Sharpening Your Greek: A Primer for Bible Teachers and Pastors on Recent Developments, with Reference to Two New Intermediate Grammars, Part II

Gregory R. Lanier
Assistant Professor of New Testament and Dean of Students
Reformed Theological Seminary, Orlando

II. Other Important Topics

We turn our attention from (some of) the complexities of the Greek verbal system to a handful of other topics about which the pastor or Bible teacher should be informed.

  1. Lexical Semantics and the Question of What a Word “Means”

A perennial challenge in the study of the GNT is determining what a given word means. Consider, for instance, the long-standing debates about significant lexemes such as δικαιόω–δικαιοσύνη, σάρξ, πιστεύω, ἱλασμός–ἱλαστήριον–ἱλάσκομαι, παρθένος, μονογενής, λογίζομαι, προορίζω, and more. Entire theological positions have stood or fallen on the “meaning” of such words. And under the surface has bubbled, in light of the postmodernist critique, a challenge to the very notion of “meaning” itself; what does it mean to “mean” anything at all? What exactly are we doing when we discuss a Greek word’s “meaning” in the first place? Recent decades have seen an explosion of work on Greek lexical semantics and lexicography. It would be nearly impossible to summarize the complexities of these sub-fields in a primer such as this. I will, therefore, simply outline some of challenges posed by lexical semantics and offer a few suggestions in light of them.

Challenges facing anyone who wants to figure out what a Greek wordmeans.”[1] There are two basic occupational hazards facing anyone who studies the GNT at the lexical level, and failure to reflect on them in a serious way leads to a variety of missteps in exegesis. The first is polysemy; the second is an error that is easy to make (unduly influenced by lexicons) in equating English glosses with the meaning of a Greek lexeme. Let us unpack both hazards in turn.

First, despite our best efforts to put Greek words into tidy, self-contained boxes, it is really without question at this point that the vast majority of Greek lexemes[2] are polysemic: they have a range of senses or possible meanings, not just one.[3] The word ἁμαρτία, for instance, does not mean “sin,” but rather it has an entire semantic range (or domain) of which “sin” (that is, the largely Judeo-Christian concept often labeled with that word in English) is only part; the range also includes various kinds of wrongdoings, errors, misdeeds, failures to achieve a goal, mental mistakes, law-breaking, and so forth. Another famous example is the keyword of the Reformation δικαιόω, which has a far broader range of meaning than we want to admit, as attested by the flexibility with which it is used in Matt 11:9; Luke 7:29; 18:14; Acts 13:38–39; Rom 3:4; 3:20; 6:7; Gal 2:16; Jas 2:24–25. The same is true for nearly every English (or French, German, Spanish, etc.) word. “Bridge,” for instance, does not mean “roadway structure spanning between two fixed points” but involves a semantic domain heavily influenced by context.[4] In short, few if any words in the GNT (or in any language) have hard-coded, inherent, single meanings; rather, their range of meaning is a function of their use within a given linguistic community. Where any given instance of a lexeme in Greek (or English) falls within its semantic range is, thus, a function of a variety of features: immediate literary context (or co-text), an author’s style and normal range of use, genre, broader context (an entire corpus), common usage in that time period and culture, and so forth. In short, determining the answer to the seemingly simple question — what does that word mean? — can be complicated.

Second, due to how Greek is typically taught — and how lexicons typically operate — most readers of the GNT hold to a view of lexical meaning that is impoverished: namely, that an English gloss (that is, a word-for-word translational equivalent) is what a Greek lexeme means. We typically learn simple equivalents for our Greek vocabulary words, which is fine for starters. However, due to the pressures of life and ministry, we often fail to move beyond them, and confusion and/or misinterpretation arises when we encounter an instance of a given word for which our memorized gloss does not fit. We can, of course, hover over it in our Bible software or look up the word in a physical lexicon, but this does not necessarily fix the problem. As several lexicographers have complained over the past decade or so, the standard Greek lexicons — while monumental works in their own right and rightly put to good use in our exegesis — also operate largely on a gloss, rather than a definition, basis.[5] That is to say, most lexicons offer a short list of one-word English substitutions for a Greek word. [6] Such glosses are, of course, quite useful, but three issues arise when one equates the meaning of a Greek word with an English gloss.

  • For most Greek words, each possible English glosse is at best a partial In light of the semantic range discussion above, linguists admit that in general few words in a source language map perfectly on-to-one to a single word in a target language, but rather one-to-many. A given English gloss may cover part of the semantic range of the Greek word, but not all; other English words with their own semantic ranges — sometimes overlapping, sometimes not — may be better fits for certain instances of the Greek word. There is, in fact, an entire network of semantic ranges that are interacting between the source and target languages. A simple illustration using the familiar Greek adjective ἀγαθός (102x GNT; 612x LXX) may help. Most of us learn “good” as its gloss, but a close study of its usage reveals that other English words are a better fit in numerous instances (narrowed to four in the diagram). However, each of these English glosses has its own range of meaning that shades into other Greek and English words beyond the one in question, and so on:

  • Further compounding this is the fact that for both κοινή and English, semantic ranges are not static but evolve as the broader language and its various cultural milieux change. One only needs to read the KJV (or Chaucer!) to uncover hundreds of English words that are no longer used today the way they once were used; some are completely unrecognizable. The same is true for Greek. While the κοινή period is somewhat well-defined, there was no point at which everyone sat down and decided what all the words meant for that time period, but rather the language was in a flux from the Homeric and Classical periods through the Hellenistic period and into the Byzantine period. Some words are more stable than others, of course, but one only need to look into the shifts in meaning of, say, δόξα or διαθήκη in secular and Jewish-Christian Greek usage to confirm the broader point. In other words, a gloss-based approach to lexical meaning involves two moving targets on top of the one-to-many problem outlined above. Both complexities can be illustrated using the famous example of σάρξ (150x GNT; 208x LXX):

First-year Greek students likely learn “flesh” as the English for this lexeme, but this only works some of the time. Even when it does, one wonders whether “flesh” (in the KJV sense) still makes sense to modern readers. Some English translations have moved on to something like “human nature” as a gloss, but even this has an expiration date, given the attempt by certain ideologies in the West to reject any notion of a “human nature” and substitute in its place a self-chosen identity that is socially or individually constructed.

  • Finally, limiting our understanding of Greek lexemes to an English word-for-word equivalent yields what one might call the reverse semantic overhang In short, any English gloss, due to its own semantic range, brings to the table connotations that are not at all part of the range of meaning of the underlying Greek word. There is, in other words, a portion of the English gloss’s range of meaning that is left overhanging when compared to the Greek word. Take, for example, the typical choice by English Bibles to use “visit” as the gloss for ἐπισκέπτομαι (Luke 1:68, 78; 7:16; Acts 15:14). In American English, “visit” carries with it a sense of, perhaps, a holiday trip to “visit” the Grand Canyon; in British English, it may carry the sense of afternoon tea with a friend. Neither of these semantic overhangs remotely capture the sense of a definitive intervention by God in salvation/judgment that the Greek conveys in these contexts. Another classic example that regularly comes up in small group Bible studies is the difference between the semantic range of the English word “fear” and the Greek φόβος (or Hebrew יראת), particularly when dealing with the common phrase “fear of the Lord.”[7] While “fear” is an adequate simple gloss for the word in many cases, the baggage of a haunted-house-or-horror-movie kind of “fear” that comes along with the word is quite out of place with what the Bible is typically communicating.

