GREEK AS AN SAE LANGUAGE: DEVELOPING ON THE MICRO-ORIENTATION PERSPECTIVE*

This paper considers the Greek language as a member of the Standard Average European (SAE) linguistic area as defined by Haspelmath (, ). After a brief presentation of the model, there follows a detailed analysis from this perspective of four selected features in Greek: relative clauses with relative pronouns, the “have”-perfect with a passive par-ticiple, participial passives, and negation. The approach applied focuses on specifics that concern standard and non-standard varieties, not only in the language system itself but also in its diachronic development. The results are then measured using Seiler’s () classification of SAE features, with an eye to enriching the classification both empirically and theoretically.


Introduction
Areal linguistics investigates the similarities among languages resulting from their common historical development on shared or adjacent territories and through the multifarious contacts between their speakers.Although areal linguistics is deeply rooted in the linguistics of the first half of the th century or even earlier, it is only since the s that a considerable number of areal linguistic studies have appeared.
Traditionally, the research has focused on smaller areas, such as the Balkan, Central European, or Circum-Baltic linguistic unions, and the specific linguistic features characteristic of the languages belonging to them.An entirely new research perspective emerged thanks to our better understanding in recent decades of non-European languages around the world.It has been recognized that, in comparison to the majority of non-European languages, the languages of Europe exhibit many features, particularly in the realm of morpho-syntax, which tend to be distributed areally.Consequently, it has been proposed to treat most European languages as a kind of linguistic unit, a unit now commonly known -following an early expression used by Whorf ()1 -as Standard Average European or SAE.There are several, slightly different approaches to this concept (cf.Stolz  for an overview), but the centre vs periphery approach eventually emerged as the most influential, hierarchizing the European languages according to the number of linguistic features present in them that conform to the SAE ideal.The degree to which these features are present in a European language is then decisive to its belonging to and position within the SAE.
In the s, extensive research was conducted from this perspective within the EUROTYP project, which focused on the areal relations between European languages and their typological profiles.In addition to eight volumes addressing individual features that were presumed to represent "Europeanisms" (Auwera ; Feuillet ; Siewierska ; Hulst ; Riemsdijk ; Dahl ; Plank ; Bernini and Schwartz ), the overall results were presented and commented upon by Martin Haspelmath in two seminal papers (Haspelmath , ).Summarizing the main results of the EUROTYP project, Haspelmath proposed a general picture of the SAE linguistic area.In Haspelmath (), a broader set of salient grammatical features was chosen, which were considered to be characteristic of the core European languages and which, according to Haspelmath, define the SAE Sprachbund: a total of  features were described that only partially intersect with the set of nine SAE features finally used in Haspelmath (),2 as shown in the list below: Haspelmath () 1. anticausative prominence 2. nominative experiencers 3. A-and-B conjunction 4. verb fronting in polar questions Haspelmath (, ) 5. definite and indefinite articles 6. relative clauses with relative pronouns 7. the "have"-perfect with a passive participle 8. participial passives 9. dative external possessors 10. negative pronouns and a lack of verbal negation 11. relative-based equative constructions Haspelmath () 12. subject person affixes as strict agreement markers 13. intensifier-reflexive differentiation When the presence of these features in European languages is combined, their areal distribution across Europe can be summarized in cluster maps (cf.Figures  and ).In the  model, the nucleus of SAE comprises German and French, with all nine features present.It is closely surrounded by a core group of Romance languages (Spanish, Portuguese, Italian, Sardinian), together with Albanian and Dutch (eight features).Modern Greek (MG), with Romanian on the southern "Balkan wing" and English on the "Germanic wing", is in the outer core with seven features.Then, there is the "Nordic group" (Icelandic, Norwegian, Swedish) and Czech (six features) and finally the "Eastern group" (South and East Slavic, Polish, Baltic, and Hungarian) with five features. 3Due to the larger number of included features, the assumed nucleus of the  SAE model was broader, encompassing German, French, Dutch, and the northern Italian dialects.The core included MG along with the other Romance, Germanic, Balkan, as well as West and South Slavic languages. 4hus, MG was closer to the SAE core in the earlier model, while slightly more distant from it in the later version.In the final model, MG exhibits features (5), ( 6), ( 7), ( 8), ( 9), ( 11) and ( 13) from the aforementioned list.
