In case the title above doesn't make it clear, this page is not intended to be a complete guide on how to learn a language, but is rather an opinion on how spaced repetition system (SRS) flashcards can most effectively aid your language learning. The language acquisition process that I outline at the start is a framework with which to illustrate where I see certain flashcards fitting into that process, including where flashcards don't seem to be significantly warranted compared to more organic styles of acquisition. If you're at the foundational stage of language learning and still working your way through textbooks, feel free to read this as a preview for the road ahead, but most of it is aimed at learners who are transitioning to passive immersion (consuming content intended for native speakers). If you are at the immersion stage, the production flashcards I recommend have slightly different goals than the recognition cards you might be familiar with, and are formulated to specifically target gaps that are harder to fill with more natural learning methods. If you happen to use, or perhaps even enjoy the recognition cards I advise against here, that's okay. I'm not saying they have no benefit. They might have more limited benefits compared to immersion, but if such cards motivate you in some way, that's honestly worth a lot, and if using them would mean the difference between continuing your studies and giving up because immersion is too intimidating for you at this point, recognition cards are probably worth using. Production flashcards aren't competing with recognition cards, immersion is, but it's up to you to choose what methods align with your goals and what methods you can stick with, and what you can stick with is actually more important than what's most effective, because even effective methods lose their effects if you can't stick with them.

To begin, I'd like to emphasize the nature and importance of practice. Practice could be viewed as "doing something, the outcome of which is becoming better at doing that thing", and it just so happens that everything you do is practice, so you inevitably get better at doing whatever you do, regardless of whether those things are "useful" or "beneficial". Following from this, if you spend time doing the things you want to get better at, you will improve at doing the things you want to get better at. Remembering this obvious-in-hindsight principle will help calibrate your learning compass and override any complicated language learning flowchart you might be trying to follow. If you want to get better at reading, read more often. More specifically, if you want to get better at interpreting what words mean in new contexts, interpret words in new contexts more often. If you want to get better at speaking, of course speak more often, but more specifically, if you want to get better at recalling certain words from memory, recall those specific words from memory more often. Some of these learning goals are effectively aided by flashcards, and some are better off being developed through immersion, but regardless of the method, you will need to manage your time to balance various kinds of practice depending on what skills you're trying to improve.

Learning a non-native language is different than learning your first language in that your goals for learning a non-native language can vary. With a first language, because it's your foundation for interfacing with the world around you, understanding and communicating with that language are essential. All skills, including the passive skills of listening and reading and the active skills of speaking and writing, must be developed in order to fully participate in the society that surrounds you. A non-native language, however, can serve whatever limited purpose you want it to, especially if you aren't surrounded by people who speak that language. You can choose to focus on specific skills rather than aim to do everything a native speaker can.

If your goal for learning a language is merely to acquire passive understanding of that language, then you're in luck, because passive skills are the only ones you need to work on. This can be achieved by first building a basic foundation from which to passively understand your target language, and then practicing passive understanding in your target language by reading and/or listening to native content (i.e. passive immersion). For adult learners, building a basic language foundation can efficiently be accomplished through explicit instruction, whether it be with textbooks, instructional videos, online or in-person courses, educational websites, personal tutors, or other guided explanatory material in which you acquire familiarity with your target language's features such as writing system(s), word order, conjugation rules, grammar, etc. Once that foundation is laid, looking up the meaning and pronunciation of words you're not fully sure of while immersing in native content is largely all that's left to do for efficiently gaining and maintaining passive proficiency in your target language.

Sounds easy, right? Unfortunately, anyone who has tried it (perhaps you included) can attest to what a massive amount of time it requires. However, difficulty due to time commitment isn't the same as difficulty due to process complexity. It simply takes a lot of time to assimilate the sheer amount of information contained within a language, and for a non-native language, there is ample opportunity to fall back on your native language and/or give up during that time. Backing off and giving up (priority shifts) are by far the biggest obstacles to acquiring your target language, not your learning techniques or what flashcards you use. This is why nearly every human being (including you) knows at least one language: When there is no other form of communication to fall back on, acquiring a language is largely inevitable due to its inescapable priority in daily life. Having another language to fall back on is the biggest difference for your priorities the second time around, and falling back on your native language is the unique crutch that is most likely to keep you from standing on your own in another language. This doesn't mean you need to stop using your native language once you've built a non-native language foundation, but again, the core difficulty of language learning lies in the time commitment, and if you expect your command of a target language to ever come close to that of your native language, expect to commit at least as much time to your target language as you do to your native language, time that now has to be split between the two (or more if you already know two languages). If you don't have that kind of time or those priority levels, recalibrate your expectations to match your commitment.

