Printer Friendly

Predicting children's word-spelling difficulty for common English words from measures of orthographic transparency, phonemic and graphemic length and word frequency.

The orthographic depth or transparency of a language, which reflects the grapheme-phoneme consistency of words, has an effect on the acquisition of literacy in young children (Seymour, Aro, & Erskine, 2003; Spencer & Hanley, 2003). Liberman, Liberman, Mattingly, and Shankweiler (1980) suggested that all alphabetic orthographies can be classified according to the transparency of their letter-to-phoneme correspondence or orthographic depth, with shallow orthographies demonstrating letters that are isomorphic to phonemes in the spoken word, and deep orthographies having letter-phoneme relations which are substantially equivocal. Based on this concept, Katz and Frost (1992) proposed their orthographic depth hypothesis, stating that literacy acquisition in a shallow language should be based on a single alphabetic process, involving the language phonology, whereas in a deep language a second logographic process, depending on visual-orthographic structure, will be involved. Seymour et al. find that this hypothesis fits well with their reading data, indicating that there appears to be a threshold of orthographic complexity which results in a step change in the way early reading is acquired. Above this threshold are the shallow European orthographies and below are those classed as deep, including Portuguese, French, Danish and English.

Seymour and Duncan (2004) indicate that their scheme is hypothetical or quasi-intuitive, acknowledging that, although the variation in orthographic complexity has not yet been submitted to a comprehensive computational linguistic analysis, there is general agreement that some orthographies are relatively shallow while others are considered to be deep, containing orthographic inconsistencies and complexities, including multi-letter graphemes, irregularities and morphological effects.

Although such quasi-intuitive approaches are useful in comparative studies that only require broad categorizations, there is an inexorable move to more refined definitions of orthographic depth or transparency that take account of the complexity of the construct including, for example, the variation in individual language characteristics according to the direction of the correspondences between graphemes and phonemes. Finnish and Turkish are highly transparent, showing one-to-one mapping in both phoneme-to-orthography (P-O or spelling) and orthography-to-phoneme (O-P or reading) directions. German and Greek are highly O-P transparent, but have greater complexity in the opposite direction, whereas English is opaque in both directions (Goswami, Porpodas, & Wheelwright, 1997), although English consistency mapping may improve when larger units (onset-rime) are considered (Treiman, Mullenix, Bijeljac-Babic, & Richmond-Welty, 1995).

In terms of refined measures of P-O regularity in English, Hanna, Hanna, and Hodges (1966; hereafter Hanna) computed values by identifying the associated grapheme for each phoneme in their 17,310 word corpus. Their results indicated that 73% of phonemes would be spelt correctly if the most regular graphemic form was used, but that only 50% of English words could be spelt correctly by their computer programme that was based on a detailed analysis of the English language. This fine grain analysis provided the basis for computing the regularity or consistency of English words for investigations of spelling, but has been surprisingly little used.

Berndt, Reggia, and Mitchum (1987; hereafter Berndt) presented their scheme that offered a quantitative description, in the opposite O-P direction, of the grapheme-to-phoneme correspondences (GPC) for English reading. The study was undertaken because 'objective information is lacking on the extent of association between specific letters (and letter clusters) and their pronunciation in words' (p. 1), and there was an implicit challenge to the regular/irregular or consistent/inconsistent dichotomous classifications of English words (Glushko, 1979). They also demonstrated that established probabilities for spelling did not reflect those for reading. For example, although the phoneme \or\ is only rarely (.15) spelt <au> (as in auction), the grapheme <au> is almost always (.95) pronounced \or\. As with Hanna, the associated values were not weighted to reflect the frequency of occurrence of individual words in the language (i.e. values were based on type and not token counts).

Carney (1994) combined the approaches of both Hanna and Berndt by taking account of the two directions of correspondence when producing type and token values for O-P (decoding) and P-O (encoding). Unfortunately, as with Wijk (1966) and Venezky (1970), detailed statistical information is not consistently presented, limiting the value of the study as a basis for quantifying orthographic depth or transparency.

Treiman et al. (1995), investigating the special role of rimes in English orthography, agreed with Berndt, pointing out that, while most researchers 'considered consistency to be a dichotomous rather than a continuous variable' (p. 113), it was possible to calculate continuous type and token consistency values for CVC words. They found a large difference in consistency between consonants and vowels, confirming Berndt's result, which had indicated that much of the O-P irregularity of English lies in the vowels. Rimes were shown to be more regular than vowels and to uniquely explain more of the variance for both reaction time and reading errors. However, the study was limited because it considered only a subset of monosyllabic words (CVCs) and computed values only for O-P consistency. The need to provide bidirectional data was demonstrated by Ziegler, Stone, and Jacobs (1996), who evaluated the consistency of French monosyllabic words and demonstrated that, although P-O for the spelling body (rime) was 79% inconsistent, O-P was only 12% inconsistent for rimes. In other words, of the 88% of words that were thought of as being 'consistent' at the O-P rime level of analysis, 77% were actually P-O inconsistent. They contrasted this with English (Ziegler, Stone, & Jacobs, 1997), which was shown to have similar P-O inconsistency (72.3%), but a higher O-P inconsistency (30.7%). The implications of this were elucidated by Stone, Vanhoy, and Van Orden (1997), who demonstrated that both consistency measures matter in visual word perception, concluding that 'perception is a two-way street'. A conflicting phenomenon was observed in foundation literacy (Spencer, 2001) when O-P measures for common words, derived from Berndt, were not significant predictors in a multiple regression model for word reading in young children, whereas P-O values were highly significant.

Kessler and Treiman (2001) revised their earlier study of consistency measures (Kessler & Treiman, 1997) to include both directions, concluding that spelling and reading are not symmetrical and that spelling is always harder because each part of the syllable is less consistent, although spelling is always helped by the sounds in adjacent parts of syllables. They conclude that the most plausible reading and spelling model is one that 'fundamentally operates on the phonemic level, but can take into account the context in which each phoneme is found' (p. 611).

There is now sufficient evidence from inter-language comparisons, using broad classifications, to support the orthographic depth hypothesis (Oney & Goldman, 1984; Frith, Wimmer, & Landerl, 1998; Paulesu et al., 2000; Spencer & Hanley, 2003). There is also increasing evidence from intra-language comparisons that capitalize on the fact that within any deep language, individual words will vary in their relative depth: some words will show one-to-one mapping in a similar manner to words in shallow languages; other words will have components of varying depth. This relative orthographic depth within a language has an effect on both reading and spelling of individual words. Fischer, Shankweiler, and Liberman (1985) and Burt and Butterworth (1996) used a three level categorization of orthographic transparency (high, medium and low) for words in their spelling studies and demonstrated that orthographic transparency had a strong effect on spelling accuracy. Perry, Ziegler, and Coltheart (2002) found a general pattern for spelling: the smaller the contingency value of a phoneme-grapheme relationship, the smaller the likelihood a word containing the grapheme had of being spelt. Treiman et al. (1995) found that their continuous measure of consistency for elements of CVC words accounted for significant percentages of the unique variance for errors in reading.

Although orthographic depth is related to reading and spelling difficulty for words, there is the added complication of grapheme complexity for deep languages and even for some shallow languages. Complexity does not map on to consistency in a straightforward manner (Laxon, Gallagher, & Masterson, 2002), for example, the three-phoneme words 'might' and 'sly' have the same consistency probability for the \ie\ phoneme (0.1), but different levels of grapheme complexity, which is manifested in their word letter-length. In an early study, Gibson, Osser, and Pick (1963) found that three-letter words were easier for young children to read than longer items, but pointed out that increasing letter-length correlates with more complicated orthographic structures within a word. Holding letter-length constant and increasing phoneme-length was found by Rey, Jacobs, Schmidt-Weigand, and Ziegler (1998) to have an intriguing facilitatory effect on word identification times that were shorter for words having a greater number of phonemes. This suggests that grouping letters into graphemes requires additional processing time and, by holding length in letters constant, increasing the number of phonemes amounts to decreasing the average phoneme complexity by reducing the number of letters per phoneme. Rastle and Coltheart (1998) described this as the 'whammy' and 'double whammy' effect of digraphs on naming latencies for non-words, which may provide support for letter by letter serial processing within their dual route cascade computer model of word recognition. This is disputed by Andrews, Woolams, and Bond (2005), who demonstrated an interaction between O-P typicality and the presence of digraphs that was not well simulated by either the dual route or parallel-processing connectionist (Plaut, McClelland, Seidenberg, & Patterson, 1996) models. As Bijeljac-Babic, Millogo, Farioli, and Grainger (2004) suggest, '... the effects of this apparently simple variable are still not completely understood' (p. 411).

Letter-length effects on naming latency were found to be modulated by word frequency by Weekes (1997), with the effect disappearing for high-frequency words. A similar effect has been observed for studies in which neighbourhood size has a facilitatory effect on low-frequency words only (Andrews, 1989; Laxon, Masterson, Pool, & Keating, 1992). Word frequency has long been a major factor controlled for in studies of reading and spelling. Zinna, Liberman, and Shankweiler (1986) found that word reading accuracy was strongly affected by word frequency, which supported the notion that the initial acquisition of word reading skills was accomplished through rote learning, with frequent words usually identified without analysis of their components. Gernsbacher (1984) concluded that over 20 years of research had demonstrated that word recognition was affected by the familiarity of a word, which was operationalized as the frequency with which a word occurs in printed English text. However, in a similar vein to Bijeljac-Babic et al.'s comments on word length, Carroll and White (1973) warned that, 'word frequency may not be the simple variable that it appears to be' (p. 563). Landauer and Streeter (1973) demonstrated that high-frequency words are likely to contain more regularly occurring phonemic and graphemic patterns than low-frequency words, and Gernsbacher suggested that they also differ along semantic and lexicographic dimensions. Balota, Pilotti, and Cortese (2001) find that 'no single variable has been studied more in psycholinguistics and memory research than word frequency' (p. 639), and note that frequency counts are usually based on extensive collections of print samples, which may be quite different from those of estimates of spoken, heard or hand-written frequency. Stuart, Dixon, Masterson, and Gray (2003), recognizing the importance of word frequency in psychological studies, comment on the need to provide recent systematic counts of the contents of children's early reading materials as it cannot be assumed that children's experience with their reading vocabularies does not change over time, nor that it is necessarily similar to that of adults.

Previous studies have demonstrated that the orthographic depth of words, and in particular the level of transparency of individual phonemes, together with word complexity and frequency have significant unique effects on spelling and reading difficulty of words for young children. The present study was undertaken to investigate the relative strength of both O-P and P-O measures of consistency at the fine-grain phoneme-grapheme level, within a narrow frequency band, in predicting spelling difficulty of words for children in the foundation phase of literacy acquisition.

Method

Participants

In the present study, spelling data were collected for all pupils in an urban Hull primary school, which performs at average national levels in English, mathematics and science. The collected data of spelling performance spanned five year groupings (ages 7 to 11 years), for a total of 207 pupils. Data on the average ages and reading quotients for two reading tests for each year group are shown in Table 1. Reading quotients were based on the administration and marking by the school literacy coordinator of Young's reading tests and NFER Group Reading Test II (NFER, 2005). The Group Reading Test (Young, 1999) and Cloze Reading Tests (Young, 1992) are widely used in UK schools for school years 1-3 and years 4-6, respectively. NFER Sentence Completion Forms A or B are recommended for years 2-4, and Forms C or D for years 5 and above (NFER, 2005). A one-way analysis of variance indicated that there were no statistically significant differences in either measure of reading ability across the age range (Young: F(4, 202) = 1.33, p = .26; NFER: F(4, 202) = 1.13, p = .35). The two tests showed statistically significant correlations ranging from 0.68 to 0.75 (p < .001).

