sudanglo:In Esperanto you have to learn the uses of words. But common sense allows you guess the meanings of derived parts of speech and other compounds. Imagine someone who knew only manĝi. Which would seem more likely as a useful word - manĝo meaning a meal (with further compounding in matenmanĝo.vespermanĝo etc) or manĝo meaning an act of eating?

But how does this "learn[ing] the uses of words" differ in significance from learning root classes? To me, it is no more effort (here I differ from RiotNrrd) to learn a root class than to learn a head word as the starting point for a derivational series. As for the specific example here, I think this is a slippery example of "common sense." One person's "common sense" might be another person's oddity. Myself, I would presume that a meal and an act of eating are equally likely meanings for "manĝo" and equally useful, so "common sense" does not get me anywhere.

sudanglo:I obviously can't accept that 'flor' changed its class (if I believe that the grammatical class of a root is a grammarians fiction).

However, it may be that in early dictionaries flori was the headword and in later ones floro.

It's no fiction, but the reality of the autorative Universala Vortaro, where you find Esperanto roots translated by words - all and without exception of the same grammatical class (although in English this is often not visible).

The case of flor' is a bit more complicated than I thought. You find floraĵo 60+ times in Zamenhof's works (and hardly in the works of other authors), but floro does appear, too, even more than 300 times.
A look at the UV reveals:

flor' fleurir | flourish | blühen | цвѣсти | kwitnąć
flor'o fleur | flower, bloom | Blume | цвѣтокъ | kwiat

So, flor' is (was) a verbal root, but from the beginning, as you wrote, also floro did exist. The explicit appearance in the UV - with grammatical ending! -, however, shows that it is a conventionalized, irregular derivation, as it appears in other such cases, e.g.

dot' doter | endow | ausstatten | надѣлять (приданымъ) | wyposażyć
dot'o dot | dowry | Mitgift | приданое | posag

sudanglo:Incidentally in the fifth edition of Millidge (revised) which was constantly reprinted from the 1920's to the 1950's we have headword bluo but headword ruĝa. Go figure!

The UV has

blu' bleu | blue | blau | синій | niebieski

ruĝ' rouge | red | roth | красный | czerwony

Figured.

But how does this "learn[ing] the uses of words" differ in significance from learning root classes?

Much of the time the results are the same whether you view Esperanto as having roots with a certain grammatical class or whether you view the derivational process as take a word (usually but not always the headword in the dictionary) conserve its meaning and replace the finaĵo with another one.

The question is which is a better description of Esperanto as it is today - leads to less tortuous analyses, makes the language seem simpler, and produces fewer exceptions.

Zamenhof may have indeed conceived the lexical roots as words, but was he right? It's a strange use of the idea of a word.

A word like Senpova under the root class theory needs a special explanation since it clearly doesn't mean senpovi-a but senpovo-a.

Under the view that the derivational process is take the meaning of a word (in this case povo rather than povi) and then derive another part of speech, senpova doesn't need a special theory of word formation. The 'sen' gives a clue that you are using pov to mean povo and not povi. Under the theory that pov inherently is a verbal root you need to explain senpova.

Again to explain a phrase like la ekfloro de Esperanto en la tridekaj jaroj all you need is to understand that this is derived from flori rather than floro but in florpoto the derivation is from floro. You don't need to see flor as a conventionalised irregularity, being both a verbal and a noun root (uncomfortable to the theory that roots have a certain grammatical class).

This simpler view is that the compounding of words in Esperanto is pragmatic rather subject to the straitjacket of grammatical theory. Is there something in the world that any particular part of speech (or other compound) can naturally name.

You can find hundreds of pairs or words with related meanings where one expresses an idea substantively and the other verbally, eg profesoro and instrui, and the derivations are consequently different. And from such observations the theory of grammatical root class was developed. But in the end it is better to have a theory that explains all uses rather than the majority, and is conceptually simpler to boot.

