3 weeks ago
November 13, 2020
Unpacking the Contentious Matter of Testing Freelance Translators
Tests for freelance language professionals get a bad rap, some of it well deserved. The least scrupulous language service providers (LSPs) have been known to save money by disguising their clients’ projects as unpaid “tests” for translators. But some professional translators may also object to tests, paid or unpaid, on principle, saying that their credentials speak for themselves.
Both LSPs and freelancers have valid arguments, and the truth is that tests, like any kind of job interview, should be a two-way street. LSPs want some kind of guarantee that suppliers are able to handle the work, and candidates should be able to get an idea of the actual work they would be assigned.
Deciding to integrate testing into the recruitment process is not a quick fix; rather, it is an investment in the pool of freelancers who project managers feel they can entrust with projects.
When considering — and then designing — a test, LSPs should think about their goals, and why a test is necessary. In some cases, LSPs and freelancers might be better served by implementing a short “probation” period, during which a translator works on a set number of (paid) projects, and will enter the LSP’s pool of vetted and trusted freelancers depending on their performance.
SlatorCon Remote returns on December 3, 2020, featuring the best of our proprietary research and network of language industry leaders.
Of course, a test may be necessary as part of a proposal to a high-profile client. In other instances, most freelancers might struggle to follow the meticulous formatting rules required for certain clients, and someone working on a trial basis might submit work that requires too much editing to meet a client’s deadline. Perhaps typical certifications are not available to candidates working in a hard-to-source language, and the LSP needs some way to confirm that candidates are qualified.
A Game Plan for Testing
The best strategy is to test to as specific a task as possible. A generic test carries the potential pitfall of a candidate later claiming the passing grade qualifies them to translate any and all content.
The best strategy is to test to as specific a task as possible
By contrast, a domain-specific test helps LSPs assess the ease with which candidates handle a specific vocabulary and concepts. (Vendor managers, also known as VMs, are often responsible for creating and maintaining a system in the vendor database that identifies which translators are qualified to handle specialized kinds of work, based on the tests they have passed.)
The test should also mimic the conditions of the work itself. If the task at hand is translating medical documents that include handwritten comments from physicians, the test should include handwritten text. Similarly, candidates should have access to the kind of resources freelancers typically have when working on these projects, such as style guides and clear instructions.
Who will create and grade the tests? In a classic catch-22, an LSP tests candidates to build a pool of freelancers to handle a specific kind of work; if an LSP lacks at least a few freelancers it can already trust, it will be difficult, if not impossible, to create a test in-house. The next best choice for an LSP truly starting from scratch might be working with another LSP that specializes in regional languages or a specific subject matter.
Language experts may charge per hour to create a test, depending on the complexity of the assignment. For graders, a flat rate of compensation per test graded is generally appropriate, with a possible slight discount for a high volume of tests to be graded in one go.
Candidates should have access to the kind of resources freelancers typically have when working on these projects, such as style guides and clear instructions
Ideally, VMs should work with trusted linguists to develop rubrics that establish what graders should be looking for when they review submissions. Categories might include translation quality, formatting, and the ability to follow instructions. How heavily each factor is weighed can depend on the ultimate purpose of the translations.
For example, if a client prioritizes quick turnarounds over perfectly idiomatic translations, graders can weigh these factors accordingly on a client-specific test. Graders should also be provided a forum where, if necessary, they can explain their reasoning beyond a numeric score, and possibly offer insight into whether the test seems too easy, too hard, or in need of tweaking.
Straight Talk About Tests
Regardless of the specifics of an LSP’s testing process, transparency goes a long way toward earning freelancers’ trust.
Starting from the first contact, recruiters should inform candidates that a test is required to work with the LSP (either in general or for certain kinds of work) so that uninterested freelancers can opt out, saving both sides time and energy.
Including estimates of the average time it should take a candidate to complete a task, as part of the test instructions, can give translators a target by which they can measure their own efficiency. Lastly, many translators have been burned by LSPs that dropped the ball in communications. Let candidates know where they stand in terms of a timeline for test submission, grading, and results.
Let candidates know where they stand in terms of a timeline for test submission, grading, and results
By all means, tests should come with a deadline, just as projects do. However, if there is a time limit (for example, candidates must finish the test within 48 hours of starting), best practice is to allow candidates to begin the test on their own schedule.
Just how much feedback should test-takers expect? It will vary across LSPs but, in general, VMs err on the side of “less is more.”
Collecting and sharing feedback on test submissions is time-consuming for both graders and VMs. For candidates the LSP has not worked with up until this point, it is unlikely to be worth the effort. Feedback can also open the LSP up to a barrage of protests from test-takers who feel they should not have been failed, or who want another chance. On the other hand, if a candidate already works with the LSP, providing feedback on a freelancer’s test can be an investment in an existing relationship.
Now about that controversial question of paid versus unpaid tests: Tests have an upfront cost to LSPs, which must pay specialists to create and then grade the tests. When it comes to paying test-takers, LSPs would do well to consider the time and effort they expect candidates to expend on a test that may or may not lead to future work, and calculate a flat rate of compensation per test accordingly.