The more efficient the instructional sequence, the greater the benefit to students and this is of particular importance for students who have lagged behind their peers because, if they are to catch up, they will need to learn more in less time. But how can such a goal be achieved? Lessons need to be focused, purposive and communication needs to be clear. However, even if these conditions are met, this will not guarantee that learning will be maximised.

__Generativity__

DI programmes teach to the general case and teaching can be said to be ‘generative’ if it enables a learner to respond appropriately to untaught situations. In this sense, generativity is very similar to the ideas of transfer and flexible knowledge. For example, if reading instruction enables student to read untaught words, it can be said to be generative. Teaching students that the grapheme ph is a spelling for the /f/ sound can be said to be a generative approach if students can then decode untaught words that contain this grapheme.

Teaching number families in basic arithmetic is also a generative approach. For example, in addition and subtraction, 2, 3 and 5 is a number family because it can produce four basic facts: 2+3=5, 3+2=5, 5-2=3 and 5-3=2. Instead of asking students to memorise these four facts, students can learn one number family alongside the relations necessary to produce the four facts. This approach is far more efficient as it has a lower memorisation load whilst also teaching students relations that can be transferred to other number families.

Teaching students to manipulate the components of a sentence is also a generative strategy and is a much more efficient approach than learning how to copy and apply single example sentences.

Most content domains are so large and complex that it is impossible to teach everything that they encompass. There are far too many possible combinations of content, responses and contexts to teach them all individually. Because of this, we need to teach to the general case, but how can this be done?

__Content Analysis__

The purpose of content analysis is to identify generalizable relations within a domain and arrange the content in such a way that learning becomes as efficient as possible. Efficiency here means maximizing the amount of learning within a given time and the goal is to produce the most learning from the least amount of teaching.

A content domain is synonymous with the topic that is being learned, examples of which include computer programming, writing analytical essays, mathematics or molecular chemistry. For every possible content domain, there will be multiple ways of conducting content analysis and deciding what to teach and some ways will be more generative than others. For example, in the content domain of spelling, teachers could decide to produce word lists for students to learn based around topic themes or commonly misspelled words; however, such decisions could be problematic as they are not generalizable. Alternatively, words could be grouped and taught according to phonic or morphographic content therefore producing a highly generative approach to spelling instruction.

Content analysis is the base upon which all other pillars of curriculum design stand. Every other part of curriculum design (creating explanations, sequencing examples etc) will depend upon the content analysis. If the content analysis isn’t up to scratch, the generativity of instruction will be minimal irrespective of how well the other aspects have been designed.

So if it is so important, how do we do it properly?

The process of content analysis should involve a cycle of logical and empirical analyses:

The process should begin through logical analysis where teachers generate possible methods of organizing the content. These should then be logically compared before being empirically tested.

Here’s what this might look like in English:

Imagine you want to teach students how to write compelling pieces of persuasive writing.

__Step 1: Engage and Generate__

You would start by reading research into effective writing instruction, perhaps choosing to read *The Handbook of Writing Research*

The handbook explains how emulating model texts and teaching students how to structure their writing can be effective approaches. So, you come up with a few possible approaches:

- Teaching classical speech structure (Exordium, Narratio, Divisio, Probatio, Peroration)
- Asking students to emulate that brilliant article that you found
- Teaching a different persuasive structure

__Step 2: Logically Compare__

By logically compare all three approaches, it quickly becomes apparent that approach b is not optimal. While the article is brilliantly polemic, it does not contain a transferable structure that students can use and, despite the fact that the prose is captivating, sardonic and nuanced, its complexity will preclude all but the most able students from aping its inimitable style. While students may benefit in other ways from reading the article, it will not be the best choice here. As a result, you logically infer that a or c may be more suitable, both containing malleable structures that could be used in a wide range of relevant contexts and tasks.

Here are some possible questions you could ask while comparing approaches:

- Is the approach clear enough for all students to understand?
- Can you create and sequence sufficient examples and non-examples in order to refine and develop students’ mental representations of what you are teaching?
- Can the approach be practiced so that students build fluency?
- Does the approach allow space for creativity: is it a straightjacket or a springboard for success?
- Is the approach applicable to the full range of relevant contexts that student will encounter?

__Step 3: Empirically Test__

Once you have finished deciding which one you will use, you will need to test whether it is efficient and generative. Just because an approach is logically generative does not mean that it will necessarily be empirically generative.

For a test to truly assess generativity, it must involve untaught content. With structuring persuasive writing, you could use Language Paper 2 Question 5 tasks that students have not encountered before so that you can see whether they can apply the approach that has been taught. To make the test as valid and rigorous as possible, you should choose an outlier style of question where the topic and task are maximally dissimilar to what is usually asked for: does the approach work with writing letters? What about topic X? You should also pay close attention to the output of the weakest students and use them as your guide as to whether the approach is successful.

__Directly examining the domain__

Reading research can be very helpful when beginning content analysis but so can analyzing the domain itself. Wherever possible, content should be chosen for its utility and whatever subject you teach, there will be some big ideas or concepts that can be applied to different units and topics, some even crossing traditional subject boundaries. As an example, the idea of convection can be applied widely in science and geography:

If you are interested, many of the ideas in this blog post are based upon *‘Features of Direct Instruction: Content Analysis’* Behaviour Analysis and Practice 2021. Thanks to @JonOwenDI who posted a link to the paper on twitter.

I would highly recommend Jon’s blog which has some fascinating posts about DI and curriculum design.