In his book Περὶ τῆς τοῦ λόγου συντάξεως Michael Syncellus includes a section on the rules for accenting prepositions that occur in anastrophe (1128–204, §§ 144–9).Footnote 1 This section is also part of the chapter on the accentuation of prepositions preserved in the Τονικὰ παραγγέλματα by John of Alexandria (26.13–28.19),Footnote 2 an important epitome of Aelius Herodian's lost work Περὶ καθολικῆς προσῳδίας.Footnote 3 Further below (1595–606, § 189, and 1614–43, §§ 191–2), Michael's treatment of the various functions of the conjunction ἤ/ἦ (διαζευκτικός, παραδιαζευκτικός, διασαφητικός, διαπορητικός, διαβεβαιωτικός) again presents very strong similarities with the corresponding unit of the chapter on the accentuation of conjunctions in the Τονικὰ παραγγέλματα (41.1–42.24).Footnote 4 In this article I first argue that Michael's sections, which have gone entirely unnoticed by students of the Cath. Pr., have drawn directly upon the Cath. Pr., and I identify the ways in which they add to our picture of Herodian's and John's aforementioned works. Then I turn my attention to John's chapter on the accentuation of prepositions: I provide additional evidence to support Eduard Hiller's view that it does not form a coherent whole, and I discuss the implications of this problematical structure for defining the contents of Book 18 and the Appendix of the Cath. Pr.
I. MICHAEL SYNCELLUS: AN INDEPENDENT WITNESS TO ΤΗΕ ΠΕΡΙ ΚΑΘΟΛΙΚΗΣ ΠΡΟΣΩΙΔΙΑΣ
The similarities between Michael and John of Alexandria in the aforementioned sections are very close indeed, and sometimes are even almost verbatim. In order to illustrate this, it will suffice to quote two pairs of passages:
Mich. Sync. 1146 (§ 145): εἰ γὰρ μεταξὺ πέσοιεν ὀνόματος καὶ ῥήματος ἢ μετοχῆς πρὸς διάφορον σημασίαν ἀναστρέφονται ἢ οὐκ ἀναστρέφονται, οἷον· ‘ΙΘΑΚΗΝΚΑΤΑΚΟΙΡΑΝΕΟΥΣΙΝ’ (Od. 1.247). εἰ μὲν γὰρ τὴν κατά πρὸς τὴν Ἰθάκην συντάξειέ τις ἵνα δηλώσῃ τὴν τοπικὴν σχέσιν κατὰ τὴν Ἰθάκην, ἀναστρέψει τὴν πρόθεσιν· εἰ δὲ πρὸς τὸ κοιρανέουσιν, ἵνα πλείονα δηλώσῃ ἔμφασιν τῆς τῶν μνηστήρων ἐπικρατείας, ὅπερ καὶ μᾶλλον δοκεῖ, ὀρθοτονήσει τὴν πρόθεσιν.
Ioh. Alex. 26.13: ἐὰν μεταξὺ πέσωσιν ὀνόματος καὶ ῥήματος ἢ μετοχῆς, πρὸς διάφορον σημασίαν ἢ ἀναστραφήσονται ἢ οὔ, οἷον ‘ΙΘΑΚΗΝΚΑΤΑΚΟΙΡΑΝΕΟΥΣΙΝ’ (Od. 1.247)· εἰ μὲν πρὸς τὸ Ἰθάκην συντάττοις τὴν κατά, ἵνα τοπικὴν δηλώσῃς σχέσιν, κατὰ Ἰθάκην, ἀναστρέψεις τὴν πρόθεσιν· εἰ δὲ πρὸς τὸ κοιρανέουσιν, ἵνα ἔμφασιν πλείονα δηλώσῃς τῆς τῶν μνηστήρων ἐπικρατείας, ὅπερ καὶ μᾶλλον δοκεῖ, ὀρθοτονήσεις τὴν πρόθεσιν.
Mich. Sync. 1631 (§ 192): διαβεβαιωτικὸς δὲ καὶ μεθ’ ὑπερωτήσεως καὶ χωρὶς ὑπερωτήσεως πολλάκις παραλαμβάνεται. μεθ’ ὑπερωτήσεως μὲν οἷον· ‘τί δὲ λαὸν ἀνήγαγεν ἐνθάδ’ ἀγείρας | Ἀτρεΐδης; ἦ οὐχ Ἑλένης ἕνεκ’ ἠϋκόμοιο;’ (Il. 9.338-9). χωρὶς δὲ ὑπερωτήσεως· ‘ἦ μάλα δὴ τέθνηκε Μενοιτίου ἄλκιμος υἱός’ (Il. 18.12), ‘ἦ μάλ’ ἐλαφρὸς ἀνήρ, ὃς ῥεῖα κυβιστᾷ’ (Il. 16.745), ‘ἦ δὴ λοίγια ἔργα τάδ’ ἔσσεται’ (Il. 1.573).
