Book of Mormon/Wordprint studies

FAIR Answers—back to home page

Wordprint studies and the Book of Mormon

Jump to details:

• Question: What do wordprint studies say about the Book of Mormon?
• Larsen and Rencher: "Our approach is sometimes referred to as the science of stylometry, which can be defined loosely as statistical analysis of style"
• Matthew Roper, Paul J. Fields, and Atul Nepal: "have applied the latest iteration of computer analyses to the unsigned editorials that appear in 1842 in the Times and Seasons"

Question: What do wordprint studies say about the Book of Mormon?

What is a wordprint?

Wordprinting, or "stylometry" as it is more commonly known, is the science of measuring literary style. The main assumption underlying stylometry is that an author has aspects of literary style that may be unconsciously used, and can be used to identify their work. Stylometrists analyze literature using statistics, math formulas and artificial intelligence to determine the "style" of an author's writing.

Because authors may write on a variety of topics, the vocabulary they use may vary considerably. Researchers often attempt to use "non-contextual words" in their analyses to avoid this problem: patterns in the use of these words (e.g. such as: and, if, the, etc.) will be less influenced by a change in subject matter.

Debate about the value of wordprints persists, though it has been used in some academic settings to identify previously-unknown authors. Readers are cautioned that the results of wordprint analysis of the Book of Mormon are only as reliable as they would be for other written works, and that "the jury is still out" as to whether wordprints can actually do what their advocates hope. The statistical analyses are not generally disputed; the points of contention revolve around the assumptions which undergird the statistics.

Initial efforts

The initial Book of Mormon wordprint studies were carried out by Larsen, Rencher, and Layton. They compared twenty-four Book of Mormon authors (each having at least 1,000 words) to each other, and concluded on the basis of three separate statistical tests that these authors were distinct from each other and Oliver Cowdery, Joseph Smith, Jr., and Solomon Spaulding.

These efforts were critiqued in Ernest H. Taves, Trouble Enough: Joseph Smith and the Book of Mormon (Buffalo, N.Y.: Prometheus Books, 1984), 225–60. John Hilton characterized Taves' review as "fundamentally flawed," and noted that his effort "therefore did nothing to add to or detract from their work."

An LDS author considered some of Larsen, Rencher, and Layton's work in Croft pointed out some flaws in their assumptions, and was cautious about whether wordprint evidence should be accepted or rejected as it then stood.

John Hilton and the Berkeley Group

A more sophisticated approach was taken by John Hilton and non-LDS colleagues at Berkeley. The "Berkeley Group's" method relied on non-contextual word patterns, rather than just individual words. This more conservative method was designed from the ground up, and required works of at least 5,000 words.

The Berkeley Group first used a variety of control tests with non-disputed authors (e.g. works by Mark Twain, and translated works from German) in an effort to:

• demonstrate the persistence of wordprints despite an author's effort to write as a different 'character'
• demonstrate that wordprints were not obliterated by translation (e.g. two different authors rendered by the same translator would still have different wordprints).

The Berkeley Group's methods have since passed peer review, and were used to identify previously unknown writings written by Thomas Hobbes.

The Berkeley Group compared Book of Mormon texts written by Nephi and Alma with themselves, with each other, and with work by Joseph, Oliver, and Solomon Spaulding. Each comparison is assessed based upon the number of "rejections" provided by the model. The greater the number of rejections, the greater the chance that the two texts were not written by the same author. Tests with non-disputed texts showed that two texts by the same author never scored more than 6 rejections; thus, one cannot be certain if scores between 1–6 were written by the same or different authors. Scores of 0 rejections makes it statistically likely the two texts were written by the same author.

However, seven or more rejections indicates that the texts were written by a different author with a high degree of probability:

# of Rejections	Certainty of being different authors
7	99.5%
8	99.9%
9	99.99%
10	99.997%

The results are striking

The results are striking:

Recall that any test over 6 indicates different authorship; 1–6 or less is indeterminate; 0 is same author. Each x represents one test.

Compare	Total Number of Tests Performed	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
Nephi vs. Nephi	3	----	----	x	----	x	x	----	----	----	----	----	----	----	----	----	----
Alma vs. Alma	3	----	x	x	x	----	----	----	----	----	----	----	----	----	----	----	----
Smith vs. Smith	3	x	----	xx	----	----	----	----	----	----	----	----	----	----	----	----	----
Cowdery vs. Cowdery	1	----	x	----	----	----	----	----	----	----	----	----	----	----	----	----	----
Spaulding vs. Spaulding	1	----	----	x	----	----	----	----	----	----	----	----	----	----	----	----	----
Nephi vs. Alma	9	----	----	x	----	----	xx	xx	x	x	x	x	----	----	----	----	----
Smith vs. Nephi	6	----	----	----	----	x	----	----	----	xx	----	x	x	x	----	----	----
Smith vs. Alma	6	----	----	----	xx	x	x	----	xx	----	----	----	----	----	----	----	----
Cowdery vs. Nephi	6	----	----	----	----	----	----	x	x	----	----	----	xx	----	x	x	----
Cowdery vs. Alma	6	----	----	----	----	----	----	----	xxxx	x	x	----	----	----	----	----	----
Spaulding vs. Nephi	6	----	----	----	----	----	----	----	----	----	----	x	x	x	----	x	xx
Spaulding vs. Alma	6	----	----	----	----	----	----	xxx	----	xx	----	----	----	x	----	----	-

Furthermore, each "rejection" is statistically independent—this means that the chance of two different authors being the product the same person can be determined by multiplying the chance of each individual failure.