There are three essential characteristics of Haspelmath's work and the EUROTYP project more generally: 1) studying only contemporary varieties of European languages through a synchronic approach, 2) investigating only standard varieties of these languages, and 3) applying the criterion of the mere presence or absence of the feature under investigation in a given language.However, recent studies in the areal typology of Europe (Murelli ; Murelli and   Fagard et al. ) have sought to extend the scope of the research into SAE by addressing the questions of diachrony, non-standard varieties, as well as the possible constraints on a feature within the language system and its use.These recent advances are based on the work of Auwera (), who has called this approach a "micro-orientation perspective".Following this trend, our aim in this paper is to show that by looking at SAE from this expanded viewpoint, we can achieve more accurate results by analyzing diachronic and stratification factors, and thus look at the situation within SAE from a more comprehensive perspective.Our objective is to show the need for a more detailed analysis using Greek as an example.We will attempt to describe and assess four selected SAE features in Greek, which we consider the most illustrative in this respect.Three of them (relative clauses with relative pronouns, the "have"-perfect with a passive participle, and participial passives) have been acknowledged by Haspelmath (, ) as being present in Greek, whereas the remaining feature (lack of verbal negation) was claimed to be absent.
2. The specifics of the selected SAE features in Greek

Relative clauses
Postnominal relative clauses formed by a relative pronoun that is inflected and syntactically integrated into a subordinate clause, a strategy that is quite exceptional outside Europe, are considered to be one of the unique European areal features.According to Haspelmath (: ), in this "European" type of relative clauses, the relative pronoun functions as a resumptive relative and, with a few exceptions,5 is derived from the interrogative pronoun in most European languages.In MG, this SAE feature is supposedly represented by the relative pronoun ο οποίος (m.), η οποία (fem.), το οποίο (n.) 'who, which, that' , a declinable relativizer which, introducing a relative clause, agrees in gender and number with the noun phrase it modifies, but it must be in the case appropriate to its function within the relative clause itself (Holton et al. : ).In what follows, we will attempt to verify whether this pronoun, when viewed from a micro-orientation perspective, does indeed fit Haspelmath's definition of this European areal feature.First, the assumption that ο οποίος was derived directly from an interrogative pronoun as in other European languages is not fulfilled in Greek.Although the question of its origin is extremely complex and complicated, recent research has tended to conclude that the second part (οποίος) is derived from the ancient indefinite relative pronoun ὁποῖος 'of such a sort, as' , which gradually acquired a definite relative function ('who, which'), 6 and not from the interrogative pronoun ποῖος.
The first part ὁ probably arose from the erroneous division of ὁποῖος into ὁ ποῖος and a subsequent hypercorrection into ὁ ὁποῖος (cf.e.g.Horrocks : -).The segmentation into an "article" and a second segment may have been due to the considerable influence of the Romance relatives il quale (It.)/lequel (Fr.) in territories with close contacts between Greek and the Romance languages.The expression ὁ ὁποῖος is first attested in the th century, especially in southern Italy, and from the th century onwards with increasing frequency in territories under Frankish or Venetian rule (Cyprus, Rhodes, Crete).Nevertheless, the theory of a pure calque induced by the Romance languages (i.e. the definite article + the interrogative pronoun ποῖος) no longer seems to be accepted in light of new data (see Holton et al. : - with a further bibliography).Even in Romance languages, moreover, the question of the origin of relative expressions is not completely clarified, as Giacalone Ramat (: -) has shown in the case of Italian in her discussion of the grammaticalization process leading to the constituency of a new relative, namely il quale.In addition to the traditional scenario in which the possible source for this relative construction is the interrogative qualis, she also considered the possibility of its origin from the correlative/comparative meaning of qualis (talis -qualis).This consideration with regard to Italian is interesting in the overall context of how the origin of this SAE feature is generally defined across European languages and corroborates the need to develop more comprehensive, diachronically oriented descriptions of the SAE relatives' characteristics. 