1 Language in contact: The case of Corinth This project aims to identify and evaluate the Latin influence (lexical, morphological and syntactic) on the Greek language as recorded in a selection of Greek inscriptions. The starting point is the epi- graphic material produced in Corinth during the imperial era, from its re - foundation as a Roman colony in 44 BC (which followed its destruction in 146 BC) until Late Antiquity (approximatel y until AD 600). 1 Corinth had been one of the most important centres during the Roman empire, for several reasons. In a recent study , Millis (2010) has reviewed epigraphic and literary evidence previously misinter- preted in its reference to the population of Corinth in Roman times and has shown that the “new” colony was repopulated not mainly with Roman veterans, as claimed in the past, but with freedmen who were Greek in origin. 2 Corinth «was a hybrid of both cultures, in which one language [i.e. Latin] became the mode of expression within the public sphere and another [i.e. Greek] within the private» (Millis 2010: 31). The peculiarity of Corinth in the oriental landscape can be exemplified by its religious organiza- tion Hupfloher (2008) presented an ove rview on how the two cultures mixed in the colony to the level that we cannot simply talk about ‘romanization’ or ‘hellenization’ because the two phenomena happened simultaneously Basing her account both on literary evidence, Pausanias and Strabo, and , above all, epigraphical evidence, mainly inscriptions in Latin, she showed how the two cultures blended in the new colony. On the one hand, the evidence reveals that in the Roman colony the gods venerated were not simply an imitation of the ones worshipped in Rom e , but they belonged to four groups: there were the ‘traditional’ Greek gods, such as Demeter, Kore and, above all, Poseidon, next to Roman gods like Iuppiter Optimus Maximus and Ianus ; there were also gods who were venerated both in the East and in the West but whose origin was neither Greek nor Roman, i.e. Moirai, Tyche, Isis and Sarapis , and, finally, imported gods from the South - East or East of the E mpire, such as Ar- temis of Ephesos and Apollo of Claros Among the cults there were imported from R ome, the imperial cult was one of the most significant. However, it was also developed in ways typical of the Greek usage, like in the case of the deification of Octavia, Augustus’ sister and Marcus Antonius’ wife. Pausanias informed us that a temple dedic ated to her existed in Corinth, but she was never venerated 1 The choice of indicating AD 600 as an (approximate) cutoff date has been made following Dickey (Submitted version), who took as a model the end date indicated by Adams (2007) and Souter (1949) in their account of the Latin language. This has been made on the assumption that after that time we cannot be certain that contacts with the West involved the “unitary” Latin, but they could have involved the Romance languages that were arising. Furthermore, the relationship between the two areas which were formerly part of the Roman Empire progressively decreased, except for the temporary ‘Restauratio Imperii’ by Justinian : D ICKEY : 10. This date is, however, just an approximate indication: some texts, espe- cially the ones published by Sironen (2016), which date to a later period, will not be excluded a priori but they will be still analysed , taking into account the peculiarities of the period they were written in. 2 More precisely, in the past, the analysis of literary evidence, which included passages from Strabo, Appian and Plutarch, convinced scholars that the colony was populated by Romans. However, those passages discuss mainly Carthage’s pop- ulation and only the similarity between the histo ry of the two colonies was used to claim that Corinth’s situation was the same as Carthage On the other hand, a comparison between Corinth and another colony, i.e. Patras, has shown Millis that while in Patras there were many funerary inscriptions for Rom an veterans, they are absent from Corinth, where there are many epitaphs for Greek freedmen instead Furthermore, Millis has demonstrated , mainly through onomastic (and, in general, epigraphic ) evidence, that the freedmen were mostly Greek. Finally, the pr edominance of Greek over Latin in graffiti on pottery, which Millis consider s the expression of private language choices, demonstrated how the Greek ele- ment was important in the new colony: M ILLIS 2010, pp. 17 - 20, 22 - 24, 26 - 27. 2 as a goddess in Rome. 3 This usage resembles the reverence towards Roman people who played an important role and reached integration in the province. 