Explanations of synoptic patterns of variation and agreement that assign to all or some of the tradition a direct oral provenance, or, alternatively, expand the number of written sources, have never been lacking in scholarship. Solutions of this sort have experienced a revival, their current advocates arguing not only that redaction-critical explanations have proven inadequate but that progress in research on orality and writing in the ancient world gives new grounds for considering oral or multiple-source hypotheses. James Dunn reargues the case for distributing the double tradition into oral (‘q’) and written (‘Q’) sources.Footnote ¹ Moving beyond Dunn's binary model, Terence Mournet argues that given the complex interrelationships of oral and written media any solution to the synoptic problem will necessarily be complex: the synoptics emerged from a plurality of oral and written sources.Footnote ² In a revival of the oral gospel hypothesis Armin Baum conceives both the double and triple tradition existing as a memory-borne body of oral tradition that Mark, Matthew, and Luke independently committed to writing.Footnote ³ Delbert Burkett in his recent two-volume analysis attributes the patterns of variation in the synoptic tradition to a prolonged history of scribal compilation of multiple written sources.Footnote ⁴

These utilization hypotheses occupy very different positions on a media spectrum defined by orality at one pole and writing at the other. Though Dunn, Mournet, and Baum appeal to distinctive features of ancient orality, and Baum programmatically invokes the factor of memory, their accounts suffer from inchoate or flawed media understandings. These include a media dualism that in its strong form considers writing and orality to be virtually autonomous modes: variability is viewed as the index property of oral tradition, whereas writing supposedly gives the tradition a fixed form subject only to copying and editing. Interaction between written and oral media if posited is done so inarticulately and ad hoc. Burkett in his utilization account assigns to oral tradition no significant role at all; his history of the tradition is almost uniformly a history of written sources and writing.

I

In practice Dunn separates oral and written lines of transmission, invoking oral tradition to account for high-variation parallels that resist reduction to redaction-critical explanations.Footnote ⁵ This media dualism finds expression in his Q/q partition of the double tradition, which with different nomenclature returns to the source-critical solution proposed by Hawkins, Streeter, Taylor, and others who when levels of variation exceeded what might plausibly be explained by editorial operations upon a written source assigned such pericopes to oral tradition.Footnote ⁶ Lacking a model for the coordination of orality and writing in gospel composition, Dunn runs into difficulties stipulating the threshold of variation for assigning a pericope oral rather than written provenance, given agreement rates that move incrementally from 8% to close to 100%. Unwittingly he also frames the synoptic data in a question-begging way. A distinctive of the written medium is its capacity to establish intelligible connections and hence a durable order out of heterogeneous materials. For his analysis Dunn breaks up the double tradition into isolated pericopes, thereby destroying their sequence and thus possible evidence of documentary provenance.Footnote ⁷

In his 2005 monograph Mournet worked mostly within Dunn's framework. Though in his 2009 essay he still refers to the ‘orality model’, he abandons Dunn's simple media dualism.Footnote ⁸ However his assertion that the complexity of ancient media realities will entail a complex solution to the synoptic problem, in addition to being overly axiomatic in approach, is a non sequitur. While Mournet acknowledges that oral and written traditions existed in ‘complex interaction’,Footnote ⁹ like Dunn he lacks a practicable account of their interface and thus risks falling back into a de facto media dualism. This explains his default to a complex, multi-source solution to the synoptic problem, for in the absence of a model for their interaction, orality and writing appear to fall into discrete media strands.

II

Baum's starting point is likewise the difficulties simple documentary utilization hypotheses face accounting for the complex profile of synoptic variation. This is not just a matter of the low-agreement parallels, but the remarkable range, or to use his word, inhomogeneity of patterns of agreement. In many cases, he argues, no redactional or stylistic rationale can be made out for variation in the triple and the double tradition.Footnote ¹⁰ On the other hand the synoptic profile corresponds to characteristic features of oral tradition and—what for Baum will amount to the same thing—to features, identified in experimental studies, of materials reproduced out of memory. Accordingly Baum argues that an oral gospel hypothesis, refounded upon contemporary research in orality and memory, elegantly accounts for synoptic relationships: Mark, Matthew and Luke draw their double and triple tradition materials independently from a common body of oral tradition.

It is understandable that the shortcomings of the copy-and-redact model that prevails in much synoptic scholarship might make Baum skeptical of the standard documentary hypotheses. He is also on the mark in identifying orality and memory as determining factors in synoptic relationships. Nevertheless the media conceptions that push Baum toward an oral gospel hypothesis are flawed and ironically the same that operate in scholarship on the hypotheses he contests. Baum takes close, visually executed copying with some margin for editing and redaction (Abschreibevorgang) as in fact the mode in which source relationships are enacted between documents. Hence he finds the unpredictably varying levels of agreement in the synoptics inexplicable on any theory of literary dependency, for why would a source be copied at randomly different rates of agreement?Footnote ¹¹ This assumption explains why Baum deems that demonstrating the inadequacy of the copying procedure, so understood, to encompass the synoptic data per se rules out the viability of the Two Source Hypothesis.Footnote ¹² With most scholarship on the synoptic problem he therefore recognizes no leading role for memory in source-utilization—in mediating written relationships—rejecting as incredible even the very limited case of some variation arising from scribes having to hold sequences of text in short-term memory as their eyes shift from exemplar to copy.Footnote ¹³ This leaves him with no way to conceive how variation might enter into the transmission of written tradition beyond the narrow margin for editorial modification and in his actual assessments forces him into categorical determinations of oral or written provenance. This media dualism is also evident in his association of memory exclusively with oral tradition and oral transmission: Baum treats ‘oral tradition’ (mündliche Überlieferung) and ‘memory activity’ (Gedächtnistätigkeit) as equivalent, mutually entailing terms and construes evidence of mnemonic features and mnemonic processing in a text as ruling out its written transmission.Footnote ¹⁴ This confounding of individual memory functionality with oral tradition processes accounts for the evidentiary weight Baum accords to experimental studies on memory. Viewed within this theoretical framework oral tradition's variability arises not so much from its cultivation and transmission in living contexts as from the mechanics and vagaries of individual memory functionality.

