The Development of Spelling

Patricia Scharer & Jerry Zutell. Handbook of Early Childhood Literacy. Editor: Nigel Hall, Joanne Larson, Jackie Marsh. Sage Publication. 2003.

In The Barnhart Dictionary of Etymology (Barnhart, 1988), the verb to spell is defined first as to ‘name the letters of. Along with the related noun form spell, meaning ‘words supposed to have magical powers, incantation, charm,’ to spell is related to an older root, ‘in part probably developed from Old English spellian to tell, declare, relate, speak’ (1988: 1044) and occurs in compound form in Gospel, the ‘good story’ of Christ’s life on earth. Spelling is clearly an important aspect of effective written communication, of our ability to tell our stories well. (Ironically, given the definition of the noun form, English spelling often also has the reputation of being mysterious and/or mystifying. Good spellers may be considered charmed in their ability to master the system, though the system itself is rarely thought of as charming!)

The perspective guiding this chapter is that spelling represents the development of the child’s gradual understanding and control over a complex system, universal in its general structure but also very specific to the child’s spoken and written language environment. Understanding how children go about figuring out the specific characteristics of their language system as readers and writers can provide insights into their sociocognitive processes, their literacy development, and the ways they learn over time in response to their instructional contexts. Exploring the relationships between reading, writing, and spelling is a significant part of literacy scholarship. Further, spelling can have a permanence that contrasts with the transience of the reading event. The speller reveals his understanding and mastery of the language’s writing system through both correct and approximate attempts. Thus, spelling has been perceived as an important indicator of literacy competence and a record of school performance that is available for both examination and criticism by parents and the general public as well as by educators.

This chapter will review research and scholarship about how young children up to the age of eight unravel some of the mysteries of how their language is represented in print. Specifically, we will deal with how words are represented through spelling systems and how students understand this relationship as manifested in their writing and spelling attempts and the change in those attempts over time. The chapter will be organized into five sections. The first section will provide an introductory discussion of how writing systems map spoken language to written forms. The second section will deal with how young children acquire basic and broad concepts about the relationship between written and spoken language. Then we will review research on learning to spell as a developmental process as well as discuss differing perspectives and critiques of stage theories. The fourth section will address significant issues including the debate surrounding invented (temporary) spellings and recent approaches to assessment and instruction. A concluding section offers future directions.

History and the Nature of Writing Systems

Before exploring how children gain control of their spelling system, it is important to review some basic principles about the nature of writing systems and how orthographies represent linguistic units in print. Coulmas asserts that ‘a conventional relation between graphical sign and linguistic unit is crucial for writing’ (1989: 27). He defines the process by which graphical signs with concrete references came to be associated with linguistic signs for those objects as ‘phonetization’ (1989: 26). Once the primary value of the written sign became a sound it was transformed from icon to symbol. Young children must come to understand this essential characteristic.

Each language uses a particular system for mapping pronunciations of significant units to a system of written symbols. In theory, writing systems can be divided into two general classes: (1) those whose symbols represent units of sound, which combine to form pronunciations of linguistic units, and (2) those whose symbols cue meanings, which are then mapped to particular pronunciations. The first class can be further divided into alphabetic systems, which map symbols to sounds at the phonemic level, and syllabic systems in which each symbol represents a spoken syllable. Examples of alphabetic systems include most Western European systems, e.g. English, French, Spanish. Korean Han’gul is an often-cited example of a syllabic system, and Chinese of a meaning-based system. Alphabetic systems have the advantage of using a small number of symbols to represent a very large number of pronunciations, but at the cost of a high level of abstractness. The Chinese system is said to be less abstract, but requires the learning of a large number of distinct characters. Syllabic systems fall between these extremes.

In reality, well-developed systems tend to be multilayered in representing relationships, with trade-offs between the demands of sound, visual pattern, meaning, and historical influences. Letter combinations represent meanings as well as sounds, and spellings sometimes preserve meanings when variations in pronunciations are predictable from the phonological context. For example, the English morpheme s, meaning plural, is spelled the same in cats and ribs though the pronunciation is /s/ in the first word and /z/ in the second. A further consideration is that the development of most writing systems has been greatly affected by historical, social, and political factors along with linguistic ones. Once writing systems were initially established for a language, both the oral language and the written language evolved over time somewhat independently but also in connection with each other. These developmental changes have had a direct impact on the relationship between spellings and pronunciations. In some cases, a relatively slow rate of orthographic change (once print had been relatively standardized) has interacted with a greater change in pronunciation over time and place to add considerable complexity to the relationship between spelling and sound.

English is a particularly noteworthy example in this respect. (See Henderson, 1990, and Venezky, 1999, for fuller historical treatments.) While English began as a Germanic language, and is typically categorized as such, the Norman Conquest led to the eventual blending of Anglo-Saxon and Norman French into a new language. Many new spelling conventions were adopted from French as well; other changes are believed to have been made by scribes in order to make handwriting clearer, sometimes at the expense of phonological accuracy. For example, the replacement of u by o in ton and woman was supposedly motivated by the desire to break the confusion of a succession of a large number of vertical strokes (Venezky, 1999). As the need to represent new ideas increased, Latin and Greek elements (prefixes, roots, and suffixes) were used to build new word forms. The sound-letter relationships in such multisyllabic words are further complicated by the English pattern of distributing stress unequally across syllables depending on the number of syllables per word, as in the following set of words from the same root: oppose, opposite, and opposition. In such cases, spelling tends to preserve and make visible connections that may not be as apparent in pronunciation.

During the Age of Discovery, contact with new peoples and other European languages in new environments led to a large number of adoptions and adaptations of foreign words that did not always correspond to typical English letter-sound relationships. Contrast, for example, the borrowings of vanilla and tortilla from Spanish. Note, too, the various pronunciations of ch in child (Anglo-Saxon), chef (French) and chorus (Greek) but the /ch/ sound in cello (Italian). The spread of English-speaking peoples to new parts of the word and the military, political, and commercial prominence of British and later American institutions and culture have led to the development of varieties of English across the globe.

