Annals of Dyslexia

, Volume 58, Issue 2, pp 97–114

What type of computer-assisted exercise supports young less skilled spellers in resolving problems in open and closed syllable words?

Authors

    • PI Research – VU Amsterdam
  • Pieter Reitsma
    • PI Research – VU Amsterdam
Article

DOI: 10.1007/s11881-008-0019-0

Cite this article as:
Hilte, M. & Reitsma, P. Ann. of Dyslexia (2008) 58: 97. doi:10.1007/s11881-008-0019-0

Abstract

Dutch bisyllabic words containing open and closed syllables are particularly difficult to spell for children. What kind of support in spelling exercises improves the spelling of these words the most? Two extensions of a commonly used dictation exercise were tested: less skilled spellers in grade 2 (n = 50; 7 years and 10 months) either received explicit syllabic segmentation cues or received spelling cues by means of a visual preview. Comparisons between pre-, post-, and retention tests of spelling skill showed that extra syllabic cues did not show a significant improvement beyond normal spelling dictation and that visual preview was most effective as compared to the other types of training. The findings suggest that word-specific knowledge can effectively be improved by exposure to the correct letter pattern during exercises in spelling and seems to result in lasting improvement of word-specific orthographic representations, at least for 5 weeks.

Keywords

ChildrenOrthographyReadingSpellingSyllablesTraining

Introduction

In alphabetic writing, both visual-orthographic and phonological information of a word are fundamental to correctly spell a word. Visual-orthographic information includes information about word-specific spelling, and the typical letter sequences or visual spelling patterns that are encountered in a particular orthography. Phonological information is used to segment words into syllables and phonemes and to map these phonemes to the corresponding graphemes. Studies of the process of learning to spell often show an age-related shift from reliance on phonological information to visual-orthographic information (Bear & Templeton, 1998; Ehri, 1992; Frith, 1985; Henderson, 1992; Notenboom & Reitsma, 2003; Schlagal, 2001). Another frequent finding is that poor spellers experience phonological difficulties in learning to spell. Actually, a dominant view in reading and spelling research is that reading and spelling problems have their origin in phonological language processing skills (Frith, 1997; Snowling, 2000; Stanovich & Siegel, 1994). Less skilled spellers frequently experience difficulties in phonological tasks involving the identification and manipulation of speech sounds and memorizing speech. These children experience a so-called phonological core deficit. Either their mental phonological representation of words is underspecified or this representation is relatively less accessible. Poor phonological skills make it hard to learn the links between phonemes and letters and to use these links to spell unfamiliar words (Frith, 1997; Snowling, 2000; Stanovich & Siegel, 1994; Vellutino et al., 1996). Practice in phonological skills may be an obvious and beneficial action in such cases. In addition, one could hypothesize that providing extra phonological information during the actual process of spelling might be helpful for learning to spell.

In a recent comparison between the effects of providing phonological cues during computer-based exercises in spelling and the effects of allowing a preview of the specific visual-orthographic pattern of the word to be spelled, a visual preview appeared to be more effective than phonological cues during practice for less skilled spellers in grade 3 (Hilte & Reitsma, 2006). It should be emphasized that children actually had to spell the words in both conditions, either after the presentation of the phonological cues or after they had previewed the specific letter pattern. From the perspective of both the phonological deficit hypothesis and developmental accounts, this finding is rather unexpected. Maybe, younger and less skilled spellers, as compared to the spellers in the earlier study, do profit relatively more from phonological support in spelling exercises. The current study, therefore, focuses on spellers of grade 2 and the question is whether providing phonological help during practice in spelling is more beneficial than allowing a visual preview or, alternatively, whether a visual preview is most effective for learning to spell.

In order to evaluate progress during the development or to assess individual differences in spelling skills, ordinarily, a dictation test is used in primary education. The teacher dictates the word, and children have to spell subsequently the word from memory. In the current study also, a dictation test is used to determine the spelling skills before and after the experimental exercises in spelling. One rather basic format of training in spelling then is to repeatedly dictate a word auditorily and providing feedback concerning the spelling produced in order to allow children to learn from the result. Because the format of the type of exercise is quite comparable with the format of the pre- and posttests (cf., Searleman & Herrmann, 1994), it can be expected that the children may profit considerably from such training. However, given the phonological deficits of poor spellers, it seems reasonable to predict that extra phonological help during the exercises may be beneficial and would result in even higher learning gains. The phonological cues provided during practice should of course be adapted to the specific problem in spelling at hand.

One difficulty in spelling is that there often are inconsistent phoneme–grapheme relations. For example, the words ‘beer’ and ‘hear’ have a similar phoneme in the middle, but different graphemes for this phoneme. In these words, you have to know the exact orthographic structure to spell the words correctly after dictation. In an opaque language, like English, there are relatively many inconsistent phoneme–grapheme relations. In a more transparent language, such as Dutch, phoneme–grapheme relations are generally one-to-one, which means that most words can be spelled by a simple strategy of phoneme–grapheme translation. Nevertheless, in Dutch too, there are words that are not easy to spell.

