Since Since again - PowerPoint Presentation

Slide1

Since Since again

Kai von Fintel and Sabine Iatridou

1Slide2

Three interconnected papers:

Iatridou 2000+ 

Iatridou

2014 in Natural Language Semantics"About determiners on event descriptions, about time being like space (when we talk), and about one particularly strange construction"von Fintel and Iatridou 2002+  “ Since Since” ms MITvon Fintel, Fox, Iatridou 2006+  von Fintel, Fox, Iatridou 2014 in The Art and Craft of Semantics: A Festschrift for Irene Heim"Definiteness as Maximal Informativeness"

2Slide3

First a discussion on the Perfect

Then a PuzzleThen a bit of a solution

Then a bit about the consequences of the solution

3Slide4

We will start out with a compositional implementation of the analysis of the Perfect developed by

Iatridou, Anagnostopoulou and Izvorski

2002. (a version of Extended Now of

McCoard 78, Dowty 72, 79). The Perfect introduces a time interval: the ``perfect time span'' (PTS).The Right Boundary (RB) of the PTS is set by Tense The Left Boundary (LB) of the PTS may be set by ``perfect adverbials''.The lower predicate (event) is predicated of the PTS.This predication is mediated by various devices, most notably operators associated with the Perfective or Imperfective.4Slide5

1 .

I have visited Cape Cod three times since 1990 a. -There is a time span (the Perfect Time Span/PTS)

- the Right Boundary of the PTS is the time of utterance

- the Left Boundary of the PTS is (some time in) 1990 - in the PTS there are 3 subintervals at which it is true that I visit Cape Cod b. t : RB(NOW, t) and LB (1990, t) and t’, t’’, t’’’  t : I visit Cape Cod at t’, t’’, t’’’ c. 1990 UT LB ✔ ✔ ✔ RB |……………………………………………………………………………….......................|

(the Perfect Time Span/PTS)

5Slide6

In

general, the LB of the PTS can be set by an adverb:2.

I have visited Cape Cod three times

since 1990Or contextually:3. I have visited Cape Cod three times (=since the beginning of my life)6Slide7

In general, the RB of the PTS is manipulated by Tense.

The Present Perfect:

4

. I have visited Cape Cod two times since 1990 1990 UT LB RB |………………✔…………………………✔…………………………………….......................| t : RB(u, t) and LB (1990, t) and t’, t’’  t : I visit Cape Cod at t’, t’’The

Past

Perfect:

5

.

When we met, I

had

visited Cape Cod

two

times

LB RBwe meet UT |…………✔………………………✔…………………………………………….......|- - - - - - - - - - - - - - - - -| t : RB(<u, t) and LB (1990, t) and t’, t’’  t : I visit Cape Cod at t’, t’’

7Slide8

The

Future Perfect: 6

.

By next Monday, I will have visited the Cape 2 times7. _LB________x______________|__x_____RB Bill is born UT Monday8. _LB________x_________x____|________RB Bill is born UT Monday9. _LB______________________ |__x__x___RB Bill is born

UT

Monday

All of these are verifying circumstances for (6)! All the Perfect does is place the event(s) inside the PTS. It does not place the event(s) on the timeline

wrt

UT.

8Slide9

This is why you can say things like :

10. She is a very good student. She will have written her essay by next Monday for sure.

For all I know, she has written it already.

9Slide10

What we have seen so far are examples of the “Existential Perfect”.

There is also the “Universal Perfect”:11. She has been living on Cape Cod since 1990.

a. -There is a time span

(the Perfect Time Span/PTS) - the Right Boundary of the PTS is the time of utterance - the Left Boundary of the PTS is (some time in) 1990 - for every subinterval of the PTS it is true that she lives on Cape Cod. b. t : RB(NOW, t) and LB (1990, t) and  t’  t : she lives on Cape Cod at t’10Slide11

Existential vs Universal Perfect at a glance:

She has visited Cape Cod twice since 1990

12’.

