Customer Experience

Is Google Translate Accurate? Evaluating Google Translate’s Accuracy

Learn how accurate Google Translate is for different languages, how Google Translate works, and the method to achieve the most accurate translations.

Veronica Pastuch's avatar
Veronica Pastuch

September 28, 2022


How Does Google Translate Work?

Whether you’re a tourist who wants to feel comfortable in a new city or looking to expand your business to another country, Google Translate is there to help. Google Translate was launched in 2006 using Phrase-Based Machine Translation (also called Statistical Machine Translation). These first translation systems worked by building statistical models of relationships between words, phrases, and sentences and then applied that model to convert text into a new language.

By 2016, Google Translate upgraded and moved to a Neural Machine Translation (NMT) engine. NMT works by using advanced algorithms called neural networks in addition to artificial intelligence (AI). NMT produces translations for large blocks of text and uses ‘deep learning’. Using simulations based on the brain, scientists can teach machines to translate closer to humans. Neural machine translation learns from mistakes and can correctly translate something on the next try.

Google Translate is a valuable tool for delivering content to people worldwide, and it currently supports 133 languages. There are several ways to use Google Translate through a computer or smartphone app. Unfortunately, not every Google Translate solution is available for every language. Here’s how to utilize it:

  • Type: Use your computer or phone keyboard to type text into the Google Translate box, and your text will automatically translate to your language of choice.
  • Write: In the mobile app, press the pen symbol and write what you want to translate instead of typing.
  • Talk: Press “Conversation” on the mobile app and begin speaking, and you’ll see your words show up in the top half of your screen. The translated version will be translated in the bottom half of your screen in real-time. Once completed, the translated text will automatically be read out loud.
  • Snap: The Google Translate mobile app also translates content from a photo. Point your camera at the text to translate, such as a street sign, menu, or document, and take a picture. After a moment, the translation is superimposed over the image.
  • See: This is similar to the “Snap” feature above, but without needing to snap a photo. Once the app has a moment to recognize the text, the translation is superimposed over the original text, viewable right on the screen.
  • Transcribe: Google Translate can “listen” to a single audio source to continuously translate speech. Tap the microphone icon at the top of the screen and speak into the app. Google Translate then translates your words into the target language. Then, tap the Speaker icon to hear the translation. Another option is to tap the Transcribe icon and start speaking. Tap the microphone icon when done. You can then select and copy the transcription to paste it elsewhere and save your transcript for future use.
  • Offline: Many of the languages supported by Google Translate can still be translated if you’re offline by allowing you to download an offline translation language file.

Testing Google Translate’s Accuracy

Advantages of Google Translate

The pros and cons of Google translate impact professional translators and anyone who chooses to use it as a translation tool. Some advantages of Google Translate include:

Speed and volume: NMT is fast and can translate millions of words almost instantaneously while continually improving as more content is translated. In an average workday, an experienced translator can translate about 2,000 words maximum (300-400 words/hour) depending on the difficulty of the text. In contrast, Google Translate can produce a translation with the same number of words in seconds. For high-volume projects, NMT can not only handle volume at speed, but it can also work with content management systems to organize and tag content. This makes it possible to retain organization and context as the content is translated into multiple languages.

Consistency and Accuracy: Translations done by NMT create accurate, contextual information. NMT translates whole sentences rather than individual words, making translations faster and more accurate. NMT also requires less memory and data to perform well.

Save time and money: AI-based translation can save time and money. While human translators won’t be replaced entirely, an AI-based translation can offer machine translation that results in easier editing. This can significantly reduce the cost and time required by human translators. In addition, for companies with limited resources, an AI-based translation offers an opportunity to scale globally.

Disadvantages of Google Translate

Google Translate is a beneficial tool, but it’s not without drawbacks.

Accuracy: Because Google Translate uses machine learning, languages not widely used in the tool will have a higher inaccuracy rate than more widely used languages such as Spanish or French. Translation quality varies based on the chosen language pair. Some of the most accurate translations are the Spanish-English language pair. Other languages have a lower level of accuracy, including Chinese, German, and Portuguese.

