Conlang Random Thread

[bradrn]

I asked this question earlier, but never got any answer to it:

I know of at least five strategies to mark complement clauses:

• Use of a complementiser particle (used in e.g. English)
• Use of an interrogative word (e.g. English)
• Addition of an affix or clitic to the predicate (e.g. Tariana)
• Other morphological change (e.g. Jarawara modifies the final vowel of the predicate)
• Other strategies which link two clauses without actually making a complement clause (Dixon calls these ‘complementation strategies’; e.g. serial verb constructions, nominalization, purposive clause linking, apposition etc.)

Aside from these, are there any other strategies used?

[Qwynegold]

You’re welcome! I do find this topic particularly interesting, so I’m always happy to talk about it.

Thanks! Syntax is something I myself dread. But now I have at least this bit resolved.

[bradrn]

You’re welcome! I do find this topic particularly interesting, so I’m always happy to talk about it.

Thanks! Syntax is something I myself dread. But now I have at least this bit resolved.

You’re welcome! I dread syntax as well — but by contrast, I find morphosyntax to be pretty interesting!

[akam chinjir]

I know of at least five strategies to mark complement clauses:

[...]

Aside from these, are there any other strategies used?

I liked Michael Noonan's chapter on Complementation in Language Typology and Syntactic Description.

One thing you don't seem to mention is verb-deranking---e.g., allowing a smaller range of tenses or aspects in subordinate clauses.

A lot of things could fall under "an affix or...other morphological change." But it would be important to think about what the morphological change represents. E.g., are you building a participle? Some sort of nominalisation? Or is it more like a mood marker or something? Also: is it a particular affix? Or a switch to a different agreement paradigm, or something like that?

You could also just not mark complement clauses, but you're asking how they can be marked, so maybe you've already considered or ruled out that possibility.

[bradrn]

I know of at least five strategies to mark complement clauses:

[...]

Aside from these, are there any other strategies used?

I liked Michael Noonan's chapter on Complementation in Language Typology and Syntactic Description.

That looks like an excellent source; how can I get access? I’ve been referring to the book Complementation by Dixon and Aikhenvald (henceforth referred to as ‘Dixon’), which is also pretty good — but oddly enough, it doesn’t have anything on how to form complement clauses.

One thing you don't seem to mention is verb-deranking---e.g., allowing a smaller range of tenses or aspects in subordinate clauses.

I didn’t mention this because it wasn’t my question. I was asking about how to form complement clauses, not what phenomena they show.

A lot of things could fall under "an affix or...other morphological change." But it would be important to think about what the morphological change represents. E.g., are you building a participle? Some sort of nominalisation? Or is it more like a mood marker or something? Also: is it a particular affix? Or a switch to a different agreement paradigm, or something like that?

You could also just not mark complement clauses, but you're asking how they can be marked, so maybe you've already considered or ruled out that possibility.

In this case, what I meant by that bullet point was that a complement clause could be marked by a separate affix specifically for complementation. But you’re right in saying that participles or moods could work too, as could nominalizations. And I don’t see the difference between a different affix and a switch to a different agreement paradigm as too interesting: they’re both fundamentally a type of morphological change. I‘m more interested in what the morphological change represents than what (morpho-)phonological shape it takes.

I think one factor I also didn’t realise is that a complement clause can be marked in multiple ways simultaneously: for instance, the complement clause in ⟨John’s playing the national anthem⟩ pleased Mary is marked both by a possessive and a participle.

At this point I think it would be useful to bring in some terminology. Dixon distinguishes between complement clause constructions and complementation strategies. The former you already know; the latter are other constructions which give a subordinate meaning but are not actually complement clauses themselves. (Dixon never actually defines the term properly, but I believe this was his intended definition.) For instance, out of your list, nominalisation and apposition (‘not marking’) are both complementation strategies rather than ‘complement clauses’ per se.

Now I think that this distinction between ‘complementation strategies’ and ‘complement clause constructions’ is pretty vague and useless. But Dixon distinguishes them, and he has an excellent overview of various complementation strategies but none on complement clause constructions. Specifically, he covers serial verb constructions, relative clauses used as complement clauses, nominalisation, apposition, clause chaining and purposive (‘in order to’) linking. So I want to know specifically about complement clause constructions which are not in that list.

[akam chinjir]

That looks like an excellent source; how can I get access? I’ve been referring to the book Complementation by Dixon and Aikhenvald (henceforth referred to as ‘Dixon’), which is also pretty good — but oddly enough, it doesn’t have anything on how to form complement clauses.

lib-sci seems to have it.

