DEVELOPMENT OF A PRODUCTIVE DERIVATIONAL PATTERN ON THE BASIS OF LOAN TRANSLATION? THE CASE OF CROATIAN ADJECTIVES FORMED WITH THE PREFIX MEĐU1

DEVELOPMENT OF A PRODUCTIVE DERIVATIONAL PATTERN ON THE BASIS OF LOAN TRANSLATION? THE CASE OF CROATIAN ADJECTIVES FORMED WITH THE PREFIX MEĐUThis paper deals with the question of the formation of Croatian adjectives with the prefix među-. While such adjectives were very rare in late 19th and early 20th century, an analysis of relevant lexicographic works and digital corpora demonstrated that their number started to become larger in later 20th century, culminating in recent decades. Today, the [među-N-Suff]Adj derivational pattern is a productive, accounting for 134 adjectives with a frequency of ten occurrences or more retrieved from the largest extant Croatian web corpus, hrWaC. On the basis of an analysis of available older lexicographic works and digital corpora, it can be concluded that međuprefixed adjectives Linguistica_2020_1_FINAL.indd 59 22. 12. 2020 15:07:34 60 first entered into Croatian as loan translations (calques) of Latin(ate) and German terms. According to more recent lexicographic works and digital corpora, later on, and especially in recent decades, which coincided with a growing English influence on Croatian, međuprefixed adjectives were probably produced as equivalents of English interprefixed adjectives. The number of međuprefixed adjectives, as well as the variety of semantic domains in which they are used, testify to the fact that the [među-N-Suff] Adj pattern is well-established and productive in contemporary Croatian. The analysis of Croatian međuprefixed adjectives in this paper could contribute to shedding more light on the question of morphological borrowing phenomena in general.

[među-N-Suff] Adj pattern appeared in Croatian, it probably continued to exist to some extent, as supported by evidence of the Riznica corpus and Benešić's dictionary. However, it was only in the mid-and late 20 th century that it really became productive, as attested to by the VRH dictionary, as well as the HNK and hrWaC corpora. That period coincides with a large English influence on Croatian (Filipović 1990;Samardžija 2002;Turk 2013), so it could be assumed that the influence of English contributed to the reinforcement of the [među-N-Suff] Adj pattern, and to the creation of a larger number of new među-adjectives.
The rest of the article is organized as follows: Section 2 provides an overview of language borrowing phenomena, with a special focus on morphology, especially that which is derivational, and the difference between MAT and PAT borrowing (Sakel 2007). Section 3 positions the Croatian language in the context of language borrowing, emphasizing especially the role of loan translations (calques). In Section 4, we provide more details about the derivation of adjectives in Croatian. Section 5 is dedicated to the methodology we have used to study the question of whether među-adjectives appeared in Croatian as a result of loan translation. Our methodology is a combination of analyses conducted on available lexicographic works and digital corpora. Section 6 presents the results of our analysis. It shows that, as predicted at the beginning, među-adjectives first emerged as a result of loan translation, and have become more numerous with time, with the highest number of occurrences recorded in contemporary Croatian, both according to lexicographic and corpus data. The majority of adjectives are formed via the [među-N-Suff] Adj pattern, and the među-prefix in these adjectives behaves like a polysemous morpheme whose meanings form a radial structure, with the prototypical (spatial) meaning at the centre (cf. Lakoff 1987). In Section 7 we provide some concluding remarks, reaffirming our hypothesis that među-adjectives have entered the Croatian language as a result of loan translations of Latin and German terms, and could subsequently have become more numerous under English influence. This is also corroborated by data from the Etymological Dictionary of the Croatian Language (2016).

BRIEF OVERVIEW OF LANGUAGE BORROWING PHENOMENA
Linguistic change stemming from language contact could be said to be an almost universal phenomenon occurring in both existing and extinct languages (cf. Grant 2019: 1). Another well-known fact is that language contact can affect the participating languages on a number of levels, and most prominently in the lexicon (e.g. Thomason 2001: 10;Turk 2013;Poplack 2018: 1;Grant 2019: 1). In this paper, however, attention is turned to a different aspect of language contact, morphological borrowing. The issue has been much less explored than lexical borrowing, but today is "far from being unknown" (Grant 2019: 17). Moreover, recent years have seen an increasing scholarly interest in the field of morphological borrowing (Gardani, Arkadiev and Amiridze 2015: 1).
