On the nouniness of V2-clauses under preference predicates

Page created by Patrick Goodwin
 
CONTINUE READING
On the nouniness of V2-clauses under preference predicates
On the nouniness of
V2-clauses under preference predicates
Frank Sode – Goethe University Frankfurt
DGfS Arbeitgruppe 7 – February 26, 2021

                                           1 / 40
Topic of the talk

The topic of this talk are embedded V2-clauses in construction with
predicates expressing a desire or preference as illustrated in (1).

(1)   a.    Es wäre    gut, ich wäre  schon zu Hause.
            It be.SUBJ good I be.SUBJ already at home
            ‘It would be good if I was already at home.’
      b.    Ich wäre      froh, ich wäre       schon zu Hause.
            I be.SUBJ happy I be.SUBJ already at home
            ‘I would be happy if I was already at home.’
      c.    Ich { wollte     / wünschte }, ich wäre      schon zu Hause.
            I { want.SUBJ / wish           } I be.SUBJ already at home
            ‘I wish I was already at home.’

                                                                           2 / 40
Topic of the talk

I will pursue mainly two claims:

  • Empirical claim: These V2-clauses are a special kind of what Williams
     (1974) and Pesetsky (1991) call complement fulfilling conditionals.
  • Theoretical claim: These V2-clauses are the counterfactual
     counterpart of the nouny factive complements of the same predicates.

                                                                            3 / 40
What are “complement fulfilling cond’s”?

Williams (1974) was the first to notice that there is something special about
sentences like (2-a). At the surface, (2-a) looks like a regular conditional;
take for example (2-b).

(2)    a.    Mary would be glad if she won the lottery.
       b.    Mary would be rich if she won the lottery.

The reasons for Williams to doubt that (2-a) is a regular conditional is that
predicates like glad are factive predicates that require a complement.
They cannot be used predicatively or attributively like happy.

(3)    a.    I am { happy / *glad }.
       b.    I am a { happy / *glad } person.

(4)    Mary is glad that she won the lottery.   ; Mary won the lottery.

                                                                                4 / 40
Williams on “complement fulfilling cond’s”?

Intuitively, the idea is that glad introduces an additional argument slot that is
left open.

(5)    a.    [ Mary would be glad        ][ if she won the lottery ]
       b.    [ Mary would be rich ][ if she won the lottery ]

Williams assumption is that this argument slot is somehow filled by a
proposition made salient in the if-clause.

The resulting reading according to Williams assumptions is:

(6)    [ Mary would be glad that she won the lottery ][ if she won the lottery ]

                                                                                    5 / 40
Williams’ idea

(7)    [ Mary would be glad that she won the lottery ][ if she won the lottery ]

  • On Williams’ analysis the if -clause plays a double role: It not only
      specifies the conditions under which “gladness” holds.
  • It also fulfills the complement requirement of glad and thus specifies
      the subject matter (Pesetsky), or what the “gladness” is all about.
  • In that sense, the conditional is complement fulfilling (Pesetsky).
  • I follow Pesetsky in calling the corresponding conditionals
      “complement fulfilling conditionals” = CFCs.

                                                                                   6 / 40
Two classes of modal evaluatives

Williams points that CFCs distinguish two classes of modal evaluatives.

                               Class 1                     Class 2
                       unlikely, convinced, . . .   good, glad, happy, . . .

      that-clauses           non-factive                    factive
       allow CFCs                  *                          X

(8)   a. *It would be unlikely if Bob left.              (Williams, 1974, p. 158)
      b. *I would be convinced if Bob left.              (Williams, 1974, p. 158)

(9)   a.    It would be good if Bob left.
      b.    I would be happy if Bob left.

                                                                                    7 / 40
Why Williams’ analysis is beautiful

• Williams analysis explains the outer appearance of a conditional (It is a
  conditional!)
• It allows to keep the standard analysis of glad as an (emotive) factive
  predicate that takes a proposition as its argument and presupposes its
  truth in the world of evaluation.
• It explains the vanishing of factivity as a textbook case of filtering in
  conditionals.
• The only thing needed is a story for the copying of the proposition
  expressed in the conditional clause.
• The details are provided by Pesetsky (1991)’s If Copying Rule.

