Base Texts for Transcriptions Again

In a previous post on “Methodology in Transcribing Greek Manuscripts” I suggested that the use of two different base texts, Nestle-Aland and a Byzantine text like Oxford Textus Receptus 1873, for independent transcriptions of a manuscript might result in the elimination of errors resulting from the use of the same base text where transcribers would risk making the same mistake in trusting their base text too much.

In any case, a reader of our blog, Anthony Pope who for some technical reason could not comment on that post sent me a relevant example I could share of an error  in the Editio Critica Maior of Acts, 28:13 (4) which would likely have been avoided using two base texts:

a περιελοντες 01*. 03. 044. 6. 93. 321. 383. 398. 665. 1127. 1609. 1838. 1842. 2200. 2774

Only the uncials listed read περιελοντες. All the minuscules listed read περιελθοντες, as can be verified from the images on the NTVMR. The error has been reported to the editors and will be corrected in the next revised edition.

Do any readers have other examples or thoughts? Share them in the comments.

Methodology in Transcribing Greek Manuscripts

About two weeks ago I participated in a meeting in Volos, Greece, with the groups that are currently editing the Novum Testamentum Graecum Editio Critica Maior (from Münster, Birmingham, and elsewhere). We had a lot of stimulating presentations and discussions. At some point in a discussion I repeated an old suggestion I have made that independent transcriptions of manuscripts which are later reconciled (i.e., compared by software with each other and checked by a third person who resolves any differences between the files in order to create one reconciled transcription) could be transcribed from two different base texts – the Nestle-Aland text and a Byzantine text.

Personally, when I worked on my dissertation on Jude I transcribed some 560 manuscripts, most of which naturally followed the Byzantine majority, and so I used the Textus Receptus (Oxford, 1873) as a base text which I then edited for each manuscript. I did not want to use the Nestle-Aland text as a base text because I reckoned I would have to edit each Byzantine minuscule much more (i.e., more work), and when editing I would also risk introducing errors.

In fact, I transcribed first on paper because the computer tools for transcribing were not as good, and the paper method actually saved me precious time. I used a base text on the left side of the paper and recorded variants by underlining the text that had variation and recording the variant on the right side of the paper. The most common variants were already listed in the right margin so that I could just put a circle around the one attestested by the manuscript I transcribed ... I saved valuable "Münster-time" (yes, in these days you had to go to the INTF in Münster to collate manuscripts, there was no NT.VMR). When I came home I entered everything into the software Collate 2.0 developed by Peter Robinson (who sent me his software and manual for free – Kudos to him, I could not have done this without his software). I learnt the paper-transcription trick, i.e., to have the basetext with prepared variants on paper, from no other than the transcription master Maurice Robinson (who, to my knowledge, has not yet entered his collation data of the pericope adulterae into digitial form, but hopefully this will happen). UPDATE: Maurice informs us in a comment that all data has been digitized and he is waiting for an interested publisher.

In any case, I know that the INTF have used the Nestle-Aland text as a base text for a long time, and at the meeting in Volos someone pointed out that it is actually good that the manuscript to be transcribed differs from the base text – this will keep the transcriber alert and they will do less mistakes and this may well be the case. On the other hand, there was an openness to consider my proposal to use two different base texts for the initial independent transcriptions, at least it could make for an interesting experiment.

Here is my own experience as of today. I have just trained a student how to read and transcribe a manuscript in Philippians with the Online Transcription Editor, not the editor which is integrated in the NT.VMR, but the freestanding OTE (http://www.itsee.birmingham.ac.uk/ote/). There you upload a base text, transcribe your manuscript (using images from NT.VMR or elsewhere) and then export to an xml.file which you name after the manuscript you transcribe. As I was proofreading the first attempts by this student today I found the following seven errors (among others) which would likely have been spotted in a reconciliation where two students had used two distinct base text according to my proposal (NA28 and Textus Receptus):

Minuscule 365

1:27 ακουω (= base text NA28);
the manuscript reads ακουσω (= Byzantine text)

2:5 φρονειτε (= base text NA28);
the manuscript reads φρονεισθω (= Byzantine text)

2:23 αφιδω (= base text NA28);
the manuscript reads απιδω (= Byzantine text)

2:30 παραβολευσαμενος (= base text NA28);
the manuscript reads παραβουλευσαμενος (= Byzantine text)

3:6 ζηλος (= base text NA28);
the manuscript reads ζηλον (= Byzantine text)

3:10 συμμορφιζομενος (= base text NA28);
the manuscript reads συμμορφουμενος (= Byzantine text)

4:15 λημψεως (= base text NA28);
the manuscript reads ληψεως (= Byzantine text).

