Sunday, January 27, 2013

Matthew 14:33 - Did They Come to Jesus?

In Matthew 14:33, after Jesus has entered the boat after walking on the Sea of Galilee, do the disciples come to Jesus, and then worship Him, or do they simply worship Him?  To put it another way:  does the word elthontes belong in the Greek text of Matthew 14:33, or not?   Among major English translations, only the King James Version, the New King James Version, the Good News Version (the paraphrase, not the translation), and the World English Bible reflect a base-text that includes elthontes here.

 Edward Miller’s 1899 A Textual Commentary on the Holy Gospels, covering Matthew 1-14, provides some valuable information.  (Miller’s book, which is more like a detailed textual apparatus than an analytical commentary, can be downloaded at http://archive.org/details/textualcommentar00mill .)  The UBS Greek New Testament is remarkable unhelpful in this case, because although this variant clearly has an impact on translation, neither the second nor the fourth edition of the UBS GNT acknowledges its existence.

The Byzantine Text, including the Textus Receptus and almost all uncials (Miller lists Phi D E F G K L M P S U V X Gamma Delta Pi) and “Much the most minuscules,” that is, almost all of them (but not 1, 22, 118, 300, and 700, and a few others) support elthontes (having come) between ploiw (ship) and prosekunhsan (worshipped).  The Nestle-Aland apparatus also mentions uncial 0106 (from the 600’s) as support for elthontes.  Miller lists several patristic writers who support the inclusion of elthontes, including Didymus (in Egypt), Chrysostom (in Constantinople and Antioch), Jerome, and Augustine.  Clearly, elthontes has very broad, diverse, and ancient support.  There is another reading, proselthontes (having come near), supported by Theta, 13, 124, 346, 503, 556, and 1424, and by the Sinaitic Syriac.  And, although Miller lists the Curetonian Syriac as support for elthontes, Willker has listed it as an ally of its Syriac relative, supporting proselthontes like the Sinaitic Syriac.

The reading ploiw prosekunhsan, without anything in between these two words, is supported by Aleph, B, C, N, 22, 579, the first hand of 892, and a few other MSS, as well as by one Old Latin MS (ff1) and the Bohairic and Sahidic versions.  With the exception of Codex N, the Greek support is mainly Egyptian. 

This is an instructive variant-unit.  Observe what has happened:  an early copyist, with the words of Matthew 14:12 fresh in his memory, added a slight editorial touch, altering elthontes to proselthontes, via the addition of pros.  Thus he produced the reading ploiw proselthontes prosekunhsan.  A subsequent copyist, inheriting this slightly altered text, accidentally skipped from pros- to pros-, omitting the letters in between and thus losing an entire word.

A sustained attempt to correct the addition of pros was probably undertaken at Caesarea:  manuscripts 118 and 209 (representatives of the Caesarean Text) read ontes.  It is as if copyists were given instructions to remove the first part of the word, and this was overdone, resulting not only in the loss of the extra pros but also of the neighboring letters elth.

While elthontes is the Byzantine reading (and is supported by Western representatives as well), and the simple lack of any word is attested by the best representatives of the Alexandrian, a relatively minor group, displaying the Caesarean Text – the text of a relatively small group of MSS that hardly ever gets its readings chosen (mainly because this is a highly harmonized text-form) – is the one that provides the key to developing a hypothesis about the steps from a form of text with elthontes to a form of text without it.  Without the intermediate step (i.e., the addition of pros-) we would see no simply mechanism by which the word could be lost.  This is one reason why, in textual criticism, it is important to seek help wherever help might be found, even if this means that the enterprise, when it begins, may appear to be a matter of a lion being helped by a mouse.     

Against this two-step hypothesis, the only alternative explanation that might be made to defend the Alexandrian reading is that a copyist, for some unfathomable reason, thought that a narrative in which Jesus and His disciples are already close together in a boat needed to state that the disciples came to Jesus before they worshipped Him.  This is rather less likely than the alternative.  (Willker balanced the theory that proselthontes is probably “a harmonization to immediate context” against the observation that the context does not naturally suggest that the disciples in the boat would need to come to Jesus before they could bow down to Him.  The same consideration applies to elthontes.)
 