Problems arise in some multi-syllabic words, for example, in the orthographic representation of phoneme durations where a set of context-sensitive phonological rules needs to be applied (cf. for other orthographies: Juul & Sigurdsson, 2005; Lehtonen & Bryant, 2005). In Dutch singular word forms, long vowel phonemes are represented by geminated, sometimes referred to as doubled, vowels (eg., noot [nut]), whereas short vowel phonemes are spelled with a single letter (eg., kat [cat]).In the plural form, however, a long vowel phoneme, which occurs at syllable boundary, in open syllables is spelled with a single letter (eg., noten [nuts]). In plural words with a closed syllable, double consonants mark preceding short vowel phonemes (e.g., katten [cats]). In other types of Dutch words that do not have long or short vowel phonemes in singular form, for example in words that contain digraphs, one can most often add -en to the singular word form for plural spelling [e.g., kuil (hole) for singular and kuilen (holes) for plural]. In analogy to these types of words, second graders tend to spell plural forms of words with long or short vowel phonemes consistently with adding the plural suffix -en to singular forms (e.g., incorrect: nooten, katen, correct: noten, katten). Even fourth graders sometimes experience problems with these variations in vowel elimination in open syllables and consonant doublings in closed syllables in plural forms (Landerl & Reitsma, 2005; Notenboom & Reitsma, 2007). Knowledge of the spelling of the singular word form may interfere with the spelling of the plural form, for example in the word noten: The singular form is noot, which has a long vowel phoneme spelling with a geminated vowel, but the plural spelling is noten with a single letter. The question then is how phonological cues can be provided during exercises in spelling when this difficulty in spelling is addressed. One plausible way would be to emphasize the short and long vowel phonemes by explicitly segmenting the word into syllables. When words are segmented in syllables, the singular word form is less easy to recognize, and attention can be focused to the phonological level and may, consequently, not interfere with applying the conditional phonological rules.

In addition to plural words with open syllables ending with the suffix -en, a group of singular words exist in the Dutch language that also have open syllables. These words do not have singular word forms that could interfere in spelling, for example, the word boter [butter], because the word boter is already a singular word form itself. The spelling of these types of words remains, however, relatively difficult because these open-syllable singular words do not occur very frequently and also because children tend to spell these words phonologically (with two vowels instead of one, e.g., booter). These types of words are also included in the experiment, as children may benefit from segmentation cues in these words too because attention can be focused to the phonological level.

In the current study, the effect of syllabic segmentation as a phonological cue during spelling exercises will be compared to the effects of dictation training. Bisyllabic words with a single intervocalic consonant and a long vowel in the first syllable are segmented right after the long vowel, e.g., mu-ren [walls]. Words with closed syllables are segmented either ambisyllabic (e.g., wek-ker [clock]) or before the intervocalic consonant (e.g., we-kker). It has been suggested that linguistically ambisyllabic segmentation is most appropriate (Booij, 1995, p. 32). However, age or knowledge about the spelling may be an important factor to consider. It has been found that 8-year-old children intuitively segment bisyllabic words with a short vowel in the first syllable (closed syllable) relatively frequently ambisyllabic, e.g., the segmentation of the English word rabbit as /raeb/ + /bit/ (57.0% and 25.6% depending whether the first syllable is stressed or not), whereas 5-year-old children produce ambisyllabic responses less frequently (29.1% and 12.6%; Gillis & De Schutter, 1996). Externally providing an ambisyllabic segmentation during practice in spelling may help the speller to realize the consonant doublet. Segmenting bisyllabic words is not a very difficult or demanding task, but the explicit pronunciation of the separate syllables will emphasize the length of the vowel. This emphasis may help children to detect the regularity that is based on the length of the vowel and the number of consonants. Moreover, segmenting the words will distract attention from the singular word form which may possibly interfere with applying the phonological rule. More generally, it is thus expected that syllabic segmentation cues during spelling exercises will be more effective compared to a normal dictation, especially for words with open and closed syllables, because it emphasizes the quality of the vowel.

This benefit from syllabic segmentation cues can be expected to be limited to words with open and closed syllables. In words with inconsistent phoneme–grapheme mappings, children need word-specific knowledge to know the correct spelling, and phonological cues can be expected to be not very supportive in these types of words. In order to test whether the expected beneficial effect of syllabic segmentation cues will only apply in words with open and closed syllables, other types of words, in which word-specific knowledge is needed, will also be included in the current experiment.

The conditions of normal dictation and practice with additional segmented speech are contrasted with a visual preview condition. In this condition, children are first asked to observe the spelling of a word carefully. Subsequently, the word disappears and the child has to spell the dictated word from memory. One reason to include this format of practice to the current experimental training study is because some previous studies indicate that poor spellers may benefit from visual-orthographic training. In a first study on the effect of visual preview (visual dictation) in grade 1, examination of a visual spelling pattern was the most effective instruction method for both skilled and less skilled spellers. In this study, visual preview was compared to four other instruction methods (Van Leerdam, Bosman, & Van Orden, 1998). However, none of these instruction methods involved a normal dictation in which a word is dictated auditorily. More recently, the effect of visual preview as a training was compared to a spelling dictation training in which extra phonological cues (spelling pronunciations) were provided (Hilte & Reitsma, 2006). The results have shown a significant difference between skilled and less skilled spellers in grade 3. Less skilled spellers appeared to profit most from visual preview, whereas skilled spellers benefited as much from visual preview as from spelling pronunciations. Thus, these studies indicate that a visual approach in practice may be beneficial for poor spellers in grade 2.

The hypothesis that visual preview may be beneficial is less obvious than one might expect. As a matter of fact, for practice with visual preview, the act of reading is explicitly required. Several studies have shown that reading a word correctly does not guarantee that one can also spell the word afterwards (Ehri, 2000; Fletcher-Flinn, Shankweiler, & Frost, 2004; Gough, Juel, & Griffith, 1992; Katz & Frost, 2001). Although in general spelling skills are highly correlated with reading skills, mostly varying from 0.50 and 0.80 (Frith, 1980), it has long been debated whether or not reading and spelling are two separate abilities (Ehri, 1997; Ehri, 2000; Frith, 1980). Spelling is more difficult because it requires the production rather than the recognition of a visual spelling pattern, and grapheme–phoneme relations are generally more consistent (84.5% for Dutch and 69.3% for English) than phoneme–grapheme relations needed in spelling (36.9% for Dutch and 27.7% for English; Bosman, Vonk, & Van Zwam, 2006). Thus, it is not very surprising that more words are read correctly but not spelled than vice versa [e.g., 37% read but not spelled and 6% spelled but not read (Shankweiler & Lundquist, 1992) and 21% versus 13% (Bryant & Bradley, 1980)].

