2.1.1 The Diachrony of Romance Influence

The problem of the areal distribution of Latin on the Balkans has been addressed in many controversial discussions, and the borderline between Latin and Greek domination before and after the bipartition of the Roman Empire has also been under examination. Almost all historians agree on the existence of the so-called “Jireček line,” more or less following the Roman Via Egnatia, which started in the Durazzo region in Albania and went across the Balkans, with Skopje (Lat. Scupi) in Macedonia to the left, Latin side. Then it turned in a northeastern direction, leaving Sofia (Serdica) on the right, Greek side.⁵

In any case, there was Latin dominance in the Moesia (Superior and Inferior) provinces of the northern Balkans, where the Slavs started their invasions by crossing the Danube in 527 ce and where later on Bulgarian and, at least in part, Proto-Romanian formed.

In view of the controversial denomination of the individual stages and the distribution of Latin/Romance varieties, claimed in the literature, Figure 1 may be seen as just one possible phylum of Balkan Romance. It is based on language contact as the main criterion for language splitting.

Open in new tab

Figure 1. Attempt at a phylum of Balkan–Latin varieties.

Here the largely homogeneous variety of the whole Empire, including the Balkans, is termed “(common) Latin.” Specific characteristics referring to the whole Peninsula are addressed as “Balkan Latin.”⁶ In the western parts, it developed into “Western Balkan Romance” on the basis of an Illyrian substrate, possibly bifurcated at least into a northern and a southern variety (Muljačić, 2000, pp. 363–374).⁷ Its counterpart was “Eastern Balkan Romance,” with its Thracian substrate. Between Western and Eastern Balkan Romance, to the west of the river Drina, there was an area without a Romanized population, termed “Romanisierungslücke” (Arvinte, 1968).

After the Slavic invasion, Western Balkan Romance transformed into “Dalmatian,” under the influence of the western varieties of South-Slavic (the Serbo-Croatian language continuum). Dalmatian was spoken in the early Middle Ages predominantly near the Adriatic coast and its hinterland in several varieties, the best known being Ragusian and Vegliote, all of them extinct (Bartoli, 1906; Muljačić, 2000; see also the article “Dalmatian (Vegliote)”).

In contrast, Eastern Balkan Romance developed into “Proto-Romanian,” due to the influence of the sub- and superstrate of the eastern varieties of South Slavic, also termed “Balkan Slavic,”⁸ which developed into the Bulgarian–Macedonian language continuum. In the 10th century, Proto-Romanian divided into (common) Daco-Romanian and south-Danubian Aromanian, whereas the other south-Danubian varieties, Megleno-Romanian and Istro-Romanian, separated only later on from common Romanian, perhaps in the 12th or 13th century.

Left-Danubian Dacia had become a Roman province in 107 ce. It was populated by Romance-speaking settlers from other parts of the Empire and by more or less Romanized local Thracian (Geto-Dacian) tribes. In 275 ce, Roman troops and administration withdrew to the right of the Danube, and with them at least a considerable part of the Romanized population. In contrast, most of the territory of what now is Romania was invaded by Slavic tribes, including its southern part, Wallachia. This is evidenced by its denomination as “Sklavēnía” (Land of Slavs) in Byzantine–Greek documents (Rosetti, 1984, p. 264).

There are three basic hypotheses for the formation of Romanian (and the Romanian people): the autochthony thesis (Proto-Romanian developed in left-Danube Dacia only), the discontinuation thesis (all Latin-speaking people left Dacia), and the “as-well-as” thesis, with Proto-Romanian developing on both sides of the Danube.⁹

At the latest in 527 ce, when Slavic people crossed the Danube and settled in Romance-speaking Moesia, the so-called Romance–Slavic “symbiosis” with its strong bilingualism started. As it seems, the Slavic migration to the south was accompanied by an opposite migration of part of the Romanized population to the north. As a result, this symbiosis continued on either side of the Danube, with exactly opposite effects: Slavic-influenced Romanian in the north and Romance-influenced Slavic (Bulgarian) in the south.

2.1.2 Romance Influence on Balkan Slavic (Bulgarian and Macedonian)

As for Bulgarian, the influence of the Romance substrate in Moesia is clearly recognizable in essential parts of its grammar, whereas Romance lexical borrowings dominate in its shepherd and dairy-farming terminology¹⁰—for example, fičor ‘young shepherd’ ← ficior, zira ‘whey’ ← zer. But they are abundant also in many other domains of material and spiritual life—for example, buza ‘cheek’ ← buză ‘lip’, Kračun ‘Christmas’¹¹ ← Crăciun—especially in local dialects.¹² Romance influence in the Bulgarian lexicon mostly goes back to the earliest centuries after the invasion of Slavic tribes in the south-Danubian territory. But Bulgarian also has a layer of Pre-Romanian (i.e., Balkan–Latin) lexical borrowings, often found in a similar form in Serbo-Croatian. They mostly belong to Christian terminology such as olъtarь ‘altar’, komъkati sę ‘to commune’, kămotrъ ‘godfather.’¹³

The Romance–Slavic symbiosis, especially the substrate effects, going back to the Slavicized Romance population, was probably the main reason for Bulgarian and Macedonian turning in the course of a long-lasting process into the most untypical members of the Slavic language family.¹⁴ The most evident parallelism of Balkan Slavic with Romance is the substitution of morphological cases by prepositional phrases, probably as a result of the substrate type of language contact (Breu, 1994; Wahlström, 2015).

Romance seems to have been also the main mediator of Balkanisms, for example, the loss/reduction of the infinitive, ultimately going back to an internal-Greek phonetic development.¹⁵ In part, they could have even been their very source, as in the case of the development of an article system in Balkan Slavic (Skok, 1934, p. 200). The morphological Genitive–Dative merger, found in Bulgarian and other Balkan languages, could be based on the corresponding traditional syncretism in the Latin a-declension of the type causae ‘cause’ dat=gen.sg.¹⁶ Object doubling by means of personal clitics seems to be another case, as similar (although not identical) doubling is found also in other Romance varieties.

The substrate influence of Thracian could have been the ultimate reason for the postposition of the definite article in Balkan Slavic and common Romanian, although internal Romance variation and an independent Bulgarian development are not excluded.¹⁷ Another possible case is the completely analytical comparison of adjectives and adverbs by means of particles (comparators)—for example, dobăr – po+dobăr = Ro. bun – mai + bun, Alb. i mirë – më i mirë ‘good – better’.

Because Romance influence is especially strong in the southwestern region of today’s (North-)Macedonia, historically a center of the Aromanian population, Romance characteristics in Macedonian are claimed to be copied mainly from this variety. Aromanian had both functions—that of an adstrate in the everyday contacts with its Slavic neighbors and that of a substrate, due to language shift of its speakers to the dominant local Slavic variety. This language shift is probably responsible for the development of an analytical perfect of the Romance type, formed with to have + passive past participle (e.g., imam dojdeno ‘I have come’), as opposed to the usual Slavic perfect type, formed with the auxiliary to be + L-participle.¹⁸ There are also mixed types of perfects in Macedonian, again showing some evidence for Romance influence (Makarova, 2016).¹⁹

Calques from Aromanian, possibly going back to the substrate type, too, are also frequent in Macedonian—for example, the lexical calques sakam ‘want, love’, corresponding to the Aromanian polysemy of voi^u, and vrne in the sense of ‘it is raining’ besides its basic meaning of ‘returning’, based on the model of Aromanian vearsă. A phraseological calque is umot mu se svrte, formed on the model of Arom. míntea il’ si šucə́ ‘he went mad’, literally “the mind turned him over” (Gołąb, 1984, p. 15).

