Should Translators Proofread Documents Translated by a Machine?

It is becoming more common for agencies and freelance translators to get requests for proofreading translations done with a machine translation tool. As we’ve discussed before, these are tools such as Google Translate—the most widely used machine translation tool available—which use algorithms to translate text that you put into them. These tools have many shortcomings, some of which we have discussed in this blog, which can be problematic for anyone seeking a professional and accurate translation service.

Because there are some very common and obvious mistakes which these tools tend to make when producing the output, you can easily spot text that has been translated by a machine. Moreover, anyone can verify whether one of these tools has been used by simply copying a paragraph of the source text and translating it to the target language in Google Translate. But in the event that someone does decide to use a machine translation tool, they’ll have to be prepared for extensive editing prior to submitting the work.

And for those who have tried this method, thinking it would save some time, it quickly becomes clear that editing machine-translated text actually uses significantly more time than just translating the document from scratch. With the lure of a potentially time-saving aid, it might just be a process that each person needs to go through, and an option that understandably would tempt a client. But in the end, a professionally-done translation is just a much better option.

Are you a translator? Click on the image below and tell us if you would accept to be paid a proofreading or editing fee to correct this translation made with Google Translator.

 

The Fallibility of the ‘100% Match’ and How it Affects Agency Rates

Continuing with the topic of Trados, or translation memory software, the ‘100% matches’ moniker warrants a closer look. The term can be both deceptive and misapplied, for reasons including the true accuracy of the matches and the way they affect translation rates, respectively. The term applies to any segment of source text that is an exact match with an already-translated segment stored in the software’s database. Different memory systems can achieve the 100% match to different degrees, but even the most effective ones can still suffer misapplication.

Take, for instance, the source text segment “cozy and modern.” If the software’s first encounter with this phrase involves the description of an apartment, it would be stored in the memory as “acogedor, moderno.” The problem arises when this is applied to the same phrase in reference to, say, a kitchen or some other feminine noun. The source language would appear exactly the same, as English does not use gender, but the exact match application would obviously be incorrect in Spanish.

And this is only one example. There are other instances in which software programs cannot necessarily account for the intricacies of language in their “exact” translations. Other examples include verbs that change depending on plural/singular or wider context differences that the software may not pick up on in the words immediately surrounding the segment. All together, these imperfect applications complicate the way agencies incorporate the ‘100% match’ moniker into their rates.

Since exact matches are technically translated by the software and not the human translator, some agencies will not pay for these segments. But the examples detailed above illustrate how even these segments must be proofread for accuracy and are far from fool-proof despite being ‘100% matches’. Even when agencies choose not to pay for these, they should at least have someone on hand to go through and proofread exact match segments. And if it’s the client who doesn’t want to pay for these translated segments, a brief explanation about the software’s imperfect output might be enough to justify a separate editing fee at a minimum.

Should translators be paid by the hour or per word?

It is far more common for translators to quote a per-word rate to their clients than an hourly rate. Yet hourly rates still persist in some particular cases. There are many reasons why per-word pricing is the norm and preferred by both clients as well as translators and translation agencies. Chief among these is that clients can better calculate costs. It’s much more difficult to get an idea of how much a translation will cost when all you know is an hourly rate, and not necessarily how long it will take to finish. And it’s particularly helpful when comparing the rates of one translator to another, when it can be difficult to accurately gauge how quickly each would complete projects relative to their rates.

Per-word rates also provide an incentive for translators to work quickly and efficiently, which is also better for the client. But it benefits the translator, too, as they can ultimately be more productive and earn more with their work. It may also encourage the use of tools to help increase efficiency. Yet despite the benefits, there is still another side to the coin for translators. Some documents may take much longer to complete, due to legibility issues or a higher difficulty level of the content. Charging by the word, in these cases, can benefit the translator less as their per-word rate undoubtedly would have taken into account how quickly they can translate. In these cases, some people choose to include a rate specifically for editing which is per-hour rather than per-word.

