Soltic ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 2012 Jorie Soltic and Mark Janse From Enclisis to Proclisis in Medieval Greek: σὲ λέγω and its Uses in the Chronicle of Morea Jorie Soltic and Mark Janse N RECENT YEARS linguists have shown an enormous inter- est in clitics, small words which have no accent of their own and consequently have to ‘lean’—κλίνω in Ancient Greek —on another word, a phonological ‘host’.1 If this phonological host is the following word, we call it a proclitic; if it selects a preceding word as host, it is an enclitic. In this paper we focus on (the phonological hosts of ) the object clitic pronouns (OCPs) in Greek, viz. clitics whose grammatical function is the (in)direct object and whose syntactic host is the finite verb. In Ancient Greek, the OCPs have an enclitic nature, whereas Modern Greek OCPs are proclitic. Thus, the phonological nature of the OCPs must have al- tered in the period in between. Indeed, in the Medieval period a certain reanalysis has occurred: [X ← enclitic OCP + finite verb] becomes [X + proclitic OCP → finite verb].2 Horrocks, one of the pioneers in the study of Medieval Greek OCPs, has made an attempt to fill in this abstract scheme:3 he holds that the particle νά plays a major role in the transition from enclisis to proclisis (section 1 below). 1 A. Zwicky, On Clitics (Bloomington 1977) 9. 2 Cf. M. Janse, “La position des pronoms personnels enclitiques en grec néo-testamentaire à la lumière des dialectes néo-helléniques,” in C. Brixhe (ed.), La koiné grecque antique I (Nancy 1993) 83–121, at 110. 3 G. C. Horrocks, “Clitics in Greek: A Diachronic Review,” in M. Roussou and S. Panteli (eds.), Greek outside Greece II (Athens 1990) 35–52, at 49–50. I JORIE SOLTIC AND MARK JANSE 241 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 We discuss here another—more concrete—environment in which the phonological switch could have originated, which does not, however, necessarily exclude Horrocks’ proposal. Our hypothesis is based on sentences in which an emphasized constituent is followed by the unit σὲ λέγω or by similar con- structions in which a second-person OCP is combined with a first-person of a verbum dicendi (section 1.4). We assume, first, that the combination σὲ λέγω as a whole has become enclitic and that afterwards a reanalysis has taken place. Speakers start supposing that σέ leans on the following word, the verb λέγω, instead of on the preceding one, since the following word con- stitutes its natural syntactic host. The phonological dependence of the OCP now perfectly coincides with its syntactic depen- dence (section 3). The immediate impetus for our hypothesis is the abundance of this construction in the fourteenth-century Chronicle of Morea. The frequent association of this so-called vernacular text with an oral tradition confirms the appropriateness of our corpus for an explanation of a change which naturally has its origins in the spoken language (section 2). 1. Enclisis vs. Proclisis 1.1 From Ancient Greek enclitics to Modern Greek proclitics Since clitics are phonologically deficient words and thus have no accent of their own, they must lean on another word. For this dependence on the position of the phonological host, we speak of proclitics: phonological host follows, vs. enclitics: phonological host precedes.4 Both types are exemplified in τὸ παράδειγµά µου.5 With regard to the clitics which function as 4 We leave aside the class of endoclitics (clitics within a word), as it is not relevant here and even its existence is disputed: S. R. Anderson, Aspects of the Theory of Clitics (Oxford 2005) 165. 5 See Μ. Janse, “Phonological Aspects of Clisis in Ancient and Modern Greek,” Glotta 73 (1995–96) 155–167, for the extra enclitic accent words can receive. 242 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 (in)direct object of a finite verb,6 the direction of clisis changed during the history of Greek: the Ancient Greek enclitic OCPs have turned into proclitic OCPs in Modern Greek.7 In Ancient Greek, in accordance with the Indo-European ‘Law’ put forward by Wackernagel in 1892, the OCPs are placed in second position in the clause, regardless of the po- sition of their syntactic host, the verb:8 τοίου µιν θάρσευς πλῆσε φρένας ἀµφὶ µελαίνας (Il.17.573) καὶ δῶρα ταῦτά τοι διδοῖ τοῖσι καὶ αὐτὸς µάλιστα ἥδεται χρεώµενος (Hdt. 3.21.5) As these examples illustrate, second position cannot be con- sidered an absolute notion and is thus not an sich important. Therefore, it is revealing to reformulate the Law from another perspective: OCPs follow after the first word or constituent in