7econd, in the standard variety of Modern Greek (Standard Modern Greek, SMG), ο οποίος competes with an indeclinable relative complementizer που which is the major relativizer in Greek. 8The parallel existence of the two relative expressions is not unique in European languages (Eng.which vs that, Cz. jenž vs co, etc. Cf. also Seiler :  for the situation in German), but the indeclinable particle relatives cannot be considered a specifically European feature, as they are attested widely elsewhere in the world.As far as the competition between που and ο οποίος in Greek is concerned, these expressions are, in many cases, interchangeable.Nevertheless, certain functional and frequency limitations caused by pragmatic or structural factors may mean one is preferred over the other (Holton et al. : ).The relative pronoun language of the non-literary papyri from the rd century bc to the th/th century ad, and in these instances rarely used as a mark of official or archaic style, especially in legal documents (Kriki : ).As an archaic survival, it rarely introduces indirect interrogative clauses in Medieval Greek vernacular texts, while as an indefinite relative with the meaning 'whoever, whichever, whatever' , it is common both in literary and non-literary Medieval Greek texts (Holton et al. : , ).
ο οποίος is more frequently used in formal discourse, especially in written texts, than the relative complementizer που.Ο οποίος is also preferably chosen instead of που with oblique/prepositional cases, even in less formal discourse, to avoid the ambiguity caused by the uninflectedness of που.Although there are as yet no quantitative studies on the frequency of που/ο οποίος, it is implicitly assumed that ο οποίος is generally less frequent in SMG than που, partly due to its "learned" character. 9n dialects of Modern Greek ο οποίος is almost entirely absent, and if present, it is thought to have been incorporated into the language of dialect speakers during the th century, resulting in the hypercorrect doublet ο οποίος που (Nicholas :  with examples and further bibliography).In Greek, Haspelmath's definition of this Europeanism would be much better matched by the pronouns τίς and ποίος, originally interrogative pronouns, which in the Hellenistic and early Byzantine periods had already expanded via indirect questions into non-restrictive and later restrictive relative clauses (Holton et al. : -).These relative pronouns, partly or almost fully grammaticalized as uninflected relativizers, are still preserved in certain peripheral MG dialects that retain many archaic features, such as in Cappadocian, Pontic, and the dialects of Southern Italy (Liosis and Kriki ; Kriki : ).
In this example, we see that the Europeanism in question does not hold a prominent position in the standard or colloquial varieties of contemporary Greek, where it competes with που, nor does it appear in dialectal varieties.On the contrary, only in dialects do we find any relatives (τίς, ποίος) through which we can diachronically trace expressions "more suitable" to Haspelmath's definition, although today they have mostly lost the parameter of inflection.A more precise definition of this SAE feature thus proves highly desirable -either a broader one (e.g.not insisting that the relatives originate from an interrogative pronoun) or, conversely, a narrower one, but one that would change the distribution of this SAE isogloss.

Constructions with passive participles
Two commonly acknowledged features of the SAE languages are the "have"-perfect, formed by the auxiliary verb "have" with a passive participle, and participial passives, formed with a passive participle and an intransitive copula-like verb ("be", "become", etc.).Again, these two features are rare in languages outside Europe (Haspelmath : -).In MG, we find such constructions with the verbs έχω 'I have' (έχω το γράμμα γραμμένο, lit.'I have the letter written') and είμαι/γίνομαι 'I am/become' + perfect passive participle (είμαι δεμένος, lit.'I am tied/I have been tied').
