Friday, May 8, 2015

Compensation for Post-Editing MT All Comes Down to Compromise

This is a guest post to provide an alternate and independent perspective on issues I have discussed in this blog. It is unedited and may or may not be consistent with my own views, but I think it is always useful to hear other views on these issues. I invite others who may wish to share their opinions on these MT related issues to also voice their opinions, especially those with very different views. I have added very few clarifications (in italics). My own views are outlined in this recent post and this older post which is still one of the most popular posts on this blog.

With thanks to @MattBramowicz for this submission.
If you ask a professional translator, chances are they will tell you that they’ve been requested at one time or another to provide post-editing or revisions for machine-translated text. When this happens, more often than not many translators either begrudgingly accept or refuse the project altogether.
This can be attributed to 2 main reasons:

  1. The machine-translated text is not very accurate, to begin with, resulting in sometimes more work and time to decipher the text than it would take to translate it outright.
  2. The rates to post-edit are oftentimes significantly less than the translator’s normal translation rates.
As technology progresses, so do the methodologies for translation practices. Of course, technology is not just instantaneously perfect. It takes time to develop, improve, and advance until it is if not perfect, at least to a level that is efficiently satisfactory. At this moment in time, machine translation is in its “needs improvement” phase. While the accuracy has come a long way since its inception decades ago, with some languages being translated more accurately than others, it has still not reached a level that would be deemed “good”. However, for it to improve, it must still be a technology that is utilized so that there is a need to warrant the time and energy it would take to improve its process.

While many translators may refuse to provide post-editing services, the fact remains the need for the service is at an all-time high. Whether there are platforms created for mobile app translations, website content management systems, or even standardized document content formatted for machine translation, there are many resources that are utilizing the ubiquitous nature of web development to promote the post-editing or “hybrid” method of translation services. Therefore, translators could and should be more open to the process.

That being said, in no way should translators be “exploited” for their services either. There should be a compromise between translation service providers and translators to ensure fair pay is given for quality work.

For a compromise to take place, both sides must understand the other’s concerns and needs. Since we’ve already established the translator’s concerns, let’s go over the translation service providers’:
Hybrid translation is usually offered as a cost-saving method to the client to provide translation services for items that either would not have been translated in the first place
  1. due to the content not being of utmost importance, or there is too much text and the client wouldn’t be able to afford professional translation for it.
  2. Hybrid translation is usually offered as less than perfect translation method, not as good as professional translation, but still much better than mere (raw) machine translation.
  3. For reasons 1 and 2, the rate they charge the client is much less than for professional translation, so the budget for translators is much lower.
Now that we know both sides, the trick is to find a solution so that all parties’ needs are met.
For starters, we can look at post-editing translation jobs as opportunities for translators to make a little extra money on projects that for all intents and purposes came about thanks to hybrid translation being an option. Therefore, translation service providers can categorize these types of projects as such, and create a separate list of available translators who are willing to make themselves available to work on these from time to time as a means to earn some extra income. That way, TSP’s are only contacting translators who are willing participants in these projects. Likewise, TSP’s can and should keep more than the usual amount of translators on file for these types of projects, so if a translator is too busy to work on a certain project, they can decline without feeling pressured into taking it because the TSP has no one else available.

Second, while translators will have to work at a lower rate than ordinary for these projects, they should still be given a rate commensurate with the work involved. Translation Service Providers should set a rate at or close to the proofreading rate of translators. While the margins may not as be as profitable as professional translation service orders, the quantity of orders placed by clients should make up for it, provided the outcome is good enough quality that they want to order again. The best way to ensure that is the case is to pay the translators enough so that they feel compelled to do the best job they can. 

Also, deadlines should be reasonably set so that translators are able to work on these projects in between their traditional and more lucrative projects. However, many TSP’s tend to offer hybrid translation as a “speedy” translation service. While this can still be true, especially when formatting is taken out of the equation, they should still set reasonable deadlines with the client. If the client is in need of a rush delivery, TSP’s can offer a rush delivery surcharge option to the client, and pay the translator a little bit more for that particular project. If that isn’t an option, the project could be split between 2 or more translators in order to meet the deadline on time without putting too much strain on one translator.

