Patrick G. Grosz

Dissertation Defense

June 30, 2011

ON THE GRAMMAR OF OPTATIVE CONSTRUCTIONS Patrick G. Grosz Oh that Apollo would but drive his horses slowly, that the day might be three hours longer; for it is too soon to depart, and that for fear of a pocky setting of the Watch. (A. Marsh. 1682. The Ten Pleasures of Marriage. London: The Navarre Society.)

Roadmap  Overview of my proposal (⇒ Section 1)  The source of desirability in optatives (⇒ Section 2)  The semantics of prototypical particles (⇒ Section 3)  The role of mood in exclamations (⇒ Section 4)  Conclusion (⇒ Section 5)

1. Overview 1.1

Empirical Scope

In brief: In this section, I survey the empirical scope of this dissertation.  Optative utterances: Utterances that express a wish, regret, hope or desire without containing a lexical item that means wish, regret, hope or desire (cf. Quirk et al. 1972, 1985, Scholz 1991, Rosengren 1993, Rifkin 2000, Asarina & Shklovsky 2008, Biezma 2011ab). ⇒ – – –

Optatives can be described in terms of three factors: Material in C (if, that, finite verb/auxiliary) Grammatical mood (counterfactual/subjunctive vs. non-counterfactual/indicative) Particles (e.g. only, at least, doch)

-1-

Patrick G. Grosz

(1) a. b. c.

d. (2)

Dissertation Defense

June 30, 2011

subjunctive (and counterfactual) optatives Daß er nur rechtzeitig gekommen wäre! that he only in.time come were Wenn er nur rechtzeitig gekommen wäre! if he only in.time come were Wäre er nur rechtzeitig gekommen twäre! were he only in.time come ‘If only he had come in time!’ paraphrase: I wish [he had come in time].

that-variant if-variant V1-variant

indicative optatives a. Daß er nur rechtzeitig gekommen ist! that he only in.time comes b. Wenn er nur rechtzeitig gekommen ist! if he only in.time comes ‘If only he comes in time!’ c. paraphrase: I hope [he has come in time].

(3) a. b. c. d. e. f. g. h.

⇒

possible particles Ach, wäre ich Wäre ich doch Wäre ich nur Wäre ich wenigstens (Ach,) wäre ich doch nur (Ach,) wäre ich nur wenigstens (Ach,) wäre ich doch wenigstens Ach, wäre ich doch nur wenigstens oh were I doch only at.least ‘(Oh,) if only I were (at least) rich!’

that-variant if-variant

reich reich reich reich reich reich reich reich rich

twäre! twäre! twäre! twäre! twäre! twäre! twäre! twäre!

Part of my proposal is that optatives should be connected to another type of exclamation, so-called polar exclamatives.

 Polar exclamatives: Utterances that express surprise, shock or amazement at a fact (not at the degree to which something holds), without a lexical item that means surprise, shock or amazement (cf. Quirk et al. 1985, Scholz 1991, Rosengren 1992, Delsing 2010). -2-

Patrick G. Grosz

(4)

Dissertation Defense

June 30, 2011

1

indicative polar exclamatives a. Dass Sie (doch / tatsächlich / wirklich) daran gedacht haben! that-variant that you doch indeed really of.it thought have ‘[It’s remarkable] that you really remembered it!’ b. Haben Sie doch (tatsächlich / wirklich) daran gedacht thaben! V1-variant have you doch indeed really of.it thought ‘[It’s remarkable] that you really thought of it!’ (based on Scholz 1991:132-133, attributing (4a) to Wilhelm Oppenrieder)

Now: The connection between the two exclamation types and the puzzles that arise.  How are these two construction types connected? – both polar exclamatives and optatives appear to be exclamations – they are typically used to exclaim (cf. Rosengren 1992, 1993; see Quirk et al. 1972, 1985, Rifkin 2000, Kyriakaki 2007, 2008, 2009 for optatives). ⇒ For instance Quirk et al. (1985:841) propose that both utterance types have the “illocutionary force of exclamations”, optatives being exclamatory wishes. – both construction types have that-variants and V1-variants, i.e. similar form types. – both constructions appear to involve a comparison between the expressed proposition and its polar opposite.  The puzzles that these constructions raise: ⇒ How does their meaning come about? ⇒ What do the prototypical particles contribute? ⇒ How do we account for their variant shapes and forms? 1.2

The analysis in a nutshell

In brief: I summarize my system, which I discuss in more detail in the later sections.  What are the ingredients?

1

English marginally allows for polar exclamatives, but a modal should or could is typically required. i. That he should have left without asking me! ii. That you could ever want to marry such a man! (Quirk et al. 1985:841) -3-

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 I argue for three core ingredients: – An expressive operator EX: An EX-utterance is felicitous if the expressed proposition exceeds a salient threshold on a contextually salient scale. (EX is loosely inspired by Gutíerrez Rexach’s 1996 EXC.)

– –

Particles (such as nur ‘only’) as non-truth-conditional propositional modifiers. Split realization of semantic mood (e.g. counterfactuality, factivity), which gives rise to the different form variants2. (This part of the analysis is inspired by Truckenbrodt 2006ab, Portner 2006.)

 The following example illustrates all three ingredients. (5)

a. Wenn es nur geregnet if it only rained ‘If only it had rained!’ b.

hätte! had

CP eo EXpreferences CP qp C[Mood] … MoodP wenn qp TP Mood[Mood:CF] ri VP T 6 es nur geregnet hätte

Mood-C agreement

2. The Source of Desirability in Optatives 2.1

Introducing the EX operator

In brief: I introduce the EX operator. EX contributes the basic meaning of exclamations.  What does an EX-utterance convey?

2

I do not assume a direct correspondence between semantic mood and morphological mood; it is orthogonal to my analysis whether morphological mood employs past tense or a specialized subjunctive marker. -4-

Patrick G. Grosz

(6)

Dissertation Defense

June 30, 2011

An utterance of EX(ϕ) conveys: i. the speaker at the point of utterance has an emotion ε (or at least an evaluative attitude ε) towards ϕ. The speaker intends to express ε, rather than describe ε. ii. ε involves a scale S on which ϕ exceeds a salient threshold (THRESHOLD(c)). ⇒

I propose that EX only combines with scales that are anchored to the speaker and are evaluative/emotive. I argue for such a restriction below.

 Illustration of the core idea: ⇒ We can construct surface minimal pairs that are ambiguous between an optative reading and a polar exclamative reading, shown in (7). (7)

a. Mein Gott, dass der Otto nicht verschlafen hat! my God that the Otto not overslept has lit. My God, that Otto didn’t oversleep! b. paraphrase of optative reading: I hope [that Otto didn’t oversleep]. c. paraphrase of polar excl. reading: I’m shocked [that Otto didn’t oversleep]. ⇒

(8)

The EX-Operator analysis allows for a uniform analysis where the difference between the two readings is linked to the contextually provided scale, cf. (8).

a. EXS (Mein Gott,) dass der Otto nicht verschlafen my God that the Otto not overslept lit. My God, that Otto didn’t oversleep! b. optative reading: [S → speaker’s preferences]

hat! has

⇒ circumstances in which Otto didn’t oversleep count as tolerable/preferable. c. polar exclamative reading: [S → prior unlikelihood according to the speaker] ⇒ circumstances in which Otto didn’t oversleep count as remarkable.  What scales can EX combine with?

-5-

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 I argue that the scales must be connected to speaker emotion. ⇒ (9) shows that EX-utterances can express different types of evaluation / emotion, but they cannot be non-emotive. (mei ‘my’ is a South German interjection.) (9)

a.

(Mei,) Dass der Otto gegangen ist! my that the Otto left is lit. That Otto left! b.  polar exclamative reading: I am surprised [that Otto left]. cf. Dass der Otto doch glatt gegangen ist!

c.  optative reading:

⇒

I hope [that Otto left].)

cf. Oh, dass der Otto nur gleich gegangen ist! ⇒

d.  adversative reading:

2.2

particles disambiguate.

I am disappointed [that Otto left].)

cf. Dass der Otto aber auch gegangen ist!

e. * unattested reading:

particles disambiguate.

⇒

particles disambiguate.

# I am convinced [that Otto left].

Motivating the EX operator

In brief: I provide conceptual motivation and a first argument for the EX operator.  What are the possible sources of the expressive force of an exclamation?3 – The EX-Op view: The denoted proposition is the argument of an expressive operator EX. (A similar, but more specialized operator is proposed by Kyriakaki 2007, 2008, 2009.) – The matrix clause deletion view (Evans 2007): There is an elided matrix clause, which embeds the utterance that is pronounced. ⇒

The matrix clause deletion view is very intuitive and rather simple; it is also appealing. I thus review it now and provide arguments against it.

Now: I argue that a matrix clause deletion approach is not feasible (in spite of its appeal).  Consider first an illustration of the matrix clause deletion approach. ⇒ Example (10a) (from Evans 2007:373) is perceived to be an if-optative without an optative particle. Example (10b) is a corresponding that-optative. 3

See appendix 3 for Biezma’s (2011ab) proposal on if-optatives. As shown in appendix 3, some of the arguments against the matrix clause deletion view also prove problematic for Biezma (2011ab). -6-

Patrick G. Grosz

Dissertation Defense

June 30, 2011

(10) a. Wenn ich deine Statur hätte! if I your build had ‘[Oh!] If [only] I had your build!’ b. Dass ich deine Statur hätte! that I your build had ‘[Oh!] That I had [but] your build!’ ⇒

Straightforwardly, the matrix clause deletion account posits LFs as in (11). It is easy to see how such an analysis appears to solve the optative problem.

(11) a. [Es wäre schön,] Wenn ich deine Statur hätte! it were lovely if I your build had ‘[It would be lovely] if I had your build.’ b. [Ich wünschte,] Dass ich deine Statur hätte! I wish that I your build had ‘[I wish] I had your build.’  The core argument against matrix clause deletion4: Consider first the baseline example of an embedded if-clause in German, modified by a verb-final because-clause. (Verb-second variants, like English because-clauses, distribute more freely and are thus irrelevant for our investigation.) (12)

4

a. Die Party wäre ein Erfolg gewesen, wenn Hans gekommen wäre, the party were a success been if Hans come were weil er immer guten Wein mitbringt. because he always good wine brings ‘The party would have been a success if Hans had come, because he always brings good wine.’  because he always brings good wine > [the party would have been a success if Hans had come] b. Es wäre besser gewesen, wenn Hans gekommen wäre, weil it were better been if Hans come were because er immer guten Wein mitbringt. he always good wine brings ‘It would have been better if Hans had come, because he always brings good wine.’  because he always brings good wine > [it would have been better if Hans had come]

See appendix 1 for two further arguments. -7-

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 The crucial observation is that answer fragments behave exactly on a par with the baseline examples. The because-clause can take scope over the elided matrix clause. (13)

a. A: Unter welchen Umständen wäre die Party ein Erfolg gewesen? under which circumstances were the party a success been ‘Under which circumstances would the party have been a success?’ B: Wenn Hans gekommen wäre weil er immer guten Wein if Hans come were since he always good wine mitbringt. brings ‘If Hans had come, because he always brings good wine.’  because he always brings good wine > [The party would have been a success if Hans had come] LF: [The party would have been a success if Hans had come], because he always brings good wine. b. A: Was wäre besser gewesen? what were better been ‘What would have been better?’ B: Wenn Hans gekommen wäre weil er immer guten Wein if Hans come were since he always good wine mitbringt. brings ‘If Hans had come, because he always brings good wine.’  because he always brings good wine > [it would have been better if Hans had come] LF: [It would have been better if Hans had come], because he always brings good wine.

