2014-08-19

Grammar is cheating: the brain vs. language learning

In this in-depth post, we’ll take an alternative look at language learning through the lens of an idea borrowed from psychology. This is aimed at both language teachers and people learning a language. We’re going to look at what goes on in the brain when dealing with new languages, and why certain methodologies either fail or succeed because of it.

System 1 and System 2

Daniel Kahneman, in his Nobel Prize-winning work on human judgement and irrationality, Thinking, Fast and Slow, paints a picture of two “systems”, or modes, of thought. It’s a simplification, but one that proves very useful in analysing why humans do the things they do. He refers to them simply as “System 1” and “System 2”.

System 1 is associated with automatic thinking. It’s the thinking that your brain does on autopilot.

System 1 is what makes you become alert and nervous after a loud noise in the house at night. It’s why hearing the word “scientist” makes you think of a white lab coat, and why people have knee-jerk emotional reactions to news stories and Buzzfeed articles. It’s what makes you unthinkingly check Facebook when you sit down at your computer. It’s why you always remember to clean your teeth before going to bed. It’s the reason you suddenly “find yourself” doing things, like singing along to a pop song from your childhood. It’s why you react to things in a certain way, even if you don’t want to.

System 2 is associated with manual thinking. It’s when the pilot takes over and controls the plane. It’s your behaviour monitor and overseer, and it requires mental effort (and actual energy: it functions poorly when your blood sugar is low). It’s what you use when System 1 won’t cut it.

For example, if I ask you, “what’s 2 + 2?”, your brain says “4” without thinking. If you have basic numeracy skills, I can also ask you, “what’s 8 × 7?” and you’ll think “56” immediately. Your brain hasn’t done any real work here; it’s relying on its neural reflexes. But suppose I ask you, “what’s 387 – 49?”. Now, your System 1 turns up a blank answer. You haven’t memorised this one. Time for System 2 to take over. You make a conscious decision to solve the problem, and then you work through it: “49 is 50 – 1… and 387 – 50 is… (8 – 5 is 3)… 337, and now we add the 1 again… 338.”

If you repeated “387 minus 49 is 338” out loud for 30 minutes every morning, that phrase would eventually work its way into your brain and you’d be able to answer the original question as effortlessly as “what’s 2 + 2?”. It would become second nature, and upon hearing the question “what’s 387 – 49?” you wouldn’t have to engage your mathematical faculties (System 2) to answer it. This is how you learned your times tables. Times tables are useful because they save mental effort and time by making the knowledge available to System 1 via repetition and memorisation.

Language works the same way. System 1 is responsible for automatic verbal responses, such as when you hear the phrase “the capital of France” and your brain says, “Paris”, or when someone knocks on the door and you say, “Who is it?”. It’s the reason your native language feels effortless to you. You receive a trigger, and your brain responds for you. Easy.

Quick, finish this sentence: “Dad’s been running around like a chicken with its…”. Whether or not you are capable of giving the answer immediately depends on your System 1. A native speaker who grew up hearing this expression can’t help but think of “head cut off”, while a learner of English would be at a loss to answer it, and might come up with logical guesses using their System 2: “Like a chicken with its rooster? Like a chicken with its friends?”. They might get the answer, but they would have to expend mental effort to do it.

The difference between someone who’s fluent in a second language and someone who stumbles through every sentence is whether their System 1 is readily providing the words they need. If you’ve ever tried to converse with someone in a second language after having not practised it for a long time, you know how frustrating it feels when your brain won’t give you the words you want. It draws a blank, just like when you’re asked about 387 – 49. You’re like the clueless foreigner in the headless chicken scenario.

The two systems in language learning

In the past, foreign language study usually involved memorising lists of verb conjugations and various grammatical “rules” and “exceptions”. Vocabulary was learned separately, in lists, and the student would glue together the individual vocabulary words with the grammatical rules they had memorised. If you studied Latin at school, you know what I’m talking about. Languages are still taught this way in most parts of the world, particularly in East Asia where there has been a heavy emphasis on grammatical drills and exercises with little or no communication practice.

In our terms, this approach relies heavily on System 2 for producing language. You build it in your head, and after squinting and frowning for a few seconds, you say it. System 1 is engaged in this approach, too — for recalling individual words (that have been sufficiently reviewed) and for recalling verb tables and rules that have been hammered into the student’s brain.

I remember discussing German with a friend’s 50-year-old mother who had studied the language at high school for 5 years. She could effortlessly rattle off the singular definite article (“the”) in all its cases: der den dem des die die der der das das dem des. Her System 1 had assimilated it perfectly. Unfortunately, it hadn’t assimilated any actual German as it’s used in context, and so she couldn’t speak German to save herself. This seems to be a very common experience of language classes for most people born before 1980.

Masc.Fem.Neut.
Nom.derdiedas
Acc.dendiedas
Dat.demderdem
Gen.desderdes

In the last few decades, classroom language teaching has undergone a major upheaval in the English teaching world, possibly because people cottoned on to the fact that they were hopeless at German even after studying it for 5 years. Now, the emphasis is on “communicative competence.” Accuracy, which used to be paramount, has effectively been replaced by fluency, and grammatical errors are given minimal attention. The advantages of this approach are that 1) it’s more fun — though hardly more fun than talking to native speakers — and 2) common words and phrases are brought to the front of the mind through constant use and repetition. However, these phrases are often laden with grammatical errors, since, as a student, your only conversation partners are linguistically in the same boat as you, and virtually no time is spent on practising correct, natural-sounding groups of words.

Strangely, despite this focus on communicative competence, grammar is still taught and learned separately, in the form of rules and drills. First, you “do some grammar”, which involves learning a rule like adjective order (e.g. nice, big, red car instead of red, nice, big car — remember, it goes opinion, size, age, shape, colour, origin, material, purpose!). Then you “get some practice in”, which is supposed to involve talking about your favourite car or something in order to practise the grammar rule. What was just learned through System 2’s manual, logical reasoning is now commanded to manifest through System 1’s natural, automatic reflexes. This is as absurd as asking someone to calculate 387 – 49, among a few other math problems, and then expecting them to effortlessly provide the answer later off the top of their head — while focusing on another task, like trying to communicate with someone!

Kahneman and others discovered that System 2 can’t be relied on to control behaviour when it’s distracted: it lets System 1 do that instead. This is why people who are concentrating intensely on something become autopilot zombies:

Imagine that you are asked to retain a list of seven digits for a minute or two. You are told that remembering the digits is your top priority. While your attention is focused on the digits, you are offered a choice between two desserts: a sinful chocolate cake and a virtuous fruit salad. The evidence suggests that you would be more likely to select the tempting chocolate cake when your mind is loaded with digits. System 1 has more influence on behavior when System 2 is busy, and it has a sweet tooth. People who are cognitively busy are also more likely to make selfish choices, use sexist language, and make superficial judgments in social situations.

— Daniel Kahneman, Thinking, Fast and Slow (2011)

This fact does not bode well for someone trying to practise a new grammatical rule while simultaneously trying to communicate in a second language. I think it’s also the reason why students tend to fare better in “real” conversation (where the cognitive task is to communicate) than in “classroom” conversation (where there is a triple cognitive task: to communicate, to use the language the teacher wants me to use, and to fight the urge to check Facebook and Twitter). The busier System 2 is, the more you end up relying on System 1 and its old habits.

Language as sport

I’ve often heard it said, among the online language-learning community, that learning a language is more like learning a sport than an academic subject. This is a good analogy, since improving at a sport mostly involves training physical and mental reflexes (System 1) while academic subjects tend to involve training reasoning skills (System 2).

When you leave a friend’s house with well, I guess I’d better get going, or you hear I’ve heard so much about you and you reply with all good things, I hope!, you’re relying on your System 1 for those pre-packaged expressions, forged into neural pathways through decades of repetition. Most of the everyday language you use consists of these “chunks” of language, mixed in various ways — and so it follows that the greater part of language learning should consist of massive repetition and exposure to these chunks in their natural context. This way, your System 1 will be ready to produce results when you need them.