Another option is for TSP’s to set up some sort of reward package for translators who consistently accept these projects and provide good services. For instance, for every 15 projects completed, let’s say, the translator receives a $50 bonus. While the amount may be relatively small, it still provides an incentive for the translators to accept as many projects as possible and it shows that the TSP appreciates their loyalty and hard work. 

Perhaps TSP’s could also pay translators an hourly rate instead of a per word rate. That way, for segments of text that are exceedingly difficult to decipher the meaning from, the translator is compensated fairly for their time. The only caveat for this option is that the end-user is usually charged a per word rate (since there’s no way of knowing how much time to charge them upfront). So there is a potential for the project to go over-budget if the time it takes to complete the translation is much longer than anticipated and surpasses the per word rate charged to the client. A possible solution to that issue would be to have the client agree to a clause that states the price they are given upfront is simply an estimate, and the final amount may be more. This is the traditional practice for services like auto repairs, home repairs, etc., and most other service-based industries, which translation services are most certainly a part of as well. However, for some reason, many clients don’t view it as such and are reticent to agree to such terms upfront.

While post-editing jobs may pay less than traditional translation projects in terms of word count, in some cases (depending on how the platform is set up and quality of machine translation), the process of post-editing can be completed much more quickly, and therefore result in a higher rate per hour than initially anticipated. In such cases, depending on the source text quantity, a translator can earn close to, if not the same rate, as traditional projects in a given time span. 

In the end, it all comes down to how the platform and post-editing process is set up by the TSP and the level of cooperation that can be achieved between the TSP and the translators.

About the Author
Matt Bramowicz is a content writer and graphic designer for Translation Cloud LLC (, a leading professional translation company located in Jersey City, NJ. You can follow him on Twitter: @MattBramowicz 


  1. An unbelievably condescending post.
    I can respect opinions that are different than my own, but this post is just one big insult to intelligence and the translation profession.
    This is clearly written from a very narrow perspective, and as a SEO bait to promote the commercial interests of a company, but I'll response anyway.
    1) We all should bear with the technology until it matures: Technology is an enabler, not the goal. If anything, MT is a tool that should help translation professionals do their work better. It is here to serve and help the professionals, not the other way around.
    If you want to sell a tool to translation professionals, show them how it makes their lives better. Any attempt to convince them that the abuse and exploitation some try to rope them into are actually in their favor is an insult to intelligence.
    2) ...the fact remains the need for the service is at an all-time high: Arbitrary statement without context. Is it really a fact? Is this data even valid? Where it is collected from and by whom? Does it take into account text sent to raw free engines such as Google and Microsoft Translate (likely by individual people translating bits and pieces on Facebook and shopping websites)?
    3) Therefore, translators could and should be more open to the process: So you are basically saying that you (as a non-translator according to your bio) know better than the translation professionals what is good for them and how the work should be done. Extremely arrogant and condescending statement. If you want to sell a tool to a professional, show them how it makes their lives easier/better instead of telling them they are closed-minded idiots who don't know anything.
    4) Referring only to agencies as Translation Service Providers: Translation agencies are not the only legitimate translation service providers in the market, and translators don't work exclusively, or even at all, with translation agencies. Individual translators, small teams of translators, and specialized translation practices are as much, if not more, of a translation service provider as the intermediaries calling themselves LSP/TSP/Whatever.
    Agencies and technology developers don't own the profession, are not authorized to speak on its behalf, and don't hold the keys to the market. If anything, they have some insight into the market segments they serve, and these segments are not the entire market.
    5) We can look at post-editing translation jobs as opportunities for translators to make a little extra money on projects that for all intents and purposes came about thanks to hybrid translation being an option: Translation professionals are not employees, hobbyists, day works, or moonlighters (yes, there are those as well, but if you are looking for these types of translators, at least be honest about it and don't use the term translators). Translation professionals are business owners (i.e. employer + employee in case of sole proprietor) working to build a sustainable business (even if for many it is not a business in the traditional sense but more of an sustainable career). Translators are not after some pocket money. If you are after the moonlighting/tax avoiding/amateur translators with a day job/living off of their parents or spouse, be clear about it.

