The -um Wildcard, and Word-Formation Beyond the Rules

The affix track makes a promise this course repeats often, because it is true: Esperanto's affixes are regular and productive, each with one fixed meaning, and together they turn two hundred roots into the working vocabulary of a much larger language. This article is about the two places the promise carries an asterisk. The first is a suffix, -um-, whose official meaning is that it has no meaning. The second is quieter and more general: swapping endings on a root always produces a grammatical word, but what that word means depends on something the endings never show - the class of the root itself.

Neither asterisk is a scandal. Both are Zamenhof being honest about where derivation stops and vocabulary begins, and knowing where that line runs will save you from a specific family of confident mistakes.

-um: the suffix that guarantees nothing

Every regular suffix computes. -il- is the tool, -ej- the place, -ist- the person, -ig- to cause; meet a new word built from them and you can price it up without a dictionary. -Um- is defined differently: it marks a word as having something to do with its root, and refuses to say what. It is the exact counterpart of the preposition je - the language's other official blank - and it exists for the same reason, covered from je's side in the tricky prepositions: a planned language leaks at the edges, and Zamenhof preferred a labelled overflow pipe to a pretence.

Here is the working roster, decoded:

Three more for the collection, because the drills below use them: foliumi (from folio, leaf) is to leaf through a book - or, these days, to browse; ventumi (from vento, wind) is to fan; and the community long ago boiled the whole joke down to its essence with umi itself - to do something-or-other, to potter about unspecifiably. A suffix with no meaning, promoted to a verb with no meaning. Esperantists find this funnier than outsiders do, which is its own small fluency test.

Read down the fourth column of the table and the pattern is the absence of one. Plenumi and malvarmumi share nothing except that each is the conventional -um word on its root. -Um- never tells you the meaning; it tells you that the meaning could not be built, and a century of usage has filled in the rest.

The pair to keep closest is malvarmumi against malvarmigi. The regular one computes: -ig- means to cause, so malvarmigi la supon is to cool the soup down. The -um one is convention: mi malvarmumis is I caught a cold. One suffix you can calculate, one you must have learned - which is the whole distinction, wearing a scarf.

The honest advice, then: treat every -um word as vocabulary, not derivation. When you meet one, the root hands you the neighbourhood (something to do with necks, something to do with air) and you look up, or are told, which house. Do not coin new ones - a fresh regular derivation is understood instantly because the affix computes, while a fresh -um word computes to nothing and leaves your listener guessing. The established set is small, high-frequency and closed for practical purposes. Learn it like you learned the irregular verbs of every other language you have studied, and enjoy the fact that this time the list fits on a beer mat.

Root class: the asterisk on one-root-four-words

Now the quieter asterisk, which does not need -um at all. Word endings showed the trick that makes Esperanto vocabulary compound: one root, four words - bona, bone, bono, boni. What that article politely deferred is the question of what the four words mean. The grammar guarantees each is well formed; the meaning follows rules of its own, and they hinge on the root's class - the part of speech the root inherently is, before any ending touches it. PMEG's vortfarado chapter builds the full theory; the practical core fits in two contrasting pairs.

The brush and the comb. Two near-identical objects on the same shelf, on opposite sides of the line. Bros- is a thing-root: the brush, broso, is the root's own meaning, and the verb brosi derives from it, meaning "do the characteristic thing with a brush". Komb- is an action-root: combing, kombi, is the root's own meaning, and the tool is what needs deriving - kombilo, comb, built with the instrument suffix exactly like tranĉilo from tranĉi and ŝlosilo from ŝlosi. Nothing on the surface of broso and kombilo warns you they were built in opposite directions; the dictionary form of the root is the fact you have to carry. This is why our word pages give you the lemma the root actually lives at - kanti and ludi are verbs whose nouns (kanto, a song; ludilo, a toy) derive, while telefono is a thing whose verb (telefoni) derives.

The hammer and the pencil. A thing-root will verb happily - but only where one action is so bound to the thing that naming it names the act. A hammer has exactly one job, so marteli, to hammer, is old and solid, like brosi and telefoni. A pencil's job is writing, and skribi already owns writing, so krajoni has no work to do and never became a word: you skribas per krajono. The grammar would accept krajoni without complaint; the language community never needed it. Derived meaning is conventional at exactly this point, which is why the confident calque ("I'll just verb the noun, like English does") produces Esperanto that is grammatical, comprehensible and slightly wrong all at once.

The forward directions, for the record, are safe and computable: an action-root's -o names the act or its product (kuro, a run; danco, a dance; kanto, a song), an adjectival root's -i means to be it (boni, to be good), a thing-root's -a means of-or-like it (ora, golden). It is the backward coinages - inventing the verb a thing-root "ought" to have, or the thing an action-root "ought" to be - that need a licence from usage. When in doubt, reach for the boring phrase (per, fari, uzi) and you will never be wrong, merely unpoetic.

And this is where -um rejoins the story: it is the community's own tool for exactly the cases where root class and the forty regular affixes both come up empty. Akvo could not verb into "to water plants" (akvi would claim to be THE water-action, and there isn't one), no affix means "apply to" - so akvumi. The wildcard is not a hole in the system. It is the system admitting, in writing, that no derivation machine covers everything, and providing the part of speech where the leftovers go.

What to take away

• The regular affixes compute; -um is officially meaningless. It marks a word as root-related in some unstated, conventional way.
• Decode the roster once, then learn it as vocabulary: plenumi, kolumo, manumo, brakumi, aerumi, akvumi, malvarmumi, komunumo, proksimume, cerbumi.
• Malvarmigi computes (to cool something); malvarmumi is convention (to catch a cold). The pair is the whole lesson in miniature.
• Do not coin -um words; do coin freely with the regular affixes.
• Roots have classes. Action roots derive their tool (kombi gives kombilo); thing roots derive their one conventional action (broso gives brosi) - and only the dictionary form tells you which way a root was built.
• A thing-root verb exists only where the thing has one settled action: marteli yes, krajoni no. Coin backwards with care; skribi per krajono is always available.

See also

• The affix system - the forty regular pieces this article is the exception to.
• Word endings - the one-root-four-words promise, here given its asterisk.
• Compounding - the other word-building machine, and when to prefer an affix.
• igi and igxi - the -ig- that makes malvarmigi compute.
• The tricky prepositions - je, the -um of prepositions: the same honest blank, one word class over.
• Nouns and adjectives - what the -o and -a endings do grammatically, whatever the root class.