On the “Pleonastic” Usage of Complement Markers in Early Modern Greek

A syntactic feature that characterizes Early Modern Greek is the “pleonastic” usage of the complement conjunction óti or pos with the mood (“subjunctive”) particle na , as well as the co-presence of the complementisers óti and pos . These co-occurrences are ungrammatical in Modern Greek, while in vernacular Late Medieval and Early Modern Greek texts they are sufficiently attested. In this paper we record a large number of instantiations of the { óti / pos } + na / óti + pos structures in order to trace the conditions of their occurrence; the examples come from extended prose texts of the 16th century as Kartanos’ “Palaia te kai nea Diathiki” (Kakoulidi-Panou 2000) or Morezinos’ “Klini Solomontos” (Kakoulidi-Panou et al. 2007), as well as an anthology of demotic prose texts of 16th century edited by Kakoulidi-Panou, Karantzola & Tiktopoulou (in press).


1.1
The Standard Modern Greek (smg) Complementation System Due to the drastic reduction of non-finite verbal forms in Modern Greek1 (mg) as a consequence of the gradual loss of the category 'infinitive'2 (cf.Joseph 1983: 37-84), the complementation system of the language makes use of a number of subordinating markers that introduce finite verbal types.Along general lines, verbs of saying, thinking, believing and other predicates that demand a clause in order to be semantically completed select complements introduced by the conjunctions 'ότι' óti and its slightly more colloquial counterpart 'πως' pos (1).The conjunction 'που' pu introduces complements after the so-called factive verbs (2),3 namely verbs that commit "the speaker to the truth of a subordinate proposition" (Matthews 22005: 125).Verbs expressing deontic and dynamic modality (obligation, necessity, willingness, ability, purpose), aspectual and causative verbs, as well as predicates expressing command or ordering select finite verbs headed by the proclitic mood particle na (3).The configuration na + verbal form is traditionally called the subjunctive4 and although the verb form after na is not morphologically marked for mood, there are good reasons to put forward that a na-clause is not merely a complement construction with modal semantic content but a genuine non-indicative mood.
( To complete the picture of complementation in mg, we might make reference to clauses introduced by μη / μήπως mi / mípos 'lest' after verbs of fearing (Holton et al. 22012: 545f.) as well as embedded indirect questions (ibid.: 540ff.).Yet, these clauses oscillate between complementation and adverbial subordination (mi / mípos-clauses entail a causal or final flavor while indirect questions basically deliver adverbial information).Since these two categories are not crucial for our analysis herein, we will not further discuss them.5

1.2
Pleonastic Complement Markers in Early Modern Greek (emg) An important feature of the complement markers óti-pos, pu and na is that they cannot occur simultaneously within a single complement clause in mg.6 Certain predicates can select either one or another complement (or even all three) but in this case the semantics of the predicate varies in accordance with the modal content of the complement.7 Still, diachronic research can reveal that a parallel occurrence of these markers has been possible in the history of Greek.8  o Theós the-God.nom.evaptísthi was-baptized.3sg.aor.ind. is in ton Iordhánin potamón the-Jordan-river.acc.m. 'As Adam heard that God was baptized in the Jordan river ….' (Kartanos,Testament: 289v) All three examples are perfectly understandable for a mg native speaker.Yet, leaving aside slight semantic and morphological aberrations, the sentences deviate from smg with regard to the fact that they select two complement markers in a way that is ungrammatical in smg.In particular, in (4) óti is ungrammatical in mg, similarly pos in (5) while the counterparts óti-pos can never appear together since they are in free distribution in mg.
Journal of Greek Linguistics 16 (2016)  The instantiations of these syntactic structures in the corpus we researched (cf.next section) may initially seem to be "pleonastic" if not "arbitrary", however it is possible to find particular patterns of distribution, as we will see in this paper.

