Journal of Experimental Child Psychology 139 (2015) 256–264
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
Brief Report
Kindergarteners’ performance in a sound–symbol paradigm predicts early reading Josefine Horbach a,⇑, Wolfgang Scharke a, Jennifer Cröll a, Stefan Heim b,c,d, Thomas Günther a,e a Child Neuropsychology Section, Department of Child and Adolescent Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, 52074 Aachen, Germany b Department of Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, 52074 Aachen, Germany c Jülich Aachen Research Alliance (JARA)–Translational Brain Medicine, 52428 Jülich, Germany d Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich, 52428 Jülich, Germany e Faculty of Health, Zuyd University, 6419 DJ Heerlen, The Netherlands
a r t i c l e
i n f o
Article history: Available online 10 July 2015 Keywords: Early reading development Sound–symbol learning Serial processing Rapid automatized naming Phonological awareness Predictors of reading
a b s t r a c t The current study examined the role of serial processing of newly learned sound–symbol associations in early reading acquisition. A computer-based sound–symbol paradigm (SSP) was administered to 243 children during their last year of kindergarten (T1), and their reading performance was assessed 1 year later in first grade (T2). Results showed that performance on the SSP measured before formal reading instruction was associated with later reading development. At T1, early readers performed significantly better than nonreaders in learning correspondences between sounds and symbols as well as in applying those correspondences in a serial manner. At T2, SSP performance measured at T1 was positively associated with reading performance. Importantly, serial application of newly learned correspondences at T1 explained unique variance in first-grade reading performance in nonreaders over and above other verbal predictors, including phonological awareness, verbal short-term memory, and rapid automatized naming. Consequently, the SSP provides a promising way to study aspects of reading in preliterate children. Ó 2015 Elsevier Inc. All rights reserved.
⇑ Corresponding author. Fax: +49 241 803380571. E-mail address: [email protected] (J. Horbach). http://dx.doi.org/10.1016/j.jecp.2015.06.007 0022-0965/Ó 2015 Elsevier Inc. All rights reserved.
J. Horbach et al. / Journal of Experimental Child Psychology 139 (2015) 256–264
257
Introduction To acquire reading, children must learn the code for representing speech as a series of visual symbols (Ziegler & Goswami, 2005). Multiple factors can influence the success or failure of reading acquisition, including genetic (Gilger, Pennington, & Defries, 1991; Pennington & Lefly, 2001), environmental (Bus, van IJzendoorn, & Pellegrini, 1995; Leseman & Jong, 1998), and cognitive factors (e.g., Hammill, 2004; Snowling & Hulme, 2008). Of the potential cognitive factors that might influence reading ability, there is evidence that phonological awareness (PA)—that is, the ability to identify, memorize, and manipulate speech sounds (e.g., Lonigan, Burgess, & Anthony, 2000; Vellutino, Fletcher, Snowling, & Scanlon, 2004)—and rapid naming of high-frequency items (RAN) (Hammill, 2004; Norton & Wolf, 2012) are important predictors of reading acquisition. Another factor associated with reading acquisition is verbal short-term memory (verbal STM) (Gathercole, Alloway, Willis, & Adams, 2006). During the initial phase of reading, a high short-term memory capacity seems to be critical in helping children to memorize speech sounds and merge them into whole words. To develop reading, the association of single sound–symbol pairs is necessary. Furthermore, decoding words requires the serial processing of the newly learned sound–symbol correspondences. Both PA and RAN have elements of these requirements. Good PA skills help to differentiate the sounds when assigning phonemes to graphemes, but tasks that have assessed PA do not require the ability to decode information in a serial manner. Tasks measuring RAN require the ability to process highly automated items in a serial manner but do not account for how learning to read requires the processing of newly learned sound–symbol pairs. Consequently, PA and RAN cannot fully explain reading acquisition. However, the ability to associate sound–symbol pairs, which mirrors the process of reading, is likely to be a critical predictor of reading acquisition. Existing studies examining the role of sound–symbol associations in reading have used paradigms that require the serial application of newly learned visual–verbal associations and correlated performance on these tasks with reading ability. For example, Aravena, Snellings, and Tijms (2013) developed an artificial orthography and showed that normal readers outperformed