Teaching word-level reading

Two significant papers were published in the last fortnight. They were:

Buckingham, J. (2020). Systematic phonics instruction belongs in evidence-based reading programs: A response to Bowers. The Educational and Developmental Psychologist, 1-9.

Colenbrander, D., Wang, H. C., Arrow, T., & Castles, A. (2020). Teaching irregular words: What we know, what we don’t know, and where we can go from here. The Educational and Developmental Psychologist, 1-8.

The first paper by Jennifer Buckingham is a clear, comprehensive overview of the research evidence for systematic phonics instruction. The second paper by the Macquarie University Centre for Reading explains the research base that exists for irregular word teaching, that is, words for which phonics instruction is not helpful. I highly recommend both papers; together they establish a helpful foundation for word-level reading instruction.

The Simple View of Reading (Gough & Tunmer, 1986) proposes that Reading Comprehension is the product of Linguistic Comprehension and Decoding. The formula is RC = LC x D. I don’t particularly like D because to me it implies word reading/word recognition (WR) comes from decoding alone. This is not the case. There are decodable words which can be learned via systematic phonics instruction and there are irregular words which need to be learned through other methods. Students also need good phonemic awareness, and a lot of practice for fluency and orthographic mapping to occur, ideally resulting in accurate, automatic word reading.

In this blogpost I am only going to cover one aspect of the Simple View of Reading, the process of learning D or WR. I have written about the meaning of words in a previous blogpost, including how we can approach teaching vocabulary for reading comprehension.

How regular is English?

In 1966, Hanna, Hanna, Hodges and Rudorf published ‘Phoneme-grapheme correspondences as cues to spelling improvement’. Based on analysis of over 17,000 words, they concluded that about 50% of words in English are predictable or regular based on phoneme-grapheme correspondences (PGCs). A further 34% are predictable except for one irregularity, most commonly a vowel. Another 12% are quite irregular in terms of English PGCs but they are predictable if students are taught morphology and etymology. That leaves 4% of words which are truly irregular and unpredictable; no amount of phonics instruction will support students to read these words. Many others have written about this since. One example is here and the complete reference is:

Berndt, R. S., Reggia, J. A., & Mitchum, C. C. (1987). Empirically derived probabilities for grapheme-to-phoneme correspondences in English. Behavior Research Methods, Instruments, & Computers, 19(1), 1-9.

Some examples of applying PGC knowledge with hope or naivety to irregular words from my early childhood were:

‘Colonel’ which I used to read as /kəloʊnəl/, /kəlɒnəl/ and /kɒlɒnəl/

‘sachet’ which I used to say as /sætʃət/

‘lasagne’ and ‘chef’ which my Nana hilariously and forever said as /ləsagni/ or /ləsandʒ/ and /tʃɛf/ despite endless correction

Given we have words that are decodable and words that are not, it is important we consider both in our word-level reading instruction. This quote from Colenbrander, Wang, Arrow and Castles (2020, p.1) is helpful:

“.. in English, the relationship between letters (also known as graphemes, which can be single or multiletter; e.g., ‘t’ as in cat, ‘ch’ as in chip) and speech sounds (phonemes) is not always regular or one-to-one. Therefore, some words would not be pronounced correctly if sounded out using regular grapheme-phoneme correspondences (e.g., does, were, yacht). These words are often referred to as irregular words, although in reality, regularity lies on a continuum from completely regular words (e.g., dog), to words with only one irregular grapheme-phoneme correspondence (e.g., find), to highly irregular words with many irregular correspondences (e.g., aisle, meringue).”

Teaching decodable words

Systematic phonics instruction sets children up to be able to decode (read) approximately 84% of words. Systematic phonics instructional programs cover one-to-one (direct) PGCs as well as those PGCs that are less regular but decodable if patterns are learned.

Phonics instruction should include systematically teaching how our 44 speech sounds are represented in print. This starts with simple or regular correspondences and moves through to complex or less regular correspondences, in words with increasingly complex word shapes (also known as phonotactics e.g. CVC, CCVC, CVCC, CCVCC etcetera) and length. There are over 200 correspondences to learn but the time and effort is worthwhile given through this we are unlocking most words for children.

As far as I know, there is no research evidence to suggest that there is any difference between teaching phonics via phoneme-grapheme correspondences (PGCs) or grapheme-phoneme correspondences (GPCs). In systematic phonics research, instructional methods that have included either approach have been found to have effect. Having said that, I do not know of any studies that have had a direct comparison between PGC and GPC instructional methods as their sole aim. I welcome any corrections about this and will update this post accordingly.