The result is a set of orthographies in which contentions between phonetic, visual, semantic, and etymological demands are resolved at the expense of simple, straightforward, sound-letter relationships (Cummings, 1988). It would seem that the demands of such a system on a beginning learner would be higher than on those learning one with simpler, more phonetically transparent and regular relationships.

Similarly, though the Chinese system has often been simplistically described as logographic, the vast majority of Chinese characters are compounds with both meaning and phonetic elements that combine to specify a distinct morphemic-syllabic unit. A compound Chinese character consists of two elements: a classifier that conveys something about the meaning of the overall character, and a phonetic element that provides information about its pronunciation. Coulmas gives the example of the phonetic element, pronounced tang, which, when combined with the classifier for water, yields the compound character meaning ‘pond’ (1989: 101).

However, an important difference between Chinese and alphabetic systems is that in Chinese the phonetic and meaning elements occur in separate parts of the character. In alphabetic and syllabic systems the meaning and pronunciation are conveyed simultaneously by the combination of symbols. The English spelling m-a-n cannot be broken into distinct parts representing the pronunciation on the one hand, and the meaning (adult masculine human) on the other. Further, the phonetic element in the Chinese character cannot be parsed into subunits that map to individual phonemes in the character’s pronunciation, which is the basis of alphabetic systems. Thus, while both kinds of systems provide information about sound and sense, they do so in very different ways. Some scholars hypothesize that such differences in how information is presented may lead to differences in how it is processed. For example, preliminary findings suggest that Chinese readers attend to phonetic information later during word reading processes than do readers of alphabetic scripts (Ju and Jackson, 1995).

How Children Acquire Early Written Language Understandings

The brief analysis above points to the complexity of the task children face as they learn about their writing system and its unique characteristics. Researchers studying preschool children for the past 30 years have sought to document, describe, and analyse the processes through which children become conventional spellers. The distinction made by Coulmas (1989) between three tiers (writing system, script, and orthography) is a useful way to consider the levels of understanding required for children to become conventional spellers. At one level, learners must grasp the general principles underlying a writing system as it relates to oral language (e.g. consistency in directionality). At another, they must become familiar with the features of the specific kind of script (e.g. letters or characters) that accompanies the language they speak. In addition, they must master the specific rules and patterns of that orthography, in terms both of internal visual rules and of how pronunciations and meanings of oral linguistic units are mapped to particular written forms. To illustrate, though English has 26 letters in its alphabet, not all letters are used in the same way orthographically; for example, English words never start with ck or end with a q. Furthermore, English in the US and English in the UK have varying spelling patterns (such as theater/theatre or honor/honour).

Given these layers of understandings, it is no small task for children to learn the complexities of their writing system and its accompanying script and orthography. This includes learning many aspects of how print ‘works’ in their language. Children must learn that written symbols are different from objects or pictures, that there are specific features of their script, that the writing system has a consistency, and also that there are complex characteristics of the writing system such as directionality and concept of word.

In this section we argue that although these aspects are often individually described in developmental ways based upon studies of children of various ages, there is also reciprocity between these understandings such that what children learn about one may support the development of a more complex understanding of another. Sulzby writes that ‘each child acquires the abilities to read and to write within a culture in which both oral and written language are being acquired simultaneously, and that the two together comprise language ’ (1996: 28). Goswami (1992) argues that children’s emerging knowledges of phonological and orthographic concepts interact, each refining the other as the child’s understanding becomes more complete. Because of the literate nature of our society, children experience both print and oral language simultaneously as they begin to understand how language carries messages and also the specific characteristics and organization of the script that they see around them.

Thus, learning to spell must be discussed within the context of learning not merely the relationship between symbols and sounds but also how symbols map on to language in a manner that begins with approximations and moves to conventions. Goswami’s (1992; 1994; 2001) work on children’s phonological awareness focuses on the relationships between learning about print and the oral aspects of language including children’s abilities to recognize, categorize, and manipulate spoken language in the absence of print. This work emphasizes that children’s phonological awareness, as well as their understandings of the match between written and spoken forms, is in the process of development. Students’ abilities to represent spoken language will be affected by their current phonological understandings. For example, in a recent review of the research on early phonological development, Goswami (2001) suggests that there is a developmental progression from phonological awareness of larger to smaller units. Her findings suggest that (1) syllables are natural units of analysis for English speakers; (2) onsets and rimes are particularly salient for young learners as their phonology becomes more segmented; (3) children are able to use onset and rime as the basis for analogy at a young age; (4) phonological awareness of onset and rime predicts later success in reading and spelling; and (5) phonemic awareness develops through instruction in alphabetic orthography. At the beginning stages of language development, phonological processing is, for the most part, holistic. As vocabularies expand

there is considerable developmental pressure to represent these words in the brain in a way that will distinguish them from other words and allow the child to recognize them accurately and quickly during speech comprehension To distinguish between these similarly sounding words both quickly and accurately, child linguists argue that children must begin to represent the sequences of sounds that constitute each known word in their brains. They must represent the ‘segmental phonology’ of the words they know. (2001: 113)

From another perspective, children must understand the concepts relative to wordness in both oral and written contexts. The concept of word operates primarily in literate societies and, in some ways, is defined by the writing system (Coulmas, 1989). For adults, the definition of a spoken word may seem simple; but, for children, it is an understanding that takes many years of experiences with oral and written language to refine. Papandropoulou and Sinclair (1974) approached their study of young children (ages four to 10) with a Piagetian lens and found that the youngest children did not differentiate between words and things; they typically described words as either objects or actions (strawberry is a word because it grows in the garden or pencil is a word because it writes). For these children, the length of the word was relative to the size or location of the object (train is a long word because it goes and goes). Later, between the ages of five and seven, children described words as ‘what you use to say about something’ (1974: 244) but limited this definition to comments (proposing full sentences as words) or labels of objects, explaining that the is not a word because you need something else like the truck to be a word. By the ages of six to eight, however, words are seen as part of a larger more meaningful expression (bits of a story). This development of a concept of word is both gradual and complex: ‘Gradually words become detached from the objects and events they refer to, and it is only fairly late in cognitive development that they are regarded as meaningful elements inside a systematic frame of linguistic representation’ (1974: 249).

