(avg. read time: 13–26 mins.)
My posts for this month are focusing on Gospel synopsis, and they are related to my work on my Gospel synopsis commentary. But they have not been direct previews of my planned volume, and so they have not been behind a paywall. There will be an exception to that at the end of this post, as I will attach what I plan to be one of my appendices, which is a comprehensive set of tables showing the degree to which portions of the Gospels are verbally similar. This will stand behind my response today.
The response here is written to Andrew Mark Henry’s work on his YouTube channel “Religion for Breakfast.” Here is the video:
I am responding to this video because Henry represents a rather popular way of presenting the Synoptic Puzzle (more popularly known as the Synoptic Problem), with all the half-truths, distortions, and other errors pertaining thereto. As such, Henry is not uniquely at fault here, but his video certainly is emblematic of the inaccuracies of common presentations of the Synoptic Puzzle that I hope to overcome in my own commentary. In the following response, I will give the relevant timecodes for the points I am responding to.
1:30 To this point, Henry summarizes well why the Synoptic Gospels are called “Synoptic” in distinction from John. But his presentation starts to become all too simplistic when he says the Synoptic Gospels not only tell the same stories but that they tell them in “the very same words.” I do not often use the term “verbatim” to describe verbal similarities in the Gospels, since it implies not only the same words, but also the same words in the same order, which does not happen in the Synoptic Gospels as often as is thought (or the areas of overlap between any of the Synoptics and John). Most of the time, the same words will not be in the “very same” order. The longest string of verbatim similarities that we can find between any two Gospels are as follows:1
Matthew to Mark: 31 words (Matt 10:21–22 to Mark 13:12–13)
Matthew to Luke: 27 words (Matt 11:25c–27a to Luke 10:21c–22a2)
Matthew to John: 6 words (Matt 21:9 to John 12:13)
Mark to Matthew: 31 words (Mark 13:12–13 to Matt 10:21–22)
Mark to Luke: 29 words (Mark 10:14b–15 to Luke 18:16b–17)
Mark to John: 7 words (Mark 11:9 to John 12:13)
Luke to Matthew: 27 words (Luke 10:21c–22a to Matt 11:25c–27a)
Luke to Mark: 29 words (Luke 18:16b–17 to Mark 10:14b–15)
Luke to John: 6 words (Luke 23:3 to John 18:33)
John to Matthew: 6 words (John 12:13 to Matthew 21:9)
John to Mark: 7 words (John 12:13 to Mark 12:9)
John to Luke: 6 words (John 18:33 to Luke 23:3)
As it is, I tend to refer to “absolute verbal similarities” for times when the same words are shared, whether or not these are in the same order. And as I have explained before, there are many instances of “weighted verbal similarities” where the Gospels may have the same words in different forms (such as different cases of nouns or adjectives, or different tenses of verbs), which I count as 0.75 for each word (as opposed to 1 for each case of absolute verbal similarity), or where they may use synonyms, which I count as 0.5 for each word. For cases where word order of similarities varies, which is the vast majority of them, I use a < sign to signify that the word order variation makes the degree of similar less than the number figure would indicate (but I have not found a good way to account for that kind of variation in giving a number). When taking all such factors into account, although these figures (especially the weighted ones) could be subject to minor adjustments (most likely downward if I were to eliminate incidental similarities), I have determined based on my own work that the degree of verbal similarity strictly among parallel texts among the Gospels is as follows:
Moreover, while there are cases of nearly verbatim similarities between texts, there is not a single case in all of the parallel texts of the Gospels where the same stories are told in “the very same words.”
1:36 Oddly enough, the first example he uses for a story is about the call of Levi (or Matthew in Matthew’s version). The story is in Matt 9:9–13 // Mark 2:13 (or 14)–17 // Luke 5:27–32, but for some reason he only references the opening verses of each (Mark 2:14 in that case). He does rightly note that most of the twenty words appear in Matthew and Luke (seventeen of them to be exact), but they do not all appear in either parallel, and each other parallel has their own peculiar similarities and differences included with the common elements all share (thirteen words).
2:00 Next, we come to the feeding of the 5,000, a story with four versions I have compared elsewhere. Overall, the level of verbal similarity between the versions of that story are as follows:
It should be obvious from the above that the claim that the Synoptics “use the same words” is incredibly reductive, as besides the comparison of Matthew to Mark, every Gospel version is more different to the other versions than they are alike when it comes to wording. Likewise, his claim that “John does not share any wording,” is also inaccurate. There is not a great deal of similarity relative to the overall word count, but the similarities are there and do not need to be ignored.
