The relative clause can't have a subject... wait, WHAT?

If you're new to these arts, this is the place to ask "stupid" questions and get directions!
Post Reply
User avatar
LittleLynx_53
rupestrian
rupestrian
Posts: 17
Joined: 13 Mar 2022 17:35

The relative clause can't have a subject... wait, WHAT?

Post by LittleLynx_53 »

Hello, I am a bit new to this still. I ended up working on relative clauses already. I have something that seems to function fairly well, but... I'm not entirely clear on what roles some of the words are taking. And anything you don't understand about your own language is a disaster waiting to happen, as I've already learned.

Some of how my language works:
Strict VSO word order with the exception of direct and indirect object: the indirect object is marked, allowing the two to be swapped around for emphasis.
Verbs ignore valency/transitivity most of the time, simply getting interpreted differently depending on the number of objects in the sentence. One exception is the passive prefix, which is important here.
Words like "into" are tacked onto the end of the object and have a tense marking that agrees with the verb.
Note: I am currently only translating directly into a gloss. (I would like to get a solid handle on the grammar before adding the extra step of remembering the nuances of vocabulary) Wish me luck: it's my first time using the leipzig rules.

So right now, "I threw the onion into the fire" is
/throw-PST I fire-DAT into-PST onion/

To say just "I threw" is
/throw-PST I/

To say something like "the onion was thrown," you use the passive prefix I mentioned, which lets you get rid of the subject.
/PASS-throw-PST onion/
Now, there's already a question of weather getting rid of the subject is optional or not, but considering what happens next, I feel like it may be obligatory.

I wanted to translate the relative clause "The fire I threw and onion into." As in, "will you help me put out the fire I threw an onion into." Or, more accurately "the fire an onion was thrown into." (I'm leaving out the "by me" for now) It comes out as:
/PSV-throw-PST fire-DAT into-PST onion/

But if you wanted to say "The onion that was thrown into the fire" the only simple option seems to be (and I'm heavily against adding any more case marking) switching the indirect/direct object order to change the emphasis.
/PSV-throw-PST onion fire-DAT into-PST/

I kind of like that solution. It does mean that the order of the direct/indirect object is a lot more important in a relative clause. It changes the meaning slightly more than just the emphasis. You can say "Don't worry: I threw the *onion* into the fire," or "Don't worry: I threw the onion into the *fire.*" but with relative clauses "Will you help me put out the fire I threw an onion into" makes sense, but "will you help me put out the onion I threw into the fire" doesn't.

BUT... there doesn't seem to be a place for "I" in either of those relative clauses. You would put "by me" or the lang's equivalent after it. So... no subject? Truly no subject, and no place for it? "Fire" can't be the subject, right? It's already an object and not in any way the thing that's *acting.* Is "By me" the subject if you choose to add it, and if so did the word order just get accidentally switched to VOS for relative clauses?


Sorry for the long post--my brain had enough trouble with this that I felt I needed to ask the question in full. And thank you for any answers in advance.
Never fight a land war in (that fictional group of countries heavily based on Asia.)
User avatar
Omzinesý
runic
runic
Posts: 3455
Joined: 27 Aug 2010 08:17
Location: nowhere [naʊhɪɚ]

Re: The relative clause can't have a subject... wait, WHAT?

Post by Omzinesý »

First
It seems you understand your lang very well. You just lack some terms describing it.

Basically, your clauses could be "impersonal passives" or just "impersonal", meaning that they lack the subject. Some languages have such voices.

Probably you could say that 'onion' is the subject, but if the construction is like dative shift of objects, it is not very elegant to call it subject in relative clauses.

How would you say "the donkey that threw the onion into the fire", where the head of the relative clause is its subject?
My meta-thread: viewtopic.php?f=6&t=5760
User avatar
LittleLynx_53
rupestrian
rupestrian
Posts: 17
Joined: 13 Mar 2022 17:35

Re: The relative clause can't have a subject... wait, WHAT?

