Zompist Bboard Again

There's probably at least one hierarchy for literally any linguistic property you can think of.

There are transitivity hierarchies (at what point do languages treat an event as requiring transitive verbs, particularly in languages in which transitivity is orthogonal to valency?). There are subjecthood hierarchies (when you have several arguments that are subjectlike, which tends by default to be the subject?) and topichood hierarchies and focushood hiearchies. There are voice hierarchies (which voice is default?). There are various hierarchies around objecthood, particularly with ditransitive (and higher) verbs. I think really any time you see the word "alignment", you can rephrase it with "hierarchy" if you want. [ergative-absolutive alignment means that patients "outrank" (i.e. are higher on the hierarchy) than agents when there is only one argument slot to fill, and/or that patients outrank agents in filling the pivot slot]. There are also relativisation hierarchies - if, for example, you read about the rise of wh-pronouns as relative pronouns in Germanic languages, you'll see they "spread" from one end of a hierarchy of relative clause types toward the other. And so on.

Thinking about it logically...

You've got two clauses: X put Y on Z, and Q broke. You want some way to indicate that Q is coreferential with either Y or Z. Let's call the argument that Q agrees with the "target". Assuming it's not purely contextual, this means some kind of target-marking. There are three options: mark this somehow in the first clause; mark it in the second clause; or mark it in the relationship of the two clauses. [or, four: mark it in more than one way.]

Marking in the first clause...

Can you mark it on X? Probably not, that doesn't seem to make much sense. It's weird to modify the semantic subject to indicate the relationship that an object has with another clause.

Can you mark it on "put"? Conceivably. You could have a property, targethood, and have verbs mark whether their object is a target or not. This is unlikely to be a thing in its own right, but it could conceivably be involved in other properties. In particular, in languages that distinguish dyadic intransitives from transitives, you could very plausible have a rule that true objects (of transitives) are in general better candidates for targethood. If you did that, making "put" intransitive could be used to signal that Y is not likely to be a target, and that by implication Z is.

[Pseudoenglish example: "I put the glass on the table and [the glass] broke" vs "I unburdened myself of the glass on the table and [the table] broke"]

Similarly, rather than just removing objecthood from Y, you could actively assign objecthood to Z, with an applicative. Or, you could be more radical, make subjects outrank objects in the targethood hierarchy, and require passive or applicative-passive voice to promote either Z or Q to subjecthood in order to make them the target.

["I put the glass on the table and [the glass] broke" vs "I burdened the table with the glass and [the table] broke"; or "the glass was placed on the table by me and [the glass] broke" vs "the table had the glass placed on it by me and [the table] broke"]

If your verb inflects to agree with the topic, the focus, or the 'trigger', these could all be used in a similar way.

Can you mark it directly on Y or Z? Sure. Again, if subjects or objects head the target hierarchy, you can require the appropriate case marking on the nouns. You could also make some other property be high in the hierarchy. Again, I'm guessing a pure targethood property is unlikely (though I wouldn't rule it out), but you could, for example, use definiteness in this way (definites are more likely to be targets than indefinites). The appropriate articles (or other definiteness marking) could thus be used to disambiguate.

["I put the glass on a table and [the glass] broke" vs. "I put a glass on the table and [the table] broke". This isn't compelling in English, because in English we also use definiteness to mark focus, and I think we have a tendency to interpret the focus as the natural target in many cases. But either of those facts could be different in another language.]

And, of course, any overt syntactic or morphological marking of topic or focus (or 'trigger') could be used here to promote one argument up the hierarchy. Either directly, or by making whichever is unmarked the target (that is, marking the topic could imply the other is the focus and hence the target, or vice versa).

["Regarding the glass, I put it on a table and [the glass] broke" vs "As for the table, I put a glass on it and [the table] broke". Or "I put this glass on the table and [the glass] broke" vs "I put the glass on this table and [the table] broke"]

Finally, can you mark anything on "on"? I don't think I've seen anything like that, but it could be possible to in some way mark or alter the heads of oblique phrases to promote or demote their contents in targethood, again probably related to some other property.

------

In the second clause....

Can you mark it on "and"? I'd have thought so, yes. Whatever the "natural" target in the first clause, you could alter your conjunction (etc) to indicate a switch in the second clause. ["I put the glass on the table and [the glass] broke" vs "I put the glass on the table BUT [the table] broke". Or more radically, you could contrast a conjoined clause with a subordinated one. "I put the glass on the table and [the glass] broke" vs "I put the glass on the table, breaking [the table]".

In addition to all the directly markable linguistic things in the first clause that a "switch" could reverse, you could also assign the default target through semantic context - near vs far, animacy hierarchy, salience in conversation, etc. Or through, as you say, just assuming that the first eligible argument is the natural target.

