Wednesday, February 01, 2006

Why spelling matters

No, this is not a reproduction of the lecture I give to my students when they hand in essays in bad English...

I want to argue here that the spelling of the Greek New Testament matters more than is usually reckoned.

Background: All editions of the GNT since printing began have sought to produce an NT with consistent spelling. Though there are various spellings in the manuscripts, editors usually plump for one and stick with it all the way through.

It is true that with words of rare occurrence different editors may accept different spellings in different places (e.g. the spellings of Rahab in Hebrews 11:31 and Matthew 1:5). There are also names for which different spellings are produced, e.g. Jerusalem or Mary. On the whole, however, when witnesses are regularly divided spelling is decided globally, despite fluctuations from text to text in the manuscript support for the general decision the editors have made. Thus Westcott and Hort spelled David's name Δαυειδ but this is always spelled Δαυιδ in Nestle-Aland. It is as if scholars woke up one morning and all of a sudden the evidence for the spelling in each passage changed.

This does not merely affect proper names, but also many verb and noun forms (e.g. forms of λαμβανω), and particularly the question of the moveable nu. There are literally hundreds of words in the GNT with varied possibilities for spelling. Some phenomena, such as the spelling of δε or αλλα and whether the final letter elides, offer a literal wealth of data.

Why does this matter? After all, spelling variants do not affect meaning.

The reason it matters are as follows:

1) Spelling may tell us something about canon history. David Trobisch has argued that the common presence of nomina sacra in manuscripts points towards a 'canonical edition'. Even if, like me, one is sceptical of his case, one can still admit that spelling may tell us something about which books were collected when. Letters by the same person might contain varied spelling (authors are not always consistent, and varying amanuenses might be used). If, however, there is a significant homogeneity of spelling this may point to a common scribe (perhaps someone who brought letters together) or author. To what extent is spelling uniform across the fourfold gospels? What about Luke-Acts? What about Hebrews in relation to the Paulines? These are all questions that could be addressed by research on spelling.

2) Spelling may tell us something about literary dependence. If there is an unusual congruence of spelling between two works, but not between others, this might show us something in relation to literary dependence. This might be used in the Synoptic question.

3) Spelling may help us detect secondary variants. Obviously, this could only be used alongside other criteria.

4) Spelling may help highlight citation. We have of course the fascinating case of the use of ειπον 'I said' in John 10:36 following the Septuagintesque (Bible-sounding) ειπα 'I said' in John 10:34. This may be somewhat akin to the difference for us between 'spoke' and 'spake'. How many more of these are there that are not recorded in editions?

5) Spelling data may tell us about the attitude of the copyists towards the text. We know from a comparison of P75 and B that spelling could be faithfully copied. How widespread was the view that the letters of the text should be copied faithfully? Can this be connected with a theology of the text?

6) Spelling data may help us differentiate between Patristic citation by memory and Patristic citation by consultation of a manuscript.

etc.

There are a number of other reasons why spelling data are important. I think that this may be a significant area of untapped data. Spelling is a domain in which theological debates are hardly likely to have made inroads. Moreover, despite the fact that scribes clearly did frequently change spelling, there is also plenty of evidence of conservatism, which is why some spellings widely used in Classical literature are hardly found in NT mss. The absence of the iota adscript, though more common in inscriptional material, is particularly noticeable. Since the spelling data taken over the whole NT gives us thousands of pieces of information it is highly unlikely that all historically useful information has been lost from our manuscripts. This is so, even if we take a pessimistic view of their transmission.

This is an area where there has been virtually no interest and virtually no research, but there is almost certainly lots to discover. It is also an area where an optimistic view of the transmission of the NT (such as an evangelical might tend to possess—though it is not exclusively found in them) is a far better stimulus to research than the pessimism that seems to dominate academic discourse about Christian origins. The critical consensus may be said to have tended to lead to complacency.

22 Comments:

Carl W. Conrad said...

One problem, or at least a consideration, in this regard is that a copyist may be writing out what he "hears" his model to be saying rather than the characters he scans. The papyri certainly show that the same word can be spelled differently by a writer within the same letter. If the model is being dictated, then the spelling may be even more hazardous.

Eric Rowe said...

Perhaps a project could be done that collates JUST spelling information. The editors could pick a few hundred key words whose spellings are known to vary among mss, and compare their spellings accross as many mss as possible. This could be done in a single volume. It could use shorthand notation developed just for this purpose. Actually, this might be a very good work to do on computer and to make available to scholars as a database whose data can then be arranged either according to word or according to ms. Then scholars could investigate any of the various issues raised in Pete's post. They could also group mss into families according to the spelling phenomena of their choice. This could allow scholars to attempt to associate textual families with geographic regions according to known conventions (I don't actually know if this would work, but it sounds reasonable to me). Also, a specialized work like this would spare the average GNT reader from having to endure such minutiae as spelling changes.

