West Coast Conference on Formal Linguistics (WCCFL) (38th : 2020)

A Flexible Scope Theory of Intensionality Elliott, Patrick D. 2020-03-07

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


73804-Elliot_P_Scope_WCCFL38_2020_handout.pdf [ 123.49kB ]
JSON: 73804-1.0389851.json
JSON-LD: 73804-1.0389851-ld.json
RDF/XML (Pretty): 73804-1.0389851-rdf.xml
RDF/JSON: 73804-1.0389851-rdf.json
Turtle: 73804-1.0389851-turtle.txt
N-Triples: 73804-1.0389851-rdf-ntriples.txt
Original Record: 73804-1.0389851-source.json
Full Text

Full Text

A Flexible ScopeTheory of Intensionality*Patrick D. Elliott†March 7, 2020 wccfl 38, ubcDownload a copy of the handout here:https://patrl.keybase.pub/handouts/wccfl38.pdf1 Roadmap• Fact: predicates are world-sensitive; in an intensional context, DPs may beinterpeted de re or de dicto:(1) George wantsintensional context⏞⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⏞⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⏞[ the Red Sox players to win the game].– De re: interpretation, Red Sox player is interpreted relative to the ut-terance evaluation world.– Dedicto: Red Sox player is interpeted relative toGeorge’swant-worlds.• Two broad camps in accounting for world sensitivity – the Binding Theoryof Intensionality (bti)1 and the ScopeTheory of Intensionality (sti).2• The bti is powerful, but must be supplemented with a binding theory forworld variables. The sti is much more restrictive, but (seemingly) under-generates.*Particular gratitude to Keny Chatain, Kai von Fintel, Julian Grove, Patrick Niedzielski, RogerSchwarzchild and three anonymous wccfl reviewers for comments that improved this work.Thanks are also due to attendee’s of the 02/26 LF Reading Group.†mit; pdell@mit.edu1Percus (2000), Heim & von Fintel (2011: chapter 8), etc.2Heim & von Fintel (2011) call this the standard theory.• Scope theory state of the art – Keshet’s split intensionality3 – succeeds in ad-dressing some of the worst over-generation issues, but others remain.• Concretely, Grano (2019) shows that the account of exceptional de re forindefinites runs into insurmountable obstacles.• I’ll aim to improve on split intensionality by presenting a new take on the sti– which I’ll call the flexible scope theory – whereby expressions can receiveexceptional de re interpretations via recursive scope-taking, facilitated by aminimal inventory of type-shifters.• The flexible scope theory will preserve a central claim of split intensionality– de re requires movement to an edge position.• The resulting theorywill bear a (non-accidental) family resemblance toChar-low’s (2014, 2019) theory of exceptionally-scoping indefinites.2 Split intensionality• The scope theory says, roughly, that an expression is interpreted de dicto ifit scopes below an intensional operator, and de re if it scopes above an inten-sional operator.• One immediate problem for the scope theory is the fact that scope islandsdo not always block de re interpretations.4(3) George thinksscope island⏞⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⏞⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⎴⏞[ every Red Sox player is staying in the Ritz-Carlton].• Keshet’s split intensionality theory is tailored to circumvent this problem.• Keshet assumes that embedded clauses basically denote extensions, and at-titude verbs are looking for intensions – in order to repair this mismatch, atype-shifter ∧ is inserted at the clause-edge.3Keshet 2008, 20114This is of course no problem for the bti, assuming free insertion of an abstraction index a the edgeof the matrix clause.(2) 1 George thinks every Red Sox player𝑤1 is staying in the Ritz Carlton.1• A QPmay be interpreted de re, while nevertheless receiving narrow quantifi-cational scope, by QR-ing to a position above ∧ but below the attitude verb.