Major method equivalent for letters

Is there an equivalent to the major method for letters?

The major method, as we all know, is a set of rules for substituting phonetic sounds for numbers, which we can then build words/images from.

Is there an established system for doing the same for letters? For instance, if you have a string of letters you want to memorize: ESCHRCH CL. Is there a recognized method for converting strings of letters into words/images?

If not, what are your first thoughts about good rules for devising such a system? Such a method could be incredibly useful. What I imagine doing is programmatically generating images for terminology, based on the consonants of any given word.

My example above was taken by stripping the vowels from the term Escherichia Coli. The program could then generate a list of words that could be used as images for the given term. One could specify the consonants that one is interested in generating the list for. For example, in ESCHRCH CL it may be sufficient to generate an image based just on the prefix ESCH, or the suffix RCH, or the stressed syllable of words.

I have found some related posts here and will link to them as soon as I read through them all.

I am not particularly interested in having a unique image for each letter - it doesn’t seem like a very robust approach. The rules based system should be dynamic (i.e. possible to generate many different images for the same letter combinations), and not based on single letters, but groups of letters and hopefully phonemes, as in the major system

2 Likes

I haven’t seen one but it’s a nice idea… Sounds might work better than letters for grouping. There are 44 common phonemes in the English language. For converting text to images that is otherwise beating you this could be quite good… Looking forward to answers as well.

1 Like

Hi!

I believe your idea is consonant with my utopic dream of creating an image-based alphabet for mnemonic purposes. I’ve mentioned elsewhere how I think that we should create an established image vocabulary that could represent the entire spoken language. I use the term “the” because, ideally, this mental language would be “concept-based” and not “word-based”, thus being totally independent of the native-language of whoever is using it. So, a program like you mentioned would be able to translate any idea into images and these could be transmitted to anyone who spoke any language.

But, as I said, this is utopic, and it seems very hard that a complete language, even a mental one, could disregard individual phonemes in constituting adequate words. Hieroglyphs, as far as I know, had many logograms for entire concepts, but also used symbols for specific phonemes, so that nonsensical or unseen words could also be represented.

With that intention, I think you should take a look at this article. Its intention is to form a syllabary for English or, roughly, a set of symbols to represent syllables. He concludes that there are 15,831 syllable-candidates, “far too many to make a syllabary worth the trouble”. Well, since people memorise 2,704 images just for cards, I don’t think 15,831 would be an impracticality in order to being able to memorise EVERYTHING. But I also don’t think it would be necessary.

There are many possible approaches to come up with such a system. I won’t pretend here to know any of them, but I am slowly (very slowly) thinking more about it. I think that, first of all, a thorough study of word and syllable frequency distributions in English should be carried out. It seems that websites such as these have already done that. By using these frequencies, such alphabet could be developed by first starting with the more frequent words or syllables. We might find out eventually interesting (and useful) facts such as this one from wikipedia:

So, with a carefully designed 00-99 number system, we can also represent 50% of everything that is written in English! That sounds nice, doesn’t it? Yet, we don’t care about that… only about cards. (Please, I am not criticizing mental athletes here, I am just making a point. I myself am very interested in speed cards.)

Then we should look into syntactics and semantics. IF THEN ELSE statements, for instance, could have specific images associated with them, so that we would not have to add three symbols corresponding to these three words every time we want to use such grammatical structure (forgive me if my expressions are not linguistically correct, I am just an enthusiast of the subject, not a scholar). Moreover, the same symbols should possibly convey different meaning depending on the context. The Egyptians used determinatives to accomplish that and we might borrow the same tactics.

These efforts would theoretically take care of the codification part, but much more than that is needed if the goal is long-term storage of general knowledge. We would really have to enter the field of Computer Science and find proper ways for data organisation and information retrieval, even thinking about data structures such as hash tables, appropriate search algorithms and the like. I am not from either the Linguistics field or the Computer Science one, so these are just my naive (and wild) guesses, of course.

