This paper examines the quality control variables available on CrowdFlower, a crowdsourced translation platform, and analyzes how each variable influences translation outcomes. The discussion covers contributor selection methods (both passive and active), the role of geography and language capability in ensuring accuracy, how pay and time constraints shape the talent pool, and the relationship between completion time and translator experience. The paper concludes that clients can achieve an optimal balance of cost, quality, and turnaround time by strategically adjusting CrowdFlower's available controls, while crowd members retain the ability to self-select assignments that suit their own preferences and capabilities.
CrowdFlower offers a number of different options for quality control. Variables include contributor settings — you can choose internal contributors, external contributors, or both. Contributors can be filtered by geography or language, which allows for greater granularity when translating into languages with many regional variants, such as French or Spanish. Contributor pools can also be filtered by capability level.
A company can use test questions to gauge the accuracy of the platform and to help identify contributors who are best suited for a given role. A company operating in a particular vertical could use test questions to find contributors who are especially strong at translating terminology specific to that industry.
Additional controls exist for time, pay, and related variables. A company might choose to pay less, but should do so understanding that this decision may reduce the overall quality of the human component of the translation. Paying more is likely to attract better talent.
The different options can influence results in a variety of ways. The accuracy of crowd workers is shaped by several factors. With respect to pay and time, quality should be directly related to how much time a contributor has to complete the work accurately, and how much the client is willing to pay for talent and effort. Workload may also factor into effective pay rates — translating an entire website in a single day will yield lower accuracy than translating a press release over the course of a week. These variables represent passive selection: the choice of settings narrows the pool of potential contributors in ways that indirectly influence quality.
Test questions and contributor selection variables help a company obtain the results it wants. When the client has more control over who contributes, the client has a greater ability to reach the specific person or people best suited for the task. The different contributor variables thus represent active selection, in which the client directly targets a particular type of contributor and selects for that profile. A company needing a translation into Spanish for the Mexican market would, at minimum, set a geographic filter — otherwise it risks receiving work from a Spanish or Argentine translator whose regional dialect may reduce accuracy. As noted, there are also use cases in which a company in a niche market seeks out any contributor who understands that market's specialized terminology and knows how that terminology translates into the target language.
"How variables shape translation turnaround speed"
"How pay is set and influenced by project factors"
"Balancing cost, time, and quality for clients"
You’re 45% through this paper. Sign up to read the remaining 3 sections.
Sign Up Now — Instant Access Already a member? Log inAlways verify citation format against your institution’s current style guide requirements.