Whenever attempting to articulate my positions about the Bible, I invariably state that the single most augmenting factor in my biblical education has proven to be a knowledge of the biblical languages. In my opinion, exegetical decisions about a translated text in the end can amount to nothing more than exegesis of an already interpreted text, viz. the interpretation of a secondary source. That said, most English translations do a good job communicating the biblical text, and furthermore, the ready availability of [seemingly] countless translations, concordances, and other biblical reference works can give even the non-specialist reasonable access to the meaning of most Greek, Hebrew, or Aramaic words and phrases used in the Bible. So what's so special about learning the languages?
Lifting the Veil
There's something especially deceptive about "a Bible"—I don't mean "deceptive" in a bad way, but when one sees a clean leather cover wrapped around thousands of uniformly thin sheets of paper, typeset with the same font and running page count all written at an 8th Grade level, it's easy to think about the Bible as it were "a" book rather than a collection of books written over the course of several hundred years and touched by as many hands. It masks not only the language barrier between the reader and the authors, but the barriers between the authors themselves.
Even more fundamentally, it masks the material witnesses of the Greek, Hebrew, and Aramaic texts that underlie most English translations—of which no two are exactly alike. I say witnesses very intentionally because the text that underlies nearly every modern translation doesn't represent any one manuscript, but instead represents a "critical" text compiled and edited by scholars through a process called textual criticism.
The Goal of Textual Criticism
This process is the means by which scholars try to reconcile differences between manuscripts and provide the most plausible "original" reading of a text. Most modern scholars agree that the goal of textual criticism—to find the "original" reading of a text—while laudable, is, quite simply, impossible to achieve. As with any historical endeavor, there exists no way to "prove" that a given reading represents the original. Instead scholars must settle for a certain amount of ambiguity and acknowledge that even in spite of overwhelming evidence, there always exists some doubt—however slight—that a reading is wrong. That said, many discrepancies between textual witnesses can be restored with a reasonable amount of certainty.
Textually Transmitted Disease: Accidental and Methodical Corruptions
Most commonly, differences between texts arose through the copying process, what scholars refer to as "textual transmission" and which happened exclusively by hand in the ancient world (note that the word "manuscript" literally means "written by hand"). Most errors of this type come about by a phenomenon known as parablepsis, literally, errors caused by "looking to the side." This shouldn't come as a surprise to anyone who has ever transcribed a text—it is quite easy to lose one's place, skip words, repeat sections, and otherwise corrupt the text being copied (the "Vorlage") as one glances back and forth "to the side." Often, these errors, rather than being corrected, were copied by the next scribe intending to faithfully reproduce his Vorlage. The image is one of a tree—the trunk representing the "original" and the branches representing divergent copies. As each copy itself is copied, it carries with it the corruptions of its ancestors and possibly introduces its own. The goal of the textual critic is to start at the outer-most branch and trace his or her path back to the trunk.
Scribes, however, were not mindless drones dispassionately copying texts. While they had a great deal of respect for the biblical text, they had no delusions that they were copying it perfectly. As scribes noticed differences between texts, they were forced to make critical decisions about what to do. Most commonly, instead of picking one of the readings, scribes would combine (or "conflate") texts. This phenomenon happens frequently with divine name/epithet combinations (like Iēsous christos, and yhwh ʾelohênû). When given the choice between Iēsous and christos, the scribe would often choose to include both. The same thing occurs with larger units even full pericopae.
Similarly, when presented with a difficult word or phrase, scribes would often add a marginal note, known as a gloss to help the reader understand—for example—an infrequently used word. Unfortunately, the margins were the home of numerous other scribal edits—including words, phrases, and pericopae that they had accidentally left out. Rater than chance leaving out part of the biblical text, scribes would sometimes incorporate these editorial comments into the text, which would then be copied by another scribe, etc.
For this reason, one of the most fundamental principles of textual criticism—especially of the Bible—is lectio brevior preferenda est, which means "the shorter reading is to be preferred." Because scribes were more inclined toward expanding texts, generally speaking and all other considerations the same, the shorter of any two readings should be preferred.
However, lectio brevior cannot account for all differences (omissions from the text being an obvious example). Sometimes two texts are just different. In these situations, scholars often use the principle lectio difficilior preferenda est, which means "the more difficult reading is to be preferred." Because scribes had a tendency to make readings easier—which is to say, to clean-up confusing grammar or harmonize with other parallel texts—generally speaking and all other factors the same, the reading that makes less sense should be given priority.
Closing Remarks
As one might imagine, the actual situation is much more complex. To extend our tree metaphor, we might imagine that through the course of some unfortunate event, we are left with only a few branches scattered around the base where our tree once stood. To a certain extent we may be able to deduce the approximate position of each branch relative to the others and see the general shape and direction of each branch, but we would have to be content with not actually being able to see the whole tree. Moreover if we always went with the shorter and more difficult texts, we would end up with an [albeit brief] unintelligible "original" text. Because of this, the overarching methodological principle of textual criticism remains that the reading that best accounts for the other readings should be preferred. It is important to remember that these "principles" are not "laws"—there are always exceptions. While there certainly are "typical" mistakes made through textual transmission, each text-critical decision should be treated in its own right, unique and without precedent.
Having recently completed Elizabeth Clarks's