The point of all of this[8] is not to cause us to throw up our hands and give up, descending into the death spiral of lexical indeterminacy. Rather, it is simply to point out the possible theoretical and practical pitfalls that may arise when the busy pastor or Greek student operates with an underdetermined view of word meaning.

Suggestions for those in ministry. What, then, should we do to shore up our grasp of lexical semantics when working with the GNT (short of getting a degree in lexicography)? I would offer the following suggestions:

  • Constantly check oneself to avoid lexical fallacies. Despite the best efforts of scholars to stamp out the abuse of biblical language in ministry, [9] common fallacies still creep into our preaching and teaching, often unnoticed. Many of these pitfalls are no doubt familiar to those in ministry, but it is worth being reminded of them. Three seem to have particular staying power: word-concept fallacy (equating a theological idea with a specific lexeme and assuming the use of the lexeme always engages the concept); etymological and/or root word fallacy (assuming that the meaning of a Greek word is the sum of its parts or its meaning at some other point in time);[10] and illegitimate totality transfer (importing the entire semantic range of a word into a given instance).[11]
  • Work towards increasing one’s feel for semantic range. This is no easy task, as, at the end of the day, it requires reading a lot of Greek. One easy step down the path is to add Louw & Nida’s semantic domain oriented lexicon to one’s toolkit.[12] It is particularly useful for understanding words whose semantic ranges overlap (and/or are synonymous) with a given word being studied. This allows one to think through the possible implications of why the author (say, Paul) used this word instead of that word, which may have worked in the context as well. What does this intentional choice among possible synonyms imply about the chosen word’s meaning in context?
  • Be thoughtful in doing word studies. Examining a lexicon or two and maybe a theological dictionary (with caveats) are great initial steps (and often sufficient, if one is pressed for time and uses the right resources). Going deeper may involve the step of typing the word into Bible software and skimming through at the verse level (and, if pressed for time, an ESV translation rather than the Greek) the various places in the GNT where the word is used. That is a perfectly fine start. But simply reading short snippets of text — especially if limited to English — for a given word may still yield a truncated view of its semantics. To grasp fully the semantic range of the word requires reading more context in Greek of each of its uses, not just the phrase in which it appears. This, of course, takes time, so one has to prioritize how deep to go, and thereby adjust his/her level of certainty in proportion to the level of depth attained.
  • Know when to pick your lexical battles. A fuller appreciation of the complexities of how words convey meaning — and how to bridge the gap from one language to another — should give us reason to be more thoughtful in how we engage in debates. Much theological energy (and often friendly fire) can be expended debating whether, say, κεφαλή means “source” or “authority,” or whether βάπτω/βαπτίζω means “dip,” “pour,” “sprinkle,” or “immerse.” But given the complexities —on the Greek and English sides of the coin — of how the semantic ranges of words interact with linguistic shifts, literary context, authorial distinctives, and so forth, we would be wise to continue reminding ourselves that definitive theological conclusions hinge on more than just an English gloss. In other words, we should be careful not to overexegete a single word if it is not warranted.
  • Avoid saying things like this word in the original literally means X in English.”[13] Whether one says this out of an urge to show off or out of genuine desire to clarify, it is misleading at best and inaccurate at worst. Not only do such definitive pronouncements instill (or stoke) a latent suspicion among one’s congregants that they cannot trust their English translation, they also run roughshod over how words work. It is far better to say such things as, “The word Paul uses here, which the ESV/NIV renders with X, can carry the sense of Y, and it is often used in contexts dealing with Z. This helps us be more precise about what he is getting at here, which is something like…” and so on. Such an approach is not only more pastoral, it is also more accurate — and more linguistically humble.
  1. CBGM and What Is Changing in the Critical GNT Editions

We turn our attention now to a topic that has not received as much attention within evangelical circles as it should: namely, the changes in goal, method, and output behind the newest critical editions of the GNT, NA-28 and UBS-5.[14] Before getting there, it is important first to survey where we have come from. As most Greek students have been taught in preceding generations, prior NA (blue) and UBS (burgundy) critical editions — which are the basis of all modern English translations apart from those in the KJV/Majority-Text tradition — have been edited using standard text-critical principles. Starting from the Westcott-Hort base text that, as is well-known, highly privileged Vaticanus and Sinaiticus (among other witnesses),[15] successive generations of scholars have revised the base eclectic text as new manuscripts (especially papyri) became available over time. The choice of which variant readings — recently estimated to be 480,000–540,000![16] — to place in the main text versus the apparatus has traditionally been made by the editors on the grounds of external evidence and internal evidence.

External evidence focuses on weighing various manuscripts — that is, actual papyrus or parchment artifacts that contain a portion of the GNT (a distinction that will become important below) — based on their quality, not just their quantity. Quality, in turn, has traditionally been pegged primarily to the “text-type” or “text-family” to which any given manuscript/artifact is deemed to belong: the so-called “Alexandrian” witnesses (א, B, A [non-gospel portions], C [some portions], and most papyri) are usually deemed superior; “Byzantine” and related witnesses are viewed with a healthy degree of suspicion; “Western” witnesses (D, old Latin, old Syriac) are early but usually expansive and, thus, deemed lower quality; and “Caesarean” (if a real text-type to begin with) are largely ignored.

Internal evidence comes alongside the external to help further adjudicate among variants found among the manuscripts, focusing on unintentional scribal errors (homoioteleuton, homoioarchton, permutation, haplography, dittography, transposition), intentional scribal changes (harmonization, conflation, correction), and authorial style or writing patterns. By studying such possible explanations, applying the so-called “canons” of textual criticism,[17] and integrating external evidence, a reader of the GNT can make an informed decision on a variant and, thus, “establish the text.” While not every pastor or Bible teacher does this every time they prepare a sermon, lesson, or translation, all (in principle) have been taught this basic approach, and it has served quite well for decades.

Furthermore, this method that students learn in their Greek training basically matches what editors themselves have traditionally done to produce the critical editions themselves. Thus, students can use the same toolset to decide (using Metzger’s Commentary as aide)[18] whether they agree or disagree with the UBS committee. Commentators, likewise, often speak in the same terms regarding “Alexandrian” witnesses and whatnot, and translations committees too have used the same method in determining when to go with NA/UBS or to modify it and make a footnote. In other words, everyone has been singing from the same sheet of music.

However, this is fast becoming no longer the case, but few outside academic circles are even aware.

Summarizing the sea change behind our GNT editions is not a simple task, for even specialists still have a hard time getting their heads around it. I will focus on articulating as simply and clearly as possible what is changing along three dimensions: goal, method, and output.

