I like learning languages a lot! While there are many ways to learn languages, there are some approaches that I think are more efficient and fun than others. This note will be a living record of my experience with and thoughts on language learning so far.

The main takeaway for others should be that teaching oneself a language using just the internet is:

1. very doable,
2. can be done in 15-30m a day,
3. can be quite fun, and
4. can yield surprisingly fast rates of progress.

Exposure-based Learning

Like many people, I took a language (Spanish) for most of high school. While it wasn't a very rigorous course, I'd generously estimate that I came away with an effective vocab size of only around 500 words. Considering the number of hours spent in class ($6 \text{ years} * 40 \text{ weeks} * 2 \text{ hours} =480 \text{ hours}$) compared to the US government's estimate of 690 hours¹ to reach some level of fluency for Spanish (a Category 1 language), coming away with just a 500 word vocab seems pretty bad.

From talking with (American) friends, it seems pretty common to have sunk multiple years into studying a language in high school with relatively little to show for it. And yet we also see folks getting to impressive levels with languages within just a few years, whether by studying on their own or via (perhaps better) courses. I don't think there are great differences in people's natural aptitudes for language learning, and think rather that the difference typically comes down to approach.

Let's consider two very different approaches to learning a language.

The first approach is to memorize words and conjugations and to study grammar to learn all the patterns. When reading or listening, adherents of this approach strive to understand 100% of materials consumed in the language. When speaking, they consciously recall conjugations and grammar rules to ensure what they're saying is correct.

The second approach is to never explicitly study or memorize any part of the language, and simply to read and listen to the language at a difficulty level just above what one can comfortably understand. Adherents of this approach aren't bothered when they don't understand every word in something they're reading or listening to, as long as they understand most of it. When they speak, they aim to express what they want to express without worrying over small grammatical mistakes.

In reality, these approaches are two ends of a spectrum and we can imagine many shades in between. Having said that, I'm convinced that approaches closer to the second are generally more efficient and fun. Here are my main arguments for why language learning by exposure is the right approach for most people and most levels:

1. It's how people learn their first language. We know that learning via exposure worked for all of us at some point to learn our first language. Many people ² believe that humans have an innate ability to learn language and that there's no real need to explicitly teach it.
2. It's more scalable. Languages are really big; there are a ton of words and grammar patterns to learn. Memorization only gets you so far, and beyond that, if you're aiming to have a conversation with someone, you're not going to have time to consciously compute proper grammar and conjugations. Learning via exposure means slowly gaining an unconscious ability to understand a language at something approaching the speed of thought.
3. It's more efficient. For a given hour that you can spend learning a language, you can expose yourself to at least an order of magnitude more words via reading and listening than via studying. Of course, you're not going to have a perfect understanding of all of the words and patterns you were exposed to at the end, but your understanding will hopefully be a tad better for each, which still scales well.
4. It's more fun. It's super satisfying to notice that you can understand sentences that would've stumped you a few days before. On the flip side, unless you're a freak, you probably find memorizing conjugations boring.
5. In my experience, it just works very well. When I started taking a more exposure-based approach to learning Spanish, I learned more Spanish in a few months than I did in the entirety of high school. Some of this was due to me putting in a lot more effort because I was having a lot more fun, but I think that's relevant to factor in too!

The exposure-based approach that I'm arguing for is largely in-line with the set of hypotheses about second-language acquisition that linguist Stephen Krashen put forth in the 1970s. If you google "Comprehensible Input Hypothesis" you'll find many others' takes on it.

Here's a grab-bag of follow-up questions that will hopefully add some nuance.

What does exposure-based learning look like more concretely? The practice of learning a language via exposure is pretty simple. You find content (stories are great!) at a level that is just beyond your current level of understanding. You'll read and listen to this content, perhaps multiple times, without getting too stuck at understanding everything. When ready, you move on to content that is slightly more difficult. Repeat.

Tools like LingQ are great for this. LingQ has collections of stories in different languages that expose the reader to increasingly difficult and diverse words and patterns. The stories are narrated so that you can listen to them too, and clicking on a word or phrase shows the translation (though you should try to understand via context first).

