Monthly Archives: November 2016

Why I love Hebrew

After the last post I feel obliged to say a few words about Hebrew as well. And as it will probably turn out, there will be more than a few by the time I finish writing this.

Japanese is like art. You look at it, you’re fascinated with it, and you may see as many meanings in it as your imagination will allow you. Hebrew, on the other hand, is like science.

In many ways, Hebrew is very similar to Western languages. It has a proper alphabet (no vowel letters or capital ones, though), it has spaces (unlike Japanese), it has grammar and spelling rules and all that stuff. What makes it special (along with Arabic, probably) is the structure of a typical Hebrew word, I mean mostly nouns, adjectives and verbs here.

A typical Hebrew word consists of a pattern and a stem. In most cases the stem is either three or four letters. The pattern is whatever left after you take out the stem, or at least in the simplest cases. Let’s take the word “shalom”, for example, which literally means “peace” or “well-being”, but often just “hi”. There are different ways to write it. The most commonly used one is


(Read right-to-left.)

However, when learning Hebrew, it makes much more sense to write it as


Note all the three differences: a tiny dot above ש, a T-shaped mark under it and a tiny dot above ו. The first dot is not very interesting: it’s just that there are two different letters that look almost the same: shin (שׁ) and sin (שׂ). These dots just mark the difference, and normally they are omitted, just like you often omit various diacritics in English loanwords like “fiancee”.

The other two marks are much more interesting. They are vowels. Turns out, Hebrew predates vowel letters, and due to religious reasons it was not feasible to add them later on. So diacritic marks known as nekudot ([neh-koo-DOT], meaning just “dots”) were invented. The T-shaped one is “a”, and the dot above ו is “o”. A funny thing, even though the letter ו (vav) is normally pronounced as “v”, the presence of that dot makes it silent, therefore turning it into a vowel letter (the letter itself is silent, but the dot is still pronounced as a vowel). There is a good reason for this: before nekudot were invented, some consonant letters were used as both consonant and vowel ones. The most commonly used are ו (v/o/u), י (y/i) and ה (h/a/e, but can only be silent at the end of a word, where it typically is silent). But sometimes י stands for “a”, or ה stands for “o” at the end of a word or other weird things happen. This weirdness comes from religious texts too, which can’t be changed, so the language had to adapt to reflect the correct pronunciation.

But that all is annoying and not really fun. What’s important is that this word has a stem, namely ש-ל-מ. Never mind that ם turns into מ and vice versa. ם is just the right way to write מ at the end of a word, and that’s purely a graphical feature specific to just five letters (of 23). But since ש-ל-מ is not exactly a word, the standard forms of letters are used. So both שלום and שלומך have the same stem.

What’s the pattern then? In Latin, it could be written as 1-a-2-o-3. In Hebrew, it’s kind of hard to write because you can’t write nekudot without letters, and all nekudot are a part of the pattern (except the dot from the letter shin, which is actually a part of the letter and therefore belongs to the stem together with the letter). The most common way to write patterns is to put some random stem in it, preferably a one that doesn’t distort the pattern (which is a pretty common occurrence known as gizra [geez-RAH]). The usual stem is ק-ט-ל, which is a bit uncomfortable because its meanings are associated with killing, but it works the best as the letters are very gizra-neutral and don’t wreak havoc on the pattern.

The order of the letters is important. It’s also a very rare thing of the pattern letters to get intermixed with the stem letters. In fact, there is only one pattern where that happens, and there are strict rules for that one. Note, however, when I say “letters” here, I mean full-fledged consonants. וֹ in שלום is no longer a letter in that sense and therefore it jumps in between ל and ם quite easily.

The whole stem-pattern business has enormous impact on the learner. In a random Western language, when you see an unfamiliar word, you can sort of guess its meaning if you know its stem. Say, if you know what “white” means, you can probably guess the meanings of “whiting”, “whiteness” and similar words. Hebrew goes much further because sometimes even apparently unrelated words share a common stem or a pattern. Say, you know that פיגוע [pee-GOO-ah] is a terrorist act. And then you see פגיעה somewhere. You instantly recognize the stem, and knowing that the pattern קטילה [ktee-LAH] usually means some action, but not as intense as the pattern קיטול [kee-TOOL], you deduce that it must be some other bad action, but not as bad as a terrorist act. An assault perhaps, or some other kind of offense. Then you look at the context:

מה לעשות במקרה של פגיעה משריפה?