Clay also describes children’s simultaneous learning across various aspects of written language, arguing that ‘The individual child’s progress in mastering the complexity of the writing system seems to involve letters, words, and word groups all at the one time, at first in approximate, specific and what seem to be primitive ways and later with considerable skill’ (1975: 19). Children begin to learn the general characteristics of the script accompanying their writing system as they begin to draw, scribble, or write mock letters loosely reflecting conventional print. For Portecorvo and Orsolini, the distinction between drawing and writing is the first phase of writing development followed by a time when ‘children explore the graphic and syntactical regularities of the notation system’ (1996: 15). In a manner that parallels the drawings of the cave dwellers, pictures carry the meanings of young children who, when asked to look at the story, may attend only to the picture (Clay, 1975). Pictures are broad and interpretive; but the symbols children must learn representing spoken language units are narrow and specific.

Harste et al. (1982) asked four-year-old children from a variety of international backgrounds to write everything they could write. Children’s writing samples were clearly influenced by their environments, as the English-speaking child wrote in cursive-like scribbles; the Israeli child’s sample looked much like the Hebrew alphabet; and the Arab child pointed out that the researchers wouldn’t be able to read her sample since Arabic has more dots than English. None of these children wrote a recognizable word or could read their message, yet their samples reflected an initial understanding of the nature of three different orthographies. Harste and his colleagues argued that these children were creating hypotheses about how written language works that were constantly being challenged and revised as they learned more and more about ‘how the grapho-phonemic, syntactic, and semantic systems of language operate in relation to one another and in relation to those things known about the world’ (1982: 65).

As children learn the specific visual qualities of the system, their attempts reveal an element of experimentation as they explore the various ways to make letter forms. They often repeat pictures, letterlike shapes, or individual words to make longer messages, a phenomenon that Clay (1975) calls the ‘recurring principle.’ Children often learn the letters of their names first; these letters are a key feature of later writings. According to Ferreiro and Teberosky, a child’s name is tremendously important and ‘in many cases, the child’s own name functions as the first stable form endowed with meaning’ (1982: 213). It is this meaning that anchors the child’s understanding that a specific written form consistently represents a particular meaning, a concept critical to further development. As children gain control over their name, they may repeat one or more letters of their name reorganized into a new pattern, employing what Clay calls the ‘generating principle’ to create messages with more complex letter arrangements. A longitudinal case study by Martens (1999) documented the importance of learning to write her name in one child’s literacy development from ages two to five. Similarly, extensive use of letters found in children’s names was documented in Bloodgood’s (1999) research, as data analysis revealed that the letters in children’s names accounted for nearly half of the random letter writing done by four-and five-year-old children as they used the same letters over and over to write their messages.

As children become more familiar with letter forms and their names, they also begin to develop hypotheses about how these forms are linked to sounds in the stream of speech, to oral words. Kamii and Manning (1999) reviewed findings in this area, beginning with Ferreiro and her colleagues’ delineation of four levels of writing revealed when analysing children’s spellings in Spanish: (1) letter strings, made up of similar letters, but of different lengths (see discussion of Clay’s work, above); (2) letter strings with a fairly fixed range of length, with more letter-like forms included and some differentiation of letters used and/or their order to indicate word differences; (3) use of one character per syllable, but usually no phonetic connection between the letter and the sounds in the syllable (though, within this level, some children begin to use a vowel for each syllable); and (4) evidence of considerable knowledge of grapho-phono correspondences. Kamii et al. (1990) found that English-speaking children generally followed the same levels as the Spanish-speaking children in the Ferreiro studies, with some important differences. At the third level, Spanish-speaking children focus on syllables, using a vowel for each; English-speaking children at this level focus on consonants. They suggest that these differences are likely due to differences in the phonologies of the two languages. In a follow-up study, Kamii and Manning (1999) asked students to write related word pairs in which one word had more syllables than the other (e.g. ham/hamster). They found that some students at the second level began to differentiate between the two words by using longer strings for the words with more syllables, though the letters might be totally different. Other students within this level used the same letters for similar parts in the words, though still using letter strings without grapho-phono correspondences. So, while children’s writings may appear somewhat random to adults, closer analysis reveals a developing system of organization and relationships even at this early stage.

While children are developing a conceptual understanding about spoken words as abstract or arbitrary labels distinct from the objects themselves, basic regularities inherent in print, and general features particular to their own scripts, they are also beginning to make links between oral and written words in more systematic ways. Morris (1993) considers the ability to match spoken words in reading with written words in text as a crucial event in learning to read and write. In his year-long study of emergent readers in three kindergarten classrooms, he documented the pivotal nature of this critical element and concluded that beginning consonant knowledge facilitates children’s matching of oral and written words, which then further facilitates phonemic segmentation, a skill supporting word recognition. Uhry’s (1999) results were consistent with these findings. Hughes and Searle (1991; 1997) also found that to ‘develop as fully phonemic spellers, the children had to establish a stable voice-print match, demonstrated by pointing accurately to words as they read memorized text’ (1991: 167).

Such connections between reading and spelling reflect fairly recent lines of research. For most of the twentieth century, scholars and educators focused on the differences in spelling and reading rather than the connections between them. This focus was reflected in the clear separation between reading/phonics instruction and spelling practice, often done at very different parts of day with separate, unconnected instructional materials used for each (Read and Hodges, 1982). More recently, research and scholarship have focused on the important similarities and connections between the word processing required for reading and spelling. Frith (1985) proposed a model in which ‘phonemic awareness in reading develops as a consequence of spelling experience’ (Goswami, 1994: 292). In fact, Morris and Perney (1984) found that a developmental spelling measure administered at the beginning and middle of first grade (ages six to seven years) was a strong predictor of sight word acquisition by the end of the year. This finding is not surprising in light of the fact that a considerable body of research has now demonstrated strong positive relationships between phonemic awareness, spelling, and success in learning to read (e.g. Ehri, 1987; Griffith, 1991; Tangel and Blachman, 1992).