As I proofread I found another very interesting variant in Philippians 1:14. The manuscript does not read πεποιθοτας (both NA28 and Textus Receptus), but πεπονθότας. If I am not mistaken this translates, "and most of the brothers and sisters, having suffered in the Lord by my imprisonment, dare to speak the word with greater boldness and without fear." This variant is not listed in von Soden's edition. I haven't checked Tischendorf's 8th edition.

Differences between ECM Acts and NA28

I need some help from your collective wisdom ...

On pp. 34*-35* of ECM Acts we find a table of Textual Changes, listing differences between ECM and NA28. This is followed by a list of passages with a split guiding line. Many of the textual changes are straightforward, but there is a grey area, signaled by the presence of two distinct lists rather than one.

(1) The entry 2,33/43 says ECM -, NA28 [και].
This is a textual change. The text that is printed is different in the two editions, though the two are not miles apart. It is not that ECM has adopted a reading that was offered as an equal possibility in NA28, which mistakenly we might assume given the square brackets around και in NA28, but the omission of και is signaled in the text. It is good to remember that square brackets do not give a full alternative but rather 'reflect a great degree of difficulty in determining the text', and 'the reading given in the text shows the preference of the editors' (NA28, 54*).
This change is slightly 'bigger' than the change in the next example.

(2) The variant unit 1,11/16 is not given as a textual change:
ECM εμβλεποντες, ΝΑ28 [εμ]βλεποντες
Both ECM and NA28 show the - identical - preference of the editors, though NA28 signals that its editors had a great degree of difficulty. There are no square brackets in the ECM, but do we have a split guiding line? Not here. The difficulty of the past was not large enough to result in a split reading in the present. Technically this is not a textual change. Still, even though there is no difference in the printed text, there is certainly a perceived difference in confidence.

(3) Now take 2,7/5, also not given as a textual change in the ECM list, though here I think there is in fact a change:
ECM - ♦ παντες, NA28 -
To me this looks like a change. A split guiding line means that the editors 'could not determine one variant which most likely represents the initial text' (ECM, 18*), which is close but not identical to the NA28 definition of the square brackets. The split guidance line is a material change to the text of NA28.

(4) The final example is 5,28/4
ECM ου ♦ -, ΝΑ28 [ου]
Since the definitions of square brackets and split guiding line differ subtly, I think technically there is a change here too. The ECM split line signals that there is no reading that is 'most likely', whilst the bracketed [ου] still signals the preference of the editors of the NA28 for the printed reading. Of course, most of us would be tempted to read ECM and NA28 as telling us the same, but I do not think they do.

Example (1) is found in the listing of textual changes in the ECM, example (2) is not listed in the print edition, and (3) and (4) are found in the listing of passages with a split guiding line.

My question after all this is, what is more accurate, to say that ECM differs from NA28 in 52 places and presents a split reading in a further 155 places, or that the text of ECM differs in 207 places from NA28? [warning: I counted the list manually, so I may be a little off, but I hope you get the question.]

Editio Critica Maior: Acts announced

The Editio Critica Maior for Acts appears in the latest German Bible Society catalogue (HT: Greg Paulson on FB). It is due for publication in August 2017. We look forward to this and encourage and congratulate all those involved.

