The Voices in Faulkner's Head

Hiatt and Hilton (1990), Deseret Language and Linguistic Society Symposium, "Can Authors Alter Their Wordprints? Faulkner's Narrators in As I Lay Dying", pp. 52-60. https://journals.lib.byu.edu/spc/index.php/DLLS/issue/view/2429

The papers we've examined so far about Book of Mormon Stylometry all arrive at at least this conclusion: styles of multiple, unidentified authors are included in the Book of Mormon. An unanswered question is, just how unusual is this? Are there examples of single authors who have created distinct wordprints for different stories or different narrators within a single work? We've seen how skilled authors failed to do so in imitating Jane Austen and Arthur Conan Doyle. We've also seen how unskilled authors succeeded under special conditions, but the Book of Mormon clearly matches the conditions of the Austen and Doyle imitators much better than the "adversarial authorship" conditions of Greenstadt's studies. Showing his willingness to be thorough and to risk his conclusions being weakened, Hilton analyzed As I Lay Dying, a work in which Faulkner narrated the text in the voices of 15 different characters. The question was, did Faulkner succeed in creating statistically different voices?

The answer to this question is yes. Most of the 15 narrators didn't say enough to be analyzed, but three did. Of course, the others could be analyzed by less exacting methods than using noncontextual word pairs, but Hilton wanted to use the most discriminating methodology he had found. You can read the details at the link provided, but here is a table summarizing the most important conclusions for us:

As a reminder, from previous examination of numerous texts from a variety of authors, Hilton found an empirical rule that if 7 or more word pairs differed in their frequencies (7 rejections) between two texts, then the texts were by different authors. If there were 1-6 rejections, then you couldn't say with >95% certainty, but most authors average 2-3 rejections between different samples of their writing. The texts used in this study were no different, for the most part. Faulkner's narrators (Darl, Tull, and Vardaman) and Faulkner's other novels (Pylon and Light in August) were written in styles very different from the control texts written by Samuel Clemens, Robert Heinlein, Samuel Johnson, and Harry Steinhauer. You can see this in the second column of the table. Every one of Faulkner's texts differs by 8.4 or more rejections. Faulkner--no matter what voice he was writing in--had a different style from all of the control authors. Each of the control authors' samples differed from other texts by the same author by 1-6 rejections, with an average for within author comparisons of 2.9 rejections.

Now I want to point out something you have to dig through the text a little to find. I've added it to the bottom of the table. Heinlein wrote a book where the chapters are split between two narrators. These two different characters used noncontextual word pairs in almost exactly the same way, and almost exactly the same way Heinlein always did. Clemens's also wrote a couple of works in the first person from two different perspectives--those of Adam and Eve. These two characters used noncontextual word pairs in the same way, and in the same way as others of Clemens's writings--even some from 3rd person perspectives. Clearly this is not a survey of all authors that ever attempted to write in different voices, but let's see how different Faulkner's attempts were from Heinlein and Clemens.

Looking at the first column, we have evidence that Faulkner could write with consistent styles. The samples from Pylon and Light in August averaged only a few rejections within the same text. Darl also narrated his two samples with fair consistency. Looking at the third column is where the real genius shows. Faulkner was able to write Tull and Vardaman so different from his other styles that he inflated the numbers in the entire last column. Pylon and Light in August may be similar to each other, but there is so much variability in Faulkner's wordprint that none of his sample texts match the average as closely as most authors. Tull and Vardaman appear to be completely different authors from Faulkner, although no one disputes their authorship.

What does this mean for the Book of Mormon? We have an example of an author, even if it is a rare example, who has changed his wordprint intentionally, consistently, and significantly at least twice in the same book. Hiatt and Hilton dug into the details of the differences and uncovered some interesting patterns. I could repeat them all, but I'll give one example and refer you to the paper for the full discussion:

Test 75 shows the percentage that "in the" occurs of all uses of "the." Vardaman consistently uses "in the" more (16.3%) than the other characters (6, 7.5, 5.7, and 7.4%). Vardaman's phrasing is influenced by his character: being a young boy, he observes and comments on things using language familiar to a young boy. He comments on his mother "in the box" and people and horses "in the river," and "in the water." Vardaman uses the prepositional "in" instead of mentioning where the object is and letting the reader remember. Faulkner apparently believed that a youngster's mind would work that way.

In creating characters, Faulkner created dialects for his characters. To quote from Hiatt and Hilton's conclusions:

Faulkner's ability to vary "non-contextual" word usage patterns seems to contradict a simple thesis of wordprinting. Yet, as we can see through our analysis of the Morton word pattern tests that were rejected, Faulkner was able to change his patterns using authorial techniques. Faulkner seems to have taken into account each narrator's socio-economic position, age, and education, as well as other intangibles like attitude and experience when creating these characters, at least as far as the within author measurements are concerned.

Therefore, although Faulkner broke the mold in creating these different narrators, his methods were discernible using this wordprinting technique.

It appears that a lot of careful thought, and probably a certain genius for language, goes into creating a character with truly different, subconscious word pattern usages. And in fact, this isn't what Faulkner did. Faulkner appears to have taken a handful of patterns most of us experience as subconscious and made them conscious. The remainder of his word pairs remain largely consistent from one text to another, even when he switches from 1st to 3rd person narration.

I guess that means the next job is to try to identify what tricks Joseph Smith used to change his noncontextual word pair ratios from Nephi to Alma to the Doctrine and Covenants to his personal writings. All I can say is, good luck. Otherwise, it's about time to accept the fact that different authors wrote different parts of the Book of Mormon, and none of them were 19th century candidates.