2:40 I am not sure where Henry got these tabulations, but based on what I have noted above, it is indubitably false to say that “77% of Mark’s precise wording appears in Matthew” and “61% of Mark’s wording appears in Luke.” In terms of tabulating the absolute verbal similarity between parallel texts of the Gospels (and so overlooking unparalleled parts), these figures are not even close. The total absolute similarity (if this is what is meant by “precise wording”) is not even 50% similarity between any two Gospels. The level of verbatim similarity (if this is what is meant by “precise wording”) is somewhere even lower.
2:56 The review of Christian tradition here is, of course, simplified, but it does capture popular traditional ideas about the authorship of the Gospels. But to say that the tradition assumes that they were independent is not clearly the case. This is not the place to go into the weeds of early Church responses to the Synoptic Puzzle, but the prime example therefrom is Augustine in his Harmony of the Gospels saying that Matthew wrote first, Mark wrote with the awareness of Matthew, and Luke wrote knowing of the other two. Whatever the merits might be of this tradition, it is not inconsistent with the notion of a literary relationship between the Gospels.
3:35 When he says that today “New Testament scholars universally recognize that the Gospel authors were copying each other,” this only attests to the limits of his research (or of what Ian Mills made him aware of, perhaps). A minority of scholars identify oral transmission and memory as the source of similarities between the Synoptics, perhaps with help from notes. Rainer Riesner represents this kind of view in The Synoptic Problem: Four Views. Is the popular consensus of scholarship in favor of a documentary/literary solution to the Synoptic Puzzle? Absolutely. Is such a view universal? No.
3:47 At this point, Henry stumbles onto a major issue with such simplistic solutions to the Synoptic Puzzle as what he advocates when he says, “the Gospel writers must have been sitting in front of documents and copying from them when they were writing their own Gospels.” As we have highlighted, even when we restrict the scope to parallel texts, the Synoptics are more different in their wording than they are similar. That alone problematizes the notion that where the Gospels are similar, they were simply copying each other. On the most charitable reading, there is more editing in these parallels than there is copying (if that is, indeed, the right way of phrasing it, which I doubt).
What is never clear in this kind of hypothesis is how such a thing would have been practically accomplished. These books would have been initially composed on scrolls, which would tend to be unrolled across the writer’s legs (if not the ground) as he wrote. And for these Gospels, we are not talking about small scrolls. Yet we are to think that the writers had one scroll they were unrolling “in front of them” while writing another one, and their eyes darted back and forth between what they were reading and what they were writing, and yet we still end up with texts more different than they are alike. Of course, it could be that they had someone read a manuscript of a text while they wrote their own text, but that does not actually address why there are so many differences interwoven in such complex and not consistent fashion with the similarities. One can simply look at the examples I have provided in my Synoptic comparisons, which are not the most extreme in terms of similarity or dissimilarity, and ask how in the world we got the texts we did if this was the process.
Moreover, even beyond the variation in word order within particular pericopes, this hypothesis faces the problem of how to explain the different orders of parallel texts. There are stretches in which the Gospels parallel each other in narrative order (as I have illustrated here), but what about those areas where they are not, on this theory, simply using another source and yet the parallels appear in different relative positions in the Gospels? For example, consider the orders of Matthew and Mark, which are more similar than those between any of the other Gospels. Matthew 8:23–34 reflects the narrative order (though not the word order) of Mark 4:35–5:20, but then Matt 9:1–17 reflects the order of a different portion of Mark 2:1–22, and then Matt 9:18–26 parallels Mark 5:21–43, then later still (after Matthew has paralleled other parts of Mark and Luke in different relative locations), Matt 12:1–14 reflects the order of Mark 2:23–3:6. Examples could be multiplied, but this alone makes one wonder what Matthew was thinking if his goal was to copy Mark where he parallels him.
3:54 In any case, he cites instances of verbatim agreement as evidence for the Gospels copying each other. The first example he cites is Matt 11:25–27 // Luke 10:21–22. It is true that this text includes a twenty-seven-word string of verbatim agreement. When he says that it strains believability that such verbatim agreement could have come from two authors independently reporting Jesus’s teaching, he is simply stating his own incredulity. They could have memorized the saying separately, and for this central portion they are identical, but the similarity beyond that varies, as the total similarity of Matthew to Luke is 54/69 words (A) (<59.5/69 [W]), and for Luke to Matthew it is 54/75 words (A) (<59.5/75 [W]). It is at least not out of the realm of possibility that two memorizations, plus particular touches, could result in such levels of similarity, especially because of the emphasis in those days of passing on the memory of a teacher’s teachings in oral/aural contexts even before they were written down.