Thursday, January 17, 2013

Daniel Wallace and Changes in the KJV


What I am writing about today is not directly related to the text of the Gospels.  Today, I want to address some misrepresentations that are being spread about the King James Version.  I am not a KJV-Onlyist, but I do not like seeing the KJV misrepresented and belittled by a writer whose descriptions of the KJV have consistently been the verbal equivalent of a funhouse-mirror.  

Some time ago, Daniel Wallace of Dallas Theological Seminary (where belief in the authority and inerrancy of Scripture is one of seven essential “Core Beliefs,” according to the DTS website at http://www.dts.edu/about/doctrinalstatement ) responded to someone's clarification of his earlier claim that the KJV has undergone more than 100,000 changes.  His response is at http://bible.org/article/changes-kjv-1611an-illustration and it is that painfully specious essay that I am writing about here.

The author of the rebuttal to which Wallace responds observed that if one sets aside changes to spelling and punctuation, changes related to differences of fonts, and changes consisting of shifts in the form of the same word (from “amongst” to “among,” for example), then there are only 421 changes from the 1611 KJV to the typical KJV that you can buy off the shelf today.  The author pointed out that Wallace, in his earlier essay, “clearly intended his reader to believe that the King James Bible of 1611 is significantly different than the King James of today.  As usual, however, the facts don’t bear the critics out.
           
Wallace could have said, "Guilty.  My statement was inaccurate and misleading.  In the text of the 66 books of the Bible, there are actually only several hundred meaningful differences between the 1611 KJV and the KJV as it is published today."  We all make mistakes.  Rare indeed are the individuals who are immune from the axiom that the more one speaks, the more one misspeaks; therefore the most generous speaker tends to be the subject of the most complaints.  

But how did Wallace respond?  With the most poorly argued essay he has ever put online (and that's saying something).  Here is the gist of his tactics: 
(1)  The person who objected to my claim wrote vigorously, like a person who has been offended or outraged, and this makes his claims questionable. 
(2)  What, me, intentionally mislead my readers?  I was just passing along information I received from other sources.  
(3)  The claim that font-changes account for the vast majority of changes to the KJV cannot be true, since a change of fonts would affect every word in the text. 
(4)  When comparing II Samuel 12:20-31 in the 1611 KJV to the same passage in a modern-day edition of the KJV, I found 41 changes, of which, after spelling-changes, punctuation-changes, and capitalization-changes are removed from the list, exactly none remain which result in a change of meaning.  Which is, of course, exactly what I wanted my readers to conclude from my earlier remark that today's KJV-text has 100,000 differences from the KJV-text of 1611!

Then he leaves the ring, stops talking about the rebuttal (which exposed the deceptive nature of his earlier statement),  and says:  “What is not admitted by KJV-only folks is that the changes in most modern translations from the KJV (though on a verbal level are certainly greater than these) do not affect the essentials of the faith.”  That is true as long as "essentials of the faith" are never defined, and everyone is allowed to move a doctrine from the “essentials” category to the “non-essentials” category at will.  Wallace knows very well that some variants that some very influential textual critics have regarded as original (at Mt. 13:35, for example) pose a real problem for the doctrine of inerrancy.  Yet he maintains his claim by gently putting the doctrine of inerrancy in the category of peripheral, rather than essential, doctrine.

Let’s briefly give this claim of his some attention:  do changes in most modern versions, where they mean something other than what the KJV means, affect the essentials of the faith?  I have already written several parts of a series of essays showing that textual variants have a significant impact on doctrine, and the series is not yet complete.  But to read those essays would take much time.  A speedier option is available:  let’s ask one of the translators of the 1881 Revised Version, the first popular English version (besides the earlier attempts by Unitarians and private individuals such as Abner Kneeland) based primarily on the Alexandrian Text, for his thoughts on this question.  George Vance Smith was a Unitarian who participated in the committee that produced the New Testament of the 1881 Revised Version of the New Testament.  And he was not a minor participant; he was a major participant:  C. J. Ellicott, another committee-member, noted that Vance Smith participated in 245 of the committee’s 407 meetings.  (For comparison:  Hort attended 362; Westcott attended 304.)