With regard to the former feature, the "have"-perfect, Haspelmath (: ) was aware of its different semantic functions in SAE languages because the perfects are at different stages within the grammaticalization process: while in English, for example, this perfect has a present-anterior meaning (I have eaten an apple.), in German it can also be used as a normal perfective past (Germ.Ich habe einen Apfel gegessen.= Eng.I ate an apple/I have eaten an apple.).In MG, it is a strictly resultative construction restricted to a very limited subset of verbs which require a direct object in the accusative (cf.Moser : -, -, ff for a historical overview and thorough analysis of additional restrictions).Even if the "have"-perfect with a passive participle in an exclusively resultative function10 was already documented in Archaic, Classical (Bentein : , -), but mainly in Early Post-Classical Greek from circa the rd century bc onwards (Bentein : ),11 it has undergone only a partial process of grammaticalization.The incompleteness of the grammaticalization process of the participle prevented it from being used with every verb and from losing its morphosyntactic restrictions in the process (Moser : -).Semantically, it is also not a "true" perfect, that is referring to an action or event that occurred in the past, but is related to or continues into the present; instead it more describes a state in the present (cf.above έχω το γράμμα γραμμένο 'I have the letter written' rather than 'I have written the letter').Consequently, this construction plays only a marginal role in the verbal system of SMG, being significantly constrained in its usage to certain classes of verbs12 and mostly replaced by the much more common periphrastic perfect construction in SMG, i.e. "have" + fossilized infinitive13 (έχω γράψει 'I have written').It follows from the above that the "have"-perfect with a passive participle in SMG does meet the definition of an SAE feature according to Haspelmath's conception.However, it is a feature with significant frequency and functional limitations, occurring in specific semantic contexts and restricted to a certain group of verbs.This raises the question to what extent the frequency or functional restrictions of an SAE feature in a language should influence its ranking or classification within the SAE Sprachbund.
The passive counterpart of the έχω 'I have' + perfect passive participle construction, i.e. the intransitive copula-like verb είμαι/γίνομαι 'I am/become' + perfect passive participle, 14 presents a different picture.This construction is much more productive in SMG than the former, being formed by a very large number of verbs.In addition to three-place and two-place predicates, this construction is also possible with a subclass of one-place predicates, the so-called unaccusative (ergative) predicates (Agouraki : ).These are semantically characterized by the fact that their subject is not actively responsible for the action of the verb, it neither undertakes any action (is not an agent) nor is involved in it (e.g., πέφτω 'I fall': γέρικα δέντρα είναι πεσμένα στο έδαφος 'old trees are fallen/lying on the ground'; ανθίζω 'I bloom': τα λουλούδια είναι ανθισμένα 'the flowers are in bloom').While unaccusative verbs can also possibly form a "have"-perfect with a passive participle (cf.έχουμε ανθισμένα λουλούδια όλο τον χρόνο 'we have flowers blooming all year round'; έχουμε πεσμένα δέντρα 'we have fallen trees'), the second subclass of one-place predicates, unergative verbs, where the subject is interpreted as an active agent or an agent actively responsible for the action expressed by the verb, cannot form the "have"/"be"-constructions with a perfect passive participle (*έχω τηλεφωνημένο / *είμαι τηλεφωνημένος 'I have phoned/*I am phoned').Thus, in the case of intransitive verbs, the choice of either έχω 'have' or είμαι 'be' depends on unaccusativity or unergativity (cf.Agouraki  with further examples).
As far as the semantic function is concerned, this construction indicates a state which is purely stative, not resulting from the previous action, and therefore cannot be classified as a genuine perfect.A resultative interpretation is also possible, but only if based on the lexical meaning of the verb itself.15Thus, if these criteria are met, this construction may be equivalent to the much more common, "competing" resultative SMG perfect έχω 'I have' + fossilized infinitive (το γράμμα έχει γραφτεί 'the letter has been written') in expressing the resultative meaning.
