Edited by: Manuel Perea, University of Valencia, Spain
Reviewed by: Eva E. Gutierrez-Sigut, University College London, United Kingdom; Heather Grantham, Washington University in St. Louis, United States
This article was submitted to Language Sciences, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study set out to explore the cognitive and linguistic correlates of orthographic learning in a group of 32 deaf and hard of hearing children with cochlear implants, to better understand the factors that affect the development of fluent reading in these children. To date, the research about the mechanisms of reading fluency and orthographic learning in this population is scarce. The children were between 6:0 and 10:11 years of age and used oral language as their primary mode of communication. They were assessed on orthographic learning, reading fluency and a range of cognitive and linguistic skills including working memory measures, word retrieval and paired associate learning. The results were analyzed in a set of correlation analyses. In line with previous findings from children with typical hearing, orthographic learning was strongly correlated with phonological decoding, receptive vocabulary, phonological skills, verbal-verbal paired-associate learning and word retrieval. The results of this study suggest that orthographic learning in children with CI is strongly dependent on similar cognitive and linguistic skills as in typically hearing peers. Efforts should thus be made to support phonological decoding skill, vocabulary, and phonological skills in this population.
For deaf and hard of hearing children who are fitted with cochlear implants, here referred to as
Although the relative performance of children with CI on different aspects of reading has rarely been directly addressed in previous research, some findings would suggest relatively more difficulties with phonological decoding processes than with orthographic word recognition (e.g.,
Recent cross-linguistic studies have demonstrated that the most important predictors of early reading development are a child’s phonological skills including letter knowledge and phonemic awareness, but also verbal fluency (e.g.,
Phonological skills are particularly important for learning to sound out words letter by letter in a process referred to as
Beginning readers typically read mainly by means of phonological decoding but as they become more experienced they gradually learn to recognize whole words by sight (
In order to read fluently by means of orthographic word recognition, children need to build up a large lexicon of orthographic representations in their long-term memory. The new orthographic representations are acquired through a process referred to as
Orthographic learning is typically measured in tasks in which children are presented with new nonsense words (Swedish example ‘
Phonological decoding has proved to be the most important predictor of orthographic learning in typically hearing children. This is, for example, demonstrated by decreased orthographic learning in studies designed to prevent phonological decoding, e.g., by simultaneously producing an irrelevant pseudo word at the time when they are exposed with the target non-word (e.g.,
Furthermore,
In several studies, children with CI have been found to perform within the normal range, not significantly different from TH peers on measures of VSTM, both when the test stimuli has been presented simultaneously (e.g.,
We set out to investigate the cognitive and linguistic correlates of orthographic learning in a sample of Swedish children with CI. We also wanted to compare the pattern of correlations between orthographic learning and cognitive/linguistic skills to the pattern of correlations between word decoding fluency and cognitive/linguistic skills in order to find out whether orthographic learning is dependent on the same underlying processes as word decoding.
It is important to increase our knowledge about orthographic learning in children with CI as it has the potential to give insights into the process of reading acquisition in this population of children and possibly understand the reported relative drop in reading performance when the children grow older (c.f.,
This study was conducted as a part of a longitudinal project on reading development in children with CI. The children were assessed on a range of cognitive and linguistic measures, which have all been found to predict reading skill or orthographic learning in children with typical reading development. The children’s performance on measures of reading, cognition and language was also compared to age-norms or previously collected results from age-matched TH children.
This research was approved by the Research Ethics Review Committee at Linköping University (Dnr 2011/295-31).