In contrast, lexical matter borrowing from Aromanian seems rather insignificant in Macedonian, with the usual exception of the shepherding terminology. Typical examples are bač ← bač^u ‘chief shepherd’, samar ‘pack-saddle’ ← səmaru, mirizuva ‘to rest (referring to sheep in shadow during hot day hours)’ ← (a)miriʒ^u < Latin meridiare. In some cases, a pre-Aromanian Balkan–Latin source has to be considered, too—for example, porta ‘door’ ← porta (≠ Arom. poartə), klisura ‘canyon’ ← clisura (Gołąb, 1984, pp. 12–13).

In modern times, Romance influence on Bulgarian and Macedonian has come mainly from French. Trummer (1998, p. 155) counts more than 2,000 lexemes, predominantly in technical domains.

2.1.3 Romance Influence on the Serbo-Croatian Language Continuum

The migration of the western branch of South Slavs, directed to the Adriatic Sea, took place at about the same time as that of the eastern (Bulgarian/Macedonian) branch. Their language divided into Slovenian and the Serbo-Croatian language continuum.

In the Serbo-Croatian area, language contact started as soon as the newcomers mixed up with the existing Balkan–Romance population, speaking varieties of Western Balkan Romance. All over the Dalmatian territory, Balkan–Romance borrowings concern culture and housing, as well as the local flora and fauna, especially fishing.²⁰ Characteristics of such borrowings from Western Balkan Romance are the preservation of the consonant clusters ct [kt] (≠ It. tt, Romanian pt) and cs, x [ks] (≠ It. ss, Ro. ps), as well as ps, mn (≠ It. [ss, nn]), and missing palatalization of the velars [k, g] before front vowels (≠ It. [tʃ], [dʒ]). Examples are trakta ‘fisher net’ ← tracta ≠ It. tratta, kapsa ‘chest, coffin’ ← capsa ≠ It. cassa, liksija ‘suds’ ← lixivia ≠ It. liscivia, surgati ‘to weigh anchor’ ← surgere ≠ It. sorgere [dʒ].

The next stage of Romance influence is characterized by Friulian borrowings, usually with a larger diffusion than Balkan–Romance terms. Most frequently they are connected with the liturgical domain, which speaks in favor of their distribution by the patriarchate of Aquileia (Ernst, 1983), p.es. križ ‘cross’ < kryžь ← crọge [ʒ] and korizma ‘Lent’ ← corèsime (quaresima).

Then Venetian became the overall donor language. Its influence was very far-reaching due to Venice’s political hegemony in the coastal area and its cultural, military, and economic strength, as well as due to its role as the local opponent against the Osman–Turkish Empire.²¹ Many Venetian borrowings have reached the Dalmatian hinterland, too. Semantically, they came from all fields in which Venetian culture was dominant—for example, brageše ‘trousers’ ← Ven. braghesse, kamara ‘room’ ← cámara, garzun ‘trainee’ ← garzon, kogo ‘cook’ ← cogo, štumig ‘stomach’ ← stòmego.

Borrowings from southern Italo-Romance dialects are rarer, and they normally concentrate in the southern coastal area of Dalmatia—for example, maret ‘calm sea’ ← maretto or tanđati ‘to tempt’ ← tandà.

Borrowings from several Italo-Romance varieties, but especially from Venetian, even entered the standard languages of the Serbo-Croatian continuum—for example, kasa ‘cash (desk)’ ← It. cassa, kaput ‘coat’ ← It. cappotto; koverta ‘envelope’ ← Ven. coverta. In part, Italianisms entered Serbo-Croatian by Greek mediation—for example, ambis ‘abyss’, rendering the Greek orthography of originally Italian abisso; see also the article “Greek in Contact With Romance.” In contrast, Venetian in some cases was the mediator for Germanic and Oriental source words, such as roba ‘merchandise’ ← It./Ven. roba ← German roub (Raub) and gabela ‘duty’ ← It. gabella (of Arab provenience).

In Dalmatia, apart from simple matter borrowing, we also find contaminations and pleonastic combinations of Romance and Slavic forms—for example, vajksenpre ‘always’ ← Slav. vavik ‘always’ × It. sempre ‘always’; baramenku ‘at least’ ← Slav. bar × Ven. almanco (with identical meanings).

Less important Romance influences on the Serbo-Croatian continuum came from the opposite side of its territory, or rather from Romanian-speaking nomadic Vlachs. Just like elsewhere on the Balkan Peninsula, such borrowings most frequently concern sheep breeding and dairy farming—for example, kaš ‘cheese’ ← caş, strgljata ‘sour milk’ ← strigl ́ata.²² But Vlach influence also plays a role in toponyms (Skok, 1934, p. 447), like the denomination of two mountains in Montenegro: Durmitor (from a dormi ‘to sleep’, originally a place, where shepherds and their sheep passed the night) and Pirlitor (from a pârli ‘to burn’, originally woodland cleared by means of fire).

Apart from this, there are old Romance borrowings in Serbo-Croatian, found, in part, also in Bulgarian, whose exact provenience, due to their form and distribution, remains unclear—for example, oltar ‘altar’ ← Lat. altare(m), poganin ‘pagan’← Lat. paganum.

In modern times, standard French (and Italian) borrowings have entered the Serbo-Croatian language continuum, often by means of German mediation (Trummer, 1998, p. 156). In this context, the infinitive suffix -ir- is especially important. It originally derives from French -er, -ir, but came into Serbo-Croatian (especially its western parts) as part of German borrowings in -ieren, like Fr. informer ‘to inform’ → German informieren → Croatian informirati.²³ Later on, it became productive also language-internally, mostly in colloquial forms such as Croatian slikirati (standard slikati) ‘to photograph’ from slika ‘picture’, nogirati ‘to shoot (soccer)’ from noga ‘foot’ (Popović, 1960, p. 617).

2.1.4 Romance Influence on Slovene

The original Romanization of today’s Slovene territory had probably been rather weak, as it left just a few words in the language of the immigrating Alpine Slavs, such as češnja ‘cherry’ ← Lat. Ceresia, and some Christian terms such as križ ‘cross’ (Šega, 1998; Trummer, 1998, pp. 158–159). The role of Romance changed, however, after the 11th century, although for political and economic reasons, borrowings from Italian, Venetian, and Friulian (and French) were very frequently mediated by German (Ožbot, 2008). But in the western-most regions, where Romance and Slavic people mixed, some kind of bilingualism came into existence. This was especially true in the patriarchy of Aquileia (Ernst, 1983) and in the Venetian-dominated part of Friuli. Here even linguistic calquing (Skubic, 1997, 2007) occurred.²⁴ In the Middle Ages, just like in West-Slavic countries (see Breu, this encyclopedia, b, section 2.1), Latin was the dominant language of education and science, which also resulted in a number of Latinisms in Slovene, though again mostly by means of German mediation.

