Today and yesterday, I memorized around 300 Italian words. I didn’t use the method of loci; instead, I connected each Italian word with its German equivalent (I am Swiss). These images are simply sitting in a white room without any specific locations. Is this an effective way to memorize a large number of words, or should I consider using the method of loci?
1 Like
You’ll have to see how you do on review tomorrow and then in a few days.
Honestly if it works for you, it’s fine. If it doesn’t, then you can try out something else.
3 Likes
To remember something, either you use it in real life, or will have to review it in some other way. Real life, in a context that is just right for your level, is often the best approach. But, besides not being always feasible, this also usually won’t ensure that you will review and retain everything you have learned.
If you’re not familiar with the method of loci, it might take some time to familiarize and get comfortable with it, but it will be worth it, and I think the sooner the better. Once you have your palace you can review it whenever wherever with no device needed.
Some people prefer approaching the learning of a language without it and just immerse themselves in the language - that’s ok too, just remember that these two things are not mutually exclusive
3 Likes
With the caveat that reviewing hundreds of words (going through the whole memory palace(s), because you’re not distinguishing the easy words from the harder words) is very long and can be… inefficient (yes I dared say that 
)
Honestly for vocabulary I feel cards with mnemonics/associations in a spaced repetition system are just as good for languages that are relatively similar to your own.
I mean I think memory palaces are awesome and that it’s worth it to train to use them, but not obligatorily the best for everything.
3 Likes
if you place all the words at random yes, but if one is studying them like he’s doing, then he’s likely learning words by subject, and the harder words will be in their context space, which will help. also in real life you usually hear things in context so there’s no need to make things harder and review the words in the aseptic environment of flashcards
2 Likes
If the flashcards you’re creating are boring then you do them wrong and you have another problem 
It’s quite easy to add example sentences on an anki flashcard to give context…
But to each their own, everybody has to experiment and see what works best for them.
2 Likes
Strongly agree 
Use memory palaces where they are the best tool (memorizing elements of a speech you have to deliver, memorizing lists of words or numbers or playing cards), but for languages the best tools are mnemonics/associations paired with flashcards (using context for the flashcards if it’s more of a grammar term).
I’ve seen spatial systems (similar to method of loci) used for être verbs in French, and parts of the body in any language. Might also work for ordered data like days of the week. But ultimately you need these to be stored as direct associations, with the spatial system a temporary intermediate step.
2 Likes
Yes this works for me! I just have a hard time with small words like: so - anche, already- già etc. I just cant turn these into a picture! Do you have any suggestions how I can do that?
“anche” (=also) sounds like “anchor” - your favorite anchor keeps saying “and also” in a loop and doesn’t know what to say next.
“così” (=so) sounds like “cozy” - how are you sitting on that big couch? So cozy
“gia’” sounds like the start of “jacket”. imagine someone putting their jacket in no time and you wonder “how did they do it already?”
1 Like
I use the memory palace and don’t use Anki. I used to use Anki but realized it was a waste of time. With the memory palace or even if you don’t use the palace, just mnemonics, you will have to immerse yourself, and immersion will be the review. The problem is that you have to force yourself to see basic things and obviously memorize basic things. It’s very difficult to stay immersed without understanding anything, wow.
In my opinion, memorize using your technique, make sentences, and read them every day. Read quickly, just try to understand and move on to the next as fast as possible. This study is just to reach the level where you can understand the context, and then you can do anything and you will learn.
I put the sentences in ChatGPT and read them there. I use the palace while I’m reading books. Without Anki, there’s more freedom.
1 Like
I used to use associations to memorise vocab but always tended to burn out under the workload, especially combined with SRS. You can learn a lot of words quickly with mnemonic images, but if you go above the default daily limit of 15 new cards, the number of reviews quickly gets really big.
In recent years I found and agreed with Stephen Krashen’s argument that conscious drilling of vocab and grammar rules is actually detrimental to fluent language acquisition. Drilling individual “facts” is a declarative memory task, but language understanding and production should rely on semantic memory, which is much faster and automatic.
2 Likes
Make a physical notebook with each word you learn, and number them as you go. This is just for later reference. Memorize earch word “as is”, or rather - out of sequence. Cibo means food in italian, so to memorize this make a substitute image of the word cibo, like Cheetos, and then mix in the meaning - food, by adding a substitute image for “food”. No need to place these visualizations in memory palaces unless you want to memorize all words you have memorized perfectly, or in sequence.
2 Likes
Interesting. I actually have found grammar and drills to work great and CI a waste of time. I think a lot depends on the language, and how its related to your mother tongue. Probably languages which are grammatically similar its counterproductive to do drills and productive to do CI.
I have found Krashen to be a little bit light on what he actually means when he comes out with stuff he comes out with. The whole N+1 “Learning takes place when messages are understood” is pretty profoundly obvious. Of course we understand when we understand. Yet I do not believe he has ever come out with any definitive methods of CI.
Like driving a car, learning to type or anything else, learned by drills or otherwise, you eventually make the transition from conscious production to subconscious production when you get competent, or fluent.
