Inter-professional Dialogue: Renato Beninatto

(To get an idea of this article in Spanish, using the Google Machine Translation tool, click here.)

My virtual café chat this time took me all the way to Lowell (Massachusetts), in the USA, to talk to Renato Beninatto, who, at the time we chatted, was still co-founder and an active member of Common Sense Advisory (CSA), “a research and consulting firm committed to improving the quality of international business and the efficiency of the online and offline operations that support it.”

I was particularly interested, among other ideas they have published, in learning more about this company’s/group’s comparison between the traditional Gutenberg-based “TEP” (from “Translate, Edit, Proofread”) translation quality concept and their proposed “PCTP” (from “Plan, Coordinate, Translate and Publish”) translation quality concept, as described in 2007 in one of CSA’s blog entries.

Although in our chat Renato basically covered the same areas as in the above mentioned article, what I tried to do differently was to “challenge” some of these ideas with my own comments or questions, based on my own experience as a freelance translator.

But in the end we (well, rather Renato) talked about a lot of things, like translation education, prices and globalization, etc., which are among the many aspects that I believe any freelance translator should seriously consider when facing the translation market and translation clients.

The following is an edited version/transcription of our chat through Skype. It turned out rather long, but I honestly think is worth a full read, because Renato managed to analyze many of the issues with a style that is both thorough and informal, a cross between an informal conversation and a presentation.

——————————————-

So Ivette, I understand you are interested in discussing my recurring topic about the Gutenbergian approach to translation quality. OK, let me start with a little bit of background [information], so that you understand where I come from.

I started my life as a freelance translator. I am originally from Brazil and I translated into Portuguese.

I actually started translating for movies, and eventually, when the IT revolution started happening, I got involved with software localization, and over time I built a translation company that had 65 employees. We worked in specialized, very large multi-million-word projects, for companies like SAP, Hewlett-Packard or Oracle and others.

Mostly into Portuguese from Brazil and Spanish. I had an office in Brazil and an office in Argentina. So I’ve had the opportunity to look at productivity from the translator’s side and the manager’s side, and also over time I ended up working as an executive in large companies like Berlitz, where I was working with very large projects.

So the background of all this starts with the frustration [caused by] inefficiency, but also with some successful projects in the past where we had multiple translators working in very large projects. Another data point I want to add to this is that one of my largest clients in Brazil at one point was Shell Oil, and I was very closely involved, from the translation side, with their ISO 9000 Certification, their first one, back in the late 80’s/early 90’s.

During this ISO Certification process, my organization had a lot of training on Total Quality Management and quality processes and procedures in the manufacturing space and applied to services also. So I have always been fascinated by and involved with the quality concept, and the efficiency concept. One of the things that you learn being a small business owner for a while is that different projects have different requirements.

The challenge that you have when you start talking in terms of generalities or generalizations in the language service industry is that we talk in terms of absolutes, of rules of quality becoming something that is definable in very narrow terms, and that a process or the same process will work for every type of project. I’ve been involved with industry associations for over 20 years and the thing that frustrates me the most is that what dominates the discussion and the debate in any type of industry organization is mediocrity.

Essentially how do we make everybody do the same, and not really how do we make everybody do the best job. I am interested in the best job, and the best job or how I would define the best job is delivering the best quality according to the client’s expectations, within the deadline, with a profit for the service provider, and at a fair price for the end-client. So paradise, everybody is happy! [Laughs...]

So let’s move on to the standard issue, the standard question. I did a series of presentations in several conferences that were titled “Quality doesn’t matter.” What I mean by that is that quality, from a sales perspective, and the way it is talked about in the translation industry, is totally irrelevant, it doesn’t mean anything.

So normally I asked the audience how many people know what TEP is, and how many apply TEP, and how many believe in TEP. I realized that “Translation, Editing and Proofing” (TEP) is like a dogma in the language services industry. People + organizations + companies + schools, they all learn this concept that a translator does the job, somebody else reviews the job (another translator reviews the job), then you do a final check, and then you have quality.

The problem with that organization and that structure is that it focuses on the process and not on the players, on the elements in the process.

What I see, based on the economic structure of the translation business, is that you have a translator, who gets a certain amount for the job he/she does, then you have an editor who gets 20% of that amount to review the work, and then the internal organization in charge of the process that does (when they do) the final check of the files before they are delivered to the client.

The assumption is that by adding processes, by adding steps to the process, you are going to improve the quality. This philosophy is based on the idea that quality is “catching errors”. Therefore, the job of the reviewer is to catch mistakes and errors that the translator made. The job of proofreading is to catch errors that the editor didn’t see, right? So it’s all a chain based on the error or the mistake, but that’s not quality.

Quality is doing it right the first time.

So the proposition of process that we talk about here in CSA, which, by-the-way, is nothing new, as it is something that I adopted 20 years ago, it’s just that it has become easier by the technology and so this topic came back in force about two years ago. The concept is “you do it right the first time.” You provide all the tools, all the infrastructure and all the information for the professional to do it right.

That is, instead of having a reviewer to catch errors, you have an infrastructure to prevent mistakes. Instead of having error correction, you start having error prevention. That’s how quality and security work in the manufacturing industry. You don’t correct defective parts, you create a process that avoids the manufacturing of defective parts.

The idea here is that you start with, first of all, some general assumptions in the concept of quality. I am not sure I stated this clearly, but if we look at the industry standards, there is a dogma that states that:

* The fewer translators you have involved in a project, the better quality you are going to have, because it is more consistent.

* You need multiple eyes to look at the translation in order to provide better quality.

* More revisions improve the quality because you catch more mistakes.

I want to blow all this concept away, because that was true when you had typewriters and telephone lines. Or when you needed everyone to be working in the same room because you wanted to avoid typos and problems like that.

Today, the infrastructure that exists allows me to state, very clearly, that I can have a better translation with 10 translators, instead of 2. I can have that translation faster, with better quality and more consistency. So that sounds like a heresy, right?

[IVETTE: Sure, but the problem is that sometimes the work that is needed to avoid those mistakes, to carry out that prevention you described, is not taken care of.]

Exactly, here is the problem. The companies that have implemented what I propose understand that this is a complete cultural shift. You cannot start tomorrow and just say “Oh, Renato says that quality doesn’t matter, so we’re going to eliminate the review process.” Absolutely not.

Let me give you an example. As a freelance translator, there are some subjects where you are usually very good at, in which you are very productive, which allow you to do even 8000 words a day, and which don’t make you feel so tired, because you don’t have to look at the dictionary every five minutes.

But there may be other topics where you do 500 words and you are already exhausted and even feel that you are the dumbest person in the world. I’m telling this from my own experience.

So my point here is to bring some concepts into this idea of cultural shift. What I propose is to work with that more productive side of the translator which allows him/her to make 8000 words a day because he/she knows very well the subject matter. So if the translator had to work on a manual about one of his/her strongest subjects, but there is a section of the manual that deals with a subject with which he/she is not so familiar (for example, Medicine), the translator would need help only for that section.

Therefore, the job of the coordinator in charge of the project would be to assure the translator support in those areas where the translator requires help. In the previous example, the section about Medicine would be carefully reviewed by a doctor, in order to guarantee the quality of the translation in this specific section. The rest of the translation could then be perfectly taken care of by the translator him/herself, as he/she is a professional who knows how to deliver final files that have been reviewed/spell-checked by yourself.

The main shift in this new process is that, first of all, you start with a strong project management organization that spends time planning.

In the traditional translation process you have a Project Manager (PM), a translator, a reviewer and an in-country reviewer from the client (to validate the translation at the ending validate phase).

What I propose is that you eliminate the editor. You have a Project Manager whose function changes from manager to “facilitator”. And you create a “community” (a discussion list, a portal, it can take any shape that you want), where you have the translators, a consultant, an expert on the topic whose job is to answer questions about the topic in the corresponding language.

Not a linguist but, for example, if you are doing an Adobe project, a graphic designer, or if you are doing an SAP project, an accountant, or if you are doing a medical devices project, a doctor. You would also have the client, as the validator, forming part of this discussion group.

Other parts of this discussion group would be the tools, which are the shared translation memory, Machine Translation if you want to use it, as well as online resources such as terminology and style guides. Everything would be shared in a single place. This can be a free tool, anything ranging from Yahoo or Google groups, a private forum or whatever you can use and create for specific projects.

Now let’s address the quantity of translators in a project. If you look at the timeline for a project, usually the Project Manager will try to distribute the work as fast as possible to everyone involved, so the translators have plenty of time to translate and the reviewers have plenty of time to review.

In the process I propose, the PM gets gets the job in and then stops and holds the project. The PM checks what are the potential problem areas that the project is going to have.

For example, the typical problem that comes from dividing the project among people is the issue of style consistency. But the truth is that style is something that is very very very easy to standardize. What usually happens, as we both know, is that most translators don’t usually read the style guide, they assume that it’s always the same thing.

To give you a classic example, we had a project in Portuguese for SAP and while we were translating part of the material, one of the glossary entries was “cost component”, which in Portuguese has an easy literal translation (“componente de custo”) and as a translator you would never look up this this combination of terms in a glossary, because they are common words of common usage.

But SAP used for this term combination the idea of “cost element” and it happened that none of the translators remembered to check this, because most translators think that the glossary is a reference guide for things that they don’t know. They don’t [seem to] understand that glossary means mandatory terminology, that it is a standardization tool.

Every project usually starts with a kick-off meeting where the PM usually calls or writes to the translator to inform him/her of the number of words to be translated, as well as the sample material, the style guide and the glossary available, all of which the translator usually checks only if he/she needs it, after having started to translate immediately.

So in my proposal there is this element of first setting the expectations. In this new process, in the kick-off meeting you bring as many translators as you can, in order to inform them of the total number of words that the whole team will work on and to let them know that, after having checked the material, there are some clear style rules for the project (for example, titles will be capitalized, the gerunds will be translated as infinitives, etc.), which everyone has to agree to follow, so that there are no doubts or discussion afterwards about the style to be followed.

