From the Bulletin Learning Difficulties Australia www.ldaustralia.org Discussion and Debate

Evidence strongly favours systematic synthetic

Jennifer Buckingham they have found the ultimate solution programs is a and stop looking for better ones. false comparison. provides a carefully- However, it is very different to propose Both phonics and presented counterargument that teachers and politicians should comprehension to the recent suggestion consider using unproven ‘alternative’ instruction are methods. Teaching practice and necessary; a that the evidence in favour education policy should be based on the finding of a of systematic phonics best available evidence unless and until positive effect of it is superseded by new information and one on

Evidence strongly favours systematic synthetic systematic favours strongly | Evidence LDA Bulletin instruction is weak, and that new evidence. outcomes does early phonics instruction Bowers reviews major meta-analyses not prove that should be replaced by of studies that have looked at the effect of the other is unnecessary. Measures of systematic phonics instruction on various are measuring ‘Structured Word Inquiry’. reading outcomes. His key criticisms are both word identification and that the strength of the measured effects comprehension factors. In the early of systematic phonics are overstated and stages of reading development, that the studies do not directly compare word identification is the stronger systematic phonics with what he calls predictor of reading comprehension, ‘unsystematic phonics’. but once decoding is fluent, language However, Bowers’ interpretation of comprehension becomes more important the findings of these meta-analyses is (Garcia and Cain, 2014). not accurate. There is stronger evidence According to Bowers (2020), ithin the community of in favour of using systematic phonics in “systematic phonics explicitly research and practice reading instruction than not using it. teaches children - that is informed by correspondences prior to emphasizing scientific evidence, there What is systematic the meanings of written words in text (as isW a general acceptance that children in or balanced need to learn the alphabetic code in phonics? instruction) or the meaning of written order to be able to read accurately The broad term ‘systematic phonics’ words in isolation (as in morphological and for meaning, and that the most describes practices for the teaching of instruction).” (p. 3) effective way to teach the code is decoding and word reading. Evidence- This is incorrect. Systematic phonics through systematic and explicit phonics based understandings of systematic does not preclude a focus on the instruction (Castles et al., 2018). phonics place it within a comprehensive meaning of words. There is no directive A recent challenge to that consensus program of instruction that includes that learning grapheme-phoneme has come from Professor Jeffrey Bowers four additional essential elements correspondences (GPCs) must precede who has published a journal article that – , , all other elements of reading instruction. claims to show that “there is little or no and comprehension. Alone, The common recommendation that evidence that systematic phonics is better systematic phonics is not a fool proof instruction comes after a than the main alternative methods used guarantee of reading success and its period of systematic phonics instruction in schools, including whole language and effectiveness is mediated by the quality (the precise optimal time for this has balanced literacy” (Bowers, 2020, p. 1). of the rest of the literacy program. not yet been determined) is based on Furthermore, Bowers says, “Once this is Systematic phonics does not scientific evidence that the phonological understood, my hope is that researchers supplant or contradict the need for pathway for decoding words is essential and politicians will be more motivated to instruction that develops language for beginning readers. While implicit consider alternative methods.” comprehension. Therefore, comparing morphological understanding is evident Bowers is right to say that the effects of systematic phonics in young children’s oral language, researchers should never consider that instruction with comprehension-based childrens’ use of morphological

30 | Volume 52, No 1, June 2020 From the Bulletin Learning Difficulties Australia www.ldaustralia.org