Goal. Within text-critical circles there has been a running debate about what exactly the goal should be for studying textual variants of the GNT. For the better part of the 20th century, the editors have attempted to sift through the thousands of Greek (and other) manuscripts to approximate, as closely as possible, the “original” text. That is: what Paul or John or whomever actually wrote (≈ the “autographs,” which are, of course, lost). This goal, however, has undergone a marked shift in the past decade. Leading text critics now propose that our goal should be identifying the “initial” text (Ausgangstext): the earliest recoverable approximation of whatever text stands as the headwaters of the subsequent stream of textual transmission.[19] This shift may seem subtle, but it exposes an important revision to our understanding of the earliest stage of transmission of the apostolic writings: there is a gap between the “autographs” and the earliest-reconstructible Ausgangstext that, based on the evidence we have, is ultimately unbridgeable. For instance, if Paul wrote in 50–60 AD and our earliest extant artifactual witnesses are early 200 AD, we can at best reconstruct the Ausgangstext as of a few decades prior to 200 AD, assuming the scribes of the extant witnesses copied older exemplars with some level of accuracy. From that Ausgangstext we can use text-critical deductions to get us closer to the original author, but such a move from Ausgangstext to “original” can only be achieved by educated inference. Put differently, the difference between what, say, Paul actually penned (or his amanuensis) — which, for evangelicals, is the nexus of inspiration/inerrancy — and what we can reconstruct as the “initial” stage of the subsequent scribal tradition may be small (or zero) … or quite large. We simply do not — and, according to some leading text critics, cannot —know.[20] No one, of course, is saying that the Ausgangstext is completely unrecognizable relative to what Paul actually wrote. Rather, the issue is an epistemological one regarding on what grounds we can make upstream deductions, and how much confidence we should have in doing so.

In light of this shift in objective, many (but not all) textual critics have further argued that the real goal of the endeavor is not really to reconstruct the Ausgangstext anyhow — though it serves as a key starting point — but rather to study textual variants primarily for the insight they give us into the reception history of each GNT writing. A textual variant, then, is not a problem to be resolved in order to get back to what Paul actually wrote but, rather (or perhaps chiefly), a window into the world of early Christianity. Variant readings show us what various scribes, and thereby the communities they represent, thought about the text and its meaning.

This latter sub-goal is not bad in itself; textual variants both large and small do provide interesting insights into how early Christians were reading the GNT (e.g., the manuscript turbulence surrounding the various endings of Mark is a fascinating glimpse into this). However, the shift in overarching goal — from “original” to “initial” text — is a substantial change. It is also one whose implications for an evangelical approach to the GNT have not been fully digested. For now, one can continue to treat NA-28/UBS-5 just as one has always treated an edition of the GNT — as an imperfect reconstruction of the autographs, about which we can have a tremendously high degree of confidence — but we will have to play a few more innings to see just how much the underlying philosophical shift in textual criticism will impact things.

Method. Here we enter into yet murkier waters, but they are important to understand. For the past several years, a talented team of NT scholars and textual critics (under the auspices of the Institut für Neutestamentliche Textforschung [INTF])[21] has been engaged in producing the Editio Critica Maior (ECM).[22] The ECM project, beginning with the Catholic Epistles (Acts is due to be released next), aims to produce the first thoroughly critical edition of the GNT based on the most comprehensive and representative set of data available. The ECM leverages technological tools as well as a far more robust set of textual data — namely, full collations of hundreds of manuscripts — unavailable to prior generations of editors of the GNT. By the time the project concludes in the 2030s, each book of the GNT will be covered by the ECM, and the results will, in turn, feed directly into the “hand editions” (NA and UBS) that are used by all. The ECM project in itself is a massively important undertaking that has been a long time coming, and its results will impact NT studies for the next hundred years.

The engine driving the ECM is the point at hand. In sharp contrast to the UBS committee of yesteryear, which, as discussed above, operated on the basis of text-types and so forth that a pastor or second year Greek student could, in principle, repeat for themselves, the editors of the ECM have adopted an entirely new method for analyzing this mountain of textual data. Developed by Gerd Mink, the Coherence-Based Genealogical Method (CBGM)[23] marks a significant advance in methodological rigor but also involves a number of substantial shifts with which we are still coming to terms. Describing the CBGM is not for the faint of heart, but I will do my best to present a (slightly-oversimplified) account.[24]

  • Texts,” notmanuscripts.”[25] The starting point for the CBGM is a subtle but significant shift in the raw data for consideration. Rather than focusing, as with the traditional approach, on tangible artifacts (and their age, provenance, etc.), the CBGM focuses on texts — that is, the words that a given artifact contains. The object of study is family tree or “genealogy” of a given text rather than the genealogy of a given tangible artifact (like P75 or Sinaiticus). A given text can be an “ancestor” of — and, thus, be the input to — another text regardless of which of the two physical manuscripts are older or whether the scribe of one used the other as an exemplar. [26] Put differently, the quality of the text of a manuscript (≈ how accurately it reflects the Ausgangstext) is independent of the piece of papyrus or parchment on which it is documented. A medieval manuscript may contain a text that is actually prior to (an “ancestor” of) the text found on an uncial that, from an artifact perspective, pre-dates it by centuries. This in itself is mostly old news; textual critics have recognized this for a long time. However, the shift is that the CBGM by and large treats the actual manuscripts as somewhat irrelevant and ascribes virtually no weight to them; the text on them is all that matters. Gone, then, are text-types, “Alexandrian” priority, text-critical weight placed on a manuscript’s paleographical dating, and so forth. This feature of the CBGM merits illustration:

In this highly simplified example, Text X is superior to (as the ancestor of) Text Y; however, Manuscript 2 (which contains Text X) is later than Manuscript 1 (which contains Text Y), due to the vicissitudes of the transmission history. Thus, the CBGM would in this case prioritize Text X over Text Y, ignoring the fact that Manuscript 2 (where it is found) may be a medieval minuscule and Manuscript 1 may be Alexandrinus. This is the reverse of the traditional method, which would tend to privilege Manuscript 1 over 2.

  • Comparison of readings to construct a local stemma for a variation unit. With this principle in mind, the critic using CBGM will analyze all the various readings for a “variation unit” (that is, a portion of, say, James 5:10) [27] to determine their family tree, which captures how one reading led to the next one(s). How does he/she determine such “genealogical” relationships and the “flow” from ancestor > descendent1 > descendent 2 > etc.? It can vary for each variation unit and the scholar, but generally a combination of the standard “internal evidence” and text-critical canons (outlined above) are used, along with data elsewhere within the CBGM (which is designed to be iterative and adjust as each decision on a given unit is made). For instance, when faced with ten options for the variation unit at the end of James 5:10, the critic may determine the following genealogy of the reading, called a “local stemma”[28]:

Usually the editor is able to construct such a stemma, but occasionally the genealogical relationship between two readings is left undetermined (until further data is incorporated).