students with dyslexia when applying newly learned sound–symbol associations in a serial context. Interestingly, the two groups differed in their ability to apply knowledge of new letter names in a serial manner even though they did not differ in their knowledge of the new letter names themselves. Elbro, Daugaard, and Gellert (2012) tested students with dyslexia on a paradigm that also required them to learn new sound–symbol correspondences and apply them in a serial context. Participants needed to learn three new symbols and decode strings of them in a serial manner. The authors found that this dynamic measure of reading was able to differentiate between dyslexic and nondyslexic readers. However, it is still unknown whether the ability to learn new sound–symbol correspondences is predictive of reading acquisition. Both of these studies were cross-sectional and so could not assess potential causal associations between the ability to decode new sound–symbol correspondences serially and reading acquisition. Moreover, participants in both studies already had several years of reading experience (mean age of 9 years in Aravena et al., 2013, and adults in Elbro et al., 2012) when tested. Currently, to our knowledge no study has assessed whether the serial application of newly learned sound–symbol pairs is predictive of reading ability before children are formally instructed to learn how to read. Therefore, the current study examined whether performance in a simple sound–symbol paradigm (SSP) at kindergarten age is predictive of reading ability in first grade. To assess the specific predictive value of the SSP, we sought to examine unique variance associated with SSP controlling for other well-established predictors of reading ability, including PA, verbal STM, RAN, early literacy, and environmental factors. Method Sample and procedures A total of 268 kindergarten children (150 boys, average age at T1 = 6 years 2 months) took part. At the end of first grade (T2), 243 children were retested. All children were native German speakers
258
J. Horbach et al. / Journal of Experimental Child Psychology 139 (2015) 256–264
recruited from regular kindergartens and had an IQ equal to or greater than 85. The level of education of mothers and fathers was assessed via a questionnaire, coded as 1 (no/low educational achievement), 2 (medium educational achievement), or 3 (high educational achievement) and averaged. The categories of level of education were defined in accordance with the German school system (low:
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
Brief Report
Kindergarteners’ performance in a sound–symbol paradigm predicts early reading Josefine Horbach a,⇑, Wolfgang Scharke a, Jennifer Cröll a, Stefan Heim b,c,d, Thomas Günther a,e a Child Neuropsychology Section, Department of Child and Adolescent Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, 52074 Aachen, Germany b Department of Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, 52074 Aachen, Germany c Jülich Aachen Research Alliance (JARA)–Translational Brain Medicine, 52428 Jülich, Germany d Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich, 52428 Jülich, Germany e Faculty of Health, Zuyd University, 6419 DJ Heerlen, The Netherlands
a r t i c l e
i n f o
Article history: Available online 10 July 2015 Keywords: Early reading development Sound–symbol learning Serial processing Rapid automatized naming Phonological awareness Predictors of reading
a b s t r a c t The current study examined the role of serial processing of newly learned sound–symbol associations in early reading acquisition. A computer-based sound–symbol paradigm (SSP) was administered to 243 children during their last year of kindergarten (T1), and their reading performance was assessed 1 year later in first grade (T2). Results showed that performance on the SSP measured before formal reading instruction was associated with later reading development. At T1, early readers performed significantly better than nonreaders in learning correspondences between sounds and symbols as well as in applying those correspondences in a serial manner. At T2, SSP performance measured at T1 was positively associated with reading performance. Importantly, serial application of newly learned correspondences at T1 explained unique variance in first-grade reading performance in nonreaders over and above other verbal predictors, including phonological awareness, verbal short-term memory, and rapid automatized naming. Consequently, the SSP provides a promising way to study aspects of reading in preliterate children. Ó 2015 Elsevier Inc. All rights reserved.
⇑ Corresponding author. Fax: +49 241 803380571. E-mail address: [email protected] (J. Horbach). http://dx.doi.org/10.1016/j.jecp.2015.06.007 0022-0965/Ó 2015 Elsevier Inc. All rights reserved.
