Back to Mormonism Researched Page

The Book of Mormon: Proven Ancient?

Research by Russell C. McGregor.

A review of "THE BOOK OF MORMON: ANCIENT OR MODERN?" from the Salt Lake City Messenger, November 1993, pp. 5-11.

As time goes by and anti-Mormon criticisms are disposed of, one would expect that the critics would improve their methodology. If the article under review is anything to go by, the Tanners do not seem to have shown any such improvement in technique. Indeed, it is as poor as anything that I have seen them produce.

After the obligatory opening paragraph very briefly summing up the coming forth of the Book of Mormon -- an opening that only serves to remind us that the article is intended for the uninformed -- the Tanners present a long quote from Orson Pratt. Then they launch into the following astonishing statement:

"Our study of the Book of Mormon has extended over a period of thirty years and has led us to conclude that it is not an ancient or divinely inspired record, but rather a product of the nineteenth century."

The sense of this sentence is that they (the Tanners) rejected the Book of Mormon only after their thirty years of study. This would be true if their position on the subject had been neutral during that time, but it plainly has not. In reality, they have been actively publishing anti-Mormon material since at least the mid-70's, to my knowledge. It is therefore safe to conclude that they rejected the Book of Mormon fairly early in that thirty-year period. Actually the statement in question is a fairly standard anti-Mormon dodge: they are telling us that their conclusions are the product of their research, when in reality their conclusions were reached a priori and have entirely guided their research.

The article goes on to cite a number of "striking parallels" between the Book of Mormon and a number of 19th-century books. These "striking parallels" are, for the most part, nothing more than the appearance of a few "word combinations" in the Book of Mormon that appear in other books. For example, the phrase "forts of security" in Alma 49:13 and 18 also appears, according to the Tanners, in A History of the American Indians by James Adair. Not having read Adair, I cannot comment; but it can hardly matter. Phrases have a way of being widely used in a given place and time. I know of no writer who routinely goes to the trouble of coining entirely original phrases in every sentence they put down. The Tanners themselves provide a case in point: in their sentence quoted above, there appears the phrase "divinely inspired record." Now it happens that the same phrase appears in the passage they quoted from Pratt. Does this prove plagiarism? Well, they didn't use quotation marks, so they weren't quoting; they simply used the phrase that fitted the intended meaning. I spotted this parallel in less than a minute.

Other phrases used by the Tanners that would readily find parallels elsewhere are: "hard to believe," "by chance," "In addition to," "there are other," "For example," "they had been," and many others. There is no doubt that a serious attempt at matching phrases from the Tanner's writings and other contemporary publications would turn up hundreds of them. Such parallels would mean nothing at all, and simply illustrate that phrases, like words, are part of the vocabulary of a language.

The second type of parallel shown by the Tanners, also using Adair's book, is a more promising one wherein a long descriptive passage is matched to a similar passage in the Book of Mormon. However, the problem arises that the Book of Mormon passage is not a single passage at all; it is a patchwork of nine different passages spliced together. But even so, the resemblance between the two quotations is very superficial. In order to make something stick, the Tanners are reduced to pointing out such details as the word "ditch" appearing in both (and even "the ditch," as if an article before a noun is somehow more significant than the noun on its own.)

And so the phrases, "which had been dug," "the inner" and "of earth" are also supposed to show something important that supports the Tanner's position. We are constrained to ask -- what, exactly, do they show?

Summing up the first part of the article, the Tanners give us their conclusion:

"We are convinced that Joseph Smith read a number of books and articles about the Indians especially books equating them with the ancient Israelites. His own mother, Lucy Smith, tells that Joseph Smith had a fervent interest in the ancient Indians before he received the plates from which he 'translated' the Book of Mormon."

The Tanners then go on to quote the often-abused statement by Lucy Mack Smith describing how Joseph used to tell the family about the ancient inhabitants of the Americas, prior to receiving the plates. Of course this does not give us a problem -- this was after Moroni's visit, and Joseph had already received that knowledge by revelation -- but why should it be a problem if Joseph did find and read some books on the subject? When we consider that an angel had appeared to him, shown him a vision of the Nephites, and promised him that he would later on receive a book written by them, would it not be entirely natural for any person to be intensely interested in anything that might shed some light on the subject?