In sum, I think there is a lot to be done about these ideas and I’d be very interested to collaborate in that respect. Right now, an indirect effort I am making is to learn a programming language that will allow me to come up with programs such as the one you alluded to. I have occasionally used high-level programming languages throughout the years, but I have never reached any level of proficiency. Now, surprisingly enough, Mnemonics is motivating me in that direction again. I think Python provides excellent packages for Linguists, such as the Natural Language Toolkit, which can in turn be used for the aforementioned purposes. Moreover, with the Web scraping technology nowadays and the current stage of the Semantic Web, programs could dynamically (like you suggested) search for new images corresponding to words and phrases, creating personal “local” image vocabularies that could be used for specific memorising endeavours – Python also provides lots of resources for that.

Forgive me if I digressed too much from your OP, it is an exciting topic…

Best!
Tammish.

5 Likes

Or… you could just come up with 44 memorable images that you can link to the phonemes and create a journey for difficult to remember phrases. The problem I see is that in order to do so you are decompressing rather than compressing the data so it could get very long, but you could be letter perfect for specific text.

Attempting to unpack the subtlety of meaning/context into images is called art :wink:

The problem/gift of art is that we interpret it individually. Common/Universal visual language seems like a fairly unrealistic objective. Smaller objective; useful method of expanding ones ability to think creatively and remember the spoken/written word more vividly/deeply would be very interesting.

Sound like fun, I think like that you still have to develop these skills in some kind of simple progression and before you could actually DO this rather than talk about this you would likely have to master the progressions of peg, loci, linking games and tricks of the mnemocist.

That the Mnemocist’s who have mastered these tricks/games have not extended these techniques to general purpose thought suggests that it is either not easy or the idea is malformed. There are people here who have created 1000 image, multi-dimensional lookups, can memorize multiple decks of cards in unreasonably short order and have generally got themselves tuned up to the point where they can use these techniques habitually. I don’t think it is simply a problem of lack of motivation or insight… Or maybe Illuminati :wink:

2 Likes

Hi Robert!

I think that for verbatim memorisation, that could definitely be a path.

I like this definition very much, nice!

Oh yes, probably so indeed. But, by aiming at the heavens, men has achieved the skies. Walking on the Moon by 1945 or smart-phones by 1990 were also “fairly unrealistic objectives” and, yet, after around 20 years…

We already have that. Example from Harry Loraine’s “The Memory Book”: for the purpose of memorising the 50 American states, you can picture an “ark and a saw floating in the air over a safety zone” for Arizona and Arkansas. Nice and useful. But not quite exciting anymore :wink:

Agreed 100%. That’s why I am embarrassing myself daily with my “around-5 min” speed cards time…

So, all that came from this sentence “Yet, we don’t care about that… only about cards”?

That was just a hyperbole for rhetorical purposes, forgive me if I was unsuccessful. The idea is definitely not easy and that’s exactly why it is so exciting. I can also imagine 100 reasons why a Dominic O’Brien, an Alex Mullen or a Lance Tschirhart would not dedicate themselves to the idea – e.g. they are not THAT nerd :slight_smile:

Anyway, maybe you’d like to check this book out: “Logic and the Art of Memory: the quest for a universal language”, by Paolo Rossi. I haven’t read it, but it is on the top of my list together with Yate’s “Art of memory”. From Amazon:

At least, I am not alone in my “fairly unrealistic” objectives :wink:

Best!
Tammish.

1 Like

Yes, I imagine the Natural Language Tool Kit in Python will be important in writing the program. Also, see Carnegie Mellon University Pronouncing Dictionary, which is used in the Natural Language Tool Kit.

I think it’s a simple matter of building the look up dictionaries for characters (single and strings) and their corresponding list of phonemes, which represent the possible substitutions that will be allowed, then generating a list of analogous words based on user input of a string of consonants. The list would be generated by first taking the user input, generating a regular expression that would match all words that match the particular phonetic pattern, then using that regular expression to search dictionaries. One could even use arbitrarily customized dictionaries, i.e. lists of cartoon characters, famous people, mythical creatures, characters from novels, historical figures, etc. and could specify that one wishes to only see results from a particular list or a particular part of speech, i.e. proper nouns, adjectives, etc. It’s may also be possible to have corresponding lists for historical figures and associated actions and objects, provided there is a resource/method for obtaining or generating lists, i.e. George Washing, chopping, cherry tree, etc.