4:30 He says the aforementioned example is not unique, which is true, but it is an extreme example. As noted above, it is the single longest string of verbatim agreement between Matthew and Luke. There are a few other strings that come close to this length, but neither this example nor those can be properly considered representative. The next example he cites is of Matt 3:7–10 // Luke 3:7–9. The quote he gives does not start with the beginning of either Matt 3:7 or Luke 3:7, but if we restrict our analysis to what he quotes in English, we do indeed get sixty-four words used in Luke and sixty-three in Matthew (he must have included a textual variant in Matthew’s case). How different are they? As Henry says, they differ only slightly, as Matthew has three words that are different and Luke has four. Is this shocking? Not really, given that the imagery is vivid, the order is easily recalled, and it appears to be designed for memorability. And yet the surrounding versions of this story beyond this quoted saying are not nearly so similar. The similarity of Matthew to Luke overall in Matt 3:1–17 is <154/335 (A) (<173.5/335 [W]). The similarity of Luke to Matthew overall in Luke 3:1–9, 15–18, 21–22 is <154/316 (A) (<173.5/316 [W]). Without this portion that is so disproportionately similar, we get <94/272 (A) (<112/272 [W]) for Matthew and <94/252 (A) (<112/252[W]) for Luke.
He says there are dozens of other examples of this kind of verbatim agreement, but this is distortive. These are extreme examples, not representative examples, and there are not dozens of similar scale to what he has noted. For my paid subscribers, I will post the full charts that substantiate the following table, but as a summary of the data, if we were to go by pericopes/units and the degrees to which they match at the verbal level, including absolute (A) and weighted (W) verbal similarity (and with the understanding that the vast majority of these have variable word order), here is how they are distributed:
I know that is a lot to sort through and the table is unwieldy, but the data I am summarizing consists of 51 pages of charts. As those more extensive charts will also show, the texts that tend to be at the higher end of verbal similarity are shorter texts, although shorter texts are also distributed all over the place. Longer texts also show variety in distribution, but they are more often represented below the 50% line in terms of absolute similarity (weighted similarity will put some of these above the 50% line, particularly when what may be incidental similarities are counted).
4:40 A further point that he thinks favors these similarities being based entirely on literary dependence and to the point of copying is that Jesus “taught in Aramaic, not Greek.” The basic claim may be mostly true, but in such multilingual contexts as Jesus spoke in, it cannot be ruled out that he taught in Greek at all. It may or may not have been the majority of the time, but in any case such an issue is beside the point. The point Henry is making is that it is implausible that two authors working independently would translate Jesus’s words into Greek in “precisely the same way.” While his statement could be quibbled with to a degree here, not least because of how he has distorted the degree of similarity between the Gospels, this still does not demand that the similarities are based only (or even necessarily) on literary dependence. Even on the view he is arguing against of the Gospel authors working independently, the expectation is not necessarily that the Gospel authors translated these teachings themselves. If the stories and teachings in the Gospels circulated orally for decades before the Gospels as we have them now were written, and in some measure they circulated outside of the regions where Jesus ministered, would not one expect them to be translated into Greek in the process of circulation? Could not authors participating in or drawing from the same or similar streams of oral history account for similarities like we see? Taking this factor seriously does not entail that there is zero textual relationship between these books, but it can lead in a more helpful and illuminating direction than insisting upon a picture of copying that does not fit the Gospels as they are.
5:00 Henry also highlights that the Synoptic evangelists share narrations of events in addition to teachings. We have already noted this with the feeding of the 5,000, and the degree of similarity is notable yet by no means inconceivable on a hypothesis that allows for at least partial explanation of similarity through appeal to orality. This story would have been told many times over with variations in level of detail given, elements emphasized, and wording used.