The relevant question is, however, whether reading transfers to spelling. Ehri (2000) has taught beginners to read specific words and has asked them afterwards to spell these words. Although some transfer from reading to spelling was found, words were not spelled perfectly correct. A greater proportion of letters was spelled correctly, 70% to 80%, as compared to the entire words, 30% to 40%. If one or two letters are forgotten, reading can still be accurate, but perfect spelling will not be possible (Ehri, 2000). In another study, Katz and Frost (2001) have shown that subjects are likely to accept the orthographic representation when it is presented two times. However, spellings that altered the phonology were rarely accepted. In particular, pre-experimental internal orthographic representations are less stable when there are many phonologically acceptable alternatives. Words with open and closed syllables are spelled in contradiction with the phonology, and they can be spelled in various phonologically acceptable alternative ways. For example, the word boter (butter) can also be spelled as booter, bootter, bootur, or boottur, which are spellings that are consistent with the phonology, which is actually not the case for the correct spelling of boter. Thus, it could be argued that children will not readily accept the spelling of words with open and closed syllables when it is presented visually because the spelling alters the phonology. In conclusion, providing a visual preview of the correct spelling does not necessarily transfer directly to the actual production of the correct spelling. Therefore, the relative advantage of phonological cues as compared to visual cues is an issue that should be resolved empirically.

In summary, in the current experiment with less skilled spellers of grade 2, we aim to study the effects of a spelling dictation training with or without an extra syllabic segmentation cue as compared to a visual preview training. Syllabic segmentation cues added to a spelling dictation training are expected to be more effective than a spelling dictation training without extra cues, especially in words containing open and closed syllables. It is hypothesized, however, that less skilled spellers of grade 2 will profit most from the preview approach like in previous studies (Hilte & Reitsma, 2006; Van Leerdam, Bosman, & Van Orden, 1998). Visual preview is expected to be most effective as a training compared to a normal dictation training with or without extra syllabic segmentation cues.

Method

Participants

A total of 50 children, 29 boys and 21 girls, participated in the training study. These children were selected as being less skilled spellers among a total of 347 Dutch children, 174 boys and 173 girls, from grade 2 of 12 primary schools. All schools were situated in the surroundings of the Dutch capital, Amsterdam. Most of the participants were native Dutch children and were of middle socioeconomic status. At the moment the present experiment started (January), the children had received 14 months of formal instruction. Prior to the experiment, spelling skills were determined by means of a Dutch standardized spelling test (Geelhoed & Reitsma, 1999). The third and fourth blocks of the test (30 words in total) were dictated and children were selected who spelled less than 16 words correctly. The remaining children spelled, on average, 24 words correctly. The norms of the test indicate that after 14 months of instruction, only 25% of the children do not reach a score of 16. Thus, the current selected group of children can be seen as poor or less skilled spellers in comparison to their age or grade level. We did not use further diagnostic measures, but it is possible that among the less skilled spellers, there were a few children with dyslexia because the scores of the children were below the 25th percentile of the spelling test. The mean chronological age of the poor spellers was 7 years and 10 months (SD = 5 months).

Materials

A list of 32 bisyllabic Dutch words was selected. Almost all the words used are normally taught in grade 2 or beyond. The 32 words consisted of eight words with open syllables, CVCVC words, 12 words with closed syllables, CVCCVC words, and 12 words with various other spelling difficulties: Five out of the other types of words have ambiguities in phoneme–grapheme correspondences (e.g., the different spellings ij and ei or au and ou both refer to a similar sound). The other words comprise letter patterns that are an exception to Dutch spelling rules. For example, diphthongs (ei, eu, and ui) are always followed by a single consonant. Another exception is that short vowels cannot be followed by geminated v or z, except in loan words (like puzzel). Thus, a closed syllable with a double v or z does not exist in Dutch words, and a single v or z always suffices. The f/v and the s/z are often exchanged and are difficult in Dutch spelling. Thus, these other types of words comprised exceptions to the rules or other ambiguities in spelling and were selected to test whether the syllabic segmentation condition is only effective in open and closed syllabic words. The 20 words containing open (eight words) and closed syllables (12 words) will be analyzed in the results to study the various practice effects in open and closed syllables. Because we expected that 36 words (12 for each type of words) would be too much words for a dictation test for poor spellers of grade 2 and because open syllable words are easier to spell than closed syllable words (Meulenkamp, 2000), we decided to use only eight open-syllable words instead of 12 (because of the design, it had to be equally divisible into four conditions).

A total of eight words were practiced by presenting the visual spelling pattern before the word had to be spelled, eight words were practiced by presenting syllabic segmentation cues, and eight words were practiced on the basis of a normal auditory dictation. The remaining eight control words were not practiced at all. Assigning specific words to conditions was varied between participants so that when data were collapsed over subjects, all 32 words appeared in each condition with almost equal frequency. In the segmentation condition, half of the subjects practiced with an ambisyllabic segmentation of the closed syllables (wek-ker [clock]), and the remaining subjects practiced with the intervocalic consonant only tied to the second segment (we-kker).