Ioh. Alex. 42.8: περισπῶσι καὶ τὸν βεβαιωτικόν· ‘τί δὲ λαὸν ἀνήγαγεν ἐνθάδ' ἀγείρας | Ἀτρείδης; ἦ οὐχ Ἑλένης ἕνεκ' ἠϋκόμοιο;’ (Il. 9.338-9). ἐνταῦθα μὲν βεβαιωτικός ἐστιν ὁ ἦ μεθ' ὑπερωτήσεως, ἔστι δὲ καὶ μόνως διαβεβαιωτικὸς χωρὶς ἐρωτήσεως· ‘ἦ μάλα δὴ τέθνηκε Μενοιτίου ἄλκιμος υἱός’ (Il. 18.12), ‘ἦ μάλ' ἐλαφρὸς ἀνήρ, ὃς ῥεῖα κυβιστᾷ’ (Il. 16.745), ‘ἦ δὴ λοίγια ἔργα τάδ' ἔσσεται’ (Il. 1.573).
However, apart from the strong links, there are also a number of discrepancies between the two authors, which concern issues of content, variant readings and arrangement.
1. Additional material
Michael's section on prepositions contains three items that are not found in John. All three can be shown to be typical of Herodian:
(1) In 1137ff. (§ 144) Michael mentions that the preposition ἀνά does not become paroxytone in anastrophe, even though it is disyllabic and consists of two morae (δίχρονος); he then proceeds to explain the rationale of the exception: ἵνα μὴ συμπέσῃ τῷ ‘ἄνα Λητοῦς | υἱέ‘ (Ap. Rhod. 2.214-15). That ἀνά in anastrophe is not susceptible of throwing back its accent is a known teaching inaugurated by Aristarchus and followed by Herodian: cf., for example, schol. Il. 5.824a. The illustration of the phenomenon through a quotation of Apollonius Rhodius is not attested elsewhere, but Herodian did use the Hellenistic poet in other cases: for example, apud Ioh. Alex. 11.33, Palimps. Fragm. f. 4v, f. 24r, f. 2v,Footnote 5 schol. Il. 21.141.Footnote 6
(2) In 1140ff. (§ 145) Michael cites Od. 6.15-16 (ᾧ ἔνι κούρη | κοιμᾶτο) and Il. 1.30 (ἡμετέρῳ ἔνι οἴκῳ) in order to illustrate that in anastrophe the disyllabic form of ἐν, i.e. ἐνί/ἔνι, is liable to throw back its accent, since it is an instance of the general rule that every anastrophized preposition that is disyllabic and consists of two morae has that potentiality. The teaching appears in Ioh. Alex. 27.33ff. too, but John cites only Od. 6.15-16 as an example. That the additional example Il. 1.30 ἡμετέρῳ ἔνι οἴκῳ adduced by Michael may ultimately have been a Herodianic choice is a possibility arising from the paroxytone nature of ἔνι in the context of the structure of Il. 1.30: the structure of Il. 1.30 follows the pattern ‘adjective qualifying noun x + preposition + noun x’, and Herodian was one of those who taught that a (δισύλλαβος, δίχρονος) preposition occurring in the aforementioned pattern actualizes its potentiality of throwing back its accent. The relevant evidence is contained in Ioh. Alex. 26.25ff., schol. Il. 18.92b1 (ὑπό) and schol. Il. 2.839b1; the passage from John reads as follows (Ioh. Alex. 26.25):
εἰ δὲ μεταξὺ πέσῃ δύο ὀνομάτων ἡ πρόθεσις, τοῦ μὲν κυρίου ὄντος τοῦ δὲ προσηγορικοῦ ἢ ἐπιθετικοῦ καὶ προσηγορικοῦ, ἐν πᾶσι τούτοις ἀναστρέφειν δεῖ τὴν πρόθεσιν.
However, there is evidence in Ap. Dysc. Synt. 442.1 and schol. Il. 2.839b1 that giving ἐνι an acute on the penultimate syllable in the context of Il. 1.30 would fit the teaching not only of Herodian, but also of Apollonius Dyscolus, Ptolemy of Ascalon and Nicias.Footnote 7
(3) In 1170ff. (§ 147) Michael discusses what happens to the accent of those prepositions that are placed immediately after their cases, that is, where no words intervene between them. The examples he employs to clarify the case include Il. 1.162 and Od. 6.12, found also in Ioh. Alex. 27.5ff., but additionally Michael offers Od. 4.555 = Od. 24.104 (Ἰθάκη ἔνι οἰκία ναίων). The new example fully conforms to the Herodianic teaching (attested in the cited passage of Ioh. Alex. and in schol. Il. 5.739b) that in such a case the preposition becomes paroxytone.