knowledge in word reading is BulletinLDA | Evidence strongly favours systematic synthetic phonics demonstrated later (Rastle, 2019). Systematic phonics can include synthetic and analytic approaches, which differ in the unit of sub-word Study Effect size (Cohen’s d or Hedge’s g) analysis. Synthetic phonics begins Overall with – the smallest (2000)/ Ehri (2001) d = 0.67 (decoding regular words) sub-word level. Children learn the d = 0.60 (decoding pseudowords) associations between speech sounds d = 0.40 (irregular words) (phonemes) and the letters or letter d = 0.51 (reading comprehension) clusters that represent them in (), and that this is a reversible Type of phonics process. They learn to synthesise the Synthetic phonics d = 0.45 (average for all phonemes and graphemes to read measures) and spell words. Synthetic phonics d = 0.35 (average for all instruction has a defined sequence measures) for teaching grapheme-phoneme correspondences. Grade level Analytic phonics uses larger Kindergarten d = 0.56 (average for all measures) sub-word units such as onset-rime First grade d = 0.54 (average for all measures) for word analysis. For example, rather Grade 2-6 d = 0.27 (average for all measures) than learning to read the word rat as a Camilli, Vargas & Yurecko d = 0.24 (average for all measures) composition of three letters and sounds, (2003) r-a-t, children would learn that the word rat is in a ‘word family’ with the rime -at, Camilli, Wolfe & Smith (2006) d = 0.123 * phonics only instruction such as r-at, s-at, c-at, and so on. There are far fewer GPCs than Torgerson, Brooks & Hall d = 0.27 / 0.38 (fixed effects/random effects; there are ‘word families’ and learning (2006) word reading accuracy) phonics at the phoneme level is more d = 0.24 / 0.35 (fixed effects/random effects; systematic and efficient than onset-rime reading comprehension) families (Vousden et al., 2011). The Suggate (2010) d = 0.5 (average for all measures) vast majority of rimes can be read using d = 0.59 (pre-reading) their component GPCs (Brooks, 2015). d = 0.42 (reading) Knowledge of phonemes is a stronger d = 0.41 (comprehension) predictor of early reading acquisition d = 0.32 (average for all measures; follow up) than knowledge of rimes (Nation & Hulme, 1997). Adesope, Lavin, Thompson, & g = 0.40 Ungerleider (2011) Is Bowers’ Galuschka, Ise, Krick & g = 0.322 (average for all measures) interpretation of the Schulte-Korne (2014) meta-analyses fair? Suggate (2016) Post-test d = 0.44 (average for all measures) In his review of evidence on systematic d = 0.48 (pre-reading) phonics, Bowers looks in detail at d = 0.45 (reading skills) meta-analyses conducted over the past d = 0.48 (comprehension) twenty years, starting with the National Reading Panel (2000) later published as Follow up Ehri et al. (2001). d = 0.25 (average for all measures) Bowers argues that the effect sizes d = 0.26 (pre-reading) in these studies are not large and do d = 0.30 (reading skills) not justify the authors’ conclusions that d = -0.03 (comprehension) systematic phonics has the strongest evidence in its favour. However, the * included unpublished and published studies effect sizes in these studies are certainly ** 8 out of 22 phonics interventions were stronger than the evidence found for computer-based phonics training any other method, including whole language. Subsequent studies have McArthur et al (2018) d = 0.51 (mixed/regular word reading accuracy) added to the evidence in favour of d = 0.67 (nonword reading accuracy) including systematic phonics in reading d = 0.84 (irregular word reading accuracy) instruction (for example, Hjetland et al. d = 0.45 (mixed/regular word reading fluency) 2019). Detailed descriptions of these d = 0.39 (non-word reading fluency) meta-analyses and an explanation of the d = 0.28 (reading comprehension) flaws in Bowers’ interpretation of them is Table 1: Effect sizes for reading outcomes associated with systematic phonics instruction provided in Buckingham (2020).

Volume 52, No 1, June 2020 | 31 From the Bulletin Learning Difficulties Australia www.ldaustralia.org

Table 1 presents the main findings the quality of classroom instruction. bases in education. Synthetic phonics, of meta-analyses and systematic reviews Nevertheless, the overall effect size is which is the most systematic form of phonics instruction and intervention. invariably and significantly positive. of phonics instruction, has been These are the same studies reviewed specifically investigated in a number of by Bowers, with the exception of the What are randomised control trials (Christensen McArthur et al (2018) study which & Bowey, 2005; Hatcher, Hulme, & supersedes the earlier study included in the potential Snowling, 2004; Johnston, McGeown & Bowers (2020). The effect sizes relate “alternatives” to Watson, 2011) and has been found to to the difference in reading outcomes be a common factor in high performing associated with systematic phonics systematic phonics schools (Joseph, 2019; Louden, 2015; instruction/intervention as opposed to OFSTED, 2010). After the introduction of non-systematic or no phonics instruction. instruction? mandatory synthetic phonics instruction The most common interpretation of What are the alternative methods in 2006 and a phonics screening check in effect sizes is that proposed by Cohen to systematic phonics, and what is 2012 in all English primary schools, there (1969): 0.2 is small; 0.5 is medium or the likelihood that they will be more was an improvement in upper primary moderate; and 0.8 is large. Some of the effective? Bowers suggests that reading in national assessments and studies in the above table are reported instruction “should focus more on the early indications of gains in international as Hedges’ ‘g’. According to Torgerson role that meaning plays in organizing assessments (Buckingham, 2016; et al. (2018), the difference between (via morphology) and that Machen et al., 2018; Double et al., 2019). these types of effect size estimates is English system (sic) makes Synthetic phonics is strongly aligned minimal. By Cohen’s interpretation, the sense once the interrelation between with cognitive scientific research and effect sizes in the table fall mostly in the , morphology, and etymology models of reading that have been found moderate range. The outlier is Camilli are considered.” (p. 23). to be highly predictive – the Dual Route et al. (2006) who achieved a small Jeffrey Bowers’ brother Peter Cascading Model (of word reading) effect size by making multiple coding Bowers has developed such a program and the (for manipulations to the studies that are – Structured Word Inquiry (Bowers reading comprehension) in particular