Materials and procedure

Previous studies (Spencer 1999, 2000) collected data for words that had been used in national surveys of the UK School Curriculum and Assessment Authority (SCAA) and had a wide variation in frequency. However, this study was concerned with the most frequent 150 words found in British adult print materials (Hofland & Johansson, 1982). These words were selected because they form the basis for adult reading, representing 50% of the total adult token count. The study was designed to track children's spelling over 5 years to the age when it was reasonable to expect children to have mastered the recognition and spelling of most of these essential words.

Carroll, Davies, and Richman (1971) and Stuart et al. (2003) note that their English word frequency counts are heavily skewed towards low frequencies and comment on the difficulty this presents for teachers of English reading and spelling, with content words (nouns, adjectives and verbs with specific meanings) being relatively infrequent compared with function words (articles, prepositions, adverbs, pronouns, conjunctions and auxiliary verbs) in the highest frequency category. Stuart et al. demonstrate that the proportion of function to content words is 6:1 for the first 100 words, changing to 2.2:1 over the 200 most frequent words. The 85 high frequency function words within the first 100 account for approximately 50% of the total token count in the Stuart et al. corpus. The 150 words derive from the Lancaster-Oslo-Bergen (LOB) adult corpus (Hofland & Johansson, 1982) and used in this study conform to the findings of Stuart et al., with a larger proportion of function words (3.3:1). This reflects the structure of the English language: the most frequent words tend to be function words.

Given the extremely skewed nature of word frequencies across reading ages, it was recognized that using target words from a frequency range from a younger age group may result in frequency distortion effects for the older pupil performance data, which would suffer from unacceptable skewness and kurtosis preventing valid analysis of the data across the age range. It was further recognized that even the selection of words from an adult corpus may result in a proportion of the very high frequency words distorting performance results for the older pupils and that should skewness and kurtosis reach unacceptable values for the full 150 words, the highest frequency words would gradually be eliminated from the analysis for all age groups until a continuous band of words was found that was not skewed or kurtic for spelling difficulty across the age range. In fact, the word spelling difficulty dependent variable results were found to be unacceptably skewed for the 150 words for two age groups (Y4 and Y6) and the process, described fully in Results, was applied to produced a band of 120 words (21-140) which was suitable for analysis. In Tables 2-6, correlations are given only for the 120 words selected for the regression analyses.

The words were administered to each of five year groups (ages 7 to 11 years), in order to investigate the robustness of previous models when applied to high frequency words, because frequency has been shown to modulate the effects of a number of word characteristics such as word length and neighbourhood size (Andrews, 1989; Weekes, 1997). The 150 words were randomly assigned to five lists of 30 words, which were administered on 5 consecutive days. Class teachers were chosen to administer the spelling tests because the children were used to receiving instructions and tests from them. It was recognized that variation in word pronunciations could be a source of error and a group session was held with the five teachers to standardize the pronunciations. In the classroom tests the word was read out, followed by an example sentence agreed at the group session and a further repetition of the word. There was no time limit for the test. Pupils wrote their answers on forms with word numbers clearly indicated. The marking for the test demanded complete mastery: pupil responses were marked as incorrect if the spelling was not perfect. The dependent variable was the arcsine transformed value for the proportion of pupils spelling each word correctly [arcsin(SQRT(Number of correct spellings/Number of subjects))].

Measures employed in the study

Graphemic and phonemic length of the word and phonetic difference

Word length can be defined as the number of sounds or phonemes in a word and in a written language with perfect orthographic transparency, having one-to-one correspondence, this measure of word length would be identical to the number of letters in a word. In less perfect orthographies, there will be a difference between the two measures, the phonetic difference (PhD) that is associated with word complexity, which can be obtained by subtracting the number of phonemes in a word from the number of letters used to represent it. In English this is usually a positive value (English orthography generally requires more letters than sounds). In this study the number of phonemes in each word and the phonetic difference, which are both related to letter length (number of phonemes + phonetic difference), were included in the regression analyses.

English is classified into 44 phonemes, 23 of which are represented by the 26 letters of the alphabet. The remaining phonemes are usually represented by a combination of two or more alphabetic characters. According to the corpus of 3,500 words used to calculate transparency values by Spencer (1999), 92% fall within the PhD range 0 to 2.

Orthographic depth for words and word components

There is no standardized way of measuring this factor and a number of approaches have been adopted and refined for this study. Hanna provides spelling correspondences for 52 phonemes (30 consonants, 22 vowels), expressing each combination as a percentage value of the total frequency of the phoneme in the 17,310 word corpus. Fry (2004) has recently commented on the utility of such a classification of phonemes and has proposed a simplified scheme of 42 phonemes (17 vowels and 25 consonants) with correspondences derived from the frequencies provided in the original 1965 report. For the present study, Fry's simplified scheme was used to create phoneme-grapheme probabilities. Each phoneme within a word was assigned a probability value according to its graphemic representation. Two transparency probability values were calculated for each word, in each case large values representing greater transparency. A mean transparency (MT P-O) value for each word was derived, together with a value for the least transparent phoneme (LTP P-O), which was obtained by identifying the component with the smallest probability value, which was generally, but not always, the vowel component. In previous studies (Spencer, 2001) the mean value was found to be a less powerful predictor than the least transparent phoneme value for both reading and spelling difficulty.

Berndt suggested that the Hanna study could form the basis for O-P correspondences and, in a similar vein to Fry, produced values for a reduced phonemic classification scheme of 45 phonemes. This scheme provides values for the calculation of mean transparency (MT O-P) and least transparent phoneme (LTP O-P) values for each of the 150 words. These were originally created by Berndt to provide information for studies of reading, however, given Stone et al.'s (1997) suggestion that both feedback and feedforward consistency influences visual word recognition, both P-O and O-P values are also included in the present study of spelling.

Taking account of McGuinness's (1997) suggestion that language metrics may actually be distorted within large corpora, Spencer (1999) derived phoneme-grapheme probability values from the 7,000 most frequent words in the LOB corpus (Hofland & Johansson, 1982), which were lemmatized by removing inflected forms if the base form also appeared in the list, resulting in a reduced sample of 3,500 words. The phonemic representation of each of these words was determined from the Oxford English Dictionary (second edition, CD-ROM version), which identifies 44 individual phonemes in the English language. Mean and least transparent phoneme P-O values were calculated for each word. Also, the data from Spencer were recalculated, to provide O-P values, as Berndt had done with Hanna's data.

Thus, from the two corpora of different sizes (Hanna et al.: 17,000 words; Spencer: 3,500 words), four probability values, which provide an estimate of transparency for each word, were derived for both feedforward (P-O) and feedback (O-P) directions for spelling the high frequency words. Correlations are high and significant within both transparency measures (Table 2), but lower and generally insignificant between the directions of correspondence. There are also consistent and significant, negative correlations between phonetic difference and P-O measures, but not between phonetic difference and O-P. Phonemic length has weak and generally insignificant correlations with all transparency measures.

Word frequency

As with measures of orthographic depth, word frequency values depend on a number of factors, including the specific language or its variants (e.g. American or British English), the corpus size and the source material from which the data are extracted. Six sources of frequency data for the words were used for 18 separate values per word:

(1) Carroll et al. (1971) (hereafter Carroll) provided eight values for each word from source materials for American grades 3-9 and for the total corpus count (US grade 1 is equivalent to UK year 2, 7-year-olds) (hereafter C3-C9).

(2) Stuart et al.'s (2003) (hereafter Stuart) research project data are available at the associated website and provide four frequency counts for words taken from British children's texts for reception (5-year-olds) to year 3 (hereafter SR, S1, S2 and S3).

(3) Hofland and Johansson (1982) and Leech, Rayson, and Wilson (2001) each provide values based on British adult written material.

(4) Reid's (1989) values were based on two separate counts of British 7- and 8-year-old children's writings (hereafter R7 and R8).

(5) Brown (1984) and Leech et al. (2001) provide values for British adult spoken word counts.

The word frequency within each corpus, which were based on different sample sizes, was recalculated when necessary as frequency per million words and log transformed for the correlation and regression analyses. Table 3 shows the Spearman correlations between the 18 log transformed frequency values for the 120 words used in the regression analyses.

All correlations are significant, with an overall mean of 0.70 and a range of 0.28 to 0.99, suggesting that care should be taken in selecting frequency values for target words for psychological studies involving children. For example, the correlations between SR and C8 to C9 are significant but low, as are the correlations between SR and British adult sources. However, there are strong correlations within sources and also between the British and American children's data for similar age groups.

Word frequency and orthographic transparency interactions

It has been suggested that word frequency may not be such a simple variable as is often anticipated and that certain phonemic and graphemic features may be differentially associated with high or low frequencies. Table 4 shows the Spearman correlations between frequency and orthographic transparency values. Both P-O and O-P transparency values, as well as complexity (PhD), have weak and generally insignificant associations with frequency. Phonemic length has significant, negative correlations with all measures of frequency. For this band of 120 frequent words the significant association is between frequency and phonemic length: more frequent words have a tendency to be phonemically shorter.

Results

The principal objective of this study was to develop a multiple regression model which would predict spelling difficulty of high frequency words for 7- to 11-year-olds, based on a number of predictor variables that have previously been identified as contributing to spelling difficulty. A secondary objective was to determine the robustness of such a model in accounting for spelling difficulty for children of varying ages who find spelling difficult, defined in this instance as the lower quartile group (N = 52) for the entire school cohort (hereafter Q1) and consisting of 40% of pupils from Y2, 32% Y3, 10% Y4, 10% Y5 and 8% Y6.

However, it was found that although the 150 most frequent words were chosen from an adult corpus, providing a suitable range of words for the older children in the study, cumulative frequency effects on the highest frequency words induced unacceptable skewness for the dependent variable for Y4 and Y6 children. In order to meet the requirements for multiple regression, a minimum sample of 100 words was required, given the number of predictor variables that were to be entered in the analysis (see Field, 2005, p. 173). Bands of words, starting with the highest rank (most frequent) word were incrementally (+1) tested for both skew and kurtosis. The 120 word band 21-140 (LOB rank position) was identified as suitable for analysis (see Table 5). There were no missing data and checks within the regression procedures indicated no departure from bivariate normality, homoscedasticity or linearity of bivariate relationships, and no multivariate outliers.

Regression analyses may enter all independent variables simultaneously, with each variable evaluated in terms of what it adds to the prediction that is different from that afforded by all the other variables, or, alternatively, stepwise regressions may be used to include or exclude predictors based purely on mathematical criteria, to provide the most parsimonious model. Tabachnick and Fidell (2001) recommend that stepwise methods are best suited to exploratory model building, although for such models it must be recognized that random sampling variation may cause variables to be included on the basis of slight differences in their semi-partial correlation. Regression models must also be based on reasonable assumptions concerning variables that are likely to contribute to the power of the model in predicting values of the dependent variable, and how these variables interact one with another, especially when variables are highly intercorrelated, which may result in unacceptable multicollinearity. Clearly, for example, the frequency values derived from Carroll's data cannot be simultaneously entered into a regression analysis because of the near perfect correlations between many values; similarly, some values of word transparency are very highly intercorrelated.

Table 6 provides the correlations between the independent predictor variables and the years 2 to 6 (arcsine transformed) measures of word spelling difficulty, and for the Q1 and total school data. It is immediately apparent that spelling difficulty across the 5 year age range has weaker associations with adult measures of word frequency than with data from children's sources, confirming Stuart's criticism of the practice of using adult frequency data as a basis for the selection of experimental materials for children. The British sources show larger correlations with the British children's data than the American source. However, the expectation that spelling difficulty for older children shows larger correlations with frequency data gathered from similar age groups is not met. For the Carroll frequencies, spelling difficulty across the age range shows the largest association with the values from the youngest children's texts (G3). A similar trend is observed for the Stuart frequencies and the Reid frequencies. There are consistent positive correlations between spelling difficulty and P-O measures of transparency, with higher values for LTP measures than for mean word values. Feedback O-P transparency for spelling shows few significant correlations. Word complexity, as measured by PhD, shows large significant negative correlations across the age range, but phonemic length is only weakly associated with spelling difficulty, confirming the suggestion that so-called word length effects (in fact, letter length) are caused by grapheme complexity.