Wrong Wordprint Methodology

The article goes on to discuss similarly unimportant parallels between the Book of Mormon and the preface to the King James Bible. Then it goes into the issue of wordprints. As this is the first critical article which I have seen that attempts to deal with that issue, it is interesting to see how they deal with it. Firstly, they argue that "the list of '24 Major Book of Mormon Authors Used in the [BYU] Study,' seems to be somewhat padded" because it includes Isaiah, the Lord, Jesus and the Father. However, even if we exclude the authors in question, we are still left with 20 distinct authors. Furthermore, since the study in question (Larson, Wayne A., Rencher, Alvin C. and Layton, Tim, "Who Wrote the Book of Mormon? An Analysis of Wordprints," _Brigham Young University Studies_, Vol. 20, No. 3, Spring 1980, pp. 225-251) intentionally set out to include all Book of Mormon author attributions of significant length, excluding passages attributed to those authors would be artificial at best.

The Tanners then quote John A. Tvedtnes of BYU, who does not endorse the wordprint evidence, and says, "I have my own reasons for rejecting those studies, however, and hope to express them elsewhere." Since Tvednes undoubtedly does _not_ endorse the research which the Tanners put forward either, this is an obvious case of loading the dice -- a popular anti-Mormon trick.

The article then produces, with a flourish, some research which supports the Tanners' position. It is introduced as follows:

"Recently another computer study of the Book of Mormon has come to our attention. It is entitled, 'A Multivariate Technique for Authorship Attribution and its Application to the Analysis of Mormon Scripture and Related Texts.' The research was done by David I. Holmes, a Senior Lecturer in Statistics at Bristol Polytechnic, and was published by Oxford University Press for the Association for History and Computing."

Let us pause at this juncture to point out that while the BYU team was mentioned only by their surnames in the Tanners' article, Holmes gets all of his credentials presented. Again, this is loading the dice, since Larsen and Rencher held senior positions in the BYU Statistics department at the time of their study. Also, it is worth pointing out that a Polytechnic in the British educational system is essentially a trade school, and does not aspire to the high scholarly pretensions of a university. This is not of itself a critical issue (and I do not mention it to undermine Mr. Holmes's credibility) -- the best thing about statistical methods is that the validity of their results is unrelated to the prestige of their practitioners -- but it is worth mentioning to show how heavily the Tanners have loaded the dice.

Holmes describes his methodology in this manner: "The most impressive statistical analysis carried out on the Book of Mormon is that undertaken by Larsen, Rencher and Layton.... The authors conclude that their results all strongly support multiple authorship of the Book of Mormon yet their whole case rests on the assumption that the frequency of occurrence of non-contextual function words is a stylistic discriminator. The article claims that there is no resemblance between the authors of the Book of Mormon and the nineteenth century authors sampled, but the case rests on usage of words such as 'unto, beyond, yea, forth, verily, lest and nay' which would all naturally be prominent in an archaic biblical-type style, but could hardly be expected to occur with the same frequency elsewhere, even in the early nineteenth century. Against this background, the aim of my research is to complement historical and scientific studies into the authenticity of the Book of Mormon by subjecting it and related Mormon scripture to stylometric analysis. In this paper it is understood that a particularly effective measure for purposes of discrimination between writers is the vocabulary richness of a text...."

There are several problems with Holmes's methodology as he describes it. First of all, he rejects the use of non-contextual filler words because "the case rests on usage of words such as 'unto, beyond, yea, forth, verily, lest and nay'." However, he has misstated the case. The keywords used by Larsen and Rencher were: and, the, of, that, to, unto, in, it, for, be, which, a, this, now, with, upon, but, from, therefore, even. The only one which Holmes got right was "unto", and that is a fairly common word even today. He then selects a methodology with these words: "In this paper it is understood that a particularly effective measure for purposes of discrimination between writers is the vocabulary richness of a text."

Notice that Holmes does not explain why he chooses "Vocabulary richness" as his preferred discriminant; it is "understood," i.e. assumed without argument, to be the best one for his purposes. Now it happens that it may indeed have some uses for some applications; but it has serious flaws in this one, and for four reasons:

1) The same writer may use different vocabularies at different times. An essay written by a sixteen-year-old will hardly use the same vocabulary as the same writer's Master's Thesis, even if the thesis was written on the same subject. Thus, Holmes's method would detect two different writers. On the other hand, since the use of non-contextual filler words is a set of word use habits that are acquired very early on, a wordprint using the Larsen-Rencher method would, providing the writer developed normally throughout his or her life, detect the similarities and ignore the differences.

