The phoneme /o/ is often present in the Albanian dialect of Opoja in environments where it is absent elsewhere in Albanian. This paper explains /o/ in Opoja by reference to the Slavic substrate present in that area of Kosovo. Language shift from Slavic to Albanian took place in the late 16th and 17th centuries in Opoja, and I argue that during this process, Slavic /o/ was identified with Albanian /ə/. This identification was facilitated by the fact that the Slavic dialect of Opoja lay directly on a major isogloss of a crucial sound change in Slavic: the loss of hyper-short high vowels, also known as the jer shift. To the south of Opoja, the Slavic dialect of Gora has /o/ and /e/ from these hyper-short high vowels (known as jers in the Slavic literature), but to the north, the dialect of Prizren shifted both vowels to /ə/. This allowed Albanian /ə/ to be identified with the Slavic jer reflexes, which were then imposed on Albanian during the process of language shift.
The Albanian dialect of Opoja, spoken in far southwestern Kosovo between Gora and Sretečka Župa, is unique among Albanian dialects due to the widespread presence of /o/ in certain environments where it is absent in other dialects of Albanian, as described in Pajaziti (2005). After presenting sociohistorical background information about Opoja, I review all of the environments where /o/ occurs uniquely in the Opoja dialect, and argue that the presence of /o/ in Opoja is due to a sound change from /ə/ to /o/. I furthermore argue that this sound change was influenced by the Slavic jer shift, in which hyper-short front and back high vowels, known as jers in the Slavic linguistics literature, were either lost or shifted to full vowels. In particular, I argue that the sound change from /ə/ to /o/ was caused by the imposition of Slavic jer reflexes during the process of language shift from Slavic to Albanian, which likely took place in the late 16th and 17th centuries. Finally, I consider the question of what information can be reconstructed about the now-extinct Slavic dialect of Opoja, and hypothesize that the Slavic dialect of Opoja may have been a one-jer dialect of a hitherto unattested type in Slavic, in which both jers shifted to /o/ - i.e., a unique transition zone between Serbian (in which both jers shifted to /ə/) and Macedonian (in which the back jer shifted to /o/ and the front jer shifted to /e/).
1.1. Sociohistorical background
Opoja (Serbian Opolje) is located in the mountainous, historically somewhat isolated southwesternmost corner of Kosovo known as the Dragash district, which is divided between the region of Gora in the south and the region of Opoja in the north (see figure 1). Due to the cultural and historical developments that have affected the Dragash region as a whole, the sociohistorical background of Opoja must be examined in the light of developments that also affected Gora. Both communities are Muslim, although widespread conversion to Islam in Gora took place significantly more recently than in Opoja. Currently, Slavic and Albanian are spoken in Gora, while only Albanian is spoken in Opoja.1 Aromanian toponyms give some indication of the pre-Slavic linguistic situation in these regions, but in the early Middle Ages, it appears that Slavic was spoken in both regions (Mladenović 2001: 51). Gora and Opoja are both located in high, rugged terrain, somewhat isolated from nearby areas, separated only by the river Plava (Lutovac 1956: 234, Mladenović 2001: 40). Furthermore, cultural ties bind the two regions. Lutovac states that relations between the Albanians of Opoja and the Slavic speakers of Gora have traditionally been good, even to the point of sporadic marriage ties, which is not the case for the relationship between Gorans and other neighboring Albanian groups (Lutovac 1956: 278-280). Given the commonalities between these communities, two major discrepancies between Gora and Opoja must be explained. First, why did Islam spread in Opoja centuries before the full Islamization of Gora? Second, why is Albanian spoken in Opoja while Slavic is spoken in Gora?
Lutovac (1956) argues that the key difference was the presence of a wealthy Turkish landowner named Kukli-Beg in Opoja in the early 16th century. According to sources, Kukli-Beg owned 117 houses in Prizren and property elsewhere in the Ottoman Empire, but, crucially, had only meager possessions in Gora (Lutovac 1956: 259). Lutovac then hypothesizes that the presence of Kukli-Beg resulted in the largely Slavic population partially converting to Islam and partially migrating elsewhere, leaving unoccupied areas into which Albanians emigrated, eventually resulting in language shift. Evidence from toponyms supports this hypothesis (Lutovac 1956: 262-263). Since the Islamization of the Albanian population was not necessarily complete by this time, this account is consistent with Pulaha’s data from the registration of 1591. According to this document, “in the nahija of Opoja there were 369 Muslim heads of households and bachelors [i.e., with names that bear no indication of language - AD] and 78 Christian heads of households and bachelors, the majority of whom have Albanian anthroponyms” (Pulaha 1984: 21-22).