                                                                              8 / 40
Pesetsky’s If Copying Rule

Following Williams (1974), Pesetsky assumes that the sentence in (10-a) has
an interpretation in the sense of (10-b).

(10)    a.   John would like it if Mary knew French.   →   [If Copying Rule]
        b.   John would like it that Mary knows French if Mary knew French.

                                                                               9 / 40
Pesetsky’s If Copying Rule

This interpretation comes about by applying the following mechanism:

(11)   If Copying Rule                                                     Pesetsky (1991)
       1.    Take a clause k of the form [IF IP] where k modifies a sentence Σ.
       2.    Copy k as k’ substituting that for IF, making appropriate changes in mood so as
             to replace irrealis with realis mood marking.
       3.    Place k’ in an argument position of Σ. Leave k as an adjunct modifer. (It gets
             interpreted as a restrictive clause, with Σ the nuclear scope.)
       4.    k’ is factive.

(12)   a.    [Σ John would like it ][ IF [IP Mary knew French ]]
       b.    k’ = [ that [IP Mary knows French ]]
       c.    [Σ John would like it ][ IF [IP Mary knew French ]]
                 z              }|             {
                 [ that [IP Mary knows French ]]

                                                                                               10 / 40
Two ways to arrive at Williams’ reading

As Pesetsky points out: There are now two ways that we can arrive at an
interpretation in the sense of Williams:
Way 1: with a fronted if-clause via a referential pronoun: (unexciting)

(13)    If Mary knew French, John would like iti .
        for g (i ) = Jthat Mary knows FrenchK

This pattern can be found with a wide range of factive predicates: see for
example Schueler (2013); Hinterwimmer (2014); Schwabe (2016).

Way 2: via the IC Rule (restricted to a certain class of factive predicates)

(14)    John would like it if Mary knew French.

              [Σ John would like it ][ IF [IP Mary knew French ]]
                   z             }|               {
                   [ that [IP Mary knows French ]]

Pesetsky only calls the second kind of conditionals CFCs.
                                                                               11 / 40
Evidence for two ways / Pesetsky’s IC Rule

Pesetsky argues that we actually need something like the If Copying Rule to
account for contrasts as the following:

While the referential construction allows for the proposition to be picked up
by referential that, (15-a), CFCs exlude referential that, (15-b).

(15)    a.   If unicorns existedi , I would love thati .   (Pesetsky, 1991, p. 74)
        b. *I would love thati if unicorns existedi .      (Pesetsky, 1991, p. 74)

His explanation: The If Copying Rule doesn’t allow to overwrite truly
referential pronouns like that.

(Pesetsky disccusses more differences.)

                                                                                     12 / 40
Extension to desire reports

The If Copying Rule not only plays a role in Pesetsky’s account of CFCs.
It also plays a crucial role in Pesetsky’s analysis of desire reports.

(16)    Pesetsky (1991)’s Post-LF analysis of want (+ ∅for -complement)
        John wants [ ∅for Mary to know French ] →
        John would like [ ∅for Mary to know French ] →          [If Copying Rule]
        John would like (it) that Mary knows French, IF Mary knew French.

For Pesetsky, ∅for -clauses as the complements of desire verbs like want are
hidden CFCs that combine with a hidden factive preference predicate.

                                                                                    13 / 40
Complement fulfilling conditionals

“Williams-Pesetsky-Class”: the class of predicates that license CFCs.

(17)   The Williams-Pesetsky-Class of predicates
       a.   Prioriety-oriented modal evaluatives: good, glad, happy, . . .
       b.   Desire verbs: want, wish, like, love, . . .

                                                                             14 / 40
Embedded V2 in German

                        15 / 40
Embedded V2-complements

Two basic assumptions in the literature
  1   Only predicates with a non-negative “assertive” meaning license
      V2-complements.