the clause by which the OCPs are attracted. These initial ele- ments often belong to the same types of words: function words such as conjunctions, interrogatives, and negations, but also emphasized words.9 Here, a demonstrative (respectively τοίου and ταῦτα) attracts µιν and τοι. Both OCPs form a phono- logical unit with their preceding constituent and are thus of an enclitic nature. In later Greek, the OCPs remain enclitic. In this connection, Janse makes a convincing case by discussing various kinds of assimilations found in the papyri, the language of which is close 6 This is a very important addition, since the Modern Greek OCPs are still enclitic in combination with imperatives and gerunds (cf. I. Philippaki- Warburton, “Verb Movement and Clitics in Modern Greek,” in Themes in Greek Linguistics I [Amsterdam 1994] 53–60). Henceforth we use the term ‘verb’ to refer to finite verbs. 7 Cf. Janse, Glotta 73 (1995–96) 155–167. 8 Cf. Janse, in La koiné grecque 83–121; D. M. Goldstein, Wackernagel’s Law in Fifth-Century Greek (diss. Berkeley 2010). 9 Janse, in La koiné grecque 83–121. The syntactic rule and pragmatic principle identified for Medieval Greek actually constitute a logical step in the history of the Greek OCP: preverbal OCP position is still triggered by the same categories of initial words (cf. section 1.2). JORIE SOLTIC AND MARK JANSE 243 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 to the spoken language, for instance ἐάµ µοι.10 This example clearly illustrates the enclitic nature of the OCP, since µoι assimilates with the preceding word, to which µoι does not belong syntactically. However, the strongest evidence for the enclitic nature of the ancient OCPs is the fact that they never open an utterance, which proves their need of a preceding host. In Modern Greek,11 on the contrary, OCPs do occur in initial position; one example should suffice: µας µιλά. The OCP µας cannot lean to another constituent than its verb µιλά, which follows, and it is thus proclitic. In longer sentences as well, Modern Greek OCPs always immediately precede their verb and procliticize to it. These preverbal proclitics can be considered as the outcome of a natural evolution, since “the phonological dependence … then coincides with their syntactic dependence.”12 The immediate cause of this modern harmonization between phonological and syntactic host is found in the Medieval per- iod, when the adjacency of OCP and verb becomes obligatory: “the clitic object pronoun ceased to be a freely moving part of the clause and instead became part of the verb phrase.”13 In 10 E.g. P.Petr.2 3.21; Janse, Glotta 73 (1995–96) 160. 11 More precisely: in Standard Modern Greek, for some modern Greek dialects show a different OCP distribution, for example Pontic and Cap- padocian. See M. Janse, “Cappadocian Clitics and the Syntax-Morphology Interface,” in B. D. Joseph et al. (eds.), Themes in Greek Linguistics II (Amster- dam 1998) 257–281; “Object Position in Cappadocian and Other Asia Minor Greek Dialects,” in M. Janse et al. (eds.), Proceedings of the Second In- ternational Conference on Modern Greek Dialects and Linguistic Theory (Patras 2006) 115–129. 12 M. Janse, “Clitic Doubling from Ancient to Asia Minor Greek,” in D. Kallulli and L. Tasmowski (eds.), Clitic Doubling in the Balkan Languages (Am- sterdam 2008) 165–202, at 176. 13 P. Mackridge, “An Editorial Problem in Medieval Greek Texts. The Position of the Object Clitic Pronoun in the Escorial Digenes Akrites,” in N. Panayiotakis (ed.), Origini della literatura neogreca I (Venice 1993) 325–342, at 339. 244 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 contrast to the Ancient Greek OCPs, the Medieval Greek OCPs now always appear next to their syntactic host, the verb. Nevertheless, since the position of the OCPs vis-à-vis the verb is not (yet) as fixed in Medieval as in Modern Greek (always preverbal), but varies between pre- and postverbal, we can ask whether the (preverbal) OCPs are already proclitic, as in Modern Greek, or whether they are still enclitic, as in An- cient Greek. This question is not an easy one: “Whether or not the Medieval Greek pronouns were still enclitic, as in Ancient Greek, or had become proclitic … is a moot question.”14 Pappas presents an intelligent solution: he proposes the Medieval Greek OCPs to be in se clitic: “Late Medieval Greek weak pronouns are always phonologically attached to the verb, either as enclitics or proclitics.”15 Revithiadou and Spyropulos agree: “In this respect, we are in total agreement with Pappas … that in the language of the texts of the 12th century and