The Priority of St. Matthew’s Gospel

Daniel J. Castellano


Part IV: Evidence of Oral and Written Tradition in Matthew and Mark
   4.1 Conceptual Limitations of a Strictly Literary Paradigm
   4.2 Equivocal Evidence of Literary Genre in the Gospels
   4.3 Signs of Oral Composition in the Gospel Texts
   4.4 Signs of Oral Transmission in the Gospel Texts
   4.5 A New Analysis of the Matthaean-Marcan Double Tradition
      4.5.1 Methodology
      4.5.2 Results
      4.5.3 Discussion
   Footnotes to Part IV

Part IV: Evidence of Oral and Written Tradition in Matthew and Mark

Improved insight into oral modes of composition and transmission compel us to reinterpret the similarities and differences among the Synoptics. In order to do this, we first need to understand something about oral characteristics that might be found in a written text. Only then can we adopt better criteria for identifying a shared tradition as oral or written. While it is generally not possible to prove that a Gospel source was oral rather than written, we can at least establish in most places that the hypothesis of literary dependence is not necessitated by the evidence. This will be seen by a pericope-by-pericope comparison of the Gospels of St. Matthew and St. Mark. Without the presumption of literary dependence, a critical pillar of support for Marcan priority crumbles, leaving us less reason to overturn the verdict of historical and external manuscript evidence.

4.1 Conceptual Limitations of a Strictly Literary Paradigm

When we move beyond a strictly literary model of the Synoptic problem, we may regard conventional characterizations of Gospel tradition as no longer apt. The question of whether a certain version is original or derivative, primitive or secondary, historical or ahistorical, need not apply to traditions of oral character. In oral traditions, there can be a plurality of equally authentic though verbally different versions. We cannot identify a rendering as more or less primitive on the basis of its length or brevity, since reciters are granted leeway to add or omit unessential detail. This poetic license is not considered to derogate from the historical character of a narrative, since oral cultures have a different notion of history than the more literal-minded. Indeed, our rigid notion of “literal truth” comes from writing, so we should take note of how this mode of thought blinds us to likely realities about the oral Gospel.

The notion of “literal truth” supposes a set of one of more propositions with a fixed verbal form, which can be tested against dialectical criteria. The written word is something you can see and visually subdivide or compare different parts for logical analysis. In oral cultures, by contrast, there is no single fixed verbal form to test for “literal truth,” nor are narratives composed of propositions for logical analysis. What matters is the local meaning, so it is of no concern if different versions of a concept are portrayed consecutively, as in deliberately paradoxical pairs of proverbs (e.g., Proverbs 26:4-5), or if different versions of unessential descriptive details are used in the same narrative. Oral expression is to be experienced as it happens, not visually scanned and compared for “consistency.”

This lack of dialectical thinking is noted by David Rubin, citing A.R. Luria’s studies of Uzbeks and Kirghiz in the 1930s. Subjects were asked: “In the far north, where there is snow, all bears are white. Novaya Zemlya is in the Far North. What color are bears there?”[1] They responded that they did not know, not having seen a bear in Novaya Zemlya. Such an answer reflects both (1) being unaccustomed to subjecting different parts of a spoken conversation to logical comparison and inference, focusing only on local meaning, and (2) a trust only in concrete experience rather than in reasoning from generalizations. The respondents are not saying anything false or illogical, but are only neglecting to amplify their knowledge with logical inference.

Even literate people may find simple inferences less obvious when problems are put forward orally rather than in writing. Yet it is not writing as such, Rubin argues, that makes us inclined to logical analysis and categorization, but our more formal education, which itself is made possible by writing. Categorization, we recall, is facilitated by the brain's visual processing system, which is invoked by literacy. We often teach logic by a set theoretic presentation, drawing diagrams of “X” and “not X” as mutually exclusive, and so forth. When we teach abstract reasoning and generalization, we still appeal to the human preference for the concrete and sensible.

The educated classes of Europe have been accustomed to a primarily literary mode of thinking since about the third century. It was at that time when European thinkers began to turn decisively toward the written word as the primary means of conveying information and formulating theories. This mode of thought was in evidence when dealing with Christian revelation, both by apologists and opponents of Christianity. Men of letters now tried to interpret Jewish and Christian tradition by strictly literary, dialectical criteria, ignoring the oral modes of composition which had allowed a broader range of expression.

Porphyry of Tyre, the famous Neoplatonic critic of Christianity, was among the first raised in the developing literary culture of the third century, so he subjected the Bible to demands for literal consistency, disregarding oral modes of composition. The vast majority of Porphyry's attacks (and those of modern critics) on Biblical truth rely on a pedantic literalism. His principal errors include: (1) denying that the same text can have more than one meaning; (2) insisting that all accounts of the same event must have logically consistent descriptive details; (3) insisting that all ethical and religious precepts must fit into one logical scheme, making no allowance for paradox, change of emphasis or context. All these errors reflect a surprisingly rigid adherence to literary modes of thinking, without regard for the oral modes of thought likely behind much of the Scriptures. Even modern skeptics may hesitate to accept Porphyry’s pretended discovery of contradictions in the teachings of Jesus, which ignores changes in emphasis or context. Still, the general practice of analyzing Christian revelation in terms of “literal truth” has remained prevalent.

By the fourth century, Christians themselves had internalized the literary mode of thinking used by Porphyry. It was at this time that the Diatessaron was composed, and various Patristic authors sought to resolve the supposed inconsistencies of the Gospels while upholding a literal sense of truth. They could do this only by appealing to an allegorical interpretation on top of the literal sense. This device had been used by Jews and Christians for centuries, in order to reconcile the Scriptures with Hellenistic dialectic.

The literary culture of the Middle Ages was confined to the few who had access to expensive texts. For the rest of Christendom, the Gospel was still received orally, though mediated by texts. In rural areas, where there were no texts beyond those required for liturgy, the Word was conveyed through pictorial art, hymns, rote prayers and devotions. It is no coincidence that the Reformation took off in the age of the printing press. It was only then that a sola scriptura hermeneutic could be practicable.

Modern Biblical scholarship has been heavily influenced by a Protestant understanding of revelation, either defending the “literal truth” of the Scriptures, or identifying deviations from such as “inconsistencies” or “contradictions.” By such analysis, both conservative and liberal scholars effectively ignore the likely oral character of the vast majority of Jewish and Christian revelation. Once this orality is properly appreciated, we can better understand why so much of the Pentateuch contains stories far older than its likely period of written composition. The prophetic books are likewise recollections of the spoken word, for the very function of a prophet is to speak, not write, the word of God, which is why the Jews have not counted the Book of Daniel among the prophets. If the substance of the Gospel tradition is likewise predominantly oral in composition, we should not be amazed that words indifferent to numerical and dialectical consistency should have been unhesitatingly regarded as true by intelligent and ethical men.

The whole notion of “literal truth” presupposes written letters (litterae), presented in a fixed form to be univocally interpreted. In oral tradition, by contrast, singers enjoy creative license to tell a story the way they want, modifying incidental details but retaining the essentials. Incidental variations would not impugn a singer with falsehood, but anyone who dared change an essential feature, which in some cases (proverbs and key phrases) required near verbatim accuracy, would be shouted down and corrected, as observed by Bailey. As noted in Part III, variability actually helps stabilize a tradition, since it facilitates recall according to the preferences of each singer, and does not make the tradition dependent on the preservation of any single variant.

The oral genres behind the Synoptic Gospels do not neatly fit into our modern categories of fiction and non-fiction. If we attempt to describe them in such terms, we might say that the narratives are “non-fiction” in the essentials, and “fiction” in the incidental details. Even this requires further qualification, as the incidental details may be based on historical facts and sometimes replicate these accurately. When they depart from history, however, there is no falsehood involved, since they are primarily intended to serve the poetic function of situating the story in concrete imagery. Recollections of dialogue do not strive for verbatim accuracy (which can only be tested by writing), but only the preservation of the general sense, as well as some key phrases or sayings.

This need not mean that we never have the ipsissimi verbi of Jesus and other New Testament figures. Sayings or proverbs are more likely to be preserved close to verbatim, and this may be confirmed by identifying the meter and rhythm of the original Aramaic. Still, for the most part, the pursuit of “original exact words” is in vain, motivated by a literary notion of accuracy.

Nineteenth-century Biblical scholars, incapable of comprehending ancient indifference to literalism, concocted hypotheses of a crazy-quilt patchwork of texts put together by some redactor clumsily trying to harmonize contradictory though individually self-consistent sources. This naively assumed that the ancients were as obsessed with literalism as are we, and only knew literary modes of composition. On the contrary, it is only in the Hellenistic era that such concern started to become manifest, taking full bloom in the third and fourth centuries. Older texts, both sacred and profane, give evidence of oral modes of composition, along with distinct notions of truth, accuracy and history.

It remains to be seen how we might know which parts of the written Gospels exhibit signs of oral or literary composition. To this end, we will briefly review modern scholarship on characteristics of literary genre and orality in texts, as discussed in Mournet.[2].

[Top of page]

4.2 Equivocal Evidence of Literary Genre in the Gospels

Most modern scholars have been wedded to the notion that literal truth is the only real kind of truth, so that the question of the historical accuracy of the Gospels gets confused with the question of their literary character. Some, like Bultmann, find that oral traditions in the Gospels are necessarily legendary or fictitious. Others, wishing to uphold the historical truth of the Gospels, argue that they follow the form of literary biography or history. Both lines of argument contain the modern prejudice that truth can only be found reliably in the written word. Ironically, this approach turns out to be ahistorical, as first-century Christians, like other men of their time, held greater regard for spoken teachings.

