Children's Written and Oral 1 Running Head: Written and Oral Spelling Children's Written and Oral Spelling

For adults, written spelling is generally superior to oral spelling. To determine whether the same holds true for children in kindergarten through second grade, we compared children’s ability to spell real words (Experiment 1) and nonsense words (Experiment 2) orally and in writing. Building on the work of Tangel and Blachman (1992, 1995) and others, we developed a reliable system to assess the overall quality of the children’s spellings. We also examined the phonological and orthographic legality of the spellings. By first and second grade, written spellings were superior to oral spellings in both overall quality and representation of phonological form. This held true for words and nonwords. The results suggest that children, like adults, more accurately analyze the linguistic structure of a spoken item when they can represent the results in a lasting, visible form than when they cannot. Children’s written and oral 3 Historically, research on spelling has lagged behind research on reading. The gap has narrowed somewhat in recent years with studies that have described the course of spelling development in English and other languages (e.g., Perfetti, Rieben, & Fayol, 1997) and studies that have examined the effects of phonemic analysis instruction on spelling achievement (e.g., Tangel & Blachman, 1992, 1995). However, we still know less about how children learn to spell than about how they learn to read. The present study was designed to fill some of the gaps in our knowledge about spelling. We sought to provide information that would add to our theoretical understanding of the spelling process and, more practically, that could be used in the design and scoring of spelling tests for children. In most classroom spelling tests, and in all standardized spelling tests of which we are aware, children’s spellings are scored simply as correct or incorrect. However, research on emergent spelling (e.g., Henderson & Beers, 1980; Read, 1986; Treiman, 1993) has revealed large differences within young children's incorrect spellings. Stage theories of spelling development (e.g., Ehri, 1986; Gentry, 1982; Henderson & Beers, 1980) explain these differences by postulating that children go through a series of qualitatively different stages during the course of learning to spell. The earliest spellings bear no relationship to the sounds in the intended word, and have been called precommunicative (Gentry, 1982). Semiphonetic spellings represent some of the phonemes in the word (e.g., “l” for elephant). Phonetic spellings (e.g., “elefut” for elephant, “chran” for train) provide a more complete representation and may capture certain features of the pronunciation that are ignored in conventional English spelling (Ehri, 1986; Gentry, 1982). During the transitional (Gentry, 1982) or morphemic (Ehri, Children’s written and oral 4 1986) stage, children increasingly rely on visual and morphological information (e.g., spelling eighty as “eightee” instead of the phonetic “ate”). Other researchers question the concept of developmental stages, while acknowledging that existing stage theories may provide a rough overall picture of spelling development (e.g., Rittle-Johnson & Siegler, 1999; Treiman & Cassar, 1997b; Varnhagen, McCallum, & Burstow, 1997). Although the debate is still ongoing, it is clear that children’s misspellings reveal varying levels and types of knowledge. These differences are masked if spellings are scored simply as correct or incorrect. Several previous studies have attempted to develop real word spelling tests that can be scored so as to capture the sophistication of young children’s errors (e.g., Burns & Richgels, 1989; Liberman, Rubin, Duques, & Carlisle, 1985; Mann, Tobin, & Wilson, 1987; Morris & Perney, 1984; Richgels, 1986; Tangel & Blachman, 1992, 1995; Zutell, 1980). In these scoring systems, spellings that represent more of a word’s phonological and orthographic features receive higher ratings than spellings that bear little resemblance to the intended word, even when both spellings are unconventional. In the present study, we tried to refine these tests and scoring systems. In addition, we sought to develop similar tests of nonword spelling. Our word and nonword spelling tests, which together we call the T-BEST (Treiman-Bourassa Early Spelling Test), were designed for young children who are beginning to master the English writing system. A further goal of our study was to examine the cognitive requirements of different spelling tasks and how these influence children’s performance. In all standardized spelling tests of which we are aware, and in most classroom spelling assessments, children write their spellings on paper. Spelling can be tested in other ways, however, for Children’s written and oral 5 example by asking children to spell words aloud. We asked in the present experiments whether the output modality makes a difference. If children can reveal their knowledge more easily or more completely in one modality than another, this would have obvious implications for assessment. Modality-related differences, if found, might also shed light on the nature of the spelling process. There is evidence that adults spell more accurately when they are permitted to write words down than when they are asked to spell the words aloud. Anecdotal support for this claim comes from the observation that people sometimes trace a word in the air or on a table if a pencil is not available. More conclusively, Tenney (1980, Exp. 2) asked college students to choose between two alternative spellings for words such as nickel (“nickel” vs. “nickle”) and insistence (“insistence” vs. “insistance”). Students performed significantly better when they were told to write the word both ways on a sheet of paper before deciding than when they were told to think as long as they liked. In two experiments, Stadtlander (1996) found that college students were more accurate when spelling words on paper than when spelling them aloud. Only a few studies have compared children’s ability to spell words orally and in writing. Using a within-subjects design, Jorm and Schoknecht (1981, Exp. 2) found a superiority for written spelling over oral spelling in a group of fourth, fifth, and sixth graders. Children below the fourth grade were not included in the Jorm and Schoknecht study, begging the question of when the superiority for written spelling emerges. Turner and Quinn (1986) compared various methods of teaching children to spell made-up words such as proat. Children heard the spelling orally and then reproduced it either orally or on paper. Recall was assessed by either an oral or a written test. Three groups of Children’s written and oral 6 children participated: 7 8 year olds, 8 10 year olds, and 10 11 year olds. The children generally did better when the form of the final test (oral) matched the form of the initial presentation (oral). The 10 and 11 year olds did particularly well when they wrote the spellings on paper before they were tested. No such advantage was found for the younger groups. At least in this word learning task, it apparently takes some time for an advantage of written spelling to emerge. In the research reported here, we studied children in early elementary school -kindergartners, first graders, and second graders in Experiment 1, and kindergartners and first graders in Experiment 2 -and asked when a superiority for written spelling begins to emerge. In Experiment 1, each child spelled two lists of real words. One list was spelled in writing and the other list was spelled aloud, with the assignments of lists to conditions balanced across children. Experiment 2 used a similar design, except that the children spelled nonsense words rather than real words. In Jorm and Schoknecht (1981)’s comparison of written and oral spelling, as in many other spelling tests, children’s spellings were scored simply as correct or incorrect. We scored the real word spellings in this way, but we also looked at the spellings in more detail. Specifically, we developed scoring systems that were designed to capture (a) the overall sophistication of the spellings, (b) their phonological accuracy, and (c) their conformity to the orthographic structure of English. Each of these scales assesses a different aspect of spelling. When combined, they allow for a comprehensive examination of the sources of information that children use when they spell in writing and when they spell aloud. Children’s written and oral 7