|
Article Excerpt
Most alphabetic orthographies, including German, Spanish, Dutch and Italian, are said to be 'shallow' or 'transparent' in that graphemes in these systems generally represent only one phoneme. However, some alphabetic systems, including English and French, are said to be 'deep' or 'opaque'. This means that individual graphemes represent a number of different phonemes in different words, and there are many exceptions to grapheme-phoneme correspondence rules. As a consequence, English orthography contains many irregular or exception words such as 'have', 'shoe' and 'one'. In a transparent orthography, the mappings from letters to sound are much more consistent, and there are very few irregular words.
It is important to discover whether the transparency of an alphabetic orthography has any effect on the way in which children learn to read. The idea that reading acquisition may differ according to the nature of the orthography has been referred to as the 'Orthographic Depth Hypothesis' (e.g. Lukatela, Carello, Shankweiler, & Liberman, 1995). There are at least two interesting issues at stake here. The first is whether children can learn to read a transparent alphabetic orthography more quickly than they can learn to read an opaque alphabetic orthography. The second concerns the qualitative nature of reading development. Do children adopt different strategies when faced with an opaque alphabetic orthography compared to the strategies that children use when learning a transparent orthography? If so, are different skills required in learning transparent and shallow alphabetic orthographies?
Investigations of children learning to read transparent alphabetic orthographies such as Turkish (Oney & Durgunoglu, 1997), German (Wimmer & Hummer, 1990) and Italian (Cossu, Shankweiler, Liberman, & Gugliotta, 1995) have shown that reading skill develops very rapidly at school, with children making relatively few errors by the end of the first year of formal instruction. Oney and Durgunoglu (1997) for instance, found that even children with relatively limited awareness of phonological units at the start of the school year in Turkey were able to read and spell with high levels of accuracy by the end of the first grade. These results were interpreted as reflecting the consistent nature of the mappings between graphemes and phonemes in Turkish orthography.
Wimmer (1993) argued that learning to read and write a transparent orthography may be characterized by more rapid development of word-decoding skills than is learning to read an opaque orthography. Consistent with this suggestion, Wimmer and Hummer (1990) found that virtually all of the children they studied possessed a strategy for accurate phonetic decoding of letter strings in German; even 'poor' readers could decode and blend a word correctly, although it took them longer to do so. Goswami, Gombert, and de Barrera (1998) found superior nonword reading in Spanish children relative to French and English children. Wimmer and Gosdwami (1994) found that Austrian children who were learning to read in German performed extremely well relative to English children at reading nonwords at a relatively early stage in the developmental process. Wimmer and Goswami argued that the Austrian children were better able to exploit the regularity of the sound-symbol correspondences than the English readers. Landerl (2000) exte nded these findings by showing that the nonword reading skills of English children were worse than those of their Austrian counterparts even when they were being taught exclusively via a phonics approach.
In addition, Wimmer and Hummer (1990) found that most of the erroneous responses to real words that were made by children learning to read German were nonsense words. In marked contrast, errors made by children reading English are frequently (the wrong) real words (see Seymour & Elder, 1986; Stuart & Coltheart, 1988). Thorstad (1991) observed a very similar pattern of findings in a study that compared the reading and spelling errors made by Italian children aged 6-11 years with those made by English children.
There is also evidence that phonological awareness skills can develop more rapidly amongst children learning to read a transparent orthography. Phonological awareness refers to the ability to hear and show overt sensitivity to component sounds in spoken words, and is known to be highiy correlated with reading ability in children (see Goswami & Bryant, 1990). It is demonstrated by successful performance on tasks such as phoneme tapping (Liberman, Shankweiler, Fischer, & Carter, 1974). Cossu, Shankweiler, Liberman, Katz, and Tola (1988) compared the phoneme segmentation abilities of Italian children with those of English-speaking American children. Prior to starting at school, the children were given tests of syllable segmentation and tests of phoneme segmentation. Both groups performed best on the syllable segmentation task. The children were tested again after starting school. This time it was found that the American children performed better on the syllable segmentation task than at phoneme segmentation, whe reas the Italian children showed the opposite pattern of performance. Most striking was the fact that the Italian children consistently outperformed the American children, particularly on the tests of phoneme awareness.
To summarize, recent research has suggested that children become competent readers of transparent orthographies such as German, Italian and Turkish within a year of the start of formal reading instruction. There is also evidence that their reading of nonwords is superior to that of English children even when reading age is matched, and that the errors they make when reading familiar words are qualitatively different. There is evidence that phoneme awareness skills may be superior in children learning a shallow orthography.
Nevertheless, it would be inappropriate to conclude that the research conducted to date provides conclusive evidence that it is easier to learn to read a transparent than an opaque alphabetic orthography. This is because there are important differences between educational practices in British schools and schools in continental Europe. In Britain, children begin to learn to read in school when they are approximately 4-5 years old. Conversely, teaching children of this age to read is actively discouraged in countries such as Austria, where formal reading instruction does not begin until the children are at least 6 years old. This means that children generally start to learn to read transparent orthographies when they are somewhat older than they are when they start to learn to read English. It is therefore impossible to be certain whether the rapid development of reading skill observed in children in these countries during the first year of reading instruction is due to the age of the children or the transparen cy of the orthography. Even in Wimmer and Goswaxni's (1994) investigation of nonword reading in English and Austrian children, the English children were on average 10 months younger than their Austrian counterparts.