My personal preference (please note that is all that it is unless there is any research to support it) is to teach PGC, because we are then directly teaching speech (phoneme) to print (grapheme). For example, in a GPC approach we might be focusing on the relationship between ‘a’ the grapheme and /æ/ the phoneme. In a PGC approach, instead we would focus on /æ/ the speech sound and teach that we represent it using ‘a’ the grapheme. The /æ/ speech sound is always represented by the ‘a’ grapheme, while the ‘a’ grapheme does not always represent the speech sound /æ/. We must remember that letters or graphemes don’t have sounds, rather they are used on their own or in combination to represent phonemes.

Irregular words

“At present, there is research support for three broad areas of instruction, delivered alongside instruction in grapheme-phoneme correspondences: sight word instruction, mispronunciation correction, and morphology. It is likely that a comprehensive, systematic approach to instruction would involve instruction in all of these elements; however, research is required to support this, and to date no studies have attempted to tease apart the individual contributions of these different elements.” (Colenbrander, Wang, Arrow and Castles, 2020, p.6)

I cannot do justice to this paper (referenced at the outset) in a brief blogpost. Please read it to capture the nuance and detailed reviews of all relevant studies and approaches. In summary, some approaches to consider are:

Morphology and etymology

Teach the grammatical structure of words (root words and affixes).

Teach the history of word meanings also known as etymology e.g. ‘ch’ in French words is pronounced /ʃ/ [shhhh].

Mispronunciation/error correction

Provide correct pronunciation to students when they mispronounce.

Teach the many vowel variations.

Sight word (irregular word) instruction

Whole word memorisation of truly irregular words can be effective and may include students seeing, hearing, spelling or visualising the word with flashcard, picture or mnemonic supports. Note that this is for a small number of words. Most words can be taught through phonics instruction.

On that note, a common criticism of phonics instructional programs is that they prevent students from learning irregular high frequency words. This is a myth. All systematic phonics programs I have used or seen used teach high frequency irregular/sight words commensurately.

Assessment of word reading ability

Given word “regularity lies on a continuum” (Colenbrander, Wang, Arrow & Castles, 2020, p.1) we need to assess word reading accordingly. A very helpful assessment, available for free online, is the Castles & Coltheart 2 (CC2). It assesses single word reading of regular, irregular (sight) and nonsense words.

People often get confused or argumentative about the inclusion of nonwords in reading assessments. The key point is that we test nonword reading, we don’t teach nonword reading. Testing reading of nonwords is the only way to accurately and reliably assess how a child’s phonic knowledge is developing, without inference from their sight word memory. There is a great paper that summarises nonword reading assessment utility and you can read it here or the full reference is here:

Castles, A., Polito, V., Pritchard, S., Anandakumar, T., & Coltheart, M. (2018). Do nonword reading tests for children measure what we want them to? An analysis of year 2 error responses. Australian Journal of Learning Difficulties, 23(2), 153-165.

When reading a word, we use one of two routes as detailed in the Dual-Route Model of Reading. One route is the lexical route. We use the lexical route when we are reading the word by accessing a memory of a stored word that we have previously seen written. Words are stored in our lexical memory when we have been taught them as whole words (sight/irregular words) or when we have read and practiced words we have learned through decoding many times, and they have become so automatic as to be stored as whole units. The other route is the non-lexical route. We use this route when we are reading a word, be it regular, irregular or nonword, that is new or unfamiliar to us. We know we are using the non-lexical route when we are sounding out a word.

If you read the word ‘bat’ you are using the lexical route. You can’t help yourself. You read it automatically without effort as a whole word. Once upon a time, unless you were schooled entirely in Whole Language, you would have read this word, the first few times at least, using your non-lexical route, to sound it out. And over time, through multiple exposures and practice, it was stored as a whole word in your lexical memory for accurate and rapid retrieval.

If you read the words ‘impignorate’ or ‘jentacular’ you are likely using your non-lexical route as these words are new or unfamiliar to you. You can’t help but to sound them out based on your phonics (and perhaps morphology) knowledge.

Nonword assessment is assessing the strength of a child’s non-lexical route. This is an essential component of reading assessment, given through phonics instruction, we can unlock 84% of words for children.

In closing

Let’s get word-level reading assessment and instruction right. The two recent publications referenced at the outset provide us with the foundational knowledge required to teach regular and irregular word-level reading.

“If a child memorizes ten words, the child can read only ten words, but if the child learns the sounds of ten letters, the child will be able to read 350 three-sound words, 4,320 four-sound words, and 21,650 five-sound words.” (Kozloff, 2002)