Translation is a pivotal action in the globalization strategy for your business. By making your product available in different languages, you potentially reach new audiences that help grow your visibility and your revenue.
Machine translation is a popular term, but many questions are often asked around this topic: What is the extent of automatic translation? What is the advantage of machine translation?
This article aims to give you all the information you need about machine translation to integrate it into your translation workflow in the best possible way.
What is Machine Translation?
Machine translation—also known as automatic translation—consists of translating a text from one language to another automatically and with the help of machine translation software.
While automatic translation gained popularity in the early 2000s, it is in the last decade that developers flipped a switch in the industry, when they started using artificial intelligence (AI) to boost machine translation speed and efficiency.
Best (and Not So Great) Uses of Machine Translation
Jumping into localizing your product is exciting. However, figuring out how to tackle the challenge can be intimidating. You have heard of machine translation before and think it is a great opportunity to see what it is all about. Let us guide you through the good and the bad ways you can use automatic translation to globalize your business:
When it is a Good Idea
Countless businesses trust machine translation with the globalization of their products. In fact, there are some aspects of translation that are perfectly handled by a machine to help you save time and money. Here are some of them:
| Aspects | Examples |
| For external and internal communication | Internally: it helps with presentations, emails, reports… for your international team to understand your brand TOV. Externally: end users understand your offer. Automatic translation of feedback builds trust and more room for reflection. |
| To assist online customer service | Via email or live chat. Simultaneous machine-translated conversations make communication smoother. |
| To improve data analysis | To access more user comments and understand what is being said about your brand globally. |
When You Should Avoid it
While machine translation is great in various situations, it has the opposite effect in others. Typically, automatic translation works great with generic topics. When content becomes too technical, specific or creative, the output gets confusing:
| Aspects | Examples |
| Your topic is too niche | If your topic is detailed and narrow, more context is needed to localize, which is not available with MT. |
| Cultural nuances are important | Content can vary from one market to another; cities, national events, currencies, weather… MT does not take that into consideration. |
| Your content is creative | Poems, songs & creative pieces might be translated word-for-word by MT, which would not make sense in other languages. |
How Machine Translation Benefits Your Business
You can see the benefits of machine translation as soon as you use it. Consider them when building your localization strategy to understand how it can help the whole translation process:
Immediate Output
You cannot get your content delivered faster than with machine translation. In fact, you input what needs to be translated, and the machine gives it back to you almost immediately. For larger volumes, it might take a little longer, but it is still faster than human translation.
Great for Urgent Matters
Given how immediate the output of automatic translation is, you should consider it when you have urgent needs that cannot be completed by a team of human translators.
Small-Budget Friendly
MT is valuable for companies that need quick translation of their offer at low cost. It is a first-time strategic approach for businesses that want to try translation without taking a big financial risk.
Easily Scalable
Small requests, big batches of content… Machine translation does it all! Automatic translation works for all types of volumes, so you can focus on small translated sections of your product, or its entirety in one go.
Wide Range of Languages
The sky is—almost—the limit! Some machine translation providers offer translations in up to 200 languages. Thanks to that, you can reach audiences from all over the world easily.
Customization Feature
You can let machine translation do its job by itself. But if you need more specific outcomes, some automatic translation providers offer a customization feature: you feed it data for more accurate results over time.
Seamless Integration
There are different layers to MT integration: it can be in your translation management system (if you use one) to assist your localization process. Another option is directly on your website, for your users to choose when they visit your page.
24/7 Availability
Machines do not sleep! It does not matter when you need your content translated, as providers are available all the time. This is particularly useful in highly urgent situations.
Limitations of Machine Translation
Pros, unfortunately, do not come without cons. Since automatic translation helps you with a great range of topics, it has limitations in others. Also consider these aspects when thinking about integrating machine translation into your localization workflow:
Privacy Issues
This aspect is probably the most important to consider as a company. When inputting your content into a machine translation provider, your data is analyzed and kept for training purposes. Sensitive information is therefore released to the World Wide Web.
Quality Issues
MT will translate everything you feed it. However, linguistic nuances such as synonyms, figures of speech, etc. are not its priority. Content is delivered with little attention to writing style, impacting the quality of the translation.
Lack of Context
As stated above, automatic translation does not have any context beyond the text you provide. Some words require different translations depending on the context, which machine translation does not account for.