[ x x | LB/1990 RB/UT13. She has been living on Cape Cod since 199013’. (--------)[-------------------------------| LB/1990 RB/UTIn the U-Perfect, the predicate holds for all points in the PTS (and possibly outside it as well)11Slide12

Basic structure:

T> Perfect > Imperfective/Perfective > vP

Following Klein 1994 and many others: What we call “Tense” is the relationship between Topic Time and Utterance Time:

TT<UT Past ___[TT ]______ UTUT<TT Future UT_______[TT ]UT  TT Present 14.12Slide13

Languages also encode the relationship between Topic Time and Situation Time. This is called “Aspect”:

ST  TT Perfective aspect [

TT

… [ ST …] … ]TT  ST Imperfective aspect [ST … [ TT …] … ] 1315.Slide14

16a. Bill insulted Tony

b. [TP Past [

AspP

Prf [VP Bill insult Tony]]] c. ∃t≺u: ∃t ⊇ t’: Bill insult Tony at t’ . d. ____ [TT …. [ST ... ]...] ________ UT17a. Bill was insulting Tony b. [TP Past [AspP Imp [VP Bill insult Tony]]] c. ∃t≺u:∃t ⊆ t’: Bill insult Tony at t’ . d. ____

[

ST

….

[

TT

...

]________ UT ST------14Slide15

Iatridou

et al 2002: The Existential versus Universal distinction is a function of the Aspectual distinction under the Perfect.Perfective  E-Perfect

18. I have read 2 books since last Tuesday

Imperfective  U-Perfect19. I have been reading two books since last Tuesday15Slide16

Stative

predicates in English yield ambiguities:20. I have been sick since 2014 (U-perfect reading easy)

The insurance company will reimburse you $500 if you have been sick in the period between January 1, 2014 and now.

21. I have been sick since January 1, 2014. I was sick for 3 weeks in the fall of 2015.This ambiguity is expected, given that the (morphologically unmarked) English statives can be read as TT  ST or ST  TT:22a. Yesterday he was sick b. yesterday  his disease or his disease  yesterday16Slide17

Basically, the PTS functions as TT in its interaction with Aspect.

Perfective: ST  TT



ST  PTS  E-Perfect [TT/PTS … [ ST …] … ]Imperfective: TT  ST  PTS  ST  U-Perfect [ST … [ TT/PTS ...

]

…

]

17Slide18

How is it all put together?

-The Perfect introduces an interval (PTS)

-The

PTS takes the TT it receives from the higher Tense as its Right Boundary and stretches backward in time. -This PTS is then fed downward as the new TT to the rest of the tree (which then interacts via the (Im)Perfective) with the ST (vP)23. 24. 18Slide19

The Left Boundary of the PTS can be determined by

“Perfect Level–adverbials”, which appear between the Perfect operator and the (Im)Perfective.

For example,

since 1990:25. 26. 19Slide20

The E-Perfect

27. Tony has visited Cape Cod since 1990

28. [

TP PRES [PERF since 1990 [ASP PRF [VP Tony visit Cape Cod]]]]29a. b. |….........................X..........................| 1990/LB UT/RB20Slide21

The U-Perfect

30. Tony has been living on Cape Cod since 1990

31a. [

TP PRES [PERF since 1990 [ASP IMP [VP Tony live on Cape Cod]]]] b. c. (xxxx) |xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx| 1990/LB UT/RB21Slide22

Since-clauses

Since 1990

Since (

preferrably definite) NP:Since the World Cup, Since Miranda’s graduation ceremony“Clausal Since”:Since we last met, Since Bill insulted TonyOur puzzle will be located within “clausal since”.22Slide23

Iatridou

(2014): the since-adverbial contains a definite description of a time interval (actually, a definite event description). In brief, there is an existential and uniqueness description in the

since

-clause.The existential presupposition:32. He hasn’t visited Cape Cod since his cat died.  the cat died33. Has he visited Cape Cod since his cat died?  the cat died34a. Have you been convicted of drunk driving in the last 10 years? b. Has it been 10 years since you were convicted of drunk driving?23Slide24

The uniqueness presupposition (actually unique salience presupposition):

35. a. Bill insulted Tony several times: once in 1990, once in 1993 and then again in

1997

… b. #And Tony has been living on Cape Cod since Bill insulted him. c. #Tony has visited Cape Cod two times since Bill insulted him. If modifiers like for the first/second/last time etc. are added then we can rescue (35b-c) by making one particular occurrence salient. 24Slide25

It is important to note that the

PTS is empty of the type of event described in the since-clause. In other words, there is no event of Bill insulting Tony between LB and RB:

36.

a. Tony has visited Cape Cod two times since Bill insulted him. b. Tony has been living on Cape Cod since Bill insulted him. LB RB |…..................................................................| B insults T UTBut saying that there is a covert for the last time gives the wrong results :37. a. Tony has visited Cape Cod two times since his cat died b. #Tony has visited Cape Cod two times since his cat died for the last timeThat is, there is no meaning of last in the since-clause; the since-clause is used felicitously only if there is a single salient event. In short, it is a definite singular.