The example he uses in this case is comparing Matthew and Mark on Judas’s betrayal of Jesus from Matt 26:47–48 and Mark 14:43–44. He says Matthew and Mark “describe Judas’s actions in the very same language.” There are a couple problems with that statement. One, these particular portions of text do not have much to say about Judas’s actions. But they are quite similar in their narration, and the words Judas says in the parallel parts are identical. Only Matthew and Mark include the brief bit of dialogue, but it is unsurprising that this much should be identical, as it was likely rehearsed many times and it is dialogue that functions to point forward to what is about to happen, meaning that remembering the following actions reinforces remembering the dialogue. Two, it is once again inaccurate to say that Matthew and Mark use “the very same language,” though they are closer than his earlier example of the feeding of the 5,000. The word order varies, as it often does, and the overall similarity for only these portions is <33/41 (A) (and <35.25/41 [W]) for Matthew and <33/45 (A) (and <35.25/45 [W]) for Mark. For the overall parallel episodes, which includes narration and dialogue, the levels of similarity between Matt 26:47–56 and Mark 14:43–52 are as follows:
Matthew: <85/179 (A)[~47.5%]; <93.75/179 (W)[~52.4%]
Mark: <85/140 (A)[~60.7%]; <93.75/140 (W)[~67.0%]
If one removes the typically extreme example he has cited, the degree of similarity drops even more:
Matthew: <52/138 (A)[~37.7%]; <58.5/138 (W)[~42.4%]
Mark: <52/95 (A)[~54.7%]; <58.5/95 (W)[~61.6%]
In neither the extreme example nor the rest of the story can it be rightly said that they tell the story in “the very same language.”
While certain texts, like those extreme examples Henry has picked, are presented in such a way—complete with inaccurate framing information I have noted at multiple points—as to seem to demand copying, this is not an accurate way of presenting the reality of Gospel parallels as a whole. On the hypothesis of simple literary dependence that I do not take for granted, the evidence would indicate that each Gospel author did less copying than editing of his sources (again, simply restricting ourselves to the consideration of where the Gospels parallel each other and not including their unique material).
The next section of the video takes up the argument for Markan priority, which is something I am not settled about one way or the other. Since I am not convinced of the kind of relationship Henry and others declare for the Synoptics, arguments based on redaction have rather negligible force in my view. And I have recently addressed the argument from editorial fatigue. I will not go through the many other arguments in favor of this majority view right now, but that is something I hope to do as part of a series on this Substack as the commentary takes another couple steps closer to completion. Still, I do want to say something about the scenario he puts forward, since it is in line with his overall argument that the Gospel authors were “copying” their sources and had them in front of them while they were writing their own texts.
7:17 Henry makes a standard point here about how it is difficult to explain Mark on the notion that Matthew and/or Luke wrote first. This seems especially so at the beginning and “end” of Mark. Mark does not have a story of Jesus’s birth and, on the most popular view, he does not include any accounts of his resurrection appearances. As to the former point, to say it makes no sense for Mark to omit the birth narrative if he knew of Matthew and/or Luke is presumptuous and requires us to forget the existence of the Gospel according to John, which also does not have a birth narrative as such. On any account, no author is compelled to include what his source includes, so this argument is merely appealing to present-day incredulity that an author would leave something out that was not as ubiquitously emphasized in gospel presentations (unlike the resurrection of Jesus). As to the latter point, Mark does not end where Henry thinks it does, as I have argued elsewhere. The Synoptic Gospels diverge in their wording more often than may be expected in their passion narratives, I have shown before that they are more different than they are alike in their accounts of the empty tomb, and there is no reason to be surprised that they tell such a variety of stories at their end (nor that they vary so much in wording even where they arguably parallel).
12:40 I have mentioned before that I do not rely on statistics about verses in describing the relations between the Gospels. These are conventions of textual division that came much later. But for the sake of the argument, let us go ahead and use them. Unfortunately, like so many others, Henry gives no source for his statistic that there are 220 verses “of high word-for-word agreement” shared between Matthew and Luke that do not appear in Mark, and I know that he has not done this work himself. For one thing, it is not clear if this is 220 verses in Matthew and 220 verses in Luke, or if the stat comes from only one of them, because it is not the combined total from both. Where they are parallel, they may not have the same number of verses. In fact, here are the instances where the number of verses is different between parallel texts in Matthew and Luke:
Matt 3:7–10 // Luke 3:7–9
Matt 4:1–11 // Luke 4:1–135
Matt 5:1–12*^ // Luke 6:20–26*^
Matt 5:25–26 // Luke 12:57–59*
Matt 5:31–32*^ // Luke 16:18*
Matt 5:38–42*^ // Luke 6:29–30
Matt 5:43–48*^ // Luke 6:27–28, 32–36*^
Matt 6:9–13* // Luke 11:1–4*^
Matt 6:19–21*^ // Luke 12:33–34*
Matt 6:22–23 // Luke 11:34–36
Matt 6:25–34 // Luke 12:22–32
Matt 7:3–5 // Luke 6:41–42
Matt 7:15–20*^ // Luke 6:43–44*
Matt 7:24–27* // Luke 6:47–49*
Matt 8:5–10, 13 // Luke 7:1–10*^
Matt 8:18–22^ // Luke 9:57–62*^
Matt 9:37–38 // Luke 10:2
Matt 10:37–38* // Luke 14:25–27*^
Matt 11:7–11, 14–19^ // Luke 7:24–35^
Matt 11:12–13* // Luke 16:16
Matt 11:20–24^ // Luke 10:12–15
Matt 11:25–27 // Luke 10:21–22
Matt 12:34–35 // Luke 6:45
Matt 13:33 // Luke 13:20–21
Matt 16:1–4*^ // Luke 12:54–56*
Matt 22:1–14*^ // Luke 14:15–24*^
Matt 23:13, 15–36*^ // Luke 11:39–52*
Matt 23:37–39 // Luke 13:34–35
Matt 24:37–41, 43–44 // Luke 17:26–36; 12:39–40*^
The texts with a * next to them signify cases of less than 50% absolute verbal agreement between the text and its parallel. The texts with a ^ next to them signify instances where a text contains at least one verse with 0 verbal similarity in the parallel text (some cases, like the beatitudes, involve multiple such verses). And if we only count these verses without the other cases of parallels, we have a total of 166 verses for Matthew and 147 verses for Luke (17:36 would not count in most totals). It does not work to simply say they share 220 verses because one or both of them is not going to come to that total.