There are various phenomena that enter the field of morphological borrowing. They are generally divided into two major types: borrowing of inflectional morphemes (e.g. Weinreich 1963: 31-33;Kovačec 1967;Minayeva 2003), and borrowing of derivational morphemes (e.g. Della Volpe 1997;Coghill 2015). Sakel (2007) distinguishes two basic types of borrowing between languages: matter borrowing (MAT) and pattern borrowing (PAT). MAT-borrowing refers to taking over both morphological material and its phonological form, while PAT-borrowing refers to replicating the organization and grammatical or semantic meaning without borrowing the form itself. This second type results from loanshifts or calques (Sakel 2007: 16). In other words, PAT-borrowing refers to the situation in which a recipient language uses its own morphological elements and organizes them in a way that resembles the structure of the source language (cf. Gardani, Arkadiev and Amiridze 2015: 3).
This paper focuses on a particular type of PAT derivation: loan translation of complex words consisting of an affix (a prefix), a (nominal) base and a suffix. A general claim in the literature on affix borrowing is that affixes are most commonly borrowed indirectly, as part of complex loanwords (Weinreich 1963: 31-32). In such a case, a number of complex loanwords enter a recipient language, and only subsequently can its speakers analyse these words into their constituent parts and eventually start using a foreign affix with native bases (cf. Seifart 2015: 512). In direct borrowing, on the other hand, an affix is recognized by the speakers of a recipient language and immediately used with native bases (ibid.). For Seifart (2015), the crucial difference between the two types of affix borrowing lies in the question of whether speakers of a recipient language understand the source language. If they do, it can lead to a situation of direct borrowing.
Loan translations or calques are a specific type of borrowing. In the case of complex words, a number of authors distinguish between an "entire calque", or the reproduction of a complex foreign word element by element (loan translation proper, Weinreich 1963: 51;Martinet 1980: 170), and a "partial calque", or the loan translation of only parts of words (loan rendition, Weinreich 1963: 51;calque approximatif, Martinet 1980: 170). When applied to words formed with affixes, examples of the first case would be loan translations of both an affix and a base (e.g. French pré-elargissement 'pre-enlargement' -> Polish przedrozszerzenie, Trajder 2007: 140), and of the second one loan translation of the affix only (e.g. French supercommissaire 'super-commissioner' -> Polish superkomisarz) (ibid.).
In the examples analysed in this paper, i.e. in Croatian među-prefixed adjectives, entire calque is at work due to the fact that both the prefix (među-), the base and (usually) the suffix are borrowed and translated, i.e. expressed with Croatian linguistic material, such as in the following example: Latin internationalis (inter-'inter-+ natio 'nation' +-alis 'adjectival suffix' = 'international') > Croatian međunarodan 'international' (među-'inter-' + narod 'people' + -an 'suffix'). Before proceeding with the analysis of adjectives, some details need to be provided about major linguistic influences on Croatian, as well as about the formation of Croatian adjectives in general.

CROATIAN IN THE CONTEXT OF LANGUAGE CONTACT
At its very beginnings, Croatian already came into contact with several substrate languages, and had lasting contact with Latin as the language of Western Christianity, administration and education (Samardžija 2002: 61). By the end of the Middle Ages, it had also established contacts with a number of neighbouring languages: Italian, Hungarian, German and later on Turkish, all of which exerted influence primarily on the lexical level, at times very strongly 3 (Samardžija 2002: 61-62). Latin loanwords from the areas of Christianity and philosophy, but also related to administration, law and new inventions, were for a long time the most numerous ones in Croatian (Samardžija 2002: 63).
Calques became a regular phenomenon in the Croatian lexicon starting from the second half of the 16 th century, and were especially related to the publication of first larger dictionaries (Samardžija 2002: 63). Older Croatian lexicographic works were usually bi-or even multi-lingual, and their source language was usually a foreign one, mostly Latin (Gostl 1995). Faced with numerous gaps on the Croatian side, lexicographers were oftentimes forced to invent equivalents themselves, which resulted in a large number of neologisms, calques, etc. (Samardžija 2002: 64).
At the time of the industrial revolution, a considerable number of new technical and scientific terms were formed on the basis of classical languages (Latin and Greek), which are sometimes called Europeisms (Croatian europeizmi) due to their presence in a number of modern European languages (Samardžija 2002: 65). At the end of the 19 th century, Croatian borrowed a number of internationalisms through its contacts with German and Italian (Samardžija 2002: 65). It needs to be emphasized that Latin was the official language in continental Croatia until 1847, followed by German until 1860 (Samardžija 2002: 66). When Croatian finally became the official language, it lacked functional diversity. It therefore saw numerous additions in the second half of the 19 th century, during which time two prominent lexicographers played a key role: Šulek in continental Croatia and Parčić in littoral Croatia (Samardžija 2002: 66-67). They both agreed on providing Croatized words for all notions where it was possible (Samardžija 2002: 67), which left the language once again with a substantial portion of calques.
