japanese verb conjugation the simple hard way - underreacted

japanese verbs are awesome. you can take the dictionary form of a verb (for example, "taberu", to eat) and turn it into many forms:taberu ("eat"): tabemasu (polite "eat"), tabemashita (polite "did eat"), tabemasen (polite "not eat"), tabemasendeshita (polite "did not eat"), tabenai (casual "not eat"), tabenakatta (casual "did not eat"), tabetai ("want to eat"), tabeta (casual "did eat"), tabete ("eat and...").nomu ("drink"): nomimasu (polite "drink"), nomimashita (polite "did drink"), nomimasen (polite "not drink"), nomimasendeshita (polite "did not drink"), nomanai (casual "not drink"), nomanakatta (casual "did not drink"), nomitai ("want to drink"), nonda (casual "did drink"), nonde ("drink and...").the art (or science) of making these forms correctly is called conjugation. i've tried to learn Japanese verb conjugation a few times before. at first, it looks simple (you just swap suffixes!), but there's a lot of nuance that can drag you down as a learner.i found a system i prefer but let me first explain why i struggled.how conjugation is usually taughta common way to explain conjugation looks like this:1.there are two groups of verbs (often called -ru and -u verbs).2.for each suffix (like -masu), you have to learn how to correctly attach it to a verb of either group (using tables like these).i found this approach to teaching deeply frustrating and unsatisfying.first, calling them -ru and -u verbs is confusing because you can't always tell the group by how the word ends. sometimes it's what you expect ("taberu", to eat, is a -ru verb; "nomu", to drink, is an -u verb). okay, so if something ends with -ru then surely it's a -ru verb?

actually, not necessarily! "kaeru" (to go home) is an -u verb.well, shit.if groups alone weren't confusing, take this: then you have to memorize a dozen of seemingly arbitrary rules like "く (ku) becomes き (ki) before masu" for different suffixes. some textbooks further separate the -u verbs into five subcategories to teach these rules.this makes japanese verb conjugation seem hard!however, beneath these surface difficulties there is a surprisingly elegant system that's obscured by how it often gets explained. in this article, i'll explain it in a "simple hard" way. this means that instead of memorizing a bunch of individual special cases, we'll try to slowly build a coherent system from a small number of primitives.it's just concatenationlet's go back to these two examples:taberu ("eat"): tabemasu (polite "eat"), tabemashita (polite "did eat"), tabemasen (polite "not eat"), tabemasendeshita (polite "did not eat"), tabenai (casual "not eat"), tabenakatta (casual "did not eat"), tabetai ("want to eat"), tabeta (casual "did eat"), tabete ("eat and...").nomu ("drink"): nomimasu (polite "drink"), nomimashita (polite "did drink"), nomimasen (polite "not drink"), nomimasendeshita (polite "did not drink"), nomanai (casual "not drink"), nomanakatta (casual "did not drink"), nomitai ("want to drink"), nonda (casual "did drink"), nonde ("drink and...").can you spot any patterns between these two?if you focus on the first example, conjugation looks straightforward.

yes, it's a lot of suffixes, but each is added by pure concatenation:tabe + ru = taberutabe + masu = tabemasutabe + mashita = tabemashitatabe + masendeshita = tabemasendeshitatabe + masen = tabemasentabe + nai = tabenaitabe + nakatta = tabenakattatabe + tai = tabetaitabe + ta = tabetatabe + te = tabetewe're going to call the left part — the part that is unchanging between these — the stem of our verb. taberu's stem is "tabe".so now you (almost) know how to do conjugation:To conjugate a Japanese verb, append the suffix to its stem. For example, tabe + masu = tabemasu.in a sense, this is pretty much the entire principle! but now we need to refine our understanding of both a "stem" and of "appending".a wildcard vowelin the first example, the stem was very simple: just tabe.let's now have a close look at our second example:nomu ("drink"): nomimasu (polite "drink"), nomimashita (polite "did drink"), nomimasen (polite "not drink"), nomimasendeshita (polite "did not drink"), nomanai (casual "not drink"), nomanakatta (casual "did not drink"), nomitai ("want to drink"), nonda (casual "did drink"), nonde ("drink and...").can you spot any patterns here?there's a few things that come to my mind:1.it's not entirely clear what nomu's stem is. is it nomu? but then, we define stem as the unchanging part — whereas the last vowel seems to alternate, like nomi or noma in some cases.2.even if we pick nom as a stem (which is weird!

you can't even write "nom" in hiragana because there's no solo "m"), that still doesn't quite work because nonda and nonde break the pattern.for now, let's ignore nonda / nonde and focus on the first issue.we'll say that nomu does have a stem.we'll write it like this: nom*.wait, what's that?here, * is like a wildcard vowel that's "waiting to be filled". it could be noma, nomi, nomu, nome, or nomo depending on context. when writing the stem, i'll always write it neutrally like nom*.but how is the actual vowel determined? let's try to spot a pattern:nom* + masu = nomimasunom* + mashita = nomimashitanom* + masendeshita = nomimasendeshitanom* + masen = nomimasennom* + nai = nomanainom* + nakatta = nomanakattanom* + tai = nomitaiit seems like each of the suffixes "smuggles" its preferred secret vowel that's normally not visible. we wanted to add "masu" but got "imasu". we wanted to add "nai" but we got "anai", and so on.this is, in fact, the case!each of these suffixes starts with a (secret vowel). it only reveals itself if there's a * wildcard before it; otherwise it disappears:nom* + (i)masu = nomimasunom* + (i)mashita = nomimashitanom* + (i)masendeshita = nomimasendeshitanom* + (i)masen = nomimasennom* + (a)nai = nomanainom* + (a)nakatta = nomanakattanom* + (i)tai = nomitaiin the case of tabe, there's no wildcard, so there's nowhere for the suffix's secret vowel to go.