  2. 6) …while translators will have to work at a lower rate than ordinary for these projects: Another arbitrary statement. Says who? On what grounds this lower fee should be offered? Just because MT proponents want to pocket more money themselves or undercut someone else? Translators are not employees (this post is clearly not aimed at those who work in-house), they do not sell their time and they do not sell words - they sell results based on their expertise and experience. Any attempt to claim that using MT requires less expertise or is a substitute for experience is not serious, to say the least.
    7) Translation Service Providers should set a rate at or close to the proofreading rate of translators: Again with the arbitrary statements. Translators are not the agencies' employees and therefore set their own prices. And even if for the sake of argument we assume that agencies set the rates, but making this statement you assume that all MT engines are equal, and that all translation projects inherently require the same level of effort. Both are clearly wrong, and therefore you above suggestion is nothing more than unethical and abusive.
    8) While the margins may not as be as profitable as professional translation service orders, the quantity of orders placed by clients should make up for it: Snake-oil argument, if I ever heard one. Simply a bad business advice. Again, translators are not employees, but running their own businesses. Project management and administration require time and effort, or in other words - overhead. The argument that you can offset lower margins by driving volume is an economic fallacy that ignores the cost (in time and money) of this overhead. This is not the path to a sustainable business; this is the path for modern slavery in a desperate attempt to stay afloat for yet another day.
    If things are so great for the commoditizing agencies, why won't they hire in-house translators and be done with it? Trying to roll all responsibility and liability into naïve and/or unsuspecting contractors, and proceeding to add insult to injury by giving them a bad business advice although the agencies know perfectly well that the they (the translators) are the ones having to pay the taxes, health insurance, retirement plan, and at least make a decent leaving just is exploitive, abusive, and unethical. Shame on you!
    9) Another option is for TSP’s to set up some sort of reward package for translators who consistently accept these projects and provide good services. For instance, for every 15 projects completed, let’s say, the translator receives a $50 bonus: Yes, translation and children who had finished their chores is the same business model. Highly inappropriate and insulting.

    1. Shai, thank your for taking the time to list your objections. I will let Matt respond directly if he wishes to.

    2. Number 10 got cut off:
      10) In these type of discussions those trying to promote post-editing often like to refer to the supply and demand, and economies of scale (both mentioned in this post). By the admission of the author, there is a shortage of post-editors, while the demand for MT is at all time high. The supply and demand model determines that if there is high demand but short supply, prices increase (even for commodities, which translation is not) and not decrease as the MT proponents, who coincidentally stand to gain from selling this technology, recommend the post-editors to do in general and in this post in particular.
      Simply a bad business advice.

      This post is not about exploring the PEMT model, where it fits in the market (and where it is not), and discussing best practices. It is a SEO/propaganda piece for the purpose of promoting a very specific and narrow intermediary agenda.

  3. Post-editing consistently takes far more time than translating.
    Post-editing often involves a complete rewrite or retranslation.
    Post-editing cannot be completed quickly, unless you're willing to let shoddy work go out.
    Technology needs to improve significantly before machine translation produces acceptable results.
    Why should translators fund the improvement process by accepting lower rates?

  4. Matt Bramowiczs, let me tell you something: you are a perfect moron who shows no respect for translators'work.Who do you think we are????
    Signed: a very furious translator

  5. The main reason why I, and I think many of my colleagues, shy away from editing machine translation is that it seems non-creative. More like regular editing (or proofreading or whatever other term you like to use for it), this is the kind of work that requires a different mindset: less writer-like and more librarian-like. With that comes a personality that is less inclined to want to work as a freelancer. As is so often the case, the technology-infatuated nerds who succesfully invent this technology know more about machines than they know about people, which causes them to overlook the psychological aspects of it all, aspects that are in fact so important that they can make or break the entire project. If you do not make machine translation editing in some form or another palatable for the people who have to do it, you may not have any machine translation to sell at all, never mind how well-designed the technical side of it may be. In practice what I expect to happen, is that edited machine translation will become more common in those areas of specialization that are already somewhat dull and repetitious, like technical translations (manuals, safety instructions) and translations that need not be all that accurate because no consequential decisions depend on them.

