The Greek suffix - ozos A Case Study in Loan Suffixation

This paper offers a morphological analysis of the borrowed derivational suffix - όζος [ózos], used in both a number of Modern Greek (MGr) dialects and in Standard Modern Greek (SMGr). It draws on an extensive corpus to examine the suffix from both a synchronic and a diachronic perspective. Our diachronic analysis emphasizes the geographical distribution, the etymological provenance of the suffix, and the loan accommodation strategies employed in various MGr dialects, thus providing some interesting etymological findings regarding the lexical stock of Modern Greek (Standard and dialects). Our synchronic analysis focuses on the stem categories with which the suffix combines and accounts for the phonological, morphological, and syntactic constraints that function during the derivational process.


Introduction
This paper provides a morphological analysis of the borrowed derivational suffix -όζος [ózos], which has not until now been systematically investigated. The suffix is used in a number of Modern Greek (MGr) dialects, mainly to form adjectives, as shown in (1): (1) a. σωματόζος [somatózos] Myconos, Paros, Zakynthos 'stout' b. αιματόζος [ematózos] Kythira 'scarlet' Here, in the present article, we draw on an extensive corpus to examine the suffix -όζος [ózos] from both a synchronic and a diachronic perspective. Our diachronic analysis focuses on the geographical distribution and the etymological provenance of the suffix -ózos and on loan accommodation strategies employed in the various MGr dialects (Haspelmath 2008;Anastasiadi-Symeonidi 1994). We find parallels for -ózos in other suffixes, such as -(ι)άρης [(i)áris] and -ίσιος [ísços] and then offer some interesting primary observations. Our synchronic analysis describes the stem categories with which the suffix combines and so accounts for the phonological, morphological, and syntactic constraints that are found to be operative in the derivational process (see Ralli 2005: 154-157). The paper attempts above all to shed light on a corpus of very interesting and hitherto unknown dialectal data and is an example of contemporary research on the etymology of the lexical stock of the MGr dialects. It also offers a case-study of a specific suffix and so attempts to confirm to a large extent the generalizations so far established by modern cross-linguistic research on language change driven by contact.
The corpus of our dialectal data is derived from the five volumes of the Historical Dictionary of Modern Greek (ilne)  and from the unpublished Archive of the Research Centre for Modern Greek Dialects of the Academy of Athens (Archive of kendi-ilne), which contains a considerable quantity of excerpted manuscript collections of oral dialectal speech and written sources, in the form of dictionaries, glossaries, and so on, accompanied by retrievable metadata.
Here we should note that our etymological suggestions for all dialectal forms rest on the phonological, morphological, semantic, and space-time criteria proposed in recent literature (see Moisiadis 2005, Katsouda 2012. More particularly, for the etymology of any dialectal form, our analysis takes account of a) the related phonological processes and developments in Greek and in Romance, b) the formal morphological similarity or difference between the Greek forms and corresponding Romance forms, c) semantic similarities and differences between cognate Romance forms and d) the socio-historical context of each Modern Greek-speaking area.

Historical and Social-Linguistic Conditions
The duration and the intensity of contact between the donor and the recipient language determines the type of structures borrowed, a factor expressed in hierarchical terms in Thomason & Kaufman's borrowability scale (1988: 74-75). Thus a closer look at the socio-historical conditions informing contact between Greek and Romance is necessary.
In what follows here we offer a short sketch of Greek history during the medieval and early modern period. We focus on areas and conditions dictating language contact so as to set the specific borrowing process we are examining here in a wider context. Further details can be found in Browning (1999) and Horrocks (2010).