Post by LittleLynx_53 »

Omzinesý wrote: 12 May 2022 21:13 How would you say "the donkey that threw the onion into the fire", where the head of the relative clause is its subject?
That's a good question...

If I just obeyed the default word order and did nothing else, it would be:
/PSV-throw-PST donkey onion fire-DAT into-PST/
Where the passive prefix is carrying all the weight of keeping the meaning different from "the donkey did x".
It also begs the question of why the pronoun "I" couldn't go in that same place in the previous example sentences.

I could make that only apply to pronouns. I'm not sure I want to do that. Or I could change the word order for that specific situation...
/donkey PSV-throw-PST onion fire-DAT into-PST/

Or should something else get marked with the passive to make things clearer?
/PSV-throw-PST PSV-donkey onion fire-DAT into-PST/
The word order solution seems maybe more elegant... Or I could combine them. Opinions welcome.
Never fight a land war in (that fictional group of countries heavily based on Asia.)
Salmoneus
MVP
MVP
Posts: 2626
Joined: 19 Sep 2011 19:37

Re: The relative clause can't have a subject... wait, WHAT?

Post by Salmoneus »

BUT... there doesn't seem to be a place for "I" in either of those relative clauses. You would put "by me" or the lang's equivalent after it. So... no subject? Truly no subject, and no place for it? "Fire" can't be the subject, right? It's already an object and not in any way the thing that's *acting.* Is "By me" the subject if you choose to add it, and if so did the word order just get accidentally switched to VOS for relative clauses?
No, the subject of "the onion was thrown into the fire" is "the onion". That's the whole point of passives: they promote the object of the active verb to be the subject, and demote the subject to an oblique [because they also reduce the valency, so there's no slot for an object anymore*]

Active verb: "I ate an onion". SUBJ - VERB - OBJ
Passive verb: "An onion was eaten by me". SUBJ - VERB - [no OBJ because we've reduced the valency] - OBL

Don't get confused between semantic roles - what is 'acting' or 'being acted upon' - and actual syntactic roles like subject and object.
Strict VSO word order with the exception of direct and indirect object: the indirect object is marked, allowing the two to be swapped around for emphasis.
Verbs ignore valency/transitivity most of the time, simply getting interpreted differently depending on the number of objects in the sentence. One exception is the passive prefix, which is important here.
Words like "into" are tacked onto the end of the object and have a tense marking that agrees with the verb.
To be clear, if something is governed by a preposition, it's not syntactically an indirect object anymore, but an oblique.

In English, we often talk about indirect objects in the semantic sense, but in reality English is able to encode similar semantic roles through either an oblique or an indirect object:
I gave the letter to you - 'to you' is an oblique argument
I gave you the letter - 'you' is a true indirect object

In English, only a small number of verbs can take indirect arguments (ditransitives), and only very specific roles can merit an indirect object position, whereas obliques can be used with any verb and almost any semantics - obliques are, in a way, 'outside' the core syntax of the verb.

This distinction doesn't really matter in most circumstances, but since you're struggling a bit getting some of these things clear for yourself it might be helpful to bear the pedantic differences in mind.

Similarly, I hope you won't take offence if I seem to be being patronising here, or saying things you already know. When we're confused, it's good to be clear from the ground up - particularly if one person doesn't know what the other person knows.

I wanted to translate the relative clause "The fire I threw and onion into." As in, "will you help me put out the fire I threw an onion into." Or, more accurately "the fire an onion was thrown into." (I'm leaving out the "by me" for now) It comes out as:
/PSV-throw-PST fire-DAT into-PST onion/
So, first things first: are you aware you're doing something very strange here? It's OK if you are, but if you're doing it accidentally it may be confusing you.

So, relativisation combines two clauses: the main clause, and the relative clause. Headed relatives (we'll ignore headless ones for now) link the two clauses by having a noun in common: there's one noun that has one role in the main clause, and another (potentially the same, potentially different) role in the relative clause. This noun is called the 'head' of the relative clause, and is the thing the relative clause is describing.

