Mnemonic Input [Experiment]

DRAFT VERSION – will be updated

Hello, I’m a polyglot and somewhat of a hobbyist in the art of memory, and speed math.

My topic is language acquisition: i have found great succeess with applying the principles of input hypothesis within my acquisition process, and i feel that I’m well versed enough to conduct my own personal journey and build a system and a strategy that will make the whole L2 acquiring process better.

One implementation is the use of mnemonics… well rather than mnemonics its the use of active recalls. Input at the beginner phase is always drawn out and relies primarily on endurance, you cannot acquire what you can’t understand and in a new language, your favorite novel a rocky terrain of confusion and pain. Many instead uses flashcards to undercut this process and create a vocabulary of active recall in order to start understanding a material they like to absorb through context, however through various testing i have found that the approach of using flashcards could be surpassed in terms of speed and permanence, and that is where i welcome memory palace into my process.

Some people think that mnemonics or memory palaces are useless, and its true in a certain way but i believe that way is the misguided view on the purpose of mnemonics in L2 acquisition. It comes from the belief that knowing a definition and then the form of that definition (word/character) equates to knowing the L2. Wrong.

If i may explain how language works i would use the layer analogy. So imagine at the very base which we would call layer 0 is a collection of all possible meanings and expressions. Aka, concepts. On the layer above that is the language which does not stand alone but is what is used to not only externally express layer 0 with other human being but to also internally syllogize/logic-ize layer 0 with the self.

Essentially layer 1 is the labels of all those sets of concepts below it. Which consequentially what makes language a language. If one has acquired language to this level then it is very fast and immediate as it requires no calculation, ONLY pure reference. Layer 1 is simply what people would call natural. Going above this layer and adding another on top of it would then make it a little more complicated, and then recalling those words becomes a three layer process which is going to be intrinsically slow because your brain will now have to translate layer 2 to layer 1 just to get the actual meaning of a word that is another layer lower, and its just a word! In the case of flashcards and mnemonics both are layer 2s. One takes the form of raw data and another uses visuals.

Remember that this layer explanation is not how the brain actually deals with languages but this is the best abstraction i could make to convey my refutes to the common dismissive remarks, it is not logical nor proveable but what i do have here is testable, and recreatable. So you know what to do.

There are two mental tests and a few anecdotes that I have observed that pushes me to these conclusions. (TBA – tldr, memorizing chinese characters with memory palace in one sitting is as effective as using flashcards periodically, and active recall is much better than artificial or conscious translations)

Now, to set aside all these pendantic explanations let’s get on to the most important part, to answer the subtextual inquiry that you have: what then is the purpose of memory palace in L2 acquisition?

I find that it is impossible to use mnemonics or flashcards to build layer 1. Essentially the reason for this is that layer 1 is actually built to layer 0 through feelings or stimulis. The process occurs organically. Aka leave it to input.

With that in mind one can use this to accept defeat and let mnemonics and flashcards be just what they are, that they cannot become a 1:1 understanding of a language. They cannot be used to acquire the language. In the analogy, let layer 2 be layer 2.

But then we know that layer 2 can produce results and could be used despite being slower and innacurate, an example is that sometimes knowing just a funny meme on the internet about a french word can help you get through that one french phrase witihout having to look it up on the dictionary extensively.

In simpler terms, mnemonics must serve to facilitate the acquisition process rather than replace it.

More tests are to be done within comparisons of artificially/digitally translating words while reading as opposed to recalling from a memory palace. If you have extensive results with this plelase mention me.

There are two categories that I approach this practically:

  1. Methodology
  2. Strategy

Methodology refers to what techniques are employed in the loci storing process, and what is it combined with in the input process.

Strategy involves breaking down how to understand context easier or how to organize the techniques in a way that it’d be efficient.

There are many methodologies and strategies that fits this concept however a few are being tested for results and are revealed in this thread.


Memory Palace (Common)

Character → Picture Association: essentially turnitng a character to its meaning using an image.


Common noun-centric:

Guessing the meaning of words is very important when it comes to input, what helps this is knowing that the chances of understanding through context is better when there are less possible meanings for an entire phrase. In other words deduction, centering memorization of words with nouns would help in well… recognizing the nouns, which also translates to recognizing the subject and knowing the subject limits what the verbs and conjunctions can possibly be.