in those cases, it just gets discarded:tabe + (i)masu = tabemasutabe + (i)mashita = tabemashitatabe + (i)masendeshita = tabemasendeshitatabe + (i)masen = tabemasentabe + (a)nai = tabenaitabe + (a)nakatta = tabenakattatabe + (i)tai = tabetaiso it seems like we can formulate a rule to refine concatenation:A suffix may start with a secret vowel. If the stem ends with a wildcard, the vowel is revealed: nom* + (i)masu = nomimasu. Otherwise, the vowel is discarded: tabe + (i)masu = tabemasu.with this refinement, we can now conjugate the vast majority of verbs correctly. now let's pause and take stock of where we are.so that's what the verb groups areso far we've identified two kinds of verbs:tabe + (i)masu = tabemasutabe + (i)mashita = tabemashitatabe + (i)masendeshita = tabemasendeshitatabe + (i)masen = tabemasentabe + (a)nai = tabenaiandnom* + (i)masu = nomimasunom* + (i)mashita = nomimashitanom* + (i)masendeshita = nomimasendeshitanom* + (i)masen = nomimasennom* + (a)nai = nomanainom* + (a)nakatta = nomanakattanom* + (i)tai = nomitailet's now connect it to the concepts you've heard before.verbs like taberu are called -ru verbs or, traditionally, "ichidan" verbs (literally meaning "one-row"). verbs like nomu, on the other hand, are called -u verbs, or traditionally, "godan" (five-row).so why are they called this way?in ichidan ("one-row") verbs like taberu, the last syllable of the stem is fixed.

it's always going to be be, no matter the suffix:it stays on a single row in the hiragana table, hence "one-row".on the other hand, in godan ("five-row") verbs like nomu, the final syllable of the stem alternates between ma, mi, mu, me, and mo:it spans all the five rows, which is why it's godan ("five-row"). the m* "wildcard" represents the entire ma/mi/mu/me/mo column.why romaji is actually goodso far, i've exclusively used romaji in these explanations.this is actually completely intentional. some purists may dislike that, but if you're able to read it without imposing the absurdities of English language pronunciation, i think it provides a much clearer intuition for Japanese verb conjugation than writing it in hiragana.for example, consider these conjugations that you can already do:nom* + (i)masu = nomimasuyob* + (i)masu = yobimasukik* + (i)masu = kikimasuor these:nom* + (a)nai = nomanaiyob* + (a)nai = yobanaikik* + (a)nai = kikanaisome textbooks will suggest you to learn them like this:む (mu) changes to み (mi) when adding ます（masu)ぶ (bu) changes to び (bi) when adding ます（masu)く (ku) changes to き (ki) when adding ます（masu)む (mu) changes to ま (ma) when adding ない（nai)ぶ (bu) changes to ば (ba) when adding ない（nai)く (ku) changes to か (ka) when adding ない（nai)which is correct but obscures the much simpler phonetical intuition:* + (a) = a* + (i) = iromaji conveys the intuition because you "see" the vowel moving. but this requires "cracking" the syllables which hiragana can't do.why romaji is actually badromaji is good for phonetical explanations but you should still "think" in japanese syllables. here's an easy way to verify your thinking.consider the word hanasu ("speak").

let's try to conjugate it. to conjugate any verb, we must first find out its stem. (you can google it, check the dictionary, or sometimes guess — more on that later.)the stem for hanasu is hanas*.now let's try to apply the rules:hanas* + (i)masu = hanasimasu (wrong!)hanas* + (a)nai = hanasanaias you can see, one of these is wrong.can you figure out why?(hint: recall the hiragana table. scroll down for answers)here's the corrected version:hanas* + (i)masu = hanashimasuhanas* + (a)nai = hanasanaithere is no "si" in the hiragana table, so s* + (i) = shi:this is why it's important that you don't actually "think in" romaji.s* + a = sas* + i = shis* + u = sus* + e = ses* + o = soi'm using romaji as a convenient way to refer to phonetics in text. however, your "mental algebra" should match the hiragana table.to verify your understanding so far, let's do an exercise.exercise checkpoint: a verb equationhere is a casual negative form of some verb we don't know:... + (a)nai = matanaiyour task is to produce its polite negative using -(i)masen:... + (i)masen = ???(hint: figure out the "..." stem first. scroll down for answers)first, let's figure out the stem. in the first "equation", i've highlighted mata, which tells us that the stem must be mat*:mat* + (a)nai = matanaiso this is a godan ("five-row") verb which alternates vowels. we're adding (i)masen, so t* will now "eat" the "i", so we need t* + (i):mat* + (i)masen = machimasenbut here's a trap — it's machimasen, not matimasen!this is not some fancy exception, it's just that "ti" literally doesn't exist on the table.