Selecting a Language Service Provider (LSP) partner can be overwhelming, especially if you’re new to translation or localization and unsure which questions to ask. So to get started, you’ll first need to determine the exact scope of your localization and translation needs. To do this you’ll need to understand the right questions to ask that will identify the capabilities of prospective LSP’s in relation to your needs.

To help aid you along the process, as well as increase the likelihood of selecting the right LSP, we’ve compiled a list of questions your company should take into consideration before and during the vendor selection process. These questions will arm you with the necessary information you’ll need to select an LSP that has the right expertise to help your company accomplish its localization and translation goals.

Questions to ask yourself:

1. What type of internal support and knowledge does your company have in preparation for translation and localization? Do you have multilingual staff that can review the translated content to assess it’s quality and liaise with the LSP to define language style, terminology?

2. How much are you willing to spend annually on localization? What kind of resources and how much man power are you willing to devote to localizing your product, service or website?

3.  Which services does your company require from the LSP? Simple translation or full scale localization? Editing and proofreading or full linguistic review? Do you need additional support like multimedia production, DTP and software engineering?

4. What are the core languages that you plan to localize into right now and in the future?

5. How much translation volume do you anticipate? How often will you need it? What kind of file formats will you be using?

6. Do you use a content management, document management or translation management system that will need to integrate with the localization process, and do you need your LSP to work with the system you have in place?

7. What type and level of customer service does your company require? Do you need multi-lingual, in-country reps to deal with multiple dispersed teams working on different languages individually, or will all correspondence be handled through a single point of contact?

Questions to ask your potential vendor:

1. Ask the LSP to verify their basic financial data to understand their commercial stability. Ask for annual reports and/or audited financial statements.

2. Who are their previous and current clients? Ask the vendor to provide a list and don’t be afraid to ask for references. You should choose to work with an LSP that has knowledge and experience in your industry.

3. What is the LSP’s capacity? Is your localization output volume manageable for the LSP? What will the turnaround/delivery time be? What’s the maximum number of words/language pairs they can deal with weekly/monthly?

4. What types of content is the LSP capable of translating? Ask about specific document or file types and their experience with them. For example; training material, FAQ’s, knowledge bases, legal and technical documentation, marketing material. Can one LSP manage everything or would you need multiple vendors?

5. How does the LSP maintain quality and service levels? How often will they review your account? Yearly? Quarterly, if at all?

6. Ask for an example quote on a sample project. This will leave you better able to estimate and gauge future project costs and ensure that there are no hidden costs in the future.

7. How will project management be handled? Where will your LSP’s project managers be located and how many would be dedicated to your company?

As your company’s translation and localization requirements change and demands increase, your need for a a long-term partnership that allows for growth and flexibility will become essential.  We hope these questions and considerations will help you establish and maintain a good working relationship with a translation and localization vendor that is best suited to your unique needs.

Want more advice and best practices for choosing a Language Service Provider? Request a  free consultation with one of our team.

Related articles

• Can We Expect all Pervasive Machine Translation in 2012? (milengo.com)
• LSPs are not just agencies! (translationmusings.com)
• Are Translation Agencies Really LSPs (Language Service Providers) or LSRs (Language Services Resellers)? (patenttranslator.wordpress.com)

Guidelines for Selecting a Language Service Provider

MT for Consumers.

MT in the consumer market has the potential to become a universal technology since more people now have constant internet access due to the prevalence of smart phones. GPS and Google Maps have already established location-based services as a ubiquitous technology, enabling a host of innovative applications and services, from simple apps in gadgets to truly global systems. So will machine translation become a universal service as well?

Unless English really does become THE world language, consumers will need multilingual text/voice interfaces on their gadgets and for their apps. Even on the Web, the dominance of English is diminishing whilst demand for free on-the-fly machine translation is growing. On-the-fly MT can become a great facilitator of truly global Web2 and community services, whilst also offering a cost-efficient solution for multilingual communication within, and between, global businesses (translation of chat logs and emails “at Internet speed”).

Couple this with reliable speech recognition and synthesis (“Siri” on the iPhone and GT’s “listen” button gives us some indication that the technology is already there), as well as the deployment of artificial intelligence “in the cloud”, and we will be able to communicate ever more with people and software in many languages.

Is MT-enabled multilingual communication going to become just as prevalent as GPS-enabled services and applications? Well there are serious privacy concerns when all our communication is processed by an invisible translation engine, but then the same concerns apply to GPS-enabled location-awareness (all our movements can be tracked).

Another factor limiting the widespread use of this kind of technology is the cost. Taking your cell phone abroad and using it to translate real time using voice input requires a lot of data to be sent back and forth, which invariably incur costs, especially when abroad. So until roaming costs for smart phones are reduced to an acceptable level, we probably won’t see the widespread adoption of real-time MT as a translation tool for the masses.

MT for Business.