This takes care of the style standardization, because there are 15 to 20 variables in style guides that are basically the same, so you can do the same with the terminology. Instead of calling it “glossary”, you call it “mandatory terminology”. You can first have all the key glossary entries discussed, reviewed, agreed upon.

Even if you have a translator who does not agree about the proposed translation for a given word, you can let the corresponding consultant expert or the final client help you decide if the word in the glossary should be used or another version based on the target country of the translator.

This is possible, as I followed this process in my translation office, with 60 translators in one room, with all the computers connected to each other using Trados. Today you can do that with the translators being anywhere in the world, because of the collaboration technology available. Back then Skype didn’t exist, but now we are having this conversation through Skype. I can immediately send you a file through Skype and ask you something about it. It’s a collaboration, it’s work that we can easily do together just using a very basic tool.

There is a vast array of technological collaboration tools available that you can use to improve your productivity. Therefore, there is no reason why you need only 2 or 3 translators to be working even in a medium-sized project.

The timeline I am describing means that, instead of having 2 translators working over 10 days, or instead of having 10 translators working over 2 days, you can have 10 translators working over 4 or 5 days, in order to have more time to do well your planning at the beginning, more time to do a check at the end, and you still have 2 days for the 10 translators to do the project. You multiply the number of translators by 5, but you reduce the delivery time by 2, which is a great productivity improvement. Instead of doing it in 10 days, you do it in 5 days. You do it with 10 translators. The actual translation happens in 2 days.

Does that make sense?

[IVETTE: Sure, I am completely familiar with the process you are describing. Of course you describe an ideal situation. Big companies (clients or translation agencies) aim at duly carrying out this process, but in some cases, even if this is their aim, the process can be bureaucratic, because sometimes these collaboration tools don't allow enough (group) dynamism.

That is, sometimes there is no quick response if a translator has a query about a given word and has to wait 3 days to get an answer until the corresponding expert is available. I think the scenario you describe sounds extremely effective, except that in order for it to be so you need a very well organized and synchronized project management team, to assure that all the persons involved are coordinated well enough.]

Correct, that is where the “cultural shift” happens. The most common misconception when we talk about this issue is, first of all, believing that the consultant role, the “validator” role is a luxury. No, it’s a requirement.

In the projects in which I worked, there was a commitment that any questions that were asked during the day had to be responded by 6 p.m. that same day, so the next morning or at night when the person who had made the query started working again, this query would be solved.

The second big mistake that happens is that this process only works with highly professional translators who are well paid and understand their role in the process. If the company is trying to apply a collaborative approach like the one I have described to reduce costs, to reduce price, to pay less to the translators, that’s a mistake. The value that you get from a process like this is when you pay the best translators the best rates in order to get a productivity improvement that will reduce the overall cost of the project.

That is, the fact that you reduce the turnaround time from 30 days to 15 days is a project management reduction by 50% in time, the overhead goes down by 50%.

[IVETTE: In this respect, I would like to add that maybe translators in this process ought to be technological-savvy and linguist experts, a special combination.]

Yes and no. Theoretically, if you have a very good translator with basic knowledge of technology, you don’t need to be an expert in Trados to do the job, just some basic understanding of “reuse” [of the tools/material], with a basic training of a couple of hours. I’m talking about tools like Idiom and The Freeway (from Lionbridge) that are totally online, in which case the translator doesn’t even need to have anything installed. Even Lingotek, where the translator works in his/her natural environment or browser, or something like that.

[IVETTE: Sorry to insist, but I have worked in some projects with tools that require complex PC configurations, like TCP or plugin requirements. That takes a lot of time for you as a mere translator to setup or rather figure out. That's what I meant when I mentioned this "special" kind of "technological-savvy linguist" who has to invest time to adapt to the client's tools, something that should be rewarded when it comes to rates.]

Yes, yes, in fact, one of the things I suggested when we wrote a report about this topic is that the translator should be paid for their training. Any time that the translator has to spend attending a webinar or participating on a conference call and so on, should be paid time, because this is also work, this is not something that you are doing for free.

[IVETTE: Of course, absolutely.]

Again, Ivette, my focus here is on real quality and real productivity. There are two environmental factors that drive this rationale.

One is that there is an increasing demand for translation. The growth rate of content is much higher than the growth rate of translators. I can double the volume of translation in one year, but I cannot double the number of translators that are available in the market. It takes many years to create a professional translator.

And, again, content is doubling every year, is going up, and up and up and up. So the only way you can approach this growth in contents with a stable number of translators is by increasing their productivity.

The other environmental factor that takes place is that prices are relatively stable. The market tends to follow this traditional sequential approach because there is no questioning about this process, since this is the way it has always been done, and nobody stops to think about ways of being more productive. But if you look at how the material is written, sometimes you have 30 to 40 writers who have developed the contents and then we believe that the translation should be done by 1 or 2 people only.

Again, there has to be a cultural shift, because the resources are scarce, it takes a long time to prepare these resources, the contents is growing, there is more demand for translation and prices are relatively stable. Therefore, if you become more productive you actually can make more money by working less. I am all for the concept that there is room for professional translators to make money. This is not a “romantic” business. It’s a business and you need people to understand and look at it as a business, not as a “craft”, as still many translators do.

In some countries like Argentina and Italy there is this idea that only a translator who has studied translation in school can be a translator.

[IVETTE: I agree with this, based on my own cultural background (Latin America and Spain). This is probably because of the fact that, as you said, the way translation has been carried to date, a traditional "Gutenberg" process of translation has been followed. This is related to another issue I wanted to comment, translation education.

This has not necessarily evolved the same way technology has made translation processes evolve. What I mean is that, in the future, translation training or education may have to be split into different schools, according to those areas that require to be more technological-savvy, like technical documentation, and those more traditional areas whose translation will still be carried following the "Gutenberg" process, like journalism, law, etc.]

OK, let me comment the translation education element here. First of all, if you look at the translation process, it means converting words (written, recorded, spoken, signed, etc.) from one language to another. That’s a linguistic skill. That’s the basic skill that you need. You need to have knowledge of two languages, to convert information from one language to the other.

That skill training is something that schools can do very well. They can teach you the conversion process very well, just like they teach you Mathematics (for example, A2 = B2 + C2). The same with languages: you cannot use double negatives in English, but you can in Spanish and Portuguese. These are specific rules that you teach to people and is something that is never going to change. It was like that 50 years ago, it will be like that 100 years from now. That is, the language conversion part is going to be the same.

The rest is work, it is not a skill. It’s HOW you make the conversion.

If you look over time and consider, for example, the translation of the Septuagint version of the Bible, it was done by hand. 70 people translating by hand the Bible, and that is how it was done for many many years: monks with ink, copying and translating texts by hand.

The first revolution were the typewriters. Of course then there were the issues inherent to typewriting. Then the first word processors came around (I remember my first PC, in which I worked with WordStar and Ventura Publisher and programs like that). These were [programs that required] skills that were very relevant at the time, but which are useless today.

Then Trados came along. I have worked with Trados, with Star, TM2 from IBM, Déjàvu, which are all tools, skills that you learn with a little bit of training and that are irrelevant and that are independent of the language capability.

So theoretically, and I stress the idea of theoretically, you could separate both skills and have a person, again, hypothetically speaking, who is a blind translator and who has an assistant sitting in front of a PC. So the assistant asks the blind translator how you say something in Korean. The blind translator responds and the assistant or a second person types the translation in Trados. The person operating the computer is only asking the translator the things that need translation. If there is a 100% match, he/she is not going to ask the translator to do the job.

Therefore, you can dissociate the language skill from the productivity skill. Again, one thing is the education part and the other is the training part. Training will vary over time. What is important today is going to be irrelevant tomorrow.

For instance, do you ever use Google Translate?

[IVETTE: Yes, for some things.]

Of course. It would be stupid not to use it and it’s technology that was not available years ago. If I had had Google Translate years ago, I would have made so much money. I would not have to tell my clients about it. I could do chemistry translations. I hate Chemistry, I always mention that, but I had some chemistry patents that always came to me and I could have had all that translated by Google Translate because it is a formula, so I would just edit it, what’s the big deal? It’s straightforward.

Again, getting back to the education element, we tend to conflate in the same issue that the university doesn’t prepare the person for the market. The fact is, the market changes and the market requirements are different from one market segment to the other.

Like you said, in journalism, in marketing or literature, you are not going to use Trados to translate a novel or a music or cinema website, because there is not going to be any re-use of the contents. Thus, if I am a translator who will be working in the latter kind of translation, why should I have as part of my education 5 credits in Trados?

Which also takes me to another discussion, as I believe you don’t even need university-level training to do translations. From my professional perspective and experience, domain-specific knowledge combined with language capabilities are more important than the language capabilities by themselves.

What I mean is that, for example, if I have a doctor who speaks German and English, I can teach this doctor how to translate in 2 months, while I cannot teach a translator to be a doctor in 2 months. To be more specific, I can teach, in 2 months, a doctor who is from Germany and did his residency in the USA what the translation techniques are. I am saying 2 months with practice, because in reality I can teach anyone [with language skills!] to translate any language probably in 6 hours.

[IVETTE: Laughing...]

I am serious. I am telling you this from experience. I am not saying that the person is going to be a good translator, but I developed a translation training in Brazil because Brazilian universities teach translators how to translate Shakespeare, whereas I had legal and IT translations that needed to be done. What translators brought to me was totally irrelevant and useless.

So what we did is that we developed a little workshop which teaches people who have language knowledge the 13 (or 11, I cannot remember) techniques that you can use in translation. In our workshops I got doctors, accountants, telephone technicians who worked in Brazil and who had lived in the USA and had good English knowledge, and taught them how to translate. The rest is practice.