2.2.1 Italo-Romance Influence on Molise Slavic

Molise Slavic (MSl.) is the micro-language of the South-Slavic minority in the southern Italian Province of Campobasso, Region of Molise, subdivided into three dialects, differing in many respects. It has been in “total language contact”²⁵ since the immigration of the ancestors of the Molise Slavs from the Herzegovina (Dalmatian hinterland) in the 16th century. MSl. belongs to the Serbo-Croatian language family, and it shares the Štokavian dialectal basis with the modern standard languages of this group. But it differs from them in having an Ikavic base, due to the development of Proto-Slavic ě > i and not ě>e as in Serbian nor ě > (i)je as in Croatian, its historically nearest cognate.²⁶

Whereas these standard languages are mutually intelligible, Molise Slavic is very different. This is mostly due to the effects of language contact with Italo-Romance varieties, serving as a source for matter borrowing and as a model for all types of pattern borrowing. In principle, Molise Slavs live in an adstrate–superstrate situation, which has lasted for approximately 500 years. All of them are bilingual. They use Italian in everyday conversation with Romance monolinguals, but it also serves as their high-code variety for schooling, writing, and administrative purposes. Their situation of total language contact has led to extremely radical changes on all linguistic levels, not only in the lexicon but also in grammar.

MSl. has no substantial restriction on spontaneous or persistent borrowing of lexical elements, which has allowed for 25–50% of loanwords in the everyday vernacular (Breu, 2017, pp. 67–71). There is still a stronghold of basic Slavic vocabulary against Romance dominance, excluding, however, abstract nouns, which have almost completely been replaced. For example, the Slavic suffix -ostь for the formation of abstract nouns from adjectives or -nie for deriving nouns from infinitives have been lost altogether. In contrast, Italianisms (given here in their Standard-Italian form, differing phonetically from the real dialectal sources) such as džuvindu ‘youth’ ← It. gioventù and gvalita ‘type’ ← qualità are used. Nouns are normally integrated into the paradigm of the gender in question—for example, It. senape f → sinapa ‘mustard (plant)’ f and timone m → timbun ‘thills’ m. There are, however, exceptions, when the source form does not match formal characteristics of the expected gender paradigm, for example, in the case of feminines with a stressed final vowel, becoming masculine in the Acquaviva dialect—for example, città f → cita m ‘city’ (Breu, 2017, pp. 63–64).

A special case is the system of numerals, with only 1–4 remaining almost exclusively Slavic. Higher numerals either show variation (5–10, 100) or are exclusively Romance. Variation is only rarely free. Normally the type of noun with which the number combines (inherited or borrowed, dialectal or standard Italian, morphologically integrated or not) decides. Dates show exclusively Romance numbers. The situation becomes still more complicated by the fact that the borrowed numerals consist of two rows of different origin (Italian standard vs. dialectal), which also influence the distribution rules (Breu, 2013).²⁷

In derivational morphology, both matter borrowing and pattern borrowing have occurred. Loan morphemes appear mainly in the formation of nouns. Examples are the suffixes -un and -ur. They entered MSl. as parts of borrowed nouns from the local Italo-Romance varieties, such as balun ‘gorge’ ← vallun, kužitur ‘tailor’ ← cuscətur. Later on, they were “naturalized”; see, for example, grihun ‘sinner’ with -un attached to the Slavic stem grih ‘sin’. Another case is the Italian suffix -at, forming de-verbal nouns, with borrowings such as fumata ‘(the) smoking’ ← It. fumata, having become productive in Molise Slavic, too—for example, smihata ‘(loud) laughter’ with -at attached to Slavic smih ‘(the) laughing’.

On the other hand, we find pattern borrowing in the case of the Slavic diminutive suffix -ic, traditionally affixed to nouns, but identified by the speakers with Italian -in in such a way that it can even be attached to adverbs. See, for example, the case of dobrica ‘rather good’ with -ic attached to dobra ‘well’, which calques It. benino, derived from bene.

A special case is the Slavic grammatical suffix -n, forming passive past participles both from Slavic verbs and from loan verbs, such as kandan (passive participle) from kandat ← It. cantare ‘to sing’, past participle cantato. Slavic -n and Italian -t are fully identified with each other. As a consequence, -n replaces -t analogically also in borrowed adjectives whose source shows this suffix, but which synchronically have lost their participial character in the model language—for example, fortunato ‘happy’ → fortunan, scostumato ‘rude’ → skoštuman.

Molise Slavic has two different layers of loanwords, older ones borrowed from the local Italo-Romance dialects and younger ones from standard Italian. Borrowed suffixes from the older layer have become productive also for younger loanwords. For example, even modern Italian nouns with the suffixes -one and -ore are integrated with the corresponding older suffixes -un and -ur—for example, cacciatore ‘hunter’ → kačatur. The reason for this unphonetic integration comes from the identification of the foreign with the naturalized suffix by means of the juxtaposition of older dialectal loanwords with their equivalents in standard Italian. This difference later on became the model for the productive transformation of the suffixes in new loanwords.

Such a type of analogically motivated loanword integration plays its role in the integration of verbs, too. Whereas in standard Italian there are four verb classes, ending in the infinitive in -are, -ire, and stressed and unstressed -ere, loan verbs are integrated only in the two classes with infinitives in -at and -it, with the two standard-Italian classes in -ere merging with the -ire class—for example, servire → servit ‘to serve’, possedere → posedit ‘possess’, promettere → primitit ‘promise’. This mechanism is based, once again, on the older layer: As in the Italo-Romance Molise dialects, the -ire and -ere classes have merged, the equivalents of modern Italian verbs such as leggere ‘to read’ have -it in Molise Slavic, here lejit ← lejjə. This difference has likewise become productive, precisely as the basis for an analogical integration of all new -ere source verbs into the -it class (Breu, 1991, pp. 47–49).

In (grammatical) morphology, matter borrowing has not occurred in Molise Slavic, but pattern borrowing is responsible for some morphomic reorganization in the ending paradigms. An important case is the imperfect, where the zero ending of the first person singular was replaced by -u, the ending of the third person plural—for example, padahu ‘I fell = they fell’—by copying a similar syncretism or at least a parallel stem in Italo-Romance varieties, especially in standard Italian.²⁸

However, the most evident influences are observed in the field of the semantic calques in the lexicon, but especially also in grammar, where they lead to mergers or the reorganization of existing grammatical categories, to new functions for individual grammemes, and even to the rise of new grammatical oppositions. Such a case of the adaptation of the semantic structure has occurred in the domain of the imperfect. Based on the model of the colloquial Italian imperfect, expressing apart from its tense-aspect functions also counterfactual mood, a similar meaning expansion has occurred—for example, padahu has adopted a parallel range of meanings such as Italian cadevo ‘I was falling, I used to fall, I would have fallen’ (Breu, 2019, pp. 400–403).

In addition, the Italian model is responsible for the rise of an article system on the base of the numeral jena ‘one’, which has become as polysemous as Italian uno, in expressing besides its numeral function an indefinite article. In contrast, there was no polysemous model in Italian for the development of a definite article. Here the article il, la never has demonstrative meaning, which in the situation of total language contact even blocked an independent development of a MSl. article from a demonstrative pronoun, quite frequent in the languages of the world. Nevertheless, MSl. has acquired a complete article system, by reinterpreting the lack of an article as a definite (zero) article, contrasting with the indefinite article—for example, jena ljud ‘a man’: Ø ljud ‘the man’ (Breu, 2019, pp. 411–414).