J. Horbach et al. / Journal of Experimental Child Psychology 139 (2015) 256–264
257
Introduction To acquire reading, children must learn the code for representing speech as a series of visual symbols (Ziegler & Goswami, 2005). Multiple factors can influence the success or failure of reading acquisition, including genetic (Gilger, Pennington, & Defries, 1991; Pennington & Lefly, 2001), environmental (Bus, van IJzendoorn, & Pellegrini, 1995; Leseman & Jong, 1998), and cognitive factors (e.g., Hammill, 2004; Snowling & Hulme, 2008). Of the potential cognitive factors that might influence reading ability, there is evidence that phonological awareness (PA)—that is, the ability to identify, memorize, and manipulate speech sounds (e.g., Lonigan, Burgess, & Anthony, 2000; Vellutino, Fletcher, Snowling, & Scanlon, 2004)—and rapid naming of high-frequency items (RAN) (Hammill, 2004; Norton & Wolf, 2012) are important predictors of reading acquisition. Another factor associated with reading acquisition is verbal short-term memory (verbal STM) (Gathercole, Alloway, Willis, & Adams, 2006). During the initial phase of reading, a high short-term memory capacity seems to be critical in helping children to memorize speech sounds and merge them into whole words. To develop reading, the association of single sound–symbol pairs is necessary. Furthermore, decoding words requires the serial processing of the newly learned sound–symbol correspondences. Both PA and RAN have elements of these requirements. Good PA skills help to differentiate the sounds when assigning phonemes to graphemes, but tasks that have assessed PA do not require the ability to decode information in a serial manner. Tasks measuring RAN require the ability to process highly automated items in a serial manner but do not account for how learning to read requires the processing of newly learned sound–symbol pairs. Consequently, PA and RAN cannot fully explain reading acquisition. However, the ability to associate sound–symbol pairs, which mirrors the process of reading, is likely to be a critical predictor of reading acquisition. Existing studies examining the role of sound–symbol associations in reading have used paradigms that require the serial application of newly learned visual–verbal associations and correlated performance on these tasks with reading ability. For example, Aravena, Snellings, and Tijms (2013) developed an artificial orthography and showed that normal readers outperformed students with dyslexia when applying newly learned sound–symbol associations in a serial context. Interestingly, the two groups differed in their ability to apply knowledge of new letter names in a serial manner even though they did not differ in their knowledge of the new letter names themselves. Elbro, Daugaard, and Gellert (2012) tested students with dyslexia on a paradigm that also required them to learn new sound–symbol correspondences and apply them in a serial context. Participants needed to learn three new symbols and decode strings of them in a serial manner. The authors found that this dynamic measure of reading was able to differentiate between dyslexic and nondyslexic readers. However, it is still unknown whether the ability to learn new sound–symbol correspondences is predictive of reading acquisition. Both of these studies were cross-sectional and so could not assess potential causal associations between the ability to decode new sound–symbol correspondences serially and reading acquisition. Moreover, participants in both studies already had several years of reading experience (mean age of 9 years in Aravena et al., 2013, and adults in Elbro et al., 2012) when tested. Currently, to our knowledge no study has assessed whether the serial application of newly learned sound–symbol pairs is predictive of reading ability before children are formally instructed to learn how to read. Therefore, the current study examined whether performance in a simple sound–symbol paradigm (SSP) at kindergarten age is predictive of reading ability in first grade. To assess the specific predictive value of the SSP, we sought to examine unique variance associated with SSP controlling for other well-established predictors of reading ability, including PA, verbal STM, RAN, early literacy, and environmental factors. Method Sample and procedures A total of 268 kindergarten children (150 boys, average age at T1 = 6 years 2 months) took part. At the end of first grade (T2), 243 children were retested. All children were native German speakers
258
J. Horbach et al. / Journal of Experimental Child Psychology 139 (2015) 256–264
recruited from regular kindergartens and had an IQ equal to or greater than 85. The level of education of mothers and fathers was assessed via a questionnaire, coded as 1 (no/low educational achievement), 2 (medium educational achievement), or 3 (high educational achievement) and averaged. The categories of level of education were defined in accordance with the German school system (low:
