Response to Jockers, Criddle, et al., Now Available

[_GlennThigpen]

[Roger wrote]
If Bruce is still paying attention to this thread, or if someone can contact him, there are several of us who would be interested to know whether he thinks indications from word-print studies are that the Book of Mormon was created by more than one author.[/Roger wrote]

[Marg wrote]
I asked Bruce as did Roger if the wordprint studies can show multiple versus single authorship for the Book of Mormon..I'm not aware of him answering that question. I think at a minimum that is something the wordprint studies should be able to show. If they show multiple authorship and it is strong evidence for multiple authorship then arguing against the Smith alone theory is a waste of time.[/Marg wrote]



I think that John Hilton’s work sheds the most light on this. He worked with equal-sized texts of 5000 words, all from the printer’s manuscript. He worked only with doctrinal texts associated with Nephi or Alma so that there would be no genre issues. Texts attributed to Nephi were stylometrically similar to each other, as were the texts attributed to Alma. However, the Nephi texts were stylometrically and statistically distinct from the Alma texts. That is, the doctrinal texts segregated by purported author.

I guess it’s possible that one writer could be talented enough to manipulate non-contextual words and word-patterns in such a way that texts due to separate fictional figures would be internally consistent but inconsistent between fictional figures in terms of unconscious stylometric patters. None of John’s Hilton’s control authors (Samuel Johnson, Mark Twain, Robert Heinlein), however, could do it except for William Faulkner. So I think the case is pretty strong that the Book of Mormon is of multiple authorship.

The other thing that supports multiple authorship is simply the extreme stylometric variation displayed by chapters of the Book of Mormon—much more than displayed by sections of Spaldings’
Manuscript Story for example.

How well would the Hilton methodology work for discerning differences between many authors since it uses 5000 word blocks of text? Could the NSC method be used for that purpose? The older Laresen, Rencher, Layton study tested twenty-four authors and found statistically significant differences.

Glenn

[_bschaalje]

Glenn said:
How well would the Hilton methodology work for discerning differences between many authors since it uses 5000 word blocks of text? Could the NSC method be used for that purpose? The older Larsen, Rencher, Layton study tested twenty-four authors and found statistically significant differences.

The Hilton methodology would work for the many-authors situation if you had enough texts and they were all of the same literary genre. Hilton worried that some of the between-author differences that Larsen-Rencher-Layton detected might have actually been differences due to genre. For example, Mormon’s writings may have differed from Jacob’s simply because Jacob’s writings were all doctrinal while the majority of Mormon’s writings were historical narrative. So Hilton stuck to Nephi’s and Alma’s doctrinal writings because there were a number of them and they were large.

The Larsen-Rencher-Layton MANOVA test simply rejected the general null hypothesis of no stylometric differences among the groups of texts attributed by the Book of Mormon to different authors, without indicating which groups were different from which other groups. Yes there were differences, but genre could have accounted for some of the differences.

The NSC method is (in machine-language terms) a ‘supervised learning’ method in which the groups are predefined, and the question is simply to which (if any) of the predefined groups does a new item (text) belong? It does not specifically answer the question ‘Were there multiple authors?’, but if the test texts are assigned to more than one training author the method implies multiple authorship. But you have to have training texts by the correct authors, in the same genre as the test texts.

Hilton’s method provided strong evidence that:
1. the author of Nephi’s doctrinal discourses was a single author because the multiple Nephi texts were as similar to each other as control texts known to be written by the same author,
2. the author of Alma’s doctrinal discourses was a single author because the multiple Alma texts were as similar to each other as control texts known to be written by the same author,
3. the author of Nephi’s doctrinal discourses was not the same as the author of Alma’s doctrinal discourses because the Nephi texts were highly distinct from the Alma texts,
4. and therefore, at least two distinct authors contributed to the Book of Mormon (i.e. multiple authors)

[_GlennThigpen]

The Hilton methodology would work for the many-authors situation if you had enough texts and they were all of the same literary genre. Hilton worried that some of the between-author differences that Larsen-Rencher-Layton detected might have actually been differences due to genre. For example, Mormon’s writings may have differed from Jacob’s simply because Jacob’s writings were all doctrinal while the majority of Mormon’s writings were historical narrative. So Hilton stuck to Nephi’s and Alma’s doctrinal writings because there were a number of them and they were large.

The Larsen-Rencher-Layton MANOVA test simply rejected the general null hypothesis of no stylometric differences among the groups of texts attributed by the Book of Mormon to different authors, without indicating which groups were different from which other groups. Yes there were differences, but genre could have accounted for some of the differences.

The NSC method is (in machine-language terms) a ‘supervised learning’ method in which the groups are predefined, and the question is simply to which (if any) of the predefined groups does a new item (text) belong? It does not specifically answer the question ‘Were there multiple authors?’, but if the test texts are assigned to more than one training author the method implies multiple authorship. But you have to have training texts by the correct authors, in the same genre as the test texts.

