Kanji and Kana agraphia in mild cognitive impairment and dementia: A trans-cultural comparison of elderly Japanese subjects living in Japan and Brazil

This study verifies the environmental effects on agraphia in mild cognitive impairment and dementia. We compared elderly Japanese subjects living in Japan and Brazil. Methods We retrospectively analyzed the database of the Prevalence Study 1998 in Tajiri (n=497, Miyagi, Japan) and the Prevalence Study 1997 of elderly Japanese immigrants living in Brazil (n=166, migrated from Japan and living in the São Paulo Metropolitan Area). In three Clinical Dementia Rating (CDR) groups, i.e., CDR 0 (healthy), CDR 0.5 (questionable dementia), and CDR 1+ (dementia) , the Mini-Mental State Examination (MMSE) item of spontaneous writing and the Cognitive Abilities Screening Instrument (CASI) domain of dictation were analyzed with regard to the number of Kanji and Kana characters. Formal errors in characters and pragmatic errors were also analyzed. Results The immigrants in Brazil wrote similar numbers of Kanji or Kana characters compared to the residents of Japan. In spontaneous writing, the formal Kanji errors were greater in the CDR 1+ group of immigrants. In writing from dictation, all the immigrant CDR groups made more formal errors in Kana than the Japan residents. No significant differences in pragmatic errors were detected between the two groups. Conclusions Subjects living in Japan use Kanji frequently, and thus the form of written characters was simplified, which might be assessed as mild formal errors. In immigrants, the deterioration in Kanji and Kana writing was partly due to decreased daily usage of the characters. Lower levels of education of immigrants might also be related to the number of Kanji errors.

Agraphia, or impairment of the ability to write, is classified into central or peripheral types in the classic cognitive model. 1 Namely "central agraphia" means spelling errors in legible words associated with disrupted word selection. "Peripheral agraphia" consists of mechanical distortions of writing. This includes "constructive agraphia," which involves omissions or additions of letters. 2 Recently, the term "dysexecutive agraphia" 3 was proposed for a form of peripheral dysgraphia in which complex aspects of writing, such as planning, narrative coherence, and maintained attention, are significantly disturbed in cases of impairment of executive functions. Frontal lobe damaged patients not only have difficulties in maintaining the effort required for writing, but also in organizing ideas when writing texts.
Language disorders are major neuropsychiatric symptoms of degenerative dementia 4 and Alzheimer's disease (AD). 5 Due to semantic memory impairment, patients with AD show "surface dysgraphia", in which the writing of dictated words with an irregular or atypical sound-spelling correspondence (e.g. blood) is significantly impaired relative to regular words (e.g. bland). 6 This is because irregular words generally cannot be correctly written without knowing the meaning. The majority of errors are thus phonologically plausible renderings of the target words (e.g. honor to ONER). 7 Luzzatti et al. 8 have reported multiple patterns of impairment in AD. However, no studies in a community have reported on dysgraphia of mild cognitive impairment (MCI), 9 which is considered to be the prodromal stage of AD and other dementia. 10 There are two kinds of scripts in the Japanese writing system, i.e., Kanji (logogram) and Kana (morphogram). Generally, Kanji words are thought to correspond to irregular words, whereas Kana words are considered to correspond to regular words in Western languages. We previously reported Kanji-predominant dyslexia, or reading impairment, in advanced AD. 11 However, dysgraphia, especially Kanji writing impairment, should be considered in relation to educational levels and the language environment. In this regard, it would be useful to examine migrants who received education at elementary schools in the mother land before emigrating to another language environment.
Brazil is the country with the largest population of Japanese immigrants. In 1997, we surveyed elderly Japanese immigrants from Miyagi Prefecture, Japan, currently living in the São Paulo Metropolitan Area, and reported on the prevalence of dementia. [12][13][14] The elderly immigrants had received education at elementary schools in Japan and emigrated to Brazil with their parents at the mean age of 14. In mostly the same year, we also surveyed Japanese elderly subjects in Tajiri, Miyagi Prefecture, Japan. 15 The two populations were examined systematically with the same neuropsychological tests. In this study we further analyzed the two sets of data, focusing on writing tasks. This is the first comparative community-based study of agraphia in elderly Japanese subjects living in Japan and in another country with a different language environment. To study the environmental effects on agraphia in mild cognitive impairment and dementia, we compared elderly Japanese subjects living in Japan and Brazil.