This account is also supported by Lutovac’s (1956: 273) findings that “of 72 lineages and 1032 households, 37 lineages and 578 houses are ‘immigrants’ (doseljenici) and 36 lineages and 455 households are of unknown background.” Of the immigrant lineages, 14 claim decent from the nearby Albanian border regions of Hasi / Podrimlje and Luma, while 18 claim descent from elsewhere in Albania. Notably, the lineages of ‘unknown’ background interviewed in Lutovac (1956) were not aware of belonging to a fis, the Albanian concept of clan organization known in northern Albania and Kosovo. Fis organization does not generally occur in modern-day Opoja, and late 19th-century accounts describe the inhabitants of Opoja as preserving many Serbian customs (Lutovac 1956: 277). This contrasts sharply with contemporary Gora, where newer lineages are comparatively less frequent, comprising only 358 of the 2200 households (Mladenović 2001: 42-43).
The population of Gora converted to Islam much later than that of Opoja. The earliest Goran mosque dates from 1751; other mosques were built in the late 18th and early 19th centuries. Islamization in Gora took place gradually; the last Christian inhabitant of Gora died in 1856, according to some sources, while a 1865 source asserts that “today there are still four brothers in Brod and their sister Mara, and two brothers are of the Turkish faith while the two other brothers and Mara are of the Serbian faith” (Mladenović 2001: 55). The most plausible direct trigger for Islamization seems to have been the neglect of the area by the Greek Orthodox religious hierarchy after the abolition of the Peć Patriarchate (Mladenović 2001: 55, Lutovac 1956: 270-271).
Intercommunal relations significantly worsened in the mid-19th century, due to the emergence of language-based ideologies of nationalism and the erosion of centralized Ottoman power. This seems to have had direct repercussions for the Gora and Opoja regions. Regarding tensions between Gorans and neighboring Albanians, Mladenović states: “In the second half of the 19th century animal husbandry began to decline because the Ottoman Empire was then descending into anarchy and the Albanians from the neighboring region of Luma raided their [the Gorans'] stocks. In Gora, memory of that time is still active” (Mladenović 2001: 41). By 1916 – i.e., after the Balkan Wars and during World War I – a report from the French Ministry of War states: “The [area] of Gora is populated exclusively by Muslims who only speak Serbian and live in constant hostility with the Albanians” (Mladenović 2001: 33). Thus, although Gora and Opoja share significant historical commonalities, they exist today as two sociologically differentiated units.
As mentioned above, the Albanian dialect of Opoja is unique in having the phoneme /o/ present in a wide range of morphological environments in which /o/ does not occur elsewhere in Albanian. Interestingly, the Slavic dialect of Gora is also unique in a Slavic context, in some regards occupying a transitional position between the Serbian Prizren-Morava dialects to its north and the northern Macedonian dialects spoken to the south and east of Gora (Lutovac 1955: 282, Mladenović 2001; see also Friedman 2002/3: 164-165 on the history of the classification of the dialect of Gora). Taken together, then, Opoja and Gora also form a linguistic unit in the sense that the dialects of each region are unique in their broader linguistic context – i.e., Gora is in some ways a unique transitional zone in South Slavic just as Opoja is a unique dialect of Albanian.
2. The phoneme /o/ in Opoja
The status of the phoneme /o/ in Opoja was first described in Pajaziti (2005), but has not yet been analyzed in depth. When /o/ occurs uniquely in Opoja, it often corresponds to /ə/ in other Albanian dialects. This provides the basis for the explanation proposed here. In this section, I review all the environments in which /o/ occurs uniquely in Opoja, dividing them into two groups: one where /o/ corresponds transparently to /ə/ in other Albanian dialects and can thus be explained as the result of a /ə/ > /o/ sound change, and one where there is no such correspondence, but it is nonetheless reasonable to hypothesize an intermediate stage of *ə.
