Evidence of Statistical Learning of Orthographic Representations in Grades 1–5: The Case of Silent Letters and Double Consonants in French

ABSTRACT

We investigated how and when French children in Grades 1–5 acquire orthographic representations for silent letters and double consonants. Linear mixed-effects modeling analyses on the spelling accuracy scores obtained for 2,519 French words were used to test our predictions. As predicted, the presence of a silent letter or double consonant had a unique detrimental effect on spelling accuracy that was not captured by the inconsistency and complexity generated by these letters, and this effect tended to decrease across grades. Important to note, exposure to more frequent silent-letter endings or double consonants had a facilitative effect over and above consistency that did not seem to change across grades. These findings suggest that children implicitly acquire representations for letters with no phonological value. The results obtained for other predictors also suggest a shift from a lower level, phoneme-based processing to a higher level processing at the word and rime levels as children acquire more reading experience.     

Deficits in Orthographic Knowledge in Children Poor at Rapid Automatized Naming (RAN) Tasks?

The degree to which orthographic knowledge accounts for the link between rapid automatized naming (RAN) and reading is contested, with mixed results reported. This longitudinal study compared two groups of 10- and 11-year-old children, a low RAN group (N = 69) and matched controls (N = 74), on various measures of orthographic knowledge. The low RAN group showed a deficit in orthographic knowledge, both at the level of subword letter sequences and of whole words, as well as an unexpected strength in orthographic learning. Our findings underline the persistence of RAN-related reading problems and raise questions about reading strategies in this group.     

Spelling Errors Impede Recognition of Correctly Spelled Word Forms

ABSTRACT

Spelling errors are typically thought of as an effect of a word’s weak orthographic representation in an individual mind. What if existence of spelling errors is a partial cause of effortful orthographic learning and word recognition? We selected words that had homophonic substandard spelling variants of varying frequency (e.g., innocent and inocent occur in 69% and 31% of occurrences of the word, respectively). Conventional spellings were presented for recognition either in context (Experiment 1, eye-tracking sentence reading) or in isolation (Experiment 2, lexical decision). Words elicited longer fixation durations and lexical decision latencies if there was more uncertainty (higher entropy) regarding which spelling is a preferred one. The inhibitory effect of frequency was not modulated by spelling or other reading skill. This finding is in line with theories of learning that predict spelling errors to weaken associations between conventional spellings and the word’s meaning.     

Development of orthographic knowledge in German‐speaking children: a 2‐year longitudinal study

There is growing evidence that children develop orthographic knowledge from the very beginning of literacy acquisition. This study investigated the development of German‐speaking children's orthographic knowledge with a nonword choice task. One nonword in each pair contained a frequent consonant doublet (zommul) and the other nonword contained an infrequent doublet (zobbul). Children (N = 54) performed at chance level in kindergarten but chose nonwords with frequent doublets significantly more often than expected by chance in first and second grade. Correlations between children's orthographic knowledge and their reading and spelling skills were not found. The results indicate that knowledge of frequent double consonants is evident in German‐speaking children from first grade onwards, but it is not related to their reading and spelling performance. This finding is consistent with the view that children in transparent orthographies rely less on frequent letter patterns during reading and spelling compared to children in deep orthographies.     

Correlates of orthographic learning in third‐grade children's silent reading

This study examined word identification, phonological recoding efficiency, familiar word reading efficiency, orthographic choice for familiar words and serial naming speed as potential correlates of orthographic learning following silent reading in third‐grade children. Children silently read a series of short stories, each containing six repetitions of a different target non‐word. They subsequently read target non‐words faster than homophones and preferred target non‐words to homophones in an orthographic choice task, indicating that they had formed functional orthographic representations of the target non‐words through phonologically recoding them during silent story reading. Target non‐word orthographic choice was correlated with all measures bar non‐symbol naming speed. The association between phonological recoding efficiency and orthographic learning lends support to the hypothesis that self‐teaching occurs through phonological recoding even in silent reading. Our findings were not generally consistent with the view that serial naming speed assesses orthographic learning aptitude.     

