Do All Phonics Rules Need to Be Explicitly Taught?

In an article published in the October 2022 edition of Language Magazine, Professor Emeritus Stephen Krashen and Senior Researcher Jeff McQuillan argue that children gain knowledge of how sounds map onto letters through attempting to understand what they read and hear rather than explicitly learning phonics rules. To support their claim, they point to an obscure and complex rule that renders the b in comb and combing as silent but the b in combination as pronounced normally.

Here’s the rule: The b is silent if the word ends in the letters mb or if the word’s ending plays a grammatical role (e.g., the -ing in combing). If neither of these conditions are met, we pronounce it.

Most readers—even most highly skilled readers— do not know this rule; yet we read these words just fine.

Krashen and McQuillan are correct in stating that not all our knowledge of how sounds map onto letters is gained via explicit instruction. Perhaps the fact that these distinguished scholars deem it necessary to state the obvious points to a failure on the part of advocates for explicit, systematic phonics instruction (myself included here) to articulate our position with sufficient clarity and nuance.

To be clear, no serious reading researcher holds the position that all the phonics knowledge readers come to know is the result of explicitly teaching rules. It requires over 500 rules if you want to program a computer to correctly convert English text to speech (Gough & Juel, 1991). That’s simply too many rules to teach in too little time. In fact, the situation may be even worse than Krashen and McQuillan depict. For many words it’s not clear what the rule is, or which patterns are even rule governed to begin with. As Seidenberg et al. (2020) astutely observe, “Is spook an exception because of book and look, or rule governed because of spoon and spool?” (p. 6).

What’s in question then is how most skilled readers come to learn these distinctions if not through consciously learning rules. Krashen and McQuillan propose a subconscious process called acquisition. In Krashen’s (1982, 2003) prior work he defines the concept of acquisition as a person’s innate capacity to acquire or pick up a language. He contends that for acquisition to occur, two conditions must be met: 1) the language input received must be comprehensible, and 2) affective factors such has a learner’s high anxiety must not impede the input from reaching the language centers of the brain. According to Krashen, acquiring a second language is similar, if not identical, to acquiring one’s first language. And therein lies the problem with the case for acquired phonics. Any cognitive process causally linked to the innate human capacity to acquire language cannot in principle be the primary causal mechanism by which foundational literacy skills are learned. This is because an innate capacity to read and write does not exist.

We’re wired for language but not for reading.

Many archeologists, linguists, and paleontologists believe the capacity for full-fledged human language evolved around 50,000 years ago (Klein, 2017). It’s possible that more rudimentary forms of language may have cropped up many thousands of years before that (Pagel, 2017). As such, humans have had ample time to evolve the biological capacity that enables the natural acquisition of our first language. Writing systems appeared much later, around 5,000 years ago (Mark, 2011)—just a blink of an eye on an evolutionary timescale. The implications of the relative newness of writing to our cultural repertoire are clear: To become literate, we must make use of cognitive capacities evolved for other purposes. What this means in relation to phonics is that everyone who learns to read must forge new connections from the speech regions of the brain to the visual regions without the help of any innate prewiring to make the job easier.

How are foundational literacy skills developed without knowing all the rules?

David Share’s (1995) self-teaching hypothesis is key in explaining how children can learn to read a seemingly overwhelming number of novel word spellings without being explicitly taught all the rules. He argues that when readers decode, or sound out an unfamiliar word, it acts as a “self-teaching mechanism or built-in teacher” allowing readers to acquire the representations of a word’s written form from relatively few exposures (p. 155). It’s through these successful attempts at independently decoding novel letter strings that the reader bonds the visual form of words (i.e., their spellings) with their spoken and semantic forms. In short, attempting to sound out words in their entirety is for Share the sine qua non of reading: the essential condition. This stands in stark opposition to a three-cueing approach to word recognition, which would have readers rely more heavily on context- and meaning-based cues and much less on decoding.

Importantly, self-teaching is also involved in identifying words that can only be partially decoded through applying decoding rules (i.e., irregular words). For example, when attempting to decode a word like swamp, the reader might produce /swæmp/, which rhymes with lamp. Upon recognizing that /swæmp/ is not a real word, the reader must attempt an alternate pronunciation for the vowel sound. In doing so, she creates a match with a word she recognizes from her listening vocabulary—/swɑmp/. This process, called set for variability, has been found to predict English word reading and decoding for both emerging and more advanced readers (Tunmer & Chapman, 2012). It also underscores the critical role of oral language development because set for variability cannot happen if the reader does not have the partially decoded word in her oral language repertoire.

Although Share remains agnostic as to the underlying cognitive mechanisms behind self-teaching, a viable hypothesis supported by many prominent reading theorists has come in the form of statistical learning (Seidenberg & McClelland, 1989; Elleman et al., 2018). Statistical learning is a subconscious process whereby learners extract statistical regularities from the data. When applied to breaking the written code, the relevant data are printed words. Like Krashen’s concept of acquisition, statistical learning does not require explicit knowledge of phonics rules. However, unlike acquisition, it’s devoid of any untenable theoretical baggage which would hold that an innate language-specific capacity undergirds early reading development. Instead, statistical learning relies on general-cognitive learning mechanisms and therefore aligns with the prevailing neurolinguistic research that maintains becoming literate is not a natural process (Dehaene, 2009).

What about explicit, systematic instruction?