The big problem about the language conversion part is insecurity. The translator doesn’t know what he/she is allowed to do: can I omit a word? Can I use a synonym? Can I change the sentence and change the word order? Do I have to be literal?, etc. Once again, this is an issue that is part of the planning and preparation. If you setup a certain set of rules upfront, you won’t have this confusion.

[IVETTE: In all fairness, I have worked with experts, too, like engineers, who have either done the translations or revised them, and many times you cannot compare the language quality result of an expert with that of a linguist, unless the expert has also done linguistics. I have sometimes been a vehement supporter of translation studies mainly because of the linguistic part, because I honestly think that the expertise of a subject-matter expert is not enough to produce a really acceptable text from the linguistic point of view.]

Yes, but again, this is where the training comes into place. To give you an example, the president of the Brazilian translator association is an engineer. He was an engineer when he came to work for me, and he had lived in the USA and Holland and thus had used English functionally for a long time, but he had never translated. He knew the topics very well, he was conversant in the topics, but since he had gone to a very good school in Brazil, I also knew he had a good background.

The first thing we did for potential translators like him was to do a Portuguese grammar test, and if he had a good result, you could assess that he could write well in his own language, which is something that a good school would provide. Assuming you are translating into your mother tongue, the linguistic skill comes from good basic education. If you finish your high school studies very well in a good school, you should be able to write in your own language without any problem. That’s the kind of empowerment that professionals with linguistic skills are able to do for commercial translations.

I know this is a very philosophical issue, education. [Of course I know that if] I start talking about [education like] this, say in Argentina, I have to prepare for [getting hit with] tomatoes, because if you are not a “Traductor Público Jurado” (certified translator) they are going to think that you are a… I don’t know, a fraud. So I had to fight a lot to get very good translators working for me there, because they didn’t have a diploma.

[IVETTE: In my case, I have learned to be flexible about this because I have always considered translation as you described it, as teamwork, so you need to evaluate each team member's skills and join them to lay good translation foundations from the start.

Obviously if one person is a better linguist and another one is a better subject-matter expert, if you combine both skills you will be able to produce a beautiful result.]

Exactly. It might be ideal to follow, it might require more work, but you have a combination. The final idea is that you create a community.

When you have 10 translators working on the same project, you can discuss, through the community, things like cross-references where one person working on a given chapter finds a reference to a paragraph from another chapter another person is translating and they help each other. So the community idea does work.

When I worked with this 60-person project, we had 2 external consultants in the project, and one was involved in making the final word decisions, but when a translator asked a question, any of the colleagues could respond. Somebody in the community could know the correct translation for the term asked and all the validator would do at the end of the day is say if the answer was correct or not. In case it were wrong (usually the exception), the translator would just have to go back and correct it, with search-and-replace, no big deal.

I had another project with 25 translators working off-site, translating all the documentation for a power plant. We hired a retired PhD in Engineering who used to manage a power plant. He worked from our office and the translators nicknamed him “God”, so whenever they had a question or any doubts about a term, if “God” had given the answer, there was no further discussion. [Laughs...]

There is a lot of value in selecting the right consultant, the right person to solve terminological issues, in order to create more confidence in the translation group.

[So this translation process “cultural shift”] is not an empirical discussion[, but last] year we actually did a series of events on vendor management and one company from Guatemala, of all places, came with a very interesting case-study, about a 30-million-word project they did for an automotive company.

They followed the SAEJ 2450 specification for translation projects. The project required metrics, lots of metrics, and they had an error tolerance of 0,05 %. Basically, aproximately one mistake per 10000 words. It was a very well monitored and measured project. One interesting anecdote about the project is that, in the middle of the project, the client told the translation agency that they had to speed up the project, in order to reduce costs, so they had decided to eliminate the edit process.

The translation agency was shocked, fearing they could not guarantee the quality, etc. etc., the usual stuff, but in the end they managed. This was an 18-month project and, after 6 months, they were instructed that the edit process would be eliminated. What happened is that quality improved by this elimination, because the editors were incorporating mistakes.

One of the things that is often forgotten in the quality process is that human nature dictates that we make mistakes in anything that we do.

Usually in the translation process you will have a translator who has a lot more information than the editor, because he has been there longer, because he is paid more, because he’s been briefed at the beginning of the project, he’s well informed and he knows what he is supposed to do.

It often happens that a translator has discussions with an editor because the latter insists on considering wrong a given translation which was, for example, in the style guide that the editor did not take the time to read, and the editor is changing things based on personal preferences.

What the translation company from Guatemala found out was that, after 6 months getting feedback from the editors and the reviewers, the translators didn’t need any additional review. They could perfectly deliver very good quality without any editing. What they removed from the process were the errors that the editors were incorporating into the translation. Beautiful story.

[IVETTE: One last question, regarding globalization and prices. If, for example, a company in a less-developed country manages to follow the productivity-enhanced translation and quality process you propose, but for at least half the price a similar company in a more developed country charges, how is the translation business supposed to adapt to this, especially freelancers, how can you compete in such a global market?]

Well, in your case, the problem is that you work into Spanish, which is the cheapest language in the translation market, and price is driven by the local market. The fact that there are 19 countries which have Spanish as their official language makes it harder for you to become competitive. And by-the-way, we found out in one of our researches that prices in Spain are lower than in Argentina.

So, again, if price is determined by the local market, the price of German translations is determined by the German-speaking market, the French one by the French-speaking countries, etc. Consequentially, translations in Swedish or Japanese are more expensive than Spanish or Chinese because the supply of translators in those corresponding markets is what determines the price.

The reality is that price has been very stable for a very long time from the client’s perspective. In the USA, a final client pays between 20 and 25 cents per word for translations into most languages. The translation agency will buy translations in Argentina between 8 and 10 cents per word, or in Spain between 6 and 8 cents per word, and that’s about what usually happens.

What has happened in the last couple of years is that there has been a lot of fluctuation in the exchange rates. So people are moving, they are doing what is called “arbitrage”, they are moving from doing translations in Europe, in German, to the USA, or doing Japanese translations in the USA instead of in Japan, because the exchange rate makes it more favorable.

So the translation market has become a global market and the pricing is not necessarily determined by where you are, but rather the value that you have.

My recommendation to freelance translators is to move away from the “commodity” market. [But I don't like to use the word "commodity", even though there is a certain element of commodity in freelance work.]

That is, you have the very good translators at the top of the market, who establish their price.

For example, there is this guy in Portugal who has a translation business, very technical translation, lots of clients, and he charges 10 cents of a euro per word to do Portuguese translations, while in Portugal people are doing translations for 4 or 5 cents of a euro per word. Whenever a company sends him an email with an offer to lower his rate, he sends them an invoice for 50 euros, charging them for “wasting my time”.

Anyway, he is an example of a translator who sets up his price. I know another similar case of a translator in Italy, who only works with international companies because she gets 12 cents per word, while the local market pays 5 to 6 cents, because she managed to carve herself a niche.

Another one who works in financial translations, which is another niche market, and he doesn’t need to worry about work because he has enough and has established his credibility in the market as an expert.

This also has to do with “personal branding”, how you sell yourself, how you position yourself in the market, vis-à-vis your competition; that is, the other translators.

Modestly, I was very successful as a translator and what happened over time when I was doing that is that I would take the time to share my knowledge with other translators, to train other translators, and, invariably, whenever any of them got a job that they could not do, they would send it to me.

Early on in my career my best marketing was establishing myself as an authority among my peers, so that my peers would refer me whenever they had a project they could not handle themselves. That allowed me to charge more.

This is a discussion I have with LSPs [Language Service Providers] all the time: move away from price and start talking about value. What is better, a cheap translator or a good translator who will deliver your translation earlier, better, with no complaints from the client, without multiple rounds of reviews, etc.?

You can manage to show your client, with a little effort, that if they pay you more, the total cost of the project will be lower if they are working with a professional.

Inter-professional Dialogue: Kirti Vashee About the Future of Translation/MT | Diálogo interprofesional: Kirti Vashee sobre el futuro de la traducción/TA

Translation is at a point that is perhaps similar
to the turning point for how stenographers/typists evolved[....]

La traducción se encuentra en un punto de cambio tal vez similar a la evolución
que se dio entre estenógrafos y mecanógrafos[...]

