Thoughts on Phonetic Systems and Memory Techniques as an Artificial Language

When I was deciding on what memory system to use, I tried to think about all the steps that the brain goes through when memorizing, and how to make it as efficient as possible. This post describes some of the things I’ve been experimenting with over last summer, and why I keep my system strictly phonetic.

I’m not saying that this is the way things should be done. I haven’t finished my system yet, so I am just thinking out loud (as usual). I’m having fun experimenting, and I don’t think that there will be any detrimental effect to my memory system if I turn out to be wrong, other than that generating the system is taking a longer time than it would otherwise.

The Basic Steps of Memorizing

To start, here are the steps that I think a person goes through when memorizing data with a typical system:

  1. Chunking the data — Cards are usually chunked as just one or two, and numbers typically two, three, or four. It seems like the larger the chunk of data, the fewer times this process has to be repeated. For decimal digits, chunks of two typically require between 100 and 300 images, and chunks of three require 1,000 images. Some people are trying to build systems with 10,000 images, which would allow the chunking of four decimal digits at a time.
  2. Retrieval of images — Each chunk has to be converted into an image or part of a compound image (for PA or PAO systems). At this point, the brain should be able to go from digits to image without having to read the numbers phonetically. I have a feeling that the brain uses words as unconscious anchor points though. More on that below.
  3. Repeat steps one and two until all of the images for a single locus are encoded. The more images that can be placed in one locus, the fewer times one has to go through the whole cycle.
  4. Retrieval of locus — the brain has to quickly search to find the current place in the memory journey.
  5. Placement of image(s) in locus — the brain deposits between one and three images (or more) in the locus, and then goes back to step #1, above

Step #2 above can be further broken down in the phonetic system I’m using:

  1. Convert numbers into letters — “057” becomes “SAIK” according to the phonetic system I’m using [PDF]. (It is basically the Ben System, but with different phonetics.)
  2. Convert the sound into a word — “SAIK” becomes “Psychic Lady” (from TV infomercials).
  3. Convert the word to an image — the word becomes a vivid mental image of a psychic lady.

After enough practice, there is no need to phonetically sound out the numbers—the brain goes directly from 057 to an image of a psychic lady.

Unconscious Mental Reference Points for Numbers

Despite not having to phonetically sound out the numbers to get to the image, I wonder if the brain uses unconscious anchor points to reference the image. Check out this excerpt from the book Outliers by Malcolm Gladwell:

Take a look at the following list of numbers: 4,8,5,3,9,7,6. Read them out loud to yourself. Now look away, and spend twenty seconds memorizing that sequence before saying them out loud again.

If you speak English, you have about a 50 percent chance of remembering that sequence perfectly If you’re Chinese, though, you’re almost certain to get it right every time. Why is that? Because as human beings we store digits in a memory loop that runs for about two seconds. We most easily memorize whatever we can say or read within that two second span. And Chinese speakers get that list of numbers—4,8,5,3,9,7,6—right every time because—unlike English speakers—their language allows them to fit all those seven numbers into two seconds.

That example comes from Stanislas Dehaene’s book “The Number Sense,” and as Dehaene explains:

“Chinese number words are remarkably brief. Most of them can be uttered in less than one-quarter of a second (for instance, 4 is ‘si’ and 7 ‘qi’) Their English equivalents—“four,” “seven”—are longer: pronouncing them takes about one-third of a second. The memory gap between English and Chinese apparently is entirely due to this difference in length. In languages as diverse as Welsh, Arabic, Chinese, English and Hebrew, there is a reproducible correlation between the time required to pronounce numbers in a given language and the memory span of its speakers. In this domain, the prize for efficacy goes to the Cantonese dialect of Chinese, whose brevity grants residents of Hong Kong a rocketing memory span of about 10 digits.” If this is correct, the length of a word that a number is mentally referenced by has an effect on memory.