25Slide26

How exactly is

the definite interpretation of the since-clause derived when its argument is clausal?

One possibility:

-an operator intervenes that takes us from a predicate of times (a proposition) to the time it is true of.But this won’t work:the times that “Bill insulted Tony” describes are times in the past of which Bill insulted Tony. That is not what since takes as its argument. What we want the argument of since to be, is the time AT which Bill insulted Tony.So let’s do exactly that.26Slide27

38.

The definite operator originates as the argument to a silent temporal preposition AT:

39

The operator delivers the time t such that Bill insulted Tony at t. In other words, the operator is a function from predicates of times to the unique time they are true of. In essence, the interpretation of the operator is the time at which. 27Slide28

The

movement of the operator is semantically motivated but we can see it in the syntax as well, through the existence of lower readings of the kind associated with movement of operators (Larson and many others) (40a) can mean (40b):

40a

. He has been to the Cape twice since Sue believes that Bill insulted him. b. He has been to the Cape twice since the time at which, according to Sue, Bill insulted him. Predictably, the movement cannot cross an island. (41a) cannot mean (41b): 41a. He has been to the Cape twice since Sue heard the rumor that Bill insulted him. b. He has been to the Cape twice since the time at which, according to the rumor that Sue heard, Bill insulted him.

28Slide29

A complication: the

clausal complement to since can also contain a predicate with the subinterval property.

42.

Tony has been happy since he lived on Cape Cod What is “the” time at which Tony lived on Cape Cod? If Tony ever lived on Cape Cod, there are many past intervals at which he lived there: all the subintervals of the maximal stretch of time throughout which he lived on Cape Cod.So, how can the definite reference succeed? We suggest (for now) that the definite here picks out the maximal interval of which the property holds. This is not unheard of for definite operators, but we will come to change it later.29Slide30

Now a puzzle:

43.

Tony has been happy since he has been taking Prozac

Interpret this as a U-Perfect, which means that Tony’s happiness extends throughout the PTS.How is the PTS determined here?RB is UT, since it is a Present Perfect. But what is its LB? Intuition: since the time that he started taking Prozac.LB should be provided by the complement of since. But: he has been taking Prozac does not mean he started taking Prozac.30Slide31

43. Tony has been happy since he has been taking Prozac

We understand (43) as conveying that Tony started being happy when he started taking Prozac and the time of his being happy and his taking Prozac extends up until and including

UT.

But remember that the inference that Tony started being happy when the Prozac started is an implicature:44a. Tony has been happy since he has been taking Prozac and possibly longer b. Tony has been living on Cape Cod since he has been taking Prozac and possibly longer31Slide32

The assertion:

the period in which he has been taking Prozac is (possibly properly) included in the period in which he has been happy.We call this the “Simultaneous Reading”

32Slide33

45.

Tony has been happy since he insulted Bill

LB RB

(-----)|------------------------------Tony is happy------------------------------------|Bill insults Tony UT46. Tony has been happy since he has been taking Prozac (SR) LB RB (-----)|-----------------------------Tony is happy-------------------------------------| ------------------------------Tony takes Prozac -------------------------------UT33Slide34

Also:

We see in (45), as already mentioned earlier, that the PTS is empty of events of the since-event type.

But in (46) the

PTS is not empty of the eventuality described in the since-adverbial.On the contrary, the since-event holds throughout the PTS.34Slide35

So it looks as if in the SR, the matrix event and the

since-event are simultaneous.But actually, this is derivative. What the since-event is simultaneous with is the matrix PTS, not the matrix event.

Transitivity:

Because Tony has been happy is a U-Perfect, the matrix event fills out the matrix PTSthe since-event is simultaneous with the matrix PTShence the since-event is simultaneous with the matrix event.And this can be verified, by changing step (a). That is, we look at what happens when we are dealing with a matrix E-Perfect:35Slide36

47. Tony

has been to the Cape two times since he has been taking Prozac.

LB RB |………………✔…………………………✔……………………………………...............| -------------------------------Tony takes Prozac ------------------------------UTIn short, in SR the since-event temporally coincides with the matrix PTS, not (necessarily) with the matrix event.We can have SR regardless of whether the matrix Perfect is a U-Perfect or an E-Perfect. 36Slide37

There is an additional argument that it is not the matrix event and the

since-event that are simultaneous:48a. Tony has been happy since he has been taking Prozac and possibly

longer

LB RB (-----)|-----------------------------Tony is happy-------------------------------------| ------------------------------Tony takes Prozac -------------------------------UTIf the matrix event extends to the left of the LB, the since-event does not extend to the left as well.37Slide38

In

SR the since-event temporally coincides with the matrix PTSSo for the SR reading, there are no requirements on the matrix clause, other than that it be a Perfect, given that it’s only then that we can have a

since

-clause to begin with.Are there any requirements on the morphosyntax in the inside the since-clause?Our sentences so far have a Perfect inside the since-clause as well: 49. Tony has been happy [since he has been taking Prozac]50. Tony has been to the Cape two times [since he has been taking Prozac]In fact, a Perfect inside the since-adverbial is not just possible but necessary for SR.38Slide39

No Perfect inside the

since-clause, no SR:51a. Tony has been happy

[

since he took Prozac] b. Tony has been to the Cape twice [since he took Prozac]52a. Tony has been to the Cape twice [since he was in the hospital] b. Tony has been happy [since he was in the hospital]53a.*Tony has been to the Cape twice [since he is in the hospital] b. *Tony has been happy [since he is in the hospital]

39Slide40

Moreover, what we see in our SR cases is that the Perfect inside the

since-clause is a U-Perfect:54. Tony has been happy [

since

he has been taking Prozac]55. Tony has been to the Cape two times [since he has been taking Prozac]This raises the question of whether the SR reading is possible with an E-reading inside the since-clause. We will return to this question later.40Slide41

Given that for the SR, the

since-adverbial must contain a Perfect, it follows that there is another PTS involved, namely the PTS of the

Perfect in the

adverbial the matrix-PTS the since-PTS.56. Tony has been happy [since he has been taking Prozac]57. Tony has been to the Cape two times [since he has been taking Prozac]And once we realize that there are two PTSs involved, we can now proceed to the most appropriate paraphrase:41Slide42

58. Tony has been happy

since he has been taking Prozac=

59.

Tony has been happy since the time since which he has been taking Prozac. We said earlier that in SR, it is the matrix PTS, not the time of the matrix event that is involved. Now we see that it is the since-PTS that is involved, not the time of the since-event. But because there is a U-Perfect in the since-clause, its predicate holds throughout the since-PTS. Hence the illusion that simultaneity holds between the matrix PTS and the time of the adverbial’s predicate.

42Slide43

What actually is the case:

in SR, it is the two PTSs that are simultaneous. Now that we have understood that, we need to find (59) in (58):

59. Tony has been happy

since the time since which he has been taking Prozac. 58. Tony has been happy since he has been taking Prozac43Slide44

Deriving the SR reading

Something that does not work

…

Do simply what we did before: a definite operator that starts out as the complement of a covert AT:60. Tony has been happy since he has been taking Prozac61. since Op λt [TP Pres [ Perf [AspP Imp [VP [VP he take Prozac] [ AT t ]]]]]]. To paraphrase (61):62. since the time t such that

there is a PTS stretching backwards from now such

that

that PTS is included in a time t′

which is a time at which Tony takes

Prozac

and

which

is identical to t 44Slide45

As before, the definite

the operator tries to find the maximal time that satisfies the description

.

In other words, it delivers the interval between now and when Tony started taking Prozac. But there are two problems with this. First problem: Remember that ‘since 1990’ means “since some time in 1990”. So, ‘since the interval between now and when Tony started taking Prozac’ would mean ‘since some time in the interval between now and when Tony started taking Prozac.’So, it would be true if he started being happy yesterday while he started taking Prozac last year. We don’t want this. [we avoid this problem with ‘since he wrote Anna Karenina’ because of the Perfective, which wants the writing-event completed. But what about ‘since he was in the hospital?]

45Slide46

Second problem:

since cannot take an interval as its argument that reaches all the way to now.

Imagine

that Tony has been taking Prozac since the beginning of this week. Under this analysis the argument of since is just the interval that we could refer to with this week. But this is not possible:63. #Tony has been happy since this week. (not to be confused with: Tony has been happy since this week started)though it is, of course what happens with SR:64. Tony has been happy since he has been taking Prozac46Slide47

What does work: take the paraphrase seriously

65. Tony has been happy

since he has been taking Prozac

=66. Tony has been happy since the time since which he has been taking Prozac.The clausal complement of since contains its own since-clause.47Slide48

But if there are two

sinces in the semantics, why do we pronounce just one?

67.

Tony has been happy since he has been taking Prozac68. *Tony has been happy since he has been taking Prozac sinceA silent since? Not very inspired, plus it would make the wrong predictions:69a *I met him after he has been taking prozac =/= b. I met him after the time since which he has been taking prozacOr, the answer to the question can be found in Larson 1987.48Slide49

Larson

1987 considers the following pair of sentences, which differ in a preposition that has gone missing in (b):70a

. By 1999, I will have lived in every city that John has lived in

b. By 1999, I will have lived in every city that John has lived He explains the elided preposition as an instance of Antecedent Contained Deletion, or ACD.What is ACD?49Slide50

In general, in VP-ellipsis, the

ellided VP is interpreted as its antecedent:71a. Miranda petted the dog and Lena did

__

too b. Miranda petted the dog and Lena petted the dog too50Slide51

The problem with ACD:

The ellipsis site is contained inside the antecedent (hence the name!). If we fill in at LF the only other VP

in the sentence, we get

a problem:72a. John suspected everyone Mary did __ b. John [suspected everyone Mary did __ ] c. John suspected everyone Mary [suspected everyone Mary did ___] d. John suspected everyone Mary [suspected everyone Mary suspected everyone Mary did___] e. John suspected everyone Mary [

suspected everyone Mary

suspected

everyone

Mary

suspected every one Mary did ___

]

51Slide52

ACD solution as

in Sag 76 and May 85:73a. John [VP suspected everyone Mary did [

VP

e]] b. John [VP suspected everyone Mary did [VP e] ]QR:74. [ [everyone Mary did [VP e]]k [John [VP suspected tk ]]Replacing missing VP with main VP:75. [ [everyone Mary did [VP e]]k [John [

VP

suspected

t

k

]

]

76.

[[everyone Mary did

[VP suspected tk ] ]]k [John [VP suspected tk ] ]52Slide53

Larson 1987: there is also PP ACD:

77. By 1999, I will have lived in every city John has livedThere is a missing preposition. B

oth instances of

lived require a preposition:78. *John has lived that cityGiven that there is PP ellipsis (all data Larson p.c.):79a. Squirrels left paw prints [on the porch] and birds left claw marks [PP e ] b. John got a bicycle [for his birthday] and Bill got a skateboard [PP e ] c. John gives his money to MIT and Mary gives her time [PP e]53Slide54

There is also PP ACD

:77’. By 1999, I will have lived [PP in [every city [that John has lived [

PP

e]]]]QR:80. [ [every city [that John has lived [PP e]]] [By 1999, I will have lived [PP in tk]]]Replacing missing PP with main PP:81. [[every city [that J. has lived [PP in tk ]]] [By 1999, I will have lived [PP in tk] ]]

54Slide55

Since

-deletion à la Larson 1987:82. Tony has been happy since he has been taking Prozac

83. Tony has been happy [

PP since Op [ he has been taking Prozac [PP e ] ] ]QR:84. [[Op [ he has been taking Prozac [PP e ]]k [Tony has been happy [PPsince tk]]]Replacing missing PP with main PP:85. [[Op [he has been taking

P.

[

PP

since

t

k

]

]

k

[Tony has been happy [PP since tk]](This is the LF-fill in version of ellipsis, which is the way Larson has it; PF-deletion will yield the same results)55Slide56

There is one difference with the Larson case: the

missing preposition is obligatorily missing, unlike in Larson’s cases:86. By 1999, I will have lived in every city John has lived (in)

87. Tony

has been happy since he has been taking Prozac (*since)Possibly this is because since simply does not strand and so only the ACD option is available.  88. ??When have you been living here since?(Of course we don’t know why some prepositions strand and others don’t)Is it the case that other prepositions that don’t strand undergo obligatorily ACD? 56Slide57

OK! So we have our syntax more or less in place.

What about our semantics?There we are not yet done.

We have worked out a syntax that is close to the paraphrase

89. Tony has been happy since the time since which he has been taking Prozac. But what interval is picked out by “the time since which Tony has been taking Prozac”? 57Slide58

Imagine that Tony started taking Prozac

at 3 p.m. on October 28, 2008.

Then any super-interval of that time will be an interval in which Tony started taking Prozac: October 28, The week of October 28, The month of October 2008, The year 2008

etcIn the context above, we can truthfully say all of the following:90a. Tony has been taking Prozac since October 28, 2008 b. Tony has been taking Prozac since October 2008 c. Tony has been taking Prozac since 2008 58Slide59

Our semantics

delivers a since at least -reading.

A sentence like (35)

can be read as saying that Tony started being happy when he started taking Prozac but that is a quantity implicature:91. Tony has been happy since he has been taking Prozac and perhaps even longer92. She has been living in Amsterdam since 2010, perhaps even longer93. If a person has been living in Amsterdam since 2010, she speaks proper Dutch59Slide60

You might say: That’s not a problem!

Just have “the time since which Tony has been taking Prozac” refer to the smallest interval! (i.e. exclude reference to any of the super-intervals)

But that won’t solve our problem either, because there is no shortest interval either (again because of the

at least since inferences):If Tony started taking Prozac on October 28, 2008, the following are also true:94a. Tony has been taking Prozac since 2010 b. Tony has been taking Prozac since 2012 c. Tony has been taking Prozac since 2014 RB/UT_____|________________________________________| 2008 2009 2010 2011 2012 2013 2014 2015 2016Because of the presence of these intervals, there won’t even be a smallest interval in the set of intervals that the operator operates on. 60Slide61

Instead, we propose that

……the operator finds in the set of intervals that interval from whose presence in the set we can deduce the presence of all the others in the set.

Intuitively

, the operator finds “the most informative” interval in the set. Imagine that Tony started taking Prozac in October, 2008. Then “the time since which he has been taking Prozac” will pick out that time such that all the other times at which he takes Prozac follow.The choice of any other interval will not will not deliver all the right entailments.61Slide62

RB_________|________________________________________|

2008

2009 2010 2011 2012 2013 2014 2015 201662The most informative intervalbecause all the intervals to the rightare entailed to be includedNot the most informative interval becausethere are no entailments about the intervals to the left,which, however, are includedIn short, the operator picks the smallest leftmost interval among the set of intervals since which he has been taking Prozac. Because from that one the presence of all the other intervals in the set of intervals since which he has been taking Prozac follows.Slide63

Note that we will also need this for questions:

95. Since when has Tony been taking Prozac?

63Slide64

Remember: We take from

Iatridou (2014) that the content of the argument of the since-clause provides a definite description.

Here:

this definite description should be interpreted as picking out the most informative interval, rather than the unique or the largest interval.This is different from a common semantics for the determiner the. For example, Link’s:96. The φ is defined only if there is a maximal object x st. φ(x). When defined the φ refers to the maximal object x st. φ(x). We saw above that this definition will bring about a presupposition failure in the case of the definite description in a since-clause, as there is no unique maximal (or minimal, for that matter) interval that will fit.

64Slide65

Instead, we

suggested that the definite description picks out the maximally informative interval. Does this mean that the argument of the

since

-clause is a special type of definite description? No!von Fintel, Fox and Iatridou (2014) argue that the definite description in a since-clause showcases exactly the canonical meaning of the, which is:97. the φ is defined only if there is a unique individual x such φ(x) is a maximally informative proposition among the true propositions of the form φ(x). When defined the φ refers to the individual x st. φ(x) is the maximally informative true proposition of the form φ(x). (henceforth MI)

65Slide66

For many cases, a Link-style semantics and MI predict the same results, but there are many cases where the predictions are different.

Eg :98.

T

he number of Greek soldiers who together can destroy the Trojan army According to a Link-style semantics, the NP in (98) is undefined: there is no maximal number of soldiers that can defeat the Trojan army. If a number of soldiers s can do it, any number larger than s can do it too. In fact, what we are picking out in (98) is the minimal number of Greek soldiers. 66Slide67

According to

MI, however, the article the is defined because it picks out the smallest

number of soldiers.

It does this exactly because the most informative number is the smallest number of soldiers in (98). Let’s say again that s is the smallest number of soldiers that can defeat the Trojan army. All the the sentences containing amounts m>s are asymmetrically entailed. And if we had picked an m larger than s then the proposition containing m would not have entailed the propositions with r, where s≺r≺m. See von Fintel, Fox, Iatridou for more cases.67Slide68

Finally, we come to a question that has lose threads to places that we have not followed yet.

