Turtle: Learning by Doing


When the turtle (a computer controlled robot) learns by doing, new procedures are established in the Logo computer language. Directing the turtle to immediate action, the user is engaged in CONCRETE PROGRAMMING: the retro-active generation of procedures which capture the previous directions to immediate action. Concrete programming is a new form of programming that has grown out of the Logo language in the past three years. It is a species apart, as different in spirit and style from planning oriented languages as it is from Cobol. This species of programming has been developed in applications for introducing children (and other computationally pre-literate people) to central ideas of computer science. In the commercial and scientific environments, concrete programming will produce a significant advance in programming productivity.

This design study, which attempts to present the first conceptually integrated example of a concrete programming facility, is supported by perspectives on the genesis of ideas set out in two papers by this author: Emerging Forms in Turtle Geometry and The Development of Objectives. As the most recent expression of an intellectual tradition, the study derives from and is sustained by years of experience at the Logo project. As preliminary to the design study, this paper contains a review and evaluation of introductory extensions to the Logo language.

One dictum congenial to the beliefs of teachers is that teaching is an excellent way to learn. Having to articulate his knowledge of a subject for explanation forces a teacher to examine some aspects of the material and issues embodied in it that he might otherwise little attend to. The essential element in learning-through-teaching is confrontation with those things little attended to; it is not the teacher student relationship.

Logo education has tried to liberate the child, his self-image, and sense of his own competence by permitting him to function as a teacher of the turtle. Using this image has been an effective tactic in teaching children turtle geometry, but it not a tactic free from risk. Specifically, the image of the teacher-student relationship fails to provide an enriched vision of the learning process. While augmenting the child’s self-esteem through switching roles, the use of the image validates a bad script, i.e. that one learns by being taught.

After asking primary grade children how they learned one thing and another, I have frequently heard this answer: “my teacher teached me.” The word “learning” as distinct in its meaning from “getting teached” is not an everyday working concept for small children. Their sense of teaching is not so rich as ours (with its overtones of education as self-development and the willingness to help in disentangling confusions or in applying ideas). The typical image is more like instruction, wherein the student “gets teached” by the instructor’s inserting knowledge into the student’s empty head. When the child teaches the turtle, this is precisely what the child does.

The idea of debugging is a significant enrichment of the image of the teacher-student relationship because, in convincing the child-as-teacher that making errors, i.e. estimates and first-order theories, is productive, the idea of debugging begins to break up the monolithicity of the image projected by instruction. The idea of debugging may help the child better understand and tolerate his own and his teachers’ errors when he returns to classes. Will it enrich his vision of the learning process inside his own head ?

If we hesitate to make clear for a child how he, as a learner, is learning in his role as teacher through action, i.e. by directing the turtle student in a domain in which he, the child-teacher is a neophyte, we fail to promulgate the richer vision of learning-through-action the child needs. Surely we do not want the child returning to classes, seeing himself as an empty-headed student, as an organismic turtle who can learn only by being taught.
This paper proceeds as an examination of attempts in the past years at the Logo project to have the turtle, even this mechanical robot, learn through doing. Thus it is a summary of introductory extensions to the Logo programming language, an attempt to evaluate them for extraction of their best ideas with the objective of developing a design for their integration. Any reader not interested in the Review and Evaluation may go directly to the final section of the paper, Design Ideas, for the configuration of abilities I recommend and my suggestions on implementing them.

Please note: The reviews this paper promises were those made for Rob, among others. This text will be extended and completed as soon as possible. RWL

Publication notes:

  • Written in January 1977. Unpublished Logo Working Paper.
Print Friendly, PDF & Email