All variables were entered into step-wise (forward) multiple regression analyses, with PhD values being entered as a series of three dichotomous variables, combining PhDs of 3 and 4 into a single variable, each measured against the one-to-one mapped condition (PhD = 0). Table 7 shows the final model for each year group (Y2-Y6), total school and Q1 group. The total variance accounted for ranged from 0.52 for years 4 and 5 to 0.72 for the Q1 group (Y2, [R.sup.2] = .66, F(6, 113) = 36.74, p < .001; Y3, [R.sup.2] = .65, F(4, 115) = 42.52,p < .001; Y4, [R.sup.2] = .52, F(4, 115) = 31.01,p < .001; Y5, [R.sup.2] = .52, F(3, 116) = 42.27, p < .001; Y6, [R.sup.2] = .58, F(5, 114) = 32.11, p < .001; total school, [R.sup.2] = .71, F(4, 115) = 65.40, p < .001; Q1, [R.sup.2] = .72, F(6, 113) = 47.84, p < .001).

In all cases, across the age range, the Spencer LTP P-O variable was selected for the most parsimonious model. The Stuart frequency source, usually S2, was selected in six of the seven analyses and complexity was a significant factor in all of the models. The results are generally predictable from the correlation analyses. However, although some variables initially account for a large proportion of the variance, as indicated by the variance change statistic ([DELTA][R.sup.2]), the influence of the variable may change because it also shares a proportion of its variance through its association with other variables. This change in influence within the model is reflected by the final model's standardized beta ([beta]) and the unique variance ([sr.sup.2]) statistics. Although the 120 words were selected from a narrow, high frequency band, the frequency variable still has a highly significant impact on spelling difficulty for all groups, with a unique contribution to variance of between 0.10 and 0.28. However, the three dichotomous complexity variables together account for a larger proportion of the variance (or the same amount) in six of the models. There is a general age trend for a reduced influence of the lower orders of complexity. Phonemic length has a small but statistically significant effect for years 2 and 3, and also for the poorest spelling Q1 group and combined school model. It appears that the number of phonemes in a word in the selected frequency band does not influence spelling difficulty of words for older children. Over the Y3 to Y6 age range, only P-O word transparency has a significant unique influence on spelling difficulty, ranging from 0.04 to 0.17, which does not accord with Stone et al.'s (1997) suggestion that perception is a two-way street.

Of course, these are not the only models that will account for a significant proportion of the variance based on the predictor variables in Tables 3-6, they simply represent the best mathematical fit between the dependent and independent variables. Similar models may be constructed using other measures of the independent variables. For example, for Y2 spelling difficulty as the dependent variable, but entering US-derived LTP values from the larger scale Hanna study and frequency values for G3 taken from Carroll, the model still predicts a substantial proportion of the variability ([R.sup.2] = .61; F(7, 112) = 24.69, p < .001). However, the model predicting the largest proportion of the variability in word spelling difficulty across the five school years is based on frequency and transparency data from the two comparatively small British studies (Spencer, 1999; Stuart et al., 2003). Nevertheless, it is pertinent to enquire to what extent the original hypothesized variables remain as significant predictors when the source of the values is changed. If the same variables, but based on different sources, produce essentially the same model, with slight variations in total and unique variances the model may be considered to be robust and may be thought, in this case, to have general educational significance, which may extend to proposals for techniques to ameliorate poor spelling. Going beyond the need for exploratory analysis, Table 8 shows the results from 15 separate regression analyses for Y2 to Y6, using a variety of combinations of variable sources, but entering all variables simultaneously into the analysis. Standardized beta values and unique contribution to variance ([sr.sup.2]) are presented, together with total variance explained (R2). The principal concern here was the use of age-appropriate values for the frequency variable and culture-commensurate transparency values. For Carroll's US sources, there are equivalent data for each of the age groups considered, apart from Y2 and Y3 which use the closest match (G3). This group of analyses also uses transparency values based on the US data from Hanna and Berndt. For the Stuart model Y3 frequency is the closest match for Y4 and above. The adult frequency values from Leech are also included to illustrate the magnitude of effects from appropriate culture, but inappropriate sources for frequency counts when dealing with children's data. For these analyses of British frequency effects the associated LTP P-O/O-P values are taken from Spencer.

The UK and US models are similar across the age range, with a reduction in the total variance predicted by the US data of 5-9%. Frequency in all of these models accounts for significant individual proportions of the variance, as does P-O transparency, and there is a similar tendency for a gradual reduction in the impact of the lower orders of complexity from Y2 to Y5. Phonemic length ceases to be a significant variable after Y3. However, the results from Leech's recent adult English word frequency count suggests that frequency is not a significant variable across the age range for the selected band of words, reflecting its general lack of significant association with spelling difficulty. This series of five analyses shows only a single significant result for adult frequency values, in contrast to the substantial contribution of frequency in all other models based on children's values, further substantiating Stuart's demands for age-related measures of frequency in psychological studies. Although complexity and transparency are similar in the three models for the five class years, phonemic length plays a more substantial role up to Y5 in the Leech model.

Although the models based on US-derived values for transparency and frequency account for slightly less total variance than the UK models, they substantially support the British results based on children's frequency data, and suggest that the general model is indeed robust and should be taken into account when developing remedial literacy theories.

Discussion

The results of the present study fit well with current theories of reading and spelling, and extend the theoretical perspective. Frequency, even within the selected narrow bandwidth, is seen as having a modulating effect on spelling difficulty for a wide range of children. However, this effect only emerges from appropriate frequency measures, which must take account of the subjects' age and culture. Feedforward consistency/regularity is confirmed as exerting an influence, in this case measured as a continuous transparency variable, at the fine-grain grapheme-phoneme level, but feedback effects seldom appear and when they do, they have a minor influence, representing 1-2% of the unique variance. P-O transparency is shown to be highly correlated with word complexity, which is responsible for the large 'whammy' effect observed by Rastle and Coltheart (1998) on reading latencies. These results, together, predict 'whammy' reading and spelling effects even in transparent languages having some regular but complex graphemes, a phenomenon that has been observed by Cossu, Gugliotta, and Marshall (1995) with Italian children. The meaning of the 'length effect' is changed by these studies, and future experimental work on this variable must recognize that within both deep and shallow orthographies grapheme complexity will exert an influence on dependent variables. In this study, word length, through the complexity (PhD) variable rather than phoneme length, proves to be a powerful predictor of word spelling difficulty. The results support Weekes' (1997) findings that many of the statistical properties of words are significantly associated, and that the effect of number of letters reported in many studies 'may be an artefact of intercorrelations between number of letters and confounding variables' (p. 448). The present study demonstrates that the complexity effect is independent of frequency and transparency.

Extrapolating from computer models for reading and spelling to human behaviour can be hazardous because crucial features may be poorly defined in the models. For example, Rastle and Coltheart (1999) indicate a lack of knowledge concerning the development in humans of their DRC model's GPC rules, and Houghton and Zorzi (2003) admit that the nature of the lexical representations in their dual-route connectionist spelling model is only vaguely defined, and that the assumption that all known words are equally well learned may be false. There are broadly two categories of simulation models: dual route serial processing and single route parallel processing. Dual route theories have a direct mental lexical route and a grapheme-phoneme conversion (GPC) route for new words. Serial models deal with the GPC route on a letter by letter approach with reference to rules that are but a 'set of hypotheses about what GPC rules skilled readers possess' (Rastle & Coltheart, 1999, p. 484). The alternative to this comes from connectionist or parallel distributed processing networks that learn by adjusting weights on connections so there is no sharp dichotomy between items that obey the rules and those that do not, and there is no lexicon (Plaut et al., 1996). Serial processing, grounded in GPC rules, occurs letter by letter, whereas connectionist models permit access to sublexical orthographic units of various sizes and structures in parallel. However, there is increasing acceptance of a developmental trend in children from serial to parallel processing (Bijeljac-Babic et al., 2004).

Rastle and Coltheart's DRC-L model, which produced the counter-intuitive effects of digraphs, may provide a bridge between computational models simulating adult reading latency performance, and children's reading and spelling performance, and this may refocus the orthographic depth hypothesis debate from transparency to complexity. The fact that adult processing time is increased for words with greater complexity suggests that children attempting to read words with new and complex graphemes will also require additional processing time as they struggle to resolve the conflicting results of serial decoding. Resolution of the problem can only be achieved, according to the model, by having access to a suitable set of GPC rules. Complex graphemes, if unresolved because the relevant rule is not in place, may block serial processing, leaving other strategies such as guessing from context, as the only way forward for the young reader (see Share, 1995).

Initially, reading of all complex graphemes may be problematic because: (1) serial processing demands are high; (2) the children have not extracted the GPC rules from their reading materials; and (3) they may not have been explicitly taught the relevant rules, having been exposed only to alphabetic names and sounds. This will change over time as a result of frequency modulation and teaching effects (see Compton, 2000). Increased frequency of contact with words should result in deduced knowledge of some GPC rules, and explicit teaching of the regular GPCs for complex graphemes should also extend this knowledge base. This increased rule set will consolidate a complex word's lexical entry by enhancing serial processing. However, most opaque complexes will remain problematic because the necessary GPC rules are neither taught nor deduced, resulting in serial processing being abandoned. Assuming common lexical representation for both reading and spelling (Burt & Butterworth, 1996), written spelling of such items will result in only partial success because there is not a fully defined lexical entry. Words lacking a well-defined lexical entry must be spelled by a phoneme-grapheme correspondence route, using PGC rules, which overlap with, but are not identical to, GPC rules.

Burt and Fury (2000) do not see this approach as a valid interpretation of the spelling process and conclude that spelling depends upon the quality of an individual's learning about the orthography of words rather than memory access or the application of spelling rules: 'failure to produce the correct spelling of a word is more likely to reflect the fact that the correct spelling has not been encoded securely in memory, rather than a difficulty in accessing the memory for the correct spelling' (p. 2). This appears to rule out dual route models, such as the upside-down spelling model of Perry (2003), a two-process system based on an inverted DRC model. However, the result from a single process model should show only strong feedback effects for spelling, whereas the current model has a dominant feedforward effect, confirmed by Weekes, Castles, and Davies (2006) who also failed to find consistent feedback effects.

This raises the further issue of differences between the rule sets for GPCs and PGCs. The precision of GPC rules may dictate the quality of the stored orthographic lexical representation that represents one route for word spelling. For words whose spelling has not been securely encoded in the lexicon, a two-process spelling system will call up PGC rules, and similarly, the precision of these rules and their overlap with GPC rules will dictate the quality of the written output. Compton (2000) provides good evidence for the progressive nature of GPC acquisition, demonstrating the increased predictive power, over time, of knowledge of complex graphemes when compared with other measures of cognitive-processing skill, including letter sound knowledge, for both reading and spelling: of all the variables considered only growth in advanced graphophoneme (complex graphemes) knowledge was found to be a unique predictor of non-word reading growth rate. Less is known about the development of PGC rules, although initially they will probably be the same as reading rules.

The present study indicates that word spelling difficulty for common English words may at first be affected by phonemic length, but that this is very minor for 8-year-olds and ceases to exert an influence on 9-year-olds. Grapheme transparency and complexity exert an influence across the primary years in English schools, with higher levels of complexity being the dominant force. In addition, increased word frequency, as usual, has a positive effect on literacy, even within the narrow confines of the common words considered in this study. However, frequency does not suppress transparency or complexity effects as has been reported in reading latency studies. These results may offer support for amelioration techniques, such as synthetic phonics in the foundation literacy phase. Grapheme transparency has an effect on word difficulty, but it is overshadowed by complexity effects: complex words, having one or more di- or trigraphs, are more difficult to read (Laxon et al., 2002) and spell than words with one-to-one mapping. Synthetic phonics teaches digraphs (TH, SH, CH) on an equal footing with single letters, and they are rapidly introduced in the foundation phase, often within 8 weeks in a reception class (5-year-olds), and 'as a result these short sequences become chunked and available for use in representing longer sequences' (Houghton & Zorzi, 2003, p. 152). This approach increases non-word reading performance in young children, but does not limit them to reading regular words, it also accelerates their acquisition of a sight vocabulary. The children learn 40+ GPCs, half of which are digraphs, and this means that words recognized by sight are richly underpinned in memory by connections between graphemes in the spelling and phonemes in the pronunciation (Johnston & Watson, 2004). This fostering of accurate or complete knowledge of word spellings during reading facilitates the retention of letter sequences for spelling even for orthographically opaque words (Burt & Butterworth, 1996).