Developmentally sensitive measures, like the one Morris used, capture changes in phoneme awareness as children move from pre-phonetic to semi-phonetic to letter name spellings and beyond (Richgels, 2001). Based on findings from a later study, Morris has proposed an interactive, sequential process of early literacy learning in which knowledge of beginning consonants (as demonstrated by the child’s early spelling attempts) provides a textual anchor for examining word form in greater detail. As the ability to match spoken and oral work accurately stabilizes, supportive text provides the opportunity for matching word pronunciations to letter sequences. Phonemic awareness is thus extended and is reflected in more sophisticated invented spellings. Words are initially learned and successfully identified mostly in supportive contexts with partial letter cues (in the pattern of first; first and last; first, last, and middle). As the specifics of the alphabetic principles are internalized, words are more fully processed, and a stable and expanding sight vocabulary gradually emerges (Morris, 1993).

Several other studies suggest that connections between reading and spelling remain strong beyond beginning stages. In a study of children’s ability to notice and report letters in reading specially taught words, Invernizzi (1992) found that the ability to recall the presence or absence of specific letters was clearly related to stage of spelling development and dependent on the complexity of the feature to which the letter belonged. For example, spellers at within-word pattern and beyond were very good at recalling the presence of e-markers, while letter name spellers only performed at chance on this feature. Bear (1992) compared first-graders’ developmental spellings to the fluency of their oral readings and found that spelling measures could account for three-fifths of the variance in reading rate. Letter name spellings were significantly and negatively correlated with fluency, while performance on selected within-word pattern features was strongly associated with fluency measures. In both cross-sectional and longitudinal studies, Zutell (Zutell, 1992; Zutell and Fresch, 1991; Zutell and Rasinski, 1989) found that reading and developmental spelling measures were highly correlated for third-and fifth-grade students, with factor analysis strongly supporting a single factor solution which Zutell labelled underlying word knowledge.

A significant finding of these lines of research is that children’s written productions that may appear random in nature actually reveal complex thinking on several dimensions, indicating that children’s initial, inchoate understandings are moving toward increasingly refined hypotheses about orthography even before they make clear letter-sound connections. Such studies begin to describe the complex and inventive nature of the development of print concepts that happen simultaneously, sequentially, and reciprocally as young children become increasingly proficient with reading and writing tasks.

How Children Acquire Proficiency as Spellers: Spelling Stage Theories

Current understandings of spelling development, once children grasp the alphabetic principle, are highly indebted to the seminal work of Charles Read (1971; 1975; 1986) and are heavily grounded in his findings and insights. Read’s initial study focused on the spelling attempts of preschool children. He discovered that children from different settings and without formal instruction often produced similar misspellings. Most remarkably, his analysis demonstrated that such errors were neither random nor the result of auditory/phonological immaturity or deficit. Many were, in fact, quite logical, given the knowledge base the children were operating with (knowledge of the names of the letters, but little sight word knowledge) and the complex way English spelling maps phonological relationships to print. For example, Read pointed out that the spelling of /dr/ blends with initial J or G and the spelling of /tr/ blends with initial H or CH was due not to ‘mishearing’ or ‘mispronunciation’ but to the ambiguity created by the fact that the pronunciation of the first element in each is affricated in the environment of the /r/; that is, it shares a feature of articulation with the related sound and the letters associated with it. The first phoneme in dragdoes sound like and is articulated like the first phoneme in jet. In similar fashion, Read found that children often misspell short vowel sounds in particular ways (for example, E for ‘short i,’ as in SET for sit, A for ‘short e’ as in BAG for beg, I for ‘short o’ as in MIP for mop). Due to vowel shift, a major historical change in the pronunciation of English long vowels, short vowels and long vowels are not paired in spelling as they are in sound. Read argued that children were using a letter name strategy; that is, they were matching the phonemes heard in words with letters whose names also include those phonemes. Since short vowels are not used as the names of letters in English, children classified those sounds as similar to the closest long vowel letter name. The vowel sound in begis more similar to the name of the letter A than it is to the name of the letter E. Thus Read argued that children were not memorizing spellings but constructing or inventing their own plausible system based on abstract phonological relationships. They were recognizing relationships that adults had learned to ignore in the course of becoming literate (Read, 1971; 1975).

While Read’s initial study was conducted with a limited number of children, the importance of Read’s observations and analyses cannot be overestimated. His later work and that of other spelling scholars confirmed and extended those early findings. As a result of his work, the study of children’s spellings gained in respect and attention. Researchers applied his techniques and insights in case studies of young children (e.g. Bissex, 1980; Gentry, 1982) and in studies of children in the early grades of formal instruction, generally confirming, often refining and elaborating on our understandings of children’s strategies (Beers and Henderson, 1977; Treiman, 1993).

The late Edmund Henderson had been collecting and categorizing young children’s spellings over several years before Read’s work was published. He recognized Read’s work as the ‘Rosetta Stone’ of spelling research in that it provided the key to deciphering and understanding the mental processes underlying children’s attempts (Henderson and Beers, 1980). Over the last 30 years Henderson, his students and colleagues, and their students and colleagues have pursued a line of research focusing on the developmental and conceptual nature of children’s understandings and learning, extending their investigations on the one hand to younger and/or less knowledgeable children, not yet fully using the letter name strategies that Read described, and on the other to mature students grappling with the complexities of the derivational and etymological aspects of English orthographies (Henderson and Beers, 1980; Morris, 1989a; 1989b; Templeton and Bear, 1992). They proposed a stage-like quality to spelling development in which attempts change over time, both in terms of how words are misspelled and in terms of which features (e.g. consonant blends, long vowel markers) and word types are spelled correctly. Patterns of development are affected by the child’s knowledge base, strategies, and the complexity and familiarity of the features and words under examination (Gentry, 1978; Henderson, 1990; Schlagal, 1992).