Evidence strongly favours systematic synthetic phonics synthetic systematic favours strongly | Evidence LDA Bulletin methodologically debatable. & Bowers, 2008). Jeffrey Bowers has (Castles, Rastle & Nation, 2018). The However, a recent paper by Kraft co-authored papers with Peter Bowers same cannot be said for whole language, (2020) explains that these effect size on the rationale for SWI (Bowers & balanced literacy, or analytic phonics. classifications were devised from Bowers, 2017) as well as participated And while there is some validity clinical studies and makes a persuasive in evaluations of the program to the argument that meta-analyses empirical case that they are not (Colenbrander et al., 2018). provide a more accurate estimate of the appropriate for applied educational There is no problem with academics effect of an intervention, there is also research. Kraft proposes the following developing reading programs. Such a good argument to be made for giving effect size interpretations: <0.05 reading programs would naturally strong consideration to the findings is small; 0.05 – 0.2 is medium or be informed by the developers’ of individual studies that investigate a moderate; and >0.2 is large. Using understanding of the best available higher quality version of the intervention these interpretations, the effect sizes of evidence. The problem with positing of interest. Meta-analyses include using a systematic phonics program are Structured Word Inquiry (SWI) as a interventions that are short in duration, almost all very large. superior alternative to systematic phonics with small numbers, and restricted Bowers’ other key criticism, aside is that there are no studies showing that instructional scope and depth. Emphasis from the relative effect sizes, is what he SWI is effective for teaching beginning should also be given to the findings of regards to be weak evidence directly reading, either with or without the sort of larger studies with implementations comparing systematic phonics with comparison group that Bowers (2020) that more closely resemble what would ‘unsystematic’ phonics. Given the says is necessary to truly prove efficacy. generally be considered ideal classroom difficulty of classifying the comparison Evaluations of SWI do not compare it with practice, such as the Clackmannanshire conditions as unsystematic phonics, systematic phonics for initial instruction. study (Johnston et al., 2011). whole language (with or without Studies of SWI show that children Bowers’ thesis rests on the flawed unsystematic phonics), balanced can benefit from instruction in argument that when held up to the literacy, and rare ‘no phonics’ teaching, morphology and etymology after one or highest possible standards of evidence, it seems reasonable and practical more years of initial reading instruction systematic phonics falls short. It is to do what almost all studies and that includes phonics (Bowers & Kirby, therefore illogical to suggest using meta-analyses have done – compare 2010; Devonshire & Fluck, 2010; “alternative teaching methods” that the presence of systematic phonics Devonshire et al. 2013; Colenbrander et have either much weaker evidence or no instruction with the absence of al., 2018). They do not provide evidence evidence base whatsoever. systematic phonics instruction. to support the argument that instruction It is one thing to say that The available evidence from multiple based on morphology and etymology researchers should consider studies shows that reading instruction could or should be an alternative to investigating unproven alternative that includes systematic phonics is systematic phonics in the initial stages methods, but it is irresponsible to more effective than instruction that does of learning to read. make the same recommendation for not. The range of effect sizes is due to There is strong evidence for the teachers. Classroom practice should numerous factors, including the duration, inclusion of systematic phonics in initial use the methods with the strongest level of systematicity, intensity, age of reading instruction. evidence available base, and at the students, beginning level of students, Systematic phonics has one of the moment that is undeniably systematic group size, instructional fidelity, and largest and most consistent evidence synthetic phonics.

