Text Networks DANIEL SELDEN University of California, Santa Cruz Semel locutus est Deus duo haec audivi quia potestas Dei et tibi Domine misericordia quia tu reddes unicuique iuxta opera sua. Unum locutus est Deus duo haec audivi quia imperium Dei est et tibi Domine misericordia quia tu reddes unicuique secundum opus suum. Jerome, Psalmi 61 Students of Late Antique letters might profitably revisit Augustine’s remarks on the advantages that conflicting versions of the Bible afforded Roman rea- ders in the early centuries C.E. “While we can enumerate those who have turned (verterunt) the Scriptures from Hebrew into Greek,” Augustine ob- serves, “those who have rendered them into Latin are innumerable. In the early times of the faith when anyone found a Greek codex, and he thought that he had some facility in both languages, he attempted to translate it (au- sus est interpretari).”1 Rather than castigate this embarrassment of codicolo- gical riches, however, Augustine—writing de doctrina Christiana—goes on to affirm such diversity in Scripture as fundamental to the constitution of the Catholic faith: The situation helps rather than impedes the understanding (intelligen- tiam), so long as readers (legentes) are not remiss (negligentes). For in- spection of the sense in several codices has often clarified passages that ————— 1 Augustine, De doctrina Christiana 2.11. For recent discussions of Augustine’s scriptural preoccupations, see Arnold and Bright 1995. Ancient Narrative, Volume 8, 1-23 2 DANIEL SELDEN are obscure. For example, one translator (interpres) renders a verse from the prophet Isaia: Et domesticos seminis tui ne despexeris; but another says: Et carnem tuam ne despexeris. Either bears witness to the other in turn (uterque sibimet invicem adtestatus est), for by means of one the other is explained (alter ex altero exponitur). Thus, caro can be taken literally, such that each may find himself admonished not to despise his body; while domestici seminis can be understood figuratively as “Chris- tians” born spiritually with us from the common seed of the Word. Moreover, by collating the sense of the translations, the most appropriate meaning presents itself (occurrit), namely, a literal precept that we should not despise those of our own blood. For when we refer do- mesticos seminis to carnem, blood relations come particularly to mind. Whence, I think, comes the statement of the Apostle: Si quo modo ad ae- mulationem adducere potuero carnem meam, ut salvos faciam aliquos ex illis, that is, by emulating those who have believed, they too might be- lieve. He calls the Jews carnem suam on account of their consanguinity.2 Augustine’s aim, then, in collating the diverse renderings of Isaia 58:7 is not—as it would be for Lachmann—to produce a normative text.3 Augustine both allows the discrepancies to stand and deliberately validates the meaning peculiar to each one, singly as well as together. The Itala refers metonymi- cally to the body, the Vulgate metaphorically to the spiritual community of Christ, while in conjunction the two renditions metaleptically reference the Jews.4 Oblivious of fidelity to any original—exemplaria praecedentis lin- guae—Augustine concludes: “Both [versions] contain something of value for the discerning reader (ex utroque magnum aliquid insinuatur scienter legen- tibus); overall, it is difficult for translators (interpretes) to so differ from each other that they do not show commonality in some area of meaning (ut non se aliqua vicinitate contingant).” Since the root *leg-, on which this passage repeatedly puns,5 denotes in Augustine’s Latin not only “to read” but also “to assemble”, “to bring together”, “to recruit”,6 lection is by im- plication here less a process of reduction than a collational and comparative ————— 2 Augustine, De doctrina Christiana 2.12. 3 See Timpanaro 1981; Weigel 1989. Cf. G. Nagy, “Editing the Homeric Text: Different Methods, Ancient and Modern” in: Nagy 2004, 75-109. 4 On the Old Latin versions of the Bible, see Rönsch 1875; Ziegler 1879; Burkitt 1896. 