Even with my poor Hebrew, I can read that roughly as “What to do in a mikra of pgia from fire?” Now I know, as I’m reading this on a news website, that there is a lot of fires in Israel now, probably due to arsons and dry weather combined. Given all that, I’d guess that “mikra” is something like “situation” and “pgia from fire” is something like an arson, and it’s probably an article explaining what to do if you suspect someone of starting a fire or if you witnessed it, or something along that lines.

Now, I’m fairly sure that פגיעה is pronounced [pgee-AH], but I’m not sure at all about מקרה. It could be [mee-KRAH] or [mah-KREH] or anything. The lack of nekudot doesn’t help to figure out the correct vowels, but knowledge of the patterns does. If I knew what pattern that is, I could probably guess pronunciation with enough confidence, just like I did with פגיעה because it’s a pretty distinguishable pattern, thanks to the way how י and ה are placed.

I really just made up all this by just looking at a random news website just now. Now let me consult my dictionary and see what I got right.

First, מקרה is pronounced [mee-KREH]. Almost got it right. I guessed the meaning right from the context, it’s really an occurrence, or a case. “מה לעשות במקרה של” [MAH lah-ah-SOT beh-mee-KREH SHEL] is literally “what to do in case of…” And I would have been even more sure if I knew how the phrase [MAH kah-RAH] (what happened?) is spelled (that’s מה קרה, but I’ve never seen it before, only heard).

Second, פגיעה is a blow (a physical or a moral one), a hit, damage, an injury. And looking at the further text, it seems I was wrong about arsons after all. A more appropriate translation would be “What to do when fire hits?” or “What to do in case of fire injury?”, which kind of makes sense, as I think, because any reasonable person probably already knows that in case of a suspected arson the only right thing to do is call the cops, but many people don’t know everything that should be done in case of fire or fire injury.

Still, as you can see, even as I encountered two totally unfamiliar words, I was able to guess their meanings to some extent.

Thanks to gzarot (plural of gizra), though, this exercise becomes much more complicated. In this case, their effect is minimal (that “e” at the end that I got wrong). But there are cases when a mere presence of a certain letter in a certain position breaks everything. Say, words like מדבר [meed-BAR], מגדל [meeg-DAL] and מערב [mah-ah-RAV] belong to the same pattern, but see how the last one is pronounced? All thanks to the letter ע which doesn’t like “i” before it and the lack of a vowel after it, but absolutely loves the “a” vowel on both sides. Gzarot get even worse with words like שומר [shoh-MER] and גר [GAR] which also belong to the same pattern… or, rather, would belong to the same pattern if it was possible to squeeze the root ג-ו-ר into it. But since it’s impossible (for no reason), the middle letter of the stem just disappears and the whole pattern is changed so drastically, that it’s easier to say it’s replaced by another pattern altogether. Thankfully, gzarot are rules, not exceptions. Meaning that if it works that way with this particular stem and this particular pattern, it will work the same way with another stem/pattern pair that satisfies the same condition. If we take the same “shomer” pattern and the  ד-ו-ר stem, we know it’ll be דר [DAR] and nothing else. Sometimes the rules are ambiguous, though, but they still usually limit the possibilities to two or three.

And yet, with all this weirdness, the most of the language vocabulary suddenly turns from a list of seemingly random words into a nice table of stems (about 4000 of them) and patterns (about 200), and a set of gzarot to learn. That’s an order of magnitude less than tens of thousand words for a typical random language. What’s especially good about it is that once you know a word, it’s easy to restore both spelling and pronunciation. If you remember the word קרה from the phrase מה קרה and then you remember the pattern מגדר, then you may misread the word מקרה as [mee-KRAH], as I did above, but even if you don’t remember how it’s pronounced at least you’ll get the spelling right (without the vowels), so you won’t get [mee-KHRAH] or [mee-GRAH] because you know the stem. And once you’ve seen that type of a word, you’ll figure out the gizra, and then you can apply it to different words as well. Now when I see מפנה, I know it’s [mee-FNEH], and not [mee-FNAH]. Ditto for משנה [mee-SHNEH]. Apparently the gizra here is that when a stem with the third letter י (which often turns into a silent ה at the end, just like here) goes into the pattern of מגדל, the last vowel changes to “e” (before the silent ה):

מִגְדָּל ← מִקְרֶה

(The one dot is “i”, the three-dots triangle is “e”, the two dots are silent in this case, and never mind the dot in the middle of ד).