 Here is what Vance Smith said in his 1881 book Texts and Margins of the Revised New Testament Affecting Theological Doctrine Briefly Reviewed: about the claim that "the great doctrines of popular theology remain unaffected, untouched by the results of the revision." -- To the writer [i.e., to Smith], any such statement appears to be in the most substantial sense contrary to the facts of the case.  And he proceeded to give reasons, citing the absence of most of First John 5:7, the adjusted wording of John 5:44, the rendering of EIS in Matthew 28:18 as “into” rather than “in,” the textual change in I Tim. 3:16, the uncertainly conveyed by marginalia at Acts 20:28, Romans 9:5, Titus 2:13, and I Peter [II Peter is meant] 1:1, the wording of Philippians 2:9, and the disappearance of the term “atonement.”  Then he proceeds:  The changes just enumerated are manifestly of great importance, and are they not wholly unfavourable to the popular [i.e., non-Unitarian] theology?  Many persons will deny this, but it is hard to see on what grounds they do so.  Or, if it be true that the popular orthodoxy remains unaffected by such changes, the inference is unavoidable that popular orthodoxy must be very indifferent as to the nature of the foundation on which it stands.       

Back to Wallace’s essay:  he writes, “My argument about the KJV is not that it has undergone radical changes in its long history . . . but that it has undergone changes — 100,000 of them.”  But look at his earlier statement (at http://bible.org/article/why-so-many-versions ): We must remember that the King James Bible of today is not the King James of 1611.  It has undergone three revisions, incorporating more than 100,000 changes!”  The impression given by this is most definitely not true, because almost all of those 100,000 changes are trivialities that do not affect the meaning of the text.  His own list from II Sam. 12:20-31 illustrates that aside from 421 alterations, the modern-day text of the KJV is the same as the text of the KJV of 1611, allowing for updates in spelling, capitalization, fonts, punctuation, and the removal of the books of the Apocrypha.  Such updates are categorically different from changes to the base-text, or changes to what the translation was intended to mean, and Wallace knows it.

He then states that supporters of the KJV should not object to the changes that are observed among versions, or among different editions of the same version, on the grounds that the KJV has undergone similar changes:  “As the adage goes,” he concludes, “Those who live in glass houses shouldn’t throw stones.”  But this completely misses the real objections about (a) shifts in the base-texts of modern versions (such as the recent shift in the NIV at Mark 1:41, where a reference to Jesus' compassion has been replaced by a statement that Jesus was indignant) and (b) renderings in modern translations and paraphrases which mean two very different things despite being based on the same base-text. 

And there the main part of Wallace’s essay concludes.  But his argumentation continues, in a meandering way, in an Appendix, in which his main subject is not the rebuttal of his his earlier statement, but some aspects of KJV-Onlyism.  So now we reach tactic #5:  Let’s change the subject. 

Not far into the Appendix, I encounter tactic #6:  misrepresent your opponent.  Wallace wrote, “When Scott details a handful of changes that are indeed trivial, he says, “Friends, this is the ENTIRE extent of the nature of the changes from the King James Bible of 1611 to the King James Bible of the present day.” As we mentioned above, that is not correct. Some of the changes in the KJV through the centuries have been fairly significant."  But in the same material to which Wallace refers, it was reported three times that there are 421 non-trivial changes in the KJV. 

So desperate is Wallace to defend his indefensible claim that he resorts to treating the loss of the word “not” in the 1631 “Wicked Bible” as an example of a change in the KJV.  But the objection is not that modern translations have been poorly printed.  The objection, blazingly obviously, is against deliberate changes.  Wallace claims that a change is a change, whether it’s accidental or deliberate, and thus “the principle is still the same” -- even if the change is accidental, and is manifestly a printing-mistake, and the person responsible for it is punished, and it is retracted after being detected.  But that is simply not true!  The changes in modern versions against which the objection has been made are not accidental; they are not printing-mistakes; they are not punished, and they are not retracted, because the people responsible for them consider them improvements, not mistakes.  This is so obvious that further comment should not be necessary.
 
Also in the appendix, Wallace recycles his claim that Matthew 23:24 in the KJV has a typo; the idea is that the text was intended to read “strain out” instead of “strain at.”  This theory has been shown to be dubious because several English writers earlier than, or contemporary with, the KJV use the phrase “strain at,” and it is highly unlikely that all their writings endured the same misprint.  Nevertheless Wallace continued to repeat his claim.  Waitasecond:  did I say his claim?  I should say, instead, that he is repeating a claim that was expressed on pages 150-151 of the 1873 book The Revision of the New Testament (by Lightfoot, Trench, and Ellicott), at about the same time that those three men were busily working on the Revised Version alongside Vance Smith.  Is it not obvious that Wallace is echoing this claim, not because it can be shown to be true, but because it is convenient propaganda against the KJV? 