(4) There is a fourth item that needs to be discussed here. Where Michael treats the disjunctive use of the conjunction ἤ (1595ff., § 189), he has the two examples that we also find in Ioh. Alex. 41.1ff. But this is one of the casesFootnote 8 in which Michael has drawn material from more than one grammarian.Footnote 9 This is because he needs to analyse the function of the disjunctive ἤ (not only its accent), and on this matter the material contained in John is quite brief. Among the four examples unattested in John, three are quite short and common,Footnote 10 but one is extensive, Homeric, and can be (partly) paralleled only in Heliodorus the metrist: Il. 1.144-6 εἷς δέ τις ἀρχὸς ἀνὴρ βουληφόρος ἔστω | ἢ Αἴας ἢ Ἰδομενεὺς ἢ δῖος Ὀδυσσεύς | ἠὲ σύ, Πηλεΐδη, πάντων ἐκπαγλότατ‘ ἀνδρῶν (Mich. 1599ff.). Il. 1.144-6 is an exceptionally straightforward example for clarifying the meaning of the disjunctive ἤ, because the presence of εἷς in Il. 1.144 excludes the possibility of taking the conjunctions in Il. 1.145-6 as subdisjunctive.Footnote 11 Heliodorus uses part of this quotation, that is, Il. 1.145, in a context where he needs to refer to the quantity of the disjunctive ἤ: Heliod. apud Choerob. in Heph. 197.22 ὅτε ὁ ἤ διαζευκτικὸς † ἐπιφέρεται ‘ἢ Αἴας ἢ Ἰδομενεὺς ἢ δῖος Ὀδυσσεύς’.Footnote 12 There is absolutely no reason to assume a link between Michael and Heliodorus, and we can be certain that the two authors used the quotation independently of each other.
2. Variant readings
Turning our attention now to the second kind of difference between Michael and John, we can make three observations: the first two concern the section on prepositions, and the third refers to the section on conjunctions.
Michael offers the true reading in the two occurrences of the Homeric quotation Il. 7.163 = Il. 23.812 in the following places:
-
(1) ὦρτο Mich. 1179 (§ 148): ἦλθε Ioh. Alex. 27.15 (ΑV)Footnote 13
-
(2) ὦρτο Mich. 1180 (§ 148): ἦλθε Ioh. Alex. 27.16 (AV)
Ἦλθε in John may either be a deliberate change or a mistake that was made subconsciously under the influence of the following Homeric quotation Od. 18.1. In either case John himself, and not the copyists of his treatise, might be thought responsible for the alteration.
(3) Herodian taught that the conjunction ἠ takes a different accent according to whether it follows the causal conjunction ἐπεί or the interrogative pronoun τί:
Hdn. apud schol. Il. 20.251–5a1 (cf. lines 251ff., ἀλλὰ τί ἢ ἔριδας καὶ νείκεα νῶϊν ἀνάγκη | νεικεῖν ἀλλήλοισιν ἐναντίον …): … περὶ δὲ τοῦ η τοῦ ὑποτασσομένου τῷ τί καὶ τῷ ἐπεί ἐδηλώσαμεν ἐν τοῖς πρὸ τούτωνFootnote 14 ὡς τῷ τί ὑποτασσόμενος μὲν ἐγκλίνεται, τῷ δὲ ἐπεί περισπᾶται.
As we see from the above scholion, Herodian employs the verb ἐγκλίνεται to describe the accent of ἠ when ἠ is placed after τί. The same verb occurs in Apollonius Dyscolus at a passage where he deals with the same subject (Conj. 255.5):
… ἐν δὲ ὑποτάξει (sc. τιθέμενος ὁ σύνδεσμος η) ἐγκλίνεται … ‘τί ἢ δὲ σὺ νόσφιν ἀπ’ ἄλλων’ (Il. 15.244).
Ἐγκλίνω in the above two passages is used in a specialized and uncommon meaning, aptly described by the editor of Apollonius Dyscolus, Richard Schneider, as ‘τὸν τόνον acutum in gravem muto’. We do not have here the much commoner meaning ‘accentum in praecedentem syllabam vel vocem reicio’.Footnote 15 In the above two passages ἐγκλίνεσθαι refers to an essentially oxytone word.
Michael 1641 (§ 192) may be thought to preserve the original Herodianic word, when he writes ὅτε δὲ συντεθῇ μετὰ τοῦ τί πύσματος, ἐγκλίνεται· ‘τίη δὲ σὺ κήδεαι οὕτως;’ (Il. 6.55), ‘ἀλλὰ τίη μοι ταῦτα φίλος διελέξατο θυμός’ (Il. 11.407). On the other hand, John at 42.21 (ΑV) offers the variant ὀξύνεται: ὅτε παραπληροῖ μετὰ τὸ τί πυσματικόν, ὀξύνεται· ‘τί ἢ δὲ σὺ κήδεαι οὕτως | Ἀργείων;’ (Il. 6.55), ‘ἀλλὰ τί ἤ μοι ταῦτα φίλος διελέξατο θυμός;’ (Il. 11.407). John seems to treat ἐγκλίνειν and ὀξύνειν as interchangeable in sense in this kind of context, since he uses ὀξύνεται in 41.22 but ἐγκλίνεται in 42.2 to describe the very same phenomenon. So ὀξύνεται in 42.21 should be retained as a conscious choice on the part of the author.
