! 1!

The Curious Implicatures of Optional Past Tense in Tlingit (and Other Languages)

Seth Cable University of Massachusetts Amherst

1. Introduction: Variation in Tense Semantics and Variation in Cessation Inferences (1) Overarching Question What tense features does UG allow for?

a. We know from IE languages that UG allows for [PAST] and (maybe) [PRES] 1

b. We know from Lillooet (Salish; BC) that UG allows for [NON-FUT] (Matthewson 2006)

c. Are there any other features (sub-features) that T-heads can bear? 2

(2) Overarching Puzzle

a. In English, past-tense statives trigger a cancellable (ergo, non-semantic) inference of ‘cessation’.

b. In some other languages (e.g., Tlingit (Na-Dene; Alaska)), this ‘cessation’ inference cannot be cancelled in the same way.

(3) Cessation Inferences in English

In certain contexts (to be detailed later), past tense statives in English (lexical statives; progressives; generics) imply that the state in question does not extend into the present.

a. (i) Who wrote that song “Sledgehammer”? (ii) Oh, I knew this… (Implies: ‘I don’t currently know the answer’) b. (i) How are you feeling? (ii) Well, I was nauseous. (Implies: ‘I’m not currently nauseous’) c. (i) Is Dave enjoying the party? (ii) Well, he was dancing. (Implies: ‘He’s not currently dancing’) d. Dave was from New Jersey. (Implies: ‘Dave is dead’) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1!The status of ‘present’ as a tense feature (rather than simply absence of tense) remains controversial (Sauerland 2002, Thomas 2014b). 2!Although this also remains highly controversial, I follow much of the semantic literature on tense in the assumption that [FUTURE] does not exist as a tense category in the languages of the world.

! 2!

(4) Cessation Inferences in Tlingit Past tense statives in Tlingit are also reported to trigger this same inference (Leer 1991).!

a. Kuk’éiyeen. 3, 4 IMPFV.good.weather.PST The weather was nice (but turned bad) (Leer 1991: 464)

“…[This sentence] means that a specific situation, namely an instance of good weather, was true in the past and is not true now” (Leer 1991: 465)

b. Sheet’káx’ áwé yéi xat téeyin at k’átsk’ux xat sateeyí. Sitka.at FOC IMPFV.1sgS.be.PST boy IMPFV.1sgS.be.SUB

I lived in Sitka when I was a child. (Leer 1991: 465) !

“…[This sentence] could be said by someone who left Sitka during childhood…” (Leer 1991: 465)

c. Xaxáyeen. 3O.IMPFV.1sgS.eat.PST I was eating it (but have stopped). (Leer 1991: 463) d. Xaxáa noojeen. 3O.1sgS.eat HAB.PST I would eat it (habitually in the past, but no longer) (Leer 1991: 463) (5) Cancellation of Cessation Inferences in English

In English, one can readily explicitly cancel the cessation inference in (3). a. I knew this years ago, and I still know it now. b. I was nauseous this morning, and I’m still nauseous now. c. Dave was dancing an hour ago, and he’s still dancing now. d. Dave was from NJ this morning, and (of course) he’s still from NJ now. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!3!I provide only the roughest of glosses for individual Tlingit words, which can be morphologically quite complex. This simplification is most radical for verbs, as I provide glosses only for their lexical content, their agreement morphology, and their TAM morphology. 4!I employ the following glossing abbreviations: 1, ‘first person’; 2, ‘second person’; 3, ‘third person’; AUX, ‘auxiliary verb’; DET, ‘determiner’; DUB, ‘dubitative’; EMPH, ‘emphatic’; EXCLM, ‘exclamative’; FOC, ‘focus particle’; FUT, ‘future’; HAB, ‘habitual’; HYPO, ‘hypothetical/subjunctive particle’; IMPFV, ‘imperfective’; Indef, ‘indefinite’; INF, ‘infinitival’; INST, ‘instrumental’; LOCP, ‘locative predication suffix’; NEG, ‘negation’; NFUT, ‘non-future’; NOM, ‘nominalizer’; O, ‘object’; OPT, ‘optative’; PERF, ‘perfect’; PFV, ‘perfective’; pl, ‘plural’; POSS, ‘possessor’; POT, ‘potential’; PRES, ‘present’; PST, ‘past’; Q, ‘question/indefinite particle’; RECIP, ‘reciprocal’; REFL, ‘reflexive’; S, ‘subject’; REL, ‘relative clause marker’; sg, ‘singular’; SUB, ‘subordinator’.!

! 3!