The canonical Gospels represent the earliest known attempts to collect Christian tradition in a complete and orderly narrative. Nonetheless, it is problematic to regard them as purely literary endeavors, as they retain much of the oral character of their sources. While Richard Burridge has forcefully argued that the Gospels have the genre of Greco-Roman bioi,[3] critical examination of their supposed literary characteristics shows a lack of clear distinction from oral compositions.

One characteristic of bioi identified by Burridge is that they open with the subject’s ancestry, or at least a prologue. Matthew and Luke both contain genealogies and nativity narratives, though in reverse order, but Mark tells us nothing of Christ’s early life, beginning directly with the preaching of John the Baptist, with only a line of introduction passing for prologue.

Burridge finds Mark’s short prologue (if it may even be called such) comparable to that of Agesilaus, but in the latter case, Xenophon actually devotes six to eight sentences discussing Agesilaus’ ancestry, mentioning an extant genealogy of his descent from Heracles. Mark, by contrast, shows no concern whatsoever about Christ’s ancestry and early life, which is hardly consistent with a Greco-Roman bios. If his line of prologue suffices to meet the criterion, then we should say this criterion is far too weak to identify a genre, even partially. One need only look at Xenophon’s Cyropaedia to find a more representative example of ancient biography. Its lengthy, detailed prologue, shows the systematic organization characteristic of literature, providing historical context, followed by the parentage and childhood of Cyrus. The Agesilaus likewise opens by discussing the subject’s ancestry, and providing some historical context (i.e., the stability of the Lacedaemonian monarchy). It does not omit the youth of Agesilaus, since the subject took the throne at an early age. So there is strong reason to doubt that Mark, at least, wrote a bios in any conventional sense.

Another characteristic of bioi is their central focus on the individual subject, following his life through some basic chronological structure. The written Gospels all start with the Baptism of Christ and end with His Passion and Resurrection, but apart from that most of the narrative is notoriously achronological. Not only do the evangelists arrange the sequence of events differently, but most of the pericopes are completely indifferent to chronology, introduced only in illo tempore, without any definite indication of time. The stories are stand-alone vignettes without any definite trajectory in the life of Christ. This is much more consistent with oral story-telling, which is simply one thing after another, sometimes using travel from place to place as a way of ordering events.

It is true that the Gospels do focus on the person of Christ, but this is Christ as teacher, healer, savior and redeemer, the revelation of which does not require a biographical presentation. If anything, the evangelists are much more interested in the public works and teachings of Christ, rather than his personal affairs. In any event, the Gospel focus on the person of Christ need not indicate a specifically literary genre, as even oral traditions contain cycles of stories focusing on revered figures.

A third supposed characteristic of bioi is to devote very little space to the subject’s early years, and much more space to his final days. Yet lack of detail about childhood hardly suffices to establish literary composition. Any ancient account, oral or written, of someone’s life would not gather much information about the subject’s childhood, an undocumented and generally unremarkable time of life, lacking the public works that bring fame. Only modern biographies emphasize childhood, motivated by a belief that this will bring greater psychological understanding.

Still, an extreme emphasis on the subject’s final days is genuinely characteristic of bioi. In the case of the Gospels, this sign is confounded by the fact that Christ’s Passion and Crucifixion were genuinely memorable events, much like the trial of Socrates, and further by the fact that Christians regarded the death and resurrection of the Lord to have central significance to the Christian message, as proved in the apostolic writings. While an inordinate focus on a subject’s last days seems an artificial concern in a bios, the focus on Christ’s last days in the Gospels seems only natural, and is hardly probative of any genre constraints. Further, the Synoptic Gospels are, if anything, remarkably cursory in their Passion narrative, given its religious significance. They have no extended Last Supper discourses to match the Symposia of Plato or Xenophon, and they devote barely a line to the carrying of the Cross.

A fourth characteristic of bioi is continuous prose narrative, as opposed to poetry, songs or dialogues. Yet the Gospel is by no means limited to such narrative, as parables and sayings are presented in extended monologue or dialogue. The Gospel contains teachings of Christ, not just the life of Christ. If anything, there is more emphasis on the teachings, while deeds are recounted only to illustrate lessons of faith. It is true that the teachings are nested in some introductory and concluding lines of connecting narrative, but these are notoriously artificial, differing by evangelist, not part of the main body of teaching tradition. They might be editorial insertions by the writer, yet this simple sequential structure (one thing after another, moving place to place, with no higher organization or definite temporal markers) is also characteristic of oral tradition, to facilitate recall.

A fifth characteristic identified by Burridge is the use of anecdotes, stories and sayings. These are certainly present in the Gospels, but such content is hardly the exclusive province of bioi. Much of the Gospel content is just a continuous collection of such vignettes linked by lines of transition, without a coherent overarching structure. As a life of Christ, there is no definite career trajectory or clear notion of chronology, as we would expect in a biography.

Another important characteristic of bioi is that they illustrate the character of the subject by examples. This “show, don’t tell” approach is characteristic of good writing, yet it is also found in oral performance, which relies on concrete examples rather than abstract qualities. Once again, the supposed evidence of literary genre in the Gospels is equivocal at best.

Burridge has modified his views, acknowledging the differences between the Gospels and the Greco-Roman bioi, yet still insisting that they are a less formal sub-genre of “popular βιοι no longer extant.”[4] Apart from the unprovability of this assertion, the problem remains that these similarities are too generic, and need not indicate any definite literary genre.

On the contrary, we find, in agreement with early form critic Karl Ludwig Schmidt, that the Gospels, with their lack of structure and chronology, resemble folk literature (Kleinliteratur), made of material that originally circulated orally. They are unique as compilations for a religious purpose, but their components belong to various existing oral genres.

The likelihood that the Gospels are not historical biographies in genre does not imply that their facts are inaccurate, since genre does not determine truth or falsity, only intentions. Many literary works intended as factual histories are full of inaccuracies, while oral traditions, though unconcerned with literal truth in many instances, may nonetheless relate accurate facts. The presence of a nativity narrative, for instance, might of itself suggest a literary genre or at least something of the writer’s purpose, but this identification does nothing to establish the narrative’s historical accuracy or inaccuracy.

The Gospels, true to their name, are concerned principally with the “good message (evangelion) of Jesus Christ.” (Mark 1:1) St. Mark identifies this evangelion with “the message of the kingdom of God” that Jesus preached and urged the repentant to believe. (Mk. 1:14) Their content is determined not so much by literary genre as by the Christian message that was already circulated by preaching in the churches. All the biographical details of Christ are presented in the context of some religious teaching, not for the mere purpose of recording dry facts. Matthew and Luke include genealogies since their readers are concerned with the teaching that the Messiah must be a son of David. The structure of each Gospel, we will see, is better explained by catechetical concerns than by the requirements of genre. Their content is best accounted for by the orality of most of their sources, signs of which remain even in these written editions.

[Top of page]

4.3 Signs of Oral Composition in the Gospel Texts

The written Gospels, if they used oral sources, might retain some of the compositional techniques characteristic of oral performances, as well as signs that the tradition was originally transmitted orally. For evidence of oral composition, we may look to the structure or arrangement of each Gospel. For evidence of oral transmission, we may compare Synoptic parallels to see if we find the “stability with flexibility” characteristic of controlled oral traditions.

First, we ought to have a general understanding of what signs of oral composition we may expect to find in texts. Common oral compositional devices include ring composition, parataxis, inclusio, and chiasmus. On a more local scale, we may also look for metre, rhythm, or alliteration, as well an emphasis on concrete imagery over abstract concepts.

Ring composition, found in epics and ballads, presents a series of themes, reaching a crescendo, and then resolving each theme in reverse order, producing a format such as A B C D E D' C' B' A'. The core meaning of the text is in the middle (E in the example), and is framed by a concentric beginning and end. To interpret the text in linear sequence is to misunderstand it. [See: Mary Douglas. Thinking in Circles: An Essay on Ring Composition (2010).] Often the ending joins up with the beginning (A with A'), completing the ring.

Parataxis is arguably the most common feature of oral composition found in ancient texts. In this structure, statements are connected side by side with coordinating conjunctions, one after another, without the subordinate clauses or themes characteristic of literary organization. Parataxis can exist in literary composition as well, but oral tradition relies almost exclusively on parataxic connections, with hardly any subordinating clauses of considerable length.

Inclusio means bracketing into one unit everything from start to end, illustrated by the form: A xxxxx A. The identity of the A's makes us focus on the intervening xxxxx’s as important.

Chiasmus, as the term suggests, means a crosswise arrangement of concepts: first A, then B, then a variant of B, then a variant of A. It may be considered a small-scale version of ring composition. Unlike ring composition, however, it does not point to a central unit, but instead shows the interconnectedness of a pair of ideas. This feature is common in the Old Testament Psalms.

An obvious indicator of oral composition would be the presence of oral devices in texts. For example, the Homeric epics exhibit ring composition and parataxis. Parataxis is also in Hebrew rhetoric. Yet, as Mournet observes, ring composition, inclusio and chiasmus are present even in texts not derivative of oral communication and transmission. [Mournet, op. cit., pp. 153-54] This means we cannot use such structures to prove that a source is oral. At most, we can show that the hypothesis of strictly literary composition is not necessary to account for the evidence.

