Complex sentences

[The Hanged Man]

Whatc different strategies do natlangs use for making different types of complex sentences? I'm asking, because in my conlangs (now there are two of them ) I can make only simple sentences for now... and I would like to change it, but I don't want to copy all strategies from european languages.

[abeygail]

Well, by complex sentences what do you mean? Do you mean obliques or relative clauses? How complex are your sentences?
Well, to start with relative clauses, Quechua (and I think Tamil) treat them as participles and treat these participles as modifiers of the head noun/nominal.
Example (In English, as I do not know how to embed in Quechua Yet...): "The red fish that I ate yesterday swims to the store on the pier" could be something like "The red fish my eating yesterday swims to the store on the pier". You should look at this wiki link for more info: http://en.wikipedia.org/wiki/Relative_clause
Also, there is a process known as Clause Chaining that I do not fully understand but I am sure that someone on this board could be of help to you on.
Hope I helped!

[fiziwig]

You can make sentences as complex as you want just by tacking more prepositional phrases onto a simple sentence.

I took the train.
I took the train from Denver.
I took the train from Denver to Boston.
I took the train from Denver to Boston on track nine.
I took the train from Denver to Boston on track nine at noon.
I took the train from Denver to Boston on track nine at noon with Harold.
Because of that telegram I took the train from Denver to Boston on track nine at noon with Harold.
Because of that telegram I took the train from Denver to Boston on track nine at noon with Harold so I could take care of the problem.

But I'm sure you had something more substantial than that in mind.

I'd start with asking myself what I'm trying to accomplish with a "more complex" sentence. It makes no sense to say you just want your sentences to be "more complex". You need to know what that extra complexity is good for. It's fine to say "I see the man." but what about "I see that the man is here." or "I see that the man with the big hat is here." or "I wish I could have seen that the man with the big hat had been here."

What kind of complexity depends entirely on what you want to be able to say. That's were you should start IMHO.

[Miekko]

You can make sentences as complex as you want just by tacking more prepositional phrases onto a simple sentence.

I took the train.
I took the train from Denver.
I took the train from Denver to Boston.
I took the train from Denver to Boston on track nine.
I took the train from Denver to Boston on track nine at noon.
I took the train from Denver to Boston on track nine at noon with Harold.
Because of that telegram I took the train from Denver to Boston on track nine at noon with Harold.
Because of that telegram I took the train from Denver to Boston on track nine at noon with Harold so I could take care of the problem.

But I'm sure you had something more substantial than that in mind.

I'd start with asking myself what I'm trying to accomplish with a "more complex" sentence. It makes no sense to say you just want your sentences to be "more complex". You need to know what that extra complexity is good for. It's fine to say "I see the man." but what about "I see that the man is here." or "I see that the man with the big hat is here." or "I wish I could have seen that the man with the big hat had been here."

What kind of complexity depends entirely on what you want to be able to say. That's were you should start IMHO.

generally complexity is taken to mean something like 'sentences embedded in sentences' or equivalent things, not 'sentences with an infinitude of linearly stacked phrases on top of each other' (heck, even wikipedia describes a complex sentence as 'a sentence with at least one dependent clause (subordinating clause)'.

[Yiuel Raumbesrairc]

I am pretty sure Golem is referring to sentences with more than one verb.

That is, adding arguments to a verb does not complexify a sentence, it just complexifies the verb's arguments.

You have two types of complexe sentences : coordinated and subordinated.

A coordinated complex sentence is a sentence like :

X does Y and W does Z.

A subordinated sentence looks about the same :

X does Y because W does Z.

However, in the second sentence, you imply a relation between both sentence. If you reverse the first sentence, you get the same meaning, but if you do it on the second...

W does Z and X does Y.
W does Z because X does Y.**

The second version of the second sentence does not mean the same thing as its first version.

In Japanese, when you want to coordinate sentences, one of the verbs must be conjugated in a conjunctive mood (one of the various uses of the -te form). That is, you cannot use their equivalent of "and" to link sentences coordonatively. There are also subordinative moods (-ba form, -tara form) which do about the same, except that, like English "because" it implies a subordination. (In Japanese, the subordinated sentence receives the subordinative mood.)

(Again, in Japanese, some subordinated sentences are actually linked to its main sentence by means of specific circumstancial complements. As we can say "At the time I left, he had already left.", Japanese will say "Ore ga deta toki ni, aitsu ga mou dete ita.")