But the sport analogy isn’t all-encompassing. There is, in fact, a place for System 2 in language use. If there wasn’t, then you’d be able to read Kant like a paperback novel. You wouldn’t need to edit your own writing for clarity. In conversation, you could come up with stuff like “that wasn’t what I was suggesting she told you the reason that she needed you to tell me not to do it was” without effort.

System 2 is what you use from a zoomed-out, top-down perspective to make your words fit together properly. It’s connecting the pre-packaged chunks and making sure they make sense. It’s your attempt to edit your words so that you sound acceptable to others, as in consciously choosing you, Bob and I instead of you, me and Bob even though the latter feels more natural. It’s thinking about and monitoring language rather than merely producing it.

When you learn other languages, System 2 is what you use to get by when your brain isn’t coming up with the goods. Remember what we said earlier: System 2 is what takes over when System 1 won’t cut it. As a non-native speaker of a language, there are inevitably going to be times when your System 1 won’t cut it, because you haven’t built the neural pathways. Your System 2 will try to come up with a best guess, an educated guess.

Helping you make better guesses is essentially what grammar rules do. They’re a way of making up for all those years of exposure you missed out on by giving you some rules of thumb to fake it. Where native speakers have had to put in hours of practice, you use a shortcut. To put it more pithily, grammar is cheating.

Say you’re a Japanese speaker learning English and you’re not sure which is correct: in the airport, on the airport, or at the airport. You haven’t heard the expression at the airport thousands of times? No problem. Just remember that “we usually use at for wide, open spaces like airports, beaches or universities.” (Sure, this doesn’t account for in the field, on the beach, and on campus, but don’t worry about that! Let’s call them “exceptions”. That makes the teacher’s job easier.)

What’s the catch for this fast track to English proficiency? Well, every time you talk about airports, you’re going to have to pause, engage your System 2 and remember that at is for wide, open spaces before you can get the words out. It’s slow, effortful, often wrong, and leaves a bad impression. But at least you didn’t have to do all those years of listening and reading!

What to do about it

The answer isn’t to reject grammar rules out of hand. They can be very useful for writing, and occasionally even during conversation. The problem is when they are relied on as a substitute for training System 1.

You can’t have all butter and no bread. There is simply no avoiding the fact that you have to be exposed to the target language. Not only that, you have to throw the same chunks of language at your brain several times before they will stick.

This can be done in a few different ways:

  • You can read and listen extensively, which will mean the most frequent chunks of the language will be repeated enough for you to absorb them.
  • You can do it intensively, which consists of reading and listening to one piece of text/audio over and over until it’s a part of you and then reviewing it later. I made a webapp to help with this.
  • You can drill the pronunciation of individual chunks, like what I meant was…, to the point where they can be recalled and produced without effort. It’s vital that you have access to the natural pronunciation so you can mimic it, and chunks have to be reviewed or used periodically for a while to maintain them.
  • You can translate samples of the language to your native language and back again, then compare with the original text in order to see what features of the language you haven’t internalised yet.
  • You can talk to advanced/native speakers of the language regularly (tutors and language exchange partners are good for this). For this to be effective, you need attention to detail and some way to record what you learned to review it later, e.g. keeping a notepad handy or recording the conversation.
  • You can use Spaced Repetition Software like Anki, which allows you to review information efficiently. This is strictly a supplementary activity and you need to know what you’re doing for it to be useful. The biggest danger here is you may come to rely on the SRS for language study, which becomes drudgery and can kill your motivation.

Activities like these should form the basis of your language learning. From there, you can practise productive skills (speaking and writing), which will train your fluency further. And you can study as much grammar as you want, because you’ve already internalised the language that you’re analysing. Using grammar rules will no longer be blindly gluing words together, but understanding why the language you already speak works the way it does. Now you’ve got some bread to put the butter on.

In summary: poor language learning involves attempting to use System 2 (grammar rules) to compensate for language that System 1 hasn’t internalised yet. Effective language learning means using System 2 (well-planned study routines) to transfer language to System 1 and make it second nature, enabling effortless communication. Studying grammar then becomes a help rather than a hindrance.