(6) Contexts Where Cessation Doesn’t Arise in English In English, the cessation inference in (3) usually doesn’t arise if there is a salient, topical past time.

a. As soon as you asked me the question, I knew the answer. b. When the doctor saw me, I was feeling nauseous. c. I just saw Dave in the kitchen. He was dancing. d. I met this really cool guy named Dave yesterday. He was from NJ. (7) Conclusion In English, the cessation inference with past tense statives is non-semantic. (Musan 1997, Magri 2011, Thomas 2014a, Altshuler & Schwarzschild 2013) (8) The Puzzle, Part One

• In Tlingit, however, one cannot explicitly cancel the cessation inference, as in (5). • In cases like (5), one must use a ‘tenseless’ verb (which can be construed as past)

(9) Scenario: Joe has been sleeping all day. He was sleeping this morning, and he’s still

sleeping now. a. Tlingit Sentence Offered: Tle yá ts’ootaat dágáawé tá Joe. then this morning indeed IMPFV.3sgS.sleep Joe

Ch’a yeisú tá. (SE) 5 just still IMPFV.3sgS.sleep

This morning, Joe was indeed sleeping. He’s still sleeping now.

b. Rejected Sentence, Containing Past Tense: # Tle yá ts’ootaat dágáawé táyin Joe.

then this morning indeed IMPFV.3sgS.sleep.PST Joe

Ch’a yeisú tá. just still IMPFV.3sgS.sleep

Comments by Speakers: • “No; -yin makes it past tense. But you’re saying that he’s still sleeping.” (SE) • “No; [the sentence with past tense] means he’s already slept and gone.” (WF)

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!5 I indicate whether a Tlingit sentence was (i) constructed by myself and judged by speakers to be acceptable, or (ii) actually spontaneously spoken by the consultants themselves. In the former case, the sentence will be followed by a ‘(C)’, for ‘constructed’. In the latter case, I will write the initials of the speaker(s) who provided the sentence: (MD) for Margaret Dutson, (SE) for Selena Everson, (WF) for William Fawcett, (CM) for Carolyn Martin, (JM) for John Martin, and (HS) for Helen Sarabia. In addition, when I provide comments recorded from the speakers, I also provide the initials of the speakers who made the comments.

! 4!

(10) Scenario: The weather has been nice all day. It was nice this morning, and it’s nice now. a. Tlingit Sentence Accepted:

Yá ts’ootaat ch’a kuwak’éi. Ch’a yeisú kuwak’éi. this morning just IMPFV.good.weather just still IMPFV.good.weather

This morning, the weather was nice. It’s still nice now. (C) b. Rejected Sentence, Containing Past Tense:

# Yá ts’ootaat ch’a kuk’éiyin. Ch’a yeisú kuwak’éi. this morning just IMPFV.good.weather.PST just still IMPFV.good.weather Comments by Speakers:

“I don’t like kuk’éiyin. It’s past tense. It means it’s gone, the weather is gone. It’s passed.” (SE)

(11) The Puzzle, Part Two In Tlingit, even if there is a salient, topical past time, the cessation inference still arises for past tense statives.

(12) Scenario: You walk by your friend Joe’s house, and see him building a boat. A few

minutes later, you bump into your friend Sue, and she asks what Joe has been up to. You want to tell her that, just a few minutes ago, you saw him working on his boat.

a. Tlingit Sentence Offered: Dziyáak Joe xwasateení, du yaagú alyéix. (MD, SE, WF) earlier Joe 3O.PFV.1sgS.see.SUB his boat 3O.IMPFV.3sgS.build When I saw Joe earlier, he was building his boat.

b. Rejected Sentence, Containing Past Tense:

# Dziyáak Joe xwasateení, du yaagú alyéixin. earlier Joe 3O.PFV.1sgS.see.SUB his boat 3O.IMPFV.3sgS.build.PST ! Comments by Speakers:

“No; that means he’s not working on it now.” (SE) (13) Scenario: We’re at a party. You spot your friend Joe in a corner. You see that he is

singing. You then go into the kitchen. There, you hear your friend Sue say ‘Oh, I wish I could hear Joe sing!’ You want to mention that you just saw him singing. Naturally, you assume that he’s still singing now, so Sue can go hear it.

a. Tlingit Sentence Offered: Ch’a yeisú xwsiteen Joe. At shí. (SE, WF) just just.now 3O.PFV.1sgS.see Joe IndefO.IMPFV.3sgS.sing I saw Joe just now. He was singing.

! 5!

b. Rejected Sentence, Containing Past Tense: # Ch’a yeisú xwsiteen Joe. At shíyin just just.now 3O.PFV.1sgS.see. Joe IndefO.IMPFV.3sgS.sing.PST

Comments by Speakers: “This one means that he’s through singing, he had been singing.” (SE)

(14) An Important Baseline

The speakers I work with do readily accept the use of the past-marking of statives, when it’s clear from context that the cessation inference holds.

a. Yéi xat gusagéink’in. Yeedát ku yéi xat kuligéi. (SE)

IMPFV.1sgS.small.PST now though IMPFV.1sgS.big I used to be small. Now, though, I’m big.

b. Yá ts’ootaat kuk’éiyin. Yeedát ku.aa tlél kushk’é. (SE) this morning IMPFV.good.weather.PST now though NEG IMPFV.good.weather This morning, the weather was nice. Now, though, the weather is not nice.

c. Gooshúk gaaw áwé ch’a yeisú táyin. nine hour FOC just still IMPFV.3sgS.sleep.PST Yeedát ku kei wdzigít. (SE, WF) now though PFV.3sgS.wake.up. At nine o’clock, he was still sleeping. Now, though, he’s woken up.

d. Scenario: You needed to get your sink fixed, and you called a plumber. But, I know a lot about plumbing, and could have fixed your sink for you. When I find out you paid for a plumber, I want to tell you that I could have done it for you.