4 The special situation of Corinth is also re- flected in the names given to religious authorities. While some titles are entirely Latin, like augur , sacerdos and sim., others show a peculiar mixture of languages: the most significant is theocolus Iovis Capit olini , where we find the use of the Greek noun θεοκόλος / θεηκόλος (“servant of a god ” ) transliterated and morphologically integrated into Latin next to the Roman god Iuppiter Capitolinus 5 Hupfloher informs us that Corinthians could have chosen an entirely Roman title, flamen Dialis , which was adopted for example in the Roman colony of Narbo, but they preferred to use a partially Greek designation adopting a n uncommon Greek term used at Olympia. 6 Fi nally, we can point out that the Isthmian G ames, which returned to Corinth short after its refoundation after a short period in Sikyon, were an opportunity for contact between the Roman colony and the provincial elite, attracting people from the East and the West. 7 As for its importance as trade - centre, t he city was located in a strategic area for commerce : it ha d two harbours, Lechaeum on the Gulf of Corinth and Cenchreae on the Saronic Gulf. 8 Also, its re - foundation in the same place as the ancien t city was due to its location: the colony «was founded for its strategic commercial position in a relatively stable area» (Friesen 2010: 2) and it (and Carthage) w ere «destined to flourish once again as commercial center s , as they had in the past» (Engels 1990: 16). Therefore, it maintained trade relations both by land, especially with the northern regions, and by the sea with the east and the west. For all these reasons, Greek epigraphic texts produced in Corinth during the Roman empire de- serve careful s tudy, to shed more light on language contact between Greek and Latin in the Greek - speaking East. State of the art In the past fifty years increasing attention has been paid to the question of language contact in antiq- uity, especially in the Roman Empire. While it has always been acknowledged that Greek has had an extensive influence on Latin, less attention has been paid to the impact of Latin on Greek. However, some works have been written dealing with this issue, especially in recent times. The s cholarly envi- ronment has recently become increasingly a ware that the Roman world was multilingual and this context has at last started to receive the attention it deserves This is also due to the results achieved in the study of contact between modern lan guages, which will, therefore, be our starting point in the presentation of the essential bibliography for any studies concerning language contact between Greek and Latin. 1. Some basic concepts borrowed from modern sociolinguistics 3 Octavia was also worshipped in Athens and Hupfloher considers it a ‘regional custom’ more than a Corinthian use because her daughter was born in Athens during her trip to Greece in 39/8 B.C.: H UPFLOHER 2008, p. 155. 4 H UPFLOHER 2008, pp. 151 - 156 , 160 5 A ccording to Hupfloher, the term is attested only once in Greek in Kent (1966, n°207) from the beginning of A.D. III, where we find θε - / ηκ ό λος / Κρ ό ν ῳ : H UPFLOHER 2008, p. 157, n.54. 6 H UPFLOHER 2008, pp. 156 - 158. 7 H UPFLOHER 2008, pp. 159 - 160. There is no agreement on when exactly the Isthmian games returned to Corinth. Hup- floher informs us that scholars usually dated the return to a time between 7 B.C. and A.D. 3 but that Kajava (2002) has published new evidence that supports an earlier date: H UPFLOHER 2008, p. 159. 8 B RILL ’ S NEW PAULY , Corinthus/Corinth 3 Modern sociolinguistic s has contributed in many ways to the study of language contact between an- cient languages. Some of the most meaningful achievements have been obtained in the study of the motivations for language choice in a bilingual (or multilingual) context. The choice of one language over another can be determined by a process called “accommodation”, which happens in two opposite manners, depending on the attitude of the actors of the conversation: the speaker can accommodate his speech to the one preferred by the addressee («convergence») or he can decide to separate himself from the addressee («divergence»), adopting a language or some features that his hearer could not understand or could understand with difficulties. 9 The question that scholars have asked thems elves while stud- ying this phenomenon is «to what extent does the individual wish to keep identifying with the group or emphasize their own individuality and pass into the other group?» (Clyne 2003: 55). Examples of convergence can be found in the capital o f Belgium (where the most widespread languages are Dutch and French): a citizen might choose to use Dutch when talking to a person coming from the Flanders, where Dutch is the most used idiom, thinking that he/she is more proficient in Dutch than in French 10 Some of the factors that influence the language choice of modern speakers might have influenced the u se of ancient Romans and Greeks, as exemplified in the work of Adams (2003 ): e.g. socio - historical factors, the status of the language in that area etc 11 Another factor which can influence language choice is the “domain” of an interaction. The concept has been defined as follows by Fishman (1965: 75, italics