Traditionally, the translation industry has served 2 communication models. Business-to-business (B2B) communications in different languages between (or within) businesses, and business-to-consumer (B2C) communications, both of which involve the localization of user documentation, web content, software, marketing collateral, product labeling etc. into many languages.

While there are currently no established business models, MT can also facilitate consumer-to-business (C2B) communications in the customers’ native languages directly to the business, instead of going through local offices, reseller networks, etc. (e.g. user/market feedback, customer/product support requests). Consumer-to-consumer (C2C) is another model, enabling multilingual communications between consumers on user forums and communities, other community initiatives, or multilingual Web2 in general.

The Players.

The Giants (Google, Microsoft, Apple…) will cover C2C needs with pervasive cloud-based services, with MT (and possible AI) running in the background to connect Web2 users, gadgets, apps and global services and to reduce language barriers. They will also offer B2C and C2B (market intelligence) solutions for businesses which will become commonplace within the market.

Communities.

Community initiatives have been essential to improving open-source  MT solutions (e.g. Moses). Some seek to serve C2C on a community basis (e.g. itranslate4.eu). It is unreasonable to expect overall standard adoption of MT without such community initiatives – A David with brainpower can be just as important in shaping the future as a Goliath with market power. However, there is little direct funding for this meaning that development is likely to be slow.

Businesses.

IT businesses with access to the appropriate technological resources will be able to leverage cloud-based MT like Google Translate API v2, and build translation portals and workflow solutions using open-source or relatively cheap software available on the Internet, integrating MT into their business processes.

The Language Industry.

LSPs like Milengo will continue to gradually integrate MT into traditional B2B and B2C processes. For example, software UI localization could be an easy target for MT (there are even solutions to translate software UI on-the-fly into languages not covered by localization). Customized MT engines will be developed for very specific domains e.g. technical, automotive, medical, and will be combined with extensive, pre-translated product terminology and specific language pairs (possibly merging RBMT with SMT for languages with complex grammar or Asian languages).

Companies that require a secure, controlled MT environment for confidentiality and data protection will be served well by LSPs that offer MT with post-editing as an alternative or complementary approach to “Translation-only” customer needs.

New Business Models.

C2C does not seem to be a viable service for LSPs, since it can easily be covered by free cloud-based services from “giants” and (for specific domains/languages) by community initiatives. C2B (translation of customer-generated multilingual content for businesses) could be interesting as a new revenue stream.

LSP added value:

LSPs will ramp up promotion of MT and the value proposition will be further defined to cover:

• customized MT engines
• that are operated in a closed environment (confidentiality)
• seamless integration of optional human linguist input
• post-editing and QA
• translation (for critical content)

Problems.

The biggest obstacle facing LSPs that wish to develop MT solutions is a lack of adequate bilingual corpora to train SMT engines. With high volumes of customer-generated content, it also becomes difficult to identify exactly what needs to be translated and businesses most probably need aggregated/filtered results. This could be based on frequency of keywords (potential AdWords) in customer-generated content. This also lends itself nicely to patterns and trends for business intelligence (data mining) to identify a hierarchy of needs, rather than translating everything in one go.

Business potential.

An MT solution is more viable where product terminology is well-defined and available translation memories from former localization projects can be used to train the MT engine. MT could enable more centralized customer support instead of relying on local offices or reseller networks. It could also be used to analyze and evaluate customer support information globally and help develop solutions to product issues reported by the customers themselves (e.g. on user forums).

Conclusion.

MT is a big topic and set to grow. It will continue to cause controversy amongst translators who are concerned for their livelihoods, whilst it is viewed cautiously by businesses worried about cost and quality. However, we are now entering a period of development where the advantages are beginning to outweigh the disadvantages.

More and more business cases will be identified for the use of MT; on the consumer side advances in mobile technology and reductions in cost of ownership will put MT in the hands of more people than ever before. Will there be a watershed moment where MT becomes as ubiquitous as Google Maps? Maybe, but one thing is for certain, MT will play an even more important role in 2012 than it ever has, and we’re excited to see what the future brings.

Related articles

• Localization Trends: 2011 review Part 2 (milengo.com)
• Statistical Machine Translation: Building Your Own Viable Engine (milengo.com)

Can We Expect all Pervasive Machine Translation in 2012?

This is the position which Blue Coat Sytems found themselves in, as they were preparing to launch their multilingual WAN optimization marketing offensive. Central to this campaign were two taglines: “Free the WAN” and “Make it Blue”. These taglines would form the foundation upon which all accompanying collateral and whitepapers would be conceptualized and built.

The idea behind “Free the WAN” is a clear allusion to freedom; the freedom from network congestion that Blue Coat solutions bring, thereby making the network “free” to deliver the content you need. Blue Coat developed a solution to reduce traffic demands on Wide Area Networks by removing redundant components, and compressing a large amount of data to make the whole process that much more efficient. In this way Blue Coat can be said to have set your WAN free. The second tagline: “Make it Blue” highlights the Blue Coat brand name and color, which is further consolidated by the visuals accompanying the text of graffiti cans spraying the taglines for this campaign.