The vast majority of languages have "externally-headed" relative clauses: that is, the 'head' of the relative clause occurs in the main clause, not in the relative clause. [so the full version of the noun is found 'externally' to the relative clause]. Instead, the head is represented inside the relative clause by perhaps a pronoun of some sort, or just by a 'gap'.

So in English:
I saw the man [that ate you] - the main clause is "I saw the man", and the relative clause, here introduced by a relativising element 'that', is "[the man] ate you". In the process of embedding this clause as a relative into the main clause, the subject has been 'gapped', dropped, because it's the same as the noun in the main clause.

[European languages do a really weird thing where the relative clause sometimes contains a pronoun that's been moved to the front of the clause, and that's also often a pronoun that doesn't occur in most main clauses. So "I saw the carpenter whom you ate" has turned "you ate the carpenter" into "you ate whom" (replaces the head noun with a pronoun) and then "whom you ate" (fronts the relative pronoun). Using pronouns is common worldwide, but using a specifically relative pronoun that is then fronted is a weird thing that only really happens in Europe]

[and yes, as these english examples indicate, it's not unusual to have multiple ways to form relative clauses in a single language]

Now, what you're doing is instead having an "internally-headed" relative clause. That is, you actually have the full noun inside the relative clause itself. [you don't show the full sentence, so I don't know whether the main clauses contains a pronoun or a second instance of the full noun, or neither - any of these is possible].

Internally-headed relatives are rare: WALS surveys 824 languages, of which only 31 primarily have internally-headed clauses [a further 29 have internally-headed relatives as one of several competing options, and a further 10 have them as a minor construction with some other option being dominant]. In all cases, internally-headed relatives seem to be an areal feature - almost all the pure examples are from the US, with clusters of mixed languages in northern south america, australia and PNG, and the himalaya.

And there's nothing wrong with using them. But you should be aware that it's quite an unusual idea, and this might make working with them less intuitive for you.



-----

The second issue here is why you're using a passive at all. Why are you using a passive in these relative clauses?

It's not unusual to have it be more difficult to relativise direct objects than subjects, and obliques than objects. Some languages don't allow relativisation of obliques, or even objects, at all. If this were the case, it would explain why "the onion I threw in the fire" requires a passive: it promotes 'onion' to object, which may allow relativisation.

However the passive does not promote the oblique, "into the fire", to subject position, or even to object position. So it shouldn't make it any easier to relativise: why do you have to change the status of "onion" in order to relativise "fire"?

Are you just being confused by the fact that passives reduce valency and gap-strategy relatives also appear superficially to reduce valency? [they don't, but because there's a gap it looks like they do when you take them out of context]




*It's also possible to have a "symmetrical passive", as found in some Indonesian languages, which does not reduce valency, so just swaps subject and object. Whether this is really a passive voice is debated, and some language have both a symmetrical 'passive' and a genuine passive. It's a rare phenomenon, in any case.
User avatar
LittleLynx_53
rupestrian
rupestrian
Posts: 17
Joined: 13 Mar 2022 17:35

Re: The relative clause can't have a subject... wait, WHAT?

Post by LittleLynx_53 »

Salmoneus wrote: 12 May 2022 23:48 So, first things first: are you aware you're doing something very strange here? It's OK if you are, but if you're doing it accidentally it may be confusing you.

So, relativisation combines two clauses: the main clause, and the relative clause. Headed relatives (we'll ignore headless ones for now) link the two clauses by having a noun in common: there's one noun that has one role in the main clause, and another (potentially the same, potentially different) role in the relative clause. This noun is called the 'head' of the relative clause, and is the thing the relative clause is describing.

