The 20/80 Experience: Initial data and analysis

input banner

The 20/80 Experience: part 1 – background

Last post, we introduced the idea of input/output as described by Krashen and Swain; and their importance or relevance in the second language classroom.  Here, we are going to look at the first batch of data I have collected in my classroom and have a little exploratory discussion.

First however, let’s talk shop.


I should first note that I do not have access to materials that would make this all easier.  To collect the data, I used a simply voice recorder on my cell phone.  I then listened to the class over again (a useful exercise, regardless of whatever happens with this project) and counted the amount of class time spent between meaningful input and output.  I also included the amount of class time done in Korean (though, not separated into input/output).

This is very time-intensive.  Classes are 45 minutes long, and it takes at least one-listen-through to estimate the time, and if I’m feeling thorough, I listen and record again.

In addition, because of how time-intensive this kind of data collection is; I only have 8, 45-minute classes in this initial exploratory dataset.  As we keep going, the amount of data I will have will increase.


Let’s start with the goal in mind, how much input/output is ideal for the ESL classroom.  Is the oft-cited 20/80 the ultimate goal? How do specific teaching strategies affect the input/output ratios of the classroom?  Let’s start with the most general numbers.

overall input

Honestly, I was surprised by these first numbers.  I expected much higher levels of Korean and lower Output.  But as it stands, the initial numbers are really quite harmonious.  English input has the largest share of classtime, but by no means has a majority.  If we were looking to move to a “20/80” classroom, moving from 37% input to 20% is the least difficult task.

The amount of Korean in the classroom is a bit of an opaque issue for me.  I am not sure at all if this is a good thing or a bad thing; something that needs to be raise, or lowered.  Much of my co-teachers teaching is done in Korean, and much of the responses from students to his teaching is in Korean as well.  This is not necessarily a “problem” that needs to be fixed.  34% of all class-time might suggest he is not using “TEE” enough (Teaching English in English), but I don’t know for sure.

As you can imagine, the amount of Korean input/output changes drastically depending on who is leading the class.  For the Korean English Teacher’s (KET) class, the percentages look like this:

KET percentage

64% of class time is quite a bit more than the overall statistic.  But, again, whether or not this is a problem is unclear.  Undoubtedly, in the KET-led class, TEE is not a priority.  Contrary to whatever the Korean Education Administration would led you to believe.  Actual academic studies suggest that the average KET’s ability to utilize TEE is affected by external factors, including school culture, social hierarchies and covert shaming.

Anecdotally, when the KET in our classroom attempts TEE, the students often express exasperation or ridicule his attempt.  They feel very uncomfortable for whatever reason, communicating with him in anything other than Korean.  This is not a reason to abandon TEE, but it is a hurdle that must be cleared if TEE is to be used.

As concerns the main focus, input/output, the KET actually has a fairly good ratio (though, of course, far below the 20/80.)  This will change when we look class-by-class below, but I speculate that the reason the percentages are so close, is that the KET utilizes rote memorization and repetition extensively.  “Repeat after me” drills are the main technique for output, though certainly not the only one.  Repetition will give a fairly even 1:1 ratio.  For the Native English Teacher, the graph is certainly very different.

NET input

In regards to the input/output, we can see very similar amounts of NET English Input and KET Korean.  One reason the NET input is lower, may be a result that the NET must struggle to negotiate meaning with the students.  The NET is in no position to simply ramble on about English.  58% however, is probably still indicative that the NET is speaking too much, and that a good portion of that input is not comprehensible to the students (and therefore, the students aren’t learning efficiently) .

The NET does elicit more student-output.  Though the quality of the output is not necessarily the right kind. The class is directed mostly through a textbook, so the type of input/output is fairly scripted.  The activities involved rarely get the students to the point of “free expression” and have many activities of rote memorization and repetition.  The NET isn’t exactly thrilled with that, but he just does what he is told.  The students do produce a certain amount of impromtu output that would qualify under Swain’s definition.  How much? Unclear just yet.

side-by-side NETKET

Class-by-class however, there seems to be random variation in the amount of input/ouput or Korean on any given day.  The data limits any sort of conclusion just yet.  But these disparate numbers with the KET using Korean far more often, and the NET using English input closer to a 40/60 ratio are a little skewed.


