No title

JOURNAL OF SEMANTICS Volume 24 Number 1

CONTENTS LUISA MARTI´ Restoring Indefinites to Normalcy: An Experimental Study on the Scope of Spanish algunos

1

MATTHIAS GERNER The Exhaustion Particles in the Yi group: A Unified Approach to All, the Completive and the Superlative

27

IRA A. NOVECK, RAPHAE¨LE GUELMINGER, NICOLAS GEORGIEFF, AND NELLY LABRUYERE What Autism Can Reveal About Every . . . not Sentences

73

Corrigendum

91

Please visit the journal’s web site at www.jos.oxfordjournals.org

Journal of Semantics 24: 1–25 doi:10.1093/jos/ffl010 Advance Access publication December 18, 2006

Restoring Indefinites to Normalcy: An Experimental Study on the Scope of Spanish algunos LUISA MARTI´ CASTL/Universitetet i Tromsø

It is widely assumed that the scope of indefinites is island insensitive, i.e., that, generally, an indefinite inside of a syntactic island, such as an adjunct clause, is capable of taking scope outside of that island. This paper challenges this assumption by studying the scope behaviour of the Spanish plural indefinite algunos (roughly, ‘some (pl.)’). It presents an experimental study that shows that the scope of algunos is not free and depends on its syntactic environment, at least in the dialect of Spanish studied here. The paper discusses some of the implications of the study for current theories of indefinite scope: it points out the problems that choice functions and singleton indefinites have with the Spanish data, and it also discusses the implications for Schwarzschild’s (2002) solution to the so-called ‘Donald Duck’ problem.

1 INTRODUCTION Consider the English indefinites in (1) (italicized): (1) a. John gave an A to every student [who recited a difficult poem by Pindar] (Farkas 1981) b. [If some relative of mine dies], I will inherit a house (Reinhart 1997; example due to E. Ruys) c. Mary dates exactly half the men [who know a producer I like] (Fodor and Sag 1982) A well-known claim is that indefinite noun phrases inside syntactic islands can take scope outside of them. (1a), with the indefinite inside a relative clause, is claimed to have a reading in which there is this one difficult poem by Pindar such that John gave an A to every student who recited it; (1b), with the indefinite inside an if-clause, is taken to suggest that there is this one relative of mine such that, if s/he dies, I will inherit a house; and (1c), again a case with a relative clause, is claimed to be true in a situation in which Mary dates less than half of the men who know a producer I like, as long as she still dates exactly half of the men who know the same producer. In all of these readings, the
The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Abstract

2 Restoring indefinites to normalcy indefinite takes scope outside the bracketed phrase. I refer to such readings as wide scope readings, even though sometimes they have been argued to be cases of apparent wide scope readings, and nonscopal analyses have been proposed for them (see section 2 for more discussion on this). The examples in (1) are taken to contrast with those in (2), where instead of an indefinite, quantifiers like every new patient or most of my relatives appear in similar contexts:

(3) a.
Which patients will a doctor examine [the possibility that we give a tranquilizer]?
b. Which patients should a doctor worry [if we sedate]? (Reinhart 1997) The conclusion drawn in the literature is that the mechanism that generates indefinite wide scope is not the same as the mechanism that generates the scope of quantifiers like every new patient. What this other scope-providing mechanism is remains open for debate and there are several proposals in the literature: choice functions (see Reinhart 1997 and others), singleton indefinites (Schwarzschild 2002), referential indefinites (Fodor and Sag 1982), specific indefinites (van Geenhoven 1998, Geurts 1999), etc. None of these mechanisms is naturally amenable to a modification that makes them sensitive to syntactic islands. In fact, that they are not naturally so is taken to be advantageous, given (1)–(3). Some of these accounts are discussed in more detail in section 2 of this paper. The material in the paper comes from an experiment that was designed to test the hypothesis that the scope of indefinites is unconstrained syntactically. The experiment tested, in particular, whether the scope of Spanish algunos (roughly, ‘some (pl.)’)1 is sensitive to syntactic islands such as if-clauses, wh-clauses, relative clauses or coordinations. We will see that for a group of speakers of Spanish from the Madrid (Spain) area, the scope of the plural indefinite algunos is indeed 1

Algunas in the feminine.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(2) a. A doctor will examine [the possibility that we give every new patient a tranquilizer] b. [If most of my relatives die], I will inherit a house (Reinhart 1997) Neither every new patient nor most of my relatives can take scope outside the complex NP or the if-clause, respectively. Complex NPs and ifclauses are islands for overt syntactic movement as well:

Luisa Martı´ 3

2 THEORIES OF INDEFINITE WIDE SCOPE If the wide scope of indefinites is not constrained syntactically, it is not desirable to use syntactic mechanisms such as quantifier raising (QR) to account for their wide scope. If we did, then we would have to explain why this mechanism is sensitive to islands in cases such as (2) but not in cases such as (1), and why the operation behaves differently from other syntactic operations such as those exemplified in (3).2 In this section I consider some of the alternatives to QR/movement that have been pursued in the literature. I do not attempt to be thorough in this brief review. Instead, I concentrate on two accounts: the choice function account and the singleton indefinite account. I discuss the choice function account because it is a good example to show how syntactically insensitive theories of indefinite wide scope fail in the account of the scopal behaviour of algunos. I discuss the singleton indefinite account because it is a good example to show how linking indefinite wide scope to other grammatical phenomena is problematic. Section 2.3 points out briefly where in the logical space other theories of indefinite wide scope are located.

2.1 Choice function indefinites Reinhart (1997) and Winter (1997) propose to use choice function variables in the account of wide scope indefinites. The important point about this kind of account is that the existential closure of these variables is not sensitive to syntactic islands; they can be closed off at 2

See Reinhart (1997) for a very helpful exposition of this argument.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

syntactically constrained. I do not challenge the generalizations concerning the English examples in (1) or comment on the status of algunos in other dialects of Spanish but, given the results obtained here, I do think that better testing needs to be done in other languages and in other dialects. The organization of the paper is as follows. Section 2 discusses some of the theories currently entertained in the account of indefinite wide scope. Section 3 presents some background information on the meaning of algunos. It also contains a brief discussion of the phenomenon tested in the experiment. The experiment itself is described in detail in section 4. The results are summarized in section 4.2. What these results mean for the theories in section 2 is discussed in section 5. Section 6 is the conclusion.

4 Restoring indefinites to normalcy any point in the derivation.3 Let us consider what a choice function approach to the semantics of algunos would say.4 This analysis says that algunos denotes a choice function variable, of type e, t>, e>, which takes a (non-empty) set of plural individuals and returns one of these pluralities as output.5 The denotation of algunos nin˜os ‘algunos children’, would be as in (4): (4) ½½algunos nin˜os ¼ f(½½nin˜os) ¼ a plural individual with children as singular subparts

(5) Estela le dijo que [le matarı´a si se llevaba algunas joyas] clitic said that clitic would.kill if clitic took jewels ‘Estela said she would kill him if he took some jewels’ (6) a. df [if take (he, f( jewels)), kill(Estela, him)] b. df [CH(f ) & (take (he, f( jewels))) / kill(Estela, him)] The sentence is predicted to give rise to the meaning ‘there is a choice function f such that Estela said that if the singular subparts of the plurality of jewels that it selects are taken by him, she kills him’ (we will see that this is an incorrect prediction of this approach). For the narrow scope reading, the analysis is as in (7): (7) a. [if df take (he, f( jewels)), kill(Estela, him)] b. [df CH(f ) & take (he, f( jewels))] / kill(Estela, him) 3 Matthewson (1999) has argued instead that choice function variables are always given maximal scope. For Kratzer (1998), (Skolemized) choice function variables are free. Chierchia (2001) and Schwarz (2001) are recent discussions of the problems and advantages of these various possibilities. All of these approaches belong to the same camp in this paper, since they all use choice function variables. Geurts (1999, 2000) discusses certain problems with choice functions. 4 Gutie´rrez-Rexach (1999a, b) pursues such an analysis for algunos. 5 I simplify and assume that in this account nin˜os ‘children’ denotes a set of plural individuals. For reasons independent of the issue at hand, it is better to say that nin˜os denotes a set of singular and plural individuals. This detail is orthogonal to the issue discussed in this section.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

This choice function variable can be existentially closed at any point in the derivation. If it is closed off outside of a syntactic island, we get a reading for the indefinite in which it scopes outside of that island; if it is closed off inside of the island, a narrow scope reading is obtained instead. This approach predicts that wide scope readings should be possible for algunos outside all sorts of islands, since the process of existential closure of choice function variables is not sensitive to islands. The (rough) LF for the embedded clause of example (5), similar to some of those used in the experiment described later in section 4, is in (6a), giving rise to the (rough) truth-conditions in (6b):

Luisa Martı´ 5

This reading can be paraphrased as follows: ‘Estela said that she would kill him if he took some jewels or other’ (the existence of this reading is not under debate).

2.2 Singleton indefinites

(8)

If some relative of mine died this year, I will inherit a house

The prediction of this account is that (8) should give rise to a wide scope reading as in (9): (9) [dx x 2 C & x is a relative of mine & x died this year] / I will inherit a house C is a singleton set containing a contextually-salient individual In (9), the indefinite some relative of mine does not move out of the syntactic island. The trick is in C being restricted to contain a single individual:7 there has to exist this unique, contextually salient relative of mine who died this year for me to inherit the house (e.g. uncle Otto). Not any relative of mine makes this reading true, which suggests that this is not a narrow scope reading. In fact, it is equivalent to a wide scope reading of the indefinite. If, on the other hand, the domain of quantification happens not to be restricted to a singleton, then any relative of mine counts and we obtain a narrow scope reading of the indefinite. Importantly, Schwarzschild argues that the assumption that the domain of indefinites can be a singleton set is needed independently. It has been known at least since Heim (1982) that simple existential quantification outside of islands is not enough; this is known since Reinhart (1997) as the ‘Donald Duck’ problem. One of the things that choice functions do is ‘pull out’ the restriction of the indefinite from the island, which begins to address the problem (cf. (6b)). However, Schwarzschild argues that that is still not enough. To see why this is so, 6 7

Breheny (2003) makes a proposal that is in some ways very close to Schwarzschild’s. This would be a single plural individual in the case of a plural indefinite.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Schwarzschild (2002) argues that indefinites are special quantifiers in that their domain of quantification (i.e. their contextual restriction) can be a singleton set.6 Indefinites don’t move, and, when their domain is not a singleton, they give rise to narrow-scope readings. When their domain is a singleton, we get (the appearance of) wide scope. Consider the English sentence in (8), with the indefinite some relative of mine embedded inside of an if-clause:

6 Restoring indefinites to normalcy consider the wide scope paraphrase of (8) in (10). In this paraphrase, the overt restriction of the noun has been ‘pulled out’ from inside the island, which is what choice function variables do: (10) I have some relative such that if s/he died this year, I will inherit a house

2.3 Other theories of indefinite wide scope Mechanisms other than those described in sections 2.1 and 2.2 have been proposed to deal with wide scope indefinites. All of these other accounts share with choice function indefinites and singleton indefinites a crucial property: they are developed on the basis of the assumption that indefinite wide scope is not sensitive to syntactic islands and, hence, are not easily amenable to a state of affairs in which the wide scope of indefinites is sensitive to them. For example, Fodor and Sag (1982) claim that free indefinite wide scope is the result of these indefinites being interpreted referentially, which makes them insensitive to the presence of other scope-bearing elements in the sentence and to the syntactic environment.10 For van Geenhoven (1998) and Geurts (1999), free indefinite wide scope is the result of these indefinites being interpreted specifically, where specificity is treated together with presupposition projection.11 In this 8 See Schwarzschild (2002: §5) for notes on how to draw the distinction between the domain restriction of indefinites and that of other quantifiers. 9 See Haida (2003) for another way of showing that the ‘Donald Duck’ problem does not go away with the help of choice functions. 10 Actually, Fodor and Sag draw a distinction between wide scope readings and referential readings, but the two are close enough for our purposes here. 11 Again (cf. footnote 10), it is not completely accurate to say that these authors equate indefinite wide scope with specificity. It is quite difficult to distinguish specificity (or referentiality in the sense of Fodor and Sag 1982, for that matter) from indefinite wide scope. As before, the two kinds of interpretations are close enough for our purposes. Or, in other words, it is relevant to discuss these theories here because, if specificity (or referentiality) and indefinite wide scope are actually the same thing, then these theories provide mechanisms to deal with indefinite wide scope and hence the issue of whether they can explain the wide scope possibilities of algunos arises.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(10) is not immediately true if some individual or other exists (e.g. Donald Duck, who is not a relative of mine). This is an improvement over the analysis in which all that there is existential quantification outside of the if-clause, with the noun restriction remaining inside the island. However, (10) is true as long as I have some relative who did not die this year, which is still too weak. If the domain restriction of some is a singleton set, however, only the relative in the domain restriction of the quantifier counts (i.e. uncle Otto).8,9

Luisa Martı´ 7

kind of account, presupposition projection and hence specificity are treated as pragmatic phenomena, and so again sensitivity to the syntactic environment is lost. One important aspect of these accounts is that they put a significant burden of explanation in the pragmatics, whereas the accounts discussed in sections 2.1 and 2.2 would seem to assign the crucial job to the semantics. In all of these cases, what is crucial, again, is that the syntax is not involved. In section 5, I return to the problems that the data on algunos raises for the accounts in sections 2.1 and 2.2 only.

3.1 Background information on ‘algunos’ Algunos is a plural indefinite determiner that has the following characteristics:12 it can appear in existential sentences (11), it is semantically plural (12), it necessarily takes scope above sentential negation (13), it induces a partitive effect (14) and it can have event distribution readings (15): (11) Hay algunos libros sobre la mesa ‘There are algunos books on the table’ (12) John: ¿Viste a algunos nin˜os jugando en el patio? ‘Did you see algunos children playing in the garden?’ Mary: #Yes, I saw one/No, I saw only one (based on Laca 1996: 243) (13) A la reunio´n no vinieron algunos profesores ‘To the meeting there didn’t come algunos teachers’ (14) Algunos estudiantes vinieron ayer a verme ‘Algunos students came to see me yesterday’ (15) Algunos estudiantes vinieron ayer a verme. PLo hicieron uno a uno. ‘Algunos estudiantes came to see me yesterday. They did so one by one’ (12) is a question about a plurality of children and hence it cannot be answered in the positive if all that you saw was one child. (13) is 12

For a more thorough description, see Martı´ (under review).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

3 ALGUNOS

8 Restoring indefinites to normalcy

(16) ½½vinieron ¼ kx.ke. came(x)(e) The event argument does not generally get saturated but is existentially bound at some point in the derivation, following standard practice (for a recent discussion, see Chung and Ladusaw 2004). In order to account for event distributivity, I use a D-operator that can be inserted in the syntax anywhere it is type-wise compatible (c.f. Landman 1989, Link 1983, Roberts 1987, Schwarzschild 1996, Beck 2000, among others). The contribution of the D-operator is in (17): (17) [[D]] ¼ kf<e,t>.kx."y [(sing ind(y) & y < x) / f(y)] What the D-operator does is collect into a set all of those individuals whose singular parts are such that they apply to a predicate like vinieron. Since vinieron comes with an event variable, what it will do is collect plural individuals into a set and then it will express that for each of their singular subparts, there is an event in which each singular part came. This gives us event distribution when the events chosen for each singular subpart are different. I propose an LF like that in (18), where

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

impossible if no teachers came, something which would be allowed if the sentence had a narrow scope reading. (14) is felicitous in a scenario in which some though not all the students came to see me. In (15), there are different events of coming to see me, one for each student. Some of these properties become very clear when one compares the behaviour of algunos with that of unos, another plural indefinite in the language. For example, unos cannot replace algunos in (14) (in the partitive scenario) or (15) (though it can in (11) and (12) without any apparent change in meaning). It is hard to say what the best translation into English is for algunos, so I have left it untranslated in the glosses, though it is possible sometimes to translate it as English some. I assume that algunos is an existential generalized quantifier that operates on plural individuals and that induces a partitive implicature. In the experiment described below, the (im)possibility of event distributivity is used in some of the test items to draw conclusions about scope. Hence, it is important to be clear about how I assume event distribution readings are generated. I take verbal predicates to contain an event argument (cf. Davidson 1967, Parsons 1991, and many others). The denotation assumed here for, e.g. vinieron ‘came (pl.)’, is as in (16):

Luisa Martı´ 9

the algunos phrase has moved above D. The truth-conditions we obtain are in (19): (18) algunos students D 1

e t1 came to see me yesterday

[[algunos students came to see me yesterday]] ¼ 1 iff there is a plural student individual x and for all of its singular subparts y, there is an event e such that y came to see me yesterday in e

Even though the experiment reported below did not test for nonevent-distribution readings, I think it is fairly uncontroversial that a sentence like algunos estudiantes vinieron ayer a verme can also be true in a situation in which there is a single event of coming for all of them; i.e. they all came at the same time.13 Event distributivity is, then, a case of scope: it is obtained whenever algunos scopes above the D-operator and the point at which event variables are existentially bound.

3.2 The phenomenon: wide and narrow scope readings of ‘algunos’ In this section I go through the wide and narrow scope readings that the sentences included in the experiment could in principle give rise to. I do this because it is necessary to be very clear about what these sentences can potentially tell us about the scope possibilities of algunos. Consider again (5), repeated as (20): (20) Estela le dijo que le matarı´a si se llevaba algunas joyas ‘Estela said she would kill him if he took some jewels: There are two kinds of readings that this sentence could in principle have, depending on the scope relation between algunas joyas and si ‘if ’. The wide scope reading of algunas joyas can be paraphrased as follows: ‘Estela said that a particular group of jewels x is such that she would kill him if he took x’. In this reading, it is not enough to take just any jewels to satisfy the antecedent of the conditional; only taking certain jewels does.14 13 In this case, we have two ways of predicting truth: via the narrow scope LF, and via the wide scope LF (when the events chosen for all the coming students are the same event). 14 This is a collective wide scope reading. In principle, a second kind of wide scope reading, a distributive one, could be available. See footnote 17.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(19)

10 Restoring indefinites to normalcy On the other hand, in the narrow scope reading, which can be paraphrased as ‘Estela said that she would kill him if he took some jewels or other’, taking any jewels is enough to satisfy the antecedent of the conditional. The two readings are logically independent. Notice that the algunos NP is embedded inside of an adjunct clause. It is particularly important to consider if-clauses in the context of this paper because if-clauses have traditionally been taken to allow wide scope readings for indefinites (recall sections 1 and 2) but, at the same time, they are adjunct clauses, and hence they are strong islands for syntactic movement. Consider next the cases exemplified in (21) and (22):

(22) Dijo que Teresa San Juan y algunos hombres habı´an atracado said that and men had robbed el Deutsche Bank the ‘He said that Teresa San Juan and some men had robbed the Deutsche Bank’ (21) is a case in which an algunos phrase is embedded inside of a whisland. Wh-islands, especially those in which the wh-word is argumental, as in this case, are weak islands for the movement of arguments. Overt extraction in these cases usually results in slight unacceptability, if at all. (22) embeds algunos hombres inside of a coordination, a strong island for syntactic movement. (23) summarizes the different possibilities that the account of event distributivity assumed here (see section 3.1) affords us for these two cases (and for the first sentence in (15)): (23) a. algunos scopes above de; D-operator used if different events are chosen / different leavings/robbings (wide scope reading) if same event is chosen / same leaving/robbing (narrow scope reading) b. algunos scopes below de; no D-operator used same leaving/robbing (narrow scope reading) Assuming that time is a necessary criterion for event identification, the important thing to notice is that, even though there is an implication relation between the wide scope reading and the narrow scope reading, only the wide scope case can match the situation in which there are

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(21) Pregunto´ que cua´ndo se habı´an marchado algunos invitados asked that when clitic had left guests ‘He asked when some guests had left’

Luisa Martı´ 11

(24) a. algunos scopes above de; D-operator used if different events are chosen / different leavings/robbings if same event is chosen / DIFFERENT LEAVINGS/ROBBINGS or same leaving/robbing b. algunos scopes below de; no D-operator used DIFFERENT LEAVINGS/ROBBINGS or same leaving/robbing If time is not an event identifier, we can, within the same event, e.g. the party viewed as a big leaving event, have different leavings taking place at different times. Hence, in the case in (24a), if we choose the same event for all the students, we can still predict that a sentence like (15) is true in the situation in which the students came to see me separately. Likewise, if we were simply to choose (24b), we would be in the same situation. The problem would be that then we would not need to generate an LF according to (24a), and sentences like (15), (21) or (22) would tell us nothing about the scope capabilities of this indefinite. However, as we will see later, the subjects of the experiment were indeed assuming that time is a necessary event identifier. For a welldefined group of speakers, examples such as (22) turned out not to be true in the situation in which there were different robbings happening at different times, one per man. If time was not a crucial event identifier, they should have said that the sentence is true in that situation: they would have been able to consider that there was a big robbing of the Deutsche Bank event, taking up the whole year, with different parts of the robbing taking place at different times. And that is not what they did. Therefore, the only logical possibilities we have to worry about are those in (23). It can’t be that in a sentence like (15) we predict the different comings scenario via a narrow scope LF (or a wide scope LF in which the same event is chosen for all of the students).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

different leavings/robbings. If all we generated was an LF according to (23b), we would not be able to predict that a sentence like (15) can be true in a situation in which different students came to see me at different times. I assume time to be crucial in event identification in these cases, and emphasis on this is placed in the contexts provided in the experiment. A reviewer challenges this assumption: one could, e.g. in the case of (21), consider that the whole party the guests are attending is one extended event, with multiple participants still leaving at different times. If time were not crucial in the identification of events, with the technology for event distribution introduced above, (23) is modified as in (24). The difference between the two is marked by small capitals:

12 Restoring indefinites to normalcy The sentences in (15), (21) and (22), then, do tell us something about the scope possibilities of algunos.15 (25) exemplifies the scope interaction of algunos with the focusbearing item una sola persona ‘only one person’. Focused items have been found to hinder syntactic movement in certain cases (Beck 2006). (26) embeds algunos inside of another strong island for syntactic movement, a relative clause: (25) Vı´ a una sola persona lleva´ndose algunos cuadros saw to one only person taking.away paintings ‘I saw only one person taking some paintings away’ todos los que consiguieran all the that obtained hija daughter who obtained some clues

The potential wide and narrow scope readings are logically independent of each other in these two sentences (notice that in (26), algunos is in the restriction of todos ‘all’). In (25), the wide scope reading would have it that it is possible that there were several different people involved in the robbing of the paintings. The reading is compatible with there being paintings that were each taken by more than one person. In the narrow scope reading, there can only have been one person involved in painting taking. In (26), for the wide scope reading to be true, there has to be a particular group of clues such that all of those who obtain them get rewarded. According to the narrow scope reading, obtaining any set of clues is enough. 4 THE EXPERIMENT

4.1 Method 4.1.1 The informants Twenty-seven subjects were tested in this experiment. All of them are native speakers of Spanish and have spent 15 This suggests that the event of building Rome or the event of pruning the roses, which, as one of the reviewers points out, would normally be temporally extended events with different pieces of the event occurring at different times, is conceptualized differently from the event of leaving the party or the event of robbing the bank, which somehow cannot be temporally extended and composed of different pieces occurring at different times. This can be seen in other examples, such as He robbed the bank once (Klaus Abels, p.c.): the sentence has the implicature that there was only one time that he entered the bank, took the money and left.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(26) Dijo que recompensarı´a a said that would.reward to algunos indicios sobre su clues about his ‘He said that he would reward all about his daughter’

Luisa Martı´ 13

most of their lives in the area of Madrid and its surroundings. All of them are undergraduate students of Translation or English Philology. The test took place at the Universidad Auto´noma de Madrid (Madrid, Spain) in April 2006. Each subject was paid a total of €25. The subjects were invited to a debriefing session after the experiment was over in which the goals of the experiment were explained and their feedback was collected. The reason for limiting the experiment to speakers of the Madrid area was the wish to introduce as few variables as possible.