Still, the prevalence of parataxis throughout the Gospels is strongly suggestive of oral sources, even though, at the same time, this material has been supplemented by parenthetic comments by the evangelists. There are certainly some elements of literary composition in the Gospels; the evangelists were not mere compilers. While we cannot strictly prove the use of oral sources, the evidence consistent with orality means that we are not bound to a strictly literary solution of the Synoptic problem.

Orality in the Gospels would by no means be anomalous among early Christian texts. The Didache was likely solely dependent on oral traditions. It did not rely on written sources even for the Lord's Prayer. John D. Harvey, in Listening to the Text (Grand Rapids: Baker Books, 1998), finds rhetorical devices in the Pauline epistles as evidence of oral character. Mournet, however, rightly notes that much of this evidence is highly structured, suggesting a more literary mode of composition. St. Paul’s letters show “orality” only in the sense of being designed for aural reception, not in following oral modes of composition.

J.D.G. Dunn has noted that much of Gospel tradition is not interpreted in light of the Crucifixion and Resurrection, suggesting that these parts were composed during the ministry of Christ. An oral mode of composition would be most practical in this case, and Dunn proposes that a process similar to that observed by K.E. Bailey was in play. Contrary to our conception of sermons, where one person speaks and all others passively listen, oral communication in some cultures is a much more dynamic affair, involving listener participation to record memorable lessons on the spot.

Rhythm, metre and other poetic devices are ubiquitous in oral traditions, so we should expect to find traces of these in the portions of the Gospels relying on oral sources. A problem arises, however, from the fact that the texts are in Greek, when the earlier oral tradition was probably in Aramaic, as this was almost certainly the language of the preaching of Jesus. In order to recover rhythm, rhyme, alliteration and other wordplay, we would have to infer the original Aramaic from the translation. This can be done with fair probability only in a limited number of cases.

Charles Fox Burney (1868-1925) purported to find copious evidence of Aramaic poetry in the Gospels. [C.F. Burney. The Poetry of Our Lord (Oxford: Clarendon, 1925).] While his Aramaic reconstructions of the sayings and parables were generally credible, his attempts to impose a fixed metre on the text were often strained. Nonetheless, more recent scholars, notably Joachim Jeremias (1900-1979), have concurred in the general perception that there is rhythm and rhyme in the sayings material, most frequently a three-beat meter (e.g., in the beatitudes). Yet this evidence falls well short of establishing a formal poetic structure, such as Homer’s iambic pentameter. It would be a gross exaggeration to say that the Gospel preaching consists of poetry, but nonetheless it shows the signs of rhyme and rhythm found even in oral prose, serving as mnemonic aids to speaker and listener.

Another poetic device frequently found in the Gospels is parallelism, often in pairs of antitheses, and sometimes with a chiasmic structure. This parallelism, observable even in translation, is rarely sustained, however, and tends to be limited to single pairs of phrases. Still, Jeremias finds here a distinctive style, as the stress is generally on the second half of the pair, not the first as in the Old Testament. [J. Jeremias. New Testament Theology (1971), pp. 20-27.]

Burney found that the Gospel of Matthew preserved the poetic form of its source material much better than did Mark or Luke. Likewise, the discourses in the Gospel of John bear this stamp of authenticity, leading Burney to conclude that the Fourth Gospel was composed in Aramaic by a disciple and eyewitness of Christ.

Indeed, some of the distinctly Matthaean material clearly relies on an Aramaic original, e.g.: “You strain a fly (galma) out of your drink, but swallow a camel (gamla).” (Mt. 23:24) Likewise, the Matthaean version of the first beatitude shows a precise understanding of the Aramaic miskenayya as “poor in spirit.” Matthew alone uses the expression “kingdom of heaven” repeatedly, reflecting the Hebraic idiom of using “heaven” as a synonym for “God.”

Aside from poetic devices, there are numerous rhetorical devices found in all the Synoptics which show a consistent style, likely attributable to the original preaching of Jesus. These are discussed by J.T. Dillon in Jesus as a Teacher: A Multidisciplinary Case Study (Eugene, Ore.: Wipf and Stock, 1995). Two of the more obvious signs are the use of abba (“my father”) for God, original to Jesus, and the prefatory amen, absolutely unique to Jesus and unattested in all other Jewish and Christian literature.

In agreement with Eduard Schweizer, Dillon finds that Jesus uses concrete imagery not merely to clarify difficult ideas, but to lead hearers to make a judgment and come to a decision. The parables have no stated moral, but are a call to action. A knowledge of Jewish religion and ethics is presumed in the audience, not something to be taught. No special learning or rabbinical Hebrew was needed to understand what Jesus taught, which was conceptually simple but challenging to one’s character. Gerhardsson’s supposition of a Christian rabbinical school misunderstands Jesus to be just another rabbi or ethical teacher, rather than someone who demands actions of men.

While the forceful personality of Jesus clearly shines through the written Gospel discourses, assuring us that the Synoptics draw upon authentic tradition from Christ’s preaching, we should not be too hasty in concluding that the poetic aspects of the Gospels are all attributable to oral tradition. Even the Greek text has a discernible rhythm in places, such as the four-beat metre found in the beatitudes. H. Benedict Green finds Matthew's poetry in the Sermon on the Mount to be a sign of literary artifice, yet rhythm and rhyme are also characteristic of oral poetry, as is the ring composition he sees in Luke's version of the sermon (with a climax at 6:37-38).[Matthew, Poet of the Beatitudes, Sheffield, 2001] The two Gospels, Matthew and John, that Burney considered to be most poetic are also the ones most widely regarded as self-consciously literary endeavors.

None of the indicators discussed can definitively show an oral or written prehistory to Gospels. We can only show that there is no reason to assume a purely literary account of their composition.

[Top of page]

4.4 Signs of Oral Transmission in the Gospel Texts

While orality or textuality of composition cannot be strictly proved from the extant Gospel texts, it has been widely believed that the Synoptic problem can only be explained by appeal to shared written traditions. The early form critics pretended to deduce the model of tradition transmission from the extant Synoptic texts, but by a thoroughly circular method, as Mournet rightly observes. Bultmann assumed that literary and oral transmission were similar, progressing by linear growth or expansion, and so inferred that some Synoptic readings were primitive while more expansive variants were signs of literary transmission. This assumption is now known to be false, reopening the question as to how, if at all, we might discern the mode of source transmission by examining the written Gospels.

Oral traditions can change or vary even before they are put into writing, depending on the flexibility allowed to each genre. Such variations follow no fixed tendency toward expansion or brevity, so we cannot use length to determine primitivity. A plurality of variants appear from the beginning of a tradition, so there is no single primitive version, but we might speak of a form as “authentic” if it is recounted by a competent witness who retains the culturally recognized essentials. Insofar as they used oral sources, the evangelists would have sought authenticity in the quality of their witnesses, not in the supposed primitivity of verbal form.

Since many first-generation Christians were still alive when the Gospels were written, we need not worry about the reliability of oral transmission across multiple generations. The availability of first-generation witnesses further emphasizes the irrelevance of the question of primitivity, as all such accounts would be equally primitive. A single witness may vary his telling in each iteration, while remaining stable in the essentials. Poetic elegance or verbosity, favored details or forms of construction may reflect the varying genius of each witness, rather than give evidence of priority.

Our knowledge of oral tradition comes from modern sociological and folkloric studies, but it might be objected that these are anachronistic when applied to the Synoptic problem. No culture has been found in the modern era that closely parallels the linguistic environment of first century Palestine, where common people were highly exposed to both oral and literary communication, in a multilingual environment (Aramaic, Greek, and possibly Hebrew).[5] Anachronism is less problematic for those aspects of oral tradition that are culturally universal, as discussed in Part III. Still, we should respect the likelihood that the unique circumstances of Palestine, at the intersection of oral and literary modes of thought, might bring about distinctively complex modes of transmission in the Gospel tradition.

‘Folklore’ strictly means what is taught by the people, whether orally or in writing. The concept arose in nineteenth-century Germany, to express regard for traditions of common people, against the rationalistic upper class culture of the Enlightenment. ‘Folklore’ acquired negative connotations, however, on account of its association with fables and other popular fictions such as those published by the brothers Grimm. Understandably, critics such as Wellhausen considered folklore to be historically unreliable, and oral folklore to be unstable and incapable of preserving information. These judgments were informed by a culturally narrow perception of folklore. Broader studies have shown that folklore can have hundreds of different genres besides myth, including the historical, and that folkloric accounts are not necessarily created de novo, but can be handed down in accordance with strict norms.

Once we correct these misunderstandings about oral tradition, there should be no objection to considering the Gospels as “folklore” in the proper sense of the term. Controlled oral tradition can be historically reliable and stable in transmission. It does not matter if we do not fully accept Bailey’s belief that this stability can span centuries, for much of the oral Gospel was put in writing within a generation of the events it describes.

When the Bible is re-examined in light of what we know of folklore, much of the supposed evidence of redaction takes on new meaning. As Alan Dundes notes in Holy Writ as Oral Lit (Rowman & Littlefield, 1999), the variations in number, name and sequence found in the Scriptures, especially the Old Testament, are characteristics of folklore. “I maintain the Bible consists of orally transmitted tradition written down. Certainly there were collations, ‘literary’ emendations, and editorial tampering, but the folkloristic component of the Bible remains in plain sight even if blind scholars have failed to recognize it as such.” [pp. 19-20]

Since folklore is not a single genre, Dundes’ hypothesis does not address the specific questions of whether the Synoptics have intertextual dependency, whether they are historical in character, nor how the Gospel tradition originated and was transmitted. All of these questions are related to the difficult problem of identifying genre.

