Intelligent ESL Materials – WERTi 2.0

Two years ago, I wrote about the WERTi (Working with English Real Texts intelligently) system.  At the time, WERTi could be used to search for any article on Reuters and then apply one of three activities to the page for either prepositions or determiners.  The innovative (and intelligent) part of this process is that a nearly infinite number of tasks can be created and students can choose texts that interest them.

The WERTi system has now been rolled into a plugin for the Firefox browser.  So now any page being viewed by your browser can have the same activities applied.  Three new structures (gerunds, phrasal verbs, and wh-questions) have also been added.

For example, below, I did a search for a list of the most important questions in life and applied the “practice” activity to the “wh-questions”:

WERTi has taken each sentence and shuffled the words.  By clicking (or typing) them in the box, the user can check to see if they have been put back into the original order.  Green sentences have been reordered correctly while red boxes indicate they have not.  The “?” can also be clicked to reveal the original sentence which appears in black.

They system is not 100% perfect.  Occasionally structures are not labeled or are labeled incorrectly, but overall it does a very good job.  Now that it’s part of my browser, I’ve been surfing with it turned on and getting lots of great ideas for how it could be used.  Visit the download page to get more information about downloading and installing the plugin.  This project is still being developed by Detmar Meurers and his team, so suggestions for additional constructions are welcome and may be added to new versions.  If you think of some suggestions or find bugs, leave a comment below and I’ll pass them along to the development team.

Projects: Update

A quick update on projects I have been and am working on:

Not actually my desk, but you get the idea. Lots to do!

1. ESL Sandbox – coming up – I had been kicking around this idea (basically, word blocks that can be dragged around the screen to form a variety of sentences), but I have decided to try to pull it together.  After talking with some of the ICALLers at CALICO, I think I can come up with a flash-based version that would perform some basic, binary analysis of word blocks as they are dragged together.  If the words can be paired in that order, they will “stick.”  If not, they won’t.  Sentence level analysis might be too much to ask, but will word pairs be enough to analyze?  Once it’s built, we’ll try it and see.

2. Twitter and Personal Learning Environments – coming up – I was recently introduced to Twitter but really became a fan during CALICO.  It was used as another layer of discussion (a backchannel) that really added to the conference experience for me.  I also learned a lot about Personal Learning Environments and other ways to apply Web 2.0 technologies in educationally useful ways.  I intend to explore these further, particularly in the context of exploring offering online classes.

3. Interactive Whiteboards – ongoing – Since building my first $50 Wiimote-based interactive whiteboard, this project has been very well received.  I’m still hoping to get another grant to put more of them in more teachers’ hands.  (If you’re in Ohio, and interested, make sure you contact me.)  In the meantime, I received a Smartboard to use Spring Quarter.  It will be interesting to see how they compare.  I talked to some people at CALICO who had used both and preferred combining the Wiimote hardware and Smart software.  The Wiimote hardware is much more portable and is easier to use with permanently mounted projectors, which are in most of the classrooms I use.  Look for more updates on how this shakes out in the spring.

4. Second Life – done – I taught an elective class in Second Life last fall.  There were about six students who stayed with the course for its four-week duration.  Overall, they enjoyed the experience but it was more of a novelty than something that could really be used regularly in the classroom.  The student in my class were obviously technophiles who took to the movements (walking, flying, etc.) very naturally.  Building was a frustrating experience because of both the precision required and difficulty with collaborating (if two people accidentally take ownership of something by editing it, neither can move it again.)  We also had trouble finding reliably friendly places to meet new people to talk with.  Second Life search feels a little like pre-Google Yahoo searching — finding something you know is easy, but finding something new is difficult.  Until these issues are resolved, I probably won’t take students there again.

Siftables: Perfect for the Sandbox

A colleague of mine, Kathy, and I have been discussing ways to create a language sandbox game.  This idea would be to create blocks with words on them that somehow interact in a way that demonstrates grammaticality.

Siftables work like colored blocks, but they can interact.

Kathy’s oldschool.  She’s been experimenting with blocks of wood painted different colors which students can manipulate.  Great for kinesthetic learners!  We discussed cutting them into puzzlepiece shapes so that each block only “fits” other words according to grammatical rules.  At first, it seems like it would be possible to make nouns with square tabs that fit into square slots on verbs, and so on.  However, as complexity increases, this becomes exponentially more difficult.  Structures as complex as nouns modified by multiple adjectives would be prohibitively difficult.

What if a computer application could be developed that would replace the wood blocks with word tiles that could be manipulated with a mouse (or an interactive whiteboard!)?  Could the tiles snap together and repel each other like magnets according to grammar rules?  Could words be tagged for part-of-speech automatically within the application?  Could different categories of words (verbs, adjectives, adverbs, specifiers, etc.) be added and deleted with the check of a box?  Could students add their own tiles seemlessly into the pile?  Clearly, some intelligence would be required of the application to implement all of these features.

Siftables react to each other. Imagine a word on each one.

As I was kicking all of this around, my friend Mike at Ohio University pointed me to siftables, which seem to be the synthesis of both ideas.  Rather than try to describe these brilliant little devices, watch the TED Talks video.

Not only could these little devices fit the bill perfectly, the way they interact could inform interactions in the language sandbox I’ve been envisioning.  Until we’ve all got pockets full of siftables to pass out in class, my $50 wiimote-based interactive whiteboard will have to do.  In the meantime, I’m hoping that having students drag word tiles across the screen will work almost as well for kinesthetic learners.

MERLOT: Multimedia Educational Resource for Learning and Online Teaching

Search MERLOT for online resources.

I came across MERLOT (Multimedia Educational Resource for Learning and Online Teaching) in a TELR workshop recently.  I’ve been thinking about how ESL classes could be moved online and MERLOT was suggested as a place I might find some learning objects (reusable components of an online course).

While it wasn’t quite as simple as that, I did find a number of good resources.  One qualification: like many online resources, MERLOT does have some dead links and out-of-date activities, but at least there were plenty to choose from.  Try a search for “English grammar” and you’ll find lots of materials and promising leads.  The bad news is that it all feels a little like Yahoo circa 1996.  Although there are options for users to rate items, this system is not widely used and, in some cases, possibly outdated.  Many of the materials appear the be the work of single teachers doing yeoman’s work to create textbooks from scratch (Net Grammar), clearinghouses for their online materials (Edict Virtual Language Center), and drill-and-kill exerices and quizzes (English Works!, Interactive Quizzes at Capital Community College’s Guide to Grammar and Writing).

Although MERLOT makes an attempt to organize these resources, one still has to do some pretty deep wading to find useful resources.  A search returns a list, like Yahoo once did.  What is needed is something more Google-like: an algorithm-based system for sorting the wheat from the chaff.  Even better would be a proliferation of intelligent CALL (ICALL) materials (like WERTi) that would include tools to create exercises from existing content.  This is more a criticism of ESL CALL materials in general, than it is of MERLOT.  Unfortunately, we’re not there yet.

Find something you like on MERLOT?  Leave a comment and share it.

Intelligent ESL Materials – WERTi

The WERTi System. Click to try it!

Many ESL professionals are familiar with the acronym CALL (Computer Assisted Language Learning), but an interesting off-shoot is ICALL (Intelligent CALL). One of my favorite examples is the WERTi system, developed by Detmar Meurers while he was at Ohio State. (To try it, log in as “anonymous” and leave the password field blank.)

This system uses XML to create three activities based on Reuters news articles: Color, Click, and Fill-in-the-blanks. The first activity makes targeted words blue, raising the student’s awareness of the targeted structure. The second makes every word clickable. When the user clicks on a target word, it turns green; mistakes turn red. The third replaces every target word with a blank that students can complete. Correct responses are again green, errors red. If users give up and ask the computer to fill in the blank, the answer is blue. Originally based around Prepositions, I suggested to Detmar that articles might also be worth practicing, so Determiners were added (more on that in a minute).

The greatest thing about this system is that the computer is exploited to create the activities, the topics of which are selected by the student. And the number of activities is virtually unlimited.

The downside is that computers are not truly “intelligent.” Consequently, a few mistakes are made. Each page is marked up in XML using the Penn Tagset. But if a word is misidentified, this will error be reflected in the activity.

Incidentally, if you want to “hack” they system to try different grammatical features, you can add the tag from the Penn Tagset to the URL. So, to change an activity from determiners to superlative adjectives, change “pos_target=DT” to “pos_target=JJS” and voila!

Some features in the tagset are probably too uncommon to be worth including; Others may not be easy to practice using these activities. But, the idea that computers can generate activities from any page on the internet is really appealing to me.