The vast majority of languages have "externally-headed" relative clauses: that is, the 'head' of the relative clause occurs in the main clause, not in the relative clause. [so the full version of the noun is found 'externally' to the relative clause]. Instead, the head is represented inside the relative clause by perhaps a pronoun of some sort, or just by a 'gap'.
Alright, I think I see what happened. That was accidental and thank you for pointing that out. I wanted to translate "the onion I threw into the fire" because it was in a conlang book I was reading in the section about valency-changing operations. (the context being that someone new walks into the room and asks "what fire are you talking about?") I showed the sentence to someone-- they go "that's not a complete sentence, that's a relative clause" and I go OHHHH and take the ENTIRE thing to be a relative clause. But pedantic differences being important here, it's a fragment of a sentence *plus* the relative clause, isn't it.
Salmoneus wrote: 12 May 2022 23:48 The second issue here is why you're using a passive at all. Why are you using a passive in these relative clauses?
Salmoneus wrote: 12 May 2022 23:48 Are you just being confused by the fact that passives reduce valency and gap-strategy relatives also appear superficially to reduce valency? [they don't, but because there's a gap it looks like they do when you take them out of context]
Um, yes. Yes I definitely am getting confused by that. I do not fully understand the syntax-word order-relativity relationship, and I think I was trying to tack on the passive to make absolutely sure there was a distinction between "I threw an onion into the fire" and "the fire I threw an onion into" without changing the word order. Do you know any good resources for getting a handle on this? Like, what mechanisms does a language need in order to have a proper passive voice? Thank you for all your help.
Salmoneus wrote: 12 May 2022 23:48 No, the subject of "the onion was thrown into the fire" is "the onion". That's the whole point of passives: they promote the object of the active verb to be the subject, and demote the subject to an oblique [because they also reduce the valency, so there's no slot for an object anymore*]

Active verb: "I ate an onion". SUBJ - VERB - OBJ
Passive verb: "An onion was eaten by me". SUBJ - VERB - [no OBJ because we've reduced the valency] - OBL

Don't get confused between semantic roles - what is 'acting' or 'being acted upon' - and actual syntactic roles like subject and object.
Does this mean that if the speakers of the language connect semantic and syntactic roles [the definition of a subject as the one that is acting, period] like I was unconsciously doing, and some kind of relativisation is going on, that the word order will be *perceived* by them to have changed in the relative clause?
Never fight a land war in (that fictional group of countries heavily based on Asia.)
Salmoneus
MVP
MVP
Posts: 2626
Joined: 19 Sep 2011 19:37

Re: The relative clause can't have a subject... wait, WHAT?

Post by Salmoneus »

LittleLynx_53 wrote: 17 May 2022 17:48
Salmoneus wrote: 12 May 2022 23:48 So, first things first: are you aware you're doing something very strange here? It's OK if you are, but if you're doing it accidentally it may be confusing you.

So, relativisation combines two clauses: the main clause, and the relative clause. Headed relatives (we'll ignore headless ones for now) link the two clauses by having a noun in common: there's one noun that has one role in the main clause, and another (potentially the same, potentially different) role in the relative clause. This noun is called the 'head' of the relative clause, and is the thing the relative clause is describing.

The vast majority of languages have "externally-headed" relative clauses: that is, the 'head' of the relative clause occurs in the main clause, not in the relative clause. [so the full version of the noun is found 'externally' to the relative clause]. Instead, the head is represented inside the relative clause by perhaps a pronoun of some sort, or just by a 'gap'.
Alright, I think I see what happened. That was accidental and thank you for pointing that out. I wanted to translate "the onion I threw into the fire" because it was in a conlang book I was reading in the section about valency-changing operations. (the context being that someone new walks into the room and asks "what fire are you talking about?") I showed the sentence to someone-- they go "that's not a complete sentence, that's a relative clause" and I go OHHHH and take the ENTIRE thing to be a relative clause. But pedantic differences being important here, it's a fragment of a sentence *plus* the relative clause, isn't it.
Yes, although pedantically the whole thing is a sentence fragment, containing a noun and a relative clause modifying that noun.

Although the original text, "the onion I threw into the fire" could also arguably be a genuine sentence, exhibiting topicalisation: "[The lemon I gave to the dogs, but] the onion I threw into the fire". But I don't know what it was in the context of the conlang book.


