Variable-force modality

(1)

Variable-force modality

Igor Yanovich

Universität Tübingen

Rutgers University February 14, 2014

(2)

The plan

1 Introduction: modal semantics and modal quantificational force

2 A first glimpse into variable force: Old Saxon and St’át’imcets

3 Variable force in Old and Middle English Single meaning vs. genuine ambiguity

4 Variable force in the Pacific Northwest

“True” variable force, and a♦without a dual

5 Variable force in Old and Modern Ukrainian

“Triangular” ambiguity

6 Conclusion: what we now know about the variable-force landscape

(3)

A new type of modal quantificational force

Modals with the force of possibility (♦): can,may

Modals with the force of necessity (): must,have to,should Variable-force modals:

sometimes are translated into English with ♦, other times with St’át’imcets: [Rullmann et al., 2008]

Gitksan: [Peterson, 2010]

Nez Perce: [Deal, 2011]

Old and Middle English: [Yanovich, 2013a]

Old and Modern Ukrainian: this talk

...older Germanic, Old Polish, Finnish, Danish, Burmese, and counting

(4)

Semantic anatomy of a modal

A modal’s semantics = modal flavor + modal force

Modal flavors: epistemic, deontic, circumstantial, ability, etc. etc.

New modal flavors keep getting discovered:

[Portner, 2009] argues forquantificationalmodality

[Yanovich, 2013b] makes a case forsuggestion/advicemodality [Knobe and Szabó, 2013] show the existence of mixed

deontic-circumstantialmodality

But until recently, we only had ♦andfor modal force

(5)

Variable-force modality is only a descriptive label

Variable-force modality need not be justonenew modal force

[Rullmann et al., 2008], [Peterson, 2010], [Deal, 2011], [Kratzer, 2012]:

at most 3 different types of variable force, but 5 incompatible analyses.

[Yanovich, 2013a]: two other semantic types of variable force in Old and Middle English

Old and Modern Ukrainian: a yet new type

The goal of this talk: build a semantic typology of variable force, based on what we know at the moment

(6)

Where we are

(7)

What variable force looks like: Old Saxon

Old Saxon môtan(cf. Dutch moeten, Germanmüssen):

(1) endi and

ûs us

is is

firinun urgent

tharf, need

<...> that that

wi we

it it

an in

thesumu this

lande land

at from

thi you linôn

learn môtin.

môtan.subj

(Heliand 2428-30)

‘And there is an urgent need for us <...>

that wemaylearn from you (=Christ) in this land.’

(2) thes that

môtun môtan

gi you.pl

neotan use

forð, forth

sô huue sô whoever

gerno gladly

uuili will

gode god

theonogean, serve, uuirkean

do

aftar after

is his

uuilleon.

will

(Heliand 1144-6)

‘Youmustuse that (=the saving force) from now on, every one of you who wants to serve God gladly and to do after God’s will.’

(8)

What variable force looks like: St’át’imcets

St’át’imcets (Salish) variable-force deontic ka: [Rullmann et al., 2008, (31)]

(3) lán-lhkacw already-2sg.subj

ka deon

áts’x-en see-dir

ti det

kwtámts-sw-a husband-2sg.poss-det

‘You {must/can/may} see your husband now.’

How to visualize this? Think road traffic control.

(9)

Three possibilities for variable-force semantics

Possibility 1: variable-force modals have semantics different from either ♦or , with no perfect translation correlate

Possibility 2: variable-force modals are ambiguous between ♦and Possibility 3: variable-force modals are regular♦s ors, but the overall system works so that their distribution ends up being wider

It turns out that eachof the three possibilities is actualized in some language.

Moreover, there are several subtypes of possibility-1 and possibility-2 systems.

(10)

Where we are

3 Variable force in Old and Middle English Collapse variable force in Old English Finnish variable force

True♦/ambiguity in Middle English

(11)

Old English *motan

(4) bruc enjoy

þenden while

þu you

mote

motan.3sg.subj

manigra many

medo rewards

(Beo 1177-8)

‘Enjoy, while youmot, many rewards’

The (near) consensus story:

1 Earliest recorded OE:*motanambiguous between♦and

2 Very few-uses in Early OE (close to 0%)

3 Slow growth of-uses, reaching 100% in the 15-16th cent.

[Ono, 1958], [Tellier, 1962], [Visser, 1973], [Goossens, 1987]...; cf. [Solo, 1977]

(12)

Questions for the standard analysis: regularity

Meaning change isregular.

For m¯otan in Germanic, there is indeed regularity:

A very similar situation form¯otanand its cognates in Old English, Old Saxon, Old High German: a modal seemingly ambiguous between♦ and, with♦prevalent.

The modern descendants of those modals (must, Germanmüssen, Dutchmoeten) are familiar modals.

But other♦modals don’t just become ♦-ambiguous!

And they don’t turn intos either.

⇒ there must be something special about*motanand its cognates

(13)

Questions for the standard analysis: specific mechanism

Meaning change involves semantic reanalysis.

