Epistemic Containment Revisited

EPISTEMIC CONTAINMENT REVISITED

Harris Constantinou UCL - University of Leeds [email protected]

Hans Van de Koot UCL [email protected]

NELS 45 @ MIT 31/10/2014

The Epistemic Containment Principle (ECP)

QPs and epistemic modal auxiliaries - ECP consistent 1.  Every student may have left (# but not everyone has).

(may > every; *every > may) 2.  Fewer than half the students must have passed, (# but

perhaps all of them did). (must > fewer than half; *fewer than half > must)

Descriptive version A quantifier [(i.e. QP)] cannot have scope over an epistemic modal.

(Von Fintel & Iatridou 2003: 174)

The Epistemic Containment Principle (ECP)

(Raised) QPs and epistemic modal raising predicates - ECP inconsistent

3.  Fewer than half the students are certain/likely to have passed (✓ but perhaps all of them did). (fewer than half > certain/likely)

4.  Fewer than half the students seem/appear to have passed

(✓ but perhaps all of them did). (fewer than half > seem/appear)

Preliminary conclusion: The ECP does not obtain in all cases in which the QP and the epistemic modal head are in the same clause.

Overview

•  Encoding the scope of quantificational elements; deriving the R(evised)ECP

•  Word order restrictions between deontic and epistemic modal

heads •  Scoping out of infinitival clauses

Encoding the scope of quantificational elements; deriving the R(evised)ECP

Scope extension mechanism (Williams 1994) Covert scope extension represented at LF as percolation of an index originating in the quantificational element. The scope of such an element corresponds to the largest constituent carrying its index, minus the element itself. (N.B. no scope extension through covert movement) QP scope taking a.  An in-situ QP takes scope over its c-command domain (default

scope rule). b.  An in-situ QP may also extend its scope by percolating an index

to a dominating node. Head scope taking c.  A head takes scope over the phrase it projects, and nothing else,

via percolation of an index to the top node of the phrase it projects (Williams’s Head scope rule).

a) QP default scope rule

QP

α

ε δ

γ

ζ η

β

α

ε:1 δ

γ:1

ζ QP1

β

α

ζ’:1 δ

ζP:1

ζ1 η

b) QP scope extension c) Head scope rule

Inherited indices, as opposed to non-inherited ones, appear after a colon

Condition on Scope Shift (CSS)(Neeleman & Van de Koot 2012)

No node may inherit more than one scope index.

β

α

ε:2 QP1

γ:1:2

ζ QP2

*

CSS only compatible with the view that LF merely represents deviations from surface scope (Reinhart 2006).

a.  [:1…[:1,:2…QP1…[:2…QP2]]] QP1 > QP2

b.  [:2…[:1,:2…QP1…[:2…QP2]]] QP2 > QP1

c.  […[…QP1…[…QP2]]] QP1 > QP2

d.  [:2…[:2…QP1…[:2…QP2]]] QP2 > QP1

CSS incompatible

CSS compatible

Predictions i) R(evised)ECP: A quantificational element α cannot have

scope over an epistemic head β, if α is in the projection of β. ii) Limited Scope extension: An epistemic modal head cannot

extend its scope higher than the phrase it projects. iii) Intervention: An epistemic modal head blocks scope

extension of a lower quantificational element.

Claim: Epistemic modal heads obey the head scope rule.

RECP (prediction i) McDowell 1987, Brennan 1993 (i.a.) show that an epistemic modal auxiliary head merges as the highest head below C(omplementiser). Under standard assumptions, subjects occur in the specifier position of the highest head below C. 2.  Fewer than half the students must have passed.

(must > fewer than half; *fewer than half > must)

C

C’

Aux’

AuxP

QP Fewer than half the

students Aux must

RECP (prediction i)

C

C’

Aux’:1

AuxP:1

QP Fewer than half

t. s. Aux must1

C

C’:2

Aux’:1

AuxP:1:2

QP Fewer than half

t. s. 2 Aux must1

Partial representation of (2) – inverse scope.

Partial representation of (2) – surface scope: CSS violation.

*

Must scopes over the QP due to index percolated up to AuxP and AuxP dominating the QP.

Any attempt of the QP to extend its scope results in CSS violation.

RECP (prediction i) Arguably, raising predicates, such as likely (see (3) and (4)), do not merge as highest heads in the extended verbal projection. Raised subjects are thus not their specifiers. 3.  Fewer than half the students are certain/likely to have

passed (✓ but perhaps all of them did). (fewer than half > certain/likely)

Partial representation of (3)

The QP outscopes likely by the default scope rule (c-command)

C

C’

Aux’

AuxP

QP Fewer than half

t. s. Aux are

AdjP:1

Adj likely:1


scope over an epistemic head β, if α is in the projection of β. ✓ ii) Limited scope extension: An epistemic modal head cannot

extend its scope higher than the phrase it projects. iii) Intervention: An epistemic modal head blocks scope


Word order restrictions between deontic and epistemic modal heads

Limited scope extension (prediction ii)

Epistemic modal heads invariably merge higher than deontics, cross-linguistically (Cinque 1999). *deontic>epistemic hierarchical order constraint illustrated

5.  * John is required to be likely to win. 6.  John is likely to be required to win.

Vikner 1988, Cormack & Smith 2002 (i.a.) suggest that the *deontic>epistemic ordering constraint is due to a conceptual restriction that forces a deontic interpretation to scope below an epistemic one.

But, epistemic modal heads can extend their scope (see (1) and (2)). Why can’t they do it in (5) in order to satisfy this conceptual restriction?


