Are there Slavonic features in the syntax of Romanian ? Two case studies †

The present paper discusses the issue of the Old Church Slavonic influence on the syntax of (old) Romanian. The starting point is the hypothesis that many of the syntactic features of Romanian previously explained by postulating the influence of (Old Church) Slavonic (especially in studies strongly influenced by ideological issues, published during the Communist period) are actually either regular transformations which occurred in the transition from Latin to Romanian, common to the other Romance languages as well, or the output of more general tendencies manifested in the history of Indo-European languages. In order to check the role of the Slavonic influence in the syntax of Romanian, we have established a working algorithm, which is applied to two phenomena from old Romanian: (i) the position of relative adjectives, and (ii) scrambling in compound verbal forms in correlation with auxiliary inversion.


Introduction
This paper investigates a controversial issue in Romanian and Romance linguistics, i.e. the existence of elements of Slav(on)ic origin in the syntax of old Romanian, eventually also transmitted to modern Romanian.The influence of Slavonic syntax on old Romanian is relevant because, as is well known, many old Romanian texts (from the 16 th century) are religious (and, to a smaller extent, legal) translations of Slavonic texts; moreover, in the relevant literature (histories of Romanian, philological studies preceding edited old texts, etc.) numerous syntactic features of old Romanian are considered to be of Slavonic origin.In the present paper, special attention is paid not to the syntactic features accidentally occurring in one old text/in a restricted number of texts but to the general features of the old language, present in many texts belonging-to the extent that this is possible-to different stylistic registers.Only two of these features (which will be accounted for in an extensive research) shall be discussed here: the position of relative adjectives and scrambling in compound verbal forms.
The subject dealt with here faces two distinct problems: (i) a linguistic problem: is it possible for the syntax of a target language to be influenced by the source language, as it happens, obviously, in the vocabulary and in semantics?and (ii) a historical and political problem: can the political/ideological context influence the interpretation of the data from a domain apparently independent of politics, such as linguistics?
Before analysing the linguistic data, several observations-related to the way in which history, and implicitly the oscillation of the political context, could influence the linguistic interpretation-are necessary.
Two opposite tendencies have manifested themselves in Romanian historiography; these tendencies have re-emerged cyclically, with each period bringing its own perspective on the past (Djuvara, 2006;Boia, 2011): on the one hand, the negation of the Slavic influence in history of Romania, especially in relation to the emergence of the Romanian language and of the Romanian population (e.g."Școala Ardeleană" = The Transylvanian School with its exaggerated Latin "purism"), and, on the other hand, the rejection of this first tendency, manifested by claiming, in the spirit of historical objectivity, the overstatement of the role of the Slavs (e.g. the members of the "Junimea" = The Youth society, Gh.Panu, I. Bogdan, A.D. Xenopol, and "Noua Școală" = The New School in history, represented by P.P. Panaitescu and C.C. Giurescu).
The historical jigsaw is made complete by other positions: the adoption of a moderate Slavism (B.P. Hasdeu, N. Iorga), the amplification of the Dacian influence (V.Pârvan), or the granting of a privileged role to Western European influence (T.Maiorescu, E. Lovinescu).The peak of the Slavization of the Romanian history was reached at the beginning of the Communist period (P.Constantinescu-Iași, L. Pătrășcanu, M. Roller); at the end of the 1960s and later on, excessive nationalism cast shadow upon the Slavic component, and emphasis was thrown once more on the Dacian ingredient.

