Do Specifications Harm Translators?

The recently updated ASTM standard F2575 - 14 (Standard Guide for Quality Assurance in Translation) defines the content of “translation specifications”. These specifications provide a way for content creators (translation requestors) and translation providers (language service providers and individual translators) to agree on what services should be provided and what the expectations of translators are. The idea behind specifications is to eliminate the inherent uncertainty the arises when a requestor asks for “translation” or “high-quality translation” by unpacking what exactly is needed and providing a way for both parties to verify that expectations are met.

The current situation in the translation industry (without specifications) is not so different from what would happen if a building contractor were to place an order for “some good screws” without specifying what sort of screw is needed (drywall, wood, or metal), what material the screws should be made from, where they will be used (indoors, outdoors, underwater, etc.), how long they will be, what sort of thread pitch they will have, what sort of head they will have (flat, plus, hex, square, etc.), etc. If the building contractor were to ask just for “good screws”, he could not complain if he got a sort of screw that he could not use.

This notion of specifications seems obvious when we consider a case like screws, and it is hard to think of any industry where goods or services would be ordered without some notion of what those services were. Yet in translation, that is exactly what we do. Customers want “translations” and do not want to be specific about what they want (other than they want it good, cheap, and fast).

For me, specifications seem like an obvious good thing. They would define what the translator or LSP agrees to perform and how to verify that it has been delivered. I cannot imagine running a business where I would never know if what I was delivering meets expectations or not. I would insist on clear specifications for my own protection. In addition, specifications would help my provide a fair price that meets the customer’s needs. Without specifications, I would be forced to guess what is needed. If the client wants something “quick and dirty” I might bid too high by assuming that they want a fully polished translation with no tolerance for any error. If the client wants a premium product and I assume something else, I might bid too low and then end up in trouble by securing a job I could not actually provide at the agreed-upon price.

Without specifications the price signals needed to find the best provider and for providers to make the best offers are distorted by the fundamental uncertainty that comes from undefined requests. “Translation” is not enough of a description to set a price. Adding even information such as domain and purpose helps, but it is only when all parties have enough information to anticipate eventualities and understand what is requested that fair prices can be set and that unscrupulous LSPs that compete solely on unrealistically low prices can be excluded.

I would think that the benefits of transparency and openness that come with using standard specifications would be apparent to all, but there is considerable opposition from some sections of the translation industry, and not just from the bottom-feeders who profit by mystifying translation and offering junk services under the rubric of “high-quality” or “top quality” translation.

An objection to specifications

The objection to specifications I have now heard from a number of prominent translators is the following: buyers of translation may say they want something other than the best translation (e.g., they are happy with information-only translation or need something very quickly and so can tolerate some errors), but in fact they will demand the best quality. The worry about specifications then is that buyers will use “lower quality” specifications to bid down prices and then, when the translator fails to meet an unspoken and unwritten higher specification, use that fact to penalize the translator financially by withholding some portion of the payment.

It would be naïve to assume that this scenario would not happen since with highly informal specifications it already does. Accordingly there is an inherent self-preservation bias for translators and LSPs to push themselves to deliver the best quality regardless of stated expectation and price. As my colleague Geoff Koby put it in discussion, there are minimum standards for what constitutes “translation” and no translator in his/her right mind would deviate too far from those standards, no matter what the client says.

So do specifications actually provide cover for bad actors to harm translators?

Specifications as a protection for translators?

My response to this criticism is that this problem already exists, so specifications would not cause it, even if some unscrupulous actors might use specifications to provide justification for withholding payments or negotiating discounts. If anything, this objection is to the misuse of specifications, not to specifications themselves. If translation were really a self-defining object that allows no variation, I could see validity of the objection, but translation is a variable activity that requires specification, especially when we break out of narrow and parochial boxes that come from working in specific environments, such as translating only medical manuals, literature, corporate financial reports, news, or any other sort of text. If our world is focused on one segment it certainly can seem to us that translation specifications are obvious and do not need to be stated, because we know what the expectations are in our corner of the world. But as translation grows and expands, it enters areas where expectations are unknown or not yet stable. The industry is in a period of tremendous flux and established ways of business are being upended and changed. To assume that we know what is expected may work most of the time, but it can also land us in a lot of hot water if we assume wrongly.

Specifications are intended to be bilateral protection for both translators and their customers. By giving concrete terms of reference, referred to in the contract, the requestor cannot then demand that the provider meet another standard of service. (Admittedly, contractual obligations can be broken, but an appropriate contractual obligation to neutral third-party review in the case of dispute can help address this problem.) It also protects the requestor from bottom-feeding providers since the specifications can define the quality assurance steps to be undertaken and can pin payment on successful completion.

Would translators ever be asked to provide less than the best quality?

This issue also assumes that professional translators would be asked to “dumb down” their work and deliver less than their best. I’m not clear that this would actually happen outside of unusual situations such as time-critical translations that do not allow normal time for revision and quality assurance. Such situations may not be totally uncommon, but they already define an exception to the rule.

Where lower specifications are more likely to apply is in situations where MT or other technologies are used and where the requestor is satisfied with results that would not meet the standards of human translators. Lower specifications might also apply in cases of non-professional crowdsourced translations or “para-translations” conducted by students or other non-professionals.

But for most professional human translation, specifications will simply provide additional detail needed to ensure customer satisfaction, not provide a route to lower quality. Because specifications can be attached to contracts, they will filter out the lower-quality providers that distort price signals and they will actually help fight the dreaded specter of commoditization in the industry that comes when providers compete solely on price without regard to their offerings.

The industry benefit of specifications

With proper specifications requestors can also move beyond filtering out bad actors to finding the resources that actually specialize in their requirements and that deliver the services they specify. Since most requestors have little understanding of translation, the task of actually arriving at a set of specifications will have a salutary client-education impact by helping make the client aware of what is involved in translation. And for those clients who don’t care to know more, providers can post their default specifications and refer to them in standard contracts so that, in the event of a dispute, the provider can point to a positive statement of services to be provided. (For example, if basic translation services do not include DTP, a clear set of specifications with a statement about what is not included in those services can help protect both parties from confusion and disappointment.)

Specifications will not solve all problems, but the current status quo with service providers who offer murky and unclear services, clients who do not understand what they are buying or what to look at beyond price, and general confusion benefits nobody except for the bottom-feeding providers the offer garbage service at a low price. If it does turn out that specifications are abused, the industry will need to improve contracts and business practices to address this problem.

But overall, having clear and unambiguous information about project requirements and expectations will benefit all parties.