Italo-Albanian: Balkan Inheritance and Romance Influence

This chapter deals with contact-induced change in Italo-Albanian and its effects on the Balkan inheritance of this minority language. The introduction is dedicated to the general characteristics of Albanian and its varieties from a historical, dialectological and geographic perspective, followed by a section on the historical and present situation of the Italo-Albanians. While Section 3 discusses the role of Balkanisms in Standard Albanian, Section 4 gives a general overview of the fate of these Sprachbund criteria in Italo-Albanian. In Section 5, contact-induced changes in the verb systems in single Italo-Albanian dialects are investigated, with special regard to the changes in the future tense, in the analytical perfect and verbal aspect, followed by a discussion of the innovative causative construction and other periphrases. Finally, changes in the domains of mood and voice in the Italo-Albanian dialects are described, most of them of recent date and, in part, not accepted by conservative speakers. All contact-induced developments in the minority language, as well as those parts of its grammar that have resisted foreign influence, are contrasted with their Standard-Albanian counterparts. As will be shown, many traditional Balkan features have been weakened or lost, whereas others have even expanded, but always in the direction of Romance models, to which Italo-Albanian functionally has adapted or which it has calqued.


Introduction
The history of the Albanians since the early middle ages has been a history of expansion and migration. From a very restricted territory in the Albanian mountains, with the Mati valley as its center, and possibly some other areas more to the east (Dardania) escaping complete Romanization, ancient Due to these developments, arb could be claimed a third variety besides Tosk and Geg or even a micro-language in its own right.2 The two most-cited isoglosses separating Geg and Tosk (including Italo-Albanian) dialects come from historical phonology and go back to two innovations in Tosk, the so-called "rhotacism" of intervocalic n > r and the loss of nasal vowels, developing into schwa in the case of the accented nasal â > ë [ə]; see for example Geg bâna 'to make aor.1sg': Tosk bëra or the Latin borrowing arena 'sand' , giving rânë in Geg and rërë in Tosk. Figure 1 is an overview of the main varieties of the Albanian language family, including Arvanitika in central and southern Greece, at present gradually dying out.
This paper intends to show to what extent foreign Romance influence on Italo-Albanian has led to a loss of the Balkan traits Albanian traditionally has in common with other Balkan languages like Macedonian, Bulgarian, Romanian (and its south-Danubian varieties) and Greek, but also of specific Albanian inheritance, not found in the other Balkan languages. Cases of resistance against the foreign influence will also play a role.3

The Historical and Present Situation of the Italo-Albanians
As it seems, internal migration of Albanians in Italy occurred already immediately after their arrival in their new home country. People from different places mixed up, which means that there is no clear-cut correlation between specific places in their original home and the individual settlements in Italy. The same is, of course, true for their language/dialects. As of 2020, about fifty villages in southern Italy still have an Albanian-speaking population; others like Brindisi di Montagna in Basilicata and Mezzoiuso in Sicily as well as (only in the last decades) Villa Badessa in Abruzzo have lost it. The greater part of these municipalities concentrate in the Calabrian province of Cosenza. Italo-Albanian as a whole has transformed considerably with respect to the lexicon, but grammar has also been affected, mainly by means of the influences of Italian and its local varieties. This micro-language is, however, by no means homogeneous, mainly due to differences in the dialects of neighboring Romance-speaking villages, serving as models for contact-induced change. In the present article, I will in some cases refer to these differences, but I will also try to elaborate on common characteristics, in which Italo-Albanian as a whole differs from the Balkan-Albanian varieties. Nevertheless, many examples are based on single dialects, especially those of Frascineto in the Province of Cosenza (northern Calabria)4 and of Portocannone in the Province of Campobasso (Molise). The number of Italo-Albanian speakers in their traditional municipalities has continuously diminished in the last fifty years, due to their emigration to neighboring cities and to the North, in addition to language shift that has been especially strong in the last decades. Unofficial estimates run from twenty up to fifty thousand persons, using this micro-language with a certain degree of competence. The demographic figures of the official censuses normally do not consider linguistic data. Therefore, they do not reflect the real number of Italo-Albanian speakers, due to the influx of monolingual Italians, especially in regions near to the coast, in addition to language shift/loss. Nevertheless, they give some information as to growth and loss of the population in Italo-Albanian municipalities during the last 150 years. The official data of the Italian National Institute of Statistics istat from Piana degli Albanesi (Sicily), Frascineto (Calabria) and Portocannone (Molise), summarized in Figs 2-4, may give some impression of the demographic development in these three municipalities. For the sake of comparison, the data for Montecilfone (Molise) are added in Fig. 5.
The difference between communities near to the coast like Portocannone, with their strong influx of monolingual Italians, and those in the hinterland like Montecilfone, likewise in Molise, becomes evident from the statistics for the latter (Fig. 5), with a continuous decline of the number of its inhabitants since 1951.5 4 For a first overview of the dialect of Frascineto, see Breu and Glaser (1979). For the Portocannone dialect, see the dictionary of Pignoli and Tartaglione (2007). An all-embracing comparison of the arb-varieties is still missing. But see the work of L. Savoia for many details, for example Savoia (2008) and Manzini and Savoia (2015). 5 For a more detailed analysis of the Molise Albanian demographic situation (compared with neighboring Molise Slavic), including attempts at estimating the number of native speakers, see Breu (2018a).  arb speakers are in a situation of "total language contact", which means that they are all bilingual, with Italian (and the local Italo-Romance varieties) dominating their vernacular as an umbrella language (Dachsprache), including official use and writing.6 To a growing extent Italian even enters their everyday communication. Actually, there is a great deal of variation with respect to language behavior and preservation in the different places, for example, with the arb dialects in Molise, Campania, Apulia, and south-western Calabria being severely endangered, whereas the varieties in northern Calabria and the adjacent part of Basilicata (both of them conserving the Orthodox Greek liturgy) are relatively stable. See, for example, Breu (1991b) for an overview of the language behavior in most villages of the northern part of the Arbëria thirty years ago and Savoia (2010: Section 4), Breu (2018a) for a discussion of the actual linguistic and sociolinguistic situation in the Albanian villages of Molise.

