en-phrases and their morphosyntactic and semantic particulars

Prepositions constitute a problematic category because they tend tohave complex semantic and syntagmatic properties, vary in case governance and are frequently in variation with each other. Reflexes of the Indo-European *en are well attested and remain productive in Greek, Classical Armenian, Gothic and Old Church Slavic among other languages. Correspondences of Greek en/eis with Gothic in, Armenian i and Old Church Slavic vŭ occur in many instances in the canonical Gospels of the New Testament. However, Greek en/eis is frequently translated with other constructions in these languages that range from prepositional phrases that contain prepositions non-cognate with *en and nominal constructions to clausal structures. This investigation examines such correlations and points out morphological, syntactic, semantic and pragmatic factors that promote these correspondences. Case syncretism and changes in the prepositional governance are among the leading reasons that prompt translators to look for translational means other than the cognate constructions. As expected, differences in the inventory of prepositions available in the languages being examined and diversity in the division of semantic space by the prepositional phrases also add to the variety of possible renditions of Greek en/eis. Among pragmatic factors that influence the translation are the compositional marking of a certain concept, the complexity of an event/situation beingmarked, and the pragmatic appropriateness of a certain formation.

(1) kai ḗrksato kērússein en têi Dekapólei [Gk] jah dugann merjan in Daikapaulein [Goth] ew sksaw k'arozel i Dekapołin [Arm] i načetŭ propovědati vŭ Dekapoli [OCS] 'and he began to preach in Decapolis' (Mk 5:20) However, Goth in, Arm i and OCS vŭ are not the only equivalents of Gk en/eis.For example, in OCS Gk en+Dat is also translated with the help of phrases with prepositions which are not reflexes of IE *en.Cf. the Gk correspondences with OCS na (governing the locative in (2a) and the accusative in (2b)) and po (governing the dative in (2c)) in the following examples: (2) a. homoía estìn paidíois kathēménois en taîs agoraîs [Gk] podobĭnŭ jestŭ dětištemŭ sědeštemŭ na trŭžištixŭ [OCS] 'it is like children who sit in the markets' (Mt 11:16) [1] In a number of passages, just like in (1) PPs with *en reflexes express spatial relations connected with Containment metaphor.This metaphor is understood as viewing a certain location (landmark) as a containment with boundaries and appears to be crucial for the development of the semantics of *en reflexes in various languages including Gk, Goth, Arm, and OCS.See, for example, the discussion of the development of different semantic roles of *en reflexes in Ancient Greek in Luraghi (2003, 82-94, 107-117).
semantics and morphosyntax of *en-phrases [191] b. hína héna apolúsō humîn en tôi páskha [Gk] da jedinogo vamŭ otŭpuštǫna pasxǫ [OCS] 'so that I will release to you one man at the Passover' (Jn 18:39) c. hupestrṓnnuon tà himátia heautôn en têi hodôi [Gk] postilaaxǫrizy svojępo poti [OCS] 'they spread out their garments along the way' (Lk 19:36) In addition, Gk PPs with en/eis are sometimes rendered with nominal phrases (e.g., nouns in the dative, the accusative and the genitive in Goth), adjectival phrases (e.g., adjectives with a possessive suffix in OCS) and even clausal structures (e.g., clauses with the conjunctions jegda or jako in OCS and with the conjunctions zi and əndēr in Arm).Such correspondences will be also exemplified and analyzed throughout the discussion.
This study concentrates on the examination of the correspondences where a Gk en/eis is matched with a non-PP in Goth, Arm and OCS.The aim of this investigation is to analyze an intricate mechanism and interrelations of different morphological, syntactic, semantic and pragmatic factors that influence the translation of PPs and to show how they endorse correlations between Gk en/eis and a non-PP in Goth, Arm and OCS despite the genetic relatedness and morphosyntactic and semantic closeness of Gk en/eis, Goth in, Arm i and OCS vŭ.The examined corpus includes the canonical gospels of the New Testament in Biblical Greek and corresponding translations into Goth, Arm and OCS.The focus is on the internal relationship between syntax, morphology, semantics and pragmatics.