1) Could you give us a quick summary of your professional ties with the translation world? I am currently with an Asia-based startup (Asia Online) that is attempting to translate millions of pages of open source knowledge content into several South Asian languages using a combination of SMT, ongoing expert linguistic corrective feedback and crowdsourcing. The objective is to produce “almost human” quality translations rather than the “gist” quality that most MT has come to be associated with. This is possible over time because of the high degree of human feedback & involvement in our technology. Previously, I was the VP Sales & Marketing at Language Weaver, an SMT pioneer and helped build their worldwide sales and marketing organization [, as well as in other companies in which my work helped me learn] the value of localization and translation in building international businesses. My full background is available here. In a recently published article at Translution.com (see this link), a definitely MT (Machine Translation)-based view is given about the future of translation, though based on the RBMT (“rules based”) MT method, as opposed to the SMT (Statistical Machine Translation) method. You commented in another Internet forum, in reaction to the previously mentioned article, that: [... A]utomated translation is definitely coming and the future is likely to be dominated by new initiatives that are based on Open Source (Moses) SMT + Crowdsourcing + Open and Free sharing of linguistic assets around platforms that encourage and facilitate web based community collaboration. 2) Why (how) can SMT offer better results than RBMT? I recently wrote an article comparing the two approaches that was published in the ATA Language Technology newsletter (see this link to read the article). I believe this is a fair overview of both technologies. RBMT has been around for a long time and the Systran French system is now over 30 years old.What you see today is the result of 30+ years of development and refinement. Both approaches continue to improve but I happen to believe that the future will belong to SMT based approaches (which are now adding more linguistics). The current SMT systems you see at Google and Microsoft are less than 5 years old and the rate of progress is clearly faster than RBMT. Since data-driven SMT systems improve as the quality and volume of data increases, we can expect that they will continue to improve continuously. As the human translated material and monolingual content in many new languages grows on the web these systems are very likely to improve. The combination of web data harvesting, better statistical algorithms and faster, more powerful computing all favor SMT. The TAUS organization has also shown that SMT domain focused systems built for large corporate buyers outperform RBMT systems in quality and productivity benefits. The experience at Asia Online also shows that when expert linguistic corrective feedback is provided on a systematic and continuous basis, it is possible to reach very compelling quality very quickly. This man-machine collaboration I think will drive the technology forward – SMT is much more suited and likely to benefit from the social networking trends on the web. The trend in the technology is towards “hybrid” MT models, but my belief is that SMT-based hybrids will emerge as the best way to get to usable quality.I predict that “active human-steered systems” will outperform any raw MT initiative, including those at Google which are also improving. Leading edge, forward thinking LSPs are beginning to understand that automated translation technology can be used in a positive and productive way and some are investing in learning how to use it in synergistic ways to help develop a competitive edge. I believe that these pioneers will emerge as industry leaders in future. Also, SMT has a much more active and vibrant Open Source movement which is encouraging a lot of new experimentation and innovation. 3) Why should Open Source be a better solution to improve the results of MT? While I cannot say with certainty that Open Source will produce “better” solutions, it is clear that a thriving open source movement helps build momentum around a technology and also accelerates its growth and acceptance. One of the major barriers to software developers initially investigating a technology, are the initial licensing fees necessary to access the core infrastructure they need to build applications. Open source eases this issue and allows/encourages more people to engage with the technology. Additionally: * Open Source allows skillful people all over the world to experiment with the technology and add to the common pool of knowledge.People in open source are much more willing to share best practice information and insights with each other. * Open Source developers hold frequent competitions to compare how they are solving the same problem and thus are constantly from each other and driving the technology forward. The Moses SMT community does this a few times a year and they get better each year. * Open Source also tends to drive and propel innovation at vendors who are competing with the innovation from the broad community and who need to also keep improving to stay ahead. * Open Source allows initiatives that are not motivated by profit, also to benefit from the technology. * Open Source experiments provide very useful data that drives understanding. I saw many presentations at AMTA 2008 that provided information for all to benefit from. The Linux OS, Firefox and even Internet Explorer are all initiatives that started as open source and as many will attest, they are all initiatives that have produced high quality, stable and competitive alternatives to the major vendor offerings. The Wikipedia is possibly the best example of what is possible with a collaborative open source movement. While not perfect and sometimes inaccurate, for most people it is the best and highest quality information source for all kinds of information. You also commented that: Automation in most other realms has resulted in transferring human skills to higher value tasks. This caught my attention because I have recently been reading about economist Joseph Schumpeter and how he popularized the term “creative destruction” (originally used by other economics-related theorists such as Werner Sombart) to, as described in the Wikipedia, “describe the process of transformation that accompanies radical innovation”, though probably more related to the times when war acted as a “destructive renewing” force. (I mention this because, if you visit or read some translation-related forums, a number of current freelance translators seem to be a bit concerned about the fact that MT may take away part of their work.) 4) Therefore, in the case of translation, would you then agree that MT (especially when used in conjunction with translation memories) is that “creative destructive” technology which in the long run will be “reducing” more and more a translator’s task to that of “editor” of legacy translations? The business of translation I believe is poised for a major change or “creative destruction” in the sense that Schumpeter used the phrase, to “describe the process of transformation that accompanies radical innovation”. However, I see MT as a building block or by-product, rather than the driver of this change. The driving force for this change is the human quest for knowledge and information that goes far beyond the focus of the localization and globalization industry. The driving force is the human need to communicate. There are many who believe that it is important that the world’s knowledge [should] not be confined to a handful of languages and that information be liberated and shared with the world at large, in local languages wherever possible. There are others who have eloquently stated the need for this, e.g. here and here. These thoughts are being echoed across the globe and it is likely, that the change agents will be people who do not come from the traditional translation industry. Basically, there is too much information, for humans to do it all, to do it in a useful timeframe and do it at an acceptable cost. The tools, technology and processes that enable these massive translation projects to happen will also impact the traditional localization industry. I expect that competent humans will always manage the translation of the most sensitive, high value and important content, but a lot of the rote and repetitive content can and should use more of these automated translation tools. MT systems that are tuned for a specific domain are significantly superior to what you see on any of the free MT systems on the web and can improve productivity on the translation of large manuals and user documentation. I expect that there will be changes at both the tools/technology level and the human process level as outlined below. Tools/Technology: There is reason to believe that TM will fold into MT, especially SMT, which is built using legacy human translations. Many buyers/LSPs will learn how to effectively use automation to gain cost and productivity advantages over their competition. The translation industry is remarkable in how little automation penetration there is, and how fragmented, primitive and proprietary the tools are. Industry leaders like Lionbridge and SDL have shockingly low market values, probably because analysts understand that their business are extremely vulnerable to advances in the technology. On March 6th 2009, LIOX had a market value of $63M on sales of over $400M! SDL value is at least slightly in excess of sales. This is terrible, even in the midst of this financial crisis. The widespread need to translate massive, high value content will drive the development of tools and more efficient systems and processes. A combination of the most agile tools vendors from the translation industry, completely new “sea change” agents like Google and the open source movement will make this happen. The whole localization industry will benefit but the current model will change. Human Process: It is likely that current model of Buyer > MLV > SLV > Freelancers will change. There are simply not enough translators to translate all the content that buyers will need to translate to stay competitive. I speculate that this will result in the use of more automation and increased use of less skilled bilingual humans for tasks like post-editing domain focused SMT output. Even the final consumers of the information can contribute to “clean up” of MT output and raise ongoing quality via crowdsourcing or other forms of targeted community collaboration. Competent human translators should not fear that they will become mere “editors” as their skills should become more valuable, to manage the community collaboration, crowdsourcing and to steer the translation quality during conversion of large bodies of content. Skilled language professionals will be a key to success on these projects. Humans who can provide linguistic guidance, help develop error detection and error correction strategies, manage and maintain linguistic assets, and guide automation processes through linguistic complexities will be in high demand. Microsoft has shown that raw MT can help build global customer satisfaction and hundreds of millions of their customers use their MT knowledge base content successfully. Given the volume, value and urgency of the knowledge base content there is simply no way that this could be done by humans in time even if cost was no object. While Microsoft spends over $200M/year on translation services, human translators only translate a tiny fraction of the content they actually make available to customers and partners. Given the rate at which user generated content is growing, the urgency and value of doing translation can only increase. The pricing models are also likely to change. It does not make sense to charge by the word when you are translating billions of words.Pricing that is subscription, project or fixed monthly cost based until a quality level is reached, make much more sense. The pioneers who get involved early will define these new models after they better understand the technology, the costs and the keys to success. Translation is at a point that is perhaps similar to the turning point for how stenographers/typists evolved, or accountants with green shades and calculating machines evolved. In both cases people previously in these jobs, do work that today is much richer and intellectually stimulating and demanding. The people who developed the greatest skills with word processors and presentation tools in one case, or spreadsheets and data analysis tools in the other, are valued for their ability to enhance and improve business information flows and are closer to the core business mission. However, marginal employees in both cases were displaced. Evolution is also an imperative that demands adaptation and competence with new tools and new skills. I think translation has a future where it is an important driving force behind international business initiatives, enabling better communication with global customers and more closely integrated into both the product creation and the global sales and marketing activities. I see a future where translation focused staff and technology is central to global initiatives of all kinds. Who knows? It may also help to dramatically reduce human conflict along the way. 5) With the current development and use of other language-related automation technologies such as voice-transcription (for example, programs like Dragon Naturally Speaking), would you say that interpretation could also be “affected”, soon enough, by automation processes equivalent to that of MT? I think the same need to communicate will drive this technology too. I have seen research from CSA (Common Sense Advisory) that indicates this is becoming more and more likely. As the translation technology improves there will be more and more support for developing this further. The ASR (Automated Speech Rrecognition) technology is the weak link today but is improving all the time. This article indicates it is imminent: like MT this technology will evolve and improve and the best interpreters will learn to use it to gain competitive advantage. 6) If technological advances are also usually aimed at lowering costs (by decreasing the amount of work carried out by human labor, which then lowers costs), how would the translator of the future (especially in the case of freelancers), whose main task may be then to edit the MT/TM-processed contents of legacy translations, be then best paid? It would seem that the traditional wordcount (or even linecount) method would not be appropriate any more. Hourly? This is a view of the world that is focused only on the translation work of the past: brochures, user manuals, GUI and some web content. All mostly static content. The future is much more about dynamic and living content. The highest value translation of the future must focus on the opinion of the global customer in the growing and influential social networks. Global companies are generating content at a rate never seen before. Global customers want and demand that they get access to more than the user manual – they want more and more. Successful global businesses will provide this information. Microsoft is a leader who has already demonstrated the value of doing this, others will follow. The future of the internet involves more languages beyond the familiar FIGS [French, Italian, German, Spanish] circle. There will always be a demand for competent translators who can help buyers deal with this new demand for more content and the pricing and business models will change and evolve with the technology. Global success in the future will require more content, more multimedia and more customer interaction in many more languages. My guess is that translation will become a more financially and intellectually rewarding line of work and those who adapt to the emerging world will prosper in every way.