Thus the chance of Nephi and Alma being the same author is found by:

chance of 7 rejections x 8 rejections x 9 rejections x 10 rejections

= 0.005 x 0.001 x 0.0001 x 0.00003

= 0.000000000000015

=1.5 x 10^-14

This is a roughly 1 in 15 trillion chance of Nephi and Alma having the same author. Hilton rightly terms this "statistical overkill".

Authors	Cumulative chance of being the same author
Nephi and Alma	1.5 x 10-14
Joseph Smith and Alma	2.5 x 10-5
Joseph Smith and Nephi	less than 2.7 x 10-20
Oliver Cowdery and Alma	6.25 x 10-17
Oliver Cowdery and Nephi	less than 8.1 x 10-19
Spaulding and Alma	less than 3 x 10-11
Spaulding and Nephi	less than 7.29 x 10-28

Wordprinting is a valid technique, then this analysis suggests that it is "statistically indefensible" to claim that Joseph, Oliver, or Solomon Spaulding wrote the 30,000 words in the Book of Mormon attributed to Nephi and Alma

As John Hilton put the matter, if wordprinting is a valid technique, then this analysis suggests that it is "statistically indefensible" to claim that Joseph, Oliver, or Solomon Spaulding wrote the 30,000 words in the Book of Mormon attributed to Nephi and Alma. The Book of Mormon also contains work written by more than one author. Critics who wish to reject Joseph's account of the Book of Mormon's production must therefore identify multiple authors for the text, and then explain how Joseph acquired it and managed to pass it off as his own.

Additional Studies

Later studies continue to debate the application of wordprint studies to Book of Mormon authorship. Studies proposing 19th century authorship include:

• A 2008 study by Matthew L. Jockers, et al., trying to defend the Spaulding-Rigdon theory. It was thoroughly critiqued by G. Bruce Schaalje, et al. in the journal Literary and Linguistic Computing.
• A 2013 study by Chris and Duane Johnson proposing that Joseph Smith based the Book of Mormon on an 1816 book The Late War Between the United States and Great Britain. The study was thoroughly critiqued by an article in Interpreter: A Journal of Latter-day Saint Faith and Scholarship.

Studies supporting ancient authorship include:

• John B. Archer, et al., "Comparative Power of Three Author-Attribution Techniques for Differentiating Authors," Journal of Book of Mormon Studies 6:1 (1997), 47–63.
• Matthew Roper, et al., "Stylometric Analyses of the Book of Mormon: A Short History," Journal of Book of Mormon Studies 21 (2012), 28–45.
• Book of Mormon Central Staff, "Is It Possible That a Single Author Wrote the Book of Mormon?" Book of Mormon Central, KnoWhy #399, posted 16 January 2018.
• John Hilton III, et al., "Gentiles in the Book of Mormon," Interpreter: A Journal of Latter-day Saint Faith and Scholarship 33 (2019), 267–288.
• Stanford Carmack, "Personal Relative Pronoun Usage in the Book of Mormon: An Important Authorship Diagnostic," Interpreter: A Journal of Latter-day Saint Faith and Scholarship 49 (2021), 5–36.

Larsen and Rencher: "Our approach is sometimes referred to as the science of stylometry, which can be defined loosely as statistical analysis of style"

Wayne A. Larsen and Alvin C. Rencher,

Our approach is sometimes referred to as the science of stylometry, which can be defined loosely as statistical analysis of style. It is also called computational stylistics. We do not use the word style in the literary sense of subjective impressions characterizing an author's mode of expression. We must deal with countable items which are amenable to statistical analysis. We look then for what is frequent but largely unnoticed, the quick little choices that confront an author in nearly every sentence. Such choices become habits, so the small details flow virtually without conscious effort. —(Click here to continue)

Matthew Roper, Paul J. Fields, and Atul Nepal: "have applied the latest iteration of computer analyses to the unsigned editorials that appear in 1842 in the Times and Seasons"

Matthew Roper, Paul J. Fields, and Atul Nepal,

For over three decades now, computer analyses (using human-written programs, of course) have been used to differentiate the writing styles of authors. Over these decades, the analyses have become more sophisticated and more accurate, though accuracy is still relegated to probability, never certainty. Matt Roper, Paul Fields, and Atul Nepal have applied the latest iteration of computer analyses to the unsigned editorials that appear in 1842 in the Times and Seasons. Did Joseph Smith write the LDS editorial comments on Stephens and Catherwood’s book on Central American ruins? Read and see. —(Click here to continue)

Notes