Merrill Swain

Are these 17 statements about language acquisition true? Answers from the research.

Here’s a list of popular assumptions about language learning and teaching from Lightbrown and Spada (2006).  I found this on leaky grammar, which is well worth checking out.  I’m responding to these statements based on research from Stephen Krashen, Wynne Wong, Bill VanPatten, and of course the stuff in Lightbrown and Spada’s 2013 text, which everyone should read.

1. Languages are learned mainly through imitation  

No.  While obviously there is imitation– especially from children– the research is clear: most language learning comes from receiving aural or written comprehensible input.  At much later levels in the L2 acquisition process, some explicit feedback– especially for writing– will help things along.

2. Parents usually correct young children when they make grammatical errors 

Depends.  Some do, some don’t, some sometimes do.  There’s no evidence to show that this practice works.

3. Highly intelligent people are good language learners 

What does “learning” mean?  Research shows that people who traditionally test high on IQ tests (which have well-documented biases) are pretty good at doing things like remembering and consciously applying grammar rules.   However, many illiterate people– who would massively bomb any IQ test– have acquired many languages.

4. The most important predictor of success in second language acquisition is motivation 

No.  While motivation may keep students “tuned in” to instruction– they will listen/read much more (i.e. receive comprehensible input) if they want to learn– all the motivation in the world will not overcome ineffective methods.  A teacher who is providing incomprehensible input, or boring tasks, will eliminate motivation in all but the eggest-headed of students, or, even if those students stay engaged, will be unable to ‘reach’ them.  Krashen has declared motivation less than important.

Related:  motivation to what?  Most students don’t especially care about Spanish, French, etc.  I sure didn’t…what I did care about was being able to meet and talk to people in Mexico, read Paz in the original, order food, meet Colombian women, etc.

5. The earlier a second language is introduced in school programs, the greater the likelihood of success in learning 

Depends.  Canadian data on French Immersion supports this theory.  However, if a language program doesn’t follow brain rules– it provides incomprehensible input, it’s boring, it scares learners, etc– more instruction is not a good idea.  We also know that adults, under the right conditions (good comprehensible input), can massively out-pace Immersion learners and kids in acquisition.  Some estimates are that 1,000 hours of comprehensible input will build functional fluency.  Immersion and early exposure do one thing MUCH better than later exposure:  get rid of accents.

6. Most of the mistakes that second language learners make are due to interference from their first language 

No; only a very few are.  Many are interlanguage.  As Lightbrown and Spada show (in their 2013 4th edition), many errors made by second-language learners do not reflect their native language.  Indeed, studies show that many interlanguage errors are common across cultures and languages.

7. The best way to learn new vocabulary is through reading. 

Yes, provided the reading is comprehensible and interesting enough that learners will do a lot of it.  Some studies suggest that over 75% of a literate adult’s vocabulary comes from reading.

8. It is essential for learners to be able to pronounce all the individual sounds in the second language. 

Essential for what?  To acquire the language?  No.  There are plenty of cases of learning that happen without speaking.  I learned some Mandarin in 1995.  I couldn’t speak it for the life of me, but after working with my Chinese boss for 6 months, I could understand sentences such as “go to the back and grab the cleaning cloth.”  However, eventually, people will want to “roughly get” pronunciation because  saying “I have a big deck” wrong can make you sound like, uhh, well…

9. Once learners know roughly 1000 words and the basic structure of a language, they can easily participate in conversations with native speakers 

Generally.  BUT: conversational success also depends on

  • whether or not the native speaker can slow down and simplify enough for the L2 learner
  • what they are talking about– a native-speaker engineer talking engineering will be incomprehensible to an L2 learner who doesn’t know anything about engineering
  • what “know” means.  “Knowing’ grammar rules and vocab lists alone won’t work– the L2 learner must have had their 1,000 words presented in meaningful, comprehensible ways, over and over.

10. Teachers should present grammatical rules one at a time, and learners should practice examples of each one before going on to another. 

No. Grammatical rules are acquired at the learner’s own rate; some rule acquisition– e.g. use of negation in English–  follows predictable patterns, while some is piecemeal (i.e. appearing to be acquired, disappearing, reappearing etc).  Presenting rules in sequence will also make for much less compelling input (see the ¡Díme! texts for an example).  Best practice:  present a comprehensible and interesting variety of multidimensional language so that whatever learners are ready to acquire is there all the time, as Susan Gross has said. VanPatten says that verb “[t]enses are not acquired as “units,” and the brain doesn’t store grammar as a textbook-stated rule.”

