Phonological recoding

An alternative way in which unfamiliar words can be read is through phonological recoding. Providing that a word contains typical letter–sound relations, knowledge of letter–sound associations can be utilised to read aloud the new word. The ability to decode novel words through the application of previously acquired letter–sound relations can also be referred to as a ‘word attack skill’ (Saunders, 2007).

Phonological recoding occupies a prominent role in the self-teaching hypothesis proposed by Share (1995). When an unfamiliar word is encountered by beginning readers, they can attempt to ‘self-teach’ and pronounce the word through use of their previously established letter–sound knowledge (Jorm & Share, 1983). Once the correct pronunciation for the new word has been deciphered, perhaps through feedback provided by a parent or teacher, the beginning reader starts to accumulate greater knowledge about the orthographic and phonological properties of that particular word (e.g., what printed letters visually form the word? how is this letter string pronounced?) which is further refined through future encounters with the word. Such repeated exposure to the word is one way in which the word can be integrated into the orthographic lexicon. Indeed, studies have shown that some beginning readers may only need to encounter a word a small number of times in order for the word to be entered into the orthographic lexicon (e.g., Manis, 1985; Nation, Angell, & Castles, 2007; Reitsma, 1983). Ultimately then, the words stored in the orthographic lexicon should be recognised quickly and accurately ‘by sight’, thereby enabling proficient reading (Ehri, 2005).

Research has shown that phonological recoding is a skill that can be used by early readers in English to identify unfamiliar words (e.g., Bowey & Muller, 2005;

Cunningham, Perry, Stanovich, & Share, 2002; Thompson, Fletcher-Finn, & Cottrell, 1999). How then can phonological recoding be achieved?

1.3.4 Acquiring phonological recoding skills

According to Ziegler and Goswami (2005) successful phonological recoding is contingent upon beginning readers finding “shared grain sizes in the symbol system (orthography) and phonology of their language that allows a straightforward and unambiguous mapping between the two domains” (p. 3). That is, beginning readers must learn how printed letters relate to sounds, but the way in which these relations are acquired (or taught) can occur at different grain sizes. Potentially, individuals can learn the associations between individual letters (graphemes) and individual sounds (phonemes); this would be learning the print–sound correspondences at a small grain size. Alternatively, larger grain sizes, such as rimes and syllables, can be used, in which combined letter strings are matched to larger spoken units.

As will be discussed further in Chapter 8, for children learning to read in English, the difficulty lies in the inconsistency of letter–sound mappings that occur in English. An identical printed letter may have several pronunciations, while the same phoneme may be represented by different letters. Other alphabetic scripts, such as German and Italian, are highly consistent and predictable in that the associations between graphemes and phonemes are largely stable, one-to-one mappings. Fortunately, greater consistency in spelling–sound relations in English can be found for many larger units such as rimes and syllables (Treiman, Mullennix, Bijeljac- Babic, & Richmond-Welty, 1995). For example, the rime ‘ing’ as found in the words ‘ring’, ‘sing’, ‘wing’, ‘bring’, is always pronounced the same way across these words.

The very fact that writing systems vary in how printed letters represent spoken units, and in the case of alphabetic scripts, can be differentiated further in terms of the consistency of these mappings, provided the background for the psycholinguistic grain size theory of reading (Ziegler & Goswami, 2005). According to this theory, reading development is directly related to, and influenced by, the orthography that a child is learning to read. With differences between scripts in the consistency of grapheme–phoneme mappings, consequently, there may also be differences in the grain sizes that are used most effectively across orthographies to read words.