beyond pronominal clitics can be either proclitics or enclitics, depending on the structure.”16 Condoravdi and Kiparsky, on the other hand, are convinced that the Medieval OCPs still have a constant enclitic nature.17 We also believe the OCPs are enclitic as long as they do not occur in initial position. However, a transition to proclisis must have taken place at a certain moment in the Medieval period. Before discussing Hor- rocks’ hypothesis concerning the concrete context responsible 14 Janse, in Clitic Doubling 181; cf. Glotta 73 (1995–96) 165; P. Mackridge, “The Position of the Weak Object Pronoun in Medieval and Modern Greek,” Jazyk i rečevaja dejatel’nost’ 3.1 (2000) 133–151, at 137. 15 P. Pappas, Variation and Morphosyntactic Change in Greek. From Clitics to Affixes (Basingstoke 2004) 13. 16 A. Revithiadou and V. Spyropoulos, A Typology of Greek Clitics with Special Reference to their Diachronic Environment (MS. University of the Aegean 2006) 30. Cf. Janse, Glotta 73 (1995–96) 155–167. 17 C. Condoravdi and P. Kiparsky, “Clitics and Clause Structure: The Late Medieval Greek System,” Journal of Greek Linguistics 5 (2004) 159–183, at 172 ff. JORIE SOLTIC AND MARK JANSE 245 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 for this change, it is necessary to outline the Medieval Greek OCP distribution rules. 1.2 Medieval Greek distribution rules As mentioned, the position of Medieval Greek OCPs with regard to their verb still varies. Postverbal position is the un- marked order, whereas preverbal position is triggered in cer- tain environments. The exact position of the OCP depends on a syntactic rule and a pragmatic principle, or in the words of the pioneer Mackridge: “the rules are primarily a matter of syntactic context and secondarily a matter of pragmatics (in this case, emphasis).”18 Indeed, preverbal OCPs occur if the verb is preceded by an emphasized constituent. Now, some words are “emphatiques de nature”19 and are thus often as- sociated with preverbal OCPs. Dover20 reckons for example emphatic personal pronouns such as ἐγώ and demonstratives like τοῦτο among these so-called “preferential words” which at- tract the OCPs into preverbal position (cf. Homeric examples), thus: κι ἂν οὕτως οὐδὲν ποίσωµεν, ὡσὰν ἐγὼ σᾶς λέγω (Chronicle of Morea P 4737, bis) ποῦ ἔνι ἄνω εἰς τὴν θάλασσα, τοῦτο σὲ λέγω, ἀφέντη (P 1666) By extension this pragmatic principle applies to ad hoc em- phasized constituents as well, for instance: 18 Mackridge, in Origini 326. For more details on the distribution of Medieval Greek OCPs see A. Rollo, “L’uso dell’enclisi nel greco volgare dal XII al XVII secolo e la legge Tobler-Mussafia,” Iταλοελληνικά 2 (1989) 135–146; Mackridge, in Origini, and “On the Placement of the Weak Per- sonal Pronoun in Medieval Greek Vernacular,” Studies in Greek Linguistics 15 (1995) 906–929; Jazyk i rec ̌evaja dejatel´nost´ 3.1 (2000) 133–151; Condoravdi and Kiparsky, Journal of Greek Linguistics 5 (2004) 159–183; Revithiadou and Spyropulos, Typology; C. A. Thoma, “Distribution and Function of Clitic Object Pronouns in Popular 16th–18th Century Greek Narratives,” in J. Rehbein et al. (eds.), Connectivity in Grammar and Discourse (Amsterdam 2007) 139–163; Janse, in Clitic Doubling 165–202. 19 Janse, in La koiné grecque 94. 20 K. J. Dover, Greek Word Order (Cambridge 1960) 20–21. 246 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 κ’ ἐµὲν ὡσαύτως µετ’ αὐτόν, πληροφορίαν σὲ λέγω, νὰ σὲ πληροφορήσωµεν, νὰ ἔνι τὸ θέληµά σου (P 193–194) It is reasonable to argue that the direct object πληροφορίαν (“assurance”) is emphasized, being an important word in this context, for it is repeated by the verb πληροφορέω (“give as- surance”) in the next verse. This verse is also illustrative with regard to the syntactic rule active in Medieval Greek. If a function word (a subordinating conjunction or a relative) precedes the verb, preverbal position is the norm. Actually, νά, which attracts the OCP σε into pre- verbal position, is not a true subordinating conjunction, but since it is etymologically derived from one ( ἵνα), it has to be treated according to this rule as well.21 1.3 Horrocks’ hypothesis The evolution from ἵνα towards νά/να constitutes the basis for Horrocks’ hypothesis: “The history of ἵνα is vital in this connection.” More precisely, the phonological evolution from ἵνα as a phonologically independent subordinating conjunction to a preverbal proclitic particle να is relevant:22 As long as να … retains sufficient phonological independence to host a following clitic, the naturally enclitic status of the pro- nouns … can remain. When, however, the particle comes effec- tively to cliticise to the right,23 there are only two possibilities for the associated clitic pronouns. Either they remain enclitic and therefore cease to appear in this position altogether, or they themselves become proclitic, forming a clitic group with the pre- ceding particle, and attach to the following verb. Quite obviously, the Greek OCPs have chosen the latter op- 21 R. Browning, Medieval and Modern Greek (Cambridge 1999) 43. 