Mongolian Vowel Harmony: The Rule

The rule

A Mongolian word draws its vowels from one set only. There are two sets. The first is а, о, у. The second is э, ө, ү. Vowels from the two sets do not appear in the same word, and a suffix attached to a word takes a vowel from that word's set.

That is the whole rule. Everything else on this page is about using it without having to stop and think.

Two short lists finish the sets. ы, ю and я go with а, о, у; you will meet ы mostly inside endings, as in таны (your) and аргын (of the method). е goes with э, ө, ү, as in ес (nine) and ерөнхий (main, chief). ё goes the other way, with а, о, у: it writes a y followed by the vowel of о, and the endings such a word takes show it — ёс (custom, etiquette) gives ёсоор (according to) and ёсны (of custom), both with first-set vowels. One vowel, и, belongs to neither set, and it gets a section of its own below.

Read the two columns aloud and the rule stops being an abstraction. Each word on the left is built out of one family of sounds and each word on the right out of the other. Grammars call the two sets back and front vowels; the labels are a teaching convenience, and the page on the Mongolian vowels says what the sets really are in the mouth. The method this course is built on calls the first set masculine and the second feminine. Those are names for two groups of letters and nothing more: they say nothing about the meaning of the word, and Mongolian nouns carry no gender of any kind.

What this replaces in English

English has no equivalent, and that is the difficulty. In an English word the vowels are unrelated to each other: photograph runs through three different vowels and nobody notices. Mongolian words are vowel-consistent, so a Mongolian speaker hears a word with vowels from both sets the way you would hear a misspelling.

The cost to you is not in recognising words. It is in producing them. You cannot learn a Mongolian suffix as one fixed string the way you learn English -ing or -ed. Ten of the most common suffixes in the language come in four shapes each, and the word you attach them to decides which shape you use: хот (city) gives хотоос (from the city) while гэр gives гэрээс (from home); ав- (take) gives авсан (took) while өг- (give) gives өгсөн (gave). Same ending, same job, different vowel. The complete set of shapes lives on one page and one page only, the four-form suffix table, so that you learn it in one place instead of meeting it in fragments.

One more thing has to be said before you trust the rule too far: harmony decides the form of most suffixes, including the ten four-form suffixes, and a small number of suffixes ignore it completely. Those are listed, counted and named on the page on the suffixes that do not follow harmony. They are few, they are frequent, and knowing them is part of knowing the rule.

Classifying a word in three seconds

In practice you never analyse a whole word. You run three steps, and after a week you run them without noticing.

1. Cross out every и in the word. It votes for nothing.
2. Look at the first vowel that is left.
3. That vowel's set is the word's set. Because harmony holds, the remaining vowels will agree with it, so you do not have to check them.

Worked through on real words: ажил (work) — cross out the и, you are left with а, so the word is in the а о у set, which is why the past form is ажилласан (worked) and not a form in э. өчигдөр (yesterday) — cross out the и, you are left with ө, so the word is in the э ө ү set, which is why "from yesterday" is өчигдрөөс. малчин (herder) gives а. шашин (religion) gives а. зочин (guest) gives о. жүжиг (a show, a play) gives ү. хөгжим (music) gives ө.

Step three is the step that saves time, and it is worth seeing why it is safe. Once you know the first real vowel, you know the set, and the set tells you what every other vowel in the word can be. A long word is therefore no slower to classify than a short one: харандаа (pencil) is settled by its first а, and хөрөнгө (property) by its first ө.

и, and the word with mixed vowels

и is the neutral vowel; й is a consonant and is transparent to harmony. Keep those two facts apart, because they are often run together, and the second one matters: й is not a vowel at all, it is the consonant that builds the diphthongs ай, эй, ой, уй, үй, and it never decides anything. In сайн (good) and цай (tea) the decision is made by the а in front of it. In малгай (hat), the same.

и is a real vowel, and what makes it neutral is that it sits comfortably in words of either set. On the а о у side: ажил, малчин, шашин, зочин. On the э ө ү side: өчигдөр, жүжиг, хөгжим. Seeing и in a word tells you nothing about which suffix to use, which is exactly why step one of the test is to strike it out.

In a word that has и plus another vowel, the other vowel commands. That is the case in both of the forms above, ажилласан and өчигдрөөс, and it is the case in every mixed word you will meet.

There is one leftover case, and it is common enough to need an answer: a word whose only vowel is и. их (big) is one; the verb stem ир- (come) is another. With nothing else to go on, such a word behaves as a member of the э ө ү set — hence ирнэ (comes, will come), ирсэн (came), ирлээ (has just arrived), all with э. So the full test is: strike out и; if something is left, it decides; if nothing is left, the word goes with э ө ү.

Compound words harmonise unit by unit

Sooner or later you meet a word that plainly contains both sets, and the first reaction is that the rule has broken. It has not. Mongolian builds words out of whole words, and each unit keeps the vowels it had on its own.

• Сүхбаатар is Сүх plus баатар (hero) — one unit from each set.
• Батгэрэл is Бат plus гэрэл (light).
• Алтанцэцэг is Алтан (golden) plus цэцэг (flower).
• Оюунбилэг is Оюун plus билэг.
• Өвөрхангай is Өвөр plus хангай.
• Тосонцэнгэл is Тосон plus цэнгэл (joy).

Personal names and place names are where this happens most, because Mongolian names are made by joining two meaningful words, and the two words are chosen for their meaning rather than for their vowels. Read such a word as two pieces and the rule is intact inside each piece. The practical consequence is that you have to split the word before you classify it, and classify the unit the suffix actually touches — the method states the principle unit by unit and does not work a declined compound through, so treat that last step as reasoning from the rule rather than as a form you have seen. Where names come from and how they are built is the subject of the page on Mongolian names.

A word carrying –гүй can look mixed in the same way — болохгүй (not possible), ажилгүй (without work) — but there no compound is saving the rule: that ending simply never changes its own vowel. It is the first of the handful of endings that stand outside the rule, and they are listed, counted and named on the page on the suffixes that do not follow harmony.

Why the rule earns its keep

Vowel harmony is usually presented as a curiosity. It is better understood as the one piece of Mongolian that gives you something back for nothing. Three returns, in order of value.

It lets you produce forms you have never heard. Learn the ablative once, classify a noun, and you can say "from" with any noun in the language without having been taught that particular word: Канад gives Канадаас (from Canada), Франц gives Францаар (in French), Монгол gives Монголоор (in Mongolian).

It lets you hear where a word ends. A sequence that changes set halfway through is nearly always two words glued together, or a word with one of the few harmony-proof suffixes on the end. Both are useful signals when you are listening to speech rather than reading it.

And it narrows guesses. When you half-hear a suffix on an ө word, there is only one vowel it can have contained. That is a real advantage in an audio course, where you often catch four fifths of a word and have to reconstruct the rest. What the ear actually receives, and how far it differs from the spelling, is the subject of the page on spoken versus written Mongolian.

The habit to build

Before you attach anything to a Mongolian word, say the word out loud, strike out и, and name the set. Out loud, at first: the point is to make the classification automatic, and it becomes automatic faster if you hear yourself do it. Three or four days of that and the ten four-form suffixes stop being forty endings and become ten endings with an address.

Then read the two pages that complete this one. The four-form suffix table gives the forms themselves, set out on the four stems ав-, оч-, ир-, өг-. The exceptions page gives the endings that will make you doubt the rule if you meet them without warning. Between the three pages there is nothing else about harmony to learn; what remains is the stacking of suffixes on a stem, which is how a Mongolian word is built.