Hilton’s method provided strong evidence that:
1. the author of Nephi’s doctrinal discourses was a single author because the multiple Nephi texts were as similar to each other as control texts known to be written by the same author,
2. the author of Alma’s doctrinal discourses was a single author because the multiple Alma texts were as similar to each other as control texts known to be written by the same author,
3. the author of Nephi’s doctrinal discourses was not the same as the author of Alma’s doctrinal discourses because the Nephi texts were highly distinct from the Alma texts,
4. and therefore, at least two distinct authors contributed to the Book of Mormon (i.e. multiple authors)[/quote]


I produced the full text of Bruce's answer for context.

So, would it be possible to test for multiple authors using the NSC method by culling Nephi for historical narrative sections and doctrinal sections to produce a historical genre and a doctrinal genre to test the one against the other to ascertain if genre does indeed make a significant difference?

Culling Nephi's words for quotes by Lehi could produce a profile for Lehi, which would be almost entirely doctrinal. The same is true for Alma and Jacob. I don't know how the Book of Enos would be classified. The words of King Benjamin could also be excerpted to produce a block of text in a doctrinal genre. In like manner, the words spoken by Helaman as well as any of the authors named by the Book of Mormon for which sufficient sized text blocks could be obtained.

Is this something that can be done and tested against each other? Such tests should obviate the problem of archaic language at least and using only doctrinal excerpts should eliminate the genre concern.

The original Jockers, et al study took the Book of Mormon on a chapter by chapter basis and did not try to separate the words of Nephi from his quotes of his father or anyone else. I am not sure just what the assumptions were, but it seems to me that in order to produce the most accurate results, that such an effort needs to be made in order to be able to check if Nephi was really quoting someone else or the Mormon was indeed quoting Alma, Helaman, etc.


Glenn

[_marg]

I guess it’s possible that one writer could be talented enough to manipulate non-contextual words and word-patterns in such a way that texts due to separate fictional figures would be internally consistent but inconsistent between fictional figures in terms of unconscious stylometric patters.

So then according to you, none of the wordprint studies that you know of, eliminate with any degree of probability...one authorship for the Book of Mormon?

[_bschaalje]

Glenn said:

So, would it be possible to test for multiple authors using the NSC method by culling Nephi for historical narrative sections and doctrinal sections to produce a historical genre and a doctrinal genre to test the one against the other to ascertain if genre does indeed make a significant difference?

Culling Nephi's words for quotes by Lehi could produce a profile for Lehi, which would be almost entirely doctrinal. The same is true for Alma and Jacob. I don't know how the Book of Enos would be classified. The words of King Benjamin could also be excerpted to produce a block of text in a doctrinal genre. In like manner, the words spoken by Helaman as well as any of the authors named by the Book of Mormon for which sufficient sized text blocks could be obtained.

Is this something that can be done and tested against each other? Such tests should obviate the problem of archaic language at least and using only doctrinal excerpts should eliminate the genre concern.

The original Jockers, et al study took the Book of Mormon on a chapter by chapter basis and did not try to separate the words of Nephi from his quotes of his father or anyone else. I am not sure just what the assumptions were, but it seems to me that in order to produce the most accurate results, that such an effort needs to be made in order to be able to check if Nephi was really quoting someone else or the Mormon was indeed quoting Alma, Helaman, etc.

I think this could be done. Larsen and Rencher did essentially this in their original study, except that they did not separate the texts by genre. But they went through the whole 1920 edition of the Book of Mormon, parsing essentially every word of text to one of twenty-some authors. Doing this again, you would have to be careful that the texts are not too small, and I don’t think it’s always obvious whether a particular section of text is doctrinal discourse or historical narrative. But now that we have the Yale edition, something like this ought to be done.

Marg said:
So then according to you, none of the wordprint studies that you know of, eliminate with any degree of probability...one authorship for the Book of Mormon?

I’m not exactly sure what you are asking (double negatives). My conclusion is that the stylometric evidence points with quite high certainty to multiple authorship for the Book of Mormon.

[_Roger]

Bruce:

Thanks for your replies.

I guess it’s possible that one writer could be talented enough to manipulate non-contextual words and word-patterns in such a way that texts due to separate fictional figures would be internally consistent but inconsistent between fictional figures in terms of unconscious stylometric patters. None of John’s Hilton’s control authors (Samuel Johnson, Mark Twain, Robert Heinlein), however, could do it except for William Faulkner. So I think the case is pretty strong that the Book of Mormon is of multiple authorship.

Common sense alone would imply that such a feat would require either a deliberate, highly skilled attempt (which would rule out the "unconscious" part) or incredible luck.

The other thing that supports multiple authorship is simply the extreme stylometric variation displayed by chapters of the Book of Mormon—much more than displayed by sections of Spaldings’
Manuscript Story for example.

Agreed. I think error/dialect patterns show the same thing. I think Metcalf's wherefore/therefore "shift" is also evidence for more than one author. Under a single author framework, the shift has to be considered a legitimate shift in preference. Under multiple authors, one author could tend to use wherefore while another would tend to use therefore.

[_marg]

Marg said:
So then according to you, none of the wordprint studies that you know of, eliminate with any degree of probability...one authorship for the Book of Mormon?

