This content has been archived. It may no longer be relevant
With all of the technical translation projects we work on at TrueLanguage, we know quite a bit about projects involving machines. This may include instructional manuals for operating the machines, schematics and parts lists for assembling machines, certificates and documentation for exporting machines, or software to be read and interpreted by machines.
Translation for machines is one thing. What about translation by machines?
We have mentioned machine translation several times in previous writings, and always closed our thoughts with the opinion that while machine translation technology is making rapid advances, it still hasn’t reached the point of replacing human translations. Yet it is an undeniable force in the language services industry, on both sides of the provider/client relationship. A language service provider has to keep abreast of developing technology and industry trends, if it wants to remain competitive. And translation buyers in the business sphere will be open to any language services option that is kinder to the bottom line. Can machine translation get the job done for less, without compromising the quality of a translation that bears the human touch?
There are advantages to machine translation. For one, the upfront costs are apt to be lower. Translation software doesn’t charge by the hour, after all. Secondly, removing human translators from the process also removes human error from the equation. Lastly, and significantly, machine translation allows for a degree of confidentiality that you can never be fully assured of when working with human resources. For keeping the content of your sensitive documents airtight, non-disclosure agreements and security clearances can’t compare to keeping the documents away from human eyes. Someone will have to look at your files at some point, but the fewer pairs of eyes, the more secure your documents.
And the downsides?
The cost of machine translation is often lower in theory than it turns out to be in practice. Documents translated by machine will still require at least one pass by a human proofreader. Depending on the quality of the machine’s work, billable time in proofreading may eat up the savings incurred by not using a human translator. You might be able to curtail the proofreading if your translations aren’t destined for an end-reader; some translations are only meant to sit in a file drawer, helping your business fulfill its documentation requirements, without ever being read. But if your document will have readers, it needs proofreaders! Another way to cut down on the need for proofreading – use the latest in machine translation technology. Yet as you might expect, cutting-edge technology doesn’t come cheap. Free automated translation services, of the type you can access through a web browser will not do the job. Language service providers who offer the best in automated translation must do so at considerable expense, and those costs may be passed on to you.
A second downside is bound up with the first – translation software on its own can’t operate with the combination of accuracy, consistency, and linguistic fluidity that a human translation team brings to the table. Roughly broken down, machine translation systems will fall under one of two headings: rules-based and statistical. A rules-based system accepts all of a language’s prescribed grammar rules, definitions, spelling, etc., and applies them to the text it is given. As you might expect, this results in a translation that’s high on consistency (playing by the rules), but prone to stilted text, possibly confusing phrasing, and inflexibility. Conversely, a statistical system learns a language by assimilating the translations that occur within it, getting better and better at translating as it’s used to translate. This system uses current, human language to develop itself, which means heightened fluency, and the potential for compromised consistency. If you want your translations to be consistent and fluent, work with people.
In the end, machine translation could be the right option for you… as long as you go in with the knowledge that it’s still very much a developing field, and that it’s not possible to be certain of high-quality translations if you cut out the human element. As long as you have human eyes reading your translations, you’ll need to have human hands involved in making them!