 In contrast, a verb-final because-clause in optatives can never take scope over the positive evaluation component, indicating that there is no elided matrix clause. (14)

* * *

Wenn Hans doch nur gekommen wäre (#weil er immer if Hans DOCH only come were because he always guten Wein mitbringt). good wine brings ‘If only Hans had come (#because he always brings good wine).’ because he always brings good wine > [I would be happy if Hans had come] because he always brings good wine > [it would be good if Hans had come] because he always brings good wine > [all would be well if Hans had come] -8-

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 Summary: Elided matrix clauses (in answer fragments) can be in the scope of verbfinal because-clause. The purported matrix clause in an optative cannot. ⇒ This argues against matrix clause deletion in optatives.  The same argument can be made for that-optatives. (15)

Ich wünschte, dass Hans gekommen wäre, weil er I wished that Hans come were because he guten Wein mitbringt. good wine brings ‘I wish Hans had come, because he always brings good wine.  because he always brings good wine > [I wish Hans had come]

immer always

(16)

A: Was würdest du dir wünschen? what would you you wish ‘What would you wish for?’ B: Dass Hans gekommen wäre, weil er immer guten Wein mitbringt. that Hans come were because he always good wine brings ‘That Hans had come, because he always brings good wine.’  because he always brings good wine > [I wish Hans had come] LF: [I wish that Hans had come], because he always brings good wine.

(17)

Dass Hans doch nur gekommen wäre, (# weil er immer guten that Hans DOCH only come were because he always good Wein mitbringt. wine brings ‘That Hans had but come, (#because he always brings good wine).’ * because he always brings good wine > [I wish Hans had come]  I wish > because he always brings good wine > [Hans had come]

2.3

EX as an expressive operator

In brief: I present two arguments for an expressive and evaluative null operator: EX.  I argue that optatives (and polar exclamatives) only have expressive (felicity conditional) meaning and do not have descriptive (truth conditional) meaning. -9-

Patrick G. Grosz

(18)

a. b.

⇒

June 30, 2011

If only it rained! EXS(rain) : E qo rain : EXS : In this sense they compare to Potts & Roeper’s (2006) expressive small clauses.

⇒ (19)

Dissertation Defense

a. b.

to myself, after making a mistake: You fool! fool(you) : E qo you : e fool: (Potts & Roeper 2006) Argument 1 for expressivity: A hearer cannot deny or affirm the meaning of an optative (see Rifkin 2000).

(20)

A: Es ist / wäre besser, wenn Otto kommt. it is were better if Otto comes ‘It is better if Otto comes!’ B:  Das stimmt nicht. /  Das stimmt. /  Finde ich auch/nicht. that is.true not that is.true find I also/not ‘That’s false.’ ‘That’s true.’ ‘I agree / disagree.’

(21)

A: Wenn Otto nur kommt! if Otto only comes ‘If only Otto is coming!’ B: # Das stimmt nicht. / # Das stimmt. / that is.true not that is.true ‘That’s false.’ ‘That’s true.’

(22)

5

# Finde ich auch/nicht. find I also/not ‘I agree / disagree5.’

A: Was ist / wäre besser? – B: Wenn Otto kommt. what is were better if Otto comes ‘What’s better? / What would be better?’ ‘If Otto comes.’ C:  Das stimmt nicht. /  Das stimmt. /  Finde ich auch/nicht. that is.true not that is.true find I also/not ‘That’s false.’ ‘That’s true.’ ‘I agree / disagree.’

Remarkably, English speakers accept I agree in such a context, but they can also use it in response to a non-verbal expression, e.g. a frown, making I agree a rather flexible expression. (Biezma 2011b) - 10 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Argument 2 for expressivity: Unembeddability.

⇒ –

Unembeddability is a hallmark of expressive content.

(23) a. towards myself: You fool! b.* I consider [you fool] / [me fool]. (cf. Potts & Roeper 2006) (24) a. (What,) Me worry? b.* I wonder / doubt [me worry]. (cf. Potts & Roeper 2006) –

(25)

Scholz (1991) posits the unembeddability generalization that optatives cannot be truly integrated; Rifkin (2000) shows that they cannot be embedded either.

a. Wenn ich reich wäre, hätte ich einen if I rich were had I a ‘If I was rich, I would have a Porsche.’ b. * Wenn ich doch nur reich wäre, hätte ich if I doch only rich were had I ‘If only I was rich, I would have a Porsche6.’

Porsche. Porsche einen Porsche. a Porsche

(26) a. Avi thinks that [if it would snow, things would be good]. b.* Avi thinks that [if only it would snow]. (cf. Rifkin 2000) 2.4

The scalar semantics of EX

In brief: I argue that EX conveys relative good-ness of the modified proposition, as opposed to optimality. EX is thus not simply a universal modal, but a scalar operator. This permits the unification with polar exclamatives.  Core Argument: Particles can provide further information on the status of the modified proposition with respect to the salient scale (e.g. how desirable is it?). ⇒ Take the following preference scale. (The scenario is based on Villalta 2007.)

6

In English this is well-formed; my discussion of only will yield insights in what may be happening here. - 11 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

(27) Context: Sofia promised to bring a dessert to my picnic. I really do not like ice cream; I prefer apple pie; my first choice is chocolate cake. preferences most desirable chocolate cake apple pie ice cream least desirable

nothing

⇒ As I argue in section 3, particles in optatives seem to be truth-conditionally vacuous presupposition triggers7 with meanings such as the following. (28)

a. the non-exclusive only that we find in optatives conveys that the modified proposition is relatively low (i.e. not much to ask for). b. at least conveys that there is a contextually salient proposition that is more preferable than the modified proposition (i.e. the speaker is ‘settling for less’) c. doch conveys that the expressed proposition is in conflict with what is actually the case; in this sense, doch merely reinforces counterfactuality

⇒ If Sofia brought ice cream, I can utter either of the following optatives. (29)

Jetzt kommt die mit Vanilleeis daher! … now comes she with vanilla.ice.cream here ‘Now she arrives with vanilla ice cream!’ a. Ach, wenn sie doch / ?nur /#wenigstens einen Schokokuchen gebracht hätte! oh if she doch only at.least an chocolate.cake brought had ‘If only she had brought a chocolate cake!’ b. Ach, wenn sie ??doch / nur / wenigstens einen Apfelkuchen gebracht hätte! oh if she doch only at.least an apple.cake brought had ‘If only she had brought an apple cake!’ c. Ach, wenn sie doch nur wenigstens einen Apfelkuchen gebracht hätte! oh if she doch only at.least an apple.cake brought had ‘If only she had brought an apple cake!’

7

It is an open question whether these elements are presupposition triggers or triggers for conventional implicatures. In the domain of optatives, this issue is hard to test, as most diagnostics fail to apply. - 12 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

⇒ What does this teach us with respect to optatives? – –

The modified proposition does not need to denote the optimal case, (29b+c). The modified proposition can denote the optimal case, (29a).

 We can shift the focus from a more optimal case to a less optimal case… (30)

Jetzt kommt die mit Vanilleeis daher! … now comes she with vanilla.ice.cream here ‘Now she arrives with vanilla ice cream!’ Ach, wenn sie doch einen Schokokuchen gebracht hätte! oh if she doch an chocolate.cake brought had ‘If only she had brought an chocolate cake!’ Oder wenn sie wenigstens einen Apfelkuchen gebracht hätte! or if she at.least an apple.cake brought had ‘Or at least if only she had brought an apple pie!’

 … but not vice versa. (31)

Jetzt kommt die mit Vanilleeis daher! … now comes she with vanilla.ice.cream here ‘Now she arrives with vanilla ice cream!’ Ach, wenn sie wenigstens einen Apfelkuchen gebracht hätte! oh if she at.least an apple.cake brought had ‘Oh, at least if only she had brought an apple pie!’ # Oder wenn sie doch einen Schokokuchen gebracht hätte! or if she doch an chocolate.cake brought had ‘Or if only she had brought an chocolate cake!’

 My proposal: Capturing this distribution by an approach that treats optatives as constructions that convey the relative height of the denoted proposition on the salient scale. We can model this in terms of a (here: satisfaction) threshold, which can shift. (32) a.

preferences most desirable chocolate cake THRESHOLD

b.

preferences most desirable chocolate cake

apple pie

apple pie THRESHOLD

ice cream least desirable

ice cream

nothing

least desirable - 13 -

nothing

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 We can formalize this as follows. (33) Lexical entry for EX For any scale S and proposition p, interpreted in relation to a context c and assignment function g, an utterance EX(S)(p) is felicitous iff ∀q[THRESHOLD(c) >S q → p >S q] “EX expresses an emotion that captures the fact that p is higher on a (speaker-related) scale S than all contextually relevant alternatives q below a contextual threshold.”

where THRESHOLD(c) is a function from a context into a set of worlds / a proposition that counts as high with respect to a relevant scale S. (34)

Definition of scale a. A scale S is defined as a set of ordered pairs of worlds (S ⊆ W × W), which are ordered by an ordering relation R, such that for every pair of worlds in S, the relation R() holds. b. For any scale S and corresponding ordering relation R, I use w7 >S w3 to mean ‘w7 is strictly higher than w3 on S’, i.e. R() ∧ ¬R(). c. For any proposition p and q, p >S q iff ∀w3 ∈ q ∃w7 ∈ p such that w7 >S w3, and it is not the case that ∀w7 ∈ p ∃w3 ∈ q such that w3 >S w7. (adapted from Villalta 2007:106, using concepts from Klinedinst 2005)

⇒

Why do we not directly rank sets of worlds? (Villalta uses Kratzer’s 1991 better possibility to rank propositions, a move that I adopt.) – While some p worlds may be more preferable than ¬p (or q) worlds, it cannot be the case that all p worlds are more preferable (as there are p worlds where the speaker has died / was never born / etc).

 An illustration: (35) a. Mein Gott, dass der nicht verschlafen hat! my God that he not overslept has lit. My God, that he didn’t oversleep! b. EX(S)(he-didn’t-oversleep) is felicitous iff ∀q[THRESHOLD(c) >S q → he-didn’t-oversleep >S q] “The speaker expresses an emotion that [he didn’t oversleep] is relatively high on S.”

qp EX(S) (that) he didn’t oversleep wo EX S - 14 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 A prediction: Optatives are felicitous whenever the denoted proposition is just about sufficient to bring about tolerable circumstances (from the speaker’s perspective). (36)

Context: I broke my right arm and I’m right-handed Oh, if only I had broken my left arm!

⇒

(37)

a. # It would be great/good/nice/ideal [if I had broken my left arm]. b.# It would be wonderful [if I had broken my left arm]. c. # It would be a good thing [if I had broken my left arm].

⇒ (38)

In contrast, it is more appropriate to paraphrase (36) by means of (38). a. It would be preferable [if I had broken my left arm]. b. It would be better [if I had broken my left arm]. c. I would prefer it [if I had broken my left arm].

⇒

(39)

The fact that (36) does not mark circumstances in which I broke my left arm as optimal follows from the fact that none of the paraphrases in (37) is felicitous.

This follows if the threshold marks the boundary between tolerable circumstances (how ever low they may be in overall desirability) and intolerable circumstances. preferences most desirable I didn’t break anything tolerable circumstances

I broke my left arm THRESHOLD

I broke my right arm least desirable

I broke both of my arms

intolerable circumstances

 Let us now investigate the particles that we find in EX-utterances a bit more closely.