In recent history, English is undoubtedly the language that has exercised by far the strongest influence on Croatian (Samardžija 2002: 72). Numerous Anglo-American elements have spread into Croatian owing primarily to the media, and have entered the language at a quick pace (Samardžija 2002: 72). Moreover, Turk (2013: 159) claims that in the second half of the 20 th century Croatian was "inundated" by loanwords from English. English influence on the Croatian language has occurred both overtly, in the acceptance and adaptation of English lexemes into the lexicon, and covertly, as loan translations, which are "really numerous" (Muhvić-Dimanovski 1992: 94), and can be found on virtually all language levels (Margić Drljača 2009).
From the abovementioned list, it can firstly be concluded that all types of adjective formation but the first one -prefix-suffix combination of relational adjectives -are rather unproductive and of very limited scope, because they are all used to form a single adjective, according to Babić (2002). Unlike these three processes, prefix-suffix combination results in a number of relational adjectives. Secondly, it can also be concluded from the aforementioned facts that the među-prefix is productive in the formation of relational, and not descriptive adjectives.
Prefix-suffix combination, or the formation of new lexemes through the simultaneous addition of a prefix and a suffix, is also called parasynthetic formation or parasynthesis (e.g. Serrano-Dolader 2015; Iacobini 2020). The term parasynthesis is mostly used today to refer to Romance verbs formed from adjectival or nominal bases (e.g. French embarquer 'to load, board' < em-'in' + barque 'boat' -er 'infinitive ending') (Serrano-Dolader 2015: 524), but some authors also use it to refer to nouns and adjectives formed through the addition of a prefix and a suffix to a base (Serrano-Dolader 2015; Iacobini 2020). It is important to emphasize that, in order for a formation to be considered a case of parasynthesis, many authors argue that there should not be an attested "intermediate stage": thus, in the above French example, there are no words such as *barquer or *embarque. Authors writing from a generative point of view explain that requirement on the basis of the binary branching hypothesis, which specifies that only one word-formation process can apply at a time (cf. Serrano-Dolader 2015). In other words, they reject the possibility of ternary structures for parasynthetic derivations (ibid.). 4 When applied to Croatian adjectives formed according to the [među-N-Suff] Adj pattern, however, this principle does not seem to work formally because a simple adjective can be found in the language for every parasynthetic one, such as in the following examples: gradski 'urban' <> međugradski 'intercity'; državni 'state' <> međudržavni 'interstate'; zvjezdani 'stellar' <> međuzvjezdani 'interstellar', etc. Still, all these adjectives are claimed to be parasynthetic or prefix-suffix formations by Babić (2002) due to the fact that their meaning cannot be construed as the sum of the prefix and an adjective: for instance, međugradski 'intercity' does not mean 'occurring between what pertains to the city' (*među-'inter-' + gradski 'pertaining to the city'), but its meaning is 'relative to what is between cities', thus među-'inter-' + grad 'city' + -ski 'suffix'. In this paper, such adjectives are considered parasynthetic formations.
The fact that Babić (2002) enumerates a number of adjectives formed through prefix-suffix combination with među-points to the conclusion that adjectives formed according to this process are fairly present and regular in contemporary Croatian. What Babić (2002) omits to specify, however, is, firstly, how productive the derivational pattern [među-N-Suff] Adj is, and secondly and more interestingly, how it emerged in Croatian. 5 It is therefore the goal of this paper to shed some light on the history of the formation of među-prefixed adjectives and to explore their productivity in present-day Croatian language.

METHODOLOGY
In order to explore the emergence of među-prefixed adjectives in Croatian and the productivity of the patterns according to which they are created, available lexicographic works and corpora were consulted. More precisely, three dictionaries were used: the Academy's Dictionary (Budmani/Maretić 1904-1910, Benešić's dictionary (1957), and VRH (2015). Brief explanations shall be given as to why the three dictionaries mentioned were chosen for the analysis presented in this paper.
The Academy's Dictionary is a rich historical dictionary of Croats, Serbs, Bosnian-Herzegovinian Muslims and Montenegrins that provides information from the earliest linguistic sources in the 12 th century up until works of the 19 th century. In the period when it was written, it was considered that these ethnic groups spoke a single language called Croato-Serbian or Serbo-Croatian. Importantly for this paper, it is also a terminological dictionary, as well as a dictionary of foreign words and loanwords.