Balkanisms in Albanian
In order to evaluate the role of linguistic change in arb with respect to its Balkan features, the traditional position of Albanian in the Balkan linguistic area (Sprachbund) has to be considered. The best-known Balkan features (Balkanisms) found in most Balkan languages, though with variations, are:7 figure 5 Demographic data from Montecilfone, 1861-2011 6 For the term "total" (or "absolute") language contact, see, for example, Breu (2011), especially Section 3, andBreu (2019: 385-386). 7 Many hypotheses exist concerning the development of these features, including a common Balkan substrate as well as Greek and Balkan-Romance influence. The concrete realization of the single Balkanisms in the different languages is by no means homogeneous, just like their presence or absence in a given language. The properties of a Balkansprachbund, if it exists at all, have been treated controversially in the literature; see, for example, Sandfeld (1930), Solta (1980), Fiedler (1989), Hinrichs (1999), Friedman (2006), Tomić (2006), Sims (2008), Kahl et al. (2012), andTrumper (2020). For an updated overview, see also Gardani, Loporcaro, and Giudici (2021). 1) The existence of a central vowel (schwa) 2) A common Balkan lexicon 3) A postponed definite article 4) Fossilization of modal verbs to particles based on the form of prs.3sg 5) Formation of numbers 11-19 of the type "one upon ten" 6) Object doubling by means of clitic pronouns in the dative and accusative 7) A completely analytical system of comparison (adjectives, adverbs) 8) Very complex verb systems, including hypotactic particle constructions 9) A modal perfect 10) Lack of an infinitive 11) Future formation with the volitive modal particle based on 'will (to want)' 12) Lack of a nominal declension a) center: genitive-dative syncretism b) but: morphological expression of a vocative Albanian shows almost all of these properties, though, of course, in their specific Albanian form. Therefore it has been claimed to be the most typical Balkan language; see Fiedler (1992) and Beci (2012). It has a stressed schwa (at least in Tosk and the standard), unlike, for example, Greek (feature 1), it has many Balkan words in common with Romanian, which, in part, are missing in other Balkan languages (feature 2), it has a postponed definite article unlike Greek (feature 3), it has more fossilized modal particles than, for example, Bulgarian and Macedonian (feature 4).
Object doubling by means of dative and accusative clitics is by far more grammaticalized than in Greek (feature 6). Likewise, in contrast to Greek, there is no alternative to the analytical system of comparison of adjectives and adverbs (feature 7) of the type më shumë 'more (much)' . The Albanian verb system is especially complex in the domain of mood, for example, in showing a morphological optative of the type qoftë 'may it be' and a jussive, expressed by means of the particle lë + subjunctive, e.g., lë të jetë 'let it be' , unlike all the other Balkan languages (feature 8). This complexity also includes the morphological mediopassive of the type blehet 'is (being) bought' , which Albanian has in common only with Greek and which normally does not figure among the Balkanisms. Unlike Greek and Romanian, it has a modal perfect, in its specific realization as an admirative of the type qenka 'oh, s/he is' (feature 9), differing from the perfect, here ka qenë 'has been' , mainly in the order of its components.
In contrast, the lack of an infinitive (feature 10) only concerns Tosk,8 whereas Geg has an analytical infinitive of the type me bâ 'to make' . But unlike Romanian, Albanian as a whole indeed does not have any synthetic (morphological) infinitive.9 A similar relativization holds true for the future, formed exclusively with the modal particle do 'will (to want)' + subjunctive (feature 11) in Tosk (and in the standard), whereas Geg traditionally prefers the inflected auxiliary kam 'to have' + its analytical infinitive, for example Tosk do të punojë, Geg ka me punue 's/he will work' .
As for the lack/loss of the nominal declension (feature 12), Albanian, unlike Balkan Slavic and Romanian, restricts the genitive-dative syncretism only to the morphological level, while syntactically these two cases are kept apart by means of the obligatory genitive connector,10 missing in the dative (feature 12a). In general, Albanian has preserved case inflection, unlike Balkan Slavic and even more than Greek, as it has a distinct ablative, e.g., grash abl.pl.indf 'of/from women' , and some dialects additionally show a locative, for example në malt loc.sg.def 'in the mountain' (Gjinari, 2007: 263). But there is no vocative case, at least not in the standard and in the everyday vernacular (feature 12b), with the exception of rare agglutinative forms with postponed -o for persons, documented predominantly in folkloristic texts, e.g., biro voc.sg.m for bir 'son' (Weigand, 1913: 28). 8 Even here, the final clause with për 'for' + subjunctive particle + participle, e.g., për të punuar '(in order' to work), is sometimes referred to as "infinitive". 9 The Romanian short infinitive, the only one used, for example, in modal constructions, is accompanied in many cases by the preposition a, for example, a cânta 'to sing' , but it also appears without it, for example, in the analytical volitive future as in voi cânta 'I will sing' or when governed by a putea 'can' as in poţi cânta (in variation with the subjunctive poţi să cânţi) 'you can/may leave' . The so-called long infinitive, here cântare '(the) singing' , is mostly used nominally. 10 Terminology varies with respect to this formally article-like linking element. Though it certainly is not an article, its form depends partially on definiteness, expressed by the definite article, in addition to other criteria like case and number agreement as well as word order, e.g., burrat (

Overview of the Fate of the Balkanisms in Italo-Albanian (arb)
Italo-Albanian continues, in principle, to have a stressed schwa, often slightly nasalized, as in zë [zə] 'to take' (Balkan feature 1), whereas unstressed schwa has been given a new interpretation as a euphonic element in consonant clusters or was replaced by e. But at least in Calabria, many speakers, in some cases even whole dialects, replace stressed ë consequently with o, e.g., ësht ~ osht 'is' , ndë ~ ndo 'if' . Some lexical Balkanisms of pre-Turkish times, especially substrate terms and Latin borrowings continue to exist (feature 2),11 whereas others have been replaced by Italianisms. Phrase-like compounds like kush isht isht 'whoever it might be' , literally "who (it) is, (it) is", have often been calqued from the Romance neighbors, which -just like in the case of lexical borrowings -in arb attributes Italian and its varieties the same role Turkish had for the languages on the Balkans. The postponed definite article has been kept in all its functions and continues to be inflected (feature 3). The Italian preponed article has not been borrowed, nor has its position been calqued.12 The traditional formation of the numerals 11-19 continues to exist in arb (feature 5), e.g., njëmbëdhjet 'eleven' , trembëdhjet 'thirteen' , but it is severely threatened by Romance borrowings of the type dhudhëç 'twelve' , kuindhëç 'fourteen' , corresponding to Italian dodici, quindici in their local phonetic form (here Portocannone). As for the Albanian vigesimal numeral system, arb has even expanded it to trizet 'sixty' and katërzet 'eighty' , including odd decimals, missing in modern Balkan Albanian or, perhaps, it has preserved an older more complete system. Examples from Portocannone are: trizet e di 11 Pre-Latin substrate terms are normally restricted to Albanian and Romanian (Solta, 1980: 39-63), e.g., Alb./arb brez (Rom. brîu) 'belt' and katund (Rom. cătun) 'village' , conserved in Calabria, but not in Molise, where katund has been replaced by hor, borrowed from Greek. Turkish elements in the Balkan lexicon, like Alb. bojë 'color' , do not play any role in arb, apart from rare third-party borrowings (Mandalà, 2012), but Greek borrowings are even more frequent in Italo-Albanian than in Standard Albanian, due to the contact situation in Greece in pre-emigration times, e.g., parathire 'window' , dhjovasenj 'to read' . 12 Italian articles appear, however, as parts of loanwords, especially in geographical denominations like L-amerka ← It. l' America, but also l-universita ← It. l'università 'the university' . Moreover, there are also rare cases of article-like functions of the Italian article attached to borrowed adjectives. Compare, for example, in the Portocannone dialect lunku ← It. l'unico 'the unique' in the definite phrase lunku male çë kish kurra njohur 'the only mountains which he had ever known' with the indefinite adjective unku ← It. unico 'unique' in ti bëhe për mua unku te shekui 'you become for me unique in the world' . For comparable cases in Sicily, see Mandalà (2005: 17). -The arrow "←" symbolizes borrowings, in contrast to phonetic or phonological developments, indicated by ">", "<". 'sixty-two = three score and two' , njëzet e dhjet 'thirty = one score and ten' , dizet e katërmbëdhjet 'fifty-four = two score and fourteen' . Romance borrowings are, however, excluded from this system, e.g., only trendun 'thirty-one' , sëtandaduj 'seventy-two' , etc. are possible and not *trizet e dhudhëç.
Object doubling in the dative and accusative continues to exist in arb (feature 6), but the specific rules seem to be slightly different, at least for the accusative. The preservation of object doubling by means of clitic pronouns and how it applies seems to be connected with the system of object doubling in southern Italian dialects. Further research has to be done in this field, in order to come to robust conclusions. In contrast, nothing has changed with regard to the analytical system of comparison (feature 7).
As for the verb system concerning features 8 through 11, arb has remained rather complex, but with adaptations to the Romance tense, voice and aspect systems. Hypotactic particle constructions have even remarkably multiplied, perhaps as a reaction to the missing infinitive in contact with Italian varieties, having an infinitive. In many arb dialects, the perfect has reduced its functions to a special type of presumptive and there is no volitive future, which in both cases seems to be due to contact-induced changes in the meanings of the verb kam 'to have' . These developments will be described in more detail in Sections 5.1 and 5.2.
With respect to the declension of nouns, it has, in principle, been preserved, in spite of the contact with caseless Romance varieties. Even the ablative has been kept, although with a reduction of its functions, and the locative, missing in standard Albanian also continues to be used. The morphological genitive-dative syncretism has not been broken up either (feature 12a), but the syntactic differentiation by means of the genitive connector has been weakened, in terms of a growing optionality of this element. On the other hand, there is a slight tendency to express the genitive analytically, i.e., based on the Italian prepositional model, but only in relatively fixed expressions with loanwords. Interestingly, in such cases the borrowed preposition dhi or dhë is used, corresponding to both Italian di 'of' and da 'from' , which means that it also expresses ablative functions.13 Based on a southern Italian model, a neo-vocative developed, typically expressed by means of omitting all sounds following the stressed vowel of proper names and other denominations of persons, e.g., Luixhi nom ≠ Lui voc 'Luigi' . Thus, in a way, the Balkanism of a vocative (feature 12b) has been introduced by means of language contact, although in a very special form.

Contact-Induced Changes in the Italo-Albanian Verb Systems
In a situation of total language contact like that of arb as a minority language, the concept of developing a grammatical diasystem becomes important, which claims that the grammars of the two languages in contact should become as uniform as possible. Two main procedures contribute to the development of such a diagrammar, the "adaptation of the semantic structure" (semantic calque), mainly by means of copying polysemies of the minority language to the majority language, and "loan translation" of periphrastic elements (syntactic calque). In the following, the effect of these procedures in the arb verb system will be shown.14

The Fate of the Future Tense
The developments in the domain of the future tense are a typical case of the adaptation of the semantic structure of arb as a recipient language, with regard to both lexicon and grammar. First of all, it should be noted that Italo-Albanian as a whole does not have a volitive Balkan future,15 unlike the Tosk branch of Albanian. Given the predominantly southern provenience of the ancestors of modern Italo-Albanians, a connection of their kam future with the Geg one seems to be excluded.
If a volitional future existed in older times, the first step of the adaptation of the semantic structure to the southern-Italian future occurred in the lexicon, in so far as the verb kam 'to have' copied16 the polysemy of its local Romance counterpart avé 'to have, must' , thus acquiring an additional deontic meaning. The second step then regards grammar. More precisely, arb calqued the deontic future of the surrounding Romance dialects with deontic avé as its auxiliary, thus getting a de-obligative kam future. As arb did not have an infinitive, it could not follow the Romance model completely, but had to replace it in its usual way by combing kam hypotactically with the main verb in the subjunctive.17 arb dialects differ with respect to the usage of the kam future, depending on the degree of deontic connotation. It seems weak in Calabria, but relatively strong in Molise. As a consequence the kam future is used freely, for example, in Frascineto (1a), whereas in Portocannone it is only possible if there is a high degree of necessity for the scheduled state of affairs or if the speaker intends to express future reference unambiguously without using a future adverbial as in (1b).18 Otherwise in both dialects the present tense is used to express the future, which is possible in local Romance dialects and in Standard Italian, too. For differentiating the future from the present, if necessary, time adverbials are added. At least in Frascineto, the kam future is also very common in epistemic statements like (1c).  Johanson (2002). In the present paper it is restricted to the special case of transferring polysemic models from the dominant to the replica language as the reason for grammatical change, based exclusively on the semantic/functional level without any interference of concrete forms. 17 Note that the forms of the future auxiliary do not necessarily coincide completely with the paradigm of the full verb kam. In Frascineto, for example, the auxiliary loses its original final vowel in the 1st and 2nd person plural, with kemi > kem and kini > kin, and in the 2nd and 3rd person singular the particle kat is used, instead of ke t, ka t. In Portocannone, the particle kat is even used throughout the future.
(Frascineto) there for ptl make.3pl procession.acc.sg.def 'There are many people out there' -'They will (probably) be there to make a procession.' The de-obligative kam-future was documented in Italo-Albanian very early. It shows up already in Matranga's Siculo-Albanian Catechism of 1592. As a consequence, some authors claim this type of future to go back to a common usage of a kam-future in Old Albanian, not restricted to the North (Altimari, 2005). Besides the problem that the arb type of this future is characterized by the subjunctive of the main verb and not its infinitive as in Old Albanian, it seems important in this respect that the kam-future is completely absent in Greek Albanian, which for historical reasons should be arb's nearest relative. It only has a volitional future (Sasse, 1991: 227-228, 416-417). As older texts in the Arvanitika varieties are missing, it is, however, unknown, whether the de-obligative future existed there at the time of the emigration of the later Italo-Albanians.
In any case, it may be concluded that even if an older kam-future survived in Italy, it was not replaced by the volitional future, due to the existing Italian model, whereas in the Balkan-Tosk varieties the Greek volitional future was calqued, which eventually could be the overall source of this Balkanism. Be it as it may, the result is the same: Italo-Albanian does not have this Balkanism, either due to contact-induced innovation or contact-induced preservation.19

5.2
The Fate of the Analytical Perfect The Albanian (active) perfect was already traditionally formed with the help of the auxiliary have, in combination with the participle. So there was no need of a formal adaption to the structure of the Romance perfect, formed in the same way. But the first, lexical step of the meaning extension of kam to its newly acquired deontic meaning has to be considered in this case, too. Actually, a type of linguistic drift occurred, in the sense that the polysemy of kam did not only affect the formation and deontic connotation of the future but also the semantics of the perfect. In other words, the meaning of the perfect auxiliary kam 'to have' extended to 'must' . As a consequence, an originally indicative perfect like ka bën 's/he has done' acquired a deontic reading, too. As it referred to the past and not to something still to happen, it developed a strong epistemic connotation, in the given case 's/he must (have) done' , which eventually became its dominant meaning.20 As a result, the perfect lost its indicative meaning altogether and turned into a new mood, a "presumptive" referring to the past. This development, not occurring in the local Romance dialects in question, was an internal change, clearly supported by the fact that the perfect in the contact varieties had only a very restricted function, excluding its usage as a (perfective) past tense, which was expressed only by the aorist. So, the arb aorist could easily take over the remaining indicative functions of the perfect.
A simple example of a presumptive in the dialect of Frascineto is (2a), resulting from an assumption of the speaker, concluded from the given situation. If he had personally observed that the rain had stopped, he would have used the aorist pundarti instead of the perfect presumptive ka pundartur. The short dialogue between mother and son in (2b) presents both forms, the aorist, referring to the past without any connotation, and the presumptive perfect, expressing a (present) epistemic assumption regarding an event in the past: At first a mother claims to not have seen a certain photograph before. Her (passively bilingual) son contradicts her in Italian, by using the Italian (indicative) perfect. In her reply she gives in, by using the presumptive (with its epistemic perfect meaning).21 20 In addition to this basic motivation for the transition of the perfect from the indicative to the modal domain, a second one came from the epistemic future perfect, which originally had the structure type have (aux inflected) + have (subjunctive) + participle, e.g., ka t ket bën 's/he will have done' . In this construction the subjunctive t ket was deleted, probably due to the model of the deletion of infinitives of auxiliaries governed by modal verbs like volere 'to want' in the Romance contact varieties (Rohlfs, 1969: 131). In our case this means: deve aver fatto 's/he must have done > deve fatto = arb ka bën. Consequently, the future perfect became identical with the (present) perfect and transferred its epistemic function into this construction. Actually, the original future perfect has become obsolete, for example, in Frascineto. So, instead of *ka t ket fërnuar t bjer shi 'it will have stopped raining / it must have stopped raining / it probably has stopped raining' , only the presumptive perfect ka fërnuar (without the subjunctive t ket) is used, keeping the epistemic meaning of the future perfect (Breu, 1991a: 57-59 Only in part of the Italo-Albanian dialects is this change of the perfect into a presumptive observed. But as Map 1 shows, it is by far the greater part of the Arbëria from Molise down to northern Calabria.22 In the remaining parts of the Arbëria the functions of the perfect changed as well, but in a different way. More precisely, there are two different areas, in which either the temporal functions of the perfect have expanded in such a way that it has replaced the aorist (south-east), or in which the perfect has reduced its functions within the aspectual domain to an experiential perfect (extreme south).23 Map 2 (based on Google Maps) shows the two isoglosses in Calabria, differentiating the modal (M) presumptive area in the northern part from the temporal (T) area in the south-east and the aspectual area (A) in the extreme south.
The different developments could be connected with differences in the structure of the arb dialects, but also with linguistic differences between the respective contact varieties. As for internal arb differences contributing to this areal tripartition, the most important feature is certainly the formation of the mediopassive, which in the presumptive area is formed with kam (preceded by the reflexive particle u), just like the active voice, and where consequently the perfect has developed its presumptive meaning, too, e.g., u kan njohur 'they probably have got acquainted' (M). However, where the mediopassive is formed with the auxiliary jam 'to be' , e.g., jam njohur like in Standard Albanian (see below), the transformation of the perfect to a presumptive was completely blocked (T). An additional factor to be considered is certainly the distribution of perfect and aorist in the Romance contact dialects. In most parts of the arb presumptive area, complete events in the past can only be expressed by the aorist in Italian. In the extreme south (southern Calabria and Sicily), the aorist even comprises the resultative function, leaving for the perfect only its experiential function. This was the exact model for the reduction of the arb perfect to an aspectual experiential. In contrast, the Italian dialects in the south-eastern coastal areas, for example in Crotone, seem to have a tendency towards replacing the aorist by the perfect.24 This could be the reason why in the south-eastern arb dialects, the perfect (whose development into a presumptive was blocked by the above-mentioned jam-perfect in the mediopassive) in the course of the last century has become a past tense grammeme, replacing the aorist.25 Interestingly enough, in Calabria the arb isoglosses dividing the kam-perfect into three types seem to coincide also with other Romance isoglosses, especially with the isogloss separating the infinitive area from the southern area, where the infinitive (under Greek influence?) is avoided (Breu, 2015: 228). If this is just a coincidence has still to be investigated.
In any case, language change in the domain of the traditional arb perfect is completely free of any Balkan heritage. If it is true that the development of a presumptive, i.e., a modal perfect, typologically unites Italo-Albanian with 24 For the historical distribution of the perfect (passato prossimo) and the aorist (passato remoto) in Italy, see Rohlfs (1969: 45-49); modern usage in local Italian is documented by the survey of Bertinetto and Squartini (1996) on the usage of these grammemes in eleven towns throughout Italy, respecting, among other things, various types of texts. Unfortunately, the network of informants in this survey is not fine-grained enough for the Arbëria, omitting for example the Crotone area. But it confirms at least the survival of the dominance of the aorist in the south, though Palermo and especially Cosenza show deviations. As for Crotone, informants confirm the preference of the perfect over the aorist in referring to past events. For the modern situation in Sicily, with influences from the (northern) standard but still a strong preservation of the old distribution of aorist and perfect in informal dialect-based usage, see Alfonzetti (2018). 25 It is worth noting that presumptive dialects are found also in places, where the Romance dialects have replaced the aorist by the perfect, as is the case in the coastal area of Molise.
In such cases the most probable explanation is the assumption of a migration of the arb population from more central territories to these areas in historical times, when they had already developed their presumptive, or, at least, prior to the changes in the corresponding Romance varieties. those Balkan languages having modal perfects, too, it is also true that this is a bare coincidence and, what is more, valid only with respect to part of the arb dialects. But just like in the Balkans, where the Bulgarian and the Macedonian renarrative (quotative) goes back to Turkish influence and should be clearly separated from the Albanian admirative, probably also due to Turkish influence in a different setting,26 the Italo-Albanian presumptive has been induced by language contact based on a totally different model, in this case induced by a non-Balkan language.27

5.3
Verbal Aspect In the domain of verbal aspect, beyond the genuine-Slavic derivational opposition of perfectivity expressed by prefixes and suffixes and preserved in Bulgarian and Macedonian, the Balkan verbal systems traditionally had much in common with the Romance opposition between imperfect and aorist in the past tense. On the other hand, analytical aspect forms are traditionally absent in the Balkan Sprachbund, which is, so to speak, a negative Balkanism, in which modern Albanian, however, does not participate. Actually, Albanian shows at least two analytical progressives, the gerund construction formed by means of the "auxiliary jam 'to be' + converb particle duke + participle", e.g., është duke punuar 's/he is working' , and the construction with the particle po, added to the present and to the imperfect tense, e.g., po punonte 's/he was working' (Buchholz and Fiedler, 1987: 167-169).
Gerund constructions continue to express simultaneity in Italo-Albanian, but mostly reduced to adverbial phrases without an auxiliary, just like in Italian, often replacing temporal subordinate clauses. The converb particles used differ locally; see example (3) from Frascineto with the particle ture. The particle po is mainly restricted in arb to periphrases with the function of an imminentive (see below). i.e., as the last residual of an originally general Albanian development of the perfect, connected with the admirative in modern Albanian and the renarrative in Balkan Slavic. Contrary to Altimari's opinion, its distribution in Italo-Albanian clearly forms an isogloss, in spite of the scattered settlements, and the principal areal influence of the Romance dialects is obvious. I admit that in some points newer developments inside Romance and the abovementioned internal migrations of the Arbëreshë have blurred this picture to some extent; see Breu (2015) for more details. In contrast, new progressive periphrases have been formed, based on southern-Italian models,28 the most frequent being coordinated or subordinated constructions with the auxiliary jam 'to be' , for example in northern Calabria and in Molise as in (4a)  In Frascineto, a special paradigm of the auxiliary jam 'to be' is used in the coordinated progressive, conflating *je e to je (prs.2sg), *jemi e to jem e (prs.1pl) and *jini e to jin e (prs.2pl), resulting in the following present-tense paradigm: jam e hin, je hin, ësht (~osht) e hin, jem e himi, jin e hini, jan e hinjen 'I am entering' etc. That this is not simply a case of phonetic simplification, but may be claimed a step towards (morphological) grammaticalization, is demonstrated by the fact that apart from the domain of the coordinative conjunction, an e does not conflate, e.g., jam e e bie 'I am beating him' , where the second e is the clitic pronoun 3sg.acc 'him/her' . But the most evident sign of grammaticalization is the reduction of the auxiliary to an uninflected particle ish in the imperfect, derived from the inflected form of the 3rd person singular, e.g., ish e prisja 'I was waiting' , ish e prisjem 'we were waiting' .30 Another way of expressing processes is based on adverbs with the meaning 'now' , e.g., nanthi in Portocannone, combined predominantly with the present of the verb, e.g., nanthi zgjohem = jam e zgjohem 'I am waking up' , thus calquing a similar usage of southern-Italian mo 'now' .31 Besides the different kinds of progressive periphrases, there is an actional periphrasis expressing approaching events (preliminary phase). This so-called "imminentive" may under certain conditions also refer to already on-going processes, but only shortly after their beginning. It is formed in the dialect of Frascineto by means of the fully inflected auxiliary jam 'to be' + particle po or preposition pë(r) + subjunctive, initiated by the subjunctive particle t(ë), e.g., ësht (prs.3sg) po ~ pë(r) t partirenj (prs.subj.3sg) 's/he is about to leave / she is leaving' . At least the prepositional variant is a calque of the Italian model sta per partire 's/he is about to leave' , etc.32 The arb developments in the domain of aspectuality are clearly independent of any Balkan heritage. But the general tendency of Albanian towards the formation of analytical aspects, untypical for the Balkan Sprachbund as a whole, has possibly promoted the acceptance of Romance aspectual constructions by the Italo-Albanian varieties.33 30 The progressive is observed in other tenses, too, for example in the future, as in Kat jet e bjer shi 'It will be raining' , mostly understood as an assumption. In this case the future particle kat (from ka = have.prs.3sg + subjunctive particle t) requires the subjunctive in both verbal parts of the coordinated progressive, jet 'be.prs.subj.3sg' and bjer 'fall.prs.subj.3sg'. 31 A parallel usage is found in Molise Slavic, in this case by adding sa 'now' , mainly to the present tense forms of the main verb, e.g., sa parčivam 'I am leaving' (Breu, 2011: 447), in contrast to the auxiliary-based progressive periphrasis, missing in this micro-language. Note that in neither of the two minority languages the Italian progressive of the type sto partendo 'I am leaving' exists, formed with the help of the auxiliary stare 'to be, to stay' + gerund, probably due to its absence in the local Romance varieties. 32 But note that in the imminentive, just like in the progressive periphrasis, the auxiliary jam 'to be' is used, essentially corresponding to Italian essere 'to be' and not rri 'to stay, to be somewhere' , corresponding to stare in Italian. This is different from Molise Slavic, showing a parallel calque of the Italian imminentive, for example, stoji za partit 's/he is about to leave' , but with the fully corresponding auxiliary stat 'to stay, to be somewhere' (and not bit 'to be') + za 'for' + infinitive. Actually, in place of jam 'to be' , used in the progressive and the imminentive in Calabria and Molise, the auxiliary rri 'to stay' also appears in Italo-Albanian, more precisely in San Marzano di San Giuseppe (Apulia, in contact with Salentino varieties) and in Barile, Ginestra, Maschito (Basilicata); see Savoia (1991: 395-396) and Altimari (2020) for more details. At least the progressive cannot be separated from the particle constructions to be discussed in the following paragraphs. This follows, for example, from its tendency to form more or less particle-like forms of the auxiliary jam 'to be' . The same is true for the auxiliary kam in the formation of the future.

