I only know of two strategies. Most European languages, like English, rely mainly intonation to keep the arguments in a sub-clause, particularly center-embedded clauses, from being accidentally to be assumed to be part of the main clause. German seems to be the only exception. All sub-clauses are SOV, so the beginning of all sub-clauses is marked with a relativizer, as expected, but the end is marked with an inflected verb. This means there's no upper limit to how elaborate a sub-clause can get in German.

I want to know other strategies used by other languages, assuming there are any. I've found that with many things, there's only a finite number of ways to do them. There's only so many possible noun phrase orders, same thing with word order. There's only about half a dozen alignments. A few hundred phones all languages can make use of. You get the idea.

Are the two I listed the only two possible strategies or are there alternatives? I feel its euro-centric to put this into a conlang, but I know of no other way.

  • Please give some concrete examples. Thanks.
    – Lambie
    Feb 14 at 23:33

2 Answers 2


Hittite marks the beginning of each clause, and doesn't allow any nesting: you have to chain them together instead of putting one inside another.

For example, to say "send back my subjects that you've taken and led away":

ÌR.MEŠ=YA=wa=z ku-ēs dā-s
subjects=my=QUOT=INT which-N.PL take-2S.PST
"My subjects that you have taken,

nu=war=as=kan kattanta pehute-t
you have led them away,

nu=war=as=mu arha upp-i
send them back to me!"

The nu particle, known as a "sentence connective", doesn't really have any meaning of its own but introduces a new clause when there isn't something else filling that role. (In the first clause, a noun is moved to the beginning for emphasis, so nu isn't needed—the fact that all the clitics stack onto it indicates that it's the start of the clause.)

Certain markers, like =war to indicate a quotation, are also repeated on each new clause head.

  • So, it forbids center-embedding basically?
    – user43283
    Feb 14 at 21:22
  • @user8600 Indeed; no recursing clauses, just iterate and use pointers back to previous clauses as needed
    – Draconis
    Feb 14 at 21:23
  • I have heard of languages that forbid center-embedding. How common is this though? I tried to check on wals, but they don't have this listed as a 'feature' as best as I can tell. Neither the words 'embed' or 'center' turn up anything. Nothing really relevant comes up if I search 'clause', though it did list something about 'double-headed clauses', but I can't find anywhere what that means.
    – user43283
    Feb 14 at 21:44
  • @user8600 I'm afraid I don't know off the top of my head. Hittite is the one language I know that does this.
    – Draconis
    Feb 14 at 21:52

Japanese allows relative clauses:

  • Generally each clause has NOUN1 PARTICLE1 NOUN2 PARTICLE2 NOUN3 PARTICLE3 ... VERB format. Subject is optional and can be skipped if it can be guessed from context (which often happens when it is "I", or "thou" for questions). VERB can be verb or adjective.

  • Relative clause is created by putting VERB before given NOUN. Any particle, which specified function of NOUN, disappears, causing potentially ambiguous meaning. E.g. VERB NOUN. (Particle ga is used to mark subject in declarative main clauses only since late Late Middle Japanese (according to Bjarke Frellesvig (2010, "A History of the Japanese Language", pages 366-368)).)

  • No relative pronouns.

  • No particles or other words mark beginning or end of relative clause.

  • VERB in relative clause should be in Attributive form (連体形), but anyway in modern Japanese, Attributive form (連体形) is identical to Terminal/Conclusive form (終止形) for all verbs except copula na / da. (In Western Old Japanese and Early Middle Japanese, Attributive form and Terminal/Conclusive form were different for adjectives, vowel-stem verbs and few irregular verbs, but not for regular consonant-stem verbs, which comprise majority of verbs. Traces of distinction for consonant-stem verbs exist in Eastern Old Japanese.)


OJ: ki takasi

NJ: ki ga takai

"tree is tall"

OJ: [takaki] ki

NJ: [takai] ki

"tree [which is tall]" / "[tall] tree"

OJ: ware [takaki] ki wo mi

NJ: watasi ga [takai] ki wo miru

"I see tree [which is tall]" / "I see [tall] tree"

OJ: ki takaku ariki

EMJ: ki takakariki

NJ: ki ga takakatta

"tree was tall"

OJ: [takaku arisi] ki

EMJ: [takakarisi] ki

NJ: [takakatta] ki

"tree [which was tall]"

OJ: uwo kupu

NJ: sakana ga taberu / uo ga taberu

"fish eats (somebody/something)"

OJ: uwo wo kupu

NJ: sakana wo taberu / uo wo taberu

"(somebody/something) eats fish"

OJ: [kupu] uwo

NJ: [taberu] sakana / [taberu] uo

"fish [who eats (somebody/something)]" or "fish [whom (somebody/something) eats]"

LMJ: ɸana kirei nari

NJ: hana ga kirei da

"flower is pretty"

LMJ: [kirei naru] ɸana

NJ: [kirei na] hana

"flower [which is pretty]"

LMJ: ware [kirei naru] ɸana wo miru

NJ: watasi ga [kirei na] hana wo miru

"I see flower [which is pretty]"

NJ: watasi wa [[umi de oyoide iru] hito wo tabete iru] same wo mite iru

"I see shark [who is eating human [who is swimming in sea]]"

(Example of sentence with relative clause inside relative clause)

NJ: kore wa [watasi ga anata kara karita] hon da

"this is book [which I borrowed from thee]"

[ ] are used to mark relative clauses.

Explanation of particles used in above examples:

ga - Subject

wo - Direct object

de - Location

kara - Source ("from")

OJ pa / NJ wa - Topic

(Formal copula de aru in modern language has identical Terminal/Conclusive form and Attributive form. da developed as contraction of de aru and is used only in terminal function in modern language. na developed as contraction of older copula naru and is used only in attributive function in modern language. But when na and da initially developed, they were used in both functions.)

Korean has some similarities (at least in using Attributive forms of verbs placed before nouns and lack of relative pronouns).

  • Funny enough, I saw a Japanese speaker demonstrate once how Japanese does allow center-embedded clauses. The only noun phrase part of the main clause that preceeded the embedded clause was marked with the topic marker wa/ha. He claimed it was obvious to him that the first noun phrase was part of the main clause and not the embedded clause, but when I asked how he could determine that he admitted he didn't know, he just knew that was the right way to interpret it. Based on your answer, maybe it was simply the fact that it was a topic, which is often also the subject/agent?
    – user43283
    Feb 15 at 10:31
  • @user8600 Topic particle は is not used inside relative clauses (but は can be found there in different functions), so NOUN+は can be generally assumed to be part of main clause.
    – Arfrever
    Feb 15 at 10:38
  • So, going back to your original post. Japanese basically only allows 'simple' relative clauses. Pronouns, subjects, and topics can only ever be part of the main clause, among perhaps other things? I have heard that English allows far more complex relative clauses than you would normally see, but honestly I've never seen someone state what English allows that other languages do not. German does allow everything English would, plus more. As I said, there's not really a limit to how complex a relative clause can be in German.
    – user43283
    Feb 15 at 10:57
  • @user8600 Japanese does not have relative pronouns, but has personal pronouns and demonstrative pronouns, and they can be placed inside relative clauses without problems. Case particles (including subject particle ga) can be placed inside relative clauses. I added more examples.
    – Arfrever
    Feb 15 at 11:19
  • Also for clarity, topic particle belongs to binding particles (係助詞), not case particles (格助詞). There are also several other types of particles (助詞).
    – Arfrever
    Feb 15 at 11:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.