Frost's timely article launches a theory of reading single words that stands in sharp contrast to previous, English-oriented models focusing solely on orthographic or phonological patterns. Two critical features render this reading model linguistically plausible and ecologically valid: It intrinsically accommodates different orthographic systems tailored to highlight the meaning-carrying units of the languages they express; and it assumes learning to be the main mechanism which picks up statistical patterns inherent in the input to recover meaning.
This commentary is about a model which has evolved in the parallel universe of spelling research and resonates with Frost's reading model, albeit being informed by different schools of thought and based on rather different evidence sources. This model of spelling (Ravid Reference Ravid2012) construes the psycholinguistic underpinnings of Hebrew spelling as a network of phono-morpho-orthographic statistical patterns in the printed input from which reader and writers extract meaning. It is grounded in four main conceptual arenas. One is the encyclopedic approach to language held by the schools of Cognitive Linguistics (Croft & Cruse Reference Croft and Cruse2004; Langacker Reference Langacker1987) and Construction Grammar (Goldberg Reference Goldberg2003), which assume the existence of linguistic networks of subtle form–function relations learned on the basis of the input coupled with general cognitive mechanisms. Another is development as the emergent process that drives learning in interaction with the environment (Elman Reference Elman2003, p. 430). The spelling model is next predicated on the typological imperative (Berman Reference Berman, Fletcher and Garman1986; Slobin Reference Slobin, Gumperz and Levinson1996), as the set of salient characteristics of a language that shapes its acquisition. Finally, the domain of literacy makes a major contribution via Olson's view of the “script as model” (Olson Reference Olson1994), which provides the link between language typology and orthographic variation in spelling development. The spelling model is informed by analyses of Modern Hebrew and its orthography, as well as by cross-linguistic developmental evidence of spelling acquisition (for reviews, see Ravid Reference Ravid2012; Sandra Reference Sandra and Jarema2007).
The spelling model delineates a path to overcome homophonous letters in Hebrew, which derive from the loss of historical phonological distinctions, such as the distinction between emphatic q and non-emphatic k (Bolozky Reference Bolozky, Kaye and Daniels1997; Ravid Reference Ravid1995; Schwarzwald Reference Schwarzwald2001). Homophony is offset by the powerful impact of the morphological root and function roles of letters, occupying different sites in the written word. For example, v can be written by either ו or ב, however the conjunction ve (“and”) is exclusively spelled as ו at word-initial position. Clear participation in the function “envelope” at both sides of the root or stem is a shortcut to correct spelling, supporting only one possible path between homophonous sound and letter, leading to early correct spelling of function letters in Hebrew (Gillis & Ravid Reference Gillis and Ravid2006; Ravid Reference Ravid2001; Reference Ravid, Joshi and Aaron2005).
This analysis of a non–Indo-European system indicates that a spelling model, like a reading model, cannot simply be based on one linguistic facet such as phonology or orthography, but should accommodate all linguistic facets of a word. While an alphabetical orthography certainly holds an intimate relationship with phonology, it is not merely a reflection of phonology in graphic form – a role restricted to linguistic transcription systems. The goal of spelling is to achieve a high-quality fit with correct orthographic representation, which by definition expresses lexical and morphological information. In many languages with alphabetical orthographies, morphology – the structural organization of meaning within the word – constitutes the architecture of hidden units mediating the complex and often opaque relationships between phonology and orthography (Ravid Reference Ravid2012). In learning to spell, learners would be looking for the same meaningful word-internal units that they have mapped out for spoken language. Spelling thus involves the implementation of morphological and morpho-phonological knowledge within the boundaries of a word.