But why would speakers reanalyze♦as ? Two explanations in the literature:

Through permission implying obligation(e.g. [Traugott, 1989])

“You may go” from an authority implies that “you must go”.

...but then any♦-deontics would be able to turn into

Through “must not” ≈“may not” (e.g. OED)

The speakers reanalyze the negative instances, and after that take care of the positive cases.

...but all♦deontics have fixed scope¬>♦([van der Auwera, 2001]), so again, any♦is predicted to be able to change into ...and besides, won’t work for German, asnicht müssenis¬>

...finally, where would the pressure to reanalyze positive cases come from?

(14)

Variable-force analysis of [Yanovich, 2013a, Ch.4]

Old English *motan

not a♦, but anon-ambiguous variable-force modal

⇓

Early Middle English *moten

♦-ambiguity, with more frequent

⇓ Early Modern English must

pure : the less productive♦-uses have been lost

(15)

How would you translate motan?

(5) Ac but

se that

ðe which

unwærlice unwarily

ðone that

wuda wood

hiewð, hews,

&

and sua so

his his

freond friend

ofsliehð, slays, him

to.him bið is

nidðearf necessary

ðæt that

he he

fleo flee.subj

to to

ðara those.gen

ðreora three.gen

burga city.gen

anre, one.dat

ðæt that

on in

sumere some

ðara of.those

weorðe become.subj

genered, saved,

ðæt that

he he

mote

motan.prs.subj libban;

live

‘But he who unwarily hews wood and by that slays his friend, it is necessary for him that he flee to one of those three cities, so that he be saved in one of them, so that hemote

live.’ (CP:21.167.15)

(6) ealneg always

hi they

wepað, weep

&

æfter after

ðæm the

wope weeping

hi they

gewyrceað obtain

ðæt that

hi they

moton motan.pres eft

again wepan.

weep

‘always they are weeping, and after the weeping they make it so that theymotonweep

again.’ (CP:54.421.14)

(16)

The main idea of the “collapse” analysis

5 is a typical “possibility example”, while 6 is a typical “necessity use”.

But in both cases, both ♦andtranslation may be appropriate.

Imagine a set of accessible worlds uniform with regard to propositionp.

Given that set, ♦p⇔p. Either statement says the same.

Now, in natural language it’s not so clean because of the pragmatics.

When people talk about necessity, they often imply there is a force imposing it.

When they talk about possibility, they often imply somebody is interested in that possibility.

⇒unlike in logic, people may find one rendering better than the other.

(17)

Possibility-necessity collapse: the intuition

(7) a. Hu how

mæg can

he he

ðonne then

beon be

butan without

gitsunge, avarice ðonne

when he he

sceal had.to

ymb about

monigra many

monna men’s

are property

ðencan, think gif

if he he

nolde would.not

ða ða when

he he

moste

motan.sg.past.subj ymb about

his his

anes?

only

(CP:9.57.19)

b. Translation by [Sweet, 1871]:

“How can he be without covetousness when he has to consult the interests of many, if formerly he would not avoid it when hehad toconsult his own interests alone?”

c. Translation by H.W. Norman, printed in [Giles et al., 1858]:

“How can he be without covetousness when he must think about many men’s sustenance, if he would not when hemightthink about his own alone?”

Not much contrast between the ♦andreadings:

it was an open possibility for the subject to think only about their own benefit, but they also actually thought only about themselves before being promoted.

(18)

Possibility-necessity collapse: the intuition

(5) A typical “possibility example”:

‘But he who unwarily hews wood and by that slays his friend, it is necessary for him that he flee to one of those three cities, so that he be saved in one of them, so that hemote

live.’ (CP:21.167.15)

would≈mote≈may

(6) A typical “necessity example”:

‘always they are weeping, and after the weeping they make it so that theymotonweep

again.’ (CP:54.421.14)

have to≈moton≈may

(19)

A focused Old English dataset: Alfredian prose

Early OE prose: core Alfredian texts (late 9th/early 10th cent.)

C(ura) P(astoralis) (edition [Sweet, 1871])

Bo(ethius) (edition [Godden and Irvine, 2009])

Sol(iloquies) (edition [Carnicelli, 1969])

Best possible shot at geographical and temporal consistency for the period.

72 instances of*motan

A caveat: though the situation in Alfredian OE seems close to that in other early Old English texts and in other early Germanic, it is not identical.

Nor should we expect it to be: both range of dialectal variation and pace of change may be significant with modals, as variationist sociolinguists showed.

(20)

Why use a focused dataset I

1 Dialectal variation may be huge

Present-Day English, the use of different deontics across the British Isles:

from [Tagliamonte and Smith, 2006]

(21)

Why use a focused dataset II

2 Change may be very fast

The deontic system of Toronto English changed in 3 apparent-time generations:

from [Tagliamonte and D’Arcy, 2007], Toronto English

(22)

Alfredian *motan: the collapse analysis

What we can say about *m¯otanin the Alfredian dataset:

Observation

In all 72 examples, virtually no contrast between the ♦andreadings.