[2] m o r p h o s y n ta c t i c c o n s t r a i n t s Despite their relatedness Gk en/eis, Goth in, Arm i and OCS vŭ do not have the same syntactic, morphological or semantic loads in the languages in question.Let's examine the variety of the semantic functions of PPs with Gk en/eis, Goth in, Arm i and OCS vŭ summarized in table 1 on page 192. 2  Table 1 shows that there is a slight difference in case governance of *en reflexes in Gk, Goth, Arm and OCS.Namely, in Gk en governs Dat and eis is combined with nominals in Acc.In Goth and in Arm the case assignment for the *en reflexes is tripartite: Goth in governs Dat, Acc and Gen, while Arm i takes Loc, Acc and Abl.Finally, OCS vŭ governs two cases-Loc and Acc.The third case assignments in Goth and Arm are results of various changes of a different nature that took place in these languages.Goth in+Gen appears to be reserved for the designation of Beneficiary (Behalf) and it seems that it is this semantic function that is responsible for the existence of this construction.The origin of i+Abl has been variously explained.We [2] In table 1 and henceforth the following abbreviations are used: Dat-the dative case, Acc-the accusative case, Gen-the genitive case, Abl-the ablative case, Loc-the locative case, Ins-the instrumental case, Voc-the vocative case.The symbol ++ marks the function with which a P is used most frequently.
OSLa volume 3(3), 2011 [192] olga thomason follow the proposal of Meillet who considered i+Abl to have the same origin as OCS isŭ/izŭ 'from', thus, the unification of i+Acc, i+Loc with i+Abl is a result of merger (Meillet 1936, 95-96). 3In this light, it is clear that the designation of Source by Arm i+Abl is not the result of an unusual semantic extension, but rather the preservation of the main meaning of the original.Synchronically, of course, Classical Arm i presents an intriguing case of the semantic merger of the three most significant spatial concepts -Location, Direction and Source.It is not surprising that Gk eis+Acc, Goth in+Acc, Arm i+Acc and OCS vŭ+Acc primarily designate Direction, since this meaning is primary for Acc (cf.data in table 2 on page 193).Note that in Gk the saliency of the concepts Location and Direction led to further disambiguation of these notions which resulted in lexicalization (Gk en vs. eis).Table 2 shows that the reflexes of *en in Gk, Goth, Arm and OCS agree at least in regards to their spatial usages: their semantics remain concentrated on the denotation of Direction, however, all of them receive a semantic extension to mark Location.
The primary function of Gk en+Dat, Goth in+Dat, Arm i+Loc and OCS vŭ+Loc is the denotation of Location (as seen in table 3 on page 194).Just like in the case of *en reflexes governing Acc, this function goes in line with original meaning of Loc which mainly marks a space where an action takes place.The repetition and persistence with which PPs governing Acc refer to Direction and those governing Loc (and cases that the locative have merged into) mark Location are remarkable.This tendency provides valid grounds for typological conclusions about the saliency of these two functions for human cognition.At the same time, since languages tend to mark these notions using the same preposition one should keep in mind the closeness of these concepts.Note that Gk is the only language among the examined ones that attempts to produce a further distinction between Location and Direction on [3] For a more detailed discussion of the origin of i+Abl see also Thomason (2005).(Luraghi 2009, 291-292).