In Appendix 1, the words are shown in the first column with their English translations in the second column. In the third column, the words are categorized into the three word groups (words with open and closed syllables and other words), and in the fourth column, written youth frequencies of the words can be found (Staphorsius, Krom, & De Geus, 1988). Some of the words were relatively more frequent in youth literature than others, e.g., vader (father) is more frequent than roeiboot (rowing boat), but words were matched on frequencies across the conditions. Effects of word frequencies on differential training effects were taken into account in further analyses. A picture of the word was presented during all practice conditions in order to minimize any unlikely confusion about the meaning of the words. Because the pictures appeared in all conditions and because one picture was used for each word and remained the same in each condition, the specific combination of a picture with a printed word could not confound any effect of practice condition. In pre- and posttests, the words were dictated in meaningful sentences, but only the target words had to be produced.

Design and procedure

A within-subjects design was used: all 50 children who participated in this study received all three practice conditions. But also, all words are practiced in every condition (spelling exercise), though varied over subjects. In each session with the computer program, the words were presented in a different order.

Subjects practiced with a special purpose multimedia program that was installed on the computers of all participating schools. Word stimuli and feedback were presented visually in the middle of the screen. An instruction text could be read in order to make sure that it was clear what the subjects had to do, but this text might have been redundant because the experimenter already explained to the children how to use the computer program before they started. Alternatively, a button on the right could be clicked to hear the (more extensive) instruction as well. Another button, on the left side of the screen, could be clicked to repeat the normal or the syllabic segmented pronunciation of the word, if necessary an infinite number of times. A large button on the right side of the middle of the screen could be clicked after typing to receive the feedback in which the correct spelling of the word is shown. The button could be clicked again, after feedback is studied thoroughly, to proceed to the next trial during the exercises. The number of items that the children had to complete before they would finish could be found on the top corner.

Prior to the training sessions, a pretest was administered to assess the spelling proficiency for the selected words of the participants. The pretest was a dictation test of all 32 words and the children had to write down the words with paper and pencil. The words were dictated within a sentence first and subsequently repeated separately. The children who were selected from the classes were tested in a group in a separate classroom. After completion of the dictation test, an explanation was provided to the children about how to practice with the multimedia training program.

During the following 4 weeks, the children had to practice independently six sessions with the training program. Children practiced about twice a week and never more than once a day. It took children about 15 min to finish one training session. Within each training session, 24 words had to be practiced. In each session, the 24 words were practiced only once. Because three types of spelling exercises (practice conditions) were examined, eight words were practiced in every condition. Among these eight words, three words were closed syllable words, two words were open syllable words, and the three remaining words were other types of words; thus, in each session, nine closed, six open syllable words, and nine other types of words were practiced. The remaining eight words were not practiced and functioned as control words. See Appendix 2 for an overview on the number of words practiced in each condition and the number of each type of word per condition. The order in which the 24 words were presented in the computer program was randomly mixed; thus, the words were not taught in blocks of eight, with each condition separated from the others. Also, the different types of words were randomly mixed in the computer sessions because otherwise, children might lose their attention to the specific phonological problems in spelling the open and closed syllables. The following three spelling exercises were evaluated:
  1. 1

    Spelling dictation condition: The sound of the word to be spelled is produced by the program, similar to a regular spelling dictation test. The pronunciation of the word can be repeated upon request if necessary by pressing a button on the left. After the participant decides she or he is ready to spell the word, the continuation button is pressed. The child then has to type the word by heart in a text box in the middle of the screen. After pressing the continuation button once again, the program gives feedback: the program shows the child’s spelling in green with a curl (for correct) or in red (for incorrect) with a cross. In both cases, the correct spelling is presented in a green color just above the typed word.

     
  2. 2

    Syllabic segmentation condition: The general format and procedure of this exercise is comparable to the normal dictation condition. The only difference is that in this exercise, the segmented syllables of words are pronounced explicitly after the normal spelling dictation of the word. Syllables are produced with a rate of about one syllable per 0.8 s.

     
  3. 3

    Visual preview: The spelling of a word is presented in the middle of the screen. Thus, the children could examine the visual spelling pattern of a word. No phonological information is provided. After the child thinks to know how to spell the word, the continuation button is pressed and the word disappears. The child then has to type the word by heart in a text box in the middle of the screen. After pressing the continuation button once again, the program gives feedback similar to the two phonological practice conditions.

     

A few days after the children had completed six training sessions, a posttest was administered. The posttest consisted of a similar spelling test as was used in the pretest. Five weeks after the posttest, the same test was administered again as a test for retention. For each word in pretest, posttest, and retention test, it was determined whether the spelling was completely correct or not. The number of words with open and closed syllables that were spelled fully correctly was analyzed in order to determine the effect of the three practice conditions compared to the control condition and to each other for open and closed syllables. Also, the type of errors before and after practice was analyzed.