With a regular♦,♦p does not entail thatphas to happen.

(8) Youmaytake this apple. But it’s not that you have to.

(9) My electric billscanbe paid online, though I never tried.

In Alfredian OE, possibilities expressed bymagan‘can, may’ and aliefed‘permitted’ work the same way, being consistent with¬p.

But notmotan!

(23)

Alfredian *motan: the collapse analysis

Analysis for motan(p)

Acc. relation: metaphysical modal base, stereotypical ordering source Presupposition: ♦p →fut(p) ifp has a chance to actualize, it will

Assertion: ♦p

Metaphysical modal base: all w⁰ sharing the history of the actualw Stereotypical ordering source: w⁰⁰ where things go normally

E.g., the person in question doesn’t win a lottery, etc.

(24)

How the collapse analysis works

(5) A typical “possibility example”:

‘But he who unwarily hews wood and by that slays his friend, it is necessary for him that he flee to one of those three cities, so that he be saved in one of them,so that he mote

live.’ (CP:21.167.15)

w: “purpose” worlds where the purpose clause is true

⇓

w⁰: metaphysical correlates for eachw, sharing its history

⇓

w⁰⁰: those metaphysical correlates where things proceed normally Presupposition: either he lives in allw⁰⁰, or doesn’t live in allw⁰⁰ Assertion: he lives in allw⁰⁰

Paraphrase: “given that either in all possible futures lives, or in all of them he dies, it’s necessary for him to flee to one of those cities so that hemay (would)live”.

(25)

How the collapse analysis works

motan(p) conveys both inevitability (in the presupposition) and openness of possibility (in the assertion)

Variable-force translation effect:

Inevitability is stressed⇒translation

Openness of possibility is stressed⇒♦translation Rarity of*motan:

Few contexts would support the collapse presupposition.

And indeed,*motanis rare in Alfredian OE:

≈70*motanvs.≈700sculan(>shall) and≈1000magan(>may)

(26)

Alternative explanations?

Could Alfredian *motanbe genuinely ♦/ambiguous?

Nope. If it were, we would find*motan not only where♦and collapse, but also where “must(p)” is different from “may(p)”

Could Alfredian *motanbe regular ♦?

My analysis says that♦andcollapse in the context where*motan occurs. So a usual♦without a collapse presupposition would be just as good.

But first, without the presupposition we cannot explain why*motan only occurs in collapse contexts.

Second, we know that atsomepoint,*motan cannot be analyzed as a pure♦any longer. So saying it was a♦in Alfredian OE doesn’t add any explanatory power.

(27)

Modal flavor of *motan

(10) Metaphysical: worlds sharing the same history It might rain every day this summer.

(11) Circumstantial: worlds where a given set of facts is true

During the next hurricane, this tree can easily fall onto my roof.

(12) Deontic: worlds where the rules are followed You may take this apple.

Circumstantial and metaphysical are close: if the facts include everything about the world, the two collapse

Deontic and metaphysical may be hard to distinguish in texts, especially when it is about what God or fate allow

I found no examples that would clearly exclude the metaphysical analysis. Hence my claim about the modal flavor. But it’s more a reasonably-supported hypothesis than a proven fact.

(28)

The shape of the Alfredian modal system

Alfredian Old English

ability circumstantial deontic

♦ magan magan non-verbal

n/a sculan sculan

metaphysical/circumstantial/deontic

♦+ collapse presupposition motan

On the one hand, we have fairly regular♦ andmodals.

On the other, we have a special, very restricted variable-force modal.

(29)

Where we are

(30)

Are there more systems like that?

Finnish is possibly a good candidate.

Primary source: the description of Finnish modals in [Kangasniemi, 1992]

Secondary source: recent work by Kehayov and Torn on modals in Balto-Finnic

Most Finnish modals are familiar ♦s or s.

E.g., voidais a regular circumstantial/deontic♦,

pitääis a regular deontic/teleological/circumstantial , and so forth.

But saada(historically from ‘to get’) may translate as a♦ or a.

(31)

Finnish variable-force modal saada

(13) Hakaluissa

square.bracket.pl.iness olevaa

be.part.prttv A A

merkkiä symbol.prttv

saavat may.3pl käyttää

use

ainoastaan only

liittomme union.gen.1pl

jäneset.

member.pl

‘The symbol A in square bracketsmayonly be used by the members of our union.’ [Kangasniemi, 1992, p.91, (7)]

(14) ... me we

saimme

have.to.pst.1pl tämän this.gen

Kariniemen Kariniemi.gen

käskystä order.ela

sittej then

jäädä stay niin

so ku like

jälkeempäin afterwards

asiaa

matter.prttv

selvittämääj clear.up.3inf.ill

...

