On Ben’s the other day there was a question from someone who asked does explicit grammar teaching “work?” Now you all are probably not the geek that I am– for me a lovely Sunday is either rock climbing or reading about language acquisition– so here’s a summary of some recent reading.
This question also dovetails nicely with my daughter reflecting on her French experience in high-school. She finished French 12 at a prestigious Vancouver mini-school– egg-heads only, s’il vous plait— with 93% and the top French student award. And she couldn’t speak or write extemporaneously. My younger daughter, who had been in immersion from grade 5 to 7, and who was a year ahead in classroom French, dropped French at the end of Gr11 in order to do it online, because, as she said “the only difference between a French class in September and in May is what is on the worksheets.” Her French 11 class was a running joke among the kids: if you were away sick for a day, the answer to the question “what did we do in French yesterday?” was ALWAYS “worksheets.” (She also got 95% in French 11) You know there is a problem when the the top kids can’t write or speak and when they would rather get through French online than in a class. Common denominator: these keen, bright students spent four years in French doing, in their words, “grammar worksheets and stupid projects.” The exception was their Grade Nine French teacher, Polly Dobie, whose unorthodox methods and self-restraint surrounding the Grammar Grind produced, in my oldest’s words, the ability to “feel what sounds right cos she spoke it a lot in class.”
OK so back to the research. The questions “does explicitly teaching grammar work?” and “which method of explicit grammar instruction– inductive or deductive– works better?” have been looked at. The results are thus
Grammar instruction is effective…but only if you want to teach people to consciously manipulate linguistic items. If so, you are better off teaching them explicitly how to do it. Research also shows that, generally, explicit (deductive– learn rule then apply) works better than inductive (see examples then figure out rule from that). Research also shows that these gains last. Woo-hoo! Let’s teach grammar!
OK. But before we go any further, and TPRSers give up the ghost, let’s look at what “works” means. Most grammar research studies basically look at teaching people specific rules about a tiny microset of vocab and grammar. For “works,” they mean that when asked to “do stuff” with this microset, like say sentences, choose options to put in the blank, decide if something is right or wrong, etc. Almost always, they do statistically significantly better than a control group (who get no “treatment”).
I have read one study (I cannot find re– Eric? help!) where researchers taught English speakers how to use the Spanish “gustar” (“to be pleasing”) and they found that, yes, teaching them rules and having them practice and then testing them “worked better” than either not teaching them anything (other than what the Spanish reading/listening passages meant) or asking them to figure the rules out from just hearing input. Basically, every grammar study ever done– even new-wave ones like VanPatten et al‘s processing instruction ones– will show the same thing. Work on Item ___, and people get better at item ___.
So far so good for the grammarians. BUT…and it’s a big but…and as Peter Croft says, “nobody likes having a big but”:
The catch? They only looked at one, very specific, very simple, very structured thing to do: use “gustar.” Now if we are teaching people for any length of time outside of a 5-hour Uni study, and we are going to include lots of systematic grammar stuff, we run into a bunch of problems:
A) boredom. Anyone old enough to remember the ¡Díme! Spanish texts knows what I mean: grammar drills blow. They also don’t work (as VanPatten (2013) notes).
B) the “number of rules” problem. There are a theoretically infinite # of rules in a language (no known language other than computer ones or Esperanto have all the rules mapped out). So if you wanted to practice all the rules it would take forever.
C) the opportunity-cost problem. If all– or some– of your energy goes into grammar drills, you are losing out on other stuff which has demonstrably proven positive effects on acquisition (basically, C.I.). A typical study only looks at gains re: one rule/item. A better design would compare acquisition of ____ (via grammar instruction and practice) with acquisition of other stuff at the same time (via comprehensible input) and of course including a control group.
Did increases in ______ acquisition result in lesser gains elsewhere? If we hammer away at teaching, say, gustar to English speakers, and we thereby lose out on teaching (and therefore acquiring, say, conocer), we are robbing Peter to pay Paul.
D) the question of causes of gains. In a typical study, people get both grammar instruction AND comprehensible input. The instruction is about grammar manipulation but they also know what all the words mean. So…do they get better at ______ because of grammar instruction and practice? Or is the grammar instruction icing on the C.I. cake? This is a consistent problem with studies that look at grammar instruction’s efficacy. The fallacy here is post hoc, ergo propter hoc (“after this, therefore because of this”). In other words, a classic correlation vs causation problem.
(The interesting question raised here is whether or not you could ever really assess the effect of grammar instruction on authentic language without the inevitable boost that comprehensible input gives…since all language is meaningful, and it is absurd to teach people to manipulate language that is not meaningful, how do you separate the effects of C.I. from those of grammar instruction?)
The question, as always, is not “does X work?”, but “how well does X work in comparison with Y?” To put it another way, “even if X works well, is X worth doing?” Blaine Ray puts it another way: “it’s not what you get, but what you could get, that matters.”
There is also a REALLY good paper on Susan Gross’ site about grammar acquisition. This paper shows that it doesn’t really matter what order you teach grammar, because people will not acquire _____ until they are ready for it. This being the case, explicit grammar instruction– other than TPRS-style “pop-ups”– is a waste of time.
Grammar instruction also fails because people– by definition– have either acquired, or not acquired, the grammar rule in question. If they have, instruction is redundant. If they havn’t, it’s because they aren’t ready to acquire it, and so, by definition, they will be unable to acquire– i.e. produce– it (although they will be able to understand it). Grammar instruction is like the worksheets that communicative or grammar teachers love: if you get the material, you don’t need them; if you don’t get the material, they won’t help you.
Krashen has said of grammar instruction that “I’m not opposed and it doesn’t hurt […] But it has extremely limited effectiveness compared to comprehensible input and there are severe contraints on how interesting it can be.”
VanPatten has noted that there are so many processes going on during language acquisition– from many kinds of decoding and processing, to output issues, to context and pragmatics questions– that it is functionally impossible to isolate any one aspect of acquisition (grammar…and anything else).
Essentially, it comes down to this: while explicit grammar instruction can teach people to do really small, specific things better, it fails at the real job of language teaching: to enable people to understand and then use real spoken and written language in all its subtle complexity, in real time. If you want to do that, the research is clear: provide loads of compelling, comprehensible input in oral and written form.