Did Matthew Abbreviate Mark?

by Jimmy Akin

in Bible, Synoptic Problem

Saint-Matthew_This post presents the results of a test I recently did in my ongoing look at the Synoptic Problem.

In what follows, I will be testing the claim that if Matthew used Mark, he abbreviated the material he found in Mark. Note the “if,” because it’s important. I am not here arguing that he did use Mark. That’s a topic to be discussed elsewhere.

Here goes . . .

 

The Issue at Hand

An important perception among biblical scholars is that, if Matthew drew material from the Gospel of Mark, he seems to have abbreviated this material.

The presumable reason for this would be to allow Matthew to have space to fit in all the other material he wanted to add to Mark and still keep his own Gospel the size of a single volume (either a single scroll or a single codex).

In The Four Gospels, B. H. Streeter gives several examples of how Matthew (apparently) shortened different sections or pericopes (per-IH-ko-PEES) of Mark. He notes how Matthew’s versions have fewer words in Greek than the corresponding pericopes in Mark.

However, in his book The Tendencies of the Synoptic Tradition, E. P. Sanders claimed that Matthew does not consistently shorten material from Mark. If you look at all the pericopes Matthew and Mark have in common, they’re fairly even in terms of overall word count. Matthew’s total word count for these pericopes is slightly shorter than Mark’s, but not by much, and most of the difference is grouped in just a handful of pericopes (see The Tendencies of the Synoptic Tradition, 82-87).

This, however, doesn’t strike me as the optimal test: Doing a straight pericope-to-pericope test could be misleading, since Matthew adds material from his own sources to pericopes.

For example, in the account of the Testing in the Wilderness, Mark has only a brief note that the event took place, but he doesn’t describe it in detail. Matthew does; he has the three “temptations” that the devil presents Christ with.

A better test, it occurred to me, would be to eliminate Matthew’s additions (like the three temptations) and see if we find that he shortened what remains.

Unfortunately, Sanders didn’t do this kind of test. Also unfortunately, I don’t know anybody else who has done this kind of test, either.

 

Why This Is Important

In Synoptic Problem studies, a good deal hinges on whether Matthew would have abbreviated the material he took from Mark, because that gives us a clue to the order of the two Gospels.

It’s much more likely, given the way ancient authors worked, that Matthew would have consistently tightened up Mark’s text than for Mark to consistently expand Matthew’s text in a sentence-by-sentence manner.

Therefore, if Matthew’s material looks like a tightened up version of Mark’s, Mark probably wrote first.

In view of the importance of the question at hand, I wanted to find the answer.

Fortunately, I realized that I had the tools available to do the test myself.

 

The Tools

Some time ago I began developing a synopsis of the four Gospels that presents both the English and the Greek text of each one in parallel columns.

(It’s not yet published, since I’m still adding new features to it, but I hope to publish it in the future.)

To develop this synopsis, I loaded the Greek and English text into a spreadsheet (Microsoft Excel) and matched the text up verse-by-verse. (Of course, the verse divisions, like the pericope divisions scholars use, are not in the original text, but they are useful.)

The advantage of having the material in a spreadsheet is that I’m able to sort and manipulate the synopsis in various ways that can’t be done with a synopsis printed on paper.

These sorting capabilities, I realized, would let me do the kind of test I wanted to do on Matthew and Mark. Using Excel, it shouldn’t be difficult to isolate the material needed from the two Gospels and then do a word count on the Greek text.

Excel doesn’t have a good word count tool (that I know of), but Microsoft Word does. (N.B. Although Greek has diacritical marks which could, in some character encodings, cause Word to think there were more words than there are, this would have applied to both texts equally and the overall result would remain valid. However, I verified that I was not using one of those character encodings so the word count should be accurate.)

Therefore, all I had to do was isolate the relevant material, paste the Greek text into Word, and see what the resulting word count was.

So I did the test.

 

Pass 1 of the Test

To isolate the relevant material, I took the following steps:

  1. I made a copy of the spreadsheet so I could manipulate it without harming the original.
  2. I struck all material related to Luke and John.
  3. I struck the longer ending of Mark (since it likely was not original and not what Matthew had in front of him).
  4. I struck all the pericopes in Mark that have no parallel in Matthew (allowing a pericope-to-pericope comparison)
  5. I struck all of the verses that Matthew contained which have no parallels in Mark. This represents the additions that Matthew would have made to Mark (thus allowing a more refined pericope-to-pericope test than the one Sanders did).
  6. I then pasted the resulting Greek text from both Gospels into Word.

Results:

  • Matthew: 8,114 words
  • Mark: 10,542 words

If Matthew used Mark, it would seem that he abbreviated the pericopes he used by 2,428 words or 23%, dropping almost one in four words.