‘... we thenhad to stay afterwards on this Kariniemi’s orders to clear the matter up ...’ [Kangasniemi, 1992, p.102, (44)]

(32)

Reasons to think saada has a uniform meaning

[Kangasniemi, 1992, p.62]: “One motivation for the use ofsaadain expressions of necessity may be the speaker’s or writer’s pursuit of irony, stating that the actor has the possibility of doing something that he or she does not want to, and moreover, thatall other possibilities are excluded” (emphasis mine)

(15) Saat saada-2sg

lähteä go

matkalle trip-All

taivaaseen.

heaven-Ill

‘Youmay/have to set out for your trip to heaven.’

[Kangasniemi, 1992, p.322-3]: “The interpretation of[15]depend[s]on whether the agent wants to perform the act or not, i.e. whether the addressee of sentence[15]wants to go for a trip to heaven ... . Thus sentence[15]could be interpreted as permission in a religious context (which was in fact the case) but as an obligation or a threat in James Bond adventure.”

(33)

Reasons to think saada is ambiguous

[Kehayov and Torn, 2005] examinedsaada’s cognates in other Finnic:

If saadaand its cognates had a uniform meaning, we’d expect every modal flavor to feature both ♦and. The table shows it is not so.

(34)

Old English *motan and Finnish saada

Old English *motan:

lexical source uncertain (but hardly ‘to get’)

non-ambiguous variable force, via the collapse presupposition very restricted with regard to modal flavors: no strong evidence for anything but metaphysical

predominantly looked like♦in OE, but later developed into Finnish saada:

lexical source: ‘to get’

semantics unclear

unclear ifsaadais (modal-force-)ambiguous or not

ifsaadahas uniform force, it’s unclear whether it is due to the collapse presupposition or something else

a wide range of modal flavors: circumstantial, deontic, epistemic predominantly looks like♦, future diachronic trajectory to be seen

(35)

Where we are

(36)

A focused Middle English dataset: ‘AB language’

Early ME prose: ‘AB language’ (first half of 13th cent.)

A group of texts written within a few miles from each other. Clearly the product of a single common writing tradition, written in the same dialect and sharing orthography.

Seinte Margarete (SM) (edition [d’Ardenne, 1977])

Ancrene Wisse (AW) (edition [Millett, 2005])

SM predates AW by several decades.

76 instances of*m¯oten.

(37)

The Middle English descendant of *motan

Try to translate Middle English *motenin this passage:

(16) Hwen-se whenever

ye you

moten moten

to to

eani any

mon man

ea-wiht give

biteachen, the

the hand

hond not ne

comes cume not

nawt out

ut

“Whenever youmotgive anything to anyone, the hand shouldn’t

come out.” (AW 2:192-3)

This is a most typical kind of use of moten in AW.

(38)

The ME dataset: Ancrene Wisse and Seinte Margarete

58 instances of moten inAncrene Wisse (only 2 in negative clauses)

5 main types of uses:

unavoidability (circ.,≈modernhave to)accounts for>50%

moral instruction (deontic,≈modernmust,ought) wish, prayer

“open possibility”

under attitudes (grant,swear, etc.), with unclear semantic import 18 instances of moten inSeinte Margarete (only 1 in a negative clause)

A slightly different distribution:

no strict demarcation between prayers and other♦types moral-instruction uses are emerging from circumstantialuses

(39)

Deontic reading

“Moral instruction”: deontic

(17) < ... >teke this, hamotyet thurh hire forbisne ant thurh hire hali beoden

yeoven strengthe othre, ant uphalden ham, thet ha ne fallen i the dunge of

sunne. (AW 3:259)

‘...besides this, shemustalso through her example and through her holy prayers give strength to others, and hold them up so that they do not fall in the filth of sin.’

(40)

...but Early ME *m¯ oten is not yet a pure

“Open possibility”: in≈5 out of 58 examples in AW, and more inSM, we seem to have a genuine existential meaning:

(18) Þah þe flesch beo ure fa, hit is us ihaten þet we halden hit up. Wa we motendon hit, as hit is wel ofte wurðe, ah nawt fordon mid alle;

(AW 3:284-5)

‘Though the flesh is our foe, it is commanded to us that we hold it up.

Woe wemaydo it as it is well often worthy of, but we should not destroy it altogether.’

(41)

...but Early ME *m¯ oten is not yet a pure

Prayers:

(19) I þe wurðgunge, Iesu Crist, of þine tweof apostles, þet Ichmoteoueral folhin hare lare, þet Ichmotehabben þurh hare bonen þe tweolf bohes þe

bloweð of chearite, (AW 1:174-6)

‘In honor, Jesus Christ, of your twelve apostles,mayI everywhere follow their teaching,mayI have through their prayers the twelve branches that blossom with love’

(42)

*m¯ oten under attitudes

In attitudes: exact meaning unclear, but not empty; close to prayers

(20) Thet ich thurh the lare of the Hali Gastmotehalden foreward, he hit yetti

me thurh ower bonen. (AW 3:644-5)

‘That I, through the teaching of the Holy Spirit,maykeep the agreement, let Him (=God) grant it to me through your prayers.’