There are two main volumes on the text itself comprising 1,180 pages (there are 1,006 verses in Acts, so around one A4 page per verse). There is a second volume on materials. And (new for Acts compared with the Catholic Epistles) is a third volume on Studien. It is interesting that this is not entitled Kommentar, since I guess I had expected something like a textual commentary. It will be interesting to see what this volume contains. It is a bit disappointing that no scholars are named, only the Institut fur Neutestamentliche Textforschung - no doubt reflecting the hard work of many and the collective mode of textual decision-making, but I hope that contributing individuals will be named on the title page inside.

A Three Part Correction to the Editio Critica Maior at 1 John 3.24

I use the Editio Critica Maior frequently and I continue to be impressed by how well it combines clarity with great detail. Yes, it takes a bit of getting used to its format (as with any large apparatus), but once you’ve got the basics, it’s great.

But occasionally, I do find that it has mistakes. 1 John 3.24 is a case in point. The ECM records the first hand of 180 and both Syriac versions as attesting a dittography of the first ἐν αὐτῷ in καὶ ὁ τηρῶν τὰς ἐντολὰς αὐτοῦ ἐν αὐτῷ μένει καὶ αὐτὸς ἐν αὐτῷ. But, in fact, a check of Aland’s edition of the Syriac shows that there is no such dittography. Both versions have the equivalent of each occurrence of ἐν αὐτῷ (ܒܗ) just once. I suspect this was a simple case of the Syriac sigla being put with the wrong variant address. But there’s more.

Since I was checking, I tried looking at the images for GA 180 in the VMR but quickly realized that what I was looking at was a manuscript of only the Gospels. This led me to Elliott’s Bibliography where I learned that part of GA 180 has a new number: GA 2918. As best I can tell, GA 2918 is in the same codex as 180 (Borg. gr. 18) but, since the Gospels are in a different hand of a different date, they now have a different GA number. Apparently the change happened after the ECM was printed, so it still reflects the old number.

Checking the images for 2918, then, I found a third problem. Ιt looks to me like the scribe of 2918 has not written the first ἐν αὐτῷ twice, but rather ἐν αὐτοῦ, his eye jumping to the wrong αυτ-. The ECM is right about the correction though; it’s just the original error that’s incorrect.

1 John 3.24a in GA 2918 (formerly GA 180) showing the deletion marks around εν αυτου

All this means that if you were only reading the ECM apparatus, you would be led astray in three ways: (1) you would have the wrong reading for 180*; (2) you wouldn’t know that GA 180 has been split numerically in two (and so wouldn’t find it in the VMR); and (3) you would wrongly think that the Syriac tradition agreed with it in error. Thankfully, with just a little digging, all three problems can be spotted. Maybe in another printing it can be fixed.

I suppose the lesson from this is that producing a good critical apparatus is complicated work with numerous points at which the process can break down. A second lesson follows from this: check your sources!

Who Said It?

A little fun for your Friday. See if you can guess the source of the following quote. (Remember, it’s no fun if you ask Google.)

“If the primary purpose of this discipline is to get back to the original text, we may as well admit either defeat or victory, depending on how one chooses to look at it, because we’re not going to get much closer to the original text than we already are. Barring some fantastic manuscript discoveries (like the autographs) or some earth-shattering alterations in text-critical method, the basic physiognomy of our texts is never going to change.”

Update: That may have been too easy. But Bart Ehrman it is, in his review of the first fascicle of the ECM (delivered at SBL, I think). 

Index of Variants Discussed in Relation to the CBGM

Local stemma for 2 Peter 3.10/48-50

Following a question on Facebook regarding whether the CBGM had influenced the conjecture at 2 Peter 3.10 (οὐχ εὑρεθήσεται), I thought it would be good to have an index of all the places where there has been discussion of the CBGM in relation to particular variants. At the moment, this is as close as we can get to having any kind of textual commentary on the Catholic Epistles. Reading through the examples from the editors really is the best way to get a sense for how the CBGM is influencing their textual decisions. So having them all in one list is handy.

I’ve split the list into discussions (a) published by those working on the Nestle/ECM text and (b) those who aren’t. The format here is pretty straightforward. After each verse reference I give the source number followed by the page number. Some of these are more illustrations than discussions, so I’ve tried to mark with an asterisk (*) those places where the source is most explicit about how the CBGM has influenced a particular decision. Most of them are, however, still very brief. Let me know if I missed any.

