Updated: October 16, 2026
I have always been interested in how the human brain works, mostly how we form short-term memories and then how they transition into long-term ones. In particular, I have wondered why and how we remember certain things with greater ease, and yet others, with utmost difficulty. For instance, clever rhymes and limericks, swearwords and songs usually imprint themselves far more easily than certain other phrases. Likewise, memory storage isn't identical in every spoken language, it seems.
Spurred by this phenomenon, I decided to try to run an experiment, without going too PhD about it. In other words, I wanted to put aside the finer aspects of neuroscience, and mostly focus on day-to-day experiences, to see how they translate to memories. In particular, I decided to focus on two topics very dear to me: languages and songs. In other words, I went back to my hobby of singing in many languages, learning new songs, and seeing how this process affects my recall ability for the new words and verses. If you'd like to read a semi-amusing, quarter-scientific article on brain latency (mine in this case), tag along then.
The goal
If we treat this as an experiment, then the purpose of the endeavor is to measure how long it takes for me to learn new songs in new languages, in particular the languages that I don't necessarily speak fluently, or at all. I'd like to see how long it takes for me to remember all the lyrics in any one song, and when those words become a permanent (long-term) memory.
The baseline
I am well aware that my knowledge of languages isn't typical or all-representative. As it happens, I am fluent in three languages (in which I also think when I talk), plus I have reasonable conversational knowledge in a fourth one, and then, excellent understanding of a fifth one albeit with somewhat basic, rudimentary speaking skills due to insufficient continuous practice.
Technically, familiarity (both native and acquired) with various linguistic families allows me to understand and quickly acquire fresh knowledge in quite a few other languages. If you speak any which Slavic language, you understand 80% of all other Slavic languages (check Interslavic, if you will). If you're fluent in Swedish, you won't need much time figuring out Danish or Norwegian. If you speak Islandic, Frisian will be a long-distant cousin. People from Brazil have almost no trouble understanding and speaking South American Spanish, whereas, it's more complicated in the other direction.
Furthermore, I can read pretty much all Latin and Cyrillic scripts, Greek, Arabic, Hebrew, and to some extent, Aramaeic. Lastly, I can sing in about 20-30 languages - some really well, some okayish and/or I'm still learning and improving.
Note: Image taken from Wikimedia, licensed under CC BY-SA 4.0.
That's my baseline, which already includes quite a bit of language skills. Hence, I'm aware that my results and my memory timings will not be applicable to other people, nor do they need to be. What will make the experiment interesting is that it is fully repeatable, and anyone can try their own luck in the same manner.
Learning process
I would go about singing in the following manner:
- Listen to the song once at normal speed.
- Listen to the song at 70-90% speed to get the sense of the phonemes.
- Transcribe the song phonetically so I would not need to learn both language spelling and words.
- Repeat 5 times at slowly increasing speed until 1x (while singing audibly).
- Go away for a few hours, and repeat the process.
- Return the next day, and continue.
Set 1: Four languages
I decided to practice my singing in Latin, Old Norwegian, Spanish, and Brazilian Portuguese. I already knew songs in Spanish and Norse, but the other two were fresh entrants. My choices of songs included:
- Fortune Plango Vulnera (Latin).
- Rolandskvadet (Norwegian).
- Cancion Del Mariachi (Spanish).
- Lambada (Portuguese).
I would cycle through these songs in no particular order, but I would never listen to any one song twice at a time. It would be one, then another, then another, and only after I'd complete the full cycle would I listen to any one of the songs again.
Note: Image taken from Wikimedia, licensed under CC BY-SA 4.0.
Initial results and impressions
What I noticed is the following:
- Initially, until my brain got comfortable with the new sounds, I would somewhat struggle with pronounciation and rhythm. But once the brain learned fully, the new words would become natural. My best example is the Finnish song Ievan Polkka with its many fast and archaic tongue twisters. But once you learn it, you can sing it quickly, without once missing a beat. That song is not included in this test repertoire.
- I would learn some of the verses quite easily, but usually, one or two words in a particular line would elude me. Usually, I would master the verses sequentially, but not always. Sometimes, it would be latter ones that get imprinted first.
- When I stepped away from the computer, my brain would continue "replaying" the songs in the back of my mind, so to speak, nonstop, all the time, including in my sleep, going over the verses over and over and over.
- After a while, the brain would suddenly "remember" the missing words, and I would have the full knowledge of songs. It would be a single word or a line, and the brain would keep repeating until it figured out the sounds. In other words, the information got logged in my head somewhere, but not in a way that I could consciously retrieve and access it. My brain needed time to do that.
It is this process that fascinated me the most, and prompted this article. As it happens, you may have noticed this yourself. If you listen to certain songs on the radio, sometimes, you would be able to sing them sort of absent-mindedly, but if asked to recite them, you'd struggle. Or you might need to hear the first word or the first line of a song to recall it fully, but without that prompt, you'd be stuck.
Well, how quickly did I learn, you ask? We shall discuss that shortly.