The exact title of Benešić's dictionary (1957) is Rječnik hrvatskoga književnoga jezika od preporoda do I. G. Kovačića (Dictionary of the Croatian Literary Language from the National Revival until I. G. Kovačić). Its intention was to be a dictionary of contemporary Croatian literary language as a collection of quotes from the most excellent Croatian writers who published between the very beginning of the 19 th century until the 1940s (Nikolić-Hoyt 2010: 63-64). It was chosen due to the fact that it covers the "middle" period between early 20 th century and the 1940s. One of Benešić's goals in compiling his dictionary was to revise and modernize the data found in earlier dictionaries, for instance by leaving out words that were no longer used in the Croatian literary language, and by introducing those that were used by Croatian modern authors (Nikolić-Hoyt 2010: 62). It should therefore serve as a good illustration of the Croatian lexicon from early 19 th century to mid-20 th century.
Finally, the VRH dictionary is the largest and most recent dictionary of the Croatian standard language (Slišković 2016: 244). It is based on older relevant lexicographic works, manuals, specialized dictionaries and digital corpora.
After an analysis of the mentioned lexicographic works, three major digital corpora were also consulted: Riznica, HNK and hrWaC. The Riznica corpus, compiled by the Institute of Croatian Language and Linguistics, contains literary and other written sources from the second half of the 19 th century to this day. 6 The corpus contains 100 million tokens (Brozović Rončević and Ćavar 2012). Due to its specificities, Riznica was searched via the među.* standard regular expression, providing all words starting with the graphic sequence među.
The HNK, 3.0 beta version, contains more than 2.3 billion words (Tadić 2009). Much larger than Riznica, it is a balanced and representative corpus 7 of standard contemporary Croatian, which contains a certain amount of faction (such as magazines, newspapers, books, diaries, novels, etc.), fiction, etc., in line with text typology standards (Tadić 2002: 442). The HNK corpus was also searched via the među.* standard regular expression, providing all words starting with the graphic sequence među. The results obtained were then organized through the Frequency -Lemma function, providing a list of među-words with their number of occurrences in the corpus. The results were manually checked.
The hrWaC (Ljubešić and Erjavec 2011) is a web corpus whose 2.2 version was crawled in 2014 from the .hr domain, so it provides us with data about very recent Croatian language usage. It is the largest extant Croatian corpus, with 1.4 billion tokens. Adjectives entering the analysis were extracted from the corpus via the following CQL order: [word="među.*"] containing [tag="A.*"] The order searches for all words beginning with the sequence među and bearing the PoS mark "A", i.e. adjective. After that, using the option Frequency -Lemma, all the obtained results were organized according to their frequency of appearance in the corpus. For the purposes of this paper, only adjectives with freq ≥ 10, which enables the finding of both high-and lower frequency lexemes, were taken into consideration. The corpus data was checked manually in order to eliminate noise such as adjectives formed via other word-formation processes (e.g. međunarodno-pravni 'related to international law'), typos (međusubni, međunardni), etc., leaving a final list of 134 adjectives. This figure itself already suggest that adjectives formed with the prefix među-are fairly numerous in contemporary Croatian, and that the derivational pattern is a rather productive one.

RESULTS AND ANALYSIS
This section first presents the results of the lexicographic analysis, followed by corpus results. Table 1 presents the među-adjectives in the Academy's Dictionary (1904)(1905)(1906)(1907)(1908)(1909)(1910). 9 The adjective međunarodan 'international' appears with the-(a)n suffix in the Academy's Dictionary and Benešić's dictionary, while in the VRH it appears as međunarodni, with the -ni suffix. An average Croatian speaker would not note any difference in meaning between these two adjectives. An average Croatian speaker with more linguistic knowledge would probably say that međunarodni is the definite form of the indefinite adjective međunarodan, with no other difference in meaning. Babić (2002: 451-456) has dedicated a whole chapter to the question of differentiating between the -(a)n and -ni suffixes, which proves in itself that the question is rather complex. These two suffixes present differences on both formal and semantic level. Put briefly, -(a)n is used to form descriptive adjectives (e.g. pametan 'intelligent'), while -ni is used to form relational adjectives (e.g. autobusni 'pertaining to buses' When one observes the adjectives from Table 1, one immediately understands that, firstly, their number is rather limited, amounting to as few as ten adjectives altogether. 12 Secondly, and more importantly, almost all of the listed Croatian adjectives are claimed to be equivalents of Latin(ate) or German specialized terms, which means these are intentional calques created for the purposes of filling specific lexical gaps. Thirdly, and probably most importantly, five of the total of ten adjectives are hapax legomena, i.e. lexemes found in a single work, whether it be the opus of an uncited writer (međusudan 'interjudicial') or, for the remaining four, terms coined by Bogoslav Šulek for the purposes of compiling his previously mentioned DST. It needs to be emphasized that Šulek played a large role in the formation of several domains of Croatian scientific terminology, which were a result of real needs for Croatian terms in specific scientific domains and also a way to resist Germanization and/or Hungarization (Samardžija 1997: 178).

