If you work in the language industry and you haven’t heard of crowdsourcing, well, you may want to think about relocating from under that rock! Crowdsourcing is everywhere. Use Wikipedia? You’re a crowdsourcing consumer. Purchase photos for a project from iStockPhoto? You guessed it, crowdsourced. And of course, consider the dominant language services model: a global network of skilled linguists contracted for specific language pairs, for specific assignments. That’s just a professional variation of the same concept. That’s right, crowdsourcing is in our industry’s DNA, so to speak. Bottom line: crowdsourcing in our industry is here to stay. What role is crowdsourcing going to play in the language industry in the very near future? Indeed, is the crowdsourced future already here? And, how can we in the industry use crowdsourcing to improve our ability to serve our clients?
First, a definition, and what better place to look for one than the ultimate crowdsourced tool, Wikipedia: “Crowdsourcing is the act of outsourcing tasks, …to [a] … large group of people or community (a “crowd”), through an open call.”
Next, let’s look at the basics. Who doesn’t remember the project triangle? I was taught by hard experience that there are always three variables – quality, time and cost – that define the project, no matter what the size or scale. Or, not to put too fine a point on it – good, fast and cheap. You can only choose two – the triangle concept reflects the idea that these three values are always interrelated – that it is not possible, or in fact, desirable, to try to optimize all three. One will always suffer.
My initial impression is that the crowdsourcing model as currently used in the language industry is focused on time and cost. Or fast and cheap, if you prefer. The orphan variable is quality. However, it is early days yet and the project triangle has only just begun to be stress-tested by the demands of the market, the power of the tools and the creativity of the leaders in this dynamic space.
What can we do to bolster the orphan? As language industry professionals, we bring a wide range of practices and tools to the art, and okay, science, of translation.
Let me suggest a few practices that I believe can begin to address the need to support what will likely be an iterative process.
Fundamental to the success of any project, regardless of type and industry, is the identification of roles, responsibilities, actions, timelines, milestones, goals and objectives. Particularly in a crowdsourced project context, planning is something best done early and often. The approach taken by Amazon’s Mechanical Turk, for example, to organize and deconstruct project tasks, into so-called Human Intelligence Tasks or HITs, reflects the need for a rigorous taxonomy structure to help organize the work-process. Such deconstruction of projects into small, unique tasks is a first and exciting step towards a systemic approach that engages networked, crowdsourced resources.
A strict application of standards and assessments that ensure skills and subject matter expertise are assigned to the project, is always a critical aspect of any language project. Today there are three types of translation available: professional (i.e., global networks of professional linguists), amateur (e.g. fans translating a product) and machine (specialized software to translate text). Definitions of quality are largely dependent on which type of translation provider is selected for a given project. Going forward, we will see the increased integration of these three types of translation, requiring even greater attention to standards and assessments to ensure quality.
The continued and refined application of translation memory (TM), as well as glossary and terminology management, has a significant contribution to make to the successful crowdsourced language project. The evolution of TM to a cloud-based, crowdsourced phenomena in its own right bears close attention in this context.
The crown jewel of the smart crowdsourced language solution. The ability of language providers to adapt and evolve legacy QA procedures to align with the dynamic nature of crowdsourced content will be key. And the more offset there is between qualification decisions (e.g. pros v fans) and the quality requirements (e.g. high, medium, etc.), the more critical this element becomes.
In conclusion, I think that crowdsourcing’s application in the language industry is here to stay. No doubt, it will evolve from what it is today, driven by the demands of the market for further integration and ease-of-use and the developers who work tirelessly to harness the collective capabilities of the social, multilingual and globally networked humans of the 21st century.