The Slavic influence on the Romanian language. Previous research
The same two opposite tendencies of historiography have operated in linguistics as well: as a reaction to the Latin purism in relation to the history of the Romanian language adopted by the Transylvanian School and its continuators, there appeared an exaggeration of the Slavic influence (at the end of the 19 th century -the beginning of the 20 th century, augmented in the early Communist period).The following issues have been emphasized: the Slavic-Romanian bilingualism (e.g. the "Daco-Slavic" language, Petrovici, 1943; see also Niculescu, 2007), the ethnic mix, including Dacian, Slavic and Latin population (Rosetti, 1968;Tagliavini, 1977), and the external Slavic influence (Cihac, 1870;Densusianu, 1961).As a reaction to Latin purism as well, the tendency to grant more weight to the influence of the Dacian language also emerged (e.g.V. Pârvan, at the beginning of the 20 th century; this idea was then strengthened, for ideological reasons, in the nationalist period of Communism).
Given these contradictory tendencies, the reality of the Slavic influence on Romanian is superficially known: on the one hand, Romanians defended their origin and national specificity (see, for similar observations related to the situation in Greece, Mackridge, 2009) but, on the other hand, they have colluded with the grand Communist ideas, overstressing the Russian/Slavic influence on the "national language" (similarly to the former Soviet countries, see Sériot, 1995).
In Romanian and Romance linguistics, there are no special studies which ascertain the consequences of the overstatement of the Slavic influence and the opposite effect, i.e. its negation, in different periods, with the exception of Zafiu (2009), which approaches the Communist period and its effects for linguistic research in general.
(1) fără de agiutoriulu celuia de susu niciu un lucru without help.defthat.gen of up no thing în calea vieției acestiia putemu să facemu (cc 2 , 6) in way.deflife.def.genthis.gen can.ind.pres.1plsă subj make.subj.1pl‡ 'without the help of God we are not able to do a thing in this life' Despite the general tendency to exaggerate the Slavic influence, even during Communism certain prestigious linguists rejected the Slavic determinant in the case of certain phenomena which are best accounted for by Latin facts; see, for example, Rosetti's (1968, p. 303) account for the vocative inflexion -e.A similar approach, this time with respect to syntax, is the goal of the present study.
Although in studies dedicated to the history of Romanian and in introductory studies to philological editions of old texts there are cursory references to syntactic phenomena of Slavic origin, syntax is the least investigated domain, because it is not always easy to decide which features can be accounted for only through Slavic influence, which features have another origin but have been strengthened by the Slavonic/Slavic influence, and which features have been falsely accounted for as an effect of Slavic influence for ideological reasons (related to the cultural ideology, for example, in the case of Transylvanian School, or to the political one, for example, in the case of the mystifications from the Communist period).

Two case studies
Given that the literature contains vague information and that it is often difficult to tell apart linguistic reality from false interpretation determined by ideological reasons, our research will focus on the syntactic phenomena considered to be of Slavonic/Slavic origin and will answer the following questions: 1.Which are (actually) the Romanian syntactic features that are the effect of Slavic influence and to what extent are they limited to old Romanian?2. What algorithm of analysis can one apply to tell apart genuine syntactic features of Slav(on)ic origin from ideological overstated influence? 3. Was the Slavic influence in the syntax of Romanian overstated for the same reasons as in history? 4. Is the study of language able to provide a more accurate historical picture for other periods than the mysterious emergence of the Romanian language and people (knowing that this period was "reconstructed" mainly using linguistic data)?

Romanian syntactic features considered to have Slavic origins
Before going a step further with the analysis, it is necessary to give a (provisional) inventory of the syntactic phenomena which have been considered to have Slavic origins.These data are taken from books dedicated to the history of Romanian (Densusianu, 1961;Rosetti, 1968;Gheție, 1997;Ivănescu, 2000), from special studies related to the Slavic influence in the syntax of Romanian (Seidel, 1958;Beneș, 1955;Copceac, 1998) and from the linguistic studies accompanying the philologically edited texts (Mareș, 1969;Rizescu, 1971;Costinescu, 1981;Teodorescu & Gheție, 1977;Chivu, 1993;Gheție & Teodorescu, 2005).‡  Beside the constructions mentioned above (see §2), the list of the syntactic phenomena having Slavic origins also contains: the subject position and the word order of constituents within the nominal phrase, the (non-)doubling of the direct and indirect objects, differential object marking with pe (< lexical pe 'on'), the ellipsis of the copula a fi 'be' , the predicative usage of the infinitive and of the gerund, the emergence of the "short" infinitive (without the ending -re, inherited from Latin), the usage of the infinitive in subjunctive-specific contexts, auxiliary and pronominal clitic inversion, scrambling in compound verbal forms, etc.If all these features were really imitating Slavonic constructions, then one would be able to say that the entire syntactic specificity of Romanian (especially old Romanian but also modern Romanian) is modelled on the syntax of Slavic languages.

Working algorithm
In order to determine in a more precise way the extension of the Slavonic/Slavic influence on the syntax of Romanian we put forth the following algorithm: if a feature occurs in at least three of the four sources indicated below, then this feature cannot be considered to have Slavic origins.The sources taken as reference points are: (i) (late) Latin (for which we use especially the information from Ledgeway, 2012 andAdams, 2013); (ii) other old Romance languages or dialects (we refer to Company Company, 2006 for Old Spanish, to Rohlfs, 1969, Buridant, 2000, Lardon & Thomine, 2009 for Old French, to Salvi & Renzi, 2010 for Old Italian, to Ledgeway, 2009 for Old Neapolitan, etc.); (iii) old Romanian original texts (from the 16 th and the 17 th centuries, which presumably have not been heavily influenced by Slavonic); (iv) South-Danubian dialects (these varieties have not been in contact with Slav(on)ic varieties, this phenomenon being subsequent to the dialectal split); however, we will use the dialectal data with care, since, in certain situations, it is possible that these varieties have been influenced by the modern Slavic languages with which they have been in contact; if such an influence is at play, then the dialectal data are not relevant for our algorithm.