1.3
Corpus The corpus we have examined for this research derives from literary and nonliterary texts of the 16th century.Generally speaking, the period from the late 15th c. until the mid-17th c. is an extremely important one for the vernacular languages in Europe, as well as for the language sciences.Historians of language sciences, like Sylvain Auroux, have noticed (1994: 71 and passim) that the late 15th c. was the beginning of the second techno-linguistic revolutionthe first one having been the invention of writing in the third millennium bc: this massive grammatisation9 process of vernacular languages of Europe (and of other continents as well), consisted of the production of dictionaries and grammars of all world languages on the basis of the Greco-Latin grammatical tradition.Actually, although most major European languages were attested in written form since the 9th c. and the European Middle Ages was aware of dialectal and linguistic diversity, until the Renaissance a kind of complementarity seems to have existed between the "sacred" languages (Hebrew, Ancient Greek, Latin)-languages of literacy, power and religion-and the vernaculars.
From another point of view, this dynamic appearance of vernacular languages as written (and not only oral) languages, is considered the first ecolinguistic revolution in Western Europe, namely a transaction that is "… to substitute an ecology of communication dominated, in formal uses, by Latin, with an ecology based on the coexistence, in more or less territorialized spaces, of this universal medium and a common language."(Baggioni 1997: 74).
Regarding the Greek language, after a poorly documented period (7th-12th c.),10 there has been nothing comparable to the quantity and quality of prose writing process attested from the 12th, a process that peaked in the late 15th c., which reflect the vernacular Greek in different degrees.A variety of narrative texts (e.g., chronicles, lives of saints, homiletic texts, prophetic and apocalyptic visions, romance, parody and satire, etc.) and non-narrative ones (e.g., legal works, regulations, portolans, educational books, dictionaries, books of arithmetic, exorcisms, letters, etc.) are available in manuscript and/or printed form.

9
The term is in French and should not be confounded with the English terms 'grammaticisation'/ 'grammaticalization' .10 Cf. Jeffreys 1996.The occurrences of "pleonastic" marker constructions are numerous within the corpus and therefore the phenomenon cannot be regarded as sporadic or a sort of erroneous stylistics.Moreover, in terms of statistics, the occurrences of the three combinations vary in frequency and with regard to the author.Thus, the óti + na appears the most within the Anthology (59 ×) in comparison to the óti + pos (12×) and the pos + na (4×, the less frequent constellation within this compilation of texts, cf. Figure 1).One may observe in turn that óti + na is preponderant in absolute numbers in Kartanos' book (more than 283×11 vs. 4 instances of pos+na complements12) whereas it is not attested in 11 The number here, and in all other cases, refers to the occurrences in which the markers are aligned next to each other, so that no other lexical element is interpolated.There are instances of "pleonastic" usages of óti + na in which the two elements are separated, cf.ex.16.These cases are most interesting for providing a structural analysis but are not easily traceable in statistic terms; therefore they are excluded from the numbers provided here.12 There are several constellations of an accentuated pos 'how' (from which the nonaccented complementiser pos 'that' derives) with the particle na.This case is irrelevant to our investigation herein and the only (simple) challenge is to discern between the interrogative and the complementing pos.It is reasonable to assume that the latter has a syntactic status parallel to óti.
Journal of Greek Linguistics 16 (2016) 202-231 Morezinos, who mostly employs the pos + na pleonastic structure (still only 14×, cf. Figure 2).As we will also discuss later in this article, the discrepancy in distribution may be suggestive of a dialectically conditioned usage of the combined markers.

