Creative Discussion using Plain Pair Groups

Creative Discussion – a key to insight and change

William Plain
Emeritus Professor, Nagoya University of Foreign Studies

Plain Pair Group Teaching (Plain PGT)
- for universities and schools
Plain Pair Group Discussion (Plain PGD)
- for decision making and staff development
- for informal or community creative discussion

A flash of insight is the spark of cosmic intelligence.
Small group sharing of insight can change the pattern of human intelligence.

only search

Home page

Teaching and testing – the terrible twins

“Evaluation and responsibility in the student-centred classroom”

How to give a useful grade in a student-centred classroom: quantity as a guide to improvement.

Like many teachers I've administered a lot of tests and I continue to "give a grade" to my students. Although I “don't believe in testing” I think I've found a way around the problem of tests and giving grades in classes (proficiency testing is a different issue). This paper will question a few values and make a few suggestions. The manner in which I give the workshop will allow the participant to experience the type of classroom where the solutions I propose can best be applied.


A. Language education: creativity or conditioning

I have a few problems with conventional testing – and teaching!

Approach to testing reflects approach to teaching

Washback effect – both ways

Teaching as ‘heavy learning’ / ‘light learning’

Teacher centred, receptive learning, authority invested in teacher

Student centred, creative learning, student controls own learning

Teaching so student learns what someone else knows / teaching so student discovers what they know

Teacher as source of knowledge, transmission, student receiving – testing to determine if reception is accurate

Teacher imparts knowledge but student has time to work out own position – cannot be directly tested, no limit to student knowledge

Classroom creates society – as in CR so in workplace etc

Only one person in CR is active, creative, self-directing

Only those who follow the rules and rise to the top can say how things are done

Each person has equal share of activity, creativity, direction

Makes use of creative potential of all members

Testing – ‘the art of precise uncertainty’

Apparent precision of exact grade

Even publicly prepared tests have very wide error range

Look good but in fact very vague

See below on test error

B. Testing stops learning

Purpose of testing (exams) is to create/perpetuate a social hierarchy

Is this what you want? – NB this is what you are doing!

Purpose of testing is to promote accuracy

But testing looks for errors (psychology of teacher testing)

Testing tests what student does not know

Testing is always limited

Only tests within certain strict limits

How can you test open ended knowledge

Can a test test what the student knows?

Including what the teacher does not know

What type of teaching can increase learning

How can testing (evaluation) help learning

The teacher: always testing

Looking for error and correcting  = “the testing tic”

You can recognise a teacher a mile off

Always correcting you

Testing is the requirement of accuracy

Teacher looks for mistakes, not for genius

Continual correction blocks creativity, encourages conditioning

Do it the way you are told

Teacher, don’t correct

In order to facilitate communication, the teacher should avoid making individual corrections. Fluency requires the freedom to make mistakes.

Space for responsibility and personal involvement

Alternate teacher direction: teacher as centre

And student interaction: teacher silence

Principle rule: don’t correct

Difficult to pass up on chance to correct an error

Ethos of education: accuracy, and evaluation

Student expects it!

Preoccupation with accuracy hinders fluency

Fluency at heart of communication

Effect on future society of continuous teacher correction

Expectation of authority and lack of creativity and responsibility

Observe errors but comment to whole class

Avoid commenting to an individual

Leave freedom to experiment, even to ‘misunderstand’

Develop capacity for independence and creativity

C. Standard forms of evaluation

Testing proficiency: how good is the student

Absolute proficiency: how good is a student at the end of the course

Penalises those who study hard but have lower initial level of proficiency

Relative proficiency: how much has a student improved during the course

Very difficult to measure a small increment in proficiency over a short period

Testing performance: how well does the student ‘perform’ during the class

Measures of performance not very reliable

Ad hoc teacher tests have little chance of being accurate

Given that professional pre-tested tests have wide ‘error of measurement’

The description is not the language (Magritte – “This is not a pipe”)

Grammar: testing the description – grammar is not the language, it is (one) contrived description of the language

Communicative: testing the use of the language – in a contrived situation

How can we test creative thinking and discovery (in the L2)??

How can I give a useful grade when I want to avoid testing?

Evaluation which does not interfere with learning

Test what the student knows, not what they don’t know

Avoid teacher interference in student’s learning

Create sense of responsibility for one’s learning

Promote creativity and insight

Involving student in evaluation – approaches to classroom management

A system of partial self-evaluation can increase the student’s involvement and sense of responsibility.

Involvement in self-evaluation

Student direction of own learning

Most classroom activities imbued with deep sense of unreality

Self-evaluation is real – and can involve L2

D. “If a student works, they will improve”

Quantity as a measure of quality

Tests use quality – how good a student is at …

This is difficult to measure accurately, and creates hierarchy of power

Hypothesis/observable fact in the classroom

The amount of work a student does is a fair reflection of what they learn during the class

If a student does little, they learn little, if they do a lot, they learn a lot

Quantity is easy to measure

Attendance/speaking English/reports/reading

Can be done by the student

Gives student control over own learning

Promotes responsibility

Student is aware of involvement in class and can choose to change

Student controls grade

Can choose grade based on how much is done

Continuous evaluation

If ongoing grade is too low during semester, can go back and do extra work before end of semester

Testing will not stop because I have written this paper

Know how to test as well as possible

Know the limitations of testing

Use it to do what you want to do

Know when not to use testing

When you don’t want to achieve what testing achieves, look for/create something else.

What type of teaching can best implement this type of evaluation?

Plain Pair Group Teaching (Plain PGT) is designed to put in practice what has been presented in this paper

The Art of Precise Uncertainty

All test scores are approximate. It is important to know just how much approximation is built into any test we use, otherwise we risk making assertions based on a test score which the test makers themselves consider incorrect.

There are two ways of looking at test error:

(a) When can I say that a student’s score represents their “true” score? – Standard error of measurement (SEM)

(b) When can I say that one student’s score is higher or lower than another student’s score? – Standard error of difference (SEdiff)

There are two (or three) levels of confidence, i.e how many times can I be wrong in comparing a student’s (a) or students’ (b) scores and still say “it doesn’t matter”.

1 Standard error = 67% i.e. I will be wrong one time in three.

2 Standard errors = 95% i.e. I will be wrong one time in twenty.

3 Standard errors = 99% i.e. I will be wrong one time in a hundred

For each test purpose, we need to judge how many times we can make a mistake in evaluating a student and still say “that is acceptable”. For example if we say being wrong for 33 students in 100 is unjust, then we need to choose a 95% level of confidence - where we will still make a mistake for 5 students in 100 – or higher (99%) if absolutely necessary.

TOEFL (scores 300-670)

1 SEM = +/- 14 points (from test booklet)

 (the following are a personal guestimate by statistical extrapolation)

2 SEM = +/- 28 points

1 SEdiff = +/- 19 points

2 SEdiff = +/- 38 points

3 SEdiff = +/- 49 points

This means we cannot confidently tell the difference (at 95% confidence) between students whose real score may be 500 but will score between 462 or 538. Anywhere between 462 & 538 is “the same score”. In addition, in comparing 100 students whose ‘real’ score is 500, 5 will score down to 451 or up to 549  - and one will still be beyond those scores.

NB This is what happens when we have a ‘cut-off point’ on TOEFL scores for inclusion and exclusion e.g. for a scholarship, often with a difference of one or two points. We have almost no chance of avoiding injustice.



Recognition and remuneration


© William Plain  1990-2016 (print) 2005 - 2016 (website)