3. Particles and Forces: Modulating EX 3.1

The Problem

In brief: I present the intricate particle–exclamation connection as my core puzzle. - 15 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 Problem 1: Particles as licensors. – Prototypical particles such as nur ‘only’, doch and wenigstens ‘at least’ do not seem to affect the modified proposition in a tangible way; all of the utterances in (40a-e) seem to entail (40f). Yet, they seem to license optativity, cf. (41). (40) a. Wäre er nur rechtzeitig gekommen! were he only in.time come b. Wäre er doch rechtzeitig gekommen! were he doch in.time come c. Wäre er wenigstens rechtzeitig gekommen! were he at.least in.time come d. Ach, wäre er rechtzeitig gekommen! oh were he in.time come e. Ach, wäre er doch nur wenigstens rechtzeitig oh were he doch only at.least in.time ‘If only he had come in time!’ f. entailment: I wish [he had come in time]. (41)

?

gekommen! come

# Wäre er rechtzeitig gekommen! were he in.time come ‘If only he had come in time!’

 Problem 2: Particles as disambiguators. – Particles typically act as disambiguating elements in ambiguous utterances. Consider three examples of that-clauses that have an optative and a polar exclamative reading. The particle nur ‘only’ always disambiguates. (42)

a. Dass EIN Mal am Wochenende die Sonne scheint! that one time on.the week.end the sun shines lit. ‘That once the sun is shining on the weekend!’ opt. reading: [I want] that for once the sun shines on the weekend! pol.ex. reading: [I’m surprised] that for once the sun shines on the weekend! b. Mein Gott, dass der heute nicht verschlafen hat! my God that he today not overslept has lit. ‘My God, that he didn’t oversleep today!’ opt. reading: [I hope] that he didn’t oversleep today! pol.ex. reading: [I’m shocked] that he didn’t oversleep today! - 16 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

c. Und dass du dich nicht schämst! and that you you not be.ashamed lit. ‘And that you’re not ashamed!’ opt. reading: [I want] that you are not ashamed! pol.ex. reading: [I’m shocked] that you are not ashamed! d. Hätte die dem doch tatSÄCHlich das Buch gegeben! hadsubj she him doch indeed the book given lit. Hadsubjunctive she indeed given him the book! (Scholz 1991:132-133) opt. reading: [I wish that] she had given him the book! pol.ex. reading: [It’s shocking that] she would have given him the book! (43)

a. Dass nur EIN Mal am Wochenende die Sonne scheint! that only one time on.the week.end the sun shines lit. ‘That once the sun is shining on the weekend!’ opt. reading: [I want] that for once the sun shines on the weekend! * pol.ex. reading: [I’m surprised] that for once the sun shines on the weekend! b. Mein Gott, dass der heute nur nicht verschlafen hat! my God that he today only not overslept has lit. ‘My God, that he didn’t oversleep today!’ opt. reading: [I hope] that he didn’t oversleep today! * pol.ex. reading: [I’m shocked] that he didn’t oversleep today! c. Und dass du dich nur nicht schämst! and that you you only not be.ashamed lit. ‘And that you’re not ashamed!’ opt. reading: [I want] that you are not ashamed! * pol.ex. reading: [I’m shocked] that you are not ashamed! d. Hätte die dem doch nur tatSÄCHlich das Buch gegeben! hadsubj she him doch only indeed the book given lit. Hadsubjunctive she only indeed given him the book! opt. reading: [I wish that] she had given him the book! * pol.ex. reading: [It’s shocking that] she would have given him the book!

 The core questions: – What are the meanings of such particles? – How do they disambiguate between different utterances? – How do they license particular utterance types? - 17 -

Patrick G. Grosz

3.2

Dissertation Defense

June 30, 2011

A generalized account for speech act indicating particles

In brief: I present a generalized approach to such apparently ‘force-related’ particles.  The core idea that I posit is that discourse particles are truth-conditionally vacuous elements that act as presupposition triggers (or possibly as conventional implicatures – for present purposes there is no need to commit to either view). (44)

truth-conditional semantics of ‘at least’, ‘only’ and ‘doch’ in optatives If defined, ||wenigstens ‘at least’ / nur ‘only’ / doch|| = λp.p

 Specifically, I assume the following meanings for the three particles. (45) a. Wenn Otto doch nur wenigstens auf seine Mutter gehört hätte! if Otto DOCH only at.least to his mother listened had ‘If only Otto had listened to his mother!’ b. meaning without particles: I’d be content if Otto had listened to his mother. c. contribution of wenigstens: I’m settling for a less than optimal option. d. contribution of nur: This is not much to ask for / it’s a relatively low preference. e. contribution of doch: This is in sharp contrast (i.e. contradiction) to reality. Now: I show that truth-conditional vacuousness of such particles should not surprise us.  A first piece of support for truth-conditional vacuousness stems from the fact that particles that are roughly synonymous can be stacked in optatives without giving rise to compositionality issues. This is shown in (46), which contrasts with (47)8. (46)

(47)

8

a. Wenn nur bloß alle Menschen if only only all humans ‘If only all humans were so happy!’ b. Wenn bloß nur alle Menschen if only only all humans ‘If only all humans were so happy!’

so so

glücklich wären! happy were

so so

glücklich wären! happy were

German

Otto hat nur / bloß / #nur bloß / #bloß nur ein Kind. Otto has only only only only only only one child ‘Otto only has one child.’ (intended reading: ‘Otto doesn’t have more children.’)

In (46), the stacking of nur and bloß seems to give rise to a strengthening effect, something that we typically find with repeated / iterated conventional implicatures (cf. Potts 2005, 2007). - 18 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 Croatian barem ‘at least’ and makar ‘at least’ can also co-occur freely in an optative. (48)

a. Kad bi barem makar danas pala kisa! Croatian when would at.least at.least today fall.part.fem rain ‘I wish it would rain at least today.’ (lit. ‘If it would rain at least today!’) b. Kad bi makar barem danas pala kisa! when would at.least at.least today fall.part.fem rain ‘I wish it would rain at least today.’ (lit. ‘If it would rain at least today!’)

Now: I outline the general principle by which particles disambiguate utterance types.  The next step is to derive the disambiguating effect of particles. ⇒ I argue that disambiguating is due to a conspiracy and arises whenever a particle is incompatible with a certain environment. – Consider an example of multiply ambiguous V1 clause, in (49a). (49)

a. Hätte die dem tatSÄCHlich das Buch gegeben … hadsubj she him indeed the book given lit. Hadsubjunctive she indeed given him the book … b. ‘If she had given him the book …’ c. ‘[It’s shocking that] she would have indeed given him the book’ d. ‘If only she had given him the book’ –

(50)

doch requires access to some individual’s doxastic states; it appears that this is possible in factual conditional antecedents and in root clauses, but not in truly integrated adjunct clauses. It thus blocks a hypothetical conditional reading: a. Hätte die dem doch tatSÄCHlich das Buch gegeben … hadsubj she him doch indeed the book given lit. Hadsubjunctive she indeed given him the book … b. * ‘If she had given him the book …’ c. ‘[It’s shocking that] she would have indeed given him the book’ d. ‘If only she had given him the book’

–

cond. p.exc. opt.

cond. p.exc. opt.

Conversely, nur ‘only’ seems to be licensed in optatives but not in polar exclamatives, due to their entailment properties (I will discuss this in some detail); therefore, it blocks a polar exclamative reading, cf. (51). - 19 -

Patrick G. Grosz

(51)

Dissertation Defense

June 30, 2011

a. Hätte die dem doch nur tatSÄCHlich das Buch gegeben … hadsubj she him doch only indeed the book given lit. Hadsubjunctive she only indeed given him the book … b. * ‘If she had given him the book …’ cond. c.* ‘[It’s shocking that] she would have indeed given him the book’ p.exc. d. ‘If only she had given him the book’ opt. –

Similarly, wenigstens ‘at least’ conveys positive evaluation of the modified proposition, blocking any interpretation that is evaluatively neutral or negative.

Now: I argue that licensing / quasi-obligatoriness is a consequence of disambiguation.  We have now a means of accounting for licensing as a consequence of disambiguation. ⇒

(52)

⇒

Summarizing the above observations, we can construct the following ‘disambiguation graph’. V1-CLAUSE

HYPOTHETICAL CONDITIONAL

+ DOCH

POLAR EXCLAMATIVE

+ DOCH + NUR

OPTATIVE

It follows from standard assumptions on rational discourse participants (cf. Lewis’s 1969 signaling games) that the following communicative strategies will be most successful: – In order to warrant successful communication, the speaker will use disambiguating particles in multiply ambiguous utterances whenever the context alone is insufficiently biased to disambiguate. – Correspondingly, the hearer will interpret particle-free utterances according to contextual bias (i.e. prior probability); given that more canonical, less marked utterances have a higher probability, examples such as the above will typically be interpreted as conditional fragments. – This yields conventional (but crucially not grammaticized) associations such as the following. - 20 -

Patrick G. Grosz

(53)

Dissertation Defense

V1-CLAUSE

June 30, 2011 HYPOTHETICAL CONDITIONAL

+ DOCH

POLAR EXCLAMATIVE

+ DOCH + NUR

OPTATIVE

 My account for licensing by particles predicts that we find well-formed optatives without any particles. ⇒ While this is marginal, relevant examples have been documented, shown in (54). (54) a. Dass ich noch einmal Venedig sehen könnte! that I still once Venice see could ‘Oh that I could see Venice once more!’ (Truckenbrodt 2006a) b. Wenn ich deine Statur hätte! if I your build had ‘[Oh!] If [only] I had your build!’ (adapted from Evans 2007, most natural stress marking is indicated by me) c. Rico schaute die Blumen an und dachte: ‘Rico looked at the flowers and thought:’ “ Wenn Stineli diese sehen könnte!” if Stineli these see could ‘If Stineli could see these!’ und stand lange unbeweglich am Zaun. ‘and stood at the fence for a long time without moving.’ (Johanna Spyri (1878): Heimatlos. Discussed in Grosz 2010b) d. WÄRE ich zuhause geblieben! WERE I at.home stayed ‘HAD I stayed home!’ (Rosengren 1993:36)

Up next: I summarize my analyses of only and at least, which serve as two case studies.

- 21 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

3.3 Two Case Studies: Only and At Least 3.3.1 Optative Only is (Non-Exclusive) Minimal Sufficiency Only In brief: I argue for two different only’s, one of which is what we find in optatives.  The core idea that I defend is that only has a second, weaker reading, which is entailed by the reading of canonical only. Take a uniformly scalar analysis for canonical, exclusive only:

⇒

(Jacobs 1983, Bayer 1996, Klinedinst 2004, 2005, Krasikova & Zhechev 2006, Riester 2006, Beaver & Clark 2008) (55) a. ||only1,C|| = λS.λp.λw : p(w) ∨ ∃q [q >S p → q(w) = 1] ∧

AT LEAST

“Presupposition 1: The modified proposition or a higher scalar alternative is true.”

b.

MOST q ∈ g(C) [q >S p] .

LOWNESS

“Presupposition 2: The modified proposition is low on the salient scale.”

c.

¬∃q [q >S p → q(w) = 1]

AT MOST

“Truth Conditional Content: There is no higher scalar alternative that is true.”

 Illustration of the canonical exclusive reading: (56)

a. Barclay is only [a lieutenant]F. (He is not yet lieutenant commander.) Scale: ensign < lieutenant < lieutenant commander < commander < … presupposed: Barclay is at least a lieutenant, which is relatively low. asserted: Barclay is at most a lieutenant. b. I only saw [Barclay]F. (I didn’t see Geordi or William.) Scale: ⊤ < {Barclay, …} < {Barclay + Geordi, Barclay + William} < … presupposed: I saw at least Barclay, which is relatively low. asserted: I saw at most Barclay. (I did not see Barclay + Geordi. …) entailed: I did not see Geordi. I did not see William.

 New: I argue that there is also a weaker, non-exclusive version, given in (57). (57) a. ||only2,C|| = λS.λp : MOST q ∈ g(C) [q >S p] .

LOWNESS

“Presupposition: The modified proposition is low on the salient scale.”

b.

p

IDENTITY

“Truth Conditional Content: only2 is truth-conditionally vacuous.” - 22 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Now: I provide evidence for both ONLY1 and ONLY2 in German conditional antecedents.  Languages like German exhibit both readings of only in conditionals. (58)

a. Wenn nur zwei Personen einsteigen, wird das Boot sinken. if only two persons get.in will the boat sink lit. ‘If only two persons get into the boat, the boat will sink.’ b. ONLY1 reading: If no more than two persons (i.e. < 3 people) get into the boat, it will sink. implies: If three people get into the boat, it will possibly not sink. c. ONLY2 reading: If (at least) two persons (i.e. ≥ 2 people) get into the boat, which is not a lot, it will sink. implies: If three people get into the boat, it will definitely sink.