Does this work from day one? I suspect that one could immediately start off and learn even their first words by reading sufficiently simple sentences. Having said that, I haven't really done this for a language yet. With Spanish, I was already at a beginner's level from high school before switching to exposure-based learning. With French I did immediately start with stories, but French shares enough words and patterns with English and Spanish that this wasn't too rough. And with Mandarin I started by using spaced repetition to memorize direct translations of the first few hundred most common characters/words.

If you think you would recognize even just the 100 most common words in a language, I would recommend jumping straight into reading stories.

Where does speaking fit in? In my experience, the ability to speak a language lags the ability to understand the language. Because speaking comes a bit later, I think it's important to shift the goal while learning to "understanding more of the language" instead of "being able to speak more of the language now".

I don't really think speaking improves vocab or grammar understanding³, but I think it's important to practice speaking a language without fear of making mistakes. In fact, I think the more one can speak confidently, the closer their speaking ability will be to their ability to understand.

Is there any place for explicit study and memorization? My belief is that study and memorization can still be useful to the extent that it serves as a hook for understanding at an unconscious level later. Really understanding grammar at an unconscious level can take a while, and if there's a particular pattern that you're having trouble understanding from context, it might be useful to spend a few minutes reading about it to give you a mental hook to use the next time you see the pattern. Though I would say that I very rarely do this, and almost always just aim to read more to expose myself to that pattern so that I gain more data points to understand the pattern in its context.

To jumpstart my Chinese, I did memorize ~150 words before struggling through my first stories. Like I said above, I'm not sure if this is needed for a language that's easier for English speakers to pick up.

Experience with Different Languages

Spanish

• Level: Conversational. Probably at C1 for reading/listening and something lower for speaking. This means I can comfortably read books and watch shows in Spanish and can hold a conversation, but it's apparent that I'm not a native speaker due to grammar mistakes and somewhat slower speech.
• Resources
  • I strongly recommend LingQ (the subscription is ~$15/month and is definitely worth it) for beginners. They have a series of 60 stories that start very simple.
  • Podcasts that are just Spanish/no English (in order of increasing level)
    • Notes in Spanish Beginners
    • Notes in Spanish Intermediate
    • Notes in Spanish Advanced
    • Radio Ambulante
  • Books
    • Olly Richards Short Stores in Spanish Beginner and Intermediate
    • The Little Prince, Spanish translation (I think it's on LingQ)
    • Reading translations of chapter books you liked is great, as knowing the story already is useful. Ex. Percy Jackson, Harry Potter
    • La Sombra Del Viento
  • Shows
    • Casa de Papel
    • While it's probably better to watch shows originally produced in Spanish, I like watching anime with Spanish dub. Spanish sub too (never English) if you're at a lower level, though this turns into reading practice.

French

• Level: Upper beginner. Can understand chapter books and shows
  • I've spent very little time practicing French, but was able to get to a level where I can understand stuff and speak really poorly (but still speak!) quickly because of the Spanish I knew. If you already know a romance language to a decent level, learning another one is essentially twice as easy.
• Resources
  • LingQ
  • Podcasts
    • InnerFrench - start at the beginning for somewhat more clear French. First episode is a great explanation of exposure-based learning. I think the episodes are also on LingQ if you want to read along.
  • Books
    • There are Olly Richard's "Short Stories in X" books for French too
    • Le Petit Prince
    • Currently reading Harry Potter series; knowing the plot is helpful for understanding new words from context.
  • Shows
    • Lupin - amazing. on Netflix. If you have even a basic understanding, you can just put French subtitles on and you'll pick up a lot from context.

Chinese (Mandarin)

• Level: Beginner! Currently reading stories at around an HSK 2 level (300 word vocab)
• Resources
  • I used an Anki deck to sorta memorize the first ~150 words. You can search "chinese" on AnkiWeb; this deck looks good to start.
    • I switched to just reading and listening quickly, but some people continue with Anki
  • DuChinese - essentially like LingQ for Chinese. LingQ has Chinese stories and I started with them, but DuChinese is just better. Same price of ~$15/month.
  • These text adventure games in Chinese look cool, but I haven't gone through them yet.
  • Am starting to practice tones with the help of this video and these audio clips
  • Shows
    • I've started watching Scissor Seven on Netflix with Language Reactor. I'm not quite at the requisite level yet where this is comfortable enough though.