I hope that Describing Morphosyntax has got its categories better sorted out than in your exposition. Subject and Object are grammatical categories, purely matters of syntax (the NP of S => NP + VP, for example). Case is (generally) a morphemic matter:Nominative, Accusative, Ergative, etc. marked by the presence (or absence) of typically dependent morphemes attached to nouns (another grammatical category). Agents and patients and experiencers (different from patients") are semantic categories, in a broad sense (I would call them pragmatic, having to do with how the sentence fits into in context). There are all sorts of interconnections among these, varying from language to language: a subject tends to be in the nominative case (this is probably definitional) and to be the closest thing to a human in the sentence. And so on.
But, since tp has no cases and no inherent connections to agents or patients, none of this seems to help with the question of how to say I am sick. Or, worse, that this dog feels soft (to me). Given the restriction that 'pilin', like a number of other words in tp, is restricted to persons, or at least sentient being, it appears to be the verb for experiences, with it subject the experience and its object the experienced (?).
Happy Jule
Re: Happy Jule
Hey, I'm just a fake linguist who studies fake languages. If I'm half intelligible I consider that success.
Re: case in an (almost) purlely isolating language
Well, you know what I meant. Languages with morphology have compounds, but isolating languages don't-- but almost no one knows periphrasis, which is the best word I've come across to describe the corresponding process in isolating languages.
From the chapter on identifying constituent order (which is basically, a question of how Subjects, Objects and Agents and Patients work)
From the chapter on grammatical relations:
mi pilin kiwen ala e soweli. (exact parallel to mi pilin pona.)
soweli li pilin sama kiwen ala tawa mi. The dog isn't really the agent here...
pilin soweli li sama kiwen ala tawa mi. This predicate version isn't too bad, since predicative sentences are usually about static situations.
mi pilin e ni: soweli li kiwen ala. Ni is referring to the whole sentence that follows.
mi pilin e soweli ni: soweli li kiwen ala. Now it sounds like I physically touched the dog. The above sentence sounds like I might have inferred the situation.
From the chapter on grammatical relations (and grammatical relations can be expressed morphologically, syntactically)
tp's specification might be vague on grammatical relations but people's usage of tp isn't, imho. People are treating the grammatical relations of tp more or less just like any other European nominative-accusative (subject/direct object) system.
Re: case in an (almost) purlely isolating language
Well, you know what I meant. Languages with morphology have compounds, but isolating languages don't-- but almost no one knows periphrasis, which is the best word I've come across to describe the corresponding process in isolating languages.
From the chapter on identifying constituent order (which is basically, a question of how Subjects, Objects and Agents and Patients work)
But English mostly chooses AVP-- b through f are uncommon and have a forced feel to them. We shoe horn everything into AVP, including the intransitive, AV and including things that don't have a strong agent.One final caution: almost any language can be claimed to allow all possible orderings of A, P, and V if enough different kinds of constructions are included.
a. Fred skins mules AVP
b. It's mules that Fred skins PAV
c. (There he sits), skinning mules, that Fred. VPA
d. Skins 'em, Fred does to them mules. VAP
e. Fred's a mule skinner. APV
f. That mule skinner's Fred. PVA
From the chapter on grammatical relations:
So some things are bang on agents or patients and some aren't. In toki pona, something has to go in the subject or agent slot, even if there isn't anything good to put there. As for how to say "This dog feels soft", we have to put something in the subject and the rest goes either after the verb, direct object or prep-phrases. Feeling the dog fits the nominative-accusative pattern even worse than I feel good. The dog isn't much of a recipient of any action and I'm not doing much other than experiencing a subjective opinion. But since I have to fit these square pegs in round holes, we get things like:Modern functional linguists would take a different approach in defining grammatical relations. From functional point of view, the obvious, though inexact, relationship between pragmatic statuses/semantic roles and grammatical relations is motivated in terms of the notion prototype plus grammaticalization (see Payne, sec. 02). A noun phrase that is both a very good semantic Agent and a very good pragmatic topic is likely to be expressed as a grammatical subject. A functionalist would say that such a noun phrase is a prototypical subject. It is the kind of noun phrase in terms of pragmatic/semantic role that provides the functional basis for the formal category of subject in the first place. It is a very useful category, therefore it makes sense that languages should have a highly automated (grammaticalized) way of expressing it.
