The varied effects of orthographic neighbors
Frost discusses density of orthographic space and its effect on letter-position encoding in reading. We suggest that density effects on reading go beyond orthographic density in the orthographic input lexicon, beyond letter-position encoding, and beyond normal reading. First, when thinking about the process of single word reading (for which we assume the dual route model; Coltheart Reference Coltheart2006; Coltheart et al. Reference Coltheart, Rastle, Perry, Langdon and Ziegler2001; Ellis & Young Reference Ellis and Young1996; Marshall Reference Marshall, Malatesha and Whitaker1984), lexical density affects reading in two ways. One way relates to the orthographic input lexicon, the other to output stages and the lexicality of response. Take, for example, a case in which the information that arrives from the orthographic-visual analyzer is partial (or underspecified). If this partial output matches a single item in the orthographic input lexicon, the reading is expected to be correct, even if the early encoding failed. For example, if the target word is frog, underspecified information about the order of the middle letters would still activate the correct entry in the lexicon. However, if the neighborhood is dense, then partial information might activate other lexical items that match the partial information. For example, for the word bared, underspecified order of middle letters might activate other entries in the orthographic lexicon, which might end up in reading the more frequent bread or beard.
The density of the lexical space has an important effect beyond the orthographic input lexicon. One can see such effects acting on later stages of reading that involve oral responses. Individuals with dyslexia tend to produce lexical responses in reading aloud. Therefore, target words for which an error creates another existing word are more liable to be read incorrectly. This is the case even when the orthographic input lexicon is not involved in reading, as in surface dyslexia. Individuals with impaired orthographic input lexicon, who therefore cannot read via the lexical route, read via the sublexical route. For these individuals, the density of the lexical space is still a crucial factor, determining whether or not they will make an error in reading. This can be most clearly viewed in the effect of potentiophones on reading in surface dyslexia. Potentiophones are words that are written differently and sound differently, but when read solely via grapheme-to-phoneme conversion, can be read aloud as the other word, which sounds differently. An example is the word none, which can be read as known if read via grapheme-to-phoneme conversion.
Potentiophones are significantly more difficult for individuals with surface dyslexia than words for which a surface-dyslexia error does not create another word (Friedmann & Lukov Reference Friedmann and Lukov2008). Because they do not read aloud via the orthographic input lexicon, the lexical density effect should be attributed to the output stages, rather than to the orthographic input lexicon. Namely, beyond the basic classification of languages into deep and shallow languages, the manifestation of surface dyslexia would depend crucially on another property: the availability of potentiophones. Thus, reading mechanisms may be identical in different languages, but the different orthographies may lead to different reading outcomes.
Moreover, the effect of orthographic-space density extends beyond letter-position encoding, to abstract letter identity. If letter-identity encoding fails, dense neighborhoods should also hamper reading. The underspecified letter-identity information would activate other entries in the orthographic input lexicon that match this partial information. Take, for example, the word fold, which has a large Coltheart's N (Coltheart et al. Reference Coltheart, Davelaar, Jonasson, Besner and Dornic1977). If the reader fails to encode the identity of a letter in the target word fold, many other words (like cold, hold, fond, and folk) can be activated. This point has a crucial effect on the manifestation of various dyslexias resulting from impairments in the orthographic-visual analyzer, including: visual dyslexia (Friedmann et al. Reference Friedmann, Biran and Gvion2012), neglect dyslexia (Vallar et al. Reference Vallar, Burani and Arduino2010), and attentional dyslexia (Davis & Coltheart Reference Davis and Coltheart2002), dyslexias in which letter-identity errors result in existing words.
Hebrew and Arabic: Similar, yet different
Hebrew and Arabic share the pivotal role of morphology, as well as the effect of morphological structure on reading. However, whereas Frost's target article treats all Semitic languages as one group with identical properties, and hence predicts similar reading patterns with respect to letter-position encoding, Hebrew and Arabic present important differences that create different reading patterns and different manifestations of dyslexia in the two languages (Friedmann & Haddad-Hanna, in press a; in press b). This point can be demonstrated by a property that is one of the main pillars of the target article: letter-position encoding.