So far a seemingly sound and well-conceived campaign, but only for English language speakers. Blue Coat recognized that a literal translation of these taglines into the nine target languages would not be effective, for a direct translation would not allude to the elements of wordplay and metaphor, and thus lose their impact for the new target audiences. What they required was a complete re-write of the taglines to both mirror the communication objectives of the English original and make the statements relevant for the local target audiences. In other words, they required a transcreation service.

Transcreation is particularly suited for tagline text and requires a talented pool of bilingual copy-writers who ensure that campaign-critical messages are rendered consistently across languages whilst remaining sensitive to local requirements. Milengo offered Blue Coat a comprehensive transcreation solution tailored to their plan for the campaign. Once Blue Coat and Milengo had drawn up a joint creative brief, we set up a team of experienced marketing copy-writers who were asked to perform the following:

• Provide three alternative transcreations for each of the two taglines, along with a literal back translation in English.
• Provide a cultural commentary for each option. This commentary would include the reasons for the choice of translation, the associations for the target market, use of literary devices such as humor, imagery, alliteration, metaphors and similes etc. In other words, just the kinds of work linguists enjoy.
• The three options would include:
  • One option that is very close to the original English.
  • Another that represents a creative leap
  • A third option as a middle ground of sorts.
• Copy-writers were asked to indicate their preferred option with a justification as to why.

After these taglines had been transcreated into the nine EMEA languages, Milengo coordinated closely with Blue Coat’s in-country language reviewers to decide which taglines should be used for each language. Once this was completed, the rest of the translations for the accompanying marketing collateral were literally “a piece of cake”.

Related articles

• Can We Expect all Pervasive Machine Translation in 2012? (milengo.com)
• Localization Trends: 2011 in review, Part 1 (milengo.com)
• Localization Trends: 2011 review Part 2 (milengo.com)

Transcreation: Adapating your Brand Message Across Cultures

First up is Jibbigo, available for iOS or Android. It’s a bilingual translation app that translates speech in almost real time between 9 language pairs. While it’s not the cheapest app out there, $4,99 on the App Store and Android Market, it works surprisingly well as long as you’re in a reasonably quiet environment. To translate, you fire up the app, select your language pair, hold down the record button for the language you’re going to speak and then wait to hear the translation. Once you’re finished speaking, Jibbigo takes a moment to think, then speaks the translation and displays accompanying translated text. You then repeat the process for the person you’re having a conversation with.

The great thing about Jibbigo is it works offline, no data connection needed. Each language has a database of around 40,000 words, and while it’s not correct 100% of the time, it’s a great app to help make sure you’re understood when you need it. You’ll need an iPhone 3GS or above or a current generation Android device with at least a 1ghz processor to run it.

Next up is Word Lens, another unique translation app that provides real time translation, but this time with printed text. Currently Word Lens is only available on iOS but an Android version is also slated to be released sometime in the future. The concept is surprisingly simple, just point your iPhone’s camera at a piece of text and it magically translates it, overlaying the translation on the screen. The effect is pretty remarkable as you’ll see in the video below. Like Jibbigo, Word Lens doesn’t need a data connection so you don’t need to worry about huge data roaming costs when you use it abroad.

Unfortunately the biggest let down is the limited language pairs, currently Spanish to English and English to Spanish. When tested against Google Translate, the accuracy wasn’t quite as good, however with Google Translate you have to speak or type in the phrase you want translated. Word Lens also seemed to get confused when the image it was capturing was busy with other information, so the cleaner and simpler the text you want to translate, the better.

With all this said, the ability of Word Lens to provide real time translated text from images is pretty impressive and we hope to see the technology improve in the future with better accuracy and more language pairs supported. If you’re off on your travels to Spain or South America and you don’t speak Spanish, this is definitely a useful app to add to your iPhone.

Finally we move to everyones favourite go-to translation service on the web, Google Translate, and their applications for both iOS and Android. Google translate has been around for a very long time now, and much of its early bugs have been ironed out. With 60 languages now supported the Google Translate mobile apps also feature speech recognition, the ability to star or ‘favourite’ translated phrases, and also SMS text message translation.

As part of a new update Google Translate now also supports conversation mode, similar to Jibbigo, where where 2 users can each take it in turn to speak and listen to translations. SMS translation is another nice addition, and overall as with nearly all Google Apps these days, it’s extremely intuitive and easy to use. Although not 100% reliable, slang can cause some interesting results, Google Translate for iOS and Android is free, and offers many of the features the other paid for apps include. The only downside? You’ll need a data connection wherever you use it, which means incurring data roaming costs.