The battle of Manzikert (1071) was a landmark in the history of the late Byzantine Empire since the Seljuk Turks (and the later Ottoman Turks) conquered large parts of Asia Minor, thus triggering the decay of the Byzantine empire. The decline accelerated with the capture by the Crusaders of Constantinople in 1204, an event that gradually brought many Greek-speaking areas under the control of Romance-speaking conquerors via the foundation of "western-ruled states" (Markopoulos 2014: 90). Rhodes came under the military rule of the monastic order, the Knights of Saint John, in 1310 and remained there until 1522. Cyprus was held by the Frankish Lusignan family from 1191 to 1489, this being followed by the Venetian occupation, which ended in 1571. The Duchy of Athens (1205-1458) was held by a Burgundian knight who had arrived with the Fourth Crusade, Otto de la Roche, to pass in 1388 into the possession of the Florentine Acciaiuoli family. The Venetian Sanudo family founded the Duchy of Naxos (1207-1579), which included most of the Aegean Islands, in particular the Cyclades. Many of the islands of the Aegean remained for centuries in the hands of the Venetians, cases in point being Myconos (1204-1390), Tinos (1204-1715), Naxos (1207-1566), and Syros (1204-1579. Moreover, the Lordship of Negroponte, which included Euboea, was subject to Venetian influence (1205-1470). Chios was occupied by Genoese (1329-1566), while Lesbos was held by the Genovese Gattelusi family (1355-1462). The Venetians held Crete for more than four centuries, from 1211 till 1669, at which time the Ottoman occupation began. In the Ionian Islands, the Venetian occupation was lengthy, lasting sometimes for more than six centuries, namely from 1207 to 1797. Such longevity of occupation can be seen in the cases of Corfu and Paxoi (1207-1214and 1386-1797), Lefkada, (1684-1797, Zakynthos (1479-1797), Cephalonia andIthaca (1500-1797), and Cythera (1207-1797). The Venetians controlled many Peloponnesian areas, such as Methoni andKoroni (1207-1500), and many areas of Epirus, such as Parga (1401-1797), Preveza (1717-1797), and of Etoloacarnania, such as Nafpaktos (1407-1540) and Vonitsa (1684-1797) (see Savvidis 2003: 187-209, Markopoulos 2012: 425-428, Markopoulos 2014. Clearly, the co-existence of Greeks and Romance speakers did not possess the same duration, intensity and features in all these areas. However, as Markopoulos observes (2014: 90), despite differences, the co-existence of two communities makes language contact inevitable, which implies the existence of bilingualism, full or partial.
In some cases, full bilingualism in Greek occurs in areas where Romance minorities mingled with the local Greek population. This happened, for example, in Cyprus by the 15th and in Crete by the 17th c. (Markopoulos 2014: 91). On the other hand, bilingualism in Romance was for some Greek speakers a means of upward social mobility. Another important cause of full bilingualism was intermarriage, very common in Crete and Cyprus, for example. Finally, partial bilingualism is to be explained as result of Romance-Greek trade relations in Mediterranean ports, which were obviously facilitated through the use of a common language (Markopoulos 2014: 91).
Bilingualism was of course mainly an urban matter (see Terkourafi 2005, Manolessou 2008, Markopoulos 2012: Romance speakers, for instance, were to be found in trade ports and only rarely in the countryside (Markopoulos 2014: 91). A typical example of this is the case of Cythera in Venetian times, where edicts were usually left untranslated for the upper class, who lived in the Fortezza and who understood the Venetian/ Italian language. On the other hand, the peasants learned of the content of every edict via interpreters, who translated them "in idioma greco a parola per parola" (Maltezou 2008: 134, Leontsinis 2005.

The Romance Suffix -oso
Although the suffix -oso is very productive in all Romance languages, the Latin -ōsus (Ernout 1949) and the Italian suffix -oso (Magni 2001) have been the most deeply studied. Notably, the suffix belongs to the very few patrimonial suffixes transmitted grammatically and semantically intact from Latin into Roman languages (Pharies 2015(Pharies : 1857. A main syntactic constraint is that, prototypically, both the Latin and the Italian suffix are added to nominal stems (Magni 2001: 4-5), e.g., (2) a. glori(a) 'glory' + -ōsus > Lat. gloriōsus 'glorious' [[glori(a)  However, Montermini (2001) notes that verbal derivatives in -oso are also attested, which, in the view of Ernout (1949: 77), are rare later formations, e.g., More specifically, the nominal stems onto which the suffixes -ōsus/ -oso are added, belong to nouns, common-concrete or abstract-and inanimate (Scalise 1994: 99), although there are new formations based on the stem of animate nouns (Magni 2001: 4-6),1 e.g., In the view of Magni (2001: 12-13), adjectives formed with -ōsus/-oso have the meaning "who/ what possesses a quantity or quality x", when x is that which the nominal stem expresses (see also Rohlfs 1969: 441, Lehmann 19775: 231, Tekavčić 1980. In addition to the prototypical meaning "full of x", the suffix has also some more peripheral meanings, such as "who does x", "who/ what is characterized by x", "who/ what is similar to x" (Magni 2001: 8).