 

Pass 2 of the Test

Although the above results should be the best way to look at the problem, a potential objection occurred to me: The above selection of material includes verses in Mark that Matthew would have omitted entirely.

It seems to me that these verses should be counted in the test (as in Pass 1). There is nothing to say that, in selecting material from Mark, Matthew couldn’t delete entire verses within a pericope. Indeed, the evidence indicates that he would have.

However, just to go the extra mile (to bend a phrase from Matthew 5:41), I decided to do a second pass of the test, eliminating the verses in Mark that had no parallel in Matthew (even though the pericopes that contained them did have a parallel in Matthew).

My prediction, if Matthew was shortening Mark, was that the Matthew material would still contain fewer words (since Matthew was tightening things up within verses as well as by deleting verses) though the result would be less pronounced.

Results:

  • Matthew: 8,114 words
  • Mark: 8,569 words

Thus if we compare just the verses that have direct parallels in both Gospels, it would seem that Matthew abbreviated these verses by 455 words or 5%, dropping about one word in twenty as he tightened up the text (aside from the whole verses he dropped).

 

Summary Thus Far

On both versions of the test, the data supports the conclusion that if Matthew used Mark, he abbreviated the material he took from it.

This is particularly clear on the better version of the test (Pass 1), but also true on the “go the extra mile” version of the test (Pass 2).

The difference in the results of the two passes indicates that Matthew would have done much of his abbreviation by dropping the contents of entire verses, while also tightening up the contents of the verses he retained.

The versions of the test that I did are a pair of rough-and-ready assessments that depended significantly on computers. A more refined, human-based, and scholarly version of this test could be performed in the future, but the results are likely to be the same in substance.

 

Verses Matthew Would Have Added to Mark’s Pericopes

Let’s complete our look at the issue by examining the verses that Matthew would have added to the pericopes he shares with Mark.

If the hypothesis is correct that Matthew abbreviated what he took from Mark to help make room for the additional information he wanted to add to his Gospel, we should find that most of the verses he added within these pericopes should be independently-sourced, value-added verses, providing new information rather than just restating what should be obvious or paraphrasing Mark in a somewhat wordier way.

By my count, there are 149 such verses. They are listed below, and you can see what they say by hovering your mouse over the individual verse citations.

 

Prophetic Fulfillments

Matthew is very interested in showing that Jesus fulfilled Old Testament prophecy.

As a result, it is not surprising that he would have added notes regarding how Jesus did so to pericopes he found in Mark. By my count, there would be 16 verses of this material:

  1. Matt. 4:14
  2. Matt. 4:15
  3. Matt. 4:16
  4. Matt. 8:17
  5. Matt. 12:17
  6. Matt. 12:18
  7. Matt. 12:19
  8. Matt. 12:20
  9. Matt. 12:21
  10. Matt. 13:14
  11. Matt. 13:15
  12. Matt. 13:35
  13. Matt. 21:4
  14. Matt. 21:5
  15. Matt. 27:9
  16. Matt. 27:10

For Matthew, these prophetic fulfillment notices counted as value-added content for the audience he was trying to reach, which particularly included Jewish Christians who would be specially interested in how Jesus fulfilled Old Testament prophecy.

 

Double Tradition (“Q”) Material

Additional value-added material is found in what is known as the “double tradition.”

This material also was seen as valuable by the Evangelist Luke, who also included it in his Gospel (hence “double tradition,” because it is found in two Gospels).

One question is where this material came from. Did Luke get it from Matthew? Did Matthew get it from Luke? Or did they both get it from a now lost source? Many modern scholars think the latter, and they have dubbed the proposed, lost source “Q.”

There are about 235 verses of the double tradition material in Matthew and Luke, but only some of them occur in the pericopes that Matthew would have taken from Mark.

To establish which ones, I compared these pericopes with the verses attributed to the double tradition in the International Q Project’s work, The Critical Edition of Q (Robinson, Hoffmann, Kloppenborg, ed.s). This may be deemed a neutral source in that it was not formulated with respect to the hypothesis we are presently testing.

