[Continuation...]

Donald J. Harlow daŭre:So we have O-roots, I-roots and A-roots (for convenience). What role does this play in word formation using affixes?

Most affixes take a certain type of stem (a root or root with affixes) and convert it to another type of stem. The argument has been made, in fact, that certain affixes cannot be attached to certain types of stem because they simply don't take a stem of that grammatical type. But remember from above that the grammatical type is determined not by the content of the stem but by the grammatical ending attached to it. This means that, in fact, every stem is of potentially every grammatical type. Consequently, when you add an affix to a stem it automatically converts that stem to the needed grammatical type just as a grammatical ending would.

Let's try an example. The suffix -EC (a characteristic described by the root) wants to take as input a stem that is an attribute and output something tangible, an object -- in other words A-stem -> O-stem. When we create the word ruĝeco (redness), the use is obvious. On the other hand, consider the word ŝtoneco. Here we have input an object word ... or is it? No, we have simply treated the root as an attribute (represented in English by the adjective "stony" ), and have created the Esperanto equivalent of "stoniness". Or what about kureco? Here I encounter a problem -- there's no English equivalent that I know of. But the meaning should be obvious -- the characteristic associated with "to run".

So, under this theory, "pur'" is an inherently adjectival root and thus need "-ig" to turn it into a verb. Sounds neat to me.

patrik:[Continuation...]

Donald J. Harlow daŭre:So we have O-roots, I-roots and A-roots (for convenience). What role does this play in word formation using affixes?

Most affixes take a certain type of stem (a root or root with affixes) and convert it to another type of stem. The argument has been made, in fact, that certain affixes cannot be attached to certain types of stem because they simply don't take a stem of that grammatical type. But remember from above that the grammatical type is determined not by the content of the stem but by the grammatical ending attached to it. This means that, in fact, every stem is of potentially every grammatical type. Consequently, when you add an affix to a stem it automatically converts that stem to the needed grammatical type just as a grammatical ending would.

Let's try an example. The suffix -EC (a characteristic described by the root) wants to take as input a stem that is an attribute and output something tangible, an object -- in other words A-stem -> O-stem. When we create the word ruĝeco (redness), the use is obvious. On the other hand, consider the word ŝtoneco. Here we have input an object word ... or is it? No, we have simply treated the root as an attribute (represented in English by the adjective "stony" ), and have created the Esperanto equivalent of "stoniness". Or what about kureco? Here I encounter a problem -- there's no English equivalent that I know of. But the meaning should be obvious -- the characteristic associated with "to run".

So, under this theory, "pur'" is an inherently adjectival root and thus need "-ig" to turn it into a verb. Sounds neat to me.

That reminds me of learning grammatical genders in other languages, since in Esperanto one has to learn the word's category in order to be able to derivete appropriate words. That's the sole difficulty in Esperanto's derivation system, actually it is not that difficult to understand.

I once came up with the idea to think of a root word with one ore more affixes as a new word, as a compound having its own meaning.
Isn't Esperanto's derivation system just a compression of its entire word pool?
I could do learn an Esperanto word to every single word of my native language without knowing this derivation system at all, could I not? Of course that's an immense effort but anyhow that's how an Esperanto denaskulo would learn it.

That reminds me of learning grammatical genders in other languages

NO, most certainly no. Grammatical genders are arbitrary attributes of words in those languages that have gender, and need to be separately learnt. And even native speakers can be unsure. A presenter on French TV in a progamme I was watching recently got his knickers in a twist in trying to decide if it was un oasis or une oasis.

Now that ŝtono means a stone, and that pura means the opposite of filthy, and kombi means to comb is ultimately arbitrary. But here you are learning the meanings of words. The derivations are from those meanings.

For most words (but not all) the mechanism is as follows. Take the basic form of the word (the one that comes first in the dictionary, the head-word). Knock off the finaĵo, add a different one, but conserve the idea of the head-word form. It is the meaning of the head-word form which determines the meaning of the derivations.

It is a very pragmatic rather than formal system. For example take manĝi. Because of the way the world is, it happens that we more often want to refer to a meal rather than an act of eating. This is why the common meaning of manĝo is just that, though it can also mean an act of eating (the formal or system-derived meaning). The two meanings are not so distant.

But in the case of pura this doesn't contain any causative idea, it describes a state so it is a leap to far to make puri mean to clean. It has to be purigi.

sudanglo:

That reminds me of learning grammatical genders in other languages

NO, most certainly no. Grammatical genders are arbitrary attributes of words in those languages that have gender, and need to be separately learnt. And even native speakers can be unsure. A presenter on French TV in a progamme I was watching recently got his knickers in a twist in trying to decide if it was un oasis or une oasis.

I didn't say it is the same it just reminds me of that.

sudanglo:Now that ŝtono means a stone, and that pura means the opposite of filthy, and kombi means to comb is ultimately arbitrary. But here you are learning the meanings of words. The derivations are from those meanings.

I still have to keep in mind what category a word's root belongs to, don't I?
Actually one already knows most of the meanings one just has to link new words with those meanings.

sudanglo:For most words (but not all) the mechanism is as follows. Take the basic form of the word (the one that comes first in the dictionary, the head-word). Knock off the finaĵo, add a different one, but conserve the idea of the head-word form. It is the meaning of the head-word form which determines the meaning of the derivations.

