Commentary #3

English is a difficult writing system for children to learn

J. Richard Hanley

Source: Hanley, J. R. (2010) ‘English is a difficult writing system for children to learn: evidence from children learning to read in Wales’ in Hall, K., Goswami, U., Harrison, C., Ellis, S. and Soler, J. (eds)
Interdisciplinary Perspectives on Leaning to Read: Culture, Cognition and Pedagogy, London and New York, Routledge, pp. 117–29.

Introduction
In 2003, Seymour, Aro and Erskine reported the results of a comprehensive investigation of written word recognition skills at the end of first grade in 14 different European countries. Results showed that children who were learning to read English performed far worse than the children of any other nationality at reading both real words and non-words with a similar structure to real words. Whereas children from most of the 14 countries read over 90 per cent of real words accurately, the children learning to read English were correct on only 34 per cent. The next lowest score was 71 per cent of words read correctly by children from Denmark.

Following Wimmer and Hummer (1990), this is just one of many studies published in the last 20 years to show that the word recognition skills of children learning to read English take longer to develop than those of children from countries such as Austria, Croatia, Greece, Germany, the Netherlands, Italy, Turkey, Serbia and Spain.
Although the reading speed of children from these countries increases as they get older, the accuracy of their decoding skills is at a very high level by the end of their first year of formal instruction.

Why do children from the UK consistently perform so much worse in these cross-cultural comparisons? Seymour et al. (2003) highlighted two important differences between the English language and European languages where children’s word recognition skills develop particularly quickly. The first is the opaque nature of the English writing system (or ‘orthography’). The second is the complex nature of the syllabic structure of spoken English. The reasons why both the spoken and written form of English might be associated with relatively slow development of reading skills are discussed below.

The English writing system
In common with all European and American languages, English employs an alphabetic writing system in which letters (or graphemes) represent the spoken sounds of words (phonemes). A phoneme is the smallest unit of sound that can affect the meaning of a word, and a grapheme is the representation of a phoneme in written form.

The problem with English is that there is less consistency in grapheme– phoneme relationships than in almost any other alphabetic writing system. Graphemes for vowels in particular can represent a large number of different phonemes in different words. Hence English is said to have a deep or opaque orthography in contrast with languages that are written in shallow or transparent orthographies where each grapheme represents the same phoneme in every word in which it appears.

There are two obvious reasons why English is not transparent. First, although the pronunciation of many words has changed over the centuries, their spelling remains frozen in its earlier form. For example the now silent k at the start of the word knight was sounded out at the time when its written form was established. Second, when foreign words are imported into English, we generally keep the written form of the word in the language from which it originated. For example, the spelling of the word café was retained when it entered English from French instead of being changed to caffay. In languages with transparent orthographies such as Spanish or Welsh, spelling reform ensures that the written form of a word is congruent with its current spoken form. Consequently, frozen spellings and spellings of imported words are altered to ensure that they are consistent with the letter–sound rules of the transparent orthography.

There are some advantages for English in not having a completely regular orthography. For example, skilled readers of English can distinguish the meanings of homophones such as colonel and kernel directly from their written form. In a transparent orthography, they would be spelled the same way. The disadvantage of an opaque orthography is the existence of many irregular words whose pronunciation cannot be predicted from their spelling. Moreover, many frequent and early-acquired English words are irregular.

The existence of irregular words means that a child learning to read English faces two potential problems that are not encountered by most of his or her counterparts in Continental Europe. When children read a word in a transparent orthography that is part of their speech vocabulary, they can reliably generate its spoken form and hence access its meaning even if they have never encountered the word in print before. Such a strategy will not be successful for many words in English because letter–sound rules will not produce the correct pronunciation. The second problem is that the existence of exceptions means that the letter–sound correspondences that apply in regular English words are likely to be more difficult for children to learn. Decoding skills may therefore take longer to develop in opaque writing systems.

The syllabic structure of English
In many languages, including Italian and Spanish, words typically contain simple syllabic structures in which a vowel is preceded by a single consonant. English is more complex because clusters containing two or more consonants can occur either at the start or end of syllables. According to Ziegler and Goswami (2005), the preponderance of consonant clusters in English affects the acquisition of literacy by making it more difficult for children to learn grapheme–phoneme consistencies.

Before they start to read, many children become aware that spoken languages have smaller units than words and can count the number of syllables that spoken words contain (Liberman, Shankweiler, Fischer & Carter 1974).
Later, awareness of the sub-syllabic units of onset and rime develops in pre-literate children, particularly in the UK where nursery rhymes are part of the culture (e.g. Bradley & Bryant 1983). However, alphabetic writing systems do not contain visual symbols for onsets, rimes or syllables. Instead they represent phonemes. As Usha Goswami [2010] makes clear, phonemes are not natural units of speech and cannot be produced or perceived in isolation.

Furthermore, sounds that are physically different in words or syllables (e.g. the /p/ sound in spoon and pit) must be mapped onto the same phoneme. As a consequence, awareness of phonemes does not develop automatically […] [and it] appears that speakers do not know about the existence of phonemes until they learn an alphabet.

Children whose languages have a simple syllabic structure may find the transition from representations based on onset and rime to representations based on phonemes easier to master. This is because onsets and rimes will frequently be single phonemes in languages where there are relatively few consonant clusters.