The results clearly indicate that orthographic inconsistencies do impact on word spelling difficulty for children in the early years of UK schooling, but that orthographic complexities have a greater impact. It follows from this that both orthographic transparency and complexity contribute to differences in foundation literacy acquisition and that future cross-cultural studies should include measures of advanced graphophoneme knowledge, as well as knowledge of alphabet sounds (Seymour et al., 2003).

References

Andrews, S. (1989). Frequency and neighbourhood effects on lexical access: Activation or search? Journal of Experimental Psychology: Learning Memory and Cognition, 15, 802-814.

Andrews, S., Woolams, A., & Bond, R. (2005). Spelling-sound typicality only affects words with digraphs: Further qualifications to the generality of the regularity effect on word naming. Journal of Memory and Language, 53, 567-593.

Balota, D. A., Pilotti, M., & Cortese, M. J. (2001). Subjective frequency estimates for 2,938 monosyllabic words. Memory and Cognition, 29(4), 639-647.

Berndt, R. T., Reggia, J. A., & Mitchum, C. C. (1987). Empirically derived probabilities for grapheme-to-phoneme correspondences in English. Behaviour Research Methods, Instruments, and Computers, 19(1), 1-9.

Bijeljac-Babic, R., Millogo, V, Farioli, F., & Grainger, J. (2004). A developmental investigation of word length effects in reading using a new on-line word identification paradigm. Reading and Writing: An Interdisciplinary Journal, 17, 411-431.

Brown, G. D. A. (1984). A frequency count of 190,000 words in the London-lund corpus of English conversation. Behavioural Research Methods Instrumentation and Computers, 16(6), 502-532.

Burt, J. S., & Butterworth, P. (1996). Spelling in adults: Orthographic transparency, learning new letter strings and reading accuracy. European Journal of Cognitive Psycholgy, 8(1), 3-43.

Burt, J. S., & Fury, M. B. (2000). Spelling in adults: The role of reading skills and experience. Reading and Writing: An Interdisciplinary Journal, 13, 1-30.

Carney, E. (1994). A survey of English spelling. London: Routledge.

Carroll, J. B., Davies, P., & Richman, B. (1971). The American heritage word frequency book. Boston: Houghton Mifflin Company.

Carroll, J. B., & White, M. N. (1973). Word frequency and age of acquisition as determiners of picture-naming latency. Quarterly Journal of Experimental Psychology, 25, 85-95.

Compton, D. L. (2000). Modeling the growth of decoding skills in first-grade children. Scientific Studies of Reading, 4(3), 219-259.

Cossu, G., Gugliotta, M., & Marshall, J. C. (1995). Acquisition of reading and written spelling in a transparent orthography: Two non parallel processes? Reading and Writing: An Interdisciplinary Journal, 7, 9-22.

Field, A. (2005). Discovering statistics using SPSS. London: Sage Publication.

Fischer, F. W., Shankweiler, D., & Liberman, I. (1985). Spelling proficiency and sensitivity to word structure. Journal of Memory and Language, 24, 423-441.

Frith, U., Wimmer, H., & Landerl, K. (1998). Differences in phonological recoding in German- and English-speaking children. Scientific Studies of Reading, 2(1), 31-54.

Fry, E. (2004). Phonics: A large phoneme-grapheme frequency count revised. Journal of Literacy Research, 36(1), 85-98.

Gernsbacher, M. A. (1984). Resolving 20 years of inconsistent interactions between lexical familiarity and orthography, concreteness, and polysemy. Journal of Experimental Psychology: General, 113(2), 256-281.

Gibson, E., Osser, H., & Pick, A. D. (1963). A study of the development of grapheme-phoneme correspondences. Journal of Verbal Learning and Verbal Behavior, 2, 142-146.

Glushko, R. J. (1979). The organization and activation of orthographic knowledge in reading aloud. Journal of Experimental Psychology: Human Perception and Performance, 6, 674-691.

Goswami, U., Porpodas, C., & Wheelwright, S. (1997). Children's orthographic representations in English and Greek. European Journal of Psychology of Education, 7(3), 273-290.

Hanna, P. R., Hanna, J. S., & Hodges, R. E. (1966). Phoneme-grapheme correspondences as cues to spelling improvement. Washington: US Department of Health, Education and Welfare [OE-32008].

Hofland, K., & Johansson, S. (1982). Word frequencies in British and American English. Bergen: The Norwegian Computing Centre for the Humanities.

Houghton, G., & Zorzi, M. (2003). Normal and impaired spelling in a connectionist dual-route architecture. Cognitive Neuropsychology, 20(2), 115-162.

Johnston, R. S., & Watson, J. E. (2004). Accelerating the development of reading, spelling and phonemic awareness skills in initial readers. Reading and Writing: An Interdisciplinary Journal, 17, 327-357.

Katz, L., & Frost, R. (1992). The reading process is different for different orthographies: The orthographic depth hypothesis. In R. Frost & L. Katz (Eds.), Orthography, phonology, morphology, and meaning (pp. 67-84). Amsterdam: Elsevier Science Publishers.

Kessler, B., & Treiman, R. (1997). Syllable structure and the distribution of phonemes in English syllables. Journal of Memory and Language, 37, 295-311.

Kessler, B., & Treiman, R. (2001). Relationships between sounds and letters in English monosyllables. Journal of Memory and Language, 44, 592-617.

Landauer, T., & Streeter, L. (1973). Structural differences between common and rare words: Failure of equivalence assumptions for theories of word recognition. Journal of Verbal Learning and Verbal Behaviour, 12, 119-131.

Laxon, V., Gallagher, A., & Masterson, J. (2002). The effects of familiarity, orthographic neighbourhood density, letter length and graphemic complexity on children's reading accuracy. British Journal of Psychology, 93, 269-287.

Laxon, V., Masterson, J., Pool, M., & Keating, C. (1992). Nonword naming: Further exploration of the psuedohomophone effect in terms of orthographic neighbourhood size, graphemic changes, spelling-sound consistency and reader accuracy. Journal of Experimental Psychology: Learning, Memory and Cognition, 18, 730-748.

Leech, G., Rayson, P., & Wilson, A. (2001). Word frequencies in written and spoken English: Based on the British national corpus. London: Longman.

Liberman, I. Y., Liberman, A. M., Mattingly, I. G., & Shankweiler, D. L. (1980). Orthography and the beginning reader. In J. F. Kavanagh & R. L. Venezky (Eds.), Orthography, reading and dyslexia (pp. 137-153). Baltimore: University Park Press.

McGuinness, D. (1997). Why children can't read: And what we can do about it. London: Penguin.

NFER. (2005). Group reading test II. London: NFER-Nelson.

Oney, B., & Goldman, S. R. (1984). Decoding and comprehension skills in Turkish and English: Effects of regularity of grapheme-phoneme correspondences. Journal of Educational Psychology, 76(4), 557-568.

Paulesu, E., McCrory, E., Fazio, F., Menoncello, L., Brunswick, N., Cappa, S. F., Cotelli, M., Cossu, G., Corte, F., Lorusso, M., Pesenti, S., Gallagher, A., Perani, D., Price, C., Frith, C. D., & Frith, U. (2000). A cultural effect on brain function. Nature Neuroscience, 3(1), 91-96.

Perry, C. (2003). Priming the rules of spelling. Quarterly Journal of Experimental Psychology, 56A(3), 515-530.

Perry, C., Ziegler, J. C., & Coltheart, M. (2002). How predictable is spelling? Developing and testing metrics of phoneme-grapheme contingency. Quarterly Journal of Experimental Psychology, 55A(3), 897-915.

Plaut, D. C., McClelland, J. L., Seidenberg, M. S., & Patterson, K. (1996). Understanding normal and impaired word reading: Computational principles in quasi-regular domains. Psychological Review, 103(1), 56-115.

Rastle, K., & Coltheart, M. (1998). Whammies and double whammies: The effect of length on nonword reading. Psychonomic Bulletin and Review, 5(2), 277-282.

Rastle, K., & Coltheart, M. (1999). Serial and strategic effects in reading aloud. Journal of Experimental Psychology: Human Perception and Performance, 25(2), 482-503.

Reid, D. (1989). Word for word: The top 2,000 words used by 7 and 8-year-olds. Wisbech: Learning Development Aids.

Rey, A., Jacobs, A. M., Schmidt-Weigand, F., & Ziegler, J. C. (1998). A phoneme effect in visual word recognition. Cognition, 68, B71-B80.

Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in European orthographies. British Journal of Psychology, 94, 143-174.

Seymour, P. H. K., & Duncan, L. G. (2004, March). Reading acquisition and dyslexia in the European orthographies. Paper presented at the Sixth BDA International Conference Dyslexia: The Dividends from Research to Practice, University of Warwick, UK.

Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, 55, 151-218.

Spencer, K. A. (1999). Predicting word-spelling difficulty in 7- to 11-year-olds. Journal of Research in Reading, 22(3), 283-292.

Spencer, K. A. (2000). Is English a dyslexic language? Orthographic transparency and multilingual teaching. Dyslexia, 6, 152-162.

Spencer, K. A. (2001). Differential effects of orthographic transparency on dyslexia: Word reading difficulty for common English words. Dyslexia, 7, 217-228.

Spencer, L. H., & Hanley, J. R. (2003). Effects of orthographic transparency on reading and phoneme awareness in children learning to read in Wales. British Journal of Psychology, 94, 1-28.

Stone, G. O., Vanhoy, M., & Van Orden, G. C. (1997). Perception is a two-way street: Feedforward and feedback phonology in visual word recognition. Journal of Memory and Language, 36, 337-359.

Stuart, M., Dixon, M., Masterson, J., & Gray, B. (2003). Children's early reading vocabulary: Description and word frequency lists. British Journal of Educational Psychology, 73, 585-598.

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). New York: HarperCollins Publishers.

Treiman, R., Mullenix, J., Bijeljac-Babic, R., & Richmond-Welty, E. D. (1995). The special role of rimes in the description, use, and acquisition of English orthography. Journal of Experimental Psychology: General, 124, 107-136.

Venezky, R. L. (1970). The structure of English orthography. Moulton: The Hague.

Weekes, B. S. (1997). Differential effects of numbers of letters on word and nonword naming latency. Quarterly Journal of Experimental Psychology, 50A(2), 439-456.

Weekes, B. S., Castles, A. E., & Davies, R. A. (2006). Effects of consistency and age of acquisition on reading and spelling among developing readers. Reading and Writing, 19(2), 133-169.

Wijk, A. (1966). Rules of pronunciation of the English language. London: Oxford University Press.

Young, D. (1992). Cloze reading test manual. London: Hodder & Stoughton.

Young, D. (1999). Group reading test teacher's manual. London: Hodder & Stoughton.

Ziegler, J. C., Stone, G. O., & Jacobs, A. M. (1996). Statistical analysis of the bidirectional inconsistency of spelling and sound in French. Behavior Research Methods, Instruments and Computers, 28(4), 504-515.

Ziegler, J. C., Stone, G. O., & Jacobs, A. M. (1997). What is the pronunciation for -ough and the spelling for /u/? A database for computing feedforward and feedback consistency in English. Behavior Research Methods, Instruments and Computers, 29(4), 600-618.