And finally, although not available outside Japan, an honorable mention must go to NTT Docomo’s real-time translation service, launched back in may. The first-of-its-kind service allows mobile phone users to communicate in near real time across languages, hearing a translated version of what the other person is saying.

Recent demos have shown a marked improvement with the system which allows 2 callers to hold a voice conversation translated between English and Japanese. Each caller not only hears the other callers voice translated, they can also see the accompanying text in both languages as a conversation thread on screen, much like text messaging or chat.

NTT claims that speech recognition rates are up to 90% for Japanese and 80% for English which is a huge improvement on the original launch rates of 15-20%. The service is currently in test until March 2012 and if successful, NTT will begin rolling it out to its 56 million Japanese customers in the second half of the year.

Related articles

• Google Translate Goes Grande On The iPad (gizmodo.com.au)
• Google Translate for iOS Gets iPad Support (mashable.com)
• Lonely Planet Launches ‘Lonely Planet Offline Translator’ apps for iOS and Android (prweb.com)

4 Essential Translation Apps for your New Smart Phone

Broadening the spectrum of translation production models to meet different quality requirements

Let’s make it clear from the very beginning: good old TEP (translation + editing + proofreading) is dead! We still see many RFPs where TEP is used as a description of the desired quality level vendors should base their quote upon, but more and more companies realize that this is not very useful. This is especially important to know since, as Michael Porter wrote in his classic book about competitive strategies, in mature industries with commoditized purchasing behavior, RFP participants have to compete for the lowest price and thus squeeze out every micro-cent in their production costs.

Vendors who start using cheaper junior translators to save money then need second-level editing and proofreading to achieve good final output. This TEP process ends up failing to deliver better quality than a qualified senior translator who doesn’t need the extra level of editing and provides excellent quality to begin with. To try and solve this issue some started interpreting TEP as ‘translation, editing and proofreading’ all done by one and the same person. Again, this makes things pretty clear: TEP is not well suited to high quality when cost is the defining factor driving a project.

Furthermore, you don’t necessarily need the same quality level for all types of translations. For many technical documents you need perfection in terminology and a good understanding of the topic, but you don’t need a high level of stylistic elegance. Even a punctuation error here and there (and some may want to kill me for saying this…) hurts much less than a translated text which doesn’t use the proper technical terms or makes clear that the translator didn’t really understand the topic.
Website marketing content needs different translation quality than website blog entries or customer support information, and therefore should use different translation and pricing models. So the good news is, if customers take the time to properly analyze their projects and their specific requirements and discuss these with their vendor, they will not only save a significant amount of money but get just the right quality and much better value than using the old-fashioned TEP model.

Machine translation surpasses the early adopter stage

While in 2010, almost everyone in the localization industry talked about Machine Translation, its actual implementation was limited to some select companies with sufficient budget, knowledge, and word volumes. By end of 2010/2011, quite a few new MT solutions appeared on the market which significantly lowered the entry-level barrier, amongst them the latest versions of PangeaMT, SmartMate and Moses as well as DIY distributions like DoMY and file/conversion packages like m4Loc. TAUS and their Data Association have been one of the most important players, helping unify the MT community and leverage knowledge and data exchange. They’ve predicted that “thousands of MT systems will blossom” soon and started TAUS Labs at end of 2011 to foster further initiatives for MT-related collaboration.

By listening to the sound of all these activities, one can easily recognize that Machine Translation is probably the hottest car in the massively changing localization marketplace. It brings new challenges, new players and new opportunities. It will be interesting to see who will take the pole position in the next few years.

Localization for casual buyers

I probably shouldn’t do it. I shouldn’t write about VC-funded competitors, who were bulging with cash in 2011, because it might make us look jealous. And we’re not. Absolutely not, really, so I’m going to write very nicely about them.

I’m talking about companies like our namesake mygengo and Motionpoint, who provide solutions for consumers that need quick and cheap translation for either a love letter or a website. Companies like them mushroomed in 2011.

Now, with the love letter, that makes total sense to me and there are already good solutions like onehourtranslation.com and translated.net who’ve been in the market quite a while, enjoying good success. Without knowing the exact numbers, I don’t believe it’s a huge market but probably in the lower two-digit million USD region. However, if you look at the management team of our friends at mygengo and the names of some pretty impressive investors, you might think it’s a totally different story.

So perhaps then, they’re here because of the huge growing market for website globalization and localization? Well yes there is a market, albeit for WordPress websites and the like where the owner decides to translate into one or 2 languages. And for this market a simple solution like motionpoint, where you don’t have to worry about the details of a full web localization process, makes sense. But again, is this a huge market?

Other more sophisticated buyers with serious ambitions for international marketing have different requirements from our understanding. They’re less interested in quick and cheap solutions, and more concerned with CMS integration, full localization of the design, multimedia content and transcreation of sales and marketing copy.

