Aim-Shoot-Skin for Language Learning

Aim–shoot–skin is a language-learning loop that starts from a passage the learner can almost follow. It does not start from a word list built in advance. Each missing word or pattern goes on a flashcard and is used in new sentences the same day. The pile of cards is compressed later, and only if the compression is worth the time.

Aiming

Aim means choosing one specific passage that is just too hard to understand and turning it into questions the learner actually wants answered. The general version of this step, writing questions the material must answer, is on Aim. In language learning the unit is one dialogue, one scene, or one stretch of talk. A word learned in order to understand that passage is remembered longer than a word memorized in advance.

The middle band is judged by feel. It is not defined by a percentage of known words. A passage in the band cannot be transcribed word for word. Its general meaning can still be worked out from words already known plus body language, tone, and situation. Those cues outside the words are what make a spoken dialogue possible to follow when the share of known words is so low that the same words on a page would be incomprehensible. Acquisition is fastest somewhere in this range. The range is approximate and is not a measured peak. Every word that cannot quite be placed is one the learner wants to resolve, because the rest of the meaning is already partly understood.

The passage is chosen by what the learner is curious about, limited by how the language will be used, the situation it will be used in, and the effect a line is meant to have. Choosing by word frequency alone amounts to building a list first. The loop rules out building a word list first and then looking for a text that fits it. A small amount of preparation that makes a passage possible to start is a separate task, and the rule does not apply to it.

To get more questions out of one dialogue, first write variations of the exchange in a language the learner already speaks: the line as it was said, the reply the learner would have given, and two or three replies the other person might give next. Three or four variations from one real exchange is a workable number.

A fluent reader working on technical writing aims for two or three unknown words in a hundred, not one in a thousand. The target band gets narrower as the learner’s comfortable band gets wider. The steps stay the same at every level, and only the material changes. The loop is one sequence: start a passage that can almost be followed, stay with it while it is only partly understood, learn the specific missing word or pattern that would complete the meaning, and the same passage then becomes comfortable. After that, the target band moves to harder material.

Shooting

Shoot means looking up exactly the missing word or pattern, making a flashcard for it, and using it in new sentences the same day. The general version of this step, building the map against the questions, is on Shoot. Careful lookup cannot make up for a badly chosen passage. If the learner did not actually want to understand the passage, the lookups serve no purpose and there are no same-day sentences for the cards to be used in.

The lookups happen during Interactive Immersion. This loop limits them: look up exactly what is missing, meaning the gaps that the passage showed.

Every new word gets a flashcard and is used in new sentences, replies, and rewrites. Use it again the same day, more than once. Flashcards are used only to fill gaps. Using the word for real is the main retrieval practice. Extra retrieval exercises are added only for items that have not yet come up in use. The card is a record of the word. It is not a second, separate place where the word gets studied.

Cards without use are forgotten. A word that never appears in new sentences gets no connections to other words, and reviewing it becomes pure maintenance. That description is this knowledge base’s own picture of overwhelm across a whole language, not a measured count of reviews. Use without a record loses words. Without a record, the learner feels fluent until a word comes up again and turns out to have been learned once and then forgotten. These two failures are why both halves are required.

Use the word in speaking, listening, reading, and writing, with more weight on the skills the learner’s goals need, and never in only one of them. Pronunciation and spoken fluency are physical skills. They need practice in producing speech, and another recognition card does not provide it.

If reviews feel like raw memorization, the words are not being used enough. With enough use, the learner stops thinking about the card and simply knows the word. A passage that was barely understandable the day before becomes comfortable. When the band moves, choose a new passage.

What the loop is for

Aim and shoot do most of the work. Skin is optional. A learner can reach fluency with aim and shoot alone.

The loop is Inquiry Based Learning adapted to a subject made of thousands of small items: questions are written first and then answered, rather than exploring and expecting learning to follow. It is the language version of the Bear Hunter System, the four-pass encode workflow. Aim is questions the material must answer. Shoot is the working map against those questions. Skin is cutting the material down to what remains useful without the source. The linked pages for those three steps hold the general versions of each step.

An immersion practice already provides the material. This loop is added on top of the three-mode immersion system at Refold Language Learning System. The three modes are preparation, interactive, and freeflow. The loop does not replace that immersion routine. Preparation is the small amount of study that makes a passage possible to start. That preparation is not the list-first habit that this loop rules out. Three Pillars of Language Learning describes the same three-mode routine.

Aim changes content selection from choosing material that is “comprehensible and compelling” to choosing a specific passage and writing variations of it. Shoot gives each lookup a use: a card plus immediate reuse in sentences. That is the limit on lookups that Interactive Immersion already asks for. Skin reduces the size of the deck, which immersion and this loop keep adding cards to. Whether any of it is actually stored in memory depends on the Rules of Effective Memorization. The loop supplies the purpose for learning each item.

The optional pass

Skin is an optional pass over the accumulated cards to find shared roots, components, or shapes, so that the deck can be learned as fewer related units. The general version of this step, cutting the material down to what remains useful without the source, is on Skin. Fluency can be reached with aim and shoot alone. Pattern work helps mainly with reading and writing, and most of all with non-alphabetic scripts.