Academy's Dictionary
A special comment should be made concerning the adjectives međudnevički 'related to međudnevica ('three-week period between the Assumption and the Nativity of the Blessed Virgin Mary')' and međusoban 'mutual'. The first adjective is derived from the noun međudnevica as a result of suffixation, and thus the prefix među-was not directly involved in its formation ([međudnevica] N + -čki 'suffix' > [međudnevički] Adj ). In other words, the prefix među-is actually present in the adjective međudnevički as part of the noun from which the adjective was derived.
The adjective međusoban 'mutual' is more interesting. According to the Academy's Dictionary (1904)(1905)(1906)(1907)(1908)(1909)(1910), at the beginning of the 20 th century, it had two possible meanings: 1) 'internal, domestic' and 2) 'mutual, occurring between people'. The first meaning is claimed to be found only in Daničić's dictionary (1863/64) as part of a quote from a 14 th century Serbian document, while the second appears in some works written by Croatian, Serbian and Montenegrin authors (see footnote 12), who were mostly active during the 19 th century. The first meaning is not recorded in the Croatian language today, 13 and it was probably never part of it, according to the diachronic information from the Academy's Dictionary (cf. Matasović et al. 2016: 600). The second meaning of the adjective međusoban, 'mutual', is the only meaning the adjective has in contemporary Croatian. The adjective was formed according to the [Pref-Pron-Suff] Adj pattern, or more precisely according to the following formula: među-'inter-' + sebe 'reflexive-possessive pronoun' + -(a)n 'suffix' > međusoban. It is the only adjective in the Academy's Dictionary (and in the analysed corpora, as will be said infra) that was formed from a pronoun. Thus, the word-formation pattern from which it resulted is an isolated one, and did not have further impact on the formation of Croatian adjectives.
It should also be added that the Academy's Dictionary lists a number of među-prefixed nouns 14 (e.g. međubrđe 'place between hills' (< brdo 'hill'); međuvođe 'place between waters' <voda 'water')), among which many toponyms (e.g. Međuhan 'village in Serbia'; Međulići 'village in Herzegovina', etc.), thus it seems that in early 20 th century the među-prefix was used only as a noun-deriving prefix. 15 All of these facts point to the conclusion that, at the beginning of the 20 th century, a period when the Croatian standard was already formed, there existed very few adjectives formed with the prefix među-, and the extant ones were mostly scientific terms related to specific domains of specialized language, as well as hapax legomena. The only two adjectives used more frequently, according to the Academy's Dictionary, which are not claimed to appear "only once" or "only with a certain author", are međusoban 'mutual', which is also the only adjective formed according to the [Pref-Pron-Suff] Adj pattern, and međunarodan 'international', formed as a result of prefix-suffix combination. The latter, however, is "a recent invention" (Budmani/Maretić 1904-1910, which is another interesting insight important for future discussion, as it may lead to the conclusion that it was in the 19 th and early 20 th century that među-adjectives derived through prefix-suffix combination under foreign influence started to be formed and used. Table 2 presents the relevant adjectives from Benešić's dictionary (1957). međusatni 'between two (school) classes' 3 međuzvjezdani 'interstellar'

Benešić's Dictionary
As Table 2 demonstrates, Benešić's dictionary (1957) lists only three među-adjectives. This fact points to the conclusions that, firstly, over the period from the start of the 20 th century, i.e. the years when the volume containing među-adjectives of the Academy's dictionary was published, until the mid-20 th century, when Benešić finished compiling his own dictionary, the pattern [među-N-Suff] Adj had not yet become a productive one. Secondly, this might have to do with the fact that Benešić's dictionary is a dictionary of the Croatian literary language, and as such is based on literary sources only, while the observed pattern might already have become more productive in other language areas, such as the press and specialized terminology.

A More Recent Dictionary -VRH (2015)
A more recent dictionary, VRH, lists the following među-adjectives.

Corpora Analysis: Riznica, HNK and hrWaC 6.3.1 Riznica
In the Riznica corpus, only two adjectives appear in the earliest period covered (between late 19 th and early 20 th century): međunarodan 'international' and međusoban 'mutual'. After that, for a long period going all the way to 1962, these two adjectives remain the only ones formed with the prefix među-which appear in the corpus, alongside several među-prefixed nouns such as međuvrijeme 'meantime'.