Cultural context: Google Translate can have a significant impact on multilingual websites. Google Translate doesn’t take the human context of a given sentence into account: it struggles with cultural nuances like traditions, gestures, locality, and more. Machine translation also can’t decipher non-text elements like images or symbols, which can have wildly different meanings in different locales.

Quality of translation: If you have multiple versions of your website for different languages, the translation quality of one site can negatively impact the higher-quality version. Potential customers in new markets also need to find you on the web if they don’t know who you are. Google Translate focuses on translating content but doesn’t create searchable, indexable versions of those translated pages for Search Engine Optimization.

Grammar and layout issues: Google Translate does not consider grammar rules, so it often creates translations that contain grammatical errors. There’s also no quality control or adaptation of the text to specific jargon, layout, industry, or market. Finally, Google Translate doesn’t consider web design when translating text, which can affect your website layout and user experience as a whole.

Google Translate vs. Human Translation

MotionPoint human translations are deployed by professional linguistics, while Google Translate is a machine-only approach that uses artificial intelligence (AI) technology.

Here is an example of how Google Translations compares to our expert Human Translations:

Spanish Segment: ¿No sabe bien qué significa traducir un sitio web? Le contamos todo sobre tecnologías de traducción, prácticas comunes y mucho más.

This text Translates to…

Human Translator: “Mystified by website translation? Get answers to your questions about translation technologies, best practices and more” by a human translator.

Google Translate: “Not sure what it means to translate a website? We tell you all about translation technologies, common practices and much more” by Google Translate.

Machine translation can be a valuable tool provided you understand its limitations

When Does Machine Translation Make Sense?

While there are pros and cons of machine translation, there are situations when using machine translation can be useful:

Where large volumes of content need to be translated: Translating websites or large documents is not always easy to accomplish by human translators only. However, machines can run uninterrupted for long periods and instantly deliver results. Human translators can then edit and review the machine translations to ensure that the final content is polished and accurate.

Where nuances are not important: Materials that offer straightforward content can benefit from machine translation. For example, software documentation or technical manuals are materials that don’t need cultural context or flowery language. Machine translators work by translating singular words or sentences in parallel. The result is a collection of translated sentences in sequence, not a cohesively translated text.

If a translation budget is limited: Companies turn to machine translation for their work because machines cost less than human labor. However, to lower costs and maintain the integrity of the translated text, some businesses prefer to use a mix of both machine and human translators.

Where content is subject to change: Content such as customer reviews, emails, or FAQs are generated quickly, have only one-off usage, or are only relevant for limited amounts of time. Therefore, their quality doesn’t necessarily need to be as high as that of professional documents.

For Accurate Translations, Use Professional Linguists

For businesses that want to leverage machine translation’s cost-effective, time-saving benefits, use a hybrid approach for accurate translation by combining NMT with expert human translation. This delivers efficiency, accuracy, and flexibility.

Machine translation will never outperform human translators, but it can still be a helpful tool. For example, many professional translators use machine translation to create the first draft of a translation, and the machine-generated draft is then post-edited by a skilled human translator. This can save time and lower translation costs. MotionPoint is a translation and localization company combining cutting-edge AI technology with expert linguists to provide well-rounded, effective content translations to meet your budget, needs, and, most importantly, global outreach goals.

Last updated on September 28, 2022
Veronica Pastuch's avatar

About Veronica Pastuch

Throughout her 15-year career leading successful multicultural teams in the website localization industry, Veronica Pastuch has directly contributed to the translation, deployment and ongoing support of more than 1,500 multilingual websites, with an absolute commitment to superior quality and customer satisfaction.

Veronica Pastuch's avatar
Veronica Pastuch

EVP of Translation Operations


Keep Learning

Here is some related information that you might enjoy

Can Neural Machine Translation Compete with Human Translation? (Part 1)

Can Neural Machine Translation Compete with Human Translation? (Part 1)

Can Neural Machine Translation Compete with Human Translation? (Part 2)

Can Neural Machine Translation Compete with Human Translation? (Part 2)

Evaluating the Pros and Cons of Human and Machine Translation

Evaluating the Pros and Cons of Human and Machine Translation

The Shortcomings of Machine Translation

The Shortcomings of Machine Translation