  • Comparison of all variation units between two given textual witnesses to determine their genealogical coherence. The next step is to use the CBGM database to compare full texts (found in manuscripts themselves, like Sinaiticus or 33 — but, recall, the actual manuscripts are somewhat irrelevant) at each variation unit. When comparing two texts — say X and Y — at a given variation unit (let us call it A, such as the final phrase of James 5:10 above), there are four possibilities. The reading for unit U found in X may be prior-to (ancestor) to that of Y (XU à YU); posterior-to (descendent) (XU ß YU); equivalent to (XU = YU); or undetermined (XU –?– YU). For instance, if X(James 5:10) reads επι τω ονοματι κυριου and Y(James 5:10) reads το ονομα κυριου, then X à Y at James 5:10. Using the technology underlying CBGM, the editor is able to compare all the hundreds of variation units for that pair of witnesses to determine the percent of times they are equivalent (=) or derivational (ß/à). The percentage of times they are equivalent is denoted within CGBM as the “pre-genealogical coherence.” The percentage of times they are not equivalent but, rather, the reading of one is an ancestor of the other factors into “genealogical coherence.” The genealogical coherence helps the editor determine whether, based on the math, the text X is on the whole an ancestor to or descendent of the text Y. Here is a real example for James as a whole, comparing two real textual witnesses[29]:
Total units X = Y % pre-geneal. coh. X à Y X ßY Undeter.
746 649 86.9% 64 19 14

Given that the number of X à Y readings is higher than X ß Y, the editor would (absent any other data from CBGM — recall, it is iterative) initially determine that X is an ancestor of Y (regardless of the dates/provenances/etc. of the manuscripts containing X and Y) and, thus, closer to the Ausgangstext.

  • From genealogical coherence of X and Y to sub-stemmata and global stemmata. Running such analyses over a small set of textual witnesses (say, W, X, Y, and Z) enables the editor to produce a tentative “sub-stemma” or family tree of those texts. That is to say, a “local stemma” (illustrated above) reflects a single reading at a variation unit, while a “sub-stemma” takes all the local stemmata into consideration to produce a small family tree of witnesses (each of which is the sum of all the variation units). From there, the CBGM can compute a global stemma for a large number of witnesses. The sub-stemmata and global stemmata, in turn, provide additional information that allows the editor to go back in and adjust their analysis of individual readings and local stemmata. For instance, let us say the CBGM determines text Q is most likely an ancestor of text R for 1 John (e.g., the percentage of variation units that are Q à R is higher than Q ß R). If, however, it is determined from other historical evidence that the scribe producing the manuscript containing Q actually used a manuscript containing R as his exemplar (that is, MQ ß MR), the editor would have to revisit both Q and R to adjust course.[30]

The net effect of these iterative computations is the generation of the set of textual readings deemed closest to the Ausgangstext. As with prior editions of NA/UBS, this “initial text” is not, of course, found on any single existing manuscript in its entirety. It is a reconstruction based on all the analysis of textual flow at the local and global level. One benefit of the CBGM is that we can now compare (for the Catholic Epistles, at present) the percentage of equivalence between the text represented by A and the text found in any witness collated in the database. A snapshot comparing the pre-genealogical coherence (equivalence) of A with the text found on a few notable manuscripts is as follows:

James 1 Peter 2 Peter 1 John 2 John 3 John Jude Total
A — Alexandrinus 89.8% 93.5% 88.9% 91.1% 89.1% 91.55 94.0% 91.2%
A — Vaticanus 96.6% 94.6% 95.8% 96.6% 98.0% 96.8% 95.0% 96.0%
A — Sinaiticus 93.1% 88.1% 86.2% 90.8% 87.3% 94.7% 88.9% 90.0%
A — 1739 92.6% 92.0% 92.7% 95.5% 94.1% 90.5% 90.5% 93.0%

As one can see, the genealogical coherence between A and this sample set is quite high, particularly for Vaticanus.

Output. As noted already, the recent revisions to NA-28 / UBS-5 have incorporated the results of the CBGM/ECM for the Catholic Epistles.[31] This in itself is significant, for the NA/UBS text has not been updated since NA-26 (1979) / UBS-3 (1975),[32] only the apparatus. There are three main groups of updates to NA-28/UBS-5 over NA-27/UBS-4. First, the critical text itself has undergone thirty-three total revisions:  five in James; eight in 1 Peter; ten in 2 Peter; four in 1 John; two in 2 John; one in 3 John; three in Jude. [33] The vast majority are minor (e.g., παιδία to τεκνία at 1 John 3:7). Only two are deemed significant:

Jude 5
NA-27/UBS-4            [ὁ] κύριος ἅπαξ λαὸν ἐκ γῆς Αἰγύπτου σώσας
NA-28/UBS-5             Ἰησοῦς      ἅπαξ λαὸν ἐκ γῆς Αἰγύπτου σώσας

2 Pet 3:10
NA-27/UBS-4             καὶ γῆ καὶ τὰ ἐν αὐτῇ ἔργα        εὑρεθήσεται
NA-28/UBS-5             καὶ γῆ καὶ τὰ ἐν αὐτῇ ἔργα οὐχ εὑρεθήσεται

Some English translations (e.g., ESV, NET[34]) had already placed “Jesus” into the text and “Lord” into the footnote at Jude 5, but KJV and NIV, for instance, read “Lord.” Future editions of NIV, say, may likely embrace the new NA/UBS reading. The variant at 2 Pet 3:10 has long proven vexing, given the peculiar use of εὑρεθήσεται here.[35] What is notable about this change is that the reading now presented as the Ausgangstext in NA-28/UBS-5/ECM is not actually found in any extant Greek witnesses, only in some Sahidic Coptic and Syriac witnesses. While this emendation has been entertained by commentators in the past as a possible solution to this crux, rarely have any translations incorporated it in the text. Now, perhaps, they will.

Second, the new editions have removed square brackets [ ] where, in prior editions (thirty-three times in the Catholic Epistles), the UBS committee indicated it was not completely convinced about a reading but, due to its pedigree in the manuscript tradition, retained it in the body text rather than place it in the apparatus. Many of these previously bracketed texts have been removed from the main text altogether and placed as a variant reading in the apparatus (e.g., ἀμήν at 2 Peter 3:18). Third, the editors have more or less replaced the brackets with a new text critical siglum, the rhomboid ◆, which marks forty-three places where two readings for a given variation unit are deemed equal alternatives; in short, the editors were unable to choose one over the other. One ends up in the main text, with the other in the apparatus, but the reader should treat them as both equally probable.[36]

Implications. Given the newness of all the above changes in goal and method, it is unsurprising that there has been little discussion outside a small circle of text critics about the long-term implications of the CBGM as the process unfolds over the next 15-20 years. Thinking through them all would be a dedicated study in its own right;[37] for now, I will simply pose a few questions/reflections.

