As with other posts in this series, this is the results of a test I took today, July 23rd, 2021.
姉 払 関 届 家 顔 残 語 守 有 住 探 例 留 団 作 背 思 師 阪 林 橋 力 普 詳
I got 関, 届, 思, 力, and 普 wrong. That's 84%, which is pretty good. I got nothing wrong from this row because I actually memorized the last 4 kanji yesterday (July 22, 2021 留, 探, 師, and 顔). On Jul 20th, I practiced as usual learning 背, 詳, 住, 例, and 団 with the memory palace technique. On Jul 21st, I learned 橋*, 払, 阪, and 守 with the same technique. The 5 misses are everything to do with losing previous rows because I haven't been practicing them. So this needs to be fixed with a more strong approach than I've been using so far. Look forward to my adventures in fixing my scripts to make my learning more functional.
* 橋 [はし] bridge is pronounced with a pitch accent. Without proper pitch accent, this can be confused with 箸 [はし] chopsticks. How do you pronounce 橋? Low to high. So ha (low) shi (high). Chopsticks is the opposite. Note that 雨 [あめ] rain and 飴 [あめ] candy** have this same relationship. If you have JapanesePod101.com, you can read more here in Perfecting Pitch Accent in Japanese.
** Candy (飴 [あめ]) isn't in kanjidic. Any idea on why that is? We can see that 菓 [カ] candy/cakes/fruit is the only kanji in kanjidic that is translated to candy. Does that mean that kanjidic is incomplete or that the 2900 kanji in kanjidic are limited to a subset of useful kanji? The thing is that if it's not in this list, it shouldn't be used in newspapers. So that just means that 飴 is relegated to culture and must either be combined with furigana or repeated in hiragana or katakana.
This is not at all uncommon. For example let's skim Gunnm 銃夢 Last Order for a moment to find a kanji that uses furigana. Page 7 has furigana for 錆, but let's go further. ナノ技術 has the furigana (テク) because they are talking about nanotechnology, which is colloquially and often scientifically referred to as "nanotech". But those are kanjidic kanji. 花崗岩堂 has furigana グラナイト・イン that is the Japanese loaning of English phrase Granite Inn. Then there's 末梢 which has furigana まつしょう which translates to peripheral. These are also kanji from kanjidic. So why do these kanji get furigana? Because 梢 and 錆 are "grade 9 kanji". Grade 9 is not a real grade, grade 9 in kanjidic just marks these kanji as not being learned in Japan and thus must be accompanied with furigana or repeated in hiragana or katakana whenever they are used. Cool, huh?
- Row 11 Test July 20, 2021 87%
社 研 英 目 所 長 究 切 声 森 教 母 客 野 私 外 借 会 貸 勉 赤 代 姉 宅 家 近 席 度
- Row 12 Test July 21, 2021 75%
二 調 止 長 大 定 茶 工 野 元 買 院 部 室 昼 曜 酒 女 違 私 気 飯 上 彼 知 歩 願 丁
- Row 13 Test July 22, 2021 75%
内 好 持 鳥 借 天 特 二 方 百 洋 書 杯 研 海 男 料 米 理 連 君 明 魚 物 子 全 初 肉
- Row 14 Test July 23, 2021 62%
都 参 開 小 食 寒 真 困 写 家 道 寄 様 暑 生 戸 閉 飯 川 変 置 忘 魚 楽 気 年 帰 消 着
On Row 11 I got 近, 席, 度, and 念 wrong. These are just difficult kanji to remember.
