It only seems arbitrary because you're using languages that evolved over thousands of years; the languages we use today didn't just pop into existence as they are. The proto-language(s) may actually have had logical connections between sounds and objects (i.e. they may have been onomatopoeic or anatomically associative).

Parts of the language have, in fact, popped into existence as they are. For the sake of argument, substitute the word "laser" for "dog" in my previous post. Or "robot".

RiotNrrd:It's both arbitrary and conventional.

There is nothing "doglike" about the sounds in "dog". It is arbitrary in the sense that any other set of sounds COULD have been used: "gleb", perhaps, or "troob", or "yebap". That we picked "dog" is arbitrary.

Once picked, it was then adopted by convention. "Dog" now means a particular kind of animal, whereas "gleb" does not. But it COULD have gone the other way around, and no one would be any the wiser.

"I'm gonna take the gleb for a walk" could have been a conventional usage. It wasn't/isn't, but the fact that it wouldn't have made any difference if it was illustrates the arbitrary nature of the sound to concept mapping.

Yebap gets my vote, it conjures up a mental image of a big hairy critter that makes bap noises.
Anytime you want to dump Esperanto and invent a new language with words that sound like what they mean I volunteer to be your first beta tester.

razlem:The proto-language(s) may actually have had logical connections between sounds and objects (i.e. they may have been onomatopoeic or anatomically associative).

I'm pretty sure that I have read of psychological experiments in which the subjects had to guess which of two words in language X (actually nonsense words) meant Y, and there were some distinct biases. They didn't just choose randomly.

What ever the origins, this acoustic/meaning grouping can be seen in Esperanto in, for example, ŝmaci, ŝmalco, sminki, ŝmiri.

If you asked a diverse group of Esperantists to guess the meaning of ŝmuzi, or ŝmarma, I would expect some common elements in their guesses.

ludomastro:@ sudanglo
I can respect your desire to have E-o be different; however, I don't see how you can realistically create a fundamental rule that wouldn't undo the accumulated history

That is not at all my intention. Rather it is to produce a better description of that accumulated history.

I fully accept the two meanings of vestejo, as I expect most Esperantists would do, using the principle that roots can be freely combined (with some order restrictions) into an acceptable word, provided it makes some logical sense.

However, this example actually directly contradicts the theory of inherent grammatical class.

I don't want theoretical considerations that prohibit what seems to me to be a totally legitimate construction capable of meaning either vest(o)ejo or vest(i)ejo, depending on context..

In any case, the proponents of root-class are forced into abandoning their idea within their own theory to account for a perfectly normal word like 'dumviva'.

This is clearly dumviv(o)a and not dumviv(i)a.

The UV may have defined viv (the root) as 'to live' in accord with Zamenhof's desire to show the essentially combinatorial nature of Esperanto.

But it is not clear that the subsequent history supports that idea that in all compounds with 'viv' that the root stands for vivi.

Edit: I suppose that what I am asking for is a better way of talking about it than to say this is a verbal root or that is noun root. Because that sort of description leads to complications and disputes.

I don't suppose it would suffice to simply extend the root-class theory by stating that some words inherently belong to more than one root-class, such that X-ejo and such have multiple possible interpretations depending on context?

Since it seems like the root-class theory explains and aids a lot of word-building, but that there are slightly ambiguous roots that don't fit the bill, or roots that have evolved through usage to span a couple of root-classes.

Maybe the root-class should be seen as the most probable interpretation rather than a inherent immutable property of a root. Then through usage, a root may have more than one probable root-class.

(Does anyone else stumble over "inherent immutable" the first few times, even when reading in your head? )

sudanglo:In any case, the proponents of root-class are forced into abandoning their idea within their own theory to account for a perfectly normal word like 'dumviva'.

This is clearly dumviv(o)a and not dumviv(i)a.

That's not obvious to me, actually. In my opinion it parses easily both ways.

dum vivo = during [one's] life
dum vivi = while [one] lives

As an analogous example, let's suppose that someone receives a certain service as they run - passing out snacks and drinks at a marathon, for example. I could easily call it a "dumkura servo". It's obviously referring to the verb root of running; while you run, you receive this "dumkura" service. While you live, you receive a "dumviva" [whatever].

ludomastro:The root-class theory is useful for the beginner (like me) but still introduces complications into the learning process. (Wait, is "komb-" a noun or a verb?)

That's not a complication introduced by the root-classes theory. As was the case well before that theory was invented, and as is evident from analysis of actual usage, it is simply true that komb- represents and action, and bros- a thing; all root classes does is reflect that reality by assigning to the root a grammatical category corresponding to its meaning.

You're quite free never to think about verb roots and noun roots and just pay attention to what the root actually means. And as erinja said, learning that meaning is usually enough to make the complication go away. It involves a little bit of extra work, granted, and that extra bit of work may be equal to or even greater than simply memorizing a category. But this is something that's unavoidable when learning another language, as well as highly preferable to completely wrong ideas about arbitrary classifications. You can't ever assume that cognates and "equivalent" words from your own native language work the same way. That we must pay attention to this fact, and perhaps experience an extra learning burden above and beyond what might be expected by a person who makes such a mis-assumption, is neither a problem in theory nor a fault in the language.

Unless you want completely free verb transitivity and root class abstraction: which brings with it its own set of problems.

The supposed "irregularity" in the perennial examples of komb- and bros- needs debunking. It isn't an irregularity at all; just a case of two roots exhibiting a different semantic bias - a bias which may well look crooked when you lift them out and put them together for comparison and assert that they should be expected to function identically because they have some connection to each other (they're both related to the grooming of hair) - but a bias which seems rather less worthy of condemnation when you give some thought to how reasonable it is to expect complete regularization of the word stock in a way that pays attention to these kinds of dubious (and innumerable) considerations.

RiotNrrd:Parts of the language have, in fact, popped into existence as they are. For the sake of argument, substitute the word "laser" for "dog" in my previous post. Or "robot".

'Laser' is an acronym and 'robot' is from a Czech loanword meaning 'forced work'.

Going back to ofnayim's OP regarding the matter, linguists have understood this ambiguity for about 100 years and are therefore not 'missing' it. No language can overcome ambiguity, because it depends on the will of the speaker and listener to deduct the meaning.

Broso and kombi, Tom, do not present any irregularity in terms of root class theory. Nor do any of the many other such pairs that you can find in the dictionary.

Such pairs, however are relevant to root-class theory, since historically such pairs (pirato/kaperi, gasto/viziti, nobla/nobelo, kuraĝa/heroo etc) have been used to substantiate root-class theory.

What does present irregularity (or at least a complication for the theory), is when a root apparently changes its class in different compound words.

Suppose 'brosejo' were used to mean a grooming parlour (ie bros(i)ejo) - that would be irregular under the theory.