32 | Volume 52, No 1, June 2020 From the Bulletin Learning Difficulties Australia www.ldaustralia.org

References Reading acquisition from novice to influence the strength of the relationship BulletinLDA | Evidence strongly favours systematic synthetic phonics Adesope, O. O., Lavin, T., Thompson, T., expert. Psychological Science in the in English. Review of Educational & Ungerleider, C. (2011). Pedagogical Public Interest, 19, 5-51. https://doi. Research, 84(1), 74-111. https://doi. strategies for teaching literacy to org/10.1177/1529100618772271 org/10.3102/0034654313499616 ESL immigrant students: a meta- Christensen, C. A. & Bowey J. A. Hatcher, P. J., Hulme, C. & Snowling, analysis. British Journal of Educational (2005). The efficacy of orthographic M. J. (2004). Explicit phoneme Psychology, 81(Pt 4), 629–653. rime, grapheme–phoneme training combined with phonic reading https://doi.org/10.1111/j.2044- correspondence, and implicit phonics instruction helps young children at risk 8279.2010.02015.x approaches to teaching decoding skills. of reading failure. The Journal of Child Bowers, J. S. (2020). Reconsidering the Scientific Studies of Reading, 9(4), Psychology and Psychiatry, 45(2), 338- 327–349. https://doi.org/10.1207/ evidence that systematic phonics is 358. s1532799xssr0904_1 more effective than alternative methods Hjetland, H. N., Lervåg, A., Lyster, S.-A. of reading instruction. Educational Cohen, J. (1969). Statistical power H., Hagtvet, B. E., Hulme, C., & Melby- Psychology Review, Online first. https:// analysis for the behavioral sciences. Lervåg, M. (2019). Pathways to reading doi.org/10.1007/s10648-019-09515-y Academic Press. comprehension: A longitudinal study Bowers, S., & Bowers, P. (2008). Colenbrander, D., Parsons, L., from 4 to 9 years of age. Journal of Understanding SWI: “Structured Murphy, S., Hon, Q., Bowers, J. & Educational Psychology. 11(5), 751-763. word inquiry” or “scientific Davis, C. (2018). Morphological http://dx.doi.org/10.1037/edu0000321 word investigation”. http://www. intervention for children with reading Johnston, R., McGeown, S. & Watson, J. wordworkskingston.com/WordWorks/ and spelling difficulties. https://www. (2011). Long-term effects of synthetic Structured_Word_Inquiry.html nuffieldfoundation.org/wp-content/ versus analytic phonics teaching on the uploads/2019/11/Davis204217920- Bowers, J. S., & Bowers, P. N. (2017). reading and spelling ability of 10 year 20Colenbrander20SSSR202018.pdf Beyond phonics: The case for teaching old boys and girls. Reading and Writing: children the logic of the English spelling Devonshire, V., & Fluck, M. (2010). An Interdisciplinary Journal, 25(6), system. Educational Psychologist, 52, Spelling development: Fine-tuning 1365–1384. 124–141. https://doi.org/10.1080/0046 strategy use and capitalising on Joseph, B. (2019). Overcoming the odds: 1520.2017.1288571 the connections between words. A study of Australia’s top performing Learning and Instruction, 20(5), Bowers, P. N., & Kirby, J. R. (2010). disadvantaged schools. Research Report 361-371. https://doi.org/10.1016/j. Effects of morphological instruction 39. The Centre for Independent Studies. learninstruc.2009.02.025 on vocabulary acquisition. Reading Kraft, M. A. (2020). Interpreting effect and Writing: An Interdisciplinary Devonshire, V., Morris, P., & Fluck, sizes of education interventions. Journal, 23(5), 515–537. https://doi. M. (2013). Spelling and reading Educational Researcher, org/10.1007/s11145-009-9172-z development: The effect of teaching 49(4), 241-253. https://doi. children multiple levels of representation Brooks, G. (2015). Dictionary of the org/10.3102/0013189X20912798 in their . British English Spelling System. Open Louden, B. (2015). High performing Double, K. S., McGrane, J. A., Stiff, Book Publishers. schools: What do they have in common? J. C., & Hopfenbeck, T. (2019). Buckingham, J. (2016). Focus on Western Australia Department of The importance of early phonics phonics: why Australia should adopt Education. improvements for predicting later the year 1 phonics screening check. reading comprehension. British Machin, S., McNally, S., & Viarengo, The Centre for Independent Studies. Educational Research Journal, 45(6), M. (2018). Changing how literacy is Research Report 22. https://doi. 1220-1234. https://doi.org/10.1002/ taught: Evidence on synthetic phonics. org/10.13140/RG.2.2.35472.61440 berj.3559 American Economic Journal: Economic Buckingham, J. (2020). Systematic Policy, 10, 217–241. https://doi. Ehri, L. C., Nunes, S. R., Stahl, S. A., org/10.1257/pol.20160514 phonics belongs in evidence-based & Willows, D. M. (2001). Systematic reading programs: A response phonics instruction helps students McArthur, G., Sheehan, Y., Badcock, to Bowers. The Educational and learn to read: evidence from the N.A., Francis, D.A., Wang, H.C., Developmental Psychologist, Online first. National Reading Panel’s meta- Kohnen, S., Banales, E., Anandakumar, https://doi.org/10.1017/edp.2020.12 analysis. Review of Educational T., Marinus, E., Castles, A. (2018). Camilli, G., Vargan, S., & Yurecko, M. Research, 71, 393–447. https://doi. Phonics training for English-speaking (2003). Teaching children to read: The org/10.3102/00346543071003393 poor readers. Cochrane Database of Systematic Reviews 2018, Issue fragile link between science and federal Galuschka, K., Ise, E., Krick, K., & 11. Article No.: CD009115. DOI: education policy. Education Policy Schulte-Körne, G. (2014). Effectiveness 10.1002/14651858.CD009115.pub3. Analysis Archives, 11(15), 1–51. https:// of treatment approaches for children doi.org/10.14507/epaa.v11n15.2003 and adolescents with reading Nation, K., & Hulme, C. (1997). Camilli, G., Wolfe, M., & Smith, M. L. disabilities: A meta-analysis of Phonemic segmentation, not onset-rime (2006). Meta-analysis and reading randomized controlled trials. PLoS One, segmentation, predicts early reading policy: Perspectives on teaching 9(2), e89900. https://doi.org/10.1371/ and spelling skills. Reading Research children to read. The Elementary School journal.pone.0089900 Quarterly, 32(2), 154-167. https://doi. Journal, 107(1), 27–36. https://doi. Garcia, J. G., & Cain, K. (2014). org/10.1598/RRQ.32.2.2 org/10.1086/509525 Decoding and reading comprehension: National Reading Panel. (2000). Castles, A., Rastle, K., & Nation, K. A meta-analysis to identify which Teaching children to read: an evidence- (2018). Ending the reading wars: readers and assessment characteristics based assessment of the scientific

