Translation

From 2013-2026 I was interested in translyrics – the process of translating song lyrics into a different language, honouring the original text whilst writing rhyme and rhythm befitting of the original soundtrack.

For 10 years, I wrote 57 original translations of songs and published them online. 40 can be sung to the tune of the original soundtrack. I created cover songs of 16 of them and published on youtube and bilibili. I also built a website to park my translyric works over the years: www.lyrikoala.wordpress.com.

How it Began

It all started when a friend challenged me to translate some lyrics by Wang Feng back in 2013. I tend to get a little obsessive when I translate—researching the meaning behind things, thinking about what the creator meant to say, thinking about what the lyrics meant for me personally.

There’s not always an elegant solution to the turning of the phrase, especially when you have quite a few criteria to the translation; but I got pretty close quite often, and when I get the perfect word combination, it’s pretty exhilarating.

0. A Brief History of Translation Styles

Early works focus on bringing out the original meaning of the songs and conveying the translator’s understanding of the song. There was a lot of imagery in Wang Feng’s songs, and that gave me quite a bit of creative freedom.

The line-by-line translation style continued on into later works. Later works have focused on making the works sing-able while also preserving most of the meaning within the original text. If early works challenged my imagination, later works were a game of words trying to rhyme, communicate the original meaning, and also keep within the syllable count.

Perhaps due to the restriction of the song form and my self-imposed restriction of line-by-line translation, later works have become more simple and lost some eloquence. I recognised this and changed my approach in 2022 when I started on the song Gloria.

The transition between Early works and Later works did, however, create some really fantastic translations that could be sung in English. Like Wang Feng’s “Born to Wander”, and Pu Shu’s “A Path in the Ordinary”.

1. The Translation Process

1. Understanding the song and rough draft: I would refer to the original lyrics and translate it roughly first. I would avoid looking at other people’s translations or google translations, since it could limit my imagination. It then takes me 3-4 days of full concentration just thinking about the song, listening to it and jotting down ideas on paper. It also helps to listen to the soundtrack and not focusing too much on the words. This establishes a general flow and captures the feeling.
2. More Ideas: Next, I would brainstorm with the lyrics I had jotted down, imagining different similes and expressions. Usually if there is something I can relate to it makes the writing process much easier. I use thesaurus and check rhyming websites if I’m stuck. I put it through google and deepL to generate extra ideas. When I have some satisfying verses I like to write them out nicely on a piece of paper, then work out the rest of the unfinished ones.
3. Tricky Verses: For especially tricky verses, ask for a second opinion. Usually, people translate/interpret differently and also express differently! I have a friend who is incredibly direct, and they help to stop the fancy-vocab tunnel-vision.
4. Logic and Grammar Check: After the song is about 80% done, it’s important to start checking the logic of the song. Work out any kinks in the song that sound awkward, fix any grammar issues or bad phrasing. Use thesaurus again.
5. Review and Publish: Come back a week later and review with fresh eyes. Often times I have to push myself to just publish something that I’m semi-happy with, and usually a few years later when I look at it again, I’ll be able to fix whatever went wrong last time!

2. Translation style and challenges

In this section, I’d like to discuss my philosophy of translating song lyrics, with reference to the frameworks brought forth by Yan Fu—”信达雅” or Faithfulness, Elegance and Expression (1898).

“信” Faithfulness

I set a self-imposed restriction for myself: line-by-line translation. This is a translation that preserves the meaning of the original lyrics as much as possible. The benefits of this is it preserves the original logic of the song. The difficulty skyrockets with this approach in terms of technicality, depending on how faithful I wanted to be to the original text (hint: I always wanted to be more faithful than the syllable count allows).

In my works, I opt to preserve the tone and feel of the original lyrics as much as possible. I have written some songs that only retain the tone and feeling (as opposed to a more literal translation). This is a much more freeing approach in terms of writing, and suits me since I did not want to completely abandon the original song and write a new one.

“雅” Elegance

Since it is to be sung, it’s important for the phrases to have matching or similar cadence. Some artists might argue that rhyming is the most important thing when it comes to song translation, but I disagree. A verse that only rhymes for rhyming’s sake is poor writing. A smiliar cadence and a half-rhyme can work too, and I would often opt to do that if it could convey the meaning better. If rhyming is impossible at the end, I would use assonance, or add more rhyming in the middle of the verse to create that illusion.

A trick I often use in my translations is switching some phrases from one verse to the other. This is usually for rhyming and enunciation purposes, and one that can help retain all the meaning in the original song. Some words are just too much of a mouthful to sing- then I would opt for a simpler expression at risk of reducing the nuances of the original.

“达“ Expression

At the end of the translation, I always consider the perspective of a native language listener – do the lyrics flow as naturally as possible, is the meaning clear, is it easy to hear/understand when listening to the music? Do I sacrifice elegance for enunciation, or reduce the faithfulness for a more localized adaptation? There is a balance when it comes to this, and my works throughout the years reflect my decisions (struggles).

My earlier works prioritises cohesiveness and original feeling over the lyrics themselves, which resulted in more imaginative and less restricted translations. There were a few years where my works prioritised literal meaning, which created more rigid translations – phrases were simplified due to restrictions, and when the original lyrics lacks logic the translated song simply fell apart. Eventually, I started to take the lyrics less literal and found it to be much more fun!

For me, the translyrics hierarchy goes: Cohesiveness > Cadence > Original Feel > Rhyming. The song sounds better when it is relaxed and with only some restrictions.

3. Copyright Challenges

There were a few challenges when it comes to getting permission to post and perform the works:

• Translyric projects are considered derivative works of the original. Unfortunately, there isn’t a very good way of getting proper copyright permission for translyric projects. It not only requires the usual copyright consent (license for the cover, own backing tracks, mechanical license to sync the music to a video), but also requires the publishers and the writers of the song to give written consent.
• Youtube changed its terms of service: users with less than 1000 followers cannot opt out of ads on their videos. Ad revenue is shared between youtube and copyright claimant.
• False copyright claim on youtube.
• Bilibili and Netease removed songs from their platform by request of Tencent.

4. The role of artificial intelligence (2025 answer)

The question every artist asks themselves is: can AI do what I do?

After chatgpt appeared, it was clear that it could not do exactly what I was doing. The pros were it was fast, it could generate different ideas and ways to write about a thought in the verse, it rhymed, it had some idea of cadence; the cons were it couldn’t figure out the syllables, the tone of writing tends towards archaic, the words used were hard to pronounce when singing, the cadence did not fit the song and it could not be performed in the original track without some serious work.

For example, I only have one song that took inspiration from chatgpt, and it was the last one of 2023: Darling, that’s not love. Chatgpt generated “star without a wish”, which was used in the final version of the song. All other lines were unusable.

However, ai did dampen my enthusiasm for this hobby, as it excels in translating. For example, it may take 3-4 days of full concentration to work out a few nice ways to construct certain verses, GPT can give you 5 iterations in seconds, despite the usability of those verses in a songs’ context.

I would say Chatgpt did help me a lot on one thing: I finally used it on my research of the manjusaka, and it was able to point me in the direction of translated sanskrit texts, which I previously had no way to go deeper in my research!

In conclusion, Ai is good for idea generation and basic research (compared to a complete newbie). Go online and you can still find many translyrics artists still doing what they love despite the rise of ai.

5. Concluding statement

It seems crazy that when I look back it’s already been 10 years! But it’s been good! 🙂