 Context sentences that bring out the respective readings are given in (59). (59)

a. Es müssen mindestens drei Personen einsteigen. it must at.least three persons get.in ‘At least three persons must get in.’ ⇒ incompatible with ONLY2 Wenn nur ZWEI Personen einsteigen, wird das Boot if only two persons get.in will the boat lit. ‘If only two persons get into the boat, the boat will sink.’

sinken. sink

b. In dieses Boot darf maximal eine Person einsteigen, so schwach ist es. in this boat may maximally one person get.in so weak is it ‘At most one person may get into this boat, that’s how weak it is.’ ⇒ incompatible with ONLY1 Wenn nur ZWEI Personen einsteigen, wird das Boot sinken. if only two persons get.in will the boat sink lit. ‘If only two persons get into the boat, the boat will sink.’  The difference between only1 and only2 is not due to scope differences. ⇒

Both readings are available when nur ‘only’ is DP-internal (presumably taking narrow scope on the numeral), shown in (60). - 23 -

Patrick G. Grosz

(60)

Dissertation Defense

June 30, 2011

Wenn der Otto [PP mit nur [einer]F Begleitperson] kommt, spielen wir if the Otto with only one person comes play we Siedler von Catan. Settlers of Catan ‘If Otto comes with only one guest, we will play Settlers of Catan.’ a. ONLY1: [If Otto does not bring more than one guest], we will play Settlers of Catan (because all other games require a larger group). b. ONLY2: [If Otto brings at least one guest, which is easy to achieve], we will play Settlers of Catan.

⇒

Both readings are available when nur ‘only’ takes propositional scope combined with wide focus on the entire proposition.

(61) a. Wenn nur [jeder einmal etwas pro Seminareinheit sagt]F, if only everyone once something per seminar.unit says hätte man schon meist zwischen 10-20 Wortmeldungen had one schon mostly between 10-20 requests.to.speak ‘[If only everyone said one thing per seminar], it would already come to 10-20 contributions.’ (online post on www.aufmuken.at, 02/11/2010)

paraphrase: [For everyone to say one thing per seminar] is an easy means to (ONLY2) achieve 10-20 contributions. b. Wenn nur [jeder einmal etwas pro Seminareinheit sagt]F, if only everyone once something per seminar.unit says kommen wir mit unserem Projekt (halt eben) nie weiter! come we with our project simply never further ‘[If only everyone says one thing per seminar], we’ll never make progress on our project!’ paraphrase: [If nothing more happens than everybody saying one thing per (ONLY1) seminar unit] we won’t ever make progress on our project. Now: I argue that optative only is the (weaker) only2. My first argument is based on the generalization that languages with ONLY2 readings can use only in optatives.  I will also refer to ONLY2 by means of the term ‘minimal sufficiency only’, as it mainly serves to mark a conditional antecedent as likely / easy to achieve. - 24 -

Patrick G. Grosz

June 30, 2011

Evidence that optative only is typically minimal sufficiency only stems from two observations.

⇒

– (62)

Dissertation Defense

First, languages that have minimal sufficiency only can also use it in optatives.

German a. Wenn bloß zwei Personen einsteigen, wird das Boot sinken. if only two persons get.in will the boat sink lit. ‘If just two persons get into the boat, the boat will sink.’ (ONLY1 / ONLY2)

ONLY1: The boat will sink if no more than two persons get in. ONLY2: The boat will sink if at least two persons get in, which is not a lot. b. Wenn Hans bloß auf Maria gehört hätte! if Hans only to Maria listened had ‘If only Hans had listened to Mary!’ (63)

Italian a. Se solo/solamente due persone salissero su questa barca, affonderebbe if only two persons enter in this boat it.will.sink lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 /ONLY2) b. Se solo/solamente John avesse if only John had ‘If only John had listened to Mary!’

(64)

Lebanese Arabic9 a. iza bass shaxs-ein tel3ou 3a-ha-sh-shaxtoura, b-teghra' if only person-dual got-3p on-this-the-boat, sink.ipfv.3sf lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 /ONLY2) b. (Ah,) law kent bass oh if was.1s only ‘If only I were rich!’

(65)

9

ascoltato/%ascoltava Maria! % listened.to(past.subj/ ipv) Mary

ghani! rich

Czech a. Kdy-by jen dva lidi nasedli na tuto lod’, potopila by se. when-subj.3 only two people get(pptc) on this boat sink(pptc) subj self lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 /ONLY2)

The difference between iza ‘if’ and law ‘if’ is tense/aspect related and should not concern us here. - 25 -

Patrick G. Grosz

Dissertation Defense

b. Kdy-by jen Honza poslechl when-subj.3 only Honza listened(pptc) ‘If only John (had) listened to Mary!’ (66)

June 30, 2011

Marii! Marie.acc

Polish a. Jeśli tylko dwie osoby wejdą na ten statek, to zatonie. if only two people enter on this ship then sink lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 /ONLY2) b. Gdyby / Żeby tylko Jan (po)słuchał if if only John listen.(perf.)pret.3sg.m ‘If only John had listened to Mary!’

(67)

Serbian a. Ako se samo dva čoveka popnu na palubu, brod će potonuti. if self only two man climb on deck ship will sink lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 /ONLY2) b. Da je samo Jovan poslušao that be.3sg only John listened ‘If only John had listened to Mary!’

(68)

Marii. Mary.gen.nom.f

Mariju! Mary-acc

Norwegian10 a. Dersom bara to personer går i denna båten, så synker'n if only two people get in this boat then sinks’it lit. ‘If only two persons get into the boat, the boat will sink.’ (??ONLY1 /ONLY2) b. Om / Hvis han bare hadde kjørt litt fortere! if if he only had driven little faster ‘If only he had driven a little faster!’

(69)

10

Dutch11 a. Als maar twee mensen in deze boot stappen, zal het zinken. if only two people in this boat step will it sink % lit. ‘If only two persons get into the boat, the boat will sink.’ (ONLY1 / ONLY2)

Interestingly, the dominant reading in Norwegian is one where the boat will sink if exactly two people get in. This reading also emerges in Dutch if the word alleen ‘only’ is used as a translation for only. 11 Dutch speakers differ in their acceptance of in (69a). This seems to be an artifact of this particular example; speakers who dislike ONLY2 in (69) accept examples like (i), which requires an ONLY2 reading. i. Als maar iedereen op OSX overstapt komen daar ook wel virussen voor. 〈trosradar.nl〉 if only everybody on OSX change come there too indeed viruses for ‘as soon as (lit. if only) everybody switches to OSX, there will be viruses on that platform as well.’ - 26 -

Patrick G. Grosz

Dissertation Defense

b. Als Jan (nou) maar naar Marie if Jan PRT only/but to Marie ‘If only John had listened to Mary!’

June 30, 2011

had geluisterd! had listened

Now: My second argument is based on the generalization that languages that disallow only in optatives also do not allow for ONLY2 readings. –

(70)

Second, languages that do not have only in optatives, also do not have minimal sufficiency only.

Brazilian Portuguese a. Se só/apenas duas pessoas entrarem neste barco, ele vai afundar. if only two people enter in-this boat he will sink lit. ‘If only two people get on the boat, the boat will sink.’ (ONLY1 / *ONLY2)

ONLY1: The boat will sink if no more than two persons get in. * ONLY2: The boat will sink if at least two persons get in, which is not a lot. b. Se ao menos / *só / *apenas o João tivesse if at least *only *only the John had ‘If at least / *only John had listened to Mary!’ (71)

ouvido listened.to

Spanish a. Si solo dos personas se montan en esa barca, se hundira if only two people self get on that boat self will.sink lit. ‘If only two people get on the boat, the boat will sink.’ (ONLY1 / *ONLY2) b. Si 〈*solo〉 Juan hubiera 〈al menos〉 escuchado if *only Juan had.sub.past at least listened ‘If at least / *only John had listened to Mary!’

(72)

a Maria! the Mary

a María! to Mary

Greek a. An mono dhio anthropi anevun s'afto to plio, tha vuliaksi. if only two people enter on-this the boat fut sink lit. ‘If only two people get on the boat, the boat will sink.’ (ONLY1 / *ONLY2) b. An toulachiston / *mono o John iche akusi tin Mary! if at.least *only the John.nom had.3sg listened the Mary.acc ‘If John had at least / *only listened to Mary!’

- 27 -

Patrick G. Grosz

(73)

Dissertation Defense

June 30, 2011

Catalan a. Si només pugen dues persones en aquesta barca, s’enfonsarà. if only get.in two people on that boat self.will.sink lit. ‘If only two people get on the boat, the boat will sink.’ (ONLY1 / *ONLY2) b. Si almenys / *només hagués escoltat if at.least *only had.subjunctive listened ‘If at least / *only he had listened to Mary!’

(a) to

la Maria! the Mary

Now: I discuss English as an apparent counter-example to my analysis / generalizations. –

English is somewhat exotic in that it belongs to a possible third class of languages, which has only in optatives, but does not have minimal sufficiency only.

(74) a. If only two people get into the boat, it will sink. ONLY1: The boat will sink if no more than two persons get in. * ONLY2: The boat will sink if at least two persons get in, which is not a lot. b. If only John had listened to Mary! I assume that English only2 has grammaticized into an optativity marker by acquiring an additional presupposition that restrict the relevant scale to a preference scale.

⇒

 The entry for English optative only is given in (75); (75b) is not unprecedented. In Nakanishi & Rullmann (2009), concessive at least is assumed to convey (75b). (75) a. ||onlyOPT,C|| = λS.λp : MOST q ∈ g(C) [q >S p] ∧

LOWNESS

“Presupposition 1: The modified proposition is low on the salient scale.”

b.

S is a bouletic ordering .

BOULETIC

“Presupposition 2: The contextually salient scale is a bouletic scale.”

c.

p

IDENTITY

“Truth Conditional Content: onlyOPT is truth-conditionally vacuous.”

But does English not have minimal sufficiency only at all?

⇒ –

Liz Coppock (p.c.) suggests (based on joint research with David Beaver) that just may have a reading similar to what I call a minimal sufficiency reading, even in simple assertions. - 28 -

Patrick G. Grosz

(76)

Importantly, just can license optatives in English, (77).

a. Oh, if he just knew how much we miss him! (= I wish he knew how much we miss him!) b. Oh, if just once I could be a guest in such a beautiful house! (= I wish I could once be a guest in such a beautiful house!) –

(78)

June 30, 2011

Just the thought of him sends shivers down my spine. (Coppock & Beaver 2011) ONLY1: Nothing but the thought of him sends shivers down my spine. ONLY2: The thought of him (and possibly other things) sends shivers down my spine, and that’s something rather minimal. –

(77)

Dissertation Defense

And there may be evidence that just has only2 readings in conditionals.

a. But it does work reasonably well, and if you use it just once, you've saved more than the purchase price. ⇒ only2 reading preferred (http://www.amazon.com/Paylak-LK6-4-Watch-Sizing-Repair/dp/B0015SHC8Y)

b. One good thing about pu-erh is that you can use the same cake over and over for multiple infusions. If you use it just once, you’re wasting tea. ⇒ only1 (http://www.face-natural.com/blog/natural-skin-care-articles/super-teas-for-natural-skin-care)

c. If just two people get into the boat, it will sink. ONLY1: The boat will sink if no more than two persons get in. ? ONLY2: The boat will sink if at least two persons get in, which is not a lot. Now: I argue for the lowness component that all readings of only share.  I have posited that all entries of only have a lowness presupposition (i.e. the modified proposition is low on a salient scale). ⇒

(79)

For canonical, exclusive only, the following examples illustate this property of only. If the domain of relevant individuals is {John, Mary, Bill, Alex, Sue, Eric} a. The meeting was only attended by [John, Mary and Bill]F, # a surprisingly high turnout. b. The meeting was attended by everyone except/but Alex, Sue and Eric, a surprisingly high turnout. (Klinedinst 2005) - 29 -

Patrick G. Grosz

⇒

(80)

⇒ (81)

Dissertation Defense

June 30, 2011

For minimal sufficiency only, examples like the following illustrate this lowness component. a. Wenn nur zwei Personen einsteigen, wird das Boot sinken. if only two persons get.in will the boat sink intended: ‘If as little as two people get in, the boat will sink.’ b.# Wenn nur zweihundert Personen einsteigen, wird das Boot sinken. if only two.hundred persons get.in will the boat sink intended: ‘If as little as two hundred people get in, the boat will sink.’ For optatives, this lowness component is supported by the following contrast. Context: I need to get from Boston to Providence as quickly as possible. To do so, I need a car. Unfortunately I do not own a car. My neighbors have a car, but it's a Porsche, so they wouldn't lend it to me. I exclaim the following. a. Oh, if only I owned a car! (mean1-5 = 4.46, sd = 0.88, n = 13) b. # Oh, if only I owned a Porsche! (mean1-5 = 2.54, sd = 1.20, n = 13)

Now: I argue that the difference between optatives and polar exclamatives (which seem incompatible with only) is connected to the entailment properties of the respective scales.  How does minimal sufficiency only eliminate a polar exclamative reading? ⇒ Szabolcsi (2004) argues that Dutch maar (which has only2 readings) is some type of positive polarity item. ⇒ Could it be that ONLY2 has properties of positive polarity items? ⇒

This would indicate that optatives are ‘positive polarity’ environments, whereas polar exclamatives are not; conversely, polar exclamatives may be ‘negative polarity’ environments.

