Zamenhof gave Esperanto one strikingly unnatural feature: the ‑A ‑E ‑I ‑O ‑U word‐class marking endings, superficially resembling the sets of “thematic vowels” in Latin‑ or Russian‐style declensional systems. Forcing words into this pattern may add that little bit of extra distortion to the roots, but it was intended to serve three purposes:
-
Class‐Marking – if all nouns end in ‑O this makes it easier for learners to find their way around a sentence.
Now, this part did work to some extent. Unfortunately… sentences are heard as streams of noises rather than as sequences of discrete words, and listeners don't benefit from a rule that ‑I marks an infinitive and ‑O marks a noun if they can't tell the marker vowel from any other I‑ or O‑sound. Sure, readers can see the spaces between words… but if it was just for readers, what was wrong with the old German trick of capitalising nouns? Besides, all the “little words” learners need most help with are left outside the system: ilI is the pronoun “they” and maltrO is the adverb “too little”.
Once you've identified the individual verbs and so on, the next step is to be able to find your way around phrases and clauses; but Zamenhof showed no sign of recognising this higher‐level structure as something learners might have trouble with, instead filling his writings with long, contorted, under‐punctuated sentences. Just for example, the first sentence of the foreword to the Fundamento doesn't get around to mentioning its subject noun until 43 words in, a few words after its main verb.
-
Class‐Switching – any given root can take any one of the endings, reducing the number of separate dictionary entries required: a verb can become an adjective, or a noun can become an adverb, without retaining any trace of its history. (The more normal approach in natural languages is for class‐switching to involve adding some sort of extra suffix, so users can easily tell that for instance the noun goodness is derived from an adjective.)
Unfortunately… this radically artificial approach has unpleasant effects. For a start, a single noun may be associated with more than one adjective – take for instance the roots dent‑, sun‑, viv‑. As nouns they mean “tooth, sun, life”, but the adjectives are ambiguous in each case between “pertaining to” and “richly supplied with” the noun (“dental/toothy”, “solar/sunny”, “biotic/lively”). Even when the class‐switching might work neatly, it often doesn't: for instance, nouns converted to verbs may end up either transitive (broso/brosi = “a brush/to brush”) or intransitive (fajro/fajri = “a fire/to be on fire”). Compare also the haphazard behaviour of nouns like timo = “fear” and naŭzo = “disgust”.
-
Class‐Abstraction – the roots themselves are classless abstractions, so there's no need to memorise whether viv‑ is basically a verb, noun, or whatever.
Unfortunately… as soon as anyone else looked at this scheme, it became apparent that it simply didn't work. The only way Zamenhof's morphological shambles could be sorted out was by abandoning the doctrine of class‐abstraction and declaring a basic category for each root in the dictionary. Thus for instance while bros‑ = “brush” is inherently a noun, komb‑ = “comb” is a verb – which is why kombo means “an act of combing”. The meaning “a tool for combing” requires the compound komb‐ilo, but “a tool for brushing” is simply the original noun broso = “a brush”. (No, ilo and ili aren't related.)
Most unfortunately of all, this attempt to break and reset Esperanto's backbone was implemented as a “quick fix”, with as few detectable alterations to the published grammar as possible, rather than as an honest attempt to eradicate the problem (which would involve redesigning the language's morphology from the roots up). Many of my Esperantist correspondents have themselves been unaware of these retrospectively imposed root‐classes – they aren't mentioned at all in “Teach Yourself Esperanto”, despite their effects on the functioning of derivational affixes. For instance, forta = “strong” is an adjective‐root, so you can add the causative affix ‑ig to get fortigi = “to strengthen, render strong”; but unua = “first” is from a numeral‐root, so unuigi means not “render first” but “unify, make one”. Similar fiddly distinctions govern the usage not only of ‑ilo (above) but of various other basic affixes such as ‑ado (sometimes used to form verbal nouns), ‑eco (sometimes needed to form abstract nouns from adjectives), and so on.
And thus Esperanto has wedged itself between the two stools of naturalism and regularity. People familiar with the Romance (or indeed Slavic) languages often expect the ‑A ending in mi estis la triA (= “I was the third”) to signal feminine agreement, which of course it doesn't. But it isn't serving any other useful purpose, either. The word tria isn't underlyingly an adjective‐root, and its syntactic function here isn't that of a modifier: it has been promoted to the position of head of a definite noun phrase. It's just that it has got stuck in a form that ends in ‑A.