I watched an eight-year-old boy ‘read’ a page of poetry from his Hindi textbook. He spoke fluently, eyes moving across the page, finger tracking the line. Taking the book from him, I opened a page randomly and asked him to read. The child was silent. He had not been reading at all, but had memorized the poem and was performing it using the picture on the page as a cue. He could barely read from the random page.
This is not unusual in Indian schools. It is what happens when an abstraction becomes so attractive that it becomes the dominant orthodoxy. This is another cautionary tale from education, building on my last column where I wrote about another such dogma—constructivism.
This idea is called the ‘whole language’ approach to learning how to read and write. Its intellectual core, associated with American scholars Kenneth Goodman and Frank Smith from the late 1960s, went something like this: children learn to speak naturally without formal instruction, simply by being immersed in meaningful talk; therefore, reading must be learnt the same way—by immersion in meaningful, engaging text.
Reading, in Goodman’s phrase, was a “psycholinguistic guessing game”: the skilled reader does not laboriously decode letters but samples the text, predicts and constructs meaning from context.
From this followed the pedagogy: surround children with rich stories and let them guess unfamiliar words from pictures and context. Do not drill them with the dull mechanics of letters and sounds, because that kills both meaning and the love of reading.
Much of this is true. Children do need meaningful and engaging text. Mindless drilling kills the love of reading before it is born; and good readers are so because they understand. As a description of the ‘purpose’ of reading that informs some aspects of pedagogy, the whole language approach is relevant.
But the fatal move was our slide from a partial theory and description to a prescription of how reading must be ‘taught’ to the exclusion of all other considerations—specifically that the systematic and explicit teaching of the code connecting sounds to script was unnecessary, even harmful.
On this point, the science turned out crystal clear. Decades of research in education, cognitive psychology and neuroscience showed that skilled text readers do not guess from context, they decode rapidly and unconsciously. It is precisely this automatic decoding that frees the mind to process meaning. Guessing from context and pictures is not what good readers do, but what poor readers are forced to do.
Whole language classrooms were in effect training children in the habits of struggling readers. Speech is natural to us, but script is a human invention a few thousand years old; the brain must be explicitly taught to crack its code. The analogy at the heart of the theory—reading is like speaking—is simply false.
The US ran this experiment at scale and paid for it. California adopted whole language in 1987. Within years, its scores on reading had collapsed to nearly the bottom of the country.
The so-called ‘reading wars’ in the US raged for a generation. The National Reading Panel, after reviewing the accumulated evidence, concluded in 2000 that systematic phonics instruction was essential.
Yet, the whole language dogma survived under a new name—‘balanced literacy,’ and its guessing strategies persisted in classrooms for two more decades. Eventually, the most influential of whole-language advocates publicly conceded new findings and schools in the US abandoned it.
Two generations of children in American public schools, disproportionately from poor communities, bore the cost of an attractive abstraction-turned-dogma.
Why should this concern us in India? Because the idea travelled. This view of early literacy—suspicious of the alphabet chart and of anything that smells of ‘mechanical decoding’—got the aura of settled progressive wisdom, affecting our teacher education and curriculum.
The irony is that our scripts are far better suited to explicit code teaching than English, whose spelling is notoriously treacherous. The akshara (syllable)-based scripts of most Indian languages are largely transparent—what’s written is what is said. A child systematically taught the aksharas and their combinations holds the key to nearly every word in the language. To withhold that key in the name of protecting meaning is not progressive, but regressive.
As with constructivism, the effects of this dogma are unevenly distributed. The child in a book-filled home, read aloud to every night, will likely also learn the reading code from the attention showered by adults. The first- or second-generation learner for whom school is the only encounter with written language, while the parent does manual labour, is unlikely to learn that code.
Fortunately for our country, over the past few years our direction has been changing, with a clear policy and curricular effort aimed at teaching the code explicitly, systematically and early—without abandoning the idea that meaning matters. Learning letters as codes for sounds is the means, while understanding meaning is the end. This is neither a compromise between doctrines, nor a new dogma, but what science has discovered and how good teachers have always taught.
The author is CEO of Azim Premji Foundation.