On Row 12 I got 定, 茶, 院, 部, 昼, 飯, and 彼 wrong. In the test just minutes before, I got 願 and 丁 wrong, which allowed me to remember them for this test. This is by a significant margin the worst test so far. Hopefully this is not a preview of rows to come. I blame part of this on lack of sleep, but most of it is losing bits of kanji. For 茶, 飯, and 彼 I was pretty close. I made the 昼 / 午 mistake again. This may haunt me for a long time. 院 and 部 are genuinely difficult. 定 is difficult to remember, but easy to draw. I think this row and the last row accurately reflect how I am forgetting kanji: synonyms (昼 / 午), loss of detail (茶, 飯, and 彼), and genuinely difficult ones I only actually memorized for a day (席, 度, 院, and 部). That's not to say that 茶, 飯, and 彼 are not difficult to remember or draw. It's that they are so frequently used in material that reading it gives strong recognition and should form a memory of how to draw. Remember that I drew 彼 on real paper in row 13. That should have formed proper memories, but it didn't. What this means is that there are going to be ~20% of kanji that I lose based solely on the method I am currently employing to learn them. I'm not losing recognition of most kanji (I can't tell but it's probably 5%), but I'm definitely losing ability to draw them. As soon as I start memorizing these kanji specifically, it's likely that I'll be able to remember them more long term.
So type 1 is synonyms, type 2 is loss of detail, type 3 is genuinely difficult ones I only actually memorized for a day.
On Row 13 I got 鳥, 特, 杯, 理, 君, 物, and 初 wrong. Using our new type system, we can see that 鳥 is 2, 特 is 2, 杯 is 2, 理 is 3, 君 is 3, 物 is 1, and 初 is 2. Note how I drew 村 for 杯, an extra pair of legs 八 for 特, I drew 正 (correct) for 理 (logic), I drew 事 (matter/thing) for 物 (thing/object/matter), and I drew 切 (cut) for 初 (first time) because they use the same radical on the right. The only one I failed to draw anything for was 君 (mister/you) which I remembered the reading for 様 (Esq./way/manner/polite suffix) but not how to draw it. This consistent 75% is proof that I need to deal with both type 2 and type 3 errors. Type 1 errors are just bound to occur and aren't bad for now.
On Row 14 I got 都, 参, 真, 写, 道, 寄, 様, 置, 忘, 帰, and 着 wrong. This is the worst I've done by a fairly large margin (62% 4 more wrong than row 13, which is tied for my worst test), so I'm going to stop and work on row 14 until I get at least 80%. 都 is type 1, 参 is type 2, 真 is type 3, 写 is a cross between type 1 and 3, 道 is type 1, 寄 is type 3, 様 is type 1, 置 is type 3, 忘 is type 2, 帰 is type 2, and 着 is type 2. These daily tests have been getting worse and worse, so it's time to work on them.
If you've gotten this far you're probably interested in what I've been doing recently. I recently wrote two command-line scripts to help in translation. The first script searches kanji from radkfile for radicals. This is a replacement for Kiten's kradselect, so I've called it kradselect1.py. The second one decomposes kanji into its radicals using the same idea but backwards. I call this kdecompose1.py. You can see them in action below.
python3 kdecompose1.py 探 探 has 儿 冖 扎 木 python3 kdecompose1.py 消 消 has 尚 汁 月 python3 kradselect1.py 冖 尚 當 儻 礑 蟐 学 輝 堂 螳 撹 棠 賞 鐺 嘗 常 栄 瞠 覚 鴬 蛍 蟷 襠 労 裳 甞 黨 償 営 党 掌 python3 kradselect1.py 月 尚 消 峭 哨 靜 肖 逍 誚 銷 削 屑 鞘 蛸 宵 霄 梢 悄 硝 鮹 趙 胱 稍 脳
There is still some work to do to get it to prioritize those kanji in lower grades, but that should be easy to do. These scripts are 30 and 33 lines respectively, depending on my 100 line library which provides full access to radicals and grades.