Build your foundation, start immersing in native content, look up what you don't know, loop back to foundational resources if and when your immersion is telling you that you might have missed or forgotten some crucial points, and don't give up on the daily priorities you set for yourself. This is the best recipe I know of for gaining and maintaining passive language proficiency, and you might have noticed that flashcards are not part of it. I'll explain why after I finish my overview of the larger process.

While the passive language skills of listening and reading require a considerable time commitment in and of themselves, the active language skills of speaking and writing take even more time to develop because they depend on a foundation of passive understanding in order to exist. If you didn't first understand the words you were trying to speak or write, you'd end up saying or writing things that you didn't yourself understand, which would be nonsense in both concept and result. This is why you can't skip straight to outputting a language: you have to be able to understand the words you're going to say.

Nevertheless, I've seen some people misunderstand, or at least miscommunicate, the idea that input (passive immersion) is the key to outputting that same language (or even that "output isn't practice" in the most egregious case). While there is a hierarchical relationship between the two skill categories (active language skills are built on top of a foundation of passive understanding), all this really means is that you have to understand the words you're going to use before you use them yourself. This doesn't mean that practicing a foundational skill (input) counts as practicing a considerably different and higher-level skill (output), and both types of practice will be necessary if you want to improve at both. This also doesn't mean that you need to have a complete or even wide-reaching foundation of passive understanding before you begin stacking output on top of it. Not only is there no danger in improving both skillsets at the same time, this will inevitably happen anyway, as you will never have a complete passive understanding of any language, especially when that language is describing unfamiliar subjects with steep learning curves and specialized vocabulary. For reasons stated above, you will never output more than you can passively understand, so there is no need to wait until your passive understanding is "good enough" to start output. Your current passive understanding of a language is already good enough to practice outputting at that level of language ability. Obviously you won't be able to discuss politics or humanity's future direction if you only know the words for food and hobbies, but you don't have to wait until you can read a newspaper editorial to talk about what you like to eat or do for fun. Yes, you'll make mistakes, but that doesn't mean that you're doing it all wrong or that you started outputting too early. Everyone makes mistakes when they speak, even native speakers. Maybe that's why we're all so afraid of talking on the phone... but I digress.

When one's language goals include the active skills of speaking and/or writing, language learning becomes a significantly longer journey due to the additional practice required, a journey from the foothills of input to the snowy peaks of output, and it is on the mountainside between input and output where I feel flashcards can give you a much-needed foothold. However, not just any style of flashcard will effectively fulfill that role, and recognition flashcards in particular are ill-suited to the task.

What are recognition flashcards, exactly? I would describe them as flashcards where the words you're trying to learn are visible on the front of the cards, and you have to recall the pronunciation and meaning(s) of those words. These cards are commonly used to aid in acquiring passive language skills, and some common styles include the following:

Word cards feature a word in the target language on the front side of the card, and can include things like readings, target- or native-language definitions, example sentences, images and audio on the back side. These cards test whether you can remember any of the meanings and/or readings of a word, with no context and usually no hints as to which of the word's meanings or readings is being tested. There are usually more meanings to a word than readings, so meanings become the bigger issue. If the back of a word card contains multiple or even all of a word's meanings, it's unclear how many of them and which ones you should be able to recall in order to pass the card, and you aren't necessarily tasked with recalling the same mix of meanings each time the card comes up. Recalling any meaning, regardless of frequency or likelihood given a particular context, is often good enough to pass the card, even if some incorrect meanings are recalled alongside the correct ones.

Even if the meanings you recall are all on the back of the card, there's no guarantee that the target word means the exact same thing as its English equivalents. To use the above card as an example, just because one of the English equivalents is "circulation", there's no guarantee that 循環 in Japanese is also used in exactly the same cases as English, such as for blood, water, air, publications and currency. Even if it turns out to be an exact match, you shouldn't start with the assumption that it is when evaluating your grasp of a word. Even if you try to remedy this by putting a target-language (in this case: Japanese) definition on the back of the card, you still need to know what contexts the word is likely to show up in, and even if you put all of those contexts on the back of the card as example sentences, that's way too much information to recall for a single test, and isn't an accurate way to evaluate word familiarity.

