This is the first step of joining a conlang community and an early stage step in designing a language. You can’t really design a language unless you gain some proficiency in it. I’ve been putting off learning all the stems in my current conlang because I hadn’t finished the syntax and morphology. I’m thinking that is an artificial roadblock. Without taking the time to memorize stems, I won’t be able to write a syntax and morphology. I need some live examples to work with, so I must memorize the words now.
Also, if you want people to actually learn your language, you will need to do something to help them out. Writing a reference grammar and dictionary that looks just like the ones academics write for dying languages will limit your pool of fans to linguistics academics who happen to like conlangs (a small subset that is). So it might be a good idea for a conlanger to learn about vocab cramming methods and teach them along with the conlang.
Anki is great. It is a space repetition system that keeps track of the accounting for you. It does a good job of moving your through large decks and eventually moving words into your long term memory. ‘Nuf said, go get it.
On to the paper methods.
Anki is the least effective for a completely brand new set of words. For example, I started using Anki for Icelandic after I already knew maybe a thousand (or two) words. It has been very effective. I am using Anki for Klingon but words are moving into the “leech” category at a scary rate. It is as if my brain is rejecting the possibility that Klingon words are even words.
This is named after a guy on the “learn any language” forum. He’s a successful polyglot, so what ever he’s doing, it might be worth copying. I don’t completely grasp the idea yet– there are several moving parts to the system and it isn’t obvious what is relevant and what is peripheral. (joke- carpenter apprentice: I keep hammering the nails in crooked! master carpenter: Ah, you are holding your tongue wrong)
The gist is that you get about 6 or 7 words. It isn’t clear if how these are compiled are important. I plan to rummage through a dictionary and compile a long list of promising sets of 6 or 7 words. I assume most people will feel that a carefully compiled list of words is better than a random set, but if you goal is to learn 10,000 words, that means a lot of stupid words like puce and mauve are going to be on the list.
Then the study session starts. You visually look at each word and see if you know it, for each of the 7. If you don’t, I suppose, you have to look on your master list (or worse, individual lookups in a dictionary). After you can read the list and mentally (or aloud) say the translation, you write down the translations. Then you repeat the process going in the opposite direction. Then the process is repeated a few more times, I think, on different days.
The technique appears to be doing something with short term memory, as 6 or 7 things is the maximum number of things you can hold in short term memory.
I suppose, after you’ve kick started your short term memory, you can now productively use anki.
The Gold Method
This method also uses dictionaries as the source. Similarly, you flip though and find a bunch of interesting words. Write them all down. Read them aloud. Next day, filter the list to the ones you still can’t translate on sight. Read aloud the list. Continue to winnow down the list day after day. Another part of the technique is to stick to bouts of 20 minutes or so– this either provides spacing or lets the brain rest.
The reasoning behind this is that the storage of words in long term memory is subconscious and out of our control. We can’t predict what activity will trigger our brain to remember something … so apparently we don’t bother to get fancy about it, we just provide our brain with several opportunities to remember the word (by selecting it from the dictionary, by copying it, by reading it aloud, by repeating the process the next day).
Both of these techniques (Gold an Iverson) appear to be optimized for short term memory, which Anki doesn’t really address.
Same as the Iverson method, I would use this in conjunction with Anki.