in the original): «domain is a socio - cultural construct abstracted from topics of communication , relationships between communicators, and locales of communication, in accord with the institutions of a society and the spheres of activity of a culture, in such a way that individual behavior and social patterns can be distinguished from each other and yet related to each other ». The idea has been defined starting from an analysis of the socio - cultural context where language choice happens. Some domains have been identified by schol- ars like Schmidt - Rohr in the study of some multilingual settings: family, religion, school, literature, military, governmental administration etc. Significantly, similar situations, called «spheres of activ- ity», were identified by anthropologists, psychologists and sociologists as « institutional contexts or socio - ecological co - occurrences » (Fishman 1965: 73, italics in the original). Among the factors which contribute to the definition of the domain, topic and role - relations have been presented by Fishman as the easiest to define. As for the topic, in a multilingual society, the choice of the language to use among people who are fluent in more than one language can rely on the specific argument that they are discussing: for example, it might be easy to handle a discussion on economy in one language due to the existence of a speci alized terminology or to the fact that the actors of the conversation have studied the issue in that specific language at school. For what concerns role - relations, «in certain societies particular behaviors (including language behaviors) are expected (if n ot required) of partic- ular individuals vis - à - vis each other » (Fishman 1965: 76, italics in the original). In the domain of the 9 This is based on one of the approaches adopted by linguists to study bilingualism, i.e. the ‘ person - oriented ’ approach. The necessity to adopt a specific type of approach is due to the fact that nowadays, in several communities of the world, «bilingualism is the norm, rather than the exception». This implies that the functioning of the languages requires some «norms for the speakers, and a functional specialization of the languages involved»: A PPEL - M UYSKEN 1987, pp. 22 - 31 (quotes from p.22) 10 A PPEL - M UYSKEN 1987, p. 28. 11 The example of Dutch can be interesting for this study: while Dutch speakers have shifted to English in areas like US, Canada and New Zealand, they were “successful” in competition to English in some areas of South Africa and in com- p etition with Malay in the Dutch East Indies because their language was the language of the rulers. C LYNE 2003, pp. 55 - 57. 4 family, for example, different linguistic choices are made between grandmother and father or grand- mother and grandfather and so o n, and this type of interactions is vital in every domain. 12 One of the most innovative additions to the study of language contact in antiquity, especially in its distinction from borrowing/loanword, is the idea of “code - switching”. Code - switching «is th e ability on the part of bilinguals to alternate effortlessly between their two languages» (Bullock - Toribio 2009: 1). This broad definition implies that code - switching can have different linguistic man- ifestations and it can be produced by speakers who have different levels of proficiency. A widely known and accepted distinction is the one between «intra - sentential» (when a single word or a sen- tence/phrase in one language is inserted into a sentence in the other language) and «inter - sentential» switching (wh en the alternation between the two languages is found at boundaries between clauses). Both require a high degree of proficiency of the speakers in both languages. 13 An example of “intra - sentential” code - switching, as presented by Myers - Scotton (2006), can be found in a conversation between two friends who speak both Spanish and English, since they come from Latin America but they live in the USA: «Porque son two fans , él le cambió los fans » (2006:240, bold in the original, used to indicate the switched items), which means «‛Because there are two fans, he changed the fans’» (2006: 240). An example of “inter - sentential” code - switching, also from Myers - Scotton (2006), is taken from a conversat ion in a shop in Kenya, held in Swahili, «the main informal inter - ethnic group language in Nairobi» (Myers - Scotton 2006: 240), with switches into one of the official languages of the country, i.e. English, between two people who have a different ethnic ori gin and, as a consequence, do not share their first language: « Stallholder: Habari, mheshimiwa. (‛Hello, re- spected sir’.) Have some vegetables. Customer: Mboga gani? Nipe kabeji hizi. (‘Which vegetables? Give me these cabbages.’) How much is that? » (Myers - Scotton 2006: 240, both italics and bold in the original). In addition to these well - recognised types of code - switching, a third type has been in- cluded by some scholars: ‘tag ( - switching ) ’ It is also known as ‘extra - sentential code - switching’ and it is «the insertion of a tag, e.g. ‘you know’, ‘I mean’, from one language into an utterance which is entirely in another language» (Hamers and Blanc 2000: 259 ) Among tag - switches , there are inter- jections and exclam ations. This third category of code - switching could sometimes be considered as a type of inter - sentential code - switching (Adams 2002: 21). 