I’m not sure about this, but I guess it depends on what a “drill” is. I think Krashen was referring to the predominantly declarative learning of individual words and grammar rules, so if we’re making analogies to other procedural skills, I’d compare it to learning driver theory, how cars work and what control should be pressed in various situations, vs actually driving. Or answering endless flashcard questions about what notes should be in the next chord after a known sequence, vs actually trying to play through a piece on the piano (or even listening to music) – in my case, I studied some music theory while learning, but performance came first and theory mostly clicked into place afterwards.
In a similar vein, when I did my compsci undergrad 20 years ago, a comment from one professor stuck with me: “you can memorise every book on computer programming, and not be able to write even simple programs… the only way is to learn by doing”.
If by “drills” you meant full sentence recognition, that would make more sense to me. It would also fit better with more recent theories of procedural learning and contextual interference, where they’ve discovered repeatedly that practicing isolated “subskills” offers limited transfer, compared to performing the real task in a realistic environment.
Perhaps you’re right, but that’s not been my experience. I spent quite a few years learning individual Chinese characters and words with Anki, based on a frequency table. At some point I reached about ~1100 characters and realised I couldn’t hold more than the most basic conversation with a native speaker, and especially struggled to follow their speech.
After that, I tried doing sentence flashcards, which worked a bit better. Eventually though, I dumped the whole thing and went for just watching Star Trek TNG dubbed in Chinese. At least for me personally, that’s been more effective AND more enjoyable.
That’s fair. I’d like to see experimental validation/rejection of his contention that studying grammar/vocab somehow impedes “natural” fluency (other than by taking away time that would be otherwise spent on listening/conversing).
I have no idea what Krashen meant, I wonder if anyone does. 
By drill I mean, exercises based on learned grammatical theory and rote material. Learning vocabulary using traditional methods, read out loud, copy, write, read out loud, etc. plus all the other methods. This is doing. Somehow people seem to have overlooked that we always had methods for learning language that work.
Learning isolated words is actually great. But if you don’t know how to apply grammar or form a sentence then that word is not going to serve you. But don’t blame isolated word learning. Its a fantastic effective method, that serves to learn a great deal of words.
Ultimately learning a language is more than learning words, sentences and grammar. All that stuff only gets you so far. To really speak you will always need higher level cultural level mastery. But at that stage you will have the mechanics under your belt and memorization and production have long left the picture.
1 Like
I’d argue that those methods feel similar to “doing” but are not doing, because they rely on conscious decisions about construction/translation of predefined, written sentences, which they rarely get to internalise (make the leap from conscious to unconscious production), because they’re constantly moving onto the next, more complex grammar constructs. That is, the proportion of the time spent “model free” listening and understanding from context – which is how we actually communicate – is quite low, relative to time spent doing those drills. So there’s not enough opportunity to formulate an implicit model, and too much time spent forming explicit models that don’t easily lend themselves to automatic fluency, and are probably forgotten before the requisite unconscious transfer can take place.
The evidence in favour of those methods’ effectiveness is mixed, at least if we define the goal as “able to hold a decent two-way conversation”. Schools have been using these methods with limited success for a very long time – anecdotally, I saw it first-hand in Irish, French and German in the 1990s, when they’d only recently made pronunciation carry any points in oral exams.
Only the most dedicated of students made significant inroads into French or German, but our Irish was near-fluent because it was the language of instruction in my school. A friend joined the school with no prior Irish language experience at all, and managed to become completely functional within a month or two simply by trying to understand the teachers, putting almost no time at all into formal study of the grammar (that didn’t really kick in until a year later).
In countries that have demonstrated exceptional second language learning results, they’re almost always watching tons of TV / Youtube in the target language – the canonical example probably being Germans and Scandinavians with ridiculously good English, although no doubt they also do put significant effort into the structured drills you mentioned.
You might argue that explicit grammar study and rote drills etc can work, but schools just don’t apply the methods correctly. That could be the case, but I’d wonder why schools seem to almost universally suck at it.
There was an interesting series of posts here about a pseudo-CI approach used in teaching Thai to English speakers. They assert that the students who achieved the best fluency outcomes were those who tried not to consciously focus on the grammar (or even vocab) of what they were hearing, and instead just try to follow the dialogue and answer in English (the idea being to allow a subconscious model of the new forms of speech to form implicitly, which might be damaged by trying to formulate correct speech too soon).
One more idea about memory palaces and languages, if it wasn’t already mentioned above:
1 Like
I would suggest using a location for each word (not a memory palace as such because each one is isolated). Pick the first location that comes to mind when you think of the word (in your native language). Then put your ‘image’ or connection for the Italian word in that location, always linking it back to the original word of course. For words that don’t have an obvious location, use a location that starts with the same letters as the word, e.g. to remember the word ‘walk’ in a certain language I might store the image in a place called ‘Walker Road’ (I use Google Street View to find these locations if I don’t already know them).
2 Likes
Good info.
How well have you found this to scale? As in how many words? Whats the accuracy like, and is there an advantage over simple SRS approach?
I didn’t suggest not using SRS - obviously you need to review the words otherwise you’ll forget them. This is just an alternative to a sequential journey, which works well here because you can find the words instantly rather than going through the whole journey, which I think is unnecessary for language learning.
2 Likes