[The Hanged Man]

heck, even wikipedia describes a complex sentence as 'a sentence with at least one dependent clause (subordinating clause)'.

And that's what I meant . Languages that I know make complex sentences other than relative clauses by using subordinate conjuctions. I'd like to know, what other ways can be. And is there a relation between the way a language uses, and its typical word order?

[Whimemsz]

One thing you can do is have verbs inflect differently based on whether they're in an independent or a dependent clause--you can even have different inflections for different kinds of dependent clauses.

For example, Algonquian languages have several different inflectional patterns (called "orders"; the number varies depending on the language), including the Independent (generally, for indicative verbs in independent clauses) and the Conjunct (generally used for subordinate clauses, though the specifics tend to vary somewhat from language to language). So, to take an Ojibwe example, while giga-bi-izhaa means simply "you'll come" (independent order, with gi- 2nd person prefix and a null suffix indicating sg.INDPT), if it's part of a subordinate clause the verb is conjugated entirely differently: for example, inga-minwendam giishpin ji-bi-izhaayan "I'll be happy if you come" (conjunct order, with -yan, 2sg.CONJ suffix and no person prefix).

[Miekko]

heck, even wikipedia describes a complex sentence as 'a sentence with at least one dependent clause (subordinating clause)'.

And that's what I meant . Languages that I know make complex sentences other than relative clauses by using subordinate conjuctions. I'd like to know, what other ways can be. And is there a relation between the way a language uses, and its typical word order?

SOV langs have a bit of a tendency to go with participles that can take more or less any of the arguments that a normal verb phrase can instead of relative clauses and such (altho' resumptive pronouns might occur), I dunno if any similar kind of tendency goes for other kinds of subordinate clauses. However, nominalized verbs in various case forms or with adpositions or particles or other subordinating devices do occur in a load of languages, such as 'I heard that he left' -> 'I heard of his leaving', etc. Apparently ancient greek often left participles ungoverned by adpositions or such and it was pretty much up to context to figure out whether a given participle referred to the cause or something else.

[The Hanged Man]

Another question: what can be diachronic source of conjuntions?

[Radius Solis]

I think the two really major diachronic sources for conjunctions are prepositions, and various deictics like demonstratives (e.g. English "that", which has indeed become a conjunction) and sequence adverbs (e.g. "then" or "now", which haven't, but could someday).

Prepositions in particular can easily become subordinating conjunctions, as has happened to many of them in English ("until Tuesday" "until something happens"). Comitative prepositions - meaning "with" - often become coordinating conjunctions meaning "and".

[psygnisfive]

For sentence complexity, look into things like extraction islands, quantificational structures, and comparative structures. They all involve wonderfully interesting complexities.

[The Hanged Man]

I thought about some grammaticalization paths. For example

same+time, time>when, while

In sentences like:
"Mike was cribbing in time of his teacher's not-watching"> Mike was cribbing while his techer wasn't watching

while, after > who, that

by reanalysis of sentences like:
Mike met a girl after her going for a walk > Mike met a girl who has went for a walk
Mike met a girl while she was cycling> Mike met a girl, who was cycling

What do you think about my ideas? Do they look plausible?

Do you know some attested grammaticalization paths for relative clause markers (not only pronouns)?

Another thing is that, most of conjunctions based on other lexical sources (like adpositions) that I can imagine look more natural to me with nominalized complements. However, most of them (at least in european langauges) are used with subordinated clauses. For example, while it is natural to use preposition "until" with a noun, when it is used as a conjunction, it does not take nominalized forms, but clauses (not "until happening of something", but "until something happens). I'm curious, how such a use could develop. At least for languages with many such conjunctions, I can simply imagine, that if a langauge develop a new conjunction, it is used like the rest, older conjunctions, just by analogy.
But why could such a feature develop for the first time, when a language develops it's first conjunctions? How could it switched from nominalized forms to clauses?

[CaesarVincens]

I thought about some grammaticalization paths. For example

same+time, time>when, while

In sentences like:
"Mike was cribbing in time of his teacher's not-watching"> Mike was cribbing while his techer wasn't watching

while, after > who, that

by reanalysis of sentences like:
Mike met a girl after her going for a walk > Mike met a girl who has went for a walk
Mike met a girl while she was cycling> Mike met a girl, who was cycling

What do you think about my ideas? Do they look plausible?