 As a matter of fact, it can be shown that polar exclamatives, but not optatives, allow for NPIs. ⇒

Consider first an ambiguous dass-clause.

(82) a. Mensch, dass dieser Kandidat einmal einen Förderpreis erhalten man that this candidate once a grant received ‘Man, that this candidate has once received a grant!’ - 30 -

hat! has

Patrick G. Grosz

Dissertation Defense

June 30, 2011

b. scenario for optative reading We are auditioning different candidates for a new job. So far, none of our candidates has ever received a grant. We are tired and desparate and really hope that the candidate who is about to enter has received a grant at least once. paraphrase: ‘[Let’s hope] that this candidate has once received a grant!’ c. scenario for polar exclamative reading We are auditioning different candidates for a new job. Our last candidate was a complete disaster and we consider him completely incompetent. However, he has received a grant once in his career, which shocks us. After he leaves, we express our shock (and dismay) at this fact. paraphrase: ‘[It’s shocking] that this candidate has once received a grant!’  As Gärtner (2010) documents, NPIs are bad in optatives, reproduced for my example in (83). (83)

Mensch, dass dieser Kandidat je(mals) einen Förderpreis erhalten man that this candidate ever a grant received ‘Man, that this candidate has ever received a grant!’ * optative reading  polar exclamative reading

hat! has

 Conversely, we have already seen that only2 blocks polar exclamative readings. (84)

Mensch, dass dieser Kandidat nur einen Förderpreis erhalten man that this candidate only2 a grant received ‘Man, that this candidate has received a grant!’  optative reading * polar exclamative reading

hat! has

 Combining both elements is correctly predicted to give rise to ill-formedness. (85) # Mensch, dass dieser Kandidat nur je einen Förderpreis erhalten man that this candidate only2 ever a grant received ‘Man, that this candidate has received a grant!’ * optative reading * polar exclamative reading

hat! has

 A generalization seems to arise: ⇒ Optatives seem to have properties of positive polarity environments. ⇒

Polar exclamatives seem to have properties of negative polarity environments. - 31 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

 Can we derive this in a theory, where the two utterance types only differ in the nature of the scale that they involve? ⇒ Szabolcsi (2004) argues that Dutch maar cannot occur in antiadditive contexts12. (86)

a. A function f is antiadditive iff f(a ∨ b) = fa ∧ fb. b. No one walks or talks = No one walks and no one talks (Szabolcsi 2004:414)

⇒ ⇒

Could polar exclamatives be antiadditive, while optatives are not? This would have to be a property of scales.

 Evidence that optatives are not antiadditive: The statement in (87a) seems unproblematic, and appears to entail (87b). So desirability cannot be anti-additive. (87) a. It is desirable that Bob or Ann comes, but it is not desirable that both come, as they hate each other. b. It is [desirable that Bob or Ann comes], but it is not independently [desirable that Bob comes] and [desirable that Ann comes], as Bob and Ann hate each other. (Ann’s coming is only desirable if Bob does not come and vice versa.)  What about unlikelihood (which I assumed to be involved in polar exclamatives)? ⇒ According to probability theory: – If a or b is unlikely, then a is unlikely and b is unlikely. – But: If a is unlikely and b is unlikely, it does not follow that a or b is unlikely.  Could polar exclamatives still involve anti-additivity? ⇒ How do we measure subjective unlikelihood? – –

What if the relevant scale was a scale of shocking-ness? The clauses in (88) feel contradictory, indicating that this might well be antiadditive.

(88) a.??It is shocking if it rains or snows, but it is not shocking if it rains and it is not shocking if it snows. b.??It is shocking if it rains and it is shocking if it snows, but it is not shocking if it rains or snows. 12

This is a rather weak requirement, which is a desirable result, as it opens the future possibility of applying our analysis of ONLY2 to only in rhetorical questions (German Was macht ihr nur für Sachen? ‘What ARE you [only] doing?’) and imperatives (German Komm nur! ‘Please, do [only] come!’). - 32 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Now: I briefly summarize the meaning of only2 and its disambiguation/licensing effect.  What is the role of only in optatives? ⇒ It marks the modified proposition as comparatively easy, i.e. not much to ask for. ⇒ It disambiguates by blocking polar exclamative readings due to their entailment properties. 3.3.2

Optative At Least is Concessive At Least

Now: I present background on the analysis of optative at least that I argue for.  The point for at least is quick to make. Nakanishi & Rullmann (2009) argue that there are two readings for at least: The canonical, so-called epistemic reading, and a noncanonical, concessive reading. (89)

epistemic at least (illustration: He has at least two children.) ||at least||(C)(p)(w) is defined iff ∃w’[Epist(w,w’) ∧ ∃q ∈ C [q > p ∧ q(w’) = 1]] UNCERTAINTY “CI / Presupposition: The speaker is unsure whether a higher scalar value holds or not.”

If defined, then ||at least||(C)(p)(w) = 1 iff ∃q ∈ C [q ≥ p ∧ q(w) = 1]

ASSERTION

“Truth Conditional Content: The modified proposition or a higher scalar value holds.”

(slightly modified from Nakanishi & Rullmann 2009, paraphrases are mine)13 (90)

concessive at least (illustration: At least we won something!) ||at least||(C)(p)(w) is defined iff ∀r,r’ ∈ C [r’ > r ↔ r’ is preferred to r] BOULETIC SCALE “CI / Presupposition 1: Relevant alternatives are ordered according to preference.”

∃q ∈ C [q > p]

BETTER ALTERNATIVE

“CI / Presupposition 2: There is a salient alternative that is more preferable than p.” 13

Nakanishi & Rullmann (2009) treat the uncertainty contribution as a conventional implicature, whereas I treat it as a definedness condition. In the scope of my project, this distinction is not at stake, which is why I uniformly model such non-truth-functional meanings as definedness conditions. Note: Nakanishi & Rullmann (2009) acknowledge Krifka (1999), Geurts & Nouwen (2007) and Büring (2008) as the basis of their analysis. - 33 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

∃q ∈ C [q < p]

WORSE ALTERNATIVE

“CI / Presupposition 3: There is a salient alternative that is less preferable than p.”

If defined, then ||at least||(C)(p)(w) = p(w)

ASSERTION

“Truth Conditional Content: Concessive at least is truth-conditionally vacuous.”

(slightly modified from Nakanishi & Rullmann 2009, paraphrases are mine)  The core diagnostic for a concessive reading is given as follows. (91)

a. # Mary didn’t win a gold medal, but she won at least a silver medal. b. Mary didn’t win a gold medal, but at least she won a silver medal. (Nakanishi & Rullmann 2009)

(92)

a. # Mary doesn’t have three children, but she has at least two. b. Mary doesn’t have three children, but at least she has two. (Nakanishi & Rullmann 2009)

(only E) (C)

(only E) (C)

Now: I provide a first argument (⇒ German) that optative at least is concessive at least.  It can now be shown that the at least that languages employ in optatives generally has a concessive reading. ⇒ First, German wenigstens ‘at least’ only has a concessive reading, whereas mindestens ‘at least’ only has an epistemic reading. (93) a. Bei dem Unfall gab es #wenigstens / mindestens fünf Tote. at the accident gave it at.least at.least five deads ‘There were at least five casualties in the accident.’ b. Maria hat kein Gold gewonnen, aber wenigstens / #mindestens Silber. Maria has no gold won but at.least at.least silver ‘Maria didn’t win gold, but at least she won silver.’ ⇒

In optatives, only the concessive at least can be used.

(94) Wenn Hans wenigstens / #mindestens auf if Hans at.least at.least to ‘If only Hans had at least listened to Maria!’ - 34 -

Maria gehört hätte! Maria listened had

German

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Now: I provide a second argument for this analysis, based on languages with a specialized concessive at least (Romanian, Polish, Russian, Finnish). ⇒

Second, in many languages, optative at least only has a concessive reading.

(95) a. Dacă Jon măcar ar fi ascultat de if Jon at.least had listened of ‘If only John had at least listened to Mary!’

Mary! Mary

Romanian

b. Sunt cel putin / *măcar patru mere in punga asta. are at least at.least four apples in bag this ‘There are at least four apples in this bag.’ c. Maria nu a câştigat medalia de aur, dar măcar a castigat medalia Maria not won medal of gold but at.least won medal de argint. of silver ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ d. Maria nu are trei copii, dar măcar are doi. Maria not has three children but at.least has two ‘Mary doesn’t have three children, but at least she has two.’ (96)

a. Gdyby / Żeby Jan chociaż (po)słuchał Marii! Polish if if John at.least listen.(perf.)pret.3sg.m Mary.gen.nom.f ‘If at least Jan had listened to Mary!’ b. W torbie są co najmniej / # chociaż cztery jabłka. in bag are at least at.least four apples ‘In this bag, there are at least four apples.’ c. Marie nie zdobyła złotego medalu, ale chociaż zdobyła srebrny. Mary not receive gold medal but at.least receive silver ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ d. Marie nie ma trojga dzieci, ale ma chociaż dwoje. Mary not has three children but has at.least two ‘Mary doesn’t have three children, but at least she has two.’

–

(97)

For Finnish and Russian, the Nakanishi & Rullmann diagnostic failed for independent reasons. a. Esli by Vanja xotja by poslusha-l if subj Vanja.nom at.least subj listen-past.m.sg ‘If at least Vanja had listened to Masha!’ - 35 -

Mash-u! Russian Masha-acc

Patrick G. Grosz

Dissertation Defense

June 30, 2011

b. V etoj muke po krajnej mere / kak minimum / *xotja chetyre jabloka. in this bag at least as minimum at.least four apples ‘In this bag, there are at least four apples.’ c. Nu on xotj/xotja by vyigral chto-to! well he.nom at.least won:m.sg something ‘Well, at least he won something!’ (98) a. olisi-pa John edes kuunnellut Maria! be.cond-PA John at.least listen Mari.part ‘If only John had at least listened to Mary!’ b. tässä pussissa on ainakin / vähintään / *edes neljä this.iness bag.iness is.3sg at.least at.least at.least four ‘There are at least four apples in this bag.’ c. No, voitti(han) hän edes jotakin. well win(prt) he at.least something ‘Well, at least he won something!’

Finnish

omenaa apple.part

Now: I provide a third argument for this analysis, based on languages without a specialized concessive at least (Hebrew, Czech, Greek, Spanish). ⇒

Third, in languages where optative at least has an epistemic reading, it also has a concessive reading.