In any case, given what I have noted already, it should be apparent that so many of these verses do not constitute cases of “high word-for-word agreement” between Matthew and Luke, otherwise they would not be so different as they are where they parallel each other. I cannot go through all of the texts here without at least forming another even more unwieldy chart. The above list is by no means a complete representation, but it still shows plenty of instances in which there is less than majority verbal agreement, including some verses where there are no cases of verbal agreement at all.
15:00 When Henry comes to trying to draw conclusions about authorship based on the majority views on the Synoptic Puzzle, he once again steps in it with his imprecision. He confidently states how unlikely it is that Matthew wrote Matthew given that the Matthean “origin story” (also stated quite imprecisely) in Matthew is copied “word for word” from Mark. The parallel stories of Matt 9:9–13 and Mark 2:13–17 are more similar than they are different, but let us be clear that they are not verbatim similar, not even in Henry’s own visual presentation that is dissonant with his words. Their similarity scores are as follows:
Matthew’s verbal similarity to Mark: <62/93 (A); <67/93 (W)
Mark’s verbal similarity to Matthew: <62/109 (A); <66.5/109 (W)
Not only are they not verbatim similar, the order of similar words varies between the texts, and there is a difference of sixteen words in overall length (nor are these sixteen words concentrated in one location). And while I am by no means committed to the idea that Matthew is literarily dependent on Mark, it is not as inconceivable as it is often assumed that an eyewitness like Matthew would have used earlier written material in composing his own account. My old professor Craig Keener has been keen to note throughout the years that Xenophon of Athens, despite being an eyewitness himself, could use accounts written by others as sources (Hell. 3.1.2; Apol. 1.1–2).
While someone watching this video will get a presentation of popular views of scholarship, anyone hoping to receive illumination on the complicated issue itself will not find what they are looking for here. I know that the writer credit for this video goes to Ian Mills, and I can only hope that Mills was not so irresponsible in how he wrote about this issue as to be at fault for all the errors of fact. I hope this is not how he lectures on the subject, but I would not be surprised if it is. I have attributed the claims throughout to Henry because it is his voice making the assertions throughout and he is the one who is ultimately in control of this channel. Regardless of who you ultimately attribute the errors to, they remain as they are, and this remains a misleading presentation of the nature of the Synoptic Puzzle, and thus the viewer is not in a good position to understand the merits and lack thereof for the various solutions presented for it.
Here and elsewhere, regardless of my misgivings about the text at many points, I am using NA28 as my base text, since it is the most common critical text used as a reference by scholars and Bible translators.
Some combination of Matt 12:41–42 to Luke 11:31–32 has a greater degree of absolute verbal similarity here, but there is not as much verbatim similarity because these sentences are arranged differently in Matthew and Luke.
As I have noted in my review of Dale Allison’s recent book, I do not think that Luke 5:1–11 and John 21:1–19 are truly parallel texts. But if we include that in our calculation for the sake of completion, we get totals of <335/2,266 (A)[~14.8%]; <480.75/2,266 (W)[~21.2%].
If we do the same addition here as in the other footnote, the totals are 335/3,062 (A)[~10.9%]; 482/3,062 (W)[~15.7%].
There is a parallel for this story in Mark 1:12–13, but it is obviously significantly condensed.