(4) George thinksevery Red Sox player 𝜆𝑥 ∧ [𝑡𝑥 is staying in the Ritz-Carlton].• In effect, ∧ serves to create a privileged position at the clause edge in whichQPs can be exceptionally interpreted de re.2.1 Problem 1: doubly-embedded de re• We can note already that a straightforward prediction of split intensionalityis that an expression can only be interpeted de re relative to the minimallycontaining scope island.• At face value, it is clear that this generalization doesn’t hold, as pointed outby Grano. Consider the following example from Grano (2019: p. 162):(5) a. There is a group of people in this room. Neither Jo nor Mary know thatthey’re in this room. Mary hopes they’re actually outside. She reports herhope to Jo, and Jo believes her.b.3 Jo thinks [that Mary hopes [that everyone in this room is outside]].• So, it seems like there are simple cases in which split intensionality isn’t suf-ficiently general.55Of course, this isn’t accidental, but rather a design feature of Keshet’s analysis. In support of this,Keshet observes (a) that counterfactuals with tautologous antecedents sound odd, and (b) if ex-ceptional de re were available, it should rescue embedded counterfactuals.(6) #If three professors were professors, the classes would be better taught.(7) #Mary thinks that if three professors were professors, the classes would be better taught.I don’t have much to say about these examples, except that they can be much improved withsome manipulation.(8) a. #If three syntacticians were linguists, the classes would be more fun.b. Mary thinks that the classes would be better taught if three syntacticians were linguists –she has no idea what they do!2.2 Problem 2: Bäuerle’s puzzle• Consider the following example:6(9) George thinks every Red Sox player is staying in some five star hotel down-town.• There is a reading with the following features:– Red Sox player is interpreted de re – George’s beliefs pertain to a groupof men who happen to be Red Sox players, potentially unbeknownst toGeorge.– Five star hotel is interpreted de dicto –George’s beliefs involve a five starhotel; the sentence may still be true even if there are no five star hotels,just so long as George believes that there are.– Some takes scope over every –George thinks that all the people in ques-tion are staying at the same five star hotel downtown.• If we take any version of the sti, such as split intensionality, this readingwould seem to place contradictory requirements on the scope of some fivestar hotel.– In order for some five star hotel to be interpreted de dicto, it shouldscope below ∧.(10) ∧ > some fsh– In order for every Red Sox player to be interpreted de re, it should scopeabove ∧.(11) every rsp > ∧– By transitivity, this means that every Red Sox player should have to takescope over some five star hotel – but this doesn’t capture the readingwe’re interested in. Whoops!• Keshet (2010) suggests that some five star hotel is a specific indefinite, andtherefore involves a choice-functional variable existentially bound.76From Keshet 2010: p. 692, ex 1, loosely based on an example in German from Bäuerle (1983).7A function 𝑓 is a choice function iff 𝑓 is of type (a → t) → a, and for any predicate 𝑃, 𝑓 𝑃 ∈ 𝑃2• Keshet’s innovation is the suggestion that the choice function returns amem-ber of the NP restrictor at the local evaluation world.