Set 2: Five languages (but six songs)
I decided to expand the singing repertoire by adding Italian. It's a language I am quite familiar and cozy with, and I can already sing in. However, I decided to add two rather lovely but not trivial songs, both due to their rich vocabulary and clever cadence, both by Toto Cotugno - Serenata and L'Italiano.
I would then repeat the learning process, singing in the four language from before, plus Italian. In fact, I expanded the test before having fully mastered the previous set. I expanded with two out of four songs mastered (basically 50% mark).
I noticed as I added the extra songs that it became easier for me to learn in all of the languages. This could be due to my past knowledge of Italian, an active language learning process, or simply because offering extra variety in fact improves the linguistic plasticity. But it also affected the entire repertoire.
Set 3: Seven languages (eight songs in total)
With one of the two Italian songs learned, and three out of four songs from the first set fully memorized, and the fourth almost complete, I decided to expand my singing experiment yet again by adding fresh material from two completely new languages:
- Greek (Pyga Se Magisses).
- Mongolian (Chingis Haanii Magtaal) - and this is a language I never ever tried before. A completely different family, completely different sounds, plus the tricky use of the throat singing technique.
What about the actual times then?
Ah well, that's why we are here! Well, on average, my results are as follows:
- Mastering the new sounds so they become natural - usually no more than 5-10 song repetitions.
- Total song recall after listening to it 10 times is about 72 hours. In other words, once I'm done learning a song, usually over a single day or two days of practice, my brain would keep repeating it on loop "internally" for the next two or three days until my brain fully assembled and created a coherent long-term memory.
- Adding more songs did not negatively affect my learning or recall abilities. In fact, when I added the second set, my long-memory "bake" time shortened to about 48 hours.
- After I added Greek and Mongolian, the times stayed at about two days, but my ability to learn new sounds shortened. Rather than having to repeat songs about 10 times, I would feel comfortable with perhaps just five repetitions. Now, it could also be because Greek and Mongolians are relatively simple to learn (for me), at least phonetically.
The language barrier
Furthermore, you might be interested to hear that, for me, the easiest memory-forming song was in Spanish. The hardest, strangely, the Portuguese. The Italian is always easy, by less so when you choose complex songs. I also found the Norwegian quite easy, whereas the Greek gave me more trouble than even the Mongolian. I don't have a good answer as to why.
In general, I find clean-sounding, phonetic languages easier to master, and I usually am more fond of such languages, but that wouldn't explain why the Greek proved somewhat tricky in this exercise. This may actually have to do with the tempo of the songs and the complexity of sung lines rather than any inherent language difficulties.
If I look across my entire repertoire of songs, then interestingly, for example, I found the Albanian, Hindi, Russian, and Czech extremely easy to master, which is probably not something most people would say. On the other than, I "struggle" with the German (and the French) for songs, which cannot easily be explained by my relative fondless and ease of learning of Norwegian or the other Romance languages, respectively.
Now, the purpose of this section ain't to brag, it's to try to understand how languages themselves affect the learning process, to which, my simple answer, for the moment, is that I simply don't know. I don't have sufficient data, and I've not yet clearly mapped the connection between my existing linguistic knowledge (including the native tongue) and the ability to learn new sounds.
If you're curious, you should also add the following singing languages to the list: English (duh), Croatian, Serbian, Bulgarian, Slovak, Romanian, Polish, Hebrew, Arabic, Georgian, Romani, in addition to all the other languages already mentioned in this article, and maybe one or two really arcane ones I cannot remember right now. Before you say X doubt, if you've attended various open-source conferences in the past decade or so, you might have seen me join the local band and sing on stage at the closing ceremonies, in local languages, where they'd let me, of course. For realz.
Next on the list: French (L'ete Indien), German (Hallo Bimmerbahn), and Swedish (Pop, Opp i Topp).
Conclusion
This article isn't too scientific, so don't bring out your pitchforks, please. The purpose of my experiment is mostly personal curiosity and lots and lots of fun. What I can tell you is that I've been learning songs my whole life, so we can probably rule out age-related effects from the equation, as I don't see any increased difficulty in the acquisition of new songs (the 332th Ferengi Rule of Acquisition, as it is). The existing volume of linguistic skills definitely plays part, so my results cannot be compared to other people, but they do give me an indication to how my own brain works, and how it assembles its data into long-term memories.
In general, I found that adding variety to the mix does not negatively impact the learning process. Quite the opposite. The more languages and songs I added, the easier it all became. I do need to expand, to see if the function implodes at some point, he he. Overall, I needed about 10 repetitions to learn a song early on, but fewer and fewer as I grew my list. The memory forming took about three days early on and leveled at about two days half way through the process and stayed like that to the end, or at least, the writing of this piece. This latency seems more important than repetitions, from what I see. Song complexity plays a big part, but I cannot quantify it just yet. Language complexity does not seem to be a major deciding factor, or I cannot break it down to a simple, linear model.
And with that, it's time to wrap this article. I hope you enjoyed it. I know I did. Bye bye.
P.S. Once more, if you find this article a bit too rough, not scientific enough, or you disagree with my methods, please remember, the name of this section is Hillbilly physics. Ergo cogito banjo. Now, bye bye, in earnest.
Cheers.