There are five more Herodianic readings in Michael that are not found in John's MSS. However, these should be treated differently from the three readings discussed above. The reason is that these five seem to have been written by John too, but in the process of the transmission of his epitome they degenerated into the corruptions that mar his text now. Some of these corruptions might have arisen before, and some after, Michael composed his treatise, but since we have no means to distinguish between them, the safest policy is to reject them all as indications for establishing that Michael is independent of John. However, these readings are useful in showing how the textual criticism of John and the reconstruction of the Cath. Pr. can profit from Michael's parallel passages. I gather them in a list of their own:
(1) Almost all of Michael's manuscripts provide the reading νοήματος in 1182 (§ 148), which corresponds to ὀνόματος given by the two main manuscripts AV of Ioh. Alex. at 27.20. Νοήματος is obviously what is required by the context, and the editor princeps, Karl Wilhelm Dindorf, who had discovered the reading in a manuscript designated by him as ‘MS περὶ προθέσεων’, endorsed its truth,Footnote 16 although he did not put it in his text.
(2) συναληλιμμέναι at Mich. 1189 (§ 149) is obviously the correct reading.Footnote 17 John's συνηλλαγμέναι (V) or συναλλαγμέναι (Α) in 28.8 is a scribal error.Footnote 18
(3) In the section on conjunctions, Michael 1621 (§ 191) includes ἐθέλω in the quotation ‘πλουτεῖν ἐθέλω ἢ πένεσθαι’. John's MSS on the other hand omit the word (at 41.10), but without it the quotation, which is designed to clarify the comparative function of ἤ (διασαφητικὸς ἤ), loses all its relevance and force; without a verb implying comparison, the phrase can equally well be construed as an example of the disjunctive use of ἤ.
(4) Michael 1625ff. (§ 191) deals with the interrogative conjunction ἦ (διαπορητικὸς σύνδεσμος), confining the matter to the case when this conjunction is used on its own to introduce a simple question; that is, he is not concerned with the case when ἦ is combined with a preceding ἤ to introduce a set of alternative questions. He phrases the last point in the words μὴ οὔσης τῆς συντάξεως διαζευκτικῆς (1625–6). The corresponding phrase in John 41.21 (ΑV), who treats the accent of the interrogative conjunction ἦ in alternative as well as simple questions, reads as follows: διαζευκτικῆςFootnote 19 οὖν οὔσης τῆς τάξεως τῶν διαπορητικῶν, ὁ μὲν πρότερος ὀξύνεται, ὁ δὲ δεύτερος περισπᾶται. This rule is elucidated by three Homeric quotations: Od. 11.172, Od. 1.175 and Il. 1.190-2. Τάξεως in the above sentence yields good sense (‘if the function of the interrogative conjunctions is disjunctive’Footnote 20 ), but is unparalleled in John. Moreover, the proper Herodianic word here must have been what Michael's text offers, that is, συντάξεως. It is not that συντάξεως yields a better sense than τάξεως: ‘if the interrogative conjunctions appear in a disjunctive structure’, the sense is equally acceptable in the two cases. The reason for preferring συντάξεως is that there is stronger evidence to commend it: it is supported by two Herodianic notes, on A 190–2 and Y 17, that refer to the same rule and employ the same set of examples as John:
Hdn. apud schol. Il. 1.190-2: ἠ … ἠε: τὸν μὲν πρότερον σύνδεσμον βαρυτονητέον, τὸν δὲ δεύτερον περισπαστέον (accuratius προπερισπαστέον)· διαπορητικὴ γάρ ἐστιν ἡ σύνταξις. καὶ οὕτως ἐν τῇ Καθόλου.
Hdn. apud schol. Il. 20.17: … τοῦτο δέ φημι πρὸς τὰ τοιαῦτα ‘ἢ δολιχὴ νοῦσος ἦ Ἄρτεμις ἰοχέαιρα’ (Od. 11.172), ‘ἠὲ νέον μεθέπεις ἦ καὶ πατρώϊός ἐσσι’ (Od. 1.175)· ἐν γὰρ ταῖς τοιαύταις συντάξεσιν ὁ μὲν πρότερος ἤ ἐγκλίνεται, ὁ δὲ δεύτερος περισπᾶται.
It is difficult to think that the omission of ἐθέλω in (3) and the variant τάξις in (4) are deliberate changes made by John; it is better to take them as errors made by the copyists, and so ἐθέλω and σύνταξις should be restored to the Τονικὰ παραγγέλματα as well as to the Cath. Pr. Adding ἐθέλω is plausible from the palaeographic point of view too, since θέλουσι was omitted in 27.28.Footnote 21 As regards σύνταξις, it is paralleled elsewhere too,Footnote 22 and the loss of a small first part in a compound word is common in the Τονικὰ παραγγέλματα: 35.16<προ>περισπωμένων,Footnote 23 35.17 <προ>περισπωμένως,Footnote 24 38.14 <προ>παροξύνεται,Footnote 25 41.9 <παρα>διαζευκτικῶν.Footnote 26
(5) διελέξατο at Mich. 1643 (§ 192) instead of John's δέξατο in 42.24 (A; om. V) is obviously the correct reading of the Homeric quotation Λ 407. δέξατο is a scribal mistake, and Dindorf rightly removed it from John's text.