• Together with the split intensionality hypothesis, this allows him to derive theproblematic reading with the following LF:(12) ∃𝑓George thinks every Red Sox player 𝜆𝑥∧ [𝑥 is staying in𝑓(five star hotel)].As demonstrated byGrano (2019), exceptionally-scoping indefinites can alsoreceive de re interpretations.8(13) a. Jo and Bill are out shopping. Bill finds a hat that he likes and considerspurchasing it. It so happens that the hat is just like mine, but neither Jonor Bill know this. Jo thinks that that the hat looks great on Bill and hopeshe’ll buy it.b.3 Jo hopes [that Bill will buy a hat just like mine ].• Keshet can account for this by QRing the restrictor of the indefinite to a po-sition above the intensionalizing operator at the edge of the scope island.(14) ∃𝑓 Jo hopes 𝑓(hat just like mine) 𝜆𝑥 ∧ [Bill will buy 𝑡𝑥]• This solution seems to work just fine, but as pointed out by Grano 2019, itwon’t generalize to cases involving more deeply embedded scope islands.9(15) a. Mary, Jo, and Bill are out shopping. Bill finds a hat that he likes and con-siders purchasing it. It so happens that the hat is just like mine, but neitherMary, nor Jo, nor Bill know this. Jo thinks that the hat looks great on Billand hopes he’ll buy it. Jo expresses her hope out loud, and Mary believesJo.b. Mary thinks [that Jo hopes [that Bill will buy a hat just like mine ]].• SinceQR is clause-bounded, the bestKeshet can do is the following LF,whichdoesn’t derive the attested reading:(16) ∃𝑓Mary thinks that Jo hopes [ a hat just like mine 𝜆𝑥 ∧ Bill will buy 𝑡𝑥].8The following example from Grano 2019: p. 162.9The following example from Grano 2019: p. 162.• In the remainder of the paper, Grano (2019) briefly lays out (and rejects)other possible moves Keshet could make. I won’t dwell on the remainder ofthe argumentation here, but I’ll take this as a prompt to try to do better.3 Scope theory redux• In this section, we’ll start fromminimal means and bootstrap a different wayof achieving world-sensitivity that (I’ll argue) slices the pie in just the rightway.• I’ll assume that predicates deliver propositions rather than truth values, i.e.,10(18) JswimK ≔ 𝜆𝑥𝑤 . swim𝑤 𝑥 e → S t• Without going into the details of DP-internal composition (yet), I’ll assumewithout argument that definite descriptions denote individual concepts, i.e.,world-sensitive individuals.(19) Jthe boyK = 𝜆𝑤 . 𝜄𝑥[boy𝑤 𝑥] S e• Thinking through the problem of how to compose (19) with (18) will be thekey to unlocking an intensional grammar with just the combinatoric poten-tial we need to achieve exceptional de re.• Below, I define a composition rule☆ (pronounced: bind) in order to accom-plish just this.1112(20) Bind (def.)𝑚☆ ≔ 𝜆𝑘 . 𝜆𝑤 . (𝑘 (𝑚 𝑤)) 𝑤 ☆ ∶ S a → (a → S b) → S b10Here, S is the type constructor for world-sensitive values.(17) S a ≔ s → aIn other words, its a function from types to intensional types. Here, s is the type of worlds.11If you’re familiar with haskell (or category theory), you’ll recognize the type signature of bind.Wherem is a monad, monadic bind is of typem a → (a → m b) → m bIn fact, our bind is just the bind of a Readermonad.12N.b. that, for our purposes, we could havemade bind rigidly typed, where a = e, and b = t. Instead,I’ve given bind a maximally polymorphic type based on what we want it to do.3• Bind takes an argument 𝑚 and a function 𝑘; it returns a new function froma world 𝑤, where:– 𝑤 is first fed into𝑚, and then...– ...the result is fed into 𝑘, and the resulting open world argument is sat-urated again by 𝑤.• Now that we have bind, I’ll assume that definite descriptions are bind-shiftedin order to allow them to compose with predicates.(21) The boy swims.(22) S t𝜆𝑤 . swim𝑤 (𝜄𝑥[boy𝑤 𝑥])(e → S t) → S t𝜆𝑘𝑤 . (𝑘 (𝜄𝑥[boy𝑤 𝑥])) 𝑤☆ S e𝜆𝑤 . 𝜄𝑥[boy𝑤 𝑥]the boye → S t𝜆𝑥𝑤 . swim𝑤 𝑥swim• Tellingly, when we have a definite description in object position, it must bebind-shifted and undergo QR in order for composition to proceed:(23) Josie hugged the linguist.