Cohen 2004: Existential Generics Shay Hucklebridge LING 720 I Empirical claims about -Generics In this paper, Cohen describes a number of cases where generics appear to receive a quasi-existential interpretation in place of the usual quasi-universal interpretation. Emphatic Affirmation Cohen notes that when a generic is uttered as an emphatic, contrastive affirmation, it appears to be interpreted existentially: (1) a. A: When Mary smokes, it is never after dinner. B: That s not true! Mary [does] F smoke after dinner. b. Mary doesn t smoke after dinner For B s reply to A to be true, it is sufficient that Mary sometimes smokes after dinner, and it does not need to be the case that this is her typical behaviour. B is the refutation of a universal statement, not a generic. So B refutes that Mary never smokes after dinner, it does not refute the generic counterpart of A given in 1b (note that the auxiliary in 1b is not focused). Since the generic statement in 1b only has quasi-universal force, it allows exceptions, and so a statement with quasi-existential force like the one in B is not sufficient to negate it. Emphatic Negation When the focused auxiliary expresses negation rather than affirmation, generics also receive a quasi-existential interpretation. (2) a. Mammals [don t] F lay eggs. b. Mary [doesn t] F smoke after dinner. The existence of platypuses (egg-laying mammals) is sufficient to render 2a false. And likewise 2b is false if Mary occasionally does smoke after dinner. Cohen argues that, in these cases, negation has wide scope over an existential statement ( x.x is a mammal and x lays eggs) If the focus occurs on another constituent, the quasi-existential reading does not hold (e.g. Mammals don t [lay eggs] F means in general, mammals don t lay eggs and is negating the generic Mammals [lay eggs] F ). 1
Focus-Sensitive Particles (3) a. Only [mammals] F bear live young b. We play soccer only if [lay eggs] F Only is an operator which requires that no relevant alternative to the prejacent is true. The prejacents of 3a and 3b are [Mammals] F bear live young and We play soccer if [the sun is shining] F, respectively. So for 3a above to be true, the following alternatives must all be false: (4) a. Birds bear live young b. Reptiles bear live young c. Fish bear live young d. etc... If the sentences in 3a and 3b are treated as quasi-universals, then 3a should not be false, as there is not class of animals other than mammals that generally bear live young. According to Cohen, this is bad, because the existence of any non-mammal that bears live young is sufficient to render 3a false. If the prejacent is interpreted existentially, then this problem goes away. This is because the alternative will also be existential, and so the existence of any non-mammal that bears live young will be a true alternative to 3a, which would cause 3a to be false. Additives Like only, additives also and too give rise to quasi-existential readings. Cohen gives the following example: (5) a. Birds lay eggs. b. Mammals also lay eggs. c. Mammals lay eggs too. If 5b or 5c is uttered following 5a, it only requires that some mammals lay eggs. It does not have to be the case that Mammals generally lay eggs for this to be true. Unrestricted Habituals (6) a. Mary smokes b. We play soccer With respect to unrestricted habituals, Cohen brings up the problem of how quasi-universal semantics yield a generalization that is too strong. In the case of 6a, this would mean that in most situations with Mary, Mary smokes in that situation. 2
He points out that the solution proposed by Krifka (that unrestricted habituals express quantification over normal situations) is too vague and does not make clear predictions. To solve this he appeals to an earlier suggestion of Krifka s, saying that these sentences are existential statements, with the existential quantifier quantifying over conditions that bring about Mary s smoking (as opposed to situations in which Mary smokes). In other words, Marys smoking is possible; if the right conditions obtain, she would smoke. II Against an -Bare Plural analysis Since the data Cohen addresses seem to have existential readings, why not treat them like existential bare plurals? Bare plurals can receive an existential interpretation in addition to their generic one: (7) a. Birds are flying overhead. b. Mechanics are available. In the case of habituals, he states that: habituals too admit of existential readings. Unlike bare plurals, habituals are not normally considered to be ambiguous between generic and existential readings they are only generic. Because of this, it would be