11. Teachers should teach simple language structures before complex ones. 

No.  The classic example is the third-person -s ending in English.  It appears simple, elementary, etc, yet is often very late acquired.  What is a “simple” structure anyway?  In Spanish, children often acquire such supposedly “complex” grammar as the subjunctive before they properly acquire the present tense.  In French and Spanish, there are supposedly “complex” verb tense items–  from the imperfect and preterite tenses– which are much more frequently used than some present tense verbs.

12. Learners’ errors should be corrected as soon as they are made in order to prevent the formation of bad habits. 

No.  There is no evidence to suggest that this works.  Error correction may actually slow acquisition because, let’s face it, it’s not fun to be shown “you’re wrong,” and, as it has been well-documented that happy secure learners = better learners, the self-consciousness that comes with error correction may impede acquisition.

13. Teachers should use materials that expose students to only those language structures they have already been taught 

This is 95% true…but learners can, and do, use metacognitive strategies to figure out new words or grammar, and/or often do so unconsciously.  While it is important that about 95% of language be 100% comprehensible, some “new stuff” is essential to growth.

14. When learners are allowed to interact freely (for example, in group or pair activities), they copy each others’ mistakes.

No.  They often make the same mistakes, but these are generally not from copying each other, but from interlanguage processes.    While language in a communicative classroom is typically impoverished– learners provide other learners with their by-definition low-level output– this does not generally cause errors.  The communicative classroom– which features lots of student talk– is, however, a bad idea, because students are not giving each other quality input.

15. Students learn what they are taught. 

Mostly, depending on the quality of instruction. In a comprehensible input language classroom– done properly— students will acquire most of the vocab and grammar that is presented, and will pick up a lot of things that are not consciously and deliberately presented.  However, if grammar is taught via the rule-then-practice method, or vocab is taught by list memorising, students will acquire much less than “what they are taught.”  Acquisition = how much repetitive, compelling and comprehensible input students get.  Students will not necessarily acquire what we teach when we teach it, as Long (1997) notes.

16. Teachers should respond to students’ errors by correctly rephrasing what they have said rather than by explicitly pointing out the error. 

This will feel better to the students than explicit correction, but there is no evidence it aids acquisition.  My view:  corrected output errors in a T.P.R.S. class are necessary to some extent because they provide better input for other learners.

and finally,

17. Students can learn both language and academic content (for example, science and history) simultaneously in classes where the subject matter is taught in their second language. 

Yes, absolutely, provided the language is 100% comprehensible.

Input – output = acquisition!

I did a workshop yesterday at Simon Fraser University and one of the standard questions came up:

    Can people learn to speak a language without “practising” speaking it?

The answer, as forty years of research and 100,000 years of evolution show, is “yes,” but sometimes stories speak louder than data. So, today, two cool stories about acquisition without output.

First, here’s a great blog entry by Trisha Moller about language acquisition. She writes:

“Recently one of my administrators shared a story with me that illustrates what using comprehensible input and repetition can do.

My administrator was teaching English in Africa to small children before he became a Social Studies teacher here in the States. He taught very young children and used fairy tales to help them to acquire the language. For months he was reading and illustrating these stories. He read The Three Little Pigs, Goldilocks, etc. He saw no indication that they were really understanding and they produced little language.

One day, one of the little boys was misbehaving and he was asked to stand outside the door for a moment as a consequence. It was hot outside and this lad did not want to be there. After the door closed he began to bang at the door and the teacher heard the following: “If you don’t open this door I’ll huff and puff and blow the house down!” My administrator was floored as that was the first English he had heard. It was spot-on for usage and the child showed that he knew just how to use it. It gets better though. Just after this, another young girl came up to the teacher and tapped him on the arm shaking her head no and said “this house is made of bricks.”

So, if you still think that TPRS/CI is not working, do not lose heart. It will take hold. Your students [if they are getting compelling comprehensible input] will acquire whatever language you are trying to teach them.”

The second story concerns my climbing partner Teresa. She’s Brazilian, raised in Brazil. Her Mom is Brazilian and her Dad is Mexican. When they were kids, Teresa and her brother heard Portuguese from Mom (and the rest of Brazil) and Spanish from Dad. Dad spoke Spanish to Mom, and Mom spoke Portuguese to Dad. They understood each other but never formally learned each others’ languages.