⇒this type of use is most frequent in the late entries of Petersborough chronicle (underask,agree,forbid,grant,decree)

(43)

*m¯ oten’s modal neighbors in the AB language

In OE, *m¯otan was outside of the “regular” modal system:

ability circ. deontic

♦ magan magan non-modal

— sculan sculan

circ./deontic

♦+ collapse presup. motan

But in the 13th cent.,*m¯otenis an integral part of the system.

moten

circumstantial necessity deontic necessity various non- ahen(>modernought)

only deontic uses, mostly reportative sculen(>modernshall)

deontic uses, both performative and reportative future uses

“subjunctive” uses (≈modernwould)

(44)

*m¯ oten in Ancrene Wisse: true ambiguity

5 main types of uses: circumstantial , deontic, prayer ♦-like use,

“open possibility” ♦, unclear use under attitudes Thereadings are straightforward.

The non-readings are less so.

Consider prayers such as “May I everywhere follow the teaching of the apostles”.

Here,mayis not a typical♦semantically. But at the same time, oncemotenloses its other♦uses completely, it is replaced in prayers withmay.

In Alfredian OE, all types of uses could be explained with one meaning.

Not anymore in the AB language! and non-cannot be unified.

⇒ ME*motenis a genuinely ambiguous variable-force modal.

(45)

Intermediate summary

Old English *motan: non-ambiguous “collapse” variable force, little flavor flexibility

Middle English*moten: truly ambiguous between differentand♦ readings, significant flavor flexibility

NB:and♦uses do not come in pairs!

Finnish saada: unclear if ambiguous or not, significant flavor flexibility

⇒ can be similar to OE *motan, to ME*moten, or to neither

(46)

Future directions for Germanic and Finnic...

Older Germanic are relatively well documented, so we can look at

*motan’s cognates, and at its development in English in more detail.

For example, what is the right semantics formôtanin Old SaxonHeliand, cf. 1 and 2? In the first approximation, OSamôtanseems to have several types of uses, including “open possibility”, “destiny”, and perhaps a rare deonticas in 2, but do they actually feature different meanings?

In Finnic, there seems to be microvariation which could shed light on the underlying semantics ofsaada and its cognates.

Two directions to pursue: 1) study of naturally generated texts; 2) fieldwork.

Once the range of possible uses in Finnish is identified, one can proceed to the smaller Finnic languages.

(47)

...and East-(South-)Asian..

[van der Auwera et al., 2009] discuss get-based modals in

South-(East-)Asian languages, and mention a few with ♦anduses.

Example: Burmese ya’, [Vittrant, 2004, p.313]

(21) di this

ña’-ne night-day

nin 2sg

yoPSinyon movies

‘Twa go

lo’

like ya’

get tE rea.ass

‘Youcango to the movies tonight.’

(22) ‘min 2sg

ko obj

t9-son-t9-ya one-clf-one-thing

me‘myan ask

khE’

pst yin if TwEP-TwEP-leP-leP

quickly

phye answer

ya’

get mE irr.ass

‘If he asks you something, youmustanswer him quickly.’

(48)

Where we are

(49)

Adding Pacific-Northwest variable force to the “map”

What we have seen:

Old English: non-ambiguous, narrow variable force Middle English: genuinely ambiguous variable force

Adding St’át’imcets, Gitksan, Nez Perce:

St’át’imcets: exclusively variable-force modals

Gitksan: a mixed system with some epistemic variable force

Nez Perce: variable-force effects for a familiar♦in an unfamiliar system

(50)

St’át’imcets

St’át’imcets (Salish family) is the only known language

where all modals are variable-force

We saw deonticka in 3. And here is metaphysical/futurekelh:

(23) lh-tq-álk’-em-an

comp-touch-string-mid-1sg.conj

ka-gúy’t-kan-a

circ-sleep-1sg.subj-circ kélh fut

tu7 then

‘If I drive Imight(accidentally) fall asleep.’ [Rullmann et al., 2008, (20)]

(24) o, oh

xílh-ts-kan do-caus-1sg.subj

kelh fut

áti7, deic

nilh foc

t’u7 just

s-lh-nás-acw

nom-comp-go-2sg.conj í7wa7

accompany

‘Oh, I’lldo it, if you come along.’ [Rullmann et al., 2008, (25)]

(51)

The place of St’át’imcets in the variable-force typology

The modal system of St’át’imcets:

deontic metaphysical/future epistemic and evidential

♦- ka kelh k’a; ku7(?); -an’(?)

Properties of variable force in St’át’imcets

“Possibility” and “necessity” readings with the same flavors, suggesting no true ambiguity

“Necessity” readings are the default ([Rullmann et al., 2008, Sec.2.4]) With negation: at least “possibly not”, sometimes also “necessarily not”

No collapse presupposition! See 23 and 24.

⇒adifferentkind of unambiguous variable force than in Old English

(52)

Gitksan

Gitksan (Tsimshian family): a mixed system, with variable force in the epistemic-and-evidential domain

circumstantial deontic observable evidence

♦ da’ak

¯hlxw anook

sgi ¯ nakw

epistemic and evidential ima(’a); gat

[Peterson, 2008]: in most contexts, imais variable-force.

