LINGUIST List 7.546

Sat Apr 13 1996

Disc: German partial-VP fronting, Re: 7.530

Editor for this issue: T. Daniel Seely <dseelyemunix.emich.edu>


Directory

  1. schuelesfs.nphil.uni-tuebingen.de, Re: 7.530, Sum: German partial-VP fronting
  2. Ulrich Koch, Re: German partial-VP fronting
  3. Ralf Steinberger, Re: German partial-VP fronting

Message 1: Re: 7.530, Sum: German partial-VP fronting

Date: Thu, 11 Apr 1996 10:59:10 +0200
From: schuelesfs.nphil.uni-tuebingen.de <schuelesfs.nphil.uni-tuebingen.de>
Subject: Re: 7.530, Sum: German partial-VP fronting
some of my ideas:

as i understand your table, i do perfeclty well fit into the a pattern!? i
do find it a bit shocking, however, that people accept sentence 6, quite
scary, you should interview these people on their reading habits and what
they do for a living... and what other languages they speak and what
foreign or second language contexts they have been immersed in for how long
and what dialect of german they speak..also, there is a big diff what
people accept in reading and writing (depending on the kind of lit they
read) and what they acutally say... as you know

re 3: the construction should be accepted by all speakers, that might be
easier if you vary length of constituents and / or content!!! - which
applies for all sentences

for the below sentences, i accept 1,2,3,5, i reject 4 and 6, i think i
accept 4 in your examples because of the nie!

1.diese titel waren mir schon immer bekannt
2.schon immer waren mir diese titel bekannt
3.bekannt waren mir diese titel schon immer
4.diese titel bekannt waren mir schon immer
5.schon immer bekannt waren mir diese titel
6.schon immer diese titel bekannt waren mir

i remember a discussion i had with a class mate in grad school about the
acceptability of left dislocations in spoken american english that differ
apparently remarkably from what linguists have claimed to be possible...

sorry this is somewhat confused, i have never worked much with german and
whenever i think about it, i start feeling dizzy and then i think about how
much work in ling is based on only a few speakers of a language!!!

susanne
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Re: German partial-VP fronting

Date: Thu, 11 Apr 1996 17:45:08 +0200
From: Ulrich Koch <kochmailhost.uni-koblenz.de>
Subject: Re: German partial-VP fronting
All,

concerning Dick Hudson's request for judgements, here are mine:

(1) Wirkliche Fehler waren ihm in seinen Vortraegen nie unterlaufen.

OK.

(2) In seinen Vortraegen waren ihm wirkliche Fehler nie unterlaufen.

OK.

(3) Unterlaufen waren ihm wirkliche Fehler in seinen Vortraegen nie.

A bit funny, but OK. The "funniness" has semantical reasons: The topic
position of "unterlaufen" puts the focus on this word. A paraphrase
in English could be: "Real mistakes had never happened to him in his talks"
(where "happened" ("unterlaufen") is stressed). But what can mistakes
do other than happen to you? Maybe you can make them on purpose...
Then (3) would be completely OK.

(4) Wirkliche Fehler unterlaufen waren ihm in seinen Vortraegen nie.

OK, but colloquial.

(5) In seinen Vortraegen unterlaufen waren ihm wirkliche Fehler nie.

OK, but even more colloquial than (4).

(6) In seinen Vortraegen wirkliche Fehler unterlaufen waren ihm nie.

Ungrammatical.

Concerning the results, Dick wrote:

> It's not at all clear how to make more general sense of these data. Why
> should (4) be linked to (5) in judgement patterns? Rejection of (4) could be
> due to a more general ban on front-shifting of subjects (which are allowed by
> A speakers), but that wouldn't explain the rejection of (5), which contains a
> front-shifted adjunct.

I don't know the different analyses of partial VP fronting in detail,
but I see the following difference: In all of (1)-(3), the front-shifted
constituent is "simple", i.e. an NP, a PP or a V. This seems acceptable
(modulo semantic effects). In (4) and (5), a partial VP is fronted which
consists of the V and *one* constituent. Some speakers accept that, others
reject it. But in (6), the fronted partial VP contains *two* constituents
besides the V. That seems to be why many speakers accept (4) and (5), but
not (6). I suppose this explanation doesn't fit well into any existing
analysis...