2.1. /o/ corresponding to /ə/ elsewhere in Albanian
A clear example where Opoja has an alternating /o/ that regularly corresponds with standard Albanian /ə/ is found in the declension of feminine nouns. This is illustrated in table 1 with the word motor (standard Albanian motër [motər]) ‘sister’ (Pajaziti 2005: 129). In the GDA (genitive / dative / ablative) cases, /o/ in Opoja appears to correspond with /e/ in standard Albanian. This likely reflects a generalization of definite endings into the indefinite paradigm, as the presence of separate endings in the indefinite paradigm is ancient in Albanian (Demiraj 1988: 274-278).
In table 1, standard Albanian was chosen for comparison because the /ə/ in those examples is found throughout Albanian. The Opoja dialect is unique in not having /ə/ in these paradigms, although /ə/ can be deleted in other Geg dialects when phonotactics permit. In the definite GDA forms of these paradigms, /ə/ may be derived from earlier *e, but the reduction of *e to /ə/ seems to have been general in Albanian (Demiraj 1988: 367). For this reason, it is clear that the /o/ in Opoja must be derived from *ə.
Another instance where /o/ in Opoja corresponds to /ə/ elsewhere in Albanian occurs in nouns with the idiosyncratically unstressed endings *-ull, *-urr, and *-ur. These nouns display multiple patterns of behavior in Albanian dialects. In the majority of Albanian dialects, as in the standard language, these nouns are exceptional only in the assignment of stress. In northern Kosovo and in a range of dialects stretching from Lake Shkodra in the northwest Albania along a compact line running southeast almost to Lake Ohrid, the vowel /u/ in these endings is preserved in the nominative indefinite singular, but deleted when other desinences are added. This results in paradigmatic vowel-zero alternations between /u/ and Ø. Another large group of Albanian dialects shifts the /u/ in these endings to /ə/, which is deleted with other desinences are added. This phenomenon occurs in a contiguous region comprising southern Kosovo, western Macedonia, and parts of central Albania around Tirana (ADGJSH: 42, 43). Opoja is located within this region. These patterns are illustrated in tables 2, 3 and 4.
Opoja, instead, has a fourth pattern. In Opoja, the *u in these suffixes becomes /o/, and vowel-zero alternations occur. The geographical location of Opoja in the midst of dialects in which *u > ə with alternations makes it extremely likely that the Opoja dialect was originally like the Has dialect illustrated in table 4 before a shift from /ə/ > /o/. The relevant data from Opoja appear in table 5.
Some instances of /o/ clearly derived from /ə/ also occur in the verbal system. This is most directly observable with the plural aorist endings, especially in consonant-stem verbs. Such verbs form paradigms with the endings -om, -ot, and -on: mat-om, mat-ot, mat-on “measure-1pl, 2pl, 3pl.aor”. These correspond directly to the aorist endings -əm, -ət, and -ən elsewhere in most of Albanian,2 including the northeastern Geg dialects that most closely neighbor Opoja (ADGJSH: 371). Furthermore, Pajaziti (2005: 169) explicitly states that these reflexes are found only in Opoja. As with the examples given above from feminine nominal paradigms, the desinences -om, -ot, and -on in Opoja clearly reflect a shift from *ə > /o/. These endings have been generalized to vocalic-stem verbs, e.g. punu+om, punu+ot, punu+on “work-1pl, 2pl, 3pl. aor” (Pajaziti 2005: 164).
The extension of -om, -ot, and -on to vowel-stem verbs is a morphological innovation specific to Opoja Albanian, a point which is potentially obscured by the presence of superficially identical past tense forms in other dialects of Geg (Pajaziti 2005: 52). In other dialects of Albanian, the aorist in the plural was formed by the addition of consonantal endings directly to the verb stem, with an epenthetic schwa added after consonant stems. The contrast between vocalic and consonantal stems can be seen in the following examples: *mat+m, *mat+t, *mat+n > matəm, matət, matən ‘measure-1pl, 2pl, 3pl.aor’ as compared to *punu:+m, *punu:+t, *punu:+n > *punu:m, *punu:t, *punu:n. Long /u:/ subsequently diphthongized to *uo, which then yielded /ua/ in Tosk, /ue/ in most Geg dialects, and /uo/ in a smaller group of northwestern Geg dialects (ADGJSH: 170). Therefore, the Opoja paradigms are superficially identical to those that occur in dialects that preserve *uo. However, Opoja does not generally preserve *uo. Instead, like the rest of Geg, *u: > *ue, which is in its first stages of monophthongization to [u:], also like the rest of Geg (Pajaziti 2005: 52-53).