Now, what happens when a noun phrase refers to a slightly less prototypical Agent, or a less proptotypical topic? As mentioned above, languages tend to have only about three grammatical relations. This indicates that pragmatic or semantic deviation can be quite significant before a nominal phrase is excluded from a particular GR. It would seem unreasonable and inefficient to have a grammatical distinction for every conceivable nuance in semantic/pragmatic roles. That would be like having entirely different word for every conceivable shade of color in the spectrum. Therefore, ‘clustering’ of pragmatic/semantic roles occur. Referents that are ‘close enough’ to the prototype are expressed by noun phrases in the same GR as are more prototype referents. Since the notion of ‘close enough’ is a judgment call on the part of language users, there is variability from language to language (even from situation to situation) as to how the roles cluster. For example, in English sentence John likes beans, the person who ‘likes’ is treated the same grammatically as the Agent of an agentive verb like ‘kill’ or ‘eat’
mi pilin kiwen ala e soweli. (exact parallel to mi pilin pona.)
soweli li pilin sama kiwen ala tawa mi. The dog isn't really the agent here...
pilin soweli li sama kiwen ala tawa mi. This predicate version isn't too bad, since predicative sentences are usually about static situations.
mi pilin e ni: soweli li kiwen ala. Ni is referring to the whole sentence that follows.
mi pilin e soweli ni: soweli li kiwen ala. Now it sounds like I physically touched the dog. The above sentence sounds like I might have inferred the situation.
From the chapter on grammatical relations (and grammatical relations can be expressed morphologically, syntactically)
So if an nominative-accusative language is making the subject of transitive and intransitive sentence the same (by location), and then we start shoehorning most of our sentences into AVP, then when we come across a sentence where we need VP (namely in experiencing something), then we end up with a semantically peculiar AV or AVP sentence.There are semantic and discourse-pragmatic factors that may motivate S and A isomorphism or S and P isomorphism [Subject/Agent, Subject/Patient]. First we discuss the semantic factors.
The semantic similarity between S and A is agentivity: if a clause has an AGENT, it will be the S or the A argument, depend on on whether the clause is transitive or intransitive.
a. Jorges stalked out of the room. S is agentive.
b. Wimple embraced the Duchess. A is agentive.
The semantic similarity between S and P is change of state. If any argument in a clause changes state, it will be either the S or the P.
a. The bomb exploded. S changes state.
b. Lucretia broke the vase. P changes state.
...
tp's specification might be vague on grammatical relations but people's usage of tp isn't, imho. People are treating the grammatical relations of tp more or less just like any other European nominative-accusative (subject/direct object) system.
Re: Happy Jule
But being a linguist of sorts with fake languages is a good way to become a good real linguist, even if only of the hobby variety. So, in that direction, it is important to be clear about a number of distinctions (and, alas, some of the technical terms that go with them). First there the branches of linguistics: phonology (sounds and how they go together to form phonemes), morphology (meaningful strings of sound and how they go together to form morphemes and these go together to form words), syntax (how words go together to form sentences -- and intermediate phrases along the way), semantics (the meaning of morphemes and how these interact to give meaning to all the structures up though sentences -- and beyond), pragmatics (tying that meaning into the context of utterance), rhetoric (dealing with the emotional and social aspects of the sentence -- this sometimes folded into pragmatics, and, indeed all these divisions are somewhat arbitrary in particular cases).
So, in particular, every language has a morphology. Languages range from the totally isolating type, where every morpheme is free, a word in itself, to the polysynthetic, where every morpheme is bound and words are always complex -- with various stops in between.
