Wednesday, March 12, 2014

Book of Mormon and The Late War similarities

With what likelihood can the narrative element similarities between the Book of Mormon and The Late War be explained by similarities in style and genre?

Duane and Chris Johnson have proposed statistical tools to show direct literary dependence of texts on earlier texts based on stylometric features. The Johnsons use this to argue naturalistic, 19th century sources for the Book of Mormon--specifically that the Late War is one of them. It looks like a solid data set, but it is a naive application of stylometry. From my reading on authorship attribution, I don't get the sense that anyone claims stylometric similarities imply derivation. They may reveal common authorship, similarity of style or genre, imitation, or plagiarism, but not derivation. The Johnsons have, in fact, provided a piece of data that is a clear counterexample to their derivation hypothesis. The Book of Nullification, published the same year as the Book of Mormon, can be neither a source for, nor a derivative of the Book of Mormon. Despite this, the Book of Nullification is nearly twice as similar to the Book of Mormon as the Book of Mormon is to the Late War. It's rather a stretch in this context to say that derivation from a common source will result in two things more similar to each other than either is to the source. Evidences beyond stylistic similarity are needed to make claims regarding derivation.

The Johnsons' data may provide insight into many other questions. I agree with the analysis of Christopher Smith (here and here) regarding a more likely significance of linguistic similarities between the Book of Mormon, the Late War, and other early 19th century texts. In essence, the Johnsons' study shows that Book of Mormon English falls cleanly within a context of early 19th century pseudo-biblical language. When this language is imitated in other books with similar genres, unusual numbers of similarities appear between various texts. Such an understanding doesn't require that Joseph Smith ever have read books for which there is no evidence of possession or even connection. Joseph would only have had to participate in the quite common practice of imitating Biblical style in his dictation--whether original or a translation.

A second approach to explain the Book of Mormon in naturalistic terms, as a 19th century product, is to demonstrate specific sources for Book of Mormon content. The idea is that if enough Book of Mormon content can be shown to match 19th century sources and ideas, then the number of apparently ancient elements in the Book of Mormon decreases, or at least becomes more questionable. A smaller number of ancient elements can more plausibly be attributed to coincidence, imagination, stylistic imitation, and blind luck--all origins acceptable to naturalistic sensibilities. This story is most believable if the number of sources can be limited to a small number of sources that Joseph Smith might have had access to with some non-zero probability.

Some people find a more general explanation of Joseph Smith as a sponge of ideas, styles, stories, and ancient myths, followed by an impressively coherent, imaginative dictation, a sufficient explanation of the Book of Mormon. To me, this kind of explanation is as magical as Joseph's own--angels, gold plates, and revelation. No one can hope to test it, scientifically. Consequently, the most rational conclusion from this competition of untestable ideas is agnosticism--not rejection of Joseph's story. For a 'naturalistic' explanation of the Book of Mormon to claim the place of rationality (if a naturalism devoid of angels is not to be your predetermined starting place), it needs to provide more direct evidence of 19th century sources that can adequately replace Joseph's own explanations.

On Patheos, RT has attempted to do just that. RT has argued that too many similar narrative elements appear in both the Book of Mormon and the Late War for Joseph Smith to not have read and absorbed the Late War. He candidly recognizes enough significant differences that he turns to the following explanation of influence:

"A more appealing explanation than strict literary borrowing is that the LW account had made a strong impression on Joseph Smith at an earlier point at some remove from his work on the Book of Mormon and that elements of the story had had time to percolate in his mind. Eventually, these elements were called forth when circumstances necessitated, perhaps even semi-consciously, and then adapted in a unique way."

Again, this relies on pure speculation that Joseph ever encountered the Late War. There are some additional problems with this view. If the influence of books like the Late War on the Book of Mormon is too indirect and soft, then it amounts to little more than claiming that the Book of Mormon language (and possibly some narrative elements) was influenced by the cultural milieu of Joseph Smith--an argument I have no problem with, since I expect it to be the case for a translator of an ancient record. Such a relationship is already established by the stylometric similarities between the Book of Mormon and many other pseudo-biblical, 19th century writings. It is also predicted by Alma's dictum that God speaks to us according to our language and understanding. RT seems to be claiming a stronger reliance on the Late War. RT points out a large number of similar narrative elements in the Late War and the Book of Mormon. In several passages these elements are found in the same order in both books. For example, one passage in the Late War might have 5 independent elements in close succession. These same narrative elements might be found in close succession in the Book of Mormon. RT claims, in essence, that Joseph Smith absorbed and remembered those plot elements, and recalled them in order in the Book of Mormon. Here are possible explanations I can imagine:
  1. A real, ancient record with genre similarities to the Late War produced parallels in sequence by chance and/or similar orders of real events.
  2. Joseph Smith invented pseudo-ancient narrative elements which fell into the same sequence as elements of the Late War by chance.
  3. Joseph Smith copied sequences of elements from the Late War, consciously or unconsciously.
For anyone wishing to claim direct sources for Book of Mormon content, similar alternatives could be proposed. Is there a way to test the likelihood of these various scenarios without resorting to appeals to ideology? I'm not sure, but here is a proposal:
  1. Identify the total number of independent textual elements in each book (e.g. the Late War might have 5000 elements, and the Book of Mormon 10000). Elements which could not logically appear in any other order are not independent.
  2. Identify the numbers of repeat textual elements (two elements that appear 20 times each in both texts are much more likely to appear in close sequence in both books. Elements appearing only once are much less likely to appear in specific orders).
  3. Identify the number of common textual elements between the two books.
  4. Simulate the creation of two books with the respective numbers of textual elements (e.g. rearrange the 5000 elements of the Late War and the 10000 elements of the Book of Mormon several thousand times).
  5. See how many common elements line up in the same orders.
At least two tests could be performed from these data. From 3, the question can be asked if the number of common textual elements is unusual for books of similar genre. As noted by RT, none of the parallels are compelling, individually. Can similar numbers of commonalities be identified between the Book of Mormon and other books about war? If not, that lowers the probability that Joseph Smith invented the war plots independently without even addressing the element of sequence. As regards the Late War, specifically, the remaining choices are an ancient war with real events, some of which paralleled the War of 1812, or direct borrowing (through memory) of plot elements from the Late War. If the numbers of common elements can be explained simply by similarities of genre, then RT's assertion of direct influence needs to be supported by the similarities of sequence.