Wallace, near the end of the Appendix, says:  “To put all this in perspective: There are approximately 25,000 changes made in the KJV of the New Testament from the original version of 1611.  But in the underlying Greek text, the numbers are significantly smaller: there are approximately 5000 changes between the Textus Receptus (the Greek text used by the KJV translators) and the modern critical texts (used as the base for modern translations).  That’s one-fifth the amount of changes that have occurred within the KJV NT itself.

But such a comparison gives a false impression, because almost all of the 25,000 changes in the KJV to which Wallace refers are trivialities involving spelling, capitalization, fonts, and punctuation, whereas a large proportion of the differences between the Textus Receptus and the Revised Texts (and the Re-Re-x27-Revised Texts) yields a translation-affecting change.    

To illustrate:  in Room A there are 24,500 bubbles [representing benign differences between the 1611 KJV and modern editions of the KJV] and 500 dollars [representing sense-changing differences between the 1611 KJV and modern editions of the KJV].  In Room B there are 3,500 bubbles and 1,500 dollars.  Do both rooms have basically the same amount of money?  No.  Likewise Wallace’s analogy is false.   

As he closes, Wallace reaffirms that he does not regard the KJV as the best translation because (1) "its underlying text is farther from the original than is the text used in modern translations;" (2) "its translation is archaic, with now over 300 words that no longer mean what they did in 1611;" (3) "four hundred years of increased knowledge of the biblical world and languages have rendered many of the KJV renderings obsolete."  To which I answer:
(1)  The KJV’s NT base-text deviates sometimes from the original text, but at least it is stable, and the deviation is almost always benign, reaffirming a truth that is taught elsewhere in the New Testament, whereas the base-text of most modern translations is unstable and, because the text-compilers erroneously used a principle of preference for the shorter reading, it has lost some content.  I would rather sail in a ship with barnacles than in a ship with holes.    
(2)  Free tools to help readers learn these 300-400 archaic words are available online.  See, for example, the first 16 pages of the PDF at http://www.tbsbibles.org/pdf_information/1-1.pdf .
(3)  Since Wallace claims that “the Bible must be translated afresh every fifty years or so” in order to remain accessible, this charge that the KJV is full of verbal antiques, if valid, will apply, 50 years from now, to our modern versions too.  But inasmuch as Wallace made this claim about obsolete renderings without giving examples, I do not grant that he has shown that many renderings in the KJV are obsolete to the point of obscuring the meaning of the base-text.  No doubt such renderings exist, but I believe it could be concisely shown that the KJV’s archaic language (such as the distinction between thee and ye, reflecting the base-text’s distinction between singular and plural pronouns) makes the KJV more precise (compared to most modern version) more frequently than it makes the meaning obscure, and I also believe that some words that have been called archaic are merely obscure, and sometimes it is necessary to use an obscure term (that is, a term that will be obscure the very first time it is encountered, before being learned) in order to make a precise translation.  If a Bible-translation contributes to the expansion of its readers’ vocabulary, this is not necessarily a bad thing.    

Yours in Christ,

James Snapp, Jr.

Sunday, January 13, 2013

First Peter 2:2 - Growing Into Salvation

Today, let’s venture outside the Gospels to explore a couple of questions involving First Peter 2:2.  One question is translational and the other is text-critical. 

The translational issue is this:  how should the word LOGIKON be represented in English?  Did Peter mean “of the word,” or did he mean something else, like “spiritual” or “reasonable?”  In Romans 12:1, LOGIKHN is represented as "reasonable" (or an equivalent term) in most versions, and so the idea that LOGIKON means “reasonable” here in First Peter is not altogether groundless. However, Peter has deliberately given his readers cause to focus upon the Word, in 1:23-25, and that focus seems to continue into the second chapter.

Although new translations, such as the HCSB and the CEB, have represented LOGIKON as “spiritual,” even Daniel Wallace has acknowledged that “of the word” is probably a better rendering, in light of the context.  See his article about this at http://bible.org/article/1-peter-22-pure-spiritual-milk-or-pure-milk-word .  He thus opposes the rendering that is found in the RSV, ESV, NIV 2011, and, somewhat surprisingly, the NET.