Is it only me who doesn’t understand how the business model behind companies like them will run the market in 2012? Your comments are welcome.

Related articles

• Can We Expect all Pervasive Machine Translation in 2012? (milengo.com)
• Guidelines for Selecting a Language Service Provider (milengo.com)
• Statistical Machine Translation: Building Your Own Viable Engine (milengo.com)

Localization Trends: 2011 review Part 2

]]> http://www.milengo.com/blog/localization-trends-2011-in-review-part-2/feed/ 0 http://www.milengo.com/blog/localization-process/ http://www.milengo.com/blog/localization-process/#comments Tue, 03 Jan 2012 15:00:02 +0000 Valarie Badame http://www.milengo.com/?p=4475 If you’re like many people we talk to, the company you work for is probably feeling the squeeze of the current economic downturn, and has been for a while. In tough times expansion is often put on the back burner and localization projects are left behind. However, don’t be fooled into thinking that localization is [...]

4 Key Tips for Simplifying the Localization Process

]]> If you’re like many people we talk to, the company you work for is probably feeling the squeeze of the current economic downturn, and has been for a while. In tough times expansion is often put on the back burner and localization projects are left behind. However, don’t be fooled into thinking that localization is a luxury that an international business can do without. In fact, if you’re cutting back on product launches and market expansion this leaves an opportunity to improve or expand the languages you already support, increasing engagement in existing markets and improving sales.

One of the biggest misconceptions is that localization is just a fancy word for translation, when in reality it covers everything from letters to layout, color to cultural suitability, style, tone, meaning, message and everything in between. The Localization process is a much more broad and comprehensive endeavor,  interdependent and layered, and when done right, can lead to excellent results.

Here are 4 areas that are key to ensuring your localization project goes without a hitch.

1. Preparation & Work Flow

The complexities of the localization process can be greatly reduced with proper preparation, in order to avoid backtracking, redundancies and excessive amount of version control. The benefits of setting up a tailored work flow will determine not only the turnaround time and productivity level, but quality as well. Quality Assurance procedures are often multi-staged, and should not only happen at the end of the project as this has a much bigger impact that correcting errors early on.

Providing clear project milestones to your localization vendor will ensure that the work flow is designed to achieve these goals and QA can be introduced at each step to ensure the project is kept on track. This also helps reduce the amount of input required from the localization vendor as expectations of quality and turnaround are set at the beginning rather than being arrived upon once translation is complete. Outlining and tailoring well defined work flows prior to localization can help when new requirements need to be dealt with, since a clearly defined project should allow time and cost margins to account for changes.

2. Terminology

Terminology lists and glossaries are the cornerstones of any localization process.  Even the smallest of localization tasks will experience the advantages of having both set up. A properly built terms base that is domain and client specific ensures that your message is consistent and uses proper language. Terminology should precede translation while terminology augmentation can be involved in a later stage.

3. Style Guide

A good terms base should be focused on domain or company specific language and doesn’t need to act as a comprehensive guideline for all localization-related events. A style guide can easily fill in these gaps that your term base has. The style guide should cover instructions on grammar and style, and also any other language preferences as well. This will render the localization process simpler since an approved and verified result can more easily be achieved from pre-determined specifications than by implementing queries and changes at a later stage.

4. File Format specification

It goes without saying that any formatting of files should be sound and intact and any required file structure and formats should be fully documented. It’s always worthwhile to do a pseudo-translation of any material before embarking on the full-fledged localization effort so the output can be assessed and any modifications can be made. Stay away from modifying file structures “on-the-fly” to accommodate the implementation of new features, for example, comments, meta-data, change tracking, etc. It creates much more work so specify and implement the requirements prior to translation.

Related articles

• 4 Key Tips for Simplifying the Localization Process (milengo.com)
• 4 Localization Strategy Blunders To Avoid (milengo.com)
• Guidelines for Selecting a Language Service Provider (milengo.com)

4 Key Tips for Simplifying the Localization Process

]]> http://www.milengo.com/blog/localization-process/feed/ 1 http://www.milengo.com/blog/localization-trends-2011-in-review-part-1/ http://www.milengo.com/blog/localization-trends-2011-in-review-part-1/#comments Tue, 20 Dec 2011 09:00:00 +0000 Roman Kotzsch http://www.milengo.com/?p=4554 Looking back on 2011 we’ve put together some of the localization trends we think will be continuing to dominate the industry in 2012. We’ll be covering these in a 2 part post, the second of which will be published in the new year. So with the new year ahead of us, what were the hottest [...]

Localization Trends: 2011 in review, Part 1

]]> Looking back on 2011 we’ve put together some of the localization trends we think will be continuing to dominate the industry in 2012. We’ll be covering these in a 2 part post, the second of which will be published in the new year. So with the new year ahead of us, what were the hottest topics in translation and localization in 2011?