Arabic orthography presents a unique orthographic feature: Letters take different forms in different positions in the word. The form of each letter is determined by its position in the word – initial, middle, or final, and whether or not it ligates to the letter that precedes it. The ligation status of a letter depends on whether the preceding letter is one of the six non-ligating letters, in which case it would not ligate to the previous letter, or one of the other letters. This factor of letter forms that are determined by letter position crucially affects letter-position encoding in Arabic. Friedmann and Haddad-Hanna (in press a; in press b) found that letter form had a substantial effect on the rate of letter-position errors of Arabic-speaking individuals with letter-position dyslexia (LPD). Twelve individuals with developmental LPD made letter-position errors almost only when the change of letter position did not require a change in letter form.
Importantly, the difference in this property between Arabic and Hebrew orthography led to differences in letter-position errors in the two languages. In a study of a bilingual Arabic-Hebrew man with acquired LPD, Friedmann and Haddad-Hanna (in press a) reported that, in reading lists of single words, he made significantly more letter-position errors in Hebrew than in Arabic. This difference was due to the fact that whereas both lists included migratable words, in Arabic a migration of the middle letters in many of these migratable words required letter-form change. Because Arabic readers with LPD do not make letter-position errors that change letter form, the participant made significantly fewer letter-position errors in the Arabic list. When given a list that only included migratable words in which the migration does not change letter form, his error rates in Hebrew and Arabic were the same (see also Friedmann & Gvion [Reference Friedmann and Gvion2005] for the effect of letter forms on the manifestation of another type of dyslexia – word-based neglect dyslexia). Thus, Hebrew and Arabic differ along dimensions that create different outcomes in reading. More generally, there are various dimensions along which orthographies differ; the exact same reading mechanism, when faced with different orthographies and languages, may give rise to different reading outcomes in normal reading and in dyslexia.
You have
Access
The research of dyslexia and patterns of reading in various dyslexias can be revealing with respect to the characterization of normal cognitive processes. One of the domains in which dyslexia is informative is the effect of the properties of the various orthographies on reading. On the basis of studies of dyslexia, we suggest that Frost's proposal regarding the interaction between the nature of an orthography and reading can be further extended along several lines.
The varied effects of orthographic neighbors
Frost discusses density of orthographic space and its effect on letter-position encoding in reading. We suggest that density effects on reading go beyond orthographic density in the orthographic input lexicon, beyond letter-position encoding, and beyond normal reading. First, when thinking about the process of single word reading (for which we assume the dual route model; Coltheart Reference Coltheart2006; Coltheart et al. Reference Coltheart, Rastle, Perry, Langdon and Ziegler2001; Ellis & Young Reference Ellis and Young1996; Marshall Reference Marshall, Malatesha and Whitaker1984), lexical density affects reading in two ways. One way relates to the orthographic input lexicon, the other to output stages and the lexicality of response. Take, for example, a case in which the information that arrives from the orthographic-visual analyzer is partial (or underspecified). If this partial output matches a single item in the orthographic input lexicon, the reading is expected to be correct, even if the early encoding failed. For example, if the target word is frog, underspecified information about the order of the middle letters would still activate the correct entry in the lexicon. However, if the neighborhood is dense, then partial information might activate other lexical items that match the partial information. For example, for the word bared, underspecified order of middle letters might activate other entries in the orthographic lexicon, which might end up in reading the more frequent bread or beard.
The density of the lexical space has an important effect beyond the orthographic input lexicon. One can see such effects acting on later stages of reading that involve oral responses. Individuals with dyslexia tend to produce lexical responses in reading aloud. Therefore, target words for which an error creates another existing word are more liable to be read incorrectly. This is the case even when the orthographic input lexicon is not involved in reading, as in surface dyslexia. Individuals with impaired orthographic input lexicon, who therefore cannot read via the lexical route, read via the sublexical route. For these individuals, the density of the lexical space is still a crucial factor, determining whether or not they will make an error in reading. This can be most clearly viewed in the effect of potentiophones on reading in surface dyslexia. Potentiophones are words that are written differently and sound differently, but when read solely via grapheme-to-phoneme conversion, can be read aloud as the other word, which sounds differently. An example is the word none, which can be read as known if read via grapheme-to-phoneme conversion.
Potentiophones are significantly more difficult for individuals with surface dyslexia than words for which a surface-dyslexia error does not create another word (Friedmann & Lukov Reference Friedmann and Lukov2008). Because they do not read aloud via the orthographic input lexicon, the lexical density effect should be attributed to the output stages, rather than to the orthographic input lexicon. Namely, beyond the basic classification of languages into deep and shallow languages, the manifestation of surface dyslexia would depend crucially on another property: the availability of potentiophones. Thus, reading mechanisms may be identical in different languages, but the different orthographies may lead to different reading outcomes.
Moreover, the effect of orthographic-space density extends beyond letter-position encoding, to abstract letter identity. If letter-identity encoding fails, dense neighborhoods should also hamper reading. The underspecified letter-identity information would activate other entries in the orthographic input lexicon that match this partial information. Take, for example, the word fold, which has a large Coltheart's N (Coltheart et al. Reference Coltheart, Davelaar, Jonasson, Besner and Dornic1977). If the reader fails to encode the identity of a letter in the target word fold, many other words (like cold, hold, fond, and folk) can be activated. This point has a crucial effect on the manifestation of various dyslexias resulting from impairments in the orthographic-visual analyzer, including: visual dyslexia (Friedmann et al. Reference Friedmann, Biran and Gvion2012), neglect dyslexia (Vallar et al. Reference Vallar, Burani and Arduino2010), and attentional dyslexia (Davis & Coltheart Reference Davis and Coltheart2002), dyslexias in which letter-identity errors result in existing words.
Hebrew and Arabic: Similar, yet different
Hebrew and Arabic share the pivotal role of morphology, as well as the effect of morphological structure on reading. However, whereas Frost's target article treats all Semitic languages as one group with identical properties, and hence predicts similar reading patterns with respect to letter-position encoding, Hebrew and Arabic present important differences that create different reading patterns and different manifestations of dyslexia in the two languages (Friedmann & Haddad-Hanna, in press a; in press b). This point can be demonstrated by a property that is one of the main pillars of the target article: letter-position encoding.
Arabic orthography presents a unique orthographic feature: Letters take different forms in different positions in the word. The form of each letter is determined by its position in the word – initial, middle, or final, and whether or not it ligates to the letter that precedes it. The ligation status of a letter depends on whether the preceding letter is one of the six non-ligating letters, in which case it would not ligate to the previous letter, or one of the other letters. This factor of letter forms that are determined by letter position crucially affects letter-position encoding in Arabic. Friedmann and Haddad-Hanna (in press a; in press b) found that letter form had a substantial effect on the rate of letter-position errors of Arabic-speaking individuals with letter-position dyslexia (LPD). Twelve individuals with developmental LPD made letter-position errors almost only when the change of letter position did not require a change in letter form.
Importantly, the difference in this property between Arabic and Hebrew orthography led to differences in letter-position errors in the two languages. In a study of a bilingual Arabic-Hebrew man with acquired LPD, Friedmann and Haddad-Hanna (in press a) reported that, in reading lists of single words, he made significantly more letter-position errors in Hebrew than in Arabic. This difference was due to the fact that whereas both lists included migratable words, in Arabic a migration of the middle letters in many of these migratable words required letter-form change. Because Arabic readers with LPD do not make letter-position errors that change letter form, the participant made significantly fewer letter-position errors in the Arabic list. When given a list that only included migratable words in which the migration does not change letter form, his error rates in Hebrew and Arabic were the same (see also Friedmann & Gvion [Reference Friedmann and Gvion2005] for the effect of letter forms on the manifestation of another type of dyslexia – word-based neglect dyslexia). Thus, Hebrew and Arabic differ along dimensions that create different outcomes in reading. More generally, there are various dimensions along which orthographies differ; the exact same reading mechanism, when faced with different orthographies and languages, may give rise to different reading outcomes in normal reading and in dyslexia.