PEOPLE, STOP POSTING. I don't like making such long posts... It... makes me feel bad for some reason.
Omzinesý wrote: ↑21 Mar 2020 10:27SAE languages are quite flexible when it comes to syntactic transitivity. You can always add an object to an intransitive verb. "sing a song", "sneeze a handkerchief over the room" etc. Similarly, a transitive verb can be used "intransitively" or with a generic implicit object "Corona kills.".
I believe that what you say is true in typical European languages, but I just wanted to point out that in Romance, especially in colloquial Spanish, there is a category of "reflexive transitive" verbs, bearing both a reflexive indirect object and a separate direct object, whose members as a whole can't be used with an implied dropped object ("Corona kills"). That may be interesting to you...
For example, in colloquial Spanish, "to finish eating [something]" is
comerse [algo].
Se comió la cena ("REFL he.ate the dinner") 'He finished eating the whole dinner'. You
have to use a direct object with these, at least a direct object pronoun (also the indirect reflexive pronoun). If you don't, then
comerse alone can only mean 'to eat oneself' (as if the subject is eating its own flesh, not something native speakers get to say much in real life though).
I think it'd also be helpful to be clear about (morpho-)syntactic transitivity (= "valency") vs. semantic transitivity. In "Corona kills", "kill" is syntactically intransitive (= "univalent") as it has no explicit object, but is semantically transitive because an action on an unmentioned direct object is implied.
In examples of some "more exotic" languages there is an subject-verb clause that is however translated 'He ate it" or something. The point is that there is a definite anaphoric object "it" supposed. So, there must be a stricter distinction between transitive verbs (for which that "it" is supposed when there is no explicit object) and intransitive verbs (for which no object is supposed).
Is the distinction a lexical one, or does it derive "from the context"?
You find both things in natlangs. Some of those languages might generally do it lexically (e.g. Ojibwe, which marks transitivity and partly valency morphologically), and some might generally be ambiguous and rely on context a lot but not all of the time (maybe the Papua Indonesian mentioned by Creyeditor). When context is relied on, you might be able to use pronouns, if you really need to clarify meaning. There are languages that extensively use both strategies, in which some verbs do the former and others do the latter (Mandarin Chinese, although with a host of further complications). See also Salmoneus' introspection about the possible meanings of "you told!" in English, where he feels it can be used in only some meanings / situations, other situations needing "you told it!" (or "you said it!").
What would it imply for creative language use? Should there be an antipassive to code lack of such "it"?
You could have an antipassive, yes (my understanding is this is what Mayan languages basically do, but I admit it's never been all that clear to me).
There are other ways though. You could lexically treat some verbs as intransitive in the basic form, and others as transitive in the basic form (whether syntactically or semantically). Then you add valency-changing affixes to change their transitivity, say,
-a for intrans. > trans. and
-ot for trans. > intrans.
And the valency-changing "affixes" could be other things too, like verbal agreement with the object, or incorporated generic object roots (for trans. > intrans., like transitive eat > intransitive eat-rice, used for any kind of food). It could involve syntactic constructions with an auxiliary verb. At least in a subset of verbs, it could involve suppletion with entirely different roots (Latin
ardēre 'to burn', as in "the building is burning", and its causative friend
ūrere 'to burn [something]').
Again, you could also just rely on context for the most part (= Salmoneus' "relatively free zero-derivation"), adding a direct object pronoun like "it" only if there is a perceived need to be carefully unambiguous.
EDIT: I finished this reply and left it without posting it for a few hours, and now I just saw Sal's reply. I edited the above a little bit, but I have a couple replies to Sal.
Salmoneus wrote: ↑21 Mar 2020 15:15Simple to you, but still bizarre and atypical compared to all other human languages!
Is it? He's just saying his supposed natlang basically marks valency morphologically, while his conlang purely marks (semantic) transitivity morphology. Sure, I imagine that if he told us what his natlang is, we'd probably find in a published grammar examples of further complications, but anyway, it doesn't sound too exotic. What'd be exotic would be using word order to mark valency... ("love 3S.NOM", "3S.NOM love 3S.GEN wife.ACC").
In some languages it's useful to talk about valency as being how many arguments a verb CAN has, in which case you really need another word for how many arguments it DOES have, but I don't know that word, at least off the top of my head.
AFAIK there aren't words for the distinction, and it has been a source of annoyance for me for years. In my conlangs, I deal with this in dictionaries by using punctuation when listing the syntax each verb uses:
lesht vt.
# to join [sth] {lesht [j]}
# to become friends [with sb]; make friends {lesht (j)}
Here, [j] (=
jaha 'something') means a mandatory argument, and (j) means an optional one. And in the translations, you can see that it's suggested to translate the second meaning as "become friends with sb" when the optional argument is present, and "make friends" when it's absent.
I probably haven't helped answer your question, but maybe I've added some background confusion....
I think you did, I mean, it basically comes down to "natlangs do all sorts of things, including multiple things at the same time" and "BTW it'd be useful for you to distinguish valency from semantic transitivity in your terminology".
EDIT 2: Okay, just saw Omzinesý's new reply.
Omzinesý wrote: ↑21 Mar 2020 17:23Languages have the thing called transitive construction. Haspelmath's definition of it as a comparative concept is: 'To kill' is the most semantically transitive verb. Transitive construction is the one where 'to kill' has an agent and a patient.
All verbs that appear in transitive construction can be called syntactically transitive. So we avoid speaking about semantic transitivity, which is a continuum, like Hopper and Thompson argue. Transitive construction is so common in English that bivalent and syntactically transitive are almost synonyms.
Well, first of all, clearly me and Salmoneus assumed you knew less than you actually do. I at least was led astray by your sentence "Similarly, a transitive verb can be used 'intransitively'", which seemed to confuse semantics and syntax, but maybe you were trying to simplify the terminology for us.
I'm not sure what your objection here is. The way I see your post, you seem to want to replace the terms "syntactic transitivity" (= valency) and "semantic transitivity" with "(in)transitive construction" and "transitivity" (no modifiers) respectively, without changing the definitions. I don't understand how semantic transitivity being a continuum is an argument against the need to have a term for the concept (plus you do keep the concept around as the unmodified "transitivity", although apparently to suggest that in some or many languages like English the two are close). I wouldn't say the use of semantically transitive verbs as monovalents ("Corona kills") involves trivial cases in English; they should be addressed.
I wonder if the quibble in Hopper and Thompson's paper (1980) is actually about some linguists and many lexicographers' tendency to treat syntactic-semantic transitivity as clear, white-and-black categories, not the usefulness of semantic transitivity as a concept. I should go read that...
However, you forced me to think semantics. It really seems that omitting the agent really changes the semantics of the verb while omitting the patient only makes the identity of the object fuzzy.
Not true of pro-drop languages, especially East Asian and Southeast Asian languages like Mandarin, Japanese and Burmese where it's very normal to drop subjects while lacking agreement conjugations for the subject, retaining the subject as implied. English shows a bit of this with its pro-drop syntax in the spoken registers ("love you" = 'I love you', or less commonly 'we love you'), but it is terribly common in those Asian languages, for all persons.