Content speaks faster and more fluently in more languages than ever.

Content is king of the Internet and became even more powerful in 2011, especially since Google’s Panda update. Inbound marketing (also called permission based marketing) has become more and more popular, and many companies are creating large amounts of content in order to gain the attention of search engines. This has resulted in a lot of content with a short life cycle. Many companies are beginning to recognize that they can easily leverage this content in other languages, to better address international clients and at the same time increase page rank. Google loves sites with a high volume of good quality content that has been linked to by other pages and sites. As companies broaden their international language support, they generate more links back to their domain thereby increasing their domain authority. However new localization workflows, technology and solutions are needed to cope with this demand.

Another type of short length and life cycle content which adds to this type of project (or “translation stream”) are customer support texts from ticketing systems and forums. IBM and Lionbridge started with a joint venture to address this market earlier in the year. As the demand for highly connected, cloud-based services increases (which are by their nature international in their scope and capabilities), more and more companies will require an efficient way to talk to their global customers.

So all in all, exciting days for the localization industry, but providers like Milengo will need to come up with fresh solutions and cannot stick to business as usual working habits.

Crowd-sourcing is still hype.

Crowd-sourcing has been hyped in the translation and localization industry for a couple of years, ever since Facebook successfully completed the localization of their platform into several languages using their community as a resource. Many companies have investigated whether or not crowd-sourcing would be a viable option for them, most of them with the simple goal of saving costs. As expected, translators and translation vendors began crying wolf, complaining that you can’t replace serious professionals like them with inexperienced youngsters who know nothing about translation. ‘Understanding two languages does not make you a linguist’, they said.

And then there’s the product. If you look at the number of successful crowd-sourcing implementations in 2011, you’ll find that in reality they’re only partially so. Crowd-sourcing translation has worked pretty successfully in non-commercial areas like political and news content or in other voluntary areas like those supported by Translators without Borders. But on the commercial side, apart from Facebook and Skype, we have only seen very few broad adaptations of the crowd-sourcing concept. And the reasons for this are simple: it is in fact true when translators claim that you cannot simply take anyone from your user community who understands the source and the target language and expect professional results. Only a certain percentage of users have the necessary language skills that are sufficient to produce a ‘reasonable quality’ translation. If you multiply this by the small percentage of community members who are actually willing to do voluntary work in exchange for some fancy badges, a free product or some other non-monetary reward, you realize that actually you need a very large community users to run a translation crowd-sourcing project. Facebook and Skype with a couple of hundred million users do have such a community, no doubt there. However, most other commercial enterprises and smaller social media platforms don’t. So thankfully, that’s good news for the beleaguered translators and vendors.

Finance tighten the purse strings: Further commodification of translation services.

It already feels like a long time ago when localization managers used to have full control over the vendor selection process and based their decisions on language and service quality. In the last couple of years there has been a growing tendency for translation services to be considered a commodity and purchasing language services has become akin to purchasing a ton of copper. The product is the same wherever you get it, the only difference is the price. And if only the price matters, then the negotiation and purchasing decision is up to the finance or procurement department.

That makes sense for mature buyers who fully understand the localization process and can manage quality assurance and work-flow optimization on their own. They understand the complexity of defining the quality and services levels they need and since they don’t require additional support services like software engineering or DTP, they buy millions of words in bulk as you would buy tons of copper.

Buyers with a slightly lower degree of sophistication (not because of inability but more because of demand) might take the wise decision not to consider localization as a commodity. Rather, they select providers as partners which help them increase the efficiency of their localization process, instead of buying words as cheaply as possible. Selecting the right partner for this kind of scenario is a totally different undertaking since you need to select the perfect vendor based on size, (so as a customer you are not too small and not too large), experience and service quality. I wonder if the most recent developments in localization RFI/RFP best practices reflect this situation?

Related articles

• Localization Trends: 2011 review Part 2 (milengo.com)
• Can We Expect all Pervasive Machine Translation in 2012? (milengo.com)
• Statistical Machine Translation: Building Your Own Viable Engine (milengo.com)

Localization Trends: 2011 in review, Part 1

]]> http://www.milengo.com/blog/localization-trends-2011-in-review-part-1/feed/ 1 http://www.milengo.com/blog/statistical-machine-translation/ http://www.milengo.com/blog/statistical-machine-translation/#comments Wed, 14 Dec 2011 13:00:46 +0000 Deepan Patel http://www.milengo.com/?p=4454 Any company investigating Statistical Machine Translation (SMT) solutions might be put off by the various data demands made by MT technology providers. Certainly the sheer volume of training data that MT providers expect you to dig around for will make a lot of companies wonder whether it is a worthwhile exercise at all. For example, [...]