Latin-and Italian-suffixed adjectives have pejorative connotations (Magni 2001: 14, 18). Ernout (1949: 80) observes that a large number of medical terms were formed with -ōsus. The suffix was added to nominal stems of nouns denoting illness, a specific harm or the affected part of the body (e.g., morbōsus 'morbid' < morbus 'disease').
Another interesting point is that both the Latin and the Italian suffix form adjectives which may have both agent and patient thematic roles, i.e., can be analyzed in terms of either active or passive verb phrases. For example, the Latin invidiōsus can be interpreted in two ways (Magni 2001: 17): '(one) who is envied (i.e., enviable)' As for the referent, it can be either animate or inanimate (Magni 2001: 18). Thus the presumed referent in (7a) is animate, while that in (7b) is inanimate: 'what provokes fear' To conclude: the suffix -ōsus/-oso forms adjectives that refer to persons or objects, meaning, in either a positive or pejorative manner, "who/ what possesses a quantity or quality x", when x is that which the nominal stem denotes. The Latin suffix occurs in many Romance languages with the same function, as for example in French (Ahn 1994) and Spanish (Prieto Prieto 2002: 1580. Finally, one may note that the suffix is productive in English, too. The suffix, borrowed by English sometime after the 15th century, has two forms, the form -ose being employed in two-syllable words and -ous in polysyllabics (Miller 2012: 166;See also Lass 1999: 3 vol., 403).

4.1
The First Evidence Latin words in -ōsus were borrowed into Greek during medieval times (5/6th c.-13th c.), as can be seen in (8) Such cases are few, probably because of the scarcity of textual sources from the 8th and 9th century, which therefore does not allow one to form a complete picture of the later stages of early Medieval Greek (see Holton & Manolessou 2010: 542-543). Nevertheless, a few points can be made. Forms such as βιτίοσον, γρατίοσος, ρελιγίοσον3 have not survived in MGr dialects. Furthermore, the majority of the loanwords in -oso imported in MGr dialects mentioned below do not have a corresponding Latin form. It is also to be noted that the meaning of the Latin loanwords above is not to be found in any of the MGr dialects. For example, the loanword κουριόσος, 'officer responsible for the invigilation of travelers' , retains the core meaning of the Latin word curiōsus 'who oversees something' . The type curioso, meaning 'who wants to know' , appears at the end of the 14th c. (Cortelazzo & Zolli 19992 Thus curioso ('curious') was loaned from Venetian into MGr dialects with this meaning after the 14th c. (see Boerio 1856).
Consequently, if one employs morphological and semantic criteria,4 it can be argued that the loan suffix -oso was taken into MGr dialects through Venetian and Italian loanwords, rather than through Latin loanwords, as we will demonstrate below.
Greek texts that document early appearances of loanwords in -ózos are rare. The following instances may be quoted: These loanwords, dating from the 15th, 16th, and 17th c., survive in MGr dialects, except for φραντσόζος, which is not attested.

4.2
The Donor Language The donor language of the majority of the loanwords in -oso in MGr dialects is mainly the Venetian dialect and the Italian language. Of course these are not the only donor languages involved because, as we mentioned in section 3, after the capture of Constantinople in 1204, Greek-speaking areas fell under Romance domination, both Venetian and Catalan, Genoese, Provençal, Frankish or Lombard.6 For example, Thus different donor languages may often be responsible for related loan dialectal forms. In many cases, it is easy to spot the Venetian dialect as the donor language, as the forms in the recipient dialects preserve that of the Venetian word, which differs from the Standard Italian form. For example: contrasting with the It. cognate form amoroso.
