Japanese from Zero (Part 4)

The following is an unedited post created on our Tumblr page. You may find the original here.

Never thought I’d need to write a part 4… But here we are!
Just another gentle reminder that I don’t expect you to memorize all of this. These are just facts you should be familiar with.

Last time, I gave you a run-through of the convoluted way in which verbs are taught in the classroom (commonly) and why. Our approach is more interested in linguistic accuracy than in making you predisposed to any form of speaking. That’s why I’m telling you all these things that classroom students would never actually discuss.

This will be the toughest, most wordy, section, so please be patient. There are a lot of words here, but they’re not inaccessible.

Dictionary Entries, Present Affirmative, and Stems

To understand verbs, we need to understand the stems of verbs. The stem of a verb is a verb without any of the conjugating prefixes. So far, when we’ve mentioned a verb, we’ve spoken of it as its dictionary entry, which is the conjugation for the present tense. Keep that bit in mind.
Let’s look at some verbs:


So, look at these verbs and think to yourself, “What do all these verbs have in common?” The answer, obviously, is that they all end in -u. And that -u is the suffix that makes verbs present tense (note that in Japanese the present and future tenses are one in the same).
The next step for us will be to get rid of the suffix and see what we’re left with.


The words I’ve starred are tricky one’s we’ll address in a moment. The other ones are stems. Strictly speaking, this is what a stem will look like without any of the mechanics of Japanese fiddling with them. However, as I have stated before, Japanese functions on morae (which I suggested you think of as syllables). Most morae need to end in a vowel; and clearly these stems don’t end in vowels. So, Japanese has a stand-in vowel that will fix this: -i.

Matsi* -> Machi

(Matsi* is not what we would see because in Japanese there is no mora that’s tsi*. When /ts/ and /i/ come together, you get /chi/. So you get Machi)

When the -u suffix is attached to create the present tense, that -i goes away, leaving you with the present tense verbs as we saw them first.

These are what the Japanese call the consonant stem verbs, what the textbooks will call -u verbs.

The verbs we starred are those that cause a bit of confusion because they seem to be the vowel stem verbs, also known as the -ru verbs. A vowel stem verb will have its stem end in -e or -i. In this situation, the mechanics will Japanese will create a weak consonant to separate the two vowels (because Japanese has a habit of having one vowel overcome the other and we need the ending suffix to remain). That weak consonant is -r. Because it’s weak, when we remove the present suffix -u, -r goes away as well.

So we really end up with this:


But what about the other two? In the case of *Shaber, the -r is actually part of the stem and so the way you will see the stem is as Shaberi.
In the case of *I, once upon a time there used to be a consonant there (probably something like an -f). Although the consonant is gone, the way the verbs behave remains intact. Meaning that *I takes on the fill-in -i suffix regardless to give us the funky looking Ii.

Of all the exceptions, the ones ending in -er and the ones with any double i go unexplained the most often. Now we have answers.

So now we know what the present tense suffix looks like and we know what the stem looks like and we even know how we get from a stem to the present tense. This is good.

Negative Suffix, and Secondary Present Suffix and Conjugations

Remember when I said I’d get back to how -i adjectives are like verbs? This is where we’ll be able to see that.

In Japanese, verbs take on a suffix that will indicate negation, -ana; and this negation is followed by a secondary tense marker. It’s not -u, but -i.

That first -a in -ana is pretty weak. If the verb’s stem ends in -e, that -e will overcome the -a, leaving you with the verb ending as -enai, the -e being part of the stem and the -nai being what’s left of the negation, of course.

In the case that the stem itself ends in any other vowel (for which the present tense suffix will be a simple -u; that’s just an easy way to remember it), so that no more vowels are overcome (which would cause us more problems), the buffer semi-vowel -w comes to the rescue and places itself between -i and
-anai. Then -w overcomes -i (because that -i is so, so weak), leaving us with -wanai.


Arai –> *Araianai –> *Araiwanai –> *Araiwanai –> Arawanai
Tabe –> *Tabeanai –> *Tabeanai –> Tabenai
Kiki –> *Kikianai –> Kikanai
Isogi –> *Isogianai –> Isoganai

So on and so forth.

Now let’s talk about that -i at the end of -anai. You remember -i adjectives and their distinctive ending of -i? Well that one and this one are the same exact suffix. How exactly it works is hard to explain and I don’t have all the answers myself. But here’s what everybody should know. The -i suffix itself indicates the present tense. If a verb or adjective takes the -i suffix, it will only conjugate in a limited number of ways. Here’s what we have:

Present affirmative: -i
Present negative: -kunai
Past affirmative: -katta
Past negative: -kunakatta
Conditional: -kereba
Volitional: -karou

If we have a negative verb, then the -i suffix cannot be again conjugated for the negative. So there are no double negative verbs. If it is an adjective, then it will not conjugate for the conditional or volitional forms.

Here’s why things are so tricky. In all conjugations except the present affirmative, there is a -kV suffix, where V is some vowel. Where does that -kV come from? And what exactly happens so that we end up with -kunai instead of *-kanai? These are questions I don’t have answers to, not yet at least. Thankfully, these conjugations, limited to the -i adjectives and the negative conjugations, are few enough for memorization.