Because words can be used in a multitude of ways, including slang usages not in a dictionary, with no limit on the number of usages per word, you actually need one card per word usage, not one card per word. People often try to study lists of the most frequent words in a language, but the more frequently a word is used, the more usages it is likely to have, which makes singular word cards that much less useful. Take the English word "take", for example. Yes, if you master this one word, you've understood one of the most fundamental words in English, but it's also one of the most versatile. To truly master the word, you would need an entire deck of cards that covers usages like take over (someone's responsibilities, a country), take away (food, possessions), take out (food, someone to dinner, trash), take in (scenery, a stray animal), take into account, take on (a characteristic, a challenger, a responsibility), take off (in a vehicle, clothes, monetary charges, food toppings), take up (a hobby, time, someone on their offer), take down (a regime, a poster, an online post), take care of (a task, a person), take stock, take ten, take risks, take back (one's words, stolen property), take a back seat (literally and figuratively), taken aback, take by force, take by surprise, take medicine, take someone to school (literally and figuratively), take one back to their childhood, take someone to task, and so on. Just having one word card for "take" isn't going to prepare an English learner for much of anything, no matter how many times they review the wikipedia page of information on the back of the card before marking the card as "good". In the end, the most pragmatic piece of information a word card is likely to impart to the learner is the word's pronunciation.

There are many drawbacks to this style of card, not the least of which, especially when viewed from the perspective of practice, is that this card style prepares you for exactly no real-life situation, nor accurately represents your practical knowledge of a word. Even if you can recall a word's pronunciation and all of its succinctly summarized meanings every time, that doesn't mean that you're familiar with how the word behaves, nor guarantees that you'll correctly interpret the word when you encounter it in real life, because word cards aren't a test of word usage, they're a test of word summaries. That's like saying you know someone because you can say their name correctly and recite some of the qualities listed on their dating profile, not even needing to list the exact same qualities every time or spend any time with them in person. You don't get to know a person by SRS-ing the information on their dating profile, you get to know them by spending time with them in a variety of situations interacting with the environment and other people, noticing their quirks, who they get along with, and how they behave in different circumstances. The fact is, you can know a person and their tendencies without being able to accurately summarize them, and you can know a word and its tendencies without being able to accurately summarize the word. In either case, you just get them because you have a feel for them and what they stand for, and that feeling comes from sharing a variety of experiences, not repeating a summary. Even if you can't summarize what the word "ensue" means, you know that it's something chaos and mayhem often do, and even if you conclude that it means something similar to "happen", you know deep down that the two words aren't exactly the same. Personally, I'd summarize it as "proceed to occur", but that's not because I found that specific definition anywhere or SRS'd the word, it's because that's what the word feels like to me, and the feeling I have about the word is more representative of my knowledge of it than my being able to summarize it. Whether it's getting to know a person or a word, spending time with them in real life is more instructive, effective, interesting, and undoubtedly closer to the reason you wanted to get to know a new person or word in the first place: to enrich your experience of life. Word cards are neither a good simulation of nor adequate preparation for experiencing language in real life.

The particular card in the screenshots above is from a JLPT (Japanese Language Proficiency Test) deck, and in an unfortunate twist of irony, even JLPT questions themselves look nothing like this. You will never be given a word in a vacuum and be expected to prove that you can sound it out and list any or all of the senses in which it can be used. Even the "stop" on a stop sign exists in the context of a traffic situation to help you interpret the word specifically as a command for vehicles to cease moving forward, context that also helps you interpret a "bus stop" sign as a location for pedestrians to gather at rather than as a command for every type of bus to cease moving forward. Recalling every sense of the word "stop" before you get in a car isn't a realistic or efficient way to prepare yourself for encountering the word on the road or taking a driving test. Spending time on the road, however, IS a realistic and efficient way to encounter and learn the things you need to know for that environment, whether you're in the passenger seat or the driver's seat. If you're studying for a test like the JLPT (a test of passive understanding), you'd be better off immersing in past exams to see how the words you need to know form into sentences and larger ideas on actual tests. In other words, spend time in the environment you're trying to prepare for, rather than trying to prepare for any environment using no environment.

In an effort to test particular word usages by giving more context, sentence cards feature a complete sentence on the front side of the card, and a translation of the sentence and/or definitions of key words on the back, with images and audio as additional options for either side. These cards test whether you understood the sentence as a unit, or perhaps if you understood a particular word usage within the sentence if the word is highlighted somehow. Although sentence cards illustrate particular word usages through context, they still have the same drawback of it being unclear what information you need to remember in order to pass the card. Do you need to recognize every component of the sentence, just the general gist of the sentence, or a particular word within it? Even if you establish clear criteria for evaluating your grasp of a sentence card, because you're seeing the exact same sentence repeatedly, you're evaluating your understanding of words based on how familiar you are with one particular context, and the context may be what you're remembering rather than the words themselves. It's situational variety that tells you most about a word, so the best you can do with sentence cards is create multiple cards for the same word with slightly different usages on each card. At that point though, it becomes clear that you're attempting (in vain) to recreate the experience of immersion through flashcards.