The downward trend of Output surprised me initially and certainly goes against my hypothesis that as time goes on, we should expect an increase in student output, simply due to the fact that the observer is also a player in the game.  We can see, however, that the overall percentage of of input for the NET is skewed by one class period where 80% of the class was pure input.  A very boring class indeed.

Whether or not the trend will only widen, contract or fluctuate  we shall have to wait and see.  My initial guess is that this has as much to do with what class is being taught, where we are in the textbook, what day of the week it is, what time of day it is, and a few other relevant factors, as it does with anything do.


Likewise, the KET’s overall percentages are heavily influenced by one particular lesson where he spent 93% of classtime speaking and being spoken to, in Korean.  (There is a lot to be said about an ESL class that spends only 4% of class time giving English input and 0.6% English output, but we’ll avoid it for now).

Interesting is the relationship of input to output.  The KET managed to have slightly higher output than input for a couple of classes, and as I said before, this is in part due to repetition drills, but that would only explain a very close 1:1 ratio.  There must be additional reasons, which could include student free response, or student-directed discussions.  Something worth examining a little closer in the future.


Of course, we don’t have much to conclude just yet, but we have a better idea of what we should be looking at.  The most opaque areas have to do with why any given class can fluctuates so widely between input/output, if this is a problem, whether or not the teachers have the ability to control it more closely (or if the problem is external to the teacher and their control).

What can be fairly assumed right now, is that that the students are experiencing very different English classes between the NET and the KET.  And I don’t just mean the obvious “one’s an American and one’s Korean” cultural difference.  It seems that very different theoretical foundations separate the co-teachers.  The question remains of how this impacts student learning.  Does the back-and-forth disrupt students routine? Or is the effect minimal?

Those questions are beyond the scope of what we are looking at here, but I still wonder if the NET-led and KET-led back-and-forth model is actually disruptive, and would it be better to either fully integrate each teacher in every lesson (requiring far more planning than the KET has time for, in this situation), or instead making the back and forth on the scale of weeks instead of days.  Maybe there is no effect at all.


The 20/80 Experience: A look at English Input/Output at one Korean Middle School


I’ve heard many times (though, interestingly, never in my university theoretical classes) that the ideal amount of teacher input to student output is 20% to 80%.  This magic number actually does not come out of SLA theory, nor educational theory (but has since been adopted).

But, what I have learned about SLA theory over the years doesn’t always seem to jive with the 20/80 rule (as we will explore below).  On the other hand, the 20/80 rule speaks in some ways to me as an educator. (I’m not one for lecturing).  Therefore, I have decided to explore the idea further.  What will follow is a series of posts, over the course of this semester (at least) documenting my class’s input and output; along with any discussion and analysis we would like to have.

This post will serve as an introduction to the actual data and analysis that will follow (with the first set coming tomorrow).  Here, I just want to set the stage.  We will give just a basic overview of what it is I am actually looking at when we say, “input/output”.  and over the course of the study, it should be made more clear and we may even delve into more specific topics.

But first, on the basic level, what we are talking about when we say “input” and “output” is defined by Krashen’s (1982) “input hypothesis” and Swain’s (1985) “output hypothesis”.  For a real introduction, check here and here.

Krashen’s hypothesis is part of a larger theory called the “monitor model” and was itself a response to the behaviorist methodology that dominated most educational theories at the time.  Krashen built his monitor model upon the foundation that Noam Chomsky built with his theory of First Language Acquisition.  Chomsky’s claim is that Language (with a capital ‘L’) is innate to all humans.  That babies do not come into the world as a “blank slate”, but with an evolutionarily-developed capacity for acquiring language.  Similar to how bats have evolved eco-location.

