7
Principles and Practice of Textual Criticism
In previous chapters we have introduced the history of the transmission of the OT text, the character of the major witnesses to the text, the challenge of working with translations, working with critical editions, and common scribal errors. The discipline of textual criticism has a high barrier to entry as it requires a great deal of background knowledge before one is in the position to make accurate text-critical decisions. In this chapter, we will bring together our previous discussions in the book and discuss the actual method of textual criticism: collecting variants, weighing them, and determining the best reading of the text.
The Goal of Textual Criticism
As we noted in chapters 2, 3, and 4, the text of the OT has had a long and complex history. In the three thousand (or so) years since the books of the OT were written, errors and intentional changes have found their way into the text. We no longer have access to the original documents, or autographs. Rather, we have multiple texts in multiple translations with a dizzying variety of readings. Even though many of these variants are not significant, some are, and we desire to know what the authors intended us to read.
When we put this present book together, it took many months of development. We each wrote sections, corresponded with our editor, and communicated with each other. We wrote, then went back and revised. Friends and colleagues offered suggestions, and we revised again. When we were finally satisfied with the result, we sent the manuscript to Baker, and the book was published. At this point, the book was out of our hands (and computers). Now the transmission, or copying, process began. Fortunately for us, technology and the printer’s expertise mean that it is unlikely that serious errors have crept in, and you are reading exactly what we intended to write.
We can also distinguish between development and transmission in the books of the OT. Some authors presumably sat down and wrote a book. Other authors used sources to construct a narrative (such as Kings). Some books, made up of collections of material such as Psalms or Proverbs, developed over time, through several revisions. In the case of Ezekiel, the prophet’s oracles were collected over a lifetime and then edited together into the book that bears his name. However each book was created, there came a time when editing and revising came to an end, it reached its final form, and it was included in the biblical canon. At this point, the transmission process began, and scribes began copying the book in order to make it available to future generations (see fig. 7.1).
The Final Form of the Text
Original Text and Versions
a = MT = G = SP (original = “Jacob”)
b = MT = G ≠ SP (original = “loved”)
c = MT = SP ≠ G (original = “his”)
d = MT ≠ G = SP (original = “son”)
e = MT ≠ G ≠ SP (original unknown)
≠ indicates a lack of agreement between textual witnesses
Our goal, therefore, is the final form of the text. This is the “published” copy that the author intended to be promulgated. Whatever processes occurred prior to this in the development phase are the domain of literary or compositional studies. Textual criticism is concerned with working back through the transmission history of the text and establishing this final, authoritative version.1 Figure 7.2 reflects a hypothetical text and the variable relationship between it and three primary textual witnesses: the MT, the Septuagint, and the Samaritan Pentateuch.2 No single witness reproduces fully and faithfully the wording of the final text that is our goal. It should also be evident that there are various possible situations of agreement and disagreement between the three witnesses.3 By careful study of the variations that exist in the wording of the three witnesses, the text critic can, at least theoretically, postulate the reading of the final, canonical version of the text.
However, we need to allow for some exceptions. There are a few texts in the OT that exhibit significant differences between some of the versions. For example, the Septuagint version of Jeremiah is one-eighth shorter (ca. 2,700 words) than the MT and certain sections occur in a different sequence.4 Similarly, the Septuagint version of Joshua is 4–5 percent shorter than the MT overall, but it also contains elements that the MT version does not, and some elements appear in a different sequence.5 Initially, scholars suggested that these differences arose in the translation process. However, the differences are scattered throughout the book, the variation in sequence suggests adjustments on a larger scale, and the scrolls discovered at Qumran contain readings found in both the Septuagint and the MT. Tov argues that these differences reflect different literary stages or revisions of the book. In his view, a book reached a final form, was viewed by a community as authoritative, and began to be copied and disseminated. At some point the text was then revised, it reached another final form, and that revision began to be copied.6 In these cases, the different textual traditions did not arise from the transmission of the text (the domain of textual criticism), but from literary shaping (prior to transmission).
In these relatively rare situations, our goal in textual criticism may be limited to the earliest form of each tradition: one reflected in the Greek Septuagint and one reflected in the Hebrew MT. The discipline of textual criticism is intended and equipped to make judgments about the transmission of texts, not to sort out textual differences between witnesses that are due to the development or revisions of texts. Therefore, we will have to turn these questions over to literary and canon critics.7 See appendix B for a fuller discussion of these issues.