One thing that newbie translators, or those branching out to do independent work, should remember is that there is a difference between source word and target word rates. The former refers to a rate based on the number of words in the original document, while the latter refers to the number of translated words upon completion. Different agencies and freelancers may choose to go with one or the other. But it is ultimately more convenient for the client to be quoted a per-source word rate. Why? It lets them know exactly what the cost will be before you even get started.

The final thing to take into consideration when deciding how to charge your clients is whether to go with a per-page rate. It is not uncommon to receive a fax (even these days) or a scanned version of a document that cannot be easily converted into text. This kind of rate is less common than the more popular per-word rates or hourly rate. Which rate works for you will depend on the type of work that you typically receive.

However, charging per page is different than the page rate that some agencies and freelancers quote, which often is a way of referring to a set amount of words (for example, 200 words per page.) So it’s always important to be clear what you mean by “per page” (physical page or predetermined number or words) and “per word” (source word or target word) when negotiating rates.

Machine Translation vs. Translation Memory Software: Different Methods, Different Results

Most people are familiar with machine translation tools such a Google Translator and Babelfish, for example. They seek to replace the human translator by taking rules that were previously established and put into the software and applying them to new phrases. These rules cover everything from syntax to grammar and individual vocabulary units.

But these tools are not alone in challenging the traditional human translator. There is also a group of tools called Translation Memory Software or Computer-Assisted Translation. These differ from machine translators in a number of ways. One primary difference is that they do not try to replace human translators, but rather to automate part of the process—a process that ultimately is finished by an individual person. And although machine translators usually require human editing—sometimes extensive editing—this is rather the result of imperfect algorithms and the development-in-process of a relatively new technology.

Translation memory software works by taking words and phrases that the human translator has encountered and translated, and saves them in a database for future use. When the same or similar phrases are encountered again, the translation memory software applies the previously logged translation. It may be an exact match, a “fuzzy” match or, in the case of new words and phrases, no match. The human translator would then go through and review the translation as a final edit.

The benefit of this type of tool is that the more you use it, the larger its database of translated words and phrases gets—essentially a sort of reference built and used by the translator him/herself. This can decrease the time used on translations while increasing consistency. Traditional machine translation tools, on the other hand, rely on developers to gradually increase and improve their algorithms and references, somewhat less reliable than a database saved from a professional translator’s previous work.

So while both types of translation tools can help increase time and cost efficiency, they both function in different ways and, naturally, can produce very different results. The question of whether each tool adds to or detracts from the translation industry as a whole should take into consideration the respective role that each plays.

Do children really learn languages faster than adults?

Traditional knowledge has been that children are better at learning languages than adults, along with a whole host of other abilities like playing a musical instrument. Heavyweights like Noam Chomsky, professor of linguistics at the Massachusetts Institute of Technology, have supported this assumption with theories regarding the critical period of human maturation, among others. Essentially, the argument states that there is a particular period that is prime for learning skills such as language acquisition, and any time before or after that is less than prime, even much less so.

For some of us, it’s nearly enough to discourage lifelong learning. There are several explanations which support this argument, among them that very young children may learn a language by forming associations rather than by mental translations, and that young brains are just more impressionable than older ones—a sort of version of the blank slate idea.

But it turns out that this may only be part of the story. Children indeed do learn more naturally through word associations, in particular if they learn two or even three languages simultaneously without the option of translating one into another. But it can be argued that adults simply possess a different set of skills that may be no less useful in picking up a new language, albeit skills which utilize different methods.

Free image courtesy of FreeDigitalPhotos.net

For example, a string of experiments presented groups of children, 12 and 13-year olds, and adults noun-verb pairs that were pronounced and spelled differently according to whether they corresponded to animate or inanimate objects. While none of the groups were given a description of the rule or guidance on how to use it, the adults fared much better in identifying the rule and applying their knowledge. The group of 12 and 13-year olds was nearly as good at identifying the rule and applying it, while the group of children struggled with it. These particular experiments highlighted the theory that adults may be better at identifying patterns and applying their knowledge than young children—something which supports the argument that children do not learn better than adults, but that they simply possess different strengths.