In his search for pertinent source-utilization analogies Baum rightly rules out the Hellenistic historians, including Josephus, whose work was controlled by the Stilideal that sources be paraphrased to a stylistic homogeneity, a mode of variation that bears little resemblance to synoptic patterns.Footnote ¹⁵ Instead, he finds his premium analogy in the Chronicler's utilization of 1–2 Kings, as well as in the use of 2 Kings in Jeremiah 52 and Isaiah 36–39, for ‘zwischen denen eindeutig eine literarische Abhängigkeit besteht’.Footnote ¹⁶ Aggregate statistics for these Hebrew Bible test cases give an agreement rate of roughly 80%, significantly above the average synoptic rate of 42%.Footnote ¹⁷ Baum concludes it is therefore unlikely that synoptic source relationships were mediated through writing.Footnote ¹⁸ However this inference is questionable on several counts. In effect Baum takes the use of historical sources by the Chronicler and the prophet books as normative for written source utilization in general and therefore the 75–80% minimum agreement rate, along with the small margins of variation from the mean, as the metric for defining the outer limit for variation in written source dependency and hence for making categorical determinations of oral or written provenance.Footnote ¹⁹ More ominously for his hypothesis, Baum's quite correct assertion in other contexts that scribes (among whom presumably belongs the Chronicler) often knew the Hebrew Bible by heartFootnote ²⁰ belies his attaching of memory activity (Gedächtnistätigkeit) exclusively to oral tradition and oral transmission. This is not just a matter of the Chronicler no longer being indisputably a model for close visual copying of sources. If written transmission occurs out of a memory-based competence instead of or in tandem with visual copying, in principle it is open to the forces of variation that Baum associates with memory and oral tradition. In other words their common basis in memory suggests that the boundary between written and oral tradition might not be impermeable.

Further, Baum is led by his identification of memory mediation with oral tradition to follow Gerhardsson in taking memorization and recitation of written texts as the template for the existence and transmission of oral tradition in memory. Baum assimilates oral tradition and memory to the norms of the written medium, more precisely, to the memory mastery of written texts that was the distinguishing marker of elite—and literate—educational and scribal settings, the case in point being the memorization of Torah and Mishnah in rabbinic circles. The Mishnah in its definitive features in fact approximated to a written, redacted text, and its mastery and cultivation in memory was strategic and prescriptive.Footnote ²¹ Along with Gerhardsson, however, Baum takes the practices used in cultivation of the Mishnah as normative for oral tradition and its transmission; accordingly he aligns the mastery of the synoptic tradition by Jesus’ disciples with scribal and rabbinic memorization practices associated with the Mishnah and the scriptures. To extrapolate from these rabbinic practices he must venture (with the earlier Gerhardsson) the widespread existence in Jewish Palestine of educational institutions where as children Jesus’ disciples would have learned these text memorization practices that as adults they then applied to oral transmission of the Jesus tradition.Footnote ²² Besides the fact that Catherine Hezser has debunked much of the purported evidence for this educational system,Footnote ²³ Baum's alignment of the transmission of the oral tradition with these text-centered memorization practices forces him to posit extrinsic incentives for the disciples to have committed it to memory. These include their reverence for Jesus, but more puzzlingly, the need to compensate for the tradition's not being written down (‘solange…ihnen die Lehre Jesu lediglich mündlich vorlag’),Footnote ²⁴ which equates to the curious claim that the precariousness of its existence as oral tradition induced the disciples to memorize it and thereby put it in the mode (on Baum's definition) of oral tradition. At any rate it is hard to square the memorization premise with the higher, uneven levels of variation Baum regards as indicative of orally mediated relationships.Footnote ²⁵ The most salient point for our purposes though is that in consequence of this text-memory model for oral tradition Baum will project the properties of the written medium upon his oral gospel.

Baum's casting of the oral in the image of the written explains his interest in the experimental literature on memory. Research on the cognitive operations of memory shows that oral tradition indeed amounts to an efficient memory system, a collection of mnemonic strategies for transmitting crucial cultural information where the prevailing medium is orality.Footnote ²⁶ However the experimental studies Baum adduces to elucidate variation in oral tradition feature subjects’ reproduction by memory, at intervals, of written texts. Variation thus construed is a matter of deviation from a fixed standard, with the experimentally derived patterns of variation, statistically matched to the synoptic profile, declared by Baum unimpeachable evidence that synoptic relations were mediated through a common body of oral tradition and not by writing.Footnote ²⁷ In effect he takes as paradigmatic of an oral relationship what is in fact a literary relationship between the fixed text of the experiment and a subject's written reproduction of that text by memory at some interval. Baum's identification of memory instrumentality with oral tradition effectively defines the latter as any text mediated by memory.