In consistent orthographies, such as German and Italian, smaller grain sizes (e.g., phonemes) alone can be used to decode words given the straightforward mapping between graphemes and phonemes in these languages. When learning to read in an inconsistent orthography such as English, it may be necessary to develop recoding strategies at more than one grain size (i.e., phonemes, onsets and rimes, syllables, whole words) to cope with the idiosyncrasies of this particular script. As Ziegler and Goswami (2005) noted, English contains a wide assortment of word types. There are regular words adhering to expected letter–sound patterns (e.g., ‘ran’, ‘lot’, ‘hen’) that can easily be decoded through use of small grapheme–phoneme mappings. Certain words share identical printed rimes which are pronounced consistently across words (e.g., ‘bake’, ‘cake’, ‘lake’, ‘make’). For such words learning the larger rime spelling–sound mappings may be helpful. Lastly, irregular words containing atypical spellings and pronunciations (e.g., ‘choir’, ‘knife’) must be learned as whole words, each with a unique letter sequence and spoken form.

Thus, flexibility in the grain sizes that are learned and applied has been suggested to be one of the keys to facilitating successful word recognition in English (see Brown & Deavers, 1999; Ziegler & Goswami, 2005, 2006). The idea of

developing recoding strategies at multiple grain sizes arguably parallels the variable nature of English. In other, more consistent, orthographies, such flexibility is not necessary, as smaller grain sizes can reliably be used.

Evidence has been gathered to support some of the central assumptions of the psycholinguistic grain size theory. In particular, the ability to read nonwords that can only be decoded correctly through the application of grapheme-to-phoneme relations has been examined. If orthographic consistency does influence the development of recoding strategies, children learning to read consistent languages through use of small grain grapheme-to-phoneme mappings, arguably, should be more proficient than children learning to read in English when attempting to decode nonwords. Several studies have reported such a difference in nonword reading performance. For example, Goswami, Gombert, and de Barrera (1998) found that English speaking children matched for reading age with French and Spanish children were noticeably poorer at reading nonwords than the children from the more consistent languages. More recently, Ellis and Hooper (2001) and Spencer and Hanley (2003) compared reading acquisition of children living in Wales who attended English speaking schools (learning to read in English) or Welsh speaking schools (learning to read in Welsh). Unlike English, Welsh is composed of more direct and reliable one-to-one mappings between its graphemes and phonemes. Throughout the experiments, the children learning to read in Welsh continually performed better than the age matched children learning to read in English on the word recognition measures (e.g., naming of regular and irregular words, and nonwords). Once more, the findings underlined how the consistency of the orthography could, on the one hand, facilitate the rapid development of word recognition skills in the case of consistent scripts, or on the other hand, at least in the early stages of reading acquisition for inconsistent scripts,

potentially obstruct the ease with which word recognition skills are acquired. However, in studies where nonwords can be read aloud through use of rime analogies (e.g., ‘bicket’ – ‘ticket’), researchers have observed there is a marked improvement in nonword reading for English speaking children (e.g., Goswami, Porpodas, & Wheelwright, 1997, Brown & Deavers, 1999). Of relevance here also are the findings from the investigations in Wales (e.g., Ellis & Hooper, 2001; Spencer & Hanley, 2003) showing that the children learning to read in English were more likely to name printed words as other actual words (i.e., to make real word substitution errors). For example, reading ‘bye’ as /be/, ‘mow’ as /moon/. Conversely, the children learning to read in Welsh tended to mispronounce the actual words to produce nonwords that were phonologically similar to the target words. Ziegler and Goswami (2005) interpreted these findings as indicative that beginning readers of inconsistent scripts do use larger grain sizes such as rimes, alongside smaller grain sizes to decode unfamiliar words, whereas in more consistent orthographies, smaller sized grapheme- to-phoneme mappings, but rarely larger grain sizes, will be applied.

In this section we have reviewed some of the ways in which unfamiliar words can be tackled. Although these approaches were presented separately, beginning readers in English at least, do not simply rely on one strategy to decipher new words; they can use more than one available strategy (Brown & Deavers, 1999; Roberts & McDougall, 2003). For example, a child who has received instruction in how to make use of analogies, and who has learned about letter–sound correspondences, will very likely employ a combination of these approaches when confronted with an unfamiliar word.