Consequently, the generality of the previous findings would be greatly extended by a study that compared the acquisition of two orthographies from within the same education system in the same country by children of a similar age. The present study therefore set out to examine reading development in Wales because some children are taught to read in Welsh (typically native speakers of Welsh) and some children are taught to read in English (typically native speakers of English). Crucially, the Welsh writing system is another example of a phonologically transparent orthography in which the mappings from graphemes to phonemes are relatively consistent and unambiguous. It is therefore possible to conduct a sensitive investigation of the effects of orthographic consistency on reading development by comparing the performance of Welsh children learning to read English with Welsh children learning to read Welsh.
A comparison of phonological awareness skills in Welsh- and English-speaking children also makes it possible to investigate further the relationship between the development of phonological awareness skills and orthographic transparency (see Cossu et al., 1988). There are fewer vowel phonemes in Italian than in English, and the syllabic structure of Italian words is generally much simpler than the syllabic structure of English words. Therefore, the relatively simple phonotactic structure of Italian may have been one reason why Cossu et al. observed superior phoneme awareness in the Italian children. Consistent with this, recent studies of phoneme awareness in Spanish, which has a similar phonotactic structure to Italian, have indicated relatively early development of phoneme segmentation skills in pre-readers (Borzone de Manrique & Signorini, 1994; Gonzalez & Garcia, 1995). The phonotactic structure of Welsh, however, is much closer to English than it is to Italian or Spanish. In Welsh syllables, as in English syllables, there are a number of consonant clusters (e.g. streic, cnau) that can appear both before and after the vowel (Awberry, 1984). This is not the case in Italian or Spanish syllables. In addition, whereas Italian has a relatively small number of vowels, there is a similar number of vowel phonemes in English and Welsh. Indeed Welsh contains some vowel (and consonant) phonemes that are not found in English. These similarities between the phonotactic complexity of Welsh and English mean that any differences that might emerge between the phonological awareness skills of Welsh and English readers can be more confidently attributed to the transparency of the orthography than is the case when comparing phonological skills of Italian- and English-speaking children.
In the Welsh alphabet, 21 of the 26 letters from the Roman alphabet represent phonemes. The missing letters are k, q, v, x, z. The use of digraphs to represent single phonemes is characteristic of almost all alphabetic orthographies, and in Welsh there are eleven consonant phonemes that are always represented by digraphs (ch, dd, ff, ll, ng, nn, ph, rh, rr, th, si). The eight most common of these digraphs are explicitly taught to children in the early stages of reading instruction as 'letters' of the Welsh alphabet. The letters that comprise these graphemes virtually never appear together as separate graphemes (i.e. as representing two phonemes). The only exceptions to this are 'si' which sometimes represents two phonemes (e.g. sinc /sink/) rather than one (e.g. siop /sop/) and 'ng' which normally represents /[eta]/ but represents two phonemes (/n/ and /g/) in two irregular words (dangos, Bangor). There are also a number of graphemes in which two letters appear together to represent a diphthong (for example a i, ae, aw, au, ei). These letters never represent two separate phonemes when they appear together in a word. Rules for graphemic parsing are therefore quite complex but the relationship is much more consistent in Welsh than in English.
The relationship between graphemes for consonants and the phonemes that they represent is almost entirely one to one. The only complicating factor is that the letters 'i' and 'w' can represent a consonant phoneme (respectively /j/ and /w,9 in addition to vowel phonemes. This situation is somewhat more complex for vowels. There are seven different graphemes that are used to represent vowels (a, e, i, o, u, w, y). Two different phonemes are represented by these graphemes in North Wales. (1) The only exception is 'y' which can also represent /[partial]/. There are, however, context-sensitive rules that dictate which is the correct pronunciation of a vowel grapheme. These context-sensitive rules are associated with the position of the letter in a word. Generally, a vowel grapheme represents a different phoneme when it appears at the start or end of a syllable from when it appears in the middle of a syllable. Sometimes a circumflex is shown when the vowel appears in the middle of a syllable to remind the reader of the different pronunciation, but in most cases the circumflex is not included on vowels in monosyllablic words (see Griffiths & Jones, 1995; Thomas, 1996, for further details).
Welsh is also highly transparent for the purposes of spelling. The only ambiguity with vowel phonemes is that both 'y' and 'u' can represent the phonemes /[begin strikethrough]1[end strikethrough]/ and /[begin strikethrough]n[end strikethrough]/. The only other ambiguities concern geminate consonants. Sometimes /n/ is written as a single 'n', and slightly less frequently as 'nn'. The same is true of the spelling of /r/ but the use of 'rr' is much rarer than 'r'.
Because vowels are generally represented by the same grapheme in alphabetic orthographies regardless of whether they are stressed or not, inconsistent stress patterns in spoken words will make it more difficult to learn the correct pronunciation of written words. Although it does not have the regular stress pattern of Italian, stress in Welsh words is more consistent than it is in English. In words of more than one syllable, the stress normally falls on the penultimate syllable, with the main exceptions being personal pronouns and verbs with certain endings where the stress is on the final syllable. Sometimes (though not always) an unusual stress pattern is marked by an accent. In addition, although all of the Welsh vowels can appear in stressed and unstressed forms, there is a smaller difference...
|