See how learning separate words turns into learning whole groups of words. What’s even better is that different gzarot often don’t interfere with each other. If I know that ע likes “a” around it and the silent ה “e” before it in this pattern, I can easily read מערה as [mah-ah-REH].

Then there are various suffixes and endings that can be attached to a word, and that could change the pattern as well, but that’s another story. It’s enough to say that there are rules for that too. And learning rules is much more pleasant process than learning seemingly random sequences of characters that are called meaningful words for some reason that escapes me completely.

Why I love Japanese

I absolutely love foreign languages, and I’m also exceptionally bad at them. It took me, what, like 20 years to get my English to this level, and that’s mostly thanks to practice. But I just had to. I used to play a lot of computer games, and Russian translations were so awful that it was easier to play in English even when I didn’t speak it. I used to read a lot of texts about programming, and they were naturally in English too. Then I had to talk to my European colleagues on phone with no interpreter. Then I watched anime with English subtitles because Russian were either unavailable or, again, ridiculously awful. So, yeah, I had to pick some English skills on the way.

Other languages are different. Like most programmers, I’m very lazy. I’d rather make my computer do my work for me, but, naturally, it can’t learn languages for me, and even if it could, it wouldn’t help me either. So… my usual way of learning a language is to find some learning material, ponder over it for a few minutes (or hours, if I feel really stubborn) and then just forget about that particular language for half a year or so. Naturally, it doesn’t work very well.

But… even when things are that way with German, French and Irish, they are just a little bit better with Japanese and Hebrew. Hebrew aside for the moment, I’d like to say a few things about Japanese.

In case you didn’t know, Japanese has two writing systems: kana and kanji. Kana is divided into two subsystems again: hiragana and katakana. Kana is a phonetic system with pretty much one-to-one correspondence between syllables and characters. There are few exceptions, but they are pretty well-defined too. Note that I say “syllables”, not “sounds”. That’s because Japanese doesn’t have separate sounds, so even if you really wanted, you couldn’t make up a word like the famous Russian “взбзднуть”, which roughly reads as “vzbzdnut’”. That’s right, six consonants at the beginning (and as far as I know, some Slavic languages even have words with no vowels at all). In Japanese you only have syllables that sound like consonant-vowel or just a vowel. And there’s also “n/m”, which is a separate syllable that is pronounced as “n” in most contexts and as “m” in the few remaining. This feature of Japanese language makes it sound very beautiful. It also makes it rather hard to adapt loanwords to it (lack of certain sounds doesn’t help either). Now you know why “the world” sounds like “za waarudo”. It goes like this:

  • Change ”th” to “z” because it’s the closest sound.
  • Change “e” to “a” because it doesn’t sound in English like “e” at all, and “a” is again the closest sound.
  • Change “or” to “a” for the same reason. Japanese does have long vowels, though, hence the “aa”.
  • There is no “L”, but Japanese “r” sounds very much like it. But there are no separate vowels, so it turns into “ru”. That’s because the “u” sound is considered the least noticeable vowel (sometimes it isn’t pronounced at all).
  • There is the “d” sound, but, alas, the syllable “du” sounds more like “zu”. That’s why “o” is used as the extra vowel here instead of “u”.

In case you wonder, hiragana and katakana differ only in their appearance, pretty much like printed letters differ from written ones. Hiragana is mostly used for regular writing (mixed with kanji), and katakana is typically reserved for loanwords and design purposes (banners, ads and so on).

And now we’re getting to my favorite part. Kanji. They are originally Chinese, and most of them still look identically to their original counterparts (I know of only one exception). They typically have several different readings, mostly grouped in two groups: original Chinese pronunciation, adapted to Japanese, and Japanese pronunciation, corresponding to the same word in Japanese like it was before there were any kanji at all. That’s why 道 can be pronounced as “dou” (read as “door” because “u” after “o” makes it long) or “michi”. To make things even worse, kanji in names often have totally different readings, which makes it next to impossible to read a Japanese name unless it comes with hiragana transcription known as furigana.