Case study I: the position of relative adjectives
In contrast with modern Romanian, in which relative adjectives (such as omenesc 'human-like' , dumnezeiesc 'divine' , românesc 'Romanian' , literar 'literary' , etc.) are obligatorily postnominal, in old Romanian (up to the 19 th century) these adjectives could also occur prenominally (2); in the earliest surviving Romanian texts, the postnominal position was already more frequent (for a more detailed analysis, see Brăescu & Dragomirescu, 2014).
(2) a. glăsi evreiasca limbă și grăi (cv, 18 v ) speak.ps.3sgJewish.deflanguage and say.ps.3sg 'he spoke Hebrew and he said' b.Dumnezeiasca slujba ce e întru sfinți (cl, 7 v ) divine.def service.defwhich is for saints 'the divine service which is for the saints' c. acesta, deaca auzi acea păgînească poruncă (svi, 2 v ) this.one if hear.ps.3sg that pagan.f.sg order 'this one, if he heard that pagan order' d. făcîndu-i običnuită și politicească cinste (cg, 304) make.ger=cl.dat.3sgusual.f.sg and political.f.sg consideration 'giving to him usual and political consideration' In Niculescu (1999, p. 189-196) it is shown that the postposition of the adjective with respect to the nominal head is a result of Slavonic influence.This observation is not surprising, since Old Church Slavonic also featured adjectives in prenominal position, but this position was stylistically marked (Gamanovich, 2001, p. 315-318;Gasparov, 2001, p. 98), in contrast to old Romanian, where the ordering of the adjectives seems to be unconstrained.
Applying the algorithm described above ( §3.2), one can easily formulate the conclusion that the prenominal position of the relative adjectives can not be directly influenced by Slavonic.(i) It is well known that in Latin adjectives had free word order (3), eventually correlated with differences in meaning.What is less clear is whether this word order variation was correlated with any information structure related/markedness differences.As Ledgeway (2012, p. 50-51) has shown, it is most probable that in Latin there were not (yet) dedicated positions for the contrastive reading of adjectives; this specialisation occurred later in the Romance varieties.

Case study II: scrambling in compound verbal forms and auxiliary inversion
In this section two apparently independent syntactic phenomena, which set apart old Romanian from modern Romanian, are presented together: scrambling and auxiliary inversion (see, for details Dragomirescu, 2013 and 2014, where it is shown that the syntactic analysis of the two phenomena is unitary and a unique parametric change can account for the changes related to these phenomena).
In modern Romanian, scrambling in compound verbal forms is only possible with adverbial clitics (known in the Romanian literature as "semiadverbs"-( 7)), while in old Romanian scrambling was much more extended (8), with phrasal constituents displacing the verbal elements of the verbal cluster.( 7 on all whishes 'when the man will go as much as he whishes' As for auxiliary inversion, in the present-day language it is strictly limited to imprecations, in the spoken language (9a), to direct questions, in certain dialectal areas (9b), and to religious speech, where it is an archaic feature (9c).In old Romanian, auxiliary inversion seems to be a free phenomenon (10) (which is however preferred in main clauses, especially in initial position, see Zafiu, 2014)

