How did I end up thinking about stock phrases, and in Christian apocryphal texts at that?
Well, I was (and still am) working on an introduction and translation to the First Apocryphal Apocalypse of John (1AAJn) for volume 2 of Burke & Landau’s New Testament Apocrypha: More Noncanonical Scriptures.
One of the things I wanted to do was take a more rigorous look at intertextuality. And by “rigorous” here I mean “write some code and compare a lot of data.” Based on way-old experimentation of mine from 10–15 years ago (anyone remember “tri-logs?”) and more recent popularization of n-grams by Google, I thought it would be interesting to compare ngrams (but form the n-grams based on Greek lemmas instead of the inflected forms in the text) between a base corpus of the LXX and NT and noncanonical works like 1AAJn. My straightforward theory was that there would be clusters of n-grams located in the noncanonical works and those would represent probable intertextual units.
It all seems well and good, and it even works out. Here’s a basic dump of the larger intertextual relations I found between the LXX+NT and 1AAJn (note LXX references use the LXX versification scheme, so Psalms is weird). The number in parentheses is the number of clustered n-grams from the LXX+NT reference in the 1AAJn reference.
- 1AAJn 8.4 possible reliance on Psalm 88:45 (7x)
- 1AAJn 8.4 possible reliance on Psalm 88:46 (5x)
- 1AAJn 12.5 possible reliance on Psalm 102:15 (12x)
- 1AAJn 12.5 possible reliance on Psalm 102:16 (6x)
- 1AAJn 12.6 possible reliance on Psalm 145:4 (15x)
- 1AAJn 13.7 possible reliance on 1 Thessalonians 4:17 (9x)
- 1AAJn 15.6 possible reliance on Psalm 50:9 (7x)
- 1AAJn 15.7 possible reliance on Isaiah 40:4 (11x)
- 1AAJn 15.7 possible reliance on Luke 3:5 (16x)
- 1AAJn 15.7 possible reliance on Isaiah 40:5 (4x)
- 1AAJn 15.7 possible reliance on Luke 3:6 (4x)
- 1AAJn 16.4 possible reliance on Matthew 24:30 (9x)
- 1AAJn 19.4 possible reliance on Psalm 101:26 (8x)
- 1AAJn 19.4 possible reliance on Hebrews 1:10 (8x)
- 1AAJn 19.4 possible reliance on Psalm 101:27 (7x)
- 1AAJn 19.4 possible reliance on Hebrews 1:11 (7x)
- 1AAJn 21.4 possible reliance on Matthew 28:19 (4x)
- 1AAJn 21.5 possible reliance on Psalm 9:18 (9x)
- 1AAJn 21.6 possible reliance on Psalm 48:15 (5x)
- 1AAJn 22.4 possible reliance on Psalm 17:42 (9x)
- 1AAJn 22.5 possible reliance on Romans 2:12 (9x)
- 1AAJn 23.3 possible reliance on Psalm 124:3 (5x)
- 1AAJn 25.3 possible reliance on Psalm 36:29 (7x)
- 1AAJn 26.2 possible reliance on Deuteronomy 32:8 (4x)
- 1AAJn 26.2 possible reliance on Odes 2:8 (4x)
- 1AAJn 27.6 possible reliance on John 10:16 (15x)
- 1AAJn 28.2 possible reliance on Psalm 105:3 (6x)
This is all fine and good, and pretty much what I expected to find. What I didn’t expect to find was how several of the smaller or even single clusters represented some sort of common almost Bible-ese phrasing. Like “the heavens and the earth” or “the moon, the sun, and the stars” or even “upon the face of all the earth” (which happens frequently in the LXX and also in 1AAJn).
These smaller clusters of ngrams didn’t seem to have particular value in the exact reference or context; it was more like the phrasing of the canonical material was being (deliberately?) used in the noncanonical material. It was using biblical-sounding language to describe biblical sorts of things in noncanonical material. It could be a clue to register, or it might be an effort to make this noncanonical thing sound canonical, or it might just be using common language (a stock phrase?) to talk about common things. I’m not sure, but it was interesting enough it seemed like someone should look into it further. So I pitched the paper. And I’m still digging.
If you have books or articles to point me to, I’m all ears. Even better (especially for articles) if you can send along PDF too. Let me know in the comments.
3 thoughts on “Why a paper on “Stock Phrases” in Christian Apocrypha?”
That’s exactly how I identified multiple versions of songs I’m all faithlife’s hymnal (or at least ranked them by simularity so that Elizabeth (?) could make a human judgment. My algorithm eliminated common words like “a” and “than” and stemmed all other words before making the comparison. I think it would have been interesting to use wordnet to canonicalize synonym groups too. I’m sure my code is still laying around somewhere in the python or dJango projects.
I have a plan to do the same analysis to paul’s letters to create a synoptic table of them. I’ve got a lot of infrastructure I have to build out before I can do that though. 😦
Wow, the typos. I really shouldn’t try and type whole paragraphs on my phone.