An annotated excerpt from the forthcoming text by Elliot Bendoly and Sasha Clark; Taylor & Francis, 2017
Amidst a data ecosystem’s collectors, analysts and users, important meaning can be distorted or lost altogether simply because of differences, often subtle ones, in the data-dialects of these agents. Even visualization—often mistaken as a foolproof means for sharing insights universally—can fail if it doesn’t take into account the differences in language, context and other factors that can exist between creator and user. This raises an appreciation for a very import concept in contemporary data analytics: “Translation,” meaning the communication of something’s meaning through an equivalent representation. The term derives from the Latin for “a carrying across,” and the gulfs that need to be crossed can sometimes be daunting.
The concept of translation has a long history in the study of languages, with one of the most evocative symbols being that of the Rosetta Stone. When I visited the British Museum with my family in 2013, I couldn’t miss this famed artifact. Seeing it in person—knowing it was the real thing and not a replica—had a visceral effect on me.
The Rosetta Stone was rediscoverd by Pierre Bouchard, a French soldier under Napoleon’s command during the Egyptian campaign in 1799. It contained portions of passages written in Greek, Egyptian hieroglyphics and Egyptian Demotic. By the time of Napoleon, however, an understanding of the literal meaning of hieroglyphics and their composite passages that adorned monuments across Egypt had been lost to time. Egyptologist Jean-Francois Champollion, an expert in languages including Greek and Demotic, ultimately deciphered the hieroglyphics two decades later. To do so he leveraged the fact, or rather the assumption at the time, that the three passages on the stone each had the same meaning. The task was not simply one that involved determining which set of hieroglyphic symbols represented which words in Greek or Demotic, however; it required developing an understanding of how these “words” were comprised. Ultimately, Champollion determined that some of the symbols indeed represented sounds (phonetic signs), while others represented entire conceptual meanings in and of themselves (ideographic signs).
The ancient Egyptians’ choice to develop a written language that made simultaneous use of both simple and highly complex signs did later European scholars no favors, but it was an artifact of their context and history. It made sense to be able to succinctly capture commonly confronted concepts in unique images while allowing for a system of signs to permit the evolution of the written language. It was efficient within their system, and certainly not the only instance of such an effort by mankind—take, for example, the Japanese joint use of Kanji ideographs along with Hiragana and Katakana phonetics.
The world is full of both simple signs that serve as fundamental building blocks of visual expression and of richer signs that are vastly more sophisticated in the meaning they are intended to express. The trick for the user of these signs is in understanding the audience to which they are attempting to convey meaning. If the audience has a frame of reference equivalent to the communicator, issues of translation aren’t raised. If, on the other hand, the audience has a different background, the task needs to be approached in a more sensitive manner. A C-suite manager can be just as stymied by a report created by modern data analyst as a French soldier can be stumbling on pretty pictures carved into a stone.
In a text book project I am currently involved with (for which this blog post will serve as a de facto introduction) the hope is to fill an obvious need in this space—specifically, to provide a little guidance for those charged with translating complex concepts gleaned from large data sets and sophisticated analysis. The kind of guidance that can facilitate translation and offer the greatest chance of successfully driving practice.
Such necessary guidance looks beyond the simple presentation of stunning infographics and requires structured discussions of audience- and problem-focus in design, considering not just end consumer audiences but also intermediate users of data visualization in analytical development. It requires recognizing differences in need, bias and effective processes across contexts and stakeholders and rationalizing the effectiveness of lean or sophisticated representations, static versus dynamic interactive ones, solitary idioms versus dashboard systems of such, as well as how to effectively design these.
Critically, this and related endeavors must be approached with a fundamental aspiration to further advance the intelligent development of visual analytics—advancement that, like most visualization efforts, will not be the final word on development but rather what we and others working in this space can only hope will open the door for a range of novel applications in practice and follow-on advancement in design. Applications that will have a pivotal impact on responsible managerial decision making, and serve to further inspire critical discussions on best practices in data visualization design.