This demonstrates that paradigms like punu+om, punu+ot, punu+on cannot reflect the original Albanian aorist pattern. Instead, the consonant-stem endings *-əm, *-ət, *-ən were generalized and applied to vowel stems as well. This must have taken place after *ə had yielded /o/. Otherwise, this morphological innovation would have created the diphthong (or vowel sequence) /uə/ in vowel-stem verbs. Over the last centuries, Albanian in general and Geg in particular has frequently eliminated /ə/ in word-internal position when doing so would not result in impermissible consonant clusters (Topalli 2007: 141-142). Therefore, it is unlikely that the sequence /uə/ would have been stable in Opoja. Regardless of when the *-ə(m/t/n) endings were generalized, however, it is clear that the Opoja past tense endings -o(m/t/n) in vowel-stem verbs must derive from forms with original *ə.
2.2 Opoja /o/ from *ə not clearly attested elsewhere in Albanian
In addition to the above-described cases, anomalous /o/ in the Opoja dialect sometimes does not transparently correspond to /ə/ in other Albanian dialects. Such instances can be explained either as morphological innovations or through an intermediate stage with *ə.
A clearly motivated morphological innovation is found in the 3sg present ending -on that occurs in a group of consonant-stem verbs, e.g. e.g. hap-on ‘open-3sg’, hyp-on ‘jump-3sg’, kap-on ‘seize-3sg’, korr-on ‘harvest-3sg’, prek-on ‘open-3sg’ (Pajaziti 2005: 159). These verbs usually have no overt singular present endings in other dialects of Albanian. However, the ending -on in Opoja is reminiscent of the 3sg ending in the highly productive class of o-stem verbs in Albanian, in which -n is added to a stem ending in -o, e.g. puno + n > punon ‘work-3sg.’ This similarity makes it extremely likely that Opoja generalized the ending -on from o-stem verbs, thereby innovating overt 3sg present morphology in these verbs parallel to that found in other verb classes.
Morphological innovations also appear in the imperfect tense paradigm. Especially relevant are the 1sg and 1pl imperfect endings, illustrated in table 6.
Consonant-stem verbs have a pattern in which -ʃom appears as the 1pl ending, as opposed to 1sg -ʃna: lið-ʃna ‘bind, tie-1sg.impf’ but lið-ʃom ‘bind, tie-1pl.impf’. The presence of /ʃ/ in the 1sg ending is not unusual for Geg and reflects a contracted form of the imperfect of ‘to be’ (iʃ-). The addition of -na in the 1sg imperfect is not common in Albanian, but does occur in a range of dialects neighboring Opoja in southern Kosovo and northern Albania (ADGJSH: 346)3. Turning to the 1pl ending -ʃom, 1pl imperfect endings in -ʃVm (where V represents a vowel) are common in Geg, leaving only the presence of /o/ to be explained (ADGJSH: 349). The ending -ʃəm occurs in some other dialects in southern Kosovo. This ending is found in the nearby dialects of Xërxa in the Rahovec region and in Nishor in the Prizren region, which makes it reasonable to propose that -ʃom in Opoja was derived directly from *-ʃəm. The imperfect is formed differently in vowel-stem verbs, in which -ʃom occurs in the first singular. In these verbs, 1sg -ʃom contrasts to 1pl -ʃim, which corresponds closely to forms found elsewhere in Geg. This is exemplified by paradigms like di + ʃom ‘know-1sg.impf’ versus di + ʃim ‘know-1pl.impf’ (Pajaziti 2005: 162). The provenance of 1sg -ʃom is not entirely clear, and may be the result of morphological reorganization internal to Opoja. Similar forms in -ʃVm occur in northern central Albania, extending northwards as far as Kosovo and into the Tetovo region of Macedonia, but Opoja would be on the extreme periphery of the distribution of these forms (ADGJSH: 352).