Can you mark it on "Q"? Absolutely! There are three basic ways to do this...

a) you can have a natural target (ascribed in the first clause either syntactically or semantically) and mark a switch in reference on the Q pronoun. If you don't have explicit switch reference* marking, you can do it indirectly - for instance, you could have a rule that you can only use a bare personal pronoun when the target is natural, and mark a switch in reference by using either a full noun or some other sort of more marked pronoun (eg a demonstrative).

b) you can have the Q pronoun in some way directly specify a property of its target, to reduce ambiguity. This could be a lexical property (noun class), a semantic property (deictic saliency, number, etc), or a syntactic property (subjecthood, objecthood, focushood, topichood, etc), and can be marked morphologically, with particles, through what sort of pronoun is allowed, or through syntax.

or c) you can treat Q differently (morpho)syntactically to indicate an agreement of syntactic properties (that is, you could by default have subject pronouns have subject targets, object pronouns have object targets, etc).

And can you mark it on "break"? Again, certainly. Either you can use the verb to mark some property of "Q", or you can as Estav says have "break" directly classify in some way its own arguments (and hence the target of Q), either lexically or morphosyntactically.

----

Finally, you can mark it in the relationship of the two clauses. The easiest way would just be to use a relative clause. Distinguishing between a conjoined clause and a relative clause goes back to what I said about marking on "and". But you could just always use a relative and attach it to different parts of the matrix clause to mark the target. Likewise with other forms of subordinated clause.


-----------


And of course, for all these strategies, they could be either compulsory, or used only for disambiguation.







*NB "canonical" switch reference marking indicates specifically whether the subject of the second clause has the subject of the first clause as its target. Fewer languages have "non-canonical" switch reference, like indicating whether the objects are the same. But we're basically talking here about some broader concept of semantic switch reference, even if it's not explicitly and specifically marked morphosyntactically as such in many or any languages.

Linguoboy wrote: ↑Thu Feb 07, 2019 10:59 pm

Pabappa wrote: ↑Thu Feb 07, 2019 6:01 pm 1) It's interesting that it's "be" instead of "have".

None of these languages have a word for “have”. They all express possession periphrasticly with “be”.

Isn't the past tense in question also derived from the past active participle, so the semantic role of the subject would never have been unclear.

For a parallel example, the Finnish perfect is likewise formed using the perfect active participle and the auxiliary "be",

Ole-n syö-nyt.
be-SG1 eat-PRF.ACT.PTCP
"I have eaten." (lit. "I am eaten.")

Because the form uses an active participle, there's no question that the grammatical subject is indeed an agent and not a patient. In fact, the impersonal passive of the perfect is formed using the perfect passive participle,

Minu-t on syö-ty.
SG1-ACC be.SG3 eat-PRF.PASS.PTCP
"I have been eaten."

As a side note, the sole nominal participant is a full blown object now and gets the appropriate object marking.

gach wrote: ↑Fri Feb 08, 2019 8:09 am

Linguoboy wrote: ↑Thu Feb 07, 2019 10:59 pm

Pabappa wrote: ↑Thu Feb 07, 2019 6:01 pm1) It's interesting that it's "be" instead of "have".

None of these languages have a word for “have”. They all express possession periphrasticly with “be”.

Isn't the past tense in question also derived from the past active participle, so the semantic role of the subject would never have been unclear.

Yup.

(Ja) zjadłem(m.)/zjadłam(f.). "I ate."
(Ja) byłem zjedzony(m.)/byłam zjedzona(f.). "I was eaten."

malloc wrote: ↑Wed Feb 06, 2019 7:23 pm One thing I have always wondered: how does rhyming work in languages with significant inflection and agreement? Consider a language like Latin where nouns and adjectives agree in number, gender, and case. It seems like rhymes would frequently turn into repeating the same inflectional form in successive lines, like -ōrum or -āvērunt or something.

This kind of rhyming is called „rymy częstochowskie” in Polish and is considered bad style.

The key thing about moods is that they are grammaticalized, even if their meanings themselves are vague. And I am not saying I don't recognize some of the distinctions referred to in the Wiki article, but I do not think they count as separate moods per se beyond cohortative versus non-cohortative and affirmative versus negative hortative, precisely because these distinctions are encoded grammatically.

The English phrase “you might want to do X” used in the adhortative sense of “I advise you to do X” or “it would be good if you did X” is grammaticalized. It isn't the literal sense of the sum of its elements (“you possibly have a desire”) and doesn't translate easily into other languages (translated word-for-word, Polish moglibyście chcieć zrobić X wouldn't have the same meaning).

(A more direct Polish equivalent would be more like lepiej to zróbcie, with an imperative form (zróbcie) and a clarifying adverb (lepiej) acting as a grammatical word.)

Pabappa wrote: ↑Thu Feb 07, 2019 3:30 pm Does anybody know how Polish evolved a gender contrast in the past tense of its verbs? (And in fact, almost everything except the present tense?) It looks like they stuck person markers on top of a word that had a fixed gender contrast, but do we know what that was? Russian may preserve the older state, in that its past tense verbs distinguish gender but have no person marking at all .... kind of like an adjective or a participle. But Polish doesn't seem to have any word forms that look like they might have been the source of the new verbs, and Russian's don't seem to have any other use besides verbs. Im just curious, this is to fulfill my own desire for knowledge. Can anyone help?