5 See Ernout and Meillet 1959, 348-50. See also P. Chantraine 1968, s.v. λέγω. 6 So the famous admonition in Confessiones 8.12: tolle, lege. On the conventional and multiple significances of the phrase, Courcelle 1968. TEXT NETWORKS 3 enterprise in which all available transcriptions remain in play. For Augus- tine, then, multiformity in the Latin renderings of Isaia constitutes neither a historical curiosity nor sheer coincidence: rather, it is what creates the possi- bility of understanding in the first place: “each [reading] bears witness to the other, so that by one the other is explained.”7 Accordingly, no variant, for Augustine, signifies positively in isolation, but only insofar as it constitutes part of a larger textual field in which both the vicinitas and divorsitas of the several γραπτά serve—for, as well as against, one another—to converge collectively on what Paul called the πνεῦμα of the sense.8 To paraphrase the Apostle from a related context:9 when it comes to multiple witnesses, there is for Augustine neither accurate nor inaccurate, orthodox nor unorthodox, au- thentic nor spurious, but “all are one in Christ Jesus”—omnes enim vos unum estis in Christo Iesu. 10 Although De doctrina Christiana itself constitutes a fixed work—at least relatively speaking11—Augustine reminds us through this programmatic ex- ample of the degree to which Roman Imperial readers had, athwart incon- gruous and often contradictory forces of canonization,12 repeatedly to con- front, both within and across languages, texts that circulated in a bewildering number of differing exemplars,13 many of which—beyond the vagaries of scribal error—possessed equal claims to authority and few of which could a priori be dismissed.14 After all, what was true of the divergences between the Itala and Vulgate stood compounded in the different versions of the Vetus ————— 7 Cf. the admonition at De doctrina Christiana 4.21: si quis in eis non superficie contentus altitudinem quaerat. 8 The play of similarity and difference that Augustine stresses indicates that we are well within the field of metaphor or allegory. On the hermeneutic implications, see Boyarin 1994. 9 Paul, Ad Galatas 3:28: non est Iudaeus neque Graecus, not est servus neque liber, non est masculus neque femina, omnes enim vos unum estis in Christo Iesu; cf. Πρὸς Ῥωμαίους 10:12: οὐ γὰρ ἐστιν διαστολὴ Ἰουδαίου τε καὶ ‘΄Ελληνος. It thus runs counter to Augustine’s own critical principles to decide between the two variant readings that trouble the climactic moment of his Confessions (cf. O’Donnell 1992, 3:62-63): et ecce audio vocem de domo vicina / de domo divina; for manifestly here the one is explained by the other. 10 Cf. De doctrina Christiana 4.29: quasi copia dicendi forma vivendi. 11 There are significant textual variants. See also n. 88 below. 12 See, inter alia, Metzger 1977; Auwers and de Jonge 2003. 13 Thomas 2003, an important case study of “multiples”, centered on the Acts of Peter. Her divergent framework broaches issues that are nevertheless key for the problematics treated here; cf. nn. 49, 99. 14 Metzger 1987. The first reference to τὰ κανονικά occurs in the canons of the synod held at Laodicea c. 363 CE. For a well presented case study, see Trobisch 1994. 4 DANIEL SELDEN Testamentum that passed translinguistically throughout the Late Antique Mediterranean world across cultural lines: Tanakh, the Targumim, Septu- agint and Peshīṭtā, Coptic, Samaritan, Armenian, and Old Ethiopic Bibles15—to name only the most prominent—all circulated within the bor- ders of the Roman state, each less a neutral translation of its Vorlage, than an appropriation apposite to ethnically divergent contexts, which brought mat- ters of local dominance, assertion, and resistance unequivocally to the fore.16 Consider, for example, three renderings of Shir ha-Shirim 1:5,17 which po- lemically convey differing cultural assumptions about skin color: Septuagint: Μέλαινά εἰμι καὶ καλή, θυγατέρες Ἰερουσαλεμ [var.: Ἰσραηλ]. Vulgate: Nigra sum sed formonsa filiae Hierusalem. Ethiopic: ÃEM : „{ : ‘O|§q : „M„gF© : „ï¢T\DíM ; ṣallām ’ana wa-śannāyt ’em-’awāleda ’iyarusālēm. Black am I and more beautiful than the daughters of Jerusa- lem. Not only, then, as Origen’s Hexapla made clear,18 did the contents of these various Testamenta differ: all incorporated nonidentical doublings—in some cases treblings —within and between their various canonical books.19 ך״נת,20 ————— 15 Metzger 1987, 218-28. 