Still, even without knowing specific genre, we might at least find signs on a local level that the folklore of the Synoptics was oral rather than written. The repetition of key words or phrases (e.g., “wide-eyed Athena”) is an unconscious reflex among singers, according to Lord. Some hold that the frequent use of kai at the start of sentences in Mark is an example of such repetition, but this simply means “and,” and such usage is characteristic of Hebrew. [See J.D. Harvey, Listening to the Text, op. cit., pp.97-118.] This is more an example of parataxis than repetition. Better examples of repetition include: “Son of God,” and “the Kingdom of God,” as well as various phrases repeated in close proximity in Mark. (2:9-12, 3:14-16, 5:29-34, 6:24-28, 8:17-21, 9:43-47, 10:23-25, 10:47-48, 10:49, 14:56-59)

Repetition of phrases and statements may also help bracket material on larger scale, to structure and organize oral traditions, keeping both speaker and listener on track, and facilitate memorization. There is scant evidence of such large-scale oral structure in the Gospel, though we do find bracketing or inclusio on a smaller scale. For example, in the third chapter of Mark, Jesus’ relatives come to restrain him, followed by the Beelzebul controversy, after which Jesus’ mother and brethren arrive. A striking example is in Mark 5:22-43, where the healing of the woman with an issue of blood (5:25-34) is bracketed by the interrupted story of Jairus’ daughter. While keeping in mind that inclusio and ring structure are not unique to oral tradition, the evidence is at least consistent with the supposition that the Gospels consist of short-subject oral traditions compiled into a large-scale literary structure.

A more distinctive characteristic of oral tradition is flexibility, i.e., variation in each repeated performance, which Lord found in the Synoptics.

I have seen reason to believe that the Synoptic Gospels exhibit certain characteristics of oral traditional literature. First, for example, their texts vary from one another to such an extent as to rule out the possibility that, as a whole, one could have been copied from another. In this respect they have the appearance of three oral traditional variants of the same narrative and non-narrative materials. It is true that on occasion the texts are so close that one should not rule out manuscript transmission; hence it may be that oral tradition has sometimes had written sources affecting the text, not merely in respect to content but also to text. [A.B. Lord, "The Gospels as Oral Traditional Literature." In The Relationships among the Gospels: An Interdisciplinary Dialogue, ed. William O. Walker, Jr., (San Antonio, Texas: Trinity Univ. Press, 1978), p. 90.]

Although most characteristics of oral tradition may also be found in textual tradition, Lord’s point about the Synoptic texts being far too different to reflect direct copying remains sound. After all, it is hardly credible that an evangelist would so thoroughly alter the wording of a text that he had in front of him, if no change in meaning or improvement in style was thereby accomplished. It is far more likely that such common tradition was mediated by orality, either through purely oral tradition, or by aural recall of a text read at another time. It is important to emphasize that Lord is not denying a role for textual mediation in the Synoptic tradition. Orality vs. textuality is not an “either/or” question. In fact, he even goes so far as to allow that the oral tradition may itself have been shaped in part by written texts.

We can only agree with Mournet that it is difficult to accept that the Gospels are strictly oral traditional literature. [Mournet, op. cit., p. 185] The Gospel of Matthew in particular shows many signs of literary artifice. Still, the literary aspects of the Gospels need not be taken as evidence of literary interdependence, unless we could clearly show that the editorial remarks or literary devices of one evangelist was copied by another. This requires us to distinguish the contributions of the evangelist from his sources, by no means a straightforward problem, as we will see in Part V. Before even treating such a problem, however, we may find that most of the verbatim agreement among the Gospels is far too weak to require the mediation of a shared text.

Although Mournet believes in the Two-Source Hypothesis, he acknowledges that one cannot prove the textual dependency of the Synoptics on one another. At best, one can establish the likelihood of such dependency by appeal to “verbatim agreement, argument from order, alternating primitivity, etc.” Ibid., p 154.] Note that “alternating primitivity” presumes that there is such a thing as a primitive version, which can only be the case in a written tradition. If we do not assume the tradition is written, we may instead expect to find a variety of versions from the beginning, never consistently longer or shorter, more or less elaborate. The so-called “alternating primitivity” observed in the Synoptics, i.e., where sometimes one, then the other evangelist is more verbose, is much more consistent with oral tradition. To explain this phenomenon with texts, you would have to concoct one of the elaborate redaction processes proposed by form critics, but unattested in the time of the Gospels.

Strictly speaking, it is important to realize that these “formal proofs” for Markan priority and the existence of Q are not “knock down” arguments that are beyond reproach. Each of these classic arguments has been subjected to rigorous examination, and some have not stood the test of time. [Mournet, op. cit., p.155.]

Mournet acknowledges that the phenomenon of Matthew and Luke following Marcan order is no longer clearcut evidence of Marcan priority. Nonetheless, he still thinks it shows some sort of literary relationship, direct or indirect. It is conceivable that some of the Gospel tradition was written down in collections of narratives, upon which the evangelists drew, but this would contradict all the ancient testimony about how the Gospels were composed, using the preaching of the Apostles and the recollections of eyewitnesses. Yet we do not need to appeal to this ahistorical hypothesis, if we can show that the shared order of the Gospel materials is no more remarkable than the consistent ordering of narratives exhibited by oral poets.

The Two-Source Hypothesis remains the consensus among most current scholars, though some posit a “deutero-Marcus” to account for the large number of “minor agreements” in wording between Matthew and Luke against Mark. Marcan priority rests on similarity of order, wording, and alternating primitivity, all of which can be explained as well or better (in most places) by a common oral tradition. The supposition of Q would be unnecessary if we allowed direct dependence between Matthew and Luke. Their independence can be sustained only if Mark and Q are posited as overlapping in content, thereby accounting for why Matthew and Luke use different alternate forms for material in Mark (i.e., one or the other follows Q rather than Mark in different places). Yet this is circular reasoning, since it assumes the existence of Q in order to show the independence of Matthew and Luke, which presupposes the existence of Q (in the context of Marcan priority and literary dependence). If we make no such suppositions, we might just as well account for the different variants of Marcan material by simply acknowledging that there are multiple variants of the tradition, as is common in oral narratives.

We can agree with Mournet that it is not possible to strictly prove the presence or lack of oral tradition in the Gospels, but at best we can establish a probability based on the presence or absence of oral characteristics in abundance. Further, we should be clear that we need not suppose that the Gospels, or any parts of them, are strictly “oral” or “literary.” Rather, it is a question of the degree of oral or literary modes of composition and transmission indicated by the extant texts, taken by themselves and in comparison with each other.

[Top of page]

4.5 A New Analysis of the Marcan-Matthaean Double Tradition

When we introduce oral composition and transmission as serious alternatives to literary dependence, the hypothesis of Marcan priority ceases to be demanded by the textual evidence. “Alternating primitivity” is much more strongly indicative of oral tradition than literary redaction, which leaves only similarities of wording and order as evidence of distinctively literary dependence. If we look at this evidence more closely, now with greater awareness of the capabilities of oral narrators, we may find that the similarities in most places are not strong enough to point to direct textual dependence.

This evaluation can be made only after systematically examining the degree of verbatim similarity among the Synoptics, pericope by pericope. Since we are presently concerned with the question of Marcan priority, we will focus on the so-called Marcan source material.

[Top of page]

4.5.1 Methodology

In this study conducted by the author in 2013, every single pericope in the Synoptic Gospels was compared for verbatim similarity, using W.F. Farmer’s Synopticon. Farmer’s text follows the 25th edition of Nestlé-Aland, but using the readings of the 28th edition would not alter the results significantly, as will be shown in an appendix. For a word to be considered identical verbatim, it must be in the same grammatical form (marked with solid colored lines by Farmer). Synonyms and grammatical variants were not included, since the hypothesis being tested is whether copying from a text is necessary to account for the degree of verbatim similarity.

A pericope, for our purposes, is the smallest possible unit that might be transmitted as a coherent teaching or narrative. Our division of Marcan pericopes follows that given in Kurt Aland’s Synopsis Quattuor Evangeliorum (Stuttgart: Deutsch Bibelgesellschaft, 1985), with the following exceptions:

  1. Mk. 4:21-25 is subdivided into two independent sayings (Light of the World; On Judging)
  2. Mk. 6:6b is distinguished from Mk. 6:7-13
  3. Mk. 9:42-50 is subdivided into v.42-48 and v.49-50 (Salt of the Earth)

These subdivisions are introduced where Aland has made a source-critical judgment to combine parts that need not be a single unit. The independence of these parts seems corroborated by their Matthaean parallels, which are in separated pairs of locations.

The identification of Matthaean and Lucan parallel texts likewise follows Aland, except we have ignored most of the secondary parallels he identifies. The reason for this exclusion is that such secondary parallels do not at all resemble the Marcan text, and thus are unlikely to be derived from it directly. They more likely come from Q or some other extra-Marcan source, which is not our present concern.

For example, Aland considers Mt. 4:24-25 to be another parallel to Mk. 3:7-12 (besides Mt. 12:15-16), but the similarity is far too fragmentary to be convincing, and at any rate cannot be invoked in support of textual dependency. Likewise I omit Mt. 9:32-34; Mt. 7:16-20; Lk. 11:17-23; Lk. 6:43-45, since no one can contend that these were derived from the text of Mark, as closer Marcan parallels exist in the same Gospels.