Spanish-speaking children’s spelling errors with English vowel sounds that are represented by different graphemes in English and Spanish words

Vowels in Spanish have direct one-to-one letter-sound correspondences, whereas vowels in English usually have multiple spellings. For native Spanish-speaking children learning to spell in English, this transition from a shallow to a deep orthography could potentially cause difficulties. We examined whether the spelling of English vowel sounds was particularly difficult for native Spanish-speaking children, and whether the errors are consistent with Spanish orthographic rules. Twenty-six native Spanish-speaking and 53 native English-speaking children in grades 2 and 3 were given real-word and pseudoword spelling tasks in English that included words containing four vowels that have different spellings between Spanish and English. Results supported our hypothesis—native Spanish-speaking children committed significantly more vowel spelling errors that were consistent with Spanish orthography. The number of vowel spelling errors not consistent with Spanish orthography did not differ between the two language groups. These findings suggest that orthographic properties of the children’s native language influence their learning to spell in a second language. Educational implications address how knowledge of this cross language influence can aide teachers in improving spelling instruction.     

Orthographic analysis of words during fluency training promotes reading of new similar words

Remediation of a serious lack in reading fluency often takes the form of repeated reading exercises. The present study examines whether transfer of training effects to untrained (neighbour) words can be enhanced by training with an orthographic focus as compared with emphasising semantics. The effect of oral versus silent reading during training is studied as well. Two groups of reading‐disabled children (mean age=7 years, 11 months) were given repeated reading training with limited exposure duration (350 ms) in which 15 target words were repeated 20 times in exercises focused on either orthography (N=26) or semantics (N=25). The children were required to either read the target words aloud or perform the exercises silently, but this requirement appeared to have no effect on the training results. The results show that untrained neighbour words benefited more from training targets with an orthographic focus than from exercises with a semantic emphasis.     

Orthographic learning during reading: the role of whole‐word visual processing

The self‐teaching hypothesis suggests that knowledge about the orthographic structure of words is acquired incidentally during reading through phonological recoding. The current study assessed whether visual processing skills during reading further contribute to orthographic learning. French children were asked to read pseudowords. The whole pseudoword letter string was available at once for half of the targets while the pseudoword's sub‐lexical units were discovered in turn for the other half. Then memorisation of the targets’ orthographic form was assessed. Although most pseudowords were accurately decoded, target orthographic forms were recognised more often when the pseudowords entire orthographic sequence was available at once during the learning phase. The whole‐word presentation effect was significant and stable from third to fifth grades. This effect was affected neither by target reading accuracy nor by target reading speed during the learning phase. Results suggest that beyond recoding skills, the ability to process the entire orthographic letter string at once during reading contributes to efficient orthographic learning.     

Orthographic knowledge predicts reading and spelling skills over and above general intelligence and phonological awareness

It is widely accepted that general intelligence and phonological awareness contribute to children’s acquisition of reading and spelling skills. A further candidate in this regard is orthographic knowledge (i.e., the knowledge about permissible letter patterns). It consists of two components, word-specific (i.e., the knowledge of the spelling of specific words) and general orthographic knowledge (i.e., the knowledge about legal letter patterns of a writing system). Among German students, previous studies have shown that word-specific orthographic knowledge contributes to both reading and spelling. The results regarding general orthographic knowledge and its contribution to reading and spelling are inconsistent. The major goal of the present study was to determine the incremental predictive value of orthographic knowledge for reading and spelling skills among German elementary-school children (N = 66), over and above the contribution of general intelligence and phonological awareness. The second goal was to examine whether there is a difference between the two subtypes of orthographic knowledge in the amount of their respective contribution to reading and spelling performance. The results show that word-specific as well as general orthographic knowledge contribute to both reading and spelling performance, over and above intelligence and phonological awareness. Furthermore, it reveals that both word-specific and general orthographic knowledge explain more variance of spelling compared to reading. Possible explanations for these results, limitations, and implications of the study are being discussed.

     

Kindergarten Children's Initial Spoken and Written Word Learning in a Storybook Context

Kindergarteners (M age = 6;2) were exposed to novel spoken nonwords and their written forms within a storybook reading context. Following each of 12 stories, the children were required to spell and identify 12 novel written nonwords and then verbally produce and comprehend the spoken version of those words. Results indicated the children acquired initial specific phonological and orthographic representations. Spoken and written word learning skills were strongly associated and both were influenced by the words' linguistic regularities. Spoken word learning ability explained 62% of the variance on a spelling measure, whereas written word learning ability predicted 42% of the variance on a reading measure. The results provide evidence that beginning readers employ simultaneously a mutually shared learning mechanism that is sensitive to statistical regularities of words when engaged in the process of learning new spoken words and their written forms.     