"periphrasis" is a word in comparative linguistics: what one language does within a word, another does using two or more words. Of course, this assumes that we can identify "what a language does" and this usually turns out to be a semantic category of some sort (or a pragmatic one). It may be the case that the two languages really are doing "the same thing" in different ways, bt it generally turns out that they are doing slightly (or grossly) different things which can sometimes be substituted for one another. tp is a totally isolating language, so, inevitably, if it does something like watt some language does with cases or tenses or whatever, it does it periphrastically. But there is no a priori reason to think that tp does do the same thing as some other language.
Subject and Object and Agent and Patient are pairs of categories from two different fields (syntax and semantics), but they are highly coordinated in many (most? all?) languages, though they may mean somewhat different things in each. So it is important to keep them separated to be clear what you are saying. Thus, the whole talk about the order of agents and patients, is misleading, since semantic categories are nonlinear for the most part, If we keep that in mind, of course, then the talk about the order of the categories can be understood as shorthand for the order of the words that bear these categories. But it is nice to know that the person doing the shorthand is aware that he is doing so.
So, in particular, every language has a morphology. Languages range from the totally isolating type, where every morpheme is free, a word in itself, to the polysynthetic, where every morpheme is bound and words are always complex -- with various stops in between.
"periphrasis" is a word in comparative linguistics: what one language does within a word, another does using two or more words. Of course, this assumes that we can identify "what a language does" and this usually turns out to be a semantic category of some sort (or a pragmatic one). It may be the case that the two languages really are doing "the same thing" in different ways, bt it generally turns out that they are doing slightly (or grossly) different things which can sometimes be substituted for one another. tp is a totally isolating language, so, inevitably, if it does something like watt some language does with cases or tenses or whatever, it does it periphrastically. But there is no a priori reason to think that tp does do the same thing as some other language.
Subject and Object and Agent and Patient are pairs of categories from two different fields (syntax and semantics), but they are highly coordinated in many (most? all?) languages, though they may mean somewhat different things in each. So it is important to keep them separated to be clear what you are saying. Thus, the whole talk about the order of agents and patients, is misleading, since semantic categories are nonlinear for the most part, If we keep that in mind, of course, then the talk about the order of the categories can be understood as shorthand for the order of the words that bear these categories. But it is nice to know that the person doing the shorthand is aware that he is doing so.
Re: Happy Jule
[iPads have definite shortcomings for replying in this venue.]
So the words that refer to the agent and patient may be in any position with respect to the verb (I'll skip the screed about how insignificant surface order is for almost anything) (btw, about the examples, some of these are needlessly awkward though only c is actually odd, and apparently mule skinners don't skin mules, the mule skinner. Skinning mules is left to butchers, taxidermists, and people at the beginning of leather making.) What is the significance here? Apparently to reenforce the connection between the semantic terns and the grammatical ones. But it actually works the other way, since the subject tends to wander independently of the agent and sometimes disappears altogether. It is much harder (though probably not impossible for English) to get the subject and object in all those positions. It is true that English and tp tend to take SVO as basic, but that doesn't guarantee that the basic order for the semantic components is AVP (though it probably is). So what we do do is shoehorn everything into SVO, just dropping off the O for intransitives. Actually we don't do all that much shoehorning in English, since we have a lot of patterns to us. We do do it in tp. since SVO is all we have (though within that we can at least PVA semantically, and maybe PAV). Some intransitives are AV, but many more are probably PV or some not yet discussed semantic category.