1) ¿Podría darnos un breve resumen de sus lazos profesionales con el mundo de la traducción? Actualmente formo parte de una iniciativa empresarial con sede en Asia (Asia Online), la cual busca traducir millones de palabras de contenido de software de código abierto (Open Source) a varios idiomas asiáticos, mediante el uso combinado de la traducción automática estadística (en inglés “SMT”, “Statistical Machine Translation”), la ayuda continua de expertos en la corrección lingüística y recursos de tipo crowdsourcing. El objetivo de la empresa es producir traducciones de calidad “casi humana”, en lugar de “la traducción aproximada” con que se suele asociar a la Traducción Automática (TA). Esto será posible a medida que pase el tiempo, gracias al alto grado de ayuda y colaboración humana con que cuenta nuestra tecnología. Anteriormente trabajé como Vicepresidente de Ventas y Marketing en Language Weaver, una empresa pionera en el desarrollo de la traducción automática tipo SMT [, así como en otras empresas en cuyo trabajo] aprendí a valorar la importancia de la localización y la traducción en la creación de empresas de alcance internacional. En esta página aparece la información completa sobre mi trayectoria profesional. En un artículo recientemente publicado en Translution.com (véase este enlace), se da una imagen definitivamente basada en la TA sobre el futuro de la traducción, aunque más bien en la TA que utiliza el método de reglas (en inglés, “RBMT” o “Rules Based Machine Translation”), en contraste con la TA que utiliza el método estadístico (en inglés “SMT”, “Statistical Machine Translation”). En otro foro de Internet usted comentó, como reacción al artículo mencionado, que: La TA definitivamente está en camino y lo más probable es que en el futuro predominen las iniciativas basadas en SMT de tipo Open Source (Moses) + Crowdsourcing + el uso compartido y el libre acceso de todos los bienes lingüísticos a través de plataformas que promuevan y faciliten la colaboración basada en Internet. 2) ¿Por qué (y cómo) puede la traducción automática tipo SMT ofrecer mejores resultados que la traducción automática tipo RBMT? En fecha reciente escribí un artículo en el que comparaba ambos métodos, el cual fue publicado en el boletín sobre tecnología y lenguaje de la ATA (véase este enlace para consultar el artículo.). Creo que en él doy una visión imparcial sobre ambas tecnologías. La TA tipo RBMT lleva mucho tiempo de existencia y el sistema francés Systran [en el cual se basa] tiene más de 30 años. Lo que vemos hoy en día [en dicho sistema] es el resultado de más de 30 años de desarrollo y refinación [de dicho sistema]. Ambos métodos siguen en continua mejora, pero yo personalmente creo que el futuro está en manos de innovaciones basadas en el método SMT. Los sistemas basados en SMT que actualmente se encuentran en uso en Google y Microsoft tienen menos de 5 años y su ritmo de avance es claramente más rápido que la TA tipo RBMT. Dado que los sistemas basados en SMT mejoran a medida que la calidad y el volumen de datos aumenta, es factible pensar que su mejora será continua. A medida que el material traducido por personas y el contenido monolingüe en nuevos idiomas aumenta en Internet, es muy factible que estos sistemas [automáticos de traducción] mejoren. La mezcla de recolección de datos en Internet, mejores y más rápidos algoritmos y sistemas informáticos más potentes juegan en favor de la TA tipo SMT. La organización TAUS también ha puesto de manifiesto que los sistemas centrados en el método SMT desarrollados para clientes tipo grandes corporaciones ofrecen un mejor rendimiento en cuanto a ventajas relacionadas con la calidad y la productividad. Nuestra experiencia en Asia Online demuestra que cuando se cuenta con la ayuda continua y sistemática de expertos en la corrección lingüística, es posible lograr rápidamente una calidad convincente. Creo que esta mezcla de colaboración entre humano y máquina impulsará la tecnología y SMT está mucho mejor adecuada y lista para obtener ventajas de las tendencias de redes sociales en Internet. La tendencia actual de las tecnologías es hacia los modelos “híbridos” de TA, pero los híbridos basados en SMT destacarán como los mejores para obtener calidad útil. Vaticino que los “sistemas impulsados por la traducción humana activa” ofrecerán un mejor rendimiento que cualquier iniciativa basada meramente en la TA, incluidas las existentes en Google en proceso de mejora. Los proveedores de servicios de traducción [en inglés, “LSP”, de “Language Service Provider”] que lideran la industria han empezado a comprender que la tecnología TA se puede utilizar de forma positiva y productiva, y algunos han dedicado recursos al aprendizaje de cómo utilizar la TA de forma sinérgica [es decir, de cooperación interactiva entre grupos]. Creo que estos pioneros destacarán como líderes de la industria en el futuro. Asimismo, la TA tipo SMT cuenta con un movimiento de Open Source mucho más activo y dinámico, lo cual ha fomentado mucha experimentación e innovaciones. 3) ¿Por qué ofrecen los programas tipo Open Source una mejor solución a la hora de obtener mejores resultados mediante el uso de TA? Si bien no puedo afirmar categóricamente que los programas Open Source ofrecen “mejores” soluciones, está claro que el movimiento de apoyo popular con que cuentan dichos programas ha contribuido a impulsar dicha tecnología y a acelerar su crecimiento y aceptación. Una de las mayores barreras a las que se enfrentan los desarrolladores de software que empiezan a investigar una tecnología es el coste inicial de licencias necesarias para acceder al núcleo de la infraestructura a partir de la cual desean crear aplicaciones. Los programas Open Source facilitan el acceso y permiten y promueven la participación de muchas personas en la tecnología. Además: * Open Source permite a talentos de todo el mundo experimentar con la tecnología y contribuir a la base común de conocimiento. Las personas que trabajan con programas Open Source están mucho más dispuestas a compartir entre ellas información de tipo “mejores prácticas”, así como sobre aspectos más profundos. * Los desarrolladores de Open Source compiten a menudo para comparar cómo resolver un mismo problema, lo cual significa que constantemente están en contacto y todo ello ayuda a impulsar la tecnología. La comunidad Moses SMT mantiene varias veces al año este tipo de actividades y cada año obtienen mejores resultados. * Open Source también suele contribuir a dirigir e impulsar las innovaciones en el caso de proveedores de servicios que compiten entre sí por las innovaciones dentro de esta amplia comunidad y que también necesitan mejorar para mantenerse al día. * Open Source permite la incorporación de iniciativas sin fines de lucro, también con el fin de sacar provecho de tales tecnologías. * Los desarrollos experimentales en Open Source proporcionan datos muy útiles, que promueven la comprensión. Asistí a varias presentaciones en la conferencia de la AMTA (asociación americana de TA) de 2008 en que se ofreció información muy útil para todos. El sistema operativo Linux, el navegador Firefox e incluso Internet Explorer son todas iniciativas que empezaron como programas tipo Open Source y, como muchos confirmarán, han generado alternativas de alta calidad, estables y competitivas en relación con las demás ofertas de los principales proveedores del mercado. La Wikipedia es probablemente el mejor ejemplo de lo que el movimiento Open Source puede lograr en un entorno de colaboración. Aunque no es un resultado perfecto y en algunos casos no es preciso, para la mayoría de las personas es la mejor fuente de información y la de más alta calidad. También ha comentado usted que: La automatización en la mayoría de los demás campos ha tenido como resultado que se dediquen las aptitudes de las personas a tareas de más alto valor. Esto me llamó la atención, ya que he estado leyendo sobre el economista Joseph Schumpeter y sobre cómo él popularizó el término “destrucción creativa” (originalmente utilizado por otros teóricos de la economía como Werner Sombart) para, en palabras de la Wikipedia, “describir el proceso de transformación asociado a la innovación radical”, aunque probablemente él se refería a la época en que la guerra fue esa fuerza “destructiva y renovadora”. (Menciono esto porque, si visitamos o leemos el contenido de algunos foros de Internet relacionados con la traducción, veremos que actualmente un número de traductores autónomos parece estar un poco preocupados por el hecho de que la TA les quite parte de su trabajo.) 4) Por tanto, en el caso de la traducción, ¿está de acuerdo en que la TA, especialmente cuando se utiliza conjuntamente con las memorias de traducción, es esa tecnología “creativa y destructiva” que a la larga “reducirá” más y más la tarea del traductor a la de “editor” de las traducciones automáticas heredadas? Creo que el negocio de la traducción está listo para un cambio importante o “destrucción creativa” en el sentido que Schumpeter utilizó esta frase para “describir el proceso de transformación asociado a la innovación radical”. Sin embargo, considero que la TA es solamente uno de los elementos de edificación, o bien un producto derivado, y no el impulsor del cambio. La fuerza impulsora del cambio es esa búsqueda del conocimiento y la información por parte de la humanidad, algo que va más allá del enfoque de la industria de la localización y la globalización. La fuerza impulsora es la necesidad humana de comunicarse. Hay muchos que piensan que es importante que el conocimiento del mundo [no debe] quedar relegado a unas cuantas lenguas y que la información debe ser de libre acceso y poder ser compartida por todo el mundo, en cuantas más lenguas locales posibles. Hay quienes han expresado esto de manera elocuente; por ejemplo, aquí y aquí. Esta forma de pensar está teniendo eco en todo el planeta y es muy probable que los agentes del cambio sean las personas que no provengan de la industria tradicional de la traducción. Básicamente el problema es que hay demasiada información para que los humanos se encarguen de su totalidad, en un período de tiempo que resulte útil y a un coste aceptable. Las herramientas, la tecnología y los procesos que permiten que estos proyectos de traducción a gran escala tengan lugar también tendrán un efecto sobre la industria tradicional de la localización. Supongo que los traductores humanos competentes siempre se las apañarán para hacerse cargo de la traducción de los documentos con contenido más confidencial, más valioso y más importante, pero una gran parte del contenido rutinario y repetitivo puede y debe ser traducido mediante más herramientas