Of the items described in the ‘Additional material’ section above, the Herodianic teaching about the preposition ἀνά together with the quotation from Ap. Rhod. 2.214-15 should perhaps be added to this list. I have drawn elsewhere a distinction between this item and the others of the aforementioned section, and I have set out the reasons why we are justified to think that John had incorporated this item in particular in his epitome.Footnote 27
3. Arrangement of material
The third area in which the two authors diverge relates to the arrangement of the subject matter. Divergence in this area is represented only in the section on the accentuation of prepositions in anastrophe. Michael begins by explaining two things: (i) that monosyllabic prepositions placed at verse-end are the only monosyllabic prepositions that in anastrophe ‘ὀρθοτονοῦσι τὴν ὀξεῖαν τάσιν’ (1129ff., § 144), and (ii) that disyllabic prepositions consisting of two morae—except for ἀνά and διά—are the only sub-set of disyllabic prepositions that in anastrophe possess ἐν δυνάμει the disposition of changing the position of their accent (1135ff., §§ 144–5); (iii) then he proceeds to describe the circumstances under which these disyllabic prepositions consisting of two morae actualize or not that potentiality (1146ff., §§ 145–9). (iv) He ends the section by referring to the rules for accenting elided anastrophized prepositions (1189ff., § 149) and (v) ‘prepositions used as verbs’ (1199ff., § 149).
On the other hand, John structures this section as follows: (iii) 26.13–27.27 + (ii) 27.28–27.34 + (i) 27.34–28.4 + (v) 28.5–28.7 + (iv) 28.8Footnote 28 –28.19.
Michael's order is better organized, since (i) and (ii), in this order, facilitate the understanding of (iii). Also, it is best to place (iv) before (v), because with (iv) we are still in the area proper of prepositions and anastrophe, whereas (v) belongs to the area where prepositions and verbs intersect, and is irrelevant to anastrophe.Footnote 29 For these reasons Michael's order might well be taken to reflect the Herodianic one.Footnote 30 John's rearrangement of material is not hard to parallel, and another example thereof has long been revealed by Eduard Hiller:Footnote 31 there is evidence to show that Herodian treated the accent of monosyllabic nouns after the accent of non-monosyllabic nouns; more specifically Herodian had put monosyllables in Book 14 of his Cath. Pr., and non-monosyllabic nouns in Books 1–13.Footnote 32 John however reverses the order.Footnote 33
Admittedly John's rearrangement in the section on anastrophized prepositions is more complex, and needs an explanation, but before we proceed in that direction, it is useful to summarize the results of our investigation so far and to try to explain them. In the above discussion we noted the very strong connections between Michael and John in the sections on prepositions and conjunctions. We also noted a number of divergences, and we saw that they follow a specific pattern: every time Michael deviates from John, he offers an item that is in line with the Herodianic teaching. Admittedly the paroxytone nature of ἔνι in the quotation Il. 1.30 does not conform to the teaching of Herodian exclusively, and elements such as ὦρτο or the quotation of Il. 1.144-6 might reflect the work of Michael himself, but the most economical hypothesis is that Michael derived all this divergent material from one and the same source of material of the Cath. Pr.
However, we must not confine ourselves to explaining only the discrepancies between John and Michael; we need a more comprehensive hypothesis to account for the striking similarities as well as the discrepancies between them, and for this purpose the most obvious hypotheses that come to mind are: (1) that Michael used John's epitome and another source of material of the Cath. Pr., now lost; (2) that Michael used an epitome of the Cath. Pr., now lost to us; this epitome was fuller than John in comprehensiveness, but its phrasing was very close to that of John; (3) that Michael consulted the Cath. Pr. directly.
What weight should be put on each of these three hypotheses? I can hardly regard (1) and (2) as satisfactory because of the total lack of evidence elsewhere that such a source or such an epitome ever existed in the real world.Footnote 34 (1) has the additional shortcoming of not being as simple as (2) and (3), since it involves postulating two documents rather than one. (3) is free from both weaknesses, and is therefore preferable. Apart from its intrinsic plausibility, there are two items of external evidence to commend it. First, Michael refers to Herodian explicitly and makes clear that he has direct access to his works. The relevant passage comes from the preface of the treatise (Mich. Sync. 1ff., §§ 1–2):Footnote 35
Ἡ περὶ τῆς συντάξεως τοῦ λόγου διδασκαλία πολλή τε καὶ βαθεῖα καθέστηκε, φιλομαθέστατε Λάζαρε, καὶ μακροῦ δεομένη χρόνου καὶ μείζονος ἢ τῆς καθ’ ἡμᾶς δυνάμεως. Πολλὰ γὰρ περὶ αὐτῆς τοῖς ἀρχαίοις διαπεπόνηται, Ἀπολλωνίῳ τε τῷ σοφωτάτῳ τὸ ἐπίκλην δυσκόλῳ καὶ Ἡρωδιανῷ τῷ πολυμαθεστάτῳ καὶ Ἀπολλωνίῳ τῷ νέῳ καὶ τοῖς ἀττικισταῖς καὶ Ἀρκαδίῳ τῷ Βυζαντίῳ, ὧν ταῖς πραγματείαις εὖ ἐντετυχηκὼς θαυμάσεις αὐτῶν τὸ πολυμαθὲς καὶ τοῦ ἑλληνισμοῦ τὴν ὑγιᾶ καὶ κατάλληλον χρῆσιν καὶ τὴν ἀκριβῆ καὶ πολύτροπον καὶ πολυσχήμονα πάσης τῆς ἑλληνικῆς διαλέκτου ποικιλίαν· ὅμως οὖν, ὡς οἷόν τέ ἐστιν, σὺν Θεῷ τῷ σοφωτάτῳ γοῦν τοι, πειράσομαι σύντομόν σοι ποιήσασθαι μέθοδον εὐρύθμου χάριν ὁμιλίας.