22 Horrocks, in Greek outside Greece 49, 50. 23 In later work, we find a more detailed description of this phonological history: ἵνα → ἱνά → νά → να, since the loss of the initial vowel ι (aphaeresis) must have been preceded by a shift of the accent to the final syllable α: G. C. Horrocks, Greek: A History of the Language and its Speakers (Lon- don 1997) 208; cf. A. N. Jannaris, An Historical Greek Grammar (London 1897) 418. JORIE SOLTIC AND MARK JANSE 247 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 tion. Eventually, the “reinterpretation”24 of clisis direction is— by analogy—extended to other modal particles such as θα (derived from θέλω να) and ἄς, as well as to the negatives. Later on, the OCPs also appear before the verb even in the absence of these preverbal particles. Finally, indicatives start to follow the model of the subjunctives and thus preverbal pro- clitic OCPs become generalized.25 1.4 Our hypothesis: σὲ λέγω However, we have found another good candidate responsible for the enclisis-to-proclisis-change. The construction under dis- cussion consists of a second-person OCP and a first-person of a verbum dicendi: σὲ λέγω, σὲ λαλῶ, and their plurals σᾶς λέγω, σᾶς λαλῶ. It should be noted that in Ancient Greek the pronoun of the second person singular is ambiguous: it could be the weak, clitic form as well as the strong, orthotonic form; a distinction is made by attributing to the emphatic form an accent: σέ.26 In the Medieval period, on the other hand, the accent on the pro- noun tells us nothing about its clitic or non-clitic nature, for it is a convention to accentuate preverbal OCPs (examples in 1.2). On the other hand, the medieval Greeks developed a whole range of unambiguous longer forms for the orthotonic pronoun of the second person singular: ἐσέ, ἐσέν, ἐσένα, ἐσέναν, σένα, σέναν. So, if the narrator had wanted to use the emphatic form, he could presumably have chosen one of these unam- biguous forms. Moreover, the context does not favour an em- phatic reading of σέ. On the contrary, σέ is never semantically emphasized, as we shall see in the examples in 3.2. Finally, in some examples from our corpus in which σε follows λέγω, it is unaccented, for example: 24 Horrocks, Greek 211. 25 Horrocks, in Greek outside Greece 51; Greek 211. 26 M. Janse, “Alexandrian Grammatical Theory in Practice. Apollonius Dyscolus and the Personal Pronouns,” in P. Swiggers and A. Wouters (eds.), Grammatical Theory and Philosophy of Language in Antiquity (Louvain 2002) 233–255. 248 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 ὅπου ἦτον δὲ αὐτάδελφη ἐκείνου τοῦ Δ∆εσπότου, τοῦ Νικηφόρου, λέγω σε, ἀφέντου δὲ τῆς Ἄρτας (Chronicle of Morea P 8064–8065) To summarize, we can conclude that σέ has a truly clitic nature in the construction under discussion. 2. Corpus 2.1 Different versions It is time to provide some information on our corpus, the fourteenth-century Chronicle of Morea. This historiographical work describes events in mainland Greece after the settlement of the Franks following the Fourth Crusade (1204). The Chronicle survives in versions in four different languages: Italian, Aragonese, French, and Greek.27 The debate over which ver- sion is closest to the original has centred round the French and Greek versions.28 However, in what follows, we are interested only in the Greek version. 2.2 Greek manuscripts The Greek version of the Chronicle is the only one not written in prose, but composed in the πολιτικὸς στίχος, the typical metre for medieval vernacular Greek texts.29 It is preserved in five manuscripts, of which we take only the two most important into account: Havniensis Fabricius 57 (H) and Paris.gr. 2898 (P).30 27 T. Shawcross, “Oral Residue and Narrative Structure in the Chronicle of Morea,” Byzantion 75 (2005) 310–333; The Chronicle of Morea: Historiography in Crusader Greece (Oxford 2009). 28 J. Schmitt, “Zur Ueberlieferung der Chronik von Morea,” in Festschrift Konrad Hoffman (Erlangen 1889) 519–538, and The Chronicle of Morea: A His- tory in Political Verse (Groningen 1904); N. A. Bees, “Zur Chronik von Morea,” Archiv für Kulturgeschichte 13 (1917) 122–124; H. E. Lurier, Crusaders as Conquerors: The Chronicle of Morea (New York 1964); P. Topping, Speculum 40 (1965) 737–742 (rev. of Lurier); M. Jeffreys, “The Chronicle of the Morea: Priority of the Greek Version,” BZ 68 (1975) 304–350, at 304; Shawcross, Chronicle. 