A short study on semantics

[mouse]

He pocked the slubber off the ground and jibberly sugged it crouse-about. It was far enough vim to decranth the creppler's dreck, so he javed one more sug abard and encreathled the lout slubber entack. Sleavening along vimly and wrastering tot abard he brandoubled accoutly into a snib all tubbered in grib. "Ack!" He voled, now tubbererd tot himself, he littened accote and detacked the slubber of its creath. He bradened the slubber atwine him and the snib. "Are you the creppler cloring this hobble?" he sneavely stickered in a tibberous voice, his knees atibberous too.
"You dack deroove the Glandorious Snib for a mibble creppler?" berubbled the snib.

Obviously, this little segment is inspired by Jabberwocky. I tried to write it both grammatically correct and using lots of sound symbolism to make it possible to guess at what's going on, although I really don't know how well I did in that latter respect.

So here's a little bit of analysis;
Slubber
Slubbers can be pocked off the ground
Slubbers can be sugged
Slubbers can be encreathled entack, and detacked
Slubbers can have a creath
Lout slubber
Vim slubber
Creppler
Crepplers have a dreck, and that dreck can be decranthed
Crepplers clore hobbles
Mibble creppler
Snib
Snibs can be tubbered
Glandorious snib
Snibs can communicate somehow
Grib is a substance of manner of tubbering something, and it is communicable
Berubbering, stickering, and voling are all methods of vocalising.

I believe that any 'full' language (so any natural language, but probably not computer languages or most animal communication systems, and probably not many conlangs) ought to be able to withstand such replacements with gibberish without being grammatically ambigious. Does anyone know if it's true? Does the amount of substitution a language can withstand depend on the language? I'd imagine heavily inflecting languages would be much more flexible in this manner.
Overall this is just a bit of nonsense, but I was curious to see if anyone could apply this to other languages and see how it comes out? I'd be particuliarly interested in seeing how Chinese and Austronesian languages would deal with this. (Can anyone do this in their conlangs, too?)

[Ser]

Why would Chinese have problems with this? True, there's almost no inflection, but it has many markers for grammatical functions.

[zompist]

Yeah, I don't see that any of your list of deductions depends on any point of inflection. You could translate the passage into spoken Chinese and probably learn the same things.

Actually, the interesting thing would be what you did with measure words. Apart from actual measures like 'kilogram', I think this is a closed and unproductive class, so you should really use existing ones, and that would actually give extra information about many of the nouns. E.g. if you use tiao2 it's likely you're dealing with a long narrow object. (If you did make up measure words, which ones went together would also give information. E.g. if slubbers and crepplers use the same measure word, they have some (very vague) resemblance.

Written Chinese would require making up characters; the problem is that there is no way to suggest the correct pronunciation (though you can give clues). What radicals you use would also give semantic clues-- e.g. using the 'plant' or 'insect' or 'word' or 'person' radicals would all suggest a general classification.

Yuen Ren Chao translated Jabberwocky into Mandarin, so it's certainly not impossible.

[mouse]

That verifies my suspicions.

Well, we expect a certain type of word to appear in that context. I was trying to put nonsense words because I wanted to see if how much of the meaning of the word you can tell by grammar alone. For example, if you saw the same verb several times you could determine the valency of the verb and then perhaps in conjunction with prepositions or particles could determine that it must be a verb of motion, or certain idiomatic phrases where we know that the word in X position must be an animate noun or an adjective of measurement. Grammatical constructions that have a semantic implication. The count-words are exactly part of what I was getting at, there.
Would it be possible to construct a loglang which entirely removes the semantic implications of grammar? And in the opposite, how much of meaning can be placed on morphology and syntax instead of individual words? I doubt it would be possible, but I'm thinking a thought expirement conlang where morphology and syntax represent meaning and the lexemes show the relations between the words (although I think that'd be pretty near impossible, since there's only so many different combinations of words and such, but many more possible phenomenon that could be refered to in a sentence. Although...maybe, using many many dummy pronouns and lots and lots of particles for word class or very general semantic sets, and lots of derivation. Definantly it would not be practical, though.). My point is that I think the distinction between grammar and semantics may not be so hard set as it seems made out to be.