Update (3/2016)

the best single source for discussions is now Klaus Wachtel’s article on Acts published in the TC journal (2015). It’s #11 under Discussions by the Editors.

Update 2/2019

This list can also be found at the end of Peter J. Gurry, “How Your Greek NT Is Changing: A Simple Introduction to the Coherence-Based Genealogical Method (CBGM),” JETS 59, no. 4 (2016): 675–689.

Editio Critica Maior: changes between the 1st and 2nd editions

The 28^th edition of the Nestle-Aland text presents, for the Catholic Epistles, the text of the 2^nd edition of the Editio Critica Maior (not yet published). A comparison of the text of ECM¹ with the text of NA²⁸ (= ECM²) indicates that there are 13 differences between the first and second editions, as follows:

Reference	NA28 (ECM2)	ECM1
Jas 1:20	οὐ κατεργάζεται	•οὐκ ἐργάζεται•
Jas 1:22	♦μόνον ἀκροατὶ	ἀκροατὶ μόνον
Jas 2:4	καὶ οὐ διεκρίθητε	οὐ διεκρίθητε
Jas 2:15	λειπόμενοι ὦσιν	λειπόμενοι
Jas 4:10	τοῦ κυρίου	κυρίου
1 P 2:5	θεῷ	•τῷ• θεῷ
1 Pet 5:1	τοὺς	οὖν
2 P 2:18	ὄντως	•ὀλίγως•
2 P 2:20	κυρίου	•κυρίου ἡμῶν•
1 J 3:7	Παιδία	τεκνία
2 J 5	γράφων σοι καινὴν	καινὴν γράφων σοι
2 J 12	ᾖ πεπληρωμένη	πεπληρωμένη ᾖ
3 J 4	ἐν ἀληθείᾳ	ἐν τῇ ἀληθείᾳ

It is striking that 8 of the 13 differences between the two editions are not marked by either a diamond (♦)—which “indicates passages where the guiding line is split in the second edition of the ECM, because there are two variants which in the editors’ judgement could equally well be adopted in the reconstructed initial text” (NA²⁸, “Introduction,” 55*)—or  bold dot(s), which sometimes signal alternative readings of equal value to the text and sometimes simply mark passages calling for “special critical consideration” (ECM, 24*). Furthermore, of the 5 readings that are marked by bold dots or the diamond, none are marked by both dots and a diamond.

A comparison of bold dotted and diamond readings throughout the Catholic Epistles produces the following results (all numbers subject to confirmation):

Total # of readings with diamond, dot(s), or both:           107

# of diamond readings with no corresponding dot(s):       19

# of readings with both diamond and dot(s):                      22

# of dotted readings with no corresponding diamond:       66

That is, of the 107 marked variants, only 22 (20.5%) share the diamond and dot(s), while 85 (79.5%) have either one or the other, but not both.

These numbers indicate that the means of signaling uncertainty in the text—dotted readings in ECM¹, and “split guiding line” readings in ECM² (= the diamond readings in NA²⁸) overlap in only about 1 in 5 instances. Of the 88 “dotted” readings marked in ECM¹, only 22 of them (25%) are marked in ECM²—a noticeable reduction. But at the same time, of the 41 diamond readings marked in ECM², nearly half,  19 (= 46%), are new to that edition—a substantial proportion of the whole. Finally, one must keep in view the fact that of the 13 differences between ECM¹ and ECM², 8 (61%) are not marked (with a diamond or dots) in either edition of ECM.

Mike Holmes

SBL Chicago I, Novum Testamentum Graecum: Editio Critica Maior pt 1

Here is my first report on the SBL in Chicago. The first session I attended was:

Novum Testamentum Graecum: Editio Critica Maior
11/17/2012
9:00 AM to 12:15 PM

Holger Strutwolf, director of the INTF in Münster presided and Zeth Green, University of Birmingham, Troy Griffitts, INTF Münster, and Hugh Houghton, University of Birmingham presented the first part on:

Technical Developments: the Workspace for Collaborative Editing and the NT.VMR
 
Below is a summary of my notes. Note that they are based on the presentations but supplemented with information available elsewhere, not least in the on-line NT.VMR 2.0.
 
1. Introduction

Collaboration for the Editio Critica Maior between the INTF in Münster and ITSEE in Birmingham came about because of the adoption of the same software (Collate developed by Peter Robinson), which began over a decade ago.

Further steps have been required to continue the collaboration. The first is the development of the New Testament Virtual Manuscript Room (NT.VMR, now in version 2.0) which includes the Kurzgefasste Liste. The website invites community collaboration to improve access to the raw material. The NT.VMR is an AHRC funded project.

A related project is the Workspace for Collaborative Editing (WCE), co-funded by the AHRC in the UK and by the DFG in Germany. The goal is to create a suite of nine on-line tools (Work packages 1-9) primarily to enable the International Greek New Testament Project and the Institut für neutestamentliche Textforschung to work together on the Editio Critica Maior of the New Testament. 

The software development is being undertaken in Birmingham by ITSEE and in Trier by KoZe (Kompetenzzentrum für elektronische Erschließungs- und Publikationsverfahren in den Geisteswissenschaften). [Presumably, all these tools will be integrated into the NT.VMR. Subsequently, the tools can be used for other textual traditions as well.]


2. Workspace for Collaborative Editing (WCE)

 

Zeth Green and Hugh Houghton presented some of the components of the WCE:

A basic stage, after determining the content (of the whole MS and of each page) and material of a manuscript and assigning it an official Greg.-Aland number, is to handle the transcription of its text (Work package 2). This transcription work is done using unicode fonts and xml encoding following the TEI initiative). Naturally, transcribers cannot work in xml directly. Therefore, the team in Trier has developed a transcription editor which saves transcriptions in xml (for an example of this output, see here).

Another stage is the collation (comparison) of manuscripts. Here CollateX (Work package 3) is replacing Collate 2.0 which was originally developed by Peter Robinson (Robinson is also involved in developing the new version). Members of the team in Birmingham are implementing CollateX into the VMR. 

Hugh Houghton gave a demonstration of CollateX. The tools seems to be very convenient. First he did a number of regularizations (telling the software to eliminate genealogically insignificant variants relating to orthography, etc, before the comparison). After regularization, Houghton demonstrated how easy it is, technically, to combine and adapt the desired length of variation-units.

However, from a philological standpoint, it is sometimes not so easy for an editor to decide what segment of text should be compared to what. To borrow something from another  presentation, this work should be based on the principle expressed by Klaus Wachtel, “as short as possible, as long as necessary,” taking into account two corresponding factors: grammatical sense and the attestations of the manuscripts.

I had a question relating to this particular step: does the software allow you to choose one variation-unit within another? This feature has not yet been developed, Houghton replied, but it will.

After running the collation of the texts as divided into the desired variation-units, CollateX will create a textual apparatus in a raw format, which is then exported in a format for further use by the editors, for example, the application of the Coherence Based Genealogical Method (software included in Work package 5) in the variation-units, in order to determine the relationship of whole texts and individual variants and to reconstruct the initial text (Ausgangstext). Part of the final steps is to produce another electronic tool for publishing the apparatus (Work package 7).

3. NT.VMR 2.0
Troy A Griffiths demonstrated the NT.VMR 2.0 (NT.VMR 1.0 here). The manuscript room has four basic components which were demonstrated:




1. The manuscript catalogue (Liste)
2. Images of New Testament manuscripts (and the possibility to add links to images, e.g., by pasting an URL)

[These two components are accessible under the meny "manuscript workspace"]

3. Indexing tools). For a detailed account of this component, see Jan Krans's account here.
4. Transcription editor – a simple interface to enable anyone to create electronic versions of the manuscript text

[The result of indexing and transcriptions are integrated into the manuscript workspace and associated with each manuscript. Apparently, Work packages 1-2 are already integrated into the NT.VMR 2.0]
 

This first part was followed by three presentations by Klaus Wachtel, Siegfried Richter, and Martin Karrer. To be continued ...