmes së Shkodrës. Tiranë: AShSh.
Beci, Bahri. 2007. Të folmet qëndrore të shqipes së veriut. Tiranë: Toena.
Beci, Bahri. 2016. Struktura dialektore e shqipes. Një qasje e re. Tiranë: Universiteti European i Tiranës.
Beddor, Patrice S., Rena A. Krakow, & Louis M. Goldstein. 1986. Perceptual constraints and phonological change: a study of nasal vowel height. Phonology yearbook 3, 197–217.
Belmonte, Vincenzo (ed.). 2005. Giulio Variboba: Vita della Beata Vergine Maria (1762). Edizione critica e traduzione italiana. Soveria Mannelli: Rubbettino.
Carignan, Christopher. 2018. Using ultrasound and nasalance to separate oral and nasal contributions to formant frequencies of nasalized vowels. Journal of the Acoustical Society of America 143(5), 2588–2601. https://doi.org/10.1121/1.5034760.
Cimochowski, Wacław. 1951. Le dialecte de Dushmani. Description de l’un des parlers de l’Albanie du Nord. Poznań: Poznańskie Towarzystwo Przyjaciół Nauk.
Çabej, Eqrem. 1976. Studime etimologjike në fushë të shqipes ii. Tiranë: AShSh.
Çabej, Eqrem. 2013. Meshari i Gjon Buzukut3. 2 vols. Tiranë: Çabej. [First published 1968, Tiranë: Universiteti Shtetëror i Tiranës, Instituti Historisë e i Gjuhësisë.]
Çeliku, Mehmet. 2020. Të folme kalimtare dhe të folme të gegërishtes jugperëndimore të shqipes. Tiranë: Akademia e Studimeve Albanologjike.
Da Lecce, Francesco Maria. 1716. Osservazioni grammaticali nella lingua albanese. Rome: Sacra Congregazione di Propaganda Fede.
Demiraj, Bardhyl. 2006. Gjon P. Nikollë Kazazi dhe “Doktrina” e tij. Prishtinë: AShAK.
Demiraj, Bardhyl. 2008. Dictionarium latino-epiroticum. Per R.D. Franciscum Blanchum. Shkodër: Botime Françeskane.
Demiraj, Bardhyl. 2012. Conci
λ i Provintiaaλ i o Cuvendi i Arbenit (Romæ 1706). Botim kritik. Shkodër: Botime Françeskane.Demiraj, Bardhyl. 2024. Reprint of Svane 1986a. Tiranë: AShSh.
Demiraj, Bardhyl, Anila Omari, & Enkelejda Kapia (ed.). 2022. Reprint of Svane 1985. Tiranë: AShSh.
Demiraj, Shaban. 1986. Gramatikë historike e gjuhës shqipe. Tiranë: Fakulteti i Historisë dhe i Filologjisë.
DPEWA = = Digitales Philologisch-Etymologisches Wörterbuch des Altalbanischen. https://www.dpwa.gwi.uni-muenchen.de/
Fiedler, Wilfried. 2004. Das albanische Verbalsystem in der Sprache des Gjon Buzuku (1555). Prishtina: AShAK.
Gosturani, Xheladin. 1975. E folmja e Hasit. In: Dialektologjia shqiptare III. Tiranë: AShSh, 211–269.
Gosturani, Xheladin. 1982. E folmja e Çeremit dhe e Valbonës (rrethi i Tropojës). In: Dialektologjia shqiptare IV. Tiranë: AShSh, 279–344.
Gosturani, Xheladin. 1990. E folmja e Ragamit. In: Dialektologjia shqiptare VI. Tiranë: AShSh, 230–284.
Gjinari, Jorgji. 1971. Vëzhgime mbi të folmen e Bregut të Bunës. In: Dialektologjia shqiptare I. Tiranë: Universiteti i Tiranës, 331–370.
Haxhihasani, Qemal. 1971. Vështrim i përgjithshëm mbi të folmen e banorëve të Çamerisë. In: Dialektologjia shqiptare I. Tiranë: Universiteti i Tiranës, 118–193.
Haxhihasani, Qemal. 1987. Vëzhgime rreth të folmes së Polisit. In: Dialektologjia shqiptare V. Tiranë: AShSh, 93–152.
Hock, Wolfgang. 2005. Zur Nasalität bei Buzuku: I. Der Vokalismus von anʃteh ‘ist’. In: Monica Genesin & Joachim Matzinger (ed.), Albanologische und balkanologische Studien. Festschrift für Wilfried Fiedler. Hamburg: Kovač, 45–55.
Hock, Wolfgang. 2006. Anlautendes *u̯ im Albanischen. Die Sprache 46, 65–80.
Hoxha, Shefqet. 1975. Vëzhgime mbi të folmen e Vilë-e-Kalisit të Lumës. In: Dialektologjia shqiptare III. Tiranë: AShSh, 130–210.
Hoxha, Shefqet. 1990. E folmja e Arrnit. In: Dialektologjia shqiptare VI. Tiranë: AShSh, 57–188.
Ismajli, Rexhep. 2006. Pjetër Budi. Poezi. Pristhinë: AShAK.
Jokl, Norbert. 1916. Beiträge zur albanesischen Grammatik. 3. Der Akkusativ-Nominativ und der Geschlechtswechsel im Albanesischen. Indogermanische Forschungen 36, 98–164.
Karagjozi Kore, Mimoza. 2014. Gjirokastra. Vështrim historik, gjuhësor, etnologjik. Tiranë: Dudaj.