14 The difference between code - switching and loanwords, whose role is significant in recent studies on lexical contact between Latin and Greek ( see below Dickey par 3 ), needs also to be out- lined. Although the boundaries between the two linguistic phenomena are blurry and they can be considered part of a continuum, in modern sociolingu istics a distinction is made based on frequency and integration. In particular, the level of integration depends on three aspects: type (semantic, pho- nological, morphological, prosodic, graphemic), degree (level of integration, which varies based on 12 F ISHMAN 1965, pp. 72 - 77. 13 B ULLOCK - T ORIBIO 2009, pp. 1 - 4. 14 Anticipating examples from ancient languages, we can mention some of the cases Adams presented in his work. He mentioned switches into Greek in Latin texts: see Petronius (37.9) where babae babae is a tag - switch in Latin script of the Greek exclamation βαβαί βαβαί These switches are recorded also in epigraphic texts: IGUR II 294 is an epitaph in Greek where we find , in the end , the tag Dis manibus ; IGUR II 308, on the other hand, is a Latin epitaph which end s with a switch into Greek (in Latin script), i.e. eupsychi tecnon udis athanatos ( ε ὐ ψύχι τέκνον , ο ὐ δ ὶ ς ἀ θάνατος ) Next t o tags found at the beginning or end of texts, Adams considers tags cases like the insertion right after ἀ γαθ ῇ τύχ ῃ of posit memoriam , with the subject of the sentence following the formula in Greek, in the epitaph IGBulg. IV 2116. This is a case that migh t be confused with inter - sentential switching, the difference being that it is a short formulaic expression: A DAMS 2003, pp. 21 - 22. 5 the in tegration type, for example, an item can be phonologically but not morphosyntactically inte- grated) and stability (while code - switches do not become part of the recipient language and are only used by bilingual speakers, borrowings are used consistently and by monolingual speakers as well). 15 Finally, one last term needs to be defined and clarified : ‘ interference ’ The term is often used in discussions on bilingualism, but its reference is often left vaguely or not at all defined Hamers and Blanc (2000: 41) inform us that «the notion of interference, if often used, has never been more clearly defined than as the inappropriate use by a speaker of elements or rules of one language while using the other». The term is used with a more specif ic meaning in studies on second - language acquisition , where it is considered one of the main causes of mistakes by a learner in his second language : it «refers to learning processes in which the L 2 learner inappropriately transfers units of his first lan- gu age to the second» ( Hamers and Blanc 2000: 41 ). When a phenomenon of interference happens, we may assume that we are not in front of what is called ‘balanced bilinguality’, i.e. «a state of bilinguality in which an equivalent competence is reached in both languages , whatever the level of competence » ( Hamers and Blanc 2000: 368) because a ‘balanced bilingual’ knows how to prevent interference from happening From this definition, it appears clear that the main characteristic of the phenomenon termed ‘interference’ is that it is unconscious, unlike accommodation and , especially, code - switching , as standardly used 16 An example of interference can be found in the use of preposi- tional phrases by second - genera tion Slovak immigrants in America studied by Meyerstein (1969). We start from the assumption that Slovak immigrants of the second generation use English as the first language In Standard Slovak , prepositions require to be constructed with a specific case while this does not happen in English. When a preposition is found followed for example by the nominative (the unmarked case), which is never the case required by prepositions in Slovak, this can be due to English interference: for example, we find the preposition od (“from”, “of”), which is normally fol- lowed by the genitive, with the nominative in the expression /od jeden de ň / (Nom.) used by immi- grants, whereas in Standard Slovak we find od jedného d ň a (Gen.), which means in English “from one day”. 17 Before presenting how the study of ancient bilingualism has developed through time and has made used on the concepts presented above, a short paragraph will be dedicated to some of the com- plications and specificities of investigating ancient bilingualism. 