Patterns of generalisation after treating sub‐lexical spelling deficits in a child with mixed dysgraphia

This paper presents a treatment study with a developmental dysgraphic girl, KM, and addresses the mechanisms by which orthographic learning of spelling rules might occur. Before treatment, KM's spelling of words and nonwords was impaired. Analyses of spelling errors indicated poor knowledge of sound‐to‐letter correspondences. Treatment focused on two spelling rules and was successful: spelling improved for both regular words and nonwords. Untrained words that included the training rules also improved, but later than nonwords and trained words. This delayed generalisation was explained through feedback mechanisms between orthographic lexicon and graphemic buffer.     

Spelling in Children With Dyslexia: Analyses From the Treiman-Bourassa Early Spelling Test

We examined the oral and written spelling performance on the Treiman-Bourassa Early Spelling Test (Treiman & Bourassa, 2000a) of 30 children with serious reading and spelling problems and 30 spelling-level-matched younger children who were progressing normally in learning to read and spell. The 2 groups' spellings were equivalent on a composite measure of phonological and orthographic sophistication, representation of the phonological skeleton of the items, and orthographic legality. The groups showed a similar advantage for words over nonwords on the phonological skeleton and orthographic legality measures. The children with dyslexia and the comparison children also showed an equivalent advantage for written over oral spelling on the composite and phonological skeleton measures. Further analyses revealed that children with dyslexia made many of the same linguistically based errors as typically developing children but also pointed to some subtle differences between the groups. Overall, the spelling performance of children with dyslexia appears to be quite similar to that of normally progressing younger children.     

Children’s orthographic representations in English and Greek

Two experiments were conducted to compare the development of orthographic representations in children learning to read English and Greek. Nonsense words that either shared both orthography and phonology at the level of the rhyme with real words (comic-bomic), phonology only (comic-bommick), or neither (dilotaff) were created for each orthography. Experiment 1 compared children’s reading of bisyllabic and trisyllabic nonsense words likebomic vs.bommick, taffodil vs.tafoddyl, and found a significant facilitatory effect of orthographic rhyme familiarity for English only. Experiment 2 compared children’s reading of trisyllabic nonsense words that either shared rhyme phonology with real words (tafoddyl) or did not (dilotaff). Significant facilitation in reading accuracy was found for shared rhyme phonology in English, with a significant speed advantage in Greek. These results are interpreted in terms of the level of phonology that is represented in the orthographic recognition units being developed by children who are learning to read more and less transparent orthographies.     

Training letter and orthographic pattern recognition in children with slow naming speed

Although research has established that performance on a rapid automatized naming (RAN) task is related to reading, the nature of this relationship is unclear. Bowers (2001) proposed that processes underlying performance on the RAN task and orthographic knowledge make independent and additive contributions to reading performance. We examined the benefits of training orthographic pattern recognition and speeded letter recognition for children in Grades 1 and 2 with slow naming speed. Children first received training in either orthographic pattern recognition or speeded letter recognition, and then switched to the other type of training. Results indicated that speeded letter recognition can improve through training, but only when preceded by training in orthographic pattern recognition. Orthographic pattern recognition training improved the accuracy and speed of reading training words, whether training occurred alone or following letter training. Letter training prior to the orthographic training provided no additional benefit. Together, these results argue for the importance of orthographic training for children with slow naming speed.     

Orthographic Learning During Oral and Silent Reading

This study examined orthographic learning in oral and silent reading conditions. Dutch third graders read, either aloud or silently, short texts containing novel target (pseudo) words. The acquisition of new word-specific orthographic knowledge was assessed several days later by comparing target spellings with homophonic spellings in tasks requiring orthographic choice, spelling, and naming. It was predicted that orthographic learning would be evident in both oral and silent conditions but stronger in the oral condition. As expected, orthographic learning was evident in both oral and silent conditions. This finding suggests that support for the self-teaching hypothesis of orthographic learning obtained in studies of unassisted oral reading can be generalized to the more common form of independent reading: silent reading. In addition, the results on the naming task provided some evidence for stronger orthographic learning during oral reading, but the two spelling tasks did not.     

Predicting Spelling Behaviour in 7 to 11 Year-olds

There is an increasing number of studies which demonstrate that readers in more transparent orthographies than English, such as Italian, Spanish, Turkish, Greek and German have little difficulty in decoding written words, while English children have many more problems. Increasingly, lack of orthographic transparency in English is seen as having a powerful negative effect on the development of reading skills in English-speaking children. There is evidence that English-speaking children who fail to acquire reading skills may fall into two distinct categories: those who would succeed in languages, other than English, that have greater orthographic consistency; and, those who would still have problems even with perfect orthographic transparency. The first, larger, group are let down by the interaction of poor teaching methods and an incomprehensible system of orthography. At Hull University’s Institute for Learning, word factors associated with poor spelling and reading have been identified. We have found three factors which account for the relative ease with which pupils can spell words: frequency of the word in the English language, length of the word, and the presence of “tricky” letters or letter combinations. Data is presented illustrating our predictive model of spelling which enables word difficulty to be calculated from the characteristics of English words. The implications the model has for teaching and learning English is elaborated, with particular reference to the most common 150 words in the English language.     

Learning New Words Affects Nonword Pronunciation in Children

In two experiments we examined how children's nonword pronunciations are influenced by learning words. In Experiment 1, children pronounced nonwords before and after learning words sharing orthographic rimes with the nonwords. These rimes varied in spelling-to-sound consistency and regularity. Children's nonword pronunciations were more sensitive to consistency and regularity after instruction than before. Experiment 2 expanded upon Experiment 1 by modifying the instruction to highlight regularity and consistency in rime unit neighborhoods and by including both younger (M age = 7.6) and older (M age = 9.92) participants. After instruction, Experiment 2 participants demonstrated greater sensitivity to rime unit consistency and regularity than Experiment 1 participants. In both experiments, the children, especially the younger participants, made more adultlike pronunciations after instruction than before. We conclude that learning words varying in consistency and regularity increased the children's sensitivity to these properties.     

Development and Prediction of Context-Dependent Vowel Pronunciation in Elementary Readers

ABSTRACT

As children learn to read, they become sensitive to context-dependent vowel pronunciations in words, considered a form of statistical learning. The work of Treiman and colleagues demonstrated that readers’ vowel pronunciations depend on the consonantal context in which the vowel occurs and reading experience. Using explanatory item-response models we examined child- and nonword-factors associated with children’s assignment of more versus less frequent grapheme-phoneme correspondences (GPC) to vowel pronunciations as a function of rime coda in monosyllabic nonwords. Students (N = 96) in grades 2-5 read nonwords in which more versus less frequent vowel GPCs were wholly supported or partially favored by the rime unit. Use of less frequent vowel GPCs was predicted by set for variability, word reading, and rime support for the context-dependent vowel pronunciation. We interpret the results within a developmental word reading model in which initially incomplete and oversimplified GPC representations become more context dependent with reading experience.     

The role of feedback from phonology to orthography in orthographic learning: an extension of item‐based accounts

Skilled readers were trained to recognise either the oral (n=44) or visual form (n=40) of a set of 32 novel words (oral and visual instantiation, respectively). Training involved learning the ‘meanings’ for the instantiated words and was followed by a visual lexical decision task in which the instantiated words were mixed with real English words and untrained pseudowords, and the instantiated words were to be considered as words. The phonology‐to‐orthography consistency (feedback consistency) of the instantiated words was manipulated to investigate the role of feedback from phonology in orthographic learning. Masked consonant and vowel‐preserving form primes were used in the lexical decision task as probes of orthographic learning. Feedback‐consistent instantiated words were recognised significantly faster in lexical decision than feedback‐inconsistent instantiated words, and facilitation was significantly greater from consonant‐preserving than vowel‐preserving primes for orally but not visually instantiated words. The results support the hypothesis that orthographic representations based on a consonant frame can be generated from the speech signal before encountering the printed forms, and that feedback from phonology is involved in the early stages of orthographic learning.