For example if you read an unknown french verb like «mange» and then a few moments later you read «gâteau».

Simply if you know that «gâteau» means cake then the verb «mange» could only mean a few things, it is something that you do with a cake. Need i say? Eat or bake. Now if you read more nouns like «le salon» after that as in “the living room”. Well, what other meanings it could possibly have? (a lot but assumption is key)

So reading: je mange un gâteau dans le salon.

You could infer the meaning better. Now as opposed to being verb-centric:

je mange un poisson dans la cuisine.

You know what «mange» is but then what is «poisson» and what is «cuisine»? well then how would you form the image in your head when you don’t know the subject. Exactly, you have to translate and it takes away from the flow or from the potential of the narrative to become compelling. This is painful and sometimes could come to naught.

This will be updated. There are several docs and journals i have written on my own progress on this experiment. Use what you may find useful and if you could help then thank you.

This version is released to outline the goal of my process and what my visions are. There’s more substance to this, once the initial testing and writeups are done this will be updated to greater detail, and will be organized accordingly.

– 0.1–


Have you experimented with memory palaces and mnemonics both in the target language? Just a thought (not tried it yet)?

I’m learning Latin through input - there is a book “Lingua Latina per se Illustrata” which is totally in Latin, not a word of English. I’ve used memory techniques to nail down annoying words, and also declensions. The emphasis is, however through input first. I only use mnemonics once I’ve grasped the general “essence” of the word, as a hint for myself should the meaning slip my mind.

I think one of the problems with modern language teaching is the belief that you can find a mapping between English and your target language. This is not true, as languages may use different constructs in the same situation and have words which have no direct translation to English.


Yup exactly what I meant also. Input being organic means that vocabulary and grammar has to map to the concepts or their denotations rather than to your L1/native languages.

The essence of the language is a stimulus that our brain associates with a form (word). That is why mnemonics should be used in a way that it can faciliatate the access of that essence rather than replace it.

From my current findings, memory palace is mrginally more efficient than the mainstream techniques and methods used today (flashcard, 1:1 translation, songs etc.)

I will take a bit longer to update here because I’m developing three different systems that might or might not work. One being a phonetic table system which i haven’t had success with, another two are some form of modified memory palace that is constructed specifically for languages.

Good luck with youp latin journey!

As of this very moment a larger scale experiment will be done using the most barebones of memory palace, noun-centric storategy, and kanji to image method. Nothing else, and the recall results will be recorded for a whole week. And then the acqusition result after that.

Frequency List: Frequency lists by Neri: The 2000 Most Frequently Used Japanese Nouns

Time starts now


I believe you’re more experienced with this than me.

Thanks for the vote of confidence. Who are you?

Hello Clotho,
I use a few dozen memory places to Learn Levantine Arabic vocabulary. I would estimate I close to 1,500 words/stations packed in places that I can actively recall.

If you wish to talk about it you can message me.

Take care.

1 Like

Thank you!

Actually, that’s very similar to my initial goal for this experiment. Your insight would help a lot. Is it fine if i send you a set of questions that you could answer if you have any free time.

Once again, thank you for getting back to me.

On the meantime, I need to get the edit feature before I could start updating this thread.

Aaron from Linguisticator did some great work in this area before he got sidetracked into VR software.

I think emotion is the key. Linking a word to an emotion such as surprise. If the experience is new, it’s retained.

I tried linking perspective with the tones in Chinese. It worked but too much logic broke it. For example, the falling to rising tone sounds unusual to me so I always remember it. Also the brushing, sweeping sounds of Mandarin and the consonant that is closest to an /r/ that reminds me of Klingon somehow.

Before you go any further, I suggest working on the new sounds that you don’t have in your L1. Get a handle on those. Make them larger than life.

Send any questions you have. I will answer the best I can.

Apologies for the lack of response. I have halted the experiment for now as I suddenly came down with COVID19 right after the initial first day of 100 characters memorized, I’ll get back to you and to this thread as soon as I fully recover. Thank you.

Get well soon buddy.

1 Like