(99) a. lu John haya le-faxot makSiv le-Mary! Hebrew if.cf John be.past.3sg to-less listen.pres.3sg to-Mary lit. ‘If John/he had at least listened to Mary!’ b. be-te’unat ha-drax-im hayu le-faxot xamiSa harug-im in-accident the-way-pl exist.masc-pl at-least five.masc casualty.masc-pl ‘There were at least 5 casualties in the traffic accident.’ c. Mary lo zaxta be-medalyat zahav, aval le-faxot hi zaxta be-medalyat Mary not won in-medal gold but at-least she won in-medal kesef. silver ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ d. le-Mary eyn SloSa yeladim, aval le-faxot yes l-a Snayim. to-Mary not.exist three children but at-least exist to-her two ‘Mary doesn't have three children, but at least she has two.’ - 36 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

(100) a. Kdy-by aspoň Honza poslechl Marii! Czech when-subj.3 at.least Honza listened.pst.ptcp Marie.acc lit. ‘If John/he had at least listened to Mary!’ b. Při té nehodě zemřelo aspoň pět lidí. at that accident died.sg.neut at.least five people ‘There were at least 5 casualties in the traffic accident.’ c. Marie nevyhrála zlatou medaili, ale vyhrála aspoň stříbrnou. Marie neg:won gold medal but won at.least silver ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ ? d. Marie nemá tři děti, ale má aspoň dvě. Marie neg:has three children but has at.least two ‘Mary doesn't have three children, but at least she has two.’ (101) a. An tulachiston o Yannis iche akusi tin Maria! Greek if at.least the Yannis.nom had.3sg listened the Maria.acc lit. ‘If Yannis had at least listened to Maria!’ b. Sto aftokinitistiko atixima, skotothikan tulachiston pende anthropi. in-the car accident were.killed at.least five people ‘There were at least 5 casualties in the traffic accident.’ c. I Maria dhen kerdhise chriso metalio, ala tulachiston kerdhise the Maria not won gold medal but at.least won arjiro/asimenjo (metalio) silver medal ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ d. I Maria dhen echi tria pedhja, ala tulachiston echi dhio. the Maria not has three children but at.least has two ‘Mary doesn't have three children, but at least she has two.’ (102) a. Si Juan hubiera al menos escuchado a María! Spanish if Juan had.sub.past at least listened to Mary lit. ‘If John/he had at least listened to Mary!’ b. En el accidente de tráfico, hubo al menos cinco víctimas. in the accident of traffic there.was at least five victims ‘There were at least 5 casualties in the traffic accident.’ c. Maria no gano una medalla de oro, pero al menos gano una medalla Maria not won a medal of gold but at least won a medal de plata. of silver ‘Mary didn’t win a gold medal, but at least she won a silver medal.’ - 37 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

d. Maria no tiene tres hijos, pero al menos tiene dos. Maria not has three sons but at least has two ‘Mary doesn't have three children, but at least she has two.’ Now: I briefly summarize the meaning of at least and its disambiguation/licensing effect.  Interim Summary: What is the function of at least in optatives? ⇒ Due to the settling for less component of concessive at least, the speaker conveys that there is a contextually salient proposition that would be better. ⇒ This restricts the contexts in which optatives can contain concessive at least, making such optatives more marked. ⇒ The disambiguation/licensing effect of at least is due to its positive evaluation component: The modified proposition is marked as positive, thus blocking nonevaluative interpretations.

4. On the Role of Mood in Exclamations 4.1

The Problems

In brief: Presuppositions and the nature of C are unaccounted for by the EX approach.  Two puzzles: – What determines complementizer selection and the possibility of V1? – How do presuppositions on the status of the denoted proposition arise? (103) a. subjunctive (and counterfactual) optatives [EXS [Wenn er nur rechtzeitig gekommen wäre]! if he only in.time come were ‘If only he had come in time!’ b. intuitive paraphrase: ‘I wish that he had come in time (and he didn’t).’ (104) a. indicative optatives [EXS [Wenn er nur rechtzeitig gekommen ist]! if he only in.time come is ‘If only he comes in time!’ b. intuitive paraphrase: ‘I want that he came in time (and he may have).’ - 38 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

With want/wish, these are typically assumed to be part of the lexical entry. An EX-Op analysis does not assume such a lexical variation in EX.

⇒

(105) Villalta’s (2007) semantics for want and wish; based on Heim (1992) a. if defined ||wantC||g(p)(a)(w) = ||wishC||g(p)(a)(w) = 1 iff ∀q[[q ≠ p & q ∈ g(C)]→ p >DESa,w q] “The speaker prefers the denoted proposition p over all relevant contextual alternatives.”

b. ||wantC||g(p)(a)(w) is defined iff ∀q[q ∈ g(C) → Doxa(w) ∩ q ≠ Ø] “want is defined iff the denoted proposition p is still a real possibility.”

c. ||wishC||g(p)(a)(w) is defined iff p ∩ Doxa(w) = Ø “wish is defined iff the denoted proposition p is false in the utterance context.”14

(Villalta 2007:108) 4.2

The Solution

In brief: I argue for semantic mood features, which co-determine the material in C.  I assume that the presuppositions that different EX utterances have are contributed by a (semantic) Mood head that contains an interpretable Mood feature. ⇒

Consider the three types of exclamations that I have been looking at.

(106) a. Daß Otto nur rechtzeitig gekommen that Otto only in.time come ‘If only Otto had arrived in time!’ b. Daß Otto nur rechtzeitig gekommen that Otto only in.time come ‘If only Otto has arrived in time!’

wäre! were ist! is

counterfactual optative

non-counterfactual optative

c. Daß Otto doch glatt rechtzeitig gekommen ist! polar exclamative that Otto doch outright in.time come is ‘[I’m shocked] that Otto has arrived in time!’ ⇒

14

To capture the difference in semantic Mood, I propose the following three Mood features.

See also Iatridou (2000). - 39 -

Patrick G. Grosz

Dissertation Defense c

(107)a. ||iMoodCF|| = λp . λw : p ∩ Doxspeaker(w) = ∅ . p(w)

June 30, 2011 COUNTERFACTUALITY

“The speaker presupposes p to be false.”

b. ||iMoodDEF||c = λp . λw . p(w)

UNMARKED MOOD

“The status of p is unresolved as of now.”

c. ||iMoodFACT||c = λp . λw : Doxspeaker(w) ⊆ p . p(w)

FACTIVITY

“The speaker presupposes p to be true.”

 Focusing on German, as a case study, I argue that C must agree with Mood, due to an uninterpretable Mood feature in C. ⇒ I propose that subsequently the value of Mood co-determines what material can occur in C, and it also co-determines the morphological mood on the verb15. (108) a. Ach, wenn es geregnet oh if it rained ‘If only it had rained!’ b.

wenn

hätte! had

CP qp C[Mood] … MoodP qp TP Mood[Mood:CF] ri VP T 6 es geregnet hätte

Mood-C agreement

Now: I present three overlapping paradigms that motivate my approach to Mood.  My argument for this proposal is based on the following correlation between semantic Mood and complementizer selection / V1. ⇒

15

Counterfactual exclamations allow for dass ‘that’, wenn ‘if’ and V1.

A similar proposal has been made for the subjunctive mood in other languages, cf. Giorgi and Pianesi (1997, 2004), Kempchinsky (1986) and Quer (1998). The relevant idea is summarized in Giorgi (2009:1847): "[T]he Italian subjunctive exhibits a sort of discontinuous morphology, including both the verbal ending and the complementizer". - 40 -

Patrick G. Grosz

(109)

subjunctive (and counterfactual) optatives a. Daß er nur rechtzeitig gekommen that he only in.time come b. Wenn er nur rechtzeitig gekommen if he only in.time come c. Wäre er nur rechtzeitig gekommen were he only in.time come ‘If only he had arrived in time!’

⇒

(110)

June 30, 2011

wäre! were wäre! were twäre!

Non-counterfactual exclamations allow for dass ‘that’ and wenn ‘if’, but not for V1.

indicative (and non-counterfactual / non-factive) optatives a. Daß er nur rechtzeitig gekommen ist! that he only in.time come is b. Wenn er nur rechtzeitig gekommen ist! if he only in.time come is c.* Ist er nur rechtzeitig gekommen tist! is he only in.time come ‘If only he has arrived in time!’

⇒ (111)

Dissertation Defense

Factive exclamations allow for dass ‘that’ and V1, but not for wenn ‘if’.

indicative (and factive) polar exclamatives a. Daß er doch glatt rechtzeitig gekommen that he doch outright in.time come b.* Wenn er doch glatt rechtzeitig gekommen if he doch outright in.time come c. Ist er doch glatt rechtzeitig gekommen is he doch outright in.time come ‘[I’m shocked] that he came in time!’

ist! is ist! is tist! is

Now: I summarize and (for reasons of brevity) sketch the implementation. ⇒

We can summarize as follows.

- 41 -

Patrick G. Grosz

(112)

Dissertation Defense

June 30, 2011

generalization on C filling and semantic mood in exclamations a. LF: C + [iMoodCF] ⇔ PF: {dass ‘that’, wenn ‘if’, V1} b. LF: C + [iMoodDEF]

PF: {dass ‘that’, wenn ‘if’}

⇔

c. LF: C + [iMoodFACT] ⇔ ⇒

PF: {dass ‘that’, V1}

To implement this proposal formally, I propose that dass ‘that’ is a spell-out of C, whereas wenn ‘if’ is a spell-out of C+Mood (after moving the Mood head as a consequence of agreement); V1 is a spell-out of C+Mood+T (again, T undergoes head movement to Mood, and subsequent head movement to C)16.

(113) a. dass ‘that’ spells out C on its own. b. wenn ‘if’ spells out [C [Mood]]. c. V1 spells out [C [Mood [T]]]. ⇒

If we assume that C agrees with Mood and Mood agrees with T, it follows that different Mood features differ in whether they entail the EPP property on C and/or Mood.

(114) a. C[–EPP] Mood[+EPP] T ⇒

dass ‘that’ … Vfin

b. C[+EPP] Mood[–EPP] T ⇒

wenn ‘if’ … Vfin

c. C[+EPP] Mood[+EPP] T ⇒

V1 …

⇒

The idea is that Mood values determine the distributional possibilities with respect to the EPP property.

5. Conclusion 5.1

Summary

 I have argued for a system with the following three components.

16

⇒

Exclamations (subsuming optatives and polar exclamatives) derive their core meaning via a scalar expressive operator EX.

⇒

Particles that interact with utterance types are truth-conditionally vacuous elements that trigger non-truth-conditional meanings; their core effect on the utterance type is due to incompatibility with alternative interpretations.

⇒

Mood is encoded by means of a semantic Mood feature, which triggers relevant presuppositions and co-determines morphological mood and the material in C.

My proposal of connecting V1 to (e.g. counterfactual) mood is loosely related to Bjorkman’s (2010). - 42 -

Patrick G. Grosz

5.2

Dissertation Defense

June 30, 2011

The big picture

In brief: I sketch the broader relevance of my approach from a bird’s-eye view.  By arguing for a generalized EX operator, together with a generalized particle semantics, I argue against theories that derive optatives, polar exclamatives and other utterances on a case by case basis (e.g. Evans 2007, but also, to a lesser extent, Kyriakaki 2007, 2008, 2009, Biezma 2011ab). (115) The Exclamation-Operator Hypothesis (The EX-Op-Hypothesis): Optatives, exclamatives and related utterances involve an operator EX (mnemonic for exclamation) that combines with a truth-conditional statement and turns it into a felicity-conditional expression of an emotion. (116) Schema of the EX-Op-Hypothesis Utterances without EX Utterances with EX

declarative statements, etc. optatives polar exclamatives ‘adversatives’ ([I hate it] if …) …

(117) The Hypothesis of Exclamative Conspiracies (The Ex-Con-Hypothesis): Optatives, exclamatives and seemingly related utterances are actually completely unrelated, and their meanings arise on a case-by-case basis, either for pragmatic reasons or due to grammaticalized triggers. (118) Schema of Ex-Con-Hypotheses (which are somewhat ‘taxonomical’) Default declarative use declarative statements, etc. Trigger for DESIRE optatives Trigger for SURPRISE polar exclamatives Trigger for DISLIKE ‘adversatives’ ([I hate it] if …) …  To illustrate the contrast, consider the following example.