  • There are a lot of positives with the CBGM. The data set alone is a substantial improvement over what we had previously. The project has made great strides towards the previously unicorn-like dream of having thousands of manuscripts digitized, collated, and analyzable. The CBGM, for the text critic, also provides a method to hypothesize how specific textual variants emerged (and possibly even when) in the transmission history.[38] Moreover, the results for the Catholic Epistles indicate just how high-quality prior editions of the GNT (going back to Westcott and Hort and their contemporaries) have been. I would argue that our confidence in the text has, in the end, gone up with the ECM’s findings.
  • The ECM project began with the Catholic Epistles in part due to their relatively more stable textual tradition. Additionally, one could argue that the implications of modifying the critical text (which had been unchanged for nearly forty years) in this section of the GNT poses the least risk of ruffling feathers. One wonders, however, just how substantial the revisions may be in the ECM for Acts, the Gospels, and Paul — which, for most in the evangelical world, tend to harbor more emotional/theological investment. We can only wait to find out.
  • Most contemporary English translations (outside the KJV-tradition) have used NA-26 or NA-27 as their base text. Presumably at some point the English translation committees will update their volumes,[39] and when they do so, how will they approach the changes made to NA-28 (or NA-29 and beyond)? Will they embrace them? How will they signal the ◆ readings in the English text and footnotes, if at all?
  • How will (or should) students learn to do textual criticism in the future? This issue is particularly challenging. As outlined above, for decades students have been taught a fairly straightforward method for weighing major manuscripts and internal evidence to determine whether they agree with the NA/UBS critical text. However, the CGBM producing the critical text that future Greek students will purchase is operating according to an entirely different method. This method is, as all readily admit, rather complex to understand,[40] let alone teach. More importantly, one would need to have access to significant analytical tools — and abandon a manuscript-focused mentality (and text-types) in favor of the more abstract text-focused mentality — in order to reproduce the thought process behind a given judgment on a textual variant in the ECM/NA-28/UBS-5.[41] Take the 2 Pet 3:10 example shown above. The old-school approach would look at the various options, weight א, B, papyri, minuscules, and Byzantine witnesses (most of which disagree) and come to some conclusion. However, this conclusion is quite unlikely to be that the lone attested witness for +οὐκ (sa in NA-27; the Syriac is not even mentioned) offers the best reading. Yet that is precisely what NA-28/UBS-5 print in the main text! The student is at a loss, then, for explaining why that reading is preferred when, on the traditional approach, it seems to be the least preferred! Even the full ECM volume — which, incidentally, is prohibitively expensive for most non-specialists to purchase — sheds little light on it, and the (apparently) forthcoming commentary volume remains to be seen. In short, we are facing a situation in which the method currently being taught to students (and taught to scholars/pastors in the past) will no longer correspond to the method underlying the new editions of the critical GNT they are/will be working with![42] It is encouraging that the total number of changes to the text itself, at least for the Catholic Epistles, was fairly small; however, the underlying method is, nevertheless, changing substantially.
  • Related to the prior point, one wonders what use Metzger’s justly famous Textual Commentary will have in the future. It constitutes, in essence, the editorial committee’s notes from how they decided among variations in the 1970s and 1980s; its A-B-C ratings (in the UBS volumes only) have also been a helpful data point for years. However, as Elliott rightly notes, for those portions of the NA/UBS editions that incorporate the outcome of the CBGM/ECM project, “the tried and trusted vade mecum of old, Metzger’s Commentary … is only partially useful.”[43] It may have helpful things to say about the internal evidence that might have impacted the ECM team’s decision for a given local stemma, but any appeal it makes to specific manuscripts is, now, almost entirely outdated.
  • Finally, how will the shift in goal, from “original” text to “initial” text impact the way Reformed/evangelical folks who hold to biblical inspiration approach the critical GNT? Majority-text/KJV-only debates aside, most inerrantists who make use of the NA/UBS volumes have functionally equated the eclectic text found therein with, for all intents and purposes, the inspired autographs. Yes, we know that the critical edition is not itself inerrant or infallible — thus necessitating the need to make one’s own text-critical judgments — but we have embraced it as the next-best-thing we have (much like our approach to the Masoretic Text). The philosophical shift underlying the ECM project, however, is meaningful. The goal is no longer positioned as “getting back to what Mark wrote” but, rather, “getting back as early as possible, given the extant data, to what the early church received as coming from Mark.” Much effort needs to be devoted to thinking through the epistemological and doctrine-of-Scripture implications of such a change with respect to the GNT text coming out of the project.

The preceding discussion has been lengthy due to the numerous complexities involved. However, anyone who aims to study or teach the GNT in years to come must be informed about what is going on with the very GNT we are studying and teaching!

  1. Pronunciation of Greek

There has been a friendly (for the most part) debate within NT circles in the past few years regarding how to pronounce κοινή.[44] Most students in the West have been taught (to the extent that a Greek teacher encourages consistent pronunciation to begin with!) the Erasmian system, whereby each consonant, vowel, and diphthong generally has a distinctive sound. It is no secret, however, that modern Greek (and Byzantine before that) sounds markedly different. Recent phonological research has concluded that Desiderius Erasmus may have gotten Greek phonology basically correct for the Classical period, but his system may very well be wrong for the Hellenistic period in which the GNT was written. In short, the older “continental” or Erasmian system which most NT Greek students learn may be wrong; pronunciation of κοινή, it appears, may be closer to Byzantine/Modern Greek than to Classical.[45]

What are the differences between what a student usually learns in the Erasmian system (taught in, say, Mounce) and what most academics now agree is more accurate for the time period? Several competing systems have been proposed. The following is a rough guide, acknowledging that there is not yet uniformity among scholars on how closely κοινή pronunciation should correlate to Modern. Other systems have been put forward that may differ in the details from this summary (e.g., Caragounis, Allen-Daitz).[46]

  • Rough breathing: not pronounced
  • Consonants
    • Differences:

  • Similarities: everything else, including γ-nasal
  • Vowels and diphthongs
    • Differences

  • Similarities: α, ε , ο, ου[47]

Why does all this matter? Many would argue that it does not, since we are usually reading the text silently rather than reading it out loud or speaking it as a living language. The primary argument in favor of retaining the Erasmian pronunciation — even if it is historically inaccurate — is pedagogical: it is considered easier for first-year Greek students to learn case endings, verb inflections, and vocabulary words if they consistently read (out loud or in their heads) them in such a way that makes clear distinctions between letters and, especially, vowels. The argument in favor of moving away from Erasmian pronunciation is twofold: respect for historical authenticity, and euphony (or ease of pronunciation). The jury is still out on this, and I imagine in most seminary classrooms Erasmian will still be used for the foreseeable, while at academic conferences we will increasingly see a Babel-like situation unfolding as some folks make the plunge while others do not.

I would argue, however, that it is important for the advanced Greek student, at a minimum, to be aware that there is a debate at all, which many introductory grammars paper over. Even if one retains Erasmian for personal or classroom use, some basic familiarity with the more accurate phonology is beneficial for understanding numerous textual variants.[48] Some variations arise via i[o]tacism: the tendency, as shown in the vowel table above, for so many vowels and diphthongs to take on the pronunciation of the iota. Others arise from similarities in omega and omicron, among others. A famous example is ἔχομεν (present active indicative) vs. ἔχωμεν (present active subjunctive) at Rom 5:1, though there are many more. One might also surmise that understanding pronunciation in a more accurate way may help with recognizing euphony, wordplay, rhyme, assonance, and so forth in the Greek text. Finally, knowing the variations also helps the budding scholar understand his or her Byzantine-studies peers — and avoid offending modern Greeks!

  1. Discourse Analysis

“Discourse” has become somewhat of a buzzword in Greek studies, but for good reason. Much work has been done in this area (also known as “text-linguistics”), helping equip the student of the GNT to move beyond simplistic word studies towards a more robust understanding of how larger sections of text “work.”[49] It is a vast and growing field, so I will intentionally limit the discussion simply to a listing of some of the various discourse features that merit one’s attention:

Forward-pointing device

Backward-pointing device

Point/counterpoint set


Historical present

Indirect speech

Change in addressee


Argument flow linkages

Constituent order shifts

Near/far distinction

Boundary markers

Thematic addition




Mainline/supporting line

As even this simple list indicates, discourse-level analysis is fairly different from (and builds upon) the local-level syntactical analysis — prepositional phrases, genitive absolutes, adjectival participles, and so on — that one typically learns in advanced Greek. Another key feature of discourse analysis is the guiding principle (almost a mantra, for some practitioners) that “choice implies meaning.” That is to say, when a Greek writer chooses to use one conjunction instead of another, or to break a standard syntactical or word-order pattern, or to use one aspect instead of another that might be expected — such choices may be significant for meaning.[50]

To learn more about discourse features and how to analyze them within the GNT, I would point the interested party to the two most helpful and practical places to start: Steven Runge’s Discourse Grammar and Stephen Levinsohn’s Discourse Features.[51] Someone in ministry who desires to advance his/her competency in Greek would benefit greatly from reading one of these volumes in addition to either of the recent intermediate grammars discussed in Part One.

