Discussion transferred from prava/vera/korekta

Sudanglo, don't you think that if this approach had been used to build the language that we'd now have hundreds of inconsistencies instead of the handful that we currently encounter?

Short answer, Cflat, No.

The derivation of compounds based on the meanings of words permeates the whole language. Inconsistencies arise through viewing the language through the lens of the unique root class theory, but not when the derivation is viewed as word-based, but not limited to always being from the same word (ie same part of speech).

To give you a flavour of how labyrinthine the analysis becomes when based on the idea of root classes, I quote from PAG (358.3), a section that deals with korekta.

...la verba flank-elemento adjektiviĝas. Sed aliflanke, la latenta a-finaĵoj postulas antaŭan substantivigo, do la adjektivigo pasas tra substantivigo. ..... Kaj ĉar la adjektivigo de verbaj radikoj pasas ĉiam tra substantivigo, la sama (temas pri ne-specifigo de pasiva/aktiva) validas ankaŭ pri verbo-adjektivoj.

But korekta doesn't mean korektita, so how this sheds light, I do not know. However it does imply that korekta could mean corrective as well as corrected (but it doesn't mean that anyway).

The thing is we are viewing the language after it was built. I believe it was created with the root class notion in mind*. Call it theory if you wish but it seems to me that it is the principal that the language was founded upon.

In the case of korekt, if it had been introduced as meaning 'correct', and not the action of correcting, I think we wouldn't have this problem. Korekta would mean correct, and korektigi would mean to correct. It is too late now to change things, and we all agree (most of us) to go on with a few inconsistencies.

*Zamenhof in the Fundamento defined the original vortaro as a set of roots, each with a definition and that definition was either substantive, adjectival, adverbal(?), or verbal (except for words with no gramatical endings). As people added to the language I assume they followed this example. In a few cases, unfortunate choices were made.

You make a valid point, Cflat.

Zamenhof is on record as conceiving the language as built by combination of elements where the grammatical finajxoj are (bizarrely) considered as words. And, if I recall correctly, the Universala Vortaro actually defines roots rather than words.

Also the root class theorists have traded on the fact that the translations of the roots were, in the various languages, all the same part of speech.

However, a proper description of Esperanto should be based on how the language is today.

I put to you a simple question. How would an Esperantist interpret the meaning of martelisto? Is he more likely to consider that based on marteli or martelo, or either depending on context, or reject the word as improperly formed.

sudanglo:You make a valid point, Cflat.
However, a proper description of Esperanto should be based on how the language is today.

Agreed; however, shouldn't it be possible to make do with the root class description, noting the exceptions? Instead of trying to explain exceptions as part of the original intent (like I think you are saying was done in PAV), just accept that it wasn't done optimally and perhaps note how it could have been done according to the root class approach (like I did with 'korekt' above).

sudanglo:I put to you a simple question. How would an Esperantist interpret the meaning of martelisto? Is he more likely to consider that based on marteli, or martelo, or either depending on context, or reject the word as improperly formed.

I would say it should be based on the root, not on words formed from the root. That is 'martel', which from the Fundamento means "hammer". The 'ist' suffix is defined in PIV as a person who occupies him/herself with the matters defined by the root. So, what would that be... someone who owns a hammer or needs it for his work (carpenter? steel worker, plumber, etc), or the guy that sells/manufactures them? We already have a word for carpenter (ĉarpentisto) and someone who sells them would be a martelvendisto, and the manufacturer would be martelfabrikisto?

So what is a martelisto? I'm not sure, PIV lists it is an error. If an Esperantist used it in a conversation I'd probably assume they meant ĉarpentisto, if that was the context (I'd assume they had forgotten the word ĉarpentisto).

But I think this touches on the issue of how roots were chosen. For the most part, the intuition of Zamenhof and the early Esperantist's seemed to have worked things out well enough. As we've said there are only a few weird cases. If we had the luxury of picking the roots we might suggest some guideline that might avoid even the inconsistent cases. For example should a root be based on an action or on the object causing or associated with the action? Our most famous example being the following:

bros' = brush
comb' = act of combing

For example, when an action is normally associated with some tool maybe the root should bow to the action, and the tool be formed from the 'il' suffix. But this is just a thought and is not based on examining various cases. In any event, it is just an academic exercise (unless you are planning on developing your own conlang, which no doubt would be academic as well).

cFlat7:So what is a martelisto? I'm not sure...

Certainly one could make the case that martelo should be martelilo. But then I'm not sure what martelo would mean. Zamenhof may not have either, and may have just decided to remove what he might have seen as a redundant syllable, and made the -ilo somewhat implicit in the martel- root when rendered into a noun.

So, whether you take the root as predominantly verbal or substantival (or adjectival or adverbial, for that matter), a martelisto would logically be a term for someone who deals with or in hammers or hammering. A ĉarpentisto (uses hammers professionally) would be a martelisto. A martelvendisto (sells hammers) would also be a martelisto. A martelfabrikisto (makes hammers) would also be a martelisto. I don't see why it should be considered an error. It would seem to be a general term that covers a variety of areas, all related to professions that deal with hammers or hammering in some way (and I'm not sure you can separate the two, really). That's how I would interpret it if I came across it, in any event.

a martelisto would logically be a term for someone who deals with or in hammers or hammering. ...... It would seem to be a general term that covers a variety of areas, all related to professions that deal with hammers or hammering in some way (and I'm not sure you can separate the two, really)

That would be my guess too, Riot. But of course that means the root class theory is not right.

Under that theory, the martel in martelisto should mean a hammer, or you can invoke a complex mass of 'vortefikaj' rules which say that other elements in a compound shift the grammatical class.

But if -isto forces the martel in martelisto to be marteli, then what of panisto (breadman).

Far simpler to forget about root class, and to see derivations as being from this or that word, with common sense and context deciding the derivation and therefore the meaning.

Of course, this in no way prevents many derivations being from a single word (ie part of speech), but it does free us from the strait-jacket of the idea of a specific grammatical class of a root, which places beyond the pale certain pragmatic and convenient compound words, or requires the invocation of convoluted theories of the influence of one root on the class of another.

sudanglo:

a martelisto would logically be a term for someone who deals with or in hammers or hammering. ...... It would seem to be a general term that covers a variety of areas, all related to professions that deal with hammers or hammering in some way (and I'm not sure you can separate the two, really)

That would be my guess too, Riot. But of course that means the root class theory is not right.

Under that theory, the martel in martelisto should mean a hammer

No, the martel in martelisto refers to a hammer or hammering. It is either 1) both or 2) neither (depending on how you look at it) verbal and/or substantival. It lies in between. Like I said, I'm not sure you can separate hammer from hammering. At least, at the root level, in Esperanto.

I don't think that invalidates the root-class theory. That theory just says that some roots are more inclined towards some grammatical types than others. Some are obviously verbal. Some are obviously adjectival. And so on. But the levels of inclinations towards classes are probably different for every root*, and really ought to be accounted for. Some roots are very strongly verbal, or adjectival, or whatever, and anyone can plainly see it. Some are maybe not so strong, don't fit neatly into one box, incline towards multiple classes, and look like exceptions. I'm not so sure they are, though. Martel appears to be one of those weaker cases - it spreads its inclinations among more than one class - but really, so what? That doesn't invalidate anything, it just makes the root martel a little complex. I don't think it really touches the root class theory as a whole at all.

All theories are artificial - it's just a question of which one you think is more useful and/or descriptive for whatever it is you are doing with it at the moment. One theory doesn't necessarily invalidate another - sometimes they just emphasize different aspects of things by organizing their elements differently; these theories are descriptive tools, not universal laws, after all. Going at the language from a word-derivation point of view is certainly another way to slice it up, and strikes me as something perfectly valid to explore. Doesn't mean root classes are wrong and shouldn't be used. Conversely, root classes being useful and/or descriptive doesn't invalidate looking at the language using word-derivation as its lens. Esperanto isn't math. It's a human language. There's some sloppiness there. No one theory is going to get it all, and different ways of describing the language can easily, peacefully, and even logically, coexist.