Results

Differential effects of the practice conditions were established by comparing the number of words (with open and closed syllables) that were spelled correctly before, directly after practice, and 5 weeks after practice for each of the three practice conditions and the non-practiced control condition. Words other than the open and closed syllable words are not taken into account in the first analyses, but are included in following analyses, because the main question was first on the effects on the learning of open and closed syllable words and not on the other types of words. The mean absolute numbers and percentages correctly spelled open and closed syllabic words for all conditions over time and standard deviations are shown in Table 1. Note that the standard deviations are relatively high, especially in pretest. Figure 1 displays furthermore the mean percentages of correctly spelled words in the open and closed syllabic words before, directly after, and 5 weeks after practice as a function of practice condition. As is clear from the figure, in the direct posttest, the percentages of words that were spelled correctly substantially increased in all four conditions and decreased slightly from posttest to retention test.
https://static-content.springer.com/image/art%3A10.1007%2Fs11881-008-0019-0/MediaObjects/11881_2008_19_Fig1_HTML.gif
Fig. 1

Mean accuracy (% correct) with confidence intervals in open and closed syllabic words of practice and control words before, directly after, and 5 weeks after training as a function of practice condition

Table 1

Mean accuracy in absolute numbers and in percentages, and standard deviations in open and closed syllabic words of practice and control words before, directly after, and 5 weeks after training for each of the three practice conditions and the control condition

 

Before

Directly after

5 weeks after

M

M(%)

SD

M

M(%)

SD

M

M(%)

SD

Visual preview

1.10

22.0

25.6

4.04

80.8

24.6

3.56

71.2

23.6

Segmentation

0.96

19.2

20.6

3.06

61.2

28.1

2.90

58.0

27.2

Dictation

1.02

20.4

23.7

3.06

61.2

29.3

3.02

60.4

30.1

Control

1.00

20.0

24.6

1.96

39.2

27.7

2.46

49.2

27.2

In order to determine specifically whether the spelling of the children improved in the different conditions over time, repeated measures analyses of variance (ANOVAs) were performed on the data of all subjects (F1) and of all items with open and closed syllables (F2) with type of training (visual preview, syllabic segmentation, dictation, control) and time (before, directly after, 5 weeks after) as within-subject variables. Repeated measures analyses showed that all main effects were significant for type of training (visual preview, syllabic segmentation, dictation, control) [F1(3,147) = 25.46, p < 0.001, ηp2 = 0.34, F2(3,57) = 19.97, p < 0.001, ηp2 = 0.51] and for time (before, directly after, 5 weeks after) [F1(2,98) = 169.36, p < 0.001, ηp2 = 0.78, F2(2,38) = 170.70, p < 0.001, ηp2 = 0.90; sphericity is violated in both subject analysis and item analysis, and therefore, Greenhouse–Geisser is applied]. In addition, the interaction effect between condition and time was significant for subjects [F1(6,294) = 12.34, p < 0.001, ηp2 = 0.20] and for items [F2(6,114) = 10.30, p < 0.001, ηp2 = 0.35]. This finding implies that there were differences in improvements between the four conditions over time. Hence, one condition improved more due to practice than another condition (see also Table 1).

To gain further information on the differences between the four conditions in open and closed-syllable words, thus to find an answer to the question which condition improved more than the other, simple contrast analyses are carried out. First, the non-practiced control words improved significantly from pretest to directly after training and also between pretest and 5 weeks after training (p’s < 0.001). This means that the words that were not practiced improved over time. But the words practiced in all three training conditions improved significantly more than the non-practiced control words from pretest to posttest (p < 0.001) and from pretest to retention test (p < 0.05). The three practice conditions improved more than the control condition and were thus effective. Further contrast analyses were carried out to find out whether there were also differences in gains over time between the three practice conditions. These simple contrast analyses on subjects and on items showed that visual previewing was significantly more effective than normal dictation (p < 0.05) and than syllabic segmentation practice (p < 0.05) both directly after and 5 weeks after training. Thus, visual preview was the most effective type of practice as compared to the other practice types directly after training but also on the long term. No significant difference in gain of spelling scores was found between normal dictation and syllabic segmentation [F1(1,49) < 1, F2(1,19) < 1]. Thus, syllabification of the target word as a cue during practice conferred no meaningful benefit to the speller.

Although significant effects were obtained for the total percentage of correctly spelled words for the words that were practiced in the various conditions, one could ask whether the effects of the practice conditions vary for open syllable words or closed syllable words. Words containing closed syllables might profit more from syllabic cues as compared to words with open syllables. Therefore, we examined the interaction between the four conditions over time and the two word types, words with open syllables and words with closed syllables. The percentage of correctly spelled words for all conditions is presented in Table 2. Before practice, words with closed syllables clearly are more difficult to spell correctly as compared to the words with open syllables. Across conditions, only 17.0% of the words with closed syllables are spelled correctly, whereas 25.5% of the words with open syllables are spelled correctly before practice. Directly after and 5 weeks after practice, the words with closed syllables are spelled at best (62.7% and 65.7% overall) compared to the words with open syllables (57.5% and 50.8%). In order to determine whether the two word types showed different gains within the practice conditions, we analyzed the data by repeated measures ANOVAs on the averages for subjects (F1) and for items (F2) with type of training (visual preview, syllabic segmentation, dictation, control) and time (before, directly after, 5weeks after) as within-subject variables and word type (open syllabic words, closed syllabic words) as between-subject variable in items and as within-subjects variable in subjects. This interaction was not significant for subjects [F1(6,294) < 1] nor for items [F2(6,108) < 1]. Thus, the type of word (open versus closed) did not affect the different effects of the practice types over time. The words with closed syllables seem to benefit most from practice, but no significant difference between open and closed syllables was found.
Table 2

Mean accuracy (% correct) of practice and control words before, directly after, and 5 weeks after training for each of the three training conditions and the control condition

 