Henderson (1990) suggested the following early levels of development. The first is pre-phonetic, characterized by letter strings that represent concepts and ideas, but without a discernible match between letters and sounds (see above for a closer analysis of patterns of attempts at this stage). The second is semi-phonetic, characterized in the early part by the use of single consonants to represent beginning or particularly salient sounds in the word. Over time children include letters for both beginning and ending sounds, with long vowels generally represented before short ones. The third is letter name (now often labelled phonetic or alphabetic), characterized by the ability to consistently match first-to-last in spoken words with left-to-right in the written forms, generally representing each phoneme in a word regularly with the logic Read described. It is at this stage that students begin to develop a working sight vocabulary for reading that also supports developing knowledge of and hypotheses about visual and meaning-based features (Ehri, 1992; Morris, 1992). The fourth is within-word pattern (also called transitional), characterized by correct spelling of well-known words and single consonants and many blend and consonant digraph patterns, and by good control of short vowel patterns. At this stage children begin to include visual features like vowel markers and other silent letter patterns into their spellings, though they may only gradually do so with accuracy on a regular basis. They are also more likely to use analogy strategies as they are able to process letters in ‘chunks’ rather than individually. Because of the particular complexity of English visual patterns and their relationships to sound and meaning, full control of these elements may require several years of experience and study (see also Goswami, 1994; 1999).

Henderson also suggested two higher-level stages: syllable juncture, in which students grapple with patterns for combining syllables, including consonant doubling and e-drop patterns; and derivational constancy, in which students deal with the morphological patterns in English multisyllabic words whose pronunciations often obscure the links in meaning and spelling among words in the same root-word families (e.g. cave, cavity, excavation). Recently the distinct and sequential nature of these stages has been called into question (Ehri, 1992: Gentry, 2000), and current descriptions view these as aspects or phases of later development that may parallel and/or overlap with each other and the later part of the within-word pattern stage (Bear et al., 2000). Other scholars have made significant contributions paralleling and complementing this line of research. One area that has received considerable recent attention is children’s understanding of morphological markers and their complex representation in oral and written form. As noted above, English orthography tends to preserve the same spelling in such markers when pronunciation changes predictably in different phonological environments. For example, the spellings of -s, -es for plural and third-person singular remain stable when the phonological environment requires /s/,/z/ or /^z/ in pronunciation (e.g. desks, swims, dresses). Studies have generally found that young spellers (six to eight years old) become aware of and gain control over this feature relatively early (Beers and Beers, 1992; Read, 1975). In a similar vein, Treiman (e.g. Treiman and Cassar, 1996) has found that in highly focused conditions young children show sensitivity to morphology in their spellings of identical sounds depending on their relationships to morphemic units. For example, nasal omission, a misspelling Read had initially noted, was more likely to occur in brand (one morpheme) than in tuned (two morphemes with the second phoneme in the final blend functioning as a morpheme).

Bryant et al. (1999) report a detailed three-year study of the acquisition of the past tense morpheme with children who began the study at ages six, seven, and eight. In English, the -ed spelling is maintained in regular verbs whether the pronunciation is /t/, /d/, or /^d/ (e.g. skipped, trimmed, wanted). On the other hand, ‘strong’ or irregular English verbs tend to vary the vowel while maintaining the direct phoneme-grapheme match for the past tense morpheme (e.g. sleep, slept; find, found). The situation is further complicated by the fact that the same final consonant blend pronunciations exist in words unrelated to past tense (soft, blind). Bryant et al.’s findings support a stage model that includes; (1) a pre-phonetic stage in which endings are not represented in a consistent way; (2) a phonetic stage in which all endings are spelled phonetically, including those for regular past tense verbs; (3) a stage of generalizations and overgeneralizations in which the -ed ending is applied to both strong verbs (SLEPED for slept) and single morpheme words (SOFED for soft); (4) a generalizations only stage in which strong verbs may be spelled with -ed, but single morpheme words are spelled conventionally; and (5) a correct stage in which strong verb endings are again spelled phonetically. They also found that children’s movement through the stages is strongly related to their specific sensitivity to grammatical distinctions, as measured by analogy tasks like those developed by Berko-Gleason (1958). As with other studies, these results support a model of learning to spell that goes beyond memorization to include constructive processes in which children use their growing linguistic knowledge to generate hypotheses and refine them over time as their base of knowledge and experience expands.

Criticisms of spelling stage theory, particularly as proposed by Henderson and his colleagues, tend to focus on details of features of a particular stage, labels for the stages, and/or boundaries between stages (e.g. Ehri, 1992; Treiman, 1993). Consequently, researchers continue to explore and refine these concepts. Others have raised questions about the inviolability of the sequential nature of stages and their ability to account for the full range of individual behaviours. On the one hand, the sequence of stages for young children (through within word pattern) have been fairly well established. On the other, Henderson’s stage model and that of his colleagues does not assume to explain every spelling that a student might generate. Individual spellings in a given stage will naturally be affected by reading experiences, word familiarity, the spelling situation (for example, writing a story or making a list), instruction, human factors (e.g. weariness, anxiety, inattention) or other situational or individual differences. Rather, the theory suggests broad steps in development justified by the overall character of children’s performance.