Perhaps I should have used [blank] for every gibberish word, but I thought that'd be too hard to read.

Also, I wasn't suggesting translating that segment there, just suggesting perhaps to write such a nonsense passage.

It is like how children learn languages, or how historians decode a passage from an extinct language.

[zompist]

Would it be possible to construct a loglang which entirely removes the semantic implications of grammar?

You can get close-- e.g. predicate calculus, where the only real information is whether something is a predicate or an argument-- plus a full grammaticalization of conjunctions, negation, and quantification.

However, natural language grammaticalize a whole lot more than this, and that's a good thing. It's a failing, not a virtue, of predicate calculus that it's extremely awkward for expressing tense/aspec/mood, or topicality, or emotional content.

My point is that I think the distinction between grammar and semantics may not be so hard set as it seems made out to be.

There is no distinction. I think what you're noticing is the concept of grammaticalization. Some aspects of the world, with semantics attached, become part of the structure of the language. Obviously plurals, for instance, relate to the semantic notion of quantity.

(Now, doing syntax, you're encouraged not to use semantic categorization-- mostly because it leads to sloppy analysis and may actually hide a language's real structure. But I think we needn't pretend that syntactic categories have no semantic content at all.)

[Ser]

Written Chinese would require making up characters; the problem is that there is no way to suggest the correct pronunciation (though you can give clues). What radicals you use would also give semantic clues-- e.g. using the 'plant' or 'insect' or 'word' or 'person' radicals would all suggest a general classification.

I think some very creative translators could try that, but Chinese already has a number of characters/syllables that are commonly associated with borrowings and onomatopoeia. It's what they use whenever news services bring foreign names to the language for example. I would expect a translator to use such characters/syllables to make up seemingly non-sensical words.

[Vuvuzela]

(Can anyone do this in their conlangs, too?)

Anma dusuvajlaveqpukutse, (W/out incorporation: Jlavere anma dusuvapukutse), grausiup laasuge suvajivvere. Suvimak ragrevle sudrege degradep łiłkigappaqquti, kesuge vitiip ripsukijavese, kiavardiptaa, egritesse jlavere lautip, ripkinnaptaa.Jliivene łiivavime laarastere, rattut kiavaptaa, sinive kaavraddavle vaakut, jinive grive duvatuverekup."Ak!" suvavole, pusuq tut duvatuveruqikup, akkut valitatse, jlavere griite laadusuvadetak. Jlavere suvavradetse, jlavere susinivlaa taasuvaadvine.
«Juratgrevluu glorip jlaatusuvle?»leepu tivverip  dustikere vasinvive, suvuiq łiativeqpek.
«Dage sinive glannuriip ragrevle mivlip liktuvaderuve?» sinive vaveruvlese.
Edit:Tiny grammar mistake

[Thomas Winwood]

As a study into how far implication in the manner Zomp mentioned can take you, I encourage you to play The Gostak.

[mouse]

However, natural language grammaticalize a whole lot more than this, and that's a good thing. It's a failing, not a virtue, of predicate calculus that it's extremely awkward for expressing tense/aspec/mood, or topicality, or emotional content.

That was my original hypothesis. There's a phrase you used in the Language Construction Kit that very precisely reflects my philosophy in just about everything from ecology to sociology to neurology to linguistics - "Very probably the degree of redundancy of human languages is pretty precisely calibrated to the minimum level of information needed to cope with typical levels of distortion." substituting 'human language' for whatever applies.