Statistical Machine Translation: Building Your Own Viable Engine

]]> Any company investigating Statistical Machine Translation (SMT) solutions might be put off by the various data demands made by MT technology providers. Certainly the sheer volume of training data that MT providers expect you to dig around for will make a lot of companies wonder whether it is a worthwhile exercise at all. For example, in most cases the following recommendations are:
• a minimum of 500,000 sentence pairs or translated units (bilingual data)
• a minimum of 300MB of target language data (monolingual data) to help build the language model and generate target language fluency.

Of course, unless you’re a very large company that has already engaged fruitfully with translation operations and therefore built up a huge library of translation assets, then you’re going to struggle to meet these *minimum* requirements. So there goes that plan out of the window…

Well actually that doesn’t have to be the end of the matter, for as the old English proverb goes: there’s more than one way to skin a cat. Or in this case, there’s more than one way to build a viable Statistical Machine Translation engine.

These data obstacles were something that ProQuest, a Seattle based company, faced as they looked to dip their toe in the Statistical Machine Translation pool. ProQuest produces high value technical publications for the scientific and technical market. As part of their extensive academic database, they have a significant volume of very technical content in Chinese at their disposal which they wanted available in English for non-Chinese speakers.

Their objective was clear; they did not require polished translations for release into the public domain, rather they wanted to make the Chinese content ‘discoverable’ to English language researchers so that they could gain a basic understanding of the academic content. If these researchers stumbled upon an article of use to them, then they would request that the original Chinese document be submitted for traditional translation.

Despite the clarity of their aim, ProQuest had very little aligned bilingual data with which to approach an MT technology provider, and request the customization of a SMT engine tailored to their requirements. However, they did have a significant amount of data available in both Chinese and English and after extensive conversations with Milengo, we were able to devise a battle plan…

ProQuest would deliver:

• Large volumes (GBs if possible) of monolingual English material in the “General Engineering” domain.
• Large volumes of Chinese articles with rudimentary translations where available in the “General Engineering” domain
• Glossaries or key search term lists available in either English or Chinese
• This data would be used to customize a Statistical Machine Translation engine that would clearly outperform openly available MT options such as Google translate, after a threefold incremental training cycle.

The Outcome?

We were able to extract as much parallel bilingual text as possible from the material ProQuest provided. From the remaining content provided, we were able to analyse high frequency English and Chinese word clusters which were then translated to create the required bilingual data for the SMT engine training. After the threefold training cycle guess what? The new ProQuest Statistical Machine Translation engine was clearly outperforming Google and Systran on ProQuest content.

Related articles

• Can We Expect all Pervasive Machine Translation in 2012? (milengo.com)
• Study Shows Impact of Clean Data and Consolidation on Statistical Machine Translation Quality (prweb.com)

Statistical Machine Translation: Building Your Own Viable Engine

]]> http://www.milengo.com/blog/statistical-machine-translation/feed/ 4 http://www.milengo.com/blog/taus-user-conference/ http://www.milengo.com/blog/taus-user-conference/#comments Wed, 07 Dec 2011 15:50:28 +0000 Rob Davies http://www.milengo.com/?p=4422 TAUS, of which Milengo is a member, recently hosted their annual User Conference in Santa Clara where researchers and speakers spoke about translation automation, localization business innovation and industry collaboration. Leading practitioners tackled the most pressing issues and opportunities facing the global translation industry. The conference reflected the high hopes of hybridization, customization and syntax, [...]

TAUS User Conference

]]> TAUS, of which Milengo is a member, recently hosted their annual User Conference in Santa Clara where researchers and speakers spoke about translation automation, localization business innovation and industry collaboration.

Leading practitioners tackled the most pressing issues and opportunities facing the global translation industry. The conference reflected the high hopes of hybridization, customization and syntax, as well as the many unanswered questions about the role and efficient use of data. Talks from two major research groups help to ground us in reality, while a panel of the leading global researchers join the debate to help shed light on what we can expect in the near future.

Videos from TAUS Santa Clara 2011 are now online here.

You can also access the individual videos at the links below.

• The multilingual web – “beyond the translate button”
• Open Translation Platforms
• TAUS members share Moses-related tools, making others’ lives easier
• The Great Interoperability Debate
• Machine Translation Use Cases
• R/time Multilingual Chat and Spoken Translation
• Researchers on Future Translation Technologies
• Collaborative Translation and Innovating Pricing

TAUS User Conference

]]> http://www.milengo.com/blog/taus-user-conference/feed/ 0 http://www.milengo.com/blog/4-localization-blunders-to-avoid/ http://www.milengo.com/blog/4-localization-blunders-to-avoid/#comments Fri, 02 Dec 2011 14:20:09 +0000 Valarie Badame http://www.milengo.com/?p=4401 The road to localization success is not an easy one. Companies oftentimes begin the process with a desire to expand their international presence, but have little experience or prior knowledge of what is required to successfully complete a full scale localization effort. Realizing that you need a strategy is one half of the challenge and [...]