2. How to study ancient bilingualism After having presented some milestones in the study of modern bilingualism, we might be wonder ing how these methodologies can be used in the study of ancient bilingualism. The difference in the material under examination is evident: modern soc iolinguistics used as main source the spoken lan- guage, which is not available when we study ancient bilingualism. This is the reason why Fedriani - Napoli (2018) for example have preferred to adopt the definition of ‘textual bilingualism ’ when deal- ing with the ancient world 18 A first distinction that might be made is between spoken and written code - switching. The definition of code - switching presented above presupposes in fact the spoken language and a certain level of spontaneity. This spontaneity cannot be expected when dealing with a written text: «textu al planning may be considered typical of numerous instances of CS in texts» (Fedriani - Napoli 2018: 15 C LYNE 2003, pp. 142 - 146; D ICKEY 2018, pp. 8 - 9. 16 A DAMS 2003, pp. 27 - 28. 17 M EYERSTEIN 1969, pp. 64 - 66, 69 - 70. 18 F EDRIANI - N APOLI 2018, pp. 129 - 131. 6 131). W hen a text is written / inscribed the idea of preserving it is implicit and «the act of writing implies some effort by itself» (Fedriani - Napoli 201 8: 132), which partially contradicts the definition of spoken code - switching quoted above from Bullock and Toribio. Furthermore, we have to think that the texts were written with the idea that they might have survived through time. T he analysis of writt en code - switching has to take into account the readers the text was aimed at and the context of the writing process. This final element involves a series of issues, because many times we do not know the scope of the texts, who read it or even who wro te it. 19 Although the purpose of the present work is the analysis of signs of bilingualism in epigraphic material, a quick account of the examples found in literary texts can be explanatory of the peculiarities of studying ancient code - switching Fedriani and Napoli have developed and studied a corpus of 50 literary texts from the Late Latin literature (A.D. III - VII) and have individuated 4 main reasons for switching into Greek in those texts. The first cause was ‘metalinguistic remarks ’, for ex ample when the author needed to explain the etymology of a word: an example has been found by the scholars in Servius ( Aen. VI, 347) where he explain s the word cortina first in Latin, cortina dicta est aut quod cor teneat [...], and then in Greek, ὅτι τὴν κό ρην τείνει ἤτοι τινάσσει A second reason was to quote literary passages, like in Macrobius ( Sat. 7,7, 16), in reference to what Homer said , [...] Homerus testis est, qui ait, μέλιτι γλυκερῷ καὶ ἡδέϊ οἴνῳ The third motivation was to reproduce a conversat ion with speakers of Greek: again Macrobius ( Sat. 1, 17) telling Pylades’ words which were σὺ βλέπεις The final one, ‘literary play’ is found by them when Ausonius ( Epistula 6) presents more than one switch between the two languages: Aequanimus quod si fueris ετ πάντα μελῷδειν / malueris , λήθη πόνου ἔσσεται ἠδὲ πενείης , “but if you are composed and prefer to sing everything, there will be oblivion of tiredness and poverty”. In this account some of the examples might seem more the effect of a literary game than a sign of bilinguality, especially the quotations of other authors’ works, while the etymological explanations are considered signs of more advanced knowledge of Greek and, there- fore, of bilin gualism 20 For what concerns also epigraphic material , not only literary , four different types of texts have been indicated by Adams (2003) as the source for studying bilingualism, each with their own peculi- arities. Proper ‘bilingual texts’, which are texts where two languages are adopted to convey the same information , although they are not often the translation one of the other. The reaso ns behind the de- cision of writing the same text in two languages can be several, and they need to be individuated by scholars: the most obvious would be to be understood by more people, but it is often due to symbolic reasons. For example, in the Gre ek East we find texts partially in Latin to show deference to the Roman power. Furthermore, a peculiar aspect of ‘bilingual texts’ is that we can often identify the version which was written first and influenced the translated one. For example , as we wi ll see below, the Greek translations of senatus consulta show many interferences of the primary version in Latin on the translated Greek version. 21 A second type of texts are the ‘transliterated’ ones, i.e. the ones which are written in one language but us ed the script of another one, like writing Greek in Latin letters or vice versa. These texts are useful to investigate the level of ‘literacy’ of the writers: while some people in the Roman Empire might have been able to write both in Greek and Latin le tters, others might have been able to speak in one language but not to write in its script. For example, the papyrus P. Lond II. 481 is a ‘Latin - Greek glossary’ written entirely in Greek script: Adams supposes 19 F EDRIANI - N APOLI 2018, pp. 131 - 132. 