When Teresa was 8, her parents split up, and her mother married an American. Her stepfather spoke functional Spanish (to Teresa’s Mom). Her Mom, however, decided that the kids should learn some English, and so it was decided that stepdad would only speak English to the kids. So Teresa and her brother heard English, Spanish from their Dad on weekends, and Portuguese, but spoke only Portuguese. She and her brother also had a steady diet of American movies (variously subtitled into Spanish and Portuguese) and Spanish movies (also subtitled, mostly into English). She and her brother often also turned on the captions for English films as they found it easier to read English dialogue than to understand it in spoken form. They watched classic Disney films and Pixar movies over and over.

During childhood and adolescence, she had loads of input in two foreign languages, but no output: “both my Dads understood Portuguese even though they didn’t speak it much, so I heard a lot of English and Spanish but I never spoke it.”

When she got to high school, Teresa was put into advanced English and Spanish classes. She decided to go to Canada at aged 19 to University. She told me “when I arrived I could understand everything no problem. Speaking was really hard though. But one day about a month after I arrived, I was asked a question in class, and without thinking I answered in English. And after that speaking was no problem.”

Teresa got a job as a Youtube channel manager for Latin America and now uses lots of Spanish in her work. She speaks English (and Spanish) with an accent…but also with basically perfect grammar.


— comprehensible input– and interesting repetition thereof– works
— talking is the result of acquisition, and not the cause. These kids learned without having to produce.
— later-acquirers will have accents…but having an accent does not matter
— no formal grammar instruction is necessary to acquire a language
— there were no expectations placed on Teresa and her brother to speak, write, etc– they just listened and watched
— they were never made self-conscious by way of correction of grammar or accent, or by being forced to speak

Cool, huh?

Do learners who like output do better than their shyer counterparts?

I was at the British Columbia Language Co-ordinators’ Association annual conference on Friday, and attended a second-languages-methods session presided over by Meike Wernicke and Sandra Zappa of U.B.C.

One of the teachers, John, said that “in my experience, the kids who speak more in class do much better on everything  from speaking to listening to writing.”  He then argued for the importance of having lots of output from L2 learners.  I of course had some questions for him.

First, the simple fact that talkers = better L2 students in and of itself doesn’t prove anything.  This is simple correlation and mistakes correlation for causation.  The new moon rises in early evening, when the sun has gone down.  Does that mean the sun’s descent causes the moon to rise?  The term for this fallacy is post hoc, ergo propter hoc (after this, therefore because of this).

Second, if there is a strong correlation, as John seems to think, we must look for a third factor (or more factors) that account for this.  Here are my ideas.

a) Krashen– and lots of other people– have researched the “affective filter” (i.e. whether L2 learners are happy, comfortable, secure etc in their learning environment).  The research from linguistics, sports, math and pretty much everywhere is clear: people who are happy, comfortable and secure (they have a “lowered affective filter”) acquire/learn much more _______ than people who are insecure, scared, uncomfortable etc.  The Yerkes-Dodson Law states that as stress increases (beyond a basic level of “necessary” stress), learning decreases.  There’s a good basic summary here.

In the Army, for recuits, drills with fake ammo, movement only, noise etc occur before people are put into live-ammo drills or combat.  You show them the skills, then, once they are wired in, you can expose them to stress.

In rock climbing, we put people on topropes– zero chance of falls of more than a foot– before we send them up to lead-climb routes (potential for death).  You learn movement, rope handling, gear placement etc, and THEN you do the scary stuff.

In music, we learn music with a teacher, at home, with sheet music or recordings…and much later we play for friends, then parties, then recitals or whatever.

In the US and Canada, one of the non-legal causes for anti-bullying and anti-homophobia initiatives is that schoolchildren under stress from classmates, insensitive teachers, etc, simply do not learn as well as their happier, more secure peers.

Krashen’s expression is that in order to acquire L2, people first need compelling comprehensible input, and that, in addition, they must be “somehow open to input.”

What might be happening in Jesse’s experience is that these “extrovert” kids have their affective filters lowered in class, and so they are “soaking up” more language than their shyer counterparts.  That they are  willing to talk more than others indicates to me that they are happy, secure, etc,  and this positive state will allow them to “receive” more language. 

In other words, they talk because they have acquired (because they are “open to input”), rather than acquiring because they have talked.


b) These outgoing, happy kids are also by definition more engaged with their peers and teacher, and so, in addition to being “open to input,” they will simple get more input, hence greater acquisition.  They will interact more, hear more, and probably be more engaged with homework that involves the target language.