This is the conundrum of recognition cards: Sentence cards illustrate particular word usages through context, but because the context can be remembered through repetition, they aren't an accurate test of linguistic understanding. If you try to solve that problem by using word cards with no context, you're testing all word usages at the same time, you aren't comprehending or evaluating words in a realistic way, and reviewing all of a word's usages on the back of a flashcard isn't a test, nor is it very engaging or any fun.

Fortunately, passive immersion is already a fully formed recognition exercise that solves this conundrum, because it has context to illustrate particular usages of words, but the context isn't being recalled from memory since you're encountering it for the first time, requiring you to think more about what the words mean. You're given all the words up front just like with sentence cards, and faced with the exact same question: Did you understand the words you just read or heard? By being more dynamic and giving you a different usage example for each test, passive immersion does a better job of testing and strengthening word comprehension than recognition cards can. Although novel (i.e. new) contexts can still be used to figure out what unfamiliar words mean without knowing them beforehand, figuring out what words mean is the first step to understanding and remembering them, and whether you figured out a word through context, a definition, or repeated exposure with slight variations, your understanding is increasing nevertheless, which is the goal anyway, not evaluating and keeping track of word familiarity for an algorithm.

But if you aren't keeping track of word familiarity, you aren't learning words as efficiently as possible through optimal spacing and i+1 word exposure, right? As we've seen above, neither type of recognition card can accurately gauge how well you know a word, since sentence cards are familiarizing you with specific repeated contexts, and word cards aren't familiarizing you with any contexts. If your flashcards aren't an accurate representation of how well you know the words on them, what kind of learning is spaced repetition even optimizing, and how helpful can an algorithm like i+1 be if it assumes that you actually know the words on the flashcards you've seen? The hapless English learner that marks "take" as a known word is in for quite a surprise when those same four letters continue to mean something different alongside other unknown words that are supposed to be the "plus ones" for each sentence. The uncertain thing about recognition cards is that even if you're getting the cards right, there's no guarantee that you're improving at anything other than the flashcards themselves. First make sure that you're practicing what you actually want to get better at, then worry about optimization. If you want to improve at understanding a language as you spontaneously encounter it, practice understanding that language as you spontaneously encounter it. Don't just practice something specific, practice what you're specifically trying to get better at.

By immersing naturally, each sentence gives you a novel structure from which to understand its constituent words that you probably haven't encountered before in that exact arrangement, with those exact nuances, with those exact conjugations, or with that exact context preceding it. Not only is this a more realistic and robust test of word comprehension, but more interesting as well, because you're exploring the bigger ideas of the article/chapter/episode as you try to understand each sentence spontaneously as the story or thesis unfolds. Understanding a word in 10 unique contexts via immersion will do far more for rounding out your comprehension of that word than encountering it in the same exact sentence (or with no sentence) 10 times in a flashcard, even if those 10 times are spaced just right with the perfect repetition algorithm.

"But spaced repetition isn't for increasing comprehension, it's to keep you from forgetting what you do comprehend", you may say. Yes, you will forget word usages if you don't encounter them often enough, but if your word cards aren't testing usages to begin with, you're not going to encounter any unless you diligently read the usages on the back of your cards, so seeing usages in immersion might be the only thing keeping you from forgetting them. Even if you are studying word usages on the front of sentence cards or the back of word cards, SRS algorithms are based on the assumption that you won't otherwise encounter the flashcard material between repetitions, and that's fine to assume for younger cards with shorter intervals, but if you're not encountering the usages on your mature cards between repetitions, those usages are either not relevant enough to you personally, you aren't spending enough time with your target language, or both. Immersion addresses both of these issues by keeping you engaged with your target language and showing you the very usages that are relevant to you and your interests, with the most important usages repeatedly revealing themselves to you as a natural part of the process. If a word usage really is that essential to you given the content you're interested in consuming, you should naturally encounter that type of usage in immersion at least as many times as you'd encounter it via SRS. If you're putting in the immersion time but still aren't encountering a word usage as often as you'd encounter it on a long-term flashcard, that usage probably just isn't that relevant to you. In most cases though, you probably aren't spending enough time with your target language, and if you're not consuming enough material, of course you're going to feel like you need a supplement. However, don't try to guess what supplements to take if your diet is insufficient in the first place. Stick to a steady diet of native material first, and if you still don't feel like you're getting the nutrition you need, then try supplementing your diet with SRS.

All that being said, if you feel like you're benefitting from recognition cards (especially during their shorter intervals) given the amount of time you can commit to language learning, perhaps using the cards until their intervals grow beyond a certain length and then suspending them would be a good middle ground that concentrates word exposures and prevents you from being crushed under an ever-increasing snowball of cards during the time you DO have, giving you time to encounter fresh material via immersion, material that will inevitably have known word usages in it for you to recognize as a review anyway.