Aland makes some source critical judgments in identifying parallels, assuming that highly dissimilar accounts are actually variants of the same pericope. For example, he thinks Mark 11:25-26 is parallel to the Lord’s Prayer. Whether this be true or not, we set this aside, since we are operating from a pre-critical perspective, concerned only with quantitative textual comparison. For similar reasons, we discard his proposed parallels to Mark 11:11, with the tentative exception of Matthew 21:17. Since these doubtful parallels have very low rates of verbatim similarity, their inclusion would have only harmed the thesis of literary dependence.

The total number of words in each Marcan pericope was counted and tabulated. Each string separated by a space in Farmer’s text counted as a word. Several dozen of the pericopes, especially the longer ones, were recounted, in most cases confirming the initial count, and never resulting in an inaccuracy greater than one word. Then the number of words preserved verbatim, first in Matthew, then in Luke, were counted for each pericope, again recounting as needed for the longer pericopes.

Since we are examining the supposition that Matthew and Luke copied from Mark, verbatim similarity was measured as a percent of the total number of Marcan words found in each Matthaean or Lucan parallel. Making the Marcan total our base eliminates underestimation of verbatim similarity that would be caused by Matthaean or Lucan additions or interpolations.

Another measure recorded was the longest string of consecutive Marcan words preserved in each Matthaean or Lucan parallel. The longest string excludes Old Testament citations, since they constitute no proof of dependence on Mark, yet these cases are noted in marginal comments. Old Testament quotes were nonetheless counted toward the word count for verbatim similarity, though the adjusted percentages excluding these citations are noted in the margin.

Although the word counts, verbatim percentages, and longest strings are all measured objectively, the evaluation of these figures for each pericope requires some discretion. Short pericopes might have misleading percentages due to small sample size. A pericope with an overall low percentage of verbatim similarity might nonetheless have highly concentrated similarity in certain parts. A majority of the shared words may be merely incidental particles, pronouns, and prepositions, or words expected to be used in any account of the subject matter.

Several principles were used to evaluate the degree of verbatim similarity. These are explained in the descriptions of each evaluation category.

TEXTUAL = certainly textual. This classification means that the degree and kind of verbatim similarity cannot be explained by solely oral tradition or oral recall of a text, but only by some direct copying of a common written source (which may or may not be one of the Synoptic Gospels themselves). The standard for certainty is set high, requiring one of the following criteria.

If the longest string is 15 words or more (excluding Scriptural citations), then the dependence is certainly textual. The basis for this cutoff is the finding by Marks and Jack that the longest consecutive string that can be spontaneously recalled is 15.1 words for texts in sentences. [Marks, M. R. and Jack, O. (1952). Verbal context and memory span for meaningful material. American Journal of Psychology, 65, 298-300] This finding of a 15-word limit has been confirmed by subsequent research.

Although oral poets can recall verse verbatim up to 50 consecutive words, such prodigies are exceptional, and there is little evidence of formal poetic devices in the Gospels (especially rhyming), which are necessary to facilitate such extensive recall. Further, as we have noted, there is no evidence that Jesus or the Apostles ever established a rabbinic school with rote memorization of teachings in fixed verbal form. Jewish schooling, however, might account for verbatim recall of longer citations of Scripture, though it could also be that the evangelists used similar editions of the Septuagint.

Pericopes with longest verbatim strings of only 14 are also classified as “TEXTUAL” if the replicated text is not a saying, but just common narrative prose, and surrounding parts of the pericope also have a strong degree of verbatim similarity. This applied to the Matthaean parallel of Mk. 6:32-44, and the Lucan parallel of Mk. 10:17-22.

textual? = probably textual. These pericopes have strong verbal similarities which, while not impossible to reconcile with purely oral recollection, are too frequent or concentrated, or covering plain prose narrative where exact wording is not essential, to make a purely oral account likely. In these cases, the Gospel similarities must be accounted by the mediation of a text, not necessarily copied, but at least recalled aurally. The shared written source may or may not be one of the Gospels themselves.

To be classified as “textual?”, a pericope must have at least 40% verbatim similarity (or a higher percentage in some concentrated portion of the pericope containing at least 30 words), and consecutive strings of at least 5 words. Pericopes are excluded from this category if the only extended verbatim similarity is in Biblical citations or key sayings or quotes likely to be recalled with similar wording even in oral tradition.

oral = probably oral. The verbal similarities are too weak to represent any attempt to copy or edit a shared text, and can be credibly accounted by shared oral tradition. Signs of orality include a low percentage of verbatim similarity (< 40%), with consecutive strings no longer than 3 or 4 words, except for occasional memorable sayings. The words that are shared are generally separated from each other, and often in different order, suggesting that much of the similarity is purely incidental, or a result of broadly similar ways of telling the story orally.

Pericopes with higher degrees of verbatim similarity are classified as “oral” when that similarity is based solely on Old Testament citations, while the rest of the pericope is verbally dissimilar. Such pericopes are oral only with respect to Gospel tradition, not absolutely.

Also, short pericopes with high percentages of similarity due to small sample size are classified as “oral” when the shared words are only those necessary to relate the same facts, plus some incidental function words.

The tradition recounted in “oral” pericopes might be partly mediated by texts, either as indirect sources, or by aural recall. For our purposes, they are all oral insofar as they are generally incompatible with the hypothesis of direct literary borrowing from another Gospel or a shared written source.

As noted previously, it is not possible to strictly prove that a text’s source is oral or written. However, we can at least show that the hypothesis of literary dependence is more awkward in some places than others, as it requires us to suppose a highly irregular practice of verbatim copying and tortured rearrangement without purpose, even in the same pericope.

none = no dependence on common tradition, oral or written. Here the verbal similarity is so weak as to be purely incidental, based on the necessities of grammar or subject matter. In these cases, we are dealing with independent, freely composed accounts of the same subject, or else the pericopes are not truly parallel at all; i.e., not attempting to relate the same story or teaching.

Since there is some discretion involved in classifying pericopes, the possibility of bias cannot be fully eliminated. One relevant possible bias is toward giving similar classifications to consecutive pericopes. Ideally, the investigator should be blinded to the order of pericopes, but this is practically impossible for anyone familiar with the Gospel texts. Automation of the categorization process is impractical, due to the need for human judgment about the kinds of words and phrases that are similar. This difficulty is not insurmountable, however, so future studies may attempt to use natural language processing software and text analytics to implement the process by an algorithm.

Even objective aspects of the measurements may be questioned, since the objectivity of a measurement is no proof of its relevance. It would be desirable to validate each of the principles described here by controlled empirical studies.

[Top of page]

4.5.2 Results

[Full table of results]

While there is room for disagreement on the classification of various pericopes, and allowance should be made for investigator bias, the raw numbers still show a clear pattern that can scarcely be ignored. Higher and lower degrees of verbatim similarity appear in the same places across all three Gospels. This is highly suggestive of the likelihood that certain blocks of tradition were mediated by texts, while the remainder can be accounted for by oral transmission.

Again, these simplistically named categories of oral and textual should not be interpreted as implying mutual exclusivity of oral and written modes. The “oral” traditions might have been written down at some point before the Gospels, though we have no direct evidence of this, while the “textual” traditions might have been mediated aurally, and recalled from memory.

Even before making comparisons of verbatim similarities, the Synoptic Gospels reveal evidence of narrative discontinuities, shown by a sudden change in scene or by one or another Gospel moving some material to a different place in sequence. These are indicated with solid line borders in the table of results. Each block of material represents a possible tradition cycle that was generally recounted as a single unit, orally or textually. We allow that one or another evangelist may have interpolated other sayings into such a block.

Intriguingly, the “oral” and “textual” pericopes tend to be grouped into blocks, whose boundaries (represented by dashed lines) often match those of the blocks of continuous narrative. More remarkably, both Matthew and Luke seem to resort to more or less exact verbatim renderings of Mark in mostly the same places. This has devastating implications for the theory that Matthew and Luke depend on Mark, for it is hardly credible that both evangelists should independently choose to copy Mark here exactly, there paraphrastically, in nearly the same places.

Let us examine our blocks of oral or written tradition, one by one, in Marcan sequence, to see if they may be credibly accounted for without reference to Marcan priority.

Mark’s Gospel begins with a narrative about John the Baptist, his preaching, and his baptism of Jesus. Matthew and Luke reproduce highly similar versions (62%-66% verbatim), likely mediated by a common text. If that common text is Mark, we should explain why both Matthew and Luke give such a loose rendering of the subsequent material on the temptation of Christ, his journey to Galilee and ministry there. Here Matthew and Luke prefer another, lengthier source about the temptation in the desert, and so follow this source for the rest of the sequence. This neither confirms nor refutes Marcan priority, since it could be that all three evangelists relied on the same written source, yet Matthew and Luke followed it more completely through the temptation, which shows continuity with the previous episode, as Satan challenges Jesus’ claim to be the Son of God.

Matthew supposedly returns to Mark’s text on the Call of the Disciples (Mt. 4:18-22; Mk. 1:16-20), only to shortly digress into the Sermon on the Mount. This material is indeed probably textual in composition, as shown by the use of the parenthetic “for they were fishermen” in both versions. The use of this material in the same sequence is hardly remarkable, since the call of the first disciples clearly took place when Jesus entered Galilee.