4.1.3 The materials There were three examples of each sentence type, and there were six sentence types. The sentence types were as follows: (27) (28) (29)

Sentence type I: algunos c-commanded by an only + numeral phrase Sentence type II: algunos c-commanded by the D-operator and the point at which event variables are existentially bound; no islands involved Sentence type III: algunos embedded inside of an if-clause

16 The questionnaire can be downloaded from the author’s webpage at http://www.hum.uit.no/ a/marti/algunos_experiment.html.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

4.1.2 The task The task carried out by the informants was a truthvalue judgement task in the form of a questionnaire.16 The questionnaire contained a total of 54 items (where item ¼ context + sentence), of which 27 were test items and 27 were fillers. The questionnaire was divided into three parts of 18 items each, which were administered on different days to avoid speaker fatigue. Short and clear instructions were provided to the subjects at the beginning of each session, and the first few items in each session were benchmark items that allowed the subject to practice (without her knowledge). In the instructions, speakers were told a short story about a city with a powerful mafia which has been involved in a number of criminal acts, and which has bribed some witnesses to not tell the truth to the police. Each item in the experiment was judged against this background and contained a description of a situation (a robbery, a murder, etc.), and then a dialog between a POLICEMAN and a WITNESS. Every item asked the informant to judge the truth of what the witness said, where what the witness said was or contained the test sentence. Algunos was used only in test sentences, never in the descriptions of the different contexts.

14 Restoring indefinites to normalcy (30) Sentence type IV: algunos embedded inside of a wh-island (31) Sentence type V: algunos embedded inside of a relative clause (32) Sentence type VI: algunos embedded inside of coordination No test items for narrow scope readings were included, since the existence of narrow scope readings for algunos is not under debate.17 (33)–(37) exemplify how some of these sentence types were tested (everything was presented to the speakers in Spanish, but here I include English for convenience):

Dialogue between policeman and witness: POLICEMAN: What did Esteban Garcı´a say to the robber? WITNESS: Le dijo que le matarı´a si se llevaba algunas joyas ‘He said he would kill him if he took some jewels’ QUESTION: Is the witness telling the truth? (YES/NO) This item tests for a wide scope reading of algunos. In this reading, it is not enough to take just any jewels (or a single jewel) to satisfy the antecedent of the conditional; only taking certain jewels does (i.e. the ones from the collection). The narrow scope reading of algunas is false in this scenario. If speakers answer this item with ‘yes’, it is then because they allow a wide scope reading of algunas. 17

Each example belonging to sentence types I, III and V was tested twice (i.e. there were two test items per example sentence in these cases), once for distributive wide scope and once for collective wide scope. This is because originally the idea was to find out also if there were differences in the availability of these readings. However, as two reviewers pointed out, the contexts that I provided in the experiment to test for wide scope distributive readings may not have made some necessary distinctions. Distributive wide scope items are ignored in the statistical analysis below.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(33) Sentence type III In the jewellery store Garcı´a Brothers they have a lot of cheap stuff and they really earn their living on this cheap stuff more than on real jewels. However, they have a collection containing a bracelet, a watch, a ring, and a pair of earrings that is altogether quite expensive, even though each of these items by themselves are not so incredibly expensive. If there were a robbery, it would be terrible if they took the collection. This morning there has been one, and Esteban Garcı´a, one of the owners, has taken out his gun and has said to the robber: ‘Take whatever you want from these jewels over here, I don’t care. But don’t take the collection. If you take it, I will kill you. Look, you can even take the earrings, but don’t take the whole collection’.

Luisa Martı´ 15

(34) Sentence type II One of the most valuable pieces in the Duchess of Teruel’s diamond necklace collection has disappeared tonight during a party. One of the guests is suspected of being responsible. The necklace was in the safe, in the study room, which is adjacent to the room where the party took place. It is important to know who left and entered the party room when. Several guests left the party room, separately, between 6 pm and 7 pm, the time during which the necklace disappeared.

(35) Sentence type VI The Deutsche Bank has been robbed several times this year. One of the mafia groups in the city is responsible. In a meeting with his associates at the end of the year, the mafia boss summarized the achievements of the year as follows: ‘dear friends, our much appreciated Teresa San Juan, accompanied in each occasion by a different new member of our group, has robbed the Deutsche Bank several times this year’. One of the boss’ projects for this year was to train all new members with Teresa, since she is such a good robber. So he made sure that each time Teresa robbed a bank, one of the new members accompanied her, a different one each time. Dialogue between policeman and witness: POLICEMAN: You say that, while you were cleaning in the adjacent room, you listened to some of the things that were said in the meeting. What did the mafia boss say? WITNESS: Que Teresa San Juan y algunos hombres habı´an atracado el Deutsche Bank ‘(He said) that Teresa San Juan and some men had robbed the Deutsche Bank’ (34) and (35) test for the wide scope reading of algunos over the Doperator and the point at which event variables are existentially bound. The contexts make sure that it is clear that there were different events of leaving the room or of robbing the bank, happening at different

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Dialogue between policeman and witness: POLICEMAN: What happened between 6 pm and 7 pm? WITNESS: Que algunos invitados salieron del salo´n. ‘Some guests left the living room’

16 Restoring indefinites to normalcy times, which unambiguously make these contexts wide scope contexts (recall discussion in section 3.2). (36) Sentence type I There has been a robbery in the art gallery Goya. The perpetrators were three women and one man. The man transported, all by himself, five small paintings, all of them at the same time. The women joined forces and transported a huge and heavy painting together, as well as a smaller yet still heavy one. Several people have witnessed the whole operation.

(37) Sentence type V The daughter of one of the richest mafia bosses in the whole city, Esteban Iturralde, was kidnapped last week. Some of the details of the kidnapping are known: where it took place, at what time, and that the girl is alive. Yesterday, Mr. Esteban met with several of his men and said: ‘There is a reward for every one of you who manages to find out certain things about my daughter. I want to know who kidnapped her, where she is, and if she is well fed. But I must know all of these things; only those who tell me everything will get the reward. Don’t come telling me that she is alive, because I already know that.’ Dialogue between policeman and witness: You say that, while you were cleaning in the adjacent room, you heard some of the things that were being said at the meeting. What did Mr. Esteban say? WITNESS: Que recompensarı´a a todos los que consiguieran algunos indicios sobre su hija. ‘(He said) that he would reward all who obtained some clues about his daughter’ POLICEMAN:

In the wide scope reading tested in (36), it is possible that there were several different people involved in the robbing of the paintings. The reading is compatible with there being paintings that were each taken by more than one person. This is the situation that is depicted in the context in (36); the women take two paintings together, the man takes

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Dialogue between policeman and witness: POLICEMAN: You say you saw the whole operation. What did you see? WITNESS: A una sola persona lleva´ndose algunos cuadros ‘(I saw) only one person taking some paintings away’

Luisa Martı´ 17

five paintings all by himself. There are, then, some paintings that were all taken by a single person (the man). The narrow scope reading of the sentence is false in this situation: in this reading, there can only have been one person involved in painting taking, but in the situation described there are three. In the wide scope reading tested in (37), there has to be a particular group of clues such that all of those who obtain them get rewarded. This reading is true in the scenario described here. The narrow scope reading is false: according to this other reading, obtaining any set of clues is enough.

4.1.5 Controlling for errors To make sure that the experiment was guarded against concluding that the experimental hypothesis is correct when in fact it is the null hypothesis that is correct (a type I error), a conservative test of the experimental hypothesis was conducted. In the truth-value judgement task carried out here, sentences were presented with a context. The context in each case is such that it makes only the wide scope reading of algunos available, in a natural and pragmatically plausible way. For those cases that are hypothesized to lack a wide scope reading, the answer in the task should be ‘no’. The test of the experimental hypothesis is conservative because it is conducted in such a way that wide scope readings, which according to the experimental hypothesis should not always be there, are given as many chances as possible. All the relevant contexts in the questionnaire are carefully constructed so as to facilitate them. Confirmation of the experimental hypothesis happens then via no responses, protecting the experiment for the subjects’ natural tendency towards yes responses. Nine different orders for the questionnaire were provided to control for order effects. They involved both shuffling test items within one session and shuffling sessions. Filler items were interspersed with the test items at regular intervals, except for the benchmark items, which were the first four items in each one of the sessions.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

4.1.4 The hypotheses The null hypothesis is that no responses are equally likely in two groups of sentence-types: if the wide scope of algunos is not constrained syntactically, then the likelihood of obtaining a no response for sentence-types I/III/V/VI should be as likely as obtaining a no response for sentence-types II/IV. The experimental hypothesis is that the likelihood is different: a no response should be more likely in I/III/V/VI than in II/IV. A two-tailed paired t-test comparing no responses for I/III/V/VI vs. II/IV was used. The hypotheses entertained in the experiment are based on no responses rather than on yes responses because of the need to control for certain interfering factors.

18 Restoring indefinites to normalcy

4.2 Results The overall group results were as follows: Table 1

Overall results

‘no’ responses in I/III/V/VI

‘no’ responses in II//IV

75.9%

30.9%

18

There is one speaker in this group that behaves according to the null hypothesis. For this speaker, a no response is equally likely for the two groups of sentence types (66.7%). 19 The general form of this hypothesis was suggested to me by Tom Roeper.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Recall that for the experimental hypothesis to be correct, a no response should be more likely in the first column of Table 1 (corresponding to I/III/V/VI) than in the second column (corresponding to II/IV). This predicted contrast was robustly significant by paired t-test: t(26) ¼ 9.52, two-tailed p < 0.0001. This means that the difference obtained between the two columns is highly unlikely to have arisen by chance. For the null hypothesis to be correct, there should have been a statistically insignificant difference between the two columns, or no difference at all. From the perspective of the experimental hypothesis, there were 15 very well behaved speakers. For these 15 speakers, Group A, the difference between I/III/V/VI and II/IV is at least 30% and the number of overall no responses is not overwhelmingly high (no more than 60% on average per speaker). For the remaining 12 speakers, Group B, the difference between the two types of syntactic contexts is not so great and/or the percentage of no responses is quite high.18 Why did the speakers from Group B behave in this way? In particular, there is a significant subgroup in Group B, comprised of nine speakers, who have an overwhelming number of no responses in both columns in Table 1 (more than 60% of no responses on average per speaker; in some cases, quite a lot more). Why is that? One possibility is that, for these nine speakers, algunos is a narrow scope indefinite. This would account for the overwhelming number of no responses, and for the fact that the number of no responses for both I/III/V/VI and II/IV is so similar. If this hypothesis is correct, the experimental hypothesis is not in trouble: these speakers are just irrelevant. Another possibility is as follows.19 In the debriefing session, one of the speakers expressed the following concern. The dialogs in the experiment are between a POLICEMAN and a WITNESS, and because of

Luisa Martı´ 19

Table 2

Results Group A

‘no’ responses in I/III/V/VI

‘no’ responses in II//IV

71.1%

8.9%

The predicted contrast for Group A was also robustly significant by paired t-test: t(14) ¼ 13.15, two-tailed p < 0.0001. Table 3

Sentence types (Group A)

‘no’ in I

‘no’ in II

‘no’ in III

‘no’ in IV

‘no’ in V

‘no’ in VI

86.7%

6.7%

84.4%

11.1%

60%

53.3%

For Group A, sentence types I, III, V and VI are islands for the scope of algunos. This is clearest for sentence types I (so´lo un N ‘only one N’) and III (if-clauses). The reason why sentence types V (relative clauses) and VI (co-ordinations) should be considered scope islands for algunos for

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

this situation, full cooperation on the part of the WITNESS is expected. In many of the descriptions of the scenarios, the number of, say, suitcases that got stolen, or the number of people that got murdered, was very precise. The WITNESS, in answering with ‘algunas suitcases’ or ‘algunas persons’ is being more vague than he could be; he could be more cooperative and give the precise number. This apparently led this speaker to answer a lot of the items with algunos with ‘no’. If this is what these speakers were doing, then the experiment suffers from a type II error: the experimental hypothesis might be right, but it is the null hypothesis that actually comes out right for these speakers. While, as explained in section 4.1.5, the experiment is guarded against a yes bias, it is not guarded against a no bias. Distinguishing these two possibilities requires a second experiment specifically designed for this purpose, something I have not undertaken. It is important to remember, however, that the results of the experiment, as shown above, are statistically significant even with Group B factored in. For the next analysis, where the results per sentence type are considered, I concentrate on Group A, since this is the group that shows the greatest sensitivity to syntactic islands and for which the contrasts are sharper. Table 2 shows the general results for Group A. Table 3 partitions the results per sentence type for this same group:

20 Restoring indefinites to normalcy these speakers is that there is more contrast in their behaviour with respect to non-islands (i.e., sentence types II and IV) than in their behaviour with respect to sentence types I and III, i.e., the sentence types that are the clearest islands. The contrast between no responses to sentence types I/III vs. sentence type V is significant by paired t-test (t(14) ¼ 2.46, two-tailed p ¼ 0.027) but much less robust than the contrast between sentence types II/IV vs. sentence type V (t(14) ¼ 4.94, two-tailed p ¼ 0.00021). The contrast between sentence types I/III vs. sentence type VI is also significant (t(14) ¼ 2.84, two-tailed p ¼ 0.012) but much less robust than the contrast between sentence types II/IV vs. sentence type VI (t(14) ¼ 4.57, two-tailed p ¼ 0.00043).20,21

The two empirical generalizations that arise in section 4 are as follows: (38) whether one looks at the overall group results or at the results per sentence type, the generalization is that algunos cannot freely take wide scope, contrary to what the received wisdom on indefinites suggests. The wide scope of algunos is sensitive to the syntactic environment (39) there is a group of speakers, Group A, for whom if-clauses, relative clauses, co-ordinations and so´lo un N ‘only one N’ phrases hinder the wide scope of algunos I concentrate on (38) in this section, and return to (39) in the conclusion. In this section I discuss the problems that (38) raises for accounts of indefinite wide scope based on choice function variables or singleton indefinites. 20 Why it is that sentence types V and VI have less incidence of no responses than sentence types I and III is an issue that I leave for further research. It is interesting to note in this connection the difference between sentence types III (if-clauses) and V (relative clauses headed by a universal quantifier): their semantics is very similar, so it must be their syntax that is at stake here. 21 Martı´ (2005) can be viewed as the preliminary work leading to the current paper. The results reported there differ from the ones here in some ways, particularly in the fact that if-clauses (sentence type III) were reported not to be islands for the scope of algunos in the earlier work (see also Gutie´rrez-Rexach 1999a, b). The empirical work done for the current paper is much better grounded than that early work and hence the results here should be taken more seriously. Because of the preliminary results in Martı´ (2005), the experimental hypothesis initially formulated for the experiment reported here predicted a contrast between sentence types I/V/VI vs. I/III/IV. The contrast between these two groups is in fact statistically significant (though less than the contrast between I/III/V/VI vs. II/IV). I think this is because of the very few no responses in sentence types II and IV; i.e. adding sentence type III to this very low no incidence group does not make the contrast insignificant.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

5 DISCUSSION

Luisa Martı´ 21

22

As pointed out earlier, Breheny (2003) makes a proposal similar to Schwarzschild’s in some respects, particularly in their common use of the domain restriction of indefinites. The same criticism that applies to Schwarzschild applies to him.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The problem that (38) poses for choice functions is one of overgeneration, since nothing prevents wide scope readings from being generated with if-clauses, relative clauses, co-ordinations or above so´lo un N ‘only one N’. This is because one of the important features of choice functions is that their existential closure can happen anywhere in the syntactic tree. One could, of course, build syntactic sensitivity into the mechanism that closes off choice function variables, but notice that the reason why theories of indefinite scope turned to choice function variables was precisely that there doesn’t seem to be anything that connects their existential closure with the syntactic environment. Of course, as is often argued in the literature, a choice function approach is not incompatible with the idea that indefinites are ambiguous between a choice function meaning and a quantificational meaning. With a quantificational meaning for algunos it might be possible to make the movement of the quantifier sensitive to islands, but notice that wide scope readings would still be generated across the board, thanks to the choice function option. As for singleton indefinites, recall that one crucial aspect of this proposal is that wide scope readings of indefinites are linked to a particular choice of domain restriction. However, whether the domain restriction of the indefinite is a singleton set or not has nothing to do with its syntactic environment. This is the reason why algunos cannot be a singleton indefinite. In order to predict wide scope readings out of the islands that allow them and at the same time not predict them in the cases that don’t, the mechanism that gives rise to wide scope readings, i.e., restriction of the domain of quantification down to a singleton, would have to be sensitive to islands. But there is no independently justified reason to think that this mechanism would be sensitive to something like that. A mixed account, just as in the case of choice functions, over-generates as well.22 The facts about algunos force us, interestingly, to reject Schwarzschild’s nice attempt at dealing with the ‘Donald Duck’ problem. The reason is that, despite the fact that something like singleton indefinites seems to be necessary to prevent truth-conditions that are too weak (recall (10)), the domain restriction of indefinites being a singleton set is not something that can naturally be made to depend on the indefinite’s syntactic environment, as suggested above. That is, linking indefinite wide scope to a solution to the ‘Donald Duck’ problem, as

22 Restoring indefinites to normalcy Schwarzschild does, seems, given the Spanish data, ill-advised: once it is established that algunos cannot freely take wide scope, Schwarzschild’s strategy to deal with the ‘Donald Duck’ problem is no longer available to us. Yet, the ‘Donald Duck’ problem does arise with algunos, so it must be that its solution lies elsewhere.23 Just as Schwarzschild does with other quantifiers, something must be built into the semantics or pragmatics of algunos that prevents its domain from being a singleton set.24,25 6 CONCLUSION

23

This conclusion is also drawn in Martı´ (2005). The recent proposal in Schlenker (2006) seems likewise problematic, since it is an attempt at relating indefinite wide scope to so-called branching readings of indefinites. 25 Both choice functions and singleton indefinites may have problems generating wide scope distributive readings, depending on what else one assumes. Since the experiment described in section 4 did not properly test for wide scope distributive readings, I do not draw conclusions about them here. 26 Let me also note that it is immaterial to the main point of the paper whether, in the end, what I have been calling wide scope readings are genuinely wide scope or not (recall section 2.3). What matters is that the interpretation possibilities of algunos are constrained by syntactic islands, no matter how those interpretations are generated. 27 Consider that, initially, algunos was also thought to be able to freely take wide scope (see for example Gutie´rrez-Rexach 1999a, b). 24

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Generalization (39) would seem to suggest that indefinites really behave like any other item that moves, since if-clauses (generally, adjunct clauses), relative clauses, focused items and co-ordinations generally hinder movement. Thus, generalization (39) indicates that algunos does not need to be considered in a special scope category, separate from items that move. The scope properties of algunos are just like the scope properties of these other items.26 What does it mean to find out that the scope of algunos is syntactically restricted? It means that a syntactically sensitive mechanism for algunos to achieve wide scope is necessary. Do we then add this mechanism to the list of scope achieving mechanisms for indefinites? Perhaps English uses choice functions and Spanish does not. Or perhaps there are different kinds of indefinites, requiring different scope mechanisms, even within the same language. Both kinds of conclusions seem problematic if our goal is to arrive at an explanatorily adequate theory of indefinite scope. Before pursuing either, then, I suggest that we review and test experimentally our empirical generalizations about the scope of indefinites in other languages, and other indefinites in Spanish,27 perhaps following methodology similar to the one used here

Luisa Martı´ 23

and in other (experimental) work on scope such as that in Pafel (2006) or Villalta (2003). Acknowledgements

LUISA MARTI´ Center for Advanced Study in Theoretical Linguistics (CASTL) Faculty of Humanities Universitetet i Tromsø NO-9037 Tromsø, Norway e-mail: [email protected]

REFERENCES Beck, S. (2000) ‘Star operators episode 1: Defense of the double star’. University of Massachusetts Occasional Papers in Linguistics 23:1–23. Beck, S. (2006) ‘Intervention effects follow from focus interpretation’. Natural Language Semantics 14: 1–56. Breheny, R. (2003) ‘Exceptional-scope indefinites and domain restriction’. In

M. Weisgerber (ed.). Proceedings of Sinn und Bedeutung VII. Chierchia, G. (2001) ‘A puzzle about indefinites’. In C. Cecchetto, G. Chierchia & M. Teresa Guasti (eds). Semantic Interfaces: Reference, Anaphora, and Aspect, CSLI, Stanford. Chung, S. and Ladusaw, W. (2004) Restriction and Saturation. MIT Press. Cambridge, MA.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

It is no exaggeration to say that it would not have been possible to write this paper without the help of a good number of people. Klaus Abels, Tom Roeper, Floyd Rudmin and specially William Snyder generously dedicated their time to me to discuss the design of the experiment. William patiently helped me with the statistical analysis. Isabel Pe´rez Jime´nez went through many of the sentences with me and provided her judgements. Gema Chocano and Amaya Mendikoetxea (both at the Universidad Auto´noma de Madrid) helped with the logistics of the experiment, no small feat. Luis Alonso Ovalle, Andreas Haida, Yael Sharvit and Michal Starke provided comments at an early stage (so early that they probably don’t recognize the paper as it is now). Yael Sharvit was also a very understanding and patient editor. Audiences at the 15th Amsterdam Colloquium and at Sinn und Bedeutung X provided valuable feedback in the initial stages. The anonymous reviewers of the paper made me realize that a more empirically sound way of collecting the data was necessary, which then prompted me to run the experiment described here; also, their input was very helpful in organizing the contents of the paper more effectively. CASTL generously funded the whole enterprise. And, of course, the participants in the experiment patiently filled out the questionnaire and obediently did everything I told them to do without having a clue why. Many thanks to everybody! Only I am responsible for whatever mistakes the paper contains.