By this reckoning, Matthew would have added about 42 verses of double tradition material to the pericopes he used from Mark:

  1. Matt. 3:7
  2. Matt. 3:8
  3. Matt. 3:9
  4. Matt. 3:10
  5. Matt. 3:12
  6. Matt. 4:3
  7. Matt. 4:4
  8. Matt. 4:5
  9. Matt. 4:6
  10. Matt. 4:7
  11. Matt. 4:8
  12. Matt. 4:9
  13. Matt. 4:10
  14. Matt. 4:13
  15. Matt. 5:13
  16. Matt. 10:6
  17. Matt. 10:7
  18. Matt. 10:8
  19. Matt. 10:15
  20. Matt. 10:16
  21. Matt. 10:19
  22. Matt. 10:24
  23. Matt. 10:25
  24. Matt. 12:27
  25. Matt. 12:28
  26. Matt. 12:30
  27. Matt. 12:32
  28. Matt. 13:16
  29. Matt. 13:17
  30. Matt. 15:14
  31. Matt. 16:3*
  32. Matt. 18:6
  33. Matt. 18:7
  34. Matt. 21:32*
  35. Matt. 24:26
  36. Matt. 24:27
  37. Matt. 24:28
  38. Matt. 24:37
  39. Matt. 24:38
  40. Matt. 24:39
  41. Matt. 24:40
  42. Matt. 24:41

(* The International Q Project lists these two verses as possibly but not definitely being part of Q in their estimation.)

Between this material and the prophetic fulfillments, so far 58 of the 149 verses that Matthew added to Markan pericopes would be independently-sourced and value-added.

 

Other New Traditions

Of course, the Old Testament and the double tradition were not Matthew’s only sources besides his proposed use of Mark. In addition to the above, there are at least 47 verses in which Matthew seems to have drawn on additional material from his own sources (eyewitness memory or otherwise):

  1. Matt. 5:14
  2. Matt. 5:16
  3. Matt. 10:12
  4. Matt. 10:13
  5. Matt. 10:17
  6. Matt. 10:18
  7. Matt. 10:20
  8. Matt. 10:21
  9. Matt. 10:22
  10. Matt. 10:23
  11. Matt. 12:7
  12. Matt. 12:11
  13. Matt. 12:12
  14. Matt. 12:22
  15. Matt. 13:12
  16. Matt. 14:28
  17. Matt. 14:29
  18. Matt. 14:30
  19. Matt. 14:31
  20. Matt. 15:12
  21. Matt. 15:13
  22. Matt. 16:2 (* This verse is plausibly grouped with a possible double tradition or “Q” saying in Matt. 16:3; otherwise the two verses would both belong to this list.)
  23. Matt. 16:17
  24. Matt. 16:18
  25. Matt. 16:19
  26. Matt. 18:10
  27. Matt. 19:10
  28. Matt. 19:11
  29. Matt. 19:12
  30. Matt. 21:28
  31. Matt. 21:29
  32. Matt. 21:30
  33. Matt. 21:31
  34. Matt. 24:10
  35. Matt. 24:11
  36. Matt. 24:12
  37. Matt. 26:52
  38. Matt. 27:3
  39. Matt. 27:4
  40. Matt. 27:5
  41. Matt. 27:6
  42. Matt. 27:7
  43. Matt. 27:8
  44. Matt. 27:19
  45. Matt. 27:52
  46. Matt. 27:53
  47. Matt. 28:4

Adding these verses to the preceding, 105 of the 149 verses that Matthew added to Markan pericopes would seem to be independently-sourced, value-added material for him.

 

Possible Extrapolations

We now come to the most problematic of our categories, which consists of material that Matthew may have been able to extrapolate from what he had before him in Mark but that also could have derived from his own sources (including eyewitness memory or other testimony).

Much of this material deals with reactions, such as how different persons or groups reacted to what Jesus said and did.

If Matthew derived it from his own sources, then it would properly be grouped with the material in the other categories we have examined—particularly, the previous category, as independent material that Matthew saw as adding value to his narrative.

If he extrapolated the material from what he found in Mark, this also added value, but in a different way—for example, helping bring out the deeper significance of what happened in particular incidents in Jesus’ ministry.

It would not, however, represent the addition of new material to Markan pericopes in the way the previous three categories would have.

By my count, there are up to 37 verses that could belong to this category.

How many you think should belong to it will depend on how much freedom you think Matthew allowed himself to extrapolate from his sources.

My own feeling is that some of this material was likely derived from independent sources (meaning that, in the ideal, it should be reclassified into one of the above categories, particularly the previous one) but that some of it would have been extrapolated from Mark.

However, for purposes of testing our hypothesis that Matthew abbreviated Mark to include value-added material, I have been as generous as possible with the extrapolation hypothesis, thus erring on the side of Matthew extrapolating from Mark.

