About the Journal

This study examines qualitative differences in children’s orthographic knowledge and their ability to develop and test hypotheses about an unknown word in two languages, Brazilian Portuguese and English. A cross-linguistic comparison allows examination of the nature of children’s spelling development under different orthographic conditions. 81 Brazilian children in grades 2, 3 and 4 (i.e. aged 9, 10 and 11 years) and 95 English children in the equivalent school grades were asked to play the game ‘Hangman’ with 4 and 5 letter words. They were also asked to justify their selection of letters. There was greater heterogeneity in the Brazilian children’s responses than in the English children’s responses across word lengths and syllabic patterns. This is interpreted in terms of the Brazilian children’s sensitivity to syllabic patterns. Children’s justifications indicated that in all three grades and in both deep and shallow orthographies they made use of lexical and non-lexical strategies. * The authors are indebted to the schools who welcomed us and the Educational Authorities in Rio de Janeiro and Oxforshire. Support for this research was provided by FAPERJ, MCT/CNPq – Ministry of Science and Technology / National Council for Scientific and Technological Development – and the Oxford Brookes University QR Fund. 62 Jane Correa, et al. This study aims to gain insight into qualitative differences in children’s orthographic knowledge and their ability to develop and test hypotheses about an unknown word. Tests of children’s spelling abilities have traditionally been the source of evidence of children’s implicit understanding of the nature of their orthography. Researchers make inferences about children’s understanding of spelling rules using spelling errors as the data (Treiman, 2004) or by using carefully devised tasks which allow inferences to be made about the use of analogies (Nation & Hulme, 1998) morphological understanding (Nunes, Bryant & Bindman, 1997; Landerl & Reitsma, 2005; Sénéchal, Basque & Leclaire, 2006) or knowledge of legal letter combinations (Cassar & Treiman, 1997). While these approaches have yielded valuable insights into the development of children’s spelling, they are school-like tasks, and as such provided a much more formal context in which children have their skills evaluated. Besides that, these tasks explicitly required children to use their orthographic knowledge as an end in itself. From the point of view of children’s action, the sole purpose of accomplishing these tasks is to have their spelling skills evaluated. Thus these tasks do not usually tell us much about children’s spontaneous use of spelling knowledge. For this we have to look at more qualitative work such as that of Read (1986) on children’s invented spelling, Downing, Coughlin & Rich (1986) on children’s judgements of correct spelling or Caravolas, Kessler, Hulme and Snowling (2005) on the strategies children report using in response to a spelling task. On the other hand, in non-school contexts children’s spelling skills are used as tools for achieving another purposeful goal, such as when children write birthday cards or amuse themselves playing Hangman. The data gathered from more qualitative studies with non-school-like tasks may also contribute to our understanding of the nature of spelling development as well as of children’s representations of different aspects of their language, especially phonological ones (Read, 1971; 1986; Ehri & Wilce, 1980). Weekes (1994) and Castles, Holmes & Wong (1997) describe children’s spelling styles in terms of the extent to which the children are lexically reliant (i.e. by means of words) or non-lexically reliant. Their definitions were based on the degree to which children made regularisation or lexicalisation errors in their spellings. A regularisation error occurs when a misspelled word could be pronounced to sound like the target word. On the other hand, if the children misspelled a word by writing another word in its place, it was classified as a lexicalisation error. Their data showed that lexically reliant readers tend to be lexically reliant spellers and non-lexically reliant readers tend to be non-lexically reliant spellers. They suggest that both strategies are available to normally developing children for deployment. Further, the fact that the groups of lexical and non-lexical spellers did not differ in chronological age or reading age suggested that these were indeed styles and strategies and not patterns of development Using Spelling Skills 63 It is now entirely non-contentious to state that reading and spelling in a deep orthography such as English (containing many irregular words) requires both an assembled spelling strategy (based on sounds) and a lexical strategy (Frith, 1980; Goswami & Bryant, 1990). Indeed, the focus of the debate on reading and spelling strategies is not now on the use of lexical or non-lexical strategies, but on the degree to which shallow orthographies (with few irregularly spelled words) may also demand a lexical strategy. Barry (1992) addressed this question, using a priming methodology to contrast non-word spelling in English, Welsh and Italian. Priming occurs when a word is more rapidly processed due to the previous presentation of the same or a similar word. Barry (1992) found that while English unsurprisingly produced significant lexical priming effects (required to cope with the irregularities), a lexical priming effect was also found for Italian and Welsh. Two major implications can be drawn from these results. First, although the orthographic regularity of certain languages allows an efficient use of assembled spelling, this does not imply that the non-lexical strategy is the sole strategy required. The second implication concerns the functional independence of the word-specific and the assembled spelling procedures to the development of the children’s orthographic abilities. It seems that lexical and non-lexical strategies are both needed in deep and in shallow orthographies, although their relative importance might vary in different languages. There is now a long history of research into the impact of the orthography on children’s reading (Bryant & Nunes, 1998) but considerably less on children’s spelling. The impetus for the research into reading has been the development of models of learning to read which incorporate phonological awareness. In the case of spelling, the transparency of the orthography – namely the regularity of the grapheme-phoneme correspondences – is also a critical feature. If children are helped to spell by learning about grapheme-phoneme correspondences, then languages with more transparent orthography will prove easier to learn to spell than those with more opaque orthography. A cross-linguistic study will allow us to answer questions about the nature of children’s spelling development under different orthographic conditions. This study aims to compare a relatively shallow and a deep orthography (Brazilian Portuguese and English) using the Hangman game, to find out what impact the orthography has on children’s understanding of their language. In the game of Hangman a child has to work out an unknown word given only minimal information about it, knowing only the number of letters it has. This task allows children to make explicit their strategies for spelling the unknown word. The game is played by suggesting possible letters (to fill in the template) or words (to match the template). Children playing the game generally propose letters, filling in the spaces until they are able to infer the correct word or confirm their hypotheses. How then do children select letters to propose? It is possible that they do so on the basis of what they know about the 64 Jane Correa, et al. rules of spelling, in other words, using a non-lexical strategy. An alternative strategy for children would be to select letters on the grounds that they knew a word of the appropriate length containing that letter. This strategy depends on easy access to the lexicon, easy generation of vocabulary, with few limits to the choice of words except the word length. The Hangman game therefore allows us to investigate children’s knowledge of their language as used in the context of a non-school-like task and to look at some of the factors which might influence children’s performance. The impact of schooling on children’s performance in the game is likely to be significant. As children spend time in school and as their literacy skills increase, their written and spoken vocabulary increases. Across a span of three school years improvements in performance are likely to be attributable to education effects. In addition some improvements might also be attributable to increased understanding of how to deploy strategies specifically tailored to achieving success in the game. In addition to education effects it is also possible to use the game to look for the impact of orthographic-specific expectations. The expectations a reader, speller, or in this case a Hangman player, has about the target word are determined by the individual’s age and education (as these are not separable), and knowledge of the orthography. Orthographies vary considerably in terms of the salient features of words. In English word length is associated with numbers of syllables, such that longer words are more likely to be of more than one syllable in length. However it is also the case that words varying in length from one to more than seven letters may all be monosyllables, e.g. ‘a’[ə], ‘be’[bi], ‘sea’ [siː], ‘deep’ [diːp], ‘flail’ [fleɪl], ‘glitch’[glɪtʃ], and ‘thought’ [θɔːt]. Other orthographies have tighter constraints on the relationship between letters and syllables, and Brazilian Portuguese is one of these. A word with seven letters could never be a monosyllabic word in Brazilian Portuguese as the language does not contain a syllable which has more than five elements (Silva, 1999;