In the late 1970's, Andrew Q. Morton took word prints of the novels of Sir Walter Scott. These showed that Scott's word usage habits remained stable despite a series of strokes. Morton holds that this means that these habits must be related to a very primitive brain function.

2) The same writer may use different vocabularies for different purposes and audiences. An article such as this requires a different vocabulary than a letter to a teenage son or daughter, which requires a different vocabulary than a Sacrament Talk, which requires a different vocabulary than a Project Plan. The same writer may need to write all four types of documents in a two-week span; although different vocabularies would be used, the writer's wordprint would be the same throughout. Thus, the Holmes method would detect four different writers; the Larsen-Rencher method would detect one writer.

3) A group of writers with similar educational attainments and a common cultural environment will tend to use a common vocabulary. This is noticeable in any "college" type of environment, where the growing vocabulary of terms introduced by a lecturer during a semester is faithfully and fairly uniformly reflected in the assignments turned in during the course of that semester.

4) The most important flaw is that testing a translated document on "vocabulary richness" is going to detect only the translator, and not the original author(s). The act of translation can be described as the complete substitution of one vocabulary (the author's) by another (the translator's). I have seen no reconstruction of the translation of the Book of Mormon -- not even the most contrived hostile ones -- that suggests that Joseph Smith's own vocabulary was in some way set aside during that process; therefore, if Joseph Smith was telling the truth about the coming forth of the Book of Mormon, then a test of that book on "vocabulary richness" is guaranteed to detect Joseph Smith and nobody else.

This problem is so significant as to completely invalidate Holmes's conclusions. It does not entirely invalidate his research, however; he does support Larsen, Rencher and Layton in disqualifying Solomon Spaulding, Oliver Cowdery and other proposed 19th-century ghostwriters; but his conclusion that there were no ancient authors cannot be supported -- because the question simply cannot be tested -- by the method he has chosen.

An illustration of the problem is in order; suppose someone with a poor scientific vocabulary were to translate a document from German. The translator comes across the compound word "sauerstoff." He renders it into common English as "sour stuff." Later on, reviewing the translation, he decides that the tone of the original warrants the more formal "acid substance." In fact, the best translation would be "oxygen". The author's scientific vocabulary has been lost in the translation, because of the limitations of the translator's vocabulary. If we were to test the translation using the Holmes method, we would find that its scientific vocabulary was far less "rich" than that of the purported original author; in fact, it would match that of the known translator; thus the translator would be found to be the author.

Now we know that Joseph Smith had only a third-grade education; his vocabulary in 1828 was rather limited. This means that it does not matter how rich Nephi's or Alma's vocabularies were; Joseph's vocabulary is the sieve through which they must be strained, and so Joseph's vocabulary is all that we can ever see. And therefore, it is all that Mr. Holmes can ever find.

The Tanners then go on to say:

"Our own computer study of the Book of Mormon has certainly not been as sophisticated as that of David I. Holmes, but we have reached similar conclusions. We approached the problem from a different angle. After noticing that the same phrases of two or more words appear time after time throughout Joseph Smith's scriptures, we used the computer to identify hundreds of these groups of words and feel that they provide powerful evidence that the Book of Mormon, the Inspired Version of the Bible, the Doctrine and Covenants and the Pearl of Great price were all the product of one mind."

Unfortunately the Tanners do not tell us their methods. This would give us some control, but they do tell us their findings, to wit:

"For example, the phrase "the lamb of God" appears only in the New Testament, John 1:29 and 36. The Mormon prophet latched onto these words and then used them twenty-eight times in the book of 1 Nephi alone! He soon grew weary of them, however, and they only appear six more times in the rest of the Book of Mormon. Smith's inclination to grab onto expressions and then repeat them is also evident in his use of "cloud of darkness." He began using this term in Alma 19:6, and then repeated it over and over in Helaman...."

This passage, summarising their findings in "Covering Up the Black Hole in the Book of Mormon," is supposed to give us the results of their computer-aided research (what could be more impartial?) In fact, loaded phrases like "he soon grew weary of them" and "Smith's inclination to grab onto expressions" show us that they are really just guessing what was going on in Joseph's mind. As a computer software developer, I would have an intense professional interest in a reliable mind-reading algorithm. It would be enormously helpful to disabled people, for example. If the Tanners really had such an algorithm, the IT world would be beating a path to their door. But if they _don't_ have such an algorithm -- and I venture to suggest that they don't -- then their computer is less an aid to their research than it is a claim to immunity from criticism. What, (they say) you don't agree with our research? Well, you must be biased, because _we_ used a _computer_! (So did I -- to write this review.)