Before

Directly after

5 weeks after

M

SD

M

SD

M

SD

Words with open syllables

Visual preview

28.00

33.75

80.00

31.94

63.00

37.54

Segmentation

21.00

26.90

58.00

36.92

49.00

38.45

Dictation

26.00

32.33

57.00

37.81

51.00

42.25

Control

27.00

36.72

35.00

36.77

40.00

37.80

Overall

25.50

32.48

57.50

39.07

50.75

39.62

Words with closed syllables

Visual preview

18.00

31.02

81.33

28.70

76.67

30.31

Segmentation

18.00

29.52

63.33

33.84

64.00

30.74

Dictation

16.67

27.97

64.00

31.47

66.67

34.34

Control

15.33

27.94

42.00

34.21

55.33

34.74

Overall

16.97

28.95

62.71

34.88

65.69

33.31

In order to know whether the effects of the practice conditions are valid exclusively for open and closed syllable words, other types of words are included in the experiment, and statistical analyses can show whether the various training effects can also be generalized to other types of words. Analyses on the other types of words show a similar pattern as the open and closed syllabic words: Visual preview appeared to be most effective in these words too, and if these words were taken into account in the repeated measures ANOVAs on the averages for subjects (F1) and items (F2) with type of training and time as within-subject variables and word type (open syllabic words, closed syllabic words, and other words) as between-subject variable in items and as within-subject variable in subjects, the interaction was not significant for subjects [F1(12,588) < 1] nor for items [F2(12,174) < 1]. These results indicate that results can be generalized to all words with ambiguities in spelling. Visual preview was thus most effective for all types of words, and relative improvements due to the dictation training were comparable to the syllabic segmentation training for all three groups of words. Also, analyses of the number of errors in spelling during the computer training show a similar pattern as the results in the dictation tests.

Furthermore, it was hypothesized that externally providing an ambisyllabic segmentation (wek-ker instead of we-kker) during practice in spelling may help the speller to realize the consonant doublet in words with closed syllables. It was thus expected that ambisyllabic segmentation during practice (e.g., wek-ker) would be more effective as compared to segmentation with a pause before the intervocalic consonant (e.g., we-kker). In order to test this hypothesis, half of the subjects received an ambisyllabic segmentation in the words with closed syllables, whereas the other half of the subjects received a segmentation with a pause before the consonant doublet. The children who heard the syllabic segmentation with a pause before the consonant doublet (e.g., we-kker) during practice spelled in the pretest 19.75%, in the posttest 69.14%, and in the test of retention 69.14% of the closed-syllable words correctly. The children who heard the syllabic segmentation with a pause in between the consonant doublet spelled in the pretest 15.94%, in the posttest 56.52%, and in the test of retention 57.97% of the words correctly. To analyze whether these improvements were significantly different for the two types of segmentation (wek-ker or we-kker), a repeated measures ANOVA was performed on all the words with closed syllables with segmentation training and time of testing as within-subject variables and type of segmentation as between-subject variable. This interaction was not significant [F(2,96) < 1], which means that the effect of an ambisyllabic segmentation was not different from the effect of a segmentation with a pause before the consonant doublet.

Finally, an analysis of spelling errors was carried out. All previous analyses were carried out on scores for the correct spelling of the whole word. However, it would be interesting to see whether indeed the special phonological rules with respect to open and closed syllables were violated relatively frequently. Analyses of the errors showed that problems in the spelling of words with open and closed syllables occur mainly in the number of vowels and consonants. In Table 3, the percentages of words spelled with one or two vowels and one or two consonants are presented for open and closed syllables before and directly after training across and between separate conditions. The correct spelling is indicated by a bold letter type. Sometimes, errors occur in other parts of the word and not (only) in the consonants or the vowels, for instance botur instead of boter. These errors were also taken into account in the calculations (that is why percentages do not sum to 100), but are not shown in the table. The data clearly show that initially, errors in words with open syllables are most often phonologically based (59.5% of all words), i.e., the long vowel phoneme in the open syllable is spelled with two vowels instead of one (booter instead of boter). After practice, a substantial change appeared: only 28.5% of the words are still spelled with two vowels. This enhancement appears in all conditions. The number of consonants in open syllables is spelled correctly in most cases before practice, and this number remains about the same after practice. In words with closed syllables, most words are spelled phonologically in pretest too, i.e., the closed syllable is spelled with one consonant instead of two (kaper instead of kapper). After practice, this percentage of errors is decreased substantially. The correct spelling with a double consonant is only spelled in 17.6% of the words before practice but in 62.3% of the words directly after practice. Such an improvement is found for all conditions. Children also improve considerably on the non-practiced control words; however, this gain is clearly not as high as in the practice conditions.
Table 3

Percentage words spelled with one or two vowels and with one or two consonants for the open and closed syllables before and directly after practice across and as a function of conditions

 

Before

Directly after

Vis

Seg

Dict

Contr

Overall

Vis

Seg

Dict

Contr

Overall

Open syllables vowel

One

31.0

27.0

28.0

30.0

29.0

78.0

62.0

62.0

44.0

61.5

Two

57.0

62.0

59.0

60.0

59.5

16.0

25.0

27.0

46.0

28.5

Open syllables consonant

One

85.0

82.0

84.0

87.0

84.5

90.0

83.0

81.0

80.0

83.5

Two

03.0

07.0

03.0

03.0

04.0

04.0

04.0

08.0

10.0

06.5

Closed syllables consonant

One

72.0

79.3

78.0

77.3

76.7

15.3

28.7

31.3

46.0

30.3

Two

18.0

18.0

16.7

16.0

17.6

82.0

66.3

64.0

36.7

62.3

The correct spelling is in bold typeface

Discussion

The current study aimed to investigate whether spelling exercises for poor spellers in the form of repeatedly dictating bisyllabic words with appropriate feedback can be made more effective especially for words with open and closed syllables. Two extensions of a commonly used dictation exercise were tested: In a first practice condition, syllabic segmentation was provided in order to call attention to the length of the vowel in the first syllable, which is crucial for applying the correct phonological spelling rule. In the second condition, the child was allowed during the exercise to inspect the written form of the word before being required to spell the word from memory. The findings show significant effects of practice, but no difference was found between the effects of practice in dictation and practice in spelling with segmentation cues. Allowing a visual preview, however, appeared significantly more beneficial than providing phonological help during practice in spelling or than practice with no additional cues. That is, the spelling performance in pretest to posttests improved most when words were practiced in the visual preview condition.