I’m not exactly sure what you are asking (double negatives). My conclusion is that the stylometric evidence points with quite high certainty to multiple authorship for the Book of Mormon.

Well I've seen Dan V. argue that J. Smith put on an act that he couldn't spell, write coherently or have knowledge re: walls around Jerusalem. It was all a well orchestrated act, according to Dan to fool his wife and the rest of the Book of Mormon translation witnesses. So since you are saying it's possible that a writer could deliberately write as different characters changing the style for each character and no wordprint study could necessarily detect this deliberate writing style..then if that were the case, that could be argued by Dan V to support the Smith alone.

[_Roger]

Glenn:

So, would it be possible to test for multiple authors using the NSC method by culling Nephi for historical narrative sections and doctrinal sections to produce a historical genre and a doctrinal genre to test the one against the other to ascertain if genre does indeed make a significant difference?

Culling Nephi's words for quotes by Lehi could produce a profile for Lehi, which would be almost entirely doctrinal. The same is true for Alma and Jacob. I don't know how the Book of Enos would be classified. The words of King Benjamin could also be excerpted to produce a block of text in a doctrinal genre. In like manner, the words spoken by Helaman as well as any of the authors named by the Book of Mormon for which sufficient sized text blocks could be obtained.

Is this something that can be done and tested against each other? Such tests should obviate the problem of archaic language at least and using only doctrinal excerpts should eliminate the genre concern.

The original Jockers, et al study took the Book of Mormon on a chapter by chapter basis and did not try to separate the words of Nephi from his quotes of his father or anyone else. I am not sure just what the assumptions were, but it seems to me that in order to produce the most accurate results, that such an effort needs to be made in order to be able to check if Nephi was really quoting someone else or the Mormon was indeed quoting Alma, Helaman, etc.

Glenn, I am no expert, but in theory, I think I like your idea. I wonder if you would run into difficulty defining what exactly would make the cut for or against being "doctrinal" but if a standard could be agreed on then I think the results could be interesting. Last I knew, Ron Dawborn and Margie Miller were attempting to separate the "religious" material from the historical content in the Book of Mormon, so they may have developed some standard for determining what is and isn't "doctrinal."

[_Dan Vogel]

Marg,

Well I've seen Dan V. argue that J. Smith put on an act that he couldn't spell, write coherently or have knowledge re: walls around Jerusalem. It was all a well orchestrated act, according to Dan to fool his wife and the rest of the Book of Mormon translation witnesses. So since you are saying it's possible that a writer could deliberately write as different characters changing the style for each character and no wordprint study could necessarily detect this deliberate writing style..then if that were the case, that could be argued by Dan V to support the Smith alone.

I would add that Joseph Smith was dictating, not writing, and he was also a first-time author developing his skill and had not settled down into a style like the seasoned authors, who also rewrote their stories many times. It would be interesting to do word print analysis on storytellers who create and deliver their stories extemporaneously. An ancient practice, this still goes on today. I remember reading about anthropologists recording performers in Russia or Hungary. I would be mostly interested in storytellers who act out the different characters, and intentionally change persona and voice.

Bruce mentioned use of the 1920 edition, which seems problematic to me. His suggestion of using Skousen’s Yale edition would be an improvement.

[_GlennThigpen]

Marg,

Well I've seen Dan V. argue that J. Smith put on an act that he couldn't spell, write coherently or have knowledge re: walls around Jerusalem. It was all a well orchestrated act, according to Dan to fool his wife and the rest of the Book of Mormon translation witnesses. So since you are saying it's possible that a writer could deliberately write as different characters changing the style for each character and no wordprint study could necessarily detect this deliberate writing style..then if that were the case, that could be argued by Dan V to support the Smith alone.

I would add that Joseph Smith was dictating, not writing, and he was also a first-time author developing his skill and had not settled down into a style like the seasoned authors, who also rewrote their stories many times. It would be interesting to do word print analysis on storytellers who create and deliver their stories extemporaneously. An ancient practice, this still goes on today. I remember reading about anthropologists recording performers in Russia or Hungary. I would be mostly interested in storytellers who act out the different characters, and intentionally change persona and voice.

Bruce mentioned use of the 1920 edition, which seems problematic to me. His suggestion of using Skousen’s Yale edition would be an improvement.

Marge, Bruce suggested that some talented authors may be able to consciously change their use of noncontextual words and phrases. In the 1993 study by Tim Hiatt and John Hilton, only one author was able to do so, and that was William Faulkner in "As I lay Dying." This book is described as a "stream of consciousness" work, which seems to have some likeness to automatic writing. It is in a narrative form by something like fifteen characters in the story.

The conclusion reached by the Berkeley Group is that an author would have to deliberately act to change his normal writing style, and then very few would be able to do so, as evidenced by the Hiatt and Hilton study.

Larsen, Rencher, and Layton used the 1920 edition. The Berkeley Group used the printer's manuscript. But I do believe the Skousen Yale edition would be the best one to use in future wordprint studies.

In any event, it appears that a lot more work needs to be done before anyone declares victory.

Glenn