I jeeyís áyá yéi nkwasaneiyín. (WF, SE)

your hand.for FOC 3O.POT.1sgS.do.PST I could have done it for you.

e. Scenario: I was supposed to leave for Sitka this morning. When I got to the airport, though, I saw that my flight was cancelled!

Yá ts’ootaat áwé Sheet’káadei kukkwatéenin. (WF, SE, MD)

this morning FOC Sitka.to FUT.1sgS.travel.PST This morning, I was going to travel to Sitka. (15) The Burning Question Why is there this difference between English and Tlingit?

! 6!

(16) One Possibility Although the cessation inference is pragmatic in English, it’s actually semantic in Tlingit. • That is, the cessation inference is somehow encoded into the lexical semantics of

‘Past’ in Tlingit.

Proponents of this Answer: Leer (1991), for Tlingit Copley (2005), for Tohono O’odham Plungian & van der Auwera (2006), for various languages

o Wolof, Tokelauan, Lezgian, Sranan, Bamana, Washow (17) ‘Discontinuous Past’ (Plungian & van der Auwera 2006)

A past tense marker for which the cessation inference has become part of its conventionalized, lexicalized meaning.

(18) Clear Implication of Discontinuous Past Markers

• The answer to (1c) is YES. • In addition to ‘[PST]’, UG also admits of a tense (sub-)category ‘[Dis PST]’

(19) Main Claims of this Talk a. Empirical Claim

Despite the facts in (8)-(13), the cessation inference in Tlingit is still pragmatic • There are other environments where the inference is cancelled

b. Analytic Claim

The cessation inference in Tlingit arises via different mechanisms from the cessation inference in English:

(i) The fact that past tense in Tlingit is optional (ii) A special principle relating to the topicality of Utterance Time

c. Typological Claim

The proposed analysis should be extended to all other putative instances of ‘discontinuous past’.

• This would capture the fact that putative cases of ‘discontinuous past’ only

ever arise in optional tense languages (Plungian & van der Auwera 2006)

d. Major Conclusion: • There is no evidence for ‘discontinuous past’ as a separate category of tense • And, so no evidence yet that the answer to (1c) YES (still possibly ‘NO’)

! 7!

2. Evidence that the Cessation Inference in Tlingit is Not Semantic 2.1 Absence of Cessation in Examples Taken From Naturally Produced Texts (20) Key Claim

It is possible to find in naturally produced Tlingit texts instances of past-marked statives where:

a. There is no contextual support for any ‘cessation’ implication, and b. Such an implication would also be inconsistent with other information in the text

(21) Tlingit Past-Tense Stative with No Cessation Implication Ch’a yeisú áa yéi téeyin du ji.eetí. Just still there 3O.IMPRV.be.PST his hand.remnants It was still there recently, the work of his hands.

(Dauenhauer & Daunehauer 1987; 100: 359) Context:

a. The speaker is referring to the petroglyph carved by Kaax’achgóok (an ancestor hero of the Kiks.ádi clan in Sitka, AK.

b. It is common knowledge that the petroglyph still exists (Dauenhauer &

Dauenhauer 1987: 330)

c. Indeed, in the immediately following line, the narrator tells the addressee that they will go visit it later.

(22) Conclusion:

If cessation were an entailment of the Tlingit past tense, then sentence (21) would be inconsistent with the immediately following line of text.

2.2 Cancellation of Cessation Inference with Explicit Statements of Ignorance!! (23) Key Claim

Although the cessation inference cannot be cancelled by explicitly negating it (9)-(10), it can be cancelled via an explicit statement of ignorance concerning the present.

! 8!

(24) Cancellation of Cessation Inference with Statement of Ignorance The following dialog was translated from a parallel English one. a. Joe: Sheet’kát kuxwatéeni, John xwasiteen. (SE) Sitka.to PFV.1sgS.travel.SUB John 3O.PFV.1sgS.see When I traveled to Sitka, I saw John. b. Sue: O! John gé áwu hú Sheet’ká? (SE) Oh John Q there.LOCP him Sitka Oh, is John there in Sitka? c. Joe: Ha, áa yéi teeyín. EXCLM there.at IMPFV.3sgS.be.PST Tlél xwasakú ch’a yeisú áa yéi teeyí (SE) NEG 3O.PFV.1sgS.know just still there.at IMPFV.3sgS.be.SUB Well, he was there. I don’t know if he’s still there. (25) Cancellation of Cessation Inference with Statement of Ignorance The following dialog was translated from a parallel English one.

a. John: Táam gé ch’a yeisú tá? (WF) Tom Q just still IMPFV.3sgS.sleep Is Tom still sleeping? b. Mary: Yeisú dziyáak táayin. still earlier IMPFV.3sg.sleep.PST Hél xwasakú ch’a yeisú shákdé tá. (MD, WF) NEG 3O.PFV.1sgS.know just still DUB IMPFV.3sgS.sleep Well, he was sleeping earlier. I don’t know if he is still sleeping. (26) Conclusion: If cessation were an entailment of Tlingit past tense, then the past-tense

sentences in (24)-(25) would be inconsistent with the following statements of ignorance.