Because words with open and closed syllables to be spelled require rather complex phonological rules, syllabic segmentation cues in a dictation training were expected to be more effective as compared to a normal dictation training without extra cues. Providing the word auditorily segmented in syllables was expected to draw attention to the length of the vowel in the first syllable and, accordingly, would facilitate the application of the spelling rules that are based on this vowel length. However, the difference between the syllabic segmentation and the normal dictation condition was not significant. The children were thus not supported by the fact that the words were for them segmented in syllables. Also, based on intuitive syllabifications of bisyllabic words with a short vowel in the first syllable by children (Gillis & De Schutter, 1996) and because the medial consonant is explicitly produced twice in the segmentation cue, an ambisyllabic segmentation was hypothesized to facilitate the finding of the correct spelling. But the results show that it did not matter where the moment of the pause was (either between or before the consonants).

Overall, the results indicate that the children do not take advantage of having syllabic segmentation presented before they attempt to spell a bisyllabic word during exercises. Two different accounts can be suggested. First, the metalinguistic task of segmenting a word into syllables is well within the capabilities of the almost 8-year-old children participating in the present study. It is well known from the research literature on the development of metalinguistic phonological skills that many kindergarten children already are perfectly able to segment words into syllables (Treiman, Bowey, & Bourassa, 2002). Although externally providing the present group of participants with a syllabic segmentation, sometimes even with explicit cues for the gemination of the consonant, may be a stimulus to a more objective analysis of the phonology, the findings show that they did not take advantage of this. Second, the particular problem of spelling a bisyllabic word in Dutch with a long or short vowel in the first syllable may not simply be solved by segmenting the word into syllables. Research findings clearly show that this spelling problem occurs rather frequent in beginning spellers and remains so for quite a long time during the development of spelling skills (Landerl & Reitsma, 2005; Notenboom & Reitsma, 2007). Instead of providing syllabification cues, another type of support during exercises is likely needed to provide opportunities to improve on spelling this particular problem. One possibility is to explicitly provide in one form or another the phonological rule every time a word has to be spelled according to these rules. This would actually resemble most of the instructional approaches in contemporary Dutch curricula for spelling. It is interesting to note though that a fundamental part of the rule involves explicit attention to the syllabic structure of the word. Another possibility is that actively learning to produce the specific orthographic form of such words is ultimately the major impetus of reducing spelling errors in this domain. Providing the correct orthographic form within a context of spelling exercises may facilitate establishing this knowledge base.

The results of the present study indicate that a visual preview during exercises in spelling contributed most to gains in spelling performance. It should be noted that this result was obtained despite the fact that a visual preview was no longer available during the posttests. Moreover, the retention data demonstrate that the effect was still significant more than 5 weeks after the final training session. Thus, although visual preview may be conceived as a condition allowing the speller to inspect the visual-orthographic form of a word rather superficially, the learning effect appears to be quite permanent and relatively the highest of all conditions. The requirement to spell the same word from memory a few moments later probably supports a strategy to heavily focus on orthographic details. Whereas during reading for comprehension, for example, only partial cues may be sufficient to identify a word, the specific circumstances in the present spelling exercise likely impose careful reading and close attention to the letter-by-letter structure of the word. Gradually, by multiple experiences, word-specific orthographic knowledge accrues and becomes functionally available to assist the speller in finding the correct spelling. Possible ambiguities from a phonological point of view are resolved by referring to or consulting the word-specific orthographic memory base. Although the learning process of normal repeatedly spelling in dictation conditions may basically proceed in a similar way, the visual preview allows for an additional opportunity to deliberately notice the visual-orthographic details. In fact, the added value of the visual preview condition may, in particular, be due to having this specific occasion to intentionally scrutinize the correct spelling without being involved at the same time with specifying the phonemic structure or applying phonological rules.

Alternatively, the irregularities in the spellings of the other types of words may have caused a side effect that may have contributed to children’s success in the preview condition. Specifically, some of the letters in the other word types could not be predicted by children’s knowledge of phoneme–grapheme regularities. Discovery of this may have caused them to pay greater attention to letter details in all of the words, including the words with open and closed syllables, during the previews in order to spell the words correctly. This idea is supported by our general finding that the visual preview condition was most effective in both the words with open and closed syllables and in the other types of words.

From a developmental perspective, it is remarkable that the present group of young poor spellers from grade 2 benefits most from visual-orthographic practice as compared to practice with phonological cues. Normally, children develop continuously from concrete phonemic analysis to the use of more abstract linguistic and orthographic representation in spelling (Ehri, 1992; Notenboom & Reitsma, 2003; Schlagal, 2001). Learning to spell may be considered as a continuous accumulation of different segments of orthographic knowledge. Considering this gradual shift from a more phonologically based approach to a visual-orthographic approach, the effect of visual preview in practice can be different in spellers from different spelling ages, and one might expect higher effects at later stages.

In contrast, recent training studies revealed that less skilled spellers in grades 5 and 6 improved as much as the skilled spellers in both the visual preview and the spelling pronunciation condition and that younger less skilled spellers showed more benefits from visual preview than from spelling pronunciations as compared to older less skilled spellers (Hilte, Bos, & Reitsma, 2005; Hilte & Reitsma, 2006). But a rigorous investigation in which the effects of visual preview during spelling exercises are developmentally examined has yet to be carried out. Meanwhile, the present findings of strong learning effects of visual preview in rather young, less skilled spellers in grade 2 may be taken as a challenge of the earlier developmental hypothesis.