(13) πιτιτόζος [pititózos] Corfu < Ven. petitoso 'selective in food' contrasting with the It. cognate form appetitoso.7 Unfortunately, chronological evidence concerning the appearance of these forms in the Venetian dialect is available for only some of the loanwords located.8 In other cases, on the other hand, it is difficult to decide whether Standard Italian or Venetian was the donor language, for two reasons, namely: a) the loanword appears in the same form in both varieties. For example, Journal of Greek Linguistics 16 (2016)  The adjective may have existed in Venetian, since the noun avantazo 'advantage' exists in Venetian, and the corresponding forms of the Italian noun and adjective are vantaggio and vantaggioso respectively. Of course, however, one should bear in mind that Standard Italian was adopted as the official language by Venice long before the unity of Italy in the second half of the 19th c. (Fanciullo 2008). So, often it is not easy to discern whether a loanword was borrowed from Venetian or from Standard Italian. Because chronologically based data is generally lacking, one has to use the first appearance of the adjective in -oso in the donor language as a terminus post quem for insertion of the loanword in MGr dialects. In the Index below, the tables display the loan adjectives in -ózos found both in MGr dialects and in Standard MGr (SMGr). They give every dialectal form followed by its geographic distribution, the meanings of all accommodated Greek forms and, lastly, the postulated donor language. The data is divided in groups formulated according to the century of the first appearance of the word in question (Cortelazzo & Zolli 19992). The majority of loanwords attested in the donor language in the 13th and 14th c. have a corresponding Latin form, except for the loanwords γρατσιόζος 'gracious' , κουραγιόζος 'courageous' , λεμεντόζος 'complainer' , βαλερόζος 'worthy, capable' , δεσπετόζος 'one who throws tantrums' , ραμόζος 'possessing many branches' , and φλατόζος 'ill because of a chronic disease' (and φλατουόζος 'of food, causing gas'), although, as already stated in the previous section, the meaning of the Greek loanwords may be very different. On the other hand, all the adjectives adopted in MGr dialects (apart from ρεντικολόζος, see Index id5) after the 15th c. are formations without a corresponding form in Latin (see Index ic, id 1-4, 6, ie 1-5, If, and Ig). Taken together, these two facts, that is, semantic differentiation from Latin and the absence of a Latin original after the 15th c., further reinforce the assumption that the language from which the words were borrowed was Italian or Venetian, rather than Latin. The loanwords φλατουόζος (see Index ib12b) and τσιριμονιόζος (see Index ic2a) belong to the rare cases listed in dictionaries of the 19th c., for example, in that of Byzantios (18743) and that of Peridis (1878) respectively.

Loanwords in -ózos in MGr Dialects
As is evident from Table 1 above, we identified just one loanword in -ózos in an area covering the Sporades, Bithynia, the Chalcidice, Ionia, Kos, Megara, Propontis, and even Crete. The Cyclades, Cyprus, South Peloponnese (Mani, Messinia), and above all the Ionian Islands are the areas where most of the loanwords mentioned above, that is 25 out of 39, are attested.
Note that the recipient dialects of the bulk of these loanwords are spoken in regions where there was intense language contact with Venetian and Italian due to historical and geographic reasons connected mainly with the Venetian supremacy, which in many areas, as we have noted, lasted more than three centuries, and with the proximity of these areas to the Italian peninsula.
With regard to the accommodation strategy, Haspelmath (2008: 8) comments that when dealing with borrowed adjectives, note should be taken of both donor and the recipient systems. Structural properties of both the donor and the recipient language are extremely important in the choice of borrowing strategy to be followed (Anastasiadi-Symeonidi 1994, Ralli 2011cf. Ralli 2012;see also Meillet 1921: 82, McMahon 1994. As for the -ózos suffix, since Greek and Romance belong to the same linguistic family, they therefore share many structural similarities, such as derivational and inflectional strategies. Furthermore, Greek and Italian have very similar phonological systems, a feature that facilitates the integration of linguistic material from one language into another. In particular, Anastasiadi-Symeonidi (1994: 195) notes that, during the stage of secondary morphological adaption, the final element of the stem and the position where stress is placed play a crucial role in the accommodation of a loan noun or adjective to an inflectional paradigm of the recipient language. The phonological similarity of the final vowel between the loanwords in -oso with the Greek nominal stems in -o facilitates their adoption in Greek. They are formed according the model of the adjectives -ος, -α, -ο, i.e., κοστόζ-ος, -α, -ο9 [kostózos/ kostóza/ kostózo] // ωραί-ος, -α, -ο [oréos/ oréa/ oréo]. At this point, it should be noted that in many dialectal varieties the neuter gender is formed with the suffix -όz-o and not -όz-iko, although in SMGr the predominant suffix is -όz-iko. The prevalence of this suffix -όz-íko, όz-ici, -όz-iko can be accounted for, if one recalls that the majority of the adjectives in -ózos possess mainly a masculine and a feminine form (i.e., κομπλιμεντόζος/ κομπλιμεντόζα 'complimentary'), as their referent does [+animate]. These forms can be substantivized by conversion and afterwards may form a new adjective with the suffix -ikos (cf. the formation of the neuter of adjectives in -(i)áris, see Anastasiadi- All loanwords function as adjectives in the recipient language, except for the noun αμορόζος and μαφιόζος, which had already been substantivized in Italian and Venetian. During subsequent stages, adapted loanwords were combined with Greek suffixes, such as -ικος, -οσύνη, -ιά, -ιάρης, -λογώ or -εύομαι, to produce new derivatives or compounds such as: As regards semantics, the recipient varieties seem to have preserved one of the meanings of the donor language, maintaining the same connotations. In some cases the [-inanimate] referent of the donor language changes to [+animate] in the recipient varieties. For example, the Ven. petitoso 'causing appetite' was adopted in Corfu with the meaning 'person selective as regards food' .