If we turn our attention from the standard variety of MG to its dialects, a completely different picture of these Europeanisms emerges.There is a predominance of "have"/"be" constructions with passive participles at the expense of the perfect constructions common in SMG (έχω γράψει 'I have written' , έχω γραφτεί, lit.'I have been written').If these SMG perfects exist in a dialect, it is due to a secondary influence from SMG (Ralli et al. : ), where the perfect έχω + fossilized infinitive appears very late, that is not before the end of the th century.Even then, it was very rare, and therefore it seems logical that the Greek dialects did not adopt it. 16s convincingly attested by the Medieval Greek sources (cf.Holton et al. : -), after the disappearance of the monolectic perfect in late antiquity, the aorist indicative took over the entire category of perfect to express the resultative meaning, being simultaneously "accompanied" by a stative passive present represented by the "have"/"be" constructions with passive participles.This diachronic picture of European-type constructions with "have"/"be" plus passive participles in Greek suggests that they have existed consistently throughout the history of Greek, probably in all its territories, undergoing a process of grammaticalization at different stages (Ralli et al. : ff).As with the relativizers discussed in section .., the two assumed SAE features are present more, or even exclusively, in the non-standard varieties of Greek than in the standard variety.

Negation
Another feature to note in the diachronic development of the SAE features is the negation of the type verb + negative indefinite (V-NI), that is, the construction in which the negation is expressed by a NI pronoun only.This type of negation is very rare across world languages.Therefore, it was considered by Haspelmath (: ) to be specific to the European area, where it is only present in Standard Modern English (nobody NI entered V ) and German (niemand NI kam V 'nobody came').The second type of negation distinguished across European languages is that in which a verbal negation (NV) and NI co-occur (NV-NI); such languages are referred to as negative-concord languages.These are further divided into strict and non-strict negative concord languages.The difference between them lies in the type of sentence constructions in which the negative concord occurs.In non-strict negative concord languages, the position of the NI relative to the predicate is important: if it precedes the predicate verb, the verb must not be negated, and negative concord does not occur (It.nessuno NI viene V 'nobody comes').If the NI occurs after the verb, the predicate verb must also be negated, and negative concord occurs (It.non ho visto V nessuno NI 'I didn't see anybody') (see Giannakidou  with examples and further bibliography).Haspelmath (: ) classifies European non-strict negative concord languages with (N)V-NI negation as a subtype of the V-NI languages in order to include them among the SAE-type core languages possessing this specific SAE feature.This is, for example, the case with modern Romance languages, which have acquired V-NI or (N)V-NI negation with varying uses in written and colloquial varieties (cf.French ne), while in Latin, the strict negative concord was formerly present (nemo NI venit V 'nobody comes'; cf.Irslinger : -).17Nevertheless, in most modern European languages, negation is formed by means of strict negative concord, i.e. the negative concord always occurs regardless of the position of the NI to the predicate (NV-NI).This is the case with Slavic, Finno-Ugric, and Balkan languages including MG (Cz.nikdo NI ne-jde NV , Gk. κανένας NI [δεν έρχεται] NV 'nobody comes'), but this type of negation is not considered to be a specific SAE feature due to its common occurrence in other world languages.Based on these criteria, then, MG does not possess this SAE feature.
However, this has not always been the case.Classical Greek was a non-strict negative concord language since the co-occurrence of an NV and an NI pronoun depended on their position relative to the verb (Willmott : ; Muchnová : ; Gianollo ).Similarly to modern languages, n-words in Classical Greek required a negator if they were in a postverbal position (Aesch.Agam. [οὐκ οἶδεν] NV οὐδείς NI 'no one knows'), whereas if they were in a preverbal position, its use was not permitted (Aesch.Prom.Desm. τοῖσιν οὐδεὶς NI ἀντέβαινε V πλὴν ἐμοῦ 'against this none dared make a stand except me').Greek in this period was thus a non-strict negative concord language ((N)V-NI) and was, therefore, one of the core SAE languages because of the possible V-NI negation. 18he evolution of Greek from a non-strict to a strict negative concord language is still a matter of debate, with one possible explanation being the change in word order in Late Ancient Greek, which allowed for two schemes of potentially neutral word order, SVO and VSO (Horrocks : -, Holton et al. : -).Emphasis could be achieved either through stress alone or by moving the originally neutral constituent into a focalizing position (along with the emphatic/contrastive stress).This became the left periphery of the clause, i.e. the preverbal position.This is also where in Medieval Greek newly developed emphatic n-words (τις/τινάς, κανείς) began to be placed as a negative focus (Horrocks ), yet with the verb negator being still present. 19Thus, Greek became a strict negative concord language and, in its modern form, diverged from the SAE core languages.However, the language is not unique in this development: strict negative concord only gradually spread and became established in other SAE languages as well, including not only in the peripheral Russian but also in the more centrally located Czech (Giger : ).