5.4
The Development of a Causative The causative as an analytic grammeme is a new development of Italo-Albanian, too. With regard to its formation, the causative construction presents a wide spectre of possibilities, which all have in common the usage of bënj 'to make' , mostly fossilized as a particle (causative marker), in combination with a linking element and the inflected form of the main verb.34 In Standard Albanian, causation is expressed by means of the fully inflected verb bëj 'to make' , governing the main verb in the subjunctive.
Simple examples from Frascineto with intransitive (or intransitively used) main verbs are (5a) and (5b). Here the causative particle bën, originally the form of the 3rd singular present, combines with the coordinative conjunction e 'and' and the main verb. Variants of bën in Frascineto are bon ~ bin.35 The inflected main verb bears all grammatical information, i.e., besides the reference to person, number and mood also tense/aspect characteristics like "present tense" in (5a) und "aorist" in (5b). The agent of the causation is not indicated in the construction proper. It may be expressed optionally, like u 'I' in (5a-b), but normally it is concluded from the wider context. In the imperfect of these two sentences the corresponding form would be bën e pinej (~pij; iprf.3sg) and bën e partirnej (iprf.3sg), with the same fossilized particle. Even in a dialect, in which the causative particle incorporates the subjunctive particle t(ë), the main verb in the causative construction may be coordinated with the causative particle, as for example bit (< bën + të) in Portocannone. But, interestingly enough, the main verb in this dialect remains in the subjunctive in spite of being preceded by the coordinative conjunction e; see jet (be.sbjv. prs.3sg) in (7a).  This strange combination of coordination and subordination may also be the reason for the possibility of a variation between nominative and accusative case in the object of the causation (= subject of the main verb), at least for some speakers. Thus in (7b), instead of the definite nominative prinxhëpi, the corresponding definite accusative prinxhëpin was also claimed to be acceptable.36 Further research on the causative construction of Portocannone seems an important task. At present, it could be classified as an intermediate stage between the original Albanian construction with the fully inflected causative verb bëj 'to make' , governing the subjunctive and the causative in the Frascineto dialect, in which the causative marker (particle) bën ~ bin ~ bon has lost all additional grammatical properties.
There are Italo-Albanian dialects still nearer to the original Albanian con-struction37 than that of Portocannone, for example, the arb dialect of San Nicola dell' Alto in the north-eastern Calabrian Province of Crotone. Turano (1989) gives a detailed description of the causative of this dialect. For the argumentation and the examples presented in the following, see Turano (2005: 24-26).
In San Nicola dell' Alto the causative verb has completely kept its inflection and still governs the full subjunctive, subjunctive particle included, and without any coordinative conjunction. But regarding the syntactic characteristics of the causative constructions it has nevertheless adapted to Italian. While in a Standard Albanian example like (8a) the object of causation djalin 'the boy' is in the accusative, doubled by the clitic object marker e, and immediately follows the causative verb, the only possibility in the San Nicola dialect is (8b). Here the object of causation jali is in the nominative (not doubled by the object clitic) and located at the end of the construction, thus functioning as the subject of the subordinate subjunctive clause. Actually, a construction similar to (8b) is possible in Standard Albanian, too, but in San Nicola it is the only one. A third type of causative construction found in Standard Albanian, i.e., by means of a complement clause like Mësuesi bën që djali të lexojë librin 'The teacher makes the boy read the book'(literally: The teachers makes that the boy reads the book)' , introduced by the complementizer që 'that' (which would be çë in arb) + subject + subjunctive, is also excluded in San Nicola. On the other hand, just like in the dialect of San Nicola, in Italian there is also only one fully grammatical way of expressing a causation, namely a construction with the infinitive and the object of causation as indirect object (dative): Il maestro fa leggere il libro al ragazzo.38 In view of the typological difference between the two varieties in contact, with the subordination via subjunctive (arb) vs. infinitive (Italian) and their respective syntactic characteristics, it may be claimed that the causative constructions in question in Italian and arb exactly match, whereas the additional possibilities of Standard Albanian are excluded in both of them.39 Coming back to the causative particle in Frascineto with complete coordination, it is obvious that in this case full grammaticalization took place with 38 Here I exclude the only marginally acceptable construction il maestro fa che legga il libro il ragazzo, corresponding literally to (8b); see Turano (2005: 25). 39 Actually, the parallelisms do not stop here. So, the Italian construction with the object of causation as an indirect object has been calqued in San Nicola by a corresponding dative and it may even be expressed in both varieties in contact by an agentive prepositional phrase of the type arb ka djali, It. dal ragazzo 'by the boy' . Both possibilities are excluded in Standard Albanian. See Turano (2005: 26-29) for other examples and more details. only one inflected verb (the main verb) and a purely analytical marker of the function cause, leading so to speak to an analytical verb form. In contrast, it could also be argued that the arb causative shows an exactly inverse (mirrored) image of the Italian causative construction, in both cases with only one inflected verb: in Italian the causative verb, in arb the main verb. In both languages there is a corresponding uninflected element, the infinitive in Italian and the causative particle in arb. In other words, arb shows both: a preservation of the Balkanism of the lack of an infinitive and a calque of the foreign model of inflecting only one of the verbal units in the causative construction. From a point of view of language economy, both languages are on a par, and it seems obvious that this inverse image was the maximum way by which arb could adapt to Italian, given its lack of an infinitive, on the one hand and the typological characteristics it had as a Balkan language, on the other (Breu, 1994: 379-381).
It is worth noting that the Frascineto particle construction may be coordinated as a whole with inflected lë 'to let' as in (9). In this case the causative marker is used more or less pleonastically.
(9) Na e lam e bin e partirti. we 3sg.acc let.aor.1sg cop cause cop leave.aor.3sg 'We let her/him leave.' Literally: "We let and made and s/he left."

