Communication is the one of the most important elements of business, and Machine Translation is a flexible tool that can be used to facilitate communication in a wide variety of scenarios and situations. Multinationals and other companies operating globally can take advantage of Machine Translation to achieve productivity gains.
This two part blog series examines two very different examples of implementing Machine Translation. This first post will look at what multinational organizations should consider before introducing Machine Translation to their business, and the second post will discuss the productivity gains and competitive advantages that can be achieved by Language Service Providers (LSPs) who adopt MT.
What is a multinational and why should it use Machine Translation?
Multinational corporations or global businesses are organizations operating in more than one country or region. The concept of an ‘international company’ has been around for hundreds of years, going back to the trading companies, which were established in the 1700s. Outside political agendas, their main purpose was to trade in spices and other commodities throughout Asia and Europe exposing traders to different languages and cultures.
Hundreds of years later, global communication is common place as more businesses operate internationally. There are no boundaries, and companies with worldwide operations require a constant flow of multilingual communication in order to maintain relationships between global employees, customers and stakeholders.
Multinational organizations typically have two types of content; external and internal. External content is created and released to the public; corporate documents, investor information, Corporate Social Responsibility (CSR) and marketing communications. On the other hand, internal content is created for use within the company, this is usually in the form of email and chat communications, memos and other internal documents.
To Translate or not to translate
Organizations without an in house translation team, often outsource the translation of external content to a reputable LSP. This ensures a guaranteed level of quality for the translation, and it also means that the process of localization is more efficient and cost effective. This is because, over time language assets in the form of translation memories, can be built up and leveraged to off-set the cost of future translations.
Internal content, however, is mostly comprised of communications between departments; emails, chats and information on sales and marketing activities. These are usually not translated professionally for a number of reasons:
- Cost – the volume to be translated can make costs unmanageable
- Confidentiality – managing sensitive information is more difficult
- Real-time translation – emails and chat conversations generally requires real-time speed
As an example, if a company is headquartered in the United States, but operates in both Asia and Europe there is a very high possibility that more than one language is used in the company’s internal communication.
Multinational companies often select working languages that must be used for internal communications and department managers are sometimes required to have a certain level of proficiency in the company’s designated working languages, which usually includes English.
Large organizations like the United Nations also have official languages. In this case, documents are not published until a translation has been prepared in each official language.
So, what happens when an email with a client’s product specifications and sales information is sent to a group of employees who speak different languages? Some of those readers may have limited knowledge of the language being used, and only be able to understand the communication, but are not familiar enough with the language to write a coherent response. This can result in them responding in their native language. Suddenly, a single conversation thread contains more than one language, with a greater potential for miscommunication.
Why use Machine Translation?
Multinationals with global operations often have issues with the quantity and flow of internal information between departments operating in different languages. If the corporate headquarters uses a different language than its global subsidiaries, corporate documents need to be translated into each language as the internal information moves down the organizational hierarchy.
Machine Translation is a solution that can provide an instant, understandable ‘gist’ of internal information across a company operating in different languages and the use of MT can serve two purposes:
- Documents that require a professional human translation are easily identified
- Internal documents can be translated instantly so employees can get an understanding of the content
In order to understand internal content, employees often might use an open source MT solution such as Google Translate. While this is useful, it does not take into consideration any proprietary jargon or writing styles specific to the organization, and it also raises the question of confidentiality.
Challenges of MT
Many organizations may be interested in taking steps to deploy their own MT systems rather than outsourcing translation jobs or asking bilinguals in the company to do ad hoc translations. Those considering MT have two options; develop their own in house system or use a cloud-based subscription model.
Implementing any new process has challenges and MT is no exception. Some challenges traditionally associated with implementing MT systems are:
- High costs
- Complex technology
- Long deployment times
How should an MT system be integrated?
Before going ahead with an MT solution, an organization needs to carefully consider what it hopes to achieve from implementing Machine Translation. The company should evaluate all the perceived benefits thoroughly, including managing any and all expectations about using Machine Translation.
Organizations thinking of implementing MT should ask:
- What is its purpose? – Will MT be used as a management tool to improve internal communication and productivity, or to make decisions on what documents require professional outside translation? The purpose should be clearly defined at the outset.
- Do we have enough language assets to build high quality engines? Bilingual language assets are a key ingredient for building MT engines. The quality of the training data will have a direct impact on the MT engines output “garbage in, garbage out”.
- Should we invest in building our own system or buy a cloud-based subscription service? MT systems can be rule-based (RBMT), statistical (SMT) and hybrid. In house development of a propriety MT system requires a heavy technology, HR and training investment, unless those assets are readily available. Cloud-based subscription models do not require such a heavy initial investment and are often more cost effective than developing and managing an in house MT system.
- Is the Machine Translation option scalable? How many language combinations will be needed? If each language pair requires its own unique engine, how simple is it to build additional engines with new language combinations? Scalability will be determined by translating capacity and the ability to add new language combinations, this would be especially important when entering different language markets or expanding the business to new regions. The MT solution should align itself with the company’s long term goals.
- How will MT be integrated into everyday workflows? Users need to be able to easily access translation functions through their existing applications like email or the company intranet system to make it accessible and viable.
- What indirect costs and planning will be involved? RBMT and hybrid systems require qualified linguists or language experts to develop and manage the engines. SMT systems use algorithms to identify probable translations based on the frequency, therefore, storage capacity is essential for the large volumes of training data required. Cloud options eliminate the need for in house technology investment, but extra costs might be incurred for going over the subscription plans, similar to the minutes allowance with mobile phone usage.
In carefully answering these questions, any organization planning to implement MT can stay focused on using the most cost-effective solution and achieve productivity gains with less miscommunication and more time savings.
The next part of this blog will look at how LSPs can leverage Machine Translation technology for productivity gains and competitive advantage.