This test would be difficult to perform, directly. It would require defining narrative elements and labeling them in many different war books. It would require identifying war books which are similar in genre, but not in style to the Book of Mormon. Alternatively, it could be done with books that match other parts of Book of Mormon genre, but are not proposed as sources for the Book of Mormon. Any way you look at it, this would be a painstaking endeavor, however, until it is done, no proper control has been performed on RT's analysis.

A second test could be performed from the data generated by 4 and 5. We can see if RT has observed a real pattern, or if the parallels are likely to arrange in common sequences just by chance. If independent elements arrange in common sequences by chance, in numbers roughly equal to those observed by RT, then claims of direct influence--even distant influence--are weakened, and claims of either a more general cultural influence or an ancient origin are strengthened.

The Johnson study has revealed a handful of other books with high stylistic similarity to the Book of Mormon, including one published in 1830 which could be neither a source for, nor derivative of, the Book of Mormon. I suspect that these other books will also show large numbers of common narrative elements with the Book of Mormon. If this is true, then justifying claims that we can identify Book of Mormon sources through similarity of narrative elements will require comparison of sequences. If plot elements are arranged in common sequences with unlikely frequency, then claims of 19th century sources are strengthened. If not, 19th century source claims are returned to the state of their evidences on historical origins, namely, psychomagical claims regarding the extensive knowledge and genius of Joseph Smith.

Problems with conducting this study
  1. We need a measure of textual elements and of their independence. For example, how many independent elements are in a battle at a river with lots of blood, rallied troops, vanquishing of a more numerous enemy, and casualty counts afterward? It seems like lots of battles have happened at rivers, blood doesn't usually precede a battle, rallying troops happens after a battle has started and lots of soldiers have been wounded, more numerous foes are sometimes vanquished (and always after these other events), and casualty counts are rarely made before a battle is over. From this perspective it would appear to be a single element. However, these same events could have occurred on a plain, or in a forest, but you probably wouldn't rally from inside a fort. Troops might not have been rallied at all after initial setbacks. The foes could have won the battle. Casualties might not have been reported at all. Other observations could have been inserted within these events. Defining narrative elements and their independence seems prohibitively difficult for making any sort of statistical claims of derivation, unless you are trying to show plagiarism.
  2. If there are multiple sources for the Book of Mormon, the Late War might have only been a primary source for the war chapters. Thus, comparing similarities with the whole Book of Mormon will introduce systematic errors. Most likely, these errors would dilute the concentration of similarities and make the probabilities of similar sequences appear much less likely than they really are.
Ideas to manage these difficulties
  1. Choose a stylometric feature, such as word n-grams, to represent similarity. Use these features to approximate a percent similarity for the two texts. Estimate a length for narrative elements. Assume that plot elements will have the same percent similarity between the texts as the stylometric feature, since both style and content are believed to be unusually similar for these two texts. Use this narrative element length to estimate the total numbers of independent narrative elements as well as the numbers and frequency of elements shared between the two texts.
  2. Perform the comparison using different portions of the Book of Mormon. Do it once with the full text. Repeat it with only the war chapters. Repeat in with only the chapters with the highest degree of similarity. The Johnsons have identified four other books with unusually high degrees of stylometric similarity to the Book of Mormon. Take only the chapters which are more similar to the Late War than they are to the other texts. See if the results change in ways that would be expected for similarities resulting by chance as the result of similar genres, or if the likelihood of RT's observed narrative element sequence similarities remains as small as RT asserts.
My suspicion is that one set of numbers will show the probability of sequence similarities to be small, and another will show the probability of sequence similarities to be large enough to be explicable simply through coincidence. We will be left to debate the assumptions behind the analyses without a practical way to decide which assumptions are most accurate. In the end we will once again be left to choose whether we prefer psychomagic or angels/revelation/golden plates--or we will have to remain rationally agnostic as to Book of Mormon origins until we have weighed other evidence from other sources.

Additional Notes:

Observations made through the Johnsons' data and other comparisons with the Late War make a few things clear:
  • Pseudo-biblical style was common enough in the early 19th century to explain the stylistic dissimilarity between Joseph Smith's personal writings and his scriptural writings.
  • Pseudo-biblical style is sufficient to explain the existence of many, short Hebraisms in the Book of Mormon text. I have doubts as to its ability to explain all, when examined in detail. I think searches for extended chiasm in the Late War confirms the improbability of accidental insertion of Alma 36, since no meaningful extended chiasms have been found in the Late War, despite someone's having taken the time to identify a huge, meaningless chiasm in the Late War.
  • The uniqueness and complexity of story in the Book of Mormon, which was noted by RT, still requires explanation. The proposed candidates still remain Joseph's imagination or revelation/ancient translation.
  • Other pseudo-biblical texts (many of which are identified by the Johnsons' extended data set) could be tested for variety in their stylometric features. Do any of them come close to the variety of styles in the Book of Mormon? Do authors just accidentally shift styles multiple times when imitating Biblical style?

No comments:

Post a Comment