Table 3 demonstrates the richness of the semantics of *en reflexes governing Dat in Gk and Goth and governing Loc in Arm and OCS.Partially this diversity of the semantic function exists as a result of syncretism of case systems which leads to rearrangement of the functions of the cases.For example, it is a well-known fact that Dat in Gk and Goth is a result of multi-leveled syncretism.Thus, in Gk the IE Dat, Loc and Ins merged and in Goth the IE Dat, Loc, Ins and Abl merged.Therefore, the semantics of the Dat in Goth and Gk are enriched with meanings of these cases: Ins brings in the denotation of Means and Agent (and further Comitative), Dat-the meaning of Recipient, Loc marks Location and Abl tends to express Source (Meier-Brügger 2003, 261-73, Luraghi 2009, 286-288).Table 3 shows that the semantics of *en reflexes governing Dat in Gk and Goth are extended to some of these functions as opposed to *en reflexes governing Loc in Arm and OCS.For example, Gk en+Dat regularly marks Beneficiary (largely due to the input of the IE Dat), while its common counterparts Arm i+Loc and OCS vŭ+Loc do not receive such an extension and the semantics and morphosyntax of *en-phrases [195] translators had to find other means to accommodate this meaning as exemplified in (3) below: (3) kalòn érgon ērgásato en emoí [Gk] þannu goþ waurstw waurhta bi mis 4  [Goth] zi gorc bari gorceac' da yis 5  [Arm] dobro bo dělo sŭděla o mĭně [OCS] 'for she did a good deed for me' (Mk 14:6) Similarly, in the instances where Gk en+Dat regularly expresses a comitative function (due to the input of the IE Ins) Arm and OCS answer with PPs other than Arm i+Loc and OCS vŭ+Loc.Goth also has a P which is not cognate to Gk en+Dat, but this is because of the different distribution of the semantic space in this language (Goth miþ+Dat is the preferred construction in such passages): (4) ei dunatós estin en déka khiliásin hupantêsai [Gk] siaiu mahteigs miþ taihun þusundjom gamotjan [Goth] et'e karoł ic'ē tasn hazaraw zdēm ownel [Arm] ašte silĭnŭ jestŭ sŭ desetŭjǫtysoštĭ sŭrěsti [OCS] 'if he is able with ten thousand to meet' (Lk 14:31) Arm and OCS were not as greatly affected by syncretism as Gk and Goth.In OCS only the Abl merged with the Gen, whereas in Arm only the Voc has been lost (just as it was in Gk, Goth and OCS).Thus, it is not surprising that in a number of cases Gk en+Dat/eis+Acc are translated by nominal constructions (NPs) in OCS and Arm.In Goth one also finds instances of correspondences with NPs, however, they are not as frequent as in OCS and Arm and appear to be semantically limited.For example, Goth Dat renders Gk en+Dat in several passages, but in all of them this PP marks Location 'in' 6 as a result of a metaphoric extension (connected with the vision of human bodies or clothes as Containment (shown in (5a) and (5b) respectively): (5) a. dialogízontai en heautoîs [Gk] mitodedun sis [Goth] 'they pondered within themselves' (Mk 2:8) b. ánthrōpon en malakoîs ēmphiesménon [Gk] mannan hnasqjaim wastjom gawasidana [Goth] 'a man clothed in soft raiment' (Mt 11:8) [4] Note that Goth in+Dat is not extended to the denotation of Beneficiary either, but this instance is not parallel to those in Arm and OCS, because here Goth practically allocates in+Gen along with other PPs (e.g.bi+Dat in the example in (3) for the designation of Beneficiary).[5] Note that Arm responds with i+Acc and not i+Loc here, because, of course, for i+Acc the denotation of Beneficiary is an expected extension since it regularly marks Recipient and Purpose.[6] Note once again the input of the IE Loc which has merged into Goth Dat (see the discussion earlier).