New Formations in -ózos
In the corpus we examined, there occur nineteen (19) new dialectal forms in -ózos.10 Meyer was the first to note, in 1895, the productivity of the suffix -ózos in MGr dialects (see Meyer 1895: 100). Because there is no other chronological 10 Anastasiadi-Symeonidi (2002) also indexes the word πανηγυριόζος 'full of festivity' , al-evidence, the end of the 19th c. and in particular the year 1895 offers a terminus ante quem for the first appearance of the words dealt with in the previous section. As Rainer (2015Rainer ( : 1777 notes, during the first step complex words are borrowed and during the second the pattern for new formations is extracted language-internally from the borrowed words.11 In our case, the first loanwords in -ózos were borrowed some time after the 13th c. and the first new formations are attested six centuries later. Six (6)  The adjectives μπουγιόζος and μπουγιόζικος are not listed in any of the dictionaries above.

New Formations in -ózos from Greek Nominal Stems
In addition to the examples above, new adjectives derived from the addition of the suffix -ózos to Greek nominal stems of feminine nouns in -a or -i were located: For further information on the new formations in -ózos, see Index ii 1-19. In some of these cases, the formation derives from the translation of the borrowed stem into Greek. For instance, the adverbs θυμόζος and αιματόζος probably followed the pattern of the Ven./Ital. rabbioso and sanguioso respectively, in which the stems rabbi(a) 'anger' and sangue 'blood' are translated in Greek (cf. Anastasiadi-Symeonidi 2009: 64).
The seventeen new formations (if we ignore the SMGr φασαριόζος and μπουγιόζος) are found mostly in the varieties current in the Ionian Islands and the Cyclades, where the criterion of Thomason & Kaufman's scale (1988: 74-75 valid. That is, the more borrowed categories a recipient language displays, the more intense the contact it has had with the donor language. In these particular varieties, the degree of exposure to the source language and the intensity of the language contact are especially high and consequently, if one leaves aside the very high number of borrowed content words, there is a large number of function words,13 the presence of which accounts both for the productivity of -ózos formations and for the occurrence of the suffix itself (see Table 2). The morphological segmentation of the Italian/ Venetian loanwords into stem and suffix that gave birth to the these new formations was triggered by several factors, namely: a) the large number of loanwords in -ózos and the existence of certain pairs of loanwords with and without the morpheme (Weinreich 19682: 31-37; see also Anastasiadi-Symenonidi 1994: 69-70). For example, both (24a) and (24b) coexist in Kythira: For instance, in the Heptanesian dialects a large number of function words is attested (i.e. αμάgο/μάgο 'at least' < mánco, ενόρδινε 'all right' < in ordine, δεμπότο 'nearly' < deboto, δελόγκο 'immediately' < delongo etc.) and many productive suffixes (i.e., -άδα < -ada, -ίνο < -ino, -όνι < -one, -ίρω <-ir(e), -άρω < -ar(e) etc. These pairs make such structures transparent for the speaker, allowing him to identify a derivational morpheme in terms of loanwords, to analyze it, and later, when the morpheme becomes productive, to use it in the formation of new words (cf. Moravcsik 1978, Gómez Capuz 1997: 85, Field 2002: 38, Johanson 2002, Aikhenvald 2007,14 b) the semantic autonomy that, according to Moravcsik (1978), Field (2002) and Johanson (2002), facilitates the borrowing process. The segmentation of loanwords into stems makes it easy for speakers to extract the core meaning of the suffix 'who/that has/presents a quantity or quality x when x is what the nominal stem expresses' (Magni 2001: 12-13), and c) the stem-based nature of Greek morphology, in that words consist of a bound element (cf. Ralli 2012).