  1. Electronic Tools

Little needs to be said about the usefulness of the “Big Three” software packages that own the lion’s share of the market within biblical studies today (Logos, BibleWorks, Accordance).[52] Though expensive, such tools are practically essential for any serious student of the GNT today.

I find it interesting, however, that both Köstenberger/Merkle/Plummer and Mathewson/Emig (from Part One) register reservations about the use or abuse of software packages, and I would add my voice to theirs. As in many areas of life, the pace of technological advancement has, perhaps, outstripped the pace of reflection about possible downsides of such technology among those who study, teach, and preach the Bible. At the risk of sounding like a crank,[53] I would lodge a few observations or cautions.

First, upon completing a difficult course of study in Greek (or Hebrew), students may find — as ministry consumes more and more time — that a software package can become a Siren singing to them. Software offers the tantalizing opportunity to quit reading and studying their GNT for real and simply resort to using the trackpad or touchscreen to hover over words on the screen and have the computer give automated parsing and glosses. Once that Siren song has been heard, it is hard to avoid turning something that was designed to be a supplemental tool into a crutch that upholds an increasingly crippled grasp of Greek. From there it becomes a prosthesis that one cannot live without. Soon enough a student can barely recognize a vocabulary word or parse a participle (let alone remember what those things do) on his/her own, and with a seemingly inexorable march the software has taken the place of one’s brain.

I am a realist, and I know that many, if not most, students of Greek follow this well-trodden path.[54] I would, however, encourage students to fight against it. Double-clicking on a word here or there to access a snippet of a lexicon, or looking at various pie charts of word frequencies, or hovering over something to find out it is genitive plural: none of these tasks in themselves confer real insight into what Paul is saying in Galatians 2, anymore than merely looking up “wherefore,” “art,” and “Romeo” can help someone truly understand Shakespeare. The tools are there to assist, not replace, one’s wrestling with the whole text in all its fullness. Poking around in the software and looking at a smattering of words in isolation is a recipe for a lifetime of lexical fallacies smuggled into sermons and masked with the false confidence of saying, “I consulted the Greek.” One wonders if such an approach, in the end, does more harm than good. I would encourage students, rather, to read a lot of Greek on paper, to handicap themselves, to throw away the crutches, to wrestle with the text – and maybe find out it is actually enjoyable! This can be done with the computer, too, but it requires Odyssean strength to avoid the Siren’s call to let the computer do all the thinking for you. In the end, what is the fun in that anyhow?

Second (and related), I would caution Bible software users about data overload. This is simultaneously a critique of the software developers. One wonders whether it is truly necessary or even helpful for a Bible software package to offer immediate access to dozens if not hundreds of commentaries and books on the same screen as the Bible. It sounds very appealing, but it takes a very disciplined person to use such resources responsibly. A few pitfalls can emerge. While it is convenient to have one-click access to what a host of commentaries say about, say, James 5:2, such convenience comes at the cost of missing what each scholar says elsewhere about James in general, or the whole chapter, or methodology, or presuppositions — none of which will show up in the little snippet that gets loaded by the software for the verse. It seems odd for us to encourage someone to study a Bible verse in context but not a paragraph from a commentary. Moreover, while a pastor or student can just as easily emerge from a library with a stack of fifteen commentaries and read everything they say about a given pericope before ever analyzing the pericope on their own, how much easier is it to do this in, say, Logos? As preachers or teachers, we are not fundamentally called to regurgitate what NICNT, PNTC, NAC, and NIVAC say about a pericope, but to expound what the Bible says to our people. That is, secondary sources are to assist us in analyzing the primary source, not become the primary source. However, I fear that for many, this one-click access to mountains of secondary sources (exacerbated, no doubt, by the tendency among the software packages to cram as much as possible to the left/right/top/bottom of the actual text window containing the Bible) can cause them to get this ordering of priorities out of balance. Finally, in a world in which slow, deliberate reflection on Scriptural truth is crowded out by the incessant torrent of data from all angles, one wonders whether, say, 5,132 resources[55] accessible via a drop-down menu is necessarily a good thing. Has too much technology replaced the reading of whole books (seeking transformation and insight) with a quick-and-easy trawl through a collection of snippets (seeking mere data)?

In short, I find Bible software packages to be wonderfully helpful, and I do not at all discourage their use per se. However, to become a reader of the GNT who is competent and responsible with the text requires a lot of effort reading the GNT (and LXX, and other Greek writings), refining one’s understanding of grammar and lexical semantics, and so on. Software can help this, but it should not replace it. At the end of the day, there are few things more enriching and satisfying than working a Greek passage over and over again until you know it cold — using the scholarly sources to check your work — and then stepping up to proclaim God’s truth. Poking around in Accordance and taking a commentator’s word for it is a meager substitute.

Circling back to where I started, our goal is to love reading and studying the Greek Testament for the rest of our lives. Doing so requires keeping our κοινή tools sharp. One does not decide to become a master carpenter, buy an expensive table saw, allow it to get completely dull and rusty, resort to ordering prefab cabinets from Ikea for the rest one’s life — and still consider oneself a master carpenter. A rusty saw becomes a dangerous saw if one does end up trying to use it, and prefab cabinets are never as fulfilling or enjoyable as the outcome of the process of laboring-sweating-breathing-sawdust-and-glue-fumes. This primer has been offered with that in mind: to encourage those in ministry and students to rekindle their love for the GNT, to sharpen their tools, and maybe, just maybe, realize that it is worth the effort.

[1] Here I am drawing in part on the helpful survey by Constantine Campbell, “Lexical Semantics and Lexicography,” Pages 72–90 in Advances in the Study of Greek: New Insights for Reading the New Testament (Grand Rapids: Zondervan, 2015).

[2] Technical terms are probably the only exception (but most of them have a range of meaning as well as a specific technical sense).

[3] However, there is a peculiar sub-movement within biblical studies in favor of monosemy (e.g., Gregory P. Fewster, Creation Language in Romans 8: A Study in Monosemy [Linguistic Biblical Studies 8; Leiden: Brill, 2013]).

[4] E.g., space of the nose between the eyes; feature of a guitar; control area of a ship; card game; portion of a song typically between verse and chorus; dental apparatus; stretch of land between two bodies of water; roadway structure that spans between two fixed points; and so on.

[5] A “definition” is a full semantic description of a lexeme’s range of meaning that can facilitate cross-language analysis. For instance, a definition of ἁμαρτία might be something like, “state of unrighteousness, moral failure, or error arising from an inward condition or an action; being considered ‘in the wrong’ by an external moral authority.” Such a definition could be rendered in any language to convey the same content. By contrast, glosses for ἁμαρτία in English might be “sin, transgression, error, misdeed, fault, wrongdoing, evildoing.”