The text-critical issue is this:  did Peter write EIS SWTHRIAN (“into salvation”) or not?  The UBS Greek New Testament does not indicate the existence of a textual variant at First Peter 2:2.  This should bother its users, inasmuch as the variant here is not only translatable, but is much more significant than some other variants which have been given space in the apparatus. The UBS editors should feel embarrassed that at this point, their apparatus provides less guidance than he footnotes in English translations such as the HCSB and NKJV.

As a translatable variant that affects the meaning of the text, this reading has a status that ought to demand our attention.  It was one of the first non-Byzantine readings to be adopted into the text by textual critics. In 1742 (226 years after Erasmus’ Greek New Testament was published, and 131 years after the KJV was published), Bengel included the words EIS SWTHRIAN in his text.  He defended this reading, stating (according to the 1858 English translation of Gnomon, Volume 5, page 53), “The copies of greatest authority have long read, EIS SWTHRIAN.  In the recent ones, an hiatus has been introduced, the eye of one or two copyists having glided from EIS to EIPER in the next verse.”    

Matthaei, in 1804, also adopted this reading; he mentioned that he had found the words in Codex Mosquensis [Codex 018, from the 800’s], and that it was also in a quotation made by John Chrysostom, and in Book 1 of Clement of Alexandria’s composition The Instructor.  Griesbach and Lachmann also accepted it.  Tischendorf had EIS SWTHRIAN in his seventh edition (1859) of the Greek New Testament, and listed as its support A, B, C, K (= 018, Codex Mosquensis, the one mentioned by Matthaei), over 50 unspecified minuscules, Clement, Cyril (I’m not sure if he meant Cyril of Alexandria or Cyril of Jerusalem), “Dam” (John of Damascus?), Augustine, and Rufinus.

However, a doctrinal objection has been raised against the reading EIS SWTHRIAN:  this reading can be construed as support for the idea of performance-based salvation.  The charge is that someone has attempted to dilute the New Testament’s message about God’s grace by inserting EIS SWTHRIAN and thus conveying that salvation is based on an individual’s performance:  by setting aside malice, guile, etc., and by desiring the pure milk of the Word, a person who is already a believer in Christ may thus become saved.  The translators of the HCSB and NASB may have been pricked by this consideration, and for that reason, although the HCSB has EIS SWTHRIAN in its base-text, it does not render this in a literal sense (“into salvation”) but says instead, “in your salvation,” and the NASB says, “in respect to salvation.”  Such renderings carry a different meaning than some other versions of the very same base text:  Abner Kneeland’s 1823 translation rendered First Peter 2:2b as, “that ye may grow thereby to salvation,” and the Watchtower Society’s New World Translation says “grow to salvation,” and even the NET, a translation produced by evangelicals, says, “grow up to salvation.”

Concern and caution about a textual variant which thus seems to promote a false doctrine is understandable.  However, these words do not promote the false doctrine that our salvation is not an accomplished result of what Jesus Christ has already done for His people.  The problem resides in how these words have been misinterpreted, not in the words themselves.  Paul told the Philippians, “Work out your own salvation with fear and trembling,” and that has been misconstrued by some readers.  But when properly interpreted, it does not mean that Paul taught that one must work to receive salvation.  

What we have here in First Peter 2:2, with EIS SWTHRIAN included, is a statement comparable to what one says of a small child who has been given a large shirt; a bystander might say, “He will grow into it.”  The idea is not that the child, by growing, will own the shirt more.  It is already his shirt, given as a free gift.  Instead, the idea is that as he grows, he will come closer and closer to fitting the shirt.  And likewise as Christians become more and more mature, and more Christ-like, nourished on the pure milk of the Word, they will come closer and closer to fitting into the shirt.  This consideration does not make EIS SWTHRIAN original, but I think it deflects the doctrinally-based objection that they cannot be original.

Now let’s consider the case for EIS SWTHRIAN:  it is very well attested; the NA-27 apparatus lists the Majority Text against inclusion, and then lists P72, Sinaiticus, A, B, C, K, P, Psi, 33, 69, 81, 323, 614, 630, 945, 1241, 1505, 1739, al (that is, and more) latt sy co; and Clement.  In addition, there is a doctrinally benign explanation for its loss:  an early copyist’s line of sight drifted from the EI of EIS to the EI that appears after SWTHRIAN (whether the following word was EI or EIPER), thus skipping everything in between. Frequently when copyists make such parableptic errors, the resultant sentence is nonsensical. But in cases such as this one, when the text remains coherent after the mistake has been made, the mistake is more difficult to detect.  Thus the shortened text was brought into existence in an early copy that had a major influence on the Byzantine Text.  