A further contact-induced development is the rise of an eventive passive on the model of the Italian venitive construction, consisting of the auxiliary venire ‘to come’ + passive participle of the lexical verb—for example, viene fatto ‘is being made’. In Molise Slavic, this corresponds to grem ipfv / dokj pfv + passive participle, here gre činjen ‘is being made’ (Breu & Makarova, 2019).

Clear contact-induced developments have occurred in the domain of gender, too. The most salient one is the loss of the neuter in nouns, which led to a two-gender system like in Italian. Former neuters have generally changed to masculines, such as more n ‘sea’ > mor m. Less frequently they became feminines, such as nebo n ‘sky’ → neba f in the Acquaviva dialect. But nevertheless, the neuter has survived as an impersonal gender, also appearing in substantivized adjectives and adverbs (Breu, 2019, pp. 409–410). This seems to a certain degree parallel to the existence of a third gender in southern Italo-Romance dialects (Paciaroni et al., 2013).

A further instance of direct influence of Italian gender on Molise Slavic is the dismantling of the irregular feminine declension class ending in a consonant. This class as such has been completely lost, but its individual members either joined the normal feminines in -a or passed over to the masculine gender. Which way was chosen depended exclusively on the gender of the corresponding Italian noun. So, stvar ‘thing’ remained feminine and changed its form into stvara, because of the feminine gender of its Italian equivalent cosa, whereas kost ‘bone’ became masculine, due to the gender of its Italian equivalent osso (Breu, 2019, p. 409).

In contrast to quite a lot of contact-induced innovations, we also find conservative features, lost in other Slavic varieties, but kept in MSl., due to their parallelism with the Romance models. One of them is the preservation of the imperfect as a separate tense/aspect grammeme, strictly differing in usage from the perfect, with which it has been replaced elsewhere in Slavic more or less completely. Bulgarian and Macedonian are the only real exceptions, due to their own contact situations. This preservation is still more surprising as in MSl. the traditional aorist has been lost, or rather has been replaced by the perfect, just like in local and north-Italian Romance varieties, where the corresponding passato remoto has disappeared. This contradicts a “diachronic constant” in the general development of Slavic, stating that the loss of the aorist (if lost at all) always precedes the loss of the imperfect. Obviously, in MSl. this constant has been replaced with the corresponding Romance one. Another case of contact-induced preservation in the domain of tenses is the pluperfect, lost in most other Slavic varieties (Breu, 2019, pp. 389–396). Still another important phenomenon is the preservation of the perfect formation with bit ‘to be’ + active l-participle, in contrast to Macedonian, where a have perfect has been induced by Romance substrate influence; see section 2.1.2.

On the other hand, the development of a de-obligative future with the help of the auxiliary jimat ‘to have, must’ + infinitive—for example, ma jist ‘s/he will (obligatorily) eat’—clearly reminds of corresponding structures with the auxiliary avé in southern Italo-Romance dialects. However, the traditional MSl. future, formed by means of the auxiliary tit ‘to want, will’ + infinitive, has not disappeared. But, as a consequence of the deontic connotation of the jimat future, it has reduced to a future of probability—for example, ča jist ‘s/he will (probably) eat’. This means that Romance influence in this field not only caused the calque of a new future type but also led to an opposition nonexistent before, in both languages in contact (Breu & Pila, 2018).

Interestingly, the Molise Slavic case system has largely been preserved, contrary to Balkan Slavic, which underlines that we have to do with different contact directions. But even the adstrate situation of Molise Slavic allowed for some structural changes in this domain. For example, the locative was lost as a consequence of the coincidence of the expression for motion and state in the Romance model varieties—for example, MSl. u Napulu acc = It. a Napoli ‘to Naples, in Naples’. On the other hand, the instrumental has been, in principle, preserved, but in its function of expressing the means by which an action is carried out the preposition s ‘with’ is added. This led to a merger with its comitative usage—for example, s nožem ‘(together) with a knife, by (means of) a knife’ ins.sg (Breu, 2019, pp. 403–408).

An interesting case of resistance is Slavic-style verbal aspect, expressed by affixes. It has been preserved as a productive grammatical category in Molise Slavic, although Italian does not have it. Even in loan verbs the traditional system remains completely intact. Telic verbs are borrowed as perfectives, forming language-internally an imperfective partner by means of suffixation. Thus we find aspectual pairs such as It. allagare ‘to flood’ → lagat pfv / lagivat ipfv, It. decidere ‘to decide’ → dečidit pfv / dečidivat ipfv (Breu, 2017, pp. 65–66).

2.2.2 Italo-Romance Influence on Resian and Other Slavic Varieties in Northeast Italy

The contact situation of Resian, a Slovene-based minority language in the Italian Region of Friuli-Venezia Giulia in the Southern Alps, is to a certain extent comparable to that of Molise Slavic. Resian has been for a long time in total language contact with Romance varieties, too, and it is open to all types of lexical borrowings, going back to Friulian, Venetian, and standard-Italian sources.

As for the system of numerals, Resian is more conservative than Molise Slavic (see section 2.2.1), with the exception of dates, which are always borrowed from Romance (at least the years). Until 39 and for stu ‘100’ the Slavic term is used exclusively. From 40 onwards, younger speakers tend to use Romance numerals, whereas elder people still keep a vigesimal numeral system of the type dwákrat dwísti ‘40’ (= twice twenty) and the special numeral patardú ‘50’, probably “five rows”; see also Steenwijk (1992, pp. 124–125).

The grammatical effects of Resian in contact are in many respects similar to Molise Slavic. Resian likewise has violated the Slavic diachronic constant mentioned in section 2.2.1 by preserving the imperfect despite the loss of the aorist. In addition, its imperfect has acquired a counterfactual reading, too (Breu, 2011, pp. 175–177), and its pluperfect is frequently used as well. Furthermore, the indefinite article has been calqued as well, and postposition of attributes is also possible, although perhaps less consistent than in MSl. Just like in Molise Slavic, the Slavic aspectual opposition of perfectivity has been preserved, despite its absence in the contact varieties. The future, formed in Resian traditionally with the auxiliary ‘to be’, nowadays contrasts with two modal futures, using ‘will’ and ‘must’ as auxiliaries (Breu & Pila, 2018). Just like in Molise Slavic, a new eventive passive has developed, expressed by the auxiliary prït pfv / parhajat ipfv ‘to come’ + passive participle of the lexical verb—for example, parhajajo wrizane ‘they are (usually) cut’ (Pila, 2021). Similarly, the pure instrumental case in its function of means has been replaced with the comitative prepositional phrase—for example, z marćëlon ‘(together) with a knife, by (means of) a knife’ ins.sg.

But there are also specific differences with respect to Molise Slavic, in part based on the different affiliation of Resian to the northwestern (Slovenian) branch of South Slavic. For example, despite the imperfect being preserved morphologically, its functions were reduced to its secondary counterfactual usage in the past few decades, whereas its indicative aspectual and temporal functions have been replaced by the l-perfect.