24 Restoring indefinites to normalcy determinacio´n. La ausencia de determinante en la lengua espan˜ola. Visor. Madrid. 241–268. Landman, F. (1989) ‘Groups I’. Linguistics and Philosophy 12:559–605. Link, G. (1983) ‘The logical analysis of plurals and mass terms: A latticetheoretical approach’. In R. Bauerle et al. (eds). Meaning, Use and Interpretation of Language. de Gruyter. Berlin. 303–323. Martı´, L. (2005) ‘Donald Duck is back, and he speaks Spanish’. In P. Dekker & M. Franke (eds.). Proceedings of the 15th Amsterdam Colloquium. Institute for Logic, Language and Computation, Universiteit van Amsterdam. 143–148. Martı´, L. (under review) ‘The semantics of plural indefinites in Spanish and Portuguese’. Ms. Universitetet i Tromsø (http://ling.auf.net/lingBuzz/ 000269). Matthewson, L. (1999) ‘On the interpretation of wide-scope indefinites’. Natural Language Semantics 7:79–134. Pafel, J. (2006) Quantifier Scope in German. John Benjamins. Amsterdam/ Philadelphia. Parsons, T. (1991) Events in the semantics of English. The MIT Press. Cambridge, MA. Reinhart, T. (1997) ‘Quantifier scope: How labor is divided between QR and choice functions’. Linguistics and Philosophy 20:335–397. Schlenker, P. (2006) ‘Scopal independence: a note on branching readings and wide scope readings of indefinites and disjunctions’. Journal of Semantics 23:281–314. Schwarz, B. (2001) ‘Two kinds of longdistance indefinites’. Proceedings of the 13th Amsterdam Colloquium, Institute for Logic, Language and Computation, Universiteit van Amsterdam. 192–197.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Davidson, D. (1967) ‘The logical form of action sentences’. In N. Rescher (ed.), The Logic of Decision and Action. University of Pittsburgh Press. Pittsburgh, PA. Fodor, J. & Sag, I. (1982) ‘Referential and quantificational indefinites’. Linguistics and Philosophy 5:355–400. Geurts, B. (1999) ‘Specifics’, in B. Geurts, M. Krifka & R. van der Sandt (eds). Focus and Presupposition in Multi-speaker Discourse. Workshop Reader for ESSLLI 99. Utrecht. 99–129. Geurts, B. (2000) ‘Indefinites and choice functions’. Linguistic Inquiry 31: 731–738. Gutie´rrez-Rexach, J. (1999a) ‘Group indefinites’. Proceedings of WCCFL 17:250–264. Gutie´rrez-Rexach, J. (1999b) ‘Spanish indefinites and type-driven interpretation’, in J.-M. Authier, B. Bullock & L. Reed (eds). Formal Perspective on Romance Linguistics. John Benjamins. Amsterdam/Philadelphia. 151–166. Gutie´rrez-Rexach, J. (2001) ‘The semantics of Spanish plural existential determiners and the dynamics of judgment types’. Probus 13:113–154. Haida, A. (2003) ‘An argument against the choice-function interpretation of Wh-in-situ’. Ms. ZAS. Berlin. Heim, I. (1982) The Semantics of Definite and Indefinite Noun Phrases. Ph.D. dissertation. University of Massachusetts at Amherst. Kratzer, A. (1998) ‘Scope or pseudoscope? Are there wide scope indefinites?’ In S. Rothstein (ed.). Events and Grammar. Kluwer Academic Publishers. Dordrecht. 163–196. Laca, B. (1996) ‘Acerca de la sema´ntica de los plurales escuetos del espan˜ol’. In I. Bosque (ed.). El sustantivo sin

Luisa Martı´ 25 Schwarzschild, R. (1996) Pluralities. Kluwer Academic Publishers. Dordrecht. Schwarzschild, R. (2002) ‘Singleton indefinites’. Journal of Semantics 19: 289–314. van Geenhoven, V. (1998) Semantic Incorporation and Indefinite Descriptions. CSLI Publications. Stanford. Villalta, E. (2003) ‘The role of context in the resolution of quantifier scope

ambiguities’. Journal of Semantics 20: 115–162. Winter, Y. (1997) ‘Choice functions and the scopal semantics of indefinites’. Linguistics and Philosophy 20: 399–467. First version received: 24.08.2005 Second version received: 08.09.2006 Accepted: 02.11.2006

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Journal of Semantics 24: 27–72 doi:10.1093/jos/ffl007 Advance Access publication December 20, 2006

The Exhaustion Particles in the Yi group: A Unified Approach to All, the Completive and the Superlative MATTHIAS GERNER City University of Hong Kong

The exhaustion particles of the Yi languages (Tibeto-Burman languages from Southwest China) are sentence-end morphemes with a surprising wealth of possible interpretations. With gradeable states they convey the meaning of superlative (‘most’), with accomplishments they function as completive particle (‘exhaustively’), and in ungradeable states, activities or achievements they act as all particles, i.e. as universal non-distributive quantifiers, on the first argument. A unified account of the all-, completive- and superlative-meanings is proposed. It is argued that all three notions basically divide their respective domain (¼ objects, events or states) into three types: a singular domain type, a quantized domain type and a homogeneous domain type. For events there is also a fourth domain type, the bounded domain type, which does not have an analogy with objects and states. Object-denotations Event-denotations Singular Quantized

Individual OD Quantized OD

State-denotations

Punctual ED Quantized ED

Ungradeable SD State with quantized comparison class Bounded — Bounded ED — Homogeneous Homogeneous OD Homogeneous ED State with homogeneous comparison class The all-, the completive- and the superlative-meanings of the exhaustion particles have in common that they are incompatible with entities from the singular domain type, (in general) pragmatically implausible with entities from the homogeneous domain type, and compatible with entities from the quantized domain type.

1 INTRODUCTION Numerous authors have pointed to parallels between nominal and verbal semantic structure (e.g. Link 1983; Bach 1986; Krifka 1989,
The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Abstract

28 The Exhaustion Particles in the Yi group

1

The term ‘exhaustion particle’ is inspired from Bjo¨rverud (1998: 82), although she uses this appellation for a different type of particle (which I characterize as sent-versatiles, see Gerner 2002: 88).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

1992), notably to the structural similarities between mass nouns (‘water’) and activity verbs (‘run’) on the one hand and between count expressions (‘three apples’) and accomplishment expressions (‘eat three apples’) on the other hand. The former similarity was re-analysed as the two reference properties cumulativity (CUM: the union of two entities belonging to the same denotation is again of this denotation) and divisibility (DIV: any part of an entity belonging to a denotation is again of this denotation). The latter similarity was interpreted as the quantized reference property (QUA: no proper part of an entity belonging to a denotation is again of this denotation). The term ‘entity’ refers here either to ‘object’ or ‘event’. While scholars have brought to the fore the link between mass terms and activities, respectively between count expressions and accomplishment expressions, correspondences between nouns and the two remaining Vendlerian classes, i.e. achievements and states, have basically gone unexplored. The Yi languages (Tibeto-Burman languages from Southwest China) have fascinating empirical data to offer through a sentence-end morpheme that I call exhaustion particle.1 The exhaustion particles are cross-categorial modifiers acting on noun phrases (as all ¼ non-distributive universal quantifier), on dynamic verb phrases (as a completive particle), and on gradeable states (as a superlative particle). The exhaustion particles show similarity in their treatment of different object-types, event-types and state-types. They reject entities of Group I (¼ individuals, punctual events and ungradeable states). Furthermore, they are pragmatically implausible with entities of Group III (¼ homogeneous object-denotations, homogeneous event-denotations and gradeable states with an implicit homogeneous comparison class). Finally, they are configurable with and act upon entities of Group II (¼ quantized object-denotations, quantized event-denotations and states with an implicit quantized comparison class). This paper is organized as follows. In section 2, I define a semantic representation language that allows treating objects, events and states on an equal footing in a semi-lattice framework (processes are classified here as events). I discuss and partly propose reformulations of four reference types that have been advanced in the literature: the singular, the quantized, the bounded and the homogeneous reference type. In section 3, I present the exhaustion particle data of Liangshan Nuosu

Matthias Gerner 29

(one of the Yi languages—the other Yi languages involve similar particles whose sound structure is briefly introduced in the Appendix) and bring to the fore its interaction with the four reference types mentioned above. In section 4, I portray a semantic account of the input and output structures of the exhaustion particles. 2 THE SINGULAR, QUANTIZED, BOUNDED AND HOMOGENEOUS DOMAIN TYPES

2.1 Formal framework

A.

Constant symbols a. For the type <e, t> we pick out five non-logical constants (or one-place predicates of expressions of type e) and will require that the first three do not overlap:


O (for objects)
E (for events)
S (for states)
T (for times)
Tat (for time atoms or time points); b. Let 8 and 7 be two-place relations (part and proper part), technically picked out as constant symbols of type <e, <e, t; c. Let < be a two-place relation, technically picked out as constant symbol belonging to type <e, <e, t; d. Let s be a function symbol of type <e, e>; e. Let k be a two-place operation (join), technically picked out as a function symbol belonging to type <e, <e, e;

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Before starting to discuss various object, event and state types, a brief sketch of the semantic representation language is in order. We will involve an extensional type-theoretic language LEXH type (EXH for exhaustion particle) at the intermediate level between the natural language (English or the Yi languages) and the denotational level. My approach here extends those of Link (1983), Krifka (1989, 1992) and others (e.g. Bach 1986) to cover states in addition to objects and events. I proceed in two steps. In this section I outline general logical notions pertaining to all three entities (objects, events and states) and mention notions relating to objects and events only. In section 2.3 I develop the notions concerning the interpretation of states. Now, the syntax of LEXH type is given as follows:

30 The Exhaustion Particles in the Yi group B.

Properties

The above characterizations do not state anything about S, the set of states, on which I have not defined a semi-lattice structure. I will discuss the interpretation of states in section 2.3.

2.2 Object and event structure In this section I discuss and describe different reference types for objects and events by making use of the resources developed in the previous section. These reference types, i.e. singular, quantized and homogeneous, have partly been defined in the literature. I want to review them here and present a couple of amendments. Furthermore a fourth reference type is defined for events only (i.e. the bounded reference property). The notions introduced in the sections below are not mutually exclusive. The singular reference type, for example, is included in the quantized (cf. section 2.2.2) and bounded (cf. section 2.2.3) reference types. As we will see in section 3, the exhaustion particle has scope over NPs in sentence initial position and over V-bar constituents. We will therefore take the view that predicates of the semantic representation language LEXH type may encapsulate whole NPs (including definite expressions, pronouns and proper names) in contrast with the more conventional assumption that they merely symbolize N-bar constituents. For the purpose of avoiding cumbersome expressions and provided that no misinterpretation occurs, I will sometimes mix up levels of representation by saying ‘quantized object’ instead of ‘quantized object-denotation’ or ‘homogeneous event’ instead of ‘homogeneous event-denotation.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

a. In LEXH type should appear formulas that express that O, E, T, and Tat do not overlap, but I will skip these formulas here; b. Formulas that express that k is restricted to O, E and T, and induces with 8 a structure of semi-lattice (see Krifka 1992: 32 for details); c. Formulas that articulate that Tat is the set of atoms of T and < is a linear order on Tat (again see Krifka 1992: 33 for details); d. Formulas that state that s is a homomorphism between the extension of E to that of T which maps every event onto its running time (cf. Krifka 1992: 33);

Matthias Gerner 31

2.2.1 Singular reference type for objects and events As for objects, I would like to arrange in this class various nominal expressions that translate into English as singular NPs such as:






singular singular singular singular singular

count expressions (e.g. ‘one potato’) proper names (e.g. ‘John’, ‘Mary’) pronouns (‘I’, ‘you’, ‘he/she/it’) definite expressions (e.g. ‘this bed’, ‘the pen’) possessive expressions (e.g. ‘John’s nose’).

Achievement Accomplishment Activityact State

in

for

at

be+Verb+ing

+ +  

  + +

+   +

 + + 

Figure 1

Vendler’s four classes.

Scholars have criticized Vendler for separating achievements from accomplishments. First, a number of authors hinted at observational discrepancies in Vendler’s account of the progressive in achievements (e.g. Leech 1971: 1–27; Comrie 1976: 43): (1) He is dying slowly. (2) He is reaching the station. (3) He is winning the match.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Their common feature is that they relate to objects that are atoms of the semi-lattice (O, 8) in a way that will be specified below. We will not include nominal expressions in this group that are built on a head mass noun (e.g. ‘an amount of water’). Regarding events, singular event denotations will refer to so-called punctual events, also termed achievements, such as English ‘sneeze (once)’, ‘type the letter p’, ‘hit the ground’, etc. Vendler (1967) defined achievements (in English) against accomplishments, activities and states as being compatible with in-adverbials (e.g. ‘in two hours’) and atadverbials (e.g. ‘at two o’clock’) and being incompatible with foradverbials (e.g. ‘for two hours’) and with the progressive (be+V+ing). See figure 1 below.

32 The Exhaustion Particles in the Yi group These examples illustrate the compatibility of the progressive with achievements (against Vendler’s assumption). Second, scholars have questioned the genuine punctual character of achievements and thus the relevance of the modifiability of achievements by at-adverbials (Verkuyl 1993: 46-50): (4) He typed the letter p at noon sharp. (5) He typed a business letter at noon sharp.

(6) (7) (8) (9) (10) (11)

S/he S/he S/he S/he S/he S/he

typed the letter p in an instant/in one hour. typed a business letter  in one instant/in one hour. reached the top in an instant/in one hour. touched the dog in an instant/in an hour. ate three sandwiches in one instant/in one hour. won the match in one instant/in one hour.

(6), (8), (9) and (11) match Vendler’s test of an achievement. However, (6) and (9) are compatible with punctual in-adverbials only,

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The main point of these critics (mainly Verkuyl) is that in the era of modern technology the limit between punctual and nonpunctual events is blurred. The typing of p can be delayed through a Please Wait command, while a business letter can be loaded from a template through a single keystroke. In my opinion, it is correct to question the genuine punctual nature of some events mentioned in this context such as ‘die’ or ‘win a match’, but example (4) seems to be of a different kind, as its punctual nature is intuitively plausible. The difference between ‘type the letter p’ and ‘win a match’ is that the former does not seem to make reference to a natural preparation phase preceding the typing of the letter p, while the notion of preparation is much more inherent in the latter. The issue underlying example (4) appears to be strictly parallel to the question of whether masses have minimal parts. The compromise reached among scholars is an explanation of Bunt (1979: 255; 1985: 45) stating that human language designs a way of speaking about things as if they do not consist of discrete parts (see section 2.2.4 for more details). The feature [6 punctual] (or as Vendler calls it: [6 momentary]) has its relevance for a classification of events. However, there may be two types of events that are merged in Vendler’s class of achievements and that should be dissociated. Consider the following examples:

Matthias Gerner 33

SNGðPÞ4dx ½PðxÞ ^ "y ½PðyÞ/x ¼ y: This reference type characterizes definite singular expressions such as ‘John’, ‘this bed’, but appears as too strong for indefinite count expressions such as ‘one tomato’, as it requires equality of entities falling under P. This concept of singularity would reject many nominal expressions that we identified on our radar screen at the beginning of this section. Singularity as sketched above in the entrance paragraph of 2.2.1, refers to the idea of countability. Countable items are defined in the literature as those that are composed of atoms (Krifka 1989: 69). For a nominal/verbal predicate P and an object/event variable z, the property of z being an atom of P is portrayed in the following way (Krifka 1989: 45; 1992: 32): ATOMðz; PÞ4PðzÞ ^ :dx ½x7z ^ PðxÞ: Building on this notion of P-atoms, we may devise an alternative version of singularity (written as SGL). The property of singularity for nominal predicates will be defined as being a global atom of (O, 8). SGLðPÞ4dx PðxÞ ^ "y ½PðyÞ/ATOMðy; OÞ:

2 In fact, the situation is even more complicated: expressions like ‘walk to the station’ must be further differentiated from ‘win a match’ (see also section 2.2.3). The latter is ambiguous with regard to the inclusion of a preparatory phase, whereas the former seems to always comprise a prepatory phase. ‘Win a match’ can be interpreted as just one point or, alternatively, as a point plus a preparation phase leading to it. The addition of adverbials such as ‘in an hour’/‘in an instance’, known in the literature as items of aspectual type coercion (e.g. Moens and Steedman 1988), imposes one of the readings ‘point plus preparation phase’ or ‘point’. ‘Walk to the station’, on the other hand, is not ambiguous in this way. It is always interpreted as including a prepatory phase (‘walk to the station in 30 minutes’/‘walk to the station in an instant’).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

while (8) and (11) are compatible with both punctual and non-punctual in-adverbials. Events such as (8) and (11) are classified among what I call in section 2.2.3 ‘bounded events’ to which also denotations like ‘walk to the station’ belong. These events are characterized by the following property: if an event e belongs to a denotation, then also each final stage e# will belong to this denotation.2 Back to singular events: singular events comprise punctual events, i.e. events that are modifiable by some expressions resembling English ‘in an instant’ and ‘at 3 o’clock sharp’ but that are incompatible with non-punctual in-adverbials (‘in an hour’). The question now is how to represent formally nominal and verbal expressions that are singular. Krifka (1992: 32) portrays the singular property of predicates as follows:

34 The Exhaustion Particles in the Yi group This formula expresses first the existence of a matching entity and then that every matching entity must be a global atom of the semilattice of objects. Being a global atom of (O, 8) appears to be equivalent to the requirement of being countable with the count value ‘one’ (Krifka 1989: 69). I suggested above that the conventional notion of punctual event should be the target concept to be captured by singular events. Intuitively, punctual events have a running time reduced to a time point, i.e. to a member of Tat. The parallelism with the singular reference property of objects is thus obvious. Singularity for verbal predicates may be defined via the mapping homomorphism s:

As an illustration, consider again the predicate P: ‘type the letter p’. Besides the existence of a ‘type p’-event, each ‘type p’-event, is reduced to a time point as expressed and required by the property SGL(‘type p’). 2.2.2 Quantized object and event denotations The quantized reference property was first introduced by Krifka. It basically states that any object or event of a certain denotation (either nominal or verbal) does not admit any proper part that matches again the same denotation. The intuition behind this reference type, in the case of nominal denotations, is that an object such as ‘five people’ or ‘four litres of water’ does not admit any proper part that again matches the same denotation ‘five people’ or ‘four litres of water’. In the case of verbal denotations, an event such as ‘eat three sandwiches’ does not allow any proper subevent to belong again to the denotation of ‘eat three sandwiches’. This reference type is represented by the following formula (see Krifka, 1989: 41; 1992: 32): QUAðPÞ4"x; y ½PðxÞ ^ PðyÞ/:y7x: Intuitively speaking, quantized objects are exactly those objects that are presented with whole and complete shape-structure boundaries. The nominal expressions referring to quantized objects include:
count expressions (e.g. ‘three apples’, ‘five cubic metres of water’, ‘2 kg of tomatoes’)
proper names (e.g. ‘John’, ‘Peter and Mary’)
pronouns (e.g. ‘I’, ‘you’, ‘they’)

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

SGLðPÞ4dx PðxÞ ^ "y ½PðyÞ/ATOMðsðyÞ; TÞ:

Matthias Gerner 35


definite expressions (e.g. ‘the women’, ‘this/that blueberry’, ‘these/ those houses’)
possessive expressions (e.g. ‘John’s books’, ‘his tea’)

(12) Yesterday I met a group of 20 tourists. Today I have seen them again. Sentence (12) could be easily said by someone in an unfocussed context, even when the actual number of tourists the speaker saw again was 18 or 19. These quantity values would be excluded, when something like ‘Today I have seen them all again’ would be said. Note that such an ambiguity only holds for relatively high quantities. If 20 tourists were replaced by say four, then ‘them’ could not be used, when the speaker only met three of them. In general, deictic definite expressions are quantized (at least when interpreted as specific), since they take a referent or antecedent as granted and provide proportional, i.e. holistic (e.g. Eschenbach 1993: 22), quantity information. Let x and y be two objects that are predicated by ‘them’ or ‘the/these/those books’. In the normal reading, it follows that x and y are identical and in particular that y is no proper part of x. On the other hand, vague or approximative expressions (e.g. ‘many books’, ‘some coffee’, ‘several pupils’) are not quantized. If, for example, in some pragmatic setting 20,000 spectators of a football match accounted for ‘many spectators’, then 19,990 spectators would do the same and thus make it the case that the phrase doesn’t have the quantized reference property. Regarding events, an important question for quantized events is whether there is anything in the event that can be gradually processed. This does not necessarily have to be a physical patient entity (Dowty, 1991: ‘Incremental Theme’), but can also be some spatial or even temporal entity. Spatial entities (e.g. stretch of road) are the entities that are processed by verbs of motion. Temporal entities are gradually

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The classification of several of these expressions in this group needs justification. While count expressions such as ‘three apples’, ‘2 kg of tomatoes’ undoubtedly have the quantized reference property, plural pronouns such as ‘they’ or plural expressions such as ‘the women’, ‘these/those houses’ leave more room for ambiguity. There is of course the fact that the quantity value of such NPs is constantly changing with each context, while the quantity value of say ‘three apples’ does not change. But even within a given context, the quantity value of definite deictic expressions can be slightly vague sometimes. Consider the following example:

36 The Exhaustion Particles in the Yi group Processed entities Verb

Physical entity

Spatial entity

Temporal entity

eat breathe type walk push a cart serve (a sentence) waste

two sandwiches 3 m3 air 100 letters — — — €100

— — — 2 km (of road) 2 km (of road) — —

— — — — — 2 years 2 hours

Figure 2

Types of processed entities.

2.2.3 Bounded event denotations Events have an extension in time. The idea of a (temporal) boundary is that an event goes on until a certain limit. It is different from the notion of ‘quantized event’. The latter always provides a measure for the totality of an event, while the former does not make any commitment about the totality of the event, but only about its final segment. There are different types of boundaries that can be expressed by a language. (i) Boundary through a resultative state: This type of boundary is richly attested in the languages of East and Southeast Asia. Most of these languages employ verb compounds V1V2

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

processed by a number of verbs, such as ‘serve time’ (i.e. serve a sentence) or ‘waste time’ (see figure 2). Whenever there is an entity in the event that can be gradually processed, it may serve as a measure function of the whole event. When the processed entity is quantized, as the examples of figure 2, it may impose some holistic spatial-temporal boundaries on the event. In addition, a measure function may also be expressed by time adverbials in expressions such as ‘work for five hours’. Such event-denotations are quantized in the above sense. Finally, singular denotations, as defined in section 2.2.1, are quantized. To prove this, suppose that P has a singular reference and that x and y are two entities with P(x) and P(y). From the singular property SGL, it follows that x and y are atoms of O: ATOM(x, O) and ATOM(y, O). From the atomic property it is obvious that : y 7 x. Consequently, P is of the quantized reference type.

Matthias Gerner 37

where V2 is a grammaticalized verb expressing a boundary-result of V1. Here are some classic examples from Mandarin:
zha˘o-da`o (‘search-arrive’ ¼ ‘find’)
r
e ng-dia`o (‘throw-fall’ ¼ ‘throw away’)
ba`n-ha˘o (‘handle-well’ ¼ ‘handle successfully’) (ii) Boundary through an expression of extent:
work to exhaustion
eat to fullness (iii) Boundary as the destination of a motion verb:

(iv) Boundary in the lexical structure of the verb: This type of event-denotation has to be distinguished from the preceding ones, since the verbs of this class seem to be ambiguous as to whether they include a preparation phase in their lexical frame or not. For example ‘win a match’ can be interpreted to mean just the decisive endpoint or, alternatively, the endpoint plus a certain preparation phase leading to this culmination.





die close the door reach the summit win a match

These diverse events share a common property: they are closed under final stages (cf. Naumann 2001: 30).3 When e is an event belonging to the denotation of ‘walk to the station’, then every final stage of e# also belongs to the same denotation. Every final stage of the event ‘win a game’ is again an event of this denotation. The notion of 3 As pointed out by one reviewer, the condition of closedness under final stages appears to be problematic for a number of examples listed in this section. For instance, it is questionable whether we should view ‘eating to fullness’ as being closed under final stages, since the consumption of very small portions of food is unlikely to imply a resultative state of fullness. Similarly, for ‘walking to the station’ it may be debatable whether the subevent of walking the last meter may still be adequately termed by this label. Naumann (2001) views events as operations between initial and resulting states. If one takes this view, then all would depend on the initial state. If the initial state already is nearfullness or vicinity to the station, then eating the last morsel or walking the last meter could be viewed in a technical way as events of final stages. The problem of fuzziness with respect to final stages appears to be strictly parallel to the property of divisibility as discussed in section 2.2.4 (i.e. the question of whether masses or activities have discrete parts). Divisibility as well as boundedness are idealized reconstructions of the reality (Krifka 1989: 40).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011


walk to the station
swim to the coast

38 The Exhaustion Particles in the Yi group ‘final stage’ refers to a special type of sub-event that is defined in terms of the underlying event. This reference type was defined by Krifka (1992: 35) and I want to reproduce it here in a slightly modified version under the name BND (‘bounded reference type’). For every predicate P, events e; e# and times t; t#, let us define the notion of ‘terminal point of an event’ and then BND: TPðeÞ ¼ t4Tat ðtÞ ^ t8sðeÞ ^ "t#½Tat ðt#Þ ^ t#8sðeÞ/t#
2.2.4 Homogeneous object and event denotations The homogeneous reference type is defined as the conjunction of two other reference properties: CUM (cumulativity) and DIV (divisibility). There has been some discussion in the literature about whether CUM alone or CUM and DIV combined can characterize mass terms against count terms. Quine (1960: 91) was one of the first scholars bringing CUM into play: ‘So-called mass terms like ‘water’, ‘footwear’, and ‘red’ have the semantic property of referring cumulatively: any sum of parts which are water is water.’ However, this portrayal also applies to plural count nouns such as ‘people’ and cannot alone be sufficient for a characterization of mass terms. Scholars have therefore adduced a second

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Although the concept of boundedness plays a prominent role for events, there is no fruitful analogy in the nominal domain. Similar to events which have a temporal extension, physical objects extend in space. However, this extension is not dynamic, but static; it is not unidimensional (like the temporal extension of events), but multidimensional. Locational NPs like ‘shoes under table’; ‘houses at this side of the river’ simply do not have the semantic properties of the events discussed in this subsection. It is not possible to define an analogous notion to that of a terminal point, since there is no canonical linear ordering in the three-dimensional space. The NPs coming closest to bounded VPs are expressions such as ‘road to Rome’; ‘rope attached to the tree’, i.e. NPs that resemble in shape uni-dimensional entities. Since these NPs play a marginal role compared to with other types of noun phrases, I will not develop a theory based on them here. It is now evident that the singular reference type is also bounded. To verify this, suppose that P has singular reference, that we have P(e), that e# is a subevent of e (e# 8 e), and that TP(e) ¼ TP(e#). As P is singular, it follows that Tat(s(e)), that s(e) ¼ TP(e) and finally that s(e) ¼ s(e#). This is only possible if e ¼ e#. Consequently we have P(e#) and finally that P has bounded reference.

Matthias Gerner 39

If x is of the denotation ‘gold’ and x# a part of x, then x# is also of denotation ‘gold’, unless we have information about x# not being of denotation ‘gold’ (Krifka 1989: 40). Gillon (1992) takes a somewhat different approach by suggesting that not semantics alone but semantics combined with (languagespecific) morpho-syntactic properties can offer a characterization of mass NPs against count NPs: ‘. . . syntactic features and the constraints imposed by their semantic interpretation . . . determine the differences between mass noun phrases and count noun phrases’ (Gillon, 1992: 599). A similar position is held in this paper, especially in the context of classifier languages of East and Southeast Asia (to which the Yi languages belong). In these languages, the distinction singular/plural is not lexically encoded in the noun. Every common noun (whether count, e.g. ‘person’, or mass, e.g. ‘water’) satisfies CUM. Whether a common noun agrees with DIV is not always immediately obvious.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

criterion (Cheng 1973): ‘any part of something that is water is water’ (DIV). This second property has been controversial, because of the problem of whether masses contain minimal parts. The problem of minimal parts can be illustrated in the following way for the mass noun ‘water’. While a molecule of water still can reasonably be classified as water, a proton of an atom of the molecule of water probably cannot (see Krifka, 1989: 40, for a related example). The problem of minimal parts is obvious not only at the nuclear level, but also with mass terms such as ‘furniture’ or ‘cattle’ whose minimal parts are palpable without special technological assistance. A number of proposals and attitudes have been advanced to treat the problem of minimal parts. Several scholars view mass entities as exhibiting minimal parts. For example, Quine (1960: 97) wrote: ‘In general a mass term in predicative position may be viewed as a general term which is true of each portion of the stuff in question, excluding only the parts too small to count.’ On the other hand, Bunt (1979: 255–256; 1985: 45–46) argued for the retention of DIV by questioning the relevance of the existence minimal parts at the linguistic level: ‘mass nouns provide a way of speaking about things as if they do not consist of discrete parts’ (Bunt, 1985: 45, italics is in the original; see also 1979: 255). Along the same line, Krifka acknowledged that DIV is an idealized reconstruction of the reality, but proposed to adopt it as a working hypothesis by suggesting a (non-monotone) default rule at work (borrowed from Reiter, 1980):

40 The Exhaustion Particles in the Yi group

33 (13a) vI55ga33 gvu33 garment NUM:3 CL ‘three garments’

(13b)

33

33

t 21 water NUM:3 bowl ‘three bowls of water’

The difference between a classifier and a measure word is that the former ‘designates’ (Croft, 1994: 163) or ‘actualizes’ (Bisang, 1999: 121) the existing natural shape-structure boundaries belonging to the concept of a noun in question, while the latter creates arbitrary shapestructure boundaries to its concept. The measure word itself is a noun or a standard measure word such as ‘meter’, ‘kilogram’, etc, whereas the classifier is abstract, has no independent lexical meaning and occurs only in the context of other nouns. (More precisely: classifiers are often historically derived from nouns; most have lost the nominal meaning, but some have preserved it.) It now makes sense to draw a limit between count and mass nouns by saying that count nouns have a shape-structure classifier and mass nouns haven’t. Departing from this property, most languages of the Chinese sphere would put a demarcation line in the above scale somewhere between ‘sand’ and ‘powder’ (in contrast to English where the limit lies between ‘raspberry’ and ‘rice’). For terms like ‘rice’ and ‘sand’ there exist a classifier hinting at the shape of a grain in languages such as Chinese or the Yi languages (the following example originates again from Liangshan Nuosu): (13c)

44

33

33

t a33 sand NUM:3 CL ‘three grains of sand’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Consider the string of entities with increasingly small minimal parts: ‘apple-tangerine-raspberry-rice-sand-powder-air’. In English one would hesitate to draw a quick demarcation line between count and mass nouns, though most people would probably draw a line somewhere between raspberry and rice. This is because the morphosyntactic properties of raspberry and rice are felt differently: three raspberries, three rices. In the languages of East and Southeast Asia the critical test for separating count from mass nouns is whether the noun in question can employ a classifier. Since in these languages a common noun merely denotes a concept, a classifier or a measure word has to intervene in quantized expressions, such as in ‘garment+three+CL’ (‘three garments’) or in ‘water+three+MW’ (e.g. ‘three bowls of water’). Consider the following examples from Liangshan Nuosu, one of the Yi languages explored in this paper.

Matthias Gerner 41

(i) gradual-process verbs where the processed entity is specified and homogeneous (cf. figure 2)





eat cake breathe air walk distances waste time

(ii) gradual-process verbs where the processed entity is unspecified
breathe normally
walk quickly
push a cart (iii) verbs that do not admit processed entities
laugh, cry
snore

2.3 Object and state structure The part-whole relationship is a canonical ordering for objects as well as for events. Do states also exhibit an inherent ordering? For atemporal states, no sensible interpretation in terms of part-whole seems possible, but comparable states can be ordered with regard to degree. As the meaning of the exhaustion particle includes that of superlative—which is a notion built on degrees—it appears plausible to formalize the

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Nouns such as ‘powder’, ‘air’ or ‘water’ do not exhibit any shapestructure classifier (though they admit plenty of measure words), and it seems appropriate to view them as ‘a way of speaking about things as if they do not consist of discrete parts’ (Bunt, 1979: 255; 1985: 45) or in other words as satisfying DIV. Homogeneous events also represent a prominent event class. For some of the activities (e.g. eat or breathe) it may appear uncertain whether they comply with DIV. Eating for example may be viewed as displaying three minimal parts: inserting of food, chewing and swallowing. This analysis would put ‘eating’ in contradiction with DIV. These reflections notwithstanding, for many native speakers it would not be odd to label any of these three sub-events (inserting, chewing and swallowing) as an eat-event. Again, this is related to the fuzziness of human language (which is ‘a way of speaking about things as if they do not consist of discrete parts’, even though they do in reality). There are three types of homogeneous events.

42 The Exhaustion Particles in the Yi group various types of states with respect to the notion of degree. However, in my opinion degrees cannot be well formalized in a way that would allow us to apply the singular, quantized and homogenous reference types to states. Instead, we will work with the notion of comparison class associated to a state. An appropriate formulation of these reference types for states will be the topic of this subsection. In the literature, there are two approaches to gradeable states. The first (cf. Cresswell 1976; Hellan, 1981; von Stechow 1984) gives to the notion of degree a proper ontology (similar to time points) and interprets a gradeable adjective as a relation between objects and degrees. For example the expression ‘Bill is tall’ has the logical representation: where d is the degree of height of Bill. In the literature, the notion of degree is either employed as an unanalysed primitive or else characterized as the equivalence class of an equivalence relation defined on the objects of a universe for a given adjective (see Cresswell 1976: 281). For example, the degree d of Bill’s height is defined as the set or equivalence class of all the objects that are not taller and not less tall than Bill, i.e. the set {a: a  (tall-equivalent to) Bill}. The associated equivalence relation is formally defined as: a  a# 4 for all b
a is taller than b 4 a# is taller than b
b is taller than a 4 b is taller than a# Klein (1980) criticizes this approach because it leaves the crucial notion of degree undefined or relies on the notion of comparative, another unanalysed primitive. From the perspective of this paper the degree approach is also unattractive, because there is no obvious way to define the reference type of states in relation to the position or range of degrees they have on a scale of degrees. The second approach to gradeable states, initiated by Klein (1980) and others, replaces the notion of degree by that of comparison class. A class of comparison is a group of object-entities, which occur in the background assumptions of the speech participants and against which a vague statement such as ‘Bill is tall’ is evaluated. It relates to the notion of ‘topic of conversation’ and coincides often with the set of things people in a conversation happen to talk about (Klein 1980: 13). When the class of comparison of Bill’s height consists of his classmates, his tallness may be evaluated in a different way than when the class of comparison is a group of basketball-club mates. In our approach, the

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(14) ktallk(d) (Bill)

Matthias Gerner 43


pos(c, P) ¼ {x 2 U(c, P) j F(x) ¼ 1} (the items that are definitely tall)
neg(c, P) ¼ {x 2 U(c, P) j F(x) ¼ 0} (the items that are definitely not tall)
undef(c, P) ¼ {x 2 U(c, P) j F(x) undefined} (the items that are neither tall nor short). Klein’s concept of a class of comparison can be easily integrated into the architecture of LEXH type. It is based on a formal set of contexts.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

comparison class of a state is a formal object and its reference type as object-denotation closely correlates with the type of a state. Before going into details, I would like to adopt the convention that the NPreferent qualified by a stative predication always belongs to the comparison class (implying that a state always compares at least the main argument to itself). Even though it may also be conceivable to exclude the main argument from the comparison class, we will opt for its inclusion. The main motivation for this convention is related to ungradeable states. Following the choice, ungradeable states like ‘pregnant’ or ‘dead’ would correlate with comparison classes that are singular or singleton (having one unique member). This in turn would allow us to view ungradeable states, punctual events and singular objects as structurally similar. Gradeable states such as ‘tall’ stand in correspondence to quantized or homogenous comparison classes. In section 3.4, we will see that the exhaustion particle is incompatible with ungradeable states and compatible with certain gradeable states (those that exhibit a quantized comparison class). Let us go on with a formal account of comparison classes. The notion of a class of comparison is context-dependent. It is defined by Klein as a function U that maps every context c and every statepredicate P onto a set of object-entities U(c, P). For example consider P: ‘tall’ and c: ‘context in which Bill’s classmates are considered as the relevant group for purposes of judging whether someone counts as tall’. The set U(c, P) is defined as {x: x is a classmate of Bill}. It does not seem appropriate to interpret P: ‘tall’ in any model as a function De / {0, 1}. This interpretation does not express the vagueness of the predicate P: ‘tall’, because it divides the whole discourse up into those that are definitely tall and those who are definitely not tall, but would leave no intermediate room for those who are neither tall and nor small. Klein therefore proposed to interpret P: ‘tall’ by a partial function U(c, P) / {0, 1} which leaves those items which are neither tall nor small undefined. A partial function F: U(c, P) / {0, 1} is determined by a partition of De consisting of three subsets:

44 The Exhaustion Particles in the Yi group Since, for reasons of simplification, our semantic representation language LEXH type is supposed to be purely extensional (see section 2.1), I adapt his definition by getting along without contexts or by saying that there is only one context (written as ). This will not hamper the analysis, as the focus in this paper will be on the various reference types a state may represent within a given context. The object-entities U(, P), pos(, P), neg(, P) and undef(, P) can be defined as predicates COMP(P), POS(P), NEG(P) and UNDEF(P). For every stative predicate P, a comparison class is thus defined formally by the following data added to LEXH type :

At the model-theoretic level, we would have to adopt Klein’s notion of a partial model. A partial model of LEXH type is defined exactly as before with the exception of stative predicates P which are interpreted as a partial functions F(P): De / {0, 1} such that
{x 2 De j F(P)(x) ¼ 1} ¼ {x 2 De j F(POS(P))(x) ¼ 1}
{x 2 De j F(P)(x) ¼ 0} ¼ {x 2 De j F(NEG(P))(x) ¼ 1}
{x 2 De j F(P)(x) is undefined} ¼ {x 2 De j F(UNDEF(P))(x) ¼ 1} For the purpose of this paper we are interested in the possible reference types the comparison class COMP(P) of P may be associated with. In fact, since COMP(P) is a predicate of object-entities, we may define three types of states that can be distinguished by means of the reference type of COMP(P): states with singular, states with quantized and states with homogenous comparison classes. First, there are states whose comparison class is singleton or singular (i.e. composed of one element). The states of this type are basically ungradeable, e.g. ‘Mary is pregnant’, ‘the books contain an introduction’, ‘those famous monuments stand in Paris’, etc. The comparison class of these predications is a singleton each time independently of whether the argument is a singular or plural NP: {Mary}, {the books} and {those famous monuments}. The second type consists of states whose comparison class is quantized for a given context. For example, the comparison class of Bill’s classmates related to the sentence ‘Bill is tall’ is (supposedly) definite and thereby quantized. The third type is composed of states whose comparison class is homogeneous for a given context. If the

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011


COMP is a function symbol of type e, t>, <e, t (from stative predicates to object predicates)
POS, NEG, UNDEF are function symbols of type e, t>, <e, t constituting a partition of COMP (I skip the formulas expressing that COMP is partitioned by POS, NEG, UNDEF)

Matthias Gerner 45

context of ‘Bill is tall’ relates for example to all the people in the world, then the associated comparison class would be of the homogenous reference type. I will provide further illustrations of these notions in the next section with data from the Yi languages (especially in section 3.4). 3 THE YI (LIANGSHAN NUOSU) DATA Since the Yi languages have quite peculiar data to offer, I take some space to develop the empirical base of this paper.

The Yi languages belong genetically to the Tibeto-Burman language family, are spoken in Southwest China and have Mandarin as a remote relative. They are verb-final, basically SOV. Furthermore they display an analytical (isolating) morphology, which means that most grammatical categories are under-specified or optional. A bare common noun like ‘table’ can be interpreted in different contexts as ‘table’ (singular), ‘tables’ (plural), ‘a table’ (indefinite), or ‘the table’ (definite). Although devices like ‘definite article’, ‘plural morpheme’ and the like are available, these tools are optional and in use only when the speaker places a focus on these concepts. The syntactical slot in which the exhaustion particle (EXH) appears is directly after the predicate (SOV+EXH) where it shares its position with a host of other verb particles expressing categories like progressive, perfect, modality, etc. The basic operation of EXH is to contribute up to three different meanings to the clause of which one or all may be erased due to the internal semantic settings of the situation or due to pragmatic inferences: (i) EXH functions as universal quantifier of the first, i.e. the sentence-initial, argument, (ii) EXH contributes the meaning of ‘completely’ (for events) or ‘most’ (for states), (iii) combination of (i) and (ii). It is important to keep in mind that these several meanings are processed in parallel. One or several meanings may be erased due to incompatibilities of EXH with the internal makeup of the sentence. If all meanings are erased, then the sentence to which EXH is attached is ungrammatical. If none is deleted, the hearer comprehends a combination of surviving meanings. While this pattern of quantification seems to be completely unfamiliar in European languages, it might not be that uncommon on a global scale. In fact in the massive collection work ‘Quantification in Natural Languages’ (Bach et al. 1995), Jelinek (1995) describes

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

3.1 Introduction

46 The Exhaustion Particles in the Yi group a similar case from Straits Salish, a native North American language.4 The data presented in this section originate from Liangshan Nuosu, the most populous of the Yi languages (spoken by 2.2 Million people in Sichuan Province), where EXH is realized as the morpheme sa55. Besides elicited sentences, I will select example sentences in this section from two published sources: Li and Ma’s conversational textbook of Nuosu (Li & Ma 1981) and Chen and Wu’s collection of 11 folk stories published in the appendix of their grammar (Chen & Wu 1998: 216–269).

In order to let the operation of EXH on clause-initial arguments appear more obvious, I will mainly assemble in this section punctual and bounded events, ungradeable states, as well as homogeneous events or states, that is events or states that cannot receive the quantification operation of EXH. 3.2.1 Individuals and duals The main claim of this sub-section is that EXH is incompatible with noun phrases that describe an individual (atom) or a dual. Examples (15a) and (15b) are homogeneous events that cannot be modified by the exhaustion particle. Since the sole argument in (15a) is an individual and in (15b) a dual NP, the whole sentence is ungrammatical. If EXH were taken away, as in (15c), the ungrammaticality would be lifted. (15a) * 33ka55 bo33 sa55 o44. name of man go EXH DP ‘Muga went all.’ (15b) *

33

ka55 si33ni21 a33ma55 ( i44) bo33 sa55 o44. name of man and name of man NUM:2 go EXH DP ‘Muga and Shama went (both) all.’