So…do they acquire because they talk, or do they acquire because of other factors?

I must say that I also find the outgoing kids often do better than their shyer conterparts, but I don’t attribute their skill to talking.  Every year I have awesome kids who HATE talking in class.  Getting Hamid, my top second-year student, to talk in class is like pulling teeth.  But he writes beautifully, and in one-on-one conversation, or during stories, he shines.

I could be totally wrong so feel free to argue.  chris(dot)stolz(att)gmail(dot)com




Is Output Useful in the 2nd Language Classroom?

I’ve been arguing with Sarah Cottrell, Kari R. and others about the role of output in a FL classroom.  TPRS is primarily an input-based methodology:  students learn via hearing (and reading) comprehensible input.  So…should students in a TPRS class talk?  And, if so, when?

You can read Krashen’s views here.

My first answer is, talking is not useful– at all– when we start with beginners.  Most people don’t like talking at first, because they quite logically feel self-conscious because they know their accent is “wrong” or “off,” and because they know that they sound like a two-year-old.  There’s lots of research that suggests that talking in TL tops the list of “things I don’t like” in FL classrooms.

If Krashen is right about the affective filter– that people need to be happy and comfy (he says “open to input”) to learn a FL– then talking (for beginners) is totally the wrong strategy.

Output (from most learners) is often flawed, which means, bad input.  Ask any Spanish teacher what happens when you first teach “gustar” (“to be pleasing” or “to like”).  When you ask “te gusta ver la tele?”  the kids answer with “Si, te gusta ver la tele.” (Do you like watching TV? Yes, you like watching TV).  This is a classic beginner mistake.  Now, we understand (and so does the kid doing the listening)…but what they need to hear– “Si, me gusta…”– they don’t.  Why would we encourage poor modeling?

Bill VanPatten (2003) argues that language acquisition happens only when a learner is processing input. “Output,” he writes, “is not a substitute for input, which must come from other speakers.”

We might say, “poor modeling, fine, the kids will pick it up eventually.”  Sure…but that’s a waste of time; why dither?  If we can provide quality input via stories and reading, acquisition is much faster.

Another problem with output exercises– the “communication gap” activities that communicative teachers use, where kids are supposed to use (and thereby acquire) the TL to acquire information they want– is that they’re boring, and they can be bypassed via L1 use.

If you are teaching “to like,” and you tell the kids “OK, look at the list of things in your book on P. ___, and ask your partner what they like,” a lot of them will just point and say “do you like ___?” in English rather than saying “Aimes-tu les chats?”  Even if they do use TL, they’ll only do it once or twice, cos, let’s face it, this is a boring activity.  Which slows the acquisition even more.  Why would you use the target language if you don’t have to?  And everyone knows it feels fake using a non-native language with other learners.Plus, this turns the teacher into a police officer dutifully patrolling the class for TL, hounding the bad English away.  Fair enough…but don’t we have better things to do?

So if early output = bad modeling and slow acquisition, is there ever a place for output?  And we are talking output other than yes/no and one-word answers, or scripted story dialogue.  I think there is…but under some conditions.

a) Output must be perfect.  If a kid says it in class, and there’s any errors at all, it has to be immediately re-cast by teacher into perfection, and then circled.

b) Output must only come when students want to do it.  It must emerge organically.  Forced output is not language– it is drama, recitation, what VanPatten calls “language-like behaviour,” but it’s not language.  Language is what you say and understand without having to think.

With beginners, this is fairly easy: we allow only super-simple perfect output from the willing initially.  With my last batch of beginners, the first sentence they learned was Rochelle juega futbol con David Beckham en Los Angeles.  I circled that (and a parallel sentence: Breleigh baila en Cork con Seamus Ennis.).

I then wrote “juego = I play” on board and asked Rochelle “Juegas futbol con David Beckham?” and she was able to answer by reading off board– I pointed to “juego” and then “futbol con David Beckham.”  I then did a pop-up: “what do the “a” and “o” on “jueg-” mean?”  Note that I could do this because it was obvious from the first minute that Rochelle was an outgoing kid who was eager to talk (as were her friends Jasmin and Rasna, and a boy named Fahim…but they were the only ones).  Breleigh however wanted nothing to do with talking– she was OK answering questions chorally and in English– but she did not want to say anything in Spanish.  Fair enough.  After a few weeks, she started to want to answer in Spanish.