29 M. Jeffreys, “The Nature and Origins of the Political Verse,” DOP 28 (1974) 141–195. 30 Shawcross, Chronicle 35. JORIE SOLTIC AND MARK JANSE 249 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 The latter contains 8191 verses, whereas the former counts as many as 9219 πολιτικoὶ στίχοι, even though its beginning is missing. In Schmitt’s edition,31 this gap and other lacunas in H are supplemented by Taurinensis B.II.I (T). Schmitt has under- taken the enormous task of making a parallel edition of H(+T ) and P. Indeed, both manuscripts tell more or less the same story, but there are some important ideological and linguistic differences. H is the older of the two, written in the late four- teenth century, and seems to reflect a somewhat anti-Greek attitude.32 P, copied much later, filters out the pro-Frankish passages that were “found distasteful” or rewrites them from a more Greek perspective.33 In combination with its “badly written Greek,” H is, there- fore, often said to be written by a non-native speaker.34 Con- sequently, one could ascribe the construction σὲ λέγω to (bad) learner’s Greek and thus easily wave aside the hypothesis we shall put forward. However, this possible criticism can be re- jected, since the construction is found—just as frequently—in P and “it is evident that P was written by a Greek.”35 Moreover, P was “probably not directly or indirectly copied from H.”36 In addition, Jeffreys, an authority in the field, is convinced that the author of H was a native Greek as well. 2.3 Oral residues It is this same scholar who has established a clear connection between the Chronicle of Morea and an oral tradition by making an elaborated study of the formulas. Jeffreys has found that the level of formulas in H ranges from 21.2% to 53.5% per hun- 31 This edition can be found on-line on the TLG. 32 Jeffreys, BZ 68 (1975) 305–306; Shawcross, Chronicle 263. 33 Shawcross, Chronicle 264. 34 More specifically, by a Graecised Frank or a so-called ‘Gasmule’, the offspring of a Greco-Frankish marriage: Schmitt, Chronicle xxxviii. 35 Schmitt, Chronicle xxix–xxx. 36 Jeffreys, BZ 68 (1975) 350. 250 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 dred lines, which surpasses all other Greek vernacular works.37 The closeness of the Chronicle to the oral tradition is a very im- portant observation for our purpose, as it is self-evident that the switch to proclisis must have originated in the spoken language, rather than in an artificially constructed written language. Shawcross confirms these oral residues in the Greek version: “Everything about the Greek version suggests that it is a text which has been highly influenced by methods of composition derived from the pragmatic concerns of oral performance and reception.”38 More specifically, she explicitly links the com- bination σὲ λέγω with the bond that exists between a storyteller and his audience during an oral performance:39 Both the second and first grammatical persons are ubiquitous in the Greek Chronicle, with the former appearing on average once every nineteen lines and the latter every twenty lines. Indeed, a constant urge is displayed by H to bring narrator and narratees into each other’s mental presence (e.g. ‘σὲ λαλῶ’, v.381; ‘σὲ λέγω’, v.1651; ‘εἶπα σε’, v.3178; ‘σᾶς ἀφηγοῦµαι’, v.446). This betrays an uneasiness about the relationship between the processes of composition and reception. Where an oral linguistic exchange consists of the production of an utterance and its hear- ing within the same spatio-temporal context, this simultaneity, attributable to the corporeal interaction of the interlocutors, is lost with the written word. An attempt is made by H to compen- sate for the loss by simulating, within the parameters of the text itself, the establishment and maintenance of a bond typical of orality. 3. Explanation and examples 3.1 Frequency However, the fact that σὲ λέγω does not belong exclusively to the written language, but on the contrary is characteristic of the spoken language, is of course a necessary but not a sufficent 37 M. J. Jeffreys, “Formulas in the Chronicle of the Morea,” DOP 27 (1973) 163–195, at 188. 38 Shawcross, Chronicle 181. 