An interesting observation in the present experiment is that the spelling of control words also improved significantly despite the fact that these words were not practiced. One obvious reason could simply be that it is an effect of repeatedly testing the same words. Another possibility is that the control words were similar to the practiced words because most of them also had open and closed syllables. Children may have noticed the similarity and could have spelled the non-practiced control words by analogy. Yet another option is that they had induced implicitly the specific rules for the spelling of open and closed syllables. Alternatively, they may have been instructed in spelling these types of words in class. The present experiment does not provide data to allow any conclusion on this issue. But whatever reason there is for the improvement of non-practiced control words, it is evident that the increase in accuracy for the practiced words was much larger than for the control words. This finding suggests that a substantial portion of the learning gains is based on the acquisition of word-specific knowledge.

It should be emphasized that in the current study, the conditions were varied within subject: Each subject received all practice conditions. Also, to account for specific word effects, each word occurred in each condition once, but varied over the subjects. This design may have caused learning effects from one condition to the other, which might have blurred our results to a certain extent. For example, by previewing the words spellings, children may have discovered the regularity of the spelling of open and closed syllables and they may have applied this knowledge implicitly in the segmentation and dictation conditions. Nevertheless, even if our results might have become contaminated by such carry over effects, significant differences still stand. It is highly plausible that the effects may have been even larger when a between-subject design was adopted. Moreover, we have deliberately selected a completely within-subject design for the fact that second grade children vary considerably in spelling ability, which might confound practice effects when we would have used a between-subject design. Although the participants in the current study were all relatively poor spellers, the variance between subjects was still very large (see for SDs Table 1), especially in pretest. A within-subject design may preclude that the differences between conditions were the results of this wide variety in spelling skills instead of the results of practice effects. In the case of a between-subjects design, the results between the practice conditions may have been overruled by initial skill differences between conditions. Nevertheless, to rule out possible contamination between conditions, a replication study in which children are randomly assigned to only one of the different conditions can further strengthen the current findings.

Another issue concerns the regularity of the spelling of the word to be spelled. One could reason that visual preview is relatively more helpful when the spelling of a word contains many ambiguities. In some previous studies (Hilte, Bos, & Reitsma, 2005; Hilte & Reitsma, 2006), highly irregular words originating from the French or English language, so-called loan words, were used. Moreover, the printed word frequencies of the words used in these studies were considerably low. In another study, visual dictation proved to be most effective in words containing a spelling ambiguity as compared to words with consistent phoneme–grapheme relations (Van Hell, Bosman, & Bartelings, 2003). A phonemic analysis of the word and employing consistent phoneme–grapheme correspondences is sufficient for correct spelling of regular words. These earlier findings thus suggest that the effect of visual preview may be limited in words with inconsistent phoneme–grapheme relations. In the present experimental training, Dutch words were used with open and closed syllables. These words are relatively consistent if you know how to apply the conditional rule for gemination, which says that double consonants mark preceding short vowel sounds. However, young Dutch spellers appear to have serious difficulties in applying this conditional rule, which automatically makes these words with open and closed syllables rather ambiguous to spell for them. This ambiguity may be comparable with the ambiguities in loan words. If this is the case, visual preview helps to resolve the ambiguities and may support less skilled spellers in learning to spell difficult words. By previewing the visual spelling pattern of these words containing open and closed syllables, children can either store the word-specific information of the words spelling or learn implicitly how and when to use the conditional rule. It remains to be studied whether learning to spell regular words without inconsistent phoneme–grapheme correspondences would also profit from visual preview.

In addition, educational implications of the current findings need to be addressed cautiously. It should be emphasized that the current study did not try to find the most optimal way of instruction. Instead, we tried to examine whether less skilled spellers could be supported by extra cues in computer-based spelling exercises in learning to spell open and closed syllables. Although visual preview appeared to be the most effective practice condition in the current experiment, it does not implicate that children should only be instructed with such a computer exercise. The computer-assisted exercise showed to improve learning in addition to the normal school curricula, but specific recommendation about instructional implications would be premature. Thus, before drawing conclusions about the benefits of the visual preview condition overall, the visual preview approach should first be contrasted with a standardized treatment in a typical classroom setting with common Dutch instructional methods (e.g., providing specific spelling rules).

It might also be possible to further optimize the benefits of the visual preview approach by adding phonological support to the preview condition. Phonological support conferred some benefit relative to the controls, so perhaps the combination of phonological and visual support in computer-based spelling exercises would be even stronger. Thus, further research is needed and may be warranted to determine whether the findings would indicate superior learning when visual preview as a computer-based exercise is more deeply imbedded in the regular school instruction methods, if possible with an additional contrast with a condition in which phonological and visual support are combined.

In conclusion, visual preview showed to be most beneficial for less skilled spellers from grade 2. By previewing the spelling of a word, orthographic memory is fed by the correct pattern of letters and is retrieved or functionally available afterwards. Spelling dictation practice is also effective, but not as effective as visual preview. Extra cues may be effective in a dictation, but syllabic segmentation cues did not show a significant contribution to a normal spelling dictation. This result provides some clear evidence for the fact that less skilled spellers, even in early stages of spelling development, can profit considerably from briefly studying the correct spelling before they spell the word from memory. Further research is warranted in order to more closely specify the effectivity of various formats of visual preview in exercises to improve the skill in spelling.

Copyright information

© The International Dyslexia Association 2008