Learning Languages From the Content You Already Read
Most people who work through a year of Spanish can produce the vocabulary of the course and not much else. The gap shows up the first time they open a real page of Spanish, whether that is a news site, a history article or the subtitles of a film. The words on that page were chosen by whoever wrote it, and a course cannot know in advance what any particular learner will want to read.
This is a solvable problem, and the solution is older than any app. Read the things you actually want to read, and keep a list of the words that came out of them.
What a course covers, and what it leaves for you
Vocabulary in any language is distributed unevenly, and courses exploit that. Studies of English text put the most frequent thousand word families at roughly 80 percent of everything written, and the figures for Spanish, French and German come out in the same range. That first thousand is what a beginner course teaches, and it is the right thousand to teach.
The trouble is the shape of the curve after that. Reading a novel or a newspaper comfortably, which researchers usually define as knowing 98 percent of the words on the page, takes something closer to eight or nine thousand families. The distance between the course vocabulary and comfortable reading is thousands of words long, and almost none of them can be taught in a sensible order, because which ones matter depends entirely on what a given person reads.
A learner who reads about archaeology needs a different four thousand words from one who follows football. Both of them will get there faster from the texts themselves than from any generic list.
Read about things you already know
The cheapest way to read above your level is to pick a subject you already understand in your own language. Background knowledge does the work that vocabulary would otherwise have to do. If you already know roughly how the Silk Road trade worked, or what happened in Japan after 1868, or how terraced farming handles a mountainside, then a Spanish article on the subject is half decoded before you start. You are recognising words rather than deducing meaning from nothing.
The same trick works twice, because a subject repeats its own vocabulary. Two articles about irrigation share most of their specialist words, so the second one costs a fraction of what the first one cost. Learners who pick a topic and read four or five pieces on it usually find the fourth one almost comfortable, which rarely happens when they hop between unrelated subjects at the same level.
A rough test for whether a text is worth your time
If more than one word in twenty is unfamiliar, you are decoding rather than reading, and decoding is slow enough that most people quit. One new word in fifty is close to the comfortable end, and one in twenty is about the working limit.
When a text is over the limit and you want it anyway, the usual repairs are to read a shorter version first, to read a children's or reference version of the same material, or to read the piece twice and only take words on the second pass. Subtitles work in the same way for video, with the advantage that the picture supplies context the way illustrations do in a book.
Keep your own list, not someone else's
Published frequency lists and shared decks have the same weakness as the course: they were built for an average learner who does not exist. A list that comes out of your own reading is shorter, more useful and more likely to be finished.
Three rules make a personal list work.
Skip the rarest words. The eye is drawn to the strangest word on the page, which is usually the least useful one. A dynasty name, a species of bird, a technical unit of weight: memorable, and gone from your life for years. The words worth keeping sit in the middle, common enough to appear in the next thing you read but not so common that you already had them.
Cap the number. Ten words from one article or one chapter is a good working limit. A list of forty never gets finished, and an unfinished list teaches a learner that the language is homework without an end. Ten taken on Tuesday and done by Friday teaches the opposite, and the thirty words left behind will come around again in the next text anyway.
Write down the line, not the definition. Cosecha: la cosecha de arroz se retrasó por las lluvias is easier to hold on to than "harvest, noun". The sentence carries information a gloss drops: which words the new one sits next to, whether it describes people or things, whether it sounds formal or ordinary.
Two short sessions, on different days
One review is enough to remind you what a word means. It usually takes a second one, with a night's sleep in between, before the word comes back on its own.
Five minutes tomorrow morning does more than twenty minutes tonight, because the interval between the sessions is doing much of the work. Anything you get right on both occasions is finished, and it should leave the list permanently.
That last part matters more than it sounds. A list that never lets go of a word turns into a treadmill, and the pile stops shrinking no matter how much work goes in. When finished words leave for good, the list visibly gets shorter, and a shrinking list is the thing that makes people start another one next week.
When the notebook stops coping
For a few articles a notebook is fine. It starts to fall apart somewhere around the tenth text, when nobody can remember which words are already done and half the review time goes on words that were learned in March.
That bookkeeping is worth handing to software, and there are several apps that will do it. One of them is Fluera, an app to learn Spanish from the things you read: paste in a passage you have just read and it pulls out the words, puts them in the order worth learning next rather than by how rare they are, and leaves out anything you already know or have already finished. People who learn languages this way let a word leave the list for good once they get it right twice on different days.
What a week looks like
One article on Tuesday, on a subject you would have read about anyway. Ten words afterwards, each with the line it came from. Five minutes on Wednesday morning, five more on Friday. By the weekend seven are finished and three need another go.
Next Tuesday, another article on the same subject, which is now noticeably easier than the first one was. The three leftovers join the new list, the seven finished ones stay gone, and the pile keeps getting smaller instead of longer.