It is on this point that additional incongruities become apparent in Baum's analogies. He invokes memorization practices connected with the Hebrew Bible and the Talmud to argue the synoptic tradition was memorized, but to explain synoptic patterns of variation he adduces experimental literature that has subjects reproducing from memory unfamiliar texts read or heard just a few times or sometimes only once. In accord with his text-centric model of oral tradition this leads Baum to a Gedächtnisfehler account of variation, arguing, for example, that the close citation of scripture in the synoptic tradition, which contrasts with high variation levels in the ambient tradition, is due to the disciples having learned the Scriptures as children, for experimental studies show that material learned in childhood is more easily and enduringly fixed in memory than material learned with greater difficulty in adulthood.Footnote ²⁸ In other words synoptic variation reflects imperfect memory mastery of the tradition. This is at odds with Baum's memorization model for the synoptic tradition, grounded in the cultural practice of memorizing great quantities of scripture, it being an easy matter therefore for Jesus’ disciples ‘to learn the tradition by heart’.Footnote ²⁹ The suggestion that high variation is caused by the absence of the synoptic tradition ‘in schriftlicher Form’ similarly makes variation a function of the limitations of memory.Footnote ³⁰

To be sure, a great deal of variation in the transmission of tradition is due, as experimental studies confirm, to memory's functional strengths and weaknesses. Baum fails though to recognize the performative forces at work in the transmission of tradition—its cultivation in living connection with the contemporary realities of its tradent community.Footnote ³¹ This puts him in the position of having to bring virtually all variation under the Gedächtnisfehler rubric.Footnote ³² Baum sees an analogy to oral tradition in F. C. Bartlett's finding that memory is better at recalling the content (Inhalt) than the wording (Wortlaut) of a narrative. But as Baum himself notes, Bartlett observed that a narrative recollected by subjects over a number of intervals assumed an increasingly stable form.Footnote ³³ Other factors than just forgetting or shedding details of a story must therefore be at work producing variation in an oft-repeated and widely shared oral tradition.

Baum applies his statistical analysis to parallel performances of oral epic and finds in rate and range of variation a correspondence to the synoptic tradition. The analogy is not without significance; oral-traditional forces of the sort Parry and Lord observed in the cultivation and transmission of Yugoslav epic are definitely at work in the synoptic tradition. But the genre connections between oral poetry and the synoptic tradition are tenuous at best. The analogy also conflicts with Baum's text-centered model for memory and oral tradition, for a definitive feature of oral poetry, a genre emblematic of pre-literate societies, is its cultivation without reference to a written text. Baum's collocation of incompatible analogies occurs because he considers Gedächtnistätigkeit and mündliche Überlieferung practically synonymous. On this view any material mediated by memory counts as oral tradition. Thus Baum can adduce scripture memorization, scribal memory practices, experimental studies of memory, and oral epic indiscriminately in support of his oral gospel hypothesis.

These problems migrate into Baum's application of his findings to the synoptic problem. Common order of pericopes (for the Mark/Matt and Mark/Luke tradition also nine out of ten sentences in common order),Footnote ³⁴ especially in the triple tradition, is most familiar as a leading argument for documentary solutions. Baum nevertheless contends that the phenomenon is not incompatible with Mark, Matthew, and Luke independently of one another drawing from a body of oral tradition transmitted by memory.Footnote ³⁵ While the materials of the Passion Narrative are organized in a narrative logic that facilitates recollection, Baum faces greater difficulties accounting on his oral hypothesis for the common order of the self-contained, heterogeneous pericopes that make up much of the synoptic tradition and in which narrative ordering forces of the sort found in the Passion are attenuated or absent. Baum appeals to parallelism and repetition, but these mnemonic devices operate on a small scale, mostly within a pericope or over short groups of pericopes. Baum in fact acknowledges their weak ordering effects as the reason most pericope inversions are found outside the Passion Narrative.Footnote ³⁶ Sayings materials are especially prone to enter into fresh combinations, not being bound together by any narrative or immutable deliberative logic. Moreover, if the writing Evangelists changed the order of pericopes, as for instance Matthew does with Markan sequences in Matthew 8–9, it is difficult to view a body of oral tradition, on Baum's own accounting inherently susceptible to variation, as able to maintain for any significant duration the common order manifest in the triple tradition. The only way to appear to circumvent this problem is to assume a fixed order for the oral gospel, in other words, to impute to the oral gospel the properties of the written medium, that is, documentary properties. This is precisely what Baum does, conceiving the oral gospel in the image of oral tradition he has created out of case studies featuring memory of written texts.Footnote ³⁷ Baum's oral gospel is in fact a phantom written gospel.

Baum conceives the tradition circulating for decades in this stably ordered cohesion, likening the tradents of the oral gospel to scribes and rabbis who could carry even larger texts in memory.Footnote ³⁸ Eventually one of these tradents, Mark, put the oral gospel down in writing, an operation Baum not surprisingly describes as a niederschreiben. The Gospel of Mark, ‘die schriftliche Fassung des Markusstoffs’, mirrors its oral exemplar, and a two-track transmission of the gospel ensues, one written, one oral.Footnote ³⁹ In its subsequent transmission the oral version undergoes additional condensation, dropping of details, and word changes, phenomena attested for seriatim memory reproductions of written texts by experiment subjects. Hence when Matthew and Luke later write down the oral gospel their versions differ from Mark's Gospel in these respects.Footnote ⁴⁰ It is because the oral gospel behaves like a written text that Baum can thus accommodate Markan priority within the oral gospel hypothesis. Matthew and Luke's direct dependence on Mark is retained, ingeniously shifted to the oral Markustradition. Footnote ⁴¹