That looks fine.

Do you know some attested grammaticalization paths for relative clause markers (not only pronouns)?

Another thing is that, most of conjunctions based on other lexical sources (like adpositions) that I can imagine look more natural to me with nominalized complements. However, most of them (at least in european langauges) are used with subordinated clauses. For example, while it is natural to use preposition "until" with a noun, when it is used as a conjunction, it does not take nominalized forms, but clauses (not "until happening of something", but "until something happens). I'm curious, how such a use could develop. At least for languages with many such conjunctions, I can simply imagine, that if a langauge develop a new conjunction, it is used like the rest, older conjunctions, just by analogy.
But why could such a feature develop for the first time, when a language develops it's first conjunctions? How could it switched from nominalized forms to clauses?

I don't think there can be any real productive, but only speculative, look at how original conjunctions came about. Other than to say that they most likely came from adpositions, which would have come from nouns.

[The Hanged Man]

I'm reviving this old thread (at first, I didn't even realise, how old it is...) because perhaps now some more people will know how to help me, perhaps giving some radically different answers. And because now I'm dealing with the topic of complex sentences again...

So, now I'm going to create a conlang that has SVO as its unmarked word order, and marks relative clauses with a conjunction. It's different subordinate clauses (equivalents of english "when", "where" etc. clauses) will also be constructed by relativization. However, I want some subordinate clauses to be formed using a verbal noun, and a preposition, meaning 'after', 'before', 'for', etc. Object of the verb could be expressed by a noun phrase in accusative (which would be unmarked), and subject could be marked by a noun phrase preceded by a prepostion meaning 'by'. The problem is I don't know if such a structure would be naturalistic in any language. And if so, would it also be consistent with characteristics of my conlang? Do you know any SVO languages that use only verb nominalization in subordinate clauses? I know that english can do that, although it's not the only strategy, as I want it to be in my conlang.

[hwhatting]

Do you know any SVO languages that use only verb nominalization in subordinate clauses? I know that english can do that, although it's not the only strategy, as I want it to be in my conlang.

I don't, but the sample of SVO languages I know is limited; mostly European IE. I'd say it's not very likely, especially as you use ponoun-based relative sentences - there's normally a lot of overlap in the construction principles of relative sentences and other subordinate sentences, and if your relative sentences use rel. pronoun plus finite verbs, I'd say it's very likely that the language would also have conjunctions plus finite verbs in subordinate sentences at least as one of the options, and not only nominalisation.

[Whimemsz]

In a number of Amazonian languages, nominalization seems to be the overwhelmingly most common strategy for creating subordinate verbs (unfortunately, it's hard to tell from some of the descriptions whether it's literally the only strategy, or just the primary one...). I mentioned this offhand in another thread several months ago, and chris_notts' response is useful and pertinent, so I'll reproduce it here:

Not just Amazonian languages. I recommend "Subordination" by Sonia Cristofaro... she distinguishes between "balanced" strategies (subordinate clause looks like a main clause) and "deranked" strategies (subordinate clause is different - either it has special morphology or it's a nominalisation or...). She then establishes a typology of different kinds of subordinate clause, and how likely they are to be be deranked versus balanced. The key insight is that the more the nature of the main clause determines the participants / tense / aspect / modality of the subordinate clause, the more likely the clause is to show deranking rather than balancing behaviour. She establishes a hierarchy of subordination types which is confirmed by her sample... I can reproduce it if it interests you. The key point is that many (most?) languages have more than one subordination strategy - English, for example, switches between use of infinitives / particles as deranking strategies and subordination using "that" etc. as balancing strategies.

Other interesting questions include how much of the inflections on finite verbs is preserved in nominalisations, whether case marking / word order / ... is the same or different in various kinds of subordinate clause (e.g. some languages mark the object and/or subject with the genitive in subordinate clauses), which arguments can be overtly expressed in different kinds of subordinate clause etc.

[merijn]

All languages that extensively use nominalizations that I know of are SOV languages, but that doesn't mean that there are no SVO languages that use nominalizations, I cannot see why there shouldn't be any. I can think of a reason why nominalizations are more common in SOV languages, however. SOV languages are more often languages that rely on case to establish who did what, quite often almost exclusively so (I would think that Japanese is a good example of such a language). If a language relies on case then it might want to case-mark its subordinate clauses and that is much easier when it is a nominalization. I am not sure how right I am with this explanation, though.