I'm not sure why some speakers (including me) think (4) is better than (5).
Apart from the complement/adjunct distinction w.r.t. partial VPs, there
could be a semantic reason: The relationship between "wirkliche Fehler"
and "unterlaufen" in (4) is clear as soon as the speaker has heard only
those three words. The relationship between "in seinen Vortraegen" and
"unterlaufen" in (5) is less clear, so the semantic representation of
the whole VP can only be constructed when the speaker has heard the
whole sentence except for "nie".

I can't explain at all the X2 group -- I don't see why (5) should be
considered OK, but not (4).

Cheers,
Ulli

Ulrich Koch, comp. science and comp. ling. student
 http://www.uni-koblenz.de/~koch/ (also in English,
 anka^u esperante)
 "Tough" constructions are tough to analyze.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: Re: German partial-VP fronting

Date: Fri, 12 Apr 1996 11:14:04 BST
From: Ralf Steinberger <ralf.steinbergersharp.co.uk>
Subject: Re: German partial-VP fronting

Re: Subject: 7.530, Sum: German partial-VP fronting

Concerning Dick Hudson's message on German PVP-Fronting I can offer two more
native speaker's intuitions, as well as a possible explanation on why sentence
(4) is more widely accepted than sentence (5). 

 (1) Wirkliche Fehler waren ihm in seinen Vortraegen nie unterlaufen.
 ok: 36/36
 
 (2) In seinen Vortraegen waren ihm wirkliche Fehler nie unterlaufen.
 ok: 36/36
 
 (3) Unterlaufen waren ihm wirkliche Fehler in seinen Vortraegen nie.
 *: 3. ok: 23. ?: 10
 
 (4) Wirkliche Fehler unterlaufen waren ihm in seinen Vortraegen nie.
 *: 8. ok: 26. ?: 2.
 
 (5) In seinen Vortraegen unterlaufen waren ihm wirkliche Fehler nie.
 *: 23. ok: 8. ?: 5.
 
 (6) In seinen Vortraegen wirkliche Fehler unterlaufen waren ihm nie.
 *: 32. ok: 2. ?: 2.
 

Native Speakers' intuition:
- -------------------------

According to my own intuition 1 and 2 are ok, 3 to 6 are not (*).
Incidentally, I think that (3) is worse than (4).

According to my colleague Dagmar Dwehus' intuition, 1 and 2 are ok, three is
just about acceptable (?) and 4 to 6 are ungrammatical (*).

- -

I shall start off by trying to explain my (apparently unusual) intuition about
(3) and will then describe the difference between (4) and (5).

Concerning (3), I think that it is not the structure of the sentence which is
the problem but rather several other factors. According to my intuition the
following, similar, sentences (7), (8), and (9) are ok:

 7 GeSEhen hat er Iren in seinen Vortraegen nie.
 8 GeFREUT hat ihn der Besuch seiner Eltern vor Weihnachten nie.
 9 GeFOLGT hat ihm sein Hund seither nie.

A major difference between (3) on one side and (7) to (9) on the other is
that, in my opinion, 'unterlaufen' cannot reasonably be stressed as it is
semantically (almost) empty (as Ulrich Koch has pointed out in his previous
message). In order to explain the acceptability differences between the
sentences (1) to (9) I have to list some basic assumptions I have about word
order:

Factors on which native speakers' intuition may be based:
- -------------------------------------------------------

A) For semantic reasons, some words or phrases cannot be stressed,
 e.g. 'unterlaufen' in (3).

B) Degree modifying (or 'focusing') structures attract the focus of a sentence
 (e.g. 'wirkliche Fehler') and are thus likely to be stressed. Some
 modifiers also attract the focus (e.g. 'nie' in (1) to (9), 'gern', and most
 uninflected adjectives when modifying the verb, e.g. 'schoen', 'laut, etc
 as in 'Er singt schoen'). Furthermore, fronted verbs attract the focus.

C) In the Mittelfeld, thematic constituents have a strong tendency to precede
 rhematic and stressed ones. Unless modifiers are rhematic themselves, they
 typically precede rhematic constituents. 
 
D) Ideally a sentence has not more than one (strongly, or contrastively) stress
ed 
 word or phrase. The more phrases have to be stressed in a sentence the more 
odd 
 or ungrammatical it becomes, probably because it becomes harder to find a
 context for it.