The Method of Textual Criticism
Textual criticism has often been called both an art and a science. It is science because it deals with data, genetic relationships between manuscripts, a knowledge of scribal practice, and so forth. It is also art because it requires discernment and evaluation. Tov writes, “to a large extent, textual evaluation cannot be bound by any fixed rules. It is an art in the full sense of the word, a faculty that can be developed, guided by intuition based on wide experience. It is the art of defining the problems and finding arguments for and against the originality of readings.”8
The method of textual criticism involves four steps. First, the text critic gathers the available manuscript evidence. Second, he or she compares the evidence and retroverts any translated versions back into Hebrew. Third, he or she evaluates the various readings and attempts to tell the “story” of which reading is most original and which readings are secondary. Finally, the text critic selects the best, most original reading or emends the text.
Collect the Evidence
First, the text critic must collect the available manuscript evidence for a particular reading. The simplest way to do this is to rely on the evidence presented in the apparatus of BHS or another critical edition of the Hebrew text. Remember, as we noted above, that not all critical editions are created equal! While BHS is the most popular and common edition, it is also the most problematic. It does not give evidence for every reading for which there are potential variants, it sometimes omits significant evidence, and it provides evidence for readings that are not significant. Also, remember that in a diplomatic edition (e.g., BHS, BHQ, HUBP) the base text represents one manuscript, and the critical apparatus presents readings that differ from that base text.9 In an eclectic edition (e.g., HBCE), the text represents the editor’s judgment of the best readings, and the critical apparatus contains readings that the editor rejected, including those in the base manuscript. The benefits of an apparatus are that the editors (with their linguistic skills) have identified and collected readings for you.
The advanced critic must instead work directly with the evidence rather than working from a critical apparatus. He or she will find the relevant reading in the critical editions of versions such as the Greek Septuagint, Aramaic targums, Syriac Peshitta, Latin Vulgate, Qumran scrolls, and perhaps others. This requires a knowledge of at least Greek, Aramaic, Syriac, and Latin. This is time consuming and requires additional linguistic skills, but it does not rely on the work of others and their potential subjective judgments.
Compare the Evidence
Once the evidence is collected, the critic compares the readings. He or she should write out in full all Hebrew readings listed and, depending upon one’s knowledge of the languages, all potential versional variants. For example, in Hosea 1:2 the Hebrew text reads תחלת דבר־יהוה, and BHS has the following note a in the apparatus: 𝔊 λόγου cf 𝔖. This means that the Septuagint translates דבר with the genitive “of the word” and that the Syriac Peshitta has something similar. In order to really understand the difference reflected in the versions, the critic should retrovert them—that is, translate the versions back into their assumed Hebrew parent texts in order to make the comparison. This allows us to identify scribal errors such as letter confusion or to note differences in vocalization.
| MT | תְּחִלַּת דִּבֶּר־יְהוָה | ||
| G | ἀρχὴ λόγου κυρίου | = | תְּחִלַּת דְּבַר־יְהוָה |
| S | ܪܝܫ ܦܬܓܡܗ ܕܡܪܐ | = | תְּחִלַּת דְּבַר־יְהוָה |
In this example, the MT (“the beginning [when] YHWH spoke”) vocalizes דבר as a piel perfect verb (“to speak”). By contrast, the Septuagint and Peshitta vocalize דבר as a noun in the standard construct phrase, “the beginning of the word of YHWH.”10
Another option for comparing textual evidence is to create a vertically aligned text in order to contrast differences visually. For example, in Hosea 11:2 the Hebrew reads קָרְאוּ לָהֶם כֵּן הָלְכוּ (they called to them, thus they went). This is awkward because the reader expects כַּאֲשֶׁר (as) in a comparative clause when the main clause is introduced with כֵּן (thus). BHS note a indicates that the Septuagint and Peshitta represent the expected comparative particle: 𝔊 καθὼς μετεκάλεσα, l כְּקָרְאִי cf 𝔖.11 The Septuagint translates “as they called,” and the Syriac is similar. The vertically aligned chart shows the correspondences between the individual words:
Hosea 11:2
| G | S | MT | |
| as | καθὼς | ܐܝܟ ܕ | |
| they called | μετεκάλεσα | ܩܪܘ | קָרְאוּ |
| them | αὐτούς | ܐܢܘܢ | לָהֶם |
| thus | οὕτως | ܗܟܢܐ | כֵּן |
| they went | ἀπῴχοντο | ܐܙܠܘ | הָלְכוּ |
In this example, the chart indicates that the Hebrew has a minus of the comparative particle.12 More advanced text critics might find this approach useful for comparisons of variants in an entire passage, without reference to the apparatus in BHS. We might add more columns for the Qumran scrolls, targum, and so on. It also allows us to color-code the text, draw boxes around certain readings, or use other kinds of formatting in our analysis.
