I’ll be honest; my presentation for MLA this year–a pilot study on how linguists talk about literary texts in the journal American Speech in the early 20th century–was probably the least polished talk I’ve ever done at MLA, or maybe anywhere. It was difficult to focus on that work knowing people were storming the capitol and not knowing what that might mean. I made my slides using the MLA template and never figured out how to change the session # info at the bottom–I just left it that way, y’all! Wow. Anyway, I thought some of you might enjoy this series of posts I’m doing to share some of the “out-takes,” as I’m calling them–bits and pieces of what I found that didn’t fit in the time, or just weren’t maximally relevant to the talk.
I’m interested in (among other things) the relationship between linguistics as a field (as broad, varied, and ever-changing a thing that is) and literary texts. Literary texts are often used as data for variationist studies, like documenting varieties of English or analyzing how those varieties change. This is both important and problematic, as we know that literary texts are not transcriptions of speech at all, and are subject to confounding effects from the author, genre, etc. (Some researchers attend to this confound; many don’t.) I know a fair amount about the long nineteenth century, and I know a fair amount about sociolinguistics of the 2000s and beyond, but I wanted to know what was happening in the 20th century–how is literature being used by linguists in that era? To limit the study, I did a pilot of sorts, looking at the first 40 years of issues of the journal American Speech, and specifically looking at studies on African American language.
The results show a real anxiety about the use of literary texts as data. I’d like to sit with one of the later articles for a bit, called “Negro Dialect in Eighteenth-Century American Drama” from 1955, by Richard Walser. Walser argues for using early American drama as a source for what early African American language was like because:
Askew or not, the dialect as written was uninfluenced, at least in the earliest efforts, by literary tradition; and there can be little doubt that most of the Negro characters were drawn from life and that their speech is a crude, if occasionally distorted, duplication of the sounds which the playwrights thought they heard. (269)
There are a few things going on here that are interesting to me. One is that, although Walser is making a case for verisimilitude, it seems to be a weak one that I’m not sure even he truly believes. The representations are “askew,” “crude,” “distorted” versions of what writers “thought they heard.”
I’m also interested in the metaphors–“drawn from life” is an artistic endeavor. “Duplication” sounds a bit more rigorous and reliable, but how do you duplicate a sound?
The crude distortions are framed here as a result of the honest pursuit of verisimilitude-inconsistencies are to be expected, right? And they are framed in opposition to the suspicious regularities of the “local color” movement, which Walser argues is the result of literary convention and genre taking over:
The ten plays impose two general observations. None of the playwrights, with the exception of Murdock, exhibits any great consistency in rendering the dialect. Spelling is capricious, even within one play; and syntax is widely variable […] The variations are understandable. Not until much later did written Negro dialect assume its literary conventionality; nor was any uniformity possible in the eighteenth century. (269)
This inconsistency=authenticity (and therefore reliability) claim is a regular one in this era. But so is the opposite. Consistency=authenticity (and therefore reliability) is also common. (These contradictory logics are also present in studies from the 1990s and 2000s, as I discovered in my dissertation.)
Overall, the goal posts for when and how to use literary texts as American English variation data shift a lot in this era. More out-takes soon!