Language and Memorization

I think that language definitely has a relation to memory. I used to study natural history and, when I was in the wilderness, I would often see plants that I had never seen before. I realized that if I had a name for the plant, I could remember it more easily. If I didn’t have a field guide handy, I would sometimes make up a name for a strange plant so that I could remember if I saw it again. (E.g., a brittle flower might temporarily be called “paper flower”.) I could then replace the fake name with the real name once I looked it up in a field guide.

I’ve also noticed that if I have a concept in my head and I give it a specific name, or I read about it having a specific name, I can pull that idea back out of my head more easily later.

Giving something a name definitely seems to lock that thing into my brain more securely and helps me recognize it when I encounter it again.

The Artificial Language

I keep my system strictly phonetic, and I’m trying to anchor the system to the language part of my brain. Every number/image gets an artificial name and it is only one syllable, without exception.

I’ve spent a long time in countries where they don’t use a Latin alphabet, and it has always been fascinating to me to try to learn other writing systems. When I discovered memory techniques, especially the Ben System, it immediately looked to me like a system for writing language.

The meaning and pronunciation of symbols is arbitrary. There is nothing that says the symbol “5” has to be pronounced “five” or have any specific meaning. A brain can be trained to read the symbols “744” as “car”, “614” as “bir”, or “470” as “res”.

My experiment is that I’m trying to take the symbols that we know as digits and card values and literally think of them as letters–not as a stepping stone to images, but as language.

Instead of my brain unconsciously referencing 477 as “four hundred and seventy seven” (8 syllables) and then converting that to “wrecking ball” (3 syllables), it references both the number and the image only as a one-syllable sound, “rek”. “Rek” becomes a new word that means an image of a wrecking ball. I still go straight from the number to the image, but remove these longer, unconscious reference points in the brain: “four hundred and seventy seven = wrecking ball”. Based on the information in the Malcolm Gladwell quote above, all my images and numbers are reduced to one syllable. It is possible to verbally recite 100 binary digits in less than two seconds without rushing.

These are the 1-syllable, unconscious reference points for 70 digits: “SOS, SOT, SON, SOM, SOR, SOL, SOB”

It is an artificial language that can be superimposed on top of these 70 symbols:


and also these symbols:

10♠ 10♠ 10♠ A♠ 10♠ 2♠ 10♠ 3♠ 10♠ 4♠ 10♠ 5♠ 10♠ 6♠

and these symbols:


With a strictly phonetic system that is made up of artificial one-syllable words, reading a line of numbers should be similar to reading a line of words in English. The sequence of 70 binary digits above becomes seven one-syllable bits of information that can be held in short-term memory almost as easily as a phone number, without even having to place them in loci.

Death of Socrates

Sounding Out the Letters

In the beginning, it is slower to phonetically spell out numbers into words, but, with practice, the jump from number to image is just as fast as with non-phonetic systems. One doesn’t look at English words and phonetically sound them out; they are instantly assembled into meanings with just a glance. When looking at a word like “car” I doubt people spell out C-A-R — the recognition is instant and unconscious.

The symbols on each of the lines below could phonetically sound out “car”, and they are all equally fast to read with practice:

  • car
  • 744
  • καρ
  • כר
  • かる

My point is that I don’t believe that phonetic systems slow people down. That the sounds are represented by symbols that would normally represent digits makes no difference.

Other Possible Applications of a Phonetic System

A phonetic system can also be applied to memorizing things other than long sequences of numbers. When my system is complete, I will have images that match up to thousands of specific sounds, not even including my separate mnemonic system for sounds. Even at this incomplete stage, I already feel like the artificial words are setting up a large number of new associations in my brain.

Having a large mnemonic vocabulary (over 2800 images) will give me artificial words that I can apply to things that I experience in life that I don’t have a word for. I’m predicting that this will help me organize my thoughts better.




Nelson 19 Sep 2021

Dude, so glad you posted this (the part about Chinese having shorter words for numbers than English, thus Chinese people having a 10 digit span instead of our measley 7).