If we were to consider EIS SWTHRIAN to be non-original, we would have to grant that it must be a scribal insertion which was undertaken at a very early date, to affect so many branches of the transmission-stream.  Several branches of the transmission-stream (represented by the Old Latins, the Vulgate, and the Armenian version, for example) support the inclusion of EIS SWTHRIAN, and it seems very improbable that this reading would spontaneously emerge three times. More probable is a scenario in which these witnesses support this reading because it was in the base-texts from which each of these witnesses was made.  When these diverse and relatively early witnesses are aligned with the Alexandrian Text, and when we also see that 50 minuscules (including members of family-2138, which, in the General Epistles, represents a text-form that existed before the production of Sinaiticus), it is clear that all these witnesses are echoing a reading which is older than any one of its individual witnesses.

So I conclude that EIS SWTHRIAN is original, and that it does not pose a real doctrinal difficulty, and that it was accidentally lost in the Byzantine text-stream via an early parableptic error.  In this passage, a precious little genuine jewel has been recovered from outside the main Byzantine text-stream.  May we all be nourished by the pure milk of the word, and being nourished, may we grow into the gift of salvation which the Lord has given.

Saturday, January 12, 2013

Matthew 11:19 - Children or Works?

The textual question in Matthew 11:19 is, at its outset, very straightforward:  did Matthew write that wisdom is justified by her TEKNWN (children) or by her ERGWN (works)? The path toward the resolution of this question takes some interesting turns.

TEKNWN is supported by over 99% of the Greek manuscripts, but the few that support ERGWN include Vaticanus and Sinaiticus and Codex W.  According to Wieland Willker's Textual Commentary on the Greek Gospels, a few Byzantine manuscripts read ERGWN, too:  202, 1319, and 2145. And minuscules 124 and 288, which are members of family-13, read PANTWN TWN ERGWN. Family-13, collectively, has its own distinct reading:  PANTWN TWN TEKNWN (all her children), which is drawn from the parallel-passage in Luke 7:35.

Metzger accounted for TEKNWN as a harmonization to Luke 7:35.  The same idea is conveyed in the NET, in a one-sided way. That theory, however, faces two obstacles.

First, harmonization in the Synoptic Gospels typically follows a trail from Matthew to one or both of the others.  Copyists simply knew Matthew's text better; it was considered to have a sort of seniority: copyists consistently placed the Gospel of Matthew first in Gospels-collections; Matthew's account was believed to have been written first; Matthew was an apostle (unlike Mark and Luke); the proportion of quotations from Matthew, in the writings of the leaders of the early church, is much greater than Mark's and Luke's.  So when harmonization from Luke to Matthew is proposed, and it is proposed that this harmonization infiltrated multiple transmission-lines, particularly strong evidence should be required, because the natural flow of the harmonization-river went in the opposite direction.

Second, if harmonization occurred from Luke 7:35 to Matthew 11:19, we would expect to see it take the form in which we see it in the harmonized Caesarean Text, represented by family-13: PANTWN TWN TEKNWN ("all her children"), not simply TEKNWN ("children").  (Regarding this, it should be noticed, as Willker has done, that the textual apparatus in NA-27 misplaced the testimony of Old Latin Codex Bobbiensis (k): its reading, AB OMNIBUS FILIIS SUIS, agrees with "all her children," drawn from Luke 7:35, not with "all her works." The placement in the apparatus in UBS-4 is correct.)

However, inasmuch as everyone agrees that Luke wrote TEKNWN, then if Matthew wrote TEKNWN, how did the reading ERGWN originate, if it is not original?  ERGWN is the less harmonious reading.

Although the extant witnesses for ERGWN are very few, two patristic quotations indicate that it was once widely known:  Ambrose of Milan, in his commentary on Luke (chapter 66), mentioned that "many Greek copies" state that wisdom has been justified "by all her works" – "omnibus operibus suis."  He does not treat this variant as if it is dangerously heretical; he somewhat casually mentions it, as if it shines light upon the meaning of the reading that he uses ("all her children"), and moves on. And Jerome, in the course of interpreting Matthew 11:19, after presenting the usual text, stated, "In some Gospels it reads, 'wisdom is justified by her works.'"  Instead of brutally condemning this variant, he states that its sentiment is correct, because wisdom does not seek the testimony of words but of deeds.