The prospects of SAE: Seiler's "realistic areal typology"
Our study has demonstrated that individual linguistic properties of Greek, previously considered in synchronic approaches as clear instances of SAE features, or, conversely, as not representing SAE features, exhibit variable behaviour if reconsidered via the micro-orientation approach.It is now tempting to incorporate our observations into Seiler's () classification of SAE features, which considers both diachronic (examining the degree of feature representation throughout language development) and stratification (investigating the similarities and differences in feature representation between standard and non-standard varieties) dimensions.In his "realistic areal typology" ("realistische Arealtypologie") concept, which has been applied to historical (Old High German), modern standard (Modern Standard German), and non-standard/dialectal (Low Alemannic) varieties of German, Seiler (: -) distinguished between expanding and recessive SAE features ("expandierende/rezessive SAE-Merkmale") in terms of their diachronic dynamics.According to Seiler, the expanding SAE features only appeared after the Old High German period or at best became further established then.This implies that Seiler's classification admits the existence of certain SAE features in the language even before Old High German.For the recessive features, again, a decline from the Old High German period can be observed (Seiler : ).
Within each of the two groups of recessive and expanding features, Seiler further discerned three subclasses (Seiler : -; cf.Table  below for a schematic overview): A. expanding features a) indifferently expanding features: they only spread in the observable history of the language, expanding equally in standard and non-standard varieties; b) from-above expanding features: they only emerged in the modern standard variety but do not occur (or are less prominent) in the historical phases of the language (Old High German) and in non-standard varieties (Low Alemannic); c) from-below expanding features: they emerged in the non-standard variety (Low Alemannic) but are absent or restricted in Old-and also Modern High German, i.e. the modern standard variety because the conservatism of its codification prevents them from expanding into it.
B. recessive features a) indifferently recessive features: declining equally in standard and non-standard varieties; b) from-above recessive features: they recede only from the standard while remaining in dialects; and c) from-below recessive features: they recede only from dialects while remaining in the standard.
The "indifferent"-Europeanisms (A.a, B.a) are supposed to be the oldest of the three subclasses, perhaps going back to language contact after the Migration Period (as proposed by Haspelmath : -), a period marked by large-scale population migrations and the transition from late antiquity to the early Middle Ages in Europe.The "from-above"-Europeanisms (A.b, B.b) probably emerged later as a "product" of the common codification strategies of the written languages, not as a result of language contact and convergence.The "from-below"-Europeanisms (A.c, B.c) are most likely the youngest innovations among the three subtypes, not being attested in the early phases of the language (cf.Seiler ).
Our analysis of four SAE features in Greek allows us to tentatively attribute these characteristics to Seiler's (sub)classes of SAE features.The first difficulty that arises when applying these criteria to Greek is the question of the presence or emergence of SAE features in the historical phases of the language.As can be seen from the previous sections, all the discussed SAE features in Greek are attested in the historical phases of the language.Most of them, moreover, were already present in Classical or even Archaic Greek (as in the case of the "have" and "be" constructions) or, conversely, only in the late Middle Ages (ο οποίος).Only one of the features discussed, namely the changes in the system of verbal negation, can be assigned to the period between Late Antiquity and the Early Middle Ages when, according to Haspelmath and Seiler, the oldest SAE features should have emerged.In Greek, then, the historical presence of SAE features attested in the earlier phases of the language should be considered in comparison to Seiler's model, and this is the case for all the subclasses of expanding or recessive SAE features.
With this in mind, we can make an attempt to correlate the SAE features in Greek to Seiler's subclasses described above.The case of Greek negation (section ..) is unambiguous: the evolution from the non-strict to the strict negative concord that took place in Medieval Greek corresponds to Seiler's definition of an indifferently recessive feature because the original SAE language feature is eventually absent in both SMG and MG dialects.