Isn't the "head-word" - as you call it - the word's root or stem?
I know you can tag affix at the root, the tag just add a further sense to the meaning.

sudanglo:It is a very pragmatic rather than formal system. For example take manĝi. Because of the way the world is, it happens that we more often want to refer to a meal rather than an act of eating. This is why the common meaning of manĝo is just that, though it can also mean an act of eating (the formal or system-derived meaning). The two meanings are not so distant.

Yes, I understand that but where do I know from whether it is the formal or the system-derived meaning?

sudanglo:But in the case of pura this doesn't contain any causative idea, it describes a state so it is a leap to far to make puri mean to clean. It has to be purigi.

As for me, I didn't have a problem with purigi. I learnt "clean - pura" and "to clean - purigi" without having in mind that "purigi" is actually derivated of an adjective. Sometimes this is easier than just to derivate it if you need it.
I guess, since English doesn't differ "clean" in the word itself whether it is a verb or an adjective, this can cause conflicts.
I know the native language influences the way one thinks so if one has been thinking all the time in a system in which a noun can also be a verb or an adjective this might be confusing when learning another language.

Isn't the "head-word" - as you call it - the word's root or stem?

Although Zamenhof referred to each individual element or building block in Esperanto as vortoj, this is an unusual use of the idea of vorto. When we count the words in a text we don't count the number of roots.

Of course there are some elements which are words. They don't require a finaĵo. But most words are compounds.

Agreed that in the Universal Vortaro there were actually definitions of roots. But dictionaries, since I don't know when but a very long time ago, have defined words not roots.

By head word (kap-vorto), I just mean the usual first word in the dictionary entry.

It all boils down to this. Is vort (for example) a word meaning 'word' with a fixed and unchangeable grammatical class (this is the grammarians trap) or is the word for word 'vorto' from which you derive other words, such as vortumi, alivorte, vortaro.

You might say that it doesn't matter how you conceptualise the structure of Esperanto.

True, in the case of a word like vorto it doesn't seem to have a much practical consequence, whether you see vort as a word with a grammatical class or vorto as the source of most vort-compounds (but not the only possible source). However, in other cases it does.

But staying with vorto, let me ask you the question - is Vortilo a valid Esperanto word. One theory says yes, the other implies no.

sudanglo:But staying with vorto, let me ask you the question - is Vortilo a valid Esperanto word. One theory says yes, the other implies no.

That's a good question. Doesn't the suffix "-il" require a verb? If yes, then "il" can't be a suffix here.
"il" is also used as a noun meaning tool so I would rather see a word compound of "vorto" + "ilo" here.
I am not sure if it was a good idea to allow suffixes to be treated as words. Of course it decreases the number of words one has to learn but where are advantages, must be disadvantages. How do you know if it is a suffix or a word compound?

I think the linguists classify, for example, -et as a genuine suffix. Vorteto is not a type of eto, it is a type of word. But most of the traditional suffixes as tabulated in the lerno-libroj act just as other lexical roots would in compounding.

So manĝilo is an ilo just manĝbastoneto (chop-stick) is a type of bastoneto

Ganove:That's a good question. Doesn't the suffix "-il" require a verb? If yes, then "il" can't be a suffix here.
"il" is also used as a noun meaning tool so I would rather see a word compound of "vorto" + "ilo" here.

Maybe not a true suffix, but I wouldn't see it as "vorto + ilo". "-il" does require a verbal meaning, but that does not mean that only roots whose headwords are verbs can be used. It means that if you put a word before "-il" then its verbal (-i) meaning is assumed. If "vorti" had a clear meaning, then "vorti" + "ilo" would be what is assumed to be meant by "vortilo". This is one flexibility of the Esperanto system.

See for example, how the suffix is described in the PMEG. In particular:

"Tio, kio staras antaŭ IL, ricevas agan signifon. La aga signifo estas tiu, kiun la radiko havas kun verba finaĵo."

(Translation: "That which stands in front of IL, receives an action meaning. The action meaning is that which the root has with the verbal ending [-i].)

One example is "buterilo". "Butero" = "butter", but "buterilo" = "buteri + ilo" (a tool for buttering: a butter knife).

("Butero + ilo" would be like a knife made of butter or something... )

Ah! If language were but mathematical! But no! Language is somehow coupled with that bothersome little brother culture - always! Language can't just kick the kid out - well, maybe language can exchange one little culture for the other like the Brits and we do somewhat. We sort of use the same language but look up, little culture kid! Look up at your big language and oops! maybe it's a twin - well, maybe not an IDENTICAL twin - not ever really born at the same time and in the same place but sooooo much alike and yet different.
So Esperanto is a language developing its own culture but this little brat nor Esperanto belong to the speakers of any particular ethnic language. Dang! If I could just get Esperanto alone for a while... But what a flit is Esperanto, and, yet, indeed a very pretty one at that! And that bothersome little bro influences the big language as well - and who can reason with culture? So who can reason with its bigger sibling? Not I!

It means that if you put a word before "-il" then its verbal (-i) meaning is assumed. If "vorti" had a clear meaning, then "vorti" + "ilo" would be what is assumed to be meant by "vortilo"

Exactly Eldanar.

And this neatly underlines that a root doesn't necessarily belong to a single grammatical class.