Now let's explore some internal evidence.  One of the reasons why ERGWN has been adopted involves a theory about the production of the Gospel of Matthew. Papias mentioned that Matthew composed the "Logia" of the Lord in Hebrew, and this has been widely understood to refer to a Hebrew or Aramaic collection of Jesus' sayings, which was utilized in the production of Matthew's Gospel.  So some researchers, as they have approached this variant, have done so with the premise that a written Aramaic document existed which contained Jesus' saying.  And in Aramaic, there is a word for "children" that, without vowel-pointing, can also be read as "works." (See Metzger's footnote in TCGNT and compare it to Nestle's critical note on the passage in his 1901 Introduction to NTTC.)

This has elicited the theory that (a) the original Greek text of Matthew 11:19 read ERGWN and that (b) this is the author's rendering of a word in an Aramaic source-document, and that (c) Luke's TEKNWN descends from a different understanding of the word in the Aramaic source- document; this was either the understanding of Luke himself, or of someone who translated the Aramaic source-document into a Greek document which was then used by Luke.

Against that theory a couple of objections may be expressed:  (a) While the theory that Matthew used an Aramaic source-document, and that Luke used a Greek translation of its contents, is capable of accounting for this difference (and some others) between their records of Christ's words, it is a relatively complex way to account for a one-word difference.  (A translator's confusion between two similar words might be, however, a plausible explanation for the Peshitta's support for "works" in  Matthew 11:19.)  And, (b) while the theory explains why Matthew could have written "works," it does not explain how or why the passage was harmonized to Luke in a way that affected the text in multiple transmission-streams.  It does not really make a scenario in which Matthew wrote TEKNWN any less plausible.

Some further details about the external evidence are in order: Edward Miller, in his 1899 Textual Commentary on Matthew 1-14, listed B2 (that is, the person who reinforced the lettering in Codex Vaticanus; B's main copyist wrote ERGWN but the lettering-reinforcer did not reinforce ERGWN, and wrote TEKNWN in the side-margin to the left of the column), C, Phi, N, Sigma, D, E, F, G, K, L, M, S, U, V, X, Gamma, Delta, Pi, and "All but all cursives" in support of TEKNWN.  He also listed the following as versional support for TEKNWN: most Old Latin copies, the Vulgate, Curetonian Syriac, Sinaitic Syriac, the margin of the Harklean Syriac, some Armenian MSS, and the Coptic version. Miller also listed the following patristic witnesses also support TEKNWN: "Irenaeus (98). Origen (iii. 211; iv. 48). Didymus (Galland. vi. 310). Basil (i. 98). Gregory Naz. (i. 871). Epiphanius (i. 681). Chrysostom (loc. vii. 179, 419). Theophylact (loc. i. 54). Nicetas (Cat. in Matt. i. 159)," and more.

To that testimony, we must add the support of Sahidic manuscripts, and the Middle Egyptian manuscript known as Codex Schiede (from the 400's), if the NA-27 apparatus is correct.  The other Middle Egyptian manuscript (Codex Schoyen) is not extant for Mt. 11:19.

Details about the testimony of Irenaeus are appropriate, because his testimony is not listed in the apparatus of UBS-4.  Irenaeus does not tell us directly if he is quoting from Matthew or from Luke, but the quotation itself tells us, in Against Heresies, Book One, chapter 8, paragraph 4.  In the course of describing the imaginative teachings of the Valentinians, and how they turn the narrative about Simeon and Anna into a symbol description of celestial beings -- Anna being a representation of the entity called Sophia -- Irenaeus states (referring to the Valentinian teachings), "Her name, too, was indicated by the Savior, when he said, 'Yet wisdom is justified by her children.'"  Irenaeus does not say, as Luke does, "all her children," and thus it is clear that he is quoting from Matthew.