The situation starts to change from the 1970s onwards, when two more adjectives appear: međunacionalni 'occurring between nations' (1971) and međuratni 'interwar' (1972). After that, the number of među-prefixed adjectives really starts to become much larger from the 2000s, with examples such as the following: međuljudski 'inter- These facts can be said to quite convincingly prove that during the whole 20 th century među-adjectives were rather rare in Croatian, with only međusoban 'mutual' and međunarodan 'international' being in use, and that it was only from the 1970s onwards that the pattern started forming a larger number of među-adjectives.
This idea leads to the question what reasons there are for the među-prefixed adjectives to have become rather frequent in the late 20 th century. Some answers are provided in the following sections. As Table 4 demonstrates, in the HNK corpus there are 51 među-adjectives with a frequency of 10 or more occurrences. That is quite a large number of adjectives formed mostly according to the [među-N-Suff] Adj pattern, which proves that the pattern became more productive starting from the 1990s. While some of the adjectives from Table  4 belong to specialized languages (e.g. međuljušturni '(of molluscs) intervalvular'; međuzrnski 'intergranular'; međuzvjezdani 'interstellar', etc.), a large number belong to general language. If these adjectives have been found in the national corpus, that points to the conclusion that adjectives formed according to the [među-N-Suff] Adj pattern have been quite widespread in various genres of standard Croatian since the 1990s.

The HNK Corpus
Finally, the hrWaC corpus shall be analysed as the largest extant corpus of the Croatian language. It will be interesting to compare the među-adjectives appearing in hrWaC, as a large web corpus, with those appearing in the national corpus, due to the different text types they include.

hrWaC
Considering the fact that hrWaC is the largest corpus that registers the most recent use of the Croatian language, its inclusion in the analysis presented in this paper was important for learning about the current use of adjectives formed with među-. What the corpus data shows is that 134 adjectives with the prefix među-and a frequency of 10+ occurrences are registered in hrWaC. 134 is a lot more than only two adjectives in the first half of the 20 th century, and much more than just several more around the 1970s. This means that the derivational pattern [među-N-Suff] Adj has become more productive with time, culminating in recent years.
Below is a list of the 50 most frequent adjectives retrieved from hrWaC. The 134 među-adjectives extracted from the hrWaC corpus is by far the largest number of adjectives appearing in any of the aforementioned sources. Such an outcome was expected, firstly, because hrWaC is the largest extant Croatian corpus. Secondly, given the fact that hrWaC is a web corpus, which also collects data from specific sources such as chat rooms and blogs, we have expected to find a certain portion of less standard or more colloquial terms, e.g. međunožni 'occurring between legs', but also more specific terms such as međuroditeljski 'interparental', which seems to appear in a single weekly newspaper; or međuizborni 'occurring between elections', which is used exclusively in political discourse. It should also be emphasized that some of the adjectives retrieved from hrWaC belong to highly specialized terminologies, such as međuzrnski 'intergranular', međubiskupijski 'interdiocesan' or međukralježnički 'interspinal', which one would not expect to find in general language dictionaries such as the VRH, but rather in specialized dictionaries or glossaries.
It can generally be said that the 134 adjectives retrieved from hrWaC point to the conclusion that među-adjectives, a majority of which were formed according to the [među-N-Suff] Adj pattern, have recently become quite numerous in Croatian. The fact also implies that the [među-N-Suff] Adj pattern continues to produce new forms, some of which have not been listed in lexicographic works yet. It would be interesting to analyse među-adjectives below the 10 occurrences threshold to see what happens at that end of the frequency scale, how many adjectives there are and what characteristics they have. It can be assumed that there would be more new adjectives that have not been listed in dictionaries yet due to the fact that the pattern seems to be quite productive.
The adjectives extracted from hrWaC shall now be analysed first morphologically, or from a purely word-formation point of view, and then semantically. A great majority of adjectives belong to the first group. They are a result of the simultaneous addition of the prefix među-and a suffix to a nominal base (cf. Babić 2002: 445). Thirteen suffixes participated in their formation, as listed in the table below.  Table 6 demonstrates that the two most frequent suffixes, which account for the formation of more than 72% percent of all adjectives, are -ni and -ski. This fact is consonant with details related to the prefix-suffix formation of relational adjectives provided by Babić (2002: 444). Babić (2002: 398) describes the -ski suffix as "one of the most productive adjectival suffixes", and -ni as "very productive " (2002: 421). With respect to the use of the -ski and -ni suffixes in the formation of adjectives, which is a highly complex question dependent on both formal (phonological) and semantic criteria, the following can be said briefly (cf. Babić 2002: 428-434):

Morphological Analysis
1) all adjectives derived from proper nouns are formed with -ski, never with -ni; 2) adjectives derived from common nouns related to living beings (humans, animals and plants) are rarely formed with -ni; 3) both -ni and -ski are used to form adjectives from common nouns related to non-living entities (both concrete and abstract), and their distribution depends mostly on phonological context, 20 4) a number of adjectives have dual forms with -ni and -ski, but with different frequency of use, some of which are also non-standard.