      (18)    a.   Die Maria { glaubt / sagt }, man hört         ihr zu.
                   the Maria { believes / says } one listens her to
                   ‘Maria { believes / says } that people listen to her.’
              b.   *Die Maria { fordert      / beweifelt }, man hört    ihr zu.
                   the Maria { demands / doubts         } one listens her to
                   ‘Maria { demands / doubts } that people listen to her.’

  2   Embedded V2-clauses are root phenomena.

Helbig & Kempter (1974); Reis (1997); Gärtner (2002); Truckenbrodt (2006)
                                       ...

                                                                                  16 / 40
Puzzling cases

A certain class of non-assertive predicates, often characterized as
“preference predicates” or “volitional predicates”, seem to license what looks
like V2-complements.

(19)    a.   Es wäre        gut, ich wäre      schon zu Hause.
             It be.SUBJ good I      be.SUBJ already at home
             ‘It would be good if I was already at home.’
        b.   Ich wäre       froh, ich wäre         schon zu Hause.
             I   be.SUBJ happy I      be.SUBJ already at home
             ‘I would be happy if I was already at home.’
        c.   Ich { wollte      / wünschte }, ich wäre       schon zu Hause.
             I   { want.SUBJ / wish          } I    be.SUBJ already at home
             ‘I wish I was already at home.’

   Helbig & Kempter (1974); Reis (1997); Frank (1998); Meinunger (2007)

                                                                                 17 / 40
Puzzling cases

A first crucial observation is that the corresponding V2-clauses are licensed
only in subjunctive mood but not in indicative mood.*

(20)    a.   *Es ist       gut, ich bin    schon zu Hause.
             It   be.IND good I    be.IND already at home
        b.   *Ich bin      froh, ich bin    schon zu Hause.
             I    be.IND happy I     be.IND already at home
        c.   *Ich will, ich bin     schon zu Hause.
             I    want I    be.IND already at home

 * for more discussion and exceptions see: Frank (1998); Meinunger (2007)

                                                                                18 / 40
The connection to CFCs

                         19 / 40
The connection to CFCs

First, all puzzling cases seem to belong in the Williams-Pesetsky-Class:

(21)    The Williams-Pesetsky-Class of predicates
        a.   Prioriety-oriented modal evaluatives: good, glad, happy, . . .
        b.   Desire verbs: want, wish, like, love, . . .

                                                                              20 / 40
The connection to CFCs

Second, these embedded V2-clauses are more naturally paraphrased with
wenn (‘if’)-clauses in subjunctive mood than with dass (‘that’)-clauses, cf.
Frank (1998); Meinunger (2007).

(22)    Es wäre       gut, ich wäre          schon zu Hause.
        It be.SUBJ good I       be.SUBJ already at home
        ‘It would be good if I was already at home.’

                            can be paraphrased as

(23)    Es wäre       gut, { wenn / ??/*dass } ich schon zu Hause
        It be.SUBJ good { if        / that       }I    already at home
        wäre.
        be.SUBJ
        ‘It would be good if I was already at home.’

                                                                               21 / 40
The connection to CFCs

“CondV2-clauses”: V2-clauses as “complements” of the predicates in the
Williams- Pesetsky-Class.

                    CondV2-clauses are a special kind of
                    complement fulfilling conditionals.

(Similar to Pesetsky’s ∅for -complements.)

On the following slides, I present evidence in favour of this claim.

                                                                         22 / 40
CondV2-clauses are CFCs

Evidence 1: Although CondV2-clauses can be paraphrased by wenn
(‘if’)-clauses; they don’t allow to be interpreted as “true” conditionals. While
the V2-clause in (24) (that can be paraphrase with a conditional clause) is
fine with a predicate from the Williams-Pesetsky-Class . . .

(24)    Der Peter wäre        froh, der Hans würde         seine Schulden
        the Peter be.SUBJ happy the Hans will.SUBJ his            debts
        bei ihm begleichen.
        with him settle
        ‘Peter would be happy if Hans paid back his debts.’                   V2

                                                                                   23 / 40
CondV2-clauses are CFCs

Evidence 1: . . . a “true” conditional interpretation is not possible.

(25)    *Der Peter wäre        reich, der Hans würde        seine Schulden
        the Peter be.SUBJ rich        the Hans will.SUBJ his       debts
        bei ihm begleichen.
        with him settle
        intended: ‘Peter would be rich if Hans paid back his debts.’         V2

This is different from V1-clauses that do allow “true” conditional
interpretations; compare (26).

(26)    Der Peter wäre        reich, würde      der Hans seine Schulden
        the Peter be.SUBJ rich       will.SUBJ the Hans his       debts
        bei ihm begleichen.
        with him settle
        ‘Peter would be rich if Hans paid back his debts.’                   V1
                                                                                  24 / 40
CondV2-clauses are CFCs

Evidence 2: CondV2-clauses don’t allow referential pronouns.
(more discussion ⇒ Appendix)

(27)   a.   Ich fände        es gut, es würde           mal wieder regnen.
            I   find.SUBJ it good it will.SUBJ once again rain
            ‘I would like it if it finally rained again.’
       b.   *Ich fände        das gut, es würde             mal wieder regnen.
            I    find.SUBJ that good it will.SUBJ once again rain
            intended: ‘I would like it if it finally rained again.’

                                                                                 25 / 40
CondV2-clauses are CFCs

Evidence 3: CondV2-clauses only allow desire object interpretations. This is
similar to an observation made by Grosz (2012) for English CFCs.

(28)    a.   It’s not the case that I want John to be here, but if he were here,
             it would be nice that he was here.
        b. #It’s not the case that I want John to be here, but it would be nice
             if he was here.                                   cf. Grosz (2012)
(29)    Ich will zwar nicht, dass der Hans kommt, aber . . .
        ‘I don’t want Hans to come but . . . ’
        a.   #es wäre       schön, er wäre           hier.
             it   be.SUBJ nice       he be.SUBJ here
             #‘it would be nice if he were here.’                            V2
        b.   es wäre       schön, wäre           er hier.
             it be.SUBJ nice        be.SUBJ he here
             ‘. . . if he were here, it would be nice.’                      V1
                                                                                   26 / 40
CondV2-clauses are CFCs

Evidence 3: They only allow desire object interpretations.

The interpretation of wünschte is in a conceptual conflict with the modal
sollte (‘should’) on a deontic interpretation.

(30)    Ich wünschte, der Hans { würde           / #sollte } seine Schulden
        I    wish.SUBJ the Hans { will.SUBJ / should } his       debts
        bei mir begleichen.
        with me settle
        ‘I wish Hans paid back the money he owes me.’

                                                                              27 / 40
CondV2-clauses are CFCs

Evidence 3: They only allow desire object interpretations.

The use of sollte (‘should’) in the V2-clause in (31-a) has the same feeling to
it. Sollte in a V1-clause is possible; but the interpretation of sollte flips to the
deliberative interpretation of sollte found in conditionals.

(31)     a.   #Ich wäre       froh, der Hans sollte seine Schulden bei
              I    be.SUBJ happy the Hans should his            debts       with
              mir begleichen.
              me settle                                                            V2
         b.   Ich wäre       froh, sollte der Hans seine Schulden bei mir
              I   be.SUBJ happy should the Hans his            debts       with me
              begleichen.
              settle
              ‘If Hans should pay back the money he owes me, I would be
              happy.’                                                              V1
                                                                                        28 / 40
Summary

• The V2-clauses under discussion only show up in the
  Williams-Pesetsky-Class that license CFCs.
• They can be paraphrased by wenn (‘if’)-clauses; not by dass
  (‘that’)-clauses.
• They don’t allow to be picked up by referential pronouns.
• They only allow for desire object interpretations similar to CFCs in
  English (cf. Grosz (2012)).

                 CondV2-clauses are a special kind of
                 complement fulfilling conditionals.

                                                                         29 / 40
Nouniness?

             30 / 40
What’s the nouniness?