Evaluate Readings
Tov suggests that the main guide for evaluating readings is common sense.13 However, there are some principles that can be helpful for formulating our thoughts.
A preliminary step is to clear away insignificant differences between the texts. For example, in the example above from Hosea 1:2, we conclude that the differences between the MT, Septuagint, and Peshitta are due to a difference in vocalization. This is interesting, but it actually confirms that the MT, Septuagint, and Peshitta all had the same consonantal Hebrew in their parent texts. Therefore, while we may want to consider the exegetical implications of different vocalization traditions, from a text-critical standpoint our job is finished. Other insignificant differences are created in the translation process. Readings created by the translator do not reflect a variant consonantal text and therefore do not need to be considered further. In addition, some apparent variants between versions are due to obligatory language differences. If the translator did not have a choice in how he rendered the text, this is not a true variant.
External Criteria
Presented with an actual variant reading, the text critic first considers external criteria. These have to do with the status of the document in which the reading is found. The critic should arrange the readings into groups. For example, in Hosea 4:15 the MT has ואל תעלו (and do not go up). The Septuagint and Peshitta also represent the conjunction “and” before אל, but scroll 4QXIIc from Qumran does not have the conjunction. Therefore, when we group the evidence we might say:
MT = G = S ≠ 4QXIIc
This means that the MT, Septuagint, and Peshitta all agree on the reading against 4QXIIc.
One of the standard maxims in textual criticism says that we should not count manuscripts, we should weigh them. A reading might be supported by five (or fifty!) manuscripts, but if they are all genetically related, then the same error might be found in all of them.14 On the other hand, a single manuscript, if it was more accurately produced or comes from a better text tradition, might preserve the better reading. In principle, all ancient readings have equal status until they are evaluated.15 However, some versions get more attention than others because of external criteria. The MT, Qumran scrolls, Septuagint, and Samaritan Pentateuch are independent and early. The Peshitta and Aramaic targums, on the other hand, may contain important variants, but because they were translated from a standardized Hebrew text similar to the MT, they are generally given a lower priority. If certain unrelated versions agree on a reading, it might be an indicator of a significant variant.
Generally speaking, older manuscripts are preferable, because the closer a document is to the original text, the less likely that there could have been corruption.16 However, keep in mind that some copyists were more careful or more conservative. For example, the Qumran scrolls are approximately one thousand years older than the MT, but they are idiosyncratic in their orthography. Therefore, the MT is often closer to the original.
In summary, the text critic should be aware that external criteria are often not valid.17 One should be very cautious about privileging older manuscripts, certain manuscript families, or text traditions.
Internal Criteria
Internal criteria are concerned with the value of the reading itself, regardless of where it was found. Here the text critic considers what may have given rise to the various readings in the witnesses, such as scribal errors or intentional changes mentioned above in chapter 6. The text critic considers whether the scribe mistook one letter for another, accidently transposed words, skipped a word, or assimilated a reading to a parallel passage. The critic should consider the relationship between the reading and its context as well as the possibility that what appears to be a corruption is actually a rare word or grammatical construction.
There are several general principles related to internal criteria, but these also must be used with great caution as they are often not valid. One well-known principle is that the “more difficult” reading (lectio difficilior) is to be preferred. This principle is based on the tendency of scribes to simplify and clarify the texts that they were copying much more often than they would have made them more difficult. The principle may sometimes help explain what has happened, but it fails to account for scribal errors that in fact make the text more difficult. As P. Kyle McCarter well states, “The more difficult reading is not to be preferred when it is garbage.”18 Also, two readings may be equally difficult or easy.19
A second principle of internal evidence relates to the length of two or more variant readings. The idea is that the shorter reading (lectio brevior) is preferable and more likely original. This general judgment is based on the normal scribal tendency to amplify a text by adding words to clarify or smooth it out. However, several of the common unintentional errors described in chapter 6 result in shorter texts. In haplography, scribes inadvertently leave out letters; in homoioteleuton or homoioarkton, scribes may leave out larger sections of material because of confusion at the beginning or end of a line. In these cases, the shorter text is caused by error and is obviously not better.