Responding to a couple of things at once: I was thinking that verb-deranking could be your only overt signal that something's a complement clause, and that this could be revealed by a different agreement paradigm (a subjunctive paradigm or something).

I think one factor I also didn’t realise is that a complement clause can be marked in multiple ways simultaneously: for instance, the complement clause in ⟨John’s playing the national anthem⟩ pleased Mary is marked both by a possessive and a participle.

That's a nominalisation, though. (You could mark it a third time it by using John's playing of the national anthem.)

[bradrn]

I think one factor I also didn’t realise is that a complement clause can be marked in multiple ways simultaneously: for instance, the complement clause in ⟨John’s playing the national anthem⟩ pleased Mary is marked both by a possessive and a participle.

That's a nominalisation, though. (You could mark it a third time it by using John's playing of the national anthem.)

No, it isn’t. It only becomes a nominalisation when of is included. This point is a bit subtle, but Dixon notes that The playing of the national anthem is acceptable, while *The playing the national anthem isn’t. You can also use adjectives with the one but not the other, e.g. The impressive playing of the national anthem vs *The impressive playing the national anthem.

[KathTheDragon]

It might be worth using the two terms "gerund" and "verbal noun" for the contructions "playing the national anthem" and "(the) playing of the national anthem". The difference can probably be characterised in terms of their internal syntax: the relation between the gerund and its dependent(s) is verbal, while the relation between the verbal noun and its dependent(s) is nominal. This would be why the verbal noun requires "of" and allows adjectives, since these are characteristic of nouns, while the gerund does not.

[bradrn]

It might be worth using the two terms "gerund" and "verbal noun" for the contructions "playing the national anthem" and "(the) playing of the national anthem". The difference can probably be characterised in terms of their internal syntax: the relation between the gerund and its dependent(s) is verbal, while the relation between the verbal noun and its dependent(s) is nominal. This would be why the verbal noun requires "of" and allows adjectives, since these are characteristic of nouns, while the gerund does not.

That classification makes a lot of sense! But I’m not sure how this relates to complementation in English.

(Of course, they don’t have to be related — but I want to stay fairly on topic so I can get an answer to my original question.)

[akam chinjir]

No, it isn’t. It only becomes a nominalisation when of is included. This point is a bit subtle, but Dixon notes that The playing of the national anthem is acceptable, while *The playing the national anthem isn’t. You can also use adjectives with the one but not the other, e.g. The impressive playing of the national anthem vs *The impressive playing the national anthem.

Good points!

[bradrn]

No, it isn’t. It only becomes a nominalisation when of is included. This point is a bit subtle, but Dixon notes that The playing of the national anthem is acceptable, while *The playing the national anthem isn’t. You can also use adjectives with the one but not the other, e.g. The impressive playing of the national anthem vs *The impressive playing the national anthem.

Good points!

Thank you! But I must admit that I would never have come up with that first point on my own: it took a while staring at the book I’m using before I found it listed as an example of how to distinguish the two. On the other hand, once I had that first point, the second was easy to figure out. (But it’s slightly wrong: A better comparison would be John’s impressive playing of the national anthem vs *John’s impressive playing the national anthem, which works just as well.)


Anyway: Now that we all understand what we’re talking about, what is the answer to my question? That is — have I already listed all the various ways to make complement clauses, or are there other ways I haven’t mentioned yet?

[akam chinjir]

The Noonan article I linked is probably more useful than anything I can come up with, but here are some thoughts.

One thing is that I'm really never sure what to do with interrogative complements, like in "I asked what you did last summer." I don't know if I've never read anything useful about these, or if I've just managed to forget it all. One complication is that I've often confused those structures with what you see in "I know what you did last summer," where I think you've got a headless relative clause, not really a complement clause at all; and I also don't know much about how this sort of thing is handled, cross-linguistically.

In general, I think you'd probably expect a marker for complement clauses to show up in what Chomskyans call the C domain. This could be a complementiser, strictly speaking (that's actually C for "complementiser"), or it could be something else (the English question words are something else). And what's going to happen is going to depend on the details of your syntax.

Like, in German, as Chomskyans normally analyse it, the verb in a matrix clause moves to C (and then something else moves in front of it); but in a subordinate clause, that position is already taken by the complementiser, so the verb stays in its more basic clause-final position, and you get complementation (or anyway subordination) signalled by word order.

The English interrogative cases also involve word order, since unlike in a matrix question you don't get subject/verb inversion. So maybe there's a substantive answer to your question, it can also be done with word order.