The ending -om also occurs in the 1pl present tense of consonant-stem verbs, e.g. korr-om ‘harvest-1pl.pres’, pjek-om ‘bake-1pl.pres’ (Pajaziti 2005: 160). Elsewhere in Albanian this ending is usually -im (ADGJSH: 329). The presence of -om as an ending in the 1pl imperfect and aorist paradigms of these verbs suggests the possibility of a morphological explanation, according to which -om would have been generalized across tenses as a 1pl marker. However, it is also possible that the /o/ in this ending arose from *ə, since some Albanian dialects in Kosovo generalize /ə/ in the present tense plural endings of consonant-stem verbs (ADGJSH: 329-333).
In Opoja, /o/ is prevalent as a theme vowel in present mediopassive paradigms, corresponding to /e/ in most other varieties of Albanian. The only exceptions are the 3sg mediopassive, where -et occurs. The 1pl ending -om is often replaced by -na, which seems to be gaining productivity as a 1pl marker (Pajaziti 2005: 161).4 This is illustrated in table 7.
Some scattered dialects of Albanian innovate /ə/ throughout the present tense mediopassive paradigm (ADGJSH: 398). These dialects are scattered across western Geg, nowhere near Opoja, but their existence suggests that this is plausible as an independent innovation. A partial parallel exists in the nearby dialect of Anadrin, where the 1pl mediopassive ending *-emi > -um and -mi (Pajaziti 2008: 199). The dual tendencies for ə > u / _m and for unstressed /ə/ to drop in Anadrin provides some support for hypothesizing an intermediate stage of at least the 1pl ending *-əmi (Pajaziti 2008: 65)
Anomalous /o/ also occurs in participles in Opoja Albanian. Like other varieties of Geg, participles are not always marked by overt morphology. Overtly marked participles in Opoja are formed either with -om or -ot. The ending -om is preferred. The ending -ot is limited to a smaller class of verbs, and appears to be an innovation limited to Opoja, although -t is also found sporadically in the nearby Anadrin dialect (Pajaziti 2008: 173).
The participial ending -om is derived from *-un, which is still found in some Kosovar dialects, although it is generally eliminated (ADGJSH: 386, Pajaziti 2005: 171). It is likely that the derivation of -om from *-un proceeded through an intermediate stage with /ə/, probably -ən. One source of evidence for this hypothesis is found in the nearby dialect of Anadrin. It appears that Anadrin and Opoja have shared some innovations not common elsewhere in Geg, such as the emergence of participles in -t (Pajaziti 2008: 174). This suggests that developments in Anadrin may provide insight for analagous phenomena in Opoja that have been obscured by subsequent developments. In Anadrin, participles in *-un display considerable variation. Original forms in -un are attested, although mostly among the older generation. The most common form of these participles is -um. However, the alternate forms -ən and -en are found; while -ən is not common, it formally mirrors a possible intermediate stage for Opoja (Pajaziti 2008: 68, 173). The ending -en may be derived from *-ən, a development that seems to have taken place elsewhere in Anadrin as well as in nearby Gjakova (Pajaziti 2008: 68, Agani 1978). Comparative evidence from Anadrin suggests an intermediate stage in *-ə(m/n) for participial endings in Opoja.
Internal evidence also points to an intermediate stage in *-ə(m/n). The Opoja dialect overtly marks feminine agreement in participles from *-un, a development shared with other Geg dialects, including Anadrin (Pajaziti 2008: 174). A development unique to Opoja is the emergence of vowel-zero alternations in these paradigms, e.g. i veʃ-om ‘dressed-masc.sg’ versus e veʃ-me ‘dressed-fem.sg’ (Pajaziti 2005: 135). Vowel-zero alternations between /o/ ~ Ø do not normally occur in Albanian, which would make such patterns challenging to explain. By contrast, alternations between /ə/ ~ Ø are quite common. An intermediate stage in *ə, which is also supported by comparative evidence, obviates the need for any special diachronic account of /o/ ~ Ø alternations.