16 See, for example, Nida 1975; Schogt 1988; Beck 2000; Apter 2005. Of the increasingly numerous anthologies on translation theory, particularly to be recommended are Venuti 2000; Tymoczko and Gentzler 2002; and Berman and Wood 2005. 17 Masoretic Text (MT): םלשורי תונב הואנו ינא הרוחש. The titles of the “translations” vary: LXX. Ασμα; Vulg. Canticum Canticorum; Eth. Mahāləya mahāləy. The ensuing example shows why it is best to preserve the distinctions between the different texts—for which their original superscriptions are the most lapidary index—rather than conflate them into some putatively homogenous work called “Song of Songs”. On the ideological stakes in fictions of uniformity, see Derrida 1972. Texts—MT: Elliger and Rudolph 1966; LXX: Rahlfs 1935; Vulg.: Weber 1994; Eth. Gleave 1951. 18 Text: Origen, Field 1875. 19 See Davies 1998. For lucid summarization, see Friedman 1997. 20 The English term “Bible”—in the singular—conceals the fact that for Late Antique communities the scriptures were consistently designated as plural—the Greek Βιβλία, for example, means “scrolls”, not necessarily all collected. Up through the present day, in fact, Jewish communities still employ no omnibus term to designated their sacred corpus TEXT NETWORKS 5 for example, includes the two disparate accounts of creation which open תישארב;21 the deuteronomic recapitulation of the commandments—already redacted twice in תומש22—modified in םירבד;23 and the recasting of the Da- vid narrative from םיכלמ in םימיה ירבד.24 Similarly, the Septuagint expands Ιερεμιας with the Ἐπιστολὴ Ιερεμιου, followed by Βαρουχ,25 which the Peshīṭtā further extends with its ܟܘܪܒܕ ܗܢܝܠܓܕ ܐܒܬܟ.26 Nor are such gener- ative phenomena restricted to the Old Testament: undoubtedly the most stun- ning and influential set of narrative variants were the four “indispensable” Gospels27—τὸ εὐαγγέλιον τετράμορφον, as Irenaeus called them28—along- side the Gospel of the Hebrews and the Gospel of the Egyptians, whose col- ————— of written scripture; rather, what English speakers refer to as “the Hebrew Bible”, He- brew itself designates as ך״נת (Tanakh), an acronym formed from the first letters of the ti- tles respectively assigned to the three traditional divisions of the collection: Torāh (“Law”), Nəvi’im (“Prophets”), Kətuvim (“Writings”) > “Tanakh”. 21 LXX. Γένεσις; Vulg. Bresith. The Hebrew title of this book exemplifies the reasons why it is ultimately undesirable to vocalize the various Semitic syllabaries, or even to break up the consonantal text. Depending on how one vocalizes תישארב, it could mean either “In a beginning” (bərēshit) or “in the beginning (bārēshit)”—two openings whose phi- losophical implications are clearly very different. The LXX gives the somewhat hesitant Ἐν ἀρχῇ; Latin allows the Vulg. to obviate the issue entirely with “In principio”; but the Coptic (Bohairic) Pentateuch reads unequivocally “in a beginning” (qen ouarCy),” clear evidence for the antiquity of the ambivalence (de Lagarde 1967, 1). Qabbalistic writers, moreover, divide the consonantal sequence up into two words: תיש ארב (“[He] created six”), so as to make the title refer to the six days of creation. In the spirit of Augustine, for Late Antique readers it was less a matter of deciding which of these ren- derings of the title was “correct”—though different rabbis certainly had different opin- ions—but rather of recognizing the differences and, within the hermeneutics of the era, demonstrating how each the possible readings reveals a different layer of meaning im- manent within the text; cf. Zohar 3:152a. Turning to the translations, one should also note that Grk. Γένεσις and Lat. Bresith do not cover precisely the same semantic field; while Γένεσις is colloquial in Greek, Bresith is in Latin not only exotic, but actually a nonsense word—the two titles thus make very different sorts of appeals to the reader. Cf. Seidman 2006. 22 MT. Shmōt (“Names”) > LXX. Εξοδος. 23 MT. Dəvārim (“Words”) > LXX. Δευτερονόμιον. 24 MT. Məlākhim (“Kings”) / Divrei ha-Yomim (“Words of the Days”) > LXX Βασιλεῖς / Παραλιπόμενον. In the Vulgate, the full incipit for the latter exemplifies one way that translators reminded Late Antique readers of the multiformity of the texts, though the equivalency given to the titles also serves to conceal the discrepancies in matter between the diverse renditions: “Liber Dabreiamin, id est Verba Dierum qui Graece dicitur Parali- pomenon.” 25 See further Tov 1976. 26 “The Ascension of Baruch”; Text: de Boer 1972- . 27 Origen, cited in Eusebius, Historia ecclesiastica 6.25.3. 28 Irenaeus, Adversus haereses 3.11.8. 6 DANIEL SELDEN lective burden, Melito of Sardis argued, was to realize through “analogy” the “provisional draft” that the Judaic Law and the Prophets represented.29 These six gospels, translated into the eight principal languages of early Christen- dom—Aramaic, Greek, Coptic, Latin, Armenian, Gothic, Georgian, and Geʿez30—yielded over fifty competing versions of the mission of Christ, none identical in wording, but all of which—even before entering into the ni- ceties of alternatives within one language31—retained parallel authority un- der the High Empire,32 even though doctrinally the communities who pro- duced and selectively deployed them differed.33 Among Rabbinica, more- over, the Mishnāh (“repetition”), compiled c. 200 C.E.,34 likewise redoubles the Torāh which, supplemented by the Gemārā (“completion”), yielded be- fore 500 the two coincident though largely autonomous Talmudim.35 Nor were such parallel compositions necessarily commensurate in length. The Vulgate, for example, reduces the life of Enoch to a single sentence [Gen. 5:21-24]—ambulavit Enoch cum Deo … et non apparuit quia tulit eum Deus—yet no less than five pamphlets and two books of independent prove- nance, attested in as many languages,36 plus a mass of Aggadic material from the Rabbis,37 flesh out the details of Enoch’s apocalyptical career. Scriptural systems such as the apostolic Gospels, the Enochic corpus, or the Lives of Pachomius—whose Coptic, Arabic, and Greek recensions all ————— 29 Text: Melito of Sardis, Hall 1979. Melito takes his cue principally from John and Paul’s Letter to the Galatians; Augustine works out the scheme in its entirety in De civitate Dei. See also, in this connection, Crawford 2008. 30 Metzger and Ehrman 2005, 52-134. 31 There were six translations into Syriac alone, at least five into Greek; notoriously, Italic recensions did not offer a consistent text: for Luke 24:4-5 alone, at least 27 variant read- ings in Old Latin manuscripts survive, bearing out Jerome’s complaint to Pope Damasus that there were almost as many versions as there were manuscripts: “tot enim sunt exem- plaria paene quot codices” (Novum opus). See Burton 2001. 32 For example, Trobisch (1994) notes: “Approximately eight hundred early copies of the letters of Paul have survived to the current day. No two copies are completely identical… [T]he result is that there probably is not a single verse of the letters of Paul that has the same wording in all surviving manuscripts” (1 and 4). 33 The best introduction remains Pagels 1979. Of the increasingly large bibliography, par- ticularly useful are Layton 1987; Bloom and Meyer 1992. The two best general surveys are Rudolph 1977, and Filoramo 1983. 34 For a concise and perceptive introduction, see Neusner 1983. At greater length there is Albeck 1971. 35 The best overview is Akenson 1998. For a critical reading, see Ouaknin 1986. 36 Overview: Vanderkam 1995. See also Nickelsburg 2001. 37 For references, see Ginzberg 1998, 7:132-66. TEXT NETWORKS 7 S Sahidic A(a-m) Arabic VBr Vita Brevis Sbo Bohairic Den. Latin VTh Vita Theodori G Greek Paral. Paralipomena H.L. Lausiac History Pachomii Fig. 1. Lives of Pachomius. Stemma codicum [after Veilleux 1980-82]. derive from different sources [Fig.1]—constitute what we might profitably call discrete, if ultimately also overlapping—“text networks”, autopoietic bodies of related compositions whose origins largely escape us and whose evolution, in the second and third centuries C.E., remained far from com- plete. Within such self-organizing fields, however, neither origin nor termi- 8 DANIEL SELDEN nus was much at issue: so the Rabbinic Bible opens emblematically with ב, the second letter of the Hebrew alphabet,38 and breaks off open-endedly with the exhortation Vəyāʿal (“May he go up”).39 In fact, what most typified the scriptural networks of the High Empire was not their stability, but rather their set (Einstellung) towards proliferation, where entropy