Assessing Comprehension: What are we looking for?
For some time now I have been looking at our Informal Prose Inventory and wondering how I can improve it. It has faithfully withstood the test of time (23 years) but may be in need of a revamp in light of what we now know about the comprehension process.
The purposes of Comprehension Assessment has not changed.
BENCHMARKING: We are seeking data to determine a level of competence which can become a benchmark for future measurement (has progress been made?)
DIAGNOSTICS: We want data that will identify weaknesses in the reader's skill set which will guide our future instruction.
We do this by using tools which will allow us to identify a reader's Instructional Reading age or level of text difficulty.
There are many differing points of view about this so let me explain further what I mean by this and why I think it is important.
If we can agree that there is an essential comprehension skill set that the reader uses to comprehend language, we will also acknowledge that any particular reader will have a different mix of these skills.
The skill set I am referring is that proposes by Hollis Scarborough in her well known and accepted Reading Rope.
To comprehend the language that the reader encounters in a passage of text, the reader accesses their knowledge of the world, their known vocabulary, their familiarity of sentence structure, their knowledge of written genre, all of which are processed and synthesised by their verbal reasoning; their ability to making inferences).
This skill set (the 5 strands of Scarborough's 'Language Comprehension Rope) has been developing intuitively since birth and children experience language and the world around them. Depending on that exposure, they will arrive in your classroom with a unique mix of these skills.
Our job as teachers is to identify that mix and provide them with practice opportunities to grow that skill set.
These are not quick fix instructional goals that we can set and slot into our weekly lesson plans.
Background knowledge comes from a diverse and varied curriculum (science, social studies etc) and will grow with time.
Likewise, the average 10 year old will have acquired a mental vocabulary of around 10,000 words and will add to that through exposure to text and oral conversation. Of course we can support his through learning word lists but this is not the foremost means of growing a vocabulary.
We can teach the rules of grammar but the research shows that this has little impact on the in-the-moment construction of meaning that occurs in the brain as it appropriates sentence structure in its quest for constructing meaning.
Similarly, you will make some headway by teaching idiom, and other figurative language features but not necessarily be rewarded by huge leaps with your learners in their reading comprehension ability.
The reality is that they need structured practice with developing this skill set until they acquire automacy.
The key to success here is that they get to practice on text which has JUST THE RIGHT AMOUNT OF CHALLENGE for them to
The Components of the 'testing' procedure
With our Informal Prose Inventory we included there parts. We look at these and some suggested changes we are contemplating.
This has always been an important part of comprehension assessment because it underlines my belief (supported by many of the experts) that a certain degree of decoding fluency is an important precursor to the act of comprehension (constructing meaning and thinking critically).
In my experience, if a reader is to be able to successfully address the comprehension challenges of a passage, then they need to be able to read that text at 97% PLUS accuracy.
That means that there should be is no more that 3 uncorrected miscues or errors out of a hundred words. Three errors in a hundred words will not sufficiently distract the reader from constructing meaning and the context should allow them to generate an accurate understanding of the passage.
Anything more, and the cognitive energy switches from the message to the reading of the words.