If you already understand the value of immersion but are studying and/or mining recognition flashcards in order to reduce your dictionary lookups for smoother immersion, I'd like to point out that spending time in a flashcard program to create and study recognition cards is putting more fits and starts in your immersion than dictionary lookups ever could (especially if you use a pop-up dictionary), and the quantity and quality of recognition practice you'll get with flashcards will be less than if you just kept immersing, so it will actually take longer for the lookups to lessen and the smoothness to materialize. Mining cards is fine if the fruits of your mining are a unique refinement of the material they were mined from, but the idea of mining fragmentary recognition material from a more robust and refined recognition activity (immersion) has the whole metaphor backwards. That's like taking a pickaxe to a chest of minted treasure in order to chip off a pile of cruder, less valuable treasure fragments, when the treasure itself will enrich you more than its fragments ever could. When you mine a word from native material, you're not chipping a gem out of random rock, you're taking a piece out of a completed jigsaw puzzle to study the piece individually rather than studying the puzzle as a whole and seeing how its pieces fit together to create a larger scene.

First, let's compare the quantity of recognition practice you'll get with each method. Even if your reading or listening speed during immersion is a modest 15 words per minute, that's effectively the same as reviewing 15 word cards per minute (one every four seconds), and that's assuming that all word recognition is qualitatively equal. Even at that pace, most of the target-language definitions, word usage examples, pitch accent graphs, images, audio, and other content on the back of your flashcards becomes irrelevant, because you can't peruse it all in the span of 4 seconds anyway. As your reading and listening speeds improve, it will only become less and less possible for you to recognize words at a similar pace via flashcards. No matter how optimized your card mining workflow is or how fast you can flip through recognition cards, your recognized word usages per minute and total recognized word usages will quickly be higher (not to mention more varied) if you simply look up the usages you don't know and keep moving through more material via immersion.

Now, is the quality of recognition practice the same for a word encountered on a flashcard and a word encountered during immersion? I think you know the answer. You're always going to earn more experience points (XP) for encountering a word in unexplored territory with a new mix of party members around it compared to grinding that same word over and over in familiar isolated battles you've already seen before. Not only are you earning more XP by exploring, you're also progressing the story of your language learning journey and experiencing the content you came to experience. Sticking with immersion will result in faster level-ups and more story progression, which ends up being the smoother of the two approaches.

If dictionary lookups are hitting you hard enough to cause a game over, change the playing field and/or upgrade your lookup equipment so that looking up words during immersion isn't as soul-crushing as grinding recognition flashcards in the first place. With OCR (optical character recognition), text hooking software, clipboard monitors, browser-based video players, etc., combined with pop-up dictionary browser extensions like Yomichan or even more full-featured lookup solutions like Migaku, Language Reactor, and Game2Text, lookups don't have to be as cumbersome as you think, and can even be handled as quickly as reviewing a recognition card. If you're doing a lot of lookups, start with content that can be viewed in a web browser or easily imported to one so that you can use a pop-up dictionary for the fastest lookups. Once your lookups start to lessen, try OCR for digital content that you can't easily text hook, like comics or (non text-based) PC games. Once you're fairly competent in your target language, then attempt content like physical books or console games, which require the most time per lookup if they aren't being digitally captured somehow. Looking up words only as you need to also ensures that you only do the minimum necessary to understand the content you're interested in consuming, without needing to guess what's going to be useful to you in the long run. If you only have to look up a particular word usage a few times in your life, you didn't need to commit that usage to long-term memory in the first place, and by looking it up only when you needed to, you saved yourself unnecessary repetitions in a flashcard program, questioning your own learning commitment as you commit to learning things you might not need in less dynamic, less interesting, and less effective ways.

To beat this dead horse one last time: Whatever your goals for your target language, I don't think that recognition flashcards have a uniquely useful place in your studies, especially at longer repetition intervals, and spending time on them will most likely only add a stale, cumbersome, demotivating, ineffective, and unnecessary chore to your learning that avoids attempting more challenging, effective, and engaging methods that you've already used to successfully learn your first language. Like I said before though, if you find recognition cards motivating and/or enjoyable, perhaps due to their familiarity, ease of use, quick reviews, visibly improving interval stats, or whatever, by all means keep using them, especially if they're keeping you from giving up. If not though, or if you feel like giving up because of such cards, don't think that you have to use them or that you're missing out because they might be a more efficient shortcut to immersion. They're not! They're a less dynamic facsimile of immersion in the best case, and a glorified pronunciation test in the worst case, which can be comforting in its own way, and if you need that comfort and don't otherwise feel bothered by having such reviews pile up, don't feel bad for using them. Just remember to make time for passive immersion too, even if that means retiring your mature cards.