Proficiency often comes down to just how frequently or how long an individual is able to practice their new language, as well as whether they receive guidance via timely corrections of their mistakes. While children who continue to learn a new language face ample opportunity to practice it over many years, generally with teachers all too willing to correct them, adults can’t necessarily replicate that learning environment. Even with full immersion in a new place where only the new language is spoken, people are often hesitant to correct an adult. Ultimately, it comes down to just finding an effective learning method according to how the individual learns—whatever their age may be—and ensuring that there is sufficient opportunity to continue refining the skill. That last factor can make all the difference between a good command of a language and proficiency.

Social Media, Latinos, and the New Marketing Environment

As the marketing atmosphere changes and evolves faster than ever with new technological developments and new ways for companies to connect with their customers, we are seeing more companies reach out to their Spanish-language audiences. One way that they are doing this is by translating their web pages into Spanish. But they’re also going further than simply providing information to the Latin American and Caribbean markets in the Spanish language.

Marketers tend to follow media use among groups very closely in order to know where they need to be marketing their products, and how they need to be marketing them. So it’s natural that they have taken note recently of a marked increase in social media use among Latino populations in the U.S. as well as throughout Latin America itself. While some social media sites that are obscure in the U.S. have a wider audience in the Latin American region, like High 5, the most popular site globally—Facebook—has become far more popular among Spanish speakers just in the past year or so.

Even a disappointing IPO earlier this year has not detained the growth that Facebook is currently experiencing in the Latino market, nor has it watered down the interest that companies have in reaching its user base. The social media analytics company, Socialbakers, published a new infographic a few months ago which shows that this user base has increased by 47% over the past year, reaching 168 million active monthly users in the region. In a word, it’s transforming the way products are marketed to Latinos.

 

Source: Socialbakers

And as this population becomes more and more the focus of companies with an international or regional reach, various kinds of information will increasingly be available in the Spanish language. Now, it is not only that Google and Facebook are available in Spanish, but the content which they disperse is, too—in the form of advertisements, web pages, videos and more. As a result, the companies that will most successfully manage this new environment and use it to their benefit, will be the ones that can seamlessly go from an English-speaking audience to a Spanish-speaking one, and back.

As with so many other professions, localization professionals and Spanish translators may very well find their new home in marketing and social media in the months and years ahead.

The Importance of Providing Written Translations of Company Policies

While US labor laws require that employers provide translations of certain kinds of information regarding company policies to Spanish-speaking employees, the laws which are currently on the books do not necessarily cover all of the information that these employees require. As a result, it is not unheard-of for employees with limited English abilities to be unaware of their rights as workers, or unable to exercise them to their fullest capacity.

The most intuitive area that affects these workers are policies regarding anti-discrimination. Unfortunately, discrimination in the workplace is still something which occurs and which is a topic of concern among labor advocates. And while companies usually provide some form of translation of their policies, as required by law, the information is sometimes incomplete. Further, the form which those translations take can also complicate matters.

A relevant case which reached the federal court in Colorado dealt with a sexual harassment complaint which a Spanish-speaking employee brought against some co-workers. Although the company had provided a Spanish-language video explaining some of the information in the companies pertinent policy, the actual policy itself—with complete information—was never provided to the employee in Spanish. In addition, the interpreter that was available at the work premises to foster communication between Spanish-speaking and English-speaking employees—including between the employee who filed the lawsuit and those she was accusing—was implicated in the complaint. As a result, the employee felt that she could not resolve the issue directly with the parties involved.

In situations such as this, it behooves an employer to provide written translations of company policies in their entirety to workers who speak Spanish. It may be the case that if these translations were provided, beyond what the law requires, a costly escalation of the case could be avoided—a benefit to everyone involved, including the company itself. Even if a similar situation never arises, the company can rest assured that they have taken sufficient measures to anticipate any possible issues, and know that they have covered their bases. Written translations also offer the additional benefit of being evidence that Spanish-speaking employees have indeed been informed of company policies and their individual rights.