2.3 Absence of Cessation Inference in Embedded Clauses (27) Key Claim

If a propositional attitude verb in Tlingit is past tense, then the verb of its complement can also bear past tense, without contributing any cessation inference/implication.

! 9!

(28) Embedded Past Stative Lacking Cessation Implication

Scenario: When I was a kid, my uncle would bring over all this really great food to our house. I naturally assumed that he made it, and that he was a really great cook. Turns out, though, that he just bought the food from restaurants downtown!

Yéi xwajeeyín ax káak kúnáx k’idéin at sa.éeyin. (SE)

3O.IMPFV.1sgS.think.PST my uncle very well 3sgS.IMPFV.cook.PST I thought that my uncle cooked really well. (29) Embedded Past Stative Lacking Cessation Implication

Scenario: When you were a child growing up in Southeast Alaska, Juneau seemed like a big city to you. Of course, as you got older and visited places like Seattle and San Francisco, you learned that Juneau wasn’t so big after all.

a. Yéi xwajéeyin Jóonoo aan tleinx áwé satéeyin. (WF) 3O.IMPFV.1sgS.think.PST Juneau city big FOC IMPFV.3S.be.PST I thought that Juneau was a big city. b. Yéi xwajéeyin Jóonoo kúnáx géiyin. (MD) 3O.IMPFV.1sgS.think.PST Juneau really IMPFV.3S.big.PST I thought that Juneau was big. (30) Conclusion:

If cessation were an entailment of Tlingit past tense, then the sentences in (28)-(29) would entail that the speaker had believed:

a. That their uncle stopped being a good cook some time in the past (28) ! b. That Juneau stopped being a big city some time in the past (29)

But, these are not the beliefs entertained by the speaker in the associated scenarios.

!(31) Key Observation

The use of [PST] tense in (28)-(29) is quite reminiscent of the simultaneous readings that are possible for embedded (plain) past in English and (some dialects of) Modern Hebrew.

a. Yosef xašav še Mariam ahava oto Yosef think-PST that Marian love-PST him Yosef thought that Mariam loved him

(i.e., Yosef thought, “Miriam loves me”) (Ogihara & Sharvit 2012)

• In Cable 2015, I argue that Ogihara &Sharvit’s (2012) analysis of (31a) could be productively extended to Tlingit sentences like (28)-(29).

! 10!

2.4 Contexts Where the Present is ‘Irrelevant’ (32) Key Claim If the context is constructed so that the ‘present’ is not conversationally relevant, then past statives no longer trigger a cessation inference. (33) Past Stative Without Cessation Inference, When Present is ‘Irrelevant’

Scenario: You’ve lived your whole life here in Juneau. Someone is interested in what Southeast Alaska was like in the 1950s. So, they ask you, “Where did you live in the 1950s?” You want to answer that you lived here in Juneau.

a. Jóonoox’ yéi xat teeyín. (CM) Juneau.at IMPFV.1sgS.be.PST I lived in Juneau. b. Aa yéi xat teeyín, Jóonoo (SE) there IMPFV.1sgS.be.PST Juneau I lived there, in Juneau. (34) Conclusion:

If cessation were an entailment of Tlingit past tense, then the sentences in (33) would be inconsistent with the assumption that the speaker still lives in Juneau.

2.5 Languages Where Cessation Inference Is Semantic? (35) Central Conclusion from Data Above

• The cessation inference associated with past-tense statives in Tlingit is not semantic.

• Therefore, the past marking of Tlingit is not a ‘discontinuous past’ in the theoretically important sense of (17)

(36) Obvious Question This Raises

Are there clearer examples of ‘discontinuous past’, languages where the cessation inference truly is part of the lexical semantics of the past-tense morphology?

! 11!

(37) A Case for Skepticism

• To my knowledge, no one has yet tested other putative instances of ‘discontinuous past’ in the environments in Sections 2.2 – 2.4

• In the following section, I will put forth an analysis of the Tlingit cessation inference

that rests upon the following key fact:

o Tlingit is a ‘superficially tenseless’ language (Matthewson 2006)

o That is, unmarked verbs in Tlingit allow for past-tense construal, and so (in this sense), past-tense marking is optional.

• If we were to extend this pragmatic account to all putative instances of ‘discontinuous

past’, it would straightforwardly capture the following fact:

o Putative instances of ‘discontinuous past’ only ever appear in superficially tenseless – i.e., ‘optional tense’ – languages (Plungian & van der Auwera 2006)

3. Cessation Inferences in English and in Tlingit I begin by presenting an analysis of cessation inferences in English, one that based heavily upon – but is importantly different from – Altshuler & Schwarzschild’s (2013) (38) Referential Semantics for Tense (Abusch 1997, Kratzer 1998, et multia alia) a. [[ [Tense PST ]i ]]w,t,g = g(i), only if g(i) < t; undefined otherwise b. [[ [Tense PRES ]i ]]w,t,g = g(i), only if g(i) = t; undefined otherwise (39) Formal Semantics of Aspect a. [[ [Aspect IMPFV ] ]]w,t,g = [ λQ<εt> : [ λt’ : ∃e. Q(e) & t’ ⊆ T(e) ] ] b. [[ [Aspect PFV ] ]]w,t,g = [ λQ<εt> : [ λt’ : ∃e. Q(e) & T(e) ⊂ t’ ] ]

! 12!

(40) The Syntax and Semantics of Tense and Aspect, Part 1 a. Sentence: Scotty was anxious. b. LF of (40a), In a Context with Salient Past Topic Time: [TP [T PST ]i [ IMPFV [ Scotty be anxious ] ] ] c. Predicted Truth-Conditions: [[(40b)]]w,t,g is defined only if g(i) < t if defined, is true iff ∃e. anxious(e) & Thm(e) = Scotty & g(i) ⊆ T(e)

‘There is an eventuality (state) e of Scotty being anxious whose run-time T(e) contains the topic time g(i).’

(41) Special Assumption: Existential Past When No Past Topic Time

When there is no salient past time antecedent for a past tense, the Tense head can be existentially bound.

(42) The Syntax and Semantics of Tense and Aspect, Part 2 a. LF of (40a), In a Context with No Salient Past Topic Time: [TP ∃i [TP [T PST ]i [ IMPFV [ Scotty be anxious ] ] ] ] b. Predicted Truth-Conditions: 6 [[(42a)]]w,t,g is true iff ∃t’ . t’ < t & ∃e. anxious(e) & Thm(e) = Scotty & t’ ⊆ T(e)

‘There is a past time t’ and there is an eventuality e of Scotty being anxious whose run time T(e) contains the past time t’.’

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!6!Note that when a pronoun is existentially bound, its presuppositional features effectively serve to place additional restrictions on the existential quantification (Cable 2013).!!

! 13!

(43) The Open Interval Hypothesis (Altshuler & Schwarzschild 2013) • The run-time of state is an open interval. • That is, if s is a state and t’ is a temporal instant contained within T(s) (t’ ⊆ T(s)),

then there is a t’’ such that t’’ < t’ and t’’ is also contained within T(s) (t’’ ⊆ T(s)). !(44) Key Generalization about Cessation Implicatures in English

In English, a cessation implicature is triggered when a past tense sentence is uttered in a context where there is no salient, topical past time (see (3) vs. (6)) (Musan 1997, Altshuler & Schwarzschild 2013)

!(45) Cessation (in English) as a Quantity Implicature (Altshuler & Schwarzschild 2013) !

• In a context where there is no salient past topic time, a past tense sentence like (40a) will have the existential truth-conditions in (42b).!

!• The present tense variant of such a sentence will have the truth-conditions in (46c).!

!• Given the Open Interval Hypothesis (43), the present tense sentence will

asymmetrically entail the (existential) past tense truth-conditions in (42). !!

• Therefore, a scalar implicature will be generated that the present tense sentence (46a) is false.!

!• Therefore, the past state (of anxiety) invoked by the past-tense sentence does not

extend into the present.!!!(46) Key Result: Present Tense Variant of (40a) a. Sentence: Scotty is anxious. b. LF of (52a): [TP [T PRES ]i [ IMPFV [ Scotty be anxious ] ] ] c. Predicted Truth-Conditions: [[(52b)]]w,t,g is defined only if g(i) = t if defined, is true iff ∃e. anxious(e) & Thm(e) = Scotty & g(i) ⊆ T(e)

‘There is an eventuality e of Scotty being anxious whose run-time T(e) contains the topic time g(i) (which is equal to the utterance time t)’

! 14!

But what is the nature of the cessation inference in Tlingit?... We’ve already seen in Section 1 that it has important differences with the one in English…

!!(47) Key Feature of Tlingit: (Superficial) Tenselessness Unmarked stative sentences in Tlingit allow for either past or present tense construal.

a. Kuwak’éi b. Kei kukgwak’éi IMPFV.weather.be.nice FUT.weather.be.nice

The weather is/was nice. The weather will be nice. (48) ‘Superficial Tenselessness’ = Unpronounced Non-Future Tense (Matthewson 2006)

Following Matthewson (2006), I assume ‘superficially tenseless’ languages have an unpronounced ‘non-future’ tense, with the semantics below.

[[ [Tense NFUT ]i ]]w,t,g = g(i), only if ¬(t < g(i)); undefined otherwise !!!(49) Key Consequence: Topic Times Covering Both Past and Present

In superficially tenseless languages, the topic time of an ‘unmarked’ sentence can be an interval overlapping the evaluation time t and some past time t’.

