Who knew that assessing reading could be so complicated?
There are a range of assessments for schools to use, each doing something slightly different. There are those that measure what we might consider the nuts and bolts of texts: word decoding, sentence length, word knowledge.
Some are built around single sentences, or cloze procedures, where students have to choose the most appropriate word. Others focus on higher-level processes - understanding narrative, inference and social and cultural understanding.
Some assessments give scores that are measured in relation to the wider population (norm-referenced); others use criteria to assess a student’s level of competency (criterion scales).
Each type of assessment gives valuable information, but none gives the whole complex picture of a student as a reader, taking into account the cognitive, social, contextual and emotional aspects of reading development.
How, then, should teachers act on the information gleaned from a reading assessment? And when an assessment is focused on just one aspect of reading, how much stock can we place in the recommendations that follow?
Let’s take Lexile Levels, for example. Lexiles are designed to give a guide to the “readability” of texts, using an algorithm that takes into consideration the average length of sentences and the frequency of vocabulary in those texts.
These readability levels are correlated with ages and stages. Students take an assessment that calculates their ideal Lexile level and this then informs the books they will go on to read - only being recommended texts within the appropriate level.
It’s a simple approach that seems like a godsend to busy teachers and school leaders who need clear progress measures.
But this simplicity comes at a cost.
If we look at text difficulty from the perspective of reading comprehension - as researcher Scott Crossley and colleagues have shown - texts that are more easily understood contain simpler (but not necessarily more frequent) vocabulary, lower text cohesion (fewer links within and across sentences) and fewer verbs. Texts that are easier to read at speed (read fluently) have fewer unique three-word phrases, less sophisticated words and fewer things like proper nouns in each sentence.
The content and structure of a text are important factors in reading motivation and engagement, yet a Lexile level is not designed to grade text along these lines.
For example, Skellig by David Almond is a wonderful story about a “strange kind of beast”, who befriends a 10-year-old boy. The creature drinks Newcastle Brown Ale, eats Chinese takeaway and is pretty objectionable. The boy is at a point of upheaval in his life; his sister is dying.
This text has a Lexile score of 490 - in the suggested range for children in Years 2, 3 or 4 in the UK or Grades 1, 2, or 3 in the US.
Undoubtedly, children of this age range could decode this marvellous story - the sentence length and frequency of the vocabulary are within their capability. But could a child in Year 2 comprehend the story in all its nuance? And would we consider the subject matter appropriate for children of this age?
In the end, although reading assessment tools are helpful, they can only ever be part of how we understand our pupils as readers.
When it comes to literacy, there will never be one simple solution. Perhaps, embracing the glorious complexity of it all is the way forward, bearing in mind that in reading - like many other parts of the curriculum - there is nothing more powerful than an expert teacher.
Megan Dixon is a doctoral student and associate lecturer at Sheffield Hallam University