    1. Thank you for your comment. I think you get to the heart of the matter and ongoing communication between the agencies, the translators and the MT developers will be key to producing palatable and effective implementations of this technology.

  6. Yawn. Yet another post written as SEO bait by someone who is clearly not a translator. Know how I know?

    "For starters, we can look at post-editing translation jobs as opportunities for translators to make a little extra money on projects that for all intents and purposes came about thanks to hybrid translation being an option."

    No self-respecting translator needs to "make a little extra money." A professional worth his or her salt is just that, a professional... and commands rates as befits a professional.

    I don't know, but maybe it's just me. Because I sure as hell don't have time to check MT documents when I have clients who pay very well for my translation services (which, by the way, require a level of creativity and cultural mindedness that machines will never be able to duplicate).

    Which brings me to my next point: is the author even aware that post-editing MT junk is oftentimes more time consuming than just translating it correctly the first time?

  7. c) There is nothing arrogant and condescending about saying translators should be more open to an emerging technology. It's a belief that people should not be close-minded to technology that's in development. If you don't want to be, then don't. I really do not understand the vitriol.
    d) When did I ever say translation agencies were the only service providers? I simply referred to them as service providers because they are.
    e) Regarding working at a lower rate, in many cases "hybrid" translations are offered at a lower rate to the customer. Therefore, it stands to reason that the rate paid to translators must be lower. As a business, you can't pay more for your supply than you are selling it for. Very basic rationale.
    f) I think most of this backlash comes from a prejudice against translation agencies, and not what I wrote in the article. I want all translators to be compensated fairly for their work, because at the end of the day, they are the ones providing the service. However, I have a job to do as well working for a translation agency, and from my experience with MT in this role, this is the perspective I have received. I am not wrong for this perspective, as I have lived it and work in it everyday. If you disagree with anything I say from the perspective of a translation agency, that is great. I'd love to hear your perspective on it. If you have a perspective as a translator, that's great as well, and would love to hear your perspective on it as well, instead of name-calling and abuse which speaks volumes about who in this conversation is the actual narrow-minded and condescending party.

    1. Matt, I'm afraid you set yourself up for the vitriol because of the condescending way you phrase some of your points. For example, "translators could and should be more open to the process". Your only reason for this is because you (as an MT service provider) need us. And because you need us, we "should and could" jump. You are unable to put yourself in our position and offer any reasons which take our perspective into account.
      Secondly, you see it as an incentive that PEMT offers "opportunities for translators to make a little extra money". That sounds almost like pocket money. If I (as a "little boy" translator) want some extra cash, how can I go about it? I could get a job shining your shoes, or I could post-edit your MT output. However, as a translator I am not an adolescent looking for extra money, I am an expert running a business and providing a service. And if I make an informed decision not to be involved in PEMT and actually dare to voice my opinion, you accuse me (and others) of "prejudice against translation agencies".
      The uncomfortable truth is that you are looking for translators who will do what is often an unpleasant and exploited task (and a task which can often take more time than a "proper" translation), but because of the budget constraints imposed by your company's marketing promises, you are forced to underpay the professionals that you want to engage. I can sympathise with the dilemma this places you in, but please do not vent your anger on those who publicly disagree with your points.

  8. I can only agree to what my esteemed colleagues Shai and Audra wrote.

    MT only works in the minds of people who think translation works like "replace every word in language A with a fitting word in language B".
    This, however, is as far away from reality as can get.

    The way translation really works is (very simplified) "read text in language A and analyse thoroughly with regard to content, style, target group, and cultural references. Then write a text in language B that keeps the content and style as far as is appropriate for your target group within their culture, and where it is not, rewrite so your target text has a similar effect on the target group."

    And now, please, show me the machine that is capable of doing this!

    1. Thomas, you are absolutely right. I've been in this business long enough (and dealt with plenty of customers who think that translation is just replacing the words with a fitting word), to know that is not how you get an accurate translation. My main point is simply that, while yes at this point MT does not make it easy for translators in most cases, as it is not the greatest way to contextualize text, culture, localization strategies, etc. that is required in most professional translation methods, it is still a worthy tool to continue developing. The internet and computerized methodologies are not going away. Everything that can be is now being digitized, and there is no reason that translation should have to stay in the "dark ages" (so to speak). Is it at the ideal stage now? Of course not. Nobody is saying that. But to discount it entirely is a fool's errand.