Besides that, if you've got a complementiser, there are various places it can end up: as a prefix or suffix, as a clitic, maybe as a freestanding particle. (I guess English "that" doesn't have to be a clitic?) But I expect that what happens with it is going to depend on your regular clausal syntax---like, if your verb regularly moves to C, you'll get a suffix; maybe it can happen that your verb moves to C only when there's an overt complementiser, and you get SVO word order switching to VSO in subordinate clauses; maybe verbs move to C but nonverbal predicates don't, so you get a difference there. Whichever way you go, the decision is presumably going to ramify throughout other parts of your grammar.

[cedh]

At this point I think it would be useful to bring in some terminology. Dixon distinguishes between complement clause constructions and complementation strategies. The former you already know; the latter are other constructions which give a subordinate meaning but are not actually complement clauses themselves. (Dixon never actually defines the term properly, but I believe this was his intended definition.) For instance, out of your list, nominalisation and apposition (‘not marking’) are both complementation strategies rather than ‘complement clauses’ per se.

Now I think that this distinction between ‘complementation strategies’ and ‘complement clause constructions’ is pretty vague and useless. But Dixon distinguishes them, and he has an excellent overview of various complementation strategies but none on complement clause constructions. Specifically, he covers serial verb constructions, relative clauses used as complement clauses, nominalisation, apposition, clause chaining and purposive (‘in order to’) linking. So I want to know specifically about complement clause constructions which are not in that list.

As for Dixon's terminology, I think an important factor in distinguishing between ‘complementation strategies’ and ‘complement clause constructions’ is that the former is a functional term, and the latter is a formal term. A complementation strategy is a method to use a predicate in the role of a nominal element in a second, matrix predicate. A complement clause is a specific type of complementation strategy, namely one that subordinates a predicate while keeping it in the syntactic shape of a clause. (And since non-clausal complementation strategies sometimes get overlooked in language descriptions, Dixon's book focuses on those.)
(I said something similar recently about the distinction between adverbial phrases and adpositional phrases.)

[bradrn]

At this point I think it would be useful to bring in some terminology …

As for Dixon's terminology, I think an important factor in distinguishing between ‘complementation strategies’ and ‘complement clause constructions’ is that the former is a functional term, and the latter is a formal term. A complementation strategy is a method to use a predicate in the role of a nominal element in a second, matrix predicate. A complement clause is a specific type of complementation strategy, namely one that subordinates a predicate while keeping it in the syntactic shape of a clause. (And since non-clausal complementation strategies sometimes get overlooked in language descriptions, Dixon's book focuses on those.)

I wish more people were this precise! I think you’ve put your finger on what was bothering me so much about this terminology, and your explanation helps enormously in clarifying what exactly these terms mean.

(I said something similar recently about the distinction between adverbial phrases and adpositional phrases.)

Yes, I do remember this now… sorry you had to explain this twice! After this much discussion about all the syntactic topics I’ve been asking about, by now I should really have been able to figure out this stuff myself.

The Noonan article I linked is probably more useful than anything I can come up with, but here are some thoughts.

In that case, I’ll have to see if I can find it anywhere.

One thing is that I'm really never sure what to do with interrogative complements, like in "I asked what you did last summer." I don't know if I've never read anything useful about these, or if I've just managed to forget it all. One complication is that I've often confused those structures with what you see in "I know what you did last summer," where I think you've got a headless relative clause, not really a complement clause at all; and I also don't know much about how this sort of thing is handled, cross-linguistically.

In general, I think you'd probably expect a marker for complement clauses to show up in what Chomskyans call the C domain. This could be a complementiser, strictly speaking (that's actually C for "complementiser"), or it could be something else (the English question words are something else). And what's going to happen is going to depend on the details of your syntax.

Like, in German, as Chomskyans normally analyse it, the verb in a matrix clause moves to C (and then something else moves in front of it); but in a subordinate clause, that position is already taken by the complementiser, so the verb stays in its more basic clause-final position, and you get complementation (or anyway subordination) signalled by word order.

The English interrogative cases also involve word order, since unlike in a matrix question you don't get subject/verb inversion. So maybe there's a substantive answer to your question, it can also be done with word order.

I don’t know much at all about Chomskyanist syntax, but this does sound like a very nice way to explain things. And ‘word order’ does seem like a very reasonable method which I missed.