Zinna, D. R., Liberman, I. Y, & Shankweiler, D. (1986). Children's sensitivity to factors influencing vowel reading. Reading Research Quarterly, 21, 465-479.

Received 5 October 2005; revised version received 5 June 2006
Appendix: Word characteristics and correct responses per word Y2-Y6 and
Q1

           Frequency     Frequency
           (Hofland and  Y2, (Stuart  LTP P-O,   MT P-O,    LTP O-P,
Word       Johnasson)    et al.)      (Spencer)  (Spencer)  (Spencer)

  the      68315         70776        0.19       0.59       0.21
  of       35716         10431        0.01       0.47       0.01
  and      27856         31535        0.93       0.97       0.34
  to       26760         23136        0.09       0.52       0.02
  in       21108         13067        0.54       0.74       0.88
  that     11188          4788        0.95       0.98       0.34
  is       10978          5794        0.54       0.63       0.15
  was      10499         12815        0.04       0.47       0.01
  it       10010         12111        0.54       0.74       0.88
  for       9299          5663        0.45       0.64       0.62
  he        8776         14043        0.19       0.57       0.04
  as        7337          3299        0.71       0.86       0.15
  with      7197          5130        0.54       0.73       0.37
  be        7186          3460        0.19       0.59       0.04
  on        7027          8319        0.93       0.93       0.40
  his       6266          5543        0.54       0.73       0.15
  at        6043          5794        0.95       0.97       0.34
  by        5796          1670        0.10       0.55       0.06
  had       5391          4828        0.95       0.98       0.34
  this      5287          3279        0.54       0.73       0.37
* not       5142          2917        0.93       0.94       0.40
* but       4956          5402        0.67       0.87       0.34
* from      4686          2022        0.83       0.91       0.40
* have      4592          3551        0.26       0.74       0.34
* are       4544          3662        0.01       0.01       0.04
* which     4467           382        0.13       0.40       0.75
* her       4030          4426        0.46       0.71       0.22
* she       3912          8037        0.19       0.22       0.04
* or        3781          1016        0.45       0.45       0.62
* you       3590         10975        0.03       0.07       0.03
* they      3579          8872        0.01       0.51       0.37
* an        3467          1650        0.93       0.97       0.34
* were      3400          3501        0.01       0.33       0.05
* there     3180          3893        0.17       0.59       0.37
* been      3116           895        0.28       0.73       0.97
* one       3088          3309        0.01       0.32       0.26
* all       2940          4798        0.09       0.11       0.02
* we        2926          4084        0.19       0.42       0.04
* their     2808          1267        0.02       0.51       0.37
* has       2802           875        0.71       0.89       0.15
* would     2682          1278        0.04       0.56       0.99
* when      2540          3169        0.13       0.65       0.46
* if        2479          1660        0.54       0.69       0.88
* so        2413          3380        0.51       0.58       0.15
* no        2393          3289        0.51       0.72       0.15
* will      2269          3058        0.09       0.42       0.88
* him       2258          2817        0.54       0.81       0.88
* who       2200          1278        0.05       0.07       0.02
* more      2183           764        0.08       0.51       1.00
* said      2074         20913        0.00       0.54       0.05
* out       2035          4084        0.70       0.82       0.57
* about     1895          1690        0.28       0.73       0.38
* what      1872          4346        0.04       0.37       0.01
* up        1860          5201        0.67       0.81       0.34
* some      1851          2605        0.08       0.56       0.30
* my        1813          4527        0.10       0.52       0.06
* only      1813           533        0.24       0.60       0.15
* them      1699          2786        0.90       0.95       0.37
* can       1675          4265        0.67       0.87       0.34
* into      1657          2364        0.18       0.65       0.35
* time      1654          2012        0.42       0.77       0.96
* than      1646           443        0.93       0.98       0.34
* could     1614          1841        0.04       0.56       0.82
* me        1554          3430        0.19       0.57       0.04
* two       1549           956        0.01       0.48       0.97
* then      1546          3028        0.90       0.94       0.37
* other     1533           674        0.15       0.44       0.06
* its       1516           885        0.54       0.71       0.84
* these     1501           593        0.07       0.60       0.15
* now       1489          1941        0.28       0.61       0.38
* do        1478          2686        0.09       0.53       0.02
* may       1447           865        0.09       0.52       0.97
* any       1416           503        0.01       0.40       0.01
* made      1294          1569        0.30       0.74       0.99
* first     1287           724        0.14       0.64       0.84
* should    1276           332        0.04       0.42       0.99
* over      1264          1479        0.15       0.47       0.15
* very      1229          2756        0.24       0.70       0.46
* our       1228          1237        0.02       0.36       0.02
* like      1205          3722        0.13       0.42       0.96
* new       1181           966        0.06       0.50       0.54
* must      1129           865        0.65       0.80       0.34
* such      1125           211        0.53       0.62       0.34
* after     1119          1157        0.15       0.59       0.06
* man       1072          2243        0.93       0.96       0.34
* much      1071           412        0.53       0.71       0.34
* before    1061           422        0.08       0.52       0.27
* most      1059           292        0.51       0.76       0.15
* where     1033          1489        0.13       0.15       0.58
* many      1029           382        0.01       0.53       0.01
* well      1009           855        0.09       0.54       0.46
* even       999           272        0.07       0.58       0.97
* also       994           211        0.13       0.50       0.02
* being      961           241        0.19       0.58       0.04
* those      957           251        0.24       0.65       0.15
* people     953          1489        0.00       0.49       1.00
* way        941           825        0.09       0.37       0.97
* back       934          2485        0.04       0.68       0.34
* too        923          2032        0.12       0.54       0.55
* good       900          1549        0.25       0.70       0.40
* little     895          3873        0.01       0.34       0.88
* down       885          3531        0.28       0.73       0.38
* did        884          1670        0.54       0.83       0.88
* how        873          1197        0.28       0.62       0.38
* last       870           624        0.42       0.69       0.06
* between    867            50        0.19       0.67       0.27
* your       853          1620        0.04       0.07       0.03
* just       836          1972        0.25       0.63       0.34
* still      823           724        0.09       0.56       0.84
* work       819           644        0.08       0.28       0.08
* see        814          3269        0.28       0.46       0.84
* know       798          1187        0.01       0.09       0.60
* make       796          1107        0.13       0.46       1.00
* might      781           201        0.09       0.66       0.97
* because    777           473        0.00       0.39       0.06
* through    773           925        0.01       0.64       0.17
* same       768           131        0.30       0.63       0.84
* own        751           201        0.17       0.55       0.60
* long       750          1147        0.58       0.75       0.40

* get        735          2615        0.86       0.90       0.46
* here       717          1710        0.06       0.51       0.37
* go         714          3581        0.51       0.69       0.15
* great      702           342        0.01       0.68       0.03
* three      697           372        0.28       0.73       0.62
* never      686           825        0.15       0.68       0.46
* life       684           231        0.42       0.66       0.96
* year       677           211        0.09       0.10       0.03
* come       676          2525        0.08       0.57       0.30
* both       675           101        0.51       0.84       0.15
* old        670          2374        0.51       0.74       0.15
* another    668           754        0.15       0.51       0.06
* without    665           121        0.54       0.77       0.37
* again      663          1247        0.00       0.52       0.05
* us         657           875        0.67       0.69       0.15
* each       654           342        0.34       0.44       0.53
* used       648           151        0.00       0.36       0.12
* came       645          2615        0.30       0.64       0.82
* under      645           443        0.15       0.68       0.34
* take       639           966        0.13       0.46       0.97
* say        635           724        0.09       0.37       0.84
  found      624           835        0.70       0.86       0.57
  though     623            80        0.01       0.50       0.17
  men        616           936        0.90       0.93       0.46
  thought    611          1036        0.03       0.66       0.50
  day        600          1821        0.09       0.54       0.97
  right      599           422        0.09       0.65       0.97
  went       594          4657        0.64       0.86       0.46
  world      594           191        0.08       0.61       0.08
  while      590           332        0.13       0.42       0.81
  course     588            50        0.03       0.25       0.33

                                                Number of correct
           MT O-P,                              responses per word
Word       (Spencer)  Phonemes  Letters  PhD  Y2  Y3  Y4  Y5  Y6  Q1

