Re:(MIDTERM/WEEK8-3)SPSS on two files

[kellogg]

I can't seem to download the files. Can you send them to my e-mail, kellogg59@hanmail.net?

⓵ the frequency b/t T and Ss → T1:Ss=3:1, T2:Ss=7:3

Be careful! This MIGHT be nothing more than the difference between the way that Shushu transcribes her data and the way in which YOU transcribed Ms. Jang's data.

Look:

a) S: 남산

T: 남산~ 

T: 또?

T: Maybe.

T:  in the first scene.

T: What are they talking(sic)? 지금

S: 고백!

Here we can see that the teacher takes five turns and the children take only two. But suppose we transcribe it like THIS?

b) S: 남산

T: 남산~ 또? Maybe. In the first scene. What are they talking(sic)? 지금

S: 고백!

Now the teacher takes only one turn and the children take two.

Shushu pays a lot of attention to UTTERANCES--she thinks it's important to record POTENTIAL changes of speaker. But Ms. Seo is much more REALIST: she ONLY records real changes of speaker. Let's look at YOUR transcription of Ms. Jang:

T: Grandma, OK. She's grandma. Whose grandma? Whose grandma is she?

S: (no response)

S1: Little Red

T: Right. She is Little Red's grandma.(Recasting) OK. How do you spell grandma? How do you spell grandma?(Repetition) OK, the boy.

S1: g

What do you think? Is it more like a) or like b)?

⓶ wps T-analysis b/t T and T1 → p=.130(n=74)

: no meaningful differance

Both of them are skilled teachers, and skilled teachers tend to tailor their input to the children's output. So the results of the NEXT test are very important.

⓷ wps T-analysis b/t S1 and S2 → no valid data

I think the problem with the t-test probably has to do with the way you coded S1 and S2, but I'm not sure. It's better if you have a SEPARATE VARIABLE (called "lesson") and then you can use that to split the results or as a variable for comparison. I'll show you how to do that when we have time on Monday.

: I don't understand why? I calculated the same amount of data for each S1 and S2. Is that a problem to have only 22 data or S1's analysed data takes up less than 20% of the whole data? I thought it's more correct to have the same number of case.

It's not necessary to have the same number of cases. The t-test works by assuming that the variances are normally distributed (or correcting for non-normal distribution) and then calculating the average DISTANCE from the mean.

When you compare AVERAGES, it doesn't matter how many cases you have (just as when you compare percentages). You should use as many cases as you have: the more data, the more accurate your inferences will be.

   Apparently, there should a big difference b/t T1/T2 and S1/S2. It can be interesting to compare them. I once did this with my own data before and after your lecture(WEEK4). It worked well, but then I did the frequency test only, at that time. I'd like to learn how to compare two lessons qualitively and quantatively besides wps, Professor! 

Well, here are SOME things that are very countable:

a) Amount of "uptake" in words.

b) Number of "new ideas" in words (usually stressed words)

c) Number of VERBS (clauses)

And here are some things that are very codable (qualitatively):

a) Whether an utterance is a command, a statement, or a question

b) Whether an utterance is an initiate, a respond, or a follow up.

c) Whether a follow up is uptake, recasting, etc.

   About mistakes and learning, I remember you said it seemed that the more Ss learn the more mistakes they make.

My daddy used to say that if you don't fall down, then you aren't skiing hard enough.

 You also asked yourself if it is the way to teach nothing to let Ss make no mistake as a joke. Well, that's why I choose the second data(장쌤, T2) to compare the one you gave to me(Shushu, T1).

Of course, the real problem with the second data is REHEARSAL. If there is rehearsal, we cannot really say that there is ANY modeling, or prompting, or recasting at all--everything that was learned was learnt long before the class began.

 Ss2 makes huge mistakes but they are in the way of development(43% of Ss2' sentences are clause-question and statement, n=13, total=30) but Ss1 makes almost no mistake(except the articles) but they don't seem to learn much, even though they listen and answer the questions in a various way to use all the character names.

I think one of the key QUALITATIVE differences we see between the two data sets is the issue we've been talking about in class.

You know that we have a "belly button problem": the questions that we ask to check comprehension are often more difficult than the texts whose comprehension they are supposed to check. So we have no way to know whether the children don't understand the text or simply don't understand the question.

Minkyeong solves this problem by RECONTEXTUALIZING--by taking questions like "Do you like fall?" out of their imaginary context and recontextualizing them in a "non-imaginary" one: Do you REALLY like fall? As a student?

Like most solutions, this solution creates new problems. When we repeat the sentence pattern, but we completely CHANGE the contextual conditions under which that sentence pattern serves as a "communicative function", then it is no longer a communicative function.

The answers that we gave to the question "who says what to whom and why?" no longer apply; it is no longer the case that Paul and Nami are trying to agree on the seasons so that they can decide if they like each other or not, or that the seasons in Australia are relevantly different from the ones in Korea.

That means that the only thing that the children can really cling to is GRAMMAR--the sentence pattern is NOT a communicative function at all, but only a grammar pattern. So we see that Minkyeong's lesson has a strong grammatical bent, and this explains a lot of the fixed I-R-F patterns ("Right" "Correct" "OK") that we see.

ANOTHER way to solve our problem (but like Minkyeong's solution, it brings new problems) is to put the children into the story and have them CONTINUE the story in some way. That is what Shushu is doing. As you point out, it is VERY effective--in producing errors! But it also DOES give us a fair a mount of indirect evidence about how well they understood the dialogue.

A THIRD way to solve the problem (and like MInkyeong's solution and Shushu's solution, it is a solution that has problems of its own) is Ms. Jang's solution. She does NOT put the children into the story: she puts the story into the children. And we can see the result: she's very interested in vocabulary and spelling and so on. Although her SUBJECT MATTER is actually pre-school material, the emphasis is actually very much like HIGH SCHOOL teaching, and for that reason it probably pleases parents and principals.

Does it please the CHILDREN? Oh, children are easy to please! They are happy when the the materials are colorful and the teacher is cheerful and there is not too much pressure to do things they don't already know how to do. But that doesn't tell us very much!

But it doesn't please ME, to tell you the truth. I cannot believe that the way to get appropriate teaching for PRIMARY SCHOOL kids is to combine two kinds of INAPPROPRIATE teaching: pre-school content and high school testing techniques!

   Then why does the T2 use this preventive method(mostly medeling)?

I disagree that this is modeling. I think she is prompting reactions that have already been rehearsed for the cameras.

 She has a fixed way of teaching to control Ss production almost every time with her own prompts(i.e. Give me a(n), Ready? One, two, three, go!). Since Ss2 are 4th graders, it can give them a very clear idea of what they are practicing(like what you commented on Minkyeong's lesson last time).

Yes.

The problem is this highly controlled way can get rid of the opportunity that Ss can THINK. They can remember but it's hard to go BEYOND that. The sentences they make are controlled, fixed and not creative. I can't be sure that they're learning. Well, they are learning but it's not an acquisition(from Krashen's point of view.)

The problem with trying to distinguish between learning and acquisition is that BOTH of them take place inside the brain. We can't CODE that or COUNT that, we can't get inside the brain to figure out what is what.

But here's something we CAN do. Vygotsky distinguishes between LEARNING on the one hand, and DEVELOPMENT on the other. LEARNING is what children do WITH outside help: so prompting and modeling are examples of learning. But DEVELOPMENT is what they do on their own: unprompted and unmodelled utterances which are then uptaken for prompting and modeling and recasting.

Both learning and development are SOCIAL, not psychological; they take place BETWEEN brains and not INSIDE them. So we can code them and we can count them!

dk

[11기신지영]

Ahha~ So, this is the secret of the skilled teacher! And I like your opinion about Jang's prompting reaction, but for Rehearsal I never do that. It's not always so. By the way, do you say 'daddy' yet? I call my father 아빠, and he likes it.