So I set it up so that the only thing they could say was meaningful to them, and perfect– they simply had to read off the board.  If they wanted vocab specific to them, I wrote it on board w/ translation.

Sarah Cotrell said to me on Twitter “My students don’t want to have language; they want to use language.”  Great…but how does quality input come from learners?  Your 4th year kids probably won’t make errors with “te gusta?” which they learned in first year…but they’ll inevitably make interlanguage etc errors with the subjunctive or whatever they have recently started seeing.

I would do some or all of the following with upper-level kids

a) limited discussion (with teacher recasts) about texts or images

b) assignments where students have to interact with native speakers and document the results (e.g. go find a Spanish speaker, and interview them, and record questions and answers with your phone or camera).

c) encourage kids to go and do stuff with the language outside of class.

Some teachers say “don’t kids in TPRS classes get bored just listening?” and I respond with “not if what they are listening to [and reading] is interesting.”  We personalise stories, we do simple Q&A about kid interests, we weave kids into stories, and we do readings about teen characters with real problems that kids care about, and we use our sense of humour above all.  We use stories because they are the most universal and oldest and most compelling form of packaging communication, period.  We always want to find out what happens in a story!

Merrill Swain often comes up here.  Swain essentially says two things: that output is important in acquisition because it “provokes” comprehensible input, and that output can make speakers aware of errors and problems, and in their desire to “fix” these problems they will acquire some language.  There are a few problems with Swain’s ideas.

Say I am in France and I am hungry, and I walk into a boulangerie.  If I stammer and point to a loaf of breadand say “Je veux acheter un…un…un…”, the boulangeur is probably going to say “baguette.”  A classic communication-gap scenario.  And I used metaliguistic strategies:  I pointed, I said “uhh” (or “eu”) over and over, etc.

Great.  I provoked output from him that became comprehensible input for me.  But there’s two problems.

a)  how am I going to remember “une baguette”?  Evidence suggests that I am going to need to hear (or read) it 20-50 times to get it hammered into my long-term memory, and more to be able to spit it out without thinking.  How are communication gaps going to do that?  This might work in an immersion environment– if the situation comes up every day, I’ll eventually pick up baguette— but we don’t have that kind of time in class.

b)  when I get input from the baker, I am getting perfect native-speaker French.   This is not what happens in a classroom, even one full of motivated, experienced and attentive students.   No matter how good the activities etc, students are still getting impoverished, error-filled input from other learners.

Swain isn’t wrong…but her theory, properly speaking, addresses learning conditions, not actual acquisition.  If I want to acquire “baguette,” I need to hear it and read it over and over– and be focused on it– otherwise my encounter with Monsieur Boulangeur will be a one-off that will find me scratching my head next time I want bread.

The second part of Swain’s theory– that output will increase the speaker’s self-awareness of problems in their grammar, vocab and pronunciation, etc, and that this will lead to acquisition– is also problematic.

Let’s say I am in India and I want some water in a restaurant.  I ask my server kanna he? and get a puzzled response.  He nods and brings me the menu.   But I don’t want a menu– I want water.  I must have screwed up.  But how did I screw up?  Did I get the word for “water” wrong?  Is my pronunciation off?  Did the guy not hear me?  Does he not speak Hindi?  (Perhaps he is a Malayalam or Bengali speaker).  As it turns out, I have the word for “water” wrong– it’s not kanna but rather panni.

First, it’s not clear what the problem is.  I know there’s a problem…but what is it?  As an adult language learner, with developed metacognitive skills, I can figure it out: wrong word.  Could a 15 year old?  A tourist with little interest in Hindi?  What if the problem is more complex, like I get a verb tense wrong, or I forget to add a crucial postposition.  Concrete nouns are easy to figure out– point and ask with raised eyebrows– how about complex grammar?

Second, even if I figure out that I got the word wrong– and I managed to ‘rescue’ the conversation by pointing to a water fountain or bottle– I will probably, as in the  boulangeur example above, probably get one mention of the word panni and that’s it.  This might work in an immersion environment, but in a classroom this is not a viable strategy.

This, to me, is the essence of the problem of language teaching: how do we provide  comprehensible input that is compelling enough that it can be repetitive enough that students can acquire it?  Of everything I’ve seen and tried– natural approach, T.P.R.S., Grammar Grind, communicative-experiential, audiolingual, “eclectic”– T.P.R.S. is the best solution to the problem, outside of an immersion environment, because it allows us to provide masses of comprehensible input that is personalised and interesting (via the personal and/or weird details, and because we use stories).