[6] This critique applies especially to Liddell-Scott-Jones (though it is still one of the best lexicons for ancient Greek) and BAGD, but the revised edition of the latter (BDAG) makes some improvements. Of the recent LXX lexicons (Johan Lust, Erik Eynikel, and Katrin Hauspie, Greek-English Lexicon of the Septuagint [Rev ed.; Stuttgart: Deutsche Bibelgesellschaft, 2003]; Takamitsu Muraoka, A Greek-English Lexicon of the Septuagint [Leuven: Peeters, 2009]), Muraoka makes the greater strides in providing both glosses and clear definitions. Brill’s massive new lexicon (Franco Montanari, Madeline Goh, and Chad Schroeder [eds.], The Brill Dictionary of Ancient Greek [Leiden: Brill, 2015]) is also gloss-oriented — and translates the Greek glosses into English from Italian, no less!

[7] This happened, in fact, recently in my own church small group when dealing with the repeated notion of “wisdom is the fear of the Lord” in the Proverbs.

[8] We did not even cover intentional/unintentional ambiguity, cross-linguistic interference, and other issues of relevance to the study of the GNT.

[9] Three classics on semantics and Christian ministry should continue to be required reading and, for the most part, have yet to be surpassed: D.A. Carson, Exegetical Fallacies (Grand Rapids: Baker Books, 1996); Moisés Silva, Biblical Words and Their Meaning: An Introduction to Lexical Semantics (Grand Rapids: Zondervan, 1994); James Barr, The Semantics of Biblical Language (Oxford: Oxford University Press, 1961).

[10] I am reminded of a particularly painful example of this, whereby a very learned and respected guest preacher at our church in England waxed poetic for about 15 minutes — using PowerPoint slides, no less — to explain how μακροθυμέω means “long” (μακρο-) of “anger” (θυμός). (It doesn’t).

[11] For example, assuming every given use of ἀγάπη means “divine, covenantal, predestining, adoptive, salvific, marital, eternal love.”

[12] J.P. Louw and Eugene Nida, Greek-English Lexicon of the New Testament Based on Semantic Domains (2 vols.; New York: United Bible Societies, 1999).

[13] Regardless of whether one is using the traditional meaning of “literally” or the contemporary colloquial use (=“actually”).

[14] Nestle-Aland: Novum Testamentum Graece (28th rev. ed.; eds. Barbara and Kurt Aland, Johannes Karavidopoulos, Carlo M. Martini, and Bruce M. Metzger; Stuttgart: Deutsche Bibelgesellschaft, 2013); The Greek New Testament (5th rev. ed.; eds. Barbara and Kurt Aland, Johannes Karavidopoulos, Carlo M. Martini, and Bruce M. Metzger; London: United Bible Societies, 2014).

[15] Brooke F. Westcott and Fenton Hort, The New Testament in the Original Greek (New York: Harper & Brothers, 1881).

[16] Peter J. Gurry, “The Number of Variants in the Greek New Testament: A Proposed Estimate,” New Testament Studies 62/1 (2016): 97–121. His estimate is the first to use a thorough and reproducible method based on extensive manuscript evidence (from the Catholic Epistles). Importantly, he puts this number in perspective by demonstrating that the high number is not really due to scribal inaccuracy but to frequency of copying.

[17] E.g., the reading that best explains the others is usually preferred; lectio brevior (shorter is usually better, as scribes tend to expand versus contract—but not always); lectio difficilior (the more difficult reading is usually better, as scribes tend to smooth things out rather than make them more obscure).

[18] Bruce M. Metzger, A Textual Commentary on the Greek New Testament (2nd ed.; Stuttgart: Deutsche Bibelgesellschaft, 1994 [1971]).

[19] The most thorough essay discussing this shift in understanding is Michael W. Holmes, “From ‘Original Text’ to ‘Initial Text’: The Traditional Goal of New Testament Textual Criticism in Contemporary Discussion,” Pages 637–688 in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis (eds. Bart D. Ehrman and Michael W. Holmes; Leiden: Brill, 2013).

[20] Klaus Wachtel, one of the leaders of the CBGM/ECM project, summarizes: “The initial text is neither the same as the archetype nor the authorial text. The archetype marks the beginning of the manuscript tradition, being itself a manuscript now lost. It is separated from the authorial text by a span of time which may be called the initial phase of transmission. We cannot say what exactly happened to the authorial text in the initial phase. Oral tradition may have had an impact, or revisions, perhaps by the author himself, or other kinds of editorial work on the text. The initial text, as a hypothesis about the text from which the manuscript tradition started, is the result of a methodical attempt to get as close as possible to the authorial text, carried out in the knowledge that this aim cannot ultimately be reached” (“The Coherence Method and History,” TC: A Journal of Biblical Textual Criticism 20 [2015]: 1–6 [2]; emphasis added).

[21] “Institute for New Testament Textual Research.”

[22] The first editions were released piecemeal in the 1990s/2000s, but the definitive edition for the Catholic Epistles was released as Novum Testamentum Graecum: Editio Critica Maior, IV Die Katholischen Briefe (2nd ed.; eds. Barbara Aland, Kurt Aland, Gerd Mink, Holger Strutwolf, and Klaus Wachtel; 2 vols.; Stuttgart: Deutsche Bibelgesellschaft, 2014).

[23] A brief description can be found at

[24] For more detail, consult Peter J. Gurry, “How Your Greek New Testament Is Changing: An Introduction to the Coherence-Based Genealogical Method (CBGM),” Presented at the Tyndale House, Cambridge, April 8, 2015 (also forthcoming in the Journal of the Evangelical Theological Society); Tommy Wasserman, “The Coherence Based Genealogical Method as a Tool for Explaining Changes in the Greek New Testament,” Novum Testamentum 57 (2015): 206–218; Gerd Mink, “The Coherence-Based Genealogical Method (CBGM)—Introductory Presentation,” Provided by the Institut für Neutestamentliche Textforschung, Münster (August 3–6, 2008) at

[25] For a summary of this categorical distinction between texts and manuscripts, see Wachtel, “Coherence Method,” 1.

[26] I am leaving the issue of contamination (which can reverse ancestor-descendent relations) out of this description in the interest of simplicity.

[27] For the Catholic Epistles, the total number of variation units is 3,043 (e.g., James is approx. 750). Variation units can be as short as one word or as long as a phrase. There are approximately 7,600 words in James–Jude, so the average variation unit is 2.5 words. One of the limitations of the CBGM is that it (at least initially) treats all variation units equally, regardless of length or relative importance.

[28] This example is derived from data found at; for more on this variant, see Editio Critica Maior, 91.

[29] Produced using the publicly-accessible CBGM tool (which is limited but still helpful), found at

[30] Another key part of the CBGM is detecting contamination, incidental/random agreements, and other noise that helps to re-shape the overall picture of the textual flow; such details are omitted here for simplicity.

[31] For a fuller discussion of the revisions to NA/UBS, see J. Keith Elliott, “A New Edition of Nestle-Aland, Greek New Testament,” Journal of Theological Studies 64/1 (2013): 47–65.