[[completely off-topic, but "what fire are you talking about!?" here also illustrates a complexity with English articles, because "the fire" should really usually be "a fire", going by the usual (incomplete) description of "the" as a definite article). But that's COMPLETELY off-topic here...]]
Salmoneus wrote: 12 May 2022 23:48 The second issue here is why you're using a passive at all. Why are you using a passive in these relative clauses?
Salmoneus wrote: 12 May 2022 23:48 Are you just being confused by the fact that passives reduce valency and gap-strategy relatives also appear superficially to reduce valency? [they don't, but because there's a gap it looks like they do when you take them out of context]
Um, yes. Yes I definitely am getting confused by that. I do not fully understand the syntax-word order-relativity relationship, and I think I was trying to tack on the passive to make absolutely sure there was a distinction between "I threw an onion into the fire" and "the fire I threw an onion into" without changing the word order.

I remember being confused by that at first myself, to be honest. Unfortunately, there's a lot of ways in which English syntax is less than immediately clear in its structure (due to the lack of overt marking of many things), which can mislead people as to how non-English languages are likely to work...

Very short intro to these two totally different things that happen to often look the same in English...

Voice

A standard bivalent verb has an agent (who does the thing) and a patient (who has it done to them). [I say 'standard' because it's not actually true of all English bivalent verbs - when "I see him", I don't actually "do seeing" to him, if anything he something to me - but it's a good enough generalisation for our purposes today]

The agent is typically the subject, and the patient is typically the object. What is a subject? There's no one answer - someone might link you to a checklist that somebody proposed - but in general:
- it's a syntactic role assigned to an argument of a verb
- it often has a distinctive case or position in the sentence
- the verb typically agrees with it, if it agrees with anything
- most verbs will have one (though in some language subjectless 'impersonal' verbs are possible)
- in two attached clauses, the subject of the first clause is often assumed to be the subject of the second clause also, allowing it to be dropped. Thus, in "the parson punched the vicar and ran away", we know it is the parson who ran away, because the second verb can only be subjectless when the subject is the same as the subject of the first clause.

And so on.

[an object is then the prototypical second argument of a bivalent verb - the one that isn't the subject]

So, in "I threw the onion", the subject AND agent is 'I', and the object AND patient is 'the onion'. And the verb is bivalent.

The passive voice decreased the valency of the verb, so now it can only have one argument. And it promotes the former object to be the subject.

So, in "The onion was thrown", the patient is still 'the onion', but now this patient is the subject, not the object. The semantic agent, the thrower, is dropped entirely.

It is often possible to add in additional nouns that are not core arguments as 'obliques' - arguments governed by prepositions - and this is true of passives in English: "the onion was thrown by me". The onion is the patient and I am the agent. But the onion is the subject and I am only an oblique argument, introduced by the preposition 'me'. You can double-check that I'm no longer the subject by the fact that 'I' am now only 'me' (the pronoun is not nominative as it would be if it were a subject).

For clarity, we could also explicitly label the valency: "I threw-2 the onion", but "the onion was thrown-1". [I'm treating 'was thrown' as the verb, though pedantically it's probably just 'was', because English has a synthetic passive. This should be relevant here.]


Relatives

In English, relative clauses often don't look like normal clauses. We can imagine that a series of transformations has taken place, and describe this as a series of stages:

Stage 1: I saw the dog, I painted the dog yellow yesterday. [two separate clauses]
Stage 2: I saw the dog, I painted it yellow yesterday. [the second clause has a pronoun that corefers to the head noun]
Stage 3: I saw the dog, I painted which yellow yesterday. [the pronoun is now a special pronoun that only occurs in relative clauses]
Stage 4: I saw the dog which I painted [___] yellow yesterday. [the pronoun is fronted to the beginning of the relative clause, leaving a 'gap' (which I've marked) where it would normally be.]
Stage 5: I saw the dog I painted [___] yellow yesterday. [the pronoun is dropped altogether]

The same process happens when the dog is the subject of the relative clause:

Stage 1: I saw the dog, the dog bit me
Stage 2: I saw the dog, it bit me
Stage 3: I saw the dog which bit me
Stage 4: I saw the dog which bit me [no change because 'which' is ALREADY at the front of the relative, because it's the subject]

[however, Stage 5 is prohibited when the head noun is the subject of the reflexive, outside of colloquial speech: ?"I saw the dog bit me"]

This likewise happens when the dog is only the subject due to a passive voice:

Stage 1: I saw the dog, the dog was hated
Stage 4: I saw the dog which was hated


But the relativisation does not itself change the valency of the verb, so there should be no ambiguity over what is the agent and what is the patient. In theory, this is also indicated on the relative pronoun itself: "the man who punched me" vs "the man WHOM I punched". But this is only indicated for people, and is largely ignored in colloquial speech.

[however, there are languages in which the head noun can't be the object of a relative clause - in which case you need a passive - and more in which an oblique in a relative clause can't be the head noun (in which case you need an applicative, a passive plus dative shift, or just say something else...]

One other wrinkle: because the relative pronoun in English moves to the front of the relative clause, it occupies the same place as - and actually replaces - the relativising particle 'that'. In many languages these may be clearly distinguished, or only one may be present.


-----

On your more general question:

- syntax is how the words in the sentence are connected. It covers which things are subjects, which are objects, and how you can tell which is which
- word order is one tool to indicate syntax (the other big tool is morphology - modifying words). English uses a lot of syntax because it doesn't use a lot of morphology; other languages may be the other way around.
- relatives don't have to have any major impact on word order, other than that the relative clause is itself something that has to be fitted into word order (generally just after the head noun the clause modifies). The relative clause can have the same word order as any other clause. However, in English the word order is altered by moving the relative pronoun to the front of the relative clause (this is a mostly European oddity).

Do you know any good resources for getting a handle on this?
Not really, sorry. Though you could read through the WALS entries on relativisation strategies, for an overview of what different languages do. Might be a bit deep end, though.
Like, what mechanisms does a language need in order to have a proper passive voice?
It's simple. To have a passive voice, you need to do three things:

- reduce the valency of the verb, so that a bivalent verb becomes univalent and only takes one necessary core argument
- promote what would have been the object of the bivalent 'active ' verb to become the subject of the new univalent 'passive' verb
- indicate in some way that this has happened, so that listeners don't get confused. This may be done through, for example, an affix on the verb, or an alternative verbal construction (English replaces the verb with a copula+participle construction), or simply through word order in some way
Thank you for all your help.
No problem. Wish I could be more helpful, rather than just jibbering on in ways that probably just make things seem more confusing
Salmoneus wrote: 12 May 2022 23:48 No, the subject of "the onion was thrown into the fire" is "the onion". That's the whole point of passives: they promote the object of the active verb to be the subject, and demote the subject to an oblique [because they also reduce the valency, so there's no slot for an object anymore*]

Active verb: "I ate an onion". SUBJ - VERB - OBJ
Passive verb: "An onion was eaten by me". SUBJ - VERB - [no OBJ because we've reduced the valency] - OBL

Don't get confused between semantic roles - what is 'acting' or 'being acted upon' - and actual syntactic roles like subject and object.
Does this mean that if the speakers of the language connect semantic and syntactic roles [the definition of a subject as the one that is acting, period] like I was unconsciously doing, and some kind of relativisation is going on, that the word order will be *perceived* by them to have changed in the relative clause?
Not necessarily, because relative clauses don't necessarily have to have changed word order. Or passivisation.

In general, speakers will be aware when passivisation alters the order of agent and patient - but the order of subject and object is more important, since that's what determines the syntax. This is probably even more in the minds of speakers of languages in which the verb agrees more obviously with the subject, or where subject and object are marked more explicitly - because English subject and objects are only marked by word order, suppletion of personal pronouns, and marginal verb agreement (-s when the subject is 3rd person singular), it's easier for English speakers to confuse agent and subject.
Post Reply