The only adjective formed according to the [među-Pron-Suff] Adj pattern, i.e. from a pronominal base, is međusoban 'mutual', which is also the only descriptive adjective formed with među-through prefix-suffix combination, according to Babić (2002: 461).
With respect to the second group of adjectives, formed through prefixation according to the formula [među-Adj] Adj, only the following three adjectives exemplify it: međuzavisan 'interdependent', međuovisan 'interdependent' 21 and međupovezan 'interconnected'. They are formed via the addition of the prefix među-to a base without the participation of a suffix. These three adjectives account for only 2.2% of all the međuprefixed adjectives in the corpus.
What interests us more here is prefixal meanings. The prefix među-realizes two types of meanings in the analysed adjectives: concrete and abstract. Its concrete meaning is 'located between two or more (concrete) entities' (e.g. međustaklen 'between two glass surfaces'). This is the core or prototype (e.g. Lakoff 1987) meaning that refers to the concrete spatial position of concrete objects.
The abstract meanings of the prefix među-in the analysed adjectives are the following: a) 'between two or more abstract entities' (e.g. međugeneracijska solidarnost 'intergenerational solidarity'), and b) 'between two or more periods of time' (e.g. međutelidbeno razdoblje 'intercalving period').
The semantic network that the prefix među-realizes with adjectives can thus be illustrated by the following image. 'jaw' > čeljusni 'related to jaw'; kazalište 'theatre' > kazališni 'related to theatre', etc. (Babić 2002: 430 'located physically between two or more concrete objects' međuzubni 'interdental' 'between two or more periods of time' međuratni 'interwar' 'between two or more abstract entities' međuvjerski 'interfaith' The abstract meaning 'between two abstract entities' relies upon the abstract is concrete metaphor (e.g. Lakoff 1987) on the basis of which we conceptualize more abstract entities through more concrete ones. In examples such as međuvjerski dijalog 'interfaith dialogue' or međukulturno razumijevanje 'intercultural understanding', for instance, we perceive abstract phenomena such as dialogue and understanding taking place between faiths or cultures as phenomena occurring between two concrete things, thus we express them with the same preposition (među 'between') or prepositional prefix među-'inter-'.
The concrete meaning 'located physically between two or more concrete objects' is metaphorically extended into the meaning 'between two periods in time' on the basis of the frequent time is space metaphor (e.g. Kövecses 2010). This metaphor enables human beings to conceptualize time phenomena on the basis of concrete, spatial phenomena of which they have better understanding. Thus, in examples from hrWaC such as međusezonska kolekcija 'interseasonal (clothing or shoes) collection' what happens between two periods of time, i.e. two seasons, is conceptualized as being physically located between two concrete objects.
In the conclusion to this part, it must be emphasized that the prefix među-is a polysemous affix which, when attached to adjectives, realizes three related meanings, both concrete and abstract, the latter of which are based on metaphor. The semantic network of the prefix među-in the analysed adjectives demonstrates that it behaves much like other lexical categories such as nouns and verbs, construing a radial structure with the prototypical sense as the centre of its semantic network (cf. Tyler and Evans 2003: 31). The semantic level of the formation of the analysed adjectives was insisted upon owing to the fact that, according to our understanding, all complex words are motivated both grammatically (or morphologically) and semantically, i.e. that derivational processes cannot be separated from the semantic ones (cf. Booij 2005;Raffaelli 2013). 22 The semantic network of this particular prefix is just an example of the complex relations that exist between word-formation processes and meanings that are created during the derivation of new complex lexemes. Therefore, some authors (e.g. Raffaelli 2018: 153) emphasize that one of the major future tasks of word-formation as a linguistic subdiscipline is to systematically study the semantic processes which accompany the formation of complex words.

CONCLUDING REMARKS
This paper explores Croatian adjectives formed with the prefix među-. While adjectives derived according to the [među-N-Suff] Adj pattern are fairly numerous in contemporary Croatian, according to Babić (2002), the author of the most comprehensive manual of Croatian word-formation, as well as lexicographic sources (the Academy's Dictionary, Benešić's dictionary and the VRH) and large digital corpora (hrWaC and HNK), from a diachronic point of view that was not the case as recently as only a hundred years ago. More precisely, both an analysis of older lexicographic works (the Academy's Dictionary and Benešić's dictionary), as well as of digital corpora covering older texts (Riznica and HNK) have showed that in early 20 th century među-prefixed adjectives were very rare. Moreover, the Academy's Dictionary (1904)(1905)(1906)(1907)(1908)(1909)(1910) specifies that almost all such adjectives were hapaxes and calques made according to Latin(ate) or German models. In mid-20 th century, the situation was rather similar according to both dictionaries and corpora, and adjectives formed with među-only seem to have become more numerous later in the 20 th century. The question is, obviously, why.
