Why Machine Translation Matters: Trends & Best Practices – by Kirti Vashee

Abstract — This document is a summary for a publication in the Proceedings of LTAC conference to be held in Rome, Italy, April 5 – 6, 2011. It provides an overview of why machine translation is important and points out trends and best practices in the use of this technology.

Keywords — machine translation, collaboration, internet trends, localization, business translation

Introduction

We live in an age where information and information access is often a key to building prosperity and economic wealth. This is true not only in the world at large, for countries that wish to build knowledge-based economies, but is also true increasingly for enterprises involved in building global businesses. It is useful to look more closely at the forces that are driving the change as it helps us to understand where these forces are leading and also better understand the impact on the business of professional translation. Access to knowledge is one of the keys to economic prosperity. Automated translation is one of those technologies that offer a way to reduce the digital divide and raise living standards across the world. As imperfect as it is, this technology may even be the key to real people-to-people contact across the globe.

A fundamental force that underlies these forces is the increasing amount of information becoming available on the internet and the growing expectation amongst the online population that information should be available. This, more than anything else drives continued interest in machine translation, as there is simply no other way to make information multilingual in the volumes that it is being demanded in, in a timely manner.

The Forces

A useful way to look at the forces influencing this is to consider both the broad forces outside of the translation industry and the forces within the international business markets that drive much of the translation activities of the professional translation industry.

External Forces

The growth of the digital universe

There are several studies conducted by universities¹ and by information technology companies like EMC, which are close to the information creation process. In an EMC funded study of the digital universe², IDC has forecasted that the total size of this universe will grow by a factor of 44 from its current size. Of course, this does not equate into a 44 fold increase in words to be translated, as much of the new content is video and graphical, but it does suggest momentum and mass on an exponential scale as the chart below also shows.

Figure 1 IDC Digital Information Growth Forecast

The social drive behind greater information access

Today, the world faces a new kind of poverty. While we in the West face a glut of information, much of the world faces information poverty. The cost for this can be high. According to the University of Limerick President Prof. Don Barry, “80% of the premature deaths in the developing world are due to lack of information”. Much of the world’s knowledge is created and remains in a handful of languages, inaccessible to most who don’t speak these languages.

As Peter Brantley³ at Berkeley says: “Mass machine translation is not a translation of a work, per se, but it is rather, a liberation of the constraints of language in the discovery of knowledge.”

This issue is also very eloquently developed by Ethan Zuckerman in his seminal essay on the subject, “The Polyglot Internet⁴”, where he says: “While there is profound need to continue improving machine translation, we also need to focus on enabling and empowering human translators. Professional translation continues to be the golden standard for the translation of critical documents. But these methods are too expensive to be used by web surfers simply interested in understanding what peers in China or Colombia are discussing and participating in these discussions. The polyglot internet demands that we explore the possibility and power of distributed human translation. We are at the very early stages of the emergence of a new model for translation of online content – “peer production” models of translation.”

The growing online population

Is causing a shift in the view of strategic languages. There is more and more evidence that the fastest growing online populations will be in Asia and Brazil and thus languages in these regions will assume an increasingly more important role as their economic momentum continues to mount.

1. McKinsey⁵ predicts that 700 million Asian users will come online in the next 5 years and represent $80B in infrastructure and online commerce opportunity.
2. Cisco⁶ predicts that most growth in the Internet-related market will occur outside of today’s high income, or “advanced,” economies.
3. The L2 Research⁷ group called China the biggest luxury market opportunity in a generation.

Internal Forces

The growing importance of customer conversations

IDC has documented² that much of the new digital universe is pertinent and important to global enterprises. The carefully calculated marketing and corporate image control that global enterprises have been used to is coming undone. Brand impressions are increasingly being formed by real customer conversations in social networks. We see that the world of marketing is undergoing a transformation and what used to be considered critical corporate messaging is increasingly viewed as “corporate-speak” and is often not trusted by the end-customers who matter the most⁸. More and more, the high value content is being created by customers and business partners and corporations have little editorial control of this content creation process.

This new content is now seen as deeply influential in the development of branding, and user-generated content (UGC) is increasingly being translated as it is seen as influencing customer purchase decisions.

