WORKFLOW

Creating hyper-localized, translated content that is ready for distribution to any region worldwide is simple when you have tools like AI-powered machine translation (MT) at your hand. Simply select your source language and target language pairs and run your translation job through your service provider’s translation engine.

It’s that easy, right? Guess again.

In theory, translating hyper-localized content should be simple, in terms of process and steps. In reality, successfully translating hyper-localized content is a highly complex process that requires the right combination of tools, workflow, and personnel to execute well.

Professional translators are now required to consider multiple dialects, formalities, and slang within a country or region to appeal to and be better understood by a wider audience. Translators must also craft the translations within the technical requirements established by their customers, such as maximum characters per line, lines per segment, and mergeable gap; minimum and maximum caption duration; and timing styles. These expectations require that translators have the right balance of precision and creativity, while maintaining the right level of nuance to best convey the intended meaning of the content’s creator.

It’s important to have the right expectations when considering hyper-localization, understanding that translations become more subjective and less literal due to the creative measures taken to meet the job’s multi-dialect and technical requirements.

In spite of the technological advances in AI-powered machine learning, the day-to-day workflow remains a key factor in ramping up the localization process to the “hyper” level.

Given the insatiable demand for localized content on streaming platforms and linear/broadcast networks, localization service providers (LSPs) need streamlined workflows that unlock the full power of AI to support efficient and cost-effective localization. There are simply not enough linguists to translate all this content, especially when content is translated between more languages than ever before.

Hyper-localization: It’s not just for content

Hyper-localization extends well beyond the sole domain of content translation and is considered the new superpower in marketing.

Everyone has been influenced by hyper-localized marketing in some shape or form. Analytics and demographical information obtained from web traffic help marketers personalize their offerings to appeal to local consumers. The following are a few examples of hyper-localized promotions:

• A national chain retailer running promotions with local sports teams to appeal to a regional fanbase
• Mom-and-pop stores cross-promoting with local colleges to offer discounts
• Retail websites geo-targeting specific cities and regions to personalize promotional offers and materials.

There are clear benefits to the right hyper-localization approach. According to the 2021 State of Personalization Survey conducted by research firm Segment, more than half (60%) of consumers say they would likely become repeat buyers after a personalized shopping experience with a retailer (up from 44% in 2017). These findings also coincide with the world’s emergence from the Covid pandemic, a time when nearly everyone globally was forced into a hyper-local mindset due to lockdowns. It’s no surprise that these habits are likely to remain for the long term.

It’s no different in content translation. Why would a viewer in the southern part of Germany want to receive content tailored to speakers a northern dialect?

Successful hyper-localization is challenging, as it requires technical tools to analyze data and track changes in customer preferences. We must also recognize the evolving changes in languages and the complex nuances often found within the same language that can make accurate translations a logistical nightmare. Varying sentence structure rules, idiomatic expressions, compound words, multi-word verbs, and sarcasm are only a few examples of these complex nuances within the same language.

So why would we take on these challenges? The answer is simple: There are significant growth opportunities. A Slator report noted that by the end of 2021, the global translation and localization sector’s market value would be $26.6 billion, with annual forecasted growth of 11.8%. A concurrent report from DATAINTELO projected growth in the global localization software market of $5.51 billion with an expected growth rate of 4.3% by 2030.

Recognition of these opportunities has led the MT market to become extremely competitive. Currently, there are more than 50 MT vendors, but as the only media-focused MT company, we firmly believe that MT will inevitably and permanently change the landscape of the global localization industry.

The media space is such a highly segmented and specialized sector that requires a level of accuracy; creativity; storytelling capabilities; a deep understanding of cultural nuance, formality, and genre; and language fluency — all while complying with each content owner or distributor’s unique technical specifications and profiles. To support these needs, specialized data and engine training pipelines are required for developing MT engines that meet these requirements.

Localization in and of itself is complicated. When you take it even further and focus on hyper-localization, it’s important to have the right partner and tools, but also respect for the creativity involved in the post-editing process.

Most MT models only translate sentences one-by-one, losing critical context that is outside of the primary sentence. In contrast, context-aware models use information throughout the entire text to better translate gender, slang, formalities, multiple-word meanings, and other language intricacies. Context awareness allows the technology to truly “localize” content instead of simply translating it word for word. Our context-awareness models are tuned for accuracy, often exceeding the expectations of translators while providing extraordinary consistency across the document as well as an entire series.

There’s still more work we need to do

For all our advancements in technology, there are still many challenges impacting the localization industry — both from a technical and business perspective.

With the growing demand for subtitled hyper-localized content, one of the biggest challenges that can’t be immediately solved is the talent gap. This challenge is pervasive across all industries that require localized messaging. It will take time to build up our human resources. AI-powered MT technology is the only effective option to help companies scale to handle unlimited market demand and work at the speed of their business.

Focusing linguists’ valuable creative and technical skills on refinement tasks through MT post-editing enables repetitive and tedious tasks to be completed through automation and MT. Not only does this allow linguists to focus on what they do best, but it also provides more time to devote toward ensuring translation accuracy and increases the throughput of individual linguists.

Localization professionals in media and entertainment are introducing AI-powered MT into their workflow; however, some professionals are still apprehensive and have many questions. Will the migration to MT result in lost jobs? How do I choose the right MT engines for my business? Will I still gain efficiency even when post-editing is required? How can I integrate MT when I’m so busy that I don’t have the mental bandwidth to learn something new?

To answer these questions, it’s critical that we work collaboratively to address these concerns. The quick answer is no — AI-powered MT is not a job-killer. XL8 views MT as a job-saver because it helps LSPs become more profitable and increase their throughput with their existing staff. Whether you are an LSP, a large production facility, or a distribution company, it’s important to explore how we can rethink our workflows together.

While it takes head space and a commitment to implement new technology, the benefit is the immediate cost savings and efficiencies gained during a recession where we’re required to do more with less.