The knowledge base, concepts, and techniques developed in the spirit of Read’s work and developmental perspectives have been extended in two other noteworthy ways. First, differences between normal functioning and low-ability readers/spellers have been re-examined in light of these understandings. Boder (1973) had suggested three categories: dysphonetic, dyseidetic, and mixed spellers. But reanalysis of her findings in light of Read’s descriptions of logical spellings called into question the true number of dysphonetic spellings in her data (Henderson, 1992). Further, with the growing recognition of the importance of instructional level in spelling assessment and instruction (Morris et al., 1986; 1995; Schlagal, 1992), recent studies have compared low-functioning spellers at their instructional level with younger spellers at the same level of ability rather than with average spellers in the same grade. Under these conditions, differences between groups are greatly reduced in comparison to earlier studies (Invernizzi and Worthy, 1989). This finding suggests that delay in learning and a need for more processing time (Abouzeid, 1992), rather than differences in processing, is a more viable explanation of disabled spellers’ difficulties. (However, Alegria and Mousty, 1997, contend that there is some evidence to support subtle visual deficits in disabled readers for which the crucial variables have not yet been fully identified.)

Secondly, there is a small but growing body of research on spelling development in languages and scripts other than English. (Though, admittedly, this literature is likely more extensive, as we are limited to reports published in English.) Results of investigations of children’s spelling patterns for Spanish (Temple, 1978; Valle-Arroyo, 1990), French (Gill, 1980), Portuguese (Pinheiro, 1995) and Greek (Porpodas, 1989) suggest a movement away from strictly phonetic strategies to increasing use of strategies based on orthographic and morphographemic knowledge. Investigations of children’s spellings in Portuguese (Nunes-Carraher, 1985, as reported by Bryant et al., 1999) and in French (Fayol et al., 1999) have found patterns for the development of morphological markers parallel to those found by Bryant et al. (1999) discussed above.

Shen and Bear (2000) examined spelling errors of Chinese students in grades one through six, both in spontaneous writing and on a specially constructed list. They found a gradual increase in graphemically and semantically based errors and a decrease in phonologically based errors as grade level increased, with young children using mostly phonological strategies. However, this finding is somewhat confounded by the fact that in the early grades students are taught Pinyin, a phonetically based system for representing Chinese pronunciations. The overwhelming phonological strategy used in the early grades was to substitute Pinyin symbols for characters. This speaks to the complex relationships between social factors and writing systems and their impact on children’s behaviours. Interestingly enough, Shen and Bear note the presence of partial phonetic spellings in the use of Pinyin in the early grades, a phenomenon they believe warrants further investigation.

On the other hand, some studies do suggest clear contrasts with English. As noted above, Ferreiro and Teberosky (1982) found Spanish-speaking children represented vowel elements in their spellings while Kamii et al. (1990) found English-speaking children at about the same stage represent consonants. They attributed this to differences in the phonologies of the language. Yet Wimmer and Landerl (1997) found that first-grade German-speaking children and dyslexic children with two years of instruction did not have the same difficulty with consonant clusters that has often been reported for English-speaking children, although the languages seem to be comparable in this respect. They conclude that the overall consistency of an orthography (German being more consistent than English), combined with an instructional regime focusing on segmentation, may have a profound impact on spelling development, and they caution that teachers and researchers working in more consistent orthographies should not base their theories and choices solely on English findings. It seems, then, that studies of developmental spelling in multiple languages have identified some similar patterns across languages but also unique qualities shaped by the characteristics of each language and, possibly, the circumstances under which they are taught and learned.

To summarize, the contributions of developmental spelling studies and related research are considerable. Such studies have focused attention on the quality of children’s spellings, provided teachers and researchers with tools for making sense of student behaviours, and increased our understanding of the conceptual nature of learning to spell in the face of the traditional focus on memorization. In addition, such research has generated approaches to assessment and instruction that respect and build on student knowledge and strategies.

Spelling Instruction: From Memorization to Conceptualization

Early instructional methods for alphabetic systems, dating back to Greco-Roman times, emphasized memorization of letters that were then applied to the spelling and pronunciation of words. This technique was commonly called the ABCDery method (Otto, 1973). Early American spelling instruction emphasized lengthy recitations and memorization; spelling research focused on the identification of practice and study techniques for learning letter combinations and words. This perspective was motivated by a characterization of English spelling as highly irregular and by a behaviourist, stimulus-response view of the learning process (Templeton and Morris, 2000).

At the beginning of the twentieth century, however, research shifted to analysing the characteristics of words relative to their spelling difficulty, resulting in ‘demon’ lists of challenging words. Teachers were encouraged to help students learn these words by visualizing (Lee and Lee, 1941) and taking tests leading to mastery. Research focused on developing frequency counts of words in reading and writing (high-frequency words as deserving the most attention), tracking student errors in terms of control of letter-sound matches, and discovering effective practice techniques and time plans to enhance memorization. High-frequency words were taught by having children copy sentences with words commonly used in children’s writing and complete dictation exercises in hopes that children would become accustomed to spelling correctly (Scharer, 1992).

By the middle of the twentieth century, grade-level consumable spelling books were a common mode of formal spelling instruction requiring that young children cycle through a different set of words each week by completing various workbook exercises culminating in a final test on Friday. These materials, however, were criticized because students were often required to study words they already knew or were too difficult for them to learn developmentally, and the activities did not help children to learn orthographic patterns or to transfer spelling lessons to their writing (Gill and Scharer, 1996; Zutell, 1994).

More recently, two lines of research have significantly influenced early spelling instruction. First, research has begun to document a positive correlation between a young child’s ability to perform tasks of phonological awareness (an awareness of aspects of spoken language such as syllables, rhymes, and individual phonemes), phonemic awareness (a conscious awareness that words are made up of phonemes that can be both isolated and manipulated), and later achievement in both reading and spelling (Adams, 1990; Ball and Blachman, 1991; Ehri, 1980; Goswami, 1999; Griffith, 1991; National Reading Panel, 2000; Richgels, 2001). For example, as reported in Goswami (1999), Bradley and Bryant (1983) studied the impact of phonological training on children (ages four to five) with poor phonological awareness. Two years of training for the experimental group focused on onset and rime through picture sorts of rhyming words and manipulation of plastic alphabet letters to create new rhyming words. One control group was taught semantic classification; the other was unseen. After training, children in the experimental group were eight months ahead of the semantic classification group in reading and a full year ahead in spelling; experimental scores were one year ahead of the unseen control group in reading and two years ahead in spelling.