The Cases under Examination
As mentioned above, the distribution of the numerous attestations of the "pleonastic" complement markers may appear at first sight to be unsystematic and arbitrary, and indeed the sample we extract shows a variety of environments in which the combined markers emerge.Still, a closer examination of the subordinating usages of these structures allows us to trace regularities in their occurrence.In the case of complement clauses, we focused on the categories of predicates that select combined markers and their syntactic properties, so that we can reach an explanation of their function within the synchronic system of emg.It should be borne in mind, however, that every text or author at that time differs from each other insofar as the syntactic system is concerned.This means that texts of different authors and from different regions may have developed distinct and not identical syntactic systems despite their belonging to the same linguistic continuum.Hence, in the absence of a well-established vernacular mg norm, the image of the Greek language at that time as a whole creates the impression of being in a state of flux or, as Hopper (1987) would put it, "emergent".13Yet, it is not our intention herein to express a functional stance against a formal one (cf.discussion in Fischer 2007: 53ff.), but rather to provide an overall account of the parallel phenomena in examination which can be possibly described, for example, by the notion of grammars in completion put forward in Kroch 1989.Besides, as long as we are seeking patterns in the usage of the "pleonastic" conjunctions, a functional analysis is plausible, especially within the work of a particular author or a certain genre, etc.
Having this theoretical background in mind, in this chapter we will examine the distribution of all three cases of "pleonastic" introductory markers, beginning with the óti + na clauses which constitute the majority of the cases we researched.
13 Cf. Hopper (1987: 148): "Because grammar is always emergent but never present, it could be said that it never exists as such, but is always coming into being.There is, in other words, no 'grammar' but only 'grammaticization'-movements toward structures which are often characterizable in typical ways.It goes without saying that many phenomena which we would agree to call grammatical are relatively stable and uniform.That is not in dispute.The point again is that any decision to limit the domain of grammar to just those phenomena which are relatively fixed and stable seems arbitrary." Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0

óti Plus na
The óti + na clauses serve as means of both adverbial and complementary subordination, although the former usage is far less frequent than the latter;14 therefore our focus lies on the complementary function of this structure.

2.1.1
óti + na Introducing Adverbial Clauses The adverbial clauses introduced by óti + na could have a final (7) or resulta-tive15 (8) content.In these functions the óti + na final clauses competed the διά + να dhiá +na 'in order that' ones (dhiá literally means 'for'), whereas óti + na resultative appeared along with the resultative clauses that were introduced by a "plain" óti (cf.Karantzola 2000: 548) The meaning of na in these Medieval Greek examples oscillates between futurity and modality.18In smg, the two predicates meaning 'to swear' and 'to hope' may select either a subjunctive complement clause or a declarative óti/pos + future tense19 (for example mg: elpízo na érthi vs. óti tha érthi (roughly and literally: "I hope (for) him to come" vs. "I hope that he will come").16 Cf.Horrocks 22010: 333f.and 349f.Bear in mind that a secure chronology for the composition of both epic poems is still disputable.17 In Jeffreys 1998: 268f. 18 On modality, cf.Roussou 2006: 38, fn.12. On the connection between futurity and modality, cf.Lyons (1977: 816): "reference to the future, unlike reference to the past, is as much a matter of modality as it is of purely temporal reference".19 The future tense in mg is formed by the futurity particle θα tha + a verbal form marked for aspect.Its distribution is identical to the mood particle na.The two particles cannot Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0 It may have been in this sort of usages that the óti + na configuration arose: Initially only óti played the role of subordinator and na could express the future tense, possibly somehow tinged with a sense of modality20 (it is anyway a common feature of futurity and modality that encode an "irrealis" reading as opposed to the "realis" one of the present or past indicative, cf.Palmer 22006: 104ff.).In emg similar constructions are attested, though they are fairly uncommon (in Kartanos in no more than 10 ex. and in the Anthology, only once could na be interpreted as principally a future marker): ( After commissives, the na headed by óti has a strong futurity reading; thus these complement clauses relate to the cases we saw in the previous section.In Kartanos, there are only 12 instances of this sort, in the Anthology, 5 ×.