5.5
Other Periphrases In Italo-Albanian, particle constructions are more widespread than in Standard Albanian. Most modal particles existing in Balkan Albanian have been preserved, for example, mund 'can' , but not do 'will' (for forming the volitive future). On the other hand, new particles have developed, derived in part from the 3rd person singular of the present, besides the causative marker bën, used in all tenses, for example also zë 'to start' , still limited to the present. Another source is the 3rd person singular of the aorist, e.g., zu 'to start' , as in u zu e dola 'I started to go out' , corresponding to present-tense zë. Others are derived from the imperfect, in addition to the already mentioned particle ish, forming the imperfect of the progressive, for example disht and (possibly mixed with the present) do( j)t 'to want' , e.g., u dojt/disht shkruaja 'I wanted to write' .
All these particles have been, at least partially, grammaticalized in the sense of forming periphrases in the domains of actionality, irreality or causativity. Like in the case of the progressive, several possibilities for linking them with the main verb exist, coordination with the conjunction e 'and' , subordination by means of the subjunctive and, less frequently, asyndetic juxtaposition.40 A further construction in this domain, worth being mentioned explicitly, is the deontic unreal construction, formed with the particle kisht, derived from the 3rd person singular of the imperfect of kam 'to have, must' . A typical example is (10), in which the particle combines with the pluperfect, formed with the fully inflected imperfect of the auxiliary kam and the participle of the main verb. Particle constructions appear also in combination with the coordinated progressive, like jem e in (11) with the above-mentioned ingressive particle zë 'to start' .

(11)
Na jem e zë e dalmi. we be.prs.1pl cop ptl cop exit.prs.1pl 'We are starting to leave.' Just like in the case of the coordinative progressive, not all auxiliaries in periphrases are reduced to particles, with differences from dialect to dialect. For example, in the dialect of San Martino the (abbreviated) example (12) was recorded, showing the inflected auxiliary zura (aor.1sg), instead of the ingressive aorist-based particle zu, here in combination with the gerundial progressives tuke sërritur and tuke qar.

(12)
Zura tuke sërritur e tuke qar, start.aor.1sg ptl scream.ptcp cop ptl cry.ptcp ika atej ku ish mëma. run.aor.1sg there where be.iprf.3sg mother.nom.sg.det 'I started crying and shouting, I ran there where mother was.' Moreover, speakers sometimes insist in a periphrasis instead of an expected simple verb. An example from Portocannone is (13) with the suppletive verb bie 'to fall' (aorist ra 's/he fell), where according to the informants adding the aorist of the actional auxiliary vete 'to go' is obligatory, in order to give this 40 As only the 2nd and 3rd person singular present have a subjunctive form different from the indicative (with the exception of jam 'to be' and kam 'to have'), it is in many cases undecidable if the main verb is subordinated or linked asyndetically. For a general classification of verbal periphrases in the dialect of Frascineto, see Breu (1994), for their position in a Balkan-orientated typology, see Fiedler (1989). See also Breu (2008)  Particle constructions are, in principle, typical for the Balkan linguistic area. This is especially true for subordination. In this regard, the new subordinating particle constructions presented in Italo-Albanian could be claimed to be just an extension of an existing type. But as claimed above, this extension was certainly promoted by the model of the infinitive constructions in the Romance contact varieties, in which also only one element is inflected (e.g., the modal verb but not the main verb in the infinitive). It is worth noting that in this domain Italo-Albanian developed its Balkan characteristics still further under the influence of non-Balkan contact varieties, which again relativizes the role of Balkanisms (see also Gardani, Loporcaro, and Giudici, 2021: Section 3).
The causative of the Frascineto type seems to have developed without any direct model. But it obviously is the combination of the two tendencies towards coordination and the spin-off of grammatical information in particles, together with the contact-supported principle of inflecting only one component in modal constructions. As for coordination, it is present in traditional Albanian, too, for example in relatively rare actional constructions (manners of action) like mori e tha 's/he suddenly said' (literally: "s/he took and said") and zu e pi 's/he started drinking' (literally: "s/he started and drank").41 So, also in this point a historically given starting point led to further expansion in the contact situation.

5.6
Recent Developments in Italo-Albanian Grammars While the Italo-Albanian developments discussed so far seem to be rather old, as they are found in the oldest documentations of the respective dialects and are normally widely used by the speakers of the older generation, there are also more recent ones, taken from my current fieldwork. Their acceptance varies from speaker to speaker and from dialect to dialect. Unlike the well-integrated older changes, going mainly back to models from local Romance dialects, the younger ones have been calqued from Standard Italian in its colloquial form.

5.6.1
The Imperfect as a Counterfactual Mood The first development to be mentioned here is the expression of counterfactuality by means of the imperfect, as a result of its extension from a tense/aspect (past imperfective) to the category of mood.
The traditional way of expressing counterfactuality in Standard Albanian is by means of the indicative pluperfect in the protasis and the conditional pluperfect in the apodosis of hypothetical sentences. This corresponds to the Standard-Italian combination of the subjunctive pluperfect (congiuntivo trapassato) in the protasis and the conditional pluperfect (condizionale passato) in the apodosis, which in colloquial speech may be replaced by the imperfect in both clauses of the counterfactual construction.
An example of the Italo-Albanian adaptation to the counterfactual imperfect of colloquial Italian is (14a).42 The complete parallelism in the two languages in contact is demonstrated by the same glosses in the second line in (14a) for both the first line in Italo-Albanian and the third line in Italian. The counterfactual imperfect is still avoided by conservative Italo-Albanians, and its frequency varies also in the single dialects. For example, even younger speakers in Portocannone use it only rarely, while it is quite common in Frascineto.