In summary, even in environments where unique /o/ in Opoja does not obviously correspond to /ə/ elsewhere in Albanian, a reconstructed intermediate stage with *ə provides the most plausible account for the emergence of /o/. This, then, leads to the question of how to explain the *ə > /o/ sound change that took place in Opoja.
3. Opoja and the Slavic Jers
As discussed in section 1.1 above, the population of Opoja shifted from Slavic to Albanian beginning in the 16th century. The Slavic dialect of Opoja would have been located precisely on the isogloss between two large areas of Slavic in which the jer shift, in which hyper-short high vowels (jers) were either lost or shifted to full vowels, had different results. To the south of Opoja is the Slavic dialect of Gora, in which the jer reflexes are /o/ and /e/ (Mladenović 2001: 131-132). To the north of Opoja are the Prizren-Morava dialects of Serbian, in which both the jers yielded /ə/. In addition to geographical proximity, the region of Opoja had distinct sociocultural links to both of the above regions. Opoja and Gora shared long-standing cultural commonalities, which were preserved well into the modern period, but significant trade ties also existed between Opoja and the nearby urban center of Prizren. For this reason, it is extremely likely that the inhabitants of Opoja were aware of both jer reflexes in the period before and during the language shift from Slavic to Albanian. The contrast between cultural links with Gora and mercantile / political links with Prizren makes it reasonable to hypothesize that the jer reflexes were sociolinguistically salient variables. This is especially likely since the initial impetus for Albanization in Opoja was linked to the settlement of a landowner from the Prizren region, as discussed in section 1.1. During this process, it appears that Albanian alternating /ə/ was identified with the Prizren jer reflex and replaced by the non-Prizren reflex /o/.
Since the Slavic dialect of Opoja has no direct attestation, its jer reflexes must be reconstructed. Three possibilities exist: (1) both jers > /ə/, as in Prizren, (2) the jers > /o/, /e/, as in Gora, and (3) both jers > /o/, a pattern not attested elsewhere in Slavic. Possibility (1) is quite unlikely. Possibilities (2) and (3) are more probable, as shown below.
It is logically possible, but extremely unlikely, that Opoja had Prizren-type jer reflexes (both jers > /ə/), but awareness of the reflexes of Gora led to the imposition of Gora-type reflexes. As previously discussed, Gora and Opoja are located in the same highland region that is sharply demarcated geographically from the lowlands of Prizren. This, combined with the existence of cultural parallels between Opoja and Gora, makes it likely that the dialect of Gora and the Slavic dialect of Opoja were closely related. Furthermore, an awareness of Gora reflexes does not by itself provide sufficient motivation for their imposition onto Albanian, especially if ties between Opoja and Gora were weakening during the period of language shift.
The remaining two possibilities are difficult to distinguish conclusively. The default hypothesis would be that Opoja had identical jer reflexes to Gora. Nonetheless, it is possible that Opoja had a transitional pattern, in which both jers regularly yielded /o/. This would be unique within Slavic, but the existence of analagous developments suggests that it is a reasonable possibility. In southern Montenegro, dialects are found in which both jers merged to /æ/, which then developed to [ea] or even [e]. Evidence from loanwords suggests that this pattern spread into northern Albania (Popovič 1958). The articulatory path from [ʊ] to [æ] attested in Montenegrin is similar in scope to the path from [ɪ] to [ɔ] proposed here for Opoja. A more direct parallel is provided by the Macedonian peripheral dialects of Drimkol-Golo Brdo and Reka, which have a rounded /å/ from *ə. In this case, /å/ is derived from etymologically different sources (å < *ə < *ǫ, *r̥, *l̥) (Vidoeski 2005).5 These developments in nearby dialects reflect sound changes similar to that proposed here for Opoja.