In contrast to Molise Slavic, the neuter gender has, in principle, been preserved. The same is true for the locative case, although there is a tendency to the loss of the opposition between motion and state here, too. Besides, there is a new way of expressing localities (state and motion) by means of a local adverb, possibly going back to a Friulian model—for example, ja si bila ta-w Rimë ‘I was in Rome’, jtyt tu-w cirkuw ‘to go to church’ (Benacchio, 2002, pp. 105–118; Steenwijk, 1992, p. 166).

A still stronger case of resistance against contact influence may be seen in the preservation of the dual as an independent number grammeme, although the plural is gradually creeping in. In contrast to Molise Slavic, there is a tendency toward the formation of a definite article (Benacchio, 2002, pp. 43–62). This is a general phenomenon in Slovene varieties, probably going back to older influences from German. Furthermore, the possibility to form abstract nouns by means of the suffix -ust (<‑ostь) continues to exist—for example, mladjust ‘youth’ and žïwjust ‘life’, besides the loanword vïta. Again contrary to Molise Slavic, verbal nouns in -një (<‑nie) are also used, such as paraćawanjë ‘preparation’ and mïslanjë ‘thought’, besides the loanword pinsir.

There are older borrowings concerning the local culture, due to the century-lasting contact with Friulian, whereas terms from all domains of modern life come from Italian and its regional varieties. The integration of nouns is mainly based on the gender of the source. Verbs are integrated in part by means of an n-suffix—for example, vïnčinat ‘to win’ ← It. vincere / Friulian vinci—or directly from the stem of the source verb—for example, It. capire ‘to understand’ → kapet. As for aspect, the same procedure as in Molise Slavic is used, with an integration of telic verbs as perfectives and the subsequent formation of an imperfective partner verb by means of w-suffixes—for example, rispondere ‘to answer’ → rišpondät pfv / rišpundawat ipfv, maledire ‘to curse’ → maladyt pfv / moladiwat (Benacchio, 2009, pp. 189–190).

In other Slovene-based varieties in northeastern Italy (valleys of Torre and Natisone) the effects of language contact are similar to Resian, although salient differences also exist, such as the complete loss of the imperfect in all its functions despite the Romance contact situation.

In all, Molise Slavic and Resian show multiple parallel contact-induced developments but also characteristic differences. While the parallel contact situations with Romance adstrate varieties are in general responsible for the parallel developments, the differences follow from the genealogical characteristics of the two minority languages (Serbo-Croatian vs. Slovenian language family) and from peculiarities of their local Italo-Romance models. In the case of Resian, former additional contact with German also plays a role.

2.2.3 Romanian Influence on South-Slavic Minority Languages

The greater part of Bulgarian minorities still present in 21st-century Romania immigrated from beyond the Danube into the southern Romanian lands of Oltenia and Muntenia in the centuries after the fall of the second Bulgarian Empire in the 14th century and the arrival of the Ottomans on the Balkan Peninsula. At that time, the former left-Danubian Bulgarian population had already been mostly Romanized. The newcomers, however, erroneously termed Sârb ‘Serbian’ and living in an adstrate–superstrate situation with Romanian in several enclaves, have kept their language, although their number has strongly diminished; see, for example, the actual situation of a bilingual community of Romanian Bulgarians situated 70 kilometers to the northeast of Bucharest, described in Trefilova (2016) in connection with mixed kinship terms.

Whereas Balkanski (2010) concentrates on predominantly ethnological and onomastic questions, including those villages where Bulgarian has disappeared later on, the effects of Romanian influence on modern Bulgarian dialects in Romania going back to the immigration wave of the early Modern Age are addressed in Mladenov (1993, pp. 369–431). Interestingly, at least in some dialects, most of the neuter nouns changed to feminine agreement (the ending itself is unchanged), probably based on the model of the opposition between the feminine gender of Proto-Romanian borrowings from Old Bulgarian neuter sources (see section 3.1.1), like baltă f ← blato n, and their correspondences in the minority language, here blato n. > f. Following the principle of proportional analogy, a gender shift is observed in the minority language not only in cases such as blato with its direct Romanian counterpart but also in ex-neuters not borrowed in Romanian, such as ušo ‘ear’ n. → f.

There is also a special group of Catholic Bulgarians in the Banat region of southwestern Romania and surroundings, which settled there only in the 17th/18th centuries. Romanian influence in their vernacular is, of course, less far-reaching than in the dialects of the earlier immigrants, but nevertheless not absent. Whereas initially, when the Banat was still part of the Austro-Hungarian Empire, Hungarian, German, and Serbian influence dominated, Romanian influence on the language of this Bulgarian minority gained ground in the eastern Banat, annexed by Romania after World War I, mostly with respect to lexical borrowings and calques (Stojkov, 1967, pp. 426–453). The loss of the imperfect before the aorist in Banat Bulgarian (Stojkov, 1967, pp. 241–245), despite the imperfect still being fully functional in Romanian, agrees with the respective Slavic diachronic constant of grammatical development; see sections 2.2.1 and 2.2.2 for the reverse order of losses in Molise Slavic and Resian, following the Italo-Romance model. As for the situation of the Banat Bulgarians and their language in the beginning of the 21st century, see, for example, Berger (2017), Kahl and Pascaru (2017), and the references therein.

There are also several linguistic islands in southwest Romania belonging to the Serbo-Croatian language continuum, with quite different origin and immigration times (Breu, 2014, p. 2109). Among them we find the Karashovans (Petrovici, 1935), also called “Torlak Croatians,” although claimed Serbians by others.²⁹ They immigrated from southern Serbia around 1400. So, for example, their analytic comparison, expressed by “po-, naj- + positive”, could be claimed a Balkan phenomenon, already present at that time. But of course, Romanian influence would also be possible.

3.1.1 Slavic (Bulgarian) Influence on Proto- and Daco-Romanian

Slavic borrowings entered Romanian in roughly three subsequent periods: centuries 6–10, centuries 11–13, and contact influences from the 13th century onward.³⁰

The first period coincides with Proto-Romanian times—that is, between the Slavic invasion of the Balkans and the migration of the Aromanians farther south. Slavic influence in this symbiotic situation came mainly from Old-Bulgarian local varieties. In the second period, on the left side of the Danube with its developing principalities of Muntenia, Oltenia, Moldova, and others, dominated by Cumans, Pechenegs, and the second Bulgarian empire, Romanian Church Slavonic³¹ or “Daco-Slavic” became the main donor language. It functioned as a superstrate, while the language of the remaining left-Danubian Slavic population continued to be a substrate for the Romanian vernacular until it finally disappeared. In this more or less diglossic situation with Church Slavonic as the high-code variety, Romanian absorbed predominantly administrative, political, and religious terminology.

In the third period, in addition to Romanian Church Slavonic, lexical material from the neighboring adstrate languages was integrated, for example, from Serbian and Ukrainian. These different sources had different results, especially at the phonetic level, and they differed with respect to their semantic domains.

