It is a truth, universally acknowledged, that a good reader will make predictions whilst reading a text. As a result, predicting has cemented itself as an essential reading comprehension strategy and pupils are thus encouraged to predict or guess what might happen next in a text and teachers encouraged to promote prediction.
It is, thus, perhaps surprising that this fundament of reading pedagogy is founded on very little research evidence and fails to comply with the essential psychology of reading. In fact, it may be that good readers actually don’t make predictions and guesses when reading text.
The strategy of ‘predicting’ appeared in the US with the development of ‘Multiple Strategies Instruction’ in the 1980s (Wilkinson and Son, 2011) under the auspices of Collaborative Strategic Reading. This programme focused on four comprehension strategies and encouraged pupils to collaborate when employing the respective teacher-designated strategy. One of the strategies was ‘brainstorming and predicting’, whereby groups of pupils worked together to guess and predict what might happen next in a story. Although the research found that, like all multiple strategies, CSR appeared to have beneficial effects for statutory testing (Kilinger and Vaugh, 1996), the individual strategies were not evaluated for efficacy, and it appeared that the ‘brainstorming and predicting’ element was often allocated to the poorer readers as the activity could always be attempted and suggestions were seldom investigated for appropriateness (Vaugh et al., 1998). It is also apposite to refer to Garcia et al.’s (2007) observation that teachers often ‘get stuck’ on a particular single strategy rather than encouraging a meaningful dialogue around a text. This seemed to be the case with predicting. (more on comprehension here)
Research into the development of comprehension makes almost no reference to the skill of prediction as an indicator of future reading comprehension outcomes. Word recognition and vocabulary (Nation and Snowling, 2004) inevitably feature heavily but so do genre knowledge (Kendeou et a., 2006), the knowledge of story structure (Cain et al., 2004), the ability to make inferences from text (Cain et al., 2004), greater teacher-managed instruction (Connor et al., 2004) and content area knowledge (Cain and Oakhill, 2006). Prediction as a skill appears nowhere in Beck, McKeown, Hamilton and Kucan’s (1997) study of narrative text comprehension. They focus on authorial intent and readers’ understanding of character creation and author fallibility. Likewise, Elder and Paul (2006) concentrate teaching comprehension instruction upon the immediately read text and its relationship with earlier read text – not relating it to predicted future text. It could be argued, however, that these approaches relate to New Criticism (Eaglestone, 2009) and are intrinsic approaches to literature -close reading being the most common. However, extrinsic approaches (historicism, revelatory aspects of literature and psychoanalysis of text and authors) are entirely post-analytical and thus preclude any prediction of story.
Predicting would seem to have little usefulness in non-fiction texts which are expository and are intended to be as clear as possible (academic papers aside). But surely readers do predict when engaging with narrative texts and are encouraged to do so by authors’ use of foreshadowing, symbolism, allegory, metaphor, pathetic fallacy and other tropes? It feels intuitively so right, yes? It would seem not. Psychology research seems unanimous that there is no clear line between story and non-story (as there is between sentence and non-sentence) and although violating story structure order (setting, theme, plot, resolution) makes narrative less easy to comprehend, readers can puzzle out stories presented in twisted ways (Rayner et al., 2013). It is readers’ schema and mental models of narrative structures and real-world knowledge that enables readers to understand discourses (Alba and Hasher, 1983) and authors to affect them (Turner, 1979). Readers, it would seem, are not predicting when reading, they are accessing their understanding of story structure, narrative, genre and wider world knowledge and applying this to the discourse through their mental models and schema. Authors occasionally encourage readers to predict, but this is usually to ensure that the prediction is wrong - as in a whodunnit or a genre shift.
So, it would appear that good readers have extensive knowledge of the world, context, story structure and genres and tend not to predict but analyse the current text being addressed and relate this to the text previously read; not the text about to be read. In terms of teaching, it would seem, therefore, that analysing the text being read and how it relates to previously text, along with exposing pupils to a rich variety of literature, world knowledge, text structures and genres is a more efficacious approach.
Follow The Reading Ape on twitter - @thereadingape.