It's funny, because when i memorize, I "whisper" what I see to myself. If I don't "whisper" it, it doesn't stay. And I've never thought about it before, but I need to clean up, I.E. shorten their words, sometimes its a mouthful: 47 is "firing a missle." Which could be shorter if I say "fire miss" or something, or even "fi-mis"...Maybe I should re-write all my actions to be like "dance" rather than "dancing".....very interesting. very.


Josh 19 Sep 2021

I wonder if it could be compressed even more:
dance -> "dan"
fire miss -> "fi"

E.g., something like Ozzy Ozbourne firing a missile at a pancake could be "oz fi pan"...


Dan 19 Sep 2021

Ramon Campayo's has records of remembering 18, 19, 20-digit numbers. If it is true that eidetic memory does not exist than Campayo must be applying a speed-reading technique (eliminating subvocalization, large chunking, etc.). I know in his book Maximize Your Memory he devotes a whole chapter to speed-reading. In fact, his description of it, that it becomes like watching a movie, stuck with me.

Has anyone tried applying these speed-reading techniques to longer memory events? I know the Tony Buzan is a proponent of speed-reading and Dominic O'Brien has a chapter on it in his How to Pass Exams (though it seemed it was a suggestion than something he practices).

Why not add Campayo's book to your reading list? It is currently available at most Barnes & Noble.


Dan 19 Sep 2021

Sorry, I meant to say Campayo has records of remembering 18, 19, 20-digit numbers in 1 second.


Josh 19 Sep 2021

That's what I was thinking when I heard about him. With the numbers-as-words idea, memorizing a lot of digits would be like holding this in short term memory after looking at it for one second:

cat    plate   candle
rhino  car     poster

In a 3-digit system, that is 21 digits.

I'm going to look for his book today, though I don't know if there is B&N in Hawaii.


Josh 19 Sep 2021

None of the large bookstores here have it, except maybe in Kaneohe, which is far. I'll look for in when I get back to CA....

Rob Clark

Rob Clark 19 Sep 2021

I really like this idea. I am hesistant to recreate a PAO to link better images to the one syllable creations. I think with a little extra time it won't be necessary- I will just know them all digit syllable, english language pronunciation and my PAO, just like I know the same digits in Spanish, French, Chinese, and German(well many of them in German).

Josh Cohen

Josh Cohen 19 Sep 2021

Thanks -- I fixed the typo.

Experimenting with a Category-Based Ben System — 19 Sep 2021

[...] I thought that maybe I could redo my entire Ben System images from scratch entirely by using categories. That would get rid of the problems I was having with the inconsistent nature of my previous attempts, as well as keep one syllable per image. [...]

Stream Capture: Taking Control of Thought Triggers with Visualization Techniques — 19 Sep 2021

[...] I change a number’s image and the psychological effect of memory techniques, and how I assign temporary names to unknown objects and ideas in order to remember them. This post expands on that idea with [...]


Geeljire 19 Sep 2021

Isn't the point of adding a word in the number --> image relationship to serve as a crutch until direct number-image translation is achieved? If so, why should word length matter in the long-run (in phonetic systems or otherwise)?

Josh Cohen

Josh Cohen 19 Sep 2021

In the long run, maybe it doesn't matter -- I don't know. In the short term it's useful for encoding sounds into images and compressing long numbers into just a few symbols. E.g., a 21-digit decimal number can be compressed into 7 syllables and held in working memory without using the method of loci. It compresses 10-digit phone numbers into four syllables.

Steven Bhardwaj

Steven Bhardwaj 19 Sep 2021

If you're looking for an enormous library of one-syllable anchors, one-syllable per thought, the Chinese language offers precisely that, as all characters can be pronounced as one syllable, and nearly all of them represent a discrete thought. For the purposes of verbal language, and thus modern standard language, the characters have been combined into two-character or three-character words, perhaps evolving from a need to reduce excess of puns. But it seems that classical written Chinese held closely to one character per thought. Just a thought...

To leave a comment, visit the Art of Memory Forum and create a new topic.