  the      0.29       2         3        1    31  46  35  39  56  52
  of       0.20       2         2        0    21  23  34  36  53  30
  and      0.77       3         3        0    30  46  35  39  56  51
  to       0.49       2         2        0    27  44  24  34  52  49
  in       0.92       2         2        0    31  46  35  39  56  52
  that     0.56       3         4        1    25  42  35  37  56  40
  is       0.51       2         2        0    31  46  35  39  56  52
  was      0.39       3         3        0    30  41  35  38  53  43
  it       0.92       2         2        0    31  46  35  39  56  52
  for      0.81       2         3        1    24  42  31  35  52  35
  he       0.52       2         2        0    31  45  35  38  55  50
  as       0.24       2         2        0    28  43  35  38  54  43
  with     0.75       3         4        1    22  39  32  39  55  33
  be       0.52       2         2        0    26  43  32  38  56  42
  on       0.69       2         2        0    31  46  35  38  56  52
  his      0.67       3         3        0    29  45  35  36  56  46
  at       0.65       2         2        0    30  46  34  37  56  51
  by       0.53       2         2        0    19  33  28  35  54  21
  had      0.78       3         3        0    28  44  34  38  55  46
  this     0.70       3         4        1    22  42  34  38  56  37
* not      0.78       3         3        0    30  45  34  39  56  51
* but      0.77       3         3        0    29  45  35  39  56  49
* from     0.84       4         4        0    27  43  34  37  55  43
* have     0.78       3         4        1    26  43  34  39  56  43
* are      0.04       1         3        2    22  38  33  37  53  31
* which    0.81       3         5        2     9  13  22  33  43   8
* her      0.61       2         3        1    28  41  31  35  51  39
* she      0.52       2         3        1    29  39  35  39  55  44
* or       0.62       1         2        1    22  42  32  30  55  35
* you      0.05       2         3        1    27  45  32  39  56  45
* they     0.43       2         4        2    20  38  30  37  52  30
* an       0.66       2         2        0    28  46  34  37  54  46
* were     0.53       2         4        2    16  28  15  25  37  19
* there    0.47       2         5        3    14  29  28  31  42  17
* been     0.99       3         4        1    12  30  30  34  49  16
* one      0.55       3         3        0    28  45  34  39  55  46
* all      0.51       2         3        1    25  37  34  38  54  35
* we       0.52       2         2        0    30  43  35  38  55  47
* their    0.68       2         5        3     8  14  19  27  36   2
* has      0.50       3         3        0    26  41  35  37  56  41
* would    1.00       3         5        2     8  27  29  35  52  10
* when     0.75       3         4        1    21  38  33  38  54  29
* if       0.93       2         2        0    28  45  35  39  54  47
* so       0.50       2         2        0    24  40  35  37  56  40
* no       0.56       2         2        0    29  43  31  37  55  49
* will     0.96       3         4        1    29  41  35  39  54  43
* him      0.96       3         3        0    30  41  35  38  56  45
* who      0.11       2         3        1    14  28  28  32  50  12
* more     1.00       2         4        2    21  37  34  36  53  27
* said     0.63       3         4        1    21  40  34  38  55  36
* out      0.77       2         3        1    22  42  35  37  55  36
* about    0.73       4         5        1    17  33  31  35  54  19
* what     0.59       3         4        1    21  36  33  37  54  26
* up       0.67       2         2        0    29  46  35  39  55  49
* some     0.71       3         4        1    19  38  34  36  53  30
* my       0.53       2         2        0    30  41  35  35  55  45
* only     0.76       4         4        0    12  29  30  29  51  11
* them     0.61       3         4        1    25  44  35  39  51  42
* can      0.71       3         3        0    29  46  34  39  56  50
* into     0.79       4         4        0    29  44  35  39  56  49
* time     0.97       3         4        1    20  40  34  36  54  31
* than     0.56       3         4        1    22  41  30  35  53  32
* could    0.94       3         5        2    15  28  26  36  52   9
* me       0.52       2         2        0    31  43  34  39  56  50
* two      0.98       2         3        1    23  44  33  35  50  37
* then     0.60       3         4        1    24  43  35  38  56  42
* other    0.38       3         5        2    13  25  30  33  46   6
* its      0.89       3         3        0    25  42  33  38  55  38
* these    0.50       3         5        2     9  30  25  32  53   9
* now      0.68       2         3        1    21  46  34  36  51  37
* do       0.51       2         2        0    26  42  34  36  55  38
* may      0.99       2         3        1    21  40  31  37  55  34
* any      0.63       3         3        0     6  21  27  29  48   6
* made     1.00       3         4        1    17  38  34  32  53  26
* first    0.94       4         5        1    14  28  30  29  54  12
* should   1.00       3         6        3     8  21  24  30  47   2
* over     0.62       3         4        1    17  39  32  35  52  25
* very     0.84       4         4        0    21  37  31  36  55  26
* our      0.30       2         3        1    17  30  31  33  51  15
* like     0.99       3         4        1    23  36  35  38  55  33
* new      0.76       3         3        0    21  39  34  35  53  30
* must     0.79       4         4        0    20  37  29  34  54  25
* such     0.64       3         4        1    18  31  30  29  49  19
* after    0.68       4         5        1    18  38  31  33  53  25
* man      0.77       3         3        0    30  46  35  39  56  51
* much     0.70       3         4        1    20  39  33  35  51  31
* before   0.82       4         6        2    10  27  27  28  48  11
* most     0.74       4         4        0    18  35  32  36  56  28
* where    0.69       2         5        3    10  26  25  27  41  15
* many     0.72       4         4        0     8  23  26  28  50   3
* well     0.82       3         4        1    27  40  33  38  54  40
* even     0.99       3         4        1    21  38  31  37  53  28
* also     0.50       4         4        0     6  26  27  33  49   6
* being    0.73       4         5        1    16  35  25  28  44  21
* those    0.40       3         5        2    10  22  20  29  43  10
* people   1.00       4         6        2    12  33  26  34  52  11
* way      0.99       2         3        1    20  37  35  37  51  30
* back     0.78       3         4        1    22  40  32  35  52  32
* too      0.76       2         3        1    10  26  23  32  46  11
* good     0.70       3         4        1    26  44  34  38  55  42
* little   0.97       4         6        2    15  33  31  35  53  16
* down     0.78       3         4        1    21  39  32  35  55  28
* did      0.95       3         3        0    28  43  33  36  54  40
* how      0.69       2         3        1    24  42  31  38  52  36
* last     0.72       4         4        0    24  43  34  39  56  41
* between  0.87       6         7        1    10  27  23  26  49   7
* your     0.18       2         4        2    20  33  35  37  50  25
* just     0.79       4         4        0    18  37  32  33  54  26
* still    0.92       4         5        1    14  34  32  35  50  17
* work     0.69       3         4        1    15  31  31  36  53  14
* see      0.91       2         3        1    25  44  33  38  54  41
* know     0.80       2         4        2    10  20  19  30  49   8
* make     1.00       3         4        1    23  38  33  37  54  30
* might    0.99       3         5        2     8  26  28  33  48   2
* because  0.51       5         7        2     8  26  27  36  51  10
* through  0.59       3         7        4     6  15  18  21  46   0
* same     0.95       3         4        1    18  36  34  35  55  25
* own      0.79       2         3        1    14  30  30  33  51  13
* long     0.80       3         4        1    21  41  34  37  53  31
* get      0.71       3         3        0    30  46  35  39  56  51
* here     0.68       2         4        2    17  33  31  33  48  19
* go       0.43       2         2        0    31  45  35  39  56  51
* great    0.67       4         5        1     9  25  21  26  48   2
* three    0.86       3         5        2    29  45  31  33  50  41
* never    0.79       4         5        1    18  35  31  34  51  25
* life     0.98       3         4        1    20  34  32  33  52  19
* year     0.23       2         4        2    16  38  33  36  53  24
* come     0.71       3         4        1    23  38  34  36  54  31
* both     0.59       3         4        1    18  29  31  33  51  17
* old      0.71       3         3        0    24  44  34  38  56  41
* another  0.50       5         7        2    11  22  27  27  49   7
* without  0.76       5         7        2    19  38  33  36  54  26
* again    0.53       4         5        1     9  28  26  30  47   4
* us       0.24       2         2        0    26  43  35  37  54  42
* each     0.64       2         4        2    13  30  32  33  50   8
* used     0.31       4         4        0     9  24  25  26  46   4
* came     0.94       3         4        1    21  40  35  37  54  33
* under    0.76       4         5        1    15  37  34  35  54  25
* take     0.99       3         4        1    16  37  30  35  53  19
* say      0.91       2         3        1    19  37  30  33  51  22
  found    0.88       4         5        1    17  30  30  36  52  15
  though   0.27       2         6        4     6  11  11  18  33   0
  men      0.81       3         3        0    29  42  35  39  55  45
  thought  0.70       3         7        4     4  13  11  20  44   0
  day      0.98       2         3        1    29  45  35  38  55  47
  right    0.98       3         5        2    12  31  31  31  48  12
  went     0.85       4         4        0    24  39  35  35  52  38
  world    0.77       4         5        1    10  30  28  31  50  10
  while    0.92       3         5        2     7  19  19  28  43   1
  course   0.58       3         6        3     4  12  11  18  30   1

*120 words selected for analysis.


Ken Spencer*

Hull University, UK

*Correspondence should be addressed to Ken Spencer, Centre for Educational Studies, Hull University, Hull HU6 7RX, UK (e-mail: k.a.spencer@hull.ac.uk).
Table 1. Age and reading quotients for pupils, years 2-6

                         Y2        Y3       Y4        Y5        Y6

Mean age                   6.93     7.93      8.86      9.46     10.83
SD                         0.32     0.30      0.33      0.34      0.30
Mean reading quotient    100.52    97.87    101.77     95.51     98.20
  (Young)
SD (Young)                11.50    13.44     15.90     10.54     12.15
Mean reading quotient    106.84    99.57    103.86    101.18    102.45
  (NFER)
SD (NFER)                 13.51    15.77     15.42     16.83     16.14
Young-NFER correlations    0.68**   0.69**    0.77**    0.75**    0.68**
N                         31       46        35        39        56

**Correlations are significant at the 0.01 level (2-tailed).
ANOVA results: Young: [F.sub.4,202] = 1.33, p = .26; NFER:
[F.sub.4,202] = 1.13, p = .35.

Table 2. Word transparency correlations (N = 120)

                     LTP P-O    LTP P-O  MT P-O     MT P-O   LTP O-P
                     (Spencer)  (Hanna)  (Spencer)  (Hanna)  (Spencer)

LTP P-O (Spencer)     -
LTP P-O (Hanna)       0.85**     -
MT P-O (Spencer)      0.78**     0.63**   -
MT P-O (Hanna)        0.74**     0.84**   0.79**     -
LTP O-P (Spencer)     0.15      -0.02     0.17       0.01     -
LTP O-P (Berndt)      0.43**     0.41**   0.32**     0.27**   0.57**
MT O-P (Spencer)      0.08      -0.10     0.19*      0.01     0.84**
MT O-P (Berndt)       0.24**     0.18*    0.27**     0.21*    0.63**
Phonetic difference  -0.50**    -0.56**  -0.44**    -0.46**   0.20*
Phonemic length       0.01      -0.05     0.29**     0.20*   -0.02

                     LTP O-P   MT O-P     MT O-P    Phonetic    Phonetic
                     (Berndt)  (Spencer)  (Berndt)  difference  length

LTP P-O (Spencer)
LTP P-O (Hanna)
MT P-O (Spencer)
MT P-O (Hanna)
LTP O-P (Spencer)
LTP O-P (Berndt)      -
MT O-P (Spencer)      0.50**   -
MT O-P (Berndt)       0.78**   0.78**      -
Phonetic difference  -0.18*    0.02       -0.13      -
Phonemic length      -0.09     0.28**      0.18*    -0.08       -

*Correlations are significant at the 0.01 level (2-tailed).
**Correlations are significant at the 0.01 level (2-tailed).

Note.

LTP P-O (Spencer)  Least transparent phoneme  Phonology-to-orthography
LTP P-O (Hanna)    Least transparent phoneme  Phonology-to-orthography
MT P-O (Spencer)   Mean transparency          Phonology-to-orthography
MT P-O (Hanna)     Mean transparency          Phonology-to-orthography
LTP O-P (Spencer)  Least transparent phoneme  Orthography-to-phonology
LTP O-P (Berndt)   Least transparent phoneme  Orthography-to-phonology
MT O-P (Spencer)   Mean transparency          Orthography-to-phonology
MT O-P (Berndt)    Mean transparency          Orthography-to-phonology

LTP P-O (Spencer)  [spelling feedforward]  (Spencer, 1999)
LTP P-O (Hanna)    [spelling feedforward]  (Hanna et al., 1966)
MT P-O (Spencer)   [spelling feedforward]  (Spencer, 1999)
MT P-O (Hanna)     [spelling feedforward]  (Hanna et al., 1966)
LTP O-P (Spencer)  [spelling feedback]     (Spencer, 1999)
LTP O-P (Berndt)   [spelling feedback]     (Berndt et al., 1987)
MT O-P (Spencer)   [spelling feedback]     (Spencer, 1999)
MT O-P (Berndt)    [spelling feedback]     (Berndt et al., 1987)

Table 3. Word frequency correlations (N = 120)

                                    Grade    Grade    Grade    Grade
                                    3        4        5        6
                                    Carroll  Carroll  Carroll  Carroll

Carroll et al.  Grade 3    Print    -
  (1971)
Carroll et al.  Grade 4    Print    0.97**   -
  (1971)
Carroll et al.  Grade 5    Print    0.88**   0.90**   -
  (1971)
Carroll et al.  Grade 6    Print    0.93**   0.95**   0.94**   -
  (1971)
Carroll et al.  Grade 7    Print    0.85**   0.88**   0.89**   0.94**
  (1971)
Carroll et al.  Grade 8    Print    0.82**   0.85**   0.89**   0.92**
  (1971)
Carroll et al.  Grade 9    Print    0.75**   0.78**   0.84**   0.88**
  (1971)
Carroll et al.  Total      Print    0.95**   0.96**   0.95**   0.99**
  (1971)
Stuart et al.   Reception  Print    0.56**   0.50**   0.39**   0.42**
  (2003)
Stuart et al.   Year 1     Print    0.83**   0.78**   0.64**   0.69**
  (2003)
Stuart et al.   Year 2     Print    0.79**   0.74**   0.59**   0.64**
  (2003)
Stuart et al.   Year 3     Print    0.78**   0.74**   0.58**   0.64**
  (2003)
Hofland &       Adult      Print    0.68**   0.72**   0.75**   0.81**
  Johansson
  (1982)
Leech et al.    Adult      Print    0.70**   0.74**   0.79**   0.83**
  (2001)
Reid (1989)     7-year-    Written  0.72**   0.69**   0.57**   0.61**
                  old
Reid (1989)     8-year-    Written  0.74**   0.71**   0.60**   0.63**
                  old
Brown (1984)    Adult      Spoken   0.68**   0.70**   0.61**   0.66**
Leech et al.    Adult      Spoken   0.72**   0.74**   0.66**   0.68**
  (2001)

                Grade    Grade    Grade
                7        8        9        Total    Reception  Year 1
                Carroll  Carroll  Carroll  Carroll  Stuart     Stuart

Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.  -
  (1971)
Carroll et al.  0.88**   -
  (1971)
Carroll et al.  0.86**   0.88**   -
  (1971)
Carroll et al.  0.94**   0.92**   0.88**   -
  (1971)
Stuart et al.   0.39**   0.28**   0.29**   0.44**   -
  (2003)
Stuart et al.   0.63**   0.53**   0.50**   0.71**   0.75**     -
  (2003)
Stuart et al.   0.60**   0.49**   0.43**   0.67**   0.74**     0.93**
  (2003)
Stuart et al.   0.59**   0.48**   0.41**   0.66**   0.71**     0.91**
  (2003)
Hofland &       0.82**   0.81**   0.81**   0.81**   0.32**     0.52**
  Johansson
  (1982)
Leech et al.    0.84**   0.84**   0.83**   0.84**   0.30**     0.50**
  (2001)
Reid (1989)     0.57**   0.46**   0.40**   0.62**   0.67**     0.85**
Reid (1989)     0.58**   0.47**   0.41**   0.64**   0.67**     0.84**
Brown (1984)    0.62**   0.59**   0.58**   0.67**   0.40**     0.67**
Leech et al.    0.66**   0.60**   0.59**   0.70**   0.48**     0.73**
  (2001)

                                                 7-year-  8-year-
                Year 2  Year 3  Adult    Adult   old      old
                Stuart  Stuart  Hofland  Leech   Reid     Reid

Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Stuart et al.
  (2003)
Stuart et al.
  (2003)
Stuart et al.
  (2003)
Stuart et al.   0.96**  -
  (2003)
Hofland &       0.52**  0.54**  -
  Johansson
  (1982)
Leech et al.    0.50**  0.52**  0.97**   -
  (2001)
Reid (1989)     0.89**  0.88**  0.49**   0.48**  -
Reid (1989)     0.89**  0.89**  0.51**   0.50**  0.99**   -
Brown (1984)    0.66**  0.71**  0.62**   0.63**  0.64**   0.66**
Leech et al.    0.70**  0.74**  0.61**   0.64**  0.69**   0.72**
  (2001)

                Adult   Adult
                Brown   Leech

Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Carroll et al.
  (1971)
Stuart et al.
  (2003)
Stuart et al.
  (2003)
Stuart et al.
  (2003)
Stuart et al.
  (2003)
Hofland &
  Johansson
  (1982)
Leech et al.
  (2001)
Reid (1989)
Reid (1989)
Brown (1984)    -
Leech et al.    0.95**  -
  (2001)

*Correlations are significant at the 0.01 level (2-tailed).
**Correlations are significant at the 0.01 level (2-tailed).

Table 4. Word frequency and word transparency correlations

                     LTP P-O    LTP P-O  MT P-O     MT P-O   LTP O-P
                     (Spencer)  (Hanna)  (Spencer)  (Hanna)  (Spencer)

Carroll  Grade 3      0.01      0.03     -0.07      -0.04    -0.01
Carroll  Grade 4     -0.01      0.01     -0.08      -0.05    -0.01
Carroll  Grade 5      0.01      0.02     -0.09      -0.03    -0.05
Carroll  Grade 6      0.02      0.04     -0.05       0.00    -0.06
Carroll  Grade 7      0.05      0.08     -0.03       0.03    -0.06
Carroll  Grade 8      0.00      0.02     -0.08      -0.03    -0.05
Carroll  Grade 9      0.03      0.06     -0.08      -0.01    -0.09
Carroll  Total        0.03      0.05     -0.07      -0.01    -0.04
Stuart   Reception    0.01      0.11      0.03       0.12    -0.15
Stuart   Year 1       0.03      0.09     -0.02       0.01    -0.06
Stuart   Year 2       0.06      0.14      0.02       0.10    -0.08
Stuart   Year 3       0.04      0.09      0.03       0.08    -0.09
Hofland  Adult        0.07      0.10      0.03       0.10    -0.07
Leech    Adult        0.04      0.05     -0.04       0.02    -0.08
Reid     7-year-old   0.04      0.13      0.04       0.13    -0.10
Reid     8-year-old   0.02      0.11      0.01       0.10    -0.11
Brown    Adult        0.00      0.04     -0.08      -0.03    -0.06
Leech    Adult       -0.06      0.00     -0.13      -0.09    -0.07

         LTP O-P   MT O-P     MT O-P    Phonetic    Phonetic
         (Berndt)  (Spencer)  (Berndt)  difference  length

Carroll  -0.06     -0.15      -0.17     -0.09       -0.39**
Carroll  -0.12     -0.16      -0.22*    -0.07       -0.41**
Carroll  -0.14     -0.19*     -0.22*    -0.06       -0.34**
Carroll  -0.14     -0.19*     -0.24**   -0.09       -0.36**
Carroll  -0.13     -0.20*     -0.23*    -0.12       -0.36**
Carroll  -0.16     -0.18      -0.23*    -0.07       -0.34**
Carroll  -0.18*    -0.21*     -0.27**   -0.13       -0.35**
Carroll  -0.11     -0.18*     -0.21*    -0.10       -0.38**
Stuart    0.03     -0.20*     -0.07     -0.24**     -0.24**
Stuart    0.00     -0.18      -0.12     -0.18       -0.45**
Stuart    0.04     -0.17      -0.06     -0.24**     -0.38**
Stuart   -0.01     -0.16      -0.10     -0.21*      -0.35**
Hofland  -0.17     -0.13      -0.19*    -0.15       -0.30**
Leech    -0.20*    -0.17      -0.22*    -0.13       -0.33**
Reid     -0.07     -0.20*     -0.17     -0.21*      -0.30**
Reid     -0.07     -0.20*     -0.16     -0.19*      -0.30**
Brown    -0.09     -0.16      -0.18     -0.13       -0.37**
Leech    -0.11     -0.21*     -0.23*    -0.11       -0.42**

*Correlations are significant at the 0.01 level (2-tailed).
**Correlations are significant at the 0.01 level (2-tailed).
Anomolies due to rounding.

Table 5. Skew and Kurtosis for word bands

                    N    Range   Year 2 [AT]  Year 3 [AT]  Year 4 [AT]

Skewness (z score)  150   1-150   0.50        -1.29        -3.65***
                    140   1-140   0.79        -0.68        -2.18*
                    130   1-130   0.43        -0.87        -2.37*
                    120   1-120   0.20        -0.97        -2.30*
                    140  11-150   0.41        -1.38        -3.50***
                    130  11-140   0.71        -0.73        -1.99*
                    120  11-130   0.37        -0.89        -2.18*
                    130  21-150   0.45        -1.21        -3.22**
                    120  21-140   0.73        -0.55        -1.76
Kurtosis (z score)  150   1-150  -1.74        -0.84         0.91
                    140   1-140  -1.63        -1.12        -0.87
                    130   1-130  -1.58        -1.05        -0.63
                    120   1-120  -1.61        -0.96        -0.73
                    140  11-150  -1.67        -0.61         1.10
                    130  11-140  -1.52        -0.86        -0.60
                    120  11-130  -1.47        -0.79        -0.36
                    130  21-150  -1.66        -0.52         0.94
                    120  21-140  -1.49        -0.71        -0.60

                    Year 5 [AT]  Year 6 [AT]  QI [AT]  Total [AT]

Skewness (z score)  -1.66        -2.63**      -0.83    -1.92
                    -0.48        -1.38        -0.38    -0.77
                    -0.64        -1.63        -0.61    -1.00
                    -0.85        -1.95        -0.70    -1.14
                    -1.59        -2.60**      -1.05    -2.44*
                    -0.40        -1.34        -0.62    -1.34
                    -0.54        -1.56        -0.84    -1.55
                    -1.12        -2.38*       -0.94    -2.16*
                     0.05        -1.06        -0.52    -1.04
Kurtosis (z score)  -0.02         1.15        -1.48    -0.25
                    -1.08        -0.33        -1.50    -1.24
                    -1.03        -0.25        -1.32    -1.10
                    -0.95        -0.01        -1.31    -1.09
                     0.11         1.31        -1.54    -0.36
                    -0.89        -0.12        -1.56    -1.51
                    -0.84        -0.07        -1.36    -1.34
                    -0.03         1.45        -1.55    -0.35
                    -0.91         0.09        -1.55    -1.44

>< [+ or -]1.96, *p < .05; >< [+ or -]2.58, **p < .01; >< [+ or -]3.29,
***p < .001.

Table 6. Log word frequency and word transparency correlations with word
spelling difficulty, Years 2-6 (Arcsine transformation)

                     Year 2   Year 3   Year 4   Year 5   Year 6

Carroll Grade 3       0.45**   0.39**   0.39**   0.48**   0.37**
Carroll Grade 4       0.41**   0.37**   0.35**   0.42**   0.31**
Carroll Grade 5       0.31**   0.27**   0.26**   0.33**   0.25**
Carroll Grade 6       0.35**   0.30**   0.29**   0.36**   0.29**
Carroll Grade 7       0.36**   0.31**   0.29**   0.37**   0.30**
Carroll Grade 8       0.29**   0.24**   0.22*    0.30**   0.23*
Carroll Grade 9       0.31**   0.24**   0.22*    0.30**   0.25**
Carroll Total         0.38**   0.33**   0.31**   0.40**   0.32**
Stuart Reception      0.52**   0.48**   0.45**   0.43**   0.45**
Stuart Year 1         0.60**   0.51**   0.51**   0.56**   0.47**
Stuart Year 2         0.59**   0.52**   0.49**   0.58**   0.50**
Stuart Year 3         0.55**   0.48**   0.45**   0.55**   0.45**
Hofland Adult         0.32**   0.25**   0.19*    0.29**   0.25**
Leech Adult           0.31**   0.25**   0.19*    0.30**   0.24**
Reid 7-year-old       0.56**   0.48**   0.52**   0.57**   0.48**
Reid 8-year-old       0.54**   0.47**   0.51**   0.55**   0.46**
Brown Adult           0.41**   0.35**   0.26**   0.43**   0.35**
Leech Adult           0.43**   0.36**   0.30**   0.44**   0.31**
LTP P-O (Spencer)     0.51**   0.55**   0.48**   0.37**   0.48**
LTP P-O (Hanna)       0.52**   0.54**   0.47**   0.39**   0.48**
MT P-O (Spencer)      0.36**   0.45**   0.30**   0.25**   0.39**
MT P-O (Hanna)        0.39**   0.43**   0.32**   0.25**   0.38**
LTP O-P (Spencer)     0.02     0.08     0.06     0.06     0.01
LTP O-P (Berndt)      0.30**   0.32**   0.28**   0.27**   0.22*
MT O-P (Spencer)     -0.04     0.00     0.02     0.02     0.05
MT O-P (Berndt)       0.14     0.16     0.13     0.13     0.15
Phonetic Difference  -0.58**  -0.57**  -0.51**  -0.47**  -0.64**
Phonemic Length      -0.32**  -0.27**  -0.24**  -0.22*   -0.04

                     Total    Quartile 1

Carroll Grade 3       0.45**   0.45**
Carroll Grade 4       0.41**   0.41**
Carroll Grade 5       0.31**   0.32**
Carroll Grade 6       0.35**   0.36**
Carroll Grade 7       0.36**   0.38**
Carroll Grade 8       0.28**   0.30**
Carroll Grade 9       0.29**   0.31**
Carroll Total         0.38**   0.39**
Stuart Reception      0.52**   0.55**
Stuart Year 1         0.58**   0.60**
Stuart Year 2         0.59**   0.60**
Stuart Year 3         0.54**   0.56**
Hofland Adult         0.29**   0.35**
Leech Adult           0.29**   0.33**
Reid 7-year-old       0.56**   0.57**
Reid 8-year-old       0.55**   0.56**
Brown Adult           0.39**   0.44**
Leech Adult           0.41**   0.45**
LTP P-O (Spencer)     0.53**   0.54**
LTP P-O (Hanna)       0.52**   0.55**
MT P-O (Spencer)      0.39**   0.40**
MT P-O (Hanna)        0.38**   0.42**
LTP O-P (Spencer)     0.04     0.03
LTP O-P (Berndt)      0.31**   0.30**
MT O-P (Spencer)      0.00    -0.04
MT O-P (Berndt)       0.15     0.14
Phonetic Difference  -0.62**  -0.60**
Phonemic Length      -0.26**  -0.31**

*Correlations are significant at the 0.01 level (2-tailed).
**Correlations are significant at the 0.01 level (2-tailed).