Some details about the testimony of Origen may be helpful: he does not, as far as I know, specifically quote Matthew 11:19.  But in Homily 14 on Jeremiah, he states: TA DE TEKNA THS SOFIAS KAI EN TW EUANGELIW ANAGEGRAPTAI, KAI APOSTELLEI H SOFIA TA TEKNA AUTHS. Which means something like, "Regarding the children of wisdom, it is also written in the gospel, `Wisdom also sends out her children.'"  This is not an exact quotation of Matthew 11:19 or Luke 7:35; it might be based on Luke 7:35 rather than Matthew 11:19, and it might be a blurry misattributed paraphrase of Proverbs 9:3. (I wonder if there any other alleged use of Mt. 11:19 in Origen's writings.)

Now then: remember that tendency of copyists to harmonize the text of Luke so as to make it conform to the text of Matthew?  We see that tendency at work in Luke 7:35.  In Codex Sinaiticus, the copyist wrote PANTWN TWN ERGWN in Luke 7:35, although the genuine text is undoubtedly PANTWN TWN TEKNWN (regardless of whether AUTHS is placed before, or after, these words). A corrector of Codex Sinaiticus overdotted the word PANTWN, thus signifying that it should be considered a scribal error, but it is not a scribal error:  the "corrector" might have thought that he was making a correction, but he was, instead, partly harmonizing the passage to Matthew 11:19 (where the word PANTWN does not appear).

The harmonization of Lk. 7:35 to Mt. 11:19 occurs in other witnesses.  In Codices D, L, M, X, Theta, and Psi, and in family-1, family-13, 2, 22, 28, 700, 1241, 1342, and some versional witnesses, including the Armenian version, the Georgian version, and the Latin text of Codex Bezae, PANTWN is absent from Luke 7:35.  This constitutes harmonization to the text of Matthew 11:19, which implies that in the lines of transmission that produced these documents, there was a copyist who was aware of a form of Matthew 11:19 that read TEKNWN; he was so used to reading TWN TEKNWN in Matthew that when he read PANTWN TWN TEKNWN, in Luke 7:35, he wrote the Matthean form of the sentence.

Thus it should be spectacularly clear that the attestation for the text of Matthew 11:19 with TEKNWN is remarkably old, remarkably abundant, and remarkably diverse, not only in terms of geographic diversity but also in terms of text-type diversity; its supporters come from Byzantine, Western, and Caesarean witnesses (and from the text represented by the Sinaitic Syriac and Curetoniac Syriac).

Now then: if ERGWN is not original in Matthew 11:19, how did it originate?  As a simple substitution, undertaken by a copyist who felt that the text with TEKNWN gave too large a platform to heretics who, he thought, would treat it, not as a non-literal personification, or as a reference to the Holy Spirit, or as a shorthand-reference to wisdom given by God (cf. James 3:17-18), but as a reference to a celestial goddess.  This overprotective tamperer, concerned lest readers imagine that a child-bearing wisdom must be a celestial person, and not a simple personification, replaced children" with
"works."
Someone might ask, "But why, if this person thus altered Matthew 11:19, did he not also alter Luke 7:35?"  I would answer, first, that either the alteration was made when the Gospel of Matthew was still circulated as an individual book, and the tamperer simply did not have a copy of Luke handy, or else he simply was not energetic.  And, second, that in one of the three ancient Greek manuscripts
that support ERGWN in Matthew 11:19, the text of Luke 7:35 /is/ altered so as to read ERGWN instead of TEKNWN. And, third, when Ambrose mentions that "many Greek copies" say that wisdom has been justified "by all her works," he is commenting on Luke, not on Matthew.  So it appears that some copyists did make the adjustment in Luke as well as in Matthew.

The attractiveness of the reading ERGWN, as a means of preventing misinterpretation, is suggested by the mild manner in which Ambrose and Jerome react to this variant; they both clearly use TEKNWN, but they are rather gentle with the reading ERGWN and attempt to salvage it as something which casts additional light upon the meaning of the passage, rather than as something which casts a shadow.

In closing: the original text of Matthew 11:19 reads TWN TEKNWN AUTHS.  ERGWN is an adjustment created by an overprotective copyist.  The same protective tendency to adjust the text so as to prevent readers from using it as the basis to picture wisdom as an actual celestial being appears to be evident in the modern versions which have represented AUTHS as "its" instead of "her."  Of
course the translators would tell you, if asked about this, that they were simply making the text more understandable to readers, so they don't get the wrong idea.  That is probably exactly what the early copyist who changed TEKNWN to ERGWN would say.