[32] When the NA and UBS projects began to share an identical text.

[33] Outlined in more detail on pp. 50*-51* in NA-28 and p. 3* in UBS-5.

[34] The NET footnote is worth citing: “The reading Ἰησοῦς (Iēsous, ‘Jesus’) is deemed too hard by several scholars, since it involves the notion of Jesus acting in the early history of the nation Israel. However, not only does this reading enjoy the strongest support from a variety of early witnesses (e.g., A B 33 81 1241 1739 1881 2344 pc vg co Or1739mg), but the plethora of variants demonstrate that scribes were uncomfortable with it, for they seemed to exchange κύριος (kurios, ‘Lord’) or θεός (theos, ‘God’) for Ἰησοῦς (though P72 has the intriguing reading θεὸς Χριστός [theos Christos, ‘God Christ’] for Ἰησοῦς). In addition to the evidence supplied in NA27 for this reading, note also 88 322 323 424c 665 915 2298 eth Cyr Hier Bede. As difficult as the reading Ἰησοῦς is, in light of v. 4 and in light of the progress of revelation (Jude being one of the last books in the NT to be composed), it is wholly appropriate.”

[35] In addition to variations on ευρισκω, other major scribal variants include various forms of κατακαυω, καυω, and αφανιζω (see Editio Critica Maior, 252).

[36] Technically, the ECM has a gap at such points and two branches of the textual tradition feeding it (Editio Critica Maior, 34*). In NA-28, one of the alternative “branches” is listed in the text and the other in the apparatus. Somewhat surprisingly, the UBS text shows the ◆ but the apparatus largely omits any mention of the alternate readings (so they are practically useless sigla). The introduction notes, “at most places…there is no apparatus unit because the textual differences are of no relevance for translation and exegesis” (4*; thereafter directing the reader to an appendix).

[37] Gurry has lodged his own methodologically-oriented reservations in “How,” 2–3.

[38] However, the theory behind CBGM, which separates text from tangible historical artifact (which has a real date and location of origin), raises questions about how one can move from conclusions on the former to conclusions on the latter. As Gurry notes, the abstraction underlying the global stemmata cannot “cannot give direct evidence about scribal habits. A potential ancestor in the CBGM is not necessarily the same as the scribe’s exemplar” (“How,” 2).

[39] The ESV “Permanent Edition” debacle notwithstanding!

[40] Stephen Carlson wryly comments, “I have to admit that the CBGM looks like a mysterious black box whose inner workings seem inscrutable” (“Comments on the Coherence-Based Genealogical Method,” TC: A Journal of Biblical Textual Criticism 20 [2015], 1).

[41] Wachtel’s comment misses the point: “if the critical apparatus is worthy of its name, it will present the evidence in a way that enables exegetes to put the editor’s decisions to the test” (“Coherence Method,” 1). Yes, that is true, but only if the exegete can reproduce the method in order to test its results.

[42] Moreover, it is worth noting that the ECM team was somewhat surprised to discover that many Byzantine minuscules — which, in times past, were usually deemed inferior on the whole (except by KJV-only folks) — actually proved to be quite close to the Ausgangstext (sometimes up to 92+% agreement). Some Byzantine manuscripts, in other words, contain an earlier text in some cases than has traditionally been assumed. Thus, as the editors note, “all passages in which the Byzantine text differed from the primary line [=Ausgangstext] had to be reconsidered” (Editio Critica Maior, 34*). This is quite a departure from the general skepticism afforded Byzantine witnesses under traditional text-critical practice.

[43] Elliott, “New Edition,” 58.

[44] Here I am drawing on Randall Buth, “Notes on the Pronunciation System of Koiné Greek” (2012), available at; Chrys C. Caragounis, “The Error of Erasmus and Un-Greek Pronunciations of Greek,” Filología Neotestamentaria 8 (1995): 151–185; Constantine R. Campbell, “Pronunciation,” Pages 192–208 in Advances in the Study of Greek: New Insights for Reading the New Testament (Grand Rapids: Zondervan, 2015). I would also like to thank my students Michael Schrimsher and Ezra Ahn for their helpful research on the topic.

[45] How does one know? Mostly via the study of inscriptions, papyri, and other artifactual evidence where an unexpected letter appears in the place of what “should” be there (η for ε; or ω for an ο).

[46] The charts provided are my own synthesis of the data provided in Campbell, “Pronunciation” (who cites John Lee) and Buth, “Notes.”

[47] Note, however, that there remains debate about ο/ω; some argue that both are pronounced long rather than short (as shown in the chart).

[48] See the extensive study by Caragounis (Development, ch. 8). For example, he lists 155 (!) textual variants in Papyrus-66 (of John) where ι is written instead of the correct ει (or vice versa); e.g., ιμι instead of ειμι, λεγι instead of λεγει, εχι instead of εχει, and so on (502–505).

[49] For a helpful survey of the state of the question, see Constantine R. Campbell, “Discourse Analysis II: Levinsohn and Runge,” Pages 163–191 in Advances in the Study of Greek: New Insights for Reading the New Testament (Grand Rapids: Zondervan, 2015). He along with others registers the caution that Runge and Levinsohn (mentioned below) often tend to focus on the clause/sentence level of analysis, that is, at a middle point between the word and the paragraph. A robust approach to a discourse would entail understanding the details at the word level, sentence level, paragraph level, and composition level.

[50] Though one should also avoid overexegeting such phenomena. The debate about “default patterns” (e.g., for constituent order in a clause) is ongoing, so what appears “abnormal” today may, in future years, prove to be “normal” on further analysis. So one should, as always, exercise caution before placing too much exegetical (let alone theological) emphasis on the fact that, say, a genitive is fronted.

[51] Steven E. Runge, Discourse Grammar of the Greek New Testament: A Practical Introduction for Teaching and Exegesis (Peabody, MA: Hendrickson, 2010); Stephen H. Levinsohn, Discourse Features of New Testament Greek: A Coursebook on the Information Structure of New Testament Greek (Dallas: SIL International, 2000).

[52] For those on a budget, I would recommend as a robust and free tool that gives the user access to a morphologically-tagged Hebrew OT and Greek NT along with an assortment of other tools.

[53] Let the record reflect, however, that my B.S. is in Computer Science; that I am a recovering web developer and SQL programmer; and that I am as much a power-user of Accordance as the next person.

[54] The fact that many students of Greek (or Hebrew) ultimately forget much of it if they do not stay on top of things does not, in the end, mean they should not undertake it to begin with. One outcome of in-depth language study is a better understanding of English and a sharpened ability to do a close reading of a text — paying attention to structure, logical flow, syntax, and so on. Even if the Greek itself is forgotten, much of these other skills may remain. That being said, mostly-forgotten Greek can be a dull knife that, if used wrongly, can do more harm than good. As I tell my first year Greek students, a little bit of Greek knowledge often means you know just enough to be dangerous! I am reminded of the prominent blogger who made an argument about how we are free to refer to Godhead as “Mother” because Spirit is neuter in Greek and feminine in Hebrew; this blogger presumed this was quite clever but did not realize that this is a fundamental misunderstanding of grammatical gender.

[55] Offered in the Logos “Collector’s Edition” package at the time of writing (at a price of approx. $10,800).