I also like the idea of a system for allowing consonants to be substituted by a particular vowel, so that when trying to memorize an acronym such as “ftp”, an “i”, for instance, could be substituted for the “t”.

Lucky for us, memory tricks work only because they take advantage of the innate incredible capacity humans have for spatial/visual memorization. There’s nothing to master, only techniques to refine. Anybody can do this, right away, with very little training. Such a program would help people understand mnemonics, why the work, and how to use them creatively.

3 Likes

I had a thought… I’ve been playing at stenography and slowly building myself a custom keyboard.

Stenography allows a typist to type over 200 wpm by compressing the English language into a set of common abbreviations, sounds, and finally the alphabet when nothing else works. Stenographers chord words (press multiple keys simultaneously to represent each word or sound). As Stenographers work over the years each creates custom chord to represent their needs (expands their dictionary).

It might be possible to use this approach and method to develop mnemonic language that would support memorizing significant portions of spoken word or text… and then again I could be completely wrong :wink:

2 Likes

I think this is very relevant and it’s funny you should mention it because I’ve been very interested in learning stenography, especially after recently following some links which I think were posted by Josh about open source stenography tools. It seems to me like its very much along the same lines that I am thinking. I will be looking into what it takes to generate a list of all possible spellings of a particular phenome soon. I already know where I’m gonna start looking: Soundex - Wikipedia

Getting a simple python program up and running where a user can manually enter in the phonemes he/she is interested in should be easy once I figure out how to use soundex and any other tools I might need. Once a phoneme is entered and a list of alternative spellings generated, then we can search a dictionary for words that feature that phoneme according to certain specifications, e.g. words that start or end with a phoneme, words that have the phoneme preceded by any vowel, etc.

I have just now begun to (finally) read “The Art of Memory”, by Francis Yates, and I found something that reminded me of this thread and that might reconcile both mine and Robert’s idea.

On page 15, she quotes the (anonymous) author of Ad Herennium when talking about “memory for words”:

(I think the second “images” he writes should be “words”, don’t you think?)

She then explains that:

She agrees with the author that such method would be impractical since even one thousand such images wouldn’t be enough to cover all the words there are. But, of course, no one here is advocating memoria verborum and we all know that 1,000 images is a piece of cake for our mental athletes, so…

… there is still hope for my “image vocabulary” :slight_smile:

I have used this tool before to come up with some substitute words for memorising and, although very limited, it allows search by soundex. Two impressions I had when using such tool. First, a program such as the one you intend to build would be really great. Second, I don’t think Soundex is the best approach, although the concept is definitely sound (no pun intended).

Litscape also has a search tool using Metaphone instead of Soundex. Metaphone is much more accurate. I think combining a phonetic search algorithm with some sort of fuzzy searching or regular expression rules would work wonders. I’m currently seeking advice on a starting point for developing a way to test these ideas, but haven’t had much luck so far. It will probably be necessary to generate a database that organizes words into groups based on their phonetic codes (different groups for soundex and metaphone) and also different groups based on fuzzy search algorithms. This way, the program could give you a list of possible images based on more than one system of rules, i.e. a list based on soundex codes, a list based on fuzzy search criteria, a list based on finding words with regular expressions that replace consonants based on a set of rules etc. If you know regular expressions you can play with the Regex Dictionary which has also given me lots of ideas.

1 Like

I do this. It’s called a letter pair image list. So like the major system has unique sounds for 0-9. I use rules that give me unique translations between AA-ZZ and words. I’ve written out things before, but most of my digital files have been lost so I’ll try to explain my system a bit again. Letter pairs were not originally my Idea. The first person I know who started using them was Chris Hardwick. He picked images for each letter pair that he used for solving Rubik’s cubes and gave them an image.