5.6.2
Contact-Induced Change in the Domain of Voice In the domain of voice, Italo-Albanian has fully adapted to the traditional system of two passives in Italian, the participial passive with the auxiliary essere 'to be' and the reflexive passive. Though traditional Albanian also has several possibilities for the expression of states of affairs in the passive voice, there are typological differences with respect to their Italian counterparts. First of all, the two Italian passives are free variants, competing with each other, whereas Albanian has a complementary distribution, as, for example, the passive present and the imperfect are expressed by means of a synthetic medio-passive, while the perfect of the passive is expressed by means of the participial construction.
As for the changes that have occurred in the Italo-Albanian voice system since the immigration, I cannot go into details here, particularly as the passive formation does not belong to the traditional balkanisms. But two important developments, for example in the dialect of Frascineto, should be noted. The first one is the adaptation of the Italo-Albanian medio-passive to the Italian reflexive (dynamic) event passive in rejecting an explicit agent. In the same way as it is ungrammatical in Italian to say *la casa si (refl) compra da me 'the house is (being) bought by me' , literally: "the house buys itself by me", it is impossible in Italo-Albanian to say *shpia bjehet (prs.mpass.3sg) ka u,43 though in Standard Albanian the corresponding medio-passive expression shtëpia blehet (prs.mpass.3sg) nga unë is perfectly acceptable (Buchholz and Fiedler, 1987: 186-187). In contrast, the Italian participial passive of the type la casa è comprata (ptcp.sg.f) da me 'the house is (being) bought by me' is compatible with an explicit agent and may be used both as an event passive and as a stative passive, expressing the result of an action in the past. Traditionally, the Albanian passive formed by means of "present or imperfect of jam 'to be' + participle" in Albanian only expresses states (Buchholz and Fiedler, 1987: 192-193). This is also the preferred function of this passive in Italo-Albanian in examples like (15)  Just like Albanian, arb uses in its stative passive instead of the invariant participle its adjectival counterpart, by adding an adjective connector, allowing for number and gender agreement, like the feminine-singular connector e in (15). This corresponds to the agreement characteristics of the Italian past participle (in both its event and stative function), here comprata (ptcp.sg.f). Starting out from this situation, the Italo-Albanian stative passive has extended its function to an event passive in the sense of 'is being bought by me' , at least for some speakers. This is a clear adaptation of the semantic structure of this construction to the polysemy of its Italian counterpart. In both passive functions an agent is allowed, again like in Italian.
While in the present the interpretation of the adjectival passive construction as an event passive, based on the Italian model, is still rather restricted in arb, it has become the only possibility, when referring to past events. In Albanian, traditionally participial (not adjectival) forms with the present and the imperfect of jam 'to be' like është blerë 'it was bought' , ishte blerë 'it had 43 Most Italo-Albanian dialects have the same restriction with respect to the impossibility of an explicit agent for the medio-passive as Frascineto, with the exception of the central Basilicata dialects like Barile, showing a transition between the traditional usage of the passive constructions and the contact-influenced changes (Turano, 2011: 36). In both contact languages the medio-passive refers to both single and habitual events. But, for example, in the dialect of Frascineto it is possible to disambiguate these two functions by using the copulative periphrasis for current processes like Kjo shpi ësht e bjehet 'This house is being bought' . been bought' are forms of the perfect and the pluperfect, alternating with the synthetic mediopassive in the present (blehet) or the imperfect (blehej) and the reflexive mediopassive, for example, in the aorist (u ble). In the Italo-Albanian dialects that have preserved the formation of the passive perfect with jam these temporal relations are still valid, but, interestingly, the participle obligatorily transforms into an adjective, preceded by the adjective connector agreeing in gender and number (Turano, 2011: 32-35), thus extending the traditional parallelism between arb and Italian in the stative present to this environment. In a similar way, in the presumptive area, where the indicative event passive referring to the past is expressed exclusively by the Aorist of jam, only the adjectival participle seems to be possible, at least in Frascineto, e.g., shpia qe (be.aor.3sg) e (con.sg.f) bjejtur (ptcp) ka Maria 'the house was / had been bought by Maria' (Frascineto), with an agreement structure parallel to Italian and contrasting with Standard Albanian. While these developments in the domain of traditional passives already belong to the very core of Italo-Albanian grammar, there is also an innovation in the passive types themselves, more precisely, the development of a venitive passive as a calque of the Italian participial passive with the auxiliary venire 'come' . In this case, a construction like Italian viene comprata 'she is (being) bought' (ongoing or habitual event), literally "she comes bought" has an exact counterpart in the arb venitive passive vjen e bjejtur. Like its Romance model -and unlike the medio-passive as the arb counterpart of the Italian reflexive passiveit allows for an agent; see example (16) from Frascineto. Like in the participial jam-passive, the participle is used in its adjectival form, agreeing with the subject in gender and number, just like the passive participle in Italian.
(16) Ajo bika vinej e mbuluar this.sg.f pile.nom.sg.def come.iprf.3sg con.nom.sg.f cover.ptcp ka Vinxhendzi. by Vincenzo.nom.def 'This pile was (being) covered by Vincenzo.' As yet, the venitive passive is still not very frequent, and some speakers reject it altogether.44 But just like in the case of the counterfactual imperfect, they are accepted by less conservative persons, who use them in colloquial speech, as it seems, more freely in Frascineto than in Portocannone. In the descriptions of other dialects it is simply not mentioned.
The introduction of agreement into the event passive is unexpected from a common Albanian point of view. It is a clearly contact-induced Romance feature, distancing Italo-Albanian from its genetic cognates, which is also true for the emergence of a venitive passive.

Conclusions
Contact-induced change in Italo-Albanian has in part reduced its Balkan characteristics, as in the cases of the volitional future and the development of aspectual periphrases. But the opposite is also true, as in some cases in which Albanian traditionally did not participate in a specific Balkanism Italo-Albanian now has the feature in question. Examples presented here were, for example, the development of a vocative or a tendency towards the loss of the morphosyntactic differentiation between genitive and dative. But all such differentiations and adaptations have to be taken as casual, as they simply depend on the structure of Italian as the dominant model for contact-induced change in Italo-Albanian, contrasting by chance with the peculiarities of "Balkan grammar" or matching them. This also generally speaks against simply counting common features of Balkan languages as such to characterize the Balkan linguistic area, but in favor of explaining every single Balkanism in terms of the mutual influences of the languages on the Balkan peninsula. The same is true for grammatical developments in Italo-Albanian that have to be investigated individually, in order to determine the possible role of foreign models in each single case. If such developments lead to a "de-Balkanization" of Italo-Albanian is a rather secondary question.
Two other criteria are of central importance for contact linguistics, namely, the preservation of a traditional Albanian structure (be it a Balkanism or not) against the model of the dominant varieties and contact-induced change by adapting to a foreign model, in spite of the resistance of the inherited Albanian structure. Among its resistant characteristics Italo-Albanian shows such typical Albanian features like the postponed definite article or the declension of nouns, but also the lack of a (morphological) infinitive. In contrast, for instance, the morphosyntax of the passive has adapted to a large degree to the Romance type, and the imperfect has developed a counterfactual function, based on a model of colloquial Italian.
However, in many cases there are grammatical developments in Italo-Albanian not contradicting neither the Albanian language type nor the specific properties of its contact varieties, but simply uniting them to a new compatible structure, different from both. An excellent example of this type is the development of a causative, based on the Albanian heritage of forming grammatical particles and on the Romance property of having only one inflected verb in modal constructions. As Italo-Albanian has been influenced by different Italian varieties, including different local dialects, the results of such common developments may form isoglosses within the Italo-Albanian linguistic area, as in the case of the perfect, having developed into three different types, depending on both inherited differences and differences in the Romance models.