The thesis of literary dependence on Mark is more difficult to sustain when we examine the cycle about Jesus in Capernaum (Mk. 1:21-35). The astonishment at Jesus’ teaching (1:21-22) and healing of the demoniac (1:23-28) have clearly textual degrees of similarity. Matthew preserves only the former (Mt. 7:28-29), in verbatim agreement with Mark 1:22 (“And they were astonished at his doctrine. For he was teaching them as one having power, and not as the scribes” [Mt.: “…and Pharisees”]), though he omits mention of teaching in synagogue. Luke retains both pericopes (Lk. 4:31-37), though more closely agreeing with Mark in the latter (4:33-37). Both Matthew and Luke depart from the Marcan text at the same point, showing only an “oral” degree of similarity on the healing of Peter’s mother-in-law, the healing of the sick, and Jesus’ departure from Capernaum. (Mk. 1:29-38).

Why should both evangelists follow Mark’s wording closely at first, yet cease to do so for the second half of the cycle? This time, it cannot be sustained that Matthew and Luke are both following Q, for they are even more dissimilar from each other than from Mark. It is conceivable that both independently found another source they preferred to Mark, yet we would have to accept many such coincidences in order to explain similar phenomena throughout the Gospels.

More likely, the Synoptic content corresponding to Mark 1:29-39 reflects a shared oral tradition. All three evangelists append this material to a relatively fixed text (there may have been variants) about Jesus teaching and healing a demoniac in the synagogue at Capernaum. There is nothing remarkable about this agreement in sequence, since it only makes sense to append this oral tradition to the other material about Capernaum.

We should note, however, that the evangelists differ on where to place the Capernaum sequence in a larger context. Matthew has it at the end of the Sermon on the Mount, which he places near Capernaum (Mt. 8:4), and mentions nothing about anything happening in synagogue. Luke has the Capernaum sequence take place after some unique material about Jesus preaching in Nazareth, which creates a chronological problem at Lk. 4:23 (“…as we have heard done in Capernaum”). Evidently, neither the oral nor written traditions about Capernaum had been part of a larger fixed sequence prior to the written Gospels.

We next come to a block of material, corresponding to Mark 1:40-3:1-6, that is likely textual in the degree of verbatim similarity. Matthew alone makes a departure from its sequence, placing the cleansing of the leper at the beginning of his material on Capernaum. Otherwise, the material is preserved in the same consecutive order by all three evangelists, which is remarkable considering the disparity of its content. This further suggests that the shared source is textual, be it one of the Synoptics or some other writing.

Within this section, Matthew and Mark have their strongest verbatim similarities in dialogue and sayings. This indicates that the evangelists used written sources with the mentality of men steeped in oral tradition, finding verbatim exactitude important only for proverbs and key lines of dialogue. This accounts for why there is so much verbal dissimilarity even in passages where an evangelist clearly had access to a written source. Nowhere in the Synoptics do we find simple “copy and paste” redaction.

Next, we find some oral traditional material parallel to Mark 3:7-19, regarding the choosing of the Twelve and healing multitudes by the sea. Matthew and Luke do not present this material in the same order as Mark, and have only weak verbatim similarity to Mark, each replicating different words, with very little verbatim triple tradition, and no Matthaean-Lukan double tradition. The verbal similarities must be regarded as incidental, being attributable to the subject matter and shared oral tradition. The fact that this non-textual tradition is in the same part of each Gospel shows that there was a generally understood, though imperfect, chronology for Gospel events that facilitated placement of narrative material (but not sayings). Such similarity in sequence can occur even when there is clearly no literary dependence.

The next block shows evidence of ring structure in Mark (3:20-35), but this structure is absent in Matthew and Luke, who do not mention Jesus’ relatives thinking him mad. (Mk. 3:20-21) We cannot determine priority from this, since the ring structure might be original to the oral tradition, or a device deliberately or accidentally introduced by Mark. The pericope on collusion with Beelzebub has moderate verbatim similarity between Matthew and Mark, suggesting mediation by a shared text at some point. Verbatim similarity is much weaker regarding the sin against the Holy Spirit, but is strong again on the pericope about Jesus’ true kindred. Luke has these three pericopes separated in different places, though in the same general area of his Gospel, and his verbatim resemblance to the Matthaean-Marcan double tradition is far too weak to suggest a literary relationship.

Matthew and Mark, by contrast, retain a relatively high level of verbatim similarity through the parable of the sower and its interpretation. Yet here and in the previous block, much of that similarity is found in lines of dialogue and key sayings, much as we would expect in oral tradition. There is no question that some aspects of the verbal form of this tradition had become relatively fixed, but this need not have been mediated by a text, much less is it necessary that one Gospel referred to another. Indeed, the intervening pericope on the reason for parables has no stronger verbatim resemblance between Matthew and Mark than with Luke, who only shows an “oral” degree of similarity in this section.

The subsequent Marcan sayings material (4:24-34) has generally weak parallels with the other Synoptics, who reproduce only some of it, and not in the same sequence or even the same part of the Gospel. The notable exception is the saying on judging (Mk. 4:24-25), which has highly similar wording among all three Synoptics, though it is situated in three different contexts. The disparity of sequence makes it unlikely that one Evangelist was copying from another (keep in mind the difficulty of spreading out long manuscripts and jumping back and forth). More likely, this saying was recalled from memory by each Evangelist or his source, with a relatively fixed verbal form.

The next block of tradition covers the stilling of the storm, the Gadarene demoniacs, and the healing of Jairus’ daughter. (Mk. 435-5:43) Here the Matthaean parallels have only weak verbatim similarity, consistent with a shared oral tradition. Luke has slightly stronger similarity with Mark in the last two pericopes, though this is concentrated in key phrases, suggesting perhaps that this part of the oral tradition was written down. If Luke’s source was Mark, however, it is strange that he replaced his version of the stilling of the storm with a shorter form.

Mark 6:1-6 has moderately strong verbal parallels with Matthew, but none of the Synoptics have this material in the same sequence. The commissioning of the Twelve and opinions regarding Jesus (Mk. 6:7-16) have only an “oral” degree of similarity among the Synoptics, and Matthew has the pericopes separate and out of sequence, making it especially unlikely that he used Mark here.

The death of John the Baptist immediately follows the opinions regarding Jesus in Matthew and Mark, in order to explain why some thought Jesus was John raised from the dead. Luke places it early in his Gospel, with the rest of his material on the Baptist. In all three Gospels, this episode is anachronistically inserted, suggesting that it is not part of the main tradition about Jesus, but a historical digression. We find somewhat stronger similarity between Matthew and Mark, suggesting a shared textual tradition, though the similarity is localized at Mark 6:17b and the key phrase, “Give me the head…” Luke’s account is entirely dissimilar and seems to be totally independent.

Mark and Luke continue with the return of the Twelve, but Matthew omits this. It seems that the block of shared oral Gospel tradition began with the commissioning of the Twelve and ended with their return. Mark and Luke retain this narrative in its entirety; Matthew omits the ending; Matthew and Mark insert a text-based source on the death of the Baptist.

The next block of tradition, from the feeding of the five thousand through the walking on water to the healings at Gennaseret (Mk 6:32-56) has stronger verbatim similarity between Matthew and Mark, but is omitted almost entirely by Luke, who has only a brief version of the feeding of 5000, with localized similarity around Mark 6:41, the blessing of the loaves and fishes. Given the iconic significance of these great miracles, it is unsurprising if early Christians committed them to writing for liturgical use, or at least established fixed forms for key lines.

Remarkably, Matthew (15:1-16:12) and Mark (7:1-8:21) continue with a cycle of miracles and teachings in the exact same sequence, yet for the most part exhibiting highly dissimilar wording. Luke has no parallels to any of this material, except a weakly similar mention of the sign of Jonah, fragmented and in different context. Only one pericope in the Matthew-Mark double tradition, the feeding of the four thousand, shows signs of strong similarity, mediated by a text, with a string of 16 consecutive words verbatim. This is in verse 2, a direct quotation; the pericope as a whole shows only moderate similarity. Again, this is an especially important miracle, which could easily have been put to writing or given a fixed form at an early date. Its omission by Luke does not prove it is a duplication of the feeding of the 5000, since Luke does not have any material from this section. Duplications tend to be characteristic of oral tradition, so it is remarkable that both the miracles of the 5000 and of the 4000 are preserved by textual traditions in Matthew and Mark. This suggests that they were recognized as distinct events from an early date, before the written Gospels.

A block of apparently textual tradition runs from Peter’s Confession to the Transfiguration. This is found in all three Gospels, except that Luke has a different version of the Transfiguration, likely relying on oral tradition that retains the key quotation at Mk. 9:5.

The subsequent material (Mk. 9:11-48) has only weak verbal similarity among the Synoptics, though they preserve the sequence of pericopes from the Coming of Elijah through the teaching on true greatness (Mk. 9:11-37). This shows that even oral traditions may have frequently associated some pericopes in a certain sequence, so agreement in sequence is no proof of textuality. Still, the last two pericopes of this material (Mk. 9:38-48) are scattered in different places in all three Synoptics, showing that these did not belong in a particular block.

Again, we find it remarkable how all three Synoptics simultaneously move from a fairly verbatim rendering to a loose oral traditional mode at the same point (Mk. 9:10). On the conventional two-source assumptions, Matthew and Luke independently decided to change the way they used Mark at the exact same point, for some strange reason.

