Novelists often write about the past, but how do particularly skilled novelists create a realistic and vivid impression of bygone eras? How do writers invent a vision of the past that rings true, at least with how we think about historical times today? Take a quirky recent book, Cloud Atlas, by David Mitchell, a genre-bending novel that spans six time periods, from the 19th century to the far-flung future: How does it manufacture its different pasts?
Over the years, scholars and lay readers have approached this question—how to convey pastness, in a way that the present accepts it as the past—from a number of angles. But to get a more definitive answer, I took a quantitative approach. And what the data reveals is far more surprising and unexpected than what I (or anyone else) has thought.
Some things that writers do are obvious. Various types of objects—horse-drawn carriages, country mansions, workhouses—appear in 19th-century novels; such objects, predictably, also appear in contemporary novels set in these times. It is perhaps self-evident that such appropriate settings and objects contribute to the creation of scenes that feel “truly” historical. It is less clear that other techniques, such as using accurate language from the time, actually work: people turn out not to be very good at recognizing anachronistic language.
But the actual way to fake a 19th-century novel, if Cloud Atlas is any guide, is much darker. Because the quantitative data reveals that if you really want to sound like a novelist from the 19th century, one of the best ways to do this is to use a lot of discriminatory language or to make your narrator sound like a racist. This approach is helpful in critically condemning discriminatory language—and I do not for one minute believe that Mitchell uses such language for any reason except to critique it—but it also runs the risk of making us believe that problems of racism are safely boxed away in the past.
This commitment to foregrounding racism to evoke the past contributes to a type of imagined language that Mitchell has vividly described. “To a degree,” writes Mitchell, “the historical novelist must create a sort of dialect—I call it ‘Bygonese’—which is inaccurate but plausible. Like a coat of antique-effect varnish on a new pine dresser, it is both synthetic and the least-worst solution.” Mitchell’s writing turns out to be an excellent test bed for learning the features of this imagined bygonese, a form also known as “Wardour Street English.”
Cloud Atlas, for those who don’t know (or who are only familiar with the movie version), is a novel that is divided into six chapters. Halfway through each chapter, though, the text “breaks” and cuts to the next in the sequence, leaving the reader hanging. In the second half of the novel, the reader is given the “missing” finales of the chapters, cascading back down in reverse order. Importantly, as shown in figure 1, the text moves forward in historical time in its first half—spiraling up toward a dystopian future in “Sloosha’s Crossin’”—before, in the second half, spiraling back down toward the past again as the novel reaches its end, where it started—in “The Pacific Journal of Adam Ewing.”
Figure 1. Narrative time, historical time, and chapter progression in Cloud Atlas. Reproduced from Martin Paul Eve, Close Reading with Computers: Textual Scholarship, Computational Formalism, and David Mitchell’s Cloud Atlas (Stanford University Press, 2019), under a CC BY-SA license.
As if this weren’t complicated enough, the language of Cloud Atlas—what we might call its “register”—changes in each chapter, so as to mirror the historical period in which it is set. For example, the first chapter, “The Pacific Journal of Adam Ewing,” is set in the style of a seafaring 19th-century narrative; “Half-Lives—The First Luisa Rey Mystery” takes the form of a 1960s crime thriller, written in a fast-paced present tense; “An Orison of Sonmi~451” is a kind of future science-fictional documentary interview, etc.
What I wanted to know, though, was this: Given that Mitchell has written about how he constructs his historical language, what does he actually do on the ground? How does Mitchell build a 19th-century prose style?
One of the first questions to ask is whether the language—in this case, of Mitchell—is accurate. Clearly, technological anachronisms (such as satellite communication) within fiction supposedly “from” a previous era would subvert the historical realism of a novel. Mitchell’s novel does not feature such clear, attention-seeking anachronism. What I did wonder, though, was whether Cloud Atlas’s 1850 section featured words that would not be accessible to an author actually writing at this time. In other words: Was there a linguistic anachronism at work in Mitchell’s writing—and how good are we, as readers, at recognizing such features?
As it turns out: we are pretty terrible at spotting words that are out of time. But Mitchell also does an extraordinarily good job of restricting his vocabulary. In order to show this, I wrote a computer program that would look up each word in the text in various etymological dictionaries, in order to ascertain its first usage date. The results were surprising (fig. 2).
Figure 2. Word distributions in Part I of “The Pacific Journal of Adam Ewing” by first usage, according to Dictionary.com and the OED. Reproduced from Martin Paul Eve, Close Reading with Computers: Textual Scholarship, Computational Formalism, and David Mitchell’s Cloud Atlas (Stanford University Press, 2019), under a CC BY-SA license.
