[LONG] Re: Language Arts Stack

[Mark Greenberg]

On Wednesday, July 27, 2005, at 10:00 AM, Marielle wrote:
>
> Cool.... it may even help me improve my English.

Thanks.  If you speak as well as you write, then nobody can fault your 
English.

> It may help have some boxed text stating "start typing some text" as
> on first exposure, what you have to do is not obvious (I read the
> readme file only after trying). [snip] it would be a pity to have
> users trash your application only because they don't understand how
> to use it...

The kids I teach (inner-city, mostly poor, many just learning English) 
have no trouble with this.  If they are confused by the lack of the 
usual textbox, they experiment by typing a key to see what happens.  
This is the normal way a youngster works through a video game.

> er hum, constructivism -- or constructionism as Papert
> prefers to call it -- is about progressive knowledge building.

I know.  I'm reading Papert's The Connected Family right now and just 
finished Constructivism and the Technology of Instruction edited by 
Duffy & Jonassen.  Allowing students to build (in Rev, of course) their 
own video games that address a learning objective... that would be 
Constructivist.  Such projects are common in my classes (except that 
the students have used HyperStudio in the past).
>
> Is the fact that each sentence must be fully typed a deliberate
> decision? [snip] Did you
> consider having a double click on a word copy it at the end of the
> typed sentence?

Yes, I considered it.  I feel that a main reason that most isolation 
grammar drills fail is because they're multiple-choice or 
fill-in-the-blank.  The students don't hear the entire flow of the 
sentence in their mind's ear.
>
> It is not obvious either what rules you use to evaluate a good
> sentence (after all, that's what you are trying to teach).

That's on purpose.  I want the students to discover for themselves what 
makes their scores increase.
> It is
> possible to get a perfect score by only using A; B? That would be a
> pity given that what you can do is very supple.

All A;B would hurt a student's score in three categories.  First, their 
score for variety would be 300 out of 2,400 because she only used one 
pattern (out of 22 possible) in all eight sentences.  Next, that 
particular pattern has a medium rating for sophistication, 190 out of 
300, if I remember right.  That's another 880 off the max score.  
Finally, that pattern rarely results in a sentence that sounds natural, 
smooth, optimal; so the sense score would be in the medium range too.

> "Our friend Andy helped make this program. He is our son's roommate"
> can become
> "Our friend Andy, our son's roommate, helped make this program"

That is a good way to combine the two sentences.  The stack will accept 
that as well as 92 other "correct" answers.
>
> It's a pity that as soon as you type a ".", the sentence disappears.
> Why not give the possibility to experiment longer with a each
> sentence. Rather than give a score for each combination, why not give
> a score for the number of different correct combinations you can make
> with a single sentence. If the purpose is to learn variety and
> sophistication, this seems to be the best way to encourage it.

Interesting.  Yes, I like it.  I think that that would be a different 
stack altogether, but it would allow the students to experiment more 
with the various ways to combine the sentences.  I'll file that idea 
away for further development.  Thanks.
>
> Also, the red letters tells me that something is wrong, but it gives
> me no cue whatsoever about what is wrong.

Those letters are not part of a correct combined sentence.

> I can start over and quit,
> but I can not skip a sentence or get access to some tutorial that
> gives me the theory behind sentence combining (or some tricks to
> solve the task at hand).

Assuming that the teacher has given the students enough basic 
instruction in sentences patterns or combining sentences, the student 
should be able to come up with at least one of the many acceptable 
answers.  If not, the teacher should help.

> WORSE. The last screen tells me that I had
> only 62.8% accuracy? How come, all my sentences were correct (though
> I did mistypings and experimented to test the program).

See, you're constructing your own theories about how to succeed at the 
game.  If you were in my class, you'd lean over to the girl sitting at 
the next computer and say, "Hey, I got 'em all right and the stupid 
game gave me only 62% accuracy.  That sucks!  How did you get 91% on 
yours?"  (That's how students talk around here.)  The girl would 
answer, "I dunno.  Maybe 'cause I didn't use the delete button very 
much."  You'd shrug your shoulders and try again, this time reaching 
88% accuracy by trying not to backspace too much.
> How do your
> program compute that?

Backspaces / Characters

>  Sophistication: 1960 what does it
> mean?

"Jack works for Intel, and he is my best friend," is a simple pattern.  
"Jack, who works for Intel, is my best friend," is beyond the beginner 
in English, but fluent students use that pattern.  "Jack, my best 
friend, works for Intel," is also a stretch for some kids.  The most 
"sophisticated" sentence construction according to the stack is 
absolute phrasing, which only the very best writers attempt in 
authentic writing.  2,000 is the goal here.

> Sense: 1760. What does it mean?

"While Jack works for Intel, he is my best friend," is awkward, vague, 
ambiguous.  The computer will accept it, but it is not the best choice. 
  It doesn't "sound" good to an experienced writer.  Other choices make 
more sense.  Again, you're shooting for 2,000.  Your score is very good 
though, really.

> Best, really, is to give
> feedback after *each* sentence and give the user the possibility to
> correct to try to achieve an even better score.

I assume that the students will have multiple tries at the game, and 
I'd rather them abstract the patterns that lead to success than dwell 
on one sentence.

> If you successfully
> expressed criterion as program functions, that means you have an
> heuristic. Rather than have the person guess what your heuristic may
> be, best is to make it available to the user when they get stuck or
> when viewing their score (after all, the purpose is not to sanction
> bad writing, but to encourage better habits).

You and I disagree on this point.  I'd rather the student guess, given 
that the teacher has introduced the basic sentence patterns, etc.  I 
don't think that there is a right way here; as a teacher, I just like 
to bite my tongue and watch the student come to a conclusion of her 
own.  Of course, I'll monitor and help her if she is hopelessly lost.
>
> As I said, your application is very cool. I am only encouraging you
> to introduce a few changes that may help it have an even greater
> impact (it deserves it).

Thanks.

		Mark