The idea that Michael consulted the Cath. Pr. directly gains further plausibility by the fact that there is independent evidence too to show that the Cath. Pr. was available for consultation as late as the ninth century: Choeroboscus and Theognostus, both of the ninth century, are already known to have used Herodian's work as a source for their own grammatical treatises.Footnote 36
II. THE STRUCTURE OF JOHN OF ALEXANDRIA'S CHAPTER ON THE ACCENTUATION OF PREPOSITIONS AND ITS SIGNIFICANCE FOR THE ΠΕΡΙ ΚΑΘΟΛΙΚΗΣ ΠΡΟΣΩΙΔΙΑΣ
Our search for an explanation of John's aforementioned rearrangement in the section on anastrophized prepositions must start from a more general consideration of the structure of his whole chapter on prepositions (25.32–28.19). Lentz pronounced the texture of this chapter to be so coherent (‘in sich so zusammenhängend’) as to exclude the possibility of a relocation from elsewhere.Footnote 37 That his claim was false was already seen by Hiller.
In fact, Lentz entirely lost sight of two structural problems. In the lines immediately preceding (iii) we read (Ioh. Alex. 26.7–12):
Ὅτι μὲν οὖν ὀξύνονται αἱ προθέσεις εἴπομεν. διὰ τί δὲ προθέσεις εἴρηνται; ἐκ τοῦ πάσης λέξεως ἐθέλειν προτίθεσθαι· λέγω δὲ ὀνομάτων, ῥημάτων, μετοχῶν, ἀντωνυμιῶν, ἄρθρων τῶν προτακτικῶν· περὶ Ἀριστάρχου, καταγράφω, καταγράψας, περὶ ἐμοῦ, περὶ τοῦ στεφάνου.
As we mentioned (above), prepositions are oxytone. But why are they called ‘prepositions’? They are called so, because they have the tendency to come before all kinds of words, that is, nouns, verbs, participles, pronouns, articles: e.g. περὶ Ἀριστάρχου, καταγράφω, καταγράψας, περὶ ἐμοῦ, περὶ τοῦ στεφάνου.
The sentence ὅτι μὲν οὖν ὀξύνονται αἱ προθέσεις εἴπομεν, with which the above section opens, is intolerably repetitious, because John has given the very same piece of information only ten lines earlier, and has absolutely no need to refresh our memory on this fact (Ioh. Alex. 25.32ff.):
Πᾶσα πρόθεσις ὀξύνεται· ἔν, εἴς, ἔξ, ἀνά, κατά καὶ τὰ λοιπά. ὅθεν καὶ ἡ ἀπό καὶ ὑπό μηδεμιᾶς ληγούσης λέξεως εἰς -ο ὀξυνομένης διὰ τὸ εἶναι προθέσεις ὠξύνθησαν … ἔτι δὲ καὶ ἡ ὑπέρ πάλιν ὠξύνθη, καὶ ταῦτα <μη>δεμιᾶς λέξεως εἰς -ρ ληγούσης μετὰ βραχείας [καὶ] ὀξυνομένης …
_______________
3 <μη>δεμιᾶς Lentz 1867, 480.2: δὲ μιᾶς AV (recepit Dindorf) 4 -ρ scripsi: -ε AV: -ερ Dindorf καὶ eieci
Every preposition is oxytone: ἔν, εἴς, ἔξ, ἀνά, κατά, etc. Hence, although no word ending in –ο is oxytone, ἀπό and ὑπό are oxytone, precisely because they are prepositions … Also, ὑπέρ is oxytone, although no word ending in a short vowel followed by –ρ is oxytone.