Japanese elderly subjects in Miyagi Prefecture
We retrospectively analyzed the database of Prevalence Study 1998 (n=625) in Tajiri, Miyagi Prefecture, Japan. The detailed methodology for selecting subjects was described previously. 15 There were 412 subjects with Clinical Dementia Rating (CDR) [16] 0 (healthy), 168 with CDR 0.5 (questionable dementia), and 45 with CDR 1+ (dementia). Compared to the CDR 0 group, the CDR 0.5 group was older and had a lower educational level, and the CDR 1+ group was older.

Japanese elderly immigrants from Miyagi Prefecture living in the São Paulo Metropolitan Area
We also analyzed the database of Japanese elderly immigrants from Miyagi Prefecture living in the São Paulo Metropolitan Area (n=166). The detailed methodology for selecting subjects was described previously. [12][13][14] They included 104 CDR 0, 49 CDR 0.5, and 13 CDR 1+ subjects, with mean ages of 76.1, 78.4, and 85.5 years, respectively. Compared with the CDR 0.5 group, the CDR 1+ group was older and had a lower educational level. Demographics of the both subjects were noted in Table 1. The immigrants in Brazil do not receive systematic learning. They received Japanese education in Japan before immigration at their mean age of 14 and their language environment had been dramatically changed. In the Japanese community they can use Kanji and Kana for writing letters, but it depends on the subjects.

Spontaneous writing
The spontaneous writing item of the Mini-Mental State Examination (MMSE) 17 was used. An A4-sized sheet of white paper was presented to the subjects for them to write whatever sentence they imagined.

Writing from dictation
The dictation domain of the Cognitive Abilities Screen-ing Instrument (CASI) 18 was used. Subjects were asked to write the dictated words "Watashi wa ie ni kaeritai" which means "I would like to go home." This sentence is written with three Kanji and five Kana characters.

Analyses
The numbers of Kanji and Kana characters were counted and analyzed with two-way (subjects, CDR groups) ANOVA, so did formal errors of characters. We herein defined the "formal" errors as any types of incorrect patterns of Kana and Kanji characters. They included "central agraphia" which means spelling errors in legible words associated with disrupted word selection, as well as "peripheral agraphia "which consists of mechanical distortions of writing.  *Compared to the CDR 0 group, the CDR 0.5 group was older and had a lower educational level, and the CDR 1+ group was older (p<0.05, post hoc after ANOVA). **Compared with the CDR 0.5 group, the CDR 1+ group was older and had a lower educational level (p<0.01, post hoc after ANOVA). CDR: Clinical Dementia Rating. Figure 1 shows the Kanji results for the spontaneous writing task. For the number of Kanji characters, a twoway ANOVA disclosed that there was no subjects (Tajiri vs immigrants) difference (F=0.917, p=0.339); however, there was a CDR group difference (F=5.669. p=0.004). For the formal errors of Kanji, there was no subjects difference (F=0.519, p=0.476); however, there was a CDR group difference (F=4.005, p=0.019). Post hoc chi-square test showed that there were more errors in the migrant CDR 1+ group (p<0.05). Figure 2 shows the Kana results for the spontaneous writing task. For the number of Kana characters, a twoway ANOVA disclosed that there was no subjects (Tajiri vs immigrants) difference (F=0.822, p=0.365), nor did a CDR group difference (F=1.513, p=2.222). For the formal errors of Kana, there was no subjects difference (F=0.713, p=0.399), nor did a CDR group difference (F=0.599, p=0.550). The migrants in Brazil and residents of Japan wrote a similar number of characters. Figure 3 shows the Kanji results for the dictation task.

Dictation task
For the number of Kanji characters, a two-way ANOVA disclosed that there was no subjects (Tajiri vs immigrants) difference (F=0.263, p=0.259); however, there was a CDR group difference (F=28.014, p<0.001). For the formal errors of Kanji, there was no subjects difference (F=0.384, p=0.536); however, there was a CDR group difference (F=5.020, p=0.007). The migrants in Brazil and residents of Japan wrote a similar number of characters, however, there tended to be more errors in the migrant CDR 0.5 and CDR 1+ groups. Figure 4 shows the Kana results for the dictation task. For the number of Kana characters, a two-way ANOVA disclosed that there was no subjects (Tajiri vs immigrants) difference (F=3.356, p=0.068); however, there was a CDR group difference (F=14.396, p<0.001). For the formal errors of Kana, there was a subjects difference (F=21.264, p<0.001); however, there was no CDR group difference (F=2.853, p=0.059). Post hoc chi-square test showed that there were more errors in the migrant with all CDR groups (p<0.01).
"Watashi wa ie ni kaeritai" which means "I would like to go home," of which the "wa" should be written 「は」 pronounced as "HA" as a particle. This was the most common error by the immigrants.