de traducción automática. Los sistemas de TA que estén adaptados a un campo específico serán mucho mejores que cualquiera de los sistemas gratuitos de TA que encontramos en Internet, lo cual puede mejorar la productividad a la hora de traducir manuales y documentación del usuario de gran volumen. Supongo que habrá cambios a ambos niveles, tanto en lo que respecta a las herramientas, como a la tecnología, así como a nivel del proceso humano, tal y como detallo a continuación: Herramientas y tecnología: hay motivos para pensar que las memorias de traducción se combinarán con la TA, especialmente la de tipo SMT, la cual se crea a partir de traducciones humanas heredadas. Muchos clientes y proveedores de servicios lingüísticos aprenderán a usar de forma efectiva la automatización, a fin de obtener ventajas a nivel de costes y productividad sobre sus competidores. La industria de la traducción es una excepción en el sentido de lo poco que ha permitido la integración de la automatización, así como en el sentido de lo fragmentado, primitivo y de propiedad restringida que son las herramientas utilizadas. Líderes de la industria como Lionbridge y SDL tienen bajísimos valores de mercado, lo cual resulta chocante, probablemente debido a que los analistas saben que este tipo de negocio es extremadamente vulnerable a los avances tecnológicos. El 6 de marzo de 2009, LIOX tenía un valor en el mercado de ¡aproximadamente 48 millones de euros a partir de ventas por un valor de aproximadamente 309 millones de euros! El valor de SDL al menos sobrepasa ligeramente el valor de las ventas. Esto es terrible, incluso en medio de esta crisis financiera. La cada vez más extendida necesidad de traducir contenidos de gran volumen y de alto valor impulsará el desarrollo de herramientas, así como de procesos y sistemas más eficientes. Esto ocurrirá gracias a la mezcla de las herramientas más ágiles de los proveedores de servicios de la industria de la traducción y a nuevos agentes “renovadores del panorama de la traducción”, como Google y el movimiento Open Source. Toda la industria de la localización saldrá beneficiada, pero el modelo actual cambiará. Proceso humano: es muy probable que cambie el modelo actual de Comprador > Proveedor de servicios multilingüísticos (en inglés “MLV”, de “Multi-Language Vendor”) > Proveedor de servicios monolingüísticos (en inglés SLV, “Single Language Vendor”) > Traductor autónomo. No hay suficientes traductores para traducir todo el [volumen de] contenido que los compradores necesitan traducir para mantenerse competitivos. Me atrevo a conjeturar que esto tendrá como resultado el uso de más automatización y del empleo de personas bilingües que no requieran tanta preparación para tareas como la postedición de los resultados basados en el enfoque de TA tipo SMT. Incluso los usuarios finales de la información pueden contribuir a “limpiar” el resultado de la TA y mejorar la calidad en curso mediante recursos tipo crowdsourcing u otras formas de colaboración orientada a la comunidad. Los traductores humanos competentes no deben temer convertirse en meros “editores”, ya que sus aptitudes adquirirán más valor, en el sentido de que serán los encargados de gestionar la colaboración de la comunidad, de los recursos tipo crowdsourcing y de guiar la calidad de la traducción durante la conversión de los grandes volúmenes de contenido. Los profesionales lingüísticos con talento tendrán un papel decisivo en el éxito de estos proyectos. Aumentará la demanda de personas que puedan proporcionar servicios de orientación lingüística, ayudar a desarrollar estrategias tanto de detección como de corrección de errores, gestionar y mantener los bienes lingüísticos y dirigir los procesos de automatización a través de las complejidades lingüísticas. Microsoft ha demostrado que el uso directo de TA puede contribuir a edificar la satisfacción de los clientes a escala global: cientos de millones de usuarios utilizan satisfactoriamente el contenido de su base de datos de conocimiento traducida mediante TA. Dado el volumen, el valor y la celeridad de uso del contenido de la base de datos de conocimiento, no existe de ninguna manera una forma sencilla de que esto pueda ser llevado a cabo por traductores humanos en un tiempo razonable, incluso si los costes no son un problema. Si bien Microsoft dedica aproximadamente más de 150 millones de euros al año a servicios de traducción, la traducción realizada por personas es solamente una ínfima parte del contenido que ellos ponen al alcance de sus clientes y socios. Dado el ritmo al que crece el contenido generado por los usuarios, lo lógico es que la celeridad y el valor de hacer traducciones aumenten de valor. Por tanto, es lógico que también cambien los modelos de tarificación/precios. No tiene sentido cobrar por palabra en el caso de tener que traducir millones de palabras. Tiene mucho más sentido idear una tarificación basada en el coste de una suscripción, un proyecto o una cantidad mensual que se cobre hasta alcanzar un determinado nivel de calidad. Los pioneros que se atrevan primero a explorar estos nuevos modelos serán los que estarán en mejores condiciones de definirlos mejor, una vez que asimilen mejor la tecnología, los costes y las claves del éxito. La traducción se encuentra en un punto de cambio tal vez similar a la evolución que se dio entre estenógrafos y mecanógrafos, o bien entre la contabilidad manual de ataño y la contabilidad con calculadora. En ambos casos las personas que realizaban estos trabajos hoy en día realizan trabajos mucho más sofisticados y más estimulantes desde el punto de vista intelectual. [Asimismo,] a las personas que han desarrollado mejor sus aptitudes de uso de programas de tratamiento de texto o de herramientas de presentación, o bien de herramientas de hojas de cálculo y análisis de datos, se las valora [hoy en día] por su capacidad de aumentar y mejorar los flujos de información empresarial y, por tanto, están en más estrecho contacto con los objetivos principales de la empresa. Sin embargo, en ambos casos, los empleados que se hayan quedado al margen [de dicho desarrollo] han terminado siendo desplazados. La evolución también es una fuerza imperativa que exige ser capaz de adaptarse y de desarrollar un nivel de cualificación adecuado para utilizar nuevas herramientas y adquirir nuevas aptitudes. Creo que en el futuro la traducción jugará un papel importante como fuerza impulsora de las iniciativas empresariales internacionales, lo cual contribuirá no solamente a lograr una mejor comunicación con los clientes de todo el mundo, sino también a que dicha comunicación se integre mejor durante la creación del producto, las ventas globales y las actividades de marketing. Veo un futuro en que las tecnologías y la mano de obra que tengan como eje la traducción jugarán un papel crucial en las iniciativas globales de cualquier tipo. ¿Quién sabe?… Puede que además de paso contribuyan a reducir de forma espectacular los conflictos humanos. 5) Dado el actual desarrollo y uso de otras tecnologías de automatización relacionadas con el lenguaje, como la transcripción de voz (por ejemplo, programas como Dragon Naturally Speaking), ¿considera usted que la interpretación puede verse pronto también “afectada” por procesos de automatización equivalentes a los de la TA? Creo que la misma necesidad de comunicación impulsará también el desarrollo de este tipo de tecnología. He leído investigaciones de CSA (Common Sense Advisory) que indican que esta tendencia va en aumento. A medida que la tecnología de traducción mejore, aumentarán los recursos para desarrollar [la tecnología relacionada con la automatización de la interpretación]. La tecnología conocida como ASR (del inglés “Automated Speech Recognition”, que significa “reconocimiento automatizado del habla”) es hoy en día un eslabón con poca fuerza, pero está en constante mejora. En este artículo se afirma que el desarrollo de esta tecnología es inminente: al igual que en el caso de la TA, esta [tecnología relacionada con la automatización de la interpretación] evolucionará y mejorará, y los mejores intérpretes aprenderán a utilizarla para obtener ventajas a nivel competitivo. 6) Si los avances tecnológicos tienen generalmente el objetivo de reducir costes, ya que si se reduce el volumen de trabajo que lleva a cabo la mano de obra humana esto a su vez ayuda a reducir costes generales, ¿cuál debe ser entonces el mejor método de remuneración del traductor del futuro (especialmente en el caso de los profesionales autónomos), cuya tarea principal puede convertirse en la edición del contenido de traducciones heredadas que se hayan procesado mediante la TA y las memorias de traducción? Parecería entonces que el método tradicional de tarifa por palabra (o incluso por línea) ya no resultaría apropiado. ¿Por hora? Esta es una visión del mundo enfocada solamente en el trabajo de traducción tal y como se llevaba a cabo en el pasado: panfletos, manuales del usuario, interfaz gráfica del usuario y cierto tipo de contenido empleado en Internet. Todo ello contenido estático en su mayoría. El futuro tiene que ver más con el contenido dinámico y “palpitante”. En el futuro, la traducción más valiosa debe estar enfocada a la opinión de los clientes de ámbito global que formen parte de las crecientes e influyentes redes sociales. Las empresas de alcance global están generando contenido a un ritmo nunca antes visto. Los clientes de ámbito global desean y exigen poder disponer de contenidos que vayan más allá de documentos como manuales del usuario: quieren más y más. Las empresas de alcance global que tengan éxito serán capaces de proporcionar tal tipo de información. Microsoft es un líder que ya ha demostrado el valor que tiene esto y otras empresas no tardarán en emular a Microsoft. El futuro de Internet tiene que ver con muchas más lenguas que las que incluye ese círculo [de referencia para proyectos de traducción comúnmente] conocido como FIGS [del inglés "French”/francés, “Italian”/italiano, “German”/alemán y “Spanish”/español]. Siempre habrá demanda de traductores competentes que puedan ayudar a los compradores a tratar con toda esta exigencia de más contenido y, por tanto, los modelos de remuneración y negocios irán cambiando y evolucionando con la tecnología. En el futuro, el éxito a escala global va a requerir más contenido, más tecnología multimedia y más interacción de los clientes en muchas más lenguas. Me atrevo a conjeturar que la traducción se convertirá en una rama de trabajo gratificante, tanto desde el punto de vista económico, como intelectual, y que quienes se adapten a este [nuevo] mundo en ciernes lograrán la prosperidad en todos los aspectos.