(24) S t(e → S t) → S t☆ ...the linguiste → S t𝜆𝑥 Josie hug 𝑡𝑥A helpful intuitionBind takes an intensional 𝑎 and turns it into a scope-taker.3.1 Exceptional de re• We now have almost everything we need to account for exceptional de rereadings of definite descriptions.• We just need one extra ingredient, which will play a similar role to Keshet’s∧ type-shifter; therefore, we’ll also call it ∧:(25) Up operator (def.)∧ 𝑎 ≔ 𝜆𝑤 . 𝑎 ∧ ∶ a → S a• All that the up-shifter does is add a vacuous world argument. Now we canderive the de re interpretation of the philosopher in the following example,without ever scoping out of the scope island:(26) Tom hopes [Sam invites the philosopher ].• Step 1: scope the bind-shifted definite description over an up-shifter in-serted at the edge of the scope-island:(27) S (S t)𝜆𝑤2𝑤1 . Sam invites𝑤1  (𝜄𝑥[philosopher𝑤2  𝑥])(e → S (S t)) → S (S t)𝜆𝑘𝑤2 . (𝑘 (𝜄𝑥[philosopher𝑤2  𝑥])) 𝑤2the philosopher☆(e → S (S t))𝜆𝑥𝑤2𝑤1 . Sam invites𝑤1  𝑥𝜆𝑥 ∧ Sam invites 𝑡𝑥• the result is a world-sensitive proposition of type S (S t), where invite is inter-preted relative to the inner world argument, and philosopher is interpretedrelative to the outer world argument.4• Step 2: Bind-shift the scope island, andQR it to the edge of thematrix clause.(28) 𝜆𝑤2 . Tom hope𝑤2   (𝜆𝑤1 . Sam invites𝑤1(𝜄𝑥[philosopher𝑤2  𝑥]))𝜆𝑘 . 𝜆𝑤2 .  (𝑘  (𝜆𝑤1 . Sam invites𝑤1(𝜄𝑥[philosopher𝑤2  𝑥])))  𝑤2☆ ...Sam invites the philosopher𝜆𝑝 . 𝜆𝑤2 . Tom hope𝑤2  𝑝𝜆𝑝 Tom hopes 𝑡𝑝• We’ve successfully derived the de re reading of the definite. This mechanismis recursive, and will therefore generalize to more deeply embedded scopeislands:(29) Mary thinks [that Jo hopes [that Bill buys the hat just like mine ]].• Step 1: scope the definite over an up-shifter:(30) 𝜆𝑤3𝑤1 . Bill buy𝑤1  (𝜄𝑥[hat𝑤3  𝑥])...the hat☆...𝜆𝑥 ∧ Bill buys 𝑡𝑥• Step 2: bind-shift the scope island and scope it over an up-shifter:(31) 𝜆𝑤3𝑤2 . Jo hope𝑤2  (𝜆𝑤1 . Bill buy𝑤1  (𝜄𝑥[hat𝑤3  𝑥]))𝜆𝑘 . 𝜆𝑤3 . (𝑘 (𝜆𝑤1 . Bill buy𝑤1  (𝜄𝑥[hat𝑤3  𝑥]))) 𝑤3☆ ...Bill buys the hat𝜆𝑝 . 𝜆𝑤3𝑤2 . Jo hopes𝑤2  𝑝𝜆𝑝 ∧ Jo hopes 𝑡𝑝• Step 3: bind-shift the result, and scope it to the edge of the matrix clause:(32) 𝜆𝑤3 . Mary thinks𝑤3  (𝜆𝑤2 . Jo hope𝑤2  (𝜆𝑤1 . Bill buy𝑤1  (𝜄𝑥[hat𝑤3  𝑥])))...☆ 𝜆𝑤3𝑤2 . Jo hope𝑤2  (𝜆𝑤1 . Bill buy𝑤1  (𝜄𝑥[hat𝑤3  𝑥]))[Bill buys the hat] Jo hopes𝜆𝑝𝑤3 . Mary thinks𝑤3  𝑝𝜆𝑝 ...Mary thinks 𝑡𝑝• Schematically, a de re interpretation for a DP embedded in two scope islandscan be derived via the following LF:(33) [island3 [island2 [island1 DP☆1 ∧ [ ...𝑡1... ] ]☆2 ∧ [ ...𝑡2... ] ]☆3 [ ...𝑡3... ] ]• An intermediate de re interpretation can be derived by only scoping theinner-most scope island.• Themechanism for deriving exceptional de re therefore, at LF, involves cyclic5scope-taking (Charlow 2019).13143.2 Evidence for scope• Since Keshet (2011), for independent reasons, rejects a scope-based theoryof de re interpretations on definites, it’s worth dwelling on what this buys us.• Romoli & Sudo observe a constraint on de re/de dicto readings of nested DPs:(34) Nested DP constraintWhen a DP is embedded inside a DP, the embedding DP must be opaque ifthe embedded DP is opaque.• If we take the following sentence, this blocks a reading where president takesnarrow intensional scope, and wife takes (exceptionally) wide intensionalscope:(35) Mary thinks the wife of the president is nice.