Furthermore, the progress of thought between Ioh. Alex. 26.12 and the beginning of (iii) is very abrupt:
Joannes enim, ubi causam appellationis (‘διὰ τί προθέσεις εἴρηνται’) exposuit, statim nulla transgressione usus de anastropha ita agere incipit: ‘ἐὰν μεταξὺ πέσωσιν ὀνόματος καὶ ῥήματος ἢ μετοχῆς, πρὸς διάφορον σημασίαν ἢ ἀναστραφήσονταιFootnote 38 ἢ οὒ κτλ’.Footnote 39
For, after setting out the rationale of their name (‘why they have been called prepositions’) and without using any transition whatsoever, John of Alexandria begins to treat anastrophe straightway: ‘if a preposition falls between a noun and a verb or a participle, it will throw back its accent or not according to its meaning, etc.’.
A smooth transition to (iii) seems to require another sentence between 26.12 and (iii), in which John would have explained that prepositions do as a rule precede their cases, as their name suggests, but can sometimes follow their cases, that is, they can be used in anastrophe.
Hiller had noticed only the second problem, and put forward the following hypothesis: he considered the part dealing with the accentuation of prepositions in anastrophe, i.e. 26.13–28.19, as a self-contained unit, which did not belong to Book 18 of the Cath. Pr., but which had been taken from the Appendix.Footnote 40 Then this part was combined with the rest of the material on prepositions which, as Hiller seems to think, all came from Book 18.Footnote 41 Here we need to remember that the Appendix tackled the impact of connected speech on accentuation. The evidence for this is contained in [Arcadius'] Table of Contents ([Arcadius] 5.1ff. Roussou, with modifications in the punctuation):
Ἐπὶ δὲ τοῖς εἴκοσι ἐν ἄλλῳ βιβλίῳ τὸ ἀναγνωστικὸν εἶδος κατὰ τὴν σύνταξιν τῶν λέξεων παραδίδοται, ἀκολούθως μετὰ τὴν δήλωσιν τῆς καθ’ ἑκάστην προσῳδίας γινομένης διδασκαλίας, περὶ διαστολῶν καὶ συναλοιφῶν καὶ τῶν ἄλλων τῶν παρακολουθούντων τῇ ἀναγνώσει· οἷον ὅτε τὸ घεύς ὀξυνόμενον οὐκ ἔχει τὴν ὀξεῖαν ῥωννυμένην ἐν τῷ ‘घεὺς δ’ ἐπεὶ οὖν Τρῶας’ (Il. 13.1), ἀλλ’ ἐγκλινομένην διὰ τὴν ἐπιφορὰν τοῦ δέ συνδέσμου. ἐν γοῦν τῷ ‘घεύς τε καταχθόνιος’ (Il. 9.457) ῥώννυται διὰ τὴν ἐπιφορὰν τοῦ τέ. ἡ δὲ οἱ ἀντωνυμία περισπωμένη ἐν τῷ ‘οἷ αὐτῷ θάνατον’ (Il. 16.47) οὐκ ἐφύλαξε τὴν περισπωμένην ἐν τῷ ‘καί οἱ ἐπευχόμενος’, τῆς συντάξεως τοῦτo ἀπαιτούσης τοῦ σημαινομένου. οὕτως ἔχει καὶ τὸ ‘μάχῃ ἔνι κυδιανείρῃ’ (Il. 6.124 al.), καὶ ὅσα τοιαῦτα.
In addition to these twenty books, following the exposition of the doctrine for accentuating individual words, there comes a separate book that deals with reading in connected speech, dividing elements, uniting elements, and, with the other matters that relate to reading: for example, the oxytone घεύς does not retain its acute in ‘घεὺς δ‘ ἐπεὶ οὖν Τρῶας’, but changes it to grave, because of the following conjunction δέ. But in ‘घεύς τε καταχθόνιος’ the acute on घεύς is retained, because of the following τε. The pronoun οἷ, which takes the circumflex in ‘οἷ αὐτῷ θάνατον’, does not retain it in ‘καί οἱ ἐπευχόμενος’, since this is what the syntax of the meaning requires. The same applies to ‘μάχῃ ἔνι κυδιανείρῃ’ and other such [prepositions occurring in connected speech].
On the other hand, Book 18, just as all the other books, treated the accentuation of words when these are used in isolation. Its content is described thus: τὸ δέκατον ὄγδοον ἀντωνυμίας ἄρθρα καὶ προθέσεις (sc. περιέχει) ([Arcadius] 4.18).Footnote 42
The core of Hiller's hypothesis, i.e. that there has been a relocation from the Appendix and a combination of the relocated material with the content of Book 18 within the framework of a single text, is correct, but we need to redefine what exactly belonged to Book 18 and what to the Appendix of the Cath. Pr. As we saw, Hiller assumed that Ioh. Alex. 26.13–28.19 belonged originally to the Appendix. This text is a composite entity consisting of (iii) + (ii) + (i) + (v) + (iv). Since items (iii) + (ii) + (i) + (iv) discuss the accent of prepositions in anastrophe, and since anastrophe is a phenomenon arising in connected speech, the assumption that Herodian had included these items in the Appendix rather than in Book 18 is a self-evident thing. However, with regard to item (v), it is reasonable to suppose that Herodian had placed it in Book 18, since (v) is unrelated to anastrophe and treats cases whose accent is invariable. Moreover, the structural problem that escaped Hiller's notice provides a strong indication that the passage 26.7–12 was part of the Appendix too: it is not only that this passage sits very oddly with the preceding material 25.32–26.7, which clearly derives from Book 18; it is also very well-suited for functioning as part of the introduction to the topic of the anastrophe in the context of the Appendix. In this context the opening sentence ‘Ὅτι μὲν οὖν αἱ προθέσεις ὀξύνονται εἴπομεν’ is quite natural, because now Herodian resumes a subject already treated in Book 18 and so needs to build a bridge to the previous treatment. He must also refresh our memory on the main lesson taught there, since the distance between Book 18 and the Appendix is quite big: that prepositions in isolation are oxytone. He then proceeds to deal with the new aspect of the subject: prepositions in connected speech, i.e. in anastrophe, and the accents thereof.