Thirty-two children with cochlear implants participated in the study. Written informed consent was obtained from the children’s parents. All of the participants were implanted at the cochlear implant clinic, Karolinska University hospital where they were also followed up regularly once per year. Between those follow up appointments, the children attended some regular speech and listening rehabilitation at their local hospitals. It should be noted that the cochlear implant clinic at Karolinska University hospital has the highest number of cochlear implant patients in Sweden with a catchment area of approximately 5 million people. The sample was thus relatively representative of Swedish children with CI although the inclusion criterion was that the children should be able to follow the national school curriculum. The results from one of the children were excluded in all analyses because of a low score on non-verbal IQ (10th percentile on Raven CPM). Seventeen of the children were girls and 15 boys and the chronological age range was 6;0 – 10;11 years (mean 8;4 years). The sample was heterogeneous in terms of cause of deafness and age of implantation for first and second CI. Etiology and age at implantation for the sample is summarized in
Etiology and age at implantation.
Mean | SD (range) | |
---|---|---|
Age at CI1 (months) | 25.9 | 18 (7–69) |
Age at CI2 (months) | 31 | 23 (8–105) |
Unilateral CI | 6 | 6/32 |
Bilateral CI | 26 | 26/32 |
9/32 | ||
Congenital CMV | 5 | |
Meningitis | 4 | |
14/32 | ||
Unspecific heredity∗ | 3 | |
Connexin 26 | 4 | |
Usher Type 1 | 2 | |
Jerve-Lange Nielsen syndrome | 3 | |
Pendred’s syndrome | 2 | |
9 | 9/32 | |
The whole group of children were fitted with their first cochlear implant at 25.9 months on average (SD = 18; range: 7 months – 69 months). Twenty-six of the children (81%) had bilateral CI:s and were implanted with their second CI at 31 months on average (SD: 23; range: 8 months – 105 months). Twenty-seven of the children used oral language as their only mode of communication and 5 children used both oral language and sign language.
Speech perception in quiet was, as measured by phonetically balanced lists, on average 69.8% (SD: 15.3; range: 44–100). Speech perception data was missing for 3 of the children. Twenty-seven of the children were integrated in schools for TH children, 12 of those children had a teaching assistant to support them in class and they other 15 integrated children did not receive any support. Five children attended classes for children with hearing loss but with teaching in oral language, and 2 children attended sign language classes with teaching in both sign language and oral language.
The participants had percentile scores ranging between 25–95 (median: 75) on the Raven Colored Progressive Matrices test (
A range of measures to assess reading ability, orthographic learning, paired-associate learning, working memory, and language skills were administered to all participants. An overview of the test measures is presented in
Tests administered together with proportions of children who performed within 1 standard deviation of the mean for their age, and means, standard deviations and range per test for the whole group of participating children.
Measures of language and cognitive skills | Test | Quantification | Proportion of children within or above 1 SD of TH age mean | Mean (SD) | Range |
---|---|---|---|---|---|
Complex working memory | Sentence completion and recall ( |
Number of correctly recalled words (maximum score = 18) | 18/32 | 9.7 (4.3) | 1–17 |
Visual short-term memory | Matrix Pattern Test ( |
Number of cells in the most difficult pattern correctly reproduced (maximum score = 8) | 15/32 | 3.8 (1.5) | 1–6 |
Phonological short-term memory | Non-word repetition ( |
Percent correctly reproduced consonants in whole test (100% = 120 consonants) | 3/22 | 58 (18) | 15–89 |
Visual–Visual PAL | Test adapted from |
Number of correct answers (maximum score = 20) | N/A | 15 (4.4) | 7–20 |
Visual-Verbal PAL | Test adapted from |
Number of correct answers (maximum score = 20) | N/A | 6.7 (5.0) | 0–18 |
Verbal-Verbal PAL | Test adapted from |
Number of correct answers (maximum score = 20) | N/A | 2.5 (3.7) | 0–15 |
Receptive Vocabulary | PPVT-III ( |
Number of correctly identified pictures (maximum score = 228) | 25/32 | 114.4 (31.5) | 51–176 |
Word Fluency | FAS ( |
Number of words starting with F, S, S produced within 1 min | 28/31 | 15.5 (7.3) | 2–30 |
Animals ( |
