An alternative take on Krashen

[Cainntear]

The YouTube algorithm recommended this video to me for LanguageJones, a guy with a PhD in linguistics. I didn't go into it expecting to hear anything new to me, but I actually did.

There was a lot of stuff that I did already know, but there was something starting about 10:20 that I think might point towards a very interesting possible truth.

I'll put my notes below, and if you want to watch with unpolluted opinions, you can come back to them later.


...

...

...


So first up, I get what he's saying about laypeople failing to understand the scientific method, but I don't see his description of hypotheses and frameworks really being relevant to this.

He describes the comprehensible input hypothesis as being one hypothesis built on various other hypotheses, and failing to deal with those other hypotheses is the flaw of most of the internet polyglot crowd. That's all well and good, but as far as I can see, Krashen's scientific appearance is little more than pseudoscience. The language education field has had a massive problem of teachers just getting away with saying whatever they think is true and giving their credentials (i.e. the fallacy of "appeal to authority"). Talking about the order he published his hypotheses as though they rely on each other just seems misguided to me -- it seems like Dr Jones has fallen for the appeal to authority. Krashen was not really an academic researcher, he simply taught languages in a university.

Also, Dr Jones refers to the universal order of acquisition stuff as though it's true; he makes no reference to the fact that Krashen's publication about universal order was based on the order students learned/acquired stuff in his own department, so there's nothing "universal" about it, and as far as I know the alternative causal hypothesis wasn't discussed: that the way they taught that particular feature wasn't effective.

Hell, if you looked at my high school French class, you could claim that the universal order is to acquire singular conjugations first and plural ones later, because that's what happened. But that wouldn't account for the fact that we would be given many more opportunities to practice singular conjugations than plural ones. This might have been deliberate, because the teachers may have identified singular as easier based on previous pupils' difficulties, but it does become a self-fulfilling prophecy.

I should also add that Jones claims Krashen was talking about the same order of acquisition as infants, but he really wasn't -- unless I'm remembering wrong he noted a different order then L1 acquisition, but dismissed it with a bit of handwavery to assure readers that the differences were natural and to be expected.


But on to the 10:20 bit, cos that introduced something new that I'm not convinced about, but which is very close to ideas I already held as true.

Dr Jones recounts something that he heard Krashen say in an interview which strongly suggests that Krashen has ADD or ADHD. He suggested that Krashen's hypotheses might in fact be what works as a method for learners with ADHD. He contrasted this with traditional academic methods which are popular with traditional academics, noting that academia has always been heavily populated by people on the autistic spectrum, and that autistic people generally love rules, right and wrong, etc.

Now I've always said that apparent learner differences are more about what blocks learning than what causes it. I've long held that conjugation tables and long, detailed grammar rules are things that don't work well, but they are appealing to learners on the autistic spectrum, so they're less likely to notice that they don't work well and therefore stick at them; so that much I agree with. But when Krashen learned the languages he went on to teach, he must have used the traditional techniques, right? So why should we assume that the techniques he used to teach would work for people like him...?
That said, I've said before that people who do well in classes without explicit grammar tend to work out the rules for themselves, and people who don't work out the rules tend to not really "acquire" a good model as they keep introducing errors.
So Dr Jones's line about Krashen's techniques kind of makes a lot of logical sense to me: people who are prone to looking for patterns/rules are going to find the rules that the instructor is hoping they will.
But then there's the issue that this isn't really an ADHD thing, because people who are on the autistic spectrum are going to spot patterns easily.

But the idea that Dr Jones pushes towards is that these techniques are really working at or towards fairly extreme ends of neurodiverse spectrums. The natural consequence of that is never said and I'm not sure if he's leaving it "unsaid", or if he hasn't really thought that far; but surely the real call should be for moderation...? People who can see patterns for themselves don't need to be told they're there and people who love logically complete rules will love grammar books, but the population is not a binary split between people-who-spot-patterns and people-who-like-written-rules. As far as I can see, the key common feature here is "rules", so we should be investigating how average, normal people get the rules best.

[Lawyer&Mom]

As an Autistic person who loves mass-input I feel misrepresented. Yes, we like rules and charts, but Autistic people are known to often be whole-to-parts learners. It makes sense that we could do well with understanding the whole sentence before we learn its underlying grammar structures. I think the real reason schools love teaching explicit grammar isn’t because everyone in academia is Autistic, but because it’s a lot easier to test explicit grammar. Schools feel the need to show quantifiable progress, and the whole watch 500 hours of Spanish and then get back to me approach makes it hard to demonstrate incremental learning. Even if you would learn a lot more!