Preparing Projects for Localized Desktop Publishing | Preparación de proyectos de localización para su publicación

I am publishing here in my blog the following article (the original in English and my translation) because, based on my own experience as a technical translator, I consider it an excellent summary of a lot of the issues that need to be taken into account whenever a client is interested in the “localization” (that is, translation adapted to a given market) of their documentation. I requested permission from its author to translate this article and publish it here (if you wish to publish somewhere else its contents or my translation, please request first the corresponding permission). All texts included in brackets in the Spanish version have been added by me, in order to clarify some of the terms used.   PREPARING PROJECTS FOR LOCALIZED DESKTOP PUBLISHING By John Watkins President and COO ENLASO Corporation www.translate.com If you search the web for guidelines for preparing projects for localization, you will find that most of the results focus on issues related to language content. For technical documentation and marketing materials, preparation and guidelines for the document publishing portion of your multilingual projects are equally important. Some customers new to translation and localization often assume that translated documents simply open up in the same format as the source files, requiring little or no manual adjustments. This is not always the case. Ways to handle text expansion, special language-specific fonts and manipulations of the final output are just a few of the decisions the customer must make. Some planning, including the development of styles and formatting guidelines, can greatly streamline the publishing process of localized files. This article includes a thorough checklist, with some examples, that covers the decisions you should make and the information you should provide before starting a publishing oriented localization project. These guidelines are aimed at projects with paper or PDF output as a final deliverable; software or website localization require somewhat different guidelines. GENERAL QUESTIONS TO KEEP IN MIND Is the project a new translation (from scratch) or an update to a legacy project? If it is an update, is it minor or major? Do you have the legacy source and target language files for any updates? What are the source and target language(s)? Specify the country or locale as well (e.g., fr-CA for Canada vs. fr-FR for France). Are style guides available for the source documents? What about for style guides for the localized documents? If not, can they be created? Do you need your localization vendor to prepare these for you? What turn-around time do you require for the localization effort, including the layout of localized files? Does the project include a customer review step of the translated content and/or formatting before final delivery? If so, has the internal review staff been identified? Have internal review turnaround times been established? Are changes to content or format expected during the project? This will influence the probability of change orders and the likelihood of project cost increases. Expected changes often have an impact on cost and delivery time. Do you have an existing Translation memory (TM) from previous projects? Are there localizable graphics (graphics that contain text that needs to be localized)? If so, can you provide the source graphic files for editing? Are there screenshots from software that appear in the document? Is the software localized (requiring the placement of localized screenshots in the target language documents)? Can you provide the screenshots or should the vendor capture the screens for you from the software? Are there specific fonts for the vendor to use, or should the vendor select appropriate fonts for the target languages? Will a single sourced document provide multiple outputs? (e.g., FrameMaker outputting PDF, HTML and Help via WebWorks or RoboHelp?). WHEN SENDING FILES TO THE VENDOR Clean folder structures: Provide an organized and clean folder structure with all source filesneeded to build the documentation (including any artwork). If possible, only include files that are used in the documentation for localization. It is always preferable if the files are “collected for output” including fonts, when they are in Quark or InDesign. Final source files: Provide source files as close to “final” as possible. Most accurate estimates are done on final files. If non-final files are provided, a vendor must re-analyze files prior to project start. Final output of source: Provide final deliverable output of the source language (typically a final PDF). This is referred to as the ’Go by’, and is used to assure consistent and high quality output for the localized documents. If final deliverable output specifications cannot be provided by the client, the vendor will create them and request the client’s approval prior to their use. Platform/Applications/Versions of source documents: Provide information on all platforms (e.g., PC, Mac, UNIX, Linux) and applications (e.g., FrameMaker, InDesign, QuarkXPress, Illustrator, PageMaker, WebWorks, Word, Excel, PowerPoint), and versions (E.g. FrameMaker 7.0 or 8.0, InDesign CS2) used in creating the sourcedocuments. PUBLISHING PHASE Pages: During documentation layout of the localized content, the vendor needs to know if the page flow has to match the source documents (e.g., whether page breaks must match, whether documents must end on an even or odd number of pages, and the maximum allowable page count. Text expansion: As a general rule, expect content to expand 20-30% for European languages. Asian languages tend to come closer to English in page length, but they typically require larger font sizes. Consider the preferred format overrides the vendor should use (e.g., kerning adjustments before adjustment of line spacing before adjustments to smaller font sizes). Fonts: Some languages require different fonts to render the characters correctly. If you have strict preferences for appropriate fonts, provide this information to your vendor early in the localization process. Note: if you have established a custom designed “corporate font,” there may not be equivalent style fonts for eastern European and Asian languages. Units/address/phone/currency: During the localization process, measurement units may be left in English units or need to be converted for the appropriate market. It is best if you provide the converted units, if they are necessary, to ensure compliance with your products. In a similar vein, you may need to provide alternative contact information (address and telephone numbers) and currency rates to be used in the localized documentation. Graphics: Preparing graphics for localization can save money during the localization process. Ensure that the graphics have “text layers” so that the text can be easily accessed and modified by your localization vendor. If these text layers are “flattened”, the vendor must budget for re-creation of text layers. Screenshots: Software documentation often contains screenshots from the software for reference. If the software is localized as well, then localized screenshots should be placed in the localized documentation. * If they will remain in English, how should the localization vendor handle User Interface references within the body text that directly refer to a screen shot? * If screenshots are localized, will the localization vendor recreate screenshots after the User Interface is localized or will you provide them? Part/artwork numbers for the localized documents: Make sure to provide project specific language codes, part numbers, and artwork numbers that replace related numbers from the source files. Style Guides: These publishing phase guidelines are often combined into style guides for each target market. Style guides provide a living document of these requirements, helping to ensure consistent quality over multiple projects for each target market. Working with a localization vendor experienced in the development of target market style guides improves your localization efficiency. DELIVERABLES File naming and folder specifications: You may have specific requirements for the delivery of localized documentation files (including file naming conventions, directory folders naming and structure. Communicate to the localization vendor about your specific needs. Change in application for deliverables: Clients may need to transfer their source content from an old application to publishing software more suited for your target language(s) (e.g., PageMaker to FrameMaker conversion) to better accommodate features like tables and cross references. Conversion to another application may also be necessary if the source application does not support your target language(s). For example: FrameMaker 8 doesn’t support Arabic or other bi-directional languages. In this case, InDesign would be a better choice. If document conversion is necessary, re-creation of the source template in the deliverable application is required prior to localization. Make sure to approve the final source template prior to starting translation. Translation memory: Using a translation memory can improve quality (by supporting consistency in the translation) while helping to reduce costs through the leveraging of previously translated text. Do you have specific translation memory format requirements? There are a number of tools available, though the largest market share belongs to Trados and the TMX translation memory format. PDF specifications: Documentation is often output to PDF format. Should PDF be Web or print ready or do you need both? If PDF must be print ready, make sure to provide resolution requirements, dpi and final page size specifications. If PDF is for print, should a registration mark be added? Are bookmarks required or other special features? When either party (customer or vendor) “guesses” what the best output specification would be, there is the risk of additional billable time required for recreating correct PDF output. SUMMARY Finding the answers to these questions eliminate many of the surprises that can occur in multilingual documentation production. Internal fact-finding to provide full specifications to your localization vendor is well worth the effort, saving time and money during the localization process. The tips provided here can help you avoid unwanted project delays or costly change orders.