Shared etymology reduces a pile of words to a smaller set of roots plus rules for combining them, for example photo, photography, photometry. Roots are not obvious even in a native language: soccer comes from association. In a foreign language, the learner has to search for roots on purpose when the target language is Indo-European. Compressing Vietnamese is a different task. It works through Sino-Vietnamese, not through the photo/photography pattern. In Mandarin, the roots are groups of characters that share components.

Read characters by their components. The mechanics are covered on How Chinese Characters Work. This loop only sets when that knowledge gets used. One example is enough to show what the sweep looks for: a box-like outer part read as a boundary, a king-shaped part inside it, and the two together read as the word for the country. Work on patterns between characters and sounds goes through Sound Series, and How Chinese Characters Work covers the rest of the component mechanics.

A card reviews one item. Reviewing cards will never show that “these six stubborn words share a part.” Finding connections requires a scheduled sweep of the whole pile. Without a schedule, it does not happen. Every few weeks, on a schedule this knowledge base sets as its own trigger, pull the entire card list and spend about half an hour looking for similarities that have not been noticed. A pass usually finds a few connections and makes several hard-to-remember words easier.

After a card has failed several times, skin is required for that item. For that card, which is called a leech, do a short pass on its roots and on related and contrasting uses instead of another review that adds nothing. Foreign-language items are the standard example. The monthly sweep is still optional. No specific number of failures is a fixed rule.

Keep the network of connections informal. Mind maps of a language quickly become too large to manage. When the map is no longer manageable, record the patterns in some other way. The loop does not depend on mind maps.

If time is short, do aim and shoot and skip skin. The half-hour search for patterns is worth the time when there is time to spare and the language is written in characters.

What is actually known

Reading comprehension without help requires knowing close to ninety-eight percent of the words. At eighty percent known words, reading is already frustrating. For that reason, the spoken feel-test must not be used for written text. The middle band is an approximate range. It does not mean knowing a measured half of the words.

Research points in one direction: words learned to solve an immediate problem are remembered better than words learned from a list. Research on generation, problem-based work, and elaborative encoding supports that direction. No trial compares “looked up to finish this dialogue” directly against “taken from a frequency list.” The loop’s working claim has not been tested by any study. The literature supports only its direction and is not a strict comparison.

The case against the loop has four parts. Reading material in which only half the words are known is incomprehensible and falls outside the middle band. The loop is useless as a replacement for the immersion routine. Treating skin as required wastes time on a pass that is optional. Refusing all list-first study, with no limit on that refusal, also rules out the small amount of preparation that makes a first passage possible to start.

Most of the cost is in writing sentences the same day. The lookups are not the main cost. Skin costs half an hour every few weeks and is worth it when the language’s script makes pattern work useful.

Signs to quit: cards pile up with no same-day sentences, which is the failure of cards without use; a monthly sweep finds nothing because the pile was never used, which is the failure of use without a record; two passages in a row are completely incomprehensible, which means the passages are in the wrong band and the loop itself is not failing.

What to check for: a passage that was barely understandable one day is comfortable the next day. Reviews stop feeling like raw memorization.

When that happens, the comfortable band is wider and the previous passage is inside it. Choose the next passage the same way: one that can almost be followed, with gaps the learner still wants to fill, and without building a word list first. Skin is still optional.

Links into the knowledge base

• Aim: the general Aim, which is questions the material must answer.
• Shoot: the general Shoot, which is building the map against those questions.
• Skin: the general Skin, which is cutting the material down to what remains useful without the source.
• Bear Hunter System: the four-pass encode workflow; this loop is its language version.
• Inquiry Based Learning: the related method in which questions are written and then answered.
• Refold Language Learning System: the three-mode immersion routine this loop is added to.
• Preparation: the small amount of study that makes a passage possible to start, which is not list-first study.
• Three Pillars of Language Learning: the same three-mode immersion routine.
• Interactive Immersion: where lookups happen; this loop limits them.
• Flashcards: used here only to fill gaps.
• Rules of Effective Memorization: whether a card is stored in memory; this loop supplies the purpose.
• How Chinese Characters Work: component mechanics; this loop only sets when the pass happens.
• Sound Series: pattern work between characters and sounds.

Open Questions

How far below the reading-coverage numbers the share of known words can drop in a spoken dialogue before the session becomes incomprehensible, when gesture and situation are carrying a real part of the meaning.

What the monthly sweep looks like for a pile of Sino-Vietnamese cards, where searching for Indo-European roots does not apply.

Sources

• Hu, M., & Nation, I. S. P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1), 403–430. Unassisted reading comprehension near 98% known words; 80% already frustrating.
• Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? The Canadian Modern Language Review, 63(1), 59–82. The range of 2% unknown and 98% known words for unassisted reading.
• Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 104(3), 268–294. Elaborative encoding; supports the direction of the claim that words learned to solve an immediate problem are remembered better.
• Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. P. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 185–205). MIT Press. Desirable difficulties; generation and effortful retrieval support the direction of the claim and are not a strict comparison.
• Karpicke, J. D., & Blunt, J. R. (2011). Retrieval practice produces more learning than elaborative studying with concept mapping. Science, 331(6018), 772–775. Retrieval and generation can produce more learning than elaborative study; this is still not a trial of a dialogue against a frequency list.