(50) Illustration from Lillooet (Matthewson 2006) a. Scenario: Theresa threw up a few minutes ago. Charlie is throwing up now.

b. wat’k’ kw [ s-Theresa múta7 s-Charlie ]. vomit.PFV DET NOM-Theresa and NOM-Charlie Theresa and Charlie threw up / are throwing up (51) Analysis of Lillooet Data, Using ‘NFUT’ (Matthewson 2006) ! a. LF of (50b): [TP [T NFUT ]i [ PFV [ [Theresa and Charlie] vomit ] ] ] b. Truth-Conditions: [[(51a)]]w,t,g is defined only if ¬(t < g(i)) if defined, is true iff ∃e. vomit(e) & Ag(e) = T+C & T(e) ⊆ g(i)

‘There is an event e of Theresa and Charlie throwing up whose run time T(e) is contained in the topic time g(i).’ ! !

!

! 15!

(52) Topic Times Covering Both Past and Present in Tlingit

Scenario: You are watching two kids, Tom and Anne. Your friend Linda is downstairs reading. First, Tom starts jumping around the room. You tell him to stop, and he does. Soon, though, Anne starts jumping around the room. At this point, Linda opens the door and asks “What is all the noise up here?”

a. Táam ka Anne át has wujik’éin. (MD) Tom and Anne PFV.3plS.jump.around Tom and Anne jumped around/are jumping around. (53) Some Observations about ‘Cessation Contexts’ in Tlingit a. We saw that the cessation inference in Tlingit disappears in contexts where:

(i) the speaker explicitly says that they’re ignorant about the present (Section 2.2), (ii) the present is clearly not conversationally relevant (Section 2.4)

b. Generalization:

In Tlingit, the cessation inference arises when the present is conversationally relevant, and the speaker is knowledgeable about it.

• Note: The cessation inference cannot be defeated in (9)-(10) and (12)-(13), contexts where:

(i) the speaker is presented as being knowledgeable about the present, and (ii) the present is conversationally relevant

o In (9) and (10), there is a conjoined assertion about the present. o In (12), the speaker knows/believes that Joe is still working, and it’s relevant

to Sue what he is doing at present. o In (13), the speaker knows/believes that Joe is still singing, and it’s relevant to

Sue if he is. (54) Proposed Pragmatic Principle, Informally Stated

If the present time t is conversationally relevant (topical), then the topic time should contain it (in addition to potentially other topical past times)

A Comment: This principle may be a specific instance of a more general principle: a. More General Principal (?):

If both t’ and t’’ are salient, relevant and topical, then the topic-time should contain both t’ and t’’

b. Evidence from English (?): 1. Discourse: Dave jumped, then Fred jumped. Mary was dancing 2. Intuition: Time of Mary’s dancing covers both jumpings.

! 16!

(55) More Formal Statement: Include Topical UT inside the TT, Whenever Possible If all the following conditions hold, then the speaker must use sentence S1, and not S2: a. Sentences S1 and S2 are identical except for their T-heads (T1 and T2). b. Both the Utterance Time t and some past time t’ < t are salient and relevant.

c. [[ T1 ]]w,t,g contains both t’ and t, while [[ T2 ]]w,t,g = t’. d. Both S1 and S2 are ‘assertable’ (i.e., speaker’s knowledge entails them). (56) The Pragmatic Reasoning Generating Cessation Inference in Tlingit

Let us assume that the speaker has used a past tense stative in a context where the present (Utterance Time / Evaluation Time) t is topical/relevant/salient.

a. By assumption, both Utterance Time t and past time t’ < t are salient and relevant.

By assumption, the speaker has used a past tense sentence S2. b. They did not use a non-future tense sentence S1, with a T-head denoting an

interval covering both t’ and t c. S1 and S2 are identical except for the T-node (T1 and T2). d. [[ T1 ]]w,t,g contains both t’ and t, while [[ T2 ]]w,t,g = t’.

e. Given (55), it must be that the non-future sentence S1 is not assertable.

f. Past tense sentence S2 entails that there’s a state of the relevant sort containing past time t’

g. Non-Future tense sentence S1 entails that there’s a state of the relevant sort

containing past time t’ and the utterance time t

h. Therefore, the speaker’s knowledge entails that that there is a state of the relevant sort containing past time t’ but either:

(i) They don’t know whether that past state continues into present, or (ii) They know that the past state doesn’t continue into present (57) Key Consequence The ‘cessation inference’ in (56hii) won’t go through if:

• The present (Utterance Time / Evaluation Time) isn’t relevant (Sect. 2.4)

• The speaker is ignorant of whether the past state continues into the present (Sect. 2.2)

! 17!

4. Conclusion

• Despite their differing behavior from cessation inferences in English, the cessation inferences of past tense statives in Tlingit are not semantic

o That is, they are not encoded into the lexical semantics of Tlingit past tense

• Instead, these inferences have the following sources:

a. The fact that Tlingit is a ‘superficially tenseless’ (optional tense) language. b. A special principle requiring the Topic Time to contain the Utterance Time,

whenever possible (55)

• It appears that this analysis of Tlingit will generalize to other languages that have been claimed to have ‘discontinuous past’ marking.