2.1.2.2
Directives The óti + na complements appear after verba dicendi, especially after the verb λέγω légho 'say, tell' .The verbs of saying that select an óti + na clause may maintain their prototypical meaning of uttering/ speaking (13), although in most cases22 they obtained a "directive"23 reading of the sort "I tell somebody to do something" ( 14 In terms of statistics, the directive verbs form the largest predicate category selecting an óti + na clause.In Kartanos, no less than 140 out of 283 óti + na complements are selected by directives (and 22× out of 59 in the Anthology), so that it is straightforward that at least in Kartanos the directives constitute the syntactic context par excellence in which the óti + na constellation is traced.
In the first place, the compatibility of óti with na is explainable by postulating that óti in emg did not necessarily introduce assertive clauses exclusively, as is the case in mg, but it has a broader character, being a generic complementiser25 compatible with the irrealis reading of a na-verbal form, i.e. a mg subjunctive, analogous to the French que or the Italian che (both can select either the indicative or the subjunctive).26Still, this could not be "the whole story"; otherwise, óti would head any na-verbal form.Therefore, in 24 In coordination, the particle na is obligatorily repeated while óti normally appears only once.Moreover, na always forms a cluster with the verbal form (only the negator μην min and unstressed clitic object pronouns may intervene between the mood particle and the verbal form, as in mg).It seems reasonable to suggest that óti is a conjunction/complementiser while na is part of the verbal paradigm.25 Cf. also Roussou 2009: 382. 26 Whether there has been an influence of Italian dialects (e.g.Venetian) on Western Greek dialects could be an interesting working hypothesis!Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0 this constellation óti seems to frequently encode "directivity" by introducing an embedded complement that expresses a non-assertive/ performative utterance.27A related line of argumentation is to associate the selection of óti + na with indirect and reported speech, in the sense that óti introduces an embedded na-subjunctive.In a non-embedded context, the na-subjunctive can express wishes, desires, requests, orders, and less forceful commands than the imperative (Holton et al. 22012: 264).Additionally, an embedded subjunctive could encode a reported imperative.We will return to óti as a marker of indirect/ reported speech in 2.3.

2.1.2.3
Causative Verbs Causative verbs resemble the directives both semantically and syntactically: On the one hand the causative verbs exhibit a sort of intentionality similar to the one of directives; on the other hand, the argument structure has the form "I cause somebody to do/become something": There are 15 óti + na complements after causatives in Kartanos.In the anthology, this usage is not attested.

27
Roughly in the sense of Austin 1962 but in a more grammatical sense.For example, reported speech after directives usually corresponds to an imperative in direct speech: "I tell / order (etc.)you/sb to do sth" > "Do sth!".In an analogous way to directives and causatives, predicates expressing deontic modality convey "intentionality" meanings from the side of the speaker if the proposition is addressed to another person (quasi "it is necessary for you to do sth" ⟩ "I want/tell you to do sth"); therefore they express directivity.Given the nature of the texts (admonitions, (divine) orders, incitements and the like), it is reasonable to postulate that deontic modality constitutes an indirect means of expressing the will of the speaker.
Certain impersonal verbs, such as φαίνεται fénete 'it seems that' or τυχαίνει tikhéni 'it is the case that' , can select óti + na complements as well, somehow following the impersonal deontic predicates.In total, there are 19 such cases in Kartanos, 10 in the Anthology.

2.1.2.5
Volitionals Volitional predicates, especially the verb (ε)θέλω (e)thélo 'I want' could select an óti + na clause: In several cases the volitional predicates seem to be followed by óti + na when directivity is involved, i.e. when a speaker addresses his will to another person ("I want you to do sth"), while in cases of subject co-reference ("I want to do sth") a bare subjunctive clause (a "plain" na-verbal form) is preferred ( 22 In sum, volitionals appear 33 times in Kartanos, so the employment of óti + na after this category does not seem to be prototypical in this text.18 out of the 33 complements have disjoint subject reference between the matrix and the embedded clause, therefore one may assume that they have a similar function as directives (from a pragmatics point of view, when I tell an addressee that I want them to do sth, I implicate that they had better do it!).However, the rest of the cases do not exhibit disjoint reference, so they could not be categorized as sensu stricto directives, unless we suppose that directivity towards an addressee is associated with intentionality in general in the case of volitional predicates (thus, expression of will is directed either to an addressee or to the subject of the will as well).We return at this point in the next section.