Krashen and McQuillan seem to argue that because not all phonics knowledge is consciously learned, explicit, systematic phonics instruction is unnecessary. This is false. In fact, the overwhelming evidence from the reading research community supports the exact opposite conclusion; explicitly and systematically teaching students how sounds connect to print is necessary for many children to gain a foothold on learning to read and write (Adams, 1990; Snow et al., 1998; NRP, 2000). This makes sense when we consider that even a rudimentary capacity for self-teaching depends on letter-sound knowledge, basic phonemic awareness, and the ability to determine the pronunciations of words based on partial decodings—i.e., set for variability (Share, 1995). Moreover, researchers suggest that targeted explicit instruction may even enhance statistical learning at more advanced stages of decoding and word recognition development (Compton et. al., 2014).

What about children who learn to read with little to no instruction?

There are children who seemly pick up reading on their own. In fact, various case studies exist that document this phenomenon. Krashen and McQuillan refer to several of them in their article. Here, it’s vital that educators understand that these examples of precocious readers are the exception not the rule. Estimates range anywhere from 1 to 2.5% of children can learn to read with little to no formal instruction (Margrain, 2005; Olson et al., 2006). This leaves most children in need of some degree of explicit, systematic code-oriented support.

Embracing nuance

I understand and in part sympathize with certain researchers’ incredulity upon being confronted with the erroneous claim that every phonics rule should or even could be explicitly taught to every student. However, arguing against a straw man version of the pro-phonics position is an affront to those in the field committed to understanding the nuances of the issue. Yes, effective code-oriented instruction does involve explicitly and systematically teaching a subset of rules, but that’s far from all. It also entails teaching skills such as phoneme identification, awareness, and pronunciation, encouraging linguistic flexibility or set for variability, and providing ample opportunities for children to build fluency through various evidence-informed practices. Research over the past 40 years demonstrates that this explicit, systematic instruction substantially aids students as they make those critical connections from speech to print thereby setting them well on their way to extracting statistical regularities from print on their own.

Furthermore, a comprehensive approach to teaching foundational skills isn’t completely phonocentric; that is, educators should consider how the English spelling system often prioritizes morphemes (i.e., the smallest meaningful units of language) over phonemes (i.e., individual speech sounds). For example, although the ending sound(s) in the words wanted, framed, and finished are pronounced /əd/, /d/, and /t/ respectively, the spelling remains the same so as to preserve the integrity of the past-tense morpheme -ed. Explicit, systematic foundational skills instruction will incorporate this and other important principles of English orthography (see here for more on this). Still even more nuance is required when discussing how much instructional time to devote to foundational skills, in which grades, and for how long.

As we continue to have these important conversations, it’s imperative we resist clinging to convenient oversimplifications and false binaries that have for too long characterized the debate around effective reading instruction. Moving forward, let’s applaud the immense progress that has been made in recent years aligning instruction with the crucial insights from reading science, and at the same time soberly acknowledge the difficult work that lies ahead. Translating scientific findings into effective educational practice is hardly a straightforward endeavor. There is still much to learn and much nuance to explore. It’s my hope that as practitioners we will embrace this nuance together with both a sense of humility and an unwavering commitment to developing proficient readers.

Resources

Adams, M.J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press.

Compton, D.L., Miller, A.C., Elleman, A.M., & Steacy, L.M. (2014). Have we forsaken reading theory in the name of “quick fix? Intervention for children with reading disability? Scientific Studies of Reading, 18, 55-73.

Dehaene, S. (2009). Reading in the brain. New York: Penguin Group.

Elleman, A.M., Steacy, L.M., & Compton, D.L. (2018). The role of statistical learning in word reading and spelling development. Scientific Studies of Reading, 1, 1-7.

Gough, P.B., Juel, C. (1991). The first stages of word recognition. In L. Rieben & C.A. Perfetti (Eds.), Learning to read: Basic research and its implications (p. 47-56). Hillsdale, NJ: Erlbaum.

Klein, R.G. (2017). Language and human evolution. Journal of Neurolinguistics, 43(Part B), 204-221.

Krashen, S. D. (1982). Principles and practice in second language acquisition. Oxford: Pergamon Press.

Krashen, S.D. (2003). Explorations in language acquisition. Portsmouth, NH: Heinemann.

Krashen, S.D. & McQuillan, J. (2022, October). The case for acquired phonics. Language Magazine, Vol. 22(2), 19-22.

Margrain, V.G. (2005). Case studies of spontaneous learning, self-regulation and social support in the early years. [Doctoral Dissertation, Victoria University of Wellington]. Retrieved from http://researcharchive.vuw.ac.nz/handle/10063/481.

Mark, J. (2011). Writing. World History Encyclopedia. http://www.ancient.eu/writing/

National Reading Panel. (2000). Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction.

Washington, DC: National Institute of Child Health and Human Development.

Olson, L., Evans, J., & Keckler, W. (2006). Precocious readers: Past, present, and future. Journal of the Education of the Gifted, 30(2), 205-235.

Pagel, M. (2017). Q&A: What is human language, when did it evolve and why should we care?. BMC Biol 15, 64 https://doi.org/10.1186/s12915-017-0405-3.

Seidenberg, M.S., & McClelland, J.L. (1989). A distributed, developmental model of word recognition and naming. Psychological Review, 96, 523-568.

Seidenberg, M.S., Borkenhagen, M.C., & Kearns, D.M. (2020). Lost in translation? Challenges in connecting reading science and educational practice. Reading Research Quarterly, 0(0), 1-12.

Share, D.L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, 55, 151-218.

Snow, C.E., Burns, M.S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young children. Washington DC: national Academy Press.

Tunmer WE, & Chapman JW (2012). Does set for variability mediate the influence of vocabulary knowledge on the development of word recognition skills? Scientific Studies of Reading, 16(2), 122–140