4 Localization Strategy Blunders To Avoid

]]> The road to localization success is not an easy one. Companies oftentimes begin the process with a desire to expand their international presence, but have little experience or prior knowledge of what is required to successfully complete a full scale localization effort. Realizing that you need a strategy is one half of the challenge and the process of educating yourself about how to avoid potential problems and pitfalls is the other.

Along the journey you are, of course, bound to make some mistakes. Minimizing those mistakes, is the key to avoiding roadblocks that can bring your localization projects to a standstill and cause product launches to be delayed or at worst, canceled.

To keep your localization projects running smoothly over the inevitable bumps, detours and slow downs, here are four localization strategy blunders that you can learn from and, more importantly, what steps you can take to keep your localization projects on track.

Mistake 1. Not having a strategy is the worst strategy

Companies that sell products or services internationally know they need to understand localization in order to tap into local markets. However many wait too long, sometimes weeks, months and even years before fully engaging with the process, instead relying on ad hoc translation jobs to see them through. Next thing they know a major competitor sweeps into a market they had their eye on with a great product, a spectacular launch and a bunch of promotional activity that is carefully tailored to the local market. Now the game of catch up begins, as everyone scrambles to pull together a plan. If this sounds familiar, you’re not the first and probably won’t be the last.

To compensate, you end up working frantically to meet deadlines, make rushed decisions that lead to delays and leave inadequate time for a thorough and proper response to the competition. This can often lead to the recruitment of vendors that are not matched with your business objectives or expectations, and in the long-term this can cost you time and lost revenue.

The problem with this scenario is that, as the saying goes, failing to plan is planning to fail. A localization project is complicated. It requires input from many different stakeholders who all need to understand how each part of the localization process influences end results. This can only be accomplished by first creating a localization strategy, and engaging everyone from engineering, product, sales and marketing as early as possible in the product development cycle. This ensures that everyone understands the end objectives and how to get there, taking an existing product or service and making it relevant, understandable, functional and successful in each of the languages it’s adapted for.

Mistake 2. No communication or Support

A company can’t expect perfect translation quality from a localization vendor if they don’t effectively communicate their brand values and product benefits. Remember, you know your products and services best. If you can communicate how they work and why your customers buy them, along with your brand, positioning and communication style, you’ll find that when you come to review your translations it’s not just the words but also the meaning, style, tone and terminology that have been adapted to fit the target language. This level of understanding from your vendor is vital if you want to run a tight ship, keeping linguistic review time down, and keeping amends to a minimum.

Managers also sometimes have difficulty obtaining sufficient support from key people within their organization because they don’t effectively communicate their localization strategy or the recommendations of their localization vendor. Without a good level of engagement across departments involved in the localization strategy, you’ll eventually find yourself struggling mid-process, trying to convey the importance of collaboration, missing your targets and wasting time and money.

Mistake 3. Going with the first big guy (or little guy) and thinking they are the best for the job.

Many first-time time translation and localization buyers search for the best known names in the localization industry and hire them without knowing what they actually need, or what questions they should ask. Sending out a one-size-fits-all ‘request for information’ to multiple vendors without performing due diligence will only complicate the selection process. Like buying any other service or product, you must understand what you’re looking for first so you can better decide what you need from a vendor.

Another problem is that many first-time buyers are viewed as “small fish in a big pond” in the eyes of some L1ON vendors. This means they may not deliver all the support the first-time buyer needs during the critical first phase of launching a product or service in a new international market .

Alternatively, some first-time buyers may feel more comfortable hiring a small, local vendor. This can lead to a situation where the chosen vendor may not have the depth of expertise and resources to support your localization plans, quickly becoming a waste of time and money as you outgrow them. So make sure, when selecting a translation and localization provider, that you give yourself enough time to perform due diligence and use an evaluation process tailored to your specific needs. If you do this you’ll find it much easier to select the right vendor at the right price.

Mistake 4. Getting sold on expensive technology you don’t need.

All Translation Management Systems are not created equal; one size does not fit all. A TMS solution that works for a mid-sized company may not work for a smaller company and vice-versa. Your company is unique and deserves a  tailored approach. Some vendors have become very successful at convincing companies that they need to buy expensive technology and encourage them to automate the process with X, Y and Z System. They claim that everything will operate much more efficiently, the long term cost savings will be significant and that implementation will be straightforward, however the truth is sometimes the opposite. Most Translation Management Systems require dedicated teams to look after them and involve the development and implementation of specialized work flows for the system to functions effectively.

Instead, it is much more practical to buy a technology solution that is scalable and flexible enough to seamlessly adapt to your existing business processes. A technology solution should not only save time and money but also be intuitive and unobtrusive allowing you to stay focused on what is important: implementing a successful localization strategy.

4 Localization Strategy Blunders To Avoid

]]> http://www.milengo.com/blog/4-localization-blunders-to-avoid/feed/ 3