Some thoughts on “The Myth of Normal Reading” (Huettig & Ferreira, 2022)
Last week, Perspectives on Psychological Science published an article that left me mystified, hardly knowing where to begin. In “The Myth of Normal Reading” psycholinguists Falk Huettig and Fernanda Ferreira argue that the psychological and educational sciences should stop chasing the “phantom of normal reading behavior”. They make a series of claims about the state of psycholinguistic reading research, and from these draw out a number of implications regarding reading research and practice.
I do not recognise the characterisation of the literature that the authors offer. It is not unusual to have differences of opinion or perspective, but what is unusual is to see a published article that is so devoid of references or other evidence to support the claims being made. There is plenty of poor-quality research in the literature, but because this article strays into vital areas of practice (specifically, dyslexia diagnosis and reading instruction), it needs a response.
The authors argue “either implicitly or explicitly, the fields of psycholinguistics and language development have generally assumed that there is something called ‘normal reading’”.
They highlight three dimensions on which there is important variation in reading.
There are different modes of reading; for example, skim reading vs. deep reading, reading in different formats, and reading in different genres. They argue that “…the field should cease calling this normal reading….”
The process of reading differs across different writing systems. They argue that “…English is an outlier in terms of both linguistic properties and orthography….findings from the reading of English do not necessarily generalize….”.
There are substantial individual differences across readers; for example, poor vs high proficiency, children vs adults, monolingual vs multilingual. They argue that the large individual differences (with readers using “…a spectrum of styles and strategies”) means that “…it makes little sense to label any of them abnormal.”
The authors imply that reading is so variable (across individuals, writing systems, and modes) that it’s hard to say much about it; and further, that this variation has been largely ignored.
The authors don’t offer any evidence for the claim that we’re chasing a “phantom of normal reading”, so I conducted a Scopus search of every article published in the last 100 years that included the words “normal reading” in the title or abstract. This yielded 591 articles, of which 98 were not relevant to reading behaviour. The remaining articles fell largely into three categories.
Articles characterising dyslexic or low proficiency readers (224 article). These articles included studies that compared participants with a developmental or acquired disorder against matched participants in the “normal reading” range or without a disorder. They also included studies characterising reading behaviour in these disorders against a “normal reading” standard, and highlighted areas of unimpaired (“normal reading”) performance in cases of acquired brain injury. Finally, several articles reported case studies of acquired brain injury that had implications for understanding of the “normal reading” system.
Articles characterising reading behaviour in different laboratory paradigms (103 articles). These articles examined a variety of reading measures under different presentation modes (e.g. mirror reversal, RSVP, visual field manipulations) or under different task instructions (e.g. skimming, presence of a concurrent task) against “normal reading” conditions.
Articles referring to typical reading behaviour (126 articles). Many of the articles in this group referred to the use of eye-tracking in sentence reading as indicative of “normal reading”. Others used the words “normal reading” in statements about reading behaviours regarded as typical (including an article by one of the authors; Apel, Henderson & Ferreira, 2012).
This analysis provides only limited evidence for the claim that we are “chasing a phantom”. There are not very many abstracts in the past 100 years using the term “normal reading” (compare “dyslexia” with 17,845 articles). Moreover, most of these articles use the term in studies of the precise variation across individuals and modes that the authors suggest is being ignored!
Indeed, there are rich traditions of reading research (going back decades!) probing individual differences, stimulus and presentation modes, and language diversity on different reading behaviours. (Prominent reading researchers that publish non-English studies include e.g. Baayen, Beyersmann, Brysbaert, Caravolas, Carreiras, Cohen, Coltheart, Crepaldi, Dehaene, Ellis, Friedmann, Frost, Goswami, Grainger, Huettig, Hulme, Jared, Joanisse, Kinoshita, Kuperman, Manis, Marelli, Melby-Lervag, Liversedge, Perea, Perfetti, Ramus, Rastle, Richlan, Rueckl, Schroeder, Seidenberg, Siegelman, Share, Taft, Tzeng, Zevin, Ziegler, Zorzi). Science is always a work in progress, but this vast body of research does allow us to make generalisations about reading behaviours that are appropriately contextualised across individuals, modes, and languages. I can only assume the authors are unaware of this body of research.
The authors draw out four “implications” from their characterisation of the literature.
The first two deal with reading research; namely, that authors should be more precise about task goals and demands in their experiments, and that greater resources should be devoted to understanding reading in languages and writing systems other than English. I have no quibble with these recommendations. The second two deal with the practice of dyslexia diagnosis and reading instruction, and these are much more problematic.
The authors argue that there is such great variability in reading across individuals, reading modes, and reading goals that it is not appropriate to judge someone against a “...problematic standard of normal reading”. Hence, we should “…stop stigmatizing individuals who read ... differently”; for example, “We must stop labeling children dyslexic in 1st or 2nd grade. Such labels stick, affect children’s confidence, and are counterproductive and misplaced”. These are big claims that have potential to affect millions of children and families, and the authors offer no evidence to support them. There are vast datasets that allow practitioners to judge where a reader sits along the normal distribution on a variety of reading measures; this is the whole point of standardised assessments. Thus, the claim that we cannot judge a reader against a normal standard is preposterous. Moreover, the reason for identifying children as dyslexic is not to “stigmatize” them but to ensure that they get the support needed in their reading and wider educational experiences.
The authors go on to argue that we need “radically increased focus on reading comprehension when teaching reading”, and that “The focus should not be … on how well people decode … but instead on what people comprehend given their own stated goals.” Once again, these are big claims and they come out of nowhere. The authors don’t offer any evidence for them, and there is no obvious link back to their observations about the state of the literature. The authors seem to be unaware of the vast body of research showing that decoding is the proximal cause of reading comprehension in alphabetic writing systems; and that poor reading comprehension in these writing systems is typically due to some combination of poor decoding and poor oral language. That’s the reason why so many hundreds of researchers and practitioners have for years been focused on improving phonics instruction (including in developing nations where poor reading is holding back whole economies; Stern et al., 2021).
This article was written by two senior scientists that I respect. However, this article is essentially an opinion piece that deals with issues of the greatest substance based on very little evidence outside of the authors’ own narrow experience. I’m very surprised to see something like this published in Perspectives on Psychological Science.
*I’m grateful to Becky Crowley for help in coding abstracts from the Scopus analysis and to Max Coltheart for discussion*
Apel, J. K., Henderson, J. M., & Ferreira, F. (2012). Targeting regressions: Do readers pay attention to the left? Psychonomic Bulletin & Review, 19(6), 1108–1113. https://doi.org/10.3758/s13423-012-0291-1
Huettig, F., & Ferreira, F. (2022). The myth of normal reading. Perspectives on Psychological Science. https://doi.org/10.1177/17456916221127226
Stern, J. et al. (2021). Learning at Scale: Interim report. Research Triangle Institute.