Table 7. Step-wise multiple regression: dependent variable--world
spelling difficulty, years 2-6, total school and quartile 1

Year 2 [F.sub.6, 113] = 36.74, p < 0.001

Total variance   0.66
Shared variance  0.27
Unique variance  0.39

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.51  0.15   0.00    3.32***  0.00              0.00
Freq S2     0.21  0.03   0.37    5.98***  0.30              0.11
LTP P-O     0.32  0.06   0.32    4.93***  0.23              0.07
PhD = 3/4  -0.45  0.08  -0.34   -5.52***  0.05              0.09
PhD = 2    -0.21  0.05  -0.32   -4.27***  0.03              0.05
PhL        -0.06  0.02  -0.21   -3.41***  0.03              0.03
PhD = 1    -0.11  0.04  -0.21   -2.92**   0.03              0.03

Year 3 [F.sub.4, 115] = 42.52, p < 0.001

Total variance   0.65
Shared variance  0.23
Unique variance  0.42

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.64  0.12   0.00    5.24***  0.00              0.00
LTP P-O     0.38  0.05   0.44    7.47***  0.34              0.17
Freq S2     0.17  0.03   0.36    5.77***  0.20              0.10
PhD = 3/4  -0.36  0.06  -0.32   -5.58***  0.07              0.10
PhD = 2    -0.11  0.03  -0.20   -3.36***  0.03              0.03
PhL        -0.04  0.02  -0.14   -2.23*    0.02              0.02

Year 4 [F.sub.4, 115] = 31.01, p < 0.001

Total variance   0.52
Shared variance  0.13
Unique variance  0.39

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.73  0.09   0.00    8.36***  0.00              0.00
LTP P-O     0.26  0.05   0.33    4.76***  0.22              0.09
Freq SI     0.17  0.03   0.40    6.06***  0.17              0.15
PhD = 3/4  -0.32  0.07  -0.32   -4.80***  0.08              0.10
PhD = 2    -0.12  0.04  -0.23   -3.28***  0.05              0.05

Year 5 [F.sub.3, 116] = 42.27, p < 0.001

Total variance   0.52
Shared variance  0.06
Unique variance  0.46

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.62  0.07   0.00    8.41***  0.00              0.00
Freq S2     0.19  0.02   0.53    8.17***  0.31              0.28

LTP P-O     0.22  0.04   0.33    5.08***  0.14              0.11
PhD = 3/4  -0.24  0.06  -0.28   -4.32***  0.08              0.08

Year 6  [F.sub.5, 114] = 32.11, p < 0.001

Total variance   0.58
Shared variance  0.14
Unique variance  0.44

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    1.09  0.07   0.00   16.49***  0.00              0.00
LTP P-O     0.13  0.04   0.24    3.39***  0.24              0.04
Freq S2     0.10  0.02   0.32    5.11***  0.15              0.10
PhD = 3/4  -0.33  0.05  -0.47   -6.94***  0.11              0.18
PhD = 2    -0.14  0.03  -0.40   -4.86***  0.04              0.09
PhD = 1    -0.08  0.02  -0.28   -3.54***  0.05              0.05

Total [F.sub.4, 115] = 65.40, p < 0.001

Total variance   0.71
Shared variance  0.27
Unique variance  0.45

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.80  0.09   0.00    8.85***  0.00              0.00
LTP P-O     0.21  0.04   0.33    5.50***  0.31              0.08
Freq S2     0.14  0.02   0.40    6.96***  0.25              0.12
PhD = 3/4  -0.35  0.05  -0.42   -7.31***  0.09              0.14
PhD = 2    -0.15  0.03  -0.34   -4.87***  0.03              0.06
PhD = 1    -0.07  0.02  -0.21   -3.01**   0.01              0.02
LTP O-P     0.06  0.03   0.12    2.29*    0.02              0.01
PhL        -0.02  0.01  -0.12   -2.06*    0.01              0.01

Quartile 1 [F.sub.6, 113] = 47.84, p < 0.001

Total variance   0.72
Shared variance  0.31
Unique variance  0.41

           B      SE B  [beta]  t         [DELTA][R.sup.2]  [sr.sup.2]

Constant    0.15  0.18   0.00    0.85     0.00              0.00
Freq S2     0.28  0.04   0.40    7.08***  0.32              0.13
LTP P-O     0.47  0.07   0.37    6.35***  0.28              0.10
PhD = 3/4  -0.57  0.09  -0.35   -6.14***  0.06              0.09
PhD = 2    -0.25  0.06  -0.30   -4.33***  0.03              0.05
PhL        -0.07  0.02  -0.18   -3.16**   0.02              0.02
PhD = 1    -0.12  0.04  -0.18   -2.65**   0.02              0.02

***p < 0.001 **p < 0.01 *p < 0.05
Note: [DELTA][R.sup.2] Change in [R.sup.2] at each step in the
regression analysis, variables in order in which they enter the step-
wise regression B, SE B, [beta], t and [sr.sup.2] Values for variables
in the final step-wise model ANOVA results are for the final step-wise
model

Table 8. Enter multiple regression models for three frequency sources
and two transparency sources: dependent variable--word spelling
difficulty, years 2-6

                                   LTP P-O (Spencer)
                                   LTP O-P (Spencer)
                                   Frequency (Stuart S2/3)

Year 2  [R.sup.2]                            0.66
        Shared variance                      0.27
        Total individual variance            0.40
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.23**         0.03
        PhD = 2                    -0.34***        0.06
        PhD = 3/4                  -0.35***        0.09
        LTP P-O                     0.31***        0.07
        LTP O-P                     0.06           0.00
        PhL                        -0.21***        0.03
        Frequency                   0.38***        0.11
Year 3  [R.sup.2]                            0.66
        Shared variance                      0.26
        Total individual variance            0.40
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.13           0.01
        PhD = 2                    -0.30***        0.05
        PhD = 3/4                  -0.37***        0.11
        LTP P-O                     0.40***        0.12
        LTP O-P                     0.11           0.01
        PhL                        -0.15*          0.02
        Frequency                   0.33***        0.09
Year 4  [R.sup.2]                            0.53
        Shared variance                      0.19
        Total individual variance            0.34
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.13           0.01
        PhD = 2                    -0.31***        0.05
        PhD = 3/4                  -0.38***        0.11
        LTP P-O                     0.29***        0.06
        LTP O-P                     0.14*          0.02
        PhL                        -0.14           0.02
        Frequency                   0.30***        0.07
Year 5  [R.sup.2]                            0.56
        Shared variance                      0.16
        Total individual variance            0.40
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.19*          0.02
        PhD = 2                    -0.23**         0.03
        PhD = 3/4                  -0.38***        0.11
        LTP P-O                     0.24***        0.04
        LTP O-P                     0.15*          0.02
        PhL                        -0.07           0.00
        Frequency                   0.47***        0.18
Year 6  [R.sup.2]                            0.59
        Shared variance                      0.11
        Total individual variance            0.47
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.32***        0.06
        PhD = 2                    -0.44***        0.10
        PhD = 3/4                  -0.48***        0.18
        LTP P-O                     0.23**         0.04
        LTP O-P                     0.12           0.01
        PhL                         0.06           0.00
        Frequency                   0.33***        0.09

                                   LTP P-O (Spencer)
                                   LTP O-P (Spencer)
                                   Frequency (Leech adult)

Year 2  [R.sup.2]                            0.57
        Shared variance                      0.14
        Total individual variance            0.43
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.29***        0.04
        PhD = 2                    -0.43***        0.10
        PhD = 3/4                  -0.40***        0.13
        LTP P-O                     0.28***        0.06
        LTP O-P                     0.05           0.00
        PhL                        -0.32***        0.09
        Frequency                   0.13           0.01
Year 3  [R.sup.2]                            0.57
        Shared variance                      0.18
        Total individual variance            0.40
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.18*          0.02
        PhD = 2                    -0.37***        0.07
        PhD = 3/4                  -0.41***        0.13
        LTP P-O                     0.37***        0.10
        LTP O-P                     0.08           0.01
        PhL                        -0.27***        0.06
        Frequency                   0.06           0.00
Year 4  [R.sup.2]                            0.46
        Shared variance                      0.11
        Total individual variance            0.35
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.17           0.02
        PhD = 2                    -0.38***        0.08
        PhD = 3/4                  -0.42***        0.14
        LTP P-O                     0.27***        0.05
        LTP O-P                     0.12           0.01
        PhL                        -0.25***        0.06
        Frequency                   0.03           0.00
Year 5  [R.sup.2]                            0.42
        Shared variance                      0.08
        Total individual variance            0.34
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.23*          0.03
        PhD = 2                    -0.32**         0.05
        PhD = 3/4                  -0.43***        0.14
        LTP P-O                     0.20*          0.03
        LTP O-P                     0.12           0.01
        PhL                        -0.19*          0.03
        Frequency                   0.21**         0.04
Year 6  [R.sup.2]                            0.51
        Shared variance                      0.05
        Total individual variance            0.46
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.36***        0.07
        PhD = 2                    -0.51***        0.14
        PhD = 3/4                  -0.52***        0.21
        LTP P-O                     0.20*          0.03
        LTP O-P                     0.09           0.01
        PhL                        -0.04           0.00
        Frequency                   0.10           0.01

                                   LTP P-O (Hanna)
                                   LTP O-P (Berndt)
                                   Frequency (Carroll G3/G5)

Year 2  [R.sup.2]                            0.61
        Shared variance                      0.25
        Total individual variance            0.35
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.25**         0.03
        PhD = 2                    -0.38***        0.08
        PhD = 3/4                  -0.34***        0.09
        LTP P-O                     0.28***        0.05
        LTP O-P                     0.09           0.01
        PhL                        -0.22**         0.04
        Frequency                   0.29***        0.06
Year 3  [R.sup.2]                            0.58
        Shared variance                      0.26
        Total individual variance            0.32
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.15           0.01
        PhD = 2                    -0.35***        0.06
        PhD = 3/4                  -0.36***        0.10
        LTP P-O                     0.30***        0.06
        LTP O-P                     0.12           0.01
        PhL                        -0.15*          0.02
        Frequency                   0.28***        0.06
Year 4  [R.sup.2]                            0.48
        Shared variance                      0.21
        Total individual variance            0.27
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.12           0.01
        PhD = 2                    -0.33***        0.06
        PhD = 3/4                  -0.36***        0.09
        LTP P-O                     0.23**         0.03
        LTP O-P                     0.11           0.01
        PhL                        -0.13           0.01
        Frequency                   0.27***        0.06
Year 5  [R.sup.2]                            0.49
        Shared variance                      0.19
        Total individual variance            0.31
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.18           0.02
        PhD = 2                    -0.25**         0.03
        PhD = 3/4                  -0.33***        0.08
        LTP P-O                     0.23**         0.03
        LTP O-P                     0.17*          0.02
        PhL                        -0.06           0.00
        Frequency                   0.39***        0.12
Year 6  [R.sup.2]                            0.50
        Shared variance                      0.10
        Total individual variance            0.40
                                   [beta]          [sr.sup.2]
        PhD = 1                    -0.33***        0.06
        PhD = 2                    -0.49***        0.12
        PhD = 3/4                  -0.49***        0.18
        LTP P-O                     0.17*          0.02
        LTP O-P                     0.04           0.00
        PhL                         0.00           0.00
        Frequency                   0.17*          0.02

*p < .05; **p < .01; ***p < .001.
COPYRIGHT 2007 British Psychological Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Spencer, Ken
Publication:British Journal of Psychology
Geographic Code:4EUUK
Date:May 1, 2007
Words:16591
Previous Article:Characterizing linguistic structure with mutual information.
Next Article:Kinship and altruism: a cross-cultural experimental study.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters