Announcements

Posted on Feb 08 2012Charles Perfetti gave three recent presentations of note
Posted on Jan 20 2012DataShop 5.2 released!
Posted on Jan 16 2012New Course: 05-438/05-838: The Role of Technology in Learning in the 21st Century

[view all announcements]

Upcoming Talks

Feb 16 2012PiER Edbag-Julia Kaufman
Feb 20 2012PSLC Seminar - Judith Harackiewicz
Mar 15 2012PIER Edbag-Derek Lomas

[view all talks]

Upcoming Events

[view all events]

Algebra

A Multimodal Interface for Solving Equations

Primary Investigator: Lisa Anthony
Co-PIs & Other Investigator(s): Jie Yang, Ken Koedinger

Cluster(s): Coordinative Learning, Enabling Technology
Course(s): Algebra

The long-term goal of this project is to develop a multimodal intelligent tutoring system that will allow students to use handwriting input to solve mathematics equations online. We believe that providing a handwriting-based interface that is more like paper, more natural and which reduces the cognitive load of the student – who is freed up to think solely about the mathematics and not about how to type math symbols or how to find things in menus – will result in better learning. This project was a one-year project, setting the stage for motivating further research in this area by answering the following questions:

1. From a usability perspective, which input method (keyboard, handwriting, speech, or combinations thereof) is more efficient and/or tends to have less errors?
2. How will students naturally interact multimodally when solving mathematical equations?
3. From a technological perspective, how accurate will the handwriting and speech recognition technologies be with or without the ability to learn from each other and improve over time?

This project has reached the end of its PSLC funding and is in the process of applying for a grant for the upcoming year. During this year, we have explored the above questions through two user studies, one with general users and one with students learning algebra. In both user populations, we found several advantages of using handwriting-based interfaces over traditional keyboard-and-mouse interfaces. Handwriting-based equation entry is both faster and more efficient than keyboard-and-mouse; in addition, we found that users rated handwriting more highly in a post-session questionnaire. We also found some preliminary evidence in support of the following: that people’s input tends to be more variable in some modalities than in others; that people tend to leave out ambiguity control phrases such as “quantity” and “open parentheses” more in speech than actual physical parentheses in handwriting; and that the errors users make in entry tend to be non-overlapping across modalities, meaning that users make errors in different places in the same equation in different modalities. The first two points will likely make recognition of user input more difficult, while the third point may in fact help alleviate this by allowing the use of co-training and co-recognition of simultaneous or redundant input streams. In our second study, we additionally explored the advantages handwriting may have over keyboard input with respect to learning. High school and middle school algebra students came to the lab and solved simple algebraic equations in one of three modalities: handwriting, keyboard, and handwriting-plusspeech. Preliminary analyses show that students experience higher pre-test to post-test gains with handwriting than with keyboard. The study was controlled for number of problems (and concepts addressed) rather than time, so while we saw similar overall learning gains, handwriting students spent about 1/3 less time than keyboard students. Thus the effective gain, if they were given equal time, we hypothesize to be greater for handwriting than for keyboard. The handwriting-plus-speech method was a shadowing method in which students simply spoke what they were writing as they solved the equations, and this was a very low performer in the study. We explored this method due to technological advantages it might provide, but it seems the pedagogical benefits of spoken self-explanations will not be apparent with a shadowing mechanism.

Project Runs: 2004-11-01 to 2005-11-01

Most recent project report:
(2006-03-16) anthony-report1.pdf
Most recent project poster:
(2006-03-16) anthony-poster1.pdf

Improving Algebra Learning and Collaboration through Collaborative Extensions to the Algebra Cognitive Tutor

Primary Investigator: Bruce McLaren
Co-PIs & Other Investigator(s): Nikol Rummel, Hans Spada

Cluster(s): Interactive Communications
Course(s): Algebra

Our research investigates whether we can improve learning with an intelligent tutoring system, the Algebra I Cognitive Tutor by adding collaborative learning elements to the software. Until now, the Algebra Cognitive Tutor, has been used strictly in one-on-one (machine-to-student) fashion.
Why do we expect that collaboration could improve students’ algebra learning with the Cognitive Tutor? Due to cognitive, meta-cognitive and social activities that are characteristic of collaborative problem-solving (e.g. mutual explanations, cognitive conflict, joint sense-making), more relevant learning events are expected to occur that result in deeper, more robust learning. However, as research in collaborative learning has also demonstrated: it is likely that such beneficial collaborative activities do not occur without support. For example, it has often been found that one partner dominates the interaction (i.e., one partner does all of the talking or solves all of the problems). Collaborative problem-solving and learning can be supported effectively with collaboration scripts. Similar to a movie script, the script designates particular activities and roles to participants for each phase of their collaboration.
Our project investigates two particular script approaches to support students’ collaboration during problem-solving with the Algebra I Cognitive Tutor: a Peer Tutoring Script (PTS) and a Collaboration Problem-Solving Script (CPS). The Peer Tutoring Script (PTS) involves student dyads taking turns tutoring one another in solving algebra problems. The Collaboration Problem-Solving Script (CPS) involves alternation of individual work, collaboration and recapitulation, and uses algebra tasks that are designed to lend themselves to collaboration. We plan to conduct two large-scale in vivo studies in the PSLC LearnLab facilities; that is, in Pittsburgh middle and high schools. One study, focusing on the CPS approach, has just been conducted in 11 classrooms. The next study will focus on the PTS approach.
The studies explore the broad hypothesis that instructionally guided (i.e. scripted) collaboration, promotes learning of both domain-specific algebra skills and collaboration skills. In other words, we will test for (a) improvements in students’ algebra knowledge, but also (b) for their ability to maintain a fruitful collaborative interaction after having gained some experience in collaborating (conditions 2) or after having collaborated with a script (conditions 3). This research is thus an explicit exploration of the complementary benefits of foundation building and sense making in that it attempts to combine the benefits of foundation building (i.e., the more "drill-and-practice approach" of individual tutoring) with sense making (i.e., the explanations and dialogue inherent in collaboration).

Project Runs: 2005-01-01 to 2007-09-30

Most recent project report:
(2006-11-22) algcollab-2006-12-revised.doc
Most recent project poster:
(2006-06-15) pslcalgcollab2006-04.ppt

Microgenetic Analyses

Primary Investigator: Julie Booth

Cluster(s): Coordinative Learning
Course(s): Algebra

Overlapping waves theory (Siegler, 1996) maintains that individuals know and use a variety of strategies which compete with each other for use in any given situation, and that with improved or increased knowledge, good strategies gradually replace ineffective ones. The current project will use microgenetic analyses to examine students’ strategy use while learning to solve algebraic equations with the Algebra Tutor and to determine whether misconceptions in students’ understanding of Algebra is associated with use of ineffective or “buggy” procedures for solving problems; it will also test how conceptual encoding exercises designed to correct these misconceptions might affect students’ strategy use while learning with the Algebra Tutor. The first goal of the project is to identify common buggy strategies that students attempt to use when solving algebraic equations with the Tutor, as well as the conceptual and procedural prerequisites that are necessary for solving these problems correctly. The second goal is to design an intervention intended to provide students with conceptual encoding exercises which map onto the identified prerequisites; this intervention will then be implemented and tested in a variety of LearnLab Algebra I classrooms. By providing students with co-training (or coordinative learning) opportunities using multiple types of instruction (procedural practice and conceptual encoding) during their lessons, we expect to see less use of buggy strategies, greater use of effective strategies, and improvements in their knowledge of algebra concepts. This is expected to lead to better performance on the Tutor problems in the lesson, as well as three types of robust learning: better transfer of the strategies to subsequent test problems, long-term retention of the better strategies for use in later relevant lessons and review exercises, and accelerated future learning in related units later in the course. The usefulness of these conceptual encoding exercises in diverse populations will also be examined.

Project Runs: 2005-09-01 to 2007-08-31

Most recent project report:
(2006-11-21) booth project report for 2006 advisory board.doc
Most recent project poster:
(2006-06-05) booth poster for site visit 2006.ppt

Knowledge Tracking

Primary Investigator: Philip Pavlik

Cluster(s): Coordinative Learning, Fluency
Course(s): Algebra

None

Project Runs: 2005-09-01 to 2008-08-31