Proto-Romanian borrowings from Slavic, in principle, continue to be present not only in Daco-Romanian but also, for example, in Aromanian, such as bob ‘grain’, rană ‘wound’, slab ‘weak’, gol ‘empty’. As the loanwords of the Proto-Romanian period mostly go back to the everyday language of the bilingual population, they include important sections of the basic lexicon as, for example, kinship terms like babă ‘grandmother’ ← baba, maică ‘mother’ ← majka, nevastă ‘bride, wife’ ← nevěsta.³² Other semantic classes are represented in the following list only by a few examples out of a tremendous quantity (Mihăilă, 1960; Popović, 1960, pp. 198–209; Puşcariu, 1943, pp. 345–361; Rosetti, 1984, pp. 283–291; Vojvodić, 2002):

society: boier ‘boyar, nobleman’ ← boljarinъ

body parts: obraz ‘face’ ← obrazъ

clothing: poală ‘skirt’ ← pola

food: drojdie ‘yeast’ ← ‘droždiję

agriculture: brazdă ‘furrow’ ← brazda

fishing: mreajă ‘net’ ← mrěja

housing, objects, and tools: prag ‘threshold’ ← pragъ, tocilă ‘grindstone’ ← točilo

time: veac ‘century, lifetime, eternity’ ← věkъ

nature and topography: dumbrava ‘forest’ ← dǫbrava

fauna: jivină ‘living being’ ← živina

flora: mac ‘poppy’ ← makъ

concrete terms: glas ‘voice’ ← glasъ

abstract terms: nădejde ‘hope’ ← nadežda

writing: cerneală ‘ink’ ← črьnilo

religion: rai ‘paradise’ ← raj

Other parts of speech were borrowed less frequently. Nevertheless, it is easy to find adjectives in the basic vocabulary of modern Romanian going back to the earliest centuries of the Slavic–Romance contact situation, such as drag ‘dear’ ← dragъ, scump ‘expensive’ ← skǫpъ, sfânt ‘holy’ ← svętъ. Adverbs and interjections include iute ‘vehement’ ← ljutě, prea ‘too (much)’ ← prě-, da ‘yes’ ← Blg./Srb. da.

Verbs of Slavic origin cover most semantic fields of Romanian, too. In formal respect, all Slavic verbs were integrated into the Romanian i-conjugation, with the ending -i(re) in the infinitive (-î- after stems in r-) and most frequently the esc-suffix in the present tense, such as a iubi (iubesc prs.1sg) ‘to love’ ← ljubiti, a plăti (plătesc prs.1sg) ‘to pay’ ← platiti, a pofti (poftesc) ‘to desire’ ← pohotěti, a risipi (risipesc) ‘to waste’ ← rassypati ‘to scatter’, a râni (rânesc) ‘muck out’ ← rinǫti ‘to push away’.

The direct source for matter borrowing was not the Slavic infinitive but, rather, the stem of the present—for example, a primi ‘to obtain’ ← prim- prs ≠ prinję(ti) inf. This is especially interesting in the case of the Slavic suffix -ov, alternating with the present in -uj as integration stem—for example, a darui (daruiesc) ‘to give’ ← daruj- prs ≠ darovati inf, a trebui (trebuie prs.3sg, obsolete trebuieşte) ‘to need, must’ ← trěbuj- prs ≠ trěbovati inf. In this case, the alternation of the infinitive with the present stem in Slavic goes back to the Proto-Slavic sound law of opening all closed syllables, here -ou̯ > -u/_C. The foreign ui- element was abstracted from borrowed verbs and became productive as an integration suffix also for loan verbs from other languages—for example, a cheltui ‘to spend’ ← Hungarian költ—although partially by means of Slavic mediation: mântui ‘to save’ ← Serbian mentovati ← Hungarian ment. It is productive even language-internally with Romance roots—for example, a preţui ‘to appreciate’, derived from preţ ‘price’ < Lat. pretium, căpătui ‘to tend’ derived from cap ‘head’, capete pl < Lat. caput.

As for phonetics/phonology, some Slavic characteristics have entered Romanian—for example, the word-initial pre-jotation of e as in el [jel] ‘he’. It has even been claimed that the Romanian consonant system has entirely been restructured by means of the contact-induced development of a correlation of palatalization in consonants, with implications also for the vowel system (Petrovici [1957], harshly criticized by Nandriş & Nandriş [1958] and Agard & Fairbanks [1958]). The Slavic phoneme h [x] was borrowed, although it is sometimes replaced by f or c [k]—for example, hâtru ‘witty’ ← xytrъ, praf ‘dust’ ← praxъ, cojoc ‘fur’ ← kožuxъ.

Slavic ě [æ] appears mostly as the diphthong ea—for example, hrean ‘horseradish’ ← hrěnъ. The Slavic nasal vowels ę [ɛ̃] and ǫ [ɔ̃] were substituted by oral vowels + nasal consonant—for example, grindă ‘beam’ ← gręda, munca ‘work, effort’ ← mǫka ‘torture’, mândru ‘proud’ ← mǫdrъ ‘wise’. The so-called jer-vowels ъ, ь either appear as full vowels, normally o, e, or they were lost (word-finally and in other weak positions)—for example, oţet ‘vinegar’ ← ocьtъ (← Latin acētum). Slavic y [ɨ] integrates as i: dobitoc ‘cattle’ ← dobytъkъ. All other Slavic phonemes, in principle, had direct equivalents in Proto-Romanian (Rosetti, 1984, pp. 303–317; Rothe, 1957, pp. 49–59).

These correspondences refer to Proto-Romanian times. In later periods, Slavic borrowings in some cases show other results, due to changes in the individual Slavic sound systems. For example, nasal vowels were denasalized in Bulgarian varieties, with corresponding results in Romanian borrowings—for example, ǫ > ă > î/â [ɔ̃ > ə > ɨ] as in OCS gǫsъ ‘goose’ > Blg. găs(ka) → Ro. găsca > gâsca.³³

In addition to the loanwords, matter borrowing includes also many affixes, which had been extracted from integrated Slavic words and became productive in word formation later on. This is especially true for suffixes such as -ac, -an, -anie, -ar, -aş, -ca, -in, -iţa, and many more. They are used to form agentive, augmentative, diminutive, singulative, deverbative nouns and to derive feminine nouns from masculine ones (gender motion), etc. The suffix -u, used for the integration and derivation of verbs in -ui, was already mentioned. As for prefixes such as po-, pre-, pro-, răz-, and za-, they express augmentative, intensifying, temporal, resultative, or inceptive meaning, often combined with a telicizing function. But in Daco-Romanian, there is no tendency to use them as grammatical aspect-forming morphemes, in contrast to Istro-Romanian (see section 3.1.4).

Linguistic calques (pattern borrowing) are the other important type of Slavic borrowings in Romanian. The main subtypes are loan translations and semantic calques. Semantic calquing is most evident in the case of parallel means of expression for different meanings. In the adstrate–superstrate situation, this is the realm of the adaption of the semantic structure of inherited Latin words to polysemies in the Slavic model. An example of such an adaptation is Latin lumen n ‘light’ > Ro. lume f, adopting also the meaning ‘world’ on the model of common Slavic světъ ‘light, world’. Another one consists in the adaptation of Latin vena ‘vein’ to the polysemic counterpart in Bulgarian (and Serbian) žila ‘vein, nerve’, with Romanian venă acquiring the additional meaning ‘nerve’.

Later on, new differentiations could occur. So, the borrowed meaning ‘world’ of lume is the normal one in the modern standard, while it has been replaced in its traditional meaning ‘light’ by lumină in the standard, and the French loanword nerve is used to disambiguate venă. Similarly, Latin vita ‘life’ > Ro. vită became ambiguous for ‘life’ and ‘animal’, due to the polysemic model of Slavic životъ. But it was replaced by viaţă ‘life’ in its original meaning, whereas vită has been reduced to the contact-induced meaning ‘animal, cattle’. In the case of joc ‘play, dance’ < Lat. iocum ‘joke’, whose ambiguity is due to the polysemic model of Bulgarian (and Serbian) igra, both meanings have been preserved in standard Romanian.

