AI series, part 2/3: How AI can boost your localization projects – from AI in MT to more inclusive texts

AI series, part 2/3: How AI can boost your localization projects – from AI in MT to more inclusive texts

AI series, part 2/3: How AI can boost your localization projects – from AI in MT to more inclusive texts 1920 1465 Bartosz Budzynski

In this three-part series, we discuss the role of artificial intelligence in translation and localization. In our first post, we talked about how it’s changing the industry. In this second post, we look at boosting your localization and translation projects with AI, from using AI in MT to translate faster, to using recommendation engines for vendor matching. The third will provide practical advice about integrating it into your workflows.


How AI can help you translate and localize better

Use AI to translate faster

AI in MT
As we discussed in the first post in this series, AI powers machine translation engines, which are now used increasingly widely. The translation and localization industry has also adopted new approaches to MT, including augmented translation and adaptive MT, which both put the translator back in the driver’s seat, using their edits to improve MT output. But what does all this mean for translation and localization on the whole? It means speed. This is all the better since we’re seeing a trend for greater numbers of smaller translation jobs with tight turnarounds.

Speech-to-text and text-to-speech
Many freelance translators already use speech-to-text software to work faster. However, the more specialized tools, such as Dragon Naturally Speaking, don’t come cheap. But there are increasing numbers of free or budget options, which are improving all the time. We’re also seeing speech-to-text be integrated into other software packages, including Microsoft and Google, through both proprietary and third-party tools. The opposite direction, text-to-speech, has its advocates too. Similarly, it’s now available in a range of formats, software, and price points.

With more and more of this technology available, it seems inevitable output quality will improve rapidly. And with so many free options out there, perhaps the cost of paid-for software will decrease over time. The end result? Presumably more translators and localizers will embrace this way of working, which in turn will increase their speed. In other words, artificial intelligence is a way of translating and localizing faster.

Use AI to achieve the right style and tone of voice

AI can help you adapt translation suggestions to a translator’s style and voice, something that Smartcat is already making use of. It makes sense then, that AI could also help us adapt to a client’s preferred style and tone of voice. Or perhaps it could do both. In other words, it could tailor suggestions to how Translator A writes for clients B, C, and D. The levels of customization seem almost limitless.

Use AI to create more inclusive and accessible texts

Inclusive writing suggestions
Using AI to sound more human might seem counterintuitive. But it’s a very real possibility. With many organizations prioritizing inclusive writing and creating inclusive writing guides, it seems that this could prove fertile ground for AI technology too. Why not use it to highlight language that’s discriminatory, or excludes certain groups, and suggest inclusive alternatives? As we humans are learning how to think and communicate more inclusively, maybe we can see AI as a tool, helping us be more aware of our habits and tendencies.

Generating alt text for images
In a similar vein, artificial intelligence is a way of automatically generating alt image text. AI uses image recognition to create this text and make images accessible to people who use screen readers, something that Facebook is already using on their platform. MemoQ is also prioritizing alt image text, with its 9.9 release enabling the import of alt text from MS Office documents. Now it’s up to content creators to embrace these tools if they’re not already systematically adding alt text themselves (which also contributes to SEO – more on that below).

The full potential of both these approaches hasn’t yet been realized. And of course, one major challenge is that AI reflects the biases of its creators. How do we mitigate this? By having a diverse group of people involved, both writing the algorithms and training the AI machines. And with translators and localizers hailing from all four corners of the globe, you could argue our industry is particularly well placed to help with this.

Use AI to create more SEO-friendly texts

Search engine optimization (SEO) is increasingly important for the success of any business with a digital presence. And while SEO as a standalone service, and SEO content and copywriting, is nothing new, there has been increasing demand in the fledgling field of SEO translation. SEO translation involves carrying out keyword research for the target market and using the chosen keywords strategically within the translation.

There are already plenty of tools to help copywriters, content writers, and webmasters improve their search engine rankings, including Ubersuggest and Google’s Keyword Planner. Another commonly used tool is Yoast, a WordPress Plugin that analyzes a blog or website for its SEO performance and provides scores and suggestions for improvements. Of course, translation providers can already make use of these tools. But with the rise of SEO translation, perhaps we’ll see tools like these start to be integrated into CAT tools, to help us create more SEO-friendly texts?

How AI can help you optimize the project workflow

Use AI to transcribe meetings

Make the most of speech-to-text tech by transcribing your online meetings. No more furious note-taking or painstaking typing up of minutes. There are now several tools that automate this for you. Otter.ai is one and can be used as a standalone app or within other software such as Microsoft Teams, Google Meet, and Zoom.

The other benefit of transcription is that it allows you to more easily extract information to share with others. Imagine hopping off a client call, and being able to quickly copy and paste the key points to share with your vendors. This will give them more context. And that makes for a happy translator, not to mention better quality texts.

Use AI to automate translation workflows

AI can help you automate repetitive tasks within your translation and localization workflows, enabling your entire team to work more efficiently. It’s also likely to increase staff satisfaction since very few people enjoy mundane work.

Many systems, like XTRF, allow you to choose which parts of the workflow you’d like to automate. And this could vary by client and project. A perfect example of this is vendor selection. Some organizations prefer not to automate this, cherry-picking the right people for each job. But even when that’s the case, they can make the most of filtering to show them all the relevant choices. Or for those who are happy to let the tech take this off their hands, they can set a range of criteria, and use cascading job requests to make sure the job is taken care of in the blink of an eye.

Use AI to categorize and tag projects and match linguists

You will have gathered by now that AI’s specialty is data. Lots of it. It’s built for processing and analyzing vast quantities of information. So maybe we can use it to help categorize and tag translation and localization projects automatically. This way, vendors could be more accurately selected – whether that selection is done by humans or machines.

Let’s look at an example. Say we have an in-depth marketing text selling online software as a service. The subject matter is software. But the style of text is marketing. Ideally, you want the person who works on it to have both in-depth knowledge of IT and experience working on marketing texts. By using tags for both ‘IT’ and ‘marketing’ in your vendor database, and on your project, you – or the system – could more easily find the perfect match.

Use recommendation engines to choose the best workflow

Recommendation engines are what powers those “you might also like this” suggestions on countless websites, including Amazon and Netflix. KantanMT predicts that translation and localization providers will harness this technology to help them choose the best workflow for a given project. It certainly does seem like a logical next step. And if – or when – this is possible, we suggest you get in on the game.

Bartosz Budzynski

Responsible for Professional Services at XTRF. A strong supporter of the open-source movement, sharing economy enthusiast, and a passionate developer. His greatest superpower is transforming business needs into technical requirements.

All stories by : Bartosz Budzynski