It is this new conceptualization of language-learning, the acquisition of language that led to Krashen’s model.  From Lightbown’s How Languages are Learned: 

“First, in the acquisition-learning hypothesis, Krashen contrasts these two terms.  We ‘acquire’ as we are exposed to samples of the second language we understand in much the same way that children pick up their first language– with no conscious attention to language form.  we ‘learn’ on the other hand through conscious attention to form and rule learning.”

“The input hypothesis is that acquisition occurs when one is exposed to language that is comprehensible and that contains i + 1.  The ‘i’ represents the level of language already acquired, and the ‘+1″ is a metaphor for language (words, grammatical forms, aspects of pronunciation) that is just a step beyond that level” (pg 6-7).

Language input in the ESL classroom is the language directed towards students in the target language.  More specifically, ESL researchers talk about “meaning-focused input”, which is the type of input that students need in order to learn.  Meaning-focused input exists under 5 conditions:

“1. Most of what the learners are listening to or reading is already familiar to them.
2. The learners are interested in the input and want to understand it.
3. Only a small proportion of the language features are unknown to the learners.  In terms of vocabularly, 95 percent to 98 percent of the running words should be within the learners’ previous knowledge, and so nly five or preferably only one or two words per hundred should be unknown to them (Hu and Nation, 2000).
4. The learners can gain some knowledge of the unknown language items through context clues and background knowledge.
5. There are large quantities of input.” (Nation, 2009:3).

Swain’s output-hypothesis was presented as a response to Krashen due to her observation in French-immersion students high listening/comprehension skills, but low speaking/writing skills.  The output hypothesis is sometimes called, “pushed” or “forced” output because part of the learning process is said to happen when a student “notices” (Lightbown, 44) their linguistic limits.

However, the 5 main points are largely the same for the output hypothesis.  What students talk about should be familiar to them, their goal should be to convey a message to another person, only a small portion of the language they use should be unfamiliar, students use other communications strategies (like miming, gesture, body language, intonation, dictionaries, input etc..) to fill in for things they are unfamiliar with and finally, they need plenty of opportunities to produce (Nation 2009:5).

In addition, Swain’s output hypothesis proposes three functions of output that cannot be provided by input.  (1) As students attempt to speak, they notice where their knowledge lacks (not just the teacher). (2) As students attempt to formulate grammatically accurate sentences, they have to formulate hypothesizes about the language they are going to use.  And (3) As students formulate and test hypothesizes about the target language, they provide opportunities for meta-linguistic observation and discussion.  Meaning, when your student is learning the simple past-tense, and they apply ‘-ed’ to ‘fly’ and get ‘flied’, this gives you and your students a chance to talk about language and not just through language (Nation,2009:5-7).

It can seem like they really are just two-sides of the same coin, but there are advocates who would push against the other.  Krashen, for example, would see “the 20/80” experience as a very flawed methodology indeed.  But that’s what this is all about, learning (and making myself feel like I am doing something, or contributing something to this little school).

Next post, the Exploratory data.


Krashen, Stephen D (1991).  “The Input Hypothesis: An Update”.  Linguistics and Language Pedagogy:  The State of the Art.  ed. James E. Alatis. Georgetown University Press.

Lightbown, Patsy M. and Nina Spada (2006).  How Languages are Learned.  Oxford University Press.

Nation, I.S.P. and Jonathan Newton (2009).  Teaching ESL/EFL Listening and Speaking.  Routledge, New York.

Pinker, Steven and Paul Bloom (1995).  “Natural language and Natural Selection”.  The Adapted Mind: Evolutionary Psychology and the Generation of Culture.  ed.  Jerome H. Barkow et al.  Oxford University Press.

Swain, Merrill (2000).  “The output hypothesis and beyond: Mediating acquisition through collaborative dialogue”.  Sociocultural Theory and Second Language Learning.  ed. James E. Lantolf.  Oxford University Press.