In the substrate constellation of Slavic speakers acquiring Romanian as their second language, to which they eventually switched, the semantic structure of their L1 is transferred directly to their Romance L2′—that is, polysemic words are replaced by Romance words but keep their original ambiguity. Thus, the final results for both types of language contact coincide.

As for loan translations, a typical case is frunt+aş ‘leader’, translated from čel+nik, with the components frunte = čelo ‘front’ + agentive suffix. Idiomatic combinations of words have been calqued from Slavic, too—for example, cale-vale ‘so-so’, literally “way (up) -down”, corresponding to Bulgarian gore-dolu, literally “up-down”.

Due to patriotic and purist tendencies, many Slavic terms in Romanian have been replaced, at least optionally, with Romance ones, partially by means of internal word-formation processes. In other cases, borrowings from French and, more rarely, Italian served as substitutes—for example, curaj ‘courage’ ← Fr. courage, secol ‘century’ ← It. secolo, optionally replacing veac.³⁴ New western terms were frequently introduced together with newly imported concepts—for example, avantaj ‘advantage’ ← Fr. avantage.³⁵ In addition, calques of all types have been formed on a French base—for example, bine inţeles ‘granted’, translating Fr. bien entendu or cunoştinţă, expanding its original meaning of ‘knowledge’ to ‘acquaintance’ on the model of Fr. connaissance (Puşcariu, 1943, pp. 474–525).

Another type of loan translation from Old Bulgarian is found in the numeral system. The numerals 11–19 were calqued from the Slavic source with the help of the preposition spre ‘on’, corresponding to Slavic na—for example, tre+spre+zece ‘thirteen’ (literally “three on ten”, cf. Bulgarian tri+na+deset). Incidentally, this formation of numerals is the only clearly Slavic-based Balkanism, present also in Albanian—for example, tre+mbë+dhjetë ‘thirteen’. But Romanian copied additionally the Slavic way to count the decades—for example, două+zeci ‘twenty’ (literally “two+tens”), corresponding to OCS dъva desęti, tri+zeci ‘thirty’ = OCS tri desęte. The numeral o sută ‘100’ is even a matter loan from OCS săto, in its turn a borrowing from Iranian.³⁶

A case of the typologically rare borrowing of a grammatical ending is the form of the vocative of nouns referring to humans and in proper names. Whereas masculine nouns express their vocative singular by means of the ending -e, which was either inherited from Latin or borrowed from Bulgarian, the ending -o of the feminines is clearly Bulgarian—for example, soro voc.sg.f of soră ‘sister’. The very preservation of a morphological vocative is a clear exception in Romance languages. It certainly goes back to the structural model of Bulgarian.

Romanian grammatical gender has been claimed to be influenced by Slavic, too, as the neuter developed into a gender of inanimates, which reminds of the existence of a grammatical category of animacy in Slavic, serving as a model (Rosetti, 1963). Actually, the Romanian neuter is restricted to inanimate nouns (apart from class-denoting terms such as dobitoc ‘domestic animal’ and the hypernym animal itself), although the opposite is not true. But the way in which it is expressed, to wit as an ambigender with a masculine singular and a feminine plural—for example, scaun ‘chair’ sg.m, scaune pl.f or animal sg.m, animale pl.f—is by no means parallel to the Bulgarian neuter, having forms of its own. Apart from this, the Slavic animacy opposition is based on case morphology: the accusative of masculine animates, coinciding with the genitive, is different from the nominative, whereas for inanimates there is as a nominative–accusative syncretism.³⁷ In addition, borrowed neuter nouns of Slavic origin did not integrate into the Romanian neuter but, rather, became feminine—for example, okno n → oknă f ‘window’, blato n → baltă f ‘marsh’, vrěmę n → vreme f ‘time, weather’ (contrary even to the neuter gender of its Latin-based synonym timp ‘time, weather’ n < Lat. tempus n). Another argument against the claim of the Romanian neuter being triggered by the Slavic animacy opposition comes from the fact that it structurally resembles the Albanian ambigenders, also mainly restricted to inanimates, such as mal ‘mountain’ sg.m ≠ male pl.f, corresponding to Romanian mal ‘river bank’ sg.m ≠ maluri pl.f. In addition, Albanian has a separate neuter gender besides the ambigenders, especially in traditional varieties and for derived abstract nouns—for example, djathë n ‘cheese’, të ngrënët n ‘the eating’. This means that the expansion of ambigenders could go back to pre-Slavic times, when the ancestors of today’s Albanians and Romanians still lived together in the Dardania region, or in other words, to the Thracian substrate, just like their partially common lexicon; see the article “Romance in Contact With Albanian,” 5.

On the other hand, inanimate loanwords of all genders in the source languages, indeed, have often become neuters in Romanian, a process still productive in modern times. Examples are French train m and minute f → Ro. tren ‘train’ sg.m / trenuri pl.f, minut ‘minute’ sg.m / minute pl.f. In any case, the formal expression of the Romanian neuters goes back to characteristics of the inflection of neuters in Latin, which are also the bases for such couples as Italian uovo ‘egg’ sg.m : uova pl.f. In the case of Romanian, this could just be a transfer of an old category to a new way of expression.³⁸

Bulgarian (in all its varieties) was by far the most important source of Slavic elements in Romanian. Serbian has been an important source of borrowings especially in the western Romanian-speaking territories. For the Romanian-based minorities in Serbia, Croatia, and Bosnia, the Serbo-Croatian varieties have even served as dominant standard languages. With respect to their linguistic and sociolinguistic situation, see Sorescu Marinković (2008) and Sikimić (2014).³⁹

3.1.2 Bulgarian and Macedonian Influence on Aromanian

The south-Danubian varieties also show manifold Slavic influences, part of them dating back to Proto-Romanian times. But Slavic influence going back to that period is more restricted in Aromanian (Macedo-Romanian) than in the other two languages of the Romanian linguistic family mentioned here, Megleno-Romanian and Istro-Romanian (Capidan, 1925; Scărlătoiu, 1980; Trummer, 1998, p. 169). This is explained by a relatively early migration of the Aromanians farther south (Greece, Macedonia, and Albania), separating them from the Proto-Romanian group.⁴⁰

The number of Slavic elements in Aromanian increased after the Proto-Romanian period as a consequence of lexical borrowings from Macedonian, especially with respect to the dialects spoken in Slavic surroundings. For example, the agricultural terminology of Aromanian in Macedonia is almost completely Slavic (Gołąb, 1984, p. 13).

3.1.3 Bulgarian and Macedonian Influence on Megleno-Romanian

Megleno-Romanian is spoken in only a few villages, mostly on the Greek side of the Bulgarian–Greek border. It has more Proto-Romanian borrowings from Slavic than Aromanian, probably due to its later emigration to the central Balkans. In addition, Slavic influence on Megleno-Romanian did not stop after this separation, due to the new contact situations with Macedonian.

