Considering word characteristics for spelling accuracy: Evidence from Korean-speaking children Young-Suk Kim a, ⁎, Yaacov Petscher b a Florida State University and Florida Center for Reading Research, United States b Florida Center for Reading Research, United States abstract article info Article history: Received 14 November 2011 Received in revised form 19 July 2012 Accepted 7 August 2012 Keywords: Multilevel cross-classified model Orthographic transparency Spelling Word characteristics Item features We examined the extent to which word characteristics (i.e., differences in orthographic transparency among words) and child characteristics (i.e., emergent literacy skills) explain variation in children's spelling, using data from young Korean children (N =168). We compared predicted probabilities of various types of words (e.g., transparent vs. two types of opaque words) in spelling at various levels of child's latent ability in emergent literacy skills. While approximately 60% of total variance in spelling was attributable to differences due to child characteristics, a substantial amount, approximately 40%, was due to differences among words. In addition, inclu- sion of language and print-related emergent literacy skills reduced about 35 to 57% of variance that is attributable to differences among children. Inclusion of word types (i.e., transparent vs. opaque words) explained more than 50% of variance attributable due to differences among words. These results suggest that orthographic depth in words is one aspect to take into consideration for spelling acquisition. © 2012 Elsevier Inc. All rights reserved. 1. Introduction It has been shown that orthographic depth matters for reading acquisition. For instance, English which has a deep orthography takes a longer time to acquire literacy skills than languages with shallow orthographies (Seymour, Aro, & Erskine, 2003). Thus, orthographic depth (Katz & Frost, 1992) is an important factor for ease or difficulty of literacy acquisition (Frith, Wimmer, & Landerl, 1998; Seymour et al., 2003; Wimmer & Goswami, 1994). Previous studies examined cross-linguistic differences—how reading acquisition is influenced by orthographic depth across languages that vary in orthographic trans- parency. However, variation in orthographic depth occurs not only across languages, but also within a language. For instance, in English, many words can be sound out using grapheme–phoneme correspon- dence rules (e.g., cat, bike) while others cannot be (e.g., w in answer). Similarly, in a language with a shallow orthography such as Korean (the target language in the present study), some words are not ortho- graphically transparent (see more details below). The differences in orthographic regularity among words have drawn attention in reading theory (e.g., dual route model, Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001) and instructional practice (e.g., phonics approach for regular words and look-say approach for irregular words in English). Despite the theoretical and instructional attention, however, there are at least three gaps in the literature with regard to early literacy skills and orthographic consistency. First, the majority of existing studies examining word characteristics (i.e., orthographic transparency) were conducted with children learning to read and write in English (e.g., Baron, 1979; Ricketts, Nation, & Bishop, 2007; but see studies with French-speaking children such as Ziegler et al., 2008). Second, the majority of these previous studies with English-speaking children examined how individual differences in child characteristics (e.g., pho- nological awareness or letter knowledge), not word characteristics, are related to reading and spelling (e.g., Treiman & Bourassa, 2000). Even the few studies which examined whether word characteristics matter, researchers compared the extent to which child variables (e.g., phono- logical awareness) explain children's word reading by using word char- acteristics outcomes (e.g., orthographically transparent versus opaque words) in separate statistical models (e.g., Cho, McBride-Chang, & Park, 2008; Nation & Snowling, 2004; Ricketts et al., 2007). However, this approach does not allow us to accurately compare the extent to which variation in child variables versus word characteristics (i.e., orthographic transparency of words) explains variation in the outcomes because the outcomes are modeled separately in multiple regressions, which do not permit partitioning of variation due to child characteristics and word characteristics. Thus, the primary goal of the present study was to fill some of these gaps in the literature by investigating whether differences in ortho- graphic transparency in words explain accuracy (predicted probabili- ties) in children's spelling using data from children acquiring literacy in Korean. We examined the contributions of child and word character- istics (i.e., orthographic transparency of words) to spelling to find out whether word characteristics as well as emergent literacy skills matter in children's spelling performance, and if so, how word characteristics Learning and Individual Differences 23 (2013) 80–86 ⁎ Corresponding author at: Florida State University & Florida Center for Reading Research, G129, Stone Building, 1114 W. Call St., Florida State University, Tallahassee, FL 32306, United States. Tel.: +1 850 644 0370; fax: +1 850 644 9085. E-mail address: ykim@fcrr.org (Y.-S. Kim). 1041-6080/$ – see front matter © 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.lindif.2012.08.002 Contents lists available at SciVerse ScienceDirect Learning and Individual Differences journal homepage: www.elsevier.com/locate/lindif