German question: Restritive vs nonrestrictive relative clauses

[zompist]

One of the readers for my syntax book made a claim about German that I'd like feedback on from the German speakers here.

He claimed that there's no syntactic difference in German between restrictive and non-restrictive relative clauses. To save typing, I'm going to write RRC and NRC. So, e.g.

Der König, der die Götter liebt, kann nicht scheitern.

is ambiguous between

The king who loves the Gods cannot fail. (RRC)
The king, who loves the Gods, cannot fail. (NRC)

(If you're shaky on the difference, in a NRC the relative clause is just a parenthetical— the sentence is logically the same as "The king cannot fail" but adds a factoid about the king. In the first sentence the relative clause is essential as it specifies which kings we are talking about.)

Now that may well be, but in English there are other differences between RRCs and NRCs. One, only RRCs can follow an indefinite pronoun:

Everyone who drank Fred's beer ended up snoring.
*Everyone, who drank Fred's beer, ended up snoring.

Two, only NRCs can follow a bare proper noun.

Bruce Wayne, who is my friend, would never be a vigilante.
*Bruce Wayne who is my friend would never be a vigilante.

Three, you can extrapose freely only from RRCs:

Bill, who I know from grade school, was just arrested.
?Bill was just arrested, who I know from grade school.

Four, you can leave out the subordinator only for RRCs:

The king the Gods love cannot fail.
*The king, the Gods love, cannot fail.

Finally, you can move our clitic possessive 's after a RRC, but not after a NRC:

The girl I like’s house is right here.
*The girl, who I like, ’s house is right here.

So, my question is, do any of these differences work for German too? And if not, are there any other things that might distinguish RRC from NRC?

[cedh]

There is generally no orthographic difference in German between restrictive and non-restrictive relative clauses, because all types of relative clauses are usually separated with commas. However, I would say there is in fact a syntactic difference, although it's less significant compared to English.

Der König, der die Götter liebt, kann nicht scheitern.

is ambiguous between

The king who loves the Gods cannot fail. (RRC)
The king, who loves the Gods, cannot fail. (NRC)

Orthographically this sentence is indeed ambiguous. In the spoken language, the two meanings would be disambiguated by prosody though: The unmarked stress pattern with the accent on König is typically interpreted as a NRC, but a RRC reading can be forced by putting stress on the article (Dér König...).

in English there are other differences between RRCs and NRCs. One, only RRCs can follow an indefinite pronoun:

Everyone who drank Fred's beer ended up snoring.
*Everyone, who drank Fred's beer, ended up snoring.

Jeder, der Freds Bier getrunken hatte, schlief ein.
Same in German; only a RRC reading is available.

Two, only NRCs can follow a bare proper noun.

Bruce Wayne, who is my friend, would never be a vigilante.
*Bruce Wayne who is my friend would never be a vigilante.

Bruce Wayne, der mein Freund ist, würde nie zur Selbstjustiz greifen.
Same in German; only a NRC reading is available.

Three, you can extrapose freely only from RRCs:

Bill, who I know from grade school, was just arrested.
?Bill was just arrested, who I know from grade school.

Bill ist verhaftet worden, den ich aus der Schule kenne.
No problem in colloquial speech. The default reading would still be restrictive ("it's the Bill I know from school, not the other Bill"), but a NRC reading is possible too ("The Bill we were talking about earlier has been arrested, and by the way, I know him from school"). In more formal written German, the construction would be avoided though if the intended reading was nonrestrictive.

Four, you can leave out the subordinator only for RRCs:

The king the Gods love cannot fail.
*The king, the Gods love, cannot fail.

Der König, den die Götter lieben, kann nicht scheitern.
Both readings possible, depending on prosody. You can't leave out the subordinator in German in either case.

Finally, you can move our clitic possessive 's after a RRC, but not after a NRC:

The girl I like’s house is right here.
*The girl, who I like, ’s house is right here.

Das Haus von der Frau, die ich mag, ist genau hier.
German doesn't have a clitic possessive marker, so this test is not applicable.

are there any other things that might distinguish RRC from NRC?

If you can insert the adverb übrigens 'by the way', it must be an NRC:

Julia, die (übrigens) in Köln aufgewachsen ist, wohnt jetzt in Berlin.
"Julia, who grew up in Cologne (by the way), is now living in Berlin."

[dhok]

There is generally no orthographic difference in German between restrictive and non-restrictive relative clauses, because all types of relative clauses are usually separated with commas. However, I would say there is in fact a syntactic difference, although it's less significant compared to English.

Is there a prosodic difference?

[cedh]

Yes:

Der König, der die Götter liebt, kann nicht scheitern.

is ambiguous between

The king who loves the Gods cannot fail. (RRC)
The king, who loves the Gods, cannot fail. (NRC)

Orthographically this sentence is indeed ambiguous. In the spoken language, the two meanings would be disambiguated by prosody though: The unmarked stress pattern with the accent on König is typically interpreted as a NRC, but a RRC reading can be forced by putting stress on the article (Dér König...).

That is:
Der König, der die Götter liebt, kann nicht scheitern. (default reading NRC)
Der König, der die Götter liebt, kann nicht scheitern. (only RRC reading possible)

[cedh]

are there any other things that might distinguish RRC from NRC?

I've got another one:

If you can replace the definite article of the head noun with the determiner derjenige/diejenige/dasjenige 'the one who/which...', it must be a RRC:

Der/Derjenige Mann, der in diesem Haus lebt, ist seit Tagen nicht gesehen worden.
"The man who lives in this house has not been seen for days." (RRC)

Der/*Derjenige Bürgermeister, der in diesem Haus lebt, ist seit Tagen nicht gesehen worden.
"The mayor, who lives in this house, has not been seen for days." (NRC)

This determiner is quite peculiar btw, as it requires its NP to contain a restrictive relative clause! (The RRC may be omitted if the noun is highly topical and the content of the RRC can be inferred from context, but it must be there semantically for the d**jenige construction to be grammatical.)

[Salmoneus]

Orthographically this sentence is indeed ambiguous. In the spoken language, the two meanings would be disambiguated by prosody though: The unmarked stress pattern with the accent on König is typically interpreted as a NRC, but a RRC reading can be forced by putting stress on the article (Dér König...).

Fascinating! Of all the words to stress to force that meaning, I'd never have guess that one!

However, I'm not sure this is a syntactic difference. All sorts of semantic implications can be created by simply emphasising one word or another, but I'm not sure this can really be called syntax. Or at least: while emphasis can be used to indicate that a given syntax is intended, that relies on us having first established that there are really two syntactic options to choose from. And the fact that two meanings imply different syntactic choices in English doesn't mean there are necessarily two different syntactic options in German.

Everyone who drank Fred's beer ended up snoring.
*Everyone, who drank Fred's beer, ended up snoring.

Jeder, der Freds Bier getrunken hatte, schlief ein.
Same in German; only a RRC reading is available.
[/quote]
But again, not really syntax. Specifically: since an indefinite pronoun semantically excludes the possibility of a definite relative modifier, the fact that semantically an RRC reading is required doesn't seem to give us any new information.

Or to put it another way: this isn't testing whether a syntactic construction can be be used in a certain context, it's testing what semantic interpretation is likely in a certain context, which is an issue of semantics, not syntax.

Two, only NRCs can follow a bare proper noun.

Bruce Wayne, who is my friend, would never be a vigilante.
*Bruce Wayne who is my friend would never be a vigilante.

Bruce Wayne, der mein Freund ist, würde nie zur Selbstjustiz greifen.
Same in German; only a NRC reading is available.

Ditto: the definition of an RRC precludes it from ever modifying a bare proper noun, semantically. You can't restrict the reference to less than one single entity...

Three, you can extrapose freely only from RRCs:

Bill, who I know from grade school, was just arrested.
?Bill was just arrested, who I know from grade school.

Bill ist verhaftet worden, den ich aus der Schule kenne.
No problem in colloquial speech. The default reading would still be restrictive ("it's the Bill I know from school, not the other Bill"), but a NRC reading is possible too ("The Bill we were talking about earlier has been arrested, and by the way, I know him from school"). In more formal written German, the construction would be avoided though if the intended reading was nonrestrictive.

Since both readings are possible, it's not clear there's a syntactic difference.
[FWIW, the "English" sentence there is very much ungrammatical for me]



I guess my point is that "can you put X in place Y?" tests work great if it's clear you're dealing with two different things, A and B. You can see how A goes in some places, and B goes in others. But if A and B are superficially identical, and the question is whether they're even two different things at all, I don't think they really tell us anything (other than about semantics). Because you can't say, "look, we can't put B in this frame!", because yes, you can, you just call it 'A' when you do...

[zompist]

Everyone who drank Fred's beer ended up snoring.
*Everyone, who drank Fred's beer, ended up snoring.

But again, not really syntax. Specifically: since an indefinite pronoun semantically excludes the possibility of a definite relative modifier, the fact that semantically an RRC reading is required doesn't seem to give us any new information.

One can certainly decide that a given phenomenon is semantics rather than syntax, but you can't just assume it. Besides, it depends on your theory. Just about all modern syntactic theories incorporate some semantics. Some try to minimize this, some try to maximize it.

Note that some languages, such as Farsi, explicitly mark RRCs. It'd be interesting to know how Farsi handles these cases.

Anyway, I'm not sure what you're saying by "an indefinite pronoun semantically excludes the possibility of a definite relative modifier". It sounds like you're saying that definiteness is a semantic notion, which I wouldn't agree with anyway— it's pragmatic. (Russell's attempt at a semantic account has plenty of holes in it.)

Bruce Wayne, who is my friend, would never be a vigilante.
*Bruce Wayne who is my friend would never be a vigilante.

Ditto: the definition of an RRC precludes it from ever modifying a bare proper noun, semantically. You can't restrict the reference to less than one single entity...

Where are you getting these rules? How do you characterize "bare proper noun" in semantic terms? It's a syntactic fact, not a semantic one.

You can easily have RRCs with a proper noun:

The Bruce Wayne I know would never be a vigilante.

Which syntactically is not bare, but semantically refers to the same Bruce Wayne. This sentence could be used to pick out one of several Bruce Wayne, but it need not— there is no implication that the speaker (or listener) knows multiple Bruce Waynes.