We agree with Frost's target article that orthographic effects in visual word recognition, including the letter transposition (or transposed letter [TL]) effect, reflect a cognitive processing strategy of optimizing cognitive resources under the specific linguistic environments in which letters are used. We also agree that existing models of visual word recognition focusing mostly on orthographic processing (while disregarding phonology, morphology, and semantics) do not capture the various known phenomena relating to how the human cognitive system processes printed words. However, we suggest that, just like other linguistic aspects such as phonology and morphology, linguistic characteristics of the orthography (e.g., the function of a letter in a word) must also be considered as an important linguistic aspect in visual word recognition models.
As Frost mentions, recent studies of alphabetic Hangul (the Korean writing system) showed no TL effects, unlike the European languages (Lee & Taft Reference Lee and Taft2009; Reference Lee and Taft2011). Other than Semitic languages (e.g., Hebrew and Arabic), which are discussed heavily in the target article, Hangul might be the only other type of alphabetic orthography that has failed to show TL effects.
For example, Lee and Taft (Reference Lee and Taft2009) demonstrated that phonological and morphological factors cannot account for such a lack of a letter transposition effect in Hangul. Hangul is a concatenating alphabetic script that is nonlinear. Words are separated into syllables, and each syllable has a predictable structure in terms of the positioning of its vowel and consonants within a square block. The vowel is always a variation of either a horizontal line or a vertical line. The consonantal onset (which is always orthographically present, though sometimes silent) appears either above a horizontally oriented vowel or to the left of a vertically oriented vowel.
The consonantal coda, when there is one, always appears at the bottom. For example, the Korean spelling of the word “Hangul” is 한글, where the first syllable 한 (“han”) is composed of the onset ㅎ (“h”) placed to the left of the vertically oriented vowelㅏ (“a”), with the coda ∟ (“n”) at the bottom. The second syllable글 (“gul”) is composed of the onsetㄱ (“g”) placed above the horizontally oriented vowelㅡ (“u”), which, in turn, is placed above the codaㄹ (“l”).
Lee and Taft (Reference Lee and Taft2009) observed little difficulty for Korean readers in making lexical decision responses to TL-generated nonwords in Hangul (e.g., 납묵 generated from the word 남북, “nambuk,” through the transposition of ㅁ and ㅂ), unlike the considerable difficulty experienced by English readers responding to nonwords of the same structure in English (e.g., “warlus” for “walrus”). Lee and Taft suggested that this was because letters are assigned to an orthographic onset, vowel, or coda position at an early stage of processing and that, unlike linear scripts, such assignment is unambiguous in Hangul. The consonant at the top or at the top left of the syllable block is always assigned to the onset slot, and any consonant at the bottom of the block is always assigned to the coda slot.
Because Hangul physically demarcates the onset and coda positions for every consonant, it is argued that it is ambiguity in assignment of a consonant to an onset or coda slot that leads to the letter transposition effect in a linear script such as English. That is, the available physical cues offered by Hangul (i.e., the onset, vowel, and coda positions, which can be unambiguously determined from the orthographic input) prevent confusion when letters are placed in an incorrect position. It suggests that models of letter processing should incorporate the involvement of subsyllabic structure, something that is currently lacking.
Also, observation of the pattern of responses arising from letter transpositions within a syllable in the Hangul script provides insight into core aspects of Korean subsyllabic structure (Lee & Taft Reference Lee and Taft2011). For example, in reading Hangul, no confusion occurred when the onset and coda of one syllable of a disyllabic word were transposed. However, confusion with the base word was observed when the two consonants within the complex coda were themselves transposed. The patterns of responses indicate that Hangul (and possibly all concatenating orthographic scripts) is processed based on its onset, vowel, and coda structure (Lee & Taft Reference Lee and Taft2011). There are more alphabetic orthographies that have failed to show the TL effect. As discussed by Frost, two good examples are the non-concatenating orthography of the Semitic languages of Hebrew (Velan & Frost Reference Velan and Frost2007) and Arabic (Andrews Reference Andrews1996; Perea et al. Reference Perea, Abu Mallouh and Carreiras2010). Taken together with the observations reported from Lee and Taft's (Reference Lee and Taft2009; Reference Lee and Taft2011) studies, it was therefore apparent that letter coding in these Semitic languages is organized in terms of the word root.
To clarify, we are not suggesting that ambiguity in the assignment of letters to their appropriate onset, vowel, or coda slot is the only source of the letter transposition effect. Rather, we suggest that letter-position information can be used differently, depending on the nature of the orthographic structure of the language. In light of this, we suggest that such characteristics of the orthography need to be considered in order to fully understand how humans process printed words. Indeed, the cognitive system that processes Korean (or English or Hebrew) is tuned to the structure of the Korean (or English or Hebrew) writing system.
We agree with Frost's target article that orthographic effects in visual word recognition, including the letter transposition (or transposed letter [TL]) effect, reflect a cognitive processing strategy of optimizing cognitive resources under the specific linguistic environments in which letters are used. We also agree that existing models of visual word recognition focusing mostly on orthographic processing (while disregarding phonology, morphology, and semantics) do not capture the various known phenomena relating to how the human cognitive system processes printed words. However, we suggest that, just like other linguistic aspects such as phonology and morphology, linguistic characteristics of the orthography (e.g., the function of a letter in a word) must also be considered as an important linguistic aspect in visual word recognition models.
As Frost mentions, recent studies of alphabetic Hangul (the Korean writing system) showed no TL effects, unlike the European languages (Lee & Taft Reference Lee and Taft2009; Reference Lee and Taft2011). Other than Semitic languages (e.g., Hebrew and Arabic), which are discussed heavily in the target article, Hangul might be the only other type of alphabetic orthography that has failed to show TL effects.
For example, Lee and Taft (Reference Lee and Taft2009) demonstrated that phonological and morphological factors cannot account for such a lack of a letter transposition effect in Hangul. Hangul is a concatenating alphabetic script that is nonlinear. Words are separated into syllables, and each syllable has a predictable structure in terms of the positioning of its vowel and consonants within a square block. The vowel is always a variation of either a horizontal line or a vertical line. The consonantal onset (which is always orthographically present, though sometimes silent) appears either above a horizontally oriented vowel or to the left of a vertically oriented vowel.
The consonantal coda, when there is one, always appears at the bottom. For example, the Korean spelling of the word “Hangul” is 한글, where the first syllable 한 (“han”) is composed of the onset ㅎ (“h”) placed to the left of the vertically oriented vowelㅏ (“a”), with the coda ∟ (“n”) at the bottom. The second syllable글 (“gul”) is composed of the onsetㄱ (“g”) placed above the horizontally oriented vowelㅡ (“u”), which, in turn, is placed above the codaㄹ (“l”).
Lee and Taft (Reference Lee and Taft2009) observed little difficulty for Korean readers in making lexical decision responses to TL-generated nonwords in Hangul (e.g., 납묵 generated from the word 남북, “nambuk,” through the transposition of ㅁ and ㅂ), unlike the considerable difficulty experienced by English readers responding to nonwords of the same structure in English (e.g., “warlus” for “walrus”). Lee and Taft suggested that this was because letters are assigned to an orthographic onset, vowel, or coda position at an early stage of processing and that, unlike linear scripts, such assignment is unambiguous in Hangul. The consonant at the top or at the top left of the syllable block is always assigned to the onset slot, and any consonant at the bottom of the block is always assigned to the coda slot.
Because Hangul physically demarcates the onset and coda positions for every consonant, it is argued that it is ambiguity in assignment of a consonant to an onset or coda slot that leads to the letter transposition effect in a linear script such as English. That is, the available physical cues offered by Hangul (i.e., the onset, vowel, and coda positions, which can be unambiguously determined from the orthographic input) prevent confusion when letters are placed in an incorrect position. It suggests that models of letter processing should incorporate the involvement of subsyllabic structure, something that is currently lacking.
Also, observation of the pattern of responses arising from letter transpositions within a syllable in the Hangul script provides insight into core aspects of Korean subsyllabic structure (Lee & Taft Reference Lee and Taft2011). For example, in reading Hangul, no confusion occurred when the onset and coda of one syllable of a disyllabic word were transposed. However, confusion with the base word was observed when the two consonants within the complex coda were themselves transposed. The patterns of responses indicate that Hangul (and possibly all concatenating orthographic scripts) is processed based on its onset, vowel, and coda structure (Lee & Taft Reference Lee and Taft2011). There are more alphabetic orthographies that have failed to show the TL effect. As discussed by Frost, two good examples are the non-concatenating orthography of the Semitic languages of Hebrew (Velan & Frost Reference Velan and Frost2007) and Arabic (Andrews Reference Andrews1996; Perea et al. Reference Perea, Abu Mallouh and Carreiras2010). Taken together with the observations reported from Lee and Taft's (Reference Lee and Taft2009; Reference Lee and Taft2011) studies, it was therefore apparent that letter coding in these Semitic languages is organized in terms of the word root.
To clarify, we are not suggesting that ambiguity in the assignment of letters to their appropriate onset, vowel, or coda slot is the only source of the letter transposition effect. Rather, we suggest that letter-position information can be used differently, depending on the nature of the orthographic structure of the language. In light of this, we suggest that such characteristics of the orthography need to be considered in order to fully understand how humans process printed words. Indeed, the cognitive system that processes Korean (or English or Hebrew) is tuned to the structure of the Korean (or English or Hebrew) writing system.