But in observable-evidence contexts, it is in the same paradigm with nakw. In such contexts, imauniformly conveys possibility.

(53)

The place of Gitksan in the variable-force typology

Properties of variable force in Gitksan

A mixed system, with variable force in the epistemic-evidential domain Unlike in St’át’imcets, the default reading seems to be “possibility”

When variable-forceimais in a paradigm withnaxw,imauniformly gets possibility translations

Interaction with negation:

Reportative evidentialgattakes clause-level scope and doesn’t interact with negation (“gat(p)” is “I heard that¬p”, but never “I didn’t hear

thatp”) [Peterson, 2010, pp. 66-8, 149-50]

General-purpose inferential evidentialimaonly gets “possibly not”

readings [Peterson, 2010, p. 45], [Matthewson, 2013, Sec. 3.1]

No♦-collapse presupposition.

(54)

Nez Perce

Nez Perce (Sahaptian): a circumstantial/deontic variable-force o’qa.

[Deal, 2011]’s analysis foro’qa

Observation 1: in downward-entailing contexts,o’qabehaves as a♦ Observation 2: no would-bedual foro’qa

Claim: o’qais a regular♦

Deriving variable force: without a dual, no scalar implicature¬ Speaker sayso’qa(p). That simply means that there’s an accessible world wherepis true.

Suppose thatp is true in all accessible worlds. In English, you can assertmust(p)in this case. So when you say insteadmay(p), it’s implicated that there are accessible¬p worlds.

But in Nez Perce, there is no way to saymust(p). Even if all worlds are p-worlds, the only expression you have iso’qa.

(55)

The place of Nez Perce in the variable-force typology

Nez Perce vs. St’át’imcets and Gitksan:

In Nez Perce,o’qawith negation only means “necessary not”

⇒not as St’át’imcets and Gitksan variable-force modals do However, the competition effect forimain Gitksan looks very similar to Nez Perce. Unfortunately, it’s virtually impossible to check Gitksan imain other DE contexts, [Matthewson, 2013, Sec.3.1]

Nez Perce vs. Old English:

Interaction with negation is similar Nez Perce has no collapse presupposition

Moreover, Old English*motanhas a would-be dual: sculan Nez Perce vs. Middle English:

Interaction with negation is different: in Middle English, both scopes attested

Modal flavors forand non-readings of ME*motendo not have the same range of modal flavors, unlike♦anduses ofo’qa

(56)

The emerging typology

Old English *motan: unambiguous “collapse” variable force Type 1a

Middle English*moten: ♦-ambiguity Type 2a

Finnish saada: could be like Old or Middle English Type 1a or 2a?

Nez Perce o’qa: usual♦, but without adual Type 3 St’át’imcets: unambiguous variable force Type 1b Gitksanimaandgat:

could be like Nez Perceo’qa, but hard to tell Type 3 or 1c?

Up next: Old and Modern Ukrainian

a yet different type of ambiguous variable force. Type 2b

(57)

Where we are

(58)

Ukrainian: a HAVE-based variable-force modal

Proto-Slavicèì³òè>Old Ukrainian èìàòè >Mod. Ukrainian мати In Old Ukrainian (14-16 centuries):

necessity (at least deontic) futurate

possibility (at least deontic)

In Modern Ukrainian (late 19-21 centuries):

deontic and epistemic necessity futurate

possibility

My sources: The book of Lutsjk castle, 1560-1;Documents from Volynj, 16th century; the letters of Lesya Ukrajinka, late 19th cent.;Fieldwork in Ukrainian Sex by Oksana Zabuzhko, born in Lutsjk.

(59)

maty and its cousins

Proto-Slavicèì³òè‘have’ ⇒future and/or obligation in many Slavic Old Bulgarian (a.k.a. Old Church Slavonic)im³ti(10-11th centuries):

futurate

very few non-futurate meanings Middle Russianim³ti(14-17th centuries):

futurate (sometimes with modal overtones) however, virtually no clear modal meanings lost by the 17th-18th century

Old Czechjmieti(13-15 centuries):

obligation futurate

Old Polishmiec(14-15 centuries):

obligation futurate

possibility — but not clear if it’s the same as in Old Ukrainian

(60)

Modern Ukrainian: deontic necessity maty

(25) Що what

ж part

до about

моєї my

повiстi, novel

то, part

далебi, truly

не not

знаю, know.1sg

як how

з with

нею it буде,

will.be бо because

не not

розумiю, understand.1sg

як how

маю maty

думати think

про about

вiдносини relations

“Зорi”

of.Zorya до to

мене me

‘Regarding my novel, I truly don’t know what will happen with it, as I don’t understand what Ishouldthink about how “Zorya”[a literary journal]views me.’

NB: a possibility translation would also make some sense here (what I may think), but hardly a future one.