Desire reports are often analysed along the lines of Heim (1992) assuming a
structure as follows:

(32)    Mary is gladFACTIVE [CP that ϕ ]st                 standard assumption

We can arrive at the same truth conditions with the following structure that
assumes a different distribution of the elements provided by Heim’s analysis
at the syntax-semantics interface; Sode (2020):
                                                               s(st )t
(33)    [ Mary is gladNEUTRAL PROi ][ @FACTIVE [CP that ϕ ]st ]i

PRO fills an argument slot of the semantic type s(st )t.

                                                                                 31 / 40
What’s the nouniness?

                                                                  s(st )t
(34)      [ Mary is gladNEUTRAL PROi ][ @FACTIVE [CP that ϕ ]st ]i

(34) ≈ ‘Mary desires what is the case (according to her beliefs) since that ϕ’

This proposal is (a) closer in spirit to what Heim literally (=by way of her
paraphrases) proposes, which can be summarized by the following slogan:

       “There is a hidden ‘because’-clause in every factive desire report.”
                     (‘because’-clause ≈ factive conditional)

(b) closer in spirit to ideas found in the syntactic literature on factive
complements (going back at least to Kiparsky & Kiparsky (1970)):
The assumption is that factive complements are
 (a) nouny (which shows up in their semantic type:
       hs, hhs, t i, t ii ≈ type of intensional objects);
 (b) involve a hidden factive operator (here: more determiner than noun).
                                                                                 32 / 40
Advantages of the analysis

The semantic type of glad is now such that it can directly combine with an
if -clause in intension (or more precisely: an if -restricted operator).

For the counterfactual case, we can simply assume that be glad combines
with would if ϕ as its semantic argument.

                                                                          s(st )t
(35)    a.   [ Mary is gladNEUTRAL PROi ][ @FACTIVE [CP that ϕ ]st ]i
                                                                    s(st )t
        b.   [ Mary be gladNEUTRAL PROi ][ would [CP if ϕ      ]st ]i

(35-b) ≈ ‘Mary desires what would be the case (according to her beliefs) if ϕ’

would is the counterfactual counterpart of the factive operator @.

(I assume this analysis can be extended to other predicates in the
Williams-Pesetsky-Class.)

For more arguments for this analysis see Sode (2020). Additional comments:
⇒ appendix.
                                                                                    33 / 40
Analysis

In analogy, we can think of CondV2-clauses as restricting a hidden
conditional operator WOULD (that I assume to be also present in regular
counterfactual conditionals of German).
                                                                     s(st )t
(36)    a.   [ ich PROi froh wäre ][ WOULD [CP er wäre hier ]st ]i
                                                                s(st )t
        b.   [ esPROi gut wäre ][ WOULD [CP er wäre hier ]st ]i            CondV2

   Analysis: CondV2-clauses are a special kind of IF-clause. They can
   only restrict a conditional operator that fills an argument slot of a pred-
   icate in the Williams-Pesetsky-Class.

(Similar to Pesetsky’s ∅for -complements.)

                                                                                    34 / 40
Conclusion

• Empirically: CondV2-clauses have the distribution and interpretation of
  complement fulfilling conditionals.
• Analysis: CondV2-clauses are semantically interpreted as IF-clauses.
  Their use is restricted: They can only restrict operators in a thematic
  argument position of predicates in the Williams-Pesetsky-Class.
• These arguments share the nouniness of the corresponding factive
  clauses.
• Confirms the suspicion expressed in Frank (1998); Meinunger (2007)
  that V2-clauses under preference predicates are of a different kind than
  V2-complements under assertive predicates.
• That is: Not all embedded V2-clauses are root phenomena.

                                                                             35 / 40
Literature I

Frank, Nicola. 1998. Präferenzprädikate und abhängige Verbzweitsätze.
   Magisterarbeit Universität Tübingen. Arbeitspapiere des SFB 340. Bericht Nr.
   128.
Grosz, Patrick Georg. 2012. On the grammar of optative constructions. Amsterdam:
  Benjamins.
Heim, Irene. 1992. Presupposition Projection and the Semantics of Attitude Verbs.
  Journal of Semantics 9.3. 183–221.
Helbig, G. & F. Kempter. 1974. Die uneingeleiteten Nebensätze im Deutschen und
  ihre Vermittlung im Fremdsprachenunterricht. DAF 11. 75–86.
Hinterwimmer, Stefan. 2014. If vs. when, wenn vs. als: Microvariation in the
   Semantics of Conditional and Temporal Complementizers in English and
   German.
Kaufmann, Magda. 2017. Japanese modals and other evaluations. MIT, Seminar
  “The Linguistics of Desire”, November 29.

                                                                                    36 / 40
Literature II

Kiparsky, Paul & Carol Kiparsky. 1970. Fact. In M. Bierwisch & K.E. Heidolph (eds.),
   Progress in linguistics, 143–173. Mouton.
Longenbaugh, Nicholas. 2019. On expletives and the agreement-movement
  correlation: Department of Linguistics and Philosophy. MIT dissertation.
Meinunger, André. 2007. In the mood of desire and hope. In Tense, Mood and
  Aspect, 155–176. Amsterdam, New York: Rodopi.
Pesetsky, D. 1991. Zero syntax. Vol. 2: Infinitives. Manuscript.
Reis, Marga. 1997. Zum syntaktischen Status unselbständiger Verbzweit-Sätze. In
  C. Dürscheid, K. H. Ramers & M. Schwarz (eds.), Sprache im Fokus, 121–144.
  Tübingen: Niemeyer.
Schueler, David. 2013. Factivity and presuppositions. University of Minnesota, Twin
  Cities, LSA Annual Meeting 2013.
Schwabe, Kerstin. 2016. Sentential proforms and argument conditionals 211–240.
  John Benjamins Publishing Company.
Williams, Edwin Samuel. 1974. Rule ordering in syntax. Cambrigde, MA:
   Massachusetts Institute of Technology dissertation.
                                                                                       37 / 40
CondV2-clauses are CFCs

Evidence 2: CondV2-clauses don’t allow referential pronouns.

(37)   Denkst du, es wäre gut, die Klausur mitzuschreiben?
       ‘Do you think that it would be good to write the exam?’
       a.   Naja, das wäre       gut, wenn wir was dafür getan
            Well it      be.SUBJ good if      we what for.that done
            hätten.
            have.SUBJ
            ‘Well, it would be good if we had studied for it.’
       b.   *Naja, das wäre       gut, wir hätten         was dafür getan.
            Well   that be.SUBJ good we have.SUBJ what for.that done
       c.   #Naja, es wäre         gut, wir hätten        was dafür getan.
            Well      that be.SUBJ good we have.SUBJ what for.that done
            #‘Well, it would be good to have studied for it.’
                                                          in the sense of (37-a)
                                                                                   38 / 40
Advantages of the analysis

There is no need for Pesetsky’s If Copying Rule anymore. The assumption of
such a rule was only necessary given the assumption that the predicates in
the Williams-Pesetsky-Class on a factive interpretation take arguments of a
propositional type.

The crucial difference in interpretation to the Williams-Pesetsky-Account of
CFCs is that the conditional clause is only used to characterize the object of
desire on this account (in agreement with Grosz (2012)’s observation) and
not to characterize the counterfactual situations under which happiness will
be achieved (as predicted by Williams’ paraphrase).

This is a good result since it has recently been argued that Williams’
paraphrase doesn’t capture the right truth conditions: Grosz (2012);
Kaufmann (2017); Longenbaugh (2019); Sode (2020).

                                                                                 39 / 40
Factive predicates

(38)   a.   Ich wäre froh, wenn ich schon zu Hause wäre.
       b.   Ich wäre froh, ich wäre schon zu Hause.
(39)   a.   Ich wäre überrascht, wenn ich schon zu Hause wäre.
       b. *Ich wäre überrascht, ich wäre schon zu Hause.
(40)   a.   Ich würde (es) wissen, wenn ich schon zu Hause wäre.
       b. *Ich würde (es) wissen, ich wäre schon zu Hause.
(41)   a.   Ich würde (es) bedauern, wenn ich schon zu Hause wäre.
       b. *Ich würde (es) bedauern, ich wäre schon zu Hause.

                                                                     40 / 40
You can also read