As an illustration of the role of networks relating orthographic patterns to morpho-phonological knowledge in gaining command of spelling, consider the erroneous spelling of meida (“information”) as הדימ rather than עדימ – that is, treating masculine meida as feminine because it ends with a stressed a. Correct spelling requires a knowledge network that would let the speller know that despite the superficial phonological similarity to feminine nouns, meida is masculine, and therefore does not end with feminine ה. Specific morphological knowledge would relate meida to root עדי (“know”), extracted from a lexical inventory of this morphological family, such as hodí'a (“inform,” עידוה), mada (“science,” עדמ), or moda'a (“notice,” העדומ). General morpho-phonological knowledge would inform one that ע typically attracts the low vowel a at word final position, based on masculine items containing ע-final roots, such as verb hirgí'a (“calm,” עיגרה) or noun masa (“journey,” עסמ). These two converging networks – the morphological family, relating meida to other words based on the same root, and morpho-phonological generalizations based on phonologically similar items with similar roots – specify a small but consistent class of ע-final masculine items that superficially resemble the vastly larger ה-final feminine class. In experienced and literate speakers/writers, this knowledge motivates the choice of ע over ה in spelling meida.
A flexible and universally applied model of spelling cannot be mired at the level of the grapheme–phoneme link and grapheme combination. As a cognitive domain, correct spelling does not take place at the level of grapheme, phoneme, or syllable, but rather at the lexical level. This attention to meaning is at the heart of the new view of spelling as – alongside with phonology – deriving from the lexicon. Empiricist approaches (e.g., Oller Reference Oller2000) regard the lexicon as the core of phonological generalizations. Phonological representations can thus be viewed as “emergent properties of … word shapes the language user encounters and stores in memory” (Beckman & Edwards Reference Beckman and Edwards2000, p. 241). In the same way, spelling is regarded as a lexical operation that fosters the production and comprehension of words – and is thus nested within a complex network of phonological, morphological, and orthographic categories and patterns. As in spoken language, learners need to extract distributional frequencies about mapping these three domains in words. Spelling phenomena such as frequent letter sequences, phoneme–grapheme correspondences, and classification of letters by morphological role (such as root or function letters) are all entailed by the fact that correct or conventional spelling supports the production and recognition of words.
Frost's timely article launches a theory of reading single words that stands in sharp contrast to previous, English-oriented models focusing solely on orthographic or phonological patterns. Two critical features render this reading model linguistically plausible and ecologically valid: It intrinsically accommodates different orthographic systems tailored to highlight the meaning-carrying units of the languages they express; and it assumes learning to be the main mechanism which picks up statistical patterns inherent in the input to recover meaning.
This commentary is about a model which has evolved in the parallel universe of spelling research and resonates with Frost's reading model, albeit being informed by different schools of thought and based on rather different evidence sources. This model of spelling (Ravid Reference Ravid2012) construes the psycholinguistic underpinnings of Hebrew spelling as a network of phono-morpho-orthographic statistical patterns in the printed input from which reader and writers extract meaning. It is grounded in four main conceptual arenas. One is the encyclopedic approach to language held by the schools of Cognitive Linguistics (Croft & Cruse Reference Croft and Cruse2004; Langacker Reference Langacker1987) and Construction Grammar (Goldberg Reference Goldberg2003), which assume the existence of linguistic networks of subtle form–function relations learned on the basis of the input coupled with general cognitive mechanisms. Another is development as the emergent process that drives learning in interaction with the environment (Elman Reference Elman2003, p. 430). The spelling model is next predicated on the typological imperative (Berman Reference Berman, Fletcher and Garman1986; Slobin Reference Slobin, Gumperz and Levinson1996), as the set of salient characteristics of a language that shapes its acquisition. Finally, the domain of literacy makes a major contribution via Olson's view of the “script as model” (Olson Reference Olson1994), which provides the link between language typology and orthographic variation in spelling development. The spelling model is informed by analyses of Modern Hebrew and its orthography, as well as by cross-linguistic developmental evidence of spelling acquisition (for reviews, see Ravid Reference Ravid2012; Sandra Reference Sandra and Jarema2007).