The whole reason I was doing this little expirement was because I wanted to demonstrate that what elements are grammaticallised are generally so across languages because they're calibrated to be be as succinct as possible while still giving all the information (and self-reinforcing redundancy) nessesary to communicate the point. I was going to do this by with a series of examples (this one here first) culminating in the two loglangs I'd mentioned, one with no gramatical semantic content and one where as much as possible the semantic content was reflected in morphology and syntax and the relationship of the participants was reflected lexically.
The idea was that when the semantic content of verbs (the hot water implied by 'boil' for example) was explicitly given nouns in a phrase, verbs are little more than shorthand for showing the thematic relation of the participants (in addition to prepositions and word order and cases and animacy and all that), and that the 'perfect' noun is the proper noun and common nouns are little more than adjectives of a sort. That all off these relations can be shown as a sort of series of sliding scales, where the information is all there in any natural language, but the amount of it expressed within the verb against within word order or topicalisation or something other slides back and forth. That any sentence should be capable of being expressed without the verb through the use of adpositions or other markers. ('I speared the deer' becomes 'From my hand (with strength) the spear in the air at into the deer) This isn't a theory- I'm not saying this is how languages arose or that this is a way of predicting anything about the function of languages. This is merely a view I have found helpful in constructing languages, I suppose. It helps me to work out this beforehand and that once dealt with this then it's time to start doing all the things like making nouns that can never take the subject role in a phrase, or mandatory grammatical marking of the edibility of a noun. I guess you could say I was trying to find a universal view of thematic alignment.

Infact, the whole reason I got interested in this puzzle in the first place was cause I was(am still) playing a 'Language game' with some of my friends. They wanted to make a conlang, but they wanted it to arise 'naturally' so we stopped talking English and started babbling with the sole intent of getting our point across. And it ended up developing a couple of very unusual features I couldn't explain. One thing for example is that it has something sort of like an Austronesian alignment combined with topic comment word order, except that the comment often comes first or something, I still don't exactly know what's going on. It probably doesn't help that after a while I started expirementing introducing odd things. Such as the beforementioned word for 'house' that means 'ruins' in past tense. That one didn't catch on. I don't think they actually knew I was trying to make a distinction, and I didn't want to mention it outright. Two of these friends want to raise their (as yet unborn) child bilingually in English and this language of ours. That promises to be extremely interesting.

I thought I was on to something new there, but it looks like I'm late to the game, and assuredly all the mistakes I would have made have already been thought of and corrected. It was fun to work out, but I'm glad to see I wasn't completely on the wrong track.

There is no distinction. I think what you're noticing is the concept of grammaticalization. Some aspects of the world, with semantics attached, become part of the structure of the language. Obviously plurals, for instance, relate to the semantic notion of quantity.

Plurals are an example I was fooling around with in my notes, imagining a language with a separate word for every commonly pluralised noun. I also was fooling around with lexicalising tense- having words that mean 'house' only in the present, or only in the past. (With interesting things like if one used the past word for 'house' in a present tense sentence it would mean 'ruins' or the past word for 'child' meaning 'adult' in the present tense (with some words not changing in meaning, and simply meaning exNOUN, and then other words not having this distinction and there being sets of prefixes to introduce this meaning to those nouns.))

(Now, doing syntax, you're encouraged not to use semantic categorization-- mostly because it leads to sloppy analysis and may actually hide a language's real structure. But I think we needn't pretend that syntactic categories have no semantic content at all.)

Thankyou for the information!

I guess this just goes to show how little linguistics knowledge I really have- just the LCK, Describing Morphosyntax, a textbook on Syntax, The World's Languages (Bernard Comrie), and whatever I've picked up here (The ZBB- The best college course I've ever had). Any recommendations on books to learn more?

(Haha! I remembered to log in in another window before I clicked Preview this time, instead of accidentally losing my post like always! I've foiled you, you dastardly forum!)

[Miekko]

What would happen in a language where lexical hierarchies are a very significant factor in determining the roles of nouns in the sentence if we tried translating something like this into it?

[Radius Solis]

Two of these friends want to raise their (as yet unborn) child bilingually in English and this language of ours. That promises to be extremely interesting.

Maybe not so much. You should take a look at this. Short version: for the most part, children don't learn more languages than they absolutely have to.

[clawgrip]

Two of these friends want to raise their (as yet unborn) child bilingually in English and this language of ours. That promises to be extremely interesting.

Maybe not so much. You should take a look at this. Short version: for the most part, children don't learn more languages than they absolutely have to.

I can vouch for this. I'm having trouble getting my daughter to speak English.