It should be noted that even the “oral” material has occasional strings of 6-8 words with verbatim similarity to Mark (and one of 10 words in Luke’s parallel to Mk. 9:37), but these are confined to memorable quotations, sayings, and key phrases. There is generally closer verbatim similarity between Matthew and Mark, but Luke agrees more with Mark in sequence.

The saying about the “salt of the earth” is completely dissimilar in verbal content and sequence in all three Synoptics. This is not the product of a shared tradition, except in the most generic sense, being related to an expression used by Jesus, perhaps in different contexts.

The next block of material, from the crossing of the Jordan into Judea through the discussion of precedence among the disciples (Mk. 10:1-45), shows strong verbal similarity between Matthew and Mark, suggesting a shared written tradition, which is certainly the case for the last pericope. Luke uses only some of this material (Mk. 10:13-31; Lk. 18:15-30), yet he agrees with Mark much more closely than does Matthew in the blessing of the children and the rich young man (Mk. 10:13-22), with long consecutive strings proving textuality. This gives us a further clue about how the evangelists used texts. Their irregular degrees of verbatim similarity shows that they did not place special value in verbatim transcription, or the texts were mediated aurally, or they worked from variant readings. The last explanation seems least plausible, given how early the Gospels were written, and how slowly variants develop. The first two explanations are mutually compatible.

The fact that Luke excludes some of this textual tradition need not imply a literary dependence between Matthew and Mark. Luke's exclusion of the first two pericopes is adequately explained by the fact that he has already covered this material elsewhere, having chosen a different sequence in the arrangement of his "travel narrative." More puzzling, perhaps, is why he omits the Matthaean-Marcan tradition about the third prediction of the Passion and the clearly textual pericope about precedence among the disciples, preferring his own versions, yet in the same sequence. The supposition that Matthew depends on Mark does nothing to help solve this mystery.

The pericopes about the two blind men and the entry into Jerusalem have only an “oral” degree of verbal similarity. Luke has the two pericopes separated by other material, while they are consecutive in Matthew and Mark. These two Gospels continue in sequence with the return to Bethany and the cursing of the fig tree, yet with only an “oral” degree of similarity.

The cleansing of the temple is unquestionably a common written tradition between Matthew and Mark. Luke has only 21 of the 61 Marcan words, but this includes a string of 9, making some textual mediation likely.

The pericope about conspiring priests and scribes (Mk. 11:18-19) is totally independent in all three Synoptics, only incidental verbal similarity “(the) priests and the scribes.” Aland rightly does not even count this as a Marcan parallel in Matthew, though he does for Luke. The three evangelists independently placed this content after the cleansing of the Temple, to reflect a widespread belief that the priestly conspiracy was a response to this perceived outrage.

Matthew and Mark follow with the withering of the fig tree, though they differ in placing the curse of the fig tree before (Mark) or after (Matthew) the cleansing of the Temple. The supplemental character of this oral material leads us to believe that the text on the cleansing of the Temple belongs to the same block as what follows.

The subsequent block of pericopes (Mk. 11:27-12:27) on discourses at the Temple is textual in character. There is sufficient verbal variation to suggest that this text was not mediated strictly by direct copying, but also by oral recitation.

Luke has the pericope about the great commandment in a different part of his Gospel, suggesting it is not part of the above written source. This agrees with what we find in Matthew and Mark, who have scant verbal similarity here, in striking contrast with the previous pericopes. Once again, we find that the three Synoptics are consistent in what is oral and what is written, which is hardly consonant with the hypothesis that Matthew and Luke independently referred to the complete written Mark. Further, we find again that agreement in sequence between Matthew and Mark can occur even when there is no shared written text. Sequence may be dictated by oral tradition, by topics of teachings, or by causes of events.

In the next section (Mk. 12:35-13:2), we find a rare block of textual material (Mk. 12:37b-44) shared by Mark and Luke yet ignored by Matthew, who shows only an oral degree of similarity. It is unlikely that Matthew has suddenly decided to render Mark very loosely, since the pericope on the widow’s mite is omitted altogether. More likely, Matthew did not know the text accessible to Mark and Luke, which means he did not have either of these Gospels before him when writing.

The prediction of the destruction of the Temple, though an “oral” pericope, likely should be considered as an introduction to the apocalyptic discourse that follows, rather than as part of the preceding cycle. The “synoptic apocalypse” (Mk. 13:3-32) itself is strongly textual in character, with the highest degree of sustained verbatim rendering anywhere in the Synoptics. Clearly this discourse was considered highly important, and owing to the difficulty of understanding it, early Christians likely deemed it prudent to record its exact words as best as possible.

Interestingly, all three Synoptics conclude the discourse with an exhortation to be watchful, yet these are all dissimilar, showing no signs of oral or literary dependence. Evidently, the idea that Christians should be watchful for the end times was widely held.

Strikingly, most of the subsequent material up through the Crucifixion is rendered orally, though this is considered to be the most important narrative of the Gospel. This only confirms that writing was not held in higher esteem than oral tradition.

There are two exceptions to this oral rendering. First, there is a strongly textual Last Supper discourse, followed by the agony in Gethsemane and the arrest of Jesus. This text is shared by Matthew and Mark, but not Luke. The latter does, however, have some significant verbal similarity with Mark in a pericope on the Last Supper, though the same degree of similarity exists for the preparation for the Passover (Lk. 22:7-14), which is not part of the shared Matthaean-Marcan text, so Luke need not have had direct access to this. Note that the sequence in pericopes for the Passion narrative is identical in all three Gospels. (Luke omits the anointing in Bethany. Aland’s proposed parallel in Lk. 7:36-50 has no verbal agreement whatsoever with the other Synoptics.) This indicates that the narrative already had some formal structure in oral tradition.

The second place with textual similarity is the Matthaean-Marcan account of Jesus on the Cross and the death of Jesus. Given the importance ascribed to a person’s last words, it is unsurprising that a more literal rendering is presented.

The subsequent material has only an oral degree of similarity, though still preserving sequence. Shared oral tradition ends with Mark 16:8, Matthew 24:8, Luke 24:9, as the women depart from the empty tomb after the angel has announced the Resurrection. The remaining resurrection narratives of Matthew, Luke, and the “longer ending” of Mark are all completely independent of each other.

[Top of page]

4.5.3 Discussion

In sum, we find that one of the supposed pillars of Marcan priority, namely verbatim agreement, is actually much more consistent with the use of common oral and written traditions, and presents problems for the supposition that Matthew used Mark as a written source. Matthew and Luke repeatedly depart from a verbatim rendering of Mark in similar locations, even when they are not using another shared text such as Q. Such coincidences suggest that they did not have the full text of Mark before them, but only had the following blocks of text in common:

  1. John the Baptist [Mk. 1:2-11; Mt. 3:1-17; Lk. 3:1-22]
  2. Call of the Disciples [Mk. 1:16-20; Mt. 4:18-22]
  3. Synagogue at Capernaum [Mk. 1:23-28; Mt. 7:28-29 (excl. healing); Lk. 4:31-37]
  4. From Cleansing of the Leper to Man with the Withered Hand [Mk. 1:40-3:6; Mt. 8:1-4, 9:1-17, 12:1-14; Lk. 5:12-6:11]
  5. Collusion with Beelzebub / Jesus’ True Kindred; Parable of the Sower / Interpretation [Mk. 3:22-27, 31-35, 4:1-9, 13-20; Mt. 12:22-30, 46-50, 13:1-9, 18-23]
  6. Preaching at Nazareth, Death of the Baptist, Feeding of 5000, Walking on Water, Healings at Gennesaret [Mk. 6:1-6a, 17-29, 32-56; Mt. 13:53-58, 14:3-36]
  7. Feeding of 4000 [Mk. 8:11-13; Mt. 15:32-39]
  8. From Peter’s Confession to the Transfiguration [Mk. 8:27-9:10, Mt. 16:13-17:9; Lk. 9:18-27 (excl. Transfig.)]
  9. True Greatness / The Strange Exorcist [Mk. 9:33-41; Lk. 9:46-50]
  10. From Crossing the Jordan to Precedence among the Disciples [Mk. 10:1-45; Mt. 19:1-20:28; Lk. 18:15-30 (only 3 pericopes)]
  11. Woe to Scribes / The Widow’s Mite [Mk. 12:37b-44; Lk. 20:45-21:4]
  12. Sermon on End Times [Mk. 13:3-32; Mt. 24:3-36; Lk. 21:7-33 (excl. false Christs)]
  13. From Foretelling of Betrayal to Jesus’ Arrest; Mk. 14:18-52; Mt. 26:21-56, Lk. 22:15-20 (Last Supper only; also preparation for Passover: Lk. 22:7-14; Mk. 14:12-17)]
  14. Jesus Derided on Cross to Death of Jesus [Mk. 15:27-39; Mt. 27:38-54]

Even these shared texts, judging from their degree of verbatim similarity, seem to have been mediated aurally rather than by direct copying. 12 of the 14 texts are found in Matthew, while only 8 are in Luke. Of the 6 texts in triple tradition, 4 are only partially reproduced (omitting at least 1 pericope) in Luke, and 1 other is only partly reproduced in Matthew.