39 Shawcross, Chronicle 157, cf. 263. JORIE SOLTIC AND MARK JANSE 251 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 condition for the claim that this construction is the instigator of the clisis-alteration. Frequency seems an important factor as well (cf. the νά-construction of Horrocks). We believe that σὲ λέγω can also satisfy this requirement, since its meaning, “I tell (you),” lends itself to regular use in spoken discourse. More- over, it has been acknowledged that the importance of minimal utterances, consisting only of a verb and an (in)direct object, can hardly be overestimated with regard to linguistic change.40 This table gives the number of occurrences in H and P: σὲ/σᾶς λέγω σὲ/σᾶς λαλῶ H 7 30 P 3 34 3.2 Distribution More revealing than these pure statistics is the distribution of the construction: preceding constituent σὲ/σᾶς λέγω σὲ/σᾶς λαλῶ H proper names 4 20 official terms of address 3 6 kin terms – 3 other – 1 P proper names 2 17 official terms of address 1 8 kin terms – 5 other – 4 The most popular position for σέ/σᾶς λέγω/λαλῶ is immedi- ately after a proper name:41 40 Cf. Janse, in Clitic Doubling 177. 41 We mark the standard caesura after the eighth syllable with the sign # (M. D. Lauxtermann, The Spring of Rhythm: An Essay on the Political Verse and 252 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 ἐνῷ ἦτον τότε βασιλεὺς # τῆς Ρωµανίας, σὲ λέγω. (H 61) κ’ οἱ Ἀλλαµάνοι, σὲ λαλῶ, # κ’ ἐσφάζονταν ἀλλήλως· (P 4076) Note that the choice between the verbs λέγω and λαλῶ is de- termined by the metrical structure of the verse: the πολιτικὸς στίχος, consisting of two parts of respectively eight and seven syllables, does not allow accents on the odd-numbered syllables (see Lauxtermann). Consequently, λέγω never ends the first part of the verse (otherwise accent would be on the seventh syllable), whereas λαλῶ is avoided at the end of the verse (otherwise accent would be on the fifteenth syllable). In both structures, the adjective ἐκεῖνος often occurs: τὸν Μπονιφάτσον, σὲ λαλῶ, # ἐκεῖνον τὸν µαρκέσην· (H 381, cf. 208 for the preceding occurrence of the name) ἐκεῖνος ὁ παµφρόνιµος # ὁ Μπαντουής, σὲ λέγω. (H 384, cf. 190) ἐκεῖνον τὸν πανάπιστον # τὸν Μούρτζουφλον σὲ λέγω (H 878, cf. 871) ἐκεῖνος γὰρ ὁ µισὶρ Ντζᾶς, # ὁ Καταβᾶς, σὲ λέγω. (H 4775, cf. 4713) This demonstrative adjective points to the ‘semi-active’ status of the referents:42 all these persons have been introduced in the thread of the story, but normally do not occur in the immediate context. In such a situation, we can appropriately translate σὲ λέγω with “I mean.”43 Chafe defines semi-active information as information which is mentioned a while before or derivable from the context.44 In these examples, the proper names are again brought to attention, ‘activated’ in Chafe’s terminology, and thus emphasized. In the verses in which the construction closes the first part of the πολιτικὸς στίχος (thus the examples ___ Other Byzantine Metres [Vienna 1999]). 42 Cf. W. L. Chafe, “Prosodic and Functional Units of Language,” in J. A. Edwards and M. D. Lampert (eds.), Talking Data. Transcription and Coding in Discourse Research (Hillsdale 1993) 33–43. 43 Cf. D. Schiffrin, Discourse Markers (London 1987). 44 W. L. Chafe, Discourse, Consciousness and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing (Chicago 1994) 30. JORIE SOLTIC AND MARK JANSE 253 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 with λαλῶ), this is even more clear, since the second part of the verse almost always specifies the reintroduced referent, usually in the form of an apposition: τοῦ Καλοϊωάννη, σὲ λαλῶ, # ἐκεινοῦ τοῦ δεσπότη. (H 1102, cf. 1087) ὥρισεν καὶ ἐκράξασιν # κὺρ Μιχαὴλ ἐκεῖνον, τὸν Παλαιολόγον, σὲ λαλῶ, # τὸν πρῶτον τῆς Ρωµανίας (H 1227–1228)45 στὸν κὺρ Μιχάλην, σὲ λαλῶ, # τὸν µέγαν Παλαιολόγον. (H 3103, cf. 1265) κι ἄρχισεν µάχην δυνατὴν # µετὰ τὸν ἀδελφόν του, κὺρ Νικηφόρον, σὲ λαλῶ, # αὐτεῖνον τὸν Δ∆εσπότην. (P 3099– 3100, cf. 3086) However, a relative clause can also add extra information: δεύτερον πάλιν ἀπ’ αὐτὸν # τὸν Μπαντουὴν ἐκεῖνον, τὸν κόντον Φλάντρας, σὲ λαλῶ, # ὁποὖτον πρῶτος πάντων· (P 421–422, cf. 402) στὸν Καµπανέσην, σὲ λαλῶ, # ἐκεῖ ὅπου ἦτον στὸ Ἄργος· (H 1540, cf. 1519) Official terms of address constitute a second category which is often found in front of σὲ λέγω. Here as well, the narrator again focuses on a referent which has already been mentioned, often to clarify it: Ὡς δὲ ὁ Δ∆εσπότης, σὲ λαλῶ, # ἐκεῖνος τῆς Ἑλλάδας (H 3815, cf. 3810) ὁ µητροπολίτης, σὲ λαλῶ, # ποῦ εἰς τὴν Ἀνάπολη ἦτον, ἐκεῖνος τοὺς εὐλόγησεν # κατὰ τὴν συνηθείαν. (P 8599–8600, cf. 7866) In the following example, only two verses have passed since the last mention of the term, which might thus seem active in- formation, yet two other proper names (τὸ Ἀνάπλιν καὶ τὸ 45 Both proper names Μιχαήλ and τὸν Παλαιολόγος are found here for the first time. However, considering the lacunal nature of the manuscript, the loss of the verses in which they were originally introduced cannot be ex- cluded. 