Is a vestejo a vest(i)ejo or a vest(o)ejo a dressing room or a garderobe? Under the grammatical root class theory it can only be one of them. But in context allowing both meanings is practical and unlikely to lead to confusion. Oni atakis la reĝon en lia vestejo - kostas 1 eŭron por lasi sian ĉapelon ĉe la vestejo.

Let us now turn to 'korekta'.

Usage tells us that this word primarily means correct rather than corrective/correctional (which latter meaning should be primary under the theory that korekt is a verbal root).

Korekta in its usual meaning is a far too useful (it is not an exact synonym of gusxta) and also obvious word for us to abandon it in order to conform to grammarians theory. And I'm quite happy to let korekta additionally mean correctional when it is obvious that this is the intended meaning.

(Incidentally if the roots really do belong to certain classes why are they never printed in the dictionary in different colours or type faces to allow easy identification?)

The practical argument that korekta is evitinda is that it may encourage people to say korektigi rather than korekti. But do they actually do this?

If the notion that the word is the primary source for derivation leaves korektigi as potentially ambiguous (is it to correct or to get corrected) is this a big problem?

sudanglo:Usage tells us that this word primarily means correct rather than corrective/correctional (which latter meaning should be primary under the theory that korekt is a verbal root).

Which principle says that the regular type should have the most tokens? In fact words with a high token frequency often show irregularities, as the latter are more easily corrected with infrequent words.

sudanglo:The practical argument that korekta is evitinda is that it may encourage people to say korektigi rather than korekti. But do they actually do this?

I've seen it happen on a few occasions, but I agree that it's not a big enough problem to warrant the 'evitinda' designation. If I were going to argue against korekta I'd just do it on the basis that it's an exception to the normal Esperanto logic (where "corrective" would be the more obvious meaning), and it's generally a good thing to minimise exceptions. It has nothing necessarily to do with conforming to a theory, but rather making Esperanto more consistent. In any case I'm happy with a few irregularities here and there; particularly if, like korekta, they have a tradition of usage.

The root-class theory has always puzzled me, and yet, as I continue laboriously constructing my very own vortaro (I have few occasions to speak l'internacian lingvon), I seem to be instinctively following the theory, while grouping all connected words under a "root" understood as a word without an ending (-a,-e,-i,-o). In practice, I learn Esperanton a posteriori (el plejposte?) from reading Eo-librojn; ekzemple, if I find that a good author or translator uses kroni and then krono as the thing put on the new monarch's head, rather than the act of putting it there, then I suppose that there is the word kronado for that act. (I have the idea that la Majstro was thinking of the francangla finajxo "-ation" when he composed the "-ado" ending.) Am I hopelessly off the rails here?

(Flanke, dum Zamenhof nomigxis "la Majstro", mi dirus al scivolaj ke li estas "komponisto" kiu komponis Esperanton kiel muzikverkon; fakte, l'internacia lingvo estas "organizita sono", laux difino cxe Edgard Varèse de la muziko gxenerale ideita. Cxu mi malpravas?)

matus1940:Am I hopelessly off the rails here?

A certain very vocal poster on this forum has apparently been dismissing Esperanto’s root-class system as a “theory” for a long time now, but since Zamenhoff himself in Unua Libro (1887) presents his entire lexicon as classed roots to which affixes either extend or reaffirm the meaning, the only theoretician here is this forum’s vocal poster.

So you’re not off the rails at all, Matus: krono is “a crown”, kroni is “to crown”, and kronado is “a coronation”.

The way it works is that most every word in Esperanto is effectively a compound, that is, krono (listed simply as kron' in Unua Libro’s vocabulary) is really kron-o. The suffix -o has various nuances depending on the grammatical class of the word it’s attached to: after a noun root, -o is redundant; after a verb root, it names the action or the result of that action (ex., faro “deed”); after an adjective, it names the abstract quality (ex., belo “beauty”).

Which all seems natural and convenient when one’s own language works the same way (in English we make “crown” a verb simply by putting it in the verbal position of a sentence); the downside is that, without knowing that kron- is a nominal root, one might assume from kroni that the gerund is krono and that the crown itself is a kronilo. So if you’re new to Esperanto, you’ll want to keep your vortaro handy.

