Learning: One step forward, then forgotten

Ever notice that students seem to learn a particular rhetorical convention or grammar point, then go back to an earlier stage of writing. I've noticed that my learning seems to take one step forward, and then regress, too.

One Step Forward
A year ago. I had finally got around to fixing the trackback on my blog. It seemed strange: Although my page template had scripts for Haloscan but not for Yahoo maps api, the html code showed the latter but not the former. Apparently, around January while playing with the design, I changed which templates were being referenced to an older version in another folder. What led me to looking for the template location was something that Mark Bernstein, the designer of Tinderbox (the software I use to write this blog) had said on more than one occasion: Look at the HTML code.

Easy enough to change folders. After fixing it, however, I created another problem. After I added google analytics script to the page template, the sidebar dropped below the posts. It made no sense that the one would lead to the other, but I took the script out anyway. Sure enough, that didn't fix it. Comparing my template with another orginal one, I saw that somehow, while pasting in the script, I had somehow deleted six important characters after the post division: </div>.

One Step Forgotten
Apparently, despite the many times Mark has advised me to look at the HTML code and the few steps I've taken in doing so, I may still forget to look at the code. On Thursday, I downloaded Tinderbox 4.0. When I looked at the web version of a post I was working on, there was no formatting. It was as if the CSS had disappeared. For some reason, I didn't compare the before and after HTML codes. If I had, I would have seen that they were different. Anyway, I emailed Mark, who replied:

Also, there's a warning in the release notes about stylesheet export -- a new tag pair


need to be redefined as empty strings for notes that are exporting stylesheets or other code where paragraph markup is unwanted.

Sure enough. That was the problem. And another problem was that I hadn't seen the release notes. Usually they're included in the DMG image, but not this time. Now, common sense might've told me to look in the Help menu, but I didn't. After all, they hadn't been there before. It took another email from Mark to direct my attention to this possibility. It's hard to see something when you're looking in the wrong direction.

Like Student, Like Teacher
I'm much like my students. That is, although everything that Mark had told me was pretty much common sense, at least for someone who was overly familiar with Tinderbox, HTML, and CSS, it wasn't so for me. And although everything that I tell my students is common sense, at least for someone like me familiar with English, writing, and language elarning, it isn't so for my students. And like me looking at HTML code, my students often zigzag in their ability to remember grammar points like subject-verb agreement and rhetorical conventions, such as framing quotations. Why would that be?

Forgetting and Learning
Anderson and Schunn in their article "The implications of the ACT-R learning theory: No magic bullets" (pdf) state that there are three learning processes (at the subsymbolic level):

1. Power Law of Learning. As a particular skill is practiced there is a gradual and systematic improvement in performance which corresponds to a power law. ...

2. Power Law of Forgetting. As time passes performance degrades, also according to a power function. ...

3. Multiplicative Effect of Practice and Retention. Most important, the Base-Level Equation implies a relationship between the combined variables of amount of practice and duration over which the information must be maintained. ...

This implies performance continuously improves with practice ... and continuously degrades with retention interval .... Most significantly the two factors multiply which means that increasing practice is a way to preserve the knowledge from the ravages of time.

As these are power laws, learning is a logarithmic function:

Picture by Hay Kranen / PD

In other words, (1) with practice, learning increases quickly; (2) with a lack of practice, retention of learning drops off quickly; and (3) the effects of (1) and (2) interact in a way that multiplies each other rather than just adds up.

In my case, I seldom look at release notes, infrequently look at HTML and CSS, and so easily forget. In the case of students, although they may be using English every day and may be writing every few days, they are only reminded every few weeks about subject-verb agreement or rhetorical conventions, when after turning in a draft, they receive my feedback. Their time on grammar and rhetorical conventions is insufficient to stabilize it. So, sometimes they remember and sometimes they don't.

Understanding these power laws is crucial to helping students improve their writing. Most of the literature on grammar feedback and error correction with L1 students say that it doesn't work, and the L2 research has had contradictory results. However, it's unlikely that students in any of this research practiced specific grammar points frequently enough to stabilize them. Of course, it may not be feasible for most students to practice enough and for most teachers to give feedback frequently enough in order to take advantage of these power laws of learning and forgetting. I have no real answers at this time, but for some of my earlier thoughts on error correction, see my series of posts beginning at Error Feedback in L2 Writing.