20 F EDRIANI - N APOLI 2018, pp. 133 - 138, 144 - 145. 21 A DAMS 2003, pp. 30 - 40. 7 that in this case either the writer or the peo ple which it was aimed at were not able to read Latin alphabet. However, in some cases the texts were written in one script for other reasons. Adams men- tions defixiones from Africa, which were written in Latin with Greek characters for magical reasons 22 The third type are ‘mixed - language texts’, which are the ones where code - switching happens. The reason why the switch happens is one of the elements that can be investigated: IG XIV 678, an in- scription from Brundisium (Italy), states Heraclas Heraclidu Alexandreus uixit an(nis) L h(ic) s(itus) , where the first half, i.e. the name, is Greek in Latin script while the second half is in Latin. It is evident that each language was used to convey a specific information, i.e. the name of the deceas ed is in Greek and the rest of the formulaic epitaph is in Latin. As we will see below in par. 5, this choice might be due to the desire to express a specific identity. 23 Finally, there are the ‘texts which implicitly reflect a bilingual situation’ , which are texts where there are some features that indicate that the writer might have been bilingual. Instances of it can be the use of spellings typical of another language: in IGUR 728, which is a funerary inscription, the name of the person to which it was d edicated, D.M. Cointo (in dative), is spelled as if his name was in Greek translation, i.e. Κοίντος instead of Quintus 24 This short account shows how many information on bilinguality can be found in written ma- terial. It also makes clear how every sit uation is peculiar and can unveil interesting aspects of the language written and spoken in the ancient world. In the following paragraph s , we will now present accounts which deal more specifically with Latin and its influence on Greek. 3. The study of the Latin lexicon in Greek One of the first thorough studies of the Latin lexicon attested in Greek texts has been offered by Daris (1991), who presented an overview of Latin loanwords found in Egyptian papyri. He found 800 loanwords over a period of 7 centuries and drew the following conclusions: Latin influence on Gree k lexicon is superficial and limited to the official language (the language of the bureaucracy); the army was the main vector of transmission; Greek - speaking Egyptians preferred to use the corre- sponding Greek forms when they existed; from the age of Diocle tian onwards, the official terminol- ogy is more affected by the influence of Latin, but the Latin presence is still artificial. 25 We will see below that Daris’ conclusions are limited to the corpus he analy s ed and their validity is reduced by the fact that he uses “loanword” to indicate every word of Latin etymology attested at least once in his material, without distinguishing between integrated loanwords and simple code - switches, as hap- pened very often in literature before the introduction of the distincti on between the two concepts. 26 Daris ’ work is still useful but Irene - Maria Cervenka - Ehrenstrasser (1996 - ) is currently work- ing on a complete lexicon of Latin loanwords in Egyptian papyri in several volumes, which will most likely replace Daris. Only two v olumes have been published so far, as far as the letter delta. The items included in the lexicon are well - explained. Each entry is organized as follows: translation of the term, the indication of the Latin word, abbreviations (when existed), etymological d iscussion, synonyms, examples from papyri, bibliography (including lexica and other literature) and thorough commen- tary. 27 22 A DAMS 2003, pp. 40 - 63. 23 A DAMS 2003, pp. 67 - 70. 24 A DAMS 2003 , pp. 70 - 72. 25 D ARIS 1991, pp. 17 - 18. 26 D ICKEY 2012, pp.58 - 59. 27 C ERVENKA - E HRENSTRASSER 1996, pp. 15 - 17. 8 A work on a specific sector of the lexicon, i.e. Greek terms for Romain institutions, has been offered by Mason (1974), which deals with how Roman political terms have been rendered in Greek. The author decided to continue Magie ’s work (1906) De Romanorum iuris publicique sermonis vo- cabulis sollemnibus in Graecum sermonem conversis but with a different perspective. If Magie’s work focu sed on the different «linguistic methods of translation» (Mason 1974: ix), Mason decided to offer a tool where to understand the distribution of the different terms found in Greek texts, which ones were more common and which ones were rare. Examples were t aken from literature, inscriptions and papyri. In general, the material cited does not go after Diocletian’s time (including Cassius Dio, Herodian and Philostratus), but the author decided to add the material from De Magistratibus Roma- nis (John the Lydian) for the significance it has on a study on official terms for Roman institutions. 