The double tradition, notorious for its disagreements in order, is more difficult to accommodate, to say the least. Baum has Matthew and Luke drawing their double tradition materials ‘ebenfalls aus derselben mündlichen Tradition’, referring in context to the ‘mündliche Markustradition’, that is, the oral gospel.Footnote ⁴² However he does not address what seems the improbability that the Evangelist Mark omitted oral gospel material coincidentally picked up independently of him and of each other by Matthew and Luke, and that just this subset of material was reorganized relative to the triple tradition by one or both of the latter. The more pressing difficulty though is that the double tradition's divergences in order are anomalous on Baum's postulate of a more or less stably ordered oral tradition transmitted as such by memory for decades. Or to view the matter from another angle, a reshuffling of materials by a writing Evangelist puts Baum in the untenable position of attributing greater stabilizing effects to the oral than to the written medium. Earlier Baum had called attention to the disordered double tradition and its disjunction in this respect from the triple tradition.Footnote ⁴³ Caught however between an acutely variant double tradition and an oral gospel defined by text-controlled memory practices Baum simply assimilates the double tradition to his oral gospel without spelling out his reasons for doing so.

These problems in Baum's analysis can be traced back to problems in his understanding of ancient media and their interrelations. Though we can outline it only briefly here, an approach predicated upon the active interface of the oral medium and the written medium can better account for synoptic patterns of variation and agreement. The written medium, in virtue of its material, spatial, and visual properties, solidifies the tradition in extended sequences of words, sentences, and pericopes. For its part the characteristic variability of oral tradition arises from its transmission, not in the stable artifactual form of written texts, but through its live utterance (its ‘performance’) in the immediacy of face-to-face contexts of reception. Antiquity was a mixed media culture: written texts circulated and were utilized in an environment of prevalent orality, and the boundary between orality and writing was indistinct. Written works were calibrated for the ear; often they were composed out of memory and by dictation, and not only in recitation but also in private reading they were reconverted to oral utterance. Perhaps more difficult for us to grasp is the extent to which antiquity was a memory culture: oral and written tradition were grounded in memory, with the manuscript serving as the external support for the primary existence of the work in memory. Because of its intimate interface with orality, written tradition, without forfeit of its capacity to consolidate materials in small- and large-scale textual sequences, might be opened up to oral enactment practices, that is, to variation.Footnote ⁴⁴ This means that variation and memory mediation could leave deep and indelible marks on, indeed, be constitutive of, documentary relationships. Within this framework Q and Mark can be seen to lead their existence at the boundary of the oral and the written medium, but with their material, written forms the sustaining substratum of their existence and utilization. This explains the persistence, amidst often significant pericope-level variation, of a common order in the triple tradition, of more than just fragments of a common order in the double tradition, and of agreements in words, word order, and sentence order in many triple and double tradition passages.

III

Burkett echoes Mournet in his belief that the ‘simpler’ literary utilization hypotheses ‘do not work’ and Baum in asserting that ‘[n]o one Synoptic served as a source for either of the other two’.Footnote ⁴⁵ However, in a manner reminiscent of the ‘complex solution’ and ‘criss-cross copying’ of Sanders and Davies and the ‘théorie des niveaux multiples’ of Boismard,Footnote ⁴⁶ Burkett resolves patterns of variation and agreement in the synoptics back into multiple written sources that over an extended history of scribal copying and compilation accrued into the Gospels. Oral tradition appears in this history mostly as pre-literary placeholder. Burkett has Mark combining two overlapping sources shared with Matthew and Luke: Proto-Mark A and Proto-Mark B, themselves separate revisions of a Proto-Mark, conflating these with ‘three minor sources’, a C source and a miracle-mission source (both shared with Matthew and Luke), a parable discourse (shared with Luke), and some ‘units of tradition’ (medium unspecified), while adding ‘a layer of redaction to the whole’.Footnote ⁴⁷ Burkett's analogy for this procedure is Tatian's gospel harmony, and his scenario of scribes copying, harmonizing, and making accumulative small-scale alterations to sources organized in intersecting stemmata approximates the history of the synoptic tradition to the transmission of manuscripts.Footnote ⁴⁸ Patterns of variation others might explain by invoking redaction or oral tradition are for Burkett the cumulative effect of this history of compilation and conflation of written sources.