Feri-va
Dumnedzău pre toți pre ceia ce-l take.care.inf=aux.fut.3sgGod dom all dom these who=cl.acc.m.3sg iubăsc și pre toți nepocăiții va piiarde (Prav., 239 v ) love and dom all nonpenitents.defaux.fut.3sgkill.inf'God will take care of all the people who love him and will kill all the people who do not repent' In previous research regarding scrambling and/or auxiliary inversion (Moldovanu, 1977(Moldovanu, -1978;;Gheție & Zgraon, 1981;Zamfir, 2007;Frâncu, 2009), it is shown that this atypical word order is determined either by the preservation of an archaic syntactic rule, which was probably general in Romanian before the first attested texts (see Moldovanu, 1977Moldovanu, -1978)), or represents the result of the Slavonic influence, by means of translations.The second explanation is supported by the existence of auxiliary inversion in Old Church Slavonic (11); unfortunately, we did not find any relevant information in the literature with respect to scrambling in Old Church Slavonic.
(11) ne srdoe li naju gorę part bě aux vы naju 'Were not our hearts burning within?' (Lk, 24, 32, apud Pancheva, 2008) Nevertheless, the application of the algorithm described in §3.2 leads to the conclusion that Slavonic influence is not a satisfactory explanation for the phenomena investigated here.
(i) It is known (Thielmann, 1885;Bauer, 2006;Adams, 2013) that in Latin the structures of the type habeo + object + past participle are attested since the archaic period, but, as extremely convincingly shown by Adams (2013, p. 646, passim), habeo was not an auxiliary, and, consequently, it did not have word order and adjacency constraints.For other forms containing an auxiliary, Latin is irrelevant, since periphrastic forms grammaticalized later in Romance, and implicitly in Romanian.The unmarked order of the habeo-structures in Latin was past participle + habeo ((12)-see Bauer, 2006, p. 293;Adams, 2013, p. 648).The possibility to insert different constituents, especially the direct object, between the past participle and habeo is largely attested in Latin (see Bauer, 2006, p. 293;Adams, 2013, p. 649).
(  Zegrean, 2012) In Megleno-Romanian, the past auxiliary is also frequently postverbal in certain varieties (19) (see Sandfeld, 1930, p. 149;Caragiu Marioțeanu, 1975, p. 282;Caragiu Marioțeanu et al., 1975, p. 207;Atanasov, 1984, p. 523-527); the future has only synthetic forms, and thus it is not relevant for the phenomenon investigated here.Atanasov (1984, p. 527) shows that the participle + auxiliary ordering is used in Megleno-Romanian when speakers report an event to which they did not take part (see 19c-d, where the context is larger), in other words auxiliary inversion marks evidentiality.This is also the case in Macedonian, a Slavic language that might have influenced Megleno-Romanian; if Megleno-Romanian was indeed influenced by Macedonia, then the data from this dialect are not relevant for our demonstration.

Conclusions
In this final section we try to answer, at least partially, the questions raised at the beginning of §3.
The data briefly presented here show that, before considering that a syntactic feature (of Romanian) is the outcome of a foreign language influence (Slav(on)ic), one should take into account other possible scenarios, which can prove more appropriate.For (Daco-)Romanian, it is without doubt that the Latin heritage, the comparison with other old Romance languages, and the comparison with the South-Danubian varieties can be more useful than the claim-more often not supported by data and influenced by ideological considerations-that certain phenomena are due to Old Church Slavonic influence.
The algorithm we have put forth and applied to two phenomena considered in the literature to have Slavonic origins, rather explained as tendencies common to the Romance languages (eventually subordinated to certain general tendencies of the Indo-European languages, hence the similarities with Old Church Slavonic and with modern Slavic languages) can be applied to the entire list of phenomena presented in section §3.1, possibly augmented with other syntactic data.
If this algorithm is not entirely satisfactory, one can also resort to a syntactic formal analysis which, independently of the data from Latin or other Romance languages, explains the diachronic differences; for example, one can invoke the generative concept parametric change (where parameter means a formal feature of a functional category, while parametric change is related to the distribution of features, which, in diachrony, can be subject to certain transformations- Roberts, 2012).In this framework, the obligatorily postnominal position of relative adjectives from modern Romanian is the result of a parametric change taking place in Romanian: the loss of a specific type of movement (phrasal movement, precisely A(djectival)P(hrase)-movement) in the nominal domain (see, for details, Brăescu & Dragomirescu, 2014).In a similar manner, the loss of scrambling and auxiliary inversion can be accounted for by a unique parametric change: the loss of a feature responsible for movement in the Complementizer domaincorrelated with the loss of auxiliary inversion-and in the little v domain (responsible for licensing Accusative case)-correlated with the loss of scrambling (see, for details and references, Dragomirescu, 2013 and2014).
Applying this algorithm and finding formal explanations for the phenomena inventoried under §3.1 will allow us to give a more coherent description of the syntactic features of (old) Romanian, in relation to Latin and Romance languages, on the one hand, and with Old Church Slavonic and modern Slavic languages, on the other hand.The expected result of this research is undoubtedly a list of syntactic features borrowed form Slavonic much more limited than it is actually shown in the existing literature.
Coming back to the grounds which determined the overstatement of the Slavonic/Slavic influence in the Romanian syntax, one can claim that the reasons for "falsification" are mainly ideological, but one can also add the lack-up until recently-of the relevant syntactic information about (Late) Latin, Romance languages (especially their old stages), and even about the South-Danubian varieties, as well as the perpetuation, without checking, of information from works which have been ideologically corrupt.