The Academy's Dictionary's explicit claim that all adjectives formed with međubut one, međusoban 'mutual', are hapaxes and equivalents of foreign terms is a very important one, because it points to a temporarily conclusion that these are not native Croatian formations. It must be added immediately, however, that there were a number of nouns in the same period that were formed with među-. In other words, it seems that među-used to be an exclusively noun-forming prefix, which was impossible to be used with adjectives before the 19 th century. When it comes to the adjective međusoban 'mutual', it was demonstrated that it was used with two meanings, the first of which, 'domestic', appears in a single 14 th century Serbian document, and the second of which, 'mutual', which is the meaning it still has in contemporary Croatian, has been registered in texts of Croatian, Montenegrin and Serbian authors mostly from the 19 th century. In other words, even the adjective međusoban 'mutual', as an isolated derivation resulting from a prefix-suffix combination with a pronominal base (i.e. from the reflexive-possessive pronoun se(be)), is a creation dating back to the 19 th century, as the rest of među-prefixed adjectives.
Apart from the case of the adjective međusoban 'mutual', the influence of foreign languages seems to be a key element contributing to the possibility of adjective formation with the prefix među-. While Latin inter-'between' was the earliest language source of Croatian među-prefixed adjective calques, followed by few German words formed with zwischen-'inter-', neither of these languages seemed to provide a large number of new adjectives in Croatian. It was only in late 20 th century, which coincided with the advent of English predominance and the spread of its influence on Croatian (Filipović 1990;Muhvić Dimanovski 1992: 94;Samardžija 2002: 72), that the Croatian language saw a large number of new među-prefixed adjectives. It is therefore arguably under English influence that numerous adjectives with među-were formed and then spread in Croatian from mid-20 th century onwards, reinforcing the derivational pattern [među-N-Suff] Adj which had already existed in the language as a result of early Latin(ate) and German calques. The Etymological Dictionary of the Croatian Language (Matasović et al. 2016) 23 also hints at this. Moreover, Ranko Matasović (p. c.), 24 one of the leading experts on the history of the Croatian language and Slavonic languages in general, considers that the pattern has recently become productive in Croatian, and that several decades ago most of među-adjectives could have been formed as calques of English adjectives.
A subsequent question would be whether this particular derivational pattern came to be through indirect or direct English influence (cf. Seifart 2015). While the question cannot be answered with certainty, it was probably a case of direct borrowing, or direct calque, as English has been a rather widespread language among Croatian speakers, and the most spoken foreign one, in the last decades. More precisely, due to increasingly intensive contacts with English inter-prefixed adjectives, Croatian speakers have probably started to calque them in the domains they needed them to fill in lexical gaps, using the Croatian prefix među-coupled with Croatian nominal bases and a suffix.
At some point in time, the derivational pattern [među-N-Suff] Adj could probably have become as "natural" as any other adjective-deriving native Croatian pattern. If this scenario were correct, it would not be a case of the introduction of a new element in the Croatian language, but of a reinforcement of an existing prefix (među-) in a new "surrounding", i.e. with adjectival bases, because the prefix had been used to form complex nouns in the 19 th century and even earlier (for instance in a number of toponyms).
The analysis of complex među-prefixed adjectives presented in this paper demonstrates that the adjective-forming pattern [među-N-Suff] Adj , which has entered the Croatian language as a result of loan translation of Latin(ate) and German terms, and was subsequently probably reinforced through the calquing of English inter-adjectives, is a productive word-formation pattern in contemporary Croatian. More precisely, today it accounts for a number of adjectives belonging to semantically various domains, according to Croatian word-formation manuals and recent lexicographic works. Moreover, it continues to produce new adjectives, as attested by large web corpora. Not only do these insights illustrate the complex influences foreign languages (such as Latin, German and English) have had on the Croatian word-formation and lexicon, but they also make a contribution, however modest, to the study of morphological borrowing as a phenomenon in general. 23 The dictionary lists only three među-adjectives: međunarodni 'interational', which it claims to be a calque of English and French international, međusoban 'mutual', derived from među 'between' and se 'oneself', and međugradski 'inter-city', for which it claims that it was formed after the English adjective inter-city (Matasović et al. 2016: 600).