Constraints on the Derivation of New Formations in -ózos
The suffix -ózos in the examples in all of the new formations given immediately above is added to nominal stems.15 On these grounds, Meyer's (1895: 100) etymology of ευλαβιόζος, which traces the word from the adjective ευλαβής ('devout'), is to be rejected. In any case, Meyer's suggestion cannot explain the presence of [i] before the suffix. In fact, the word derives from the noun ευλάβεια 'devoutness' .
The case of σωματόζος makes it clear that new formations obey one main morphological constraint on the recipient language, that is, that the suffix -ózos is added only to the augmented allomorph of nouns in -ma. Thus we find σωματ-όζος, rather than *σωμ-όζος. Neuter nouns in -ma (Ralli's category κτ8, see 14 See also Anastasiadi-Symenonidi (1985: 91) on the productivity of the derivational loan suffix -e and Ralli (2012) on the productivity of the loan suffix -áro. Both argue that the coexistence of the suffixes with other etymological relative loanwords in the recipient language triggered the segmentation of the loanwords into stems. 15 On constraints on derivation, see Ralli (2005: 154: 157 Ralli 2005: 121) possess two allomorphs, ending in -ma and -mat-, although only the second can participate in derivational and compositional processes. Another significant point is that the specific suffixation process in MGr does not obey the constraint of the distinguishing lexical features of the stems. The suffix -ózos shows a preference for both stems of foreign origin (Italian, Venetian, Turkish) (see Index ii 3,6,7,9,11,15,17,18,19) and native Greek stems (see also Index ii 1,2,4,5,8,10,14,16).
The reason why this suffixation became productive in the MGr dialects may spring from the fact that the suffixes -ώδης (i.e. σωματώδης 'stout' , θυμώδης 'very angry' , αιματώδης 'with much blood') or -ης (i.e. ευλαβής 'devout') are: a) [+ learned], b) display morphological irregularities in the formation of certain cases of the singular and plural, leading even native speakers to produce erroneous inflectional forms. For example, a frequent production error is *το ευλαβή άτομο Nom. Sing. [evlaví átomo] instead of το ευλαβές άτομο Nom. Sing. [evlavés átomo] 'devout person' , or *του σωματώδη άντρα Gen. Sing. [somatóði ándra] instead of του σωματώδους άντρα Gen. Sing, [somatóðus ándra] 'of stout man' (see Savvaidou 2012), c) do not discriminate the masculine from the feminine forms, i.e., σωματώδης άντρας [somatóðis ándras] 'stout man' -σωματώδης γυναίκα [somatóðis ʝinéka] 'stout woman' (cf. Anastasiadi-Symeonidi 1985: 100-101). 16 This innovative suffixation process was thus not facilitated by the presence of any words already existing in the recipient language, which of course means that Aronoff's blocking constraint (1976: 43-44) still stands. Non-Standard MGr dialects drew on alternative strategies that arose from the various historical and sociolinguistic conditions that shaped these dialects in order to deal with the inflectional difficulties inherent in the unwieldy learned -ώδης suffix. SMGr, on the other hand, which had less intense contact with Romance languages, had a much greater tolerance for items of learned origin and a conversely lower tolerance for borrowed words, a situation that did clearly not promote the spread of the -ózos suffix.
Lastly, one notes that the core meaning of the loan suffix is maintained in the most derivatives in -ózos. The new formations express, in either a positive or pejorative manner, that somebody or something has the quality or quantity of the base. The new formations have both active (for example, τροπόζος 'who has very good manners') and passive readings (for example, γινατόζος 'willful or done with obstinacy'), while the referent is mostly animate (ευλαβιόζος 'devout'), but may also be inanimate (αιματόζος 'full of blood'). 16 Anastasiadi-Symeonidi (1985: 100-101)'s explanation of the establishment of the the suffix -e in Koine Greek employs the same argument.