28 The work consists of two main parts: an alphabetical «Greek - to - Latin lexicon», where each Greek item is followed by the Latin correspondent and it is explained through a se ries of examples which present the context where the specific word was used, and a «discussion of selected terms», where words which deserved further analysis and contextualization are discussed in more detail. In this second part, there are, for example, terms which were widespread in Latin official texts, such as Lat. senatus , which was translated with the Greek adjective σύγκλητος (βουλή), “called together, sum- moned” (LSJ), while βουλή alone was adopted to indicate the small local councils. Other rarer v ari- ants were γερουσία and συνεδρίον, which is found in Polybius. 29 Mason significantly noticed that most of the words recorded in his lexicon are not loanwords: «perhaps 150 of the approximately 1200 items in the glossary are Latin words» (1974: 3). The two main chapters are preceded by a useful introduction and followed by a «Latin - to - Greek reverse index». Although it is almost fifty years old, Mason’s study is still of great value when dealing with Greek documents which mention Roman magistrates or, in general, Roman institutions. The Greek - to - Latin lexicon is well - orga nized, com- plete and easy to use, and the analysis of selected terms guides the scholars in understanding the differences between items which were adopted to translate the same Roman institution. A long - awaited innovation in the approach to the study of borrowings and lexical interference of Latin on Greek has been made thanks to the contributions of Eleanor Dickey. Her explanations on the methodology to be used to distinguish loanwords and codeswi tches are of extraordinary value and deserve to be presented in detail in the following pages. First of all, she has been the one who outlined that there cannot be an agreement among scholars in the number of Latin loanwords found in Greek because, as ment ioned above, in many contributions «“Latin loanword” means “word of Latin ety- mology attested at least once in a Greek text”» (Dickey 2012: 59). She also explained why the focus has been placed, in the past, mainly on Greek papyri: in literary texts as well as epigraphy the Latin words attested may have been used just because the topic concerned Roman realities or because they were the translation of Latin official texts. However, although many examples are found in papyri, some might be codeswitches made by bilinguals. Her studies started with this idea and aimed to distinguish “real” (“integrated”) loanwords from codeswitches. Some of the differences are the fol- lowing: loanwords are not a feature typical of bilingual speakers because they are often used by mon- olinguals while code - switches can be only made by people who can switch from one language to the other; the same word can be rendered in different ways when it is a codeswitch while it is adapted in only one way when it is a loanword; «a loanword has a specific, consistent place in the lexicon of the 28 M ASON 1974, pp. ix - xi. 29 M ASON 1974, pp. 121 - 124. 9 borrowing language, while a codeswitch often alternates with a native term that would normally be used for the same idea» (Dickey 2018: 9). 30 The first thing to do to distinguish the two phenomena in ancie nt texts was to find some criteria to make this distinction. The easiest marker would have been «graphemic adaptation - or the lack thereof, namely graphemic distinction» (Dickey 2018: 11) since it is difficult to detect phonetic ad- aptation in written te xts: code - switches would have been only the cases where the words were spelt using Latin alphabet instead of the Greek one. However, codeswitches written in the Latin alphabet are very rare. 31 The adoption of the same criteria used in modern socio - lingui stics, which have been briefly presented above (par. 1.1), i.e. frequency and integration, is useful but more complex when dealing with ancient material. «Morphological adaptation», for example, was very frequent in Greek without considering whether the wo rds were codeswitches or integrated loanwords: «Latin nouns and adjec- tives almost always appear in Greek not with Latin endings, but with the endings appropriate to the equivalent declension in Greek: - us becomes - ος, - um becomes - ον, - am becomes - αν, etc » (Dickey 2012: 61, italics in the original). 32 However, there are cases in which morphological adaptation did not happen or was not frequent: in case of single words, the ones belonging to Latin third declension were frequently only transliterated without adaptation, due to the stronger differences between these declensions in Greek and Latin (for ex. Dickey, p. 22, mentions the use of βρέβεμ for Lat. brevem in Plutarch, Moralia 281d in Nachstädt et al. 1935:314); on the other hand, phrases remained often unadapted, independently from their being loans or code - switches, although examples of adaptation of at least the second member of the phrase are attested. 