But that was the bad part. The good one is that, unlike most other languages in the world (except Chinese, naturally), Japanese words actually do make sense. Think about it. Does the word “two” make any sense to you? Why “two”? In Russian it’s “два”, which suggests a common origin. Without doing an etymological research, we can hardly say what that origin is. But even if we know it, we still can’t answer the basic question: why on earth does it sound like that? Why not “boom” or “swin”? Surely such a common word must be short, but there is no saying why it should sound or be written this way or that way. In Japanese, there is no reason why the word sounds like it does, but there is a good reason why it should be written like this:

It shouldn’t surprise you much that one, two, three are written as 一、ニ、三. Of course, things are not always that simple. For example, I have no idea why four is written as 四. There is certainly an etymological reason, but who needs one when you can clearly see two strokes in a square, and two squared is certainly four! So unlike Western languages, which lose original word meanings over time, kanji actually gain new meanings, and it’s up to your imagination how many explanations you can find.

It gets even better. Most kanji are actually composed of elements, and every element has its own meaning. There are just a couple hundred elements, as compared to about two thousand kanji in Japanese alone (much more in Chinese).

It makes learning Japanese (well, the kanji part at least) a very unusual thing. Instead of the usual cramming, you may ponder a lot over every character, trying to find new meanings behind it and see its internal beauty. There are many works already done on this topic, for example Henshall mnemonics or a very good book in Russian by Adyl Talyshkhanov. But there’s one thing about them: you gain much more by doing the same work yourself than by reading explanations written by someone else. And even if you do, it’s better to rediscover meanings again and again, rather than write them down once and refer to them each time. I’ve written down a lot of them (in Russian), but hardly remember any. For example, I encountered this word recently:


The characters correspond roughly to “super”, “ability” and “strength”. The most literal (and pretty appropriate) translation would be “superpower”, as in ESP and all that kind of stuff.

I have no problems whatsoever remembering the 力 part. It’s purely graphical: just imagine a strong man bending his arm to show off his muscles. However, the first two characters often leave me puzzled.

Let’s take 能 for example. It consists of 厶 (I, myself), 月 (month, moon, but Talyshkhanov sometimes refers to it as “meat”) and two of 匕 (which my dictionary lists as “spoon”, but it looks very much like a sitting person as well). Now, Talyshkhanov explains this riddle as “some people have the ability to see plain meat in this kanji, others have the skill to see a romantic crescent moon”. Ability, skill, talent—these are the meanings of this kanji. I don’t like that explanation much, and not just because it lacks “myself”—it can be easily reworded along the lines of “I may have the ability to see a moon here, but those people are just sitting there seeing nothing more than just meat”. But it just doesn’t click either way.

But I seem to have difficulty trying to find other explanations. The good part is, the more difficulty you have, the more time you spend on it, the better you remember it in the end. This process may get very enlightening. Say, forget about sitting people and remember the original “spoon” meaning. Now I can say that I have the ability to eat the Moon with two spoons. This is ridiculous enough to easily remember, but isn’t much enlightening. Or I could think how I worked for a whole month with two spoons to develop some ability. Or maybe I worked on my abilities so hard that I only ate two spoons in a whole month. Repeat that until just looking at this character makes you think of various abilities, even if it takes you spoon-feeding yourself with such nonsense for a whole month.

The first character, 超, is even more ridiculous. It consists of 土 (soil, earth, ground), something that isn’t considered a separate element, but looks suspiciously like a shoe or like a road from the 道 kanji, and then there are 刀 (sword) and 口 (mouth). If that is of any help, the ground combined with the “shoe” makes 走 (to run), and the sword combined with the mouth makes 召, which has ridiculously many meanings, such as seduce, call, send for, wear, put on, ride in, buy, eat, drink and even catch cold. For a laugh, try to imagine how you would do all that things with a sword and a mouth, like seducing someone with your mouth, holding a sword in your hand to make sure that your words come through.

That, however, doesn’t help us understand why 超 stands for transcend or super. Of course, one may think that running is walking super-fast and putting a sword in your mouth transcends all boundaries of reasonable, but it still lacks something. It’s a bit more reasonable to imagine a samurai with a sword, running and shouting something. Such a picture certainly suggests that he transcended a certain level of samurai-ness.

The whole word put together leads us to thinking of a strong superman with a sword running on the ground in moonlight, holding a couple of spoons in his mouth, because apparently it’s his special skill, or something like that.