We said that to have a SR, we need a Perfect inside the since-clause, and the

since-

Perfect that we have been using in all our examples so far has been a U-Perfect:99. Tony has been happy since he has been taking Prozac100. Tony has been to the Cape two times since he has been taking Prozac68Slide69

We find the same phenomenon in other languages. Interestingly, but also predictably, we find it in languages where the U-Perfect reading is conveyed without a Perfect

Morphosyntax:

101.

Tony nimmt seit 1990 Prozac. Tony takes since 1990 Prozac ‘Tony has been taking Prozac since 1990.’ 102. Tony ist glücklich seit er Prozac nimmt. (Matrix U-meaning) Tony is happy since he Prozac takes ‘Tony has been happy since he has been taking Prozac.’ 103. Tony hat Cape Cod zweimal besucht seit er Prozac einnimmt. (Matrix E-Perfect) Tony has Cape Cod twice visited since he Prozac in-takes

‘Tony has visited Cape Cod twice since he has been taking Prozac’

69Slide70

But nothing in our analysis makes it necessary that the Perfect in the

since-clause be a U-Perfect. In principle, everything that we have said should work with an E-Perfect as well. Does it?No, it appears:

104. *Tony has been happy since he has visited Cape Cod

Why should this be?Recall our LF for the SR reading:105. [[Op [he has been taking P. [PP since tk]]k [Tony has been happy [PP since tk]]70Slide71

The PF-deletion view of ellipsis might make the point a bit clearer:

106. Tony has been happy since he has been taking Prozac107.

Tony has been happy [

PP since Op [ he has been taking Prozac [PP since e]] ]QR:108. [[Op [he has been taking Pr. [PP since e ]]k [T. has been happy [PPsince tk]]]PF deletion under identity:109. [[Op [he has been taking P.

[

PP

since

t

k

]

]

k

[T. has been happy [PP since tk]]71Slide72

Look at (107) again:

107. Tony has been happy [PP since

Op

[ he has been taking Prozac [PP since e] ] ]So there are two movement chains:-QR for ACD resolution: the underlined constituent-the movement of Op Both these movements would also happen in the ungrammatical (77/82):109. *Tony has been happy [PP since Op [ he has visited Cape Cod [PP since e] ] ]The problem lies in the movement of Op in the existential Perfect!

72Slide73

110a

. Op [ he has visited Cape Cod [PP since

e] ] Abstracting away from preposition stranding, pied piping would yield: b. Since Op [ he has visited Cape Cod [PP e] ] A’-movement of the since-clause is impossible in E-Perfects, while it is possible in U-perfects:111a. Since when have you been living on Cape Cod? b. *Since when have you visited Cape Cod twice?73Slide74

So compare:

112. Tony has been happy [PP since

Op

[ he has been taking Prozac [PP since e] ]]113. *Tony has been happy [PP since Op [ he has visited Cape Cod [PP since e] ] ]In (112) there is a U-Perfect and the green chain is licit.In (113) there is an E-Perfect and the red chain is illicit.(Again, abstracting away from the choice of P-stranding vs. Pied-Piping)

74Slide75

Where we are:

-we have an analysis for how the SR is derived. What is needed is a PTS inside the since-clause. Then the matrix and since-PTSs are co-extensive.

-If all we need from the

since-clause to obtain SR is a PTS, that is, a Perfect inside it, in principle any Perfect should do.-However, we saw that an E-Perfect inside the since-clause yields ungrammaticality. We connected this ungrammaticality to the inability to extract the since-clause in an E-Perfect.-But why should there be such an inability?75Slide76

What would (114) mean, if it were grammatical?

114 *Since when have you visited Cape Cod twice?

It would mean:

What is the time since which you have been to Cape Cod twice?76Slide77

Fox

and Hackl (2004) discuss this impossible extraction from an E-Perfect as reported in a previous version of the current paper.

They

argue that its unacceptability is due to the fact that it is not possible to satisfy the presupposition of the definite description in the since-clause. The reason is that the domain of time is dense. As a result it is not possible to find “the time since which an event happened”. On the other hand, with a U-Perfect this extraction is fine because the definite description picks out the time at which the living on the Cape started. 77Slide78

Summary!

78