Modal Indefinites in Mi’gmaq∗ What do we mean by ‘random’? Gretchen McCulloch [email protected] Semantics of Underrepresented Languages of the Americas (SULA 8) University of British Columbia, May 16, 2014

1

Introduction

1.1

What is a modal indefinite?

• Most indefinite items make a simple claim about existence. • For example, the indefinite natuen ‘someone’ in Mi’gmaq (Algonquian, Canada) asserts that there exists some person such that Mary talked to that person, as shown in (1). Mali gelulapnn nat-uen-n Mary talked.to NAT-who-OBV ‘Mary talked to someone ’ i.e. Mary talked to a person.1

(1)

• However, Mi’gmaq has an additional indefinite with a more complex meaning, shown in (2). • The indefinite ta’n pas’g wen ‘just anyone, any old person’ asserts not only that there exists some person that Mary talked to, but also that for all contextually-relevant people, she could have talked to that person. ∗

Also spelled Mi’gmaw, Mi’kmaq, Mi’kmaw, Micmac. This paper uses the Listuguj orthography and reflects the dialect spoken in Listuguj, Quebec. The apostrophe after a vowel indicates a long vowel and elsewhere indicates a schwa. Many thanks to Janine Metallic, MaryAnn Metallic, Janice Vicare and Joe Wilmot for their time and patience in teaching me the language. Also thank you to Luis Alonso-Ovalle, Jessica Coon, Alan Bale, and the members of the Mi’gmaq Research Group at McGill for comments. 1 The extra -n on -uenn/wenn is obviation marking, indicating a less important (non-topical) third person argument. See Manyakina (2012) for discussion. Many additional morpheme and glossing details are omitted, especially from verbs, for readability.

1

May 15, 2014

Mali gelulapnn ta’n pas’g wen-n Mary talked.to that just who-OBV ‘Mary talked to just anyone, any old person’ i.e. Mary talked to a person, and she could have talked to anyone.

(2)

• The meaning of ta’n pas’g wen can thus be divided into two parts: an existential part that is the same as natuen, and an additional layer that is paraphrased with a modal (‘she could have talked to anyone’), despite the fact that there is no overt modal in the sentence. • Indefinites with modal content (‘modal indefinites’) are interesting because they contain a modal meaning in the nominal domain, unlike typical modals such as can, should, must, etc. which are associated with verbs. • Modal indefinites have been studied cross-linguistically under several names: – Dayal (1998) – English any as a ‘free choice item’ – Kratzer & Shimoyama (2002) – German irgendein as an ‘indeterminate pronoun’ – Choi & Romero (2008) – Korean amwu-na as an ‘existential free choice item’ – Alonso-Ovalle & Menéndez-Benito (2010, in prep.) – Spanish algún and uno cualquiera, as ‘epistemic indefinites’ and ‘modal indefinites’ → For the purposes of this paper, I refer to all of these items as modal indefinites. Outline Section 2 Overview of indefinites in Mi’gmaq from a typological perspective. Section 3 Rejecting previously-described types of modality for ta’n pas’g+wh in episodic sentences2 • German irgendein and Spanish algún have an epistemic modality (speaker ignorance) (Kratzer & Shimoyama 2002, Alonso-Ovalle & Menéndez-Benito 2002) • Korean amwu-na has counterfactual modality (Choi & Romero 2008) • Spanish uno cualquiera has a particular flavour of agent-oriented modality (AlonsoOvalle & Menéndez-Benito in prep.) Section 4 Comparison with English “methodless” just Section 5 Modality of ta’n pas’g+wh: agent, if any, is not aware of how they came to the decision to act in a particular way. Section 6 Conclusion and further directions 2

I do not compare ta’n pas’g+wh with English any because any is ungrammatical in episodic sentences. For example *Mary talked to anyone. However, see Section 4 for a comparison with English just.

2

May 15, 2014

2

General description of indefinites • Indefinite pronouns in Mi’gmaq are formed from the same base as interrogatives (wh-words). • Using the typological descriptors in Haspelmath (1997), there are four series (columns in the table below) and 6 ontological categories (rows), although not all predicted forms are permitted, especially in the lower series. • Forming indefinites from wh-words, as seen here in Mi’gmaq, is common cross-linguistically. Haspelmath (1997, p. 26) suggests that a majority of the world’s languages have indefinite pronouns derived from interrogatives (63 out of his 100-language sample). Table of Mi’gmaq indefinite pronouns3 null/interrogative nattampas Person wen natuen ta’n pas’g wen Thing goqwei natgoqwei ta’n pas’g goqwei Place tami natami ta’n pas’g tami Manner talnatalta’n pas’g talProperty talamu’g natalamu’g ta’n pas’g talamug Amount tasig *natasig *ta’n pas’g tasig Determiner tegen +N *nattegen +N ta’n pas’g tegen +N

(3)

• Bare items, as in the first column of the table, are found sentence-initially as wh-words (4) and non-initially in downward-entailing contexts including polar questions (5), conditionals (6), and negation (7) as negative polarity items. (4)

wen telim’sg’s? who told.you.IRR? ‘Who told you?’