254 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 Ἄργος τε) have been introduced in verse 2880. Thus, the need for a repetition of ὁ Μέγας Κύρης is justified: τὴν χάριν, τὴν ἐχάρισε # τότε τὸν Μέγαν Κύρην, τὸ Ἀνάπλιν καὶ τὸ Ἄργος τε, # ὁµοῦ <τὰ> δύο κάστρη, ἦτον διὰ τὴν συνδροµὴν # ὅπου ἔποικεν ἐτότες ὁ Μέγας Κύρης, σὲ λαλῶ, # εἰς τὸν πιασµὸν Κορίνθου (P 2878– 2881) In the following examples as well, the term even occurs on the preceding verse, but it is now put into relief: τρεῖς χρόνους γὰρ ἐκράτησεν # ὁ πρίγκιπας τοὺς τόπους τοῦ πριγκιπάτου, σὲ λαλῶ, # ὁλῶν τῶν ἐκκλησίων (H 2654– 2655) Ἠφέρασιν τοῦ βασιλέως # τὸ στέµµα καὶ τὸν σάκκον, ἐστέψασιν κ’ ἐντύσαν τον # ὡς βασιλέαν, σὲ λέγω (H 984–985) One more example of the construction after an official term: βάλετε ἕναν ἀπὸ ἐσᾶς # νὰ διάβῃ ἐκ τὸ φουσσᾶτο τοῦ Δ∆εσποτάτου, σᾶς λαλῶ, # προφώνεσιν νὰ ποιήσῃ (H 3916– 3917, cf. 3905) Kin terms also are followed by σὲ λέγω. Again, extra infor- mation on the referent is often given in the form of appositions: στὸν ἀδελφόν του, σὲ λαλῶ, # τὸν ρῆγαν τῆς Φραγκίας, ἐκεῖ ὅπου ἐπαραδιάβαζεν # µετὰ τὸν σύγαβρόν του (H 6075– 6076, cf. 6006) νὰ εὐλογηθοῦσιν τὰ παιδία, # νὰ ἐπάρῃ ὁ υἱὸς τοῦ ρῆγα τὴν θυγατέρα, σὲ λαλῶ, # τοῦ πρίγκιπα Γυλιάµου (H 6406– 6407, cf. 6403) τὸν πόλεµον τὸν ἔποικεν # µετὰ τὸν Κουραδῖνον, τὸν ἀνηψίον, σὲ λαλῶ, # βασιλέως Φερδερίγου ὁµοίως καὶ ἐξάδελφος # τοῦ ρόϊ Μαφρὲ ἐκείνου. (P 6774–6776) µετὰ συµβίβασιν καλὴν # ὁ κόντος εὐλογήθην τὴν συγάµπρισσάν του, σὲ λαλῶ, # τοῦ δοῦκα τὴν γυναῖκαν. (P 8027–8028) These three categories account for the majority of examples and all have in common that they reintroduce a concept which usually is further specified. Consequently, all these preposed constituents are subject to a certain degree of emphasis. As mentioned in section 1.2, emphasized constituents regularly at- tract OCPs into preverbal position according to the pragmatic JORIE SOLTIC AND MARK JANSE 255 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 principle active in Medieval Greek. The few other examples do not necessarily contradict this explanation, for instance: εἰς τὸν ἐχτρόν του, σὲ λαλῶ, # αὐτὸν τὸν Μέγαν Κύρην (P 3357) Here τὸν ἐχτρόν (“the enemy”) can be considered as an— unofficial—title, which is elaborated in the second part of the verse. In the following example, the time adverbial ἐτότες, which is also made more concrete, is emphasized: λοιπὸν ἐτότες, σὲ λαλῶ, # εἰς τὸν καιρὸν ἐκεῖνον (P 1030) Our hypothesis, then, is that the construction σὲ λέγω is first attributed a sort of prosodic unity and that it then receives—in its totality—an enclitic character.46 In this connection, it is useful to look at editorial practice: Schmitt always puts the con- struction between commas, which suggests that he perceives it as a sort of unit. Furthermore, verba dicendi are often said to con- stitute a unit on their own.47 Another good point of comparison may be the discourse markers ‘you know’ and ‘I mean’, as discussed by Schiffrin.48 These short English expressions attract the attention of the listener in spoken discourse. As mentioned in 2.3, the same communicative function has been identified for σὲ λέγω by Shawcross (Chronicle 181): “frequent interventions in the narra- tive involving the first and second grammatical persons, to insist upon the delineation of a fictional communicative situ- ation.” Moreover, the precise lexical meaning of the verb ‘to know’ is no (longer) relevant. The same in fact applies to σὲ λέγω, which we could appropriately translate ‘I mean’. There- 46 Arguably, this prosodic unit can be equated with the intonation unit (cf. Chafe, in Talking Data 33–43). 47 Cf. M. H. B. Marshall, Verbs, Nouns and Postpositives in Attic Prose (Edin- burgh 1987) 26 ff. (ἔφη); M. Janse, “De la Loi de Wackernagel et ses extensions en Latin,” TEMA: Techniques et méthodologies modernes appliquées à l’Antiquité 1 (1994) 114 (inquit, inquam, aiebat). 48 Schiffrin, Discourse Markers; however, B. Fraser, “An Approach to Dis- course Markers,” Journal of Pragmatics 14 (1990) 383–395, at 392, no longer considers these expressions DMs, but labels them “parallel pragmatic markers.” 