[Cainntear]

As an Autistic person who loves mass-input I feel misrepresented. Yes, we like rules and charts, but Autistic people are known to often be whole-to-parts learners. It makes sense that we could do well with understanding the whole sentence before we learn its underlying grammar structures.

Yeah, I'm similar... as I was trying to say (but wasn't clear enough), I don't think Dr Jones's explanation really fits, because people on the autistic spectrum are often better at identifying a pattern from a lot of data than neurotypicals, and the difference within the autistic spectrum can seem pretty huge anyway.

In fact, I think you've kind of pushed my thinking a bit further on.
People on the spectrum love rules, people not on the spectrum prefer social conventions. I think I've seen people use this as justification for why comprehensible input is better for neurotypicals.

But that's assuming that Krashen's explanation is right, and that people are avoiding rules and "acquiring" language -- presumably based on social convention. If people who do well in Krashenesque classrooms are still working with rules, just rules they've worked out for themselves, (and my experience is that people who do well with input are very likely to be able to answer questions about the rules) that means that there's at best no evidence to say it works any better for average people.

I think the real reason schools love teaching explicit grammar isn’t because everyone in academia is Autistic, but because it’s a lot easier to test explicit grammar. Schools feel the need to show quantifiable progress, and the whole watch 500 hours of Spanish and then get back to me approach makes it hard to demonstrate incremental learning. Even if you would learn a lot more!

I don't think it's necessarily worth trying to establish a "real reason", because quite often the truth lies in terms of "contributory factors".

On the simplest level, the biggest obstacle to change of practices is "well I did it that way, and I'm successful, so that must be the right way" and that alone is a very real reason: the majority of teachers were shown conjugation tables and explicit rules, so the majority of teachers do the same.

But then we've got to add in selection bias: school teachers in most countries need degrees in the subjects they teach, so the only people who become teachers are people who the methodology works for.

This seems straightforward, right? Even without any diagnosis, we're effectively talking neurodiversity -- hell, even if the neurodiversity isn't on any diagnosable category.

But with universities being widely recognised as attracting autistic individuals (and Dr Jones claiming that this is particularly pronounced within linguistics, a sphere he has personal experience of), it seems reasonable to conclude that the bias is towards autistic traits.

Hell, one of the main reasons I personally got big into languages was the fact that I wanted to learn how to talk to people. One of the huge things that I got out of it was the linguistics course impressing upon my the complexity and subtlety of the grammatical rules interacting, and the notion of pragmatics over grammaticality -- a pattern may seem to be logical based on the rules, but if no-one says it, no-one says it... and that's in many cases because there's never any reason to say it.
(Recent example: the discussion of didn't use to and native speakers being uncomfortable with "use" rather than "used".)

So I'm basically aware that things working for me isn't proof it works for everyone, and while other teachers would double down on "it worked for me", I'd be constantly looking for what worked for the majority of my students.

[Iversen]

Cainntear and I have several times discussed language learning styles, and as far as I remember he was rather dismssive of them. Are they now sneakily coming back under the cover of a reference to autism versus ADHD?

[Cainntear]

Cainntear and I have several times discussed language learning styles, and as far as I remember he was rather dismssive of them.

Indeed, and there's very little research supporting the notion. I've always held the believe that the real difference in learners is simply their relative ability to learn despite particular suboptimal teaching approaches.

Are they now sneakily coming back under the cover of a reference to autism versus ADHD?

Well there was always a tendency for claims of their existence using extremes of neurodiversity as justification. I always felt that was a bit like pointing at professional athletes and claiming them as the justification for different coaching methods, but professional athletes are extreme outliers, so not necessarily good models to follow.

At the risk of acting like a stuck record, I think one of the greatest strengths of MT's approach is that he had a clear grammar focus and taught it in a way that didn't force people to remember long rules word-for-word (which only some people like doing and can do successfully) and didn't force people to trying to work the rules out for themselves (which only some people like doing and can do successfully). I prefer being given rules but can still work them out for myself... but MT taught me that I could still learn quicker without following either of those approaches.

[Le Baron]

Krashen was not really an academic researcher, he simply taught languages in a university.

From where did you ascertain this? As far as I know he was a faculty member of linguistics departments doing linguistic research. I've never heard of him merely teaching languages. He can barely speak the two languages I've seen him mention (Spanish and French). I don't think he is a languages teacher.