33 Another element often used to understand whether the word was an “integrated loanword” or not i s its survival in Modern Greek. This criterion might be useful when the word was attested both in the ancient and modern language but it involves some difficulties: Latin detained its role of lan- guage of culture long after the fall of the Roman Empire and words of Latin origin might have been borrowed by Modern Greek also in Medieval times, and the lexicon of a language constantly evolves so Latin - derived words could appear in every phase of the history of a language. 34 In addition to those valid principl es, the context of use needs to be taken into account: «some contexts make it clear that a word is not part of the writer’s language, by labelling it as a Latin word and/or by explaining what it means and thereby indicating that readers are assumed not to know the word’s meaning» (Dickey 2012: 61) and, conversely, in some cases, it is evident that the word had become of common use. For example, Dickey mentions several passages from grammarians who used a Latin word to explain other Greek words, demonstratin g that the Latin word was commonly used and widely understood, such as in the following case (Dickey 2012: 62): 30 D ICKEY 2018, pp. 7 - 9. 31 D ICKEY 2012, p.60; 2018, pp. 11 - 12. Examples of lack of graphemic adaptation have been found by Dickey in lat e authors like the legal writer Theophilus Antecessor (AD VI), whose use of loanwords and not transliterated words Dickey thoroughly analy s ed in her contribution from 2018. Theophilus Antecessor used to write common loanwords in Greek alphabet and rare Latin words, or at least words which he perceived as foreign, in Latin alphabet. However, some incon- sistencies have been found in his usage: som e words have been written sometimes in Greek and sometimes in Latin alphabet. In these cases, an analysis of context helped in understanding his usage, reaching the conclusion that it depended on what he considered of foreign origin in a specific context a nd it does not reflect the modern distinction between loan- words and code - switching: D ICKEY 2018, pp. 12 - 18, 33 - 34. 32 D ICKEY 2012, pp. 60 - 61; 2018, pp. 19 - 20. 33 D ICKEY 2018, pp. 22 - 33. 34 D ICKEY 2012, pp. 63 - 64. 10 ἀ ργυροθ ή κη, τ ὸ ν ῦ ν ἀ ργεντ ά ριον καλο ύ μενον· παρ ὰ Διοκλε ῖ · ἔ στι δ ὲ ο ὗ τος τ ῆ ς ἀ ρχα ί ας κωμ ῳ δ ί ας ποιητ ή ς. ([Herodian], Philetaeru s 194) Ἀ ργυροθ ή κη is the thing now called ἀ ργεντ ά ριον; the word is found in Diocles, and he is a poet of Old Com- edy. (bank; Latin argentaria) Other cases presented by Dickey are taken from texts which are known for having been composed for fruition by Greek - speakers only who were expected to understand also the words of Latin origin attested, like the Greek version of the New Testament. 35 Dickey is currently about to publish a ‟ Lexicon of Latin loanwords in Ancient Greek” where she applie s the criteria mentioned above to identify integrated loanwords. The conclusion she pre- sented on the preliminary results of her study, published in previous articles, are the following: the assumption that Latin loanwords in Greek are mostly words concernin g the semantic fields of politics, army, law, administration revealed to be wrong when the discussion is on “integrated loanwords”, which are often «common, everyday concepts, in some cases what linguists refer to as “basic vocab- ulary”» (Dickey 2012: 66); “integrated loanwords” appeared in every semantic field; the idea that most loanwords entered the Greek language from the age of Diocletian (see Daris above) appeared to be wrong, because in Dickey’s corpus they started to be used between II BC and AD II. 36 The criteria and methodologies offered by Dickey are of great significance for the present study and the final version of her lexicon will be a fundamental tool for every scholar working with language in- terference between Latin and Greek. 4. Studies co ncerning epigraphic material Before approaching an overview of some of the main studies that worked on contact between Greek and Latin in epigraphic texts, we will briefly present some of the «theoretical considerations» outlined by Langslow (2012a: 290) which need to be taken into ac count when working with «linguis- tic evidence», especially, but not only, epigraphical, which he presented giving examples of Latin in contact with Italic languages (mostly, but not only, Oscan and Umbrian). First of all, in historical sociolinguistics, som e caution is required when drawing conclusions on the time and way “language shift” happened when the only thing we know for sure is that there was some bilingualism which led to language death. Assistance can be sought from typological considerations, whi ch for example were successfully used, according to Langslow, by Mullen when studying language contact between Gaul- ish, Latin and Greek in Gaul. 37 Another i