As we mentioned above, the most important principle is that of common sense. The critic must reason, weigh, reconstruct, and do his or her best to discern which of the competing readings is probably more original. Any principles or rules are abstract and do not make our work more objective.
In the example from Hosea 11:2 above, we can make three observations. First, the Hebrew text is awkward and ungrammatical. Second, both the Septuagint (καθώς) and Peshitta (ܐܝܟ ܕ) contain the comparative particle. Because the Peshitta was translated primarily from a Hebrew source text, this probably constitutes an independent witness.20 Third, it is possible that the [k] sound of an original כְּ or כַּאֲשֶׁר was lost due to its similarity with the [k] sound of the next word, קָרְאוּ.
Select the Best Reading
In this fourth and final step, the text critic identifies the reading that is the best and most original. One helpful question is: Which reading explains the origin of the other ones? The text critic’s task at this point is to state that reading “x” is the best reading and to give the reasons for this decision. These reasons will include as full an explanation as possible for the probable development of the various secondary readings that are “competing” with the original reading. Perhaps the critic can tell a kind of story of what happened in the copying process. Did a scribe confuse a ח with a ה? Did the original reading seem at odds with the context, so that the scribe adjusted it for consistency? Or did a scribal error create an inconsistency in the context? The explanation should clearly indicate why a particular reading is preferred and why the secondary readings should not be viewed as original. Ernst Würthwein calls explanations of the origin of alternate readings “control of decision.”21
Most of the time textual difficulties can be resolved with the extant manuscript evidence. But there are times when the process described above does not lead to a convincing solution. At least two kinds of situations can be visualized. First, there is the case of two or more readings that are equally plausible. While some would emend in this case, others would not.22 We are of the conviction that emendation is not required or advisable in such situations. To paraphrase an old Spanish proverb—“a known evil is better than a good thing yet to be known.”23 But a case of not knowing which of two or more attested readings is original is far better (in our opinion) than the “unknown evil” of suggesting an emendation that has no attestation in any manuscript tradition whatsoever. In other words, it is better in these cases that exegesis rest on a plausible and attested reading than on a nonattested reading.
A second kind of situation where emendation is used is the case of a text where no attested reading makes sense.24 In such situations, emendation may be resorted to, but even there it must be used with caution. Bruce Waltke states the procedure for cases like these:
One may attempt a conjecture concerning the true reading—a conjecture that must be validated by demonstrating the process of the textual corruption from the original to the existing text-forms. Such conjectures, however, can never be used to validate the interpretation of the whole passage in that they will have been made on the basis of an expectation derived from the whole.25
McCarter tells the text critic to ask two important questions when suggesting emendations: “Does the proposed emendation explain all the transmitted readings? Is it suited to its context?” If either of these questions must be answered no, McCarter suggests that the textual problem must be declared “unsolved.”26
It is important to remember that selecting a reading other than that of the MT is not an emendation. Whenever a text critic has manuscript evidence for a reading, it is simply the “better reading.”27
Finally, the text critic should state clearly what difference the various readings would make for exegesis. The overall goal for the student of Scripture is to determine what the text says. Thus textual criticism must serve exegesis. In many cases the existence of variant readings is of little consequence for exegesis, even though all variant readings are important for the study of the transmission of the text. For example, cases of varied orthography (spelling) will often have little or no impact on exegesis. If this is the case, the text critic needs to state this. If, on the other hand, one or more of the variants make a major difference to exegesis, that too should be indicated. Be wary of the temptation to choose the reading that best supports your exegesis. This action would make your textual criticism and your exegesis unconvincing.
In the example from Hosea 11:2, the Septuagint and Peshitta probably represent the original reading, which contained the comparative particle as we would expect. The particle כְּ or כַּאֲשֶׁר in the MT tradition was lost because of confusion with the pronunciation of the following word, קָרְאוּ. There are really no implications for exegesis in this situation. Even with its awkward, ungrammatical reading, it is clear what the MT intends to say because of the particle כֵּן (thus). The comparison between the call and Israel abandoning YHWH in this passage is obvious, even in the MT’s truncated reading.
Conclusion
This chapter has outlined a procedure for solving textual problems in the OT. The variants must be collected, compared, and evaluated. The reading with the best evidence is chosen as the best reading. In cases where no textual witness makes sense, the text critic may suggest an emendation. The following chapter will illustrate the use of these principles by examining the textual difficulties that are found in the Hebrew text of the book of Ruth.