I was coaching TPRS to a couple of French teachers at a local high school yesterday and they were telling me about their department-mandated French exam, which includes 1/3 specific grammar questions (e.g. “which is the right form of conjugating avoir in the 3rd person pluperfect” bla bla). What a load of B.S. but ANYWAY one of them asked me, post me-asking-a-story, why do TPRS teachers minimize grammar instruction?
Well, first, we do teach grammar. We do our pop-ups– “grammar commercials”– and we explain whatever the kids want to know.
More importantly, however, we mimize grammar teaching because it has very little real return. If you want to geek out and read How Languages Are Learned (Lightbrown & Spada), or look at Ortega and Lourde (2000) you can wade through the details.
Basically, it’s like this: There are an unknown– but enormous– number of rules for grammar in any language. Some are simple (e.g. in English you cannot say I enjoy to run, because– for whatever reason– enjoy does not take an infinitive, but rather a noun, gerund etc). Some are complex (e.g. I am a handsome employed professional sounds OK; I am an employed handsome professional does not), and not even me, ueber-language-geek, can explain why. But there are so many that if we spent specific time teaching–as in, demonstrating, modeling, explaining, practicing etc– them, we could spend all our time on them.
We know from forty years of research– and there is no disagreement in the scholarly literature about this– that compelling comrehensible input drives acquisition more than anything else. While almost everyone wants some explanation of why “___ has to happen with ___”–and the research supports some explicit grammar instruction– the question remains, how much grammar teaching is enough?
I’ll give you a simple example: In German, articles change depending on whether they are masculine, feminine or neuter, singular or plural…and…whether they are with a noun that’s the subject, direct object, indirect object or possessive of the sentence. You say “Ein Mann hat eine Frau” (a man has a wife), but “Die Frau hat einen Mann.” That ein changes to einen because Mann in the second sentence is the direct singular masculine object.HOLY CRAP, IS THIS EVER COMPLICATED!
So how do you learn inflection? You have two options:
a) You do what my high-school teacher, Frau Hedda Thatcher, made us do: memorise a massive table of the endings. “Vat iss se datif plural off Hund?” she would ask. “Vat is se plural nominatif masculine form of the woman auf Deutsch?”
b) You do what kids do, which is get input– in students’ case, with restricted vocab– and you let your brain figure it out. If you hear “Der Hund ist gross” (the dog is big) and then “Der Mann hat den Hund,” (the man has the dog) your brain is getting a bit of a pattern– der becomes den with the word Hund after the verb “has.” You don’t know inflection– you can’t explain it, and you will probably make mistakes using der in sentences– but it’s a start.
You will eventually hear something like “Der Mann geht mit dem Hund” (The man walks with the dog) and your brain (unconsciously) will note that “with mit, the word der becomes dem.”
Now, once…whatever. A couple of hundred times? Your brain will– for reasons you won’t be able to articulate or even be aware of– pick the rule up. German kids do it, and so do visitors who study German and hear lots of comprehensible input. Krashen managed it and his German is pretty good. Blaine Ray asks a German story where he says “Das Hotel war in Vancouver,” and then “Das Maedchen war in dem Hotel.” Here, Blaine changes das to dem.
Now here’s what’s interesting. In order to figure the rule out, your brain also has to figure out noun gender. But– and here is the kicker– there is no way to tell what gender a German noun is. There are a few rules– e.g.-ung nouns are feminine– but mostly it’s totally ambiguous. So how does the brain do it?
Well, basically, it cross-correlates a zillion data points. If X in position Y has -en added, then gender likely Z. If X in position Y has -em added, gender likely A. Store hypothesis; apply next time data gets input to test validity; if confirmed, rule is more likely true, etc. The research makes this very clear: language learners, in learning a new language, make many rule-generalisation (and other) mistakes which are not influenced by their native language. The brain, in other words, has its own built-in sorting and predicting “software” which kicks in. (Yes, learners do also make native-tongue-influenced mistakes, but surprisingly few). You can read Bill VanPatten’s simplified description of language acquisition here.
It’s a lot like solving Sudoku. Anyone who does Sudoku knows that every puzzle has an “anchor point:” once you start writing down possibilities in each square, you will find one totally clear, one-option-only blank square, and once you have this, you can start cross-solving.
You solve Sudoku by going through the puzzle and, for each blank square, you write in (in pencil) all the possibilities for that square. You will at some point– in even the hardest puzzle– come across an “anchor point,” that is, a blank square where there is only one possible answer. When you find the anchor point, other possibilities start eliminating themselves. If there is an 8 here, that means there cannot be a 3 there, etc etc. This BTW is why Sudoku– like studying grammar– is boring. There is no real thinking, just a mechanical process of elimination.
With a bit of clarity regarding a few simple vocab items– and they don’t even have to be explained– and a ton of structured input, the brain will sort things out. It will generate and test rule hypotheses, discard some, acquire others, etc. It will notice something like “OK, der Hund goes at start of sentence, den Hund after hat. So der becomes den, and der goes with Hund. As loads more data comes in, the brain will check the new input against its hypotheses about what goes where, and confirm or discard. If the brain then hears Der Mann hat den Kaffee getrunken (the man drank the coffee), it gets a cofirmation about its rule hypothesis, and this gets archived.
Eventually it will calculate the patterns for what goes where, and simply apply them. Now, because we don’t have 4,500 hours like German kids do to listen to Mutti and Vatti, we have to slow down, simplify our vocab, and be more repetitious. But we let students’ brains do basically what kid brains do.
The beauty of it is is that, while it takes forever to explain these rules, and to practice them (which is boring), we don’t need to! (When my classmates finished high-school German– and most of them dropped it after Grade 11, because it was tedious and stupid and not necessary for University admission– they could neither speak nor write extemporaneously. But man, could they ever decline articles and explain pronoun order.)
The reason TPRS teachers don’t over-explain (or make students consciously practice grammar rules) is that there are so many, and they are sooooo complex, that the practicing would take forever. Instead, we provide comprehensible input, lots of reps, answers to grammar questions, and we let the brain’s (literally) trillions of wired-to-acquire neural connections do their work.
The research supports this. While grammar teaching does “work” in the sense that people can learn to consciously remember (and apply, and recognise right/wrong uses of grammar rules), the opportunity cost is super high. Time spent practicing grammar = time not spent getting comprehensible input. Yes, you get “returns” on the grammar-teaching investment, but these are tiny compared to the returns on interesting comprehensible input. To make best use of class time, C.I. is the ticket.
Basically, comprehensible input is two birds for one stone: you are getting your grammar instruction “smuggled in,” so to speak, while students are focusing on the fun stuff– like what happens to the character who wants three boyfriends but is offered seven (and must choose!), what language the horse speaks, etc.