(61)

Modern Ukrainian: future maty

(26) Сiчова Sich

кна-кна kna-kna

зайнята is.occupied

страшенно terribly

зборами with.gathering

радикалiв, of.radicals

що which мають

maty бути be

близько close.to

апрiля, April

через because.of

те that

кна-кна kna-kna в in

ажитацiї, excitement

немов as.if перед

before

виборами.

elections

‘The Sich kna-kna(family term for Ukrayinka’s brothers —IY)is greatly interested by the gathering of radicals whichwilltake place some time around April, and because of that the kna-kna is excited as if before the elections.’

Not pure future, but ratherplanned futureandpredicted future.

NB: a necessity translation would also make some sense here (the elections must occur around April), but not a possibility one.

(62)

Modern Ukrainian: possibility maty

(27) Ну, well

та this

es this

ist is

eine an

alte old

Geschichte, story

i, and

певне, surely

вона it

Вам to.you

так so

вже already сприкрилась

bored

досi, until.now

але but

мене me

жаль pity

бере, takes

що that

у at

нас us

на in

Українi Ukraine нiяк

in.no.way не not

скiнчаться end

одвiчнi eternal

сiї those

спори, quarrels,

та and

й part

як how

мають maty скiнчитись,

end

коли if

сперечники quarrelers

одно one

одного another

не not

розумiють.

understand

‘Well,es ist eine alte Geschichte, and surely by now you’ve had enough of it already, but still it pities me that for us in the Ukraine, those eternal quarrels never end, and indeed howcouldthey end if the quarrelers don’t understand each other.’

No reading “it’s abstractly possible” for such examples

Instead: “There are enough resources for the possibility to be realizable”.

NB: a future translation possible (how they would end), but not a necessity one.

(63)

Relationships between different meanings of maty

For ME m¯oteninAncrene Wisse, we often had clear demarkation:

deontic-for ethical contexts

circumstantial-for practical contexts with-♦for prayer contexts

etc.

But the different meanings ofmatyare connected to each other.

future (ex. 26)

ww

(ex. 25) ^//♦(ex. 27)

gg

(64)

Variable-force ambiguity in Middle English vs. Ukrainian

In Middle English (and presumably in other Middle Germanic as well), the variable-force ambiguity didn’t exist for too long.

and non-readings were distributed by context, and pretty much isolated from each other

Non-meanings became marginal and died out

Within Slavic, not all languages turned they HAVE-word in a variable-force modal. Most just made out of it a futurate and a deontic .

But in Ukrainian, once the ♦--futurate ambiguity arose, it was present in almost the same form for half a millenium.

Why such a difference in stability?

⇒ Perhaps the interconnectedness of meanings makes “Type 2”, ambiguity-based variable force more stable.

(65)

Semantics for different maty

Obligation maty(p): in all worlds where the current world’s obligations are met, p takes place.

Future maty(p): in all worlds that develop according to the current plans or predictions,p takes place

Possibility maty(p): the preconditions are met for bringingp about in every accessible world (where one would try doing so)

There are discussions in the literature as to whether ability modals are pure♦s, and the conclusion is that they are in fact more complex. See [Portner, 2009, pp. 201-3] and references therein.

(66)

Where we are

6 Conclusion: what we now know about the variable-force landscape

(67)

Variable force on the map, before

(68)

Variable force on the map, now

(69)

The landscape of variable force

Three major types:

Type 1: unambiguous variable force (or variable force proper) Type 2: genuine♦-ambiguity

Type 3: familiar♦ormodality in a system unusually shaped Types 1 and 2 clearly havesubtypes with different semantics:

Type 1: St’át’imcets vs. Old English Type 2: Middle English vs. Ukrainian

Variable force iswidely distributed geographically.

No clear correlation between the type of variable force and geography:

Type 1: Pacific Northwest and Europe Type 2: Northern Europe and Eastern Europe

(70)

Healey, Daniel Donoghue, Regine Eckardt, Kai von Fintel, Olga Fischer, Martin Hackl, Irene Heim, Sabine Iatridou, Natasha Korotkova, Ian MacDougall, Lisa Matthewson, Paul Portner, Katrina Przyjemski, Donca Steriade, Sali Tagliamonte, and Elizabeth Traugott. If not for Lauri Karttunen, I wouldn’t have learned of [Kangasniemi, 1992].

The traffic control example is due to Frank Veltman.

Some parts of this work were presented at University of Ottawa, Georgetown, Rutgers, NYU,Systematic Semantic Changeat UT Austin, SALT at UC Santa Cruz, and at University of Amsterdam. The project benefitted greatly from the comments I received at those venues. All remaining mistakes are my responsibility only.

Corpora used:

York-Toronto-Helsinki Parsed Corpus of Old English prose (YCOE) Penn Parsed Corpus of Early Middle English (PPCEME)

Parsed Corpus of Early English Correspondence (PCEEC) Russian National Corpus, historical part (www.ruscorpora.ru)

The extensive commentary toBoethiusin [Godden and Irvine, 2009] was of great help in identifying the correspondences between the Latin original and the OE translation.

(71)

Carnicelli, T. A. (1969).

King Alfred’s version of St. Augustine’s Soliloquies.