Te-Form and Affirmative Past

Speaking of oddities, let’s talk about the Te-form. The Te-form of a verb is similar to the participial form in the Indo-European languages. It’s not exactly like a participle, but it is similar enough to give us an idea of what it can do.

The -Te Form gets its name from the fact that, when conjugated, the verb will end in -Te. At a glance, it looks very irregular. This is why they teach students a song that tells them how the Te form of a verb will look like. Let’s look at the paradigm.

-u, -tsu, -ru –> -tte
-mu, -bu, -nu –> -nde
-ku –> -ite
-gu –> -ide

Believe it or not, any song that teaches this is worth learning. It doesn’t take long and it’s a good reference.
Now, let’s try to simplify things so that we have anything less than 4 long rules.

/ku/ and /gu/ are the exact same syllables except in one thing: voicing. The consonants, /k/ and /g/, are the same except that /g/ requires one to move one’s vocal chords and /k/ does not. For proof, I’ll refer you to the IPA chart where you’ll find /k/ and /g/ as velar (the part of the mouth where the sound is formed) plosives (meaning they’re produced by little bursts).
How does this help us? Well, /t/ and /d/, the only things differentiating /ite/ and /ide/, are voiced/unvoiced counterparts as well, just like /k/ and /d/. You’ll find /t/ and /d/ in the IPA chart as alveolar plosives.
So, what we have here is /t/ in both /ite/ going assimilation. Assimilation is when the features of one phoneme pass onto another.
(We saw this with “*senpai” becoming “sempai” and we see it in English all the time. For example, say the word “dogs”, that /s/ might be written as an /s/, but you’re moving your vocal chords when you say it (unless you’re from Long Island, in which case something else is going on called Hypercorrection) so that it sounds like a /z/.)
So /t/ becomes voiced when /k/ is voice, i.e. when /k/ becomes /g/. So -ku and -gu are exactly the same and their differing -Te forms are due only to assimilation. Cool.

Note that an -i exists in -ku and -gu’s -Te forms? Is that the stem’s -i? Yes. What seems to happen is that the -te suffix enforces that weak -i, which then makes the final consonant of the true stem (the stem without the -i; in linguistics we’d call this the underlying form) go away.


Kiki + -te = *Kikite –> *Kikite –> Kiite
Isogi + -te = *Isogite –> *Isogite –> Isoide (through assimilation)

Now let’s talk about -mu, -bu, and -nu (or /mi/, /bi/, /ni/, depending on how you want to see it). The truth is that there is only one verb that’s stem ends in /n/, which is shini, meaning “to die”. Here’s what I’m hypothesizing: originally there were other verbs ending in /nu/, but they eventually started turning into /mu/, and then some turned into /bu/. /b/ and /m/ represent two sounds that are made in the same place (bilabial) but are different types. /m/ is nasal (you make the sound with your nose) and /b/ is plosive. So what this means in my theory is that historically verbs ending in /nu/, /mu/, and /bu/ will conjugate the same way for the Te-form because once upon a time they were the same. (Also, one may hypothesize that there is a second strain of bone fide verbs that have stems ending in /m/ and that those also got turned into /b/ eventually.)

[The reason I set it up this way is because I can’t see a good reason for there being only one verb with a stem ending in /n/. This can all be explained with /n/ being a standalone thing and /m/ and /b/ being separate. But having them in one bunch is very nice.]

So a nasal is stronger than a vowel. Remember, the only mora that doesn’t have a vowel is precisely a nasal (-n, ん). That -i that was enforced by -te last time will now be eliminated regardless. And finally assimilation (like last time), will take place so that /t/ becomes /d/ because nasal are always voiced.

In the case that you want to see /b/ as a spinoff of /m/ in this case, you will have /b/ turn back into /m/ and then /m/ turn into /n/ because it’s assimilating the alveolar positioning of /t/ and /d/.

Shini + -te = *Shinite –> *Shinite –> Shinde (through assimilation)
Yomi + -te = *Yomite –> *Yomite –> Yonte (through assimilation) –> Yonde (through assimilation)
Yobi + -te = *Yobite –> *Yomite (historic reversal) –> *Yomite –> *Yonte (through assimilation) –> *Yonde (through assimilation)


So, last group: where the endings of the stems are /r/, /ts/, or a vowel (that’s not -e or -i, which can be the so called -ru verbs). These are continuants, which is to say that they are sounds that you can extend for as long as you’d like (as long as there is air in your lungs). So these are going to behave differently from /k/ and /g/; and they are also different from nasals, which are also continuants.

Here the -i that’s added to the stem will go away as well, as with nasals. /r/ and /ts/ are not quite going to disappear, but become a glottal stop. The double consonant in the romanization of Japanese signifies a glottal stop.

In the case of vowels, which don’t go away, I haven’t figured out a way to make a glottal stop appear except as a bi-product of the forces of the stem and the -te, with the -i becoming a glottal stop. Why a glottal stop? Possibly through assimilation because glottal stops are considered plosives, like /t/.