When I started doing that, I wanted my system to be more robust. I wanted a way to get every English word to be translated to a specific letter pair, but for every letter pair to have many possibilities for words. That sort of happened. The letters starting with X are more like acronyms since there are not many words starting with X. It’s still a success since I have an image for every pair of letters.

  1. For any word, the letter pair is determined by looking at the first syllable.
    2.the first and last sound in the first syllable become the first and second letter.
  2. I ignore suffixes even if they are in the first syllable. For example, Rock=RK so rocks is also RK. There would be too much trouble otherwise. I would not like to consider visually remembering whether a rock or a pile of rocks fell on someone. It should make no difference.

note-because the syllable determines the pair. My letter pairs occasionally have an entymology mapping which makes somewhat of a connection between possible images and the characteristics that they have. Some of these are obvious like chain and chainsaw for CN. Some surprise me like remembering that there was something green, but having no idea what image it was. Thinking, hey Piccolo, a character from a show, is green and then, remembering that it was a pickle(also green).

After that you just make hard and fast rules about how you will split up the sounds between the letters.
The trick is to differentiate between S and Z, C and K, J and G, and q and K and anything else that would give you trouble.

For example, I use the English rule for having the u vowel before q to separate u and k. I never have a u vowel before the letter k. so I can use that sound with a u vowel to fill out words with q which would otherwise be a tricky letter. For example, NQ=Nuke (nucleur bomb) KQ=Kosqiusco mustard MQ=Mosquito JQ=Juke box

There are 4 combinations for consonants (C) and vowels (V) in a pair and it helps to understand this.

1.CC is the the easiest combo and is a closed syllable. Luckily, these are also usually the best source of variety for words because the vowel in the middle of the first syllable doesn’t affect the pair. This is also the subset of words that the major system stays in. For example. TP can make a lot of words. Trap,Tarp,Tramp, Trampoline, Tempist, Temper, Tripped, Tip(as is the apex of),Tip(as in buying someone off), Triple, Triple H(a person), Troops, Trump card, Type, Typist, etc. The thing about theses CC pairs is that even though they all have so many choices. I have favorites that I use most of the time for memo. For TP, that would be trap and tarp. Others just pop into my head occasionally. The occasional ones are usually very strong though because something about the previous word or my day made me think of it.

2.CV these letter pairs are still easy to find but there are less choices of the first syllable. In some cases, there is only one choice. If you know a language that uses mostly open syllables than it is very easy to take loan words. Spanish, Japanese, and Chinese are coming to mind. Examples would be LO= Lohan (lindsey) CA=Caley Le=Lee RU=Ruby loan words could be things like Casa, Lo mein, Kakashi(scarecrow).

  1. VC These are easier since the vowel isn’t going to cause problems with the syllable in the front of the words. They are easy to find words for. For example IC=Ice, Icicle, Ice Cream, Ice cream truck, etc. EG=egg EK= elk the words are there so it’s nice.

  2. VV these are more trouble. There are words to use but some can be hard to find. I remember having a hard time finding a word for EO and I was using an acronym for a bit until I remembered Eowyn from the lord of the rings. I still use arconyms and just subvocalize the letters for some of these. UO=UO That’s brilliant, right?! but as a fan of running, UO means University of Oregon and I can remember a runner from there.

Basically these rules get you 95% of the way there and you make exceptions or just know your list/have a list for the rest. These holes shouldn’t really make sense to anyone other than you. An example would be OQ= Officier Qualities= a Canadian guard with the bearskin hat. Sure, I broke some rules, to get to an awesome image, but it’s not an image I could easily get to otherwise and I don’t think there is a really great word for OQ. These become reflexive.

Some other rules would be to use the most common name for things to avoid synonym troubles and to have habits for always using first names or last names for certain people. These rules really don’t have to make much sense to anyone other than yourself. So if you decide that friends are first names and celebrities are last names, then that’s perfectly fine. When I use journey’s I stick to concrete images and visual adjectives, nouns and verbs. Most of verbs would have objects attached to them like the verb Shoveling.