So, if passive immersion is already more effective for acquiring passive language skills than studying recognition flashcards, what other kinds of flashcards are there that might have some unique benefit to offer? If you're wanting to develop the active language skills of speaking and/or writing, production flashcards, which you may have heard of if you've heard of cloze deletion, are effective at increasing both your active and passive vocabularies in a language, and they do something that passive immersion cannot: they prompt your recall of words via meaning and context without immediately showing you the words you're trying to learn. It's like showing you a section of a jigsaw puzzle with a piece missing and asking you to pull that piece from your memory and align it to fit the surrounding pieces. This increases your mastery of how words sound, how they look, and how they're used, eventually helping you to output full sentences on your own by practicing usage of one piece at a time.

Production cards don't give you the word you're trying to learn on the front of the card, but give you everything but that word, fill-in-the-blank style, including the surrounding sentence(s), word definitions, and/or sentence translations for context, with hints to guide you to think of that particular word when multiple other viable options exist. "Other viable options" is what I call the synonym problem of production cards, when it isn't clear exactly which of several suitable words could fill in a blank. My solution to this problem has been to create a hint system for blanks that rules out similar words through character counts and per-position disallowed characters.

For example, if you're trying to fill in the blank for "The thief [.≠e..] through the window and rolled to his feet", and the definition for the missing word is "to move suddenly downward from a higher to lower level, typically head first", the learner can guess from context that the word needs to be past tense, and if the answer is four characters (noted by the three periods representing "any character", plus one "not the following character" ≠), and the second character is not (equal to) "e", then the answer can't be "fell", so it might be... let's see... "dove". After deciding on what might fill in the blank, checking the back of the card reveals the full sentence including the missing word: "The thief dove through the window and rolled to his feet." This tests whether the learner can recall and use the word "dive" as a verb in the past tense, which will not be confused with the noun "dove", a type of bird with a different pronunciation, nor the noun "dive", one sense meaning "a shabby establishment where you can order alcoholic drinks". These other usages might be possible if this were a recognition-style word card with nothing but "dive" or "dove" on the front, but production cards sidestep this pitfall and allow the learner to practice recalling particular word usages and conjugating them if necessary, similar to what would happen if the learner were constructing their own sentence and needed to recall and use an appropriate word for the concept they already have in mind. This is why it's okay to have meanings on the front of production cards, including translations: When you're outputting (i.e. speaking or writing), you already know the meaning of what you want to say, you just don't necessarily know how to say it. If you didn't know the meaning of what you were trying to say, you wouldn't have anything to say in the first place, so giving the meaning on the front of a production card is a fair starting point for guided output. Production cards are requesting a word given its meaning rather than requesting a meaning given the word.

"But production cards use sentences for context like some recognition cards. Don't those both have the same issue (of testing context recognition more than word recognition)?" If you ask the question more vaguely without the parenthetical bit, it sounds like it might have a good point, but if you include the exact problem with sentence recognition cards in the question, it starts to point to the answer: production cards aren't testing word recognition; they're testing word recall. You can't fake knowing a word if the word is missing and needs to be filled in by you. Even if you remember the context, you still have to come up with the missing word yourself, and if you can, it proves that you recognize which sounds and/or symbols mean what, and that you deemed such sounds and symbols appropriate for the situation, which is exactly what you need to do when outputting. You can't use a word unless you have a context to use it in, and by learning an appropriate context for recalling a word, you're actually aiding your learning rather than sabotaging it. Sure, the more contexts you learn for recalling a word, the better, but learning at least one context is better than none, and a familiar context doesn't short-circuit recall the way it short-circuits recognition. Here's an illustration of the difference context plays in recognition vs recall: If you say ありがとう ("thank you") to someone in Japanese, and they say something back with a smile that sounds more complicated than いいえ ("not at all"), you can guess from context that the string of sounds is probably "you're welcome", that longer phrase you can't quite remember just yet. Just because the context reliably suggests that the person said "you're welcome" doesn't mean you've learned the word for it, but if someone says ありがとう ("thank you") to you and you respond with どういたしまして ("you're welcome"), there is no doubt in anyone's mind that you know the situational meaning, basic pronunciation, and proper context for that word, because you just proved it by recalling and using it appropriately yourself.