OSLa volume 3(3), 2011 [196] olga thomason Gk eis+Acc is rarely matched with Acc in Goth.It occurs only in passages where the Gk PP is combined with a verb of speech and has a Malefactive function as in (6a) (directional accusative) or where the complement of the Gk eis was the noun aiṓn 'space of time, duration' and the PP expresses the meaning 'forever' as in (6b) (temporal accusative): (6) a. hòs d' àn blasphēmḗsēi eis tò pneûma tò hágion [Gk] aþþan saei wajamereiþ ahman weihana [Goth] 'but whoever blasphemes against the Holy Spirit' (Mk 3:29) b. kaì ou mḕ apólōntai eis tòn aiôna [Gk] jah ni fraqistnand aiw [Goth] 'and they will never perish' (Jn 10:28) As expected, in Arm and OCS correspondences of Gk en+Dat/ eis+Acc with NPs are more regular.Ins becomes an absolute leader among cases in Arm and OCS when it comes to render Gk en+Dat.NPs in this case correspond to Gk en+Dat not only in instances where the Gk construction has causal functions (Means, Agent), the original meaning of Ins, as exemplified by (7a), but also where the Gk PP expresses spatial notions (e.g., Path which is also expressed by the Ins as illustrated in (7b)).Note also how in (7a) the Goth variant agrees with the Gk original in rendering causal semantics.Such concurrence, of course, is due to the parallel case syncretism in these languages and it is not unexpected that the correspondence Gk en+Dat : Goth in+Dat : Arm Ins : OCS Ins with causal functions occurs repeatedly and becomes one of the most stable correspondences involving *en reflexes.Gk eis+Acc corresponds to the OCS and Arm Dat in a number of occurrences where this PP marks Beneficiary (or Malefactive), a regular function for Dat, as shown in (8a), or Goal, as illustrated in (8b).Note the closeness of these functions which differ primarily because the former is associated with the notion of animacy and the latter does not.These are not the only functions which the correspondence semantics and morphosyntax of *en-phrases [197] set Gk eis+Acc : Arm Dat : OCS Dat covers, 7 but they are the most frequent ones.
(8) a. ei mḗti poreuthéntes hēmeîs agorásōmen eis pánta tòn laòn toûton brṓmata [Gk] bayc' et'e ert'ic'owk' gnesc'owk' bawakan žołovrdeand kerakowr [Arm] 'unless we go and buy (Arm: sufficient) food for all these people' (Lk 9:13) allà taûta pánta poiḗsousin eis humâs [Gk] nŭ si vĭsja sŭtvoretŭ vamŭ [OCS] 'but they will do all this to (=against) you' (Jn 15:21) b. eis oudèn iskhúei [Gk] oč' imik' azdic'ē [Arm] 'it will be good for nothing' (Mt 5:13) misthṓsasthai ergátas eis tòn ampelôna autoû [Gk] najetŭ dělatelŭ vinogradu svojemu [OCS] 'to hire workers for his vineyard' (Mt 20:1) There are also instances where Gk en+Dat/ eis+Acc are rendered with constructions other than their prepositional cognates, but those are not due to specifics of the case systems in the languages in question.Consider the following examples: In (9a) en+Dat is rendered by an adverb in Arm and OCS and by an adjective in its strong masculine accusative singular form (derived from a preterit participle based on a preterit-present verb kunnan) in Goth.In (9b) the Gk relative clause construction is simplified in OCS (even though relative constructions are also used in OCS) and the whole clause is translated with the help of an adjective that contains a possessive suffix -ĭsk-.Such correspondences are common in the New Testament.Note that Arm and Goth follow the Gk original without any changes.Examples like these, [7] Gk eis+Acc is matched with the Dat in Arm and OCS in a few instances where this Gk PP expresses directional notions, topic or has a temporal meaning.

olga thomason
showing interference of the lexical and grammatical means available in a given language are important parts of the text that demonstrate the creativity of translators and their attempts to deliver the meaning of the original passage, stay true to it, but not at the expense of the linguistic specifics of their own languages.
Passages where the complement of the Gk en/eis is an infinitive receive a discrete treatment in Goth, Arm and OCS.In all four languages there are examples of infinitives serving as a subject or as a complement of a finite verb, however, not all of them allow constructions where an infinitive becomes the complement of a P. Undoubtedly, this is connected with the nature of an infinitive as a category and its morphosyntactic specifics in each language.