Besides that, if you've got a complementiser, there are various places it can end up: as a prefix or suffix, as a clitic, maybe as a freestanding particle. (I guess English "that" doesn't have to be a clitic?) But I expect that what happens with it is going to depend on your regular clausal syntax---like, if your verb regularly moves to C, you'll get a suffix; maybe it can happen that your verb moves to C only when there's an overt complementiser, and you get SVO word order switching to VSO in subordinate clauses; maybe verbs move to C but nonverbal predicates don't, so you get a difference there. Whichever way you go, the decision is presumably going to ramify throughout other parts of your grammar.

Now I’m really confused. Before, you said that the verb can move to C unless there’s a complementiser here — but now, you seem to be saying that the verb may move to C only if there is already a complementiser here. What have I missed?

[akam chinjir]

Now I’m really confused. Before, you said that the verb can move to C unless there’s a complementiser here — but now, you seem to be saying that the verb may move to C only if there is already a complementiser here. What have I missed?

Oh, just different rules in different languages. It might be as simple as, the German complementiser is listed as a free form, so it can't combine with the verb; but in those other languages the complementiser is listed as an affix, so it needs something to move to it.

[bradrn]

Now I’m really confused. Before, you said that the verb can move to C unless there’s a complementiser here — but now, you seem to be saying that the verb may move to C only if there is already a complementiser here. What have I missed?

Oh, just different rules in different languages. It might be as simple as, the German complementiser is listed as a free form, so it can't combine with the verb; but in those other languages the complementiser is listed as an affix, so it needs something to move to it.

That makes a lot more sense now — I thought that e.g. ‘verb can move to C unless a complementiser is there already’ was a universal rule, so thanks for correcting me.

One thing is that I'm really never sure what to do with interrogative complements, like in "I asked what you did last summer." I don't know if I've never read anything useful about these, or if I've just managed to forget it all. One complication is that I've often confused those structures with what you see in "I know what you did last summer," where I think you've got a headless relative clause, not really a complement clause at all; and I also don't know much about how this sort of thing is handled, cross-linguistically.

I’ve just realised I never actually responded to this from your previous post — I had an idea of what to say, realised it was incorrect, and then forgot I never wrote anything and submitted my post without checking. So I guess I should respond to it now.

Anyway: I suspect that both your sentences use a headless relative clause rather than a complement clause. Mainly this is because you can transform ‘what you did last summer’ into the NP ‘the thing that you did last summer’ without affecting the meaning of either sentence; although I have read next to nothing about headless relative clauses, I suspect that this is a rather key attribute distinguishing headless relative clauses from other types of complement clause. Dixon would probably call this sort of thing a ‘complementation strategy’: in particular, this is a relative clause being used to express a subordinate meaning. By contrast, a restatement of your example with a ‘proper’ subordinate clause would probably be something like I asked you to tell me about what you did last summer.

(Ask is actually a pretty interesting verb in terms of complementation. It is almost certainly what Dixon would call a ‘Primary-B’ verb — that is, it can take either an NP or a complement clause as an object, while most verbs can only take one or the other. This is why both the NP what you did last summer and the clause to tell me about what you did last summer can be used as objects of ask, as well as a simpler NP like the question (as in I asked the question). But ask is interesting because it is ambitransitive, and the complement clause goes in the indirect object slot rather than the direct object slot, like what happens with most other Primary-B verbs such as know. So this results in a situation where the implied object of a complement clause such as to speak can change depending on transitivity: for instance, I asked you is pretty similar in meaning to I asked, but I asked you to speak is not the same as I asked to speak (= I asked if I could speak, using yet another type of complementation strategy)! But I digress.)

[akam chinjir]

Anyway: I suspect that both your sentences use a headless relative clause rather than a complement clause. Mainly this is because you can transform ‘what you did last summer’ into the NP ‘the thing that you did last summer’ without affecting the meaning of either sentence;

I'm pretty sure "I asked the thing that you did last summer" (or "I asked that which you did last summer") is ungrammatical for me, fixable with "about."

[bradrn]

Anyway: I suspect that both your sentences use a headless relative clause rather than a complement clause. Mainly this is because you can transform ‘what you did last summer’ into the NP ‘the thing that you did last summer’ without affecting the meaning of either sentence;

I'm pretty sure "I asked the thing that you did last summer" (or "I asked that which you did last summer") is ungrammatical for me, fixable with "about."

For me it’s grammatical. Slightly odd, but perfectly grammatical — and no less grammatical for me than ‘I asked what you did last summer’, which is also a bit odd (needs an ‘about’). Think of it like ‘I asked the question’. The second one is ungrammatical for me as well though.