- 43 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

(119) a. Mein Gott, dass der nicht verschlafen hat! my God that he not overslept has lit. My God, that he didn’t oversleep! b. paraphrase of optative reading: I hope [that he didn’t oversleep]. c. paraphrase of exclamative reading: I’m shocked [that he didn’t oversleep].  An EX-Op approach: (120) a. EXS (Mein Gott,) dass der nicht verschlafen my God that he not overslept lit. My God, that he didn’t oversleep! b. optative reading: [S → speaker’s preferences]

hat! has

c. polar exclamative reading: [S → prior unlikelihood according to the speaker]  A matrix clause deletion approach (which I view as an Ex-Con approach): (121) a. that-polar exclamative Mein Gott, [es überrascht mich, [dass der nicht verschlafen my God it surprises me that he not overslept ‘My God, I’m surprised that he didn’t oversleep!’ b. that-optative Mein Gott, [ich hoffe, [dass der nicht verschlafen hat]]! my God I hope that he not overslept has ‘My God, I hope that he didn’t oversleep!’

hat]]! has

 The option of deriving expressive utterances from (descriptive) proposition by virtue of a generalized EX operator allows us to view optatives and polar exclamatives as utterances that directly express (rather than describe) an emotion (Rosengren 1993). (122) Broader Cognition: Grammar:

Cognition Utterances without EX

Emotion Utterances with EX

 I have already argued against a matrix clause deletion approach. See appendix 3 for a review of Biezma’s (2011ab) approach, which also classifies as an Ex-Con approach. - 44 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

6. Acknowledgments I am especially grateful to my co-advisors Kai von Fintel and Sabine Iatridou, and to my committee members Irene Heim and David Pesetsky. I also wish to thank Danny Fox, Martin Hackl and Norvin Richards for numerous discussions and extensive comments that have contributed to this dissertation project. This project has also benefited from exchanges with Adam Albright, Josef Bayer, Sigrid Beck, Bronwyn Bjorkman, Gennaro Chierchia, Cleo Condoravdi, Norbert Corver, Ev Fedorenko, Hans-Martin Gärtner, Ted Gibson, Caroline Heycock, Gerhard Jaeger, Angelika Kratzer, Manuel Križ, Sven Lauer, Victor Manfredi, Diane Massam, Lisa Matthewson, Jason Merchant, Rick Nouwen, Pritty Patel-Grosz, Masha Polinsky, Paul Portner, Donca Steriade, David Stifter, Josh Tenenbaum, Hubert Truckenbrodt, Rob Truswell, Igor Yanovich and Ede Zimmermann. For the translations and judgments that underlie the cross-linguistic data in this presentation, I would like to thank: Sarah Ouwayda (Lebanese Arabic), Eulàlia Bonet (Catalan); Martina Gračanin-Yuksek (Croatian); Ivona Kučerová, Radek Šimík (Czech); Norbert Corver, Marc van Oostendorp, Erik Schoorlemmer, Coppe van Urk (Dutch); Mikko Kupula (Finnish); Lobke Aelbrecht (Flemish); Lila Daskalaki, Marios Mavrogiorgos, Dimitris Michelioudakis (Greek); Micha Breakstone, Hadas Kotek (Hebrew); Roberta D’Alessandro, Giuliano Bocci, Alessia Defraia, Michelangelo Falco, Gaetano Fiorin, Ilaria Frana, Daniele Panizza (Italian); Ingrid Lossius Falkum, Sverre Stausland (Norwegian); Grzegorz Michalski, Bozena Rozwadowska, Bartosz Wiland (Polish); Ana Bastos-Gee, Cynthia L. Z. DeRoma, Ana Paula Quadros Gomes (Portuguese); Gianina Iordachioaia, Andreea Nicolae (Romanian); Aysa Arylova, Natasha Ivlieva, Alexander Podobryaev, Igor Yanovich (Russian); Miloje Despić (Serbian); Luis Alonso-Ovalle, Paula Menendez-Benito (Spanish).

7. Appendices 7.1

Appendix 1: Further Arguments against Matrix Clause Deletion

 Scholz (1991) and the unembeddability generalization: ⇒

Scholz observes that optatives can never be truly embedded, shown in (123b). (Apparent embedded optatives must therefore be parenthetical constructions.)

(123) a. Ach, wenn es doch nur mich oh if it DOCH only me ‘Oh, if only I had been hit!’ - 45 -

getroffen hätte! hit had

Patrick G. Grosz

Dissertation Defense

b. Ach, wenn es (*doch nur) mich getroffen hätte, wäre oh if it DOCH only me hit had were schlimm gewesen. bad been ‘Oh, if only I had been hit, it would be half as bad.’ (Scholz 1991:9) ⇒

June 30, 2011

das halb so that half as

Concern: While Scholz is generally right, there are some marginally acceptable counterexamples to the unembeddability generalization.

 Rifkin (2000) and the optative/conditional asymmetries: ⇒ Optatives have a more limited distribution than conditionals with consequents. (124) a. [If Sue had time, she would ski Mt. McKinley], and [if she had money, things would be good]. b.* [If Sue had time, she would ski Mt. McKinley], and [if only she had money]. (based on examples from Rifkin 2000; modified to create a minimal pair) (125) a. Avi thinks that [if it would snow, things would be good]. b.* Avi thinks that [if only it would snow]. (Rifkin 2000) ⇒

Concerns: It is not clear to what extent these facts argue against matrix clause deletion. Conjoining an elliptical construction with something else is not always felicitous and answer fragments cannot be embedded either.

(126) Context: The hearer just picked up a poisonous mushroom and wants to taste it. a. Don’t! ? b. * Don’t, and wash your hands! c. (?) Don’t put it in your mouth, and wash your hands! (127) a. A: Under wich circumstances would the party have been fun? B: (The party would have been fun) if John had come. b. A: Under wich circumstances would the party have been fun? B: Avi thinks that [the party would have been fun if John had come]. c. A: Under wich circumstances would the party have been fun? B: * Avi thinks that [if John had come]. - 46 -

Patrick G. Grosz

7.2

Dissertation Defense

June 30, 2011

Appendix 2: The formal semantics of the prototypical particles

 The formal semantics of nur ‘only’, wenigstens ‘at least’ and doch are given below in more detail. (128) a. Wenn Otto nur auf seine Mutter gehört hätte! if Otto only to his mother listened had ‘If only Otto had listened to his mother!’ b. presuppositions of nur: This is not much to ask for (i.e. it’s low on some scale). c. formalization: ||only2,C|| = λS.λp : MOST q ∈ g(C) [q >S p] . LOWNESS p

IDENTITY

“only2 is a truth-conditionally vacuous element (different from canonical only), which triggers a presupposition that the modified proposition is low on a contextually provided scale.”

(loosely based on Guerzoni’s 2003 nur2; I assume that only is always scalar, cf. Jacobs 1983, Bayer 1996, Klinedinst 2004, 2005, Krasikova & Zhechev 2006, Riester 2006, Beaver & Clark 2008) (129) a. Wenn Otto wenigstens auf seine Mutter gehört hätte! German if Otto at.least to his mother listened had ‘If Otto had at least listened to his mother!’ b. presuppositions of wenigstens: This is tolerable but there is a better option. c. formalization: ||wenigstensC||g,c = λS.λp : S is a bouletic ordering ∧ BOULETIC ∃r ∈ g(C) [r >S p] ∧ ∃q ∈ g(C) [p >S q] . SECOND CHOICE p IDENTITY “wenigstens is a truth-conditionally vacuous element (corresponding to English concessive at least), which combines with a bouletic scale and presupposes that there is a contextually salient proposition that is more preferable than the modified proposition, as well as a contextually salient proposition that is less preferable.”

(based on Nakanishi & Rullmann’s 2009 concessive at least) (130) a. Wenn Otto doch auf seine Mutter gehört hätte! if Otto DOCH to his mother listened had ‘If only Otto had listened to his mother!’ b. presuppositions of doch: This is in sharp contrast (i.e. contradiction) to reality. - 47 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

c. formalization: ||dochC||g,c,w = λp : ∃q ∈ g(C) [p ≠ q & ¬[p(w) ∧ q(w)]] ∧ p ∩ Doxspeaker(w) = ∅ ∨ ¬p ∩ Doxspeaker(w) = ∅ .

CONFLICT FAMILIARITY

p

IDENTITY

“doch is a truth-conditionally vacuous element, which triggers a presupposition that the truth/falsity of the modified proposition is established and that the modified proposition conflicts with some contextually salient proposition.”

(based on Grosz 2010a, Kratzer & Matthewson 2009) 7.3 Appendix 3: A Brief Review of Biezma (2011ab) 7.3.1 The Core Ideas In brief: Biezma derives desirability from an implicit Question under Discussion.  Biezma (2011ab) argues that (131a) is a true conditional and that (131b) is also a conditional, but one where the consequent has been omitted. (131) a. If only I had been taller, I would have played in the NBA. b. If only I had been taller! (adapted from Biezma 2011a)  She assumes that optatives like (131b) have a structure as in (132); they are type expressions. (‘⇒’ is a place-holder for the semantics of the modal ψ.) (132)

λq.p ⇒ q qo i qp qp qi ψ 6 antecedent

(Biezma 2011b:116)

 She argues for the following pragmatic derivation of desirability: ⇒

Optatives are conditionals where an pro-form in the consequent position is abstracted over, creating an expression. - 48 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

⇒

Furthermore, they are conditionals with reverse topicality, i.e. the antecedent is the information-structural focus and the elided consequent is the topic.

⇒

What is actually desired is the (elided) consequent, and not the (overt) antecedent.

⇒

Particles such as only (and at least) have two functions: First, they select the entire antecedent proposition as focus; second, under the assumption that optatives are answers to an implicit question under discussion, they indicate that the optative is a mention-some answer.

⇒

Desirability is derived from the idea that mention-some answers can felicitously answer goal-oriented questions, (133), but not neutral questions, (134), which require exhaustive answers. Desirability is thus inferred from the implicit question under discussion, which must be goal-oriented.

(133) a. How do we bring β about? b. How would we have brought β about? c. How do I get to [β the supermarket / play in the NBA / #die]? ⇒

desire for β is implied/entailed

(partially adapted from Biezma 2011a) (134) What are the circumstances that would bring about β? (Biezma 2011a) ⇒ desire for β is not implied/entailed; β can be negative/neutral 7.3.2

The Main Concerns

Now: It is unclear how to derive the difference between optatives and answer fragments.  First concern: It is not clear how the distinction between overt questions and implicit questions under discussion derives. ⇒ We have seen that optatives differ from question fragments in that question fragments provide an adjunction site above the elided matrix clause, which is absent in optatives. (135) a. A: Unter welchen Umständen wäre die Party ein Erfolg gewesen? under which circumstances were the party a success been ‘Under which circumstances would the party have been a success?’ - 49 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

B: Wenn Hans gekommen wäre weil er immer guten Wein if Hans come were since he always good wine mitbringt. brings ‘If Hans had come, because he always brings good wine.’  because he always brings good wine > [The party would have been a success if Hans had come] LF: [The party would have been a success if Hans had come], because he always brings good wine. b. Wenn Hans doch nur gekommen wäre (#weil er immer if Hans DOCH only come were because he always guten Wein mitbringt). good wine brings ‘If only Hans had come (#because he always brings good wine).’ Now: It is not evident that the locus of the wish is really exclusively in the consequent.  Second concern: It is not clear that the locus of the wish is really exclusively in the consequent (and not in the antecedent). ⇒ To argue that a speaker desires the consequent and not the antecedent, Biezma presents the following example. (136) A: B:

A:

⇒

If only I had been taller, I would have played in the NBA That would not have been necessary, you were such a great player! What would have made a difference was if you had been in a better college team. Yeah…!, you are right…, If only I had played for UCLA, I would have played in the NBA (Biezma 2011a)

But explicit wish-statements do not necessarily behave differently.

(137) A:

I wish I had been taller. Then I would have played in the NBA

B:

That would not have been necessary, you were such a great player! What would have made a difference was if you had been in a better college team.

A:

Yeah…!, you are right…, I wish I had played for UCLA. Then I would have played in the NBA. - 50 -

Patrick G. Grosz

⇒

Dissertation Defense

June 30, 2011

In contrast, the optative itself does seem to convey a wish, which cannot be canceled even if there is a desirable consequent.