Thus, in Gk infinitives are fully incorporated in the verbal stem system, they have voice and tense.Gk inflects not the infinitive itself, but the preceding article.Goth and Arm form their infinitives from the present stems, but Arm inflects its infinitives while Goth does not.In most cases OCS infinitives are formed directly from the roots and none of them are inflected.Morphological specifics of infinitives in Gk, Goth and Arm allow them to occur as a complement of Ps.These languages make use of such constructions with different frequencies.Gk PPs with an infinitive as a complement are used either to mark Purpose or to express temporal values.Thus, Gk en+infinitive denotes Time 'while, during' (Gk pro+infinitive expresses Time 'before'), while Gk eis+infinitive designates Purpose (Gk pros+infinitive also has this function).
OCS does not allow an infinitive to be the complement of P and has to use other means to translate the Gk original in such cases.Thus, Gk en+infinitive is repeatedly translated into OCS with the dative absolute construction (consisting of a participle and a noun/pronoun, both in the dative case, and typically expressing accompanying circumstances) which is regularly used in this language to mark a period of time during which something happened: (10) ephobḗthēsan dè en tôi eiseltheîn autoùs eis tḕn nephélēn [Gk] ubojašęže sęvŭšĭdŭšemŭ imŭ vŭ oblakŭ [OCS] 'they feared when they were entering the cloud' (Lk 9:34) Less frequently, Gk en+infinitive is matched with relative clauses introduced by the OCS conjunctions jegda as in (11a) or jako as in (11b): 8 (11) a. kaì egéneto en tôi epaneltheîn autón [Gk] i bystŭ jegda vŭzvrati sę[OCS] 'and when he came back' (Lk 19:15) [8] Gk pro+infinitive designating Time 'before' is translated in a similar fashion.It is matched in OCS with relative clauses introduced by the OCS conjunction prěžde.
[200] olga thomason b. kaì dúnamis kuríou ēn eis tò iâsthai autón [Gk] ew zawrowt'iwn TN̄ēr i bžškel znosa [Arm] 'and the power of the Lord was to heal him (Arm: them)' (Lk 5:17) But other means are also available and, as in the case with OCS, the Arm translator sometimes uses clausal structure or infinitive alone to render the meaning of the Gk en/eis+infinitive.But in addition to that, Arm has Ps other than i that can take an infinitive as a complement.Thus, Gk en+infinitive in several passages is translated with Arm ənd+infinitive (in its accusative form) as shown in (16a).It is also sometimes rendered with a temporal clause introduced by the conjunctions minč'(ew) or ibrew as illustrated in ( 16b) and (16c) respectively: (16) a. en dé tôi hupágein autón [Gk] ew ənd ert'aln nora [Arm] 'and when he was leaving…' (Lk 8:42) b. kaí egéneto en tôi poreuˊesthai eis Ierousalḕm [Gk] ew ełew minč'ew ert'ay na yĒM [Arm] 'and when he went to Jerusalem' (Lk 17:11) c. kaí en tôi eisagageîn toús goneîs tó paidíon [Gk] ew ibrew acin cnawłk'n zmanowkn [Arm] 'and when the parents brought in the child' (Lk 2:27) Gk eis+infinitive is occasionally translated with the infinitive alone in Arm: (17) kaí paradṓsousin autón toîs éthnesin eis tó empaîzai [Gk] ew matnesc'en zna het'anosac' aypanel [Arm] 'and they will give him to the Gentiles to be mocked' (Mt 20:19) From the examined instances we see that the translators in Goth, Arm and OCS strived to remain true to the text of the original and preserve the clausal meaning of the constructions en/eis+infinitive.The fact that we find a different set of correspondences in examples where an infinitive becomes a complement of a P supports the propositions of those linguists that argue that infinitival constructions are sentential and not just a type of a simple VP (Koster & May 1982).
[3] s e m a n t i c c o n s t r a i n t s In a number of passages Gk en+Dat and eis+Acc are rendered with non-cognate PPs in Goth, Arm and OCS due to semantics factors.Having examined all such instances in the canonical gospels of the New Testament we can argue with certainty that in such cases no matter what kind of semantics Gk en+Dat and eis+Acc may have, they can be translated by the non-cognate PPs which may be of different kinds, but all of them will have one thing in common: the function expressed by Gk en+Dat and semantics and morphosyntax of *en-phrases [201] eis+Acc will be either their primary semantic role or at least a frequent one.