[merijn]

In a number of Amazonian languages, nominalization seems to be the overwhelmingly most common strategy for creating subordinate verbs (unfortunately, it's hard to tell from some of the descriptions whether it's literally the only strategy, or just the primary one...). I mentioned this offhand in another thread several months ago, and chris_notts' response is useful and pertinent, so I'll reproduce it here:

Not just Amazonian languages. I recommend "Subordination" by Sonia Cristofaro... she distinguishes between "balanced" strategies (subordinate clause looks like a main clause) and "deranked" strategies (subordinate clause is different - either it has special morphology or it's a nominalisation or...). She then establishes a typology of different kinds of subordinate clause, and how likely they are to be be deranked versus balanced. The key insight is that the more the nature of the main clause determines the participants / tense / aspect / modality of the subordinate clause, the more likely the clause is to show deranking rather than balancing behaviour. She establishes a hierarchy of subordination types which is confirmed by her sample... I can reproduce it if it interests you. The key point is that many (most?) languages have more than one subordination strategy - English, for example, switches between use of infinitives / particles as deranking strategies and subordination using "that" etc. as balancing strategies.

Other interesting questions include how much of the inflections on finite verbs is preserved in nominalisations, whether case marking / word order / ... is the same or different in various kinds of subordinate clause (e.g. some languages mark the object and/or subject with the genitive in subordinate clauses), which arguments can be overtly expressed in different kinds of subordinate clause etc.

There is quite some interesting information in that thread, for instance the hierarchy in one of Chris Notts follow-up posts. I think that one thing that is missing is that deranking is also more common in my experience if the subject of the sub-clause is also an argument of the matrix clause especially if it is the subject of the matrix clause.

[merijn]

I think the two really major diachronic sources for conjunctions are prepositions, and various deictics like demonstratives (e.g. English "that", which has indeed become a conjunction) and sequence adverbs (e.g. "then" or "now", which haven't, but could someday).

Prepositions in particular can easily become subordinating conjunctions, as has happened to many of them in English ("until Tuesday" "until something happens"). Comitative prepositions - meaning "with" - often become coordinating conjunctions meaning "and".

Another common source of conjunctions is the verb "say". For instance in Zulu the main subordinating conjunction is "ukuthi", which is the infinitive of "say". As an infinitive it has nominal morphology and this is exploited by the language by using it in for instance possessive constructions or locative constructions.

[The Hanged Man]

I'd say it's not very likely, especially as you use ponoun-based relative sentences - there's normally a lot of overlap in the construction principles of relative sentences and other subordinate sentences, and if your relative sentences use rel. pronoun plus finite verbs, I'd say it's very likely that the language would also have conjunctions plus finite verbs in subordinate sentences at least as one of the options, and not only nominalisation.

Well, almost all language that I make, or I'm planning to make, don't use relative pronoun. I wrote that the language uses relative conjunction (that is: a relativizer), which doesn't have to be relative pronoun, but merely a conjunction. As one, it won't agree on number (which is the only grammatical category encoded on nouns in my language) with relativized noun or noun phrase, and it won't take its prepositions as well. Instead, there will be resumptive pronoun in relative clause, in place of the noun phrase, and for that, personal pronouns, and/or some anaphoric or demonstrative pronouns will be used. Or there will be nothing in that place, just a gap.

[hwhatting]

Well, almost all language that I make, or I'm planning to make, don't use relative pronoun. I wrote that the language uses relative conjunction (that is: a relativizer), which doesn't have to be relative pronoun, but merely a conjunction. As one, it won't agree on number (which is the only grammatical category encoded on nouns in my language) with relativized noun or noun phrase, and it won't take its prepositions as well. Instead, there will be resumptive pronoun in relative clause, in place of the noun phrase, and for that, personal pronouns, and/or some anaphoric or demonstrative pronouns will be used. Or there will be nothing in that place, just a gap.

Thanks for correcting me here - I misread your post. But even if you have a relative conjunction paired with finite verb forms in your relative sentences, this makes it extremely likely that your language also has subordinating conjunctions plus finite verbs at least as one of the available strategies.