Harvard University Press, Cambridge, MA.

d’Ardenne, S. (1977).

The Katherine Group edited from MS. Bodley 34.

Société d’Edition “Les Belles Lettres”, Paris.

Deal, A. R. (2011).

Modals without scales.

Language, 87(3):559–585.

Giles et al., editor (1858).

The whole works of king Alfred the Great: with preliminary essays illustrative of the history, arts, and manners of the ninth century.

Bosworth & Harrison, London.

Godden, M. and Irvine, S. (2009).

The Old English Boethius.

Oxford University Press.

Goossens, L. (1987).

Modal tracks: the case ofmaganandmotan.

In Simon-Vanderbergen, A.-M., editor,Studies in honour of Rene Derolez, pages 216–236. Vitgeuer, Gent.

Kangasniemi, H. (1992).

Modal expressions in Finnish.

Suomalaisen Kirjallisuuden Seura, Helsinki.

Kehayov, P. and Torn, R. (2005).

Modals in finnic.

Talk at the 38th Societas Linguistica Europea,

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CCYQFjAA&url=http%3A%2F%2Fkodu.

(72)

ut.ee%2F~pdkehayo%2Fmodal_power_presentation.ppt&ei=J_P6UvXuCM6B7QbK2IHoBQ&usg=

AFQjCNG2M5f4CNlgmdC7HAV_76ZdV_FjNA&sig2=iaO7w5VjW6zgkCs0DkGfPg&bvm=bv.61190604, d.ZGU&cad=rja.

Knobe, J. and Szabó, Z. G. (2013).

Modals with a taste of the deontic.

Semantics and Pragmatics, 6(1):1–42.

Kratzer, A. (2012).

Modals and conditionals.

Matthewson, L. (2013).

Gitksan modals.

International Journal of American Linguistics, 79(3).

Millett, B. (2005).

Ancrene Wisse. A corrected edition of the text in Cambridge, Corpus Christi College, MS 402, with variants from other manuscripts.

Drawing on the uncompleted edition by E.J.Dobson, with a glossary and additional notes by Richard Dance.

Ono, S. (1958).

Some notes on the auxiliary*motan.

Anglica, 3(3):64–80.

Peterson, T. (2008).

Pragmatic blocking in gitksan evidential expressions.

In Schardl, A., Walkow, M., and Abdurrahman, M., editors,Proceedings of the 38th Meeting of the North East Linguistic Society, pages 219–232, Amherst, MA. GLSA Publications.

Peterson, T. (2010).

Epistemic Modality and Evidentiality in Gitksan at the Semantics-Pragmatics Interface.

PhD thesis, University of British Columbia.

(73)

Portner, P. (2009).

Modality.

Rullmann, H., Matthewson, L., and Davis, H. (2008).

Modals as distributive indefinites.

Natural Language Semantics, 16(4):317–357.

Solo, H. J. (1977).

The meaning of*motan. A secondary denotation of necessity in Old English?

Neuphilologische Mitteilungen, 78:215–232.

Sweet, H. (1871).

King Alfred’s West-Saxon version of Gregory’s Pastoral Care, volume 45 and 50 ofEarly English Text Society.

Tagliamonte, S. and D’Arcy, A. (2007).

The modals of obligation/necessity in Canadian perspective.

English World-Wide, 28(1):47–87.

Tagliamonte, S. and Smith, J. (2006).

Layering, competition and a twist of fate. deontic modality in dialects of English.

Diachronica, 23(2):341–380.

Tellier, A. (1962).

Les verbes perfecto-présents et les auxiliaires de mode en anglais ancien: (VIII^eS. - XVI^e S.).

C. Klincksieck, Paris.

Traugott, E. C. (1989).

On the rise of epistemic meanings in English: an example of subjectification in semantic change.

Language, 65(1):31–55.

van der Auwera, J. (2001).

On the typology of negative modals.

(74)

In Hoeksema, J., Rullmann, H., Sánchez-Valencia, V., and van der Wouden, T., editors,Perspectives on negation and polarity items, pages 23–48. John Benjamins, Amsterdam.

van der Auwera, J., Kehayov, P., and Vittrant, A. (2009).

Acquisitive modals.

In Hogeweg, L., de Hoop, H., and Malchukov, A., editors,Cross-linguistic Studies of Tense, Aspect, and Modality, pages 271–302. John Benjamins, Amsterdam.

Visser, F. T. (1963-1973).

An historical syntax of the English language.

E. J. Brill, Leiden.

Vittrant, A. (2004).

La modalité et ses corrélats en birman dans une perspective comparative.

PhD thesis, Paris 8.

Yanovich, I. (2013a).

Four pieces for modality, context and usage.

PhD thesis, MIT, Cambridge, MA.

Directed by Kai von Fintel, Sabine Iatridou (committee chairs) and Irene Heim.

Yanovich, I. (2013b).

Symbouletic modality.

Talk at CSSP 2013. Handout available at

http://web.mit.edu/yanovich/www/papers/Yanovich-CSSP- symbouletic-modality- talk.pdf.