• Therefore, there is no evidence for ‘discontinuous past’ as a special (sub)category of tense.

• Therefore, the answer to our ‘overarching question’ in (1c) can still be ‘NO!’

(58) Extensions / Further Applications

In Cable (2015), I show that the analysis in Section 3 can also account for the special inferences observed for past-marked perfectives, both in Tlingit and in other alleged ‘discontinuous past’ languages (Plungian & van der Auwera 2006).

Acknowledgements:

Deepest thanks and gratitude are owed first and foremost to Tlingit elders Margaret Dutson (Sháax’ Sáani), Selena Everson (Kaséix), William Fawcett (Kóoshdaak’w Éesh), Carolyn Martin (K’altseen), John Martin (Keihéenák’w), and Helen Sarabia (Kaachkoo.aakw). I am deeply grateful for all that they have taught me regarding the Tlingit language. Special thanks are also owed to Lance Twitchell (X’unei), James Crippen (Dzéiwsh), Matthew Rolka, and Alice Taff, for their crucial logistical support of this study. Finally, I would like to dedicate this work to the memory of Dick Dauenhauer (Xwaayeenák), who with his wife Nora Marks Dauenhauer (Keixwnéi), has done more than anyone else to advance the study and preservation of Tlingit language and culture.

This material has benefited greatly from comments I received at the University of Connecticut and at Semantics of Under-Represented Languages in the Americas (SULA 8; University of British Columbia). Special thanks to Henry Davis, Rose-Marie Déchaine, Lisa Matthewson, Hotze Rullmann, Ryan Bochnak, Jürgen Bohnemeyer, Kathryn Davidson, Amy Rose Deal, Peter Jacobs, Angelika Kratzer, Andrew McKenzie, Peter Klecha, Laura Kalin, Neda Todorovic, and Jonathan Bobaljik.

This research was made possible through grants from the Jacobs Research Funds and the University of Massachusetts HFA Research Council. In addition, this material is based upon work supported by the National Science Foundation, under Award No. BCS-1322770.

! 18!

References Abusch, Dorit. 1997. “Sequence of Tense and Temporal De Re.” Linguistics and Philosophy 20: 1-50. Altshuler, Daniel and Roger Schwarzschild. 2013. “Moment of Change, Cessation Implicatures, and Simultaneous Readings.” In Chemla, E., V. Homer, and G. Winterstein (eds) Proceedings of Sinn und Bedeutung 17. 45-62. ENS Paris. Boas, Franz. 1917. “Grammatical Notes on the Language of the Tlingit Indians.” University of Pennsylvania University Museum Anthropological Publications 8:1. Cable, Seth. 2013. “Beyond the Past, Present and Future: Towards the Semantics of Graded Tense in Gikuyu.” Natural Language Semantics 21: 219-276. Cable, Seth. 2015. “The Tlingit Decessive and ‘Discontinuous Past’: The Curious Implicatures of Optional Past Tense.” Under revision for NLLT. Copley, Bridget. 2005. “O’odham Cem: When the Actual World Isn’t Inertial.” In Becker, Michael and Andrew McKenzie (eds) Proceedings of Semantics of Underrepresented Languages in the Americas (SULA 3). Amherst, MA: GLSA. Dauenhauer, Nora Marks and Richard Dauenhauer. 1987. Classics of Tlingit Oral Literature, Volume 1: Haa Shuká, Our Ancestors: Tlingit Oral Narratives. Juneau, AK: Sealaska Heritage Foundation Press. Kagan, Olga. 2011. “The Actual World is Abnormal: On the Semantics of the Bylo Construction in Russian.” Linguistics and Philosophy 34: 57-84. Kamp, Hans and Uwe Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer. Kamp, Hans, Uwe Reyle, and Antje Rossdeutscher. 2013. “Perfects as Feature Shifting Operators.” Manuscript. Institut für Maschinelle Sparachverarbeitung and Universität Stuttgart. Kratzer, Angelika. 1998. “More Structural Analogies Between Pronouns and Tenses. In Strolovitch, Devon and Aaron Lawson (eds) Proceedings of Semantics and Linguistic Theory 8. Ithaca, NY: CLC Publications. Leer, Jeff. 1991. The Schetic Categories of the Tlingit Verb. Dissertation. University of Chicago. Magri, Giorgio. 2011. “Another Argument for Embedded Scalar Implicatures, Based on Oddness in Downward Entailing Environments.” Semantics and Pragmatics 4(6): 1-51 Matthewson, Lisa. 2004. “On the Methodology of Semantic Fieldwork.” International Journal of American Linguistics 70: 369-415. Matthewson, Lisa. 2006. “Temporal Semantics in a Superficially Tenseless Language.” Linguistics and Philosophy 29: 673-713. Moens, Marc & Mark Steedman. 1988. “Temporal Ontology and Temporal Reference.” Computational Linguistics 14: 15-28. Musan, Renate. 1997. “Tense, Predicates, and Lifetime Effects.” Natural Language Semantics 5: 271-301. Ogihara, Toshiyuki and Yael Sharvit. 2012. “Embedded Tenses.” In Binnick, R. (ed) The Oxford Handbook of Tense and Aspect. Oxford: Oxford University Press. Plungian, Vladamir A. and Johan van der Auwera. 2006. “Towards a Typology of Discontinuous Past Marking.” Sprachtypologie und Universalienforschung 59: 317-349. Sauerland, Uli. 2002. “The Present Tense is Vacuous.” Snippets 6: 12-13. Singh, Raj. 2011. “Maximize Presupposition and Local Contexts.” Natural Language Semantics 19: 149-168.

