Math trigger warning: For the Americans — there are numbers in this post, but you don’t have to calculate any yourselves.
So far we have shown that paying attention to the proportion of unique words in a short passage like the PA shows that it is not significantly different from other undisputed Johannine passages of similar length. Stories about unique topics tend toward unique vocabulary. However, the same metric is also incapable of showing that several passages from the Synoptic gospels were not from John. It appears that unique words are not a sensitive enough, or informative enough, metric to do what we want it to. Granted neither author in the dispute appeals to unique words alone, but in concert with other observations. What we might say is that the PA continues to pass Heil’s test, unlike, say, the longer ending of Mark or a passage from Josephus. It is just that passing this test is a fairly low bar.
One problem is that it considers only unique words, and then not with any sense of proportion. For example, παραγίνομαι appears in John 8:2. It also appears one other time in John 3:23, so it is not unique to the PA, but it is hardly Johannine either. However, it appears 12 times in the Synoptic gospels. We would expect it to appear only about three times as often in the Synoptics if the proportions were equal. Indeed Luke uses the particular form, παρεγένετο, three times on its own (8:19; 11:6; 19:16). So this term would seem to be substantially more in line with Synoptic usage than Johannine.
Therefore I set out to get a sense of how the weight of each word shifted the balance of the passage. John has roughly 15,471 words, while the Synoptics have a combined 49,024. I say “roughly” because I did not do a very thorough job of vetting the count (e.g. taking into account textual variants): I copied and pasted NA28 into Word and did a word count. Still, John has about 24.1% of the words in the gospels. If its usage of a particular word is proportional, then it should appear in John about 24.1% of the time. Significantly more, and we can say that it is statistically Johannine. Significantly less, and it is statistically Synoptic. An example of a significantly Johannine word is ἐρωτῶντες in 8:7. John uses ἐρωτάω 27 times outside of this passage (55.1%), the Synoptics only 22 (44.9%). A 95% confidence interval around the sample proportion of 55.1% does not go low enough to include the population proportion of 24.1%, so John uses the verb significantly more often than the Synoptics.
I went through the words of John 7:53-8:11 and did word counts on the ones that do not appear more than 500 times in the gospels – no methodological decision, just laziness on my part. Obviously if one were to apply a method like this with more rigor, a count of every word should be done. At any rate, this turned up 76 words, with the following results:
|Significantly Synoptic Words||29||38.2%|
|No Significant Difference||38||50.0%|
|Significantly Johannine Words||9||11.8%|
While half the words do not swing significantly one way or the other, more than three times as many words are significantly Synoptic than are Johannine. The character of the vocabulary is more in line with the Synoptics than with typical Johannine usage. Many of the Johannine words are not terribly interesting: “again” (πάλιν), “how” (ποῦ), “no one” (οὐδείς), and “now” (νῦν). But the argument regarding these words could go either way: on the one hand, they are not typical Johannine words, but on the other hand, they are not likely to be conscious markers of style or interest. In other words, they may be terms that an author is more likely to use as a tic, and so could still be stylistically useful.
By comparison, I did the same analysis with John 2:1-12, with the following results:
|Significantly Synoptic Words||24||30.4%|
|No Significant Difference||30||38.0%|
|Significantly Johannine Words||25||31.6%|
Here there is one more significantly Johannine word, with the results less skewed toward the middle. The Johannine words are also more substantial: disciple, Jews, water, know (οἶδα), keep, manifest, glory, and believe are among them. “Cana” is recalled in another passage (John 4:46), indicating that where the end of John 4 goes, 2:1-12 goes with it. It also appears in 21:2, and while some would argue that John 21 is not absolutely original to John (although never absent in the manuscripts), that it connects Nathanael (1:43-51) to Cana (2:1-12) suggests that the story in question was already in the manuscript even if John 21 were added after the fact. So despite some words that are odd for John but integral to the story (wedding, mother, call, and wine among them), the Johannine words still take the lead in a substantial way.
Another way to look at these words is to actually weigh them. For example, John uses the word “house” 4 times outside of PA, while the Synoptics use it 56 times. But the Synoptics have more than three times as many words as John, so of course it might appear there more often! However, if we scale the Synoptic number back proportionally to John, then 56 would scale down to 17.67, still more than four times more often than in John. Put another way, the difference is 13.67 in favor of the Synoptics. Paying attention to absolute numbers (although scaled) gives a sense of relative weight. If John uses “Cana” twice outside of John 2:1-12 while the Synoptics never use it, and John uses οἶδα 82 times to the Synoptics’ 71 times (22.4 scaled), they are both significantly Johannine, but οἶδα is more heavily Johannine. It carries more weight.
Applying this to the vocabulary of both stories draws the following results, with a negative result leaning toward the Synoptics and a positive result toward John:
The score for John 2:1-12 leans quite a bit more heavily toward John than does the PA toward the Synoptics, but the PA still leans toward the Synoptics. While the wedding at Cana has a significant result (p = 2.8%), PA is not quite significant at α = 5%. Then again, this was rather sloppily done. With more attention to detail, PA’s p-value could feasibly slip below 5%. A more refined study would be necessary before drawing any hard conclusions, but it would appear that, when taking more information about the vocabulary of each passage into account, the PA is less like John than unique words alone would suggest.
I don’t think the PA is original to John. The external evidence alone convinces me of that. If the language were more characteristically Johannine, I might agree that it was written by John’s group as part of a later addition, or at least in imitation of John in order to be fit into John. But some manuscripts of John lacking the thing would need to have gone out. Since the story floats around so much, it is more likely that it was a bit of free-floating tradition that ended up in various places in John more often than in Luke or attached to the end of the gospels.
People really love this story. It is an example of Jesus showing love and compassion to a lone woman about to have her bones literally crushed in by rocks, ganged up on by a team of people just trying to make a point. Bracketing it and putting it to the side makes them uncomfortable, and I appreciate that.
That it is not original to John is perhaps not the lesson we should take away from it. Say John was completed in the 90s, without this story. It may have survived for centuries before finding various places to be fit into John (and outside of it).
That means that the story survived for that long without the authority of being incorporated into a gospel. There were probably one-sheet versions of it floating around, but people probably just liked the story and told it to each other. They liked a story where the pronouncement was, “The one without sin should cast the first stone” — and nobody does. They liked the story despite its lack of authority, and they worked against the manuscript tradition in order to preserve the story and to give it authority.
That story does not make me uncomfortable.