increased in the course of each new (re)inscription.40 Hence the vast majority of such wri- tings were not only pseudepigraphic—so among the Amorā’im the verb בתכ does not mean “he composed”, but rather “he transcribed”:41 such protocols tended to accrete either seriatim, like the variant reworkings of Estēr,42 or cu- mulate rhizomatically around a core set of tales,43 in the way that the later Acts of the Apostles constitute spin-offs of the Πράξεις of Luke.44 In fact, as late as the fifth century C.E., Jerome was still actively rendering ἀκονόνιστα βιβλία into both Latin and Greek,45 to be read—as Filaster put it—morum causā a perfectis.46 Like the variant verses from Isaia, then, which De doctri- na Christiana glosses, each member of the text network figured—so Augus- tine puts it elsewhere—in regione dissimilitudinis,47 that is, by way of simi- larity to and difference from the other works that concomitantly comprised the field—in effect a transtextual projection of Saussure’s synchronic notion of linguistic “value”.48 Thus, al-Suhrawardī’s version of Risālat al-Tayr acquires its historical significance less singularly or diachronically—that is, through an anarchic “fluidity in [its] narrative trajectory”49—than associa- ————— 38 Cf., inter alia, Genesis Rabbah 1.10f. 39 II Chronicles 36:23. 40 Pace Steiner 1998. 41 Wyrick 2004, 51-58. 42 See Gruen 1998, 177-88. 43 Cf. Deleuze and Guattari 1980. Gilles Deleuze and Félix Guattari use the term “rhizome” to designate phenomena that allow for multiple, non-hierarchical entry and exit points in data, representation, and interpretation. They oppose the rhizomatic model to arborescent forms of ordering, which work with dualist categories and binary choices. A rhizome works with horizontal and trans-species connections, while an arborescent model works with vertical and linear connections. 44 For the complexities of the corpus and translations of the texts, see Schneemelcher 1992, 2:75-482. See, further, Pervo 1986 and 2006; Bavon 1999; Molinari 2000; Klauck 2005. 45 Jerome, De viris illustribus 2. 46 Filaster, Liber de haeresibus 88. 47 Augustine, Confessiones 7.10; for the history of the phrase, see Courcelle 1963. 48 De Saussure 1980, 155-69; Trubetzkoy 1939; Todorov 1971; Jakobson 1987. Cf. Lalle- man 1998. 49 So Thomas 2003, 40-71, who accordingly—though not unusefully—stresses the “per- formative tendency” of multiform texts (“Multiforms … result from oral circulation, or from a usage of texts that views them as resources for retelling the story in another per- TEXT NETWORKS 9 tively in relation both to the earlier Arabic renditions credited to Ibn Sīnā and al-Ghazālī, as well as to Farīd al-Dīn ʿAttār’s formidable Persian recast- ing of the tale as Mantiq al-Tayr, written several decades later.50 On the hori- zon of sixth-century Islām (A.H.), all of these divergent renderings explicitly held—as T.S. Eliot maintained—“a simultaneous existence and compose[d] a simultaneous order,”51 which, insofar as the complex was autopoietic,52 also remained dynamically tensive. Or such, at all events, is the scriptural disposition that the Apocryphon of Iōhannēs stages in its opening plot frame: It happened one day, when Iōhannēs … had gone up to the temple (!!rpe < M. Eg. [r-pr]), that a Pharisee named Arimanios approached him ————— formance” [59]). She conceives this performativity, however, as external to, rather than immanent within, the text; see below p. 12. It thus seems overhasty to conclude “that it was the general line of the story, rather than the specific text at any give point [i.e., fabula divorced not only from sjužet, but from any textual embodiment whatsoever], which was the significant aspect of these works” (41); for fabula and sjužet, see Thomas 1998, 74-75. 50 For the texts, see Corbin 1954; Razavi 1997; Avery 1998, 551-59; Thackston 1999. This set makes clear that the self-organization of the text network does not coincide with genre; cf. Goldhill 2008. 51 Eliot 1975, 38. Cf. Guillén 1971: “Literature is characterized not so much by the opera- tion of full systems as by a tendency toward system or structuration. Thus it appears that the historian is led to evaluate, for every century or phase in the history of his subject, a persistent, profound “will to order” within the slowly but constantly changing domain of literature as a whole … The structures of this order are no more alien to the [narratives that the redactor] produces than the linguistic code is to the utterances of his speech” (376 and 390). 