T. R. Girill Society for Technical Communication/Lawrence Livermore National Lab. firstname.lastname@example.org Technical Writing: Fact, Fiction, and Writing Skills Technical writing calls for different "skills and knowledge" than creative (fiction) writing, and it is certainly evaluated differently. Hence, I always argue that technical writing is best learned with the help of teaching techniques that reflect these differences. The long explanation for this appears in "Technical Writing Explained" (http://www.ebstc.org/TechLit/handbook/twe2.html), one section of "Technical Writing in Science Class: The Handbook" (http://www.ebstc.org/TechLit/handbook/handbooktoc.html). But in July, 2010, a brief study on "Distinguishing Fact from Fiction" appeared that offers a dramatic "short explanation" for why technical writing skills deserve their own specific development for high-school students. Word Use Patterns Diverge J. T. Stevanak and Lincoln D. Carr compared the features of 200-word samples from twentieth-century novels (fiction) and news items (nonfiction). (Stevanak and Carr, "Distinguishing Fact from Fiction: Pattern Recognition in Texts Using Complex Networks," Arxiv.org, 15 July 2010, online at http://www.arxiv.org/abs/1007.3254) They looked at these text chunks very abstractly, representing each as an "unweighted, undirected semantic network" or graph in which * each vertex is a word (completely IGNORING any of the usually important differences in conjugation, pluralization, or punctuation), and * each edge connects two vertices (words) whenever they appear within 4 words of each other in the sample, and * the degree of each vertex (word) is the number of edges connected to it: the more often a word occurs in a text sample the more (unique) neighbors it can have and the more edges in can accumulate (giving it a higher "degree"). Stevanak and Carr found by plotting scatter diagrams that while fiction and nonfiction samples certainly overlapped, the degree distribution of words (vertices) for novels (fiction) and news items (nonfiction) differed in statistically significant ways. Novels tend to have a degree distribution that reflects relatively few different words each used relatively often. News items tend to have a degree distribution that reflects relatively many different words each used relatively seldom. Looks Different Even at a Distance Even at this rather extreme level of abstraction, which ignores all the usual "literary" cues about style and composition, these scientists found that they could very reliably predict whether a blind sample of 200 words was fiction or nonfiction. Their predictions reached 74% accuracy for novels and 69% for news items (when the "word distance" was 4), just by looking at the degree distribution of a sample's words. This difference between fiction and nonfiction is quite basic since it reflects (only) the different frequencies with which words are used, and hence clustered as neighbors, in the two kinds of text. While this may seem an esoteric difference at first, it meshes nicely with other important linguistic features. Beginning foreign-language students usually find stories easier to read than newspapers, for example, because news items (nonfiction) have more diverse vocabulary (and hence less reuse of words) than fiction stories. Likewise, Dr. Seuss famously composed "Green Eggs and Ham" using only 50 unique words (see http://en.wikipedia.org/ wiki/Green_Eggs_and_Ham). This is a story for children to be sure, but it still shows how much meaningful fiction one can construct with very little word diversity. Implications for Teaching What does this imply for teaching your students effective writing? In light of this fiction/nonfiction difference in the degree distribution of their words the need to build separate skills to draft and revise nonfiction text should not be surprising. Practice reading and writing fiction, with its own characteristic word-use pattern, often leaves students unprepared to effectively read and write the prose common in science and engineering-- technical nonfiction: * Technical text structure usually involves hierarchical analysis rather than progressive disclosure. * The usual goal is to help readers act on the text content ("far transfer" problem solving) rather than to entertain or inspire them. * The standards of success involve usability much more than self-expression. * The skill set is disciplined much more by overt disclosure (of both structure and meaning) than by subtlety or indirection. No one would--or should--use technical writing guidelines like those at http://www.ebstc.org/TechLit/analysis0.html http://www.ebstc.org/TechLit/analysisGd.html to craft fiction. But to design effective nonfiction they are crucial.