Number of animals produced within 1 min | 31/31 | 14.9 (4.0) | 9–23 | |
Expressive vocabulary | Boston Naming Test ( |
Number of correctly named pictures (maximum score = 60) | 22/32 | 32.9 (10.7) | 7–48 |
Phonological Skills | Phoneme deletion ( |
Number of correctly manipulated words (maximum score = 12) | N/A | 9.3 (3.8) | 4–12 |
Non-word decoding fluency | LäSt (Elwér, Fridolfsson, Samuelsson, and Wiklund, C. (2009) | Number of correctly read non-words in 2∗45 s (maximum score = 126) | 31/32 | 48.2 (23.0) | 11–98 |
Word decoding fluency | LäSt ( |
Number of correctly read words in 2∗45s (maximum score = 200) | 31/32 | 84.0 (38.9) | 16–147 |
Orthographic skills | Orthographic Choices ( |
Number of correctly identified spellings (maximum score = 40) | 19/25 | 30.3 (9.3) | 10–40 |
Orthographic Learning | Test adapted from |
Number of correctly spelled non-words (maximum score = 16) | 16/32 | 6.31 (4.3) | 0–15 |
The Scandinavian version of the
In this task, the child was asked to learn pairs of spoken nonsense words using two sets of four CVC nonsense words (vak, dap, lut, hab; jom, neg, tem, and pog). The child was first asked tp repeat the nonsense words and then the experimenter said the associations twice, for example, ‘dap’ goes with tem’, [2-s interval], ‘dap goes with tem’. After all pairs were introduced, the child was asked ‘which other word goes with dap?’ The responses were recorded by the examiner and feedback of the same kind as in the initial presentation was provided, for example, “Do you remember? dap goes with tem”. The procedure was repeated five times, making a total of 20 learning trials/responses.
In this task, the child was asked to learn shape – nonsense word pairings. A different set of shapes was made in the same way as those used in the visual-visual PAL condition. The nonsense words each contained three phonemes in CVC format. First, the experimenter asked the child to repeat the four non-words (e.g., vob, lep, dok, and haf) in order to make sure that he/she was able to pronounce them. The experimenter then held a set of four cards with different eight-point shapes. She showed the child one card at a time and said the associations twice. For example, ‘this shape goes with lep, [2-s interval], this shape goes with lep’. After all four pairs were introduced, the experimenter presented one card at a time and asked the child ‘which nonsense word goes with this shape?’ The experimenter recorded the child’s responses and feedback was given on each trial, in the same format as in the initial presentation, for example, “Do you remember? This shape goes with lep”. This procedure was also repeated five times, making a total of 20 learning trials/responses.
This condition of the task was assessed as a comparison to verbal-verbal and visual-verbal PAL. This test was included to ensure that the effects of PAL were not general across modalities. In this task, children were asked to learn which shapes went together. Two sets of cards with different eight-point shapes were used.
Visuospatial short-term memory was assessed in a
The third edition of the Peabody Picture Vocabulary Test (PPVT-III;
Phonological short-term memory was assessed in the
Complex working memory was measured in the Sentence Completion and Recall task (
Word fluency was assessed with the
Phonological skills were assessed in a phoneme deletion task (
All children were tested in connection to a regular follow-up appointment at the cochlear implant clinic, Karolinska University Hospital. They were assessed individually by a clinical speech language pathologist who was familiar to them. The tests were administered in two 1-h sessions, one per day on two consecutive days. All of the tests were presented in random order across both test sessions. The test instructions were given orally.
Raw scores were used as outcome measures in all analyses of the data except for Raven’s CPM for which percentile was used in the analyses.
In order to reduce the number of measures in the correlation analyses, a composite measure of word fluency (FAS and Animals) and expressive vocabulary (the Boston Naming test, BNT) was computed based on z-scores of these test measures. This composite measure, referred to as
For the test measures where we had reference data from TH children,
For some children, there are missing data in some of the tests. This is because the test times were set and in some cases we needed to prioritize among tests in order to keep the time limits.
All of the children except for one performed at ceiling on the word fluency measures (FAS and Animals) and more than two thirds of them performed within or above 1 SD of the mean of TH reference children on the expressive vocabulary test (Boston Naming). Twenty-five out of 32 children performed within 1 SD on the receptive vocabulary test (PPVT). The results were particularly low on the non-word repetition test which was used to measure phonological short-term memory. Only three out of the 22 children who completed the non-word repetition test performed within or above 1 SD of TH mean.
About 50 percent of the children performed within or above the 1 SD limit on the measures of complex and visual short-term memory capacity. All of the children had scores within or above 1 SD of TH mean on the reading fluency tests.
The relationships between orthographic learning, reading measures, and cognitive/linguistic skills were analyzed in second order Pearson partial correlation analyses with age and non-verbal IQ partialled out. These correlations are displayed in
Partial correlations between orthographic learning, reading and cognitive/linguistic skills.
1 | 2 | 3 | 4 | 5 | 6 | 7 | 9 | 10 | 11 | 12 | 13 | 14 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Orthographic Learning | 1.000 | ||||||||||||
2. LäSt words | 0.628∗∗∗ | 1.000 | |||||||||||
3. LäSt non-words | 0.729∗∗∗ | 0.858∗∗∗ | 1.000 | ||||||||||
4. Orthographic Choices | 0.462∗ | 0.609∗∗∗ | 0.617∗∗∗ | 1.000 | |||||||||
5. PPVT-III | 0.636∗∗∗ | 0.313 | 0.393∗ | 0.221 | 1.000 | ||||||||
6. Word retrieval | 0.823∗∗∗ | 0.736∗∗∗ | 0.691∗∗∗ | 0.406∗ | 0.595∗∗ | 1.000 | |||||||
7. Phoneme deletion | 0.651∗∗∗ | 0.591∗∗ | 0.611∗∗ | 0.364 | 0.460∗ | 0.647∗∗∗ | 1.000 | ||||||
8. PAL verbal-verbal | 0.660∗∗∗ | 0.269 | 0.406∗ | 0.310 | 0.521∗∗ | 0.548∗∗ | 0.265 | 1.000 | |||||
9. PAL visual-verbal | 0.409∗ | 0.399∗ | 0.414∗ | 0.222 | 0.404∗ | 0.479∗∗ | 0.268 | 0.544∗∗ | 1.000 | ||||
10. PAL visual-visual | 0.399∗ | 0.604∗∗ | 0.481∗∗ | 0.329 | 0.343 | 0.510∗∗ | 0.494∗∗ | -0.003 | 0.287 | 1.000 | |||
11. Matrix pattern | 0.457∗ | 0.443∗ | 0.492∗ | 0.093 | 0.244 | 0.386 | 0.346 | 0.125 | 0.251 | 0.637∗∗ | 1.000 | ||
12. Sentence completion and recall | 0.338 | 0.418∗ | 0.272 | -0.013 | 0.430∗ | 0.592∗∗ | 0.197 | 0.360 | 0.524∗∗ | 0.593∗∗ | 0.219 | 1.000 | |
13. Non-word repetition pcc | 0.447∗ | 0.411 | 0.474∗ | 0.003 | 0.397 | 0.718∗∗∗ | 0.535∗ | 0.163 | 0.280 | 0.611∗∗ | 0.240 | 0.738∗∗∗ | 1.000 |
Bonferroni correction for multiple comparisons was used in order to avoid Type-I errors. The significance level applied here was then α = 0.001. The correlations of interest were those between orthographic learning and word decoding fluency on the one hand and cognitive/linguistic measures on the other hand. The correlations that were significant at α < 0.001 are summarized and discussed below. However, for comparison, correlations that were significant at the conventional levels, α = 0.05 and α = 0.01 are marked with asterisks in
Orthographic learning was significantly correlated with reading fluency (decoding of words and non-words), receptive vocabulary, word retrieval, phonological skills (phoneme deletion) and verbal-verbal PAL. Neither the correlation between orthographic learning and speech perception in quiet nor the correlations between orthographic learning and age at implantation of first or second CI, were significant after Bonferroni correction.
Word decoding fluency was strongly correlated with non-word decoding fluency, orthographic skills and word retrieval.
The aim of this study was to investigate the cognitive and linguistic correlates of orthographic learning in Swedish children with cochlear implants. A second aim was to compare the pattern of correlations between orthographic learning and cognitive/linguistic skills to the correlations between word decoding fluency and cognitive/linguistic skills in order to find out whether orthographic learning is dependent on the same underlying processes as word decoding.
The results showed strong associations between non-word decoding fluency and orthographic learning. Non-word decoding is a measure of childrens phonological decoding skills and this finding is thus in line with the self-teaching hypothesis (e.g.,
The word decoding fluency test was also strongly correlated with orthographic learning. Phonological decoding may be the active mechanism in this relationship as well because children may use phonological decoding for reading words that they are not familiar with. When seeing a familiar word, on the other hand, the reader typically recognizes it immediately and performance in word decoding fluency tasks may thus be dependent on both phonological decoding skill and automatic word recognition. A relationship between orthographic learning and phonological decoding has recently been found in children with moderate – profound hearing loss (
Four other cognitive and linguistic skills were strongly correlated with orthographic learning: receptive vocabulary, word retrieval, phonological skills (phoneme deletion), and verbal-verbal PAL.
The correlation between receptive vocabulary and orthographic learning is in line with previous results from TH children (
Word retrieval involves the quick retrieval of phonological and semantic information from long-term memory (e.g.,
Verbal-verbal PAL is the ability to associate two pieces of verbal information, in this case two nonsense words. Some previous studies have found associations between verbal-verbal PAL and reading measures in TH children (
Research on TH children suggests that cross-modal visual-verbal PAL is important for early reading because reading involves arbitrary association between visual and verbal information, for example when learning the names and sounds of letters (
Word decoding fluency was strongly correlated with three of the cognitive/linguistic skills: non-word decoding fluency, orthographic skill and word retrieval. The correlation between orthographic learning and non-word decoding fluency stresses the importance of phonological decoding in fluent reading for children with CI at a relatively early stage of reading acquisition (age 6–10). Previous findings from children with TH indicate that word recognition is strongly dependent on phonological decoding early in reading development (
The association between non-word decoding and phoneme deletion was expected as phonological skills are well known to be related to phonological decoding, in particular for children early in reading development (e.g.,
In previous research, rapid naming has been found to be a strong predictor of word reading fluency (
All in all orthographic learning and word decoding fluency were both strongly associated with phonological decoding fluency and word retrieval in this sample of children with CI. Orthographic learning was also correlated with receptive vocabulary, phonological skills and verbal-verbal PAL, whereas word decoding fluency was significantly correlated with existing orthographic knowledge. It is likely that the task of orthographic learning after only a brief exposure requires a broader range of skills than the reading task for this sample of children. Although the current study does not allow for conclusions about causality, the skills that were strongly correlated with orthographic learning in this study may indicate possible areas in which intervention programs may be applied in order to improve orthographic learning and thereby reading skill in children with CI. This is an important area for future research.
It should be noted that we did not find significant relations between orthographic learning and the demographic variables, age at implantation of first or second CI or hearing levels. The effects of these and other demographic variables such as etiology and educational setting should be further investigated in a larger sample of children with CI.
MW, UL, LA, EK, EÖ, and BL were involved in the planning, and writing phases of this study. UL, EÖ, and LA collected the data. MW prepared the draft. UL, LA, EK, BL, and EÖ contributed by reading and commenting on drafts of the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.