Two competing theories

* 21

TP

VP

T’

T to

V be

AdjP:1

Adj likely1

V required

VP

TP:1

VP:1

T’:1

T to

V be

AdjP:1

Adj likely1

V2 required

VP:1, :2


Prediction of theory 2: A deontic head should block scope extension of quantificational elements that we know can extend their scope (i.e. QPs).

7.  A doctor will make sure that we give every new patient tranquilizer. (∃>∀; ∀>∃) (Reinhart 1997:350)

8.  A doctor will make sure that we are required to give every new patient tranquilizer (by circulating the instructions for medication at the beginning of the shift).

(∃>∀; ∀>∃)

Conclusion: The fact that a deontic head does not block QP scope extension is consistent with theory 1 only (i.e. limited scope extension for epistemic heads).


scope over an epistemic head β, if α is in the projection of β. ✓ ii) Limited scope: An epistemic modal head cannot extend its

scope higher than the phrase it projects. ✓ iii) Intervention: An epistemic modal head blocks scope


Scoping out of infinitival clauses

Intervention (prediction iii)

Usually, a QP may scope out of an infinitival clause in English.

9.  A different student tried/wanted to read every book. 10.  At least one American tourist hopes to visit every

European country this year. 11.  A different student decided to report on every article

on the reading list.

12.  At least one professor believes Mary to have read every book.

13.  Some actress believed every movie to be interesting. 14.  Someone expects Sue to marry every boy.

For all of the above, it has been reported in the literature (Kennedy 1997, Wurmbrand 2011, Lechner 2012) that ∀ may scope over ∃.

control

ECM


Partial representation for the inverse scope reading of (13)

TP:1

VP:1

T’:1

T

V believed

TP:1

QP every movie1

QP Some actress

Intervention (prediction iii) If a QP is found in the (infinitival) complement clause of an epistemic modal head, it should not be able to extend its scope past that head (as this would create a CSS violation). 15.  There’s likely to be (exactly) one student absent.

(likely > ∃; *∃ > likely) Bobaljik and Wurmbrand (2012)

16.  There seems to be someone from NYC at John’s parties.

(seem > ∃; *∃ > seem) Bobaljik and Wurmbrand (2008)


Partial representation of (15) – surface scope

*

TP

VP

T’

T to

V be

AP

QP one

student

A likely1

AP:1

A absent

Partial representation of (15) – inverse scope: CSS violation

TP:2

VP:2

T’:2

T to

V be

AP:2

QP one

student2

A likely1

AP:1:2

A absent

Concluding remarks Assumptions: a.  (Covert) scope extension represented at LF as index percolation. b.  Minimality constraint on (covert) scope extension; CSS c.  Epistemic modal heads obey Williams’s (1994) head scope rule.

Facts corroborating each prediction:

Ability of QPs to outscope an epistemic raising predicate but not an epistemic auxiliary.

*deontic – epistemic hierarchical order constraint.

(In)ability of QPs to scope out of infinitival clauses.

Predictions:

RECP

Limited scope extension

Intervention

References •  Bobaljik, J.D. and S. Wurmbrand. 2008. Word order and scope:

Transparent Interfaces and the 3/4 signature. (draft paper) •  Bobaljik, J.D. and S. Wurmbrand. 2012. Word order and scope:

Transparent Interfaces and the 3/4 signature. Linguistic Inquiry 43.3:371-421.

•  Brennan, V. 1993. Root and Epistemic modal auxiliary verbs. Ph.D. Thesis, University of Massachusetts, Amherst.

•  Cinque, G. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford Studies in Comparative Syntax. Oxford and New York: Oxford University Press.

•  Cormack, A. and N. Smith. 2002. Modals and negation in English. In: Barbiers, S., Beukema, F., van der Wurff, W. (Eds.), Modality and its Interaction with the Verbal System. John Benjamins, Amsterdam: pp. 133–164.

•  Fintel, K.V., and S. Iatridou. 2003. “Epistemic Containment.” Linguistic Inquiry 34 (2): 173–98.

•  McDowell, J.P. 1987. Assertion and Modality. PhD dissertation, University of Southern California.

•  Kennedy, C. 1997. Antecedent-contained deletion and the syntax of quantification. Linguistic Inquiry 28 :662–688.

References •  Neeleman, A., and H. Van De Koot. 2012. “Towards a Unified Theory

of Contrast and Scope.” In The Syntax of Topic, Focus, and Contrast: An Interface-Based Approach, edited by Ad Neeleman and Reiko Vermeulen, 39–76. Berlin: Mouton de Gruyter.

•  Reinhart, T. 1997. Quantifier scope: How labor is divided betweenQR and choice functions. Linguistics and Philosophy 20:335–397.

•  Reinhart, T. 2006. Interface Strategies. Cambridge, MA: MIT Press. •  Vikner, S. 1988. Modals in Danish and event expressions. Working

Papers in Scandinavian Syntax 39, 1–33. Lund University. •  Williams, E. 1994. Thematic Structure in Syntax. Cambridge, MA: MIT

Press. •  Wurmbrand, S. 2011. QR and selection: Covert evidence for

phasehood. Proceedings of NELS 42, University of Toronto.


This account explains why, contrary to the usual case, a QP cannot extend its scope past an infinitival clause boundary when this clause is the complement of an epistemic raising head.

Further prediction: A raising predicate without an epistemic flavor should not block scope extension.

7.  A doctor will make sure that we give every new patient tranquilizer. (∃>∀; ∀>∃) (Reinhart 1997:350)

17.  A doctor will make sure that we are likely to give every new patient a tranquilizer (by only unlocking the cupboard with the tranquilizers). (∃>∀; *∀>∃)

18.  A doctor will make sure that we happen to give every new patient a tranquilizer (by instructing us, off the record, to do so). (∃>∀; ∀>∃)

Appendix

Epistemic Containment Revisited

Documents