Most word order (and other) regularities are preferences, or tendencies,
rather than hard rules. Different tendencies can prefer different orders. The
more preferences are being contradicted in a given sentence the more
ungrammatical the sentence is. Different preferences have a different weight,
i.e. they have more or less impact on acceptability decisions. There are quite
a few more preferences which are not relevant for the sentences we are
discussing here.

Sentence (3):
- -----------

 (3) Unterlaufen waren ihm wirkliche Fehler in seinen Vortraegen nie.
 *: 3. ok: 23. ?: 10
 
According to (B), 'nie' in (3) is likely to be stressed, and so are 'wirkliche
Fehler' and the past participle 'unterlaufen' in VF position. According to (D)
this cumulation of stressed elements in such a short sentence makes the
sentence very marked and difficult to place in any context.

In addition to this, according to (B) 'unterlaufen' should be stressed, but
according to (A) it cannot reasonably be stressed. (3) is therefore not a good
sentence with respect to (A).

Finally, (C) is violated in (3) because 'wirkliche Fehler' precedes the
modifier 'in seinen Vortraegen'. In order not to violate (C) we could stress
the modifier as well, but that would get us even more into trouble with
respect to (D).

I believe that these reasons are stronger than the one mentioned in Dick's
'notes' concerning 'splitting collocations':

 Notes:
 Re A: 12 of the 17 B speakers found (3) and (5) semantically odd, but
 this would be easy to explain as the result of splitting a collocation
 "Ein Fehler X unterlaufen", `a mistake happen to X' (i.e. X make a
 mistake).

The fact that collocations can be split without doing too much harm is shown
by (10), which seems much better than (3).

 (10) Ein FEHler ist ihm schon LANge nicht mehr unterlaufen. 

Probably one could formulate another (weaker?) 'preference' concerning
collocations:

E) The different parts of collocations tend to occur together. The stronger
 the collocation, the bigger is the effect on the acceptability of a
 sentence if the collocation is split up.

Sentences (4) and (5):
- --------------------

The preferences (A) to (F) explain why people prefer (4) to (5):

 (4) Wirkliche Fehler unterlaufen waren ihm in seinen Vortraegen nie.
 *: 8. ok: 26. ?: 2.
 
 (5) In seinen Vortraegen unterlaufen waren ihm wirkliche Fehler nie.
 *: 23. ok: 8. ?: 5.
 
According to (B) one of the VF elements should be stressed in both
sentences. In (4) there are (only) two stressed phrases ('wirkliche Fehler'
and 'nie'), whereas in (5) there are three ('in seinen VORtraegen', 'wirkliche
Fehler', and 'nie'). (4) is thus better with respect to (D) (cumulation of
stresses).

The collocation in (5) is split, which violates (E). It is not split in (4).

There may be another preference which, if it is true, makes both (4) and (5)
less good than (1) to (3):

F) Sentences without fronted verb are better than (more natural than/less
 marked than/preferred to) sentences with verb fronting. The more arguments
 and modifiers are fronted with the verb the less good the sentence is. 

And possibly there is another preference, which could be an extension of (E):

G) Fronting a verb with its argument is better than fronting a verb with
 modifiers.

According to (G) (4) would be preferred to (5) as well. 

(4) is thus rejected by less people than (5) because it violates less
preferences.

The Ideal Sentence (11):
- ----------------------

According to my intuition (11) has the least marked word order. All variations
of (11) cause stresses and restrict the sentence in its context:

 (11) (dass) ihm in seinen Vortraegen nie wirkliche FEHler unterlaufen sind. 

Conclusion:
- ---------

With respect to word order there are only few hard rules but many
preferences. The preferences interact and may prefer different orders. In
order to find out about the 'hard word order rules' (if there are any) the
preference violations should be kept the same so that the sentences are
comparable. The 'grammaticality', or better 'acceptability' of the sentences
(1) to (6) is probably just a question of grade (see the native speaker's
intuition gathered by Dick). Apparently people have a different threshold with
respect to what they allow. Spoken language is more liberal with respect to
word order, presumably because we make up the sentences while we speak. I
personally allow all the sentences (1) to (6) in the sense that I understand
them but I would *write* only (1) and (2), if even those.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue