Shallice and Warrington (1986) first proposes the distinction between peripheral and central dyslexia. Peripheral dyslexia occurs at the level prior to visual recognition of words, whereas central dyslexia occurs at the level from visual word recognition to phonology or semantics.

Central dyslexia includes phonological dyslexia (the selective impairment of nonword reading), surface dyslexia (the selective impairment of irregular word reading), and deep dyslexia (phonological dyslexia plus semantic paralexia) (Table 1). Different manifestations in these dyslexias are presented by kanji and kana reading (Table 2).

Similarly, agraphia is divided into central (linguistic) agraphia and peripheral (motor) agraphia (Ellis and Young, 1997). Central agraphia refers to impaired access to the visual images required for writing, whereas peripheral agraphia refers to the interruption of converting from a visual image to a kinesthetic pattern of words.

Central agraphia includes phonological agraphia (the selective impairment of nonword writing), lexical/surface agraphia (the selective impairment of irregular word writing), and deep agraphia (phonological agraphia plus semantic paragraphia) (Table 3). Some authors identified lexical agraphia with surface agraphia (Ullrich and Roeltgen, 2011); others subdivided surface agraphia into a semantic type and orthographic type (Beeson and Rapcsak, 2015). However, lexical agraphia should be distinguished from surface agraphia in that lexical agraphia results from impairment of the orthographic lexicon or its connection to auditory input/output, whereas surface agraphia is caused by impairment of the semantic memory. This view is supported by the fact that regularization (phonologically plausible kanji writing, described later, e.g., 新文 [shinbun], a pseudohomophonous new word for 新聞 [shinbun], “newspaper”) errors are observed in the surface agraphia of semantic dementia (Sakurai et al., 2006a) but not in lexical agraphia or kanji agraphia resulting from focal lesions of the temporal and parietal lobes (Sakurai et al., 2000b; Sakurai et al., 2008a).

On the other hand, peripheral agraphia includes apraxic agraphia (the inability to realize well-formed graphemes despite the preserved visual images of words) and allographic agraphia (confusion between uppercase and lowercase) (Table 3). Depending on the word type (nonword, regular word, and irregular word), kanji and kana writing also show dissociable differences (Table 2).

Specifically, we previously reported cases of somesthetic dysgraphia (Sakurai et al., 2007; Mitani and Sakurai, 2024). This agraphia type is characterized by illegible grapheme formation due to impaired deep and combined sensation (stereognosis, two-point discrimination, graphesthesia, etc.), resulting in insufficient proprioceptive feedback to the motor programming. Somesthetic dysgraphia may be classified as peripheral agraphia (Table 3).

Kana alexia/agraphia resulting from disrupted grapheme–phoneme conversion is considered phonological dyslexia/agraphia, because greater impairment is expected for kana nonword reading/writing than for kana word reading/writing. In contrast, kanji agraphia resulting from impaired character recall is considered lexical agraphia, as kanji words with irregular reading are expected to be more difficult to read or write than those with regular reading.

We proposed a classification of alexia/agraphia in accordance with the results of lesion studies (Sakurai et al., 2010; Table 4). As previously described, alexia and agraphia are first divided into aphasic and nonaphasic types. Nonaphasic alexia/agraphia is further divided into a primary type (i.e., caused by lesions related to reading and writing) and a secondary type (i.e., caused by lesions unrelated to language that affect reading or writing performance). In primary nonaphasic alexia/agraphia, kanji or kana selectivity is added to each of the alexia/agraphia types (second and third columns in Table 4).

The classification has several points that have not been highlighted before. These points are discussed in the following session in accordance with the anatomical substrates.

Pure alexia is further divided into the splenium type or classical type, which involves the splenium of the corpus callosum, and nonsplenium type or nonclassical type (Kawamura, 1988).

Iwata was the first to report alexia with agraphia for kanji or posterior inferior temporal (PIT)–type alexia with agraphia (Iwata, 1984, 1986). Subsequently, several case studies on the PITC lesion were published (Kawamura et al., 1987; Kawahata et al., 1988; Soma et al., 1989; Sakurai et al., 1994) and revealed the following features: (1) both single-character kanji and multiple-character kanji words were poorly read or written because of impaired character recall (Sakurai et al., 2000b; Sakurai, 2004), (2) agraphia was more profound and persisted longer than alexia did (Sakurai et al., 2000b), (3) kana reading was mildly impaired and showed letter-by-letter reading (Sakurai et al., 2000b), (4) kinesthetic reading was not always successful (Sakurai et al., 2000b), and (5) kanji reading was affected by complexity, frequency, and familiarity (Sakurai et al., 2006b), whereas errors in kanji writing were mostly due to impaired character recall (lexical agraphia), which was influenced by familiarity (Sakurai et al., 2008a).

Twenty years after the first report by Iwata (1984), similar cases of alexia with agraphia from a PITC lesion were reported in Western countries (Rapcsak and Beeson, 2004). In these cases, alexia and agraphia were more pronounced in irregular words, with coexistence of letter-by-letter reading. Because the essential deficit lies in impaired word recall, we designated this syndrome as “orthographic alexia with agraphia” (Sakurai, 2004).

Two neuroanatomical problems have been noted (Sakurai, 2011). First, there are some similarities between PIT-type alexia with agraphia and fusiform-type pure alexia. That is, the PIT-type alexia with agraphia has letter-by-letter reading, as described above, whereas the fusiform-type pure alexia has kanji agraphia. The symptomatic distinction is that agraphia is more severe in alexia with agraphia for kanji and alexia is more severe in pure alexia for kanji (Sakurai et al., 2006b). Neuroanatomically, these two types have nearly the same lesion. The only difference is that the fusiform-type pure alexia has a lesion more medial to the PIT-type alexia with agraphia (Figure 4). Kawamura (1988) first indicated that the nonclassical (nonsplenium) type pure alexia is characterized by a lesion slightly more medial to the common lesion site of the PIT-type alexia with agraphia.

Second, in some individuals, alexia with agraphia for kanji resolved to pure agraphia for kanji in a similar lesion (Soma et al., 1989; Tsuruta et al., 1992; Komori et al., 2009). Furthermore, isolated agraphia for kanji occurred in nearly the same lesion (Yokota et al., 1990). Thus, some authors argued that alexia with agraphia for kanji proves to be pure agraphia for kanji or lexical agraphia in Western countries (Soma et al., 1989). We reported two patients with damage to the posterior part of the middle and inferior temporal gyri who exhibited pure agraphia for kanji (Sakurai et al., 2008a). Most of the writing errors consisted of no response due to impaired character recall. The kanji writing was influenced by character complexity, frequency, and familiarity. Taking these cases into consideration, we thought that alexia with agraphia for kanji and pure agraphia for kanji have separate neural substrates; that is, the lesion of pure agraphia for kanji is dorsolateral to that of alexia with agraphia for kanji in BA 37. This difference is consistent with the separate activation between reading and writing tasks observed in the imaging studies, which is described in the mechanism session.

Patients who have progressive Gogi (word-meaning) aphasia, the Japanese equivalent of semantic variant primary progressive aphasia (as described earlier), exhibit anomia and surface dyslexia/agraphia. Characteristic errors are regularization (phonologically plausible errors, e.g., “pint” as “/pint/”) or legitimate alternative reading/spelling of components (Patterson et al., 1995; Weekes et al., 2003). Confusion between on-reading and kun-reading has been reported in Japanese patients with semantic dementia (Sasanuma and Monoi, 1975; Sakurai et al., 2006a; Sakurai et al., 2021), as follows:

In contrast to the above-mentioned on-reading predominance, we reported a woman with alexia with agraphia for kanji who exhibited selective impairment of on-reading after hemorrhage in the left posterior middle temporal gyrus (Yoshida et al., 2020). The patient read 98 single-character kanji with kun-reading at an accuracy of 98% (time: 5 min), which decreased to 71% (time: 23 min) with on-reading. She also correctly read on-reading words at 52%, kun-reading words at 87%, and specific-reading words at 82%. She tried to recall the on-reading by associating two-character kanji words containing the target character (e.g., 白 [shiro], kun-reading meaning “white” → 紅白 [kou-haku], on–on reading meaning “red and white” → 白 [haku], on-reading).

The on-reading predominance over kun-reading in semantic dementia and the reverse pattern in this on-reading alexia suggest double dissociation between the on-reading process and the kun-reading process. Given that kun-reading is directly linked to semantics, the kun-reading pathway corresponds to the lexical–semantic route (V2 → A2 in Figure 2; O → S → P in Figure 5), whereas the on-reading pathway corresponds to the direct pathway from the orthographic lexicon to the phonological lexicon ([a] in Figure 2; O → P in Figure 5). The patient with on-reading alexia may have damage to the direct pathway. The on-kun dissociation demonstrates the existence of two pathways—direct and indirect—from the orthographic lexicon to the phonological lexicon.

Dejerine (1891) first described alexia with agraphia due to an angular gyrus lesion. He proposed that the angular gyrus stored visual images of words and letters. However, recent lesion-to-symptom studies have implicated that PITC, not the angular gyrus, is the site for visual images of words (Sakurai et al., 2000b; Rapcsak and Beeson, 2004). It should also be noted that Dejerine’s original case had damage that extended to the lateral occipital gyri. Although the concept of “angular” alexia with agraphia seems established (Benson and Ardila, 1996), it remains unclear whether alexia with agraphia occurs in the circumscribed lesion of the angular gyrus (Yamadori, 1982).

In Western countries, it is known that damage to the angular gyrus causes lexical agraphia (Roeltgen and Heilman, 1984). In Japanese, angular alexia with agraphia presents with kana (phonological) alexia and kanji (lexical) agraphia (Kawamura, 1990; Miyake and Tanaka, 1992). We reported a patient with pure agraphia of kanji caused by a restricted infarction of the angular gyrus (Sakurai et al., 2000b) and a patient with alexia with agraphia (kana alexia and kanji agraphia) due to subcortical hemorrhage in the angular and lateral occipital gyri (Sakurai et al., 2010). Thus, damage to the lateral occipital gyri posterior to the angular gyrus may give rise to alexia symptoms in angular alexia with agraphia. In support of this view, a patient did not show alexia until he had a second infarction in the lateral occipital gyri in addition to the first infarction in the angular gyrus (Kuriyama et al., 2006). Kleist (1934) was the first to claim that alexia in the angular gyrus lesion results from extension of the lesion to the middle occipital gyrus, posterior to the angular gyrus.

Kana agraphia or phonological agraphia occurs after damage to the supramarginal gyrus (Sakurai et al., 2010). Most errors are phonological (one or more substitutions in kana characters). In particular, transposition (changing the sequential order of two characters, e.g., ふしんせつ) ([hushinsetsu], “unkind” → ふんしせつ [hunshisetsu], nonword) was observed. This suggests disrupted sequential processing of characters in spelling/writing.

Some authors have reported selective or pure kana agraphia from the parietal lobe lesion including the supramarginal gyrus (Tanaka et al., 1987; Fujii et al., 1995). However, a detailed examination showed that kanji writing was mildly impaired due to defective character recall (Sakurai et al., 2010).

Kana agraphia is accompanied by mild to minimal phonological dyslexia of kana in the supramarginal gyrus lesion (Sakurai et al., 2010). In this case, similar to angular alexia with agraphia, transposition errors (i.e., changing the character sequences of the word) are observed. This indicates that sequential phonological processing of kana characters is performed from the angular/lateral occipital gyri to the supramarginal gyrus. It should be noted that in patients with damage to the supramarginal and angular gyri, verbal short-term memory is also impaired (Sakurai et al., 1998), which might affect the capacity for sequential processing.

Phonological dyslexia/agraphia and conduction aphasia share the same lesion of the supramarginal gyrus (Coslett, 2011; Ullrich and Roeltgen, 2011). In fact, these symptoms appear in the course of conduction aphasia (Alexander et al., 1992b; Beeson et al., 2010).

Apraxic agraphia is defined as (1) illegible grapheme formation that cannot be accounted for by sensorimotor dysfunction (Alexander et al., 1992a; Rapcsak and Beeson, 2002; Ullrich and Roeltgen, 2011), (2) improvement in grapheme formation with copying (Ullrich and Roeltgen, 2011), (3) preservation of oral spelling or typing (Ullrich and Roeltgen, 2011), and (4) disordered writing stroke sequences. We have added the fourth feature to these criteria (Sakurai et al., 2007). For example, the patient wrote a character as if he had drawn a line figure. Although the writing stroke sequence disorder is also observed in European languages (Coslett et al., 1986), it is easily observed in the Japanese kanji, which require complex sequential writing strokes.

The neural substrate for apraxic agraphia is located in the superior parietal lobule (Alexander et al., 1992a) or intraparietal sulcus area (Rapcsak and Beeson, 2002), where it is assumed visuokinesthetic and sequential motor engrams for letters and words are stored (G/K in Figure 7) (Sakurai et al., 2007). A meta-analysis of 18 imaging studies for handwriting identified an anterior SPL/IPS area (peak activation, x = −32, y = −38, z = 56 in the MNI coordinates) specific to handwriting (Planton et al., 2013). However, its role in handwriting is unknown. A lesion-to-symptom analysis suggests that anterior SPL/IPS damage causes disordered writing stroke sequences (Sakurai et al., 2007). In a broad sense, the G/K area includes more extensive cortices in the posterior SPL/IPS; inferior parietal, superior occipital, and precuneus gyri; and stores visuospatial attributes of characters (described in the anatomically constrained model of spelling/writing). This hypothesis might explain why some patients with apraxic agraphia exhibit deficits in letter imagery (impaired character recall) and severe deformity in writing characters (Crary and Heilman, 1988; Sakurai et al., 2007).

From a neuropsychological perspective, apraxic agraphia results from damage to the parietal G/K area per se (Sakurai et al., 2007) or disconnection of the output from the G/K area to the frontal hand area (H) (Otsuki et al., 1999). The disconnection type differs clinically from the cortical type in that disordered writing stroke sequences are more pronounced whereas grapheme deformity is minimal.

Isolated agraphia is caused by damage to the left posterior middle frontal gyrus and surrounding cortices (Abe et al., 1993; Tohgi et al., 1995; Sakurai et al., 1997; Keller and Meister, 2014). This area has been traditionally called Exner’s area (Roux et al., 2010). Agraphia is characterized by kana or literal agraphia: letter substitution, omission, and perseveration are frequently observed. If the lesion involves the inferior frontal gyrus, deficits in kana or letter imagery (impaired character recall) occur (Sakurai et al., 1997). If the lesion involves the adjacent precentral gyrus, co-occurrence of apraxia of speech or kanji (or lexical) agraphia may result (Rapcsak et al., 1988; Saito et al., 1997; Sakurai et al., 1997; Kezuka et al., 1999).

The previously mentioned meta-analysis (Planton et al., 2013) revealed two separate frontal regions for handwriting: the posterior superior and middle frontal gyri (BA 6) specialized for handwriting and the inferior precentral/posterior inferior frontal gyri (BA 6) related to more general linguistic processing. The role of these areas is unknown. We consider these two areas to be functionally connected. That is, handwriting requires at least three sources of information (Sakurai et al., 2018): (1) visuokinesthetic information of a word or letter (the indirect orthographic route, V3 → V5 → V6 in Figure 2; O → (b) → G/K → (d) → H in Figure 7), (2) the whole-word graphic image that is converted from phonemes in the temporoparietal area (Sakurai et al., 2010) (the direct orthographic route, (d) in Figure 2; O → (b) → H in Figure 7), and (3) phonological information of the word transmitted from Wernicke’s area (the phonological route, A3 → A5 → (c) in Figure 2; P → (a) → H in Figure 7). These three information sources reach the premotor and motor hand area (H) and are linked to produce motor execution of words or letters. As described in the next section, the posterior superior/middle frontal gyri may receive visuokinesthetic and whole-word information, whereas the inferior precentral/posterior inferior frontal gyri may receive grapheme-linked phonological information,.

Some authors claimed that the phoneme-to-grapheme conversion was localized in the inferior precentral/posterior inferior frontal gyri (Omura et al., 2004; Ogura et al., 2013). However, it is more likely that the graphic images of characters are stored in the angular/lateral occipital gyri (described earlier) and that phoneme–grapheme conversion is performed in these cortices. What occurs in the motor and premotor area (H) is not conversion from phoneme to grapheme but rather a connection or integration of phoneme, grapheme, and visuokinesthetic information.

The inferior longitudinal fasciculus is a major network that connects the occipital lobe and the temporal lobe. The primary and association visual cortices (V in Figure 5) connect the orthographic lexicon (O) and further the anterior temporal semantic storage site (S) through the inferior longitudinal fasciculus. The phonological lexicon (P) and orthographic lexicon (O) have reciprocal connections via U-fibers. The middle longitudinal fasciculus connects the posterior superior temporal gyrus (P; Wernicke’s area; phonological lexicon) and the semantic storage site (S) (Makris et al., 2009), which is used for direct semantic access from Wernicke’s area.

The frontoparietal networks are composed of the superior longitudinal fasciculus (SLF). The SLF is divided into three parts: the dorsal branch that connects the superior parietal lobule and superior frontal gyrus (SLF I), the middle branch that connects the intraparietal sulcus area and superior and middle frontal gyri (SLF II), and the ventral branch that connects the inferior parietal lobule and inferior frontal gyrus (SLF III) (Parlatini et al., 2017). The input of phonological information to the Heschl’s gyri goes to the Wernicke’s area (P) and then enters the arcuate fasciculus/SLF III to the motor and premotor hand area (H). The LOC and the orthographic lexicon (O; VWFA) have a direct connection to the intraparietal sulcus area (G/K) via the vertical occipital fasciculus (Jitsuishi et al., 2020; Jitsuishi and Yamaguchi, 2020). The intraparietal G/K also has an interactive connection to the frontal motor and premotor hand area (H) through SLF I or II.

Spatial agraphia is the disturbance of graphic expression due to impaired visuospatial perception. It is characterized by (1) the addition of strokes, (2) slanted rather than horizontal lines of writing, (3) writing that occupies the right side of the paper, and (4) the insertion of blanks between a word (Hécaen and Albert, 1978). These features resemble writing impairment due to hemispatial neglect (Seki et al., 1998), which Hécaen and Albert (1978) acknowledged. Thus, in cases of hemispatial neglect, the concomitant agraphia may be called spatial agraphia.

Constructional agraphia is a writing disorder that assigns the spatial location of a word or a letter inappropriately. Errors comprise the omission of a part of a letter and an imbalanced spatial position and size of letters or words, which are referred to as dislocation and disproportion errors (Sakurai et al., 2007). These types of errors are marked in kanji and katakana (derived from a part of kanji) (Ota and Koyabu, 1970), which are graphically complex and require multiple-stroke sequences. Kleist first coined this term and thought it to be accompanied by constructional apraxia, which was attributed the left angular gyrus lesion (Kleist, 1934). However, as Kleist himself acknowledged, constructional apraxia does not always occur alongside constructional agraphia. The above illegible grapheme formation errors are also observed in apraxic agraphia. Thus, constructional agraphia may be a part of apraxic agraphia.

In the English literature, constructional agraphia is considered equivalent to spatial agraphia (Ullrich and Roeltgen, 2011). However, as described above, spatial agraphia results from a lesion in the right hemisphere (Ardila and Rosselli, 1993), whereas constructional agraphia may result from a lesion in the left parietal lobe. The problem here is whether it is appropriate to refer to deformed grapheme formation due to the right hemisphere as “spatial agraphia,” whereas deformation due to the left hemisphere is called “apraxic agraphia.” Because hemispatial neglect also occurs in the left hemisphere, we may designate deformed graphemes in the left hemisphere damage as “spatial agraphia.” Furthermore, if disproportion and dislocation deformities alone are observed in the left hemisphere damage, we may label this condition “constructional agraphia.” These types of deformity can occur when damage involves the posterior intraparietal area and the superior occipital gyri (G/K area in the broad sense in Figure 7).

Researchers have reported isolated alexia with agraphia or pure agraphia following damage to the thalamic nuclei. Both alexia with agraphia (Araki et al., 1990; Ito et al., 2022) and pure agraphia (Aiba et al., 1991; Ohno et al., 2000; Sakurai et al., 2011; Toyokura et al., 2011; Maeshima et al., 2012; Osawa et al., 2013) have similar lesions of mediodorsal and ventral lateral (VL)/ventroposterolateral (VPL) nuclei. Generally, there is greater impairment in reading or writing in kanji than in kana. Some patients with damage to VL/VPL nuclei exhibited poor grapheme formation (Sakurai et al., 2011; Toyokura et al., 2011; Osawa et al., 2013). Studies using single-photon emission computed tomography revealed hypoperfusion in the frontal or parietal lobe, suggesting diaschisis. The precise mechanism has recently been explained by the disconnection of association fibers between the thalamic nuclei and the connected cortical region (Katsuse et al., 2025).

We reported patients with damage to the VL/VPL nuclei who exhibited kanji agraphia with grapheme deformity and micrographia (i.e., a progressive reduction in character size during writing) (Sakurai et al., 2011). Poor grapheme formation and micrographia might be caused by a disruption of the cortico-subcortical motor circuit, which consists of the putamen, thalamus, premotor cortex, and sensorimotor cortex.