Mark has all 14 blocks only because we have defined our scope with respect to his Gospel. There could be cases, however, where Mark records the textual tradition less completely. This can be tested by comparing the triple tradition passages. In the block on John the Baptist, we find Matthew and Luke to be in much closer agreement with each other than with Mark, as they share an additional 96 words, including strings of 24 and 20 consecutive words! Likewise, in the triple tradition pericope about Beelzebub, Matthew and Luke share text (Mt. 12:26-28, 30) not found in Mark, while Luke has nothing of Mark’s text not found in Matthew. If Matthew and Luke are independent, this proves that Mark used a textual source that was also known to Matthew and Luke, who render it more completely, so common textuality is no proof of Marcan priority. In the other four triple tradition text blocks, there are only two places where Matthew and Luke share additional text (Mt. 9:7b; 24:27).

Matthew tends to agree more closely with Mark, but this does not prove Mark is his source. We know from the triple tradition blocks that Luke tends to omit entire pericopes from his textual sources. Since he only takes excerpts or paraphrases, while Matthew and Mark render texts more completely, it is only natural that the latter two should agree more.

Aside from these blocks, Mark and its Synoptic parallels consist of common oral tradition. Out of the 117 Marcan pericopes (excluding the prologue and longer ending), only three are not found in the other Synoptics: Jesus thought to be beside himself (3:20-21), the parable of the seed growing secretly (4:26-29), and the blind man healed at Bethsaida (8:22-26). 12 Marcan pericopes have no Matthaean parallel with shared oral tradition, and 32 have no Lucan parallel.

Judging from the diverse locations of verbal coincidences, the “oral Gospel” covered the preaching and deeds of Christ from his baptism by John until the discovery of his empty tomb. The nativity and post-Resurrection narratives in the Synoptics show no indication of shared oral or literary dependence. These are all unique accounts that were not part of the commonly preached Gospel. Topics covered by shared texts are also found in oral tradition, as can be seen when Matthaean-Marcan textual tradition is recorded as oral tradition (in different order) in Luke (See blocks 5-6, 10, 13-14), or Marcan-Lukan text is oral in Matthew (blocks 9, 11). Here “oral” tradition might be mediated by a different written source, but the underlying orality explains the diversity of variants, which is not attributable to copying errors in such a short time. The material on John the Baptist and the Synoptic apocalypse might have been purely textual tradition. Still, it is clear that the scope of oral tradition was not limited to sayings, but included miracles and other deeds of Christ, reflecting the diversity of oral genres.

Agreements in sequence among the Synoptics are consistent with our identification of oral and written traditions. Naturally, the “textual” blocks agree in pericope sequence, on the supposition that the Evangelists had shared written sources, or variants of the same. Aside from the Passion narrative, where the sequence is dictated by that of events, there is relatively little agreement in sequence for the material we have identified as “oral,” as we should expect from its genre and from the limitations of human memory. There is no fixed large-scale chronology, but only local sequences that can be explained by the supposition of shared oral and written sources no less than by that of direct literary dependence.

Our table of results highlights the pericopes where Matthew and Luke definitely agree with Marcan sequence. In the Marcan pericopes, Matthew and Luke never agree against Mark in sequence. This is at least suggestive that Mark determined the shared sequence, a point in favor of Marcan priority. The point is stronger when Matthew and Luke follow Mark’s sequence in teachings and miracle stories, where the chronology is not self-evident. Luke’s sequence in Chapters 4-6 suggests he had access to Mark, or at least a portion of it. On the other hand, most of this sequence is accounted for by a shared common text. Also, the sequence at Capernaum is fairly logical and could have been preserved orally.

Our present concern, however, is whether Matthew, not Luke, used Mark to order his material. Matthew agrees with Marcan order only for early material on John the Baptist (Mk. 1:2-20; Mt. 3:1-4:22) and the start of Christ’ mission, and material after the death of John the Baptist. The coincidence for the sequence of early material is unremarkable, since the material on John the Baptist comes from a shared text, which Matthew and Luke render more completely than Mark. The temptation in the desert, early mission in Galilee and call of the disciples follow an obvious order that need not imply any literary dependence.

Matthew completely ignores Marcan order in Chapters 5-13 (Mk. 1:21-6:13), but shows close agreement in sequence in the remainder of the Gospel. This follows two narrative threads. First, there is a narrative about the identity of Jesus, from Herod’s inquiry to Peter’s confession of faith. This sequence includes the miraculous feeding of 5000 Jews and 4000 Gentiles, which revealed Christ’s global Messianic mission. Second is an extended Passion narrative, starting with Jesus’ prediction of his death as he told his disciples they must go to Jerusalem. This may be subdivided into (1) a sequence of predictions of the Passion up to the entry in Jerusalem, (2) the preaching in the Temple, including the apocalypse, and (3) the Passion narrative proper, from the Last Supper onward.

The more impressive agreement in sequence is in the first narrative, which contains diverse miracles and teachings in Galilee and Syropalestine. Matthew and Mark’s agreement is too sustained to be coincidental. Luke appears to be ignorant of this arrangement of material, as he omits much of it, and what he relates has only weak verbal similarity, and in different order. Here is the strongest case for some sort of literary dependence between Matthew and Mark, yet their sustained verbal dissimilarity suggests this dependence was not direct. This section consists of both oral and written sources, yet the oral portions are never more than four pericopes long, well within the bounds of human memory for non-narrative genres. Some groups of pericopes, such as Matthew 15:1-31, seem to belong together as an integral whole, making the coincidence in order less remarkable.

In the subsequent extended Passion narrative, the coincidence in order is less probative of literary dependence, since the sequence of events is sufficiently memorable to be recalled orally, especially by specialists entrusted with reciting the tradition. The teachings embedded in this narrative (Mt. 19:13-30, 21:23-33, 24:3-36) have textual sources to account for the shared order. The only notable exception is four consecutive oral pericopes from the coming of Elijah to the teaching on true greatness. Sustained agreement among oral pericopes occurs only in the narrative after the trial before the Sanhedrin, a sequence of events so firmly impressed on the Christian mind, that it is hardly remarkable that this should be remembered even without the aid of written sources.

When we remove the assumption that the verbatim similarity between Matthew and Mark indicates direct literary dependence, our interpretation of the agreement in sequence also changes. It could just as well be the case that Matthew rather than Mark is the originator of the sequence. Mark’s position as the “middle term” between Matthew and Luke could be explained by Luke having access only to Mark and not Matthew. Mark’s debt to Matthew for the sequence could be indirect, if his oral sources preached in accordance with it.

None of the above disproves Marcan priority, but only shows that the evidence for such is much more equivocal than is generally supposed. This leaves the question of whether there is sufficient reason to overturn the consensus of external evidence in favor of Matthaean priority.

On the other hand, if our alternative interpretation of the internal evidence is no less contradictory of historical testimonies, then it would seem the latter should also be set aside from deciding the issue. For example, if we assert that Mark used written sources, this appears to contradict ancient testimonies that his Gospel was composed from the preaching of St. Peter. If we consider the Patristic witnesses to have given fundamentally inaccurate accounts of how the Gospels were composed, we can hardly appeal to them to support Matthaean priority.

We have seen, however, that much, if not all, of the textually mediated tradition was also circulated orally. If St. Mark used some already written sources in the aforementioned fourteen blocks, or recalled these from memory, he would hardly be departing from his intention of recording the preaching of St. Peter, since these would have been valid representations of the “oral Gospel” preached by the Apostles. It would not matter that such writings were not in Peter’s exact words, since each teller of oral tradition is permitted to change wording while retaining substance. A possible exception is the material on John the Baptist, which preceded Peter’s encounter with Jesus, requiring Mark to refer to another source.

Still, if we allow St. Mark the latitude to use available written sources, however fragmentary, why should we not admit the same for St. Matthew? Is his apostolic authorship necessarily impugned if he sometimes copied from St. Mark to save labor? Now that we have a better understanding of how oral and written traditions were transmitted, it no longer seems necessary to agree with the critics that Matthaean dependence on Mark is incompatible with apostolic authorship of the First Gospel. As a matter of probable fact, nonetheless, Matthew’s knowledge of Mark, if any, is confined to excerpts. His Gospel as a whole is by no means derivative of Mark, as even its Marcan parallels are mostly derived from shared oral tradition, not from the Gospel of Mark as we have it.

The only remaining way to make a strong positive case for Marcan priority would be to show that Matthew (as well as Luke) makes use of language that is specific to Mark. This would imply that the shared written tradition, or at least the version of it used by Matthew, comes from Mark. First we would have to be able to distinguish which aspects of Mark’s Gospel come from the evangelist rather than his sources.

Continue to Part V

Footnotes to Part IV

[1] A.R. Luria Cognitive Development: Its Cultural and Social Foundations (Cambridge, Mass.: Harvard Univ. Press, 1976), p.109. Cited in: David C. Rubin. Memory in Oral Traditions. (Oxford Univ. Press, 1995), p.312.

[2] Terence C. Mournet. Oral Tradition and Literary Dependency: Variability and Stability in the Synoptic Tradition. (Tübingen: Mohr Siebeck, 2005)

[3] Richard Burridge. What are the Gospels? A Comparison With Graeco-Roman Biography (Cambridge: Cambridge Univ. Press, 1992).

[4] Richard Burridge. What are the Gospels? A Comparison With Graeco-Roman Biography, 2nd ed.(Wm. B. Eerdmans Pub. Co., 2004), p.204.

[5] JA Emerton "The Problem of Vernacular Hebrew in the First Century A.D. and the Language of Jesus", JTS 24 (1973): 1-23

© 2014-15 Daniel J. Castellano. All rights reserved.