2.1.3
óti Plus na and Obviation Effect?If we postulate that directivity is in play when we deal with óti + na complementation in Kartanos, who provides us the most cases of this structure (in our view statistically this is a justifiable assumption), a syntactic corollary is that necessarily óti + na clauses entail disjoint reference between the subject of the matrix and the embedded clause.A tentative "working hypothesis" is to examine if óti contributes both to the directive reading of the overall construction as well as to subject obviation.
Consider the following phenomena in two languages of the "Balkansprachbund" (i.e. the languages of the Balkan that share a number of similar-lexical, morphological and syntactic-features, cf.Tomić 2004 and2006), namely in Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0 Romanian and in Albanian.Both languages make use of a sort of complementiser that introduces the complement subjunctive clause.29In Romanian, the lexical element ca appears along with the subjunctive, which is marked by the mood particle să.The "subjunctive complementiser" ca has a variety of func-tions30 on the left periphery of the embedded clause such as topic, focus or plainly introducing a subjunctive complement after certain sets of predicates (26).In Albanian, the "relativum generale"31 që can also introduce subjunctive complements after several verbal categories, as in ( 27).
(  ' (In Terzi 1992: 17) There are some interesting facts to be considered concerning the examples above: in both Balkan languages the "subjunctive complementisers" are widely used in the case of disjoint subject reference, like in cases the examples manifest.In Romanian the occurrence of ca obligatorily triggers obviation (cf.Terzi 1992: 105f. andTomić 2006: 517), while in Albanian the picture is more complicated, since që introduces subjunctive complements even in cases of subject co-reference (cf.Tomić 2006: 589ff.).32Nevertheless, both languages have in common that ca and që cannot occur after certain control verbs.According to Tomić (2006: 518 for Romanian; ibid.: 588 for Albanian), the Romanian ca has stricter distribution than the Albanian që, since the former is excluded in control environments, while the latter is blocked only after the verbs mund "can/may/possible" and duhet "must/should/need/ought/ be necessary".
What does the Balkan case suggest for emg?Although it is obvious that every language has developed a different frame of usage concerning the "subjunctive complementisers", there are certain similarities.The complementiser plus subjunctive complement construction is favoured in obviation constructions and is basically excluded after particular control predicates.
If this is the case, how can we account for the counterexamples after volitional verbs like the ones we saw in (23-24)?First of all, volitional predicates are not obligatorily control verbs or obviation verbs.33Since the subjects of their complement verbs can exhibit both joint or disjoint reference to the matrix subject they differ both from the intentional and causative verbs we examined before, while they somehow resemble verbs such as the commissive omnío 'I swear' or elpízo 'I hope' (cf.ex.11 and 12 from Late Byzantine Greek.In our emg corpus omnío may select óti + na clauses whereas elpízo plus óti + na is extremely rare34).On the other hand, volitional and intentional verbs share the semantic property of expressing the willingness of the subject.Willingness and ability are parts of the so-called dynamic modality (cf.Palmer 22006: 70 and 76ff.), and while ability forms control structures, willingness can be either directed towards the subject of the verb or another person, and the latter case is then manifested by obviation.Intentional and causative verbs necessarily have complements with disjoint reference.Volitional verbs in turn can select either co-referent or non co-referent complement subject, as already said, so in a syntactic system in which the "complementiser slot" is open for the subjunctive Romanian and Albanian we discuss here.A more systematic comparison between emg and these languages from our side will be the topic of future research.33 One should consider control and obviation as belonging to a continuum, in which absolute control lies on the leftmost edge while non obligatory control on the rightmost one.Deontic modal verbs or verbs expressing competence are placed on the left extreme, volitional verbs that do not demand control, on the right.In-between there are verbs of more or less partial control like the ones meaning "to try" and the like.Cf. Landau 2004 andfor Greek, Roussou 2006: 123. 34 There is only one attestation in the Anthology: 146.In Kartanos the verb elpízo is altogether rare.
Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0complements, the volitional verbs tend to "imitate" the directive ones.If this is the case, óti before na mainly marks disjoint reference and is preferred in directive/ obviation sentences, but given that the valency of the volitional verbs may allow the appearance of a complementiser, this may "pop up" also in cases of co-reference.It should be borne in mind though that the system of emg is in flux and varies with regard to the region or the author, and even within a single text, consistency cannot always be found.If this approach is right, we would expect that also in emg there must be some restriction in the distribution of the óti + na clauses as is the case in Albanian.Indeed, this construction rarely occurs after the control modal verb μπορώ (m)boró 'I can, I am able to' .Still, even in this case we can trace counterexamples within the whole corpus: is in ton pólemon the-war.acc.'… until (nobody) could stand in the war.' (Anthology: 134) Nevertheless, these counterexamples are scarce, the case in (28) after the control verb (m)boró is unique in the Anthology.Similarly, in Kartanos there is only a single case of this control verb selecting an óti + na clauses while in about 160 instances, (m)boró selects a "bare" na-subjunctive.Interestingly, there are no instances of óti + na complements after the aspectual control verb αρχίζω arkhízo 'I begin to' .Thus, in emg, óti was not only a generic complementiser, which is not necessarily bound with assertive complementation, but in its function as an introductory marker of a complement could give rise semantically to directivity and, consequently, syntactically to obviation (although we should underline the fact that in this case syntax is the impact of a semantic function and not vice versa).Further research on this topic is required.