Publico aquí en mi blog el siguiente artículo (el texto original en inglés y mi traducción), ya que, a partir de mi propia experiencia como traductora técnica, me parece un excelente resumen de muchas de las cuestiones que un cliente interesado en la “localización” (es decir, la traducción de documentos adaptada a un mercado concreto) debe tener en cuenta. Solicité a su autor la debida autorización para traducirlo y publicarlo en este blog (si desea publicar este artículo o mi traducción en otro sitio, por favor solicite primero la debida autorización). Los textos que aparecen entre corchetes han sido añadidos por mí, a fin de clarificar ciertos términos utilizados.   PREPARACIÓN DE PROYECTOS DE LOCALIZACIÓN PARA SU PUBLICACIÓN Por John Watkins Presidente y Director de Operaciones ENLASO Corporation www.translate.com Si hacemos una búsqueda en Internet sobre guías relacionadas con la preparación de proyectos para su localización [traducción y adaptación de documentación a un entorno e idioma determinado], nos encontraremos con que la mayoría de los resultados se centran en temas que tienen que ver con el contenido lingüístico. En el caso de documentos técnicos y de marketing de proyectos en varios idiomas, la preparación y las directrices de la parte relacionada con la publicación de los documentos tiene tanta importancia como todo lo demás. Algunos clientes que requieren por primera vez la traducción y localización a menudo dan por sentado que los documentos traducidos se podrán abrir en el mismo formato que los archivos en lengua original y que se requiere poco o ningún ajuste manual, lo cual no es siempre el caso. Por ejemplo, cómo resolver la extensión del texto, el uso de fuentes [tipos de letra] especiales para un idioma específico y el tratamiento del resultado final son solamente algunas de las decisiones que el cliente debe tomar. Un mínimo de planificación que incluya el desarrollo de directrices sobre estilo y formato puede optimizar en gran medida el proceso de publicación de los archivos localizados. Este artículo incluye una lista exhaustiva, con ejemplos, de cuestiones sobre las que los clientes deben tomar una decisión, así como sobre la información que el cliente debe facilitar antes del inicio de un proyecto de localización que va a ser publicado. Estas directrices están orientadas a proyectos cuya entrega final se realizará de forma impresa o como archivos PDF. La localización de software y de sitios web requiere directrices diferentes. CUESTIONES GENERALES A TENER EN CUENTA * ¿Trata el proyecto de una nueva traducción o es la actualización de un proyecto previamente traducido? Si se trata de una actualización, ¿es poco o mucho volumen? ¿Se dispone de los archivos en lengua original y lengua terminal de la versión anterior? * ¿Cuáles son las lenguas originales y terminales? También es necesario especificar las diferencias dialectales por país o región. Por ejemplo, “fr-CA” para francés de Canadá, en contraste con “fr-FR” para francés de Francia. * ¿Se dispone de guías de estilo para los documentos en lengua original? ¿Y para los documentos en lengua terminal? De no existir dichas guías, ¿es posible crearlas? ¿Requiere el cliente que su proveedor de localización prepare dichas guías? * ¿En qué plazo se requiere que se termine la localización, incluida la maquetación de los archivos localizados? * ¿Incluye el proyecto un proceso de fases de revisión (por parte del cliente) del contenido o formato traducido antes de la entrega final? En este caso, ¿cuenta el cliente con personal que lleve a cabo dichas tareas de revisión? ¿Ha establecido el cliente, a nivel interno de la empresa, plazos de entrega de las revisiones? * ¿Se realizarán durante el proyecto cambios en el contenido o el formato? Esto aumentará la probabilidad de recargos debidos a cambios, así como de costes del proyecto. Los cambios previstos generalmente influyen en los costes y plazos de entrega. * ¿Se dispone de una memoria de traducción (TM) de proyectos anteriores? * ¿Es necesaria la localización del texto contenido en imágenes o gráficos? En este caso, ¿puede el cliente proporcionar los archivos gráficos correspondientes para editar su contenido? * ¿En el caso de software que aparece en el documento, se dispone de capturas de pantalla del software? ¿Ha sido localizado el software cuyas capturas de pantalla se insertarán en los documentos en lengua terminal? ¿Puede el cliente proporcionar las capturas de pantalla o deberá el proveedor de localización obtenerlas a partir del software? * Para los documentos en lengua terminal, ¿requiere el cliente el uso de fuentes específicas o debe el proveedor de localización seleccionarlas? * ¿Se crearán varios archivos con formato diferente a partir de un solo documento original? Por ejemplo, un archivo FrameMaker de origen a partir del cual se generen archivos PDF, HTML y de Ayuda mediante el uso de WebWorks o RoboHelp. ENVÍO DE LOS ARCHIVOS AL PROVEEDOR Estructuración clara de directorios y carpetas: Es necesario utilizar una estructuración organizada y clara de todos los archivos de origen (incluidos los archivos gráficos) que se requieren para crear la documentación. De ser posible, es mejor que se incluyan solamente los archivos de la documentación que requieran localización. Siempre es preferible que los archivos estén debidamente “agrupados para su localización”, lo cual significa que incluyan las fuentes en el caso de programas de edición como Quark o InDesign. Versión final de los archivos en lengua original: El cliente debe proporcionar los archivos en lengua original en la “versión más final” posible. Es más fácil realizar una previsión de gastos basada en la versión final de los archivos en lengua original. Si el cliente no proporciona archivos en versión final, el proveedor de localización tendrá que volver a analizar los archivos antes de que empiece el proyecto. Documentos finales de referencia en lengua original: El cliente debe proporcionar documentos finales (usualmente un archivo PDF final) en lengua original. Sirven de referencia “a seguir” y se utilizan para garantizar uniformidad y alta calidad de los documentos en lengua terminal. Si el cliente no puede proporcionar especificaciones sobre los documentos finales de entrega, el proveedor de localización las creará y solicitará su aprobación antes de utilizarlas. Plataforma/aplicaciones/versiones de los documentos en lengua original El cliente debe proporcionar información sobre todas las plataformas de sistema operativo (es decir, PC, Mac, UNIX, Linux) y aplicaciones (es decir, FrameMaker, InDesign, QuarkXPress, Illustrator, PageMaker, WebWorks, Word, Excel, PowerPoint), así como de las versiones (por ejemplo, FrameMaker 7.0 u 8.0, InDesign CS2) utilizadas para la creación de los documentos en lengua original. FASE DE PUBLICACIÓN Paginación: Durante la maquetación del contenido de los documentos localizados, el proveedor de localización necesita saber si la paginación tiene que ajustarse a la paginación de los documentos en lengua original. Es decir, si los saltos de página deben coincidir, si los documentos deben terminar en un número de página par o impar, así como el número máximo permitido de páginas adicionales. Extensión del texto: Por regla general, se calcula que el contenido del texto se extiende de 20% a 30% en el caso de lenguas europeas. En el caso de idiomas asiáticos, el texto suele ocupar casi el mismo espacio que el idioma inglés, pero generalmente se requiere una fuente de tamaño más grande. Hay que tomar en cuenta los cambios de formato que el proveedor de localización debe hacer. Por ejemplo, los ajustes de espacio para fuentes antes del ajuste del espacio de interlineado, antes de los ajustes a tamaños de fuente más pequeños. Fuentes: Algunos idiomas requieren diferentes fuentes para que los caracteres se muestren correctamente. Si el cliente tiene preferencias estrictas sobre las fuentes apropiadas que se deben emplear, deberá facilitar dicha información al proveedor de localización durante el inicio del proceso de localización. Nota: si el cliente ha establecido el uso de una fuente personalizada y de diseño exclusivo “de la empresa”, es posible que no existan fuentes equivalentes en el caso de los idiomas de Europa del Este y de Asia. Unidades de medida/direcciones/teléfonos/divisas: Durante el proceso de localización, es posible utilizar las unidades de medida del idioma inglés [sistema anglosajón] o convertirlas a las unidades del mercado correspondiente de la lengua terminal. En caso necesario, lo mejor es que el cliente proporcione los equivalentes resultantes al convertir los valores a la unidad de medida de la lengua terminal, a fin de garantizar que se ajusten debidamente al producto del cliente. Asimismo, es posible que el cliente tenga que proporcionar la información alternativa de contacto (dirección y números de teléfono) y las divisas que se deben utilizar en la documentación localizada. Imágenes y gráficos: La preparación previa de las imágenes o gráficos para su localización puede ahorrar costes durante el proceso de localización. Asegúrese de que las imágenes o los gráficos tengan “capas de texto” que permitan al proveedor de localización acceder fácilmente al texto y modificarlo. Si dichas capas están “fusionadas” [no son accesibles], el proveedor de localización deberá incluir en su presupuesto el coste de la edición de las capas de texto. Capturas de pantalla: La documentación de software a menudo contiene capturas [imágenes] de pantallas del software, las cuales sirven de referencia. Si el software se va también a localizar, las capturas de pantalla localizadas deberán insertarse en la documentación localizada. * En caso de que se dejen en inglés, ¿cómo debe el proveedor de localización tratar las referencias a la interfaz de usuario incluidas en el texto de la documentación que hagan referencia directamente a una pantalla del software? * Si se localizan las capturas de pantalla, ¿deberá el proveedor de localización volver a crear las capturas de pantalla una vez localizada la interfaz de usuario o se encargará el cliente de proporcionar dichas capturas? Códigos de referencia de las piezas o del arte final: El cliente debe asegurarse de proporcionar los códigos de referencia específicos, como códigos de idioma, números de piezas o artículos, así como del arte final que reemplazarán a los códigos de los archivos en lengua original. Guías de estilo: Las directrices sobre la fase de publicación a menudo incluyen guías de estilo para cada mercado de destino. Las guías de estilo constituyen un documento práctico sobre este tipo de requisitos, ya que ayudan a garantizar la uniformidad de calidad en varios proyectos y para cada mercado de destino. La eficiencia del trabajo de localización aumenta cuando se trabaja con un proveedor de localización que tenga experiencia en el desarrollo de guías de estilo para un mercado de destino. ENTREGA DE ARCHIVOS FINALES Nomenclatura de archivos y especificaciones de los directorios y carpetas: Puede haber requisitos específicos para la entrega de los archivos de la documentación localizada, incluidas las convenciones de nomenclatura de archivos y de directorios y carpetas, así como su estructuración. Informe al proveedor de localización cuáles son sus requisitos específicos al respecto. Cambio de la aplicación para los archivos finales: Es posible que los clientes requieran la conversión del contenido en lengua original (en formato de una aplicación más antigua) al formato de una aplicación que se adapte mejor a la lengua terminal (por ejemplo, de PageMaker a FrameMaker), a fin de lograr un mejor ajuste de ciertas características del contenido, como tablas y referencias cruzadas. La conversión al formato de otra aplicación también puede ser necesaria en caso de que la aplicación utilizada para los archivos en lengua original no sea compatible con la lengua terminal. Por ejemplo: FrameMaker 8 no es compatible con los documentos en árabe ni en otros idiomas bidireccionales. En este caso, InDesign sería una mejor elección. * Si se requiere la conversión de los documentos, antes de proceder con la localización será necesario volver a crear la plantilla del documento en lengua original en la aplicación que se utilizará para los archivos en lengua terminal. Antes de empezar con la traducción, hay que asegurarse de haber obtenido la aprobación de la plantilla final de los archivos en lengua original. Memoria de traducción: El uso de una memoria de traducción puede mejorar la calidad de la traducción, ya que, por un lado, facilita la coherencia [terminológica] de la traducción y, por otro, ayuda a reducir costes gracias al aprovechamiento de textos previamente traducidos. ¿Requiere el cliente un formato específico de memoria de traducción? Existen diferentes herramientas de memoria de traducción, aunque el formato para Trados y TMX son los que se usan mayoritariamente en el mercado. Especificaciones para el formato PDF: A menudo se solicita la entrega de los archivos finales en formato PDF. ¿Deben los archivos en formato PDF adecuarse para su publicación en Internet o para su impresión o para ambos casos? En caso de que el formato PDF tenga que adecuarse a la impresión, es necesario saber de antemano los requisitos de resolución, puntos por pulgada (DPI), así como el tamaño final de página. Asimismo, ¿deberá agregarse una marca de registro? ¿Se requieren marcas de edición u otras características especiales? Si se deja en manos del cliente o del proveedor de localización “adivinar” las especificaciones sobre el formato de los archivos finales, se corre el riesgo de que se requiera (y se cobre) tiempo adicional para volver a crear correctamente el formato PDF final. RESUMEN Obtener la respuesta apropiada para cada una de estas cuestiones evitará muchas de las “sorpresas” que a veces tienen lugar durante la producción de documentación en varias lenguas. Merece la pena dedicar [tiempo y] esfuerzos a nivel interno en la empresa del cliente para proporcionar todas estas especificaciones al proveedor de localización, ya que con ello se ahorrará tiempo y dinero durante el proceso de localización. Las recomendaciones que se proporcionan en este artículo pueden ayudar a los clientes a evitar retrasos en los proyectos o costosos recargos debidos a cambios.