(Sorry for repeating the points after the previous posters, I wrote those examples before seeing there are more messages.)

“to be” + past active participles.

Synchronically, robiłam (“I was doing (fem.)”) ← robiła-m ← robiła-*jeśm [do-PAST.ACTIVE.PARTICIPLE-FEMININE be-PRESENT-1SG]

Compare French, where the past composed forms were formed from “to have” + past passive participles, and (in certain cases) the gender contrast is retained, but with respect to the patient and not the agent.

Cf. jestem znana (“I am known (fem.)”) from znana (passive participle) vs znałam (“I knew (someone)”) from znała (active participle).

If I'm not wrong, Slavic originally had a four-way contrast in participles, with respect to voice as well as tense. Some of the forms are conserved in modern Polish:
· znająca “knowing” (present active participle)
· znajoma “familiar” (an adjective), historically “being known” (present passive participle)
· znała “she knew (somebody)” (past tense verb), historically “having known” (past active participle)
· znana “known” (generic passive participle), historically “having been known” (past passive participle)

None of these languages have a word for “have”. They all express possession periphrasticly with “be”.

Russian, right, but Polish has a word for “have” (mieć) and (afaik) it's been inherited from Slavic.

Okay, many thanks to all of you. I had not actually known that active participles existed as a distinct word form in any language.... i had just taken the word "participle" to mean a passive participle. This makes much more sense now.

And I see there is a new active participle in modern Polish to replace the old one, whereas the passive participle didn't develop a 2nd use, so it remains as it was.

Pabappa wrote: ↑Fri Feb 08, 2019 3:45 pm Okay, many thanks to all of you. I had not actually known that active participles existed as a distinct word form in any language.... i had just taken the word "participle" to mean a passive participle. This makes much more sense now.

And I see there is a new active participle in modern Polish to replace the old one, whereas the passive participle didn't develop a 2nd use, so it remains as it was.

Ok, now that I'm reading up on that, Slavic had originally a five-way distinction, with two past active participles:
· *znaję (present active) → Pol. znający (present active participle)
· *znajemŭ (present passive) → Pol. znajomy “familiar” (a derived adjective)
· *znavŭ (past active) → Pol. forms like poznawszy “having got to know” (adverbial participle)
· *znalŭ (second past active / “L-participle”) → Pol. znał (3rd person past tense)
· *znanŭ (past passive) → Pol. znany (passive participle)

Pabappa wrote: ↑Fri Feb 08, 2019 3:45 pm Okay, many thanks to all of you. I had not actually known that active participles existed as a distinct word form in any language.... i had just taken the word "participle" to mean a passive participle. This makes much more sense now.

I'll be honest English is no help here, due to the syncretism of the two forms, but actually for languages with voice contrasts I'm pretty sure English is unusual for not distinguishing voice on its participles.

Frislander wrote: ↑Sat Feb 09, 2019 8:05 ambut actually for languages with voice contrasts I'm pretty sure English is unusual for not distinguishing voice on its participles.

??????

English forms passives with a participle but the participle doesn't express voice in any of its other uses, as opposed to say Latin participles.

mèþru wrote: ↑Sat Feb 09, 2019 12:07 pm English forms passives with a participle but the participle doesn't express voice in any of its other uses

Clearly it does. "The scalded cat" and "the scalding water" do not indicate simply a difference of tense.

It just happens that the passive participle is used in English is used to form the perfect, and the active participle in English is used to form the progressive, and the progressive is most commonly used to express the non-habitual present.

Travis B. wrote: ↑Sat Feb 09, 2019 2:34 pm It just happens that the passive participle is used in English is used to form the perfect, and the active participle in English is used to form the progressive, and the progressive is most commonly used to express the non-habitual present.

Yeah that's what I was getting at, and I think particular distribution this is a particularly English phenomenon, though I'd phrase it the other way - the past/perfect participle is also used to form the passive.

But the participle itself is inherently passive, as in my example, while the other is inherently active.

Dutch has a similar arrangement, where the perfect is formed with the passive participle and the progressive is formed with the active participle.

methru, Frislander: I still don't know what you mean. As Salmoneus says, voice *is* expressed in both English participles. And this is done in a very similar way to Latin/French/Arabic participles, it's not a particularly English phenomenon.

Though the -ed participle is passive when used as an adjective, when it's used verbally it may be either passive ("I was loved") or active ("I have loved").

These desiccated little participle systems are nothing compared to Ancient Greek, whose participles mark both tense (present, future, and aorist) and voice (active, middle, and passive).

Well, I feel stupid now.
I don't know why I wrote that; I looked up this stuff plenty of times before and actually am helping someone who works with this stuff in comp ling.

Basic English language question: in English, what do you call the thing that you're doing with an answer when you're answering something? Do you "give" an answer, do you "make" an answer, do you "provide" an answer, or something else entirely?

give or provide both work, but provide is a bit more formal (maybe even snooty, but I still use it)