Further, based upon their meta-analysis, the US National Reading Panel (2000) concluded that phonemic awareness training interventions of 20 hours or less supported kindergarten children as spellers. But, while promising, this analysis was unable to determine specific teaching techniques that were most effective and engaging for both teachers and students or to document positive effects for disabled spellers. Recommendations regarding instruction are further complicated by studies documenting the reciprocity between the development of phonological and phonemic awareness and learning to read and write (Adams, 1990; Goswami, 2001; Silva and Alves-Martins, 2002). Silva and Alves-Martins (2002) found that either two weeks of phonological awareness training or guided discussions of children’s invented spellings enabled young children to move from presyllabic writing to syllabic writing.

While specific training interventions appear successful, McGee and Richgels write with concern that recommendations for phonological or phonemic awareness training in classroom contexts may result in scripted programmes ‘so divorced from actually reading and writing of authentic texts for real purposes as to be counterproductive for those students who already have phonemic awareness, or are on their way to acquiring it in other, more functional and contextualized ways’ (2000: 212). Consequently, educators are challenged to determine an appropriate curricular sequence, emphasis, and instructional balance between attention to phonological awareness, phonemic awareness, reading instruction, and writing instruction to support children’s literacy development.

A second influential line of research is the three decades of developmental spelling studies, which have also led to innovative ideas about instruction. The issue of correctness in writing, for example, takes on a new perspective as invented spellings are viewed as windows into a child’s thought processes, and assessment targets children’s developmental stages in ways that can effectively influence instruction (Richgels, 2001). Looking at each child developmentally calls into question practices of having the same grade-level curriculum for every child in the class in favour of providing instruction based upon the needs of groups and individuals. In a year-long study using spelling materials matching students’ developmental levels, Morris and his colleagues (1995) found that low-achieving students using materials at the appropriate level scored higher then low spellers taught with more difficult materials. Thus, developmental research calls for teachers to identify students’ appropriate levels and plan instruction to meet those needs. This instruction places new demands upon teachers’ expertise when compared to whole class instruction using published spelling programmes (basals). Both of these issues—invented spelling and meeting the needs of individuals as spellers—are addressed below.

Invented Spelling

Although issues relative to invented spelling often focus on the early years of school, spellers of all ages use invented spelling any time they approximate an unknown word. For young children 50 years ago, writing meant copying sentences and independent writing was not encouraged until children’s conventional spellings matched the messages they wrote to ensure accuracy. More recently, however, instructional recommendations have focused on encouraging young children to create and write messages using the knowledge they have of the system, no matter how incomplete. Consequently, preschool and kindergarten children find writing materials in play centres (Morrow, 1990), opportunities for journal writing, and teachers who encourage and celebrate their early writing attempts.

Concerns that children will not learn conventional spellings if allowed to write words unconventionally have been countered with arguments describing the advantages of encouraging young children to see themselves as writers and write stories and messages each day (Clarke, 1988; Morris, 1981; Read, 1986). Adams’ review of research, for example, concluded that:

the process of inventing spellings is essentially a process of phonics The evidence that invented spelling activity simultaneously develops phonemic awareness and promotes understanding of the alphabetic principle is extremely promising. (1990: 387)

Additional studies indicate that students encouraged to invent spellings seem to write longer, higher-quality texts and are more successful in spelling unfamiliar words correctly (Gettinger, 1993; Griffith et al., 1992).

A key component of the arguments surrounding invented spelling focus on the role of instruction, reflecting concerns that children will not learn the conventions of print without a daily spelling programme. Although whole language supporters such as Wilde (1990) have documented the spelling gains of students who read widely and write daily without formal spelling instruction, Richgels (1995) argues that children need both: (1) the opportunity to learn about grapho-phonemic relationships through both wide reading and many opportunities to write their own messages; and (2) planned instruction modelling conventional spellings and teaching orthographic features based upon student needs.

Such instruction, however, need not focus on worksheets or isolated drills. During interactive writing lessons, for example, the teacher models conventional spellings and teaches specific orthographic features as the teacher and children craft their message together; work together to write the message word by word, by saying the word slowly and writing the sounds that are heard; share the pen such that children write the letters they know; and reread the message to ensure accurate representation of the text (McCarrier et al., 2000). ‘What can you show us?’ is another instructional context for kindergarten children to develop both phonemic awareness and early concepts about letters, sounds, and words (Richgels et al., 1996) by asking children to examine a text with enlarged print (such as a poem, a big book, or a language experience chart) and demonstrate what they know about the text. Such instruction teaches concepts such as letter or word identification, letter-sound correspondences, rhyming words, or other text features. Spelling instruction in these contexts is placed securely within the contexts of reading and writing meaningful texts.

An important point, however, is that attention to early concepts must be addressed instructionally as it is not sufficient to assume that all children will learn them incidentally. Morris (1993) compared two different approaches to kindergarten instruction—one with a language experience approach and one emphasizing oral language and play. By May, 84% of the children in the kindergarten classroom using a language experience approach were successful with the concept of word task; only 50% of the children in the classroom emphasizing oral language and play understood the concept of word. Similarly, 71% of the students in the language experience classrooms could segment words into phonemes, but only 18% of the students in the language/play classroom could successfully perform the same task. This study highlights the importance of teaching young children concepts about words, which will then facilitate phonemic segmentation, letter-sound relationships, and word recognition through activities such as shared writing, language experience stories, shared reading, and helping children to write their own texts.

Matching Instruction with Assessment

Planning instruction based upon each student’s needs first requires that teachers have appropriate tools for determining developmental levels. The line of developmental spelling research has yielded spelling assessments such as Schlagal’s (1989) Qualitative Inventory of Word Knowledge (QIWK), spelling inventories for primary and intermediate learners by Bear et al. (2000), and Ganske’s (2000) Developmental Spelling Analysis (DSA). Each of these assessments provides lists of carefully selected words that gradually increase in both familiarity and feature difficulty. Analysis of children’s spelling attempts on these assessments enables teachers to determine each child’s developmental level and gain preliminary information about the orthographic features the child controls.