2.2
pos Plus na In our corpus, the distribution of pos + na differs with regard to each author or text.As afore-mentioned, in smg pos is a slightly more colloquial alternative to óti.In Kartanos, pos + na clauses appear in the same contexts as the óti + na constructions do (cf.ex.(5) in 1.2.), although the occurrences of pos + na are far less frequent within Kartanos' work.Similarly, in the Anthology, óti + na outnumbers pos + na (cf. Figure 2).In a single case in the whole corpus, óti introduces a pos + na complement (29).This usage of óti, rare as it may be, is once again suggestive of the function of óti as an indirect/reported speech marker.( 29 In Morezinos, on the contrary, there are no instantiations of óti + na clauses, while pos + na is the most preponderant "pleonastic" constellation.The reason why óti + na is not attested is the overall absence of óti in the work of the Cretan author.This fact could be an indication of a dialectal distinction between the Corfiot and the Cretan texts, namely that pos have replaced óti in the Cretan dialect, while it was already competitive to óti in Western dialects.35Further research on this issue is a desideratum for us and will be presented in the future. The "pleonastic"pos + na construction is not only numerically less common than the óti + na clauses within the rest of the 16th c. corpus, but it is also selected by a more restricted set of predicates, namely by verbs expressing thought or judgement/ belief (verba putandi, cf.30) or verbs like elpízo 'I hope' .This employment of pos + na is illustrated in (31)  '… and he hopes you will come on your own will because he believes that you will also desire him.'(Morezinos,Klini Solom. 286/260v) The semantic link between these two predicate categories is that they are both future-oriented.The usage after elpízo is a continuation of the older "pleonastic" usage of a conjunction along with the mood particle na (cf.ex. ( 12)).In 2.1.2.we remarked that verbs like elpízo can select either a na-subjunctive or a conjunction + future tense clause (elpízo na érthi vs. óti/pos tha érthi).A similar case is the smg verb πιστεύω pistévo 'believe' , which semantically corresponds to the verb kríno found in (31) and can take a na-or an óti/pos + tha comple-ment36 (dhen pistévo na érthi ávrio vs. óti/pos tha érthi ávrio 'I do not believe that (s)he will come tomorrow').
Yet, given the paucity of attestations of pos + na, as well as its distribution, it is obvious that this structure did not develop an association with directivity in Morezinos.

