KINTO Tech Blog
Localization

Language Localization at KINTO Technologies (Part 2)

Cover Image for Language Localization at KINTO Technologies (Part 2)

Introduction

On this second part about localization at KINTO Technologies, we would like to share with you what the team has accomplished so far.

The task at hand

On my previous article I explained how as data type software translations are often stored as a key-value pair and today we will go through how we manage this data.
Having these pairs of keys and terms in mind, we proceed to determine how localization project managers handle the base English and request our Language Service Providers (LSPs) to deliver us back the translations.

It is common to see -also in my experience as a freelance translator- engineers create an Excel sheet with one column of translation keys and a second column with the source language terms that are then sent to translation by adding as many columns as target languages needed. There are three main problems I see with this management style:

  1. Lack of context
  2. Inconsistency issues down the line
  3. Prone to human error

First, Excel sheets are simple rows of text without any context for the translators: is the term a button? Are there any space limitations design-wise? In the case of storytelling, who is speaking and to whom in what kind of situation?

The second problem is that those terms without being fed into a Translation Management System (TMS) will not be recorded anywhere. That can potentially lead to inconsistencies in quality and customer experience in the long run: are we calling it "reservation" or "booking"? And that is only with the source language; imagine this issue multiplied by the number of languages.

Finally, it has a high risk of causing human errors on the development side. When the translations are stored in Excel files, they will eventually need to be painstakingly copy-pasted manually into the source code one by one (the "localizable.strings" file in our example previously). This can lead to copy-paste errors, typos, or even bugs if there is no proper QA to do the final check.

How we solved it

To solve these issues, we have decided not to manage translations in Excel and instead introduced a dedicated tool to manage the translation of our products, called Lokalise.

Lokalise is a Translation Management System that manages translation tasks and allows us to have all terms centralized in one place. Another benefit of the tool is the ability to maintain a Translation Memory (a database that records how sentences have been translated before) and start a glossary creation -a vocabulary list- that can be applied across all our products. It also has the capability to connect to GitHub where the source files are stored, which allows us to "pull" the keys that our developers create for a new screen or new functionality, thereafter, "pushing" back all the target languages translations' key-and-term pairings with a pull request to GitHub.

Automating the process this way not only raises productivity but also reduces the working time for both developers and the localization team. Furthermore, we can also maintain consistent quality, while ensuring that the integrity of the terms is kept and handed over to our engineers accurately. Another enormous benefit of using Lokalise is that it allows us to import screenshots for each of the terms from our design tool Adobe XD to Lokalise, so translators have visual context when translating in the tool and are not blindly typing away on an information-deprived Excel sheet.

Below are the flow diagrams before and after implementing Lokalise for our Global KINTO app, where we also needed to deal with file conversions to have it available for both iOS and Android. Lokalise also takes care of the format conversion when exporting, saving us time:

Next Steps

One of our next steps is to think about scalability. We were able to automate the updates in the source files through Lokalise, but as the pull requests to each of the files are done one by one, it can quickly become a heavy task as the number of GitHub repositories and products grow over time. One solution we are considering is to create an in-house small management tool that would go in-between Lokalise and GitHub, allowing us to centralize all the pull requests with the updates needed on the source files.

As part of our efforts to strengthen our QA processes, another plan is to create a series of style guides, akin to our existing Brand Guidelines. This would further ensure that even translations which are requested to our language service providers are easily understood, maintain the brand's tone and manner without gaps between translators, or even between languages.

There are still more ideas and exciting projects awaiting us on the horizon, such as the potential to apply yokotenkai (horizontal deployment) of our activities to other KINTO services around the world. I might talk about these at a later point, but for now, I thank you for your interest in my post! I hope I was able to spark a bit of curiosity on this topic, and that next time you download an app, use a streaming service or play a video game, you think about the technical complexities at play for multi-region content.

Facebook

関連記事 | Related Posts

Maya.S
Maya.S
Cover Image for Language Localization at KINTO Technologies (Part 1)

Language Localization at KINTO Technologies (Part 1)

Maya.S
Maya.S
Cover Image for KINTOテクノロジーズにおける言語ローカライゼーション(後編)

KINTOテクノロジーズにおける言語ローカライゼーション(後編)

Rina.K
Rina.K
Cover Image for Promoting Output in Tech Blog

Promoting Output in Tech Blog

Maya.S
Maya.S
Cover Image for KINTOテクノロジーズにおける言語ローカライゼーション(前編)

KINTOテクノロジーズにおける言語ローカライゼーション(前編)

JL
JL
Cover Image for My experience as an application engineer in KINTO Technologies

My experience as an application engineer in KINTO Technologies

Cover Image for As a Mom and a Lead at a Cross-cultural Team

As a Mom and a Lead at a Cross-cultural Team

We are hiring!

【PjM】プロジェクト推進G/東京

プロジェクト推進グループについてプロジェクト推進グループでは、​TOYOTAのクルマのサブスクリプションサービスである『 KINTO ONE 』をはじめ、国内向けサービスのプロジェクト立ち上げから運用保守に至るまでの運営管理を行っています。

【分析プロデューサー】分析G/東京・名古屋

分析グループについてKINTOにおいて開発系部門発足時から設置されているチームであり、それほど経営としても注力しているポジションです。決まっていること、分かっていることの方が少ないぐらいですので、常に「なぜ」を考えながら、未知を楽しめるメンバーが集まっております。