When I make mnemonic sentences, which is actually how I remember everyone’s phone numbers since I don’t have contacts on my phone. I ignore all of the connective function words like prepositions, pronouns, particles, and determiners. Those words just hold the sentence together. Abstract words are usable though. For example, There is a guy on this forum whose # has the word “gracefully” in it and it’s not as a letter pair. Then, for phone numbers, I link something in the sentence to the person. It usually has to be something that actually exists like a friend’s ex or a shared experience like where I met them.

The obvious next step to expand this system is to know your common list of vocab and then know when you are stretching it. I like one syllable words because they are quick to say and easy to remember. “Gracefully” is not a one syllable word or standard word of mine. It’s an uncommon 2 syllable word with an ignored suffix. That’s like a red flag that it may be a 4 letter word which can be a 3 or 4 digit word based on whether all of the letters convert to numbers which only 12 of my 26 letters do. Of course, if you need a 3 digit number at the end of a 10 digit phone number. you can always find a word for a 4 digit number and ignore the useless last number. So to continue the example Grace-fully=90 55.

Another expansion of the system is for random words. For the abstract words that come up in the event, It’s helpful to know enough images for the syllables that you can do the CVC (consonant vowel consonant) of the syllable correctly. This breaks the abstract word into a string of syllables and if you have images for that, then it’s no sweat. I’m not going to say this is the best way for all cases or that I know the CVC for every syllable because it isn’t and I don’t. I know enough though where is helps for random words which is a very freestyle event. An example against this is “diplomacy”. This cases is a boon since you can imagine a diploma with a C on it. It’s an average diploma and then you go to the next word. But for most abstract words, it works well and a list of preset suffix images makes it better.

I think a last idea to note is about how often I manage to stick to a preset list for letter pairs for memory sport. I have noticed that when I start practicing a ton, I end up having less and less variety in my choices and for speed cards, it becomes a preset list. When I practice less or stop practicing for a while, I start using whatever comes to mind more often. Over time the first string images change. I had this documented by rewriting my lists over time. I could watch a good movie and the characters would have some appearances later in the week. Maybe I forget my list but remember the rules so I never have an issue creating an image or I’ve changed my mind too much and it toggles between previous images that I used. I don’t think it matters too much If you stick to a preset list or not since practicing will do most of the optimizing.

TL;DR- 1.Make rules for sounds/lexicon for each letter 2. Mostly use letter pair images for memory sport 3. expand to 2 syllables for 4 letters when you can for non-timed, more creative and long term memory. 4. Try splitting tough abstract random words into syllables and using your knowledge of letter pairs to memorize them.

6 Likes

ALeman, thank you for the post, it was very interesting. Only the part about the abstract words was a bit unclear. Can you expand a little on how you use syllable images, CVC etc to memorize abstract words? Maybe a few examples would be helpful.

Thank you @ALeman. I’m trying to develop a “Letter Pair” system that is based on PAOW technique. I’m avoiding a ‘phonetic based system’ like Major System…

Your post gave me some important hints…

Truthling, just to help you in your search:
This is called 'Grapheme to Phoneme" correspondence.

English is horrendous for spelling. I think English might have more G2C options than any other language with Roman script?

Here’s a detailed article:

There are about 44 phonemes in English and each has up to ~5 spelling variations associated with it. Kids have to learn them all. It’s quite a feat.
I’m a teacher and it’s part of my job, along with teaching them how to speak English, to help them remember all these patterns. Funnily enough, I’ve noticed that some of the materials I’ve bought use mnemonic techniques but it’s very haphazard and as a teacher you often have to wade through shoddy mnemonic devices and often invent your own. Even after preparing you’ll find ourself trying to come up with them on the fly!

Anyway, I digress.
People have also created languages for various purposes. In terms of minimalism there is Toki Pona with something like less than 10 letters, direct spelling and around 100 words. The design of this could be relevant to this subject even if the language itself isn’t.

2 Likes