[orlandohill]

The repeated attacks against Krashen could really do with some accurate citations, otherwise it's just a form of harassment.

it seems like Dr Jones has fallen for the appeal to authority.

Where and when did Krashen make the claim that his arguments were valid due to his position of authority, rather than the merit of the arguments themselves?

[cpnlsn88]

Without going into detail I wasn't that impressed by this video.

Krashen's essential work appeared in the 1970s or at least a decent while ago. I think that broadly most of the theoretical framework is correcct and robust. If he's been discredited by research then we can just point to that research and be done. It wouldn't be a matter of opinion, it would be settled. Most language learning approaches don't generate or base themselves on hypothese that can be evaluated and confirmed or rejected.

Science can only move forwards by taking the starting point as that established by research and move on to clarify from there.

An approach can be clarified with an improved application. I am not saying that Krashen's approach is correct in toto - no, research continues but it should be a research that can be based on the currently attained work.

Krashen had his basic hypotheses but developed further hyptheses that were less well operationalised but still important and worthy of exploration.

For instance the conduit hypthesis and the pleasure hypothesis. These are interesting hypotheses and there is one outstanding fact (I think) that the biggest threat to language learning is drop out either because it's too hard or not enjoyable enough.

Pick up a New Testament Greek textbook to see how painful a grammar and rules based approach can be (likewise Latin). Now that's not a living language and I know the issues are different, so the example is a little extreme. Anyway for many learners they aren't enjoying language learning enough to continue it, or it's just too hard. Old fashioned approaches led to people doing languages at school but unable to use the language after their exam and eventually giving up and now (in the UK) fewer and fewer people take up foreign languages at school.

So, to summarise, Krashen is, I think, a starting point. Unless there's research out there that hasn't been drawn to my attention (possible) I think his basic points hold.

By the way his initial work was inspired by Chomskyian linguistics and, yes, they were very fashionable but now in less favour. Krashen hadn't much direct conact with him but was strongly influenced by his work (I think at some point he had read all of his published work and understood it) and has recently done some webinars with him which again make clear his warm appreciation for Chomsky's work. In the end I don't think you need to subscribe to Chomsky's tenets in order to find that Krashen's hypotheses have empirical support.

I should say I always quibbled with i+1. I don't really think it matters because the extensive reading approach has differently quantified this in reading to 98-99% of a written text. Of course reading at an earlier level can be made comprehensible by a teacher doing actions, pictures etc

[Granrey]

For what I have seen recently. Krashen has modified his theory a bit by adding the word "interesting". Meaning the input not only has to be comprehensible but being interesting helps a lot as well.

To me his theory is very very obvious and in reality is not as helpful as people make it to be.

[emk]

Very little research on language learning is especially rigorous. A lot of studies involve, say, 6 students who do 1-3 study sessions over the course of 30 days. This is a consequence of how much it would cost to recruit and pay a larger number of people to study through A2 or higher. Relatively few of the commonly-reported studies even cover the equivalent of the first 30 days of Assimil.

So I think the next best step is to look at the opinions of people who succeed in teaching or learning languages. For example, FSI instructors, teachers who help run 5-10 year language programs, successful hobbyist learners, etc. Or even ask people in the Peace Corps, who frequently need to learn languages from scratch with no courses.

To make an analogy, there are few scientific studies about the best way to make wooden cabinets. But if you find a carpenter that makes good ones, they could probably offer reasonable advice.

So, getting back to Krashen. I think the input hypothesis captures a really important idea: doing lots of reading and listening will dramatically improve your language skills, even if your studying is half-baked. And there's no other way to do explain how people actually acquire real languages, because grammar is big. Much bigger than anyone ever gets taught. There are weird patterns in English that I use automatically, and that I've never thought about at all.

But I also think Krashen is flatly wrong, in a very important way: no amount of input will actually suffice for many people to produce output. There are plenty of heritage learners (and early AJATT students) with excellent listening and reading skills, but who would struggle through an A2 conversation. I know someone who happily watches gritty cop dramas in her heritage language, but who doesn't even have A2 output skills.

I should say I always quibbled with i+1. I don't really think it matters because the extensive reading approach has differently quantified this in reading to 98-99% of a written text. Of course reading at an earlier level can be made comprehensible by a teacher doing actions, pictures etc

The real limit of extensive reading is how many unknown words and incomprehensible sentences a student will tolerate. The second book I read in French, I understood maybe 60% of sentences well, could guess at another 30%, and honestly had little idea what was happening in the remaining 10% of sentences. But not everyone will enjoy this, and for many people, 98% is a good number.