4 EXH seems to be very reminiscent of Jelinek’s Straits Salish quantifier m*k#w. There is however a difference between the Yi exhaustion particles and Salish m*k#w. The Salish quantifier is not reported as taking a stative predicate in its scope with the meaning ‘most’. Otherwise, there seems to be much similarity. Witness (Jelinek 1995: 512–514):

m*k#w 1 ‘*w’ sa-t-B c* scˇeenxw. ALL 1pNOM LINK eat-TR-3ABS DET fish (i) ‘We ate all the fish.’ (ii) ‘We all ate the fish.’ (iii) ‘We ate the fish up completely.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

3.2 Objects

Matthias Gerner 47

(15c)

33

ka55 bo33 o44. name of man go DP ‘Muga went.’

Example (16) is ungrammatical because the event type is bounded and because the clause-initial argument is a noun phrase with one unique referent in the world (displaying the singular reference type). However, if (16) was uttered in the famous Chinese prehistorical myth in which the ancient world had exactly 10 suns, the sentence would be acceptable and translate as ‘The suns have all risen.’

The following examples illustrate that natural numbers above two are compatible, and those below two are incompatible with the operation of EXH. The event itself is bounded and cannot be modified by the exhaustion particle. i21 bu55 a33 si21 bo33 sa55 (17a) *th 21 33 a33d 33 book DEM:DIST NUM:2 CL 1P SG take go EXH ‘I took all two books away.’ 33 (17b) th 21 33 a33d 33 bu55 a33 si21 bo33 book DEM:DIST NUM:3 CL 1P SG take go sa55 o44. EXH DP ‘I took all three books away.’

Example (18a) exhibits a clause-initial argument whose number interpretation is ambiguous. The number reading that would be compatible with EXH is pragmatically (culturally) slightly implausible. 33 i33 ndzi33 i21 sa55 (18a) #tsh 21 3P SG+POSS house river side be at EXH ‘His/her house is all at the river side.’ 33 (18b) tsho21 o44 i33 ndzi44 i21 sa55. 3P PL house river side be at EXH ‘Their houses are all at the river side.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

33 la33 sa55 (16) *ho33bu33 sun come out come EXH ‘The sun has all risen.’

48 The Exhaustion Particles in the Yi group The noun phrase tsh 21 i33 in isolation has two readings: ‘his/her house’ (singular) or ‘his/her houses’ (plural). Since it does not happen often in Liangshan that someone has several houses and all of his houses are at the riverside and since the singular reading ‘his/her house’ stays in conflict with the quantification operation of EXH, the natural reflex of Nuosus is to judge (18a) as pragmatically implausible. The noun phrase tsho21 o44 i33 in (18b) has two readings in isolation: ‘their house’ and ‘their houses’. The presence of the exhaustion particle in (18b) imposes the plural meaning ‘their houses’, which is a quantized NP. (Quantized NPs can be modified by EXH, see next sub-section.)

21 -su33 th 21 (19) tso33 hi55 3P PL NUM:8 ART¼CL-NOML book ‘The eight people are all reading books.’

33

h 21 sa55. see, watch EXH

Sentence (20) is a quantized event involving the gradual verb ‘drink’. Its quantized Incremental Theme is in clause-initial position. This setting creates two readings, which turn out to be equivalent in meaning. (20)

i21 i33 a33 ka44 ndo33 water NUM:2 MW:bottle 1P SG COV:arrange drink 55 sa o44. EXH DP (i) ‘All of the two bottles of water was drunk by me.’ (ii) ‘The two bottles of water were completely drunk by me.’ I

33 h 55

As mentioned in section 3.1, Nuosu has a host of optional noun modifiers such as ‘demonstratives’, ‘definite articles’, etc. (21a) and (21b) exhibit two quantized NPs, the first modified by the plural definite article and the second by a plural demonstrative. Both modifiers are compatible with EXH. (21c), (21d), and (21e) involve vague noun quantifiers and cannot interact with the exhaustion particle.5 5 In Nuosu, the category of definite article is made up of collocations [classifier + su33] (where su33 is a nominalization particle).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

3.2.2 Quantized objects Objects that have the quantized reference property are configurable with the exhaustion particle. EXH contributes then the meaning of a universal non-distributive quantifier. Example (19) describes a homogeneous event with a count NP in initial position. The operation of EXH is to refer to the ‘entire object’.

Matthias Gerner 49

(21a) kh 44 g 33-su33 i21kho33 hi33 o44 dog ART¼COLL-NOML door LOC:outside 33 55 o sa . exist, have EXH ‘The dogs are all outside the door.’

(21c) *kh 44 dog o33 exist, have ‘That dog is

a33d 44 i33 i21kho33 hi33 o44 DEM:DIST CL door LOC:outside sa55. EXH all outside the door.’

(21d) *kh 44 g 33 i21kho33 hi33 o44 dog COLL door LOC:outside 33 55 o sa . exist, have EXH ‘Some dogs are all outside the door.’ a44 i33 g 33 i21kho33 hi33 o44 (21e) *kh 44 dog many COLL door LOC:outside 33 55 o sa . exist, have EXH ‘Many dogs are all outside the door.’ 3.2.3 Homogeneous objects The exhaustion particle cannot quantify mass bare nouns in sentence-initial position (pointing to homogeneous objects), unless it makes sense that the totality of this mass in the world may be portrayed in the predication. The following sentence taken from a Grimm brothers-style fairy tale6 is just of this kind. In this story, the emperor and his daughter share a common love for gold and silver. In (22), a longstanding dream becomes reality: the transformation of all organic and inorganic material of the world into gold and silver. The predication involves a gradual verb with an incremental argument in clause-initial position. The totality of a mass-object of a certain type 6

The example originates from the story ‘‘The emperor and his daughter’’ (Chen and Wu 1998: 266).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(21b) kh 44 a33d 44 g 33 i21kho33 hi33 o44 dog DEM:DIST COLL door LOC:outside o33 sa55. exist, have EXH ‘Those dogs are all outside the door.’

50 The Exhaustion Particles in the Yi group (here ‘all existing wood and stones/rock’) constitutes a homogenous object. EXH contributes two meanings to the clause, which happen to be equivalent in meaning. (22)

33 h 55 h 33 33 s 33bo33 ma55 o u i21 tree, wood stone change silver gold develop sa55 o44. EXH DP (i) ‘All the wood and stones changed into silver and gold.’ (ii) ‘Wood and stones changed completely into silver and gold.’

(23)

33

33

go33 sa55. sea water cold EXH ‘The sea water is coldest.’

3.3 Events There are four event types that interact with the exhaustion particle in various ways, i.e. punctual events, quantized events, bounded events and homogeneous events. EXH is fully configurable only with quantized events. It is pragmatically implausible with homogeneous events and outrightly incompatible with punctual and bounded events. To make this obvious I will mainly assemble sentences where the clause-initial argument is a singular NP. 3.3.1 Punctual events The exhaustion particle is incompatible with punctual events. The Nuosu verb tso33 ‘meet, run into’ is a punctual verb that only holds for an instant. It admits the operation of EXH only when the clause-initial argument is not singular, as in (24a), but not as in (24b) and (24c). (24a)

tso33 sa55 o44. tsho21 o44 a33ma55 3P PL name of man meet, run into EXH DP ‘They all met Shama.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The next example is taken from a world more closely resembling ours. It involves a gradeable state and a mass bare noun in clause-initial position (i.e. ‘sea water’). From the two meanings the exhaustion particle can convey, the object quantification meaning is erased because it would result in an interpretation such as ‘all existing sea water in the world is cold’, thus appearing to be pragmatically odd. Consequently, the exhaustion particle only conveys the meaning of superlative here. EXH suggests a context where the temperature of a small and definite number of items (or liquids) is compared.

Matthias Gerner 51

(24b) * a33ma55 tsho21 o44 tso33 sa55 o44 name of man 3P PL meet, run into EXH DP ‘Shama all met them.’ (24c) * 33ka55 a33ma55 tso33 sa55 o44 name of man name of man meet, run into EXH DP ‘Muga all met Shama.’

(25a) *ts 33 a44tsI33 33 sa55 o44. 3P SG sneeze EXH DP ‘S/he sneezed completely.’ (25b)

o21 o44 a44tsI33 1P PL sneeze ‘We all sneezed.’

33

sa55 o44. EXH DP

tI21 u55 sa55 (26a) *ts 33 3P SG electricity touch EXH ‘S/he got completely an electric shock.’ (26b) tsho21 o44 tI21 u55 sa55. 3P PL electricity touch EXH ‘They all got an electric shock.’ Other punctual verbs include mo33 ‘perceive’ or expressions like ‘jump over the gate’, etc. They reveal the same pattern of quantification. 3.3.2 Quantized events Quantized events are events that stop when the quantized processable entity is completely processed. For this situation type the exhaustion particle contributes two distinct meanings to the clause that may collapse. First it quantifies the sentence-initial argument (which might be either the initiator of the processing or the processed entity). Second it conveys the meaning of ‘completely’. These two readings are equivalent if the sentence-initial NP is the processed entity. Witness the event ‘eat ten nuts’ and its correlates: ma33 dz (27a) tsho21 o44 s 21 i33 tsi44 3P PL nut NUM:10 CL eat

33

sa55 o44. EXH DP

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The same applies to the following two pairs of punctual events. The first pair involves the intransitive verb ‘sneeze’, the second pair the transitive verb ‘get an electric shock’.

52 The Exhaustion Particles in the Yi group (i) ‘They all ate ten nuts.’ (ii) ‘They completely ate up ten nuts.’ (iii) ‘They all ate up ten nuts.’ (27b) tsh 33 s 21 i33 tsi44 ma33 dz 3P SG nut NUM:10 CL eat ‘S/he completely ate up ten nuts.’

33

sa55 o44. EXH DP

In (27a), two distinct non-equivalent readings are involved. The reading (27a(i)) expresses that the clause-initial argument is universally quantified, whereas (26a(ii)) conveys a completive meaning. (27b) only communicates a completive meaning and not also quantification of the clause-initial argument, since the latter is singular. In (27c) the position of the arguments are inversed. The patient role appears in focus position. This construction is best approximated in English by a passive. Two readings are conveyed that eventually collapse, since the processing of all input (27c(i)) is basically equivalent to the completion of the process (27c(ii)). Example (28) was selected from a folk story7 in which a Nuosu man whose name is Redisofu deals with a female demon and attempts by several methods to bring about her death. One of his attempts is described in (28) where he uses bamboo rods to beat her. (28) ma33 da33 gu33 vi33 t i33 sa55. bamboo rod NUM:9 MW:load whip, here: use up EXH (i) ‘[Redisofu] used up all nine loads of bamboo rods in beating [her].’ (ii) ‘[Redisofu] completely used up nine loads of bamboo rods in beating [her].’ The idea is that of a fierce battle where Redisofu used up successively nine loads of bamboo rods lying beside him prepared for this occasion. When the first rod of the first load broke after repeatedly striking the demon, he took up the second rod of the first load and so forth. When the last rod of the ninth load eventually broke, Redisofu overcame the demon. The exhaustion particle does not quantify 7

Selected from Chen and Wu’s collection of folk stories (Chen & Wu 1998: 237–252).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

ma33 tsh 33 dz 33 sa55 o44. (27c) s 21 i33 tsi44 nut NUM:10 CL 3P SG eat EXH DP (i) ‘All of the ten nuts were eaten by him/her.’ (ii) ‘Ten nuts were completely eaten up by him/ her.’

Matthias Gerner 53

(29a)

a33 kha33 33 tsh 33 sa55 o44. 1P SG face wash EXH DP ‘I have completed washing my face.’

(29b)

o21 o44 kha33 33 tsh 33 sa55 o44. 1P PL face wash EXH DP (i) ‘We have all washed our faces.’ (ii) ‘We have completely washed our faces.’ (iii) ‘We have all washed our faces completely.’

When the sentence-initial argument is singular as in (29a), then, again, the sole meaning conveyed is ‘completely’. When it is plural, three readings are implied: (29b(i)), (29b(ii)) and (29b(iii)). 3.3.3 Bounded events The exhaustion particle is incompatible with bounded events, because in contrast to quantized events, no measure of the entire event is given, only an endpoint. In Liangshan Nuosu, the directional verb la33 ‘come’ constitutes a bounded event involving an inherent endpoint, which is implied even when no explicit destination is spelled out. The same does not apply to its counterpart bo33 ‘go’ which is homogeneous. Consider (30a) where EXH suggests a definite-plural reading of the argument tsho33 ‘person’. In a context where one would deliberately allow all humans 8 Sentence (29a) was selected from Li and Ma’s conversational textbook (1981: 23). The sentence is uttered during a dialogue between a certain Gaga and someone whose name is Wuda. In this dialogue Gaga tries to be of assistance to Wuda during his basic morning activities. Gaga asks Wuda if he needs soap to which Wuda replies (29a).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

‘Redisofu’ since it is the name of an individual (and, moreover, it is a zero-anaphora NP), but it relates to the ‘nine loads of bamboo rods.’ (28) conveys two meanings, which turn out to be equivalent due to the contextual setting: (i) ‘Redisofu exhausted all nine loads of bamboo rods’ (i.e. the beating had to stop because Redisofu ran out of instruments); (ii) ‘Redisofu completed all the beating’ (the beating was completed when the nine loads of bamboo rods were used up). In the event ‘wash one’s face’, one’s face is a quantized expression. The event stops when the washing of all parts of the face is terminated. In a normal and healthy pragmatic setting, the washing of a quantized entity is a progress event like ‘eat an apple’ where the internal argument is successively processed. In English the Vendlerian test indicates that the event is an accomplishment: ‘I washed my face in ten minutes/ #for ten minutes.’ The same holds for Nuosu.8

54 The Exhaustion Particles in the Yi group to be referents of tsho33, the involvement of EXH would result in pragmatic oddity. If it was replaced by the singular personal pronoun tsh 33 ‘he/she’, as in (30b), the sentence would become ungrammatical. The reduplication of sa55 in (30a) is the way in Nuosu to express alternative question.9 (30a) tsho33 la33 sa55 sa55 o44? person come EXH EXH DP ‘Did all the people come?’

Generally speaking, every directional verb with an explicit destination represents a bounded event, as in (31). (31) * a33 o21 o33 33 sa55 o44 1P SG Xichang go down EXH DP ‘I have completely gone down to Xichang.’ Other verbs with inherent endpoint in the lexical coding include intransitive verbs such as s 33 ‘die’ and transitive verbs such as go55 ‘close’. 33 33 k 21 go55 ta33 sa55 o44. (32) i21kho33 door wind PASS close STP EXH DP ‘All the doors were closed by the wind.’ Another important group of bounded events in Nuosu and other East Asian languages is formed by compound verbs V1V2 with a main verb V1 and a grammaticalized resultative verb V2 (see also section 2.2.3). The resultative state expressed by V2 appears as a boundary of the whole event. Again, the exhaustion particle cannot modify bounded-events predicates of this kind (it would only quantify the clause-initial argument).

(33)

33 55 21 44 o21 o44 ka sa55 o44. 1P PL name of man search GET EXH DP ‘Muga found us all.’

3.3.4 Homogeneous events Homogeneous events are divisible and cumulative. Similar to homogeneous objects, EXH is incompatible 9

(30a) was selected from Li & Ma’s conversational textbook (1981: 5). The question is uttered by the leader of an agricultural commune who wonders whether the full number of co-workers have shown up in order to start the daily work.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

la33 sa55 sa55 o44 (30b) *tsh 33 3P SG come EXH EXH DP ‘Has he all come.’

Matthias Gerner 55

bo44 d 33 li33 sa55. (34a) *tsh 33 3P SG mountain(ous area) climb go up EXH ‘S/he walks completely in the mountains.’ d 33 li33 sa55. (34b) tsho21 o44 bo44 3P PL mountain(ous area) climb go up EXH ‘They all walk in the mountains.’ Example (35) involves a gradual verb whose incremental sole argument is a mass (¼ homogeneous) noun: ma33ha33 ‘rain’. Out of context, in absence of any overt boundary, the hearer of (35) looks at the maximal boundaries of ‘all existing rain’ and evaluates (35) as something odd like ‘all the existing rain poured out’ or, equivalently, ‘the rain has completely poured out’. Yet, (35) might be said in a context where a heavy precipitation takes place. In this case, the argument in (35) would be covertly quantized: ‘the rain related to the current precipitation’. (35) reveals thus that the involvement of EXH in homogeneous events with gradual verbs depends on pragmatic factors. i21 sa55 o44. (35) #ma33ha33 rain develop EXH DP ‘It has stopped raining.’ In a similar way, the verb thi33 ‘dawn’ is a gradual verb. Its sole argument 33 ‘sky’ is a homogeneous object-denotation and includes 10

‘Liangshan’, the area where the Nuosu live, just means ‘Cool Mountains’.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

with this semantic structure for events, or pragmatically implausible at best. In English, homogeneous events also do not combine with the concept of completive: ‘??Peter ran completely’. In fact, there are two types of homogeneous event-denotations. Firstly, there are those which do not admit incremental (or processable) arguments such as example (34) and, secondly, there are denotations which can incorporate incremental homogeneous object-denotations. Examples (35) and (36) belong to this subtype. EXH is disallowed in the former case, but allowed in the latter for certain pragmatic settings. In the following example, the Nuosu expression ‘climb mountains’ does not refer to the reaching of the summit of one particular mountain, but more to the walk in a mountainous area that can last for hours.10 It does not process incremental arguments. EXH is disallowed in (34a) since the clause-initial argument is a singular NP, but permitted in (34b) with a plural NP in first position.

56 The Exhaustion Particles in the Yi group meanings related to Nuosu metaphysics. It refers to all sky layers in the heavenly realms. A universal quantification over all the sky layers involved in an episodic event is unacceptable or pragmatically questionable. (36a)

33

thi33 o44. sky dawn DP ‘It was dawning.’ thi33 sa55 o44 sky dawn EXH DP ‘It was completely dawning.’

(36b) #

33

We distinguish between states whose comparison class is singular, quantized or homogeneous. Comparison classes correspond to collections of individuals or items and thus can be viewed as formal objects whose reference properties were discussed in section 3.2. Since comparison classes are implicit, a sentence in isolation may overtly manifest a choice between singular and non-singular reference, but within the non-singular reference type the membership in the quantized or homogeneous domain is often not obvious, unless certain additional assumptions about the pragmatic settings are known. Nonetheless, the examples in section 3.4.2 will show that there is some solid empirical ground for distinguishing between the quantized and homogeneous reference types for states. 3.4.1 Ungradeable states EXH does not elaborate on the semantic structure of ungradeable states, but quantifies only, if possible, over the clause-initial noun phrase. It could be argued that ungradeable states do not display any comparison class, but we will rather say that they exhibit a comparison classes that is singleton or singular – composed of the item referred to by the sentence-initial NP (see section 2.3 for more information). An important portion of non-gradeable states consists of positional predications like ‘sit on a rock’, ‘live in Beijing’, etc. When the clauseinitial NP is singular, ungrammaticality occurs. 33 ma55 tho55 i33 sa55. (37a) * a33 1P SG rock, stone LOC:on sit EXH ‘I all sit on the rock.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

3.4 States

Matthias Gerner 57

(37b)

33 o21 ma55 tho55 i33 sa55. 1P PL rock, stone LOC:on sit EXH ‘We all sit on the rock.’

33 (38a) tso21 o44 3P PL hair

tsh 21 i33 ndi55 NUM:1 CL have(esp. for body parts) ‘Each of them has one hair.’ 33 33 I tsh 21 i33 ndi55 (38b) *tsh 33 3P SG hair NUM:1 CL have(esp. for body parts) ‘He has all one hair.’ I

33

sa55. EXH sa55 EXH

33 33 I ndi55 sa55 i21s 33. (38c) tsh 21 3P SG POSS hair have EXH still ‘He still has all his hair.’

Another group of ungradeable states concerns gradeable adjectives that are modified by an intensifier. Example (39a) exhibits the gradeable adjective (a33)vu55 ‘green’ together with an expressive. Expressives form a distinct part-of-speech in Sino-Tibetan languages and add a descriptive value to the adjective.11 The adjective plus the expressive denote a non-gradeable state. In (39a), the argument is plural, whereas (39b) is incompatible with the exhaustion particle, since it displays a singular argument. Witness: vu55 33vu55lo33lo33 sa55. (39a) s 33bo33 g 44-su33 tree ART¼COLL–NOML sap-green EXH ‘The trees are all sap-green.’ 11 English approximations for expressives are compound words like ‘brand-new’, ‘crash-hot’, ‘trigger-happy’ and the like.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

A similar rule also applies to ungradeable states such as ‘contain three apples’, ‘have one hair’. The picture in (38a) is that of a group of people having one hair each. The two-place predicate ndi55 ‘have’ is ungradeable. It requires the clause-initial argument to be plural in order to be configurable with the exhaustion particle, as in (38c). In version (38c), EXH suggests a plural-definite reading of the clause-initial NP. If the personal pronoun was dropped in (38c), the initial NP would exhibit a homogenous reading (resulting in pragmatic oddity). The noun phrase in (38b) is singular and produces an incompatibility with EXH.

58 The Exhaustion Particles in the Yi group (39b) *s 33bo33 ts 44 bo33 vu55 33vu55lo33lo33 sa55. tree DEM:PROX CL sap-green EXH ‘This tree is most sap-green.’ If (39b) was stripped of the expressive, we would have a superlative sentence, since the adjective (a33)vu55 ‘green’ is gradeable. Superlative sentences are explored in the next section. (Without the expressive, when occurring as sole predicate, vu55 must carry the prefix a33.)

3.4.2 States with an implicit quantized or homogeneous comparison class The common interpretation of gradeable states such as ‘Bill is tall’ is to assess the implicit comparison class of states of tallness against which ‘Bill is tall’ is evaluated. The assessment of comparison classes is pragmatically determined. Basically, every clause describing a gradeable state can be embedded into a context with a quantized comparison class and with a homogeneous comparison class. The exhaustion particle is compatible with readings of a quantized comparison class, but often poses problems for readings of a homogeneous comparison class. These problems are similar to those I have already described for homogeneous objects (section 3.2.3) and homogeneous events (section 3.3.4). In (40a), the beauty of a garment is appraised against that of a class of non-specified other clothes. The hearer infers that a number of garments s/he is aware of are at disposition and that the one in question (‘that garment’) is the most beautiful. The implicit comparison class is the ensemble of the garments that are implicitly compared. The most likely scenario is that this class is specific definite and thus quantized in the mind of the speaker and the hearer. In the discussion of example (12) (see section 2.2.2), we already presented a similar argument for definite personal pronouns. Another setting would be the one where the beauty is evaluated against all existing garments in the world. The associated comparison class could be defined as the set of things which satisfy the property of being garments. This property or comparison class would be homogeneous or at least cumulative, its acceptability be questionable and only appropriate for hyperbolic contexts (‘the garment is the most beautiful in the world’). The totality of a mass-object of a certain type (here ‘all existing wood and stones/rock’) constitutes a homogenous object. EXH

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

(39c) s 33bo33 ts 44 bo33 a33vu55 sa55. tree DEM:PROX CL green EXH ‘This tree is greenest.’

Matthias Gerner 59

contributes two meanings to the clause, which happen to be equivalent in meaning. a33d 44 gu44 a55 sa55. (40a) i33ti44 garment DEM:DIST CL beautiful EXH (i) ‘That garment is the most beautiful.’ (ii) ‘The garment is the most beautiful in the world.’

g 44-su33 a55 sa55. (40b) i33ti44 garment ART¼COLL–NOML beautiful EXH (i) ‘All the garments are beautiful.’ (ii) ‘The garments are most beautiful.’ (iii) ‘All the garments are most beautiful.’ (iv) ‘Garments are most beautiful.’ In (41a), difficulties arise in assigning a pragmatically plausible comparison class, since the most plausible type of people against which the hunger of guests can be evaluated is the group of hosts. The comparison class would consist of two entities: the group of guests and the group of hosts. For the sake of illustration we assume an additional interpretation namely that the context entails the presence of diverse groups of people at the banquet different from guests and hosts (see interpretation (41a(ii))). The interpretation that the guests are the hungriest group of people among the groups of people in the world is, again, pragmatically difficult to evaluate (see interpretation (41a(iii))). In another reading, the clause-initial NP is universally quantified (see (41a(i))). i21mo21 mi55 sa55. (41a) di21vi21 g 44-su33 guest ART¼COLL–NOML belly hungry EXH (i) ‘All the guests are hungry.’ (ii) ‘The guests are the hungriest.’ (iii) ‘All the guests are the hungriest.’ (iv) ‘The guests are the hungriest.’

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

When the argument is marked as a definite plural, EXH imposes two readings. In the first reading the argument is quantified, in the second reading a superlative meaning is expressed (with an associated quantized comparison class consisting of items that people may wear, for example shoes, hats, scarves etc). A third reading is pragmatically implausible and would put the garments in contrast to all other objects in the world.

60 The Exhaustion Particles in the Yi group In (41b), the most credible interpretation is that the hunger of the person in question is evaluated against the hunger of a definite (and quantized) number of people. i21mo21 mi55 sa55. (41b) tsh 33 3P SG belly hungry EXH ‘S/he is the hungriest [among us].’ Consider the following two-place gradeable state ‘Anyuo loves Muga’. The implicit comparison class consists of all the admirers of Anyuo. This class is plausibly definite and therefore quantized. 33

sa55. EXH

4 SYNTHESIS In this concluding section, I intend to bring the various topics discussed in sections 1 to 3 together. A formalization of the various quantificational notions is indicated without being spelled out in detail. In section 4.1, I present a synoptic overview of the NP and VP types that may receive the operation of EXH. In section 4.2, a cross-categorial characterization of the semantic contribution of EXH is proposed in terms of universal quantification. I suggest that Barwise and Cooper’s concept of NPs being generalized quantifiers (cf. Barwise & Cooper 1981) can be extended to the interpretation of VPs. The completive and superlative meanings of EXH can be derived from this generalized interpretation of VPs.

4.1 The syntax and semantics of the input structures of EXH The analysis presented in section 3 reveals that the exhaustion particle has scope both over whole NP projections in sentence-initial position and V-bar constituents. The noun phrases attested above are of various kinds:
Person names [examples (15a + b)]
Personal pronouns [example (24a)]
Definite possessive constructions [examples (18a + b)]
Indefinite quantifier constructions [examples (21d + e)]
Definite article constructions [example (21a)]
Demonstrative constructions [example (21b)]
Determiner-numeral constructions [examples (17a + b) and (19)].

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

33 55 (42) a33 33 ta44 ka he33 name of woman COV:put name of man love ‘Muga loves Anyuo the most.’

Matthias Gerner 61

Regarding verb phrases, it is interesting to note that EXH occurs immediately after the predicate, but before other aspect and modality particles (in the above example only evidence for the exhaustion particle occurring before the perfect particle is given, cf. examples 19 or 32). Syntactically, without giving specifically syntactic arguments, it may therefore be assumed that EXH occupies the role of an adjunct expanding V-bar into V-bar. The major input structure to EXH is as follows:
[VP [V# (NP) V]]: Complement-Verb collocations [examples (23) and (19)].

4.2 The semantic contribution of EXH In this section we will interpret the semantic contribution of the exhaustion particle as a generalized (universal) quantifier of individuals, events and states. From figure 3, it follows that only quantized domains NP: Object denotation Singular VP: Event denotation Singular Quantized Bounded Homogenous VP: State denotation Singular Quantized Homogenous

Quantized

Homogenous

EXH * (-,") * *(#)

Example (24c) (27b) (16) (15a)

EXH (",-) (",") (",-) (",-)

Example (24a) (27a) (33) (19)

EXH # (-,") # #

Example

* (-,") #

(37a) (40a) (40a)

(",-) (",") (",-)

(21a) (40b) (40b)

# (-,") #

(38c) (23) (23)

(35) (30a) (22)

*Ungrammatical and #pragmatically odd but occasionally fully acceptable. (","): universal (i) object quantification, (ii) event/state quantification, (iii) (i) + (ii). (",-): universal object quantification (and no universal event/state quantification). (-,"): universal event/state quantification (and no universal object quantification).

Figure 3

Input structures for EXH.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Semantically, we have seen in section 3 that there are restrictions on the range of NP and V-bar constituents allowed to serve as input of EXH. The ground rule is that either NP or V-bar (where V-bar denotes events or states) must have the quantized reference property in order to yield a grammatical sentence. When only one of NP or V-bar has the quantized reference property, then the one that doesn’t satisfy QUA cannot be quantified. The following chart assembles a synoptic overview of the semantic interaction between the sentenceinitial NP and the V-bar constituent.

62 The Exhaustion Particles in the Yi group are fully configurable with EXH. In this section, we will therefore describe the semantic contribution of EXH to the sentence for quantized objects (section 4.2.1), quantized events (section 4.2.2) and states with a quantized comparison classes (section 4.2.3) only. It is important to keep in mind that EXH does not articulate the property of being quantized, but that its inputs are predicates of the quantized predicate type.

(43) ‘All students work hard.’ (44) ‘No student worked hard.’ In the set-theoretic interpretation language the above predicates are realized by the two sets student and work-hard. The truth of (43) and (44) are defined as follows (see also Keenan 1996: 43): (45) Sentence (43) is true iff student 4 work-hard (46) Sentence (44) is true iff student \ work-hard ¼ B It is suggested now that the operation of EXH on the sentenceinitial NP could be interpreted as a generalized quantifier as defined in (45). Because of space limitations, formal details won’t be spelled out here. 4.2.2 The completive as a universal quantifier It is now intuitive to extend Barwise & Cooper’s (1981) concept of generalized quantifier to the treatment of verb phrases. Let us recall that the exhaustion particle can express up to three distinct meanings: (i) universal quantification of the sentence-initial NP; (ii) the meaning of completive, i.e. a kind of universal quantifier of events; (iii) combination of (i) and (ii). These meanings are once more illustrated below (cf. example 27a). (47) tsho21 o44 s 21 i33 tsi44 ma33 dz 3P PL nut NUM:10 CL eat (i) ‘They all ate ten nuts.’ (ii) ‘They completely ate up ten nuts.’ (iii) ‘They all completely ate up ten nuts.’

33

sa55 o44. EXH DP

In analogy with the sketch given in section 4.2.1, we introduce two verbal predicates: eat-ten-nuts and they. The predicate eat-ten-nuts is

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

4.2.1 EXH as universal quantifier of NPs Since the seminal work of Barwise & Cooper (1981), NPs are interpreted as Generalized Quantifiers, which means that their noun determiners are interpreted as various relations that hold between the denotation of their NP and the denotation of the predicate with which they occur. For instance, in

Matthias Gerner 63

a label for all events under consideration in a given context c that are of the type ‘eat-ten-nuts’. The predicate they characterizes all events in the context c in which the referent of ‘they’ performs something (i.e. all events which admit ‘they’ as their agent):
eat-ten-nuts: the verbal predicate corresponding to the V-bar ‘eat ten nuts’
they: the verbal predicate defined by those events that admit ‘they’ as their agent.

(48) "e [eat-ten-nuts(e) / they(e)] The sentence ‘They ate up completely the nuts’ can be interpreted as true in a context c if and only if every event of type eat-ten-nuts is an event which is done by them (i.e. belongs to they). This interpretation is intuitively correct, especially from the perspective of EXH’s semantics. Compare the above sentence with the glosses ‘They ate up exhaustively ten nuts’, ‘They exhausted the ten-nuts-eating events’ and ‘Every ten-nuts-eating event was done by them’. The completive meaning of the exhaustion particle is interpreted as a kind of generalized quantifier on events. Again, we won’t present formal details in this paper. Also, we won’t describe the above case (iii) where EXH functions as a combined generalized quantifier of objects and events. This would require defining for each (noun or verb) phrase a kind of higher-order predicate in LEXH type able to combine EXH’s quantifier operation on objects and on events. 4.2.3 The superlative as a universal quantifier The next step is to extend this analysis to the operation of the exhaustion particle to states. The difference between states and events is that gradeable states rarely express overtly the reference type of their comparison class (whether quantized or homogenous) which is a matter of pragmatic inference, whereas events often encode their reference type membership overtly. Again, let us recall that the exhaustion particle always expresses three or four distinct meanings, when associated with states: (i) universal quantification of the sentence-initial NP; (ii) the meaning of ‘most’, if the comparison class has the quantized reference type (‘most’ is a kind of universal quantifier over states or, more precisely, over comparison

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The intuition behind this construal is to simplify the interpretation of the event quantifier ‘completely’ in a way that mirrors the operation of Generalized Quantifiers in the nominal realm. We can view EXH in (47) paraphrased by ‘completely’ as a universal quantifier of events under consideration in a given context c.

64 The Exhaustion Particles in the Yi group classes in the sense defined below); (iii) combination of (i) and (ii); (iv) when no clear pragmatic information about the comparison class can be inferred, then the default is to interpret the sentence with respect to all-inclusive comparison class which has homogenous reference. However the default interpretation results sometimes in pragmatically odd meanings. Consider again the example of section 3.4.2. (49) i33ti44 a33d 44 gu44 a55 sa55. garment DEM:DIST CL beautiful EXH (i) ‘That garment is the most beautiful.’ (ii) ‘The garment is the most beautiful in the world.’


beautiful: the stative-verbal predicate corresponding to the V-bar ‘beautiful’
that-garment: the nominal predicate referring to the physical object ‘that garment’. As the interpretation of the superlative relies on the associated concept of comparison class, we may not formalize it as a universal quantifier on states in an analogue way to the completive. We will rather analyse the superlative as a universal quantifier of comparison classes. The intuition is to paraphrase the effect of ‘most’ as the requirement for that-garment to belong to every possible non-empty way of delineating comparison classes for c and for beautiful. Let U be a function variable of type e, t>, <e, t symbolizing possible comparison classes for beautiful in context c. In section 2.3, we defined the predicate constants POS(P), NEG(P) and UNDEF(P). They should intervene in the formalization of the superlative. However, for the purpose of not overloading the outlook of the formulae, we will only consider the simplified situation of where all

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Because of practical limitations, we only concentrate on some formal considerations for EXH’s meaning of superlative, i.e. the above case (ii) which is illustrated by (49(i)). We suppose that the superlative state (49(i)) holds in a fixed context c. We aim for a formalization of the superlative via comparison classes which were defined in section 2.3 as formal objects of LEXH type. Let us therefore introduce the stative-verb predicate beautiful and the object predicate that-garment. Following the interpretation (49(i)), it is assumed that the predicate beautiful is quantized (i.e. associated with a quantized object-comparison class COMP(beautiful)). It can be understood as a label for all states under consideration in a given context c that match the type ‘beautiful’. The nominal predicate that-garment portrays one particular garment picked out in context c as ‘that garment’:

Matthias Gerner 65

members of the comparison class are ‘beautiful’ (i.e. the situation of COMP(P) ¼ POS(P)). Furthermore, let x and y be two object variables. The superlative may be characterized as follows:12 (50) "U [dx U(beautiful)(x) U(beautiful)(y))]

/

"y

(that-garment

(y)

/

This paper introduces a rare pattern of universal quantification that exists in the Yi languages (4 Tibeto-Burman languages) spoken in China. The data presented are from Liangshan Nuosu, the main language of the group, and invite reflections on the commonalities between universal noun quantification, completive and superlative. I have shown that these three notions, although modifying different entities (i.e. objects, events and comparison classes), take a similar analytic view of their respective domains. They distinguish between singular, quantized, bounded and homogeneous domain types for which they demonstrate different compatibilities. The only domain type fully configurable with these notions is the quantized domain type. The semantic contribution of the exhaustion particle is to express universal quantification whenever possible (see figure 4). Objectdenotation

Eventdenotation

Statedenotation

Singular Quantized

Individual OD Punctual ED Ungradeable SD Quantized OD Quantized ED SD with quantized comparison class Bounded — Bounded ED — Homogeneous Homogeneous Homogeneous SD with OD ED homogeneous comparison class

Figure 4 12

Contribution of EXH * " * #

Types of processed entities.

This formalization of the superlative stems from constructive discussions with Paul Portner.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

That (50) is an adequate rendering of ‘That garment is the most beautiful’ can be appreciated through the intermediate gloss ‘That garment is beautiful compared to every set of items evaluated in some context c as beautiful’. It is obvious now that the superlative meaning of EXH will be understood as a sort of generalized (universal) quantifier on comparison classes. Again, we will skip the formalization of the case where EXH is a combined operator of superlative and universal quantifier of sentenceinitial arguments. 5 EPILOGUE

66 The Exhaustion Particles in the Yi group Acknowledgements I gratefully acknowledge the very helpful comments made by Paul Portner, associate editor of Journal of Semantics. Warm regards also go to Noel Johnston for remarks on earlier versions of this paper. The Yi data in this paper embody my own field research conducted in 1999–2006. I wish to express special gratitude to native Yi speakers serving as informants.

APPENDIX A: THE EXHAUSTION PARTICLES OF OTHER YI LANGUAGES I have also surveyed the exhaustion particles of 10 other Yi languages of Southwest China. The properties of these other exhaustion particles resemble very much those of Liangshan Nuosu. A small difference exists for the superlative, which several Yi languages replace by a process- superlative (‘become most/extremely’). Below are listed contrastive data from Shizong Kopho (spoken in Yunnan Province). Shizong Kopho’s exhaustion particle g 33 functions in a similar way to Liangshan Nuosu’s sa55 with regard to activities, accomplishments, achievements and non-gradeable states, but differs from it with gradeable states. The exhaustion particle g 33 may dynamize adjectives (e.g. ‘bright’) or two-place verbal predicates (e.g. ‘know’ in the sense of ‘learn’) as long as these adjectives or verbs include in their lexical coding the possibility of a dynamic initialization phase, but g 33 is forbidden in gradeable states that do not allow dynamization. For the purpose of illustration, I will first briefly present examples for the exhaustion particle g 33 in events, before discussing in more detail its involvement in states. (51) Homogenous event denotations13 s 21 affair d 55 CIRC:after

k 55 DEM:PROX gu21dou21sa13 emperor of heaven

die33 CL i55 then

m 13 do

g-33 EXH

13 Sentence (51) was selected from an unpublished mythological flood story which attempts to retrace the origin of the Han, Tibetan and Yi (Kopho), the three predominant people in Yunnan province and environs. The emperor of heaven announces to the pre-flood generation through a messenger that after the destruction of their agriculture, a universal flood will immerse the whole world. Sentence (54) is part of this announcement.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

MATTHIAS GERNER Department of Chinese, Translation & Linguistics City University of Hong Kong 83 Tat Chee Avenue, Kowloon, Hong Kong e-mail: [email protected]

Matthias Gerner 67

i13mu33 k 55 po33 o33 a13 ocean NUM:9 CL REL water th 21 th 13 tie21. put SET OUT FUT ‘After finishing this [¼ the destruction of their agriculture], the emperor of heaven will pour out nine oceans [on the earth].’ (52) Quantized event denotations

(53) Bounded event denotations e33 o21 o53fu33 g-33 die13. gold 1P SG obtain EXH DP ‘I have obtained all the gold.’ In ungradeable states, Kopho’s exhaustion particle g 33 closely resembles Nuosu’s sa55. This means that it cannot be employed in gradeable states when the argument(s) are singular. When at least one argument is plural, then it quantifies universally this plural argument. For example: (54) Ungradeable States i33b 33 na55v 55 ni33 3P PL there sit ‘They are all sitting there.’ na55v 55 ni33 * i33 3P SG there sit ‘S/he is all sitting there.’

tu55 g-33. DUR EXH tu55 g-33. DUR EXH

The situation diverges when the state is gradeable. Kopho’s exhaustion particle dynamizes the gradeable state whenever this is possible and injects the meaning ‘become most/extremely’. Let us first consider a Kopho example.14 14

Sentence (55) is taken from an unpublished folk story which is a Kopho counterpart of the European fairy-tale ‘The frog-king’. In the Kopho story, the frog married the daughter of a local Kopho but kept his frog appearance and only took his frog skin off on market-days to change into a handsome young man. On one of these market-days, his wife’s sister found the skin and destroyed it in the fire. From this day on, the frog was ‘condemned’ to continue his life as a man. When his frog skin was burning, half the sky was lit up.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

i33 u21mu33 ta21 gu33 the21 g-33. 3P SG road NUM:1 CL run EXH ‘S/he ran a complete road section.’

68 The Exhaustion Particles in the Yi group (55) Gradeable States nu33 m 21 le21 ta21 pi55 t h 33 burn and sky QUAN:all NUM:1 MW: half 33 33 33 mu bu g- . RES:so that bright EXH ‘[The fire] is burning so much that half the sky is lit up extremely brightly.’

(56) Gradeable States i33 bo33li21 s 55 g-33. 3P SG glass resemble EXH ‘S/he has now become as fragile as glass.’ Now, not every gradeable state is dynamizable. In (57a), the state ‘this flower is red’ cannot be dynamized by the exhaustion particle g 33, i.e. (57a) is rejected by native speakers. This does not mean that the adjective ni33 ‘red’ is undynamizable per se. A field of tulips can redden by the day, especially when more and more tulips start the blossoming process, but a single tulip, once it opens its petals, cannot become more red, at least in the Kopho understanding of nature. Witness: (57a) *vi55li33 k 55 p 13 ni33 g-33. flower DEM:PROX CL red EXH ‘This flower has become extremely red.’ On the other hand, ni33 ‘red’ denotes a gradeable adjective in the following comparison structure. p 13 vi55li33 n 55 p 13 (57b) vi55li33 k 55 flower DEM:PROX CL flower DEM:DIST CL ma21bu21 o13. ni33 red more ‘This flower is redder than that one.’ The ability to allow dynamization depends on the argument(s) which occur in the state. For example it is possible to dynamize ‘the tree is big’ (‘the tree has become big’), but it is not possible to dynamize

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

The next example involves the exhaustion particle with the twoplace predicate ‘resemble’. It describes, in a hyperbolic way, the extreme fragility of a person.

Matthias Gerner 69

‘the mountain is big’, since mountains are not expected to grow. This is attested in Kopho by the following pair of examples: h 33 ne33 e mo13 g-33. (58a) *b 13 mountain DEM:DIST CL high EXH ‘That mountain has become extremely high.’

(58b)

i13 ne33 d 33 mo13 g-33. tree DEM:PROX CL high EXH ‘That tree has grown extremely high.’

Name of language

Particle

All

Completive

Superlative

Process-superlative

Liangshan Nuosu Weining Neasu Longlin Ngopho Luoping Nase Shizong Kopho Mile Axi Mile Azhee Gejiu Nesu Weishan Lalo Yongren Lolo Wuding Aluphu

sa55 tsha33 k 33 g 21 g 33 k 22 ku33 k 55 a33 g]33 gu21

+ + + + + + + + + + +

+ + + + + + + + + + +

+ +   

  + + +

 

+ +

+ 

 +

Figure 5

Synoptic table of exhaustion particles in 11 Yi languages.

APPENDIX B: THE SOCIO-LINGUISTIC SETTINGS OF THE YI LANGUAGES By the Yi group I refer to those Burmese-Lolo languages (inside the Tibeto-Burman language family) which belong to the Chinese Yi ( ) nationality. In China there are 7 million ethnic Yi (1990 census) distributed over Yunnan, Sichuan, Guizhou and Guangxi provinces.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

After presenting the analysis of two exhaustion particles (i.e. Liangshan Nuosu’s sa55 and Shizong Kopho’s g 33), I am now able to draw a synopsis table of the exhaustion particles in all the Yi languages surveyed. It appears that the sound structure of these particles concentrates around two cognate roots: One is sa55 (with minor a33) and the other k 33. The data available for Mile Axi’s exhaustion particle k 22 did not allow me to achieve a definitive understanding of its operation in gradeable states. The same is true for Weishan Lalo’s a33 (see Bjo¨rverud 1998: 82–83, where I could not make any decision on the basis of the data available in her grammar) (see figure 5).

70 The Exhaustion Particles in the Yi group The 11 Yi languages surveyed in this paper geographically represent all branches of the Yi group. If unintelligibility were a criterion to separate languages, then I would estimate the total number of Yi languages to ca. 130 (see figure 6). Language

Location

Liangshan Nuosu Weining Neasu

Number of speakers

Nuosu total: Bradley (1997): NL/Nosu 2.2 M CS: Northern dialect of Yi 300,000–600,000 Bradley (1997): NL/Nasu CS: Eastern dialect of Yi

10,000–40,000

Bradley (1997): NL/Nasu CS: Eastern dialect of Yi

100,000–300,000 Bradley (1997): NL/Nasu CS: Central dialect of Yi 20,000–40,000 Bradley (1997): NL/Nasu CS: Central dialect of Yi ca. 82,000 ca. 60,000 ca. 370,000

565,000 50,000–150,000 ca. 41,000

Bradley (1997): CL CS: South-East dialect of Yi Bradley (1997): CL CS: South-East dialect of Yi Bradley (1997): NL/Nisu CS: Southern dialect of Yi Bradley (1997): CL/Lalo CS: Western dialect of Yi Bradley (1997): CL/Lalo CS: Western dialect of Yi Bradley (1997): NL/Nasu CS: Eastern dialect of Yi

/ separates elements of different hierarchical layers. , and ; separate elements of equal rank. . CS means Chinese sources. NL means Northern Loloish. CL means Central Loloish. SL means Southern Loloish.

Figure 6

Profile of the Yi languages surveyed.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

China/Sichuan/ Liangshan China/Guizhou/ Bijie/Weining, Hezhang; China/ Guizhou/Liupanshui/ Shuicheng Longlin China/Guangxi/ Ngopho Baise/Longlin; China/Guizhou/ Liupanshui/Pan Luoping China/Yunnan/ Nase Qujing/Luoping Shizong China/Yunnan/ Kopho Qujing/Shizong; China/Yunnan/ Honghe/Mile Mile Axi China/Yunnan/ Honghe/Mile Mile Azhee China/Yunnan/ Honghe/Mile Gejiu Nesu China/Yunnan/ Honghe/Gejiu, Jinping, Jianshui, Lu¨chun Weishan China/Yunnan/Dali/ Lalo Weishan, Nanjian Yongren China/Yunnan/ Lolo Chuxiong/Yongren, Dayao Wuding China/Yunnan/ Aluphu Chuxiong/Wuding

Genetic affiliation

Matthias Gerner 71 LIST OF ABBREVIATIONS 1P PL

CL COLL

Classifier Collective

MW MW:bottle

COV

Coverb

MW:load

COV:arrange

Coverb with verbal origin if any Coverb with verbal origin if any Demonstrative Demonstrative: distal distance to speaker Demonstrative: proximal distance to speaker

1P SG 3P PL 3P SG 3P SG+POSS

COV:put DEM DEM:DIST DEM:PROX

EXH

Dynamic perfect particle Exhaustion particle

GET

GET versatile

LOC

Location particle

LOC:on

NOML

Location particle with gloss Location particle with gloss Measure word Measure word with its value Measure word with its value Nominalisation particle

NUM

Numeral

NUM:9 PASS

Numeral with its value Passive particle

STP

Stative perfect particle

LOC:outside

REFERENCES Bach, E. (1986) ‘The algebra of events’. Linguistics and Philosophy 9:5–16. Bach, E., Jelinek, E., Kratzer, A. & Partee, B. (eds) (1995) Quantification in natural languages. Volume I and II. Kluwer Academic Publishers. Dordrecht. Barwise, J. & Cooper, R. (1981) ‘Generalized Quantifiers and Natural Language’. Linguistics and Philosophy 4:159–219. Bisang, W. (1999) ‘Classifiers in East and Southeast Asian languages: Counting and beyond’. In J. Gvozdanovic (ed), Numeral Types and Changes Worldwide. Mouton de Gruyter. Berlin. 113–185. Bradley, D. (1997) ‘Tibeto-Burman languages and classification’. In David

Bradley (ed.), Papers in Southeast Asian Linguistics No.14: TibetoBurman Languages of the Himalayas. Pacific Linguistics A-86. Australian National University. Canberra. 1–72. Bunt, H. (1979) ‘Ensembles and the formal semantic properties of mass terms’. In F. Pelletier (ed), Mass Terms: Some Philosophical Problems. Reidel. Dordrecht. 279–294. Bunt, H. (1985) Mass Terms and ModelTheoretic Semantics. Cambridge University Press. Cambridge. Chen, K. & Da, W. (1998) Yiyu Yufa [Yi grammar]. Zhongyang Minzu Daxue Chubanshe [Central University of Nationalities Press]. Beijing.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

DP

ART

First person plural pronoun First person singular pronoun Third person plural pronoun Third person singular pronoun Third person singular possessive pronoun Article

72 The Exhaustion Particles in the Yi group Massentermen, Pluraltermen und Aspektklassen. Fink. Munich. Krifka, M. (1992) ‘Nominal reference, temporal constitution and thematic relations’. In Sag et al. (eds), Lexical Matters. CSLI Publications. Stanford. 29–53. Leech, G.N. (1971) Meaning and the English Verb. Longman. Harlow. Li, M. and Ming, M. (1981) Liangshan Yiyu huihua liubai ju [Six hundred sentences of dialog in Liangshan Yi language]. Sichuan Minzu Chubanshe [Sichuan Nationality Press]. Chengdu. Link, G. (1983) ‘The logical analysis of plural and mass terms: A latticetheoretical approach’. In R. Ba¨uerle, C. Schwarze & A. von Stechow (eds), Meaning, Use, and Interpretation of Language, Mouton. Berlin. 302– 323. Moens, M. and Steedman, M. (1988) ‘Temporal ontology and temporal reference’. Computational Linguistics 14(2):15–28. Naumann, R. (2001) ‘Aspects of changes: A dynamic event semantics’. Journal of Semantics 18:27–81. Quine, W. (1960) Word and Object. The MIT Press. Cambridge, MA. Reiter, R. (1980) ‘A logic for default reasoning’. Artificial Intelligence 13: 81–132. Vendler, Z. (1967) Linguistics in Philosophy. Cornell University Press. Ithaca. Verkuyl, H. J. (1993) A Theory of Aspectuality. Cambridge University Press. Cambridge. Von Stechow, A. (1984) ‘Comparing semantic theories of comparison’. Journal of Semantics 3:1–77. First version received: 4.11.05 Second version received: 24.5.06 Accepted: 30.6.06

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Cheng, Ch.-Y. (1973) ‘Comments on Moravcsik’s paper’. In K. Hintikka, J. Moravcsik & P. Suppes (eds), Approaches to Natural Language, Reidel. Dordrecht. 286–288. Comrie, B. (1976) Aspect. Cambridge University Press. Cambridge. Cresswell, M. J. (1976) ‘The semantics of degree’. In B. Partee (ed), Montague Grammar. Academic Press. New York, NY. 261–292. Croft, W., (1994) ‘Semantic universals in classifier systems’. Word 45:145–171. Dowty, D. (1991) ‘Thematic proto-roles and argument selection’. Language 67:547–619. Eschenbach, C. (1993) ‘Semantics of number’. Journal of Semantics 10:1–31. Gawron, J. M. (1995) ‘Comparatives, superlative and resolution’. Linguistics and Philosophy 18:333–380. Gerner, M. (2002) Predicate Compounding in the Yi Group: The Continuum of Grammaticalization. Studia Typologica Monographien 3, Sprachtypologie und Universalienforschung. Akademie Verlag. Berlin. Gillon, B. (1992) ‘Towards a common semantics for English count and mass nouns’. Linguistics and Philosophy 15:597–639. Hellan, L.(1981) Towards an Integrated Analysis of Comparatives. Narr. Tu¨bingen. Jelinek, E. (1995) ‘Quantification in Straits Salish’. In E. Bach, E. Jelinek, A. Kratzer & B. Partee (eds), Quantification in Natural Languages. Kluwer Academic Publishers. Dordrecht. 487–540. Keenan, E. (1996) ‘The semantics of determiners’. In S. Lappin (ed), The Handbook of Contemporary Semantic Theory. Blackwell. Oxford. 41–63. Klein, E. (1980) ‘A semantics for positive and comparative adjectives’. Linguistics and Philosophy 4:1–45. Krifka, M. (1989) Nominalreferenz und Zeitkonstitution. Zur Semantik von

Journal of Semantics 24: 73–90 doi:10.1093/jos/ffl009 Advance Access publication January 12, 2007

What Autism Can Reveal About Every . . . not Sentences IRA A. NOVECK, RAPHAE¨LE GUELMINGER, NICOLAS GEORGIEFF, AND NELLY LABRUYERE Universite´ de Lyon and CNRS, France

The sentence Every horse did not jump over the fence can be interpreted with the negation taking scope over the quantifier (i.e. not every horse jumped) or with the quantifier Every taking scope over the negation (ultimately providing the reading no horse jumped). Beginning with Musolino, Crain and Thornton (2000), much work has shown that while adults typically adopt a Not every reading in ‘2-of-3’ contexts (e.g. where 2-of-3 horses jump over a fence), children do not and often produce None readings instead. In line with suggestions from Musolino and Lidz (2003, 2006), we propose that this developmental effect relies to a great extent on pragmatic capacities. In the present work, we aim to replicate Musolino et al.’s (2000) results with 4year-olds and adults while including verbally competent autistic participants. Syntactic skills among verbally competent autistic participants are assumed to be unimpaired while their pragmatic deficiencies have been well documented. Our results show an adult preference for the Not every reading in 2-of-3 contexts and equivocality among children and autistic participants. This is in line with the expectation that syntax makes the two readings equally available and that adults, unlike young children and autistic participants, are efficient at exploiting the context in order to come up with a single consistent reading.

1 INTRODUCTION Negations are notorious for creating ambiguous readings. For example, the sentence in (1) can be understood in two different ways. (1) Every horse did not jump over the fence. When the negation takes scope over the quantifier (Not every), one arrives at the interpretation in (2) and when the quantifier takes scope

Ó The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Abstract

74 What Autism Can Reveal About Every . . . not Sentences

1

The reading in (3) can be understood more readily as None of the horses jumped over the fence and will be referred to as such (the None reading). 2 The ‘first parse’ allocates scope because the quantificational expression (Every horse in (1)) ccommands the negation (did not). In their earlier work, the authors emphasized how children are susceptible to an ‘‘isomorphism’’ effect in which the surface syntax (Every. . .not) determines the preference for the None reading.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

over the negation (Every . . . not or None)1 one arrives at the interpretation in (3). (2) Not every horse jumped over the fence. (3) All horses are such that they did not jump over the fence. The present work was initially inspired by Musolino et al. (2000, Experiment 2) who reported that adults are more likely than children to accept the Not every interpretation, as in (2), after being shown that two of three horses jumped over a fence. Meanwhile, children demonstrate ambiguity between the two readings (Musolino & Lidz 2006, Experiment 1) or a preference for a None reading, as in (3) (Musolino et al. 2000). After several follow-up investigations, Musolino & Lidz (2003, 2006; Lidz & Musolino, 2002) currently hypothesize that Every . . . not sentences are syntactically ambiguous but that either performance factors play a role among children (Musolino & Lidz, 2006) or that scope relations favour a None reading among children (Musolino et al. 2000).2 That is, children make their judgements by choosing (at random) one of the two possible readings or because there are reasons that encourage a None reading (based on contextual factors or parsing mechanisms). Critical to these claims is the notion that an adult, unlike a child, has access to each reading as well as its potential import and that preferring one (over the other) involves some extra attentional abilities at the point where the utterance is integrated with context. Musolino & Lidz (2003, 2006) invoke pragmatics in three different ways in order to explain why adults obtain the interpretation in (2) more often than children, even though the parser makes both readings available. One suggestion is the Charity account, or ‘Principle of Charity’ (Musolino & Lidz, 2003), according to which adults seek out a true interpretation; this makes the true Not every reading more attractive than a false None reading in their tasks. A second suggestion they consider (Musolino & Lidz 2006: 819–822) is what we will call the Metalinguistic Scalar account, according to which the two interpretations arise via the syntax; through a scalar implicature (which we describe below), the weaker interpretation (Not every) leads a listener to the reject the stronger one (None), even if the latter is compatible with the context. Finally, a third suggestion (Musolino &

Ira A. Noveck et al. 75

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Lidz, 2006: 842) is that, in minimally elaborated circumstances, adults favour the Not every reading because a speaker who means to convey the None proposition could have said precisely that by using the more economical form of words, e.g. by saying None of the horses jumped over the fence. Assuming that (other things being equal) speakers do not put their audience to gratuitous effort, the fact that the speaker did not use a more economical form of words suggests she meant to convey the Not every proposition. In order to account for why adults, and not children, make the Not every interpretation, one would have to assume that it is the result of acquiring the Manner maxim over the course of language development. In this case, the Not every interpretation that emerges from Every . . . not sentences may well be viewed as automatic (or as a default). We will refer to this as the Acquired Manner account, in line with Grice’s Manner maxim. To round out the picture for the pragmatics of Every . . . not sentences, below we present our own Processing account, which is drawn from Sperber & Wilson’s (1995) Relevance Theory. Relevance theory claims that (a) the speaker’s meaning is inferred from the linguistic meaning of the words and expressions used taken together with the specific context at hand and that (b) effort factors play a role in utterance interpretation. We propose that the Not every interpretation of (1) in the presented experimental context (where 2of-3 horses represent a subgroup) is more attractive than a None reading to an adult mostly because the 2-of-3 context is more readily consistent with a Not every reading. When a participant gets to the word not in Every horse did not jump over the fence, a Not every interpretation with respect to the 2-of-3 context allows a sophisticated hearer to even anticipate the rest of the sentence (. . . jump over the fence). With a None interpretation, a listener would be hard pressed to anticipate what it is that all of the horses are not doing. On the other hand, children—who are less efficient than adults at processing utterances and integrating them in context—generate the two readings as best they can and essentially choose randomly between them. Note that this account does not say that adults seek out a true interpretation per se like the Charity account, but that an interpretation that requires less effortful procedures will be more readily adopted. As Musolino & Lidz (2006) themselves point out, the findings and suggestions from the papers on Every . . . not sentences resonate with the work showing that young children are less likely than adults to produce scalar implicatures (see Noveck 2001; Papafragou & Musolino 2003; Chierchia et al. 2004). According to standard neoGricean accounts (Horn 1984; Levinson 2000), implicatures arise when

76 What Autism Can Reveal About Every . . . not Sentences

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

a speaker’s use of a relatively weak term allows the addressee to understand that the speaker has chosen not to articulate a stronger, more informative one from the same scale. Consider the weak quantifier some. From a semantic point of view, some is compatible with all, but quite often in conversation some will be understood as meaning some but not all because the speaker’s choice of words is taken to mean that a stronger term from the same scale (all) is being avoided. Adults are more likely than children to transform some into some but not all and arguably because they are more apt at considering the stronger alternative in the first place or at considering the speaker’s mental state in context (for summaries, see Noveck 2004; Noveck & Sperber in press). That said, it is important to note that the two phenomena are distinct in two critical ways. First, a sentence like the one in (1) is genuinely ambiguous whereas scalar terms are not. That is, a some but not all enrichment is extra-linguistic. At the level of the linguistic code, some’s meaning remains unchanged and could suffice for providing responses on scalar implicature tasks even without the enrichment. This distinction—between the linguistic ambiguity for Every . . . not sentences and the extra-linguistic enrichment for scalars—explains why one finds different sorts of developmental patterns across these paradigms. Whereas adults ‘can be turned into’ children and vice versa in tests with the syntactically ambiguous structures (Musolino & Lidz 2003) children consistently remain ‘less adult’ than adults in the scalar tasks (Noveck 2001; Papafragou & Musolino 2003; Pouscoulous et al. in press). The developmental patterns in the scalar paradigm indicate that the pragmatic enrichment there is essentially optional; when it arrives it is part of a more involved pragmatic reading. Second, a Not every reading—which is generally preferred among adults in the Every . . . not studies—is actually weaker than the None reading, whereas the work on scalar implicatures generally shows that older participants aim for more informative readings (some but not all is more informative than some). It appears then that there is no simple pragmatic rule about adults and the way they abide by informative strength. In light of the above comparison between Every . . . not sentences and weak scalar utterances (as well as the data they produce), we are doubtful about the Metalinguistic Scalar account. This account assumes that adult participants hear (1), generate interpretations (2) and (3), and order them with respect to informativeness before opting for the weaker of the two (which leads to the rejection of the stronger of the two); moreover, the background information is considered immaterial (the adoption of the weaker interpretation arises even if the stronger is

Ira A. Noveck et al. 77

3

Note that rejecting a reading differs from ordering readings in terms of preference. We do not use the expression Asperger’s because the French system does not use the same diagnostic criteria as found elsewhere. However, as will be seen by their responses to control problems, the autistic participants are verbal, high functioning teenagers. 4

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

contextually plausible). There are three reasons why we consider it unlikely that adults would opt for the weaker reading as a first step. First, it is important to point out that—unlike in the case of scalar implicatures where a weak spoken utterance is privileged because the stronger option was not explicitly said—both readings generated by the Every . . . not sentence are justified by what was said. This leads to the observation that the listener’s adoption of the weaker of the two readings in the Metalinguistic Scalar account of (1) appears arbitrary. Second, an account that relies on the resulting rejection of the stronger of the two encoded meanings would amount to an inherent contradiction between what is said and one of its direct interpretations.3 Third, the idea that a listener would reject a strong interpretation when it is provided seems unlikely. Many have shown (Hirst & Weil 1982; Moore et al. 1990; Noveck et al. 1995, Chierchia et al. 2004) that children, at least, prefer stronger utterances over weak ones in the absence or presence of validating context. For example, Noveck et al. (1995) showed how five-year-olds prefer to follow the advice of the speaker who uses has to over another who uses might when there is no way to determine which is correct (compare The peanut has to be under the cup versus the peanut might be under the box) and Chierchia and colleagues have shown that four-year-olds prefer stronger descriptions over weak ones when both options are valid (for example they choose all over some to describe a scenario that would best be described with all). In short, it is not clear what would motivate a listener to adopt the weaker of the two readings as a first step and, furthermore, empirical tests show a general preference for stronger readings. Thus, there is little reason to assume that the Metalinguistic account can work to the extent that it is proposed as a linguistically driven mechanism that automatically adopts a Not every reading when the stronger None is available as well. Regardless of the account that is ultimately supported in this growing literature, our aim here is modest, which is to simply validate the role played by pragmatics in interpreting sentences and to demonstrate that any explanation for the developmental effect cannot rely on syntax alone. We thus investigate the same sentences as Musolino et al. (2000) and Musolino & Lidz (2006) and, critically, we add a third group that is known for having pragmatic deficiencies: verbally competent autistic participants.4 We include such participants because their

78 What Autism Can Reveal About Every . . . not Sentences

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

syntactic skills are assumed to be relatively unimpaired (Tager-Flusberg 1981), while their pragmatic deficiencies have been well documented (e.g. see Dennis et al. 2001, Happe´ 1995). Not only are they challenged when confronted with metaphors, idioms, lies and jokes, the literature also shows that verbally competent autistic participants are not as efficient as typically developing controls at using context for more low level tasks like the disambiguation of sentences containing homographs (Happe´ 1997; Joliffe & Baron-Cohen, 1999). If utterance interpretation of Every . . . not sentences requires even a modicum of pragmatic processing, it follows from the previous research that sentences like (1) ought to be difficult for autistic participants, just as they are for young children. Another issue that we address (indirectly) concerns methodology. In their first investigation, Musolino et al. (2000) reported that the None reading is preferred among children, even though the two readings are made available by the syntax. Their more recent work (Musolino & Lidz, 2006) went on to show that features of their (Truth Value Judgement or TVJ) task sway responses among children and that (at least part of ) the children’s preference for None readings in the Musolino et al. (2000) study is due to background information provided by the task. When a given negative context indicates that first The horses ruled out jumping over a barn before two of three jumped over a fence, five year olds’ responses indicate that they prefer a None interpretation for the test sentence Every horse did not jump over the fence (i.e. over 80% say ‘No’ with respect to the test sentence, arguably because it is not the case that None jumped, two did). In contrast, when given affirmative prefacing context that indicates that first Every horse jumped over a log before two of three jumped over a fence, five year olds’ responses indicate that they are more likely to prefer a Not every interpretation for Every horse did not jump over the fence (only 40% say ‘No’ with respect to the test sentence this time; 60% indicate ‘Yes’ and arguably because they agree that not all of the horses jumped over the fence, one did not). It appears then that the children are sensitive to contrasting information. When the horses start out by failing, the children largely reject the notion that says the horses failed again (making them likely to adopt the None reading, which they then reject as false); when the horses start out by succeeding as a group, the children are more likely to detect that this time not all the horses succeeded (making them more likely here to adopt the Not every reading, which they accept as true). Adults, on the other hand, show a consistent preference for Not every readings regardless of the background information provided.

Ira A. Noveck et al. 79

The present experiment, using more standard methodology, does not focus participants’ attention in any way by presenting prior information. This might well bring about less biased results and potentially reveal the ambiguity among children and autistic participants that is assumed to be inherent in sentences like (1). It is expected that adults will continue to manifest a preference for a Not every reading. 2 EXPERIMENT

(4) All the children are not in the pool. In order to address this goal, we present stories in which ultimately 2-of-3 protagonists are shown doing something or in which 2-of-3 objects share a feature. For example, if a story were to concern the sentence above, the critical test sentence would concern the case where two of three children are in the pool. If the None reading is adopted, then participants ought to respond negatively but if the Not every reading is adopted, they ought to respond positively. Given that the sentence is ambiguous, one ought to expect children and autistic participants to prefer either (a) the initial parse (the None reading), or else; (b) to show evidence of being equivocal about the two interpretations. In any case, one should expect typical adults to prefer the Not every reading, which leads to a true response in this context. Aside from items that measure comprehension, we also introduce two controls. To verify that the sentence itself is not the source of difficulty in this task, we present it in a context where both the None reading and the Not every reading yield the same (false) conclusion. This can be done for the sentence in (4) by providing a second context, one in which all (3-of-3) children are in the pool. One can safely conclude that the sentence itself is not problematic if participants are consistent in their treatment of (4) under such conditions. A second control is to present the same sentence while removing negations, whose presence may be a source of difficulty (see Prado & Noveck 2006). Evaluating a sentence such as All the children are in the pool successfully and in the two conditions mentioned earlier (where 2-of-3 v. where 3-of-3 are in the pool) would demonstrate that both children and verbally competent autistic participants can readily evaluate the quantifying aspect of these sentences. With respect to affirmative

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

We present one experiment that investigates the way three different groups, 4-year-olds, autistic participants as well as adults, interpret the critical experimental item, as exemplified by (4):

80 What Autism Can Reveal About Every . . . not Sentences Table 1 Truth value of a story’s final slide as a function of the quantified statements (QN and QP) and context (2-of-3 and 3-of-3) Presentation condition Statement

3-of-3

?

F

F

T

Notes. The ? indicates that this is the main test statement and that the truth value depends on the participant’s reading. It would be true with a Not > every reading (Not all the children are in the pool) and false with an Every > not reading (None of the children are in the pool).

sentences like All the children are in the pool, participants ought to say false when two of three children are in the pool and to say true when all three are. These controls are not expected to cause difficulties for any of the participants and also serve to obscure the purpose of the task. Table 1 summarizes the task. Again, our main focus is to verify Musolino and colleagues’ developmental effect with respect to sentences such as (1) in 2-of-3 contexts with our own materials and to determine the extent to which verbally competent autistic participants resemble young children.

2.1 Method 2.1.1 Participants Nineteen 4 year olds (average age: 4 years and 7 months) from a public school in Lyon, 20 adults recruited from the Universite´ de Lyon, and 15 verbally competent autistic participants from a Lyonnais clinic (ITTAC) participated. The characteristics of the verbally competent autistic participants are summarized in Table 2. All the participants were French native speakers.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

QN: All the children are not in the pool. QP: All the children are in the pool.

2-of-3

Ira A. Noveck et al. 81 Table 2 Characteristics of the verbally competent autistic participants Number of participants

Mean CA

Mean MA

IQ total

IQ verbal

Per cent passing Sally Ann task

15 (13 male; 2 female)

16;3

8;9

53.94

57.87

0.67

Notes. CA, chronological age; MA, mental age. Mental age was calculated from Echelle d’Intelligence pour Enfant de Wechsler, 3rd edition.

2.1.3 Test sentences In one condition (the 2-of-3 version) the final frame shows that two of the three people or objects make up a subgroup. For example, one story describes three friends who play together. The final frame in this condition concludes with two in the pool and one reading a book (see Table 1 and the Appendix). After each story, the experimenter presents a quantified statement to be evaluated, to which participants are required to say True or False. For example, for the Pool story in the Appendix, where the children are playing by a pool and then two of them decide to jump in, the critical experimental statement is (4), which we present in French, as (5). (5)

Tous les enfants ne sont pas dans la piscine.

A Not every reading of negation (Not all children are in the pool) in this 2-of-3 context ought to prompt a true response and a None reading (None of the children are in the pool) a false response. We now turn to the alternative context in which quantified statements were presented. Stories were prepared so that they could appear in an alternative version where all three protagonists (or objects) follow through on something (the 3-of-3 condition). Thus, in the conclusion of the example above, all three children end up in the pool. In this case, the statement in (4) is false regardless of one’s interpretation. Note that the syntax is obviously the same; only the context has changed. This control aims to verify that the sentence itself is not a source of difficulty. If the sentence itself is hard to process then it would be

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

2.1.2 Materials Ten stories (in French) were prepared that were accompanied by drawings in colour by a professional artist following our specifications. Each of the stories concerned three people or three things and each was designed to have four standard frames completed by one of two possible outcomes. It follows that each story presented five frames to each participant (see Appendix for examples).

82 What Autism Can Reveal About Every . . . not Sentences

(6) Quantified assertion preceding negation (QN): All the children are not in the pool. (7) Quantified (positive) assertion without negation (QP): All the children are in the pool. (8) Factual statement that could be true or false: The children in the pool were playing with a ball.6 Participants always received either a QN statement or a QP statement along with a single factual statement. Overall, each participant received six QN statements (with half being in the 2-of-3 condition and half in the 3-of-3 condition) and four QP statements (again, with half being in each context condition) that were distributed among the ten stories so that we could do analyses based on participants as well as stories. 2.1.4 Procedure Each participant was engaged individually by the second author. She read a prepared text as she presented the story’s pictures at sensible, prescribed times. Each frame was accompanied by one or two short sentences. The factual statement was the first test item presented and it was followed by one of the quantified statements.

5

Seven factual statements were designed to lead to a true evaluation and 3 to a false one. This was the question for the story presented in Table 1. To appreciate a false statement, see Appendix. 6

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

difficult to provide consistently correct responses even in the 3-of-3 context. All the stories were prepared so that they could be part of either condition. The stories were broken up into 4 groups so that a different random set of five stories represented the 2-of-3 condition in each with the remainder in the 3-of-3 condition. To obscure the purpose of the experiment, the above sort of statement was intermixed with positively quantified statements as well (see (7) below). For example, the above story could be followed by All the children are in the pool (which is false in the 2-of-3 context and true in the 3-of-3 context). Finally, as a way to verify comprehension of each story, a quantified statement was always accompanied by a factual statement that also required an evaluation. One factual statement accompanied each story and it was permanently linked to a story.5 In summary, participants—across the task—were confronted with three sorts of statements as participants were presented each of the ten stories:

Ira A. Noveck et al. 83 1 0.9 0.8 0.7 0.6

Adults Autistic Part. Children

0.5 0.4 0.3 0.2 0.1 0 2-of-3 QN

2-of-3 QP

3-of-3 QN

3-of-3 QP

2.2 Results We analysed the results with respect to choice proportions (with a correct response to the critical QN question being the adult ‘true’ response) and we analysed the results both with respect to individuals and to stories.7 We first summarize the control items to demonstrate that all the participants generally understood the stories and the quantified statements before turning to the sentence type of interest, QN in a 2-of-3 context. Rates of correct performance were high for all of the control problems. Adults’, verbally competent autistic participants’, and children’s rates of correct responses to the factual statements were M ¼ 0.98, M ¼ 0.96, and M ¼ 0.86, respectively.8 As can be seen in Figure 1, rates of correct responses to the QP statements were comparably high for adults, verbally competent autistic participants, and children (M ¼ 0.95, M ¼ 0.88, and M ¼ 0.92 overall, respectively). Finally, rates of correct responses to the QN statement in the 3-of-3 condition were high and comparable among adults, verbally competent autistic participants and children (M ¼ 0.93, M ¼ 0.82, and M ¼ 0.81, respectively). Figure 1 displays the mean percentage of 7 By convention we refer to F values obtained with participants as a random factor as F1 (or t1) and with items as a random factor as F2 (or t2). 8 Rates of correct responses to the factual statements were significantly higher among the adults than the children, t1(37) ¼ 3.06, p < .005, t2(18) ¼ 3.61, p < .005, but not significantly higher than among the autistic participants). The significant differences were due to the fact that adults’ responses were at ceiling. Children’s responses to factual statements were still well above levels predicted by chance.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Figure 1 Mean percentage of correct responses among adults, autistic participants and 4-yearold children to quantified sentences that were affirmative (QP) or negative (QN) and across situations where two-of-three characters (or objects) were part of a completed action or where all three were. Adults’ responses (reflecting a Not every reading) were considered correct in the 2-of-3 QN case.

84 What Autism Can Reveal About Every . . . not Sentences

3 DISCUSSION This experiment was designed to replicate Musolino et al.’s (2000) and Musolino and Lidz (2006) developmental effect with Every . . . not sentences, and, by introducing verbally competent autistic participants to this paradigm, we aimed to verify that typical adult interpretations of such sentences rely on pragmatic processes. We did indeed replicate the effect among typically developing (children and adult) participants, but in a way that was slightly different than what was found in Musolino et al. (2000) and in Musolino and Lidz (2006). In the present study, children adopted either reading seemingly randomly (with a slight nonsignificant preference for None readings), reflecting the ambiguity of the sentence, and adults—as one would predict—preferred the Not every interpretation. Also, we showed that the verbally competent autistic participants and the children are very similar with respect to their responses. Responses to the control items indicate that neither the quantifier all nor the complex QN sentence is problematic for children and autistic participants and that they have a very high level of

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

correct responses to the two sorts of quantified statements across the two conditions. In contrast, the QN statement in the 2-of-3 condition did not yield comparable rates of ‘true’ responses. As can be seen in Figure 1, rates of ‘true’ responses were higher for the adults than for either the children or the autistic participants (M ¼ 0.88, M ¼ 0.45, M ¼ 0.40, respectively). A one-way ANOVA showed a significant main effect due to cohort group, F1(2, 51) ¼ 13.199, p < 0.001, F2(2,27) ¼ 10.7, p < 0.001. Adults were significantly more likely to respond true to this statement than the verbally competent autistic participants t1(33) ¼ 3.96, p < 0.001; t2(18) ¼ 3.68, p < 0.005 and the children, t1(37) ¼ 5.87, p < 0.001; t2(18) ¼ 4.48, p < 0.001. Children and verbally competent autistic participants were statistically indistinguishable and both were at chance levels ( p’s > 0.5). Slightly over a quarter of the children (5 of 19) consistently preferred the None reading (i.e. in all three trials) and slightly over 10% (2 of 19 children) consistently preferred the Not every reading, with the remainder choosing inconsistently. Six of 15 autistic participants (40%) consistently preferred the None reading and three (20%) consistently preferred the Not every reading, with the remainder choosing inconsistently. Sixty-five percent of the adults consistently chose the Not every reading and the remainder chose this reading in 2 (of 3) trials.

Ira A. Noveck et al. 85

9 Musolino and Lidz (2006, Experiment 2; see also Pouscoulous, Noveck, Politzer & Bastide, in press) have provided evidence showing that adults, unlike children, are more likely to reject statements that are literally presented as Not every in scenarios that express ‘‘None’’ and for the pragmatic reason just stated (Not every is underinformative with respect to a scenario describing None).

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

comprehension overall. This shows that the QN statement presents a challenge for the non-adults uniquely in the 2-of-3 cases, and arguably because it is ambiguous. Overall, we take this as evidence in favour of the idea that a typical adult’s Not every reading involves some amount of pragmatic processing. Below, we consider how to account for the pragmatic processing of these Every . . . not sentences by first reviewing the two less controversial accounts suggested by Musolino and colleagues, the Acquired Manner and Charity accounts, as well as our own Processing account. This is followed by the presentation of another hypothetical test case that can eventually serve to differentiate the Acquired Manner account from the other two. According to the Acquired Manner account, the Not every reading of sentence (1) occurs practically by default because sophisticated speakers reject the None reading. If the speaker wanted the listener to treat, e.g. All the children are not in the pool as None of the children of in the pool, the speaker would have put it that way to start with. Of course, this account assumes that only the adults in the present study will have acquired this capacity. According to the Charity account, adults are more apt at adopting an interpretation that keeps the utterance true and, in fact, adults appear to do that. According to the Processing account we applied, the Not every reading in a 2-of-3 context is preferred over a None reading because the former entails less effort when integrating it into the context. That all three accounts should be consistent with the data is not surprising. Here we entertain an experimental scenario which could test between the Acquired Manner account and the two others. It would involve presenting an Every . . . not test sentence in a context where no protagonist is doing what is described in the verb phrase (consider a case where 0 of 3 children can be found in a pool). Such a situation makes a None reading of the Every . . . not sentence (Every child is not in the pool) true semantically. In fact, it makes the Not every reading true as well (None entails Not every). However, according to the Acquired Manner account, adults routinely interpret Every . . . not sentences with a Not every reading (while presumably bypassing None readings). This reading (Not every child is in the pool) has the potential to prompt a false response in 0-of-3 contexts by way of a scalar implicature because Not every is under-informative when None is appropriate.9 The Acquired

86 What Autism Can Reveal About Every . . . not Sentences

10 Whereas the Charity account argues that pragmatics aims for true responses, our Processing account does not rule out a response indicating a Not every reading with an implicature (i.e. a false response) in a 0-of-3 context. Our account would only expect the more effortful interpretation to be less preferred. Further work would be required to distinguish these two accounts experimentally.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Manner account should predict a noticeable proportion of false responses (at least at rates comparable to those found when an utterance is literally expressed as Not every). The Charity account would predict that adults would adopt a reading that would make the utterance true, which is delivered by the None reading. Our own account would predict that the adults would prefer the None reading in this hypothetical 0-of-3 case and for two reasons. First, it is more immediately consistent with the presented context (here anticipating the rest of the sentence in the pool at the word not is consistent with a None reading). Second, even if participants were to consider a Not every reading, it does not automatically prompt a scalar implicature (see Noveck 2001, Bott & Noveck, 2004; Pouscoulous et al. in press).10 Choosing an appropriate reading for an Every . . . not sentence differs markedly from the interpretation procedure concerning scalars. The work on scalar terms has shown that the semantic interpretation of weak scalars (such as some) can suffice. It is only when a text or a task encourages participants to narrow the interpretation (of some to some but not all) that reading times or response times slow down on experimental tasks (Bott & Noveck 2004; Noveck & Posada 2003; Breheny et al. 2006). For example, when presented ‘underinformative’ items such as Some cows are mammals, those participants who answer true do so at a pace (roughly 800 msecs) that is comparable to that of control items (including to false controls, such as All mammals are cows), but those who answer false to ‘underinformative’ items (indicating that they invoked a narrower meaning for some) take roughly 1400 msecs; i.e. significantly longer than all the controls (Bott & Noveck, 2004, Experiment 1). Data from developmental studies and reaction time studies support the notion that pragmatic enrichment in scalar implicature tasks is optional or that it arrives as part of a more involved procedure. In the case of Every . . . not sentences, it is in the addressee’s interest to choose one reading or otherwise the sentence remains ambiguous. Unlike for the case of scalars, Every . . . not sentences require disambiguation and this depends on pragmatics. One might ask why the children in Musolino and colleagues’ studies are not consistently equivocal (as a group) like the children in the current study. We suspect that the difference is due to experimental procedures. Whereas Musolino et al. (2000) presented stories in which three characters were first shown failing to do something (e.g. in the horse-jumping

Ira A. Noveck et al. 87

APPENDIX Two examples stories, translated from French, each with a 2-of-3 and 3-of-3 ending, its factual item (test of comprehension) and quantifier test items. Pool (1) It is vacation. Thomas, Eric and Sandrine are playing with a ball in the garden (the picture features a pool). (2) It is very hot in the sun, the kids are exhausted and they are perspiring a lot. (3) Thomas and Sandrine want to go swimming. They get undressed to go inside the pool. Eric does not feel like going swimming. (4) Thomas and Sandrine jump into the pool while Eric sits down next to a tree. (5) (2-of-3) It is a lot of fun to play with the ball in the pool. Eric prefers to read a book in the shade. (5) (3-of-3) It is a lot of fun to play with the ball in the pool. Eric decides to join them.

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

scenario, all three declined jumping over a barn before considering the fence instead), our task did not. In general, truth-value judgment tasks require its participants to integrate prior information in the form of base rates of success or failure from a previous event. Our task places no particular focus on what the three protagonists or objects did as a group prior to the ending. The outcomes concerning the three were largely incidental to the conclusion. Nevertheless, in all of these studies one finds that adults come up with a Not every reading more routinely than children. To summarize, the results from this study confirm that adults rely on pragmatic processes to determine which of two readings is most applicable when hearing Every . . . not sentences. A contribution of this work is that it puts one in the position to extend the list of processing difficulties in autism to syntactic ambiguities. This modest claim is supported by the autistic participants’ consistent performance with respect to the comprehension questions and the multiple unambiguous control sentences (making these participants comparable to both the children and adults) and their being inconsistent solely with respect to QN sentences in the 2-of-3 contexts (making them comparable solely to the children). It remains to be seen how to best describe the pragmatic abilities that distinguish healthy adults from both typically developing children and autistic participants. Nevertheless, this work confirms that there is much to be gained by viewing Every . . . not sentences through a pragmatic lens.

88 What Autism Can Reveal About Every . . . not Sentences Fact: The children in the pool are playing with a ball. (True). QN: All of the children are not in the pool (Tous les enfants ne sont pas dans la piscine). QP: All of the children are in the pool (Tous les enfants sont dans la piscine).

Fact: The child is wearing (long) pants (False; he is wearing shorts). QN: All the chicks are not born (Tous les poussins ne sont pas ne´s). QP: All the chicks are born (Tous les poussins sont ne´s). Acknowledgements The authors wish to thank Robyn Carston, Bart Geurts, Julien Musolino, Anne Reboul and three anonymous reviewers for their comments with respect to an earlier version of this paper, Jeffrey Lidz and Sandra Waxman for conversations during this work, as well as the team RD&P in Lyon (Jean-Baptiste Van der Henst, Mathilde Bonnefond, Coralie Chevallier, and Je´roˆme Prado) for their helpful suggestions on all phases of the work. We also wish to thank those who participated in the experiment as well as the ITTAC in Villeurbanne for their collaboration. This work was supported by a grant from the Fondation de France awarded to the first and third authors and coordinated by Marc Jeannerod. Correspondence concerning this article should be addressed to Ira Noveck, L2C2, Institut des Sciences Cognitives, 67 Blvd. Pinel, 69675 Bron, France (email: [email protected].) IRA A. NOVECK Laboratoire sur le Langage le Cerveau et la Cognition Universite´ de Lyon 1 and CNRS RAPHAE¨LE GUELMINGER Laboratoire sur le Langage le Cerveau et la Cognition Universite´ de Lyon 1 and CNRS

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Eggs (1) Stephane spends his vacations at his grandmother’s, who owns a farm. (2) He spends his time playing with the hens and the chicks. (3) He feeds them and cleans the chicken coop every morning. (4) Three hens are in the process of brooding (couver), the three eggs should not take much longer before they hatch. (5) (2-of-3) The next day, while cleaning the coop, Stephane notices that two new chicks were born. There is still one more egg. (5) (3-of-3) The next day, while cleaning the coop, Stephane notices that the three new chicks were born.

Ira A. Noveck et al. 89 NICOLAS GEORGIEFF Centre de Neuroscience Cognitive Universite´ de Lyon 1 and CNRS NELLY LABRUYERE Centre de Neuroscience Cognitive Universite´ de Lyon 1 and CNRS

REFERENCES D. Schiffrin (ed), GURT. Georgetown University Press. Washington DC. Joliffe, T. & Baron-Cohen, S. (1999) ‘A test of central coherence theory: Linguistic processing in high-functioning adults with autism or Asperger syndrome: Is local coherence impaired?’ Cognition 71:149–185. Levinson, S. (2000) Presumptive Meanings: The Theory of Generalized Conversational Implicature. MIT Press. Cambridge, MA. Lidz, J. & Musolino, J. (2002) ‘Children’s command of quantification’. Cognition 84(2):113–154. Moore, C., Pure, K. & Furrow, D. (1990) ‘Children’s understanding of the modal expression of speaker certainty and uncertainty and its relation to the development of a representational theory of mind’. Child Development 61:722–730. Musolino, J., Crain, S. & Thornton, R. (2000) ‘Navigating negative quantificational space’. Linguistics 38(1): 1–32. Musolino, J. & Lidz, J. (2003) ‘The scope of isomorphism: Turning adults into children’. Language Acquisition 11(4):277–291. Musolino, J. & Lidz, J. (2006) ‘Why children are not universally successful with quantification’. Linguistics 44(4):817–852. Noveck I. A. (2001) ‘When children are more logical than adults: Experimental

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Atlas, J. (1989) Philosophy Without Ambiguity. Clarendon Press. Oxford. Bott, L. & Noveck, I. A. (2004) ‘Some utterances are underinformative: The onset and time course of scalar inferences’. Journal of Memory and Language 51(3):437–457. Carston, R. (2002) Thoughts and Utterances. Blackwell. Oxford. Chierchia, G., Guasti, T., Gualmini, A., Meroni, L. & Crain, S. (2004) ‘Adult’s and children’s semantic and pragmatic competence in interaction’. In I.A. Noveck and D. Sperber (eds), Experimental Pragmatics. Palgrave Macmillan. Basingstoke. Dennis, M., Lazenby, A.L. & Lockyear, L. (2001) ‘Inferential language in high-function children with autism’. Journal of Autism and Developmental Disorders 31(1):47–54. Happe´, F. (1995) ‘Understanding minds and metaphors: Insights from the study of figurative language in autism’. Metaphor and Symbol 10(4):275–295. Happe´, F. (1997) ‘Central coherence and theory of mind in autism: Reading homographs in contexts’. British Journal of Developmental Psychology 15:512–521. Hirst, W. & Weil, J. (1982) ‘Acquisition of the epistemic and deontic meaning of modals’. Journal of Child Language 9:659–666. Horn, L. R. (1984) ‘Toward a new taxonomy for scalar inference’. In

90 What Autism Can Reveal About Every . . . not Sentences Pouscoulous, N., Noveck, I. A., Politzer, G. & Bastide, A. (in press). ‘Processing costs and their impact on the development of scalar implicature’. Language Acquisition. Prado, J. & Noveck, I. A. (2006) ‘How reaction times can elucidate matching effects and the processing of negation’. Thinking and Reasoning 12(3):309–328. Sperber, D. & Wilson, D. (1995) Relevance: Communication and Cognition. Basil Blackwell. Oxford. Tager-Flusberg, H. (1981) ‘On the nature of linguistic functioning in early infantile autism’. Journal of Autism and Developmental Disorders 11:45–56. Submitted: 15.05.06 Accepted: 01.08.06

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

investigations of scalar implicature’. Cognition 78(2):165–188. Noveck, I. A. (2004) ‘Pragmatic inferences related to logical terms’. In I.A. Noveck and D. Sperber (eds), Experimental Pragmatics. Palgrave. Basingstoke. 301–322. Noveck, I. A, Ho, S. & Sera, M. (1996) ‘Children’s understanding of epistemic modals’. Journal of Child Language 23(3):621–643. Noveck, I. A. & Sperber, D. (in press). ‘The why and how of experimental pragmatics: The case of ‘‘scalar inferences’’ ’. In N. Burton-Roberts (ed), Advances in Pragmatics. Palgrave. Basingstoke. Papafragou, A. & Musolino, J. (2003) ‘Scalar implicatures: experiments at the semantics-pragmatics interface’. Cognition 86(3):253–282.

Journal of Semantics 24: 91 doi:10.1093/jos/ffl012

Corrigendum Created Objects, Coherence, and Anaphora ERIC McCREADY This journal, volume 23 (2006), number 3, pp. 251-279

Ó The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Downloaded from jos.oxfordjournals.org by guest on January 1, 2011

Due to an unfortunate misunderstanding, the printed version of this article contains several superfluous instances of the uparrow ([), which should have been replaced with blanks. The error occurs at pages 262, 263, 265, 266, and 268. The article should not have contained any instances of this symbol, at all.