Klingenschmitt, Gert. 2022. The Albanian numerals. Mit drei zusätzlichen Beiträgen zum Albanischen aus der Perspektive der Indogermanistik, ed. Bardhyl Demiraj & Stefan Schaffner. Wiesbaden: Harrassowitz.
Kristoforidhi, Konstandin. 1872. Dhjata e re. Constantinople: Boyagian.
Matzinger, Joachim. 2006. Der altalbanische Text [E] Mbsuame e Krështerë (Dottrina Cristiana) des Lekë Matrënga von 1592. Eine Einführung in die albanische Sprachwissenschaft. Dettelbach: Röll.
Muça, Fatmir. 1987. E folmja e krahinës së Konispolit. In: Dialektologjia shqiptare V. Tiranë: AShS, 281–362.
Omari, Anila. 2005. Pjetër Bogdani. Cuneus Prophetarum (Çeta e profetëve). Botim kritik me një studim hyrës, faksimile të origjinalit, transkriptim e shënime. Tiranë: AShSh.
Omari, Anila. 2016. Leksiku i Veprës së Pjetër Bogdanit. Tiranë: Qendra e Studimeve Albanologjike.
Pajaziti, Naser. 2005. E folmja e Opojës. Prishtinë: Instituti Albanologjik i Prishtinës.
Pedersen, Holger. 1900. Die Gutturale im Albanesischen. Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der indogermanischen Sprachen 36, 277–340.
Pekmezi, Georg. 1908. Grammatik der albanesischen Sprache (Laut- und Formenlehre). Vienna: Albanesischer Verein ‘Dija’.
Petriti, Çelik. 1987. Vëzhgime mbi të folmen e Strelcës (krahina e Gorës, rrethi i Korçës). In: Dialektologjia shqiptare V. Tiranë: AShSh, 206–280.
Ressuli, Namik. 2013. Il “Messale” di Giovanni Buzuku2. London: Centre for Albanian Studies. [First published 1958, Città del Vaticano: Biblioteca Apostolica Vaticana.]
Sasse, Hans-Jürgen. 1991. Arvanitika. Die albanischen Sprachreste in Griechenland. Teil 1. Wiesbaden: Harrassowitz.
Schumacher, Stefan. 2005. Albanisch munde /mundë/ ‘können’ und semuhem, -ete /sëmuhem, -etë/ ‘krank werden’. In: Monica Genesin & Joachim Matzinger (ed.), Albanologische und balkanologische Studien. Festschrift für Wilfried Fiedler. Hamburg: Kovač, 151–168.
Schumacher, Stefan, & Joachim Matzinger. 2013. Die Verben des Altalbanischen. Belegwörterbuch, Vorgeschichte und Etymologie. Unter Mitarbeit von Anna-Maria Adaktylos. Wiesbaden: Harrassowitz.
Sciambra, Matteo. 1964. La ‘Dottrina cristiana’ albanese di Luca Matranga. Riproduzione, trascrizione e commento del Codice Barberini Latino 3454. Città del Vaticano: Biblioteca Apostolica Vaticana.
Shkurtaj, Gjovalin. 1974. E folmja e Hotit. In: Dialektologjia shqiptare II. Tiranë: AShSh, 344–428.
Shkurtaj, Gjovalin. 1975. E folmja e Kelmendit. In: Dialektologjia shqiptare III. Tiranë: AShSh, 5–129.
Shkurtaj, Gjovalin. 1982. E folmja e Rranxave të Mbishkodrës. In: Dialektologjia shqiptare IV. Tiranë: AShSh, 144–278.
Sokolova, Bojka. 1983. Die albanische Mundart von Mandrica. Wiesbaden: Harrassowitz.
Svane, Gunnar. 1980. How to read Budi’s “Speculum Confessionis”. Studia Albanica 1, 121–183.
Svane, Gunnar. 1985. Pjetër Budi, Dottrina Christiana (1618). With a transcription into modern orthography and a concordance. Text. Aarhus: Institut for Lingvistik.
Svane, Gunnar. 1986a. Pjetër Budi, Speculum Confessionis (1621). With a transcription into modern orthography and a concordance. Text. Aarhus: Institut for Lingvistik.
Svane, Gunnar. 1986b. Pjetër Budi, Rituale Romanum (1621). With a transcription into modern orthography and a concordance. Text. Aarhus: Institut for Lingvistik.
Topalli, Kolec. 1995. Theksi në gjuhën shqipe. Tiranë: Enciklopedike.
Topalli, Kolec. 1996. Për historinë e hundorësisë së zanoreve në gjuhën shqipe. Tiranë: Enciclopedike.
Topalli, Kolec. 2017. Fjalor etimologjik i gjuhës shqipe. Tiranë: Qendra e studimeve albanologjike.
Topalli, Xhabir. 1974. Disa vërejtje për të folmen e krahinës së Pukës. In: Dialektologjia shqiptare II. Tiranë: AShSh, 282–343.
Totoni, Menella. 1971. Vëzhgime rreth të folmeve të Kurveleshit. In: Dialektologjia shqiptare I. Tiranë: Universiteti i Tiranës, 31–117.
Weigand, Gustav. 1913. Albanesische Grammatik im südgegischen Dialekt (Durazzo, Elbassan, Tirana). Leipzig: Barth.
Xhaferri, Haredin. 1990. E folmja e Tomorricës. In: Dialektologjia shqiptare VI. Tiranë: AShSh, 329–370.