(5)

telim’sg’s wen? told.you.IRR who ‘Did anyone tell you?’

(6)

nemij wen, tlimitis you.see.SUBJ who, tell.me ‘If you see anyone, tell me.’

(7)

Ma’li mu nemiagup’n wen-n Mary not saw who- OBV ‘Mary didn’t see anyone.’ • The nat- series in the second column is composed of the wh-word plus an indefinite marker nat- (probably related to na, a focus particle). It is found in specific (8) and nonspecific (irrealis) (9) constructions. 3

Note that -uen in natuen ‘someone’ is a orthographic variant of wen ‘who, person’.

3

May 15, 2014

(8)

nat-uen pegising’p (= pegising’p nat-uen) NAT -who arrived ‘Someone arrived.’ Context 1: The speaker knows who arrived (Someone arrived...it’s Mary). Context 2: The speaker does not know who arrived (Someone arrived, but I don’t know who).

(9)

amujpa nat-uen piluwei pipanimatis have.to NAT-who else you.will.ask ‘You’ll have to ask someone else’ Context: I don’t have what you’re looking for, and I’m not sure who would have it. • The third column, the ta’n pas’g series, is composed of a wh-word plus an indefinite marker ta’n pas’g. Items in the ta’n pas’g series are found in both episodic sentences such as (10) and modal sentences such as (11).

(10)

Mali megng’p ta’n pas’g tegen wi’gatign Mary picked that just which book ‘Mary picked just any book, any old book’ i.e. Mary picked a book, and she could have picked any book.

(11)

Mali gis-megng ta’n pas’g tegen wi’gatign Mary can-pick that just which book ‘Mary can pick any book’ i.e. Mary can pick a book, and she can pick any book

• Morphologically, ta’n pas’g consists of the complementizer ta’n ‘that, which’ plus pas’g ‘just, except, only’ and a wh-word.4 (12)

a.

Piel mat’nas’n ta’n pas’g tegen ji’nm-ul? Peter fight that just which man- OBV ‘Did Peter fight just any man?’

b.

Moqwa, pas’g matnapn ji’nm-ul ta’n masgelmatl No only fought man- OBV that he.hates.him ‘No, he only fought the man that he hates’

• The ta’n in ta’n pas’g cannot be acting as a complementizer because it is allowed in sentences with only a single verb. • For the remainder of this paper I focus on the modal portion of the meaning of ta’n pas’g+wh, which I have paraphrased above with the modal can. 4

Ta’n pas’g is sometimes reduced to tampas in rapid speech. For example, (10)-(11) could also be said as Mali (gis-)megng’p tampas tegen wi’gatign. Pas’g is probably also related to pasna ‘but, however’ (c.f. gisna ‘or’). Unlike indefinite items in the languages discussed below, the wh-word with ta’n pas’g is obligatory: *Mali megng’p ta’n pas’g wi’gatign.

4

May 15, 2014

3

Modal meanings incompatible with ta’n pas’g

3.1

Kratzer & Shimoyama: German irgendein speaker/agent indifference

• In their analysis of irgendein, Kratzer & Shimoyama (2002) propose that its covert modal is an epistemic operator, approximately equivalent to ‘as far as the speaker knows...’.5 (13)

Maria hat irgendein Buch gewählt. (Katja Münster, p.c.) Mary has IRDENDEIN book chosen ‘Mary picked any old book’ Speaker ignorance/indifference: As far as the speaker is aware, Mary could have picked any book. Agent indifference: As far as Mary cared, she could have picked any book.

• If the type of modal under which ta’n pas’g+wh is found had a meaning associated with speaker ignorance, then we would expect that ta’n pas’g+wh should be felicitous in the scenario in (14), where the speaker does not know who Mary picked but based on the speaker’s knowledge, anyone was a possible option (and where agent indifference is ruled out by context). • Instead, it is infelicitous. (14)

Speaker orientation: Mary is on a dating show where she picks the contestant she likes the best. Since I didn’t catch the show last night, I don’t know who she picked, but I know she did so intentionally. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase: As far as the speaker is aware, Mary could have picked anyone

• A second option for the modal associated with ta’n pas’g+wh is agent indifference, or bouletic modality, which would be paraphrased ‘as far as Mary’s preferences were concerned, any individual was a possible option’ and have a covert modal that resembles ‘want’. • This option is not a factor in Mi’gmaq ta’n pas’g+wh: the sentence with ta’n pas’g wen is felicitous regardless of whether Mary’s preferences are satisfied (15a) or not (15b). (15)

Background: Mary is on a game show where she has to pick someone to give money to. She is blindfolded, and spun around in the middle of a circle of people to pick someone randomly.

5

Choi (2005) points out that irgendein has several readings even without an overt modal, as shown in (13). Choi continues by suggesting a slight modification to Kratzer & Shimoyama’s account whereby the agent indifference reading is derived from the speaker ignorance/indifference one. For the purposes of this paper, however, I will examine speaker-related meanings in this subsection, and agent-related ones below.