Yesterday I went for an hour long walk listening to JapanesePod101.com lessons and came up with a plan to learn all G1-8 kanji to 90% recognition. I don't need 90% writing skill, 80% of G1-3 kanji would be plenty, which is where I'm going right now (Sept 17, 2021 is my prediction for G1-3 + all the kanji from this course I'm currently on, but that assumes 6 days per week which I have not signed up for yet). So when will I finish G1-8? At this rate, Nov 2, 2022. It's pretty clear that this rate is slow and the number of kanji in G8 is high. How many kanji are in G8? 1130. How many have I learned already? 17. In case you're curious as I am, here they are: he 彼, marriage 婚, slope 阪, slow 遅, overnight stay 泊, refreshing 涼, busy 忙, difference 違, heart 奥, token 符, counter for cupfuls 杯, exchange 替, pay 払, universal 普, cloudy 曇, trust 頼, and detailed 詳. Notice anything about them? They are important words. Ones that you wouldn't want to wait years to learn. This sort of thing actually makes a lot of sense and since I've always possessed the tools to make it happen, when I learn grade 8 kanji (assuming I do), I will organize them by priority and simplicity. Which should be prioritized first? If I sorted by strokes, I'd be learning easy kanji that have a lot less value. If I sorted by common radicals, I'd have easy kanji that have little value. If I sort by common usage in large corpus, I'd be able to learn the most valuable kanji but the ease of learning would be decreased. So my plan is to have 5 easy kanji and 10 common kanji -- with the 5 easy kanji being picked from the most common, so that I won't be learning unuseful kanji until I've run out of easy kanji to learn. At that point it'll just be a race to the finish line -- or I'll have enough kanji to ignore the rest until I am spending a lot of time translating.
You might have noticed how much effort I've now put into learning Japanese (2 months and 9 days so far). What do I intend to get out of it? It's difficult to say when learning a language whether a long vacation to Japan will be the purpose of learning so much of the language (rapid verbal communication is unnecessary for translating news, manga, or visual novels but useful for watching anime, news, and films). So what about a job translating? Since there are so many native Japanese speakers who know English well, what benefit would it be to have someone who is a native English speaker and knows Japanese at an advanced level? This problem I call the "second-rate translator problem". By learning both languages early, a native speaker of two languages has an advantage over the person who learns the second language later. So first rate translation requires this type of first-rate translator. So who wants a second-rate translator? Someone who cannot afford a first-rate translator and cannot use machine translation. In my home town there was a large population of immigrants rapidly introduced. My mother helped translate a few things for them. The number of translators available for their language in the city we lived in was insufficient because of a cultural phenomenon. This is true today, so any language you learn where there's a large immigrant community will provide an opportunity for a second-rate translator if you wish to help. There isn't a very large immigrant community of Japanese who don't speak English (or English speakers in Japan), so that won't apply to me. It does bother me a bit that I don't have a great plan for when I'm pretty good at Japanese. My whole life I've been a beginner and intermediate at languages (German at age 16, Japanese at 25, Brazilian Portuguese at 29, Spanish at 32, Dutch at 38, Latin at 39) never actually getting to a point where I am conversing but frequently learning. I've been able to use my knowledge of these languages, just not regularly.
One thing I realized is that long, long before I finish Grade 8, I'll finish JapanesePod101.com lessons (assuming 4 lessons per day 6 days per week, which isn't easy). It should be interesting to see what comes of it. Of course when I finish listening to all the lessons, I can relisten to all the dialogues to improve my listening skill.
Until next time!
I'll hopefully post all tests as I pass them. Wish me luck.
Learning to Read and Write Kanji
Learning to Read and Write Kanji Row 5
Learning to Read and Write Kanji Row 6
Learning to Read and Write Kanji Row 7
Learning to Read and Write Kanji Row 8
Learning to Read and Write Kanji Row 9
Learning to Read and Write Kanji Row 10
Learning to Read and Write Kanji Row 11
Learning to Read and Write Kanji Row 12
Learning to Read and Write Kanji Row 13
Learning to Read and Write Kanji Row 14
Learning to Read and Write Kanji Row 15
Learning to Read and Write Kanji Row 16
Learning to Read and Write Kanji Row 17
Learning to Read and Write Kanji Row 18
Learning to Read and Write Kanji Row 19
Learning to Read and Write Kanji Row 20
Learning to Read and Write Kanji Row 21
Learning to Read and Write Kanji Row 22
Learning to Read and Write Kanji Row 23