[3.1] Gk en+Dat/eis+Acc: primary function translated by primary function
In the instances Gk en+Dat and eis+Acc are used with their primary functions (the denotation of Location 'in' and Direction 'into' respectively) they are translated with non-cognate PPs in Goth, Arm and OCS which also have the designation of Location 'in' and Direction 'into' as their main/recurrent semantic roles.These are the passages where Gk en+Dat is rendered with Goth ana+Dat, Arm ənd+Acc/Loc and OCS na+Loc (as exemplified in (18a)-( 18c)) and Gk eis+Acc is translated with Goth du+Dat, Arm z-+Acc and OCS na+Acc 11 (as illustrated by ( 18d)-( 18f)).( 18) a. oi patéres hēmôn tò mánna éphagon en têi erḗmōi [Gk] attans unsarai manna matidedun ana auþidai [Goth] 'our fathers ate the manna in the wilderness' (Jn 6:31) b. kaì èn hólēi têi oreinêi tês Ioudaías [Gk] ew ənd amenayn lernakołmnn Hrēastani [Arm] 'and in (=throughout) the whole mountain region of Judea' (Lk 1:65) c. hóti ho misthòs humôn polùs en toîs ouranoîs [Gk] jako mĭzda vaša mŭnoga na nebesĭxŭ [OCS] 'for your reward is great in heaven' (Mt 5:12) d. húpage eis tòn oîkón sou [Gk] gagg du garda þeinamma [Goth] 'go to your house' (Mk 5:19) e. kaì pâs eis autḕn biázetai [Gk] ew amenayn ok' zna brnabarē [Arm] 'and everyone rushes into it' (Lk 16:16) f. hópōs ekbálēi ergátas eis tòn therismòn autoû [Gk] da izvedetŭ dělateljęna žetvǫsvojǫ[OCS] 'so that he send out workers into his harvest' (Mt 9:38) For example, in (18d) and (18e) Gk eis+Acc corresponds to Goth du+Dat (another regular marker of Direction in Goth) and Arm z-+Acc only because these constructions are in free variation with the respective *en cognates.The factor that sponsors the correspondence Gk *en cognate : Goth/Arm/OCS non-cognate is of a purely semantic nature and connected with differences in the inventory of the prepositional systems in the respective languages.This disparity gives rise to different assignments of semantic functions to the members of the prepositional systems in these languages.These types of correspondences support the idea of the idiomatic nature of the Goth, Arm and OCS translations since the translators seem to make the [11] There are some other factors that influence the choice of P in this kind of correspondences.They are discussed in section [4] below.
[202] olga thomason choice of the constructions that they use not only based on the text of the original, but also on the basis of the resources available to them in the languages with which they are working.
[3.2] Gk en+Dat/eis+Acc: secondary function translated by primary function As expected, we find a number of passages where Gk en+Dat/ eis+Acc used with their secondary functions are translated into Goth, Arm and OCS with the help of phrases with non-cognate Ps.As it is stated in section [3], in such instances the Gk constructions are rendered with non-cognate PPs for which the semantics expressed by the Gk counterpart are primary or as a minimum frequent.Examples of this type once again support the idea of the idiomatic nature of Goth, Arm and OCS translations.