• As Romoli & Sudo (2009) observe, the sentence is intuitively false in thefollowing context: Mary sees Bono Vox on TV with his wife Alison Hewson.Mary wrongly believes that he is the president, and furthermore, that the nicewoman next to him is his sister. Thus, the wife-relation is actually true, but thecharacterization of Bono Vox as the president is not• On the bti, this reading is easy to generate. On a scope theory however,such as the one outlined here, the corresponding LF will inevitably involvean unbound trace (here: 𝑡3):(36) * [island [the wife of 𝑡3]☆2 ∧ [the president]☆3 𝑡2 is nice ]☆1 Mary thinks 𝑡3.13It’s not a coincidence that the combinatorics for exceptional de re bear a resemblance to Charlow’s(2019) account of exceptionally-scoping indefinites via cyclic scope.The type-constructor S, alongside the bind-shifter and up-shifter constitute a typed instantia-tion of the Readermonad.Charlow shows in detail how the bind associated with a given monad M can be interpreted asa method for lifting a value of typeM a into a scope-taker.14I’ve framed the analysis here in terms of quantifier raising, but a completely isomorphic could begiven in a fragment which uses continuations (Barker 2002, Barker & Shan 2014) as an in-situscope-taking mechanism, as in Charlow 2014. It’s hard to find genuinely syntactic evidence forthe movements posited here, so ultimately this may be a better way to go.4 Towards an account of Bäuerle’s puzzle4.1 Extending the fragment• There are (at least) two outstanding issues with the current state of our frag-ment:– We haven’t said anything yet about DP-internal composition.– We haven’t said anything yet about what QPs denote, so we aren’t in aposition to address Grano’s challenge.• It will turn out that resolving the former issue will also give us a natural an-swer to the latter.• Let’s begin by thinking about how a definite determiner composes with itsrestrictor.(37) JtheK ≔ 𝜆𝑅 . 𝜄𝑥[𝑅 𝑥] (e → t) → e• If we want our rule to be maximally general, we need a way of composingsomething of type S ((a → b) → c) with something of type a → S b to giveback something of type S c.• We’ll accomplish via a new operation, which we’ll call c-lift. C-lift capturesa similar intuition to bind – bind provides a way of lifting intensional val-ues into intensional scope-takers; c-lift provides a way of lifting intensionalscope-takers into scope-takers with an intensional return type.15(40) C-lift (def.)𝑚✶ ≔ 𝜆𝑛𝑤 . 𝑚 𝑤 (𝜆𝑥 . 𝑛 𝑥 𝑤) S ((a → b) → c) → (a → S b) → S c15For the haskellers/category theorists in the audience, you’ll notice that, although extremely usefulfor lifting natural language determiners, this is not a very familiar type-signature. In fact, this isoperation requires something strictly stronger than a monad, namely a monad 𝑚 for which anoperation inject can be defined:(38) inject ∶ (a → m b) → m (a → b)As far as I can tell, a “natural” implementation of inject should be subject to the followingidentity law:(39) 𝜆𝑓 . 𝜆𝑥 . fmap (𝜆𝑘 . 𝑘 𝑥) (inject 𝑓) = 𝑖𝑑6• Once we up-shift our determiner, we can c-lift it to achieve the following re-sult – a function froma restrictor to an individual concept of type (e → S t) → S e.(41) (∧  JtheK)✶ = 𝜆𝑛𝑤 . 𝜄𝑥[𝑛 𝑥 𝑤] (e → S t) → S e• Exactly the same trick will generalize to the quantificational determiners.(42) JeveryK ≔ 𝜆𝑟𝑠 . ∀𝑥[𝑟 𝑥 → 𝑠 𝑥] (e → t) → t• If we first up-shift it, and then c-lift the result, we get the following meaning;a function from a predicate to an intensional scope-taker:(43) JeveryK ✶∘∧ = 𝜆𝑟𝑤 . 𝜆𝑠 . ∀𝑥[𝑟 𝑥 𝑤 → 𝑠 𝑥] (e → S t) → S ((e → t) → t)• Composing this meaning with a restrictor, e.g., boy, will result in the follow-ing:(44) 𝜆𝑤 . 𝜆𝑠 . ∀𝑥[boy𝑤 𝑥  → 𝑠 𝑥] S ((e → t) → t)• The internal composition for a QP therefore follows from the following LF:(45) S ((e → t) → t)(e → S t) → S ((e → t) → t)✶ every∧e → S tboy4.2 Back to Bäuerle’s puzzle• We now have all the pieces we need to account for Bäuerle’s puzzle.(46) George thinks every Red Sox player is staying in some five star hotel down-town.