The spelling model delineates a path to overcome homophonous letters in Hebrew, which derive from the loss of historical phonological distinctions, such as the distinction between emphatic q and non-emphatic k (Bolozky Reference Bolozky, Kaye and Daniels1997; Ravid Reference Ravid1995; Schwarzwald Reference Schwarzwald2001). Homophony is offset by the powerful impact of the morphological root and function roles of letters, occupying different sites in the written word. For example, v can be written by either ו or ב, however the conjunction ve (“and”) is exclusively spelled as ו at word-initial position. Clear participation in the function “envelope” at both sides of the root or stem is a shortcut to correct spelling, supporting only one possible path between homophonous sound and letter, leading to early correct spelling of function letters in Hebrew (Gillis & Ravid Reference Gillis and Ravid2006; Ravid Reference Ravid2001; Reference Ravid, Joshi and Aaron2005).
This analysis of a non–Indo-European system indicates that a spelling model, like a reading model, cannot simply be based on one linguistic facet such as phonology or orthography, but should accommodate all linguistic facets of a word. While an alphabetical orthography certainly holds an intimate relationship with phonology, it is not merely a reflection of phonology in graphic form – a role restricted to linguistic transcription systems. The goal of spelling is to achieve a high-quality fit with correct orthographic representation, which by definition expresses lexical and morphological information. In many languages with alphabetical orthographies, morphology – the structural organization of meaning within the word – constitutes the architecture of hidden units mediating the complex and often opaque relationships between phonology and orthography (Ravid Reference Ravid2012). In learning to spell, learners would be looking for the same meaningful word-internal units that they have mapped out for spoken language. Spelling thus involves the implementation of morphological and morpho-phonological knowledge within the boundaries of a word.
As an illustration of the role of networks relating orthographic patterns to morpho-phonological knowledge in gaining command of spelling, consider the erroneous spelling of meida (“information”) as הדימ rather than עדימ – that is, treating masculine meida as feminine because it ends with a stressed a. Correct spelling requires a knowledge network that would let the speller know that despite the superficial phonological similarity to feminine nouns, meida is masculine, and therefore does not end with feminine ה. Specific morphological knowledge would relate meida to root עדי (“know”), extracted from a lexical inventory of this morphological family, such as hodí'a (“inform,” עידוה), mada (“science,” עדמ), or moda'a (“notice,” העדומ). General morpho-phonological knowledge would inform one that ע typically attracts the low vowel a at word final position, based on masculine items containing ע-final roots, such as verb hirgí'a (“calm,” עיגרה) or noun masa (“journey,” עסמ). These two converging networks – the morphological family, relating meida to other words based on the same root, and morpho-phonological generalizations based on phonologically similar items with similar roots – specify a small but consistent class of ע-final masculine items that superficially resemble the vastly larger ה-final feminine class. In experienced and literate speakers/writers, this knowledge motivates the choice of ע over ה in spelling meida.
A flexible and universally applied model of spelling cannot be mired at the level of the grapheme–phoneme link and grapheme combination. As a cognitive domain, correct spelling does not take place at the level of grapheme, phoneme, or syllable, but rather at the lexical level. This attention to meaning is at the heart of the new view of spelling as – alongside with phonology – deriving from the lexicon. Empiricist approaches (e.g., Oller Reference Oller2000) regard the lexicon as the core of phonological generalizations. Phonological representations can thus be viewed as “emergent properties of … word shapes the language user encounters and stores in memory” (Beckman & Edwards Reference Beckman and Edwards2000, p. 241). In the same way, spelling is regarded as a lexical operation that fosters the production and comprehension of words – and is thus nested within a complex network of phonological, morphological, and orthographic categories and patterns. As in spoken language, learners need to extract distributional frequencies about mapping these three domains in words. Spelling phenomena such as frequent letter sequences, phoneme–grapheme correspondences, and classification of letters by morphological role (such as root or function letters) are all entailed by the fact that correct or conventional spelling supports the production and recognition of words.