One drawback of production cards is that it's difficult and time-consuming to formulate hints that rule out every possible word except one, and I'm continually thinking of similar words that could fit in a particular blank and editing my hints, especially as my vocabulary expands in my target language. Sometimes I use a thesaurus to rule out every last obscure synonym with the same number of characters, but that can quickly become tiring and borderline unnecessary as some synonyms are rare and unnatural sounding. More often than not, my hints improve as I actually use the cards. If I answer a card incorrectly with a suitable word that wasn't ruled out by my current hints, I edit the hint to rule out the suitable word I thought of, bury the card, and try it again with better hints next time. In this way, production cards are always improving as your language ability improves, but the cards are never truly finished, much like your learning. If production cards gain more traction in language learning communities, perhaps one day we'll see automation of the synonym-slaying process in projects like Language Reactor and Migaku, which would be a huge boon for production card creation. If I'm allowed to dream even bigger, we might even see active immersion tools that perform synonym-free cloze deletions in immersion material you've never seen before, eliminating the ambiguity of repetition's role in assisting recall and assuring that you can use words in novel contexts. Honestly, I would use this to improve my active vocabulary in English as well, and I'm a native English speaker.

Another potential pitfall of the hint system is getting thrown off by the other ruled-out possibilities before coming up with a guess yourself. If you first try to imagine all the other possible answers that are ruled out with not-equal-to (≠) hints, it could be intimidating and lead to self-censoring while trying to think of words yourself. Ideally you would make your first guess based purely on meaning and context, check your guess against the hints, and if your guess is ruled out by the not-equal-to (≠) hints or character count, think of another possibility and check the hints again, repeating the process until you think of a suitable answer that isn't ruled out. Even if you recall a word that ends up being ruled out, recalling the word is still legitimate practice at making connections to similar vocabulary and using the word yourself, so there is little disadvantage in recalling more words.

Another drawback is that translations, despite their usefulness for generating multiple cards from a single sentence and testing grammar structures that are hard to pin down with a definition, are hardly ever one-to-one equivalents due to differences in language conventions. For example, in the example screenshots above, one might change the "Please rest your weary bones" translation to the equally unlikely but less native-sounding "Please rest your tired body" in order to be more one-to-one with the Japanese, but since testing the transitive usage of "rest" (i.e. to rest something) is the core point of the card, anything beyond that is more of a stylistic choice, but it can nevertheless be tricky to strike a balance. Thankfully, for non-grammar cards with only one vocabulary item per sentence, a simple definition of the missing word will usually suffice, and translations don't feel necessary in such cases.

Another characteristic of production cards that will quickly become apparent is that they're more difficult to answer than recognition cards, and thus take more time to answer, especially if you're trying to improve speaking and writing at the same time. This is to be expected, since recognition cards are essentially giving you most of the word information on the front of the card, and only ask you to mentally fill in some details, however vaguely, about what the words mean or how to pronounce them. Production cards, on the other hand, only tell you what a word means and where in a sentence it can be used, and require you to enter all other details about the word into a text prompt: how it sounds, how it looks, and how it's conjugated. If you're used to reviewing recognition cards at a pace of 20 per minute, you might be disheartened to see that you're only reviewing something like 3-6 production cards per minute. Personally, as someone who writes my answers by hand into the cards' front-side text box using a Wacom drawing tablet, I only review about 2-3 cards per minute. If you're concerned about your stat graphs and getting the most cards "done" per minute, that's fine if those things motivate you, but if actual results are what you're after, production cards will bring you results that even immersion cannot.

If you're like me and trying to improve both speaking and writing via production cards, there will be times when you remember how a word sounds but not how to write it, and vice versa, and need to decide whether to pass or fail the card. This isn't so different than recognition cards for which you remember the meaning but not the reading, and vice versa. In such cases, I would make the decision based on your priorities, and also keep in mind that burying cards is also an option. For me, knowing how to pronounce a word seems most important since that allows me to at least type the word phonetically and choose a form I recognize from the autocorrect options. So if I remember a word's pronunciation but write it incorrectly, I'll usually bury the card and try writing it correctly next time, not failing the card since I can already recall the most important information, but not passing it either and potentially waiting a long time to practice the word's written form. If I write it correctly the next time it comes back up after being buried, I'll then pass the card and increase its interval. Conversely, if I can write a word but forgot how to say it, I'll usually fail the card and practice it at more frequent intervals to reinforce its pronunciation. For words with multiple acceptable written forms given a specific character count, I've included an "AlternateAnswers" field that our Discord quiz bot makes use of, but I don't currently know of a way to get Anki to accept multiple possible answers, so you'll have to make your own evaluations in such cases. If in doubt, you can always bury the card and write it a different acceptable way next time.

"But how do you know that the words you're learning to output will come in handy?" you might ask. Unfortunately, there is no way to guarantee what you'll need to know for output, and taking guesses at the words you might need is an unavoidable part of the preparation, because output simply requires more preparation than input. When you encounter a word you don't know in passive immersion, you can just look it up and keep reading or listening at whatever pace you like. However, when you encounter a concept you don't know how to express in conversation, there are no pop-up dictionaries (yet) that display the word you're looking for, so you can either A) make the person you're talking to wait while you look up an appropriate word, or B) explain the concept using less specific words that come to mind. In nearly all cases, you're probably going to choose option B, so the more words you have available to explain concepts you don't know how to succinctly express, the smoother your output will be.