The backbone of this articulation of sources is Proto-Mark in its three editions. Burkett's utilization hypothesis thus balances upon a master premise, namely, a late ‘layer of redaction’ in Mark, unknown to Matthew and Luke and thus analogous to the M materials. Like Mark, therefore, Matthew and Luke can only have drawn their triple tradition from Proto-Mark A and Proto-Mark B, the subsequent editions of Proto-Mark.Footnote ⁴⁹ On close examination, though, the evidence for the late layer of Markan redaction turns out to be suspect. Burkett claims for Mark special usages of πολύς that are absent from Matthew and Luke, for example the seven occurrences of πολύς and its cognates in Mark's Healing of Jairus's Daughter (Mark 5.21–43) that are missing in Matthew and Luke's versions (Matt 9.18–26//Luke 8.40–56).Footnote ⁵⁰ However the ὄχλος πολύς that meets Jesus upon his return from across the sea at the beginning of the Jairus episode in Mark 5.21 drops out of Matt 9.18 because in Matthew 8–9 Jesus’ return from the Gadarene exorcism now leads directly into the Healing of the Paralytic (Matt 9.1–8//Mark 2.1–12). Accordingly Matthew's version of the episode (Matt 9.18–26) now follows directly upon the Feasting, Fasting, and Wineskins sequence (Matt 9.14–17//Mark 2.18–22) and the simple MattR connective ταῦτα αὐτοῦ λαλοῦντος αὐτοῖς. From Burkett one receives the impression that the other omissions of πολύς in this pericope are likewise anomalous lacunae in an otherwise copied text, but in fact they occur in the course of Matthew's drastic reduction of the pericope (140//374) that shears away whole sections of Mark's pleonastic descriptions in support of Matthew's Christological narrowing of focus upon Jesus and his disciples. Luke's omissions in this pericope occur likewise in the course of condensing Mark (285//374) or in the syntactic restructuring of sentences. In other words they occur, not singularly, but in other sorts of intelligible rewriting operations upon the Markan material—for example modifying ὄχλος πολύς to οἱ ὄχλοι (Mark 5.24//Luke 8.42; see Mark 5.31//Luke 8.45)—rather than as with the M materials making an anomalous set of substantive omissions while otherwise reproducing Markan sequences.Footnote ⁵¹

The parallels to ὁ δὲ ἐξελθὼν ἤρξατο κηρύσσειν πολλὰ καὶ διαϕημίζειν τὸν λόγον at the conclusion of the Healing of the Leper (Mark 1.45//Matt 9.31//Luke 5.15), cited by Burkett as a case of the absence in Matthew and Luke of Mark's adverbial use of πολλά, are similarly instructive.Footnote ⁵² In Matthew the Markan episode (Mark 1.40–45) is foreshortened, closing with Jesus’ command to go and see the priest (Matt 8.4//Mark 1.44). The parallel to Mark 1.45a appears at the conclusion of the Healing of Two Blind Men (Matt 9.27–31) in an enhanced function as an inclusio demarcating the entire Matthew 8–9 Messiah of Deed section and anticipating the Matthew 10 Mission Instruction: οἱ δὲ ἐξελθόντες διεϕήμισαν αὐτόν (Matt 9.31). This use of Mark 1.45 is so conspicuous an element of Matthew's programmatic arrangements of the Markan materials in Matthew 8–9 it is hard to understand why Burkett regards the failure of Matt 9.31 to reproduce woodenly Mark's πολλά to be decisive for illuminating the Matt-Mark source relationship.Footnote ⁵³ For his part Luke has ὁ λόγος as the subject rather than with Mark the object of the sentence, part of an alternative Lukan construction beginning with διήρχετο δὲ μᾶλλον ὁ λόγος περὶ αὐτοῦ and continuing into a mass healing scene, and so here also Burkett's solitary fixation upon Mark's πολλά is curious. But to back up somewhat, our discussion of Matthew's use of Mark 1.45 in Matt 9.31 draws attention to the triple tradition parallels to the constructed M pericope, the Healing of the Two Blind Men (Matt 9.27–31), which are found in Mark 10.46–52, Luke 18.35–43, and Matt 20.29–34. Burkett identifies πολλοί in Mark 10.48, absent from Luke 18.39 and Matt 20.31, as an element of Mark's late ‘layer of redaction’. However in the same verse Luke replicates Mark's adverbial use of πολύς in πολλῷ μᾶλλον ἔκραζεν. Burkett limits the adverbial use of πολύς characteristic of the purported redactional layer to ‘neuter accusative used adverbially’,Footnote ⁵⁴ but as with πάλιν below, this is special pleading, a preemptive adding of narrow qualifiers until an exclusively Markan subset is defined and declared an element of the late layer of Markan redaction.

That a distinctive Markan use of πάλιν to coordinate external connections among pericopes, or more precisely, ‘to indicate the repetition of an action that occurred in a previous pericope’Footnote ⁵⁵ (54% of all cases of πάλιν in Mark), should all but disappear in Matthew and Luke is not surprising given that the occurrences in question are by definition to be found at the Markan seams in service, moreover, to Mark's paratactic style of sequencing (thus they might be better construed as elements of the Gospel's formative redaction). Matthew and Luke's appropriation of Mark into their own literary enterprises entails extensive recasting of pericope introductions and connections. This explains why a number of the uses of πάλιν in Mark to establish connections among internal pericope elements survive and are attested for Matthew (Mark 12.14//Matt 21.36; Mark 14.39–40//Matt 26.42; Mark 14.69–70//Matt 26.71–73). Luke often rewrites the entire Markan introduction in Lukan style; in other words he does not register isolated omissions of πάλιν. Though not noted by Burkett, three of these fifteen occurrences of πάλιν omitted by Luke fall into the Great Omission (Mark 7.14, 31; 8.1), in two others Luke gives the Q doublet (Luke 11.29//Mark 8.13; Luke 16.18//Mark 10.10), in another he and Matthew agree in omitting all of Mark 3.20. In any case there are just eight occurrences of πάλιν in all of Luke–Acts (3/5), compared with twenty-eight for Mark and seventeen for Matthew, so Luke is hardly neutral toward the word, Burkett's claim to the contrary notwithstanding.Footnote ⁵⁶ In short the absence of πάλιν from Luke is indiscriminate; its non-occurrences have no relation to some subset of Markan usage.