Thus this issue of the relevance of corporate websites matters, because increasingly customers are making decisions about products long before they get to the corporate website. Valuable content is linked to influencers and dynamic conversations that naturally evolve in online social spaces. This content influences purchasing behaviour and helps to form brand impressions and build brand loyalty. It is unwise to ignore it, as this is where brands, market dominance and leadership positions are increasingly being built. Remember that the whole point of localization is to enhance and drive international business initiatives.

So while one definition of higher value translation is the extent of transformation during the translation process, I think the more important driver is the value of the content at the business level especially to final customer. My view of some of the emerging high value content:

• conversations that are trusted by potential customers at various stages of the purchase process (e.g. Amazon, C-Net, Orbitz, Travelocity etc.);
• conversations and content that help build customer loyalty (this could include support and reseller community content as well);
• articulate and unfiltered opinions and reviews on the customer experience (Amazon, Orbitz, Expedia etc.);
• leading Bloggers who influence and help form brand impressions;
• content that is co-created with customers that often facilitates comparison with competitors;
• content that encourages collaboration with customers and key partners (e.g. Dell IdeaStorm).

Figure 2 IDC Digital Universe Estimate of Enterprise Related UGC

The growing importance of open innovation and collaboration

In a world where the customer’s voice can overpower the corporate marketing office, we need to rethink how we do things. We are also seeing that co-creation of products and services with customers can be a huge momentum builder. Facebook has shown that engaging users in translating interfaces can also help accelerate building the customer base in those languages. Their international growth has been closely linked to their crowdsourcing translation efforts even though crowdsourcing can be less predictable. Dell Computer actively solicits and nurtures active customer feedback in a program called IdeaStorm to develop new products. Collaborating with customers and partners is emerging as a way to engage customers, build loyalty and accelerate penetration in new markets. This requires new approaches and new levels of transparency to be able to successfully engage with customers in an active dialogue.

The rise of Asia and BRICI

Historically, Western Europe and FIGS have dominated the professional translation world. While FIGS will remain important, in the future it looks like Brazilian Portuguese and Spanish will be the most important European languages and Chinese, Hindi, Indonesian and other Asian languages will become increasingly more important and strategic as global revenue generating translation investments. Many companies now are expanding their base of languages and the FIGS-CJK view is slowly receding as the world shifts from a G-7 to a G-20 view with a new economic power structure.

The IMPACT

So if we add all this up, we can see that global enterprises are facing a content deluge with dynamic content coming from both internal and external sources, and large volumes of this content are expected to be translated increasingly faster to have any value in competitive situations. Global enterprises that quickly identify high value content and make it multilingual will find that this can drive international revenues and that translation can be a strategic tool to building long-term competitive advantage.

Figure 3 The growing volumes of dynamic content

However this is a time of revolution, and the TEP (Translate-Edit-Proof) and SDL (Software and Documentation Localization) mind sets are not likely to be adequate to meet these new translation challenges. The old approach worked for static, low volume content but new thinking and new approaches are required to deal with the data deluge today. Automated translation is an absolute necessity in the new world but this is not the MT of yesteryear that many are still implementing and describing at localization conferences today.

In the new world, data has to flow from content creation to consumption as seamlessly as possible, delivered to where it is needed at desired quality levels. This means that humans need to be part of the production process and are the key to producing the best quality. The future is about much better man-machine collaboration. The automated translation tools need to be learning and getting better all the time. They need to be responsive to skilled human linguistic steering and corrective feedback. They need to be focused on dynamic streams of content not just static, packaged translation projects like user documentation. They need to understand that with flowing content, upstream clean-up efforts will flow through the production line and make every downstream process easier and more efficient. They see the information cycle as an organic system and thus build the collaboration infrastructure to address the whole problem. They need to see MT systems improve with human steering, in weeks, not months or years. They need to be MT systems that can be trained and managed by skilled professionals to get you to “good enough” production quality and “fast enough” to have a positive impact on your business, not the black box MT of yesteryear. It is likely that content streams will be characterized for processing in schemes like the one shown below:

1. critical and legal content may be processed via high-quality professional translation channels (mostly human);
2. large volume product-related knowledge content may be processed via customized MT and post-editing;
3. random comments and social media feedback could be processed by customized MT systems to identify influencers who may receive additional processing.