In case six paragraphs of caveats and warnings weren't enough to pique your interest in production cards, their advantages include testing just one word usage, phrase, or grammar concept at a time, and keeping the tested words off of the front of the cards so that they have to be produced from memory. There's enough context, meaning, and hints on the front of the card to facilitate recall of the target word without simply letting you repeat the words you're most familiar with to complete the sentences (as you might be tempted to do with more spontaneous but less structured freeform output). This unique feature of production cards allows them to expand your active vocabulary even in your native language (if you were so inclined to make/study flashcards for your native language)! Freeform output is still of course invaluable for developing your speaking and writing abilities, but it's a huge leap to go from 0% fill-in-the-blank passive understanding of a language to the 100% fill-in-the-blank required of freeform output, especially when you want to structure your output to use more word variety and expand your active vocabulary. Viewed this way, production flashcards are an intermediate 5%-20% fill-in-the-blank stepping stone between passive language recognition and freeform language output that allows you to practice recalling specific vocabulary, strengthening your active recall and passive recognition of the target words at the same time. If you know how to properly use a word when you recall it for a production card, your passive recognition of that same word will only be reinforced by its active usage, because usage of a word is only possible when you recognize what it means. By aiming higher up the mountainside of the language acquisition process, production cards' output benefits seep down to strengthen your foundational understanding as well. If a word has made it into your active vocabulary, it's also rooted in your passive vocabulary by necessity, and such words are harder to forget than words you've only recognized passively but don't actively use yourself.

Another benefit of production cards that is specifically helpful for languages like Japanese that feature non-alphabetic writing systems like kanji is that they serve as a "spelling" test to reinforce your command of the writing system in a more practical way than cards testing only one character at a time. Testing one character at a time is certainly a good first step for learning the characters themselves, but most non-native learners stop there and end up slowly forgetting character by character. Even if you remember how to write all the characters of a language, it doesn't mean you know how to spell, and for me personally, production cards are the only reason I can spell anything at all with kanji when writing Japanese. The usefulness of writing skills in the digital age is certainly debatable, but practically speaking, it does speed up dictionary searches in the absence of a pop-up dictionary or OCR, especially when you don't know the pronunciation of a word, because you can use a handwriting keyboard without searching by radicals or needing to think of a word that uses each kanji you're trying to input phonetically. Philosophically speaking, the idea of only phonetically inputting a language that has semantic characters amounts to relying on autocorrect for the maintenance of basic literacy, but that may not be a thought that troubles you. I'm not going to insist that any of this makes learning how to write and spell "worth it", but aiding such learning is an advantage of production cards nonetheless.

TL;DR: While recognition flashcards have some limited benefits for improving your passive understanding of a language, those benefits aren't unique to such cards, and passive immersion actually does a better job of broadening and strengthening your passive language skills because it is literally practicing the very thing you're trying to get better at: understanding your target language spontaneously as you encounter it in its myriad permutations. However, neither method specifically addresses building an active vocabulary for output. Even exercising one's active vocabulary through freeform output (e.g. verbal conversation or written compositions) can result in using the same familiar words repeatedly. If you're trying to expand and strengthen your active vocabulary in preparation for or in addition to freeform output practice, help bridge the gap between input and output with production flashcards, a top-down approach to building and strengthening your active and passive vocabularies at the same time.

If you're interested in trying production flashcards and happen to be learning Japanese, that's the language I've been learning too, and using the cards I've already made for Japanese might give you a better idea of how such cards could benefit your learning and save you the time of creating cards yourself, especially if you're unsure whether they're right for you. You can find a collection of my Japanese production flashcards here. If this link isn't working (saying something like "This item is currently unavailable."), it likely means I've updated it within the last 24 hours, and it will become available once Ankiweb's 24-hour probationary period is over, so try the link again tomorrow. In order to use the cards, you'll need the flashcard program Anki. More information about particular decks in the collection can be accessed from the navigation menu at the top corner of this page. These pages give more detail about particular card layouts, and may include guides on how to suspend cards that you're not interested in or not ready for yet.

If you have any questions, comments, or feedback about the cards, feel free to let me know on our Discord server. We have a production card quiz bot there, so you can try answering production cards with others in a multiplayer format.

Thanks for reading, and the best of luck to you on your language learning journey!

Oct. 30, 2024 Update:

I've attempted to summarize the various language skills exercised per type of activity as they pertain to Japanese in the following chart. I hope it can aid you in choosing activities based on the skills you're trying to develop. (Click the top right corner to open it in a new window):