Volume 52, No 1, June 2020 | 33 From the Bulletin Learning Difficulties Australia www.ldaustralia.org

research on reading and its implications for reading instruction. Bethesda, MD: National Institute of Child Health and Human Development. OFSTED (2010). Reading by six: How the best schools do it. UK Office for Standards in Education, Children’s Services and Skills. Rastle, K. (2019). The place of morphology in learning to read in English. Cortex, 116(July 2019), 45-64. https://doi.org/10.1016/j. cortex.2018.02.008 Suggate, S. P. (2010). Why what we teach depends on when: Grade and reading intervention modality moderate effect size. Developmental Psychology, 46(6), 1556–1579. https://doi. org/10.1037/a0020612 Suggate, S. P. (2016). A meta- analysis of the long-term effects of phonemic awareness, phonics, fluency, and reading comprehension interventions. Journal of Learning Disabilities, 49(1), 77–96. https://doi.

Evidence strongly favours systematic synthetic phonics synthetic systematic favours strongly | Evidence LDA Bulletin org/10.1177/0022219414528540 Torgerson, C. J., Brooks, G., & Hall, J. (2006). A systematic review of the research literature on the use of phonics in the teaching of reading and spelling (DfES Research Report 711). Department for Education and Skills, University of Sheffield. Torgerson, C., Brooks, G., Gascoine, L., & Higgins, S. (2018). Phonics: Reading policy and the evidence of effectiveness from a systematic ‘tertiary’ review. Research Papers in Education, 1-31. https://doi.org/10.1080/02671522.201 7.1420816 Vousden, J. I., Ellefson, M. R., Solity, J., & Chater, N. (2011). Simplifying reading: Applying the simplicity principle to reading. Cognitive Science, 35(1), 34-78. https://doi.org/10.1111/j.1551- 6709.2010.01134.x

Dr Jennifer Buckingham is Director of Strategy and Senior Research Fellow at MultiLit. She is also the director of the Five from Five reading project.

34 | Volume 52, No 1, June 2020