Tagalog Translation Project: How An Unfamiliar Script Was Able To Benefit From Machine Translation

Let’s get a quick orientation on Tagalog before jumping into the project itself. Tagalog is a language belonging to the Austronesian family. While not as well-known as other Asian languages, it still has around 22 million native speakers residing in the Philippines. The ethnic group of the Tagalog people consists of about a quarter of the population, but the majority of Philippine people speak Tagalog as a second language.

Just like other languages spoken in the Philippines, today Tagalog is written with the Latin alphabet. But this wasn’t always the case. Before the Spanish started to colonize the Philippines back in the 16th century, Tagalog was written in an abugida called Baybayin. 

When we received the request to translate a company’s sizeable chat backlog from Tagalog to English using Machine Translation (MT), this difference between the written and the spoken language was our primary challenge.

The Challenge

While the written language of Tagalog may now be communicated through a Latin script, it is still pronounced in its original way. The client needed a translation of the chat backlog using MT since the document was intended for internal use, and accuracy and quality were not a priority. 

If the text would have been intended for external use, we would have recommended the use of a Post-Editing Process where a qualified linguist applies their expertise to the MT output.

Since the language was not written in its original characters, we faced the challenge that the MT-Engine couldn’t recognize the language written in the company’s chat history, and so an MT could not be produced.

The Solution

MT-Engines should always be selected based on the language pairs to achieve the most optimal translation output. And while we had selected the correct MT-Engine, we still had to find a way to revert the Latin script to its original without resorting to the time-consuming job of setting a native linguist the task of going through the entire log.

Eventually, we found a Google Neural MT-Engine that was compatible with the Tagalog – English language pair and could also produce a transliteration to the original script. With this transliteration, we were able to deliver a Tagalog to English auto-translation.

Make sure you select an LSP who is willing to go the extra mile to find a solution for your translation needs and who can apply the technical insights to actually find it.

Are you facing a similar challenge translating rare languages? We can help! Contact us, we remain available 24/7.

Contact Us