The second quote seems to be the modern version of medieval psychologism, tossing away the gains in linguistics in the intervening centuries. That a grammatical category should be specified by a fuzzy set of semantic or pragmatic categories is doing linguistics backward. It may be that, when all the work is done, we can see such categorizations. And it may be that this other approach is handy for teaching certain kinds of fluency, but that is way later. And maybe the fluency issue is a good one to look at in dealing with tp problems. If the dog feels fuzzy then the dog is a patient (I feel the dog) and also whatever the subject of a stative verb is going to be (I think that he is fuzzy) (This case is complicated by the fact that only sentient being can feel and then that is very tactile: "The dog feels fuzzy" could mean that it has its nose in velour.). Of your sample translations, them first has the wrong subject and probably the wrong verb, even if we went with the 'pilin ike' version of being sick. The second dodges the second problem but not the first. I frankly don't understand the third one "Feeling a dog is like softness to me". The next to last is pretty good except it leaves out your feeling the dog. The last is fine, but I wish we could make it shorter somehow.
So the words that refer to the agent and patient may be in any position with respect to the verb (I'll skip the screed about how insignificant surface order is for almost anything) (btw, about the examples, some of these are needlessly awkward though only c is actually odd, and apparently mule skinners don't skin mules, the mule skinner. Skinning mules is left to butchers, taxidermists, and people at the beginning of leather making.) What is the significance here? Apparently to reenforce the connection between the semantic terns and the grammatical ones. But it actually works the other way, since the subject tends to wander independently of the agent and sometimes disappears altogether. It is much harder (though probably not impossible for English) to get the subject and object in all those positions. It is true that English and tp tend to take SVO as basic, but that doesn't guarantee that the basic order for the semantic components is AVP (though it probably is). So what we do do is shoehorn everything into SVO, just dropping off the O for intransitives. Actually we don't do all that much shoehorning in English, since we have a lot of patterns to us. We do do it in tp. since SVO is all we have (though within that we can at least PVA semantically, and maybe PAV). Some intransitives are AV, but many more are probably PV or some not yet discussed semantic category.
The second quote seems to be the modern version of medieval psychologism, tossing away the gains in linguistics in the intervening centuries. That a grammatical category should be specified by a fuzzy set of semantic or pragmatic categories is doing linguistics backward. It may be that, when all the work is done, we can see such categorizations. And it may be that this other approach is handy for teaching certain kinds of fluency, but that is way later. And maybe the fluency issue is a good one to look at in dealing with tp problems. If the dog feels fuzzy then the dog is a patient (I feel the dog) and also whatever the subject of a stative verb is going to be (I think that he is fuzzy) (This case is complicated by the fact that only sentient being can feel and then that is very tactile: "The dog feels fuzzy" could mean that it has its nose in velour.). Of your sample translations, them first has the wrong subject and probably the wrong verb, even if we went with the 'pilin ike' version of being sick. The second dodges the second problem but not the first. I frankly don't understand the third one "Feeling a dog is like softness to me". The next to last is pretty good except it leaves out your feeling the dog. The last is fine, but I wish we could make it shorter somehow.
Re: Happy Jule
I've actually been experimenting with constituent order in TP.janKipo wrote:[iPads have definite shortcomings for replying in this venue.]
So the words that refer to the agent and patient may be in any position with respect to the verb (I'll skip the screed about how insignificant surface order is for almost anything) (btw, about the examples, some of these are needlessly awkward though only c is actually odd, and apparently mule skinners don't skin mules, the mule skinner. Skinning mules is left to butchers, taxidermists, and people at the beginning of leather making.) What is the significance here? Apparently to reenforce the connection between the semantic terns and the grammatical ones. But it actually works the other way, since the subject tends to wander independently of the agent and sometimes disappears altogether. It is much harder (though probably not impossible for English) to get the subject and object in all those positions. It is true that English and tp tend to take SVO as basic, but that doesn't guarantee that the basic order for the semantic components is AVP (though it probably is). So what we do do is shoehorn everything into SVO, just dropping off the O for intransitives. Actually we don't do all that much shoehorning in English, since we have a lot of patterns to us. We do do it in tp. since SVO is all we have (though within that we can at least PVA semantically, and maybe PAV). Some intransitives are AV, but many more are probably PV or some not yet discussed semantic category.