Other evidence Burkett cites for the ‘late layer of redaction’ fares no better. The patterns in Matthew and Luke's omissions of Mark's periphrastic εἰμί constructions are not particularly remarkable.Footnote ⁵⁷ Burkett's argument that Matthew and Luke knew only half the Markan version of the Healing of the Epileptic Boy (Mark 9.14–29) is not persuasive.Footnote ⁵⁸ Matthew and Luke abbreviate a redundant pericope and both show awareness of motifs in the material they omit.Footnote ⁵⁹ The upshot is that a late ‘layer of redaction’ in Mark does not coalesce out of the evidence (some minor agreements in omission) adduced for it, which even taken uncontested and in aggregate does not rise to a ‘whole layer of Markan redaction’.Footnote ⁶⁰ Burkett puts it in these magnified terms because he has tied his utilization hypothesis to Mark's late layer of redaction being analogous to the M materials.Footnote ⁶¹

Matthew and Luke's omission of single words like πολύς or πάλιν is anomalous if the relevant utilization operations of copying, compiling, and conflating are regarded, as Burkett evidently does, as rather wooden in execution. This conception of the scribal functionality in question also explains the approach he takes to the problem of the major divergences from Mark's order in Matthew 8–9. Burkett parses the differences in pericope order in Matthew and Mark's parallel tradition into three sequences and identifies these as sources (Proto-Mark A; sources B and C) that Matthew and Mark shared and independently combined, ‘usually following the order of their source’.Footnote ⁶² In effect Burkett has carried out in the three requisite sequences a formal reconciliation of the divergent order of the Matthew–Mark tradition, repeating what Vincent Taylor did in fifteen sequences with the Matthew–Q tradition, and then hypostasized the three sequences into sources. Burkett makes clear that his line of inference assumes the existence of Proto-Mark, which he deduced from the late layer of Markan redaction: ‘If neither Matthew copied Mark nor Mark copied Matthew, then only one alternative remains: both copied the same source or sources. If they shared only a single source, we would not expect to find a great deal of divergence in the order of their shared material’.Footnote ⁶³ Without Proto-Mark it is a matter of Matthew in Matthew 8–9 transposing two or three sequences of Mark, for which, Burkett argues, no convincing redaction-critical explanation exists.Footnote ⁶⁴

The confidence Burkett expresses in this solution makes sense if sources were utilized somewhat woodenly, by scribes adept at compiling but not so much at composing. To fully resolve the strands of variation and agreement in the triple tradition back into parallel and overlapping documents, amalgamated in an extensive history of scribal compilation, requires a remarkable proliferation of sources. When finished Burkett has ramified the triple tradition into Matthew, Mark, Luke, Proto-Mark, Proto-Mark A, Proto-Mark B, sources B, C, K, L, M, Parable Discourse, Lord's Supper Sayings, Trial Source, Elements of a Passion Narrative, and Mission-Miracle source, in total sixteen sources.Footnote ⁶⁵ Oral tradition is for all practical purposes absent from this history; Burkett is ambivalent even about the medium of the ‘isolated traditions and collections of traditions’ to which he consigns some of the triple tradition remainders.Footnote ⁶⁶

In the Unity and Plurality of Q (2009) Burkett argues that the patterns of variation in the double tradition are likewise the outcome of scribal compilation of sources, the significant number of verbatim agreement parallels in the tradition confirming that Matthew and Luke themselves functioned ‘primarily as compilers and editors rather than as composers’.Footnote ⁶⁷ Compilation and composition, construed as categorical alternatives, take on a higher visibility in this second volume as the leading analytical binary. Burkett adopts it from a 2007 essay by John Kloppenborg that rates the scribal function in synoptic relations as largely transcriptional. While Kloppenborg recognizes that written transmission of a work likely was affected by its oral cultivation, other than noting some Greco-Roman rhetorical practices bearing on the composition and reproduction of texts he does not pursue this line of analysis.Footnote ⁶⁸ His primary interest is in rebutting Dunn and Mournet's claim that high levels of variation in a synoptic text correlate to oral sources; accordingly his approach is to map the synoptic patterns of variation and agreement onto a taxonomy of known cases of utilization of written sources in antiquity. A survey shows that these fall roughly into two groups, those characterized by ‘rhetorical paraphrase’, and those characterized by ‘wooden copying’ (a rubric that includes compilation).Footnote ⁶⁹ ‘Wooden’ is a loaded term, weighted toward a certain conception of the scribal procedures in question; nevertheless Kloppenborg binds his analysis to these pronounced alternatives. Rhetorical paraphrase is ruled out as the mode of Matthew and Luke's source-utilization, for such would leave few close agreements in the double tradition. This leaves as the pertinent analogy the case studies grouped around ‘wooden copying techniques’.Footnote ⁷⁰ Matthew and Luke's ‘transcription of Q’ thus reflects scribal competence at the level of the librarius, defined as ‘a copyist, transcriber of books’,Footnote ⁷¹ an occupational cohort to which Kloppenborg attributes a clerical level of competence of the sort that in addition to transcription includes the ability to craft administrative documents such as ‘petitions, deeds, leases, and loans’, a functionality characteristic of the ‘hundreds of lesser scribes servicing towns and villages’.Footnote ⁷² Kloppenborg has long maintained that Q's provenance lies among village- or lower-level administrative scribes; his analysis here leads him to locate the Evangelists among this group as well.