Attestations for these words are offered in both SMGr dictionaries (Babiniotis Dictionary 20124, Triantafyllidis Dictionary 20077, and Anastasiadi-Symeonidi Reverse Dictionary 2002) and in electronic corpora (ilsp, sek, potheg). Their ability to form secondary formations (-ózos vs. -όzikos) has been statistically examined, though the intuitions of individual speakers regarding each particular lexical item may vary, thus slightly affecting the validity of conclusions drawn from these statistics. Of these, nine are loanwords and just one (φασαριόζος) is actually a new formation. All the adjectives and the one noun in -ózos in SMGr refer mainly to animate referents. Furthermore, both adjectives and noun form secondary adjectives in -ikos, such as γουστόζικος, φουριόζικος, φασαριόζικος, μαφιόζικος, which also refer to inanimate referents.

Comparison with Other Borrowed Suffixes
The case of the borrowed suffix -ózos is not unique. Since contact between Greek and Latin and other Romance languages was so intense, several suffixes forming adjectives were adopted by Greek (smg and MGdialects), such as -(ι)άρης [(i)áris] and -ίσιος [ísços]. The Latin suffix -(i)aris ⟨-(ι)άρης⟩ forms denominal adjectives or nouns, i.e., A comparison of these suffixes with -ózos makes it clear that the borrowing was triggered by the influx of a large, possibly even massive, number of loanwords in -όzos, -(i)áris and -ísios, where the core meaning of the suffixes is extractable. Their absorption into Greek was also aided by the borrowing, at the same time, of their noun base, which made the point of contact between stem and suffix clearly visible (cf. Anastasiadi-Symeonidi 2009: 63-64; see also above in 5.1), as in, for example, (28) a. γουστ-όζος 'tasteful' -γούστο 'taste' (< Ven. / Ital. gusto) b. καμπ-ίσιος 'of the plain' -κάμπος 'plain' (< Lat. campus) c. ταβερν-ιάρης 'publican' -ταβέρνα 'tavern' (< Lat. taberna) Furthermore, some of the new formations in -ózos and -ísios arose from the translation of the noun base, with which the suffix was then combined (Anastasiadi-Symeonidi 2009: 64; see also above in 5.1), e.g., (29) a. θυμ-όζος (< rabbi-oso 'ireful' < rabbia = θυμός 'anger') 'ireful' b. λιβαδ-ίσιος (< prat-ensis 'of the meadow' < partum = λιβάδι 'meadow') 'of the meadow' All three suffixes are [-learned] and are also combined with [-learned] noun bases. Adjectives (or nouns) in -όzos, -(i)áris, and -ísios are not used in academic or learned terminology and can be expected to occur in informal speech and in colloquial registers.  (2008), the influence of Old French limited or wiped out the native English suffixes -lice and -bǽre (see also Trips 2014: 404-406;cf. Scherer 2015cf. Scherer : 1785. In fact, in terms of type frequency (Scherer 2015(Scherer : 1783, the suffix -ózos is not as productive as -(i)aris and -isios in SMGr, although, in the view of Anastasiadi-Symeonidi (2009: 69-70), even the suffix -ísios is apparently retreating today in SMGr, in the face of other competitive suffixes (-ίνος, -ίστικος, -ιάτικος etc.). To conclude, it is clearly the case that in the borrowing process of all three [-learned] suffixes, even if they were borrowed at different times, the same conditions applied. Initially complete lexical items, set in a transparent structure, were borrowed, although subsequently the suffix was detached and combined with other stems, to create new forms.

Conclusions
The suffix -ózos was adopted into Greek via the adoption of Venetian and Italian loanwords. Such words must have come into Greek after the 13th c., and how they were accommodated depended on the properties of the recipient language. Examples from many regions throughout Greece have been identified, but most are to be found in the Ionian Islands and the Cyclades, where there was intense contact with the donor languages. In these dialectal varieties, the loan suffix was fairly productive, as a significant number of new adjectives, whether positive or pejorative, testify. The new formations, however, were derived only from nominal stems from both foreign and native origin and retained the core meaning of the loan suffix. In general, the history of suffix -ózos in Greek, when seen in comparison with those of the [-learned] suffixes -(i)áris and -ísios, confirms recent linguistic theories concerning the mechanisms of loan suffixation. We have also shown that the intensity of contact and the structural similarity between source and target language are of crucial importance to the process of lexical borrowing and that lexical and suffixal borrowing precedes grammatical borrowing. We have also shown that the path that the process of borrowing is usually assumed to follow is indeed correct, that is, that it does indeed consist of a progress from the adoption of whole lexical items consisting of transparent constituents (stem and suffix) via the detaching of the loan suffix, to the acquisition of the ability to combine with native stems.