erinja:I don't know if every paper in this field has an English version. Much of the work is published in Esperanto, which is an excellent chance to improve your Esperanto.

Yes I do have great hopes for improvement by this method . And you're indeed correct, much of the material do not seem to have translations. Luckily for my work, I find written Esperanto quite comprehensible even at my level (and it is indeed getting more and more accessible the more i read), it is just that producing grammatically correct Esperanto myself is tricky, and it is important that the grammar yields grammatically correct results.

Questions about whether you can or cannot say something in a certain way in Esperanto are of general interest to all learners. So Forum members will always be happy to debate this.

However, what you label as 'ontological' matters, and I would call issues of meaning or semantics, seem to me to be intimately bound up with the way Esperanto works.

All words in Esperanto which aren't members of the small group that do not need a finaĵo are constructed according to a principle of as many roots as you need, and are sufficient, to make your meaning clear. Of course, this means clear to a human being equipped with normal intelligence and a knowledge of how the world is.

This is sometimes referred to as la principo de neceso kaj sufiĉo.

And even words with are compose of a single root and a finaĵo are just such compounds.

The reason why la viro tie estas mia amiko is OK and la viro tie estas dormo is not OK, seems to me perfectly explicable on semantic grounds.

On its face, someone might say that Esperanto is not a good model for 'natural' languages, because it is regular and 'natural' languages are not.

Although the word building property of Esperanto is more regular than other languages, the idea of having a set of suffixes isn't unique to Esperanto; it's just that Esperanto deploys it on a wider scale, and with more regularity than other languages. In a sense, you could say, for example, that the Esperanto suffix -ec- corresponds, to a degree, to -ness in English. You won't find every single valid -ness word in an English dictionary, we apply the suffix when we need it. redness, goodness, hardness. We don't use it when there's another noun form of an adjective that we can apply (beauty, not *beautifulness; as an analogy to Esperanto, you might say that beauty is a noun root in English, whereas it is an adjective root in Esperanto)

However, as sudanglo says, the principle of "neceso kaj sufiĉo" also applies. Although Esperanto is formed in a regular way, it is spoken by humans, not computers, and the way that we parse it out requires a human brain. We choose the interpretation of a word that intuitively makes the most logical sense in a human life. In some cases, it might be quite different from the way that a computer would parse a word according to the strict definitions of a root and its affixes (a virino is therefore a woman, not a female man)

And a human knows the difference between a koleg/o (a colleague) and a kol/eg/o (a big neck). Those words that can be parsed in two different ways require context to distinguish.

It's hard to see how the regularity of the correlatives would present a problem. They are a closed system, so unlike Esperanto's affixes, you can't just tack the correlative component parts onto any old word. You'd just put them into the system as individual words, just as you'd put English correlatives as individual words (what, that, when, then, where, there, rather than putting in wh/, th/, /at, /en, /ere, etc)

sudanglo:
The reason why la viro tie estas mia amiko is OK and la viro tie estas dormo is not OK, seems to me perfectly explicable on semantic grounds.

Obviously both are grammatically correct, the second one just doesn’t make any sense, just like "green colorless ideas sleep furiously".

erinja:It's hard to see how the regularity of the correlatives would present a problem. They are a closed system, so unlike Esperanto's affixes, you can't just tack the correlative component parts onto any old word. You'd just put them into the system as individual words, just as you'd put English correlatives as individual words

Hi Erinja,

Thank you for more good insights and suggestions!

With the correlatives however, getting them into the system is not hard at all, the manner you describe is kind of how I implemented them at present, but I would like the grammar model whenever reasonable to reflect the Esperanto grammar, in order for the user who interacts with it to get the facts right about Esperanto: How and where and if it differs from the underlying model of language the system represents.