(138) Context: I love snowboarding and I want to go snowboarding as often as possible. a. If it snowed tonight, we would go snowboarding tomorrow … but I really don’t want it to snow, because I hate shoveling the sidewalks. b. If only it snowed tonight, we would go snowboarding tomorrow … #but I really don’t want it to snow, because I hate shoveling the sidewalks. c. If it snowed tonight, we would go snowboarding tomorrow, and I wish we would go snowboarding tomorrow … but I really don’t want it to snow, because I hate shoveling the sidewalks. Now: There are concerns with respect to analyzing optatives as focused constituents.  Third concern: It is not clear that the information-structural reasoning of Biezma’s is correct. ⇒

In German, conditionals that truly integrate the antecedent (i.e. conditionals in which the antecedent occupies the pre-verb-second SpecCP position) typically only have one focused element, Reis & Wöllstein (2010).

(139) a. Under which circumstances would you drive a Bentley? [Wenn ich MillioNÄR wäre]F, würde if I millioNAIRE were would ‘If I were a millionaire, I’d do it.’ b. What would you do if you were a millionaire?

ich I

Wenn ich Millionär wäre, [würde ich if I millionaire were would I ‘If I were a millionaire, I’d drive a bentley.’ (Reis & Wöllstein 2010:148) ⇒

es it

tun. do

BENTley fahren]F. BENTley drive

Optatives with an (apparent) overt ‘consequent’ typically have separate focusbackground structuring in the ‘antecedent’ and in the ‘consequent’. This seems connected to the unembeddability of optative clauses.

(140) a.* Wenn ich doch nur MillioNÄR wäre, würde if I doch only millioNAIRE were would ‘If only I were a millionaire, I’d drive a bentley.’ - 51 -

ich I

Bentley fahren. Bentley drive.

Patrick G. Grosz

Dissertation Defense

b. Wenn ich doch nur MillioNÄR wäre, ich if I doch only millioNAIRE were I ‘If only I were a millionaire, I’d drive a bentley.’ c. Wenn ich doch nur MillioNÄR wäre, dann if I doch only millioNAIRE were then ‘If only I were a millionaire, I’d drive a bentley.’ ⇒

June 30, 2011

würde would

BENTley fahren. BENTley drive.

würde ich BENTley fahren. would I BENTley drive.

Deaccentuation of an overt ‘consequent’ is not possible if the antecedent is an optative, which is different from regular conditionals.

(141) a. Under which circumstances would you drive a Bentley? Wenn ich MillioNÄR wäre, dann würde ich Bentley fahren. if I millioNAIRE were then would I Bentley drive ‘If I were a millionaire, I’d drive a Bentley.’ b.* Wenn ich doch nur MillioNÄR wäre, dann würde ich Bentley fahren. if I doch only millioNAIRE were then would I Bentley drive. ‘If only I were a millionaire, I’d drive a Bentley.’ ⇒

Conversely, in the rare counter-examples to the unembeddability generalization that I mentioned before, the required intonation pattern seems to go counter to the pattern that Biezma argues for.

(142) a. [Wenn ich doch nur könnte], würde ich if I DOCH only could would I ‘If only I could, I would come immediately.’ b.* [Wenn ich doch nur KÖNnte], würde ich if I DOCH only could would I ‘If only I could, I would come immediately.’

soFORT immediately

kommen. could

sofort

kommen.

immediately

could

Now: The status of the consequent as topic is debatable as it is not always retrievable.  Biezma’s theory assumes that the consequent (which is the topic and embedded in the Immediate Question under Discussion) is always active and retrievable. ⇒ It is not completely evident that this is the case. ⇒

In the following example, the speaker is hard-pressed to provide a relevant consequent though it is clear that a positive evaluation is expressed.

- 52 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

(143) Context: The apartment is a mess. There are lots of things I can't find. Currently, I'd like to send back a Netflix DVD, but I can't find the shipping envelope. I haven't seen it for days. Out of instinct I lift a book that's lying in the middle of a heap of stuff and – to my great surprise – I find the envelope under it. DAS war jetzt aber einfach! Wenn nur ALLES so einfach wär! that was now but easy if only all so easy were ‘My, that was easy! If only everything would be that easy!’ ⇒

When asked what the consequence would be, it is natural to come up with one of the following, but none of them seems quite natural in this context.

(144) a. ?? … b. ?? … c.?? … d. ?? … e. ?? … f. ?? … ⇒

then I would be happy. then my life would be much easier. then I would have everything I want. then my apartment wouldn’t be this messy. then my wishes would come true (cf. Biezma 2011b:118) then I would save so much time.

This calls into question the idea that consequents are always salient / retrievable.

8. References Asarina, Alya & Kirill Shklovsky. 2008. Optativity in English and Other Languages. Handout for MIT Ling-Lunch. Bayer, Josef. 1996. Directionality and Logical Form. On the Scope of Focusing Particles and Wh-in-situ. Dordrecht: Kluwer. Beaver, D., and B. Clark. 2008. Sense and Sensitivity: How Focus Determines Meaning. Oxford: Blackwell. Biezma, Maria. 2011a. Optatives: deriving desirability from scalar alternatives. SuB 15. Biezma, Maria. 2011b. Anchoring Pragmatics in Syntax and Semantics. PhD dissertation, UMass / Amherst. Bjorkman, Bronwyn. 2010. ‘The Syntax of Inverted Conditional Antecedents’. LSA handout. Büring, Daniel. 2008. The least at least can do. In WCCFL 26. 114-120. Coppock, Elizabeth, and David Beaver. 2011. Mere-ology. Ms, Lund / UT Austin. - 53 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Delsing, Lars-Olof. 2010. Exclamaives in Scandinavian, Studia Linguistica 64, 16-36. Evans, Nicholas. 2007. Insubordination and its uses. In Finiteness. Theoretical and Empirical Foundations, ed. by Irina Nikolaeva. Oxford: Oxford University Press. Gärtner, Hans-Martin. 2010. ‘SAI and NPI-Licensing: The Case of Optatives’, handout for Oberseminar “Neuere Arbeiten zu Syntax und Semantik”, Universität Tübingen. Geurts, Bart, and Rick Nouwen. 2007. At least et al.: The semantics of scalar modifiers. Language 83, 533-559. Giorgi, Alessandra. 2009. ‘Toward a syntax of the subjunctive mood’, Lingua 119, 18371858. Giorgi, A., Pianesi, F., 1997. Tense and Aspect: From Semantics to Morphosyntax. Oxford University Press, New York. Giorgi, A., Pianesi, F., 2004. Complementizer deletion in Italian. In: Rizzi, L. (Ed.), The Structure of CP and IP. Oxford University Press, New York, pp. 190–210. Grosz, Patrick. 2010a. German doch. An element that triggers a contrast presupposition. CLS 46. Grosz, Patrick. 2010b. Facts and Ideals: On the Relationship Between Conditionals and Optatives. SuB 15. Guerzoni, Elena. 2003. Why Even Ask? On the pragmatics of questions and the semantics of answers. PhD dissertation, MIT. Gutiérrez Rexach, Javier. 1996. ‘The semantics of exclamatives’. In Syntax at sunset. UCLA Working Papers in Linguistics, ed. by E. Garret and F. Lee. UCLA. Heim, Irene. 1992. ‘Presupposition projection and the semantics of attitude verbs’, Journal of Semantics 9, 183-221. Iatridou, Sabine, 2000. The grammatical ingredients of counterfactuality. Linguistic Inquiry 31, 231–270. Jacobs, J. 1983. Fokus und Skalen: Zur Syntax und Semantik der Gradpartikeln im Deutschen. Tübingen: Niemeyer. Kempchinsky, P., 1986. Romance subjunctive clauses and logical form. Ph.D. dissertation, UCLA. Klinedinst, Nathan. 2004. ‘Only Scalar Only’. Handout for the –< Presupposition & Implicature Workshop, October 5, 2004, Paris. Klinedinst, Nathan. 2005. ‘Scales and Only’. Unpublished generals paper, UCLA. Krasikova, Sveta and Ventsislav Zhechev. 2006. “You Only Need a Scalar ‘Only’ ”, in C. Ebert and C. Endriss (eds.) ZAS Papers in Linguistics 44, Proceedings of Sinn und Bedeutung 10, Berlin.

- 54 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Kratzer, A. 1999. Beyond Ouch and Oops. How descriptive and expressive meanings interact. Handout for Cornell Conference on Context Dependency. Cornell University. Kratzer, A., and L. Matthewson. 2009. Anatomy of two discourse particles. Handout for SULA 5. Harvard/MIT. Kratzer, Angelika. 1991. Modality. In Semantics: An International Handbook of Contemporary Research, ed. Arnim von Stechow and Dieter Wunderlich, 639-650. Berlin / New York: de Gruyter. Krifka, Manfred. 1999. At least some determiners aren’t determiners. In K. Turner ed., The Semantics/Pragmatics Interface from Different Points of View, 257-291. Oxford: Elsevier. Kyriakaki, Maria. 2007. The meaning of Na and Conditional Wishes. Unpublished generals paper, University of Toronto. Kyriakaki, Maria. 2008. The meaning of NA and Conditional Wishes. Presented at the Canadian Linguistic Association. University of Vancouver, Vancouver. May 31 – June 2, 2008. Kyriakaki, Maria. 2009. When desiderative exclamations meet the conditionals: The case of Greek. Paper presented at Chronos 9th, International Conference on Tense, Aspect and Modality, Paris Diderot University, Paris, France. September 2 – 4, 2009. Lewis, David. 1969. Convention. Cambridge, MA: Harvard University Press. Nakanishi, Kimiko, and Hotze Rullmann. 2009. Epistemic and Concessive Interpretations of At Least. CLA, Carleton University, May 24, 2009. Portner, Paul. 2006. Dependent Contexts in Grammar and in Discourse: German Verb Movement from the Perspective of the Theory of Mood Selection. Theoretical Linguistics 32, 353-368. Potts, Chris. 2007. The expressive dimension. Theoretical Linguistics 33, 165–198. Potts, Christopher, and Tom Roeper. 2006. The Narrowing Acquisition Path: From Expressive Small Clauses to Declaratives. In Ljiljana Progovac, Kate Paesani, Eugenia Casielles, Ellen Barton (eds.): The Syntax of Nonsententials: MultiDisciplinary Perspectives. John Benjamins. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford Studies in Theoretical Linguistics. Oxford: Oxford University Press. Quer, Josep. 1998. Mood at the Interface. The Hague: Holland Academic Graphics. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik. 1972. A Grammar of Contemporary English. London: Longman. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. - 55 -

Patrick G. Grosz

Dissertation Defense

June 30, 2011

Reis, Marga, and Angelika Wöllstein. 2010. ‘Zur Grammatk (vor allem) konditionaler V1-Gefüge im Deutschen.’ Zeitschrift für Sprachwissenschaft 29, 111-179. Riester, Arndt. 2006. Only Scalar. In Janneke Huitink & Sophia Katrenko (eds.): Proceedings of the Eleventh ESSLLI Student Session, 64-75. Rifkin, Jay. 2000. If only if only were if plus only. In CLS 36-1, 369–384. Rosengren, Inger. 1992. Zur Grammatik und Pragmatik der Exklamation. In Satz und Illokution Band 1, ed. by Inger Rosengren, 263-306. Tübingen: Niemeyer. Rosengren, Inger. 1993. Imperativsatz und “Wunschsatz” – zu ihrer Grammatik und Pragmatik. In Satz und Illokution Band 2, ed. by Inger Rosengren, 1-48. Tübingen: Niemeyer. Scholz, Ulrike. 1991. Wunschsätze im Deutschen – Formale und funktionale Beschreibung. Satztypen mit Verberst- und Verbletztstellung. Tübingen: Niemeyer. Szabolcsi, Anna. 2004. ‘Positive Polarity – Negative Polarity’, NLLT 22, 409-452. Truckenbrodt, Hubert. 2006a. ‘On the semantic motivation of syntactic verb movement to C in German’. Theoretical Linguistics 32, 257-306. Truckenbrodt, Hubert. 2006b. Replies to the comments by Gärtner, Plunze and Zimmermann, Portner, Potts, Reis, and Zaefferer. Theoretical Linguistics 32, 387410. Villalta, Elisabeth. 2007. Context Dependence in the Interpretation of Questions and Subjunctives. PhD dissertation, Universität Tübingen.

- 56 -