Regarding Carl's comment, I'm not sure why a scribe who writes what he hears would be more likely to use multiple spellings of a word than one who writes what he sees. The only difference would be that the former would necessarily be the source of the variant spellings, whereas the latter could be copying them from his exemplar.

Anonymous said...

Are spelling variants catalogued in the new Editio Critica Maior volumes?

Stephen C. Carlson said...

Swanson's editions includes spelling variants.

If I recall correctly, Tim Finney's dissertation on Hebrews also looked at the interaction between spelling variants and text-types.

Tommy Wasserman said...

"Anonymous" asked whether the Editio Critica Maior records spelling variants.

Yes. Such readings are listed with the letter identifying the variant (a, b, c, etc) followed by "o" for "orthographical" variant (i.e. ao, bo, co)...

An interesting thing is that ortographical variation at some point may take on geneological significance when MSS are very closely related.

P J Williams said...

Carl, Thanks for your point. However, hazardous spelling is not a problem unless spelling is hazardous in all variables, which is pretty unlikely.

Eric, my book after next may (DV) be on this subject. However, I'd be very glad if someone pinched the idea and wrote the book themselves because then I would save some time and be able to write another book.

Stephen,
Thanks for mentioning this. Tim Finney's thesis is available here:
http://gamma.ei.virginia.edu/tfinney/thesis/

I've not read it (yet).

I'm a paid up Swanson fan, though he lumps the Byzantine mss together—for obvious reasons.

Daniel Buck said...

Spelling indeed matters.
Is it just my computer, or is there really a mss with the spelling λαμβανω?

maurice a robinson said...

While I do not consider minor orthographic issues to be significant (e.g., final Nu/Sigma or phonetic assimilation such as -LHMF-/-LHF-), other spelling issues do appear to have some significance within text-critical matters.

Such particularly is the case in regard to the spelling of "Jerusalem": even a single author will vary the spelling in his own work (e.g., Lk-Ac in the Byzantine text has 69x IEROUSALHM and 24x IEROSOLUMA [NA/UBS 64x and 26x respectively]).

Also, in the Byzantine Textform the dominant spelling of Moses in the Gospels through Ac 7:44 is MWS-; but from that point onward it becomes MWUS-. The mystery to me is why Luke apparently changed the spelling in the midst of Acts (NA/UBS simply standardize the spelling to MWUS- throughout).

P J Williams said...

Prof. Robinson,

Thanks for yours. You wrote: 'I do not consider minor orthographic issues to be significant (e.g., final Nu...)'

It is, however, extremely unlikely that all the authors of the NT used nu in every instance, as in your edition. What are the chances that they all did that independently?

If my impression, based on NA27, is correct, Luke omits it more often, e.g. εδοξε in Luke 1:3. The mss support for the presence of nu fluctuates. Should that not be reflected in editions of the GNT? Although, clearly, many scribes later tended to omit the nu, early mss overwhelmingly tend to have it (including P52). Do we have much evidence that scribes added nu?

maurice a robinson said...

PJW: It is, however, extremely unlikely that all the authors of the NT used nu in every instance, as in your edition.

In imitation of Falstaff, I will take the better part of valor and assert that (a) I really don't know what the NT authors did at such (minor) orthographic points; (b) the MSS send mixed signals on this particular matter, and cannot necessarily be relied upon, regardless of date or texttype; and (c) it thus becomes a matter of guesswork or presumption to make a definitive declaration in this regard.

PJW: What are the chances that they all did that independently?

I really do not know what the original authors did in such cases, so can make no estimate.

PJW: If my impression, based on NA27, is correct, Luke omits it more often, e.g. εδοξε in Luke 1:3.

I am cautious in this regard, and suspect that minor orthographic matters were of no particular concern to the NA27 editors (they may feel free to correct me if I am in error on this point). It is possible that the orthography that appears in NA27 (or UBS4) may simply have followed that which appears in W-H, perhaps unintentionally remaining inconsistent as regards minor orthographic standardization of text.

PJW: The mss support for the presence of nu fluctuates. Should that not be reflected in editions of the GNT?

Which MS or group of MSS then will serve as the standard, since inconsistency reigns?

PJW: Do we have much evidence that scribes added nu?

Counter-question: do we not have ample evidence that scribes omitted final Nu, particularly in cases where their exemplar may have had a (faint) suspension bar that was overlooked? This is one reason why I do not consider the Nu-movable issue one that impacts NT textual criticism.

P J Williams said...

MR: 'it thus becomes a matter of guesswork or presumption to make a definitive declaration in this regard.'

I'm not at this stage urging people to be 'definitive'. I'm saying that, despite the difficulty involved, this is a matter that needs to be searched out. We cannot just start with the assumption that the data will show no patterns that will give us historical information. To do so also involves a form of presumption.

In Luke 1:3, W-H and NA would obviously have been influenced by the fact that major manuscripts—Aleph, B, D omitted nu while generally having it elsewhere.

MR: 'Which MS or group of MSS then will serve as the standard, since inconsistency reigns?'

I appreciate that you believe the eclectic method to be problematic. I am not therefore suggesting that your edition should follow Alexandrian mss for the omission of nu. However, I am sure that the level of support for nu varies even within Byzantine mss. Ideally this would be reflected in an edition of the Byzantine text, as other spelling matters are. To me it was a matter of great interest that your edition sometimes presented one and sometimes another spelling of the name Moses, and also that some of these locations contained major variants within the Byzantine text and some not. Up to this point our editions of the GNT are artificially consistent in spelling. We need more inconsistency.

MR: 'Counter-question: do we not have ample evidence that scribes omitted final Nu, particularly in cases where their exemplar may have had a (faint) suspension bar that was overlooked? This is one reason why I do not consider the Nu-movable issue one that impacts NT textual criticism.'

I see this as an argument for my position! Yes, we have abundant evidence of scribes using the suspension bar, and the extreme likelihood that such suspension bars were overlooked. But these both produce changes in the same direction: from greater presence of nu to lesser presence of nu. However, we have cases where there is a consensus across early mss against the presence of nu. Here 'transmissional factors' (if I may borrow your phrase) might be thought to suggest that the reading without nu existed long prior to the mss themselves.

See 'competition' post to follow.

maurice a robinson said...

One (long) final comment on this matter:

PJW: We cannot just start with the assumption that the data will show no patterns that will give us historical information. To do so also involves a form of presumption.

If the MS data reflects inconsistency, on what ground can one determine patterns to exist? Should we simply follow the oldest MSS, and overlook possible local-text or individual scribal irregularities? Were the issue to involve proper name orthography, I would consider it significant. I am not so persuaded in the case of movables, nor very much persuaded in the case of routine elision (as witness the continued manuscript fluctuation between ALL’ and ALLA, regardless of the supposed “rule” regarding elision).

PJW: In Luke 1:3, W-H and NA would obviously have been influenced by the fact that major manuscripts—Aleph, B, D omitted nu while generally having it elsewhere.

Tim Finney, I suspect, would say that such might only reflect localized Egyptian usage, and may not reflect that which appeared elsewhere. Swanson (Apparatus A) shows that EDOXEN with the movable Nu is read by A E F K S Theta Lambda Omega f1 2 565 1424. At least for the Byzantine uncials (as well as some non-Byzantines), this does seem to indicate that the omission of that movable Nu in Aleph B D might be merely coincidental, localized, or at least of no major significance within the grand scheme of things.

PJW: I am sure that the level of support for nu varies even within Byzantine mss.

Perhaps, but to what degree is uncertain. Certainly the later “rule” regarding omission or inclusion is followed most strictly in the post-9th century minuscules (as should be expected). From what I have seen, the Byzantine uncials seem generally to retain the movable Nu in most cases (as witness the Lk 1:3 example).

PJW: To me it was a matter of great interest that your edition sometimes presented one and sometimes another spelling of the name Moses .... [O]ur editions of the GNT are artificially consistent in spelling. We need more inconsistency..

In some cases I obviously concur, and agree that perhaps some printed editions go too far by way of orthographic standardization. In other cases, however, I demur, and relegate the “minor” orthographic matters to what I consider their proper (non-significant) place.

PJW: I see this as an argument for my position! Yes, we have abundant evidence of scribes using the suspension bar, and the extreme likelihood that such suspension bars were overlooked.

If so, then in Lk 1:3 it is possible that EDOXE might be a by-product of mere accident, and the issue becomes moot.

PJW: But these both produce changes in the same direction: from greater presence of nu to lesser presence of nu.

I am not convinced of unidirectional alteration. Should Nu have vanished from a word in an early papyrus or uncial due to scribal activity, a later scribe copying from that exemplar equally might restore that which was “more familiar”. This I think would occur more frequently than other alterations that otherwise would affect meaning..

PJW: 'transmissional factors' ... might be thought to suggest that the reading without nu existed long prior to the mss themselves.

Of that I remain unconvinced, and will merely note that this issue generally should not affect the Byzantine uncial testimony (which to a high degree parallels the standardized pattern in our edition). If some pattern can be discerned among the Alexandrian or Western MSS, then by all means pursue it.

The Buck Stops said...

"Is it just my computer, or is there really a mss with the spelling λαμβανω? "

It must have just been my computer, choking on an overdose of Greek orthography.

All is normal now.

daniel said...

A Hebrew scholar here, lobbying for equal time.
The name 'Daniel' is spelled two different ways in the Hebrew scriptures. Its meaning is very much dependent on the spelling.

DaNiEL: All 3 references in Ezekiel
means "God's Judge"
DaNIEL: Elsewhere (including Chaldean)
means "God is my Judge"

So the question is, did Ezekiel and/or his amanuensis use a unique spelling for Daniel, or was Ezekiel referring to some other proverbially wise and righteous (but otherwise unknown) person?
An NIV footnote suggests the latter, a man they call (spurning the massoretic pointing) 'Danel'.

P J Williams said...

Daniel, thank you for the reflections on your name!

It is good to bring Hebrew into the question. There has been useful work on the spelling of the OT done by A. Dean Forbes, F.I. Andersen, D.N. Freedman and James Barr.

There are several points to be made in relation to the names:

1) We do not actually know for certain what they meant. The interpretation of hireq joining parts of names is rather difficult. It can express a possessive relationship (as in Abinadab) or not (as in Peniel).

2) It may be that the meaning of the names in Ezekiel and Daniel is the same. Simply having different spelling does not mean that the names are not the same. Witness the spelling of David as DWD and DWYD.

3) The suggestion that the figure Dan'el appears in Ezekiel is based on identification with the figure in the Ugaritic Aqhat cycle. For critical remarks on this identification see:

Harold H. P. Dressler, 'The Identification of the Ugaritic DNIL with the Daniel of Ezekiel', Vetus Testamentum 29 (1979) 152-61.

In brief, the figure in the Ugaritic text shows none of the personal characteristics implied by the context of Ezekiel 14:14.

4) Some scholars of Ugaritic even suggest that the figure in the Aqhat cycle should be vocalized as Dan'ilu.

daniel said...

"It can express a possessive relationship (as in Abinadab) or not (as in Peniel)."

You lost me there. Doesn't Peniel mean Pen-Face i-of El-God?

Peter M. Head said...

Granted that we can't be sure about the results until the work has been done I think this investigation is definitely worth doing.
Since collations of manuscripts began (let's say with Sepulveda's 365 readings from Vaticanus sent to Erasmus in 1533) there has been a gradual process of including more and more information into the collation. Paying detailed attention to spelling is simply (like paying attention to punctuation and other such things) the next stage before us. Exactly what it will tell us about the textual history of the GNT we don't know yet, mainly because it hasn't been done.
The resources are presumably out there for this research and may soon become available. IGNTP and Munster are using the Collate program which made its first impact by, among other things, including all the information about the spelling of every word in 85 (or so) manuscripts of Chaucer's Canterbury Tales. So the information (presuming that it is being entered carefully) should be there.

P J Williams said...

Peniel—the 'i' joins two words in a construct-genitive relationship (which is possessive, but not what I was meaning by possessive).

Abinadab—the 'i' represents a possessive pronominal suffix (what I was meaning by possessive).

Sorry for expressing myself in a confused manner.

Daniel Buck said...

"Abinadab—the 'i' represents a possessive pronominal suffix (what I was meaning by possessive)."

OK, you're contrasting posessive pronominal infix with genitive construct. Ergo, DaNIEL doesn't necessarily mean "God is my judge?"

On a tangental note:
Why do modern GNT's even include such niceties as accent marks, breathing, iota subscripts, etc? I can see their value in a Byzantine collation, since they are found in increasing numbers as you move up in time through the minuscules. But in a next purporting to be based on the Oldest Uncials, what grounds is there for including what the Uncials and papyri didn't?

P J Williams said...

Daniel, If we're aiming for authenticity then we should not use minuscule script. If minuscule is used then so should accents be. Arguably it is better to stay with minuscules and accents. After all, you can fit more on a page than with an uncial, and it is also much easier to scan read. Capitalization is a great help to scanning, which is why languages without it (e.g. Hebrew) are harder to scan.

Peter M. Head said...

Aren't we aiming for authenticity?

Tim Finney said...

I think that spelling gives clues to MS provenance. My PhD dissertation analyses spelling variation and substantive variation. It is online at http://purl.org/tfinney. By the way, the analysis is based on full transcriptions of all papyrus and majuscule MSS of Hebrews known when I did the work (1992-99).