Foreign Language Maintainence

the context

I’ve been doing a lot more Chinese study lately than in recent years. Overall I’m a rather dissatisfied with my level. The language was my second major in college—got to do a whole study abroad and everything during my senior year. Admittedly though, I didn’t put much more effort into it than required. (I did love my asian studies classes that I took for the major, though). At the absolute peak of my Chinese study (December 2019) I might have passed the HSK4 test.

For those not in the know, the HSK is a Mandarin language curriculum administered by the PRC. Most Mandarin courses, including the ones I’ve taken, are loosely structured around it. They’ve updated the test since, but when I was in college, the level 4 test was roughly equivalent to an A2 fluency test according to the Common European Framework. For those that go outside, that’s about the language skill level where you can have conversations about rehersed topics with someone talking to you like a toddler. Youtube “polyglot” level. Not particularly impressive, the same amount of classroom time and much less immersion nearly carried me to a B2 (general conversation proficiency) in Spanish back in high school. Oh well, that was the speed of the course and I did get A’s in my classes, I just didn’t really go above and beyond.

Regarding the HSK, the vocab list for each level is about twice as large as the prior one:

Note that my 7th semester was a study abroad in Shanghai with 16h/wk classroom immersion

This pace really isn’t indicative of the experience someone who is actively learning the language will have. Sure, I double-majored with Chinese and did well, but my school wasn’t really offering language majors as part of an intensive translation program, rather they were more geared towards supplementary education like how I engaged with them. Chinese was the only class i had every semester, sure, but I always had a full course load with 3 more. All to say, I think I can move a lot faster when I’m not focusing much on learning anything else.

the situation

I’ve been on my latest studying binge for a little over a month now. I’ve mostly been using Skritter, which is a spaced-repitition (SRS) flashcard app like Anki, but with a writing input box to test character memory and stroke order. The essence of SRS is an algorithm that presents you with cards you get wrong more frequently, and cards you get correct less frequently. Let’s say I learn 10 new words and add them to my deck today. On my first review, I get 8 cards right and 2 cards wrong. All of those cards will be returned to my review queue, but the ones I got wrong will be placed closer to the top. I’ll probably see those 2 I got wrong again tomorrow, but I might see the cards I got wrong again in 2 days.

This result is reinforced by further reviews: if I get one of my problem cards wrong again tomorrow, I’ll see it again the next day until I get it right. But, if I get one of my strong cards correct the next time I see it, that card will be moved even further back in my review queue. Eventually, you get to the point where you are being consistently quizzed on all of your vocabulary, but you only see your strong words once every few months. If at any point you get a word wrong, it will kick that card back into your immediate “problem” queue so you can re-learn it.

SRS is a great way to consume flashcards and introduce vocabulary to a study rotation when you start from scratch and are only adding 10ish words per day, but when you haven’t used one in a while, it’s kinda brutal. In that case, you’re trying to work your way through an established backlog of vocabulary (in my case around 1200 word), which presents an issue. In order to really use an SRS app, you need to clear your review queue every day. If you don’t it’ll be part of your queue for tomorrow. If I add a batch of cards today, even if I get all of them right, I’m going to see all of them again this week.

This isn’t so bad in a language like Spanish where the writing is phonetic, this card has all the information I need to learn a Spanish word:

Maestro(a) » Teacher

For a less-phonetic language like English, you might include a phonetic transcription on the definition side, or break it out into two cards:

1. Spelling // Definition

Teacher » Maestro

2. Spelling // Pronunciation

Teacher » ti-chәr

Mandarin has some added complications. The writing system is morphemic, so you need to learn how to write each morpheme individually. There’s a lot of character re-use once you get into more advanced vocab, but for the first 1500 common words you usually have to learn at least one new character per word. Therefore, you need a card to test your ability to write the characters in that word. Those characters aren’t phonetic (not reliably, at least), so you’ll need a card to test you know how to read them properly. Mandarin is tonal, so you need to test that you’re getting those tones right too. And, of course, you need a definition card. So, to fully learn a word in Mandarin you need four flashcards:

1. Writing // Definition

老师 » Teacher

2. Writing // Reading

老师 » laoshi

(Pinyin phonetic transcription)

3. Writing // Tones

老师 » 3 1

(Skritter has you draw the tone contour as an answer which is fun)

4. Definition // Writing

Teacher » 老师

(Ideally, you input this answer by actually writing it)

Four cards per word.

If I add ten words, that’s forty new cards in my deck.

Forty new cards that I’m going to see again multiple times in the coming weeks even if I get them all right.

If I’m adding words every day, you can see how the size of my daily review stack can massively outpace the vocab I learn.

HSK4 is about where the “tolerable input” threshold starts, so in theory all I really need to get to that threshold is polish up vocab (that I’ve already learned) to that level. Tolerable input is how I like to call language material that is actually interesting to consume. One of the worst things about language learning is that initial bit where you’re limping through heavily simplified stories and very explicit practice material. Once you get to a certain point though, you can start reading and listening to more naturalistic stuff like tv shows. Usually still practice material, but at least it can be complex enough to be engaging in its own right. Youtube is a great place to get tolerable input. With Chinese I still heavily rely on mandarin subtitles, but I’m getting better. For reading I like to use Lingq.

the plan

I’d love to be at hsk5 (2.0, not the new one) by the end of 2025. I think that might be doable. it’s only double the vocabulary I’ve yet acquired through 6 semesters of study at home and one abroad. hsk5 is generally considered to be the lowest level of working proficiency, though. I’d love to have that.