Apart from the lexicon, two contact-induced developments in the domain of grammar show the high degree of Macedonian influence.⁴¹ The most famous phenomenon is loan endings in the first and second person singular of the present of the Megleno-Romanian verb—for example, aflum, afliş ‘to find’, corresponding exactly to the Macedonian endings -m, -š. But they are used only in part of the local dialects and only after consonant clusters ending in a liquid.⁴² Exclusively in the dialect of Ţărnareca, another ending has been borrowed from Macedonian, the gerund suffix -e̯ái̯ḱi, e.g. nirḑe̯ái̯ḱi ‘walking’ (Atanasov, 2002; Puşcariu, 1943, pp. 271–276).

3.1.4 Croatian Influence on Istro-Romanian

Istro-Romanian has been spoken on the Istria peninsula under Croatian linguistic domination for centuries. As a result, we find in the Istro-Romanian varieties far-reaching borrowing effects, not only in the lexicon⁴³ but also in grammar.

The Istro-Romanian system of numerals has been Slavicized to a great extent, with only the basic numerals remaining Romance; for the opposite development of a Romanized Slavic system of numerals, see section 2.2.1. A mixed system of Istro-Romanian numerals is presented in Sârbu and Frăţilă (1998, pp. 24, 40), with the numerals 1–7 of exclusively Romance origin (e.g., påtru ‘4’) and 8, 10, and 1,000 with Romance–Slavic variation (e.g., opt ~ osăn ‘8’). The numeral devet ‘9’, the teens (e.g., dvanaist ‘12’), and the decades (e.g., dvaiset ‘20’),⁴⁴ including sto ‘100’, are purely Slavic. But this view is a bit oversimplified. On the one hand, there is some dialectal difference; on the other hand, it has to be highlighted that variation occurs also with regard to the numerals 5–7, which, however, is not unconditioned (free). For example, the borrowed alternatives are restricted to combinations with borrowed terms indicating time or units of measurement. So, we get in the example ‘five years’ an opposition of Slavic-based pet let versus Romance činč anj (Kovačec, 1971, pp. 219–220).⁴⁵

An interesting case of grammatical borrowing is the emergence of neuter adjectives in -o. In a first step, Croatian neuter adjectives such as dobro n ‘good, well’, novo ‘new’ were borrowed, forming a gender opposition with masculine and feminine ones. This newly acquired loan gender was then conferred to stems of Romance origin such as bun ‘good’ m, bura f, here as buro n (Sârbu & Frăţilă, 1998, p. 23). In the beginning, the innovative neuter adjectives possibly did not agree in gender with nouns, but they were substantivized as in ceva slåbo ‘something bad’ or were used in subjectless sentences or as adverbs. This would correspond to the situation of neuter adjectives in Molise Slavic; see section 2.2.1. But later on, agreement with borrowed neuter nouns emerged, excluded in Molise Slavic.⁴⁶

The most striking effect of Croatian influence on Istro-Romanian may be seen in the development of a Slavic-style category of verbal aspect, based on aspectual pairs of perfective and imperfective verbs.⁴⁷ As it seems, aspect entered Istro-Romanian first by means of complete aspectual pairs borrowed from (Čakavian) Croatian, such as cosi/pocosi ‘to mow’ ipfv/pfv ← kosit/pokosit, formed by prefixation, and poberi/pobiręi ‘to collect’ pfv/ipfv ← pobrat/pobirat, formed by stem alternation.⁴⁸ As a consequence of borrowing also aspectual pairs formed by means of suffixation, Istro-Romanian has absorbed the complex morphonological system of Čakavian verbs, too, with all its derivational, suppletive, and inflectional alternations. See Kovačec (1971, pp. 125–130) for many oppositions of this type that read just like parts of a Čakavian–Croatian grammar—for example, (pfv/ipfv) obidi/obhaiui ‘to bypass’, oživi/oživl’ui ‘to liven up’, priscoči/priscacui ‘to jump (over)’, uloží/ulaɣui ‘to insert’. Together with the formal oppositions, the full range of their aspectual functions may have been copied, too.

In principle, borrowed Croatian prefixes allowed for an extension of the aspect opposition to Romance, as in the case of torce/ potorce ‘to spin’ ipfv/pfv or mânca/namânca ‘to eat’ ipfv/pfv. Romance–Slavic suppletive pairs, containing prefixed Slavic perfectives, also occur, such as bę/popi ‘to drink’ ipfv/pfv.

Things are more complicated for presumed aspectual pairs on the basis of Romance roots, perfectivized by means of Slavic prefixes and secondarily imperfectivized by suffixation, normally with the help of the borrowed suffix -(a/i)v, as in zadurmi-zadurmivęi ‘to fall asleep’. It is not excluded that such “imperfectives” are restricted to the expression of iterative/habitual states of affairs, which is almost sure in the case of derivations from verbs with a Romance prefix, which can hardly be conceived as perfectives, such as ânvešti-ânvescavęi ‘to dress’. They clearly resemble the iterative Aktionsarten, derived from simplex imperfectives by means of this very suffix in Čakavian, as in Čak. kopat : kopivat ‘to dig : to dig usually’.⁴⁹ Actually, such actional pairs have been borrowed as a whole, too, in the given case as copęi : copivęi. In addition, the whole iteration procedure has been calqued for Romance roots, resulting in oppositions such as durmi : durmivęi ‘to sleep : to sleep usually’.

As long as Romance-based secondary “imperfectives” cannot be demonstrated to express “ongoing processes” in opposition to “states of affairs conceived as a single whole,” at least for loan verbs, the threshold into ordinary Slavic aspect has not been crossed.

The Istro-Romanian infinitive ending -ęi typically occurs on verbs borrowed from Croatian. Language contact was involved also in the rise of this form itself, as the phoneme ę [æ] (preceding the infinitive ending -i) developed by a new interpretation of an original a after palatal consonants in loanwords.⁵⁰ Borrowed infinitives in -ęi are independent of the aspectual affiliation of the source verb, as is shown by copęi ‘to dig’ ← Čak. kopat ipfv or pisęi ‘to write’ ← Čak. pisat ipfv, on the one hand, and by ocopęi ‘to dig up’ ← okopat pfv, on the other hand. Where -ęi superficially seems to be an imperfectivizing suffix, it is always attached to stems already imperfectivized by other means, for example, by stem alternation or with the help of the suffix -(a/i)v. In rare examples such as hiti / hitęi ‘to throw’ pfv/ipfv, in which -ęi, indeed, has imperfectivizing (or iterativizing) function, this is not an independent Istro-Romanian development, but it goes directly back to the same opposition in the Croatian source, here Čak. hitit pfv : hitat ipfv.

Infinitives in -ęi have become productive for the integration of borrowings from non-Slavic languages as well, just like -uí in Romanian—for example, fruştikęi ‘to breakfast’ ← German frühstücken or skuzęi ‘to excuse’ ← It. scusare (Sârbu & Frăţilă, 1998, p. 32).⁵¹

In addition to -ę́i, -úi (stress is different from Romanian -uí) also appears in Istro-Romanian borrowings from Croatian. It goes back to source forms in -ova inf, -uj- prs like otputui ← otputovat(i) pfv ‘to leave’. Among them, imperfectives derived from perfectives prevail, just like in Čakavian—for example, ocopęi/ocopui ‘dig up’ ← Čak. okopat/okopovat pfv/ipfv.