Kloppenborg's taxonomy is illuminating: besides showing that ancient authors were perfectly capable of using their written sources inconsistently it leaves little doubt that Matthew and Luke conceive their task as being one of copying and compiling their sources. The difficulties lie in his inferences to level of scribal competence and provenance. Neither the ‘wooden copying’ nor the ‘rhetorical paraphrase’ case studies he adduces display the curiously mixed patterns of variation and agreement emblematic of the synoptics.Footnote ⁷³ These strains upon the analogy, as well as the categorical construal of the compositional alternatives, induce Kloppenborg finally to equivocate, to impute both ‘wooden copying’ and ‘rhetorical paraphrase’ to the scribes of the double tradition.Footnote ⁷⁴

For his part Burkett seizes on Kloppenborg's approximation of synoptic source utilization practices to ‘wooden copying techniques’Footnote ⁷⁵ and sees in his own hypothesis of multiple sources the obvious solution to the conundrum of high variation passages that forces Kloppenborg's awkward attribution of ‘rhetorical paraphrase’ to low-ranking scribes.Footnote ⁷⁶ Burkett finds validation and a prototype for his approach in Mark-Q overlap passages such as the Beelzebul Accusation, where the patterns of variation and agreement demonstrably arise from Matthew's conflating Mark (Proto-Mark A) and Q where Luke follows Q.Footnote ⁷⁷ This and similar compiling operations, such as Matthew's and Luke's different strategies for attaching Q material to Markan contexts, he argues, account for much of the variation in the double tradition.

Burkett posits a unified documentary Q that encompasses most of the double tradition, identifying a core sequence of eighteen pericopes in more or less common absolute order. To this he adds a number of additional units on the basis of motif affinities with the material in this core.Footnote ⁷⁸ Because it lacks these affinities he designates a double tradition sequence ‘On Confession’, constituted of identically ordered but in Matthew and Luke differently positioned elements, a distinct source. In view of pervasive disagreements in wording this source likely descended to Matthew and Luke as such in variant versions.Footnote ⁷⁹ Burkett echoes Baum in his opinion that redaction-critical explanations of high-variation parallels must impute source-utilization inconsistency to Matthew and Luke.Footnote ⁸⁰ Instead (with the exception of ‘On Confession’ above) variation in the double tradition arises from Matthew and Luke combining Q with overlapping sources. Since Mark–Q overlaps are limited, and since Matthew and Luke ‘were functioning essentially as compilers’, Burkett must postulate that M and L are written sources that in addition to single-attestation M and L passages contained M and L versions of numerous Q pericopes.Footnote ⁸¹ Here Burkett takes up and extends the expedient adopted by Streeter and Taylor, who in cases of double tradition variation exceeding what could plausibly be accounted for on editorial grounds conjectured an M source with overlap passages.Footnote ⁸² In Matt 7.15–20, for example, Burkett has Matthew removing ‘for a good tree does not bear bad fruit, nor again a bad tree bear good fruit’ from its position as the opening premise of the Q version (Luke 6.43–45) to adjust it to the order of an M version, for Matthew ‘favored the order of M and adapted the Q material to M’.Footnote ⁸³ Matthew–Luke agreements in Judge Not give the substance of the Q version while uniquely Lukan elements such as ‘do not condemn and you will not be condemned’ come from Luke's conflation of Q and L versions.Footnote ⁸⁴ Matthew draws his Non-Resistance parallel (Matt 5.38–42) from M and conflates M and Q versions of the Golden Rule (Matt 7.12), giving preference to the ‘fixed place’ the latter held in M over its position in Q, whereas Luke gives the Q versions in both wording and probable position (Luke 6.29–30, 31).Footnote ⁸⁵

That Matthew and Luke are scribal tradents whose task it is primarily to consolidate the tradition in its received written forms through compiling and re-contextualizing materials thus brought into living conjunction, and that much variation is due to these operations, is not in dispute. The difficulties in Burkett's analysis are caused by his partial-spectrum view of ancient media. Though he gestures toward the possibility that M and L are oral tradition, or, more indeterminately, a ‘body of materials’ or ‘strands of the Gospel traditions’,Footnote ⁸⁶ Burkett's scenario of scribes woodenly compiling and conflating requires that M and L function de facto as written sources equivalent in documentary and textual coherence to Mark and Q. Their elevation to the status of written sources on the order of Mark is therefore ad hoc; Burkett subjects M and L materials to none of the proofing (coherence in order and conception) he applied with all due methodological rigor to the Q materials to establish their documentary integrity.Footnote ⁸⁷ Its dependence upon such contrivances leads the larger hypothesis deeper into problems. As clinching evidence that M was a written source Burkett argues that Matthew's transposition of the Golden Rule from Q 6.31 to Matt 7.12 is explicable only if Matt 7.12 was the ‘fixed place’ of the M version of the saying ‘in a document parallel to Q’.Footnote ⁸⁸ But if Q 6.31 is displaceable, despite on Burkett's own telling being embedded at a sequential position in a written document Q—in a ‘fixed place’ to borrow his expression—then its position in Matt 7.12 can tell us little about whether M was a written source or indeed whether an M version of the Golden Rule ever existed.Footnote ⁸⁹ We see the deleterious effects of Burkett's ‘wooden copying’ conception of synoptic scribal functionality: that the Golden Rule is differently positioned in Luke and Matthew can only be due to its being differently positioned in two distinct sources. The placement and wording of an element in the Gospels reflects the influence of a particular source; double tradition material found in Matthew with different placement or wording indicates the influence of M.Footnote ⁹⁰ Burkett's ancillary assertion that Matthew ‘favored the order of M’, fitting Q materials to the M source much as he inserts Q materials into Mark,Footnote ⁹¹ entails the irony that M, a phantom source, one for which Burkett brings no evidence of its documentary reality, plays a role equivalent to Mark in Matthew's use of Q—a source whose documentary integrity Burkett established by careful analysis. Moreover Matthew's exigency in the case of Mark and Q is to fit Q sayings aptly to a narrative framework; M has no narrative framework that Burkett has shown.