For what it’s worth, you’re not the only one to be frustrated by this confusion — in Ido and other Esperanto reform projects, one goes from krono “a crown” to kron-iz-ar “to crown” to kron-iz-o “a coronation”, martelo “a hammer” to martel-ag-ar “to hammer” to martel-ag-o “a hammer strike”, and so on. Sometimes the smaller innovations of these Esperantidos get absorbed into Esperanto itself, but, so far, not this one.

I seem to be instinctively following the theory

In many cases the theory that a root is in effect the same as the corresponding word with a particular part of speech ending leads to the same result as seeing derivations coming from the meanings of words (not roots).

The way we label the world (at least in the European languages) tends to divide up the world into things, actions (or states) and characteristics.

To explain the meaning of to hammer one makes reference to a hammer. To explain the meaning of to strike or to beat does not require reference to a particular implement used to strike or beat. To explain the meaning of big, we point to things having the quality of being big (we contrast, for example, a mansion with a cottage, a giant with a dwarf).

The problem with the theory that a root belongs to a specific grammatical class is that it entails a convoluted superstructure of word formation rules to explain words in which the root no longer has its 'official' class.

In any case, the ordinary Esperantist, might well scratch his head when asked to say what the class of a particular root is, though he well understands and uses correctly the words containing that root.

It wasn't until some 80 years after the Unua Libro that the Academy adopted the root class theory, previously preferring to see word construction as governed by the principle of neceso kaj sufiĉo.

The only 'benefit' of the root class theory is that it gives the linguists something to pontificate about. But provided that we ignore their 'ex cathedra' pronouncements, and happily pursue our pragmatic path, these particular cooks won't spoil our broth.

sudanglo:It wasn't until some 80 years after the Unua Libro that the Academy adopted the root class theory, previously preferring to see word construction as governed by the principle of neceso kaj sufiĉo.

Interesting. Never heard of that. But I suspect I never will outside of this forum, since the early grammar books and dictionaries I have (published just sixteen years after Unua Libro) don’t bear your claim out:

So, for example, the derivation of frat'in'o, which is in reality compound of frat “child of the same parents as one’s self,” in “female,” o “an entity,” “that which exists,” i.e.,that which exists as a female child of the same parents as one’s self” = “a sister,” — is explained by the grammar thus: the root for “brother” is frat, the termination of substantives in the nominative case is o, hence frat'o is the equivalent of “brother”; the feminine gender is formed by the suffix in, hence frat'in'o = “sister.”

Esperanto: The Student’s Compete Text Book, 1903

Particular attention must be given to the fact that it is to the root of a word that the prefixes and suffixes are added. When it is stated that the final letter “i” indicates the infinitive, the letter “o” the noun, the letter “a” the adjective, the letter “e” the adverb, the letter “j” added to form the plural, etc., the pronouns mi, li, vi, etc., do not interfere with the statement, for they are complete words; the letters m', l', and v' are not roots. The word “do” is not a noun, because d' is not a root. The word plej is not a plural, because ple' is not a root. The word meti, to put, has nothing to do with the diminutive suffix et, because m' is not the root.

English-Esperanto Dictionary, 1906

Multaj vortoj estas faritaj el radikoj per la aldono de specialaj gramatikaj finiĝoj.

Rimarku la vortojn tiel faritajn el la radiko demand-: demando, demandas, demanda, demande.

Oni faras multajn vortoj per la aldono de unu radiko al alia, aŭ kun iliaj gramatikaj finiĝoj (ekz. nigratabulo), aŭ sen la gramatika finiĝo de la unua vorto (ekz. ĉerizarbo). Ĉi tiuj estas kunmetitaj vortoj

Tra la Jaro, 1914

And so on.

In any case, even if what you claim is true, then “root class theory” has been officially adopted and is the rule of the land in Esperantio. The many pronouncements and denouncements from your own high cathedra suggest that might be happier with a language project of your own.