--------------------------------------
* Maybe even measurable, if you make certain assumptions about how frequency of use might correlate to root class.

No, the martel in martelisto refers to a hammer or hammering. It is either 1) both or 2) neither (depending on how you look at it) verbal and/or substantival. It lies in between.

But if you look at it that way, Riot, surely it does invalidate the root class theory, which says that roots are of a certain grammatical class. Though to be fair to the elaborations of the theory such as are to be found in PAG, through a process of 'vort-efiko', other elements in a compound can change the root class according to somewhat labyrinthine rules.

Note, that once you allow hybrid roots, then korekta no longer needs to be classified as evitinda. Korekt is then both an adjectival root and a verbal one.

Of course the early Esperantists were using the language and being understood without reference to a dictionary for every compound they created, and this continues today. So some theory of word-building is implied by their usage. I just don't think that root class gives a good account.

The process of word-building in Esperanto seems to me to be very much in the business of taking a word, knocking of the (compulsory) ending but preserving the idea of the word with the ending you have knocked off and then adding a different ending as required (or embedding it in a multi-element compound, rather than taking a root (with a defined class) and then incorporating it in the compound.

IMHO the process of word-building in Esperanto is pragmatic, reflects strongly our knowledge of the world, and those thing we wish to speak about, and it simply isn't necessary to erect a theoretical superstructure based on the grammatical class of a root when we have recourse to the meanings of words.

sudanglo:
The process of word-building in Esperanto seems to me to be very much in the business of taking a word, knocking of the (compulsory) ending but preserving the idea of the word with the ending you have knocked off and then adding a different ending as required (or embedding it in a multi-element compound, rather than taking a root (with a defined class) and then incorporating it in the compound.

Regarding this word-building approach:

If this is an idea for the present, then what value does it provide? The root-class "theory" we all agree works for aprox. 99.9 percent of the cases. The few inconsistencies can be easily memorized. So another "theory" doesn't really seem necessary or useful.

In any event, for the few inconsistencies we do have, this word word-building approach has the problem that if you start with a root (e.g. in a word like martelisto), you have no clue as to which "word" it was based on. All you have is the bare root, in this example 'martel'. So you not only have to memorize the root meaning, but you have to memorize which word it was based on. And each inconsistent case can possibly be based on a different word (grammatical class).

And intuition in this area can't reliably be trusted, as we know it can vary from culture to culture. Experienced Esperantists have no need for the approach as they will already know all the problem words, and beginners I believe might find the word-building approach somewhat confusing and not very secure.

Of course, since you say "word-building" then the approach isn't intended for people trying to figure out what a word means, but instead how words should be created/built. But hasn't the word-building in Esperanto mostly already been done? And if there are new words to be created, shouldn't we use a method that puts the new words into the 99.9% category instead of potentially creating new oddities?

this word word-building approach has the problem that if you start with a root (e.g. in a word like martelisto), you have no clue as to which "word" it was based on

How so?

Actually in a modern dictionary of Esperanto, you can't actually look up martel and find it classed explicitly as a substantive root. However you will find martelo defined (as the tool) and marteli defined (as using the tool).

You then have the option of seeing martelisto as based on marteli, as vendisto is based on vendi, or as based on martelo, as panisto is based on pano (not pani).

[The reason why pani is not used as a verb is that there is no real-world activity (at the moment) that is consistent with the b]meaning[/b] of pano.]

The idea that each root has a specific grammatical class leads to a mass of complications. Have you ever looked at the PAG section on word-building and seen how it gets its knickers in a twist trying to fit the actual usage into the strait jacket of the specific class of a root theory?

Even PAG admits that senpova is based on povo and not on povi, but achieves this by 'vortefikaj' rules that shift the grammatical class.