2.3
óti Plus pos From a mg native speaker's point of view, the juxtaposition of the complementisers óti + pos is intuitively the "least acceptable"37 among the three construc-36 Verbs of opinion, such as pistévo, usually select na-complement if negated or in questions, cf.Roussou 2006: 68f. 37 Jannaris (1897: §1755) writes that the combination of the concurrent "declarative particles" are "of no rare occurrence" in popular speech and reports the following sentence: μου εμήνυσε ότι πως ο αδερφός του αρρώστησε 'he sent me word that his brother fell ill' .We remain rather dubious about this claim, firstly because as we said in smg this construction is not acceptable, secondly because Jannaris does not seem to make a distinction between the complementiser πως and the interrogative πώς (in his analysis in polytonic Greek he shifts from interrogative πῶς to complementiser πώς without mentioning any differentiation).Nevertheless, further research in dialectic texts may yield more such cases.In accordance with the selecting predicates, we observe that the óti + pos clauses occur in "narrative" contexts with reference to events to which the speaker or the subject of the predicate of the main clause has 'non-firsthand' knowledge (other terms: 'non-confirmative' , 'indirective ' , and 'mediative' , cf. Aikhenvald 2004: 25).On these grounds, it may be suggested that most óti + pos clauses encode a certain degree of evidentiality.Despite the fact that evidentiality is chiefly associated with a verbal category or it is expressed with particles, there is no reason to exclude the possibility for a language to manifest evidentiality in the domain of a Complementiser Phrase Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0(cp).39As Aikhenvald (ibid.: 69) puts it, "there are hardly any morphological limitations on how evidentiality can be expressed".Syntactically, in turn, it may be rather erroneous to postulate a univerbation of the two conjunctions.Although a univerbation in the head of a cp is possible (compare, e.g., ag διότι 'because; for the reason that' out of δι' ὅ, τι < διὰ ὅτι, i.e. from a configuration of preposition + neutral relative pronoun, cf.Schwyzer & Derbrunner 1950: 661), a univerbation of two complementisers should have been difficult at this early stage of emergence of this structure because a grammaticalisation process within phrases normally takes place when two or more lexical items that have a distinct categorical status form a collocation within a particular context (cf.also Hopper & Traugott 22003: 134f.).Since óti and pos shared the same categorical status when they were used separately, a univerbation at the early stage could not be possible (although it could be possible in a later stage, if the structure was well established in the language).It is then preferable to analyse óti as a sort of a marker analogous to the reported/ indirect speech marker we postulated for the óti + na clauses.
Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0 The "pleonastic" óti in emg does not undertake the function of ὅτι recitativum, since it does not introduce direct speech.Nevertheless, the role of nt ὅτι as a marker of either direct or indirect speech could well have influenced the ecclesiastical scholars since they constantly dealt with the language of the Bible.
Thus, we can put forward the idea that in the collocation óti + pos the first element plays an introductory role while the complement clause is an adjunct in apposition.If this analysis holds, the óti is a constituent of the main clause, quasi an argument, and therefore it is no longer a genuine conjunction, given that its content is less grammatical and more lexical, somehow "returning" to the etymological roots of óti (⟨ ag ὅ τι, a neuter relative pronoun).41And so the clause had the following structure: [x say that (óti): [that2 (pos) .…]].Surely, etymology did not play a crucial role here, but the persistent presence of the ὅτι recitativum in the literal tradition of Greek was giving a stylistic model that had an impact up to the emg.Furthermore, the conjunction pos could convey an evidential reading in this configuration.