In my search of Dictionary.com, I found just six anachronistic words that would definitively not have been available to the narrator of the first chapter of Cloud Atlas, and therefore are not “realistic”:
-
- home-town [1910–1915]
- spillage [1920–1925]
- lazy-eye [1935–1940]
- returnees [1940–1945]
- Latinos [1945–1950]
- A-frame [1960–1965]
Yet the Oxford English Dictionary disagrees. “Hometown,” it tells us, was coined in 1851; “returnee” in 1870; and “A-frame” as far back as 1827. The OED also yields a number of other terms coined after our cutoff date that Dictionary.com does not reveal. The OED, through my automatic approach, gives us:
-
- bizarreness [1920]
- slumped [1937]
- pulsed [1942]
- colour [1944]
- scuttlebutt [1945]
There are many complexities to this. (Why does the OED indicate that “colour” first appeared in 1944? Because that sense of the word refers to musical timbre.) We are left, though, with just three terms that I feel can be said to have been absolutely inaccessible to both Mitchell’s historical author and the intradiegetic editor: “spillage,” from ~1934; “Latino,” from ~1946; and “lazy-eye,” from ~1935.
“Lazy-eye,” in particular, is a startling find. This term sounds like a pejorative slur for people with amblyopia that would have been coined well before 1935. It’s something we would expect to find in literature from 1850, not in contemporary parlance—so this term works quite well for Cloud Atlas.
The fact that we cannot recognize which words are appropriate to a time period brings to the fore a problem that has vexed historical fiction and its study for many years: To what extent is accuracy to the historical record actually important? And if the language is not totally historically accurate, what other markers might signify to a contemporary reader that the work is from the past?
First and foremost, to achieve his historical style, Mitchell does use archaic language, as we might predict. Within the first few lines of the text, we encounter “Indian,” used to refer to any non-European; a “hamlet,” or a settlement; “trowzers,” a variation of “trousers”; a “Pea-jacket,” a jacket of 18th-century origin; an ampersand (&), used as a conjunction instead of the more common “and”; and the term “eyrie,” meaning a homeland.
How does David Mitchell build a 19th-century prose style?
However, I wondered whether Mitchell might simply be using uncommon language to create the perception of a stylistic affinity with Victorian-era prose. For when readers do not know when words were actually first used, it might make sense to present them with a range of words that they are less likely to have encountered. This unfamiliarity might lead these words to be construed, then, as being older than those used in day-to-day speech. Or, more simply put: the less familiar the language, the more archaic it might sound.
In order to test this, I took a set of magazine articles from 2004. I then removed all words from Cloud Atlas’s Ewing chapter that are found in these contemporary magazines, leaving the words that occur only in Cloud Atlas. The result is, indeed, a set of terms that are unusual to the modern ear.
In particular, though, Cloud Atlas’s Ewing chapter falls back on offensive racial addresses in order to achieve its historical style and its critical focus on the legacies of colonialism. For instance, Mitchell’s text gives us: “Blackamoor,” “blackfella,” “darkies,” “harridan,” “womenfolk,” “bedlamite,” “mulatto,” “quadroon,” and “mixedblood.” In fact, colonial terms of racist abuse occur in the Ewing section of Cloud Atlas at a far higher frequency than in a broader contemporary literary corpus. I should stress once again, however, that these terms are used in the text for purposes of critique—not to condone such language and its imperial origins.
There’s a great deal of debate about whether specific terms in texts are good markers of their genre. For instance, the critic Ted Underwood has recently pointed out that science fiction novels are more clearly defined by their use of the adjectives of scale than by spaceships and so forth (although they do often have the latter). In Cloud Atlas, though, racial epithets serve to build an empire-based racist charge in the language, which contributes to our belief that this writing could really have come from 1850. This is strengthened in the narrative by the outmoded colonial-era notions of “tropical medicine,” in which the white man may fall prey to the diseases of the warmer climes.
All of which is to say that Cloud Atlas makes for an excellent case study of how contemporary writers build an aesthetic of writing from the past. We like to imagine that language choice is all-important and that every word counts. But as I found out, we aren’t very good at recognizing which words could have been used in 1850 and which words could not.
The manner in which such writing sets itself apart from other contemporary works—through, say, racist and colonial markers—often denotes how we think we wrote back then. In many ways, this also rests upon a belief that we are better now and that things always tend to improve. If we are not careful, there can be a dangerous tendency to believe that the use of offensive racial epithets is a problem conveniently stowed out of sight in the 19th century. But there are telling points to consider for our times: it looks to me as though at least one of the ways that we evaluate the writing of the past lies in how authors write about those from other demographics.
The question then becomes: How will future writers pretend that they are writing in the 21st century? What will be the signature marker of our own style, for novelists in times to come?
This article was commissioned by Richard Jean So. 