52 On autopoietic or self-organizing systems, see particularly Maturana and Varela 1975, and Luhmann 1997. Contemporary systems theory provides among the most useful mod- els for the study of text networks insofar as it recognizes both the openness of all self- constituting systems, as well as their recursive tendency towards closure—cf. Pankow 1976, 16: “All living systems and all supersystems which are built from living systems are open systems. They are open with respect to the matter, energy, and information which they exchange with their environment”; Jantsch 1979, 65: “Autopoiesis has been called a kind of system dynamic through which the system regenerates its own compo- nents and thus itself. It implies closure.” Literary text networks can thus be seen as auto- poietic insofar as they remain open, yet constitute themselves through a closed (i.e., cir- cular) organization of scribal processes created through the recursive interaction of its own products; see also Altmann and Koch 1998, an excellent collection of articles on autopoietic systems in language, literature, and culture. For the “exorbitance” which is concomitantly endemic to such forms of structuration, see Derrida 1967, Barthes 1973, de Man 1979, Austin 1975 (on “perlocutionary” effects), Luhmann 2001. Also useful, in this connection, for a poetics of the multiform are Starobinski 1971 and Riffaterre 1978. 10 DANIEL SELDEN and said, “Where is your master (cah < M. Eg. [sš]), whom you used to follow?” He said to him, “He has gone back to the place from which he came.” The Pharisee said, “With deception (plany [< πλανή]) this Nazarene deceived you (af-plana [< πλανᾶν] !mmwtn [pl.]), and he filled your ears with lies, and closed your hearts, and turned you from the traditions (paradocic [<παράδοσις]) of your fathers.” When I heard these things, I turned away from the temple (hieron [< ἱερόν]) to the mountain, a desert place (toou !njaeie). And I grieved greatly in myself, saying, “How was the savior appointed, why was he sent into the world. . . and who is his father, and of what sort is the aeon to which we shall go?”53 The Coptic cah, literally “scribe”, in conjunction with the Greek derivative paradocic, a written document or, in a more technical parlance, Rabbinic interpretation,54 suggests that what the Apocryphon stages here is an allegory of its own textual divarication. In his ascent to the tabernacle, Iōhannēs en- counters a Pharisee—that is, a strict adherent to the doctrine of the dual Torāh55 [< Heb. pārush, “separated” (i.e., unto a life of purity)]—who, bear- ing the ominous Zoroastrian sobriquet “Arimanios”, 56 taunts him for having swerved (af-plana) from the “accounts of his Fathers,”57 at which point Iōhannēs turns away from the House of YHWH to stake his ground in unoc- cupied territory on the margins of the cultivated fields (toou !njaeie).58 In the code-switching between Egyptian and Greek that typifies the Coptic idiom,59 the scriptural repository that Iōhannēs first approached under native nomenclature as !rpe, he leaves behind him as the now alien ἱερόν. When at this point the heavens open, the Savior appears to Iōhannēs, and recounts to him a complex variant of the classic Gnostic myth,60 the text proceeds the- ————— 53 Waldstein and Wisse 1995, 12-15. 54 See Lampe 1961, s.v. παράδοσις. 55 See Neusner 1995-2000, 1:117-216, and 2:35-62; Neusner and Chilton 2007. 56 Phiroze Vasunia has suggested to me in personal communication a potential pun on πέρσης behind varicaioc, a possibility not to be ruled out considering the wild syncre- tisms found in contemporary voces magicae; see, for example, Versnel 2002. There is a comparable pun, derived from Old Persian, in the Coptic Cambyses Romance; see Kam- merzell 1987. 57 Cf. The chapter “Clinamen, or Poetic Misprision” in Bloom 1973, 19-45. For further details see Bloom 2003. 58 On the geographic symbolism in the Egyptian context, see inter alia Assmann 2000, 217- 42; O’Connor and Quirke 2003. 59 See Reintges 2001. 60 Layton 1987, 5-22.