The article also rehashes the old stand-by argument that B. H. Roberts lost faith in the Book of Mormon. Even if that were true, it would mean nothing -- people lose their faith every year -- but it has been so often and effectively shown to be wrong that to treat it again here would be to flog a dead horse.

It is just a pity that the critics of the Church do not tighten up their standards. The fact that they do not leads inevitably to the question -- is it possible that sloppy scholarship and flawed methodology are prerequisites to being an anti-Mormon? If it is, can we conclude from this that rigorous scholarship and sound methodology tend to undermine the anti-Mormon position? Surely, if the critics had some _good_ material to bring against the Book of Mormon, they would do so; can we, in fact, conclude that such a poor performance is well-nigh proof that the Book of Mormon is what it claims to be? I believe that this conclusion is a safe one. If it is not, then it falls to the critics to show that it is not. And the only way they can do that is if they dramatically lift the quality of their research.

Kerry A. Shirts's Afterword:

I enjoyed this analysis very much. What I can add, though little, is, I believe, significant enough to be understood. Wordprint studies have been around for awhile, and though the Rencher, Larson groups pioneering work has been refined through the decades, the very latest, newest study is worthy of a thorough understanding. John L. Hilton, of the Berkeley Group of computer scientists, independently of Larson and Rencher, have refined the wordprints to almost perfection. The very latest information as of 1997, on this remarkable phenomenon, is summed up thusly: (from John L. Hilton's article, "On Verifying Wordprint Studies," in Noel B. Reynolds, ed., Book of Mormon Authorship Revisited: The Evidence for Ancient Origins," FARMS, 1997, pp. 225-253)

The new format and technique for wordprinting tests have used the most highly skilled authors (such as Twain, Johnson, and Heinlein) which show that when these authors, with all their literary skill deliberately tried to imitate other authors simply couldn't do so, and were detected. "Wordprinting is practically immune to deception by a forger." (p. 228)

The use of non-contextual words such as "the", "and", and "a", are used because they are capable of being exchanged or even dropped without losing the overall meaning of sentences, paragraphs, etc. In addition this technique improves statistical accuracy. Using the noncontextual words typically make up approx. 20% to as high as 45% of the total text, which certainly provides a high number of statistical "events." (p. 228).

Now interestingly, LDS scholars have not just accepted this scientific testing without finding flaws, as D. James Croft has noted in his study and critique of Larson and Rencher. He concluded that as of 1981 there were areas which were vulnerable to criticism, and that at that time it was best to reserve judgment as to the multiple authorship of the Book of Mormon. He listed several areas that were vulnerable to criticism. John Hilton and the Berkley Group concentrated on those very areas, and with the help of Croft (A believing Latter Day Saints), worked out techniques to overcome the vulnerability, and refined the process of using wordprint studies. Several other applications on other literature and books have been used with this refined technique with absolutely incredible results! It works. As of 1997, the technique is nearly flawless, and certainly far stronger, and better than Larson and Rencher's original technique, not to take away from them at all. Their contribution is immense and has been the foundation of subsequent wordprint studies, by both LDS and Non-Mormon scholars and computer analysts.

The technique includes using word clusters as well as individual noncontextual words, and uses large segments of text, up to 30,000 words. The computer analysts tested Nephi and Alma extensively, leaving out the phrase "and it came to pass" as it was so very repetitious and excluded the quotes from the Bible since that was not Nephi himself talking. The conclusions (they have the graphs and results - really fun to see!) are that independent measure of statistical confidence that Alma and Nephi are separate authors beased on the new refined wordprints method are greater than 99.5% on one test, 99.99%, on another, 99.997% on still another! It is virtually certain that Alma and Nephi are separate authors in the BofM. It is also conclusive that Solomon Spaulding, Joseph Smith, Oliver Cowdery have been literally eliminated from contention as authors of the BofM. The Book of Mormon measures multiple authorship. with internal and external consistency even though the writings were translated. That was also taken into account and refined and new techniques for dealing with translated materials were tested to ensure accuracy. (p. 241)