Let us now return to the way the two grammarians arranged their material in the section on anastrophe. Can we better understand why the two arrangements differ? On the basis of the revised version of Hiller's hypothesis both Michael and John combined material from two different sections of the Cath. Pr. Book 18 and the Appendix within the scope of a single text. We can think that in the process of this combination they made different choices of arrangement: Michael chose to follow Herodian closely in arranging the items he derived from the Appendix in the order (i) + (ii) + (iii) + (iv); he also seems to have thought fit to add item (v) from Book 18, but as he realised that this is not in line with the other four, he chose to put it at the very end, after (iv), thus preserving the essential unity of (i)–(iv).
Now John had to borrow more material from Book 18, as his topic was primarily accentuation and not syntax like Michael's. He opened his chapter on the accentuation of prepositions with the section 25.32–26.7, which treats the accent of prepositions when these are used in isolation; he derived it, as we mentioned, from Book 18. He then needed to deal with the accent of prepositions in connected speech, so he added 26.7–12 from the Appendix as an introduction to the phenomenon, without however taking care to construct a seamless transition at 26.7 and 26.12. His further thought processes that determined the ordering of the material are not clear, but we can make an attempt at reconstructing them as follows: after 26.12, his next addition seems to be based on his assessment of the relative importance of items (iii) and (i): since accenting disyllabic anastrophized prepositions constitutes the main bulk of the teaching about the accentuation of anastrophized prepositions, he decided to begin with item (iii). To (iii) he was then bound to subjoin item (ii), since both (iii) and (ii) refer to disyllabic prepositions. He was then left with the very short (i), which refers to the rule for accenting monosyllabic anastrophized prepositions. His choice to proceed to (v) of Book 18 might have been determined by the fact that (v) treats unelided words just as all the previous items. He closed his chapter with (iv), which refers to elided words.
CONCLUSIONS
In this article the identification of Michael Syncellus as a further source for the reconstruction of Herodian's Cath. Pr. and the close analysis of the structure of John of Alexandria's chapter on prepositions (25.32–28.19) have yielded a number of results that concern both Herodian's work and the Τονικὰ παραγγέλματα by John of Alexandria. It is now time to summarize them.
With the aid of the new source, we can enrich the material or improve the readings of no fewer than ten passages in Lentz's reconstruction of the Cath. Pr. (1867). More specifically, we can include four more poetic quotations in Lentz's text: 480.16, Ap. Rhod. 2.214-15; 480.18, Il. 1.30; 484.1, Od. 4.555 = Od. 24.104; 517.5, Il. 1.144-6. Also, we can substitute five readings and add a sixth one: 484.4 ὦρτο for ἦλθε; 484.5 ὦρτο for ἦλθε; 484.8 νοήματος for ὀνόματος;Footnote 43 518.2 συντάξεως for τάξεως; 520.10 ἐγκλίνεται for ὀξύνεται; 517.8 we must add ἐθέλω after πλουτεῖν. Moreover, with regard to the distribution of material among different books, Lentz's decision to assign to Book 18 of the Cath. Pr. the teaching about the accentuation of anastrophized prepositions, which he found in John of Alexandria, was misguided. As it has been argued, John of Alexandria derived this teaching (26.7–28.4 and 28.8–19) as well as the (first part of the) introduction to it (26.7–12) from the Appendix, and combined them with 25.32–26.7 and 28.5–7, which he had taken from Book 18. Concerning the ordering of the teaching in the Appendix, Michael Syncellus must be our guide, since there is good reason to believe that he, and not John of Alexandria, reflects the Herodianic arrangement of the material.
Finally, with regard to the Τονικὰ παραγγέλματα, we would be justified to put in John's text the teaching about the preposition ἀνά together with Ap. Rhod. 2.214–15 (27.33).Footnote 44 We are also entitled to include νοήματος (27.20), συναληλιμμέναι (28.8), ἐθέλω (41.10), συντάξεως (41.21) and διελέξατο (42.24).
In his seminal review of Herodianic scholarship in 1993 Andrew Dyck stated that ‘much basic philological work remains to be done on Herodian's epitomators and excerptors’.Footnote 45 The case of Michael Syncellus confirms this statement, and shows how much we stand to benefit from such a work.
