Review of the Statistical Restoration Greek New Testament

I'm always excited when a new Greek New Testament comes out. It gives another opportunity to look carefully at the subtle differences between editions. This is a review of the hard copy of the Statistical Restoration Greek New Testament (SRGNT), edited by Alan Bunning (GlossaHouse, 2023), based mainly on a careful consideration of the text of Matthew.

The SRGNT is different from other editions because it is 'the first computer-generated text derived directly from the earliest manuscript witnesses using an algorithmic statistical model to simulate a reasoned-eclecticism approach' (back cover).

Bunning has done a great job entering primary data at www.GreekCNTR.org, a site which I used almost daily for two years when I was evaluating early orthographic information for the Tyndale House Greek New Testament.

At the very least such a computer-generated text, by differing little from the editions produced by human judgement, gives us assurance about the many parts of the Greek New Testament about which there is no dispute.

The edition aims to replace 'the subjective theological bias of human editors with the use of objective statistical and computational methods' (Introduction, p. i). However, computers also do stupid things when humans ask them to, which is why this edition contains hundreds of accentual mistakes and misplaced commas, most of which would have been caught by any careful human editor.

Admittedly wrong accents and commas aren't the most important aspects of a Greek New Testament, but they can be annoying.

In Matthew 5:39 we get this nonsensical comma:

ἀλλʼ ὅστις σε ῥαπίζει εἰς τὴν δεξιὰν σιαγόνα, σου στρέψον αὐτῷ καὶ τὴν ἄλλην·

Or γάρ beginning a clause in Matthew 10:23:

ἀμὴν, γὰρ λέγω ὑμῖν

Or δέ beginning a clause in Matthew 15:14:

ὁδηγοί εἰσιν τυφλοί τυφλὸς, δὲ τυφλὸν ἐὰν ὁδηγῇ, ἀμφότεροι εἰς βόθυνον πεσοῦνται.

There's an interrupting comma after the article in Matthew 25:20:

Καὶ προσελθὼν ὁ, τὰ πέντε τάλαντα λαβὼν, προσήνεγκεν ἄλλα πέντε τάλαντα 

The good news is that such mistakes can all be corrected in a subsequent edition. This is a first edition and can be built upon. As it says in the Introduction (p. ii) 'The SR can be regenerated in less than a minute reflecting all of the latest evidence'.

However, the algorithms are also capable of churning out nonsense readings in more substantial matters. In double square brackets, Matthew 16:2 ends thus: Εὐδία πυῤῥαζει γͅͅὰρ ὁ. The last word of the verse is the definite article and it has no accompanying noun. Good luck translating that!

Again, this is nothing that can't be fixed. But I would argue that this edition overvalues the goal of "scientific objectivity". That's not a problem if it's just an edition within a market place of editions. It can happily be used as a control on other more subjectively produced editions. But it is a problem if you remain convinced that a computer-generated text will give you the best edition.

Here are two examples of where the computer-generated text has produced something inferior to what humans would produce:

1) In Matthew 6:5 the SRGNT produces a switch from the 2nd person singular to the 2nd person plural:

Καὶ ὅταν προσεύχῃ, οὐκ ἔσεσθε ὡς οἱ ὑποκριταί

This unlikely reading may arise because the external attestation for the two verb forms has been treated as unconnected.

2) For Christ's dereliction cry the SRGNT has Ἐλωί, Ἐλωί, λεμὰ σαβαχθάνι in Matthew 27:46 and Ἐλωῒ, Ἐλωῒ, λεμὰ σαβαχθάνι in Mark 15:34. Literally the only differences arise through editorial inconsistency about diaeresis and accents. However, the deeper problem is having the same spelling of the Semitic expression for 'my God' in Matthew and Mark, which does nothing to explain the underlying manuscript differences.

But more positively the work of this edition can definitely help improve other editions. Despite taking great care in considering the orthography of the New Testament, here's a variation I had previously missed, and now know through the SRGNT:

Luke 10:13 Οὐαί σοι, Βηθσαϊδά

but Matthew 11:21 Οὐαί σοι, Βηθσαϊδάν.

The THGNT has Βηθσαϊδά in both cases. Arguably it's a detail we missed and the final nu should be there in Matthew.

So I want to thank Alan Bunning and his collaborators and cheer them on in their task, even as I remain firmly persuaded of the value of the human element in making good editorial decisions.

#SRGNT, #AlanBunning, #THGNT