The hypothesis that /o/ was the only jer reflex in Opoja also avoids a difficulty that proposal (2) encounters. Namely, if Opoja had /o/ and /e/ from the jers, as in Gora, it would be necessary to explain why /o/ emerged as the alternating vowel that replaced Albanian *ə. If Opoja had both alternating /o/ and /e/ from the jers, a sound change from *ə to /e/ would be more likely, since alternations between /ə/ and /e/ in various environments are fairly common in Albanian. If, instead, Opoja had only /o/ from the jers, its imposition onto *ə follows automatically. Such an account would involve the direct imposition of Slavic (morpho)phonology onto Albanian during the period of language shift, and is much simpler than a process in which one of two alternating vowels was chosen.
This paper presents an argument that unique /o/ in Opoja Albanian is derived from *ə, and that the sound change of *ə to /o/ took place through imposition of substrate Slavic jer reflexes onto Albanian /ə/ during language shift from Slavic to Albanian beginning in the 16th century. This conclusion is of note in the context of Albanian studies because it offers a possible explanation for a heretofore unexplained dialect feature. It is also of note for the history of Slavic insofar as we can extract information from Opoja Albanian about the Slavic substrate of that region. It is possible, as discussed in section 3, that the Slavic substrate dialect of Opoja belonged to a previously unattested type of dialect in which both jers yielded /o/.
Furthermore, this development is of note from the point of view of contact linguistics more broadly. First, it is a case study of how substrate imposition can result in what synchronically appear to be non-obvious sound changes. Furthermore, it contributes to research dealing with the possibility of reconstructing information about substratum languages through indirect evidence (e.g. Lehiste 1965). A particularly striking parallel is with Pannonian Slavic: Richards (2003) uses the evidence of Slavic loanwords into Old Hungarian to develop a description of the Pannonian Slavic substrate (see also Greenberg 2004). Another parallel is with Myznikov’s (2004) use of Finnic loanwords into Russian dialects to delineate separate zones of substrate influence from Finnic. In both of these cases, studies of the substrate both provide additional perspective on the multiple layers of language contact and change characteristic of linguistically diverse regions and provide additional information about the earlier dialect differentiation of the substrate languages.
Agani Hilmi. 1978. Ë-ja në të folmen e qytetit të Gjakovës. Studime gjuhësore I. Dialektologji. Prishtina. IAP. 207-225.
JosephBrian. 2004. Typological and Areal Perspectives on the Reshaping of a Macedonian Verbal Ending. In JosephBrian and JohnsonM. A. eds. Macedonian Studies. Papers from the 5th International Macedonian-North American Conference on Macedonian Studies 1-4 May 2003 at The Ohio State University (Ohio State Working Papers in Slavic Studies 4): 143–151.
JosephBrian. 2006. On Connections Between Personal Pronouns and Verbal Endings in the Balkans. In RothsteinR.ScattonE. and TownsendC. eds. Studia Caroliensia. Papers in Linguistics and Folklore in Honor of Charles E. Gribble.Bloomington: Slavica Publishers: 177-88.
Richards Ronald O. The Pannonian Slavic Dialect of the Common Slavic Proto-Language: The View from Old Hungarian. Ivanov Vyacheslav V. and Brent Vine UCLA Indo-European Studies Volume 2. Los Angeles: UCLA Program in Indo-European Studies.
1 The Albanian-speaking population in Gora is likely of recent provenance, and should therefore be distinguished from the Albanian-speaking population of Opoja (Mladenović 2001: 36-38, Friedman p.c.)
2 Northwestern Geg and southeastern Tosk have paradigms in which the order of the segments in these endings is switched, e.g. -mə, -tə, -nə (with various adjustments of the final vowel) (ADGJSH: 371).
3 This reflex is found in 16 points on the ADGJSH dialect map, including most saliently Gjakova, but extending well into Albania.
4 The 1pl ending -na is likely influenced by the nominative pronoun na 'we.' The form na – as compared to ne in Standard Albanian – is almost ubiquitous in Geg (ADGJSH: 275). Similar influence of pronouns on verbal inflections has also been documented in dialects of Macedonian, Bulgarian, and Greek (Joseph 2004, Joseph 2006).
5 The articulatory path from a rounded /å/ to [ɔ] or even [o] is not problematic; these sounds are phonetically close, differing primarily in the degree of rounding. Parallels exist in nearby dialects of Albanian (Gjinari and Shkurtaj 2003: 185-187) and also in the Rhodope dialects of Bulgarian (Stojkov 1968: 87-91).