Machi + te = *Machite –> *Machite –> *Machte –> Matte (through assimilation)
Shaberi + te = *Shaberite –> *Shaberite –> *Shaberte –> Shabette (through assimilation)
Arai + te= *Araite –> *Aratte (through assimilation)

And the -ru verbs, the ones with stems ending in vowels (and there was no consonant that historically got deleted), you just add -te and that’s it.


Tabe + te = Tabete

So that accounts for most things. The hypotheses proposed aren’t quite perfect, but they get the job done.

For the affirmative past, it’s the exact same story as the Te-form except that the last vowel is not -e but -a.

So we have Kiita, Isoida, Shinda, Yonda, Yonda (realize now that I used two verbs that end up as homophones), Matta, Shabetta, Aratta and Tabeta.

Potential Form

(The potential case describes an action as being able to be done, not actually being done. It’s the difference between “I speak English.” and “I can speak English.”)

Much, much, much, much, much, much, much simpler than that royal mess you just saw.

The suffix is -eru. For the so called -u verbs (which are those with stems ending in consonants and vowels ), you just add -eru and you’re good to go. As tends to happen, the -i added to the stem will go away. It’s just really, really weak.

For the -ru verbs (the ones ending in -e or -i and historically do not have a consonant ending that’s gone away), we end up with double vowels, to which the mechanics of Japanese responds by putting in a whole mora. Why a whole mora? Because of that habit I told you about where vowels are trying to overthrow one another. The mora inserted is -ra. /r/ is that same weak /r/ in -ru and /a/ is chosen for unknown reasons (probably because it’s strong). And then, to avoid yet another fight between vowels, a consonant /r/ (same as before) is placed between -ra and -eru, creating -rareru.


Kiki + eru = *Kikieru –> Kikeru
Tabe + eru = *Tabeeru –> *Taberaeru (a buffer insertion) –> *Taberareru (another buffer insertion)

That -u at the end is your present tense suffix, which behaves just as we’ve described it before, keeping in mind that that /r/ is one of those buffers that will go away when going conjugation (in layman’s terms, it works like an -ru verb)

Passive Form

Comparable to the potential form. The suffix is -areru.

For the -u verbs (which you are already very familiar with), you add -areru, the -i added to the stem goes away, and are pretty much good to go. The only exception to this is the same as with the negative, where a -u verb ending in a vowel (which will be either /a/, /o/, or /u/) will bring in the buffer semi-vowel (which means it lies somewhere between a consonant and a vowel) /w/ before -areru.


Arai + -areru = *Araiareru = Arawareru (buffer insertion)

For -ru verbs, we have another case of vowels coming together and Japanese trying to separate them with /r/.


Tabe + -areru = *Tabeareru –> Taberaru (buffer insertion)

So yes, the passive and potential forms of -ru verbs look exactly the same.

Causative Form

(Causative is a valency operation, a operation that affects the agents. “Billy melted the ice.” vs. “Billy caused the ice to melt.”)

The ending is -aseru. Just like last time (are you seeing a pattern here?) you will see the insertion of a /w/ for the -u verbs with vowels. In the case of -ru verbs, the added consonant won’t be /r/ but /s/. Why /s/? Probably because of euphony (literally, because it sounds nicer).


Arai + aseru = *Araiaseru –> *Araaseru –> *Arawaseru (buffer insertion)
Tabe + aseru = *Tabeaseru –> Tabesaseru (buffer insertion)

Imperative Form

(The imperative mood is when you command someone to do something: “Eat!”)

The suffix is -e.
But things get funky with -ru verbs, where the suffix is actually something else: -o. Why? Because lots of -ru verbs end in /e/. So to avoid double /e/ another vowel has to be used in a process called dissimilation (it’s like assimilation but opposite). So now we have -o.

However, something funny happens with -o, namely that we need a buffer between the vowels of the ending of the stem and the suffix -o and there are two of them. One is /r/, which is what we’d expect by now. However, /y/ is also used, probably because it’s used for the volitional form as well and they are similar in meaning.


Arai + -e = *Araie –> Arae
Tabe + -e = *Tabee –> *Tabeo (dissimilation) –> Tabero OR Tabeyo (buffer insertion)

Volitional Form

(It’s like the imperative in that it is a mood, but it expresses desire, not command. It’s also known as the presumptive or the hortative)

The ending is -ou. It’s just like the imperative form except that the -ru verb’s buffer insert is only /y/.

Conditional Form

(This is a different case entirely from the indicative. Instead of being indicative, this is somewhat like a subjunctive or optative. If you do not know what either of those are, in English it’s created periphrastically by “would X” where X is a verb.)

The ending is -eba. For -ru, verbs, the buffer insert is /r/.


Arai + -eba = *Araieba –> Araeba
Tabe + -eba = *Tabeeba –> Tabereba (buffer insertion)

There is a second conditional form with is simply the past affirmative + ra:

Tabetara, Arattara, Igoidara, etc.

And that’s right about it in terms of conjugation.

Congratulations, we’re done.

Next time we do a strict grammar lesson, we’ll be talking about the periphrastic tenses.
My next post will probably be a funny scene from K-On! that teaches a few vocabulary words. After all this, best to keep things short.