Lack of Localization
Machine translation is, as its name implies, good at translating. Localization is not one of its strengths. This means that it will not change information that needs to be edited for specific markets, creating inconsistencies in the translated product.
Not Always Self-Sufficient
While using machine translation helps you save time and money in your localization process, it sometimes needs an extra pair of human eyes to check quality, consistency, and to edit what was not localized.
🧠What about AI?
Machine translation and AI translation are two different concepts. Using technologies like machine learning, AI translation takes more factors into consideration, like date and time format and measurement units, for more accurate translation.
Types of Machine Translation
Machine translation is no longer a generic system. Different types of MT exist for different purposes. Three of the most common types of machine translation are:
| Type | Key aspects |
| Rule-based (RBMT) | Earliest versionPredefined linguistic rulesHuman input neededLower quality |
| Statistical (SMT) | Statistical model of relationships between words in a text.Applies the same model to target languagesLower quality |
| Neural (NMT) | Most recent versionNo predefined rulesEncodes and decodes the source textMore languages availableHigher qualityFaster output |
Automated vs Machine Translation
Not to be confused with automatic translation, automated translation is a different process from machine translation. Both terms are sometimes used interchangeably, but they do not refer to the same step. Learn the difference between the two:
Automated translation—automating the translation process on a localization content management system. It organizes files, locales, and assigns tasks to the relevant human translators (or machine translation if it is the strategy used). Automated translation is not the action of translating. It is the action of organizing the localization workflow.
Machine translation—translating content with the help of a machine. It is the action of translating.
Most Popular Machine Translation Providers
You are ready to dive into your first translation project, and machine translation is the winner. Many of them are available on the market, but which one is the best for your product? Have a look at the most popular ones:
Google Translate
Probably the most widely used provider. Google Translate uses neural translation technology to offer translations in a wide variety of languages.
| Pros | Cons |
| Users make it better by translating with itFree to useNo translation volume limit | Public platform = not suitable for sensitive dataNot customizable |
Amazon Translate
Like Google Translate, Amazon Translate uses neural technology to provide more accurate translations to its end users.
| Pros | Cons |
| Pay-per-use model. Great for different project volumesTranslation in bulk thanks to its application programming interface (API) | Requires coding skills |
DeepL
DeepL is built on a machine learning model. This means one text gets better translated over time, providing high-quality output for end users.
| Pros | Cons |
| A paid plan allows unlimited translation on browser, desktop and mobileMachine learning model is better at taking context in consideration | The per-user translation limit is low on some paid plansOffers fewer languages than its competitors |
As machine translation is not a one-size-fits-all process, exploring all the alternatives and comparing them to your needs helps you pick the right provider for your localization project. Embracing the qualities and flaws of machine translation software allows you to get the most out of it without being disappointed down the line.
How to Implement Machine Translation
Choosing automatic translation to globalize your product is one aspect of your translation project. Implementing machine translation in your workflow requires taking many factors into account to understand what your localization journey will look like.
| Factors | Questions to consider |
| Your goals | What exactly do you want translated?What does it represent in volume?Is it the entirety of your product? |
| Existing content | What is already available on your website/app?Is it translatable with machine translation? |
| Target languages | Which markets do you want to globalize your business to?How many languages does that represent? |
| Timeline and budget | When do you need your content translated for?What is your budget for this project? |
| Your team and strategy | Will a translation team work on your project?Do you have a clear and solid strategy? |
| Machine translation provider | Which machine translation provider works best for your needs?Do you work via a TMS? Or directly via the MT provider? |
| Run translation testing | Did you run translation audits after the first batch of content translated?What feedback did you collect? |
The more information you have about your localization project, the more efficient your machine translation process will be. Having clear goals and KPIs helps you understand what tools you need to build a bulletproof translated product.
💡Did you consider?
Going for 100% human translation or 100% machine translation is not the only solution. Some businesses choose hybrid machine translation to get the best out of both worlds: fast turnaround, lower budget, and human post-editing to ensure high quality and consistency!
Lokalise AI and Machine Translation
Lokalise AI is human-like translation at superhuman speed. By evaluating the quality of different translation engines (Google Translate, Microsoft Translator, and similar services) with the power of artificial intelligence, you always get the best option for your product, immediately. You save up to 80% in translation costs with 80% fewer post-edits needed, resulting in output 10 times faster than human translation. Your tone of voice remains intact and the quality of your translations is not altered.