Here are the 37 verses:

  1. Matt. 3:2
  2. Matt. 3:14
  3. Matt. 3:15
  4. Matt. 9:26
  5. Matt. 12:5
  6. Matt. 12:6
  7. Matt. 12:23
  8. Matt. 15:23
  9. Matt. 15:24
  10. Matt. 15:25
  11. Matt. 15:31
  12. Matt. 16:12
  13. Matt. 16:27
  14. Matt. 17:6
  15. Matt. 17:7
  16. Matt. 17:13
  17. Matt. 18:3
  18. Matt. 18:4
  19. Matt. 21:10
  20. Matt. 21:11
  21. Matt. 21:20
  22. Matt. 21:43
  23. Matt. 21:46
  24. Matt. 22:33
  25. Matt. 22:38
  26. Matt. 22:40
  27. Matt. 22:46
  28. Matt. 26:25
  29. Matt. 26:44
  30. Matt. 26:50
  31. Matt. 26:53
  32. Matt. 27:21
  33. Matt. 27:25
  34. Matt. 27:26
  35. Matt. 27:43
  36. Matt. 28:2
  37. Matt. 28:3

Because of the ambiguous nature of this material, we cannot establish a definite number of verses that would have independently-sourced, value-added material.

If none of these verses were counted that way, we would still have 105 of 149 verses being independently-sourced, value-added material.

If all of them were counted that way then we would have 141 of 149 verses counted that way.

The truth is likely between these two figures.

 

Bridging Material

Our final category is what I am calling bridging material. These are verses that are likely extrapolated by Matthew to bridge one section of his narrative with another, based on what he had in front of him in Mark.

For example, Matthew 26:1 reads, “When Jesus had finished all these sayings, he said to his disciples. . . .” This bridges the end of the sayings discourse in Matthew 23-25 with the material that follows it in chapter 26. It is thus something Matthew likely derived neither from Mark (because it isn’t there) nor from other, independent sources but was extrapolated for literary purposes to bridge one section of his narrative with another.

By my reckoning, there are 7 such verses in the pericopes that Matthew would have taken from Mark:

  1. Matt. 8:1
  2. Matt. 13:18
  3. Matt. 14:18
  4. Matt. 15:29
  5. Matt. 22:34
  6. Matt. 26:1
  7. Matt. 27:36

With a high degree of probability, these represent verses that Matthew did not derive from independent sources but used to bridge material he found in Mark with other sections of his narrative.

They do not, however, affect the totals arrived at above: It would still appear that between 105 and 141 of the 149 verses that Matthew added to the Markan pericopes he used came from independent, value-adding sources.

 

Conclusion

While scholars might argue with the specific numbers offered above, it remains true that Matthew would have added between approximately 105 and 141 independently-sourced, value-added verses among the approximately 149 verses he would have added to the Markan pericopes he used.

In view of this, Matthew would not have extrapolated enough material from what he found in Mark to overturn the conclusion arrived in the first part of our test: It still appears that Matthew significantly shortened the material he found in Mark to include independently-sourced, value-added material.

Future research may change the numbers involved somewhat, but it is unlikely to change the fundamental conclusion.

If you liked this post, you should join Jimmy's Secret Information Club to get more great info!


What is the Secret Information Club?I value your email privacy

{ 2 comments }

The Masked Chicken January 1, 2016 at 8:37 am

Dear Jimmy,

I have been doing computational linguistics, recently, in my study of humor processing of jokes and I can tell you that there is much more powerful free software available to do what you want to do and much more. Excel is okay, but it is very simple. R is the premier free statistical software program. It is extensively documented, but the language is a bit odd. There are hundreds of free program packages available. Specific to Natural Language studies is the excellent Natural Language Toolkit, written in Python. It does many types of analysis using tokenized text (text split into different parts of speech). One can incorporate entire corpora (language lists), such as the Bibl,, for use in statistical analysis. There is a guided tutorial that comes with it. Another huge area in data mining is called, Formal Concept Analysis. This lets one couple attributes and objects in a chart and produce lattices showing the relationships between concepts. There are on-line site for this and excellent free software. Given that Matthew and Mark were written at different times, phenomena such as linguistic broadening and contraction also have to be looked at, etc.

If you want to try some of these more high powered software, let me know.

The Chicken

wineinthewater January 7, 2016 at 1:44 pm

It seems that one of the premises to this whole exercise is whether Matthew was actually shortening. Do we know that Matthew pushes against the size restrictions of contemporary literary technology? If he did not, then that removes one of the best explanations for *why* Matthew would abbreviate Mark. If that particular kind of difference cannot be ascribed to abbreviation, then that challenges the assumption that Matthew had Mark in front of him.

One theory that I like (in my glorious amateurness 😉 ) is that the shared material actually comes from the oral tradition, a sayings source. In this hypothesis, there is no Q, but rather an oral source shared by Mark, Matthew and Luke. Differences in those three sources result from the differences that arise from different carriers of that oral tradition. In that way, what is called Q is actually two highly harmonious versions of the oral tradition. And Mark’s brevity can be explained by being an earlier as well as different codification of that oral tradition. By the time Matthew and Luke come along, the oral tradition has been further fleshed out by the addition of additional witnesses gathered over the course of missionary work.

This also accounts for the apocryphal Thomas.

Previous post:

Next post: