Modal Existential Constructions in Spanish

1

Modal Existential Constructions in Spanish

0. Introduction

This paper examines two issues in the semantics of wh-‐constructions. On the one

hand, I look at so-‐called Modal Existential Constructions (MECs) in Spanish. These

are non-‐interrogative wh-‐constructions that occur as complements of existential

predicates and are interpreted as narrow-‐scope indefinites (Caponigro 2003)1. In

this paper, I show that (i) the universal that MECs only contain subjunctives or

infinitivals does not hold in Spanish, (ii) the infinitive/subjunctive mood alternation

in MECs can be derived from independent factors in the grammar and (iii) the

restricted syntactic distribution of MECs to object position can be derived from its

semantics and not postulated. These factors have never been explained in the

literature but have merely been noted as being characteristic of MECs.

On the other hand, the paper focuses on MECs with “who”. Spanish has two wh-‐

words for “who”, namely quien and quienes, which are Spanish for “who-‐singular”

and “who-‐plural”, respectively. The difference between quien and quienes has never

been studied in detail. Spanish is the only Romance language to have such

singular/plural distinction for “who”, but the function of the two wh-‐elements has

been taken to be just a singular/plural distinction. The data in this paper show that

these two pronouns are not merely a singular/plural alternation. The paper

contains a thorough description of the facts highlighting some puzzling semantic

restrictions and an analysis of the data ensues.

1 In Caponigro (2003) MECs are called existential Free Relatives.

2

The paper is organized as follows: §1 presents an overview of MECs and

provides a great deal of data on Spanish MECs. §2 discusses MECs with quien and

quienes and provides a preliminary analysis of the meaning of the two wh-‐elements.

In §3, I discuss quien and quienes in more detail and provide a final analysis of the

meaning of these two wh-‐pronouns. §4 looks at the mood in MECs and provides

novel data that show that Šimík’s universal that MECs only contain subjunctives or

infinitives cannot be maintained. In §5 I look at the similarites between bare plurals

and MECs in Spanish and provide a possible explanation for MECs restricted

distribution. §6 provides a semantic derivation of a MEC and I conclude the paper

with the findings and implications in §7.

1. Modal Existential Constructions (MECs)

A modal existential construction (MEC) is a construction that contains a wh-‐element

in object position with the embedded verb in either the infinitive or the subjuntive

mood.

1. No tengo [qué decir]2. Not have.1S what say.INF “I don’t have anything to say” (lit: “I don’t have what to say”)

MECs are found in most European languages (Romance, Baltic, Slavic , Albanian,

Greek, Finno-‐Ugric, and Basque), in Semitic languages (Caponigro 2003, 2004, Grosu

2004) and they have recently been found in Mixtec as well (Caponigro et al 2013) In

2 Glosses: 1: first person, 2: second person, 3: third person, PL: plural, S: singular, SUBJ: subjunctive, IND: indicative, INF: infinitive, reflx: reflexive pronoun, masc: masculine, fem: feminine, neut: neuter gender, CLT: clitic, recip: reciprocal.

3

the European family, the exception is the Germanic family, where only Yiddish and

New York English have been claimed to have MECs (Grosu 2004).

Cross-‐linguistically, there is a great deal of variation regarding which wh-‐words

can be used in MECs. Languages cluster into five different groups depending on

which wh-‐element they allow in MECs: (i) languages with no restrictions

whatsoever and all wh-‐elements can participate in MECs (e.g., Catalan, Greek,

Hungarian), (ii) languages that do not allow “why” (e.g., Hebrew, Slovenian), (iii)

languages that do not allow “how” and “ why” (e.g., Russian), (iv) languages that

disallow “when” and “why” (e.g., Portuguese), (v) languages that disallow “when”,

“how”, and “why” (French) (Šimík 2011)3. Spanish belongs to group (i), the most

liberal group, allowing all wh-‐words to ocurr in MECs. In this section, I present data

from Spanish that show MECs with all wh-‐elements available.

2. En Bahía tengo [con quién salir]. (who)

In Bahia I have with who go.out.INF “In Bahia I have somebody/people to go out with”

3. Para mi primer trabajo ya tengo [sobre (what) For my first paper already have.1S about qué escribir]. what write.INF “I already have something to write about for my first paper”.

4. Compré muchas cosas, y no tengo [dónde (where)

I bought many things and not I.have where ponerlas]. put.INF.them “I bought many things and now I don’t have anywhere to put them”

3 Italian also disallows “when”, “how” and “why” (Caponigro p.c)

4

5. Compró un auto nuevo y ahora no tiene [cómo (how) bought.3S a car new and now not have.3S how pagarlo]. pay.INF.it “He bought a new car and now he doesn’t have a way to pay for it”.

6. Yo no dije que lo iba a hacer, así (why) I not said.1S that it went.1S to do.INF so que no tengo [por qué hacerlo]. that not have.1S why do.INF.it “I didn’t say I was going to do it so there’s no reason why I should do it”

7. Quería viajar pero no tengo [cuándo (when) wanted.1S travel.INF but not have.1S when hacerlo]. do.INF.it “I wanted to start yoga but I don’t have any time to do it

Note that, although MECs and free relatives look like very similar constructions,

they are different. Unlike MECs, free relatives in Spanish cannot take “why” (8), and

the wh-‐word for “what” is que in MECs (9a) but lo que in free relatives (9b).

8. *Leo fue a Madrid [porque Mario fue.] Leo went.3S to Madrid why Mario went.3S Intended reading: “Leo went to Madrid for the same reason that Mario went”

9. a. No tengo [qué/*lo que comer]. not have.1S what/the.neut that eat.INF “I don’t have anything to eat”

b. Quiero [*que/ lo que ella tiene] want.1S what the.neut that she have.3S “I want what she has”

Although the examples in (2-‐7) all contain the same predicate, i.e., “have”, MECs

can occur under other predicates. Grosu (2004) identifies two major classes of

predicates under which MECs can be embedded. These are (i) predicates that

5

denote assertion of existence (e.g., “be” and “have” types) and (ii) predicates that

denote coming into being, view or availabity and/or its causation (e.g., “choose”,

“look for”, “find”, “send”, “obtain”, “arrive”).

10. Está buscando [con quién dejar a los chicos]

is seeking with who leave.INF to the.PL kids “He’s looking for someone to leave the children with”.

11. Finalmente encontró [de qué hablar.] finally found.3S. of what talk.INF Finally, he found something to talk about.

12. Hay [en dónde refugiarse si se there’s in where shelter.INF if reflx.3S larga a llover]. start.3S to rain.INF “There’s a place where we can shelter should it rain”.

13. Consiguieron [dónde mandar a los chicos obtained.3PL where send.INF to the.PL kids en el verano] in the summer “They managed to find a place where to send the children in the summer”

Cross-‐linguistically, it looks like all languages with MECs allow them with the

predicates in (i), whereas only a subset of these languages allow predicates in (ii) to

be used in MECs. The data in (10-‐13) show that Spanish is one of the languages that

allow MECs with a greater variety of predicates besides the predicate tener “have”.

A relevant question is whether these constructions are not merely embedded

interrogatives of the form “I don’t understand what to do”. The surface form

between MECs and embedded interrogatives is very similar. Both constructions

contain a wh-‐element and the infinitive can be used like in the example. However,

they differ in the types of wh-‐elements that can occur in them. In embedded

6

interrogatives, we find complex wh-‐phrases, where the wh-‐phrase has a NP

complement such as “which boy” (15, 17), but these are not allowed to occur in

MECs (14, 16).

14. *No encuentro [qué amigos sepan surfear]. not find.1S what friends know.3PL.SUBJ surf.INF (“I can’t find which friends know how to swim”)

15. No sé [qué amigos saben surfear]. not know.1S what friends know.3PL.IND surf.INF “I don’t know which friends can surf"

16. *No tengo [con qué personas vivir] not have.1S with what people live.INF (“I don’t have with which people to live”)

17. Me pregunto [con qué personas vive Juan] me ask.1S with what people live.3S.IND Juan “I wonder which people Juan lives with”

The predicates in (i) and (ii) are predicates that trigger the definiteness effect

(DE). The DE goes back to Milsark (1974) where he observes that only “weak” noun

phrases such as “an NP”, “some NPs”, and “two NPs” can appear as the complement

of English existential constructions, while “strong” noun phrases such as “the NP”,

“all NP”, and “most NPs” cannot. This pattern was then found to be true in many

other languages. Spanish is one of these languages so existential predicates cannot

take a definite DP (18-‐19).

18. *Tengo los amigos que conocí en Bariloche. have.1S the.PL friends that met.1S in Bariloche (I have the friends I met in Bariloche)

7

19. *En la sala hay las personas para la reunión. in the room there’s the people for the meeting (There are the people for the meeting in the room)

Szabolcsi (1986) looks at the DE in Hungarian and she groups the predicates that

trigger the DE into four groups, but all of the predicates contain an existential

component. The predicates that can take MECs as their complements are exactly the

predicates that can only take “weak” NPs in Milsark’s terms, or in other words,

indefinite DPs. So the wh-‐constructions in MECs share with indefinites the fact that

they can be embedded under the same type of existential predicates. However, the

availability of DE-‐predicates to take a MEC as their complement is language

dependent and no language has been found so far to freely take all DE-‐predicates

(Grosu 2004).

The fact that MECs are interpreted similarly to an indefinite DP can be shown

with the paraphrases that a MEC construction allows or does not allow. (20a) is an

example of a MEC in Spanish; (20b) shows that a MEC can be replaced with an

indefinite DP and the result is grammatical and the meaning is kept relatively the

same; (20c) shows that if the MEC is replaced with a definite DP (“the person”), the

sentence becomes ungrammatical.

20. a. Todos mis amigos tienen con quién hablar. All my friends have.3PL with who speak.INF “All my friends have someone to talk to”.

b. Todos mis amigos tienen a alguien con All my friends have.3PL to someone with quién hablar. who speak.INF “All my friends have someone who they can talk to”

8

c. *Todos mis amigos tienen a la persona All my friends have.3PL to the person con quien hablar. With who speak.INF (“All my friends have the person to talk to”.)

Besides having to be interpreted as indefinite, MECs must be interpreted as non-‐

specific DPs. In other words, specific-‐indefinite DPs cannot be used to paraphrase or

replace a MEC with. The indefinite DP in (20b) cannot be interpreted as one specific

person that all my friends talk to, instead, the interpretation is that each of my

friends can talk to a different person.

Another defining characteristic of MECs is the verb forms that can occur in them.

Many researchers have observed that MECs are characterized by containing a verb

form in either the infitive or the subjunctive, if a language possesses this form (or

some functionally equivalent verb form) (Grosu (2006), Pesetzky (1982), Suñer

(1984), Rappaport (1986), Rivero (1986), Rudin (1986, Grosu (1989, 1994), Grosu

and Landman (1998) and Izvorski (1998)). In his dissertation, Šimík (2011) goes

even further and proposes an implicational universal based on this observation that

rules out the possibility that a language can use another verb form different from an

infinitive or a subjunctive.

21. Šimík’s Implicational Universal 3

If a language has the infinitive mood, it uses it in its MECs. Otherwise, it uses the

subjunctive (or its functional equivalent) (Šimík 2011: 62)

In his language sample of sixteen languages, Šimík found no exceptions to this

implicational universal. Based on this, he develops a typology of MEC-‐languages.

9

Regarding the mood or verb form used in MECs, the following pattern emerges: (i)

languages that use the infinitive (e.g., Russian), (ii) languages that use the

subjunctive (e.g., Greek), and (iii) languages that use both (e.g., Hungarian).

At a first glance, this observation holds in Spanish as well (Spanish was indeed

part of Šimík’s sample). Although all examples so far contain an infinitive in the

MEC, Spanish must use the subjunctive when the subject of the MEC is different

from the subject of the main clause (this applies to the use of subjunctives in general

in Spanish, so it is not a special feature of MECs).

22. Gustavo no tiene [quién le cocine]. Gustavo not have.3S who him cook.3S.SUBJ “Gustavo has no one who can cook for him.”

23. La empresa busca [quién la financie]. the company seek.3S who it finance.3S.SUBJ “The company is looking for someone who can finance them”

24. No encuentro [en quién confíes.] not find.1S in who trust.2S.SUBJ “I can’t find who you trust”

The MECs in examples (22-‐24) contain a verb in the subjunctive. In each case the

subject of the embedded clause is different from the subject in the main clause. As I

said above, this fact is not restricted to MECs. Spanish requires the subjunctive

whenever the subject of an embedded clause is different from the subject of the

main clause (25) (provided the predicate of the main clause requires the

subjuntive). When the subject of the embedded clause is co-‐referential with the

subject of the main clause, the infinitive must be used (26).

10

25. Voy al banco para que me vendan una casa. go.1S to.the bank for that me sell.3PL.SUBJ a house “I’m going to the bank so that they can sell me a house”

26. Voy al banco para vender una casa. go.1S to.the bank to sell.INF a house “I’m going to the bank to sell a house”

However, this is not the complete picture for Spanish MECs. In this paper, I show

that Spanish does in fact allow MECs with the indicative mood, an observation that

has never been reported so far. Though before we proceed to these data, we must

first look at MECs with the wh-‐element for “who” in Spanish.

2. MECS with quien and quienes.

Spanish is the only Romance language with two wh-‐pronouns for “who”, namely

singular quien and plural quienes. As far as I know, there has never been a study

looking at the distinction between these two wh-‐elements neither in MECs nor in

other wh-‐constructions. Therefore, in this section, and for the rest of the paper, we

will focus on MECs with the wh-‐words quien and quienes.

As I said above, Spanish has two wh-‐words for “who”, a morphologically singular

form quien and a morphologically plural pronoun quienes. Examples (27-‐34) show

MECs with quien “who singular”. (27-‐29) show the pronoun quien as subject within

the MEC, in (30-‐31) quien is the object of a preposition, (31-‐32) show quien in direct

object position, and in (33-‐34) quien is the indirect object.

11

27. No encuentro [quién sepa hablar not find.1S who.sg know.3S.SUBJ speak.INF Mixteco]. Mixtec “I can’t find anyone who can speak Mixtec”

28. No hay [quién lo sepa todo] Not there’s who.sg it know.3S.SUBJ all “There is no one that knows everything”

29. Busco [con quien discutir la situación]

seek.1S with who.sg discuss.INF the situation “I’m looking for anyone who could solve the situation as quickly as possible”.

30. Necesito [en quién confiar]. need.1S in who.sg trust.INF “I need someone I can trust”

31. No tengo [a quién culpar por el error].

not have.1S to who.sg blame.INF for the mistake “I don’t have anyone to blame for the mistake”

32. Tenemos [a quién llamar en una emergencia]. have.1PL to who.sg call.INF in an emergency “We have someone we can call in an emergency”

33. Necesito [a quién confiarle mi secreto.].

need.1S to who.sg trust.INF.3S.CLT my secret “I need someone I can trust with my secret”

34. Siempre tengo [a quién pedirle always have.1S to who.sg request.INF.3S.CLT dinero]. money “I always have someone I can borrow money from”

The data (27-‐34) all contain singular quien. Although the sentences with

singular quien may suggest a singular interpretation, the use of quien does not rule

out a plural interpretation. For example, in (30), the speaker may resort to more

than one person when asking for money, if the first person cannot lend them any

12

money, the speaker may go on to another person on his “money lender” list and so

on. So when they say “I always have someone who I can borrow money from”, there

might be more than one person available at the time of the utterance in the

speaker’s mind. This means that rather than being singular, quien is unmarked for

number.

Let us look at the same sentences but with quienes to contrast their

interpretation to the sentences with quien.

35. No encuentro [quiénes sepan hablar not find.1S who.pl know.3PL.SUBJ speak.INF Mixteco]. Mixtec “I can’t find anyone who can speak Mixtec”

36. Busco [con quienes discutir la situación]

seek.1S with who.pl discuss.INF the situation “I’m looking for anyone who could solve the situation as quickly as possible”.

37. Necesito [con quiénes hablar de la reunión]. need.1S with who.pl talk.INF of the meeting “I need people I can talk to about the meeting”

38. No tengo [a quiénes culpar por el error]. not have.1S to who.pl blame.INF for the mistake “I don’t have anyone to blame for the mistake”

39. Tenemos [a quiénes llamar en una emergencia]. have.1PL to who.pl call.INF in an emergency “We have someone we can call in an emergency”

40. ?Necesito [a quiénes confiarles mi secreto]. need.1S to who.pl trust.INF.3PL.CLT my secret “I need someone I can trust with my secret”

13

41. Siempre tengo [a quiénes pedirles always have.1S to who.pl request.INF.3PL.CLT dinero]. money “I always have someone I can borrow money from”

The sentences with quienes can be uttered in a smaller number of situations than

those with quien because quienes only allows a plural interpretation. For example,

(35) can be uttered in a context in which the speaker needs more than one Mixtec

speaker. (35) would be infelicitous in a context where the speaker only needed one

Mixtec speaker. In fact, another possible paraphrase of (35) would be (42) below.

42. No encuentro [un grupo de personas que hable not find.1S a group of people that speak.3SG.SUBJ

Mixteco]. Mixtec “I can’t find a group of people who speak Mixtec”.

Any of the sentences with quienes in (35-‐41) can potentially be paraphrased

using a plural DP like “people” or “a group of people”. This suggests that Spanish has

a wh-‐pronoun quien that is unmarked for number, so it can refer to a singular

and/or a plural individual, and a plural wh-‐pronoun quienes, which can only refer to

a plural individual. This is summarized in table (1).

14

Table 1. Meaning of quien and quienes

In this section, I presented data with quien and quienes which allowed me to

come up with a preliminary generalization of the meaning of each of the two wh-‐

words, noting that quien is unmarked for number whereas quienes can only be used

with a plural interpretation.

3. More on quien versus quienes

Based on the observations in the previous section, I will now examine different

predicates with quien and quienes to further understand the difference between the

two wh-‐elements and propose a semantic definition.

If quien can refer to both plural and singular individuals, the question arises as to

what happens with predicates that force only one interpretation. For example, with

predicates that only take a plural argument such as collective predicates or

reciprocals, is quien also available?

43. Hay quién/quienes se congrega(n) en la iglesia. there’s who.sg/who.pl 3.reflx gather3S(PL) at the church “There are people who gather at the church”

44. Hubo quien/quiénes se juntó/juntaron a rezar. there.was who.sg/who.pl 3.reflx gathered3S/PL to pray “There were people who got together to pray”

quien quienes

Singular ✓

Plural ✓ ✓

15

45. Hay *quién/quienes se *ayudó/ayudaron there’s whosg/who.pl 3.recip helped.3S/PL durante el tornado. during the tornado “There are people who helped each other during the tornado”

46. Hay *quién/quiénes se *grita/gritan por nada. there’s who.sg/who.pl 3.recip yell.3S/PL for nothing “There are people who yell at each other for nothing”

The data (43-‐44) contain collective predicates, namely the Spanish forms for

“gather” congregarse and juntarse. With these two predicates, we find that both

quien and quienes are possible. In (45-‐46) we have two reciprocal predicates

ayudarse “help each other” and gritarse “yell at each other”. In this case, only quienes

is available and the use of quien results in ungrammaticality.

Note that the collective predicates in (43-‐44) are both reflexive verbs and the

predicates in (45-‐46) are reciprocals. The data then suggest that with collective

predicates, quien can be used for plural reference with reflexive verbs but not with

reciprocals. Reciprocals only allow plural quienes. Another collective predicate is

vivir juntos “live together”.

47. Hay *quien/quienes vive(n) juntos por muchos años. there’s who.sg/who.pl live3S(PL) together for many years “There are people who live together for many years”

With vivir juntos, the only possibility is quienes with quien resulting in

ungrammaticality. In this case, the predicate vivir juntos is made up of a verb and a

modifier juntos “together”. The modifier is inflected for plural number, so this might

16

be the reason why it is not compatible with quien, suggesting that quien does not

allow for plural modifiers4.

The picture that these data provide is rather complex. It is not just the case that

quien can be used to refer to either singular or plural individuals and quienes can

only refer to plural individuals. It is true, though, that quienes never allows a

singular interpretation probably because the plural morphology on it enforces

plurality. With morphologically singular quien, the story becomes less clear. On the

one hand, we saw that it can refer to a singular or a plural individual. With

predicates that require a plural subject, quien is allowed only with reflexive

predicates but not with reciprocals. But when the predicate contains a

morphologically plural marked form like vivir juntos (i.e., juntos is a plural adjective)

then quien is completely ruled out and only quienes is allowed. The picture becomes

even more complex with the following data.

48. *Hay quien se congrega en las iglesias. there’s who.sg 3S.reflx gather.3S at the churches (“There are people who gathered at churches)

49. *Hubo quien se juntó a rezar en las

there.was who.sg 3S.reflx gathered.3S to pray.INF at the casas. houses (“There were people who gathered to pray at people’s homes)

This is the same data as (43-‐44) but now I have added a plural locative modifier

that forces a distributive reading, where there are multiple events at different

4 The modifier juntos is also inflected for gender but this is not relevant for our discussion.

17

locations. That is to say, there are not only a lot of people that gathered like in (43-‐

44) but now there are a lot of people that gathered in different locations; there are

many gathering events. Surprisingly, now we see that quien is ruled out. Crucially,

the sentences become grammatical with the use of quienes.

50. Hay quienes se congregan en las iglesias.

there’s who.pl 3.reflx gather.3PL at the churches “There are people who gathered at churches

51. Hubo quienes se juntaron a rezar en las There.was who.pl 3.reflx gathered.3PL to pray.INF at the casas. houses “There were people who gathered to pray at people’s homes

Based on these observations, I propose that the meaning of quien is the set of all

atomic and plural human individuals whereas quienes denotes the set of all plural

human individuals. Under this analysis, the meaning of quienes is a subset of the

meaning of quien. This explains why in some instances both quien and quienes can

appear in the same sentence without necessarily changing the meaning of the

sentence. I formalize this notion below in definitions (1) and (2).

Definition (1)

⟦quien⟧M, w : {x: human(x) ⋀ ∀ y human(y) ⟶ y=x ⋁ ¬x=y }

Definition (2)

⟦quienes⟧M, w: {x: human(x) ⋀ ∀ y human(y) ⟶ ¬x=y }

18

Definition (1) says that quien denotes the set of human individuals and that for

every human individual in a given model M and world w, if x is human and y is

human, then x and y can refer to the same individual, giving us the singular

interpretation of quien, or y can be different from x, in which case we obtain the

plural interpretation. Definition (2) is clearly a subset of the definition in (1), where

in a given model M and world w, x is human and for every y that is human, x and y

must be different from each other (i.e., they must refer to a different individual) and

in the same way as with quien, this gives us the plural meaning of quienes.

The analysis we arrived at and definitions (1) and (2) allow us to explain most

of the data. For example, we can account for why it is ungrammatical to use quienes

when the wh-‐element refers to only one singular individual, since quienes only

refers to plural individuals. This proposal also handles cases in which only a plural

interpretation is available like in (47), where vivir juntos “live together” requires a

plural subject. In order to explain these cases, we can resort to the notion of scalar

implicatures. More specifically, we can argue that the use of quien triggers the

implicature that the predicate is compatible with both singular and plural

interpretations but the predicate vivir juntos is only compatible with a plural

interpretation, so the implicature is not met and there is a semantic clash between

quien and vivir juntos5.

5 Another possible explanation is that, although quien is semantically plural, it is morphologically singular, and the overt plural marking on juntos requires overt morphologically plural marking on the subject. This is also seen with a DP like la gente “people” which, though semantically plural, is morphologically singular and triggers singular agreement.

i. La gente triste/*tristes the people sad.sg/sd.pl “Sad people”

19

Likewise, collective predicates are grammatical with quien because quien

denotes the set of all atomic and plural individuals so the meaning of collectives is

compatible with the plural denotation of quien.

With reciprocals the explanation is not so straightforward. On the one hand, the

data in (45-‐46) suggest that reciprocals and quien are not compatible. However, the

following data suggest otherwise.

52. No hay quien se salude con un not there’s who.sg 3S.reflx greet.3S.SUBJ with a beso en Alemania. kiss in Germany “There are no people in the USA that greet each other with a kiss”

53. Hay quien se besa en la calle. there’s who.sg 3S.reflx kiss.3S.IND in the street “There are people who kiss each other in the street”

It is not clear to me whether it is reciprocals per se that are not compatible with

quien (in which case (52-‐53) should not be grammatical) or something else in the

data is making the sentences with quien sound odd or ungrammatical rendering

(45-‐46) ill-‐formed. These data deserve an analysis of their own which goes beyond

the scope of this paper, so I will leave the puzzle between reciprocals and quien for

future investigation.

This section presents a detailed description of different examples with quien and

quienes with collective, reciprocal and distributive predicates. The generalizations

stemming from the data allowed us to conclude that the meaning of quienes is a

subset of the meaning of quien and that both denote sets; quien denotes the set of all

20

atomic and plural individuals, whereas quienes denotes the set of all plural

individuals. We were not able to account for the behavior of reciprocals in the data

and this was left aside for further research.

4. Mood in Spanish MECs

In §1, we discussed the cross-‐linguistic observation that languages seem to only use

the infinitive or the subjunctive with MECs. In this section, I present data from

Spanish that show that infinitives and subjunctives are not the only mood choices

available for MECs. The data is organized based on the type of predicates that

introduce MECs. Following Grosu’s (2004) classification, predicates of type (i) are

“be” and “have” types, and predicates of type (ii) are predicates describing coming

into being, view or availability or causation of these.

A) Predicates of Type (i):

54. a. Hay [quién cree en el Diablo] (ind) there’s who believe.3S.IND in the devil “There are people who believe in the Devil” b. *Hay [quien crea en el Diablo] (subj)

there’s who believe.3S.SUBJ in the devil (“There are people who believe in the Devil”)

55. a. Hay [quien dice siempre que no.] (ind) there’s who say.3S.IND always that no “There is someone who always says no” b.* Hay [quien diga siempre que no] (subj) there’s who say.3S.SUBJ always that no (“There’s someone who always says no”)

21

56. a. Tengo [quien quiere trabajar los martes] (ind) have.1S who want.3S.IND work.INF the Tuesdays “There is someone who wants to work on Tuesdays” b. *Tengo [quien quiera trabajar los martes] (subj) have.1S who want.3S.SUBJ work.INF the Tuesdays (There is someone who wants to work on Tuesdays)

57. a. *No hay [quien cree en el Diablo] (ind) not there’s who believe.3S.IND in the devil (There are no people who believe in the Devil) b. No hay [quien crea en el Diablo] (subj) not there’s who believe.3S.SUBJ in the devil “There are no people who believe in the Devil”

58. a. *No hay [quien dice siempre que no] (ind)

not there’s who say.3S.IND always that no (“There isn’t anyone who always says no”) b. No hay [quien diga siempre que no] (subj) not there’s who say.3S.SUBJ always that no “There aren’t always people who say no”

59. a. *No tengo [quien quiere trabajar (ind) not have.1S who want.3S.IND work.INF los martes the Tuesdays (‘There is someone who wants to work on Tuesdays”) b. No tengo [quien quiera trabajar los martes] (subj) not have.1S who want.3S.SUBJ work.INF the Tuesdays (There is someone who wants to work on Tuesdays)

60. a. Siempre hay [con quién hablar]. (inf) always there’s with who speak.INF “There’s always someone to talk to”

22

b. No hay [con quién hablar] (inf) not there’s with who speak.INF “There’s no one to talk to”.

From the data in (54-‐60) the following generalization arises: with predicates of

type (i), if the MEC is affirmative, the indicative must be used, and the subjunctive

renders the sentence ill-‐formed. When the MEC is negative, the reverse holds; the

subjunctive must be used, and the indicative renders the sentence ill-‐formed. The

infinitive is possible for both affirmative and negative sentences (60).

1. Predicates of type (ii):

61. a. *No encuentro [quién dice siempre que no] (ind) not find.1S who say.3S.IND always that not (I can’t find anyone who always says no)

b. No encuentro [quién diga siempre que no] (subj) not find.1S who say.3S.SUBJ always that not “I can’t find anyone who always says no”

62. a. *Necesitamos [quién nos ayuda con esto] (ind)

need.1PL who us help.3S.IND with this (We need someone to help us with this) b. Necesitamos [quién nos ayude con esto] (subj) need.1PL who us help.3S.SUBJ with this “We need someone to help us with this”.

63. a. *Busco [quién puede actuar y cantar] (ind) seek.1S who can.3S.IND act.INF and sing.INF (“I’m looking for someone who can act and sing”)

b. Busco [quién pueda actuar y cantar]. (subj) seek.1S who can.3S.SUBJ act.INF and sing.INF “I’m looking for someone who can act and sing”

23

64. a. Conseguí [con quién ir a la boda] (inf) obtained.1S with who go.INF to the weding “I got someone to go to the wedding with" b. No conseguí [con quién ir a la boda] (inf) not obtained.1S with who go.INF to the wedding “I couldn’t find anyone to go to the wedding with”

The data in (61-‐64) suggest the following generalization: with predicates of

type (ii), the indicative can never be used and the subjunctive or the infinitive are

the only options regardless of whether the sentence is affirmative or negative.

The question that arises from these data is whether this observation can be

derived from other aspects of the grammar. As I mentioned above, the

infinitive/subjunctive alternation in MECs is not an isolated fact restricted to MECs,

but it’s present in other types of embedded constructions (see (25) and (26) above)

so it is a very general pattern of subjunctive use. I will show in this section that the

same can be said about the indicative/subjunctive alternation. We can simply derive

the possible use of the indicative in MECs from more general mechanisms in the

language that are not construction specific.

Let us first look at negation. The generalization for predicates of type (i) was

that the MEC can only take the subjunctive when it is negated, otherwise it takes the

indicative. This is a very well-‐known pattern of subjunctive versus indicative use in

Spanish. There are many predicates that must take the indicative in an affirmative

sentence (65a) and (66a) but the subjunctive when negated (64b) and (65b).

24

65. a. Creo que hoy es/*sea jueves. believe.1S that today be.3S.IND/SUBJ Thursday ‘I think it’s Thursday today” b. No creo que hoy sea/*es jueves. not believe.1S that today be.3S.SUBJ/IND Thursday “I don’t think it’s Thursday today”

66. a. Es cierto que Julio toma/*tome.

be.3S. true that Julio drink.3S.IND/SUBJ “It’s true that Julio drinks” b. No es cierto que Julio tome/*toma not be3S true that Julio drink.3S.SUBJ/IND “It’s not true that Julio drinks”

With predicates of type (ii) the issue is a bit more complex but we nevertheless

find a very clear picture of a more general pattern of subjunctive/indicative

alternations in the language. Recall the generalization about predicates of type (ii)

was that the indicative can never be used with these predicates, the only options

being the subjunctive or the infinitive. Let us begin by looking at other wh-‐

constructions and the verb forms that can occur in them.

The subjunctive in Spanish is also used in headed relative clauses with non-‐

specific indefinites. If the relativized noun is a specific indefinite then the indicative

must be used (Rivero 1975, Leonetti 1999, Perez Saldanya 1999).

67. a. Busco (*a) una persona que hable ruso. seek.1S to a person that speak3S.SUBJ Russian “I’m looking for a person that speaks Russian” b. Busco *(a) una persona que habla ruso. seek.1S to a person that speak3S.IND Russian “I’m looking for a certain person that speaks Russian”

25

A good paraphrase of (67a) is “I’m looking for a person that speaks Russian but

I don’t know them, nor do I know whether I’ll find them”. (66b) means “I know

someone and I’m looking for them, and something characteristic of them is that they

speak Russian”. In other words, (67a) is a non-‐specific DP whereas the DP in the

headed relative clause in (66b) is referential.

Another important difference between (66a) and (66b) is the presence of the

marker a in (66b). Crucially, note that (66a) lacks a. This will become important to

our analysis of MECs so it is worthwhile to look at what other types of expressions

can take the marker a and which ones do not before we draw any conclusions about

MECs in particular.

68. a. Conozco *(a) Federico. know.1S to Federico “I know Federico” b. Conozco (*a) Buenos Aires . know.1S to Buenos Aires “I know Buenos Aires” (meaning: I’ve been to Buenos Aires)

69. a. No encuentro *(a) mis amigos. not find.1S to my.PL friends “I can’t find my friends”

b. No encuentro (*a) amigos. not find.1S to friends “I can’t find any friends”

70. a. No busco *(a) nadie. not seek.1S to nobody “I’m not looking for anybody”

26

b. Conseguí *(a) un profesor de China que habla

found.1S to a teacher from China that speak.3S.IND inglés. English “I found a teacher from China that speaks English”

In (67a) the direct object is a human being and referential so a is used and, as

the asterisk shows, the lack of a results in ungrammaticality. The reverse is true for

(67b), where “Buenos Aires” is an inanimate object. In (68a), the direct object is a

human definite DP “my friends” and the marker is obligatory. (68b) contains a bare

plural object and a is ruled out. In (69a) the object is a quantifier and the marker is

also obligatory and (69b) contains an indefinite referential DP and a is grammatical.

The generalization that emerges then is that a surfaces depending on the

semantics of the object DP. Therefore, a is used with animate/human-‐like

referential expressions like definite DPs or animate/human-‐like (indefinite)

quantified DPs. Bare plurals do not take a. Hopper and Thompson (1980) note that

whether the DP is definite or indefinite, a specific and extant referent must be

available.

Another difference between a-‐marked DPs and a-‐unmarked DPs is the

subjunctive/indicative alternation they display. In general, in headed relative

clauses the verb will be in the subjunctive if the relativized DP is not marked by a

and in the indicative if it is marked by a.

71. Necesito una persona que sepa leer. need.1S a person that know.3S.SUBJ read.INF “I need a person that can read”

27

72. Conozco a una persona que sabe leer

know.1S to a person that know.3S.IND read.INF “I know someone who can read”

73. Necesito a Juan. need.1S to Juan “I need Juan”

(71) is a headed relative clause that contains the subjunctive, there is no a

preceding the relativized noun una persona “a person”. In (72) the relative clause

contains a verb in the indicative and the head noun is preceded by a (i.e., a una

persona). (73) shows that the verb necesitar “need” can indeed take a complement

with a if its complement requires it (in this case because Juan is an animate definite

DP).

These data show then that the alternation between subjunctive and indicative

forms in wh-‐clauses is closely tied to the semantics of the head noun, namely

whether the head noun is animate, (in)definite, a quantifier or a bare plural and

whether the main clause is affirmative or negative. The relationship between

specificity of the relativized noun and the mood in the embedded clause has

received a lot of attention in the literature (see e.g., Klaus von Heusinger & Georg A.

Kaiser 2003 and references therein). What has not been discussed is the interaction

of these factors and the mood in MECs specifically, which are a subset of the possible

relative clauses in Spanish. I will now show that the subjunctive/indicative

alternation found in MECs can be independently motivated from the meaning of

MECs without having to resort to construction-‐specific rules such as “the mood in a

MEC must be subjunctive or infinitive”.

28

Recall the data in (61-‐64) where it was observed that MECs with predicates of

type (ii) could not take the indicative, and only the subjunctive or the infinitive

yielded a well-‐formed sentence. However, the ungrammatical sentences in this data

set all become grammatical when the wh-‐word is preceded by a.

74. No encuentro a quién dice siempre que no. not find.1S to who say.3S.IND always that not I can’t find the person who always says no.

75. Necesitamos a quién nos ayuda con esto. need.1PL to who us help.3S.IND with this “We need the person that helps us with this”.

76. Busco a quién puede actuar y cantar. seek.1S to who can.3S.IND act and sing

“I’m looking for the person who can act and sing”

In each case in (74-‐76) the speaker has someone in particular in mind. For

example, in (74) I know this person that is always saying no, but I just cannot find

them. These data show that the verbal mood in a MEC is not dictated by whatever

property we might want to ascribe to MECs (e.g., modal force) but it is a more

general property of the grammar of Spanish related to the semantics of DPs, which

is completely independent of a specific construction.

Having shown the correlation between the referential marker a and the

presence of the indicative, I will now show that the presence of the subjunctive in

MECs is predicted from the meaning of MECs as set-‐denoting expressions in the

same way as bare plurals.

29

We briefly noted with (68b) that bare plurals cannot be marked by the

referential marker a. Bare plurals are not referential expressions; they are set-‐

denoting expressions that must undergo existential closure. If the marker a marks

referentiality, it is no surprise that bare plurals cannot be preceded by it.

77. No tengo (*a) empleados. not have.1S to employees “I don’t have empoyees”

78. Conseguimos (*a) voluntarios.

got.1PL to volunteers “We got volunteers”

79. Buscamos (*a) ingenieros.

seek.1PL to engineers “We’re looking for engineers”

(77-‐79) show that bare plurals can never be preceded by a. This is exactly the

same pattern that MECs with subjunctives display; subjunctive MECs can never be

preceded by a. As I just noted above, if a wh-‐pronoun is preceded by a, the indicative

must be used.

80. Busco (*a) quién sepa de computadoras. seek.1S to who know.3S.SUBJ of computers “I’m looking for someone who knows about computers”

81. Encontré (*a) quien me arregle el televisor. found.1S to who me fix.3S.SUBJ the television “I’ve found someone who can fix my TV”.

82. No tengo (*a) quién me planche la ropa. not have.1S to who me iron.3S.SUBJ the clothing “I don’t have anyone who can iron my clothes”

30

These data show the reversed situation from (74-‐76). Subjunctive MECs pattern

with bare plurals in not being able to be preceded by the marker a. In fact,

Caponigro (2004) proposes an analysis of MECs as set-‐denoting entities, which is

the meaning of bare plurals. Therefore, the presence of subjunctives in MECs can be

traced back to the same explanation as with bare plurals: it’s the semantics of the

complement that forces the subjunctive to appear in the embedded clause and not

the construction itself. In other words, the absence of referentiality in MECs is

expressed through the lack of the referential marker preceding them and the

appearance of the subjunctive inside them.

In this section I discussed the different moods that can appear in MECs and I

presented novel data that show that in Spanish the indicative can appear in

affirmative MECs with Grosu’s predicates of type (i). By looking at the referential

marker a as well as the semantics of bare plurals I also showed that the appearance

of the subjunctive in MECs can be derived from independent factors in the grammar

eliminating the need to postulate construction specific rules and giving an account

of what, thus far, has been taken as an epiphenomenon of the construction.

5. The Distribution of MECs and Bare Plurals

In the previous section, I established the correlation that MECs and bare plurals

trigger the same morphological marking inside and outside the wh-‐clause. Outside

the clause, we found that bare plurals and MECs are not marked by the referential

marker a and inside the clause we found that verbs appear in the subjunctive. These

two observations led us to believe that these two constructions are semantically

31

similar. In this section, I show that the parallel between bare plurals and MEcs can

be found in other parts of the grammar, an observation that will help us understand

the restricted distribution of MECs to object position.

As I mentioned in §1, MECs are restricted to object position. Plann (1980) first

observed that MECs cannot appear in subject position in Spanish.

83. [*Quien me cuide a mis hijos] vive cerca. who.sg me care.3sg.SUBJ to my children live.3S near (“The person who takes care of my children lives nearby”)

84. [*Quien me hable] me pone nervioso. who me speak.3S.SUBJ 1S.reflx put.3S nervous (“The person that would speak to me makes me nervous”)

The observation that MECs cannot appear in subject position holds universally

when they realize external arguments. Some languages allow MECs in subject

positions when the subject is an internal argument (e.g., passive voice in Russian)

(Šimík 2011)6. In Spanish, MECs can never appear in subject position.

The distribution of bare plurals in Spanish (and in Romance in general) patterns

in this respect with the distribution of MECs. Bare plurals, like MECs, cannot appear

in subject position (86); they are only licensed in object position (85).

85. Tengo mascotas. have.1S pets “I have pets”.

6 The conditions under which these are grammatical are more complex. For a detailed discussion see Šimík (2011) and Pesetsky ( 1982).

32

86. *Mascotas necesitan tiempo. pets need.1PL time (“Pets demand time”)

This observation strengthens the similarity between MECs and bare plurals and

it allows us to offer an explanation for why MECs cannot appear in subject position,

at least in Spanish. If these two constructions are semantically equivalent, meaning

that both are set-‐denoting expressions, then the prediction is that they should share

the same syntactic distribution. These data support this prediction.

It is true, however, that the distribution of bare plurals in Spanish is more

complex, so, for example, bare plurals can appear as subjects of unacusative verbs

and non-‐focused inverted subjects (Torrego 1989). But this is not a problem for our

observation because the number of predicates that can embed MECs is very limited.

There is no unacusative predicate, as far as I know, that can take a MEC in Spanish

so the fact that MECs do not appear as subjects of unacusative predicates is simply

because these are not the types of predicates that can appear with MECs.

In this section, I compared the syntactic distribution of bare plurals with the

syntactic distribution of MECs and I showed that the apparent odd behavior of MECs

regarding their unacceptability in subject position can be derived from their

semantic status as set-‐denoting expressions, which, as is the case with bare plurals,

are banned from subject position with ordinary transitive predicates.

33

6. A Semantic Derivation of a MEC

This section presents a simplified semantic derivation of a MEC. Consider (87).

87. Hay quién1 t1 ama a Carla. There’s who love.3S A Carla “There’s someone who loves Carla”.

[1]: love(x,c)]

[2]: ⟦quien⟧: ƛx[human(x) ⋀ P(x)]

[3]: ƛx [human(x) ⋀ love (x,c)]

[4]: ∃y[ƛx [human(y) ⋀ love(y,c)]](y)

[5]: ∃y[human(y) ⋀ love(y,c)]

34

[1] says that “love” is a transitive predicate, [2] is the meaning of quien, which is

a set containing all individuals that are human and have property P. [3] is the

meaning of quien ama a Carla “who loves Carla”. It states that there is an individual

who is human, and who loves Carla. [4] is the result after applying existential

quantification over [3] and [5] is the result after applying ƛ-‐abstraction to [4]. It

states that there is some specific individual such that it is human and loves Carla.

7. Conclusion

In this paper I looked at Modal Existential Constructions with quien and quienes to

investigate several phenomena related to these constructions. The findings in this

paper demonstrate that a universal along the lines of Šimík’s implicational universal

that languages will only use infinitivals or subjunctives is too strong and does not

capture the fact that Spanish allows three types of verb forms in MECs, namely

infinitival, subjunctives and indicative forms. I also showed that the alternation of

these verbal forms can be derived from independent factors of the grammar like

referentiality of the direct object and consequent presence or absence of the marker

a, the presence of negation in the matrix clause and the asymmetry between co-‐

referential and non-‐co-‐referential subjects between the matrix and the embedded

clause. Also, I demonstrated that the type of predicates in the matrix clause plays a

crucial role in determining which of these factors is/are relevant. With predicates of

type (i), I showed that negation triggers the subjunctive, whereas for predicates of

type (ii) referentiality of the object plays a crucial role in the choice between

indicative or subjunctive. Crucially, all these factors play a role in determining the

mood of the verb forms in Spanish embedded clauses in general, not just in MECs;

35

there is nothing special about MECs in this regard. Finally, I showed that the

distribution of MECs to object position can be traced back to their status as set-‐

denoting expressions. I demonstrated that this characteristic parallels the behavior

of bare plurals in Spanish (i.e., they are not marked by a), which, like MECs, are

banned from subject position.

Regarding the wh-‐pronouns quien and quienes, I first provided a detailed and

thorough examination of the possible contexts in which each of them can occur in a

MEC. Based on the description of the facts, I showed that there is more to them than

just a singular/plural morphological alternation. In particular, I proposed that

morphologically singular quien denotes the set of all atomic and plural individuals,

rendering it unmarked for number. On the other hand, morphologically plural

quienes denotes the set of all plural individuals. This analysis enables us to account

for all the data except reciprocals, for which the data did not seem very consistent so

reciprocals were left aside for further investigation.

Finally, the findings in this paper provide an independent explanation for the

seemingly inherent properties of MECs, while the descriptions and generalizations

from the data on quien and quienes provide novel data for researchers working on

Spanish and/or wh-‐constructions in general.

36

References

Caponigro Ivano, Torrence Harold, and Cisneros Carlos (2013). Free relative clauses in two Mixtec languages. "International Journal of American Linguistics" 79(1): 41-‐96.

Caponigro, Ivano (2004), The semantic contribution of wh-‐words and type shifts:

Evidence from free relatives crosslinguistically, in Robert B. Young (ed.), Proceedings from Semantics and Linguistic Theory (SALT) XIV, 38-‐55. Ithaca, NY: CLC Publications, Cornell University.

Caponigro, Ivano (2003). Free Not to Ask: On the Semantics of Free Relatives and Wh-‐

Words Cross-‐linguistically. Ph.D. dissertation. University of California, Los Angeles.

Grosu and Landman (1998). Strange relatives of the third kind. Natural Language

Semantics 6/ 2. 125-‐170.

Grosu A. (1994). Three Studies in Locality and Case. London and New York: Routledge.

Grosu, A. (1989). “Pied piping and the matching parameter”. The Linguistic Review 6/1. 41-‐58.

Izvorski, R. (1998). “Non-‐indicative wh-‐complements of existential/possessive predicates”. In P.N. Tamanji and K. Kusumoto eds. NELS 28. 159-‐173.

Hooper, Joan B. (1975). “On Assertive Predicates” . Syntax and Semantics vol 4, ed. by James Kimball 91-‐124. New York: Academic Press.

Hopper, Paul and Sandra A. Thompson (1980). Transitivity in grammar and discourse. Language. 56.251-‐99.

Klein, Flore. (1975). "Pragmatic Constraints in Distribution: The Spanish Subjunctive." Chicago Linguistics Society 11: 353-‐365.

Leonetti, Manuel (1999). El artículo. In: I. Bosque & V. Demonte (eds.). Gramática descriptiva de la lengua española. 1: Sintaxis básica de las clases de palabras. Madrid: Espasa Calpe, 787-‐883.

Milsark, Gary L. (1974). Existential sentences in English, Ph.D. thesis. MIT, Cambridge, MA.

Mejías-‐Bikandi, Errapel (1994). “Assertion and speaker’s intention: A pragmatically based account of mood in Spanish”. Hispania, 77: 892-‐902.

Plann, Susan Joan (1980). Relative clauses in Spanish without overt antecedents and related constructions. Los Angeles, CA: University of California Press.

37

Pérez Saldanya, Manuel (1999). El modo en las subordinadas relativas y adverbiales. In: I. Bosque & V. Demonte (eds.). Gramática descriptiva de la lengua española. 2: Las construcciones sintácticas fundamentales. Relationes temporales, aspectuales y modales. Madrid: Espasa Calpe, 3253-‐3322.

Pesetzky, D. (1982) Paths and categories. Ph.D. Thesis, MIT. Rappaport, G. (1986). “On a persistent problem of Russian syntax: Sentences of the

type mne negde spat”. Russian Linguistics 10. 1-‐31.

Rivero Maria Luisa (1975). Referential Properties of Spanish Noun Phrases, Language 51(1): p. 32-‐48, 1975.

Rivero, M.L. (1986). “Dialects and diachronic syntax: Free relatives in Old Spanish”.

Journal of Linguistics 22. 443-‐454.

Rudin, V. (1986). Aspects of Bulgarian Syntax: Complementizers and Wh-‐Constructions. Columbus, Oh: Slavica Publishers.

Šimík, Radek (2011). Modal existential wh-‐constructions. University Library Groningen][Host].

Szabolcsi, A. (1986). From the definiteness effect to lexical integrity. In: Abraham

and de Meij eds. Topic, Focus, and the Configurationality, Amsterdam: John Benjamins. 321-‐348.

Torrego, Esther (1989). “Unergative-‐Unaccusative Alternations in Spanish. MIT

Working Papers in Linguistics 253-‐272. Cambridge, Mass. von Heusinger, Klaus. & Georg A. Kaiser (2003). “Anymacy, Specificity and

Definiteness in Spanish” in von Heusinger, Klaus. & Georg A. Kaiser (eds), Proceedings of the Workshop “Semantic and Syntactic Aspects of Specificity in Romance Languages”, 41-‐65. Universität Konstanz: Arbeitspapier 113. Fachbereich Sprachwissenschaft.

Modal Existential Constructions in Spanish

Documents

Transcript of Modal Existential Constructions in Spanish