Compare the following passages: (19) a. ei dunatós estin en déka khiliásin [Gk] siaiu mahteigs miþ taihun þusundjom [Goth] 'whether he is able with ten thousand' (Lk 14:31) b. kai periêgen en hólēi têi Galilaíāi [Gk] ew šrǰēr ȲS̄ənd amenayn kołmn Gałiłeac'woc' [Arm] 'and he (Arm: Jesus) went throughout the whole region of Galilee' (Mt 4:23) c. en gàr toútōi ho lógos estìn alēthinòs [Gk] o semĭ bo slovo jestŭ istinĭnoje [OCS] 'for about this the word is true' (Jn 4:37) In (19a) the comitative function of Gk en+Dat is rendered with Goth miþ+Dat for which this meaning is primary.The Gk passage in (19b) where eis+Acc denotes Path (a less frequent function for this construction) is translated in Arm with ənd+Acc (a regular marker of Path in this language).OCS regularly designates topic with o+Loc and that is why this PP becomes a counterpart of Gk en+Dat in (19b).All these instances in (19) exemplify how synchronic peculiarities in the distribution of the semantic load in each of these languages increase the number of possible counterparts for Gk en+Dat/ eis+Acc, in other words, how internal content affects the external form.
[4] p r a g m at i c fa c t o r s Reasons for translation of Gk en+Dat/ eis+Acc by means other than their cognates could be not only purely grammatical (which were illustrated earlier), but could also reflect personal preferences, the interpretations of a translator and the pragmatic complexity of a concept and the translator's vision of the notion which is being discussed in a certain passage.Consider (20) where Gk eis+Acc is rendered with OCS na+Acc: OSLa volume 3(3), 2011 semantics and morphosyntax of *en-phrases [203] (20) anébē eis tò óros [Gk] vŭzide na gorǫ[OCS] 'he went up onto the hill' (Mt 14:23) It is feasible to suggest that (20) is an example that shows how the same situation can be interpreted through the accentuation of different domains.Thus, OCS na+Acc does not only act as a marker of Direction, it also specifies the notion Surface whereas Gk eis+Acc designates Direction and accentuates the notion of Containment.It would be wrong to propose that Gk is not sensitive to the division between Containment and Surface, because one frequently finds epí+Acc in passages expressing motion on top of some surface: e.g.pâs ho pesṑn ep' ekeînon tòn líthon 'everyone falling upon that stone' (Lk 20:18).It appears that the opposition Surface vs. Containment is not absolute in Gk in a sense that the PPs that are selected to designate these concepts do not do so exclusively.In other words, both constructions-Gk eis+Acc and epí+Acc-can mark direction connected with the notions Containment and Surface.The difference is of a quantitative nature.Gk eis+Acc expresses the first meaning more frequently whereas Gk epí+Acc tends to designate the second function with greater regularity.
We find a similar opposition in Goth where the opposition Containment vs. Surface is expressed by PPs in+Acc vs. ana+Acc, in Arm where this set is marked by PPs i+Acc vs. i veray+Gen respectively and in OCS where these notions are set apart by vŭ+Acc and na+Acc respectively.Undoubtedly, in cases like (20) the meanings of the NPs are responsible for the endorsement of the concepts Surface and Containment, but nominal semantics are not the only items that contribute here.One cannot predict the usage of Ps based on the lexical input of nominal constituents.Namely, it would be wrong to claim that if an NP denotes Surface (desert, road, etc.) then Gk epí, Goth ana, OCS na will be necessarily used (as it happened in the OCS translation in (20)).In addition, it is important to note that the prefix in Gk verb an-'upwards, above, on the top' is also connected with the notion Surface whereas the OCS vŭz-'upwards' does not exhibit such an obvious connection.And, of course, Ps also have their own semantic valency that in turn influences their syntactic particulars.Thus, our data supports the main proposition of various models of Construction Grammar suggesting the undivided continuum between form and meaning where meaning is induced by a concept and realized by a pragmatic meaning.
This proposition is also supported by the ideas expressed by Nikitina (2008) where she points out the importance of pragmatics in such cases.Having analyzed variation in spatial goals markers, Nikitina concluded that even if a language has lexicalized means to mark a certain concept (English prepositions into expressing goal in her case) a speaker does not have to denote this notion exclusively by this lexeme.This notion could be marked in a compositional manner; so that it is not a P, for example, but other members of a construction (verbs, particles) contribute to olga thomason the whole picture.Thus, the meaning is inferred not just from one particular component of a passage, but from a combination of sentence members whose meaning is restrained by such factors as context and the conceptualization of the event in question.In our case the conceptual pair Surface-Containment could be marked not only by the PPs mentioned above, but also, of course, by NPs or prefixes (for example, in (20) Surface is signaled by the Gk prefix an-) and then a translator is given a choice in his selection of a P since the concept is already marked by something else.
