May 20, 2019

Graphically Displaying Synoptic Data

A surprising thing popped up on my newsfeed today. (Yes, I still consume my RSS newsfeed daily. Call me a dinosar.) Mark Goodacre has reanimated his NTBlog for the purpose of reflecting on Matthew Larsen's book (which I have not seen). Mark @Goodacre has since microblogged (ie. Tweeted) on one particular feature in Larsen's book as a teaser for his next blog post, namely, the displaying of proportional synoptic data between two gospels. How much of Mark is in Matthew? How much of Matthew is in Luke? Etc. He has repurposed (colorized) Larsen's "proportional Venn diagram".


I do so enjoy such discussions, but seeing Larsen's graphic prompts me to revisit some work I have done for years but never completed on models for visualizing synoptic data.
I would like to publicly come out (momentarily reanimating my own blog in the process) in favor of rectangular Venn diagrams for comparing Gospel Synoptic Data. To explain:
Venn diagrams are most commonly thought of, and produced, as overlapping circles. This is because circles are gorgeous and tremendously satisfying.
However, a Venn diagram can be comprised of any shape (or variety of shapes). And given that the goal of comparing synoptic data is to, well, compare it, a rectangular application of Venn diagrams serves much better. Two primary reasons:

1. Visualizing proportionate size

Circles are fine for highlighting "relative" sizes, but not proportional ones. The areas of circles are very difficult to ascertain. To illustrate, look at these two diagrams:
In the first diagram, what proportion of the blue shape is overlapping with the red shape? What proportion is uniquely blue, or uniquely red?
One third. One third each. It's easy to tell. Immediately.
Now, what proportion of the blue circle is overlapping with the red circle?
Uh... er... Hmmm... "Honey, where are my calipers?"

2. Compare diagrams to each other

In the same way that rectangular shapes are easier to get a sense of their not just relative, but also proportionate size within a diagram, two or more diagrams beside each other can also be better evaluated in rectangular format.
To illustrate, let's look at the synoptic data that Mark Goodacre and Matthew Larsen are wrestling with. They are looking at how many "stories" or pericopes in one gospel are also in another. There will be stories unique to each, along with some overlap. This can be seen in the graphic tweeted above, where the unique blue Matthew passages and the unique red Mark stories overlap with purple. Likewise, yellow Luke pericopes and blue Matthew pericopes in another graphic would also share green overlap.
Ignoring for the moment how horrible it is to do this kind of comparison based on the pericope divisions in the Aland Synopsis (which doesn't inspire me to read Larsen), let's have a look at all three Venn diagrams (Mt-Mk, Mt-Lk, Mk-Lk) in the much more useful, boxier formats:


There are a lot of decisions that go into making such a diagram. I won't go into all those. But look how easy it is to compare the relative differences in each section. I don't think I need to say more about the graphics… they speak for themselves.

See here for a followup post on visualizing a comparison of all three gospels using a rectangular Venn diagram.

Couple notes:
I should say that my count of Aland pericopes shared among Mark and Luke was hasty and is bound to be off some, but I find that data point so useless, I didn't care to spend more time on it.

I've thought Mark Goodacre was brilliant for years, and consider myself his student though I've never been in his classroom, but I knew I was going to like Mark the first time I read his stuff many years ago and saw that he advocated the same primary-color-based system for marking a gospel synopsis that I had begun using in seminary (to Gene Boring's dismay). Though, I prefer to use 60% transparent white versions of the primary colors for true blending, because, you know, ugly.


1 comment:

TonyTheProf said...

There is some interesting work going on in computational linguistics looking at textual comparison - one example

https://onlinelibrary.wiley.com/doi/full/10.1111/j.2041-5370.2016.12042.x
DOCUMENTING HOMERIC TEXT‐REUSE IN THE DEIPNOSOPHISTAE OF ATHENAEUS OF NAUCRATIS

In particular, I note "The possibility of aligning Athenaeus’ reuses to the Homeric texts shows the challenges that philologists have to deal with when trying to establish a rigorous method for annotating text‐reuse phenomena."

The section on "Example: reordering and collision" I found especially interesting.

https://www.etrap.eu/historical-text-re-use/
Historical text reuse: what is it?

Another interesting article with a rather neat diagram.

Text reuse within a single language also shows in a table how the same concept is formulated and reused in the Ancient Greek authors Alcaeus, Plato and Theocritus.

And not to be parochial and Western
https://academic.oup.com/dsh/article-abstract/33/3/670/4583485?redirectedFrom=fulltext

Text reuse in early Chinese transmitted texts is extensive and widespread, often reflecting complex textual histories involving repeated transcription, compilation, and editing spanning many centuries and involving the work of multiple authors and editors.

In conclusion, once we move from direct reuse of one text by one author of another to the area of near similarity, there are still computational mechanisms which are being developed to show the likelihood of sources being reused.