Bear et al. (2000) encourage teachers to find out what children are ‘using but confusing,’ demonstrating as yet incomplete knowledge of spelling concepts, and to target their instruction accordingly. For example, children who represent each word with letter-like shapes demonstrating few grapho-phonemic relationships are not yet ready for formal spelling instruction but can benefit from rhyming games, shared readings of poems, songs, and stories, sorting activities with pictures, and attention to learning the letters of their names. Based on his series of studies on concept of word, Morris (1993) concluded that formal spelling instruction should be delayed until the child has a firm concept of word to avoid both confusion and frustration.

Spellers who begin to demonstrate the accurate use of initial and final consonants and some evidence of short vowel understandings can benefit from studying CVC word patterns through word sorts, word hunts, word families and rhyming words (Ganske, 2000). Johnston encourages teachers to carefully consider when to teach word families, arguing that ‘well-timed instruction with word families can help children solidify tentative understandings, sort out current confusions, and move along to new understandings in the most efficient manner’ (1999: 67). As children develop a concept of word and begin to move beyond the sounds of letter names that they hear, they may begin to demonstrate some knowledge of silent letters which signals a need for instruction featuring more complex word sorts with VCe or CVVC patterns. The teacher’s role, then, is to build the instructional programme by organizing instruction to teach spelling features and concepts based upon students’ developmental levels.

Such instruction, however, contrasts sharply with the widespread use of basal spellers, typically one book for each grade level with a specific sequence of word lists and features. Such materials have been criticized for a one-size-fits-all approach that does not meet the needs of many children in a particular grade. Schlagal (2002) recognizes both the limitations of singular, grade-level spelling materials and the challenges faced by teachers attempting to respond to the individual needs of each student. He proposes that teachers use the resources found in a range of spelling basals, with their organized, progressively more difficult lists, to plan instruction for small groups of children with similar abilities, thereby matching instruction with ability to scaffold learning.

Future Directions

Recent research has provided significant information about how young children learn to spell. However, further studies are needed to clarify questions related to process, development, and instruction. There is relatively little information on developmental differences or similarities within and across different language systems. Future studies should help distinguish those aspects of learning about printed language that are more or less universal and those particular to specific types of orthographies. Not only are studies needed within and across different language systems, but studies are also needed about how children’s learning is affected by their exposure to one or more systems simultaneously. With the increased globalization of society, many students are learning to speak and write in two languages at the same time—in their native language and a second language in which they will be expected to function in a literate manner. It is also important to distinguish between situations in which the second language is learned as a foreign language (i.e. situations in which the primary mode of oral and written communication at home, school, and community is the native language) and contexts in which the second language is the principal means of communication in school and the larger community (e.g. for immigrant populations). These would be rich, though complex, contexts in which to more carefully explore developmental issues.

Other important questions relate to differences between normally developing and disabled readers/ spellers. Preliminary studies suggest that differences may be more related to time for learning than to differences in perceptual/cognitive processing and/or abilities. Further research may help to resolve this discussion. Increasingly advanced technology available for the study of brain activity during language and literacy processing may yield significant insights about brain function with important instructional implications for students experiencing learning difficulties.

A large proportion of spelling studies done thus far have focused on children as they begin formal schooling, but relatively few have been done with younger students. This emerging line of research has documented that important concepts about spelling are formed early and are developing during preschool and the early years of formal education. Studies are needed to examine the implications of instruction for three-to five-year-old children by identifying appropriate assessment tools, sensitive to young children’s needs; by exploring instructional practices in preschool settings that foster accelerated development; and by examining appropriate practices to provide early intervention for children at risk. Such studies could contribute to current questions about the relationship between development and instruction; that is, how does instruction foster or impede spelling development?

A theme throughout this review has been that learning to spell is a complex task as children learn to understand oral language and its functions, develop concepts about the visual aspects of print, and learn the specific relationships between oral and written language unique to their orthographic system. The line of research followed by Morris (1981; 1989a; 1989b; 1992; 1993) exploring the relationship between a learner’s concept of word and spelling has begun to provide insights into the relationships between complex factors. More research is needed to identify how learning about the orthographic system through reading and writing relate to one another over time.

Correlational and experimental studies on phonological and phonemic awareness and their relationship to spelling development have been gaining interest in the research community. But many of these studies have been criticized for limitations in their design. In a review of experimental methodology on interventions to develop phonological awareness, Troia (1999) reviewed 39 studies and found that most had serious methodological flaws. In fact, only seven of the studies met two-thirds or more of the evaluative criteria; all had at least one fatal flaw. Such reviews point to the challenges of designing valid and reliable research to inform the field about early spelling teaching and learning. Experimental studies often focus on single aspects of oral or written language in fragmented, isolated ways, leaving important questions about the application of findings to complex classroom contexts. New and creative methodologies are required, building on the experimental research of the past, to inform future classroom instruction.

Lines of early spelling research have typically focused on children and their learning processes. However, given the critical role of teachers in helping children to become conventional spellers, research must also focus on teachers and what they need to know and do in classrooms of diverse learners. At one time, instructional materials and professional development activities provided teachers with rather scripted, dogmatic (and sometimes erroneous) information about the nature of the orthographic system. Later, attention to spelling was limited, in preference to a focus on more holistic methods of literacy instruction. Currently there is renewed interest in providing teachers with accurate and useful information about the nature of the writing system and developmental processes. Models of spelling instruction based on developmental research recognize the range of abilities found in classrooms and reject one-size-fits-all, grade-level programmes. Such models require that teachers have a new level of knowledge about: how spelling systems work, how children learn, how to assess students’ knowledge, and how to organize classrooms to facilitate learning for diverse learners. Teacher education programmes might increase their efforts to develop, implement, and document innovative pre-service and in-service programmes that provide teachers with the knowledge, skills and dispositions they need to make such effective instruction a reality in their classrooms.