• I’ll assume that the meanings of every Red Sox player and some five star hotelare assembled via c-lift:(47) a. Jevery Red Sox playerK = 𝜆𝑤𝑘 . ∀𝑦[rsp𝑤 𝑦 → 𝑘 𝑦]b. Jsome five star hotelK = 𝜆𝑤𝑘 . ∃𝑥[fsh𝑤 𝑥 ∧ 𝑘 𝑥]• if we take an intensional QP, and c-lift it again, we derive something thatscopes at an intensional abstract:(48) Jevery Red Sox playerK ✶ = 𝜆𝑘 . 𝜆𝑤 . ∀𝑦[rsp𝑤 𝑦 → 𝑘 𝑦 𝑤]• Inside of the embedded clause, some five star hotel scopes to a position belowthe up operator, via c-lift.• every Red Sox player scope to a position above the up-shifter via bind, leavingbehind a higher-type trace below the existential’s scope site.16(49) Step 1: compute the value of the embedded clause𝜆𝑤2𝑤1 . ∃𝑥[fsh𝑤1  𝑥 ∧ ∀𝑦[rsp𝑤2  𝑦 → 𝑦 staying-in𝑤1  𝑥]]...every Red Sox player☆𝜆𝑄𝑤2𝑤1 . ∃𝑥 [fsh𝑤1  𝑥∧ 𝑄 (𝜆𝑦 . 𝑦 staying-in𝑤1  𝑥)]𝜆𝑄 ...∧ 𝜆𝑤1 . ∃𝑥 [fsh𝑤1  𝑥∧ 𝑄 (𝜆𝑦 . 𝑦 staying-in𝑤1  𝑥)]...some five star hotel✶𝜆𝑥𝑤1 . 𝑄 (𝜆𝑦 . 𝑦 staying-in𝑤1  𝑥)𝜆𝑥 𝑡𝑄 (𝜆𝑦 𝑡𝑦 staying in 𝑡𝑥)16In this upgraded fragment, bind is still essential in order to allow for semantic reconstruction.7(50) Step 2: scope the embedded clause out via bind𝜆𝑤2 . George thinks𝑤2  ⎛⎜⎜⎝𝜆𝑤1 . ∃𝑥⎡⎢⎢⎢⎣fsh𝑤1  𝑥∧ ∀𝑦 [rsp𝑤2  𝑦→ 𝑦 staying-in𝑤1  𝑥]⎤⎥⎥⎥⎦⎞⎟⎟⎠𝜆𝑘𝑤2 . 𝑘 ⎛⎜⎜⎝𝜆𝑤 . ∃𝑥⎡⎢⎢⎢⎣fsh𝑤1  𝑥∧ ∀𝑦 [rsp𝑤2  𝑦→ 𝑦 staying-in𝑤1  𝑥]⎤⎥⎥⎥⎦⎞⎟⎟⎠ 𝑤2every Red Sox player is staying insome five star hotel☆𝜆𝑝 George thinks 𝑡𝑝5 Specificity and transparency• More generally, this system divorces intensionality and quantification in asystematic way. Consider the famous constellation of readings for the fol-lowing sentence, as discussed by Fodor (1970).(51) Mary wants to buy an expensive coat.a. 3Non-specific opaqueNarrow quantificational and intensional scopeb. 3Specific transparentWide quantificational and intensional scopec. 3Non-specific transparentNarrow quantificational and wide intensional scoped. 7Specific opaqueWide quantificational and narrow intensional scope5.1 Non-specific opaque• This is easy – an c-lifted QP scopes below want.(52) Mary wants [ an expensive coat ✶ (𝜆𝑥 PRO buy 𝑡𝑥) ]• Quantificational and intensional effects scope together.5.2 Specific transparent• This is easy too – an c-lifted QP scopes above want.(53) an expensive coat ✶ (𝜆𝑥Mary wants PRO buy 𝑡𝑥).• Quantificational and intensional effects scope together.5.3 Non-specific transparent• There are at least two ways we could achieve this:• (i) A bind-shifted QP scopes abovewant, and the quantificational part of themeaning reconstructs:17(54) an expensive coat ☆ (𝜆𝑄Mary wants (𝑄 (𝜆𝑥 PRO buy 𝑥))).• (ii) A c-lifted QP scopes to the edge of the embedded infinitival over an up-shifter, which in-turn is bind-shifted and scopes above want.18(56) an expensive coat ✶ 𝜆𝑥 ∧ PRO buy 𝑥 (𝜆𝑝Mary wants 𝑝).• Intensional effects can out-scope quantificational effects.5.4 Specific opaque• There is no obvious way of achieving wide quantificational and narrow in-tensional scope on this system.17Achieving the so-called “third reading” via semantic reconstruction was proposed in Heim & vonFintel 2011: chapter’s nomotivation for positing pied-piping in this particular case since the complement ofwantisn’t a scope island.(55) A different student wants to attend every seminar. ∀ > ∃8• One possibility we could entertain is that certain expressions can leave be-hind a type S e. This allows, e.g., definite descriptions to totally semanticallyreconstruct.• Consider, e.g., the following example:(57) The most expensive coat, Mary wants to buy.• The most expensive coat can be interpreted de dicto. In order to account forthis, we can totally semantically reconstruct the definite description.19(58) ...𝜆𝑤 . 𝜄𝑥[mostExpensiveCoat𝑤 𝑥]the most expensive coat...𝜆𝑖 ...Mary ...wants 𝜆𝑤′ . PRO buy𝑤′  𝑖𝑤′𝜆𝑘𝑤′ . (𝑘 (𝑖 𝑤′)) 𝑤′☆ 𝑖...𝜆𝑥 PRO buy 𝑥• However, based on the machinery we’ve introduced for lifting classical GQsinto intensional operators, determiners always induce quantification over(extensional) individuals.