[bradrn]

I’ve been reading that Noonan article, and it’s proved incredibly informative — thanks for the recommendation akam chinjir! I think that now I finally have enough information to figure out what I want my conlang’s complement system to look like:

• There are two ways to form complement clauses: ‘s-like’ complement clauses (to use Noonan’s terminology), and nominalization.
• S-like complement clauses are structured the same as a regular independent clause, but with the complementiser clitic =C added to the end. (I haven’t figured out what the words actually are yet — I’m just sketching the grammar for now.) These complement clauses are only used as objects of a transitive verb or subject of an intransitive verb (the language is ergative). Although the regular word order is SOV, s-like complement clauses are regularly extraposed to the end of a sentence e.g. ‘I know that you see it’ 1s-ERG 3s-1s-know [2s-ERG 3s-ABS 3s-2s-see]=C.
• Nominalization is marked by the nominalization affix -NZN on the verb (which then can be marked for noun case etc. as the head of the clause). The absolutive argument of the verb is marked as a possessor using the genitive; the ergative argument (if any) remains in the ergative case. e.g. ‘I want to go there slowly’ 1s-ERG [1s-ERG there-GEN 1s-go-NZN-ACC slowly] 3s-1s-want.
• Generally, s-like complements are used in situations where the complement clause could refer to any time (e.g. ‘I know that you see/saw/will see it’), whereas nominalized clauses are used in situations where the clause is restricted to a specific time (e.g. ‘I make you do it / *do it yesterday’). (Noonan calls these ITR and DTR respectively.) Nominalization is also used when the complement clause is used as a subject, since s-like clauses cannot be used there. Beyond this I don’t want to get too detailed: this is only meant to be a protolanguage, and I don’t want to get too bogged down in its details.

I do have some questions about this though:

1. Most importantly, does the above system sound realistic?
2. Noonan states that ‘all languages have some sort of reduced complement type in opposition to the indicative’. But what exactly is a ‘reduced complement type’? Noonan defines it fairly straightforwardly as ‘any complement type that has fewer syntactic and inflectional possibilities than an indicative main clause’, but then never refers to this definition again and keeps on talking about ‘grammatical relations’, whatever those are.
3. Should my s-type complement clause take case marking? Or is this a special case where languages usually omit case?
4. For my nominalization construction, since the absolutive argument is marked as a possessor, I have a situation where there appears to be an ergative argument but no absolutive argument. Is this realistic?
5. Out of the two complement clause types I have described, I want one to have a restricted range of tense/aspect combinations available (which I’ve heard is pretty normal in complement clauses). For which clause type is this most plausible? What sort of restriction should I use? (I suspect the answer to this is in that Dixon article I keep on talking about — maybe I’ll check it later when I have more time.)

[cedh]

1. Most importantly, does the above system sound realistic?
2. Noonan states that ‘all languages have some sort of reduced complement type in opposition to the indicative’. But what exactly is a ‘reduced complement type’? Noonan defines it fairly straightforwardly as ‘any complement type that has fewer syntactic and inflectional possibilities than an indicative main clause’, but then never refers to this definition again and keeps on talking about ‘grammatical relations’, whatever those are.
3. Should my s-type complement clause take case marking? Or is this a special case where languages usually omit case?
4. For my nominalization construction, since the absolutive argument is marked as a possessor, I have a situation where there appears to be an ergative argument but no absolutive argument. Is this realistic?
5. Out of the two complement clause types I have described, I want one to have a restricted range of tense/aspect combinations available (which I’ve heard is pretty normal in complement clauses). For which clause type is this most plausible? What sort of restriction should I use? (I suspect the answer to this is in that Dixon article I keep on talking about — maybe I’ll check it later when I have more time.)

1. Yes, it looks plausible.
2. I would think Noonan's "reduced complement types" cover mostly the same ground as Dixon's non-clausal "complementation strategies". Noonan seems to focus on morphological differences to a finite main clause verb though; a subjunctive clause, for instance, might fit Noonan's category because it appears in a non-indicative mood, even though it is still a clause.
3. IIUC your s-type complement clauses can only appear in an absolutive role, so it seems likely they wouldn't take overt case marking. (But they probably could, if you wanted that.)
4. Yes, cf. English (otherwise not ergative, of course): Chelsea's defeat by Liverpool
5. In your language, the nominalization construction is the deranked/reduced variant because it's not a full clause, so I'd expect this construction to have fewer internal possibilities compared to the s-type construction (but obviously more external roles that it can be used for).