! 19!

Story, Gillian. 1966. A Morphological Study of Tlingit. Language Data: Amerindian Series 7. Dallas, TX: Summer Institute of Linguistics. Story, Gillian and Constance Naish. 1973. Tlingit Verb Dictionary. Fairbanks, AK: Alaska Native Language Center. Thomas, Guillaume. 2014a. “Nominal Tense and Temporal Implicatures: Evidence from Mbyá.” Natural Language Semantics 22: 357-412. Thomas, Guillaume. 2014b. “The Present Tense is Not Vacuous.” Journal of Semantics DOI:10.1093/jos/ffu010 Williams, Frank, Emma Williams, and Jeff Leer. 1978. Tongass Texts. Fairbanks, AK: Alaska Native Language Center. ! Appendix: Linguistic and Methodological Background (59) The Tlingit Language (Lingít): A Few Bullet Points

• Traditional language of the Tlingit people of Southeast Alaska, Northwest British Columbia, Southwest Yukon Territory (shaded area in map below) 7

• Member of the Na-Dene language family; distantly related to Athabaskan languages (e.g. Navajo, Slave, Hupa). Shares the complex templatic morphology of this family.

• Highly endangered; ≤ 200 speakers, all over 60, mostly over 70. Several fluent/near-

fluent second language learners; a few raising their children in the language.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!7!Map graphic by X’unei Lance Twitchell. Available at http://tlingit.info/.

! 20!

(60) Notes on the Data and Methodology

• Unless otherwise noted, all data in this handout were obtained through interviews with native speakers of Tlingit (2012, 2014, 2015)

• Six fluent elders have participated; all are residents of Juneau, AK; all are speakers of

the ‘Northern Dialect’ of Tlingit

o Margaret Dutson (Sháax’ Sáani) o Selena Everson (Kaséix) o William Fawcett (Kóoshdaak’w Éesh) o Carolyn Martin (K’altseen) o John Martin (Keihéenák’w) o Helen Sarabia (Kaachkoo.aakw).

• Interviews lasted two hours and were held in a classroom at the University of Alaska

Southeast; 2-4 elders were present at each interview.

• Speakers were asked to translate English sentences paired with particular ‘scenarios’, as well as to judge the ‘correctness’ (broadly speaking) of constructed Tlingit sentences relative to those ‘scenarios’ (Matthewson 2004).

• The scenarios were described to speakers in English, both orally and with

accompanying written text. (61) Important Terminological Note: Past = ‘Decessive Epimode’

The Tlingit verb forms that I have been labeling ‘past tense’ are referred to by specialists as being in the ‘decessive epimode’

(62) Morphology of the Decessive Epimode Verbs in the decessive appear with a [-I] classifier and a suffix. The form of the suffix depends upon the kind of clause headed by the verb.

a. Decessive Suffix in Main Clause: -(y)een / -(w)oon

Exact appearance depends upon three morphophonological rules (Leer 1991) • In Northern Dialect, the vowel can also surface as short: -(y)in / -(w)un

b. Decessive Suffix in Subordinate Clause: yéeyi

• This marker can also modify nouns, where it means ‘ex-, former’. c. Decessive Suffix in Atributive Clause: -(y)i / -(w)u

• The morphophonological rules mentioned in (62a) also apply to this suffix.

! 21!

(63) Detailed Illustration of Decessive Morphology a. (Plain) Stative Imperfective b. Decessive Stative Imperfective Kuwak’éi Kuk’éiyeen Ku-ya- k’éi Ku-∅-k’éi-yeen AREAL-CL[-D,∅,+I]-√good AREAL-CL[-D,∅,-I]-√good-DEC

IMPFV.weather.be.nice IMPFV.weather.be.nice.DEC The weather is/was nice. The weather was nice (but has turned bad). (64) Early Semantic Descriptions of the Decessive

• Boas (1917: 84) and Story (1966: 143) simply describe the decessive as a past tense

• Naish (1966: 91) and Story & Naish (1973: 356) add the statement that it describes past times ‘when the situation was other than it is’.

(65) The Semantic Description of Leer (1991: 460-478)

“[Decessive] generally means that [the sentence] was true at some point in the past, but is no longer true at present.” (Leer 1991: 461)

!Note:

• The main claim of this talk is that the suffixes in (62) are ‘just’ optional past tenses o (Boas was right!)

• Therefore, I gloss them throughout as ‘PST’ (past tense)