• Since we derive meanings for QPs of type S ((e → t) → t), then if they canleave behind traces of this same type, we predict that QPs should be able tototally reconstruct too. This seems right:(59) An expensive coat, Mary wants to buy.19Technically, the QP semantically reconstructs to an intermediate position, where it is bind-shifted.• Based on the machinery we’ve introduced however, there’s no way to derivea quantifier over individual concepts from a classical GQ. Intuitively, this iswhat we would need to derive narrow intensional scope and wide quantifi-cational scope.• Conjecture: there are no natural language quantifiers which quantify overindividual concepts.6 Conclusion• Starting from the assumption that definite descriptions denote individualconcepts, and predicates return propositions, we’ve shown the following:– A natural operation that shifts a description into a scope-taker (☆),alongside an operation for deriving trivially intensional meanings (∧)automatically gives rise to exceptional de re; this is because, much likedefinite descriptions, scope islands can be bind-shifted.– A natural operation for shifting determiners (✶), automatically givesrise to a system in which quantificational and intensional scope are di-vorced – either quantificational and intensional effects scope together,or intensional effects outscope quantificational effects (the third read-ing).• The result is a flexible scope theory which inherits many of the advantages ofclassical scope theory (Romoli & Sudo’s generalization, etc.), while avoidingunder-generation pitfalls (exceptional de re, Bäuerle’s puzzle, etc.).• World-sensitivity slots neatly into a broader family of “effects” that may takeexceptional scope via recursive scope taking.ReferencesBarker, Chris. 2002. Continuations and the Nature of Quantification. Natural Lan-guage Semantics 10(3). 211–242.Barker, Chris & Chung-chieh Shan. 2014. Continuations and natural language (Ox-ford studies in theoretical linguistics 53). Oxford University Press. 228 pp.9Bäuerle, Rainer. 1983. Pragmatisch-semantische Aspekte der NP-Interpretation.In Manfred Faust et al. (eds.), Allgemeine Sprachwissenschaft, Sprachtypologieund Textlinguistik: Festschrift für Peter Hartmann, 121–131. Tübingen: GunterNarr.Charlow, Simon. 2014. On the semantics of exceptional scope. Dissertation.Charlow, Simon. 2019. The scope of alternatives: indefiniteness and islands. Lin-guistics and Philosophy.Fodor, JanetDean. 1970.The linguistic description of opaque contents.MassachusettsInstitute of TechnologyThesis.Grano,Thomas. 2019.Choice functions in intensional contexts: RehabilitatingBäuerle’schallenge to the scope theory of intensionality. In Richard Stockwell et al.(eds.), Proceedings of the 36th West Coast Conference on Formal Linguistics,159–164. Somerville, MA: Cascadilla Proceedings Project.Heim, Irene & Kai von Fintel. 2011. Intensional semantics. Lecture notes.Keshet, Ezra. 2008. Good intensions : paving two roads to a theory of the de re / dedicto distinction. Massachusetts Institute of TechnologyThesis.Keshet, Ezra. 2010. PossibleWorlds andWide Scope Indefinites: A Reply to Bäuerle1983. Linguistic Inquiry 41(4). 692–701.Keshet, Ezra. 2011. Split intensionality: a new scope theory of de re and de dicto.Linguistics and Philosophy 33(4). 251–283.Percus, Orin. 2000. Constraints on Some Other Variables in Syntax. Natural Lan-guage Semantics 8(3). 173–229.Romoli, Jacopo & Yasutada Sudo. 2009. De re/de dicto ambiguity and presupposi-tion projection. In Arndt Riester & Torgrim Solstad (eds.), Proceedings of Sinnund Bedeutung 21. Universität Stuttgart.10


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items