Journal of Greek Linguistics 16 (2016) 202-231 c.Though it is difficult to indicate an initial context in which the óti + na and pos + na structures arose, we may hypothesize that in cases in which the meaning of na was not purely modal but also leant towards futurity, the complementiser óti became compatible with na in the same way it can co-occur in smg with the future particle tha.In later phases, óti assumed a more generic syntactic role, being compatible not only with assertions but also with modal expressions.In certain varieties, especially in Western Greek dialects, óti + na is selected extensively by directive verbs, encoding an embedded indirect nonassertive mood (mostly an imperative) and marking the distinction between the matrix and the embedded subject.
d.The marker óti both in the óti + na and the óti + pos conjunctions can be described as a marker introducing indirect speech.This is fairly straightforward after, for example, verbs of saying or verbs of order that select óti + na, or verbs of narration and perception that select óti + pos, but it can also be argued for other categories of predicates, such as volitionals.This role of óti is reminiscent of the ὅτι recitativum of New Testament, the influence of which might have played a role.Nevertheless, in terms of syntax óti must have had a different position and function in these two constructions.In óti + na complements, óti is probably a complementiser, whereas na forms a cluster with the verb as in mg.In coordinating clauses óti should not be repeated, contrary to na which is obligatory before a verb in these clauses.The case of óti + pos is more obscure.Given the scarcity of the examples, it is difficult to postulate a univerbation of the two markers or a duplicate complementiser.Our (tentative) analysis is that semantically óti introduces the indirect speech while pos is somehow associated with evidentiality in this particular constellation.Syntactically, óti may be interpreted as an argument-like element, and the pos-clause is an adjunct in apposition to this óti.That both óti and pos were to a great extent used as complementisers in complementary distribution explains why this construction did not get established in the language.
A number of issues, such as a more precise syntactic analysis of these structures, the dialectic differentiation concerning them on the basis of the available emg texts and a systematic comparison with earlier phases of Greek, as well as with other languages with which emg was in contact, are objects of ongoing research.Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0
Consider the examples below (4-6) from the 16th c.: 250 extracts from all kinds of prose texts mentioned above-unpublished for the most part-are included in the work Δημώδης πεζός λόγος του 16ου αιώνα (Kakoulidi-Panou,Karantzola & Tiktopoulou, in press, c. 500 pages).They stem from various regions of the Greek-speaking world (mainly Crete, Ionian Islands, Constantinople and mainland Greece), providing valuable information about the linguistic situation at that time, with respect to the dialectic and stylistic variety or the overall development of emg.Because of its representativeness, this anthology will be our main body of reference for the study of the distribution of "pleonastic" use of complement markers in emg.In addition, our corpus comprises two more extensive texts: Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0a.The "Παλαιά τε και Νέα Διαθήκη" (Paleá te ke néa Dhiathíki, cf.Kakoulidi-Panou (ed.): 2000) of the Corfiot Ioannikios Kartanos (Venice 1536, c. 145,000 words), the first translation in vernacular Greek of excerpts of the Bible; b.The "Κλίνη Σολομώντος" (Klíni Solomóntos, cf.Kakoulidi-Panou, Karantzola & Chalvatzidaki, (eds.).2007) of the Cretan Ioannis Morezinos (1598, c. 82,000 words), which records sixty miracles of the Virgin Mary, in Cretan dialect (ms Xiropotmou, 1602).
Markopoulos 2009) να τον σώσω ζωντανόν … pistévis believe.2sg.pres.ind.Do you believe I will make it to see him alive?' (Kartanos, Testament 279v)Along with these cases, in our emg corpus the óti + na complement clauses are selected by the following sets of predicates: -occur, while tha does appear after an óti-pos or a pu complementiser (contrary to na).In Late Medieval and Early Modern Greek tha is not yet attested (cf.Markopoulos 2009).
http://creativecommons.org/licenses/by-nc/4.0ungrammatical in smg but obviously they may be acceptable under certain contextual conditions (e.g. after verba dicendi but not after directives).More research on this topic is needed.Downloaded from Brill.com 07/14/2024 10:33:59PM via Open Access.This is an open access article distributed under the terms of the prevailing CC-BY-NC license at the time of publication.http://creativecommons.org/licenses/by-nc/4.0