256 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 fore, we can rightly consider σὲ λέγω as an (enclitic) unit. Consequently, {σὲ λέγω} can be—as a unit—attracted by the preceding emphasized constituent (E), in accordance with the pragmatic principle operative in Medieval Greek (cf. 1.2). Schematically, this becomes: [E ← {σέ + λέγω}]. Speakers must have reanalyzed this construction into the more natural structure [E + σέ → λέγω], where “the phonological depen- dence … is in perfect harmony with the syntactic depen- dence.”49 The occurrence of σὲ λέγω after an emphasized verb/clause has probably played a principal role with regard to this evolution, for instance: κ’ εἰ µὲν τὸ ποιήσεις, σὲ λαλῶ, # ἀπάνω µου τὸ παίρνω (P 6333) In such cases, it is far more natural to assume that σέ pro- cliticizes to the verb λαλῶ, to which it also syntactically belongs (as indirect object),50 than that it encliticizes to the verb ποιή- σεις. In a second stage, [σέ → λέγω] must have become possible even if no preceding constituent was present. One such example is attested in the Chronicle of Morea: ἀλλὰ διὰ συντοµώτερον # καὶ διὰ κοντοὺς τοὺς λόγους, σὲ λέγω καὶ πληροφορῶ, # µὲ ἀλήθειαν σὲ τὸ γράφω (H 1094– 1095)51 By analogy, this pattern must have extended to OCPs other than the second person and to other verbs. Chila-Marko- poulou52 draws attention to two examples in the Chronicle itself: 49 Janse, Glotta 73 (1995–96) 159. 50 In Medieval Greek the indirect object is no longer expressed by the dative, which has fallen out of use, but by the genitive or accusative: T. Lendari and I. Manolessou, “Η εκφορά του έµµεσου αντικειµένου στα µεσαιωνικά ελληνικά. Γλωσσολογικά και εκδοτικά προβλήµατα,” Studies in Greek Linguistics 23 (2003) 394–405. 51 Approximately the same verse is found in P: σὲ λέγω καὶ πληροφορῶ, # ἀλήθεια σὲ τὸ γράφω (P 1095). 52 D. Chila-Markopoulou, review of Pappas, Variation: Journal of Greek Linguistics 5 (2004) 210 n.6. JORIE SOLTIC AND MARK JANSE 257 ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 σὲ θέλει ἐβγάλει ἐκ τὸν Μορέαν, # ὅπου οὐδὲν ἔχεις δίκαιον (H 4125)53 µὲ ἤφερεν ἡ ὄρεξις # κ’ ἐπίασα τὸ βιβλίον (H 7638)54 Ultimately, the scheme [OCP → verb], is generalized: the for- mer enclitics have become true proclitics. 4. Conclusion Since OCPs which function as (in)direct objects of a finite verb are enclitic in Ancient Greek, but proclitic in Modern Greek, a change from enclisis to proclisis must have occurred in the period in between: the Medieval period (cf. 1.1). Hor- rocks relates this transition to the phonological development of ἵνα into να (cf. 1.3). In this paper, we have discussed another environment in which the change could have originated: [emphasized con- stituent + OCP σέ/σᾶς + verbum dicendi λέγω/λαλῶ] (cf. 1.4). We take this to be a plausible context for a phonological change, since σὲ λέγω—as an oral residue—clearly belongs to the spoken language of that period (2.3). Moreover, we can imagine that an utterance with the meaning ‘I tell (you)/I mean’ was very frequently used by fourteenth-century speakers, as is confirmed by the abundance of examples in the two most important manuscripts of the Chronicle, H and P (3.1). The majority of these examples are found immediately after a proper name, after an official—or by extension a less formal —title, or after kin terms, all constituents which introduce new referents (3.2). According to the pragmatic principle operative in Medieval Greek (1.2), these emphasized constituents attract the (still) enclitic OCPs into preverbal position. We assume {σὲ λέγω} to have become an enclitic unit and thus to be attracted as a whole into this position. However, the syntactic dependence of σέ on λέγω must have led to a reanalysis: instead of leaning on the preposed constituent, the OCP now 53 Cf. σὲ θέλει ἐβγάλει ἐκ τὸν Μορέαν, # ὅπου οὐκ ἔχεις δίκαιον (P 4125). 54 Cf. µὲ ἤφερεν ἡ ὄρεξις # κ’ ἔπιακα τὸ βιβλίο (P 7638). 258 FROM ENCLISIS TO PROCLISIS IN MEDIEVAL GREEK ————— Greek, Roman, and Byzantine Studies 52 (2012) 240–258 forms a phonological unit with the following verb. The result of this natural evolution is found in Modern Greek, where pre- verbal proclitic OCPs are the norm. Our hypothesis, which focuses on one so-called ‘minimal utterance’, is decidedly more concrete than Horrocks’, yet both can co-exist, as they are based on the same principles: an original—frequently occurring—situation in which an enclitic OCP leans on a preceding word (να/emphasized constituent) and is followed by its verb; a reanalysis/reinterpretation of the direction of clisis of the OCP—it is no longer enclitic on the preceding word, but becomes proclitic on its natural host, the verb; extension by analogy and finally generalization. January, 2012 Dept. of Linguistics Ghent University Blandijnberg 2 Ghent 9000, Belgium jorie.soltic@ugent.be mark.janse@ugent.be