The other two features, the "have"-perfect and "be"-construction with a passive participle (section ..), are very close to what Seiler calls a "from-above" recessive Europeanism.They are still present and prevailing in MG dialects, while in SMG, they are restricted in frequency and function, ceding to the more productive and highly prevalent active/passive perfect forms (έχω γράψει, έχω γραφτεί).
A similar but inverse case is the relative pronoun ο οποίος (section ..): its prevalence in SMG (albeit in competition with που) and its absence from MG dialects suggest "from-above" expansion.Even if Seiler's definition of the "from-above" features as a relatively recent sociolinguistic phenomenon, possibly induced by the codification process, does not match the Greek feature which is already attested in the Middle Ages, we may still view it as induced "from-above": in Medieval Greek, it is first attested in higher registers such as notarial and legal documents (Manolessou ), and so it emerged and spread into the language from the more educated strata of society.This "from-high-register" spread can be likened to the "from-above" spread, even though there was no codified language at the time.Table  summarizes the investigated SAE features in Greek according to Seiler's classification, with the caveat that for "from-above"-and "from-below"-Europeanisms we admit the possibility of their being older Europeanisms, also present in earlier phases of the language.We also believe -given that in the development of modern Greek, the codification process generally did not play a major role -that, for the "from-above"-Europeanisms, it is desirable to consider also high, but not yet codified varieties, as the possible source of their origin and the means of their spread.
In general, the lesson to be learned from Seiler's definition of the latter two subclasses (A.b, A.c, B.b, B.c) and from the evidence introduced in this paper is that crossing the diachronic and stratification criteria is not an easy task: Seiler (), probably due to the limitation of studying material only from German varieties, may have failed to address all the possibilities.But in spite of this, or perhaps because of it, his approach remains deserving of attention and further rethinking.In any case, the research into the question of a language swinging towards or, conversely, moving away from the defined SAE ideal helps us to understand better the dynamics of the areal profiling of European languages.In this context, the question may arise as to what the cultural and social implications of SAE are and whether "the SAE linguistic area really exists, and if so, in which terms", as one of the reviewers of this text rightly asks.It is important to note that the existence of a linguistic union in general has a different validity than the existence of another group, often considered complementary, namely the language family.While the latter can be clearly defined on the basis of regular sound correspondences between specific elements (roots, morphemes in general), the language union is to a much greater extent a scientific construct whose acceptance or non-acceptance depends on how well this model helps us to make sense of the multifaceted linguistic reality.Seiler's approach, and -we hope -our modification of it, is an attempt to provide a more detailed description "from the bottom", i.e. from individual language varieties.Only the sum of such descriptions will possibly provide the basis for a more significant change in the understanding of the SAE linguistic area in terms of its cultural and social aspects, and possibly of language unions in general.Our creed is to remain cautious and not draw overly generalizing conclusions based on data from only one or two languages.

Figure 1 :
Figure 1: Haspelmath (: ) Kortmann ; Seiler ;3Haspelmath (: ) established that five Europeanisms as a minimum were necessary to include a language in the SAE linguistic area, with the remaining languages exhibiting only one or two features. 4MG lacked only one of the  features defined in the  model, namely feature (10).Of the remaining features included in this model that were absent from the  model, features (1), i.e. a high percentage of anticausative verb pairs, and (2), i.e. expressing experiencer arguments by nominative subject, were prominent in MG when compared to the other SAE languages.Cf.Haspelmath () andBossong ()for detailed studies of these phenomena across European languages.

Table 1 :
The dynamics of SAE features according to Seiler's () classification.Legend: + = feature present/expanding; -= feature absent/recessive; -/(+) = feature absent/recessive or present, but in limited frequency/function or in parallel/competition with another expression

Table 2 :
The dynamics of SAE features in Greek based on Seiler's () classification.Legend: HP = historical phase of the language; SMG = Standard Modern Greek; D = dialect(s); + = feature present/expanding; -= feature absent/recessive; -/(+) = feature absent/recessive or present, but in limited frequency/function or in parallel/competition with another expression