As we noted earlier, Burkett attributes variation in the material ‘On Confession’ to what amounts to an On Confession^Matt and an On Confession^Luke, but he rejects the Q^Matt and Q^Luke expedient invoked in some Two Document Hypothesis scholarship because he finds it anomalous these should contain high agreement alongside high variation parallels.Footnote ⁹² Here Burkett's media assumptions appear with particular clarity: given wooden compiling practices it is variation that individuates sources and plural sources that account for variation; high-agreement parallels indicate Matthew and Luke are copying the same source.Footnote ⁹³ Against this backdrop the incongruities in Burkett's utilization model become quite forcibly evident, for there are a number of occasions he allows the Evangelists a rather free hand in redaction. Typically he does so ad hoc, to resolve cases of variation that anomalously remain after a thorough application of the compile-and-conflate model. ‘The Evangelists’, he writes, ‘sometimes made changes to the wording of Q that were not necessitated by the influence of some other source or context… [T]hey clarified the meaning of Q, in the process becoming interpreters as well as editors. In yet other cases they adapted Q to make it relevant to the situations of their own communities’.Footnote ⁹⁴ Examples include Matthew's editing of the Sign of Jonah ‘to express his interpretation of it as a reference to Jesus’ resurrection’,Footnote ⁹⁵ and Matthew and Luke's different rendering of certain elements of the otherwise high-agreement John the Baptist sequence.Footnote ⁹⁶ Some triple tradition elements cannot be aligned with the three sequences of Proto-Mark A, and sources B and C, ‘probably because either Mark or Matthew disturbed the sequence by moving the material from its original place in one of the sources’.Footnote ⁹⁷ Mark composed in his own hand a significant expansion of the Exorcism story (9.14–29) and in other places ‘revised and expanded the material that he had received’.Footnote ⁹⁸ It is difficult to see that Burkett in permitting significant redaction even in a limited number of cases has not cut the ground out from under his utilization hypothesis, there being no reason in principle that redaction should not be invoked in other cases of variation in order and wording. In any case, compilation and conflation demanded such high levels of skill and ingenuity (as Burkett shows) that it seems arbitrary to deny that the range of scribal competence in question extends to redaction. Acknowledging this has the advantage of validating many of Burkett's analyses in detail while disencumbering the utilization account of its multiple sources. It also takes care of the difficulty that he is limited to parsing variation synchronically into a panoply, and diachronically into a regress, of separate sources. He cannot account for the origins of tradition variants, such as doublets, in the first place.Footnote ⁹⁹

IV

Despite the problems in his utilization hypothesis, Burkett has put his finger on important features of scribal agency in the history of the tradition, chief among these that Matthew and Luke's source utilization amounts to a programmatic consolidation of the tradition. There can be little doubt that Mark and Q likewise are the scribal artifacts of consolidating enterprises, in their case of mostly pre-documentary tradition. These large-scale scribal projects entailed modifications in wording and order in the course of the necessary adjustments and assimilations to new documentary frameworks, as Burkett shows. However his own analysis brought to light a surplus of effect and surplus of variation in these operations that his narrowly conceived scribal functionality could not encompass. The problem lies in his correlation of source copying, compilation, and conflation to a modest scribal competence. Rather than wooden operations in which we mark the toils of low-ranking scribes, these are the requisite technical skills for the full enactment, or better, reenactment of the tradition.Footnote ¹⁰⁰ When it comes to the history of the gospel tradition, therefore, what is needed is further research on scribes whose training made them capable of functioning as tradents, that is, competent not just in transmission but in cultivation of a written tradition. Burkett focuses on one side of the scribal ethos—the preservation and copying of texts, the consolidation and transmission of an authoritative tradition—to the neglect of the other side: cultivation of a living tradition, bringing the tradition constantly into engagement with the realities and predicaments of its tradent community.Footnote ¹⁰¹

This approach is particularly pertinent to the problem of the wide range and uneven levels of variation in the synoptic tradition, for it draws attention to strategic and contingent factors in the rise of variation. Though doubtless random or autonomous factors also are at work, much variation arises because scribal transmission of the tradition occurs not in isolation but in reciprocal contact with the social, historical, and theological realities of particular tradent communities. Given the contingent nature of those realities, this relationship will affect the tradition differentially, placing new interpretative demands upon some elements, leaving the received form of other elements intact. Redaction is the means by which written tradition is articulated for new or altered contexts of reception; its wide range may be understood as the effect of the persistence of oral practices and habits into the scribal cultivation of the written tradition in a mixed media cultural environment of pervasive orality. However it is important not to exaggerate the play of these factors within the written tradition. As Burkett shows, Matthew and Luke's utilization of Mark and Q is a matter of consolidation, with much variation due to solving technical problems of coherent integration of two normative written sources. Moreover, when we contemplate Mark and Q and then Matthew and Luke, we must probably think less of a continuum of constant redactional modifications of the written tradition and more in terms of a very few but programmatic redactions that mark great crisis and transitional moments in early Christianity, turning points that have called forth fresh consolidations and re-articulations of the tradition. Validating this approach, however, requires not just a thorough, differentiated analysis of scribal function in antiquity to locate and describe more accurately the scribes of the Jesus tradition, but also close, piecemeal analysis of the tradition itself.