Classifying Vignettes

In thinking about the intersections between journalism and literature in the nineteenth century, the space between genres—the space, we suggest, that the vignette occupies—proves crucial to our understanding of the development of a number of twentieth-century literary genres. In the nineteenth century, the hard and fast line between literature and journalism was not yet set and the roots of many contemporary literary genres such as realist fiction and literary journalism can be found in the interstitial space between the two modes. In their mix of fact and fiction, sentimental language and realistic detail, vignettes act as a kind of meeting place between literature and journalism, as well as past and future literary genres. In this section we show how the hybridity of vignettes manifests not only to human readers, but also probabilistically. In short, vignettes tend to confuse computational classifiers as thoroughly as human readers.

Theorizing the Vignette

Scholars of French literature have posited the fait divers—a more reliably sensationalistic cousin of the vignette—as a central influence on early realist fiction in France. We situate the vignette as an equally important genre in antebellum American letters, both influential in the development of sentimental fiction and a precursor to the prose writing later styled literary journalism. Like much reprinted material, vignettes largely circulated anonymously. Rather than being penned by any particular author, they were—as we argue in chapter one of this book, authored by the network of newspaper exchange itself. Vignettes were defined by their circulation and were constantly revised, recontexualized, and reauthored for new audiences. By borrowing from journalistic and fictional conventions, vignettes embody the complicated negotiation between objective truth and subjective fiction that underlay much of the period’s literature.

The novel Uncle Tom’s Cabin, for example, was a viral sensation in its own right. Originally serialized in the National Era newspaper, Harriet Beecher Stowe described the work as “a series of sketches.” These sketches were quite literally fiction, a novel published serially and then in collected volumes. Nevertheless, abolitionists took up the novel’s depictions as truthful representations of slavery, while pro-slavery groups attempted to prove the novel factually inaccurate in order to discredit its arguments. In her preface to Uncle Tom’s Cabin, Stowe (1998b) claimed the work was in a real sense true, writing, “what may be gathered of the evils of slavery from sketches like these, is not the half that could be told, of the unspeakable whole. In the northern states, these representations may, perhaps, be thought caricatures; in the southern states are witnesses who know their fidelity. What personal knowledge the author has had, of the truth of incidents such as here are related, will appear in its time.” Stowe does not mean journalistic fidelity, at least not in a twenty-first-century sense, but does maintain that her novel’s scenes speak truths. She is not claiming that her scenes literally happened, but that they are the kinds of scenes that happen.

Stowe also defended the truth of Uncle Tom’s Cabin against critics by publishing The Key to Uncle Tom’s Cabin, a book that comprises eyewitness testimonies, selections from Southern newspapers, and other evidence to prove the scenes and characters of her novel, if not literally true, are nonetheless scenes that could be true or characters who could be real people. In the first chapter of The Key to Uncle Tom’s Cabin, Stowe writes,

At different times, doubt has been expressed whether the scenes and characters pourtrayed in “Uncle Tom’s Cabin” convey a fair representation of slavery as it at present exists. This work, more, perhaps, than any other work of fiction that ever was written, has been a collection and arrangement of real incidents, of actions really performed, of words and expressions really uttered, grouped together with reference to a general result, in the same manner that the mosaic artist groups his fragments of various stones into one general picture. His is a mosaic of gems—this is a mosaic of facts. (Stowe 1998a)

Through the paired works of Uncle Tom’s Cabin and its Key (and indeed, in the field of responses to Stowe’s work in between) we can understand that nineteenth-century writers and readers saw fact and fiction not as diametrically opposed modes, but instead as points along a continuum. Fictional works could speak religious, moral, and political truths, and journalistic works could borrow the language, structures, and characterizations of fiction.

Through their form and situation, vignettes demonstrate similarly deep entanglements between the newspaper’s informational mode and the emotional mode of contemporaneous fiction. When we read nineteenth-century fiction through the lens of the newspaper vignette, we can see how deeply this short, ephemeral form penetrated the language and scene craft of a great many authors during the period. The newspaper’s vignettes and sketches offered a continually expanding, stylistically diverse, topically hybrid exploration of local, national, and even global scenes. We might consider the rapidly-shifting scenes and perspective of authors as diverse as Fanny Fern and Herman Melville as borrowing and expanding the hybridity of their contemporary newspapers.