5

May 15, 2014

a.

Agent preferences satisfied: All of the people in the circle were strangers: Mary had no preference who she chose. Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (true): As far as Mary’s preferences are concerned, she could have picked anyone

b.

Agent preferences unsatisfied: The circle contained strangers and Mary’s mother: Mary preferred to pick her mother, but couldn’t control this. Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (false): As far as Mary’s preferences are concerned, she could have picked anyone

3.2

Circumstantial modality

• Since agent preference is not a factor, another possible option is circumstantial modality: ‘as far as the circumstances were concerned, any individual was a possible option’. • The examples below show two contexts in which some contextually-relevant individuals are ruled out by the circumstances, and ta’n pas’g+wh is still felicitous. (16)

Background: Mary is on a game show where she has to pick someone to give money to. She is blindfolded, and spun around in the middle of a circle of people to pick someone randomly. a.

Circumstantial 1: Some of the people in the circle were standing behind other people, so Mary couldn’t reach them. Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (false): As far as the circumstances were concerned, Mary could have picked anyone.

b.

Circumstantial 2: Some of the people in the circle were children. Mary had her arms out at shoulder height, so she couldn’t pick the children because they were too short. Mali meginap’n ta’n pas’g wen-n 6

May 15, 2014

Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (false): As far as the circumstances were concerned, Mary could have picked anyone.

3.3

Choi & Romero: Korean amwu-na has counterfactual modality

• For Choi & Romero (2008), Korean amwu-na has counterfactual modality: for example, (17) can be paraphrased ‘John picked up a book and if the set of books had been different, then he would still have picked up a book.’ • By their logic, since John would have picked up a book regardless of the content of the set of books, it can be inferred that John doesn’t care about which book he picks, but this agent indifference is only an implicature, and the literal meaning of amwu-na only includes a counterfactual modal. (17)

John-un amwu-chayk-ina cip-ese congi-uy-ey noh-ass-ta. John-TOP AMWU-book-OR pick-and paper-TOP - LOC put-PAST- DEC ‘John picked up a random book and put it on the pile of paper.’ (Choi & Romero 2008: 51a) Paraphrase (true): John picked up a book, and if the set of books had been different, he would still have picked up a book. (paraphrase mine)

• Choi & Romero derive their notion of counterfactuality for amwu-na on von Fintel’s (2000) analysis of whatever, as found in examples like (18). (18)

Zack simply voted for whoever was at the top of the ballot (von Fintel 2000) = Zack voted for the person at the top of the ballot, and if the person at the top of the ballot had been different, he would still have voted in the same way.

• If Mi’gmaq ta’n pas’g+wh has counterfactual modality, we would expect that it should be fine in contexts where the agent has decided on a particular way of picking an arbitrary person. • For example, in (19), Mary is blindfolded and has decided to pick the first person she touches, whoever that may be. • Clearly, if the set of people to choose from were different, Mary would still pick the first person she touches, so this scenario satisfies Choi & Romero’s counterfactual modality. • However, the sentence with ta’n pas’g+wh is infelicitous in this context. (19)

Counterfactual 1: Mary is on a game show where she has to pick someone to give money to. She is blindfolded, and spun around in the middle of a circle of people to pick someone randomly. 7

May 15, 2014

Mary has mentally committed to picking the first person she touches, whoever that may be. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (true): Mary picked a person, and if the set of people had been different, she would still have picked a person. • Similarly, in the scenario in (20), Mary does not know the order in which the candidates are listed on the ballot and has decided to vote arbitrarily by picking the first candidate on the ballot, whoever that may be. • If the set of candidates were different, Mary would still vote for the candidate at the top of the ballot, so this scenario is also counterfactual according to Choi & Romero’s criteria and yet the sentence with ta’n pas’g+wh is infelicitous. (20)

Counterfactual 2: Mary is voting in an election where she doesn’t care who wins, so she’s decided to choose a candidate randomly. Mary doesn’t know what order the candidates will be listed on the ballot (alphabetical, reverse alphabetical, by party, etc) so she has decided that she will choose randomly by voting for whichever person is at the top of the ballot. (modelled on von Fintel 2000: 32) #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (true): Mary picked a person, and if the set of people had been different, she would still have picked a person.

• Note that the example in (20) does not differ in outcome from that in (21): in both cases Mary ends up picking the candidate at the top of the ballot without caring who she picks. • The only difference is in how she approaches the selection process: when she pre-commits to a method that will result in a specific person, even though she doesn’t know who the specific person is, ta’n pas’g+wh is unacceptable; but when she is unaware of how she is deciding which person to pick, ta’n pas’g+wh is fine. (21)

No method: Mary is voting in an election where she doesn’t care who wins, so she’s decided to choose a candidate randomly. Mary has decided that she will choose randomly by closing her eyes and pointing at the ballot. Her finger lands on the person at the top of the ballot. Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’ Paraphrase (true): Mary picked a person, and if the set of people had been different, she would still have picked a person. 8

May 15, 2014 • In fact, we can extend the notion of a randomizing method to a wide variety of devices, including drawing names out of a hat, spinning a roulette wheel, flipping a coin, rolling a die, and shuffling cards. • All of the sentences in (22)-(27) are counterfactual by Choi & Romero’s definition: they are all compatible with the paraphrase ‘Mary picked a person, and if the set of people had been different, she would still have picked a person in the same way.’ • However, ta’n pas’g+wh is not felicitous in any of them, which is consistent with the observation above that it is sensitive to whether the agent has used any particular method deciding who to pick. (22)

Randomizing method 1: Mary picked someone randomly by drawing names out of a hat. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

(23)

Randomizing method 2: Mary picked someone randomly using a roulette wheel with people’s names on it. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

(24)

Randomizing method 3: Mary picked someone randomly by flipping a coin. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

(25)

Randomizing method 4: Mary picked someone randomly by rolling a die with names on the sides. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

(26)

Randomizing method 5: Mary picked someone randomly by shuffling a deck of cards with names on them and drawing the top card #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

(27)

Randomizing method 6: Mary picked someone randomly by scattering a deck of cards with names on them face-down on the surface of a table and picking one up. #Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

9

May 15, 2014

3.4

Alonso-Ovalle & Menéndez-Benito: Spanish uno cualquiera indicates deliberate random choice by agent

• Another possible meaning would be deliberate random choice by the agent as with Spanish uno cualquiera, as analyzed by Alonso-Ovalle & Menéndez-Benito (in prep.). • In (28a), we see that uno cualquiera is grammatical with the agentive verb coger ‘take’, as expected by either the counterfactual or the agent’s random choice account. • However, in (28b), we see that uno cualquiera is not grammatical with the non-agentive verb tropezar ‘stumble’, which is correctly predicted by an account that takes into consideration the fact that the subject of a non-agentive like stumble cannot have chosen to do so. (28)

a.

Juan cogió una piedra cualquiera. Juan took a rock CUALQUIERA ‘Juan took a random rock.’ (data from AO&MB, paraphrases mine) Paraphrase (true): Juan took a rock, and as far as his decisions were concerned, he could have chosen to take any rock 3

b. #Juan tropezó con una piedra cualquiera. Juan stumbled with a rock CUALQUIERA ‘Juan stumbled on a random rock.’ My paraphrase based on AO&MB (false): Juan stumbled on a rock, and as far as his decisions were concerned, #he could have chosen to stumble on any rock 3 • If ta’n pas’g+wh has the same condition on deliberate random choice by the agent as uno cualquiera, then we would expect it to be excluded with non-agentive verbs, but fine with any verbs that involve a deliberate (agentive), random action. • We know from above that ta’n pas’g+wh is fine with ‘pick’, which is agentive, but the example below shows that ta’n pas’g+wh is marginally acceptable with non-agentive verbs (‘bump into’), not wholly excluded like uno cualquiera. (29)

Accidental: Mary is really absent-minded. Yesterday, she was walking down the street while texting, so she bumped into (accidentally hit) a lot of people. ?ulagu Mali pet-tesguap’n ta’n pas’g wen-n yesterday Mary accidentally-hit.PAST that just who-OBV ‘Yesterday, Mary hit anyone accidentally, bumped into anyone’

• Furthermore, the examples in (22)-(27) where the agent outsources the outcome of their decision to a randomizing method (drawing names, spinning a wheel, etc.) are perfectly fine with uno cualquiera because the agent has still decided on an act with a random outcome.6 6

These readings may be more available when it is specified that the agent does not know the outcome of the decision until the choice has already been made using the randomizing method, e.g. the name drawn from the hat must be the winner even before the agent has unfolded the slip of paper, so that the agent does not have the option to decide non-randomly between accepting or rejecting the single result returned by the randomizing method.

10

May 15, 2014

(30)

Randomizing method: Mary picked someone randomly by drawing names out of a hat. Maria cogió una persona cualquiera Mary picked a person CUALQUIERA ‘Mary picked just anyone, any random person’ (Alonso-Ovalle, p.c.)

• We can therefore see that Mi’gmaq ta’n pas’g+wh has a type of modality that is distinct from that of German irgendein, Korean amwu-na and Spanish uno cualquiera. In the following section, however, I outline how ta’n pas’g is strikingly similar to a particular use of English just which is not typically thought of as .

4

English “methodless” just

4.1

Overview

• There is a particular use of English just plus an indefinite that is strikingly similar to Mi’gmaq ta’n pas’g+wh, which I will call “methodless” just, as in (31).7 (31)

a. b. c.

Was Mary picky about which apple she took? Nah, she just grabbed one. How’d you get the door to stay open? Oh, I just used a book. How did Mary get that book off the highest shelf? Oh, she just asked a tall person.

• By contrast, English a random, like Spanish uno cualquiera, is fine in a randomizing method context. • Both methodless just and a random are, as expected, acceptable in a context where no method is specified (32). (32)

No method: We have a variety of apples on a table. Mary picks one for no discernable reason. a. Mary picked just any apple. ACCEPTABLE b. Mary just picked an apple. ACCEPTABLE c. Mary picked a random apple. ACCEPTABLE

• Using the sentence Mary just took an apple with this same meaning of just (‘How did she decide which one to take? I dunno, she just took an apple.’), then we find that it is unacceptable in the same type of randomizing method context as ta’n pas’g wen above, as shown in (33)-(34).8 7

Compare temporal just (= recently) as in Mary just/recently arrived and minimizer just (= only) as in How much would you like? Just/only a little bit. Methodless just must be unstressed and is always pronounced without the final /t/, whereas for the other types the /t/ is optional: pronouncing the /t/ in (31) forces the temporal reading. 8 It seems slightly simpler to get these interpretations in English using a concrete indefinite noun like ‘an apple’, whereas in Mi’gmaq it is simpler to do so with a wh- item like wen ‘who, a person’, probably because Mi’gmaq lacks articles.

11

May 15, 2014

(33)

Randomizing method: We have a variety of apples arranged around a lazy-susan (rotating wheel) in the middle of the table. Mary has decided that she’ll take whichever apple lands closest to her after the wheel has been spun. a. #Mary picked just any apple. U NACCEPTABLE b. #Mary just picked an apple. U NACCEPTABLE c. Mary picked a random apple. ACCEPTABLE

(34)

Randomizing method 2: Mary picked someone randomly by drawing names out of a hat. a. #Mary just picked someone. U NACCEPTABLE b. Mary picked a random person. ACCEPTABLE

• As far as I’m aware, we do not have a semantics for just that is distinct from only, although one is clearly necessary given the differences between how just and only behave in scenarios like Randomizing method (33) and (35). (35)

Randomizing method: We have a variety of apples arranged around a lazy-susan (rotating wheel) in the middle of the table. Mary has decided that she’ll take whichever apple lands closest to her after the wheel has been spun. a. *Mary picked only any apple. b. #Mary only picked an apple. (=All Mary did was pick an apple.) c. #/?Mary picked an only apple. (?= Some apples were in groups and some were alone. Mary picked one of the alone apples.)

• Although I do not propose a semantics for methodless just, in the following subsections I outline how it is similar to ta’n pas’g+wh and therefore what criteria such an account would need to take into consideration

4.2

Methodless just patterns with ta’n pas’g+wh with non-agentive verbs

• With non-agentive verbs, ta’n pas’g+wh isn’t quite wholly excluded, but is also not quite fully acceptable. (36)

Non-Agentive: Mary is really absent-minded. Yesterday, she was walking down the street while texting, so she bumped into (accidentally hit) a lot of people. a. ?ulagu Mali pet-tesguap’n ta’n pas’g wen-n yesterday Mary accidentally-hit.PAST that just who-OBV ‘Yesterday, Mary hit just anyone accidentally, bumped into any old person’

• While it is possible to use just with a non-agentive verb, such as Mary just dropped an apple, in this case just tends to get an eventive reading. Attempts to force a methodless reading on non-agentive just (‘#How did she decide which one to drop?’) also result in marginal grammaticality.

12

May 15, 2014

4.3

Methodless just patterns with ta’n pas’g+wh under negation

• In downwards-entailing contexts (under negation, in polar questions, and in the antecedent of a conditional), the method sensitivity of both ta’n pas’g and just are preserved. • Compare the equivalent sentences without ta’n pas’g and which lack English ‘just’ in the glosses, which are unacceptable in these scenarios. (37)

a.

Context: Mary went to a party and talked only to her friends, not to any random person. Mali mu gelulagp’n ta’n pas’g wen-n Mary not talked.to that just who- OBV ‘Mary didn’t talk to just anyone, any old person #Mali mu gelulagp’n wen-n Mary not talked.to who- OBV ‘Mary didn’t talk to anyone (at all)

b.

U NACCEPTABLE

Context: We know that Mary talked to her friends at the party, but we’re wondering whether she also talked to random other people she didn’t already know. Mali gelul-as-’n ta’n pas’g wen-n? Mary talk.to.IRR that just who- OBV ‘Did Mary talk to just anyone, any old person? #Mali gelul-as-’n wen-n? Mary talk.to.IRR who- OBV ‘Did Mary talk to anyone (at all)?

c.

ACCEPTABLE

ACCEPTABLE

U NACCEPTABLE

Context: Mary went to a party and talked only to her friends. She did not have fun, and we know that this is because she’s a very extroverted person and will only have fun if she talks to random people she doesn’t already know. We’re discussing what Mary should do differently in order to have fun at the next party. lpa Mali gelulaj ta’n pas’g wen-n, gise’-tew if Mary talk.to.SUBJ that just who- OBV will.have.fun ‘If Mary talks to just anyone, she will have fun #lpa Mali gelulaj wen-n, gise’-tew if Mary talk.to.SUBJ who- OBV will.have.fun ‘If Mary talks to anyone (at all), she will have fun

13

ACCEPTABLE

U NACCEPTABLE

May 15, 2014

4.4

Methodless just patterns with ta’n pas’g+wh in subject position

• This far, I have only addressed the distribution of ta’n pas’g+wh in object position. In Mi’gmaq, ta’n pas’g+wh is acceptable as the subject of an agentive verb as in (38a), but is entirely ungrammatical in the subject position of non-agentive verbs as in (38b). (38)

a.

Ta’n pas’g wen gelus’p that just who spoke ‘Anyone spoke, someone just spoke.’ Context: there was a community meeting without a panel or guest speaker, so anyone who attended was free to speak

b. *Ta’n pas’g wen pet-nisiep that just who accidentally-fell (intended) ‘Anyone fell (over) accidentally, someone just fell over.’9 • The intuitive reading for these sentences is that with the agentive verb there is some sort of external entity giving permission for anyone to speak, and that this reading is unavailable for the non-agentive verb because it does not make sense to give permission for an event that is performed accidentally. • It is not clear how the denotation that I have proposed up until this point accounts for reference to an additional external entity of this nature. • Nonetheless, ta’n pas’g+wh in subject position continues to pattern with an English indefinite plus methodless just. • We can see below that this construction in English is fine as the subject of an agentive verb like talk in (39a), marginally acceptable in a context where someone may have fallen deliberately, for example someone with a job as a stuntman as in (39b), but completely excluded when it is clear that the falling is accidental as in (39c). (39)

a. b. c.

Was there a keynote speaker at the meeting? Nah, someone just talked. Was it the stuntman who fell? *?Nah, someone just fell. Was it Mary who fell over? *Nah, someone just fell over.

• Here again, ta’n pas’g+wh and methodless just differ systematically from uno cualquiera and a random. • Both uno cualquiera and a random are equally grammatical as the subject of agentive and non-agentive verbs, but the random choice reading is not available at all in this position. Instead, the sentences and glosses in (40) have a reading that the subject is unremarkable or pejorative (Alonso-Ovalle & Menéndez-Benito, in prep). 9

The equivalent sentence with a different subject, e.g. Mali petniesiep ‘Mary fell accidentally’ is fine.

14

May 15, 2014

(40)

a.

Un estudiante cualquiera habló. a student CUALQUIERA spoke ‘An unremarkable student spoke, some random student spoke.’

b.

Un estudiante cualquiera tropezó. a student CUALQUIERA stumbled ‘An unremarkable student stumbled, some random student stumbled.’

• For the purposes of this paper, I do not attempt to provide a comprehensive account of decision-sensitive modal indefinites in subject position or under negation and other downwardsentailing environments. • However, I find it overall a positive indicator that the indefinites that I previously identified as patterning together in object position continue to behave alike in different positions, despite their obviously different composition (while pas’g can be translated ‘just’ or ‘only’, this does not account for the ta’n or the wh word).

5

Proposal for the meaning of ta’n pas’g • So far: The additional modal meaning that distinguishes ta’n pas’g wen from natuen cannot have epistemic, bouletic, circumstantial, counterfactual, nor agent’s random choice modality. • Descriptively, ta’n pas’g+wh is acceptable in contexts where there is no method by which the agent made a decision between the various available targets of the action. • More detailed environments where ta’n pas’g wen is acceptable, unacceptable, and marginal, using the following two sentences: the ‘bump’ sentence in (41a) is found in the non-agentive scenario in (42), and the ‘pick’ sentence in (41b) is found in the agentive scenarios in (43).

(41)

a. ?ulagu Mali pet-tesguap’n ta’n pas’g wen-n yesterday Mary accidentally-hit.PAST that just who-OBV ‘Yesterday, Mary hit just anyone accidentally, bumped into any old person’ b.

(42)

Mali meginap’n ta’n pas’g wen-n Mary picked that just who- OBV ‘Mary picked just anyone, any old person’

Non-Agentive Scenario a. N ON -AGENTIVE : the subject has not decided to act, so there is no agent. Context: Mary is really absent-minded. Yesterday, she was walking down the street while texting, so she bumped into (accidentally hit) a lot of people. Acceptability: ?

15

May 15, 2014

(43)

Agentive Scenarios a. M OST O PTIMAL : the agent decides based on the greatest payoff Context: Last night, Mary was on a dating show where she picked the contestant she liked the best. Acceptability: 7 b.

C ONSULTANT: the agent decides based on another person’s advice about the greatest payoff Context: Last night, Mary was on a dating show where she picked the person who was recommended to her by a matchmaker. Acceptability: 7

c.

R ANDOM M ETHOD : the agent decides based on the outcome of an independent randomizing method Context: Yesterday, Mary was on a game show where she had to pick someone arbitrarily to give money to. She picked someone randomly by spinning a roulette wheel. Acceptability: 7

d.

A RBITRARY: the agent decides for no discernible reason Context: Yesterday, Mary was on a game show where she had to pick someone arbitrarily to give money to. She was blindfolded and spun around in the middle of a circle of people, and chose one of them. Acceptability: 3

e.

F IRST R ANDOM : the agent decides on the first option, whoever that may be Yesterday, Mary was on a game show where she had to pick someone arbitrarily to give money to. She was blindfolded and spun around in the middle of a circle of people, and chose one of them. She decided to pick the first person she touched, whoever that may be. Acceptability: 7

(44)

Summary Table Verb Situation ‘bump’ N ON -AGENTIVE ‘pick’ M OST O PTIMAL ‘pick’ C ONSULTANT ‘pick’ R ANDOM M ETHOD ‘pick’ A RBITRARY ‘pick’ F IRST R ANDOM

Acceptable? Rational? Method? ? no no 7 yes yes 7 yes yes 7 no yes 3 no no 7 no yes 16

May 15, 2014 • Key to summary table: – ‘Acceptable’: whether or not ta’n pas’g+wh is allowed in a given scenario. – ‘Rational’: whether the agent acted rationally, i.e. chose the option that would lead to the greatest payoff. – ‘Method’: whether the agent decided who to pick according to any namable set of criteria. • Intuitively, the factor that the unacceptable agentive scenarios have in common is the presence of a particular method or set of criteria that the agent used in making the decision to pick a particular person. (For the moment, I put aside the N ON -AGENTIVE scenario and focus on the Agentive ones.) • Tampas+wh is only acceptable where the agent did not use a method, whether rational (e.g. M OST O PTIMAL) or random (e.g. R ANDOM M ETHOD), in deciding which person to pick. In fact, one might have assumed that in the A RBITRARY situation Mary when blindfolded would have picked the first person she touched, but if the context specifies that Mary consciously decided to pick the first person she touched, then the sentence with ta’n pas’g wen is no longer acceptable (F IRST R ANDOM). • The contrast between R ANDOM M ETHOD and F IRST R ANDOM versus A RBITRARY makes it clear that the relevant distinction is not about the randomness of the outcome but rather about the agent’s lack of awareness of the method used to decide who to pick. • Since ta’n pas’g wen is acceptable only in A RBITRARY, I argue that it is the agent’s reliance on a method for deciding that is the most relevant distinction to be made. • The three components of the meaning of ta’n pas’g wen as summarized in (45) are: an existential (a), a decision (b), and a lack of method for the decision (c). (45)

5.1

Mary picked ta’n pas’g wen. a. There was a picking event by Mary of a person. b. Mary decided to cause this picking event of this person. c. Mary does not know how she decided to pick that particular person

Deliberate events are preceded by an agent’s decision

• Following the model of Alonso-Ovalle & Menéndez-Benito (in prep.), I assume that any deliberate event, such as the picking of a person a, must be preceded by a decision on the part of the agent Mary to pick person a, and that this decision is itself an event (albeit a mental one) that has content in that it determines the set of possible events, and its minimal properties involve the agent choosing from a set of possible events.

17

May 15, 2014 • The possible events that Mary has to decide between are not all possible actions (e.g. picking a, picking no one, going to the movies); instead, she decides between possible picking events: the picking of a, the picking of b, and the picking of c. Intuitively, this is desirable, since ‘Mary picked ta’n pas’g wen’ does not mean ‘Mary picked a person, and she could have done anything’, but rather, it means ‘Mary picked a person and she could have picked anyone’. • Separating the agent’s decision between possible picking events from the picking event itself allows for an important distinction in that we can place conditions upon the way in which the decision is performed that are distinct from the way in which the picking is performed.

5.2

Even arbitrary decision-making criteria are still criteria

• While methods of performing a picking event might be purely physical and should have no bearing on whether ta’n pas’g+wh is acceptable (e.g. picking a person with a stick versus with one’s hand), methods of deciding which person to pick are mental events that reflect the agent’s intentions.10 • By making reference to how a decision is made, we can see that the unacceptable scenarios (M OST O PTIMAL, R ANDOM M ETHOD, etc.) are similar in that for each of them the agent relies on some sort of method in deciding between possible options. • A method is a function that, given the input of a set of possible options, such as picking a, picking b, or picking c, returns an output of a single option that forms the content of the agent’s decision. • One method might be the M OST O PTIMAL method as in (43a), which returns the person who leads to the greatest payoff for the agent. • Another method might be the T HIRD PARTY method as in (43c), which returns the person whose name is landed on by a roulette wheel. • In all cases with a method, the agent’s decision is to accept or reject the person returned by the method. • By contrast, when no method of any kind is specified in the context, as in (43d) A RBITRARY, the agent’s decision must be between all of the possible options.11 • Spanish uno cualquiera is similar to ta’n pas’g+wh and just in that it is excluded in payoffsensitive situations, but it is sensitive to the outcome of the agent’s decision and not how the decision itself is made, provided that the agent commits to selecting the result of the randomization sight unseen (Alonso-Ovalle, p.c.). 10

If we do not account for the agent’s intentions, then we over-predict that any picking event with a random outcome will be acceptable, that is, that ta’n pas’g+wh should be acceptable in R ANDOM M ETHOD and F IRST R ANDOM, which is not the case. 11 On the other hand, the decisions in T HIRD PARTY and A RBITRARY overlap because in both cases the agent does not act rationally, i.e. to maximize the expected payoff. This is the context in which uno cualquiera should be allowed if we follow Alonso-Ovalle & Menéndez-Benito.

18

May 15, 2014

6

Conclusion • Although Mi’gmaq is typologically unrelated to languages in which modal indefinites have previously been described, I have shown that Mi’gmaq indefinites can have modal content, like their counterparts in Spanish, Korean, and German. • Within the broader class of modal indefinites, Mi’gmaq ta’n pas’g+wh also fits into smaller sub-classes: like Spanish uno cualquiera and Korean amwu-na, ta’n pas’g+wh has both a modal meaning that refers to the agent and a composition that includes a wh-word, while German irgendein and Spanish algún have neither. • In addition to these similarities, I have shown that ta’n pas’g+wh has a slightly different flavour of modality from other modal indefinites. • Like uno cualquiera, ta’n pas’g+wh is sensitive to the decision made by the agent to engage in a particular event, but while uno cualquiera specifies that the outcome of the decision must be arbitrary, ta’n pas’g+wh specifies that the arbitrary portion is instead the process of coming to the decision, that is, the agent is not aware of any association between the decision and a method or criteria. Thus, in a context where the agent decides to pick a person based on drawing names–sight unseen–out of a hat, uno cualquiera is fine but ta’n pas’g+wh is unacceptable. • This distinction between randomness as an outcome versus randomness as a process is one that to my knowledge has not been made in previous literature (although I would welcome additional references in the question period), and yet the same notion of the agent’s ignorance of how a decision to act is made also accounts for a particular meaning of English just which I have also not found in any literature. • This usage, which I call methodless just, is found in sentences like ‘Mary just took an apple’ when it has a meaning similar, but not identical to, ‘Mary took a random apple’. I propose that methodless just has a random process meaning like Mi’gmaq ta’n pas’g+wh, while a random has a random outcome meaning like Spanish uno cualquiera. • We need a formal semantics for both methodless just and ta’n pas’g+wh that accounts for the behaviour described above: just is not identical to only and ta’n pas’g+wh is not identical to other modal indefinites, and the two are strikingly similar to each other. • Furthermore, although it is normally only the pas’g part of ta’n pas’g+wh that is paraphrased with ‘just’ or ‘only’, we clearly need to account for both the solitary meaning of pas’g and the compositional meaning of ta’n pas’g+wh, which may in turn shed light on methodless just. • Finally, this whole account calls into question the characterization of modal indefinites as belonging to the nominal domain: although items like irgendein and uno cualquiera fit quite well as determiners or determiner+adjective, both just and ta’n pas’g are clearly not determiners (and indeed must co-occur with a determiner such as a, any or tegen ‘which’), although they are not verbal either. 19

May 15, 2014

References Alonso-Ovalle, Luis and Paula Menéndez-Benito. 2003. Some Epistemic Indefinites, in M. Kadowaki and S. Kawahara (eds.), Proceedings of the North East Linguistic Society, 33, GLSA, Amherst, MA, pp. 1-12 Alonso-Ovalle, Luis & Paula Menéndez-Benito (in prep.) Probing into Random Choice Modality: Spanish Uno Cualquiera. Alonso-Ovalle, Luis & Paula Menéndez-Benito. 2010. Modal indefinites. Natural Language Semantics 18(1). 1-31. Choi, Jinyoung, 2005. Another type of free choice effect: Korean Amwu-N-lato. In Alderete, John, Chung-hye Han, & Alexei Kochetov (eds.), Proceedings of WCCFL 24, pp. 88-96. Somerville, MA: Cascadilla Proceedings Project. Choi, Jinyoung & Maribel Romero. 2008. Rescuing existential free choice items in episodic sentences. In O. Bonami & P. Cabredo Hoffner (eds.), Empirical Issues in Syntax and Semantics 7, vol. 7, 77-98. Dayal, Veneeta. 1998. Any as Inherently Modal. Linguistics & Philosophy, 21(5), 433- 476. Lewis, M. Paul, Gary F. Simons, & Charles D. Fennig (eds.). 2013. Ethnologue: Languages of the World, Seventeenth edition. Dallas, Texas: SIL International. von Fintel, Kai. 2000. Whatever. In Brendan Jackson & Tanya Matthews (eds.), SALT X, 27-40. Ithaca, NY: CLC Publications. Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Oxford University Press. Kratzer, Angelika. 1996. Severing the external argument from its verb. In J. Rooryck & L. Zaring (eds.), Phrase structure and the lexicon, 109-137. Dordrecht: Kluwer. Kratzer, Angelika & Junko Shimoyama. 2002. Indeterminate pronouns: The view from Japanese. In Y. Otsu (ed.), Proceedings of the 3rd Tokyo Conference on Psycholinguistics, 1-25.

20