This means that I would like to represent the correlatives not as unrelated words scattered over many part-of-speeches (PoS), since this is not how the Esperanto grammar deals with them, but as a related group of words. So when you for example ask your system to show you some lexical items, you will in some way get informed that these words are part of the correlative system, and other words are not. This also have practical implications, the possessive forms need to tag along some of them when they are introduced into the lexicon, since some of the internal PoSes they will be tied to have rules deriving them into other internal PoSes, that might take possessive forms. All internal PoSes are identified with a type representing their inflection patterns and category properties, which means that if you for example want to inflect some adverbs for case (such as tie->tie-n), you'll simply have to let all your adverbs inflect for case (kiel->kiel-0), a process that can get rather redundant if a word that shows full inflection sneaks its way into a PoS that otherwise how no inflection or vice versa (c.f ĉiu - nur). It is easily dealt with by changing the overall language model, or making an exception for Esperanto, but this is exactly spot on; GF is a multilingual system intended to capture grammar "universals" and "universal patterns" in order to facilitate language implementation by using them as underlying structure shared by all the encoded languages, again whenever reasonable. This is why the correlatives get so interesting to me, is it reasonable and constructive to cut the Esp. grammar view of the correlatives into pieces and just mould it into the system's point of view, or can the systems point of view be used but enriched in the Esperanto implementation so we get the best out of both worlds, or does the Esperanto grammar simply say "No, I won't lend myself to this, you'll have to make an exception for me because I just don't look anything like the language model you've made up". At present, I have encoded them scattered, but tagged them as correlatives and in case they behave similar to pronouns, with their possessives.

I guess I would say that Esperanto's correlatives are just a more regularized version of something that already exists in other European languages. As speakers of those languages we don't think of our correlatives as something formed systematically from a table, but you can actually make a correlatives table for many languages, with a surprising degree of regularity (albeit with irregularities and "missing" words)

For example, this English-Latin correlative table isn't really so different from Esperanto's correlative table, in a lot of ways.

erinja:I guess I would say that Esperanto's correlatives are just a more regularized version of something that already exists in other European languages. As speakers of those languages we don't think of our correlatives as something formed systematically from a table, but you can actually make a correlatives table for many languages, with a surprising degree of regularity (albeit with irregularities and "missing" words)

For example, this English-Latin correlative table isn't really so different from Esperanto's correlative table, in a lot of ways.

Well, I tend to follow the motto "where there are patterns to find, there are patterns to use", and in this case, they are luckily in plain view . (Besides, I think both me and my computer like the same code written only once if possible, so it can be transparant, compact, easy to rewrite and easy to reuse if needed for other purposes by me or somebody else. Also I think from a work load point of view, my compiler prefers collecting values from preprocessed code, instead of unknowingly doing the same stuff again and again, in worst case during run time. So from my point of view, I'm all to happy with regularity!)

darkweasel:

sudanglo:
The reason why la viro tie estas mia amiko is OK and la viro tie estas dormo is not OK, seems to me perfectly explicable on semantic grounds.

Obviously both are grammatically correct, the second one just doesn’t make any sense, just like "green colourless ideas sleep furiously".

Yet, isn't there an important difference though between a nonsense phrase which at a certain time doesn't make sense, but in theory could make sense with the extension of the meaning of certain words, and one that could never make sense.

Green ideas didn't make sense when the famous nonsense phrase was invented but makes perfect sense today (ideas of the green party).

Could the 'o' in dormo ever come to mean a concrete thing rather than an act or state.

La viro tie estas nia venonta manĝo does make sense in a cannibalistic society
(despite manĝ being a so-called verbal root). But what would have to happen for dormo to be something that could be a man? Wouldn't this require the Universe to change?

Forigite

Hang about, HB. 200 years ago we had fairy stories and fantastical tales in which characters had the power of flight, and after a collision or an explosion objects went flying through the air.

So all that has happened is that technological advance has produced a way of your mother flying to Paris without using a witch's broomstick, or a magic carpet. It wasn't unintelligible to speak of your Mother flying 200 years ago.

But for dormo to be something concrete which a man could be would require a category change and is not something that could be brought about by technological invention.

However much you argue that X-o estas Y-o is a valid piece of grammar, there would be many examples in Esperanto where that wasn't possible and the explanation would be semantic.

Whether tiu X-o estas Y-o is 'well-formed', depends on X and Y, and calling that structure grammatically correct seems to serve little purpose other than to let you know that with some X's and some Y's that produces an OK sentence.

Kredu min, mi estas pravo!