Likewise does the vignette rest on readers’ understandings of literary conventions. Because the form is so dramatically truncated—one scene, with one set of characters, and very little exposition—it relies on readers to fill in the edges with their own knowledge of the genres from which the vignette draws and to which it contributes. If we look again at “A Religious Courtship,” for instance, we might recognize how the language of the sentimental novel—"conceived a most sudden and violent passion"—stands in for longer stretches of novelistic narrative. There is a shorthand at work in many vignettes. They read like peeks into larger stories and bigger scenes, and readers likely understood them in relationship to familiar literary examples.

Modeling Hybridity

As we outline above, we posit vignettes as a genre that straddles the border between journalism and fiction during the nineteenth century. We can demonstrate that through readerly engagements, as we have throughout this chapter, choosing phrases or rhetorical framings in particular vignettes that seem to borrow from one more or the other, as with the young man—perhaps from Albany, perhaps visiting Germantown—who woos in church. Computationally, we would expect the vignette to be tractable primarily through its relationship to the two genres from which it ostensibly borrows. Recall that computational genre classification is best suited not to generating a single, objective definition of a genre, but instead at comparing sets of texts to other sets: e.g. how much does this text look like that group, which we have designated as poems? For historical inquiry, a genre model can give us some perspective on how readers in a given time period might have understood one text in relationship to others. In the case of the vignette, we are in some ways seeking to ask the genre question slant: would readers of nineteenth-century vignettes be as uncertain about these texts’ relationship to journalism and fiction as we are?

In order to model the vignette’s hybridity computationally, we assembled three sets of texts: news items identified from our corpus by the computational genre classification method described above; a small collection of short fiction, derived from the Wright American Fiction project at Indiana University; and a set of vignettes, which we identified from our reprint corpus by manually reading through and hand-tagging the texts. To begin, we trained a classifier to be able to tell the difference between fiction and news with about 90% accuracy. Then, we introduced the vignettes as an unknown genre and asked the classifier to determine the probability that each vignette could be classified as fiction and news. Just as in our prior experiments, each “unknown” text (or vignette in this case) was tested against each genre. Thus it would be possible for a vignette to be determined to be news with a 65% probability and fiction with a 30% probability. Interestingly, however, in most cases the classifier assigned a strong probability to one genre or the other.

The above visualization bears this trend out. A few of the texts show probabilities of belonging to both news and fiction, but even in these cases the probability leans strongly in one direction or the other. Overall, if we choose only the most probable genre for each text, the classifier most often splits vignettes between news and fiction within a range of 35/65 to 50/50. On average 40% of the vignettes have a strong probability of being fiction and 60% have a strong probability of being news. That is, taken as a group, vignettes are most often classified as news, but barely so. In the majority of our cases, the split is nearly even.

Ultimately, both close and distant readings confirm our reading of vignettes as a hybrid genre that exists in the space between news and fiction in the nineteenth century. Whether classified by human readers or computationally, vignettes are a bit like fiction and a bit like news. Their similarity to fiction in style leads us to theorize about the relationship between vignettes and sentimental fiction: that they both influenced and were influenced by the dominant form of storytelling in the nineteenth century. Their topical relationship, however, points strongly to their medium: vignettes speak to the same themes and concerns present elsewhere in the newspaper, in those texts readers would have recognized as news.

We do not claim—nor could we—that every nineteenth century reader would have been baffled by every vignette they encountered. As we have seen in this chapter, some vignettes may well have been compressed accounts of real events, while others were more patently fictional, and still others were something in between. We suspect savvy readers would have been attuned to these valences and been able to read across a spectrum. What our experiments do show, however, is the variegated reality of that spectrum for nineteenth-century readers. Vignettes signified simultaneously as both journalism and fiction, functioning as interstitial, transitional texts. They exemplified the newspaper medium’s own hybridity, and stitched together its diverse genres through their form and content. In order to parse the truth of any given vignette, readers had to negotiate between the poles we describe here and intellectually situate that text in relationship to the explicit news and fiction that was often printed in the same issue.