Consequently, splitting an Italian or Spanish word into its onset and rime will often automatically produce two phonemes. It may therefore be relatively easy for Italian or Spanish children to learn the relationship between the letter sounds that they are taught in school and the words that these letters represent when the words are written down. However, only 5 per cent of English [syllables] have a CV structure (De Cara & Goswami 2002), which means that English onsets and rimes will both typically contain more than one phoneme. English children may therefore need much more explicit training before awareness of phonemes develops.
[…]

Research study of learning to read in Wales
Between 1996 and 2004, my colleagues and I conducted a research programme that investigated some of the reasons why early reading skills develop relatively slowly in children learning English. In this investigation, the ease of learning to read English was compared with learning to read Welsh (Hanley 2010; Hanley, Masterson, Spencer & Evans 2004; Spencer & Hanley 2003, 2004). The research was conducted in Denbighshire in North Wales, where 27 per cent of the population described themselves as Welsh speakers in the 1991 Office of Population Censuses and Surveys. Some towns in this county contain both English-and Welsh-speaking Primary schools, and parents can choose which school their child will attend. […]

If they attend Welsh-speaking schools, children are taught to read in Welsh, a transparent alphabetic orthography in which letter–sound relationships are relatively consistent and irregular words are virtually non-existent […]
Despite the differences in the transparency of the orthographies, the Welsh-and English-speaking children in our studies lived in the same area of North Wales, commenced reading instruction at the same age, and were taught by similar methods of instruction. Welsh syllables also contain consonant clusters that can occur either at the beginning or end of words. It is therefore possible to compare the acquisition of a shallow and deep orthography in children of a similar age whose languages contain words with complex syllabic structures, all of whom receive phonic instruction. Wales therefore offers a unique opportunity to investigate the influence of orthographic consistency on reading development. […]

When the children were ten years old, we compared their reading accuracy on a set of 60 words that varied according to their regularity and their frequency (i.e. how often they occur in written English). For example, the words horse (ceffyl), tooth (dant)and grill (gril) are regular words of high, medium and low frequency, and bowl (bowlen), glove (maneg) and sword (cleddyf) are irregular words of high, medium and low frequency. The English children read the regular words and the high frequency irregular words as accurately as did the Welsh children. […]

This suggests that the decoding skills of the English children have by now caught up with those of their Welsh counterparts. Significantly superior performance by the Welsh children was only observed on the medium and low frequency irregular words. The lower frequency irregular words will have been encountered less often in print and many of them do not yet appear to be part of the English children’s sight vocabulary. If English children try to use decoding skills to read these words, they will pronounce them incorrectly. The absence of irregular words in Welsh means that Welsh children will be able to read aloud correctly the Welsh equivalents of these words even though they are equally unlikely to have encountered them in print very often.

It therefore appears to be the case that the opaque nature of the English orthography slows down the acquisition of decoding skills, but even when these skills have caught up at ten years old, children learning to read English have not received sufficient print exposure to many irregular words to allow them to be read accurately. Because of the absence of irregular words, a much larger reading vocabulary is available to readers of Welsh immediately they have developed competence in decoding. […]

Pedagogy
Improving children’s reading attainment in the future by reform of the English orthography is clearly a utopian pipe-dream. […]
A more realistic strategy than reforming the orthography might be to acknowledge that English probably requires more extensive training in phonics than transparent orthographies. […]
[However]itis[…] possible that English requires a different type of phonic reading instruction. According to grain-size theory (Ziegler & Goswami 2005), teaching should not be exclusively focused on grapheme–phoneme correspondences. Children should also be made aware of correspondences between larger phonological and orthographic units. Most obviously, the correct pronunciation of many irregular words can only be taught by word-specific training. Nevertheless teaching a child to read an irregular word such as friend will allow him or her to read all words that contain ‘friend’ as their root morpheme (e.g. friend, friendship, friendly, friendlier, unfriendly, befriend, friendliness,etc.).

In some irregular words, however, there are higher order consistencies that provide information about how the irregular portion of the word should be pronounced. In particular, there are important orthographic consistencies at the level of onsets and rimes even in words that are irregular in terms of their grapheme–phoneme mappings (Treiman, Mullennix, Bijeljac-Babic & Richmond-Welty 1995). For example, the pronunciation of the ea vowel in health differs from the regular pronunciation of ea (as in heat). However, ea is pronounced as it is in health in all words in which –ealth is the rime segment wealth, stealth,etc.). Consequently, if they are taught the correspondences between rime segments and their pronunciation, Ziegler and Goswami argue that children should be able to successfully decode words that contain consistently spelled rime segments even if they are irregular. A teaching schedule that concentrates exclusively on grapheme–phoneme relationships ignores this important source of information about the English writing system. […]

Morphemes (defined as the smallest units of meaning in a language) contain another important source of information about how English words are written because English orthography often preserves morpheme identity at the expense of phonology. For example, the is always spelled the same way even though its pronunciation differs according to whether it precedes a vowel or consonant. In a highly impressive series of studies, Nunes and Bryant (2006) provide a powerful demonstration that interventions based on the teaching of morphemes to children significantly improve their spelling ability. For example, they point out, teaching children that the plural inflection at the end of a word is consistently represented by the letter s should prevent them from spelling rocks as rox even if they have learnt to spell fox correctly. […]

In conclusion, it is evident that English is a particularly difficult writing system to learn. The transparent alphabetic orthographies that are commonly used in other European countries including Wales allow children to develop phonological awareness and decoding skills much more easily. […] It appears that children learning English require more extensive phonics instruction than is required for children learning transparent orthographies. […] However, the unpredictable nature of the English writing system can be reduced if children’s attention is[…] drawn to the relationships between larger units than graphemes and phonemes (Nunes & Bryant 2006; Ziegler & Goswami 2005). Knowledge of the relationships between the orthography and phonology of onsets, rimes, and morphemes is likely to make it easier for children to achieve mastery over the notoriously complex English writing system.