Figure 4 Developing production lines for streaming content

As more senior corporate executives realize that translation is strategic, and that translation technology properly used can generate substantial revenue in global markets, they will start to look at solving these new kinds of translation problems. Executive managers are unlikely to be excited by the possibility of getting user documentation done faster and cheaper. However, as they start to get their hands around the fact that customer conversations are important (many are struggling with this), and that they need to respond with speed and agility to build global customer relationships, we can expect to see a new kind of executive who will seek to make flowing content multilingual. They will care that real data interchange standards exist and will likely remove products that do not comply. Handling the flow efficiently will become the focus, and the most visionary localization managers may even have senior roles in making this happen. Speed and agility are key, and customer engagement across the globe will require a real understanding of how to make these dynamic content focused translation systems work.

All of this cannot happen without the efforts of skilled language industry professionals but new skills are likely to be needed. The focus is likely to shift to flowing streams of content related to customer conversations about products and services.

Figure 5 New human skills required

We are living in a time of great change. In times of change there is often an opportunity for new paradigms and new leaders to launch. Not so long ago Microsoft grabbed the desktop market away from IBM and Google grabbed away the web search market from Microsoft. New leaders with new visions are the change agents that make this happen, and they are often dismissed by the established status quo and “leaders” of the time.

Machine Translation is likely to be a key core competence for industry leaders in the future as there is no way to deal with the flowing streams of conversations and content of the future without it. However, rather than be passive users of public MT engines, professionals need to learn how to use the technology for their specific goals and objectives. They need to learn how to use it to work for their needs.

Figure 6 Taking control of machine translation

The skills required will include the following (and many are just emerging so this is a great opportunity for innovators and leaders):

1. customization of MT systems for specific business purposes;
2. corpus analysis and assessment skills;
3. evolutionary approaches to making high value content multilingual;
4. rapid quality assessment skills;
5. linguistic steering of automated translation systems;
6. community and crowd collaboration management and administration to do a variety of linguistic work;
7. more structured approaches to post-editing MT to enable rapid error identification and correction;
8. continuously evolving and learning MT systems that produce on-going improvements in translation quality;
9. much better and more robust data interchange standards will likely develop.

As high-level executives begin to see the huge value and market enabling power of translating large amounts of relevant content, we can expect to see that translation will be viewed as a much more strategic core competence⁹. As this happens, translation professionals could become facilitators and enablers of many key conversations between global enterprises and their customers.

Kirti Vashee
Vice President Enterprise Translation
Asia Online
Santa Monica, CA USA
kirti.vashee@asiaonline.net
k.vashee@gmail.com

References

1. Lyman, P. & Varian, H. R. (2003), How much Information?, School of Information Management and Systems, Berkeley, University of California.
   http://www2.sims.berkeley.edu/research/projects/how-much-info-2003/
2. IDC (2010), A Digital Universe Decade – Are You Ready?
   http://gigaom.files.wordpress.com/2010/05/2010-digital-universe-iview_5-4-10.pdf,
   http://www.emc.com/collateral/demos/microsites/idc-digital-universe/iview.htm
3. Bentley, P. (2008), Losing what we don’t see: Translation
   http://blogs.lib.berkeley.edu/shimenawa.php/2008/11/02/losing-what-we-don-t-see-translation
4. Zuckerman, E. (2008), The Polyglot Internet
   http://www.ethanzuckerman.com/blog/the-polyglot-internet/
5. Daga, V., Manuel, N. & Narasimhan, L. (2010), Riding Asia’s Digital Tiger, McKinsey Quarterly, September 2010
   https://www.mckinseyquarterly.com/Marketing/Digital_Marketing/Riding_Asias_digital_tiger_2667
6. Rueda-Sabater, E., Derosby, D., Johnston, J. & Nancy Murphy (2010), The Evolving Internet, Monitor Group
   http://www.monitor.com/Portals/0/MonitorContent/imported/MonitorUnitedStates/Articles/PDFs/Monitor_GBN_Evolving_Internet_Cisco_2010_August.pdf
7. Galloway, S. & Mullen, M. (2011), The biggest opportunity for luxury brands in a generation
   http://www.europeanbusinessreview.com/?p=2391
8. Owyang, J. (2007), How to evolve your irrelevant corporate website
   http://www.web-strategist.com/blog/2007/05/29/web-strategy-how-to-evolve-your-irrelevant-corporate-website/
9. Vashee, K. (2011), The changing face of localization (Professional translation)
   http://kv-emptypages.blogspot.com/2011/03/changing-face-of-localization.html

Tags: slider