    2. Thomas, I think an assumption that many translators make is that MT is considered a wholesale replacement for translators. The only places where this might be the case is for content that simply would not be translated if it were not for MT.

      And you are right MT is very literal in an idiot savant kind of way -- if the right data has been used to train it it can be very useful in translating CERTAIN kinds of content and help to get work done more efficientyly. It should never be used where linguistic finesse or cultural context and understanding is required. I would suggest that much of what is on corporate websites describing mass market products is fair game since it is material that often has a shelf life of months and then is discarded forever e.g. material about printers, computers, gardening tools, etc.. A lot of this material is not worth the care and attention that a competent translator would bring to the table.

  9. This comment has been removed by the author.

  10. Well this post certainly garnered a lot of hate. As any situation
    where you have individuals in different roles, there will be
    different perspectives. I will try to address some of the comments
    below, as I think a lot of what I said was misinterpreted, as the
    negative comments seem to contradict what I even stated in the first

    1) Audra - Many self-respecting translators are looking for work
    and do need to make "a little extra money." Know how I know? I get
    literally 150 applications a day from translators looking for work.
    Many of them may not be that great, but many of them are also
    professional translators who are worth their salt because I've hired
    plenty of them over the years and they've consistently provided
    quality work. Also, there are plenty of new translators starting
    out who still need to build their client-base and don't have
    numerous projects to work on every day. Your situations is the not
    the same for everyone.

    2) Shai - This post was not meant to be condescending at all. I'm
    sorry if you felt that way, but over and over again I stated how
    translators should not be exploited for their work and that
    compensation should be fair. What did you find so condescending
    about that? Yes, this article is written from the perspective of
    someone who works for a translation agency, not from a translator.
    Of course our perspectives would be different, since we play
    completely different parts in the translation process. That's the
    great thing about having different perspectives, as it allows people
    to see the entire picture, not just the one from their point of

    a) Regarding technology, that's just the way it is. GPS was shoddy
    when it started, giving roundabout directions, sometimes telling you
    to go down streets that are dead ends... If people took your
    attitude, expecting it to be perfect right off the bat and give up
    on it, then we wouldn't have an improved GPS system we do now.
    Every antibiotic created has a human-trial phase, with mixed
    results. We lost many dogs, monkeys and even humans in our space
    program that was being developed in the 60's. Again, that's just
    how it works. Nothing is perfect from the get-go. All I'm doing is
    encouraging translators to not give up on it at this point when we
    all know, and I've admitted to multiple times in my article, it's
    not a perfect technology at this point, especially for the

    b) "the need for the service is at an all-time high: Arbitrary
    statement" This is not an arbitrary statement, since when in the
    past has it ever been higher? If you can give me a time in the past
    when the need for the service has been higher than it is today
    (which doesn't stand to reason since it's a progressing technology).

  11. Machine translation is a technical mechanism, machine translators have been used for multiple translations. My Personal opinion, machine translation is not reliable for professional assistance. Majority of translating agencies gives the services to consumer for human translators.
    Translation of Documents

    1. Who cares about your "Personal opinion" if it is not in any way original? That added just about NOTHING to the discussion.

  12. Hello there

    As we all know , translation is being popular day by day. we all need translation that make the things easy to

    I love so much your blog.
    Thanks for sharing here.

  13. Hello there

    As being a businessman , i can understand how much it is more valuable for us. I have translation business and i always being updated everywhere to get the best for my business.

    Thanks for sharing this article with all of us , it might be helpful for us..
    Very Nice..!!!Keep it Up..!!!

  14. Hello there

    I read this blog that relates to translation , it is really amazing to read. Professional/Document Translation has been very necessary because every processional needs to translate their document in easy understandable Language.
    Very Useful info.
    Thanks for sharing with us.

  15. A lot of Professionals are following your blog that proves it that how much valuable your article for them.