While rendering a passage, a translator analyzes the text based on his own prior experiences of the surrounding world and communications.Thus, using the semantic complexity and relativity of the concept in question, a translator has the option of choosing a point of view (rendition).This is another reason why in spite of a high frequency of correspondences between Gk en/eis and their cognates in the languages under consideration, one finds a number of correspondences between Gk en/ eis and non-cognate PPs.Factors related to the conceptualization of situation/event/entity also influence the translation.Compare the following correspondences: In (21a)-(21c) the original Greek text receives different renditions which occur not because of some grammatical particular of a language into which the manuscript is being translated.The reason is of a pragmatic nature connected with the complexity of the conceptualization of the situation in question and the fact that many situations can be viewed from different perspectives.These perspectives are made possible by contiguity of ideas, events, and experiences.And it is for this reason that translators reinterpret the original Gk passages, thus increasing the level of variation in means used to translate Gk en+Dat/ eis+Acc.Going back to (20), here we can also find an instance of such a rendition.The general conceptualization of an event is also at play here, since going up the hill will result in being on its top (Surface), however, a person/object which is moving up the hill is comparatively small and (especially from the distance) could be viewed as being a part/inside of the hill (Containment).The passages in ( 20) and ( 21) show that pragmatic and semantic factors are closely interconnected and it is difficult at times to draw strict semantics and morphosyntax of *en-phrases [205] boundaries between them.
The pragmatic appropriateness/inappropriateness of a certain construction in a particular context is another factor that affects translation.Compare the following correspondence set: (22) ouaì dè taîs en gastrì ekhoúsais [Gk] gore že neprazdĭnyimŭ [OCS] 'and alas to those who are with child' (Mt 24:19) The Gk en+Dat expressing the meaning 'with child' is rendered with the OCS adjective 'pregnant' (lit.'not empty') not because OCS does not have a construction structurally parallel to that of Gk.In fact it does and we find Gk en+Dat translated with a construction with OCS vŭ in several passages, cf. ( 23): (23) sullḗmpsēi en gastrì [Gk] začĭneši vĭ črěvě [OCS] 'you shall conceive in the womb' (Lk 1:31) However, it is likely that the translator chose the rendition with the adjective because a literal interpretation of the Gk phrase is seen by him as culturally inappropriate.Apparently, only Northern Slavic languages demonstrate the semantic extension '(with) a stomach' > 'pregnant' and Old Russian brjuxataja 'pregnant' is considered to be a rude term (Trubachev 1976).
[5] c o n c l u s i o n s As we have seen, Gk en+Dat/eis+Acc receives an abundant selection of correspondences in Goth, Arm and OCS despite the fact that there exist cognate Ps in these languages that have relatively similar grammatical functions.The assortment of correlations ranges from phrases with Ps which are non-cognate with Gk en+Dat/eis+Acc and NPs in a variety of case forms to clausal structures.This variety is made possible for a number of reasons which are morphological, syntactic, semantic or pragmatic in nature.As it is demonstrated in this article, it is not uncommon to find instances where several of these factors influence the translation at the same time, demonstrating the complexity of the interrelations among them.The main factors that affect the translation include discrepancies in prepositional case governance and results of syncretism, differences in the inventory of prepositions available in a given language and as its outcome diversity in division of semantic space by PPs.Several pragmatic aspects that sometimes affect the choice of translational means, such as the possibility of a compositional marking of a certain concept, the complexity of an event/situation being marked, and the pragmatic appropriateness of a certain construction should also be taken into account.

table 1 :
Semantic load of various reflexes of *en in Gk, Goth, Arm and OCS.

table 2 :
Semantic load of *en reflexes governing Acc in Gk, Goth, Arm and OCS.