The second quote seems to be the modern version of medieval psychologism, tossing away the gains in linguistics in the intervening centuries. That a grammatical category should be specified by a fuzzy set of semantic or pragmatic categories is doing linguistics backward. It may be that, when all the work is done, we can see such categorizations. And it may be that this other approach is handy for teaching certain kinds of fluency, but that is way later. And maybe the fluency issue is a good one to look at in dealing with tp problems. If the dog feels fuzzy then the dog is a patient (I feel the dog) and also whatever the subject of a stative verb is going to be (I think that he is fuzzy) (This case is complicated by the fact that only sentient being can feel and then that is very tactile: "The dog feels fuzzy" could mean that it has its nose in velour.). Of your sample translations, them first has the wrong subject and probably the wrong verb, even if we went with the 'pilin ike' version of being sick. The second dodges the second problem but not the first. I frankly don't understand the third one "Feeling a dog is like softness to me". The next to last is pretty good except it leaves out your feeling the dog. The last is fine, but I wish we could make it shorter somehow.
AVP: soweli li moku e waso. The animal eats the bird. Standard operating procedure.
APV: soweli e waso li moku. Still good. (Working on an TP/Pirahan APV conlang, actually)
PVA: e waso li moku soweli. soweli looks like an adverb (German fressen vs. essen, perhaps?).
PAV: e waso soweli li moku. soweli looks like an adverb. With dummy pronoun: waso la soweli li moku ona.
VAP: li moku soweli e waso. soweli looks like an adverb. With dummy pronoun: soweli la li moku ona e waso.
VPA: li moku e waso soweli. soweli looks like an adverb. With dummy pronoun: waso la li moku ona soweli. Perhaps?
Re: Happy Jule
I know that you know that only the first of these is in toki pona, but I am not sure whether you think that some of the others might be reasonable additions to tp to ease up the restriction to SVO order (not AVP order, for reasons laid out above). By and large, think they are unlikely candidates because to much fundamental information is lost in the transformations. But --keeping it well out of Sonja's sight -- it might be worthwhile to figure out what would be needed minimally to allow these shifts in order, and maybe even some shifts in meaning (passives, for example). An easy one, requiring nothing new, would be to allow 'la' phrases to be topic or focus: waso la soweli li moku e ona. Moving the verb will be harder.
Re: Happy Jule
The final quote is farrago of mixed categories, with S, A , and P appearing sometimes a grammatical concepts, sometimes as semantic, even in the same sentence (" the agent is agent" is apparently not meant to be a tautology. As grammatical category, agent is apparently ergative, with subject playing the absolutive role ( though patient apparently does too -- mainly in lieu of object). We are told that the mark of patient is change of state, but in the first sentence, where the subject changes state twice in one short sentence, it is said to be agent. And so on. I don't see any reason to base any linguistic study of tp on this book, no matter whether Mato has generally reported it accurately, these direct quotes pretty much show it to be confused, if not criminally negligent.
Re: Happy Jule
Oh, I have no intention of changing Toki Pona in that way - especially since it pollutes the statistical analyses some jan pi TP love. I just find the small vocabulary of TP makes thinking about constituent order easier (the topic in your example is one of the things that struck me immediately). I do know the difference between AVP and SVO - I shall try to be clearer in the future. The only one that could work as an alternative (for poetry, perhaps) are SOV - everything else requires a postpositive 'la' and a resumptive pronoun, which IMO is cheating unless 'la' is a topic marker (which it isn't, since it doesn't increase the prominence of the fronted material or even state that the material in the 'la' clause is important). Confiteor: I did just use a resumptive pronoun in 'musi pi sike ma sin': ko li anpa kama la ona li ante tawa telo lete.janKipo wrote:I know that you know that only the first of these is in toki pona, but I am not sure whether you think that some of the others might be reasonable additions to tp to ease up the restriction to SVO order (not AVP order, for reasons laid out above). By and large, think they are unlikely candidates because to much fundamental information is lost in the transformations. But --keeping it well out of Sonja's sight -- it might be worthwhile to figure out what would be needed minimally to allow these shifts in order, and maybe even some shifts in meaning (passives, for example). An easy one, requiring nothing new, would be to allow 'la' phrases to be topic or focus: waso la soweli li moku e ona. Moving the verb will be harder.
Re: Happy Jule
Since the guy who wrote the book you cite does not seem to know difference between SVO and AVP, I wonder where you got it.
I do think the rigid place structure of tp is unnecessary for keeping the language simple, so a few transformations might make for some variety.
I'd use'kama' rather than 'ante tawa' in the last, for the usual agent/ patient reasons. Also, don't need Kama with anpa, but, if you have, it comes first.
I agree about revised ''la'. I'm not sure what resumptive pronoun is, but there doesn't appear to be anything wrong with your examples
I do think the rigid place structure of tp is unnecessary for keeping the language simple, so a few transformations might make for some variety.
I'd use'kama' rather than 'ante tawa' in the last, for the usual agent/ patient reasons. Also, don't need Kama with anpa, but, if you have, it comes first.
I agree about revised ''la'. I'm not sure what resumptive pronoun is, but there doesn't appear to be anything wrong with your examples
Re: Happy Jule
At least the author in the example in the book I was showing wanted to express the point that if there are many was to say something in a language, one might be able to find a variety of SVO/AVP orderings that were reasonably grammatically correct. (and I wanted to use this example to express that the typical syntactic roles don't line up tidy with the semantic roles-- the do-er of a sentence might be off in an oblique phrase instead of in the subject phrase)Mako wrote:I've actually been experimenting with constituent order in TP.
AVP: soweli li moku e waso. The animal eats the bird. Standard operating procedure.
* APV: soweli e waso li moku. Still good. (Working on an TP/Pirahan APV conlang, actually)
* PVA: e waso li moku soweli. soweli looks like an adverb (German fressen vs. essen, perhaps?).
* PAV: e waso soweli li moku. soweli looks like an adverb. With dummy pronoun: waso la soweli li moku ona.
* VAP: li moku soweli e waso. soweli looks like an adverb. With dummy pronoun: soweli la li moku ona e waso.
* VPA: li moku e waso soweli. soweli looks like an adverb. With dummy pronoun: waso la li moku ona soweli. Perhaps?
waso li kama moku. soweli li kama jo moku. waso li moku -- Subjects only. No objects.
moku li pali pi soweli tawa waso. -- "verb", agent, patient
waso li kama moku tan soweli - patient, verb, agent
waso li kama moku tan moku soweli - patient, verb, agent (bird became food because of the animal's eating)
soweli la waso li pana e moku. agent - patient -verb.
waso la soweli li kama jo e moku. patient - agent - verb -
? soweli en waso li moku e wan tawa tu. -- agent and patient first and then verb.
If the syntactic roles and semantic roles have to line up exactly, then none of the above should be intelligible at all to anyone. They don't and don't have to line up, except out of stylistic concerns. Since they don't have to we can shoe horn the agent and patient into unusual positions and still be understood. Anyhow, the idea works well enough for me, so like a clock that's right twice a day, that's good enough for me. It explains why descriptions of state are forced into agent - patient paradigms- why they look peculiar when they do and why one shouldn't worry much about it.
Re: the author of Describing Morphosyntax not knowing the difference between agents and patients.
And I don't feel like taking on the whole what linguistics are currently thinking-- I simply don't have the resources to presume that everything that I read in "Describing Morphosyntax" (or Chomsky, or Pinker or anywhere) is wrong and then rebuild it all up again from scratch. And even if I grievously misunderstand them, I figure that a misunderstood authority (backed up by the usual evidence-- broad community acceptance, got published by a respectable university press, has Phd, etc) beats whatever I can coble together on my own resources alone.
Any how, I suppose we can agree that we probably don't understand each other and move on to something more interesting like what number systems haven't been proposed in toki pona yet.