necessary to come up with an explanation for how, and when habituals are actually interpreted existentially. He also notes that the bare plural in these existential generics can be replaced with a definite determine, and that this does not change the meaning significantly: (examples adapted) (8) a. A: Nobody in America travels to Cuba. B: That s not true! The American does travel to Cuba. b. The Martian mammal doesn t lay eggs. c. Only the Martian mammal bears live young. Definites cannot typically be interpreted as an simple existentially quantified variable, and so further explanation is needed for cases where they appear in quasi-existential generics. Unlike regular existentials, the quasi-existential generics that Cohen discusses still appear to be law like, and express generalizations that aren t temporary. Suppose that during this year s mating season, a group of puritan zoologists took the trouble to prevent any male platypus from getting near any female platypus, and any male echidna from getting near any female echidna, so that this year no mammals laid eggs. In this scenario, the following sentences would all remain true: 3
(9) a. Mammals [don t] F lay eggs. b. A: Birds lay eggs. B: Mammals also lay eggs / Mammals lay eggs too. III Focus and Generics Generics are evaluated with respect to a set of alternatives. (10) Mammals bear live young 10 is evaluated with respect to other ways of reproducing. So its alternative set is {lay eggs, undergo mitosis, etc...}. Different sets of alternatives can change the truth conditions of a sentence: (11) People buy cheap goods from thieves a. {cheap, medium priced, expensive,... } b. {from thieves, at garage sales, at auctions,... } Depending on what alternative set 11 is evaluated with, the meaning changes. If set a. is selected, then the sentence has the meaning in general, if a person buys something from a thief, the price will be cheap. If the set in b. is used instead, then the meaning of the sentence will be something like if a person buys something cheap from some source, it will generally be bought from a thief. The means of selecting the relevant alternative set is through focus. If cheap is focused in 11, then the alternative set will be a. Likewise, if from thieves is focused, the alternative set will be b. Cohen proposes that in the existential generics, no alternatives are introduced. This occurs because there is either no focus, or the focused part is not associated with the generic quantifier. (12) a. A: Nobody in America travels to Cuba. B: That s not true! Americans [do] F travel to Cuba! b. A: When Mary smokes, it is never after dinner. B: That s not true! Mary [does] F smoke after dinner. c. Mammals [don t] F lay eggs. d. Mary [doesn t] F smoke after dinner. In 12, the focus is on the auxiliary, and is contrastive. In Cohen s view, this indicates that the focus is not associated with the generic quantifier, and so cannot be the source of alternatives. 4
When focus is contrastive, it does not associate with focus-sensitive operators (13) A: John always takes [Mary] F to the movies. B: No! [Peter] F always takes Mary to the movies. In 13 the focus on Mary is associated with the focus-sensitive operate always (whenever John takes someone to the movies, it is always Mary). In B, the focused element is not associated with always. According to Cohen although Peter is focused, B is not saying that whenever someone takes Mary to the movies, it is always Peter, but rather that whenever Peter takes someone to the movies, it is always Mary (and that A is wrong in saying that John, rather than Peter, behaves in this way). This is likewise true with generics, for Cohen: (14) a. A: People buy [cheap] F goods from retailers. B: No! People buy cheap goods [from thieves] F. b. A: Mary smokes [after dinner] F. B: No! [Kate] F smokes after dinner. The sentences in B are not evaluated with respect to alternatives of their focused material, since the focus in these cases is serving a contrastive role. Therefore, the B sentences are interpreted as corrections to the claims of A. In these generics, the B sentences also inherits the focus structure of A as second-occurrence focus, and this secondary focus does associate with the generic operator (or with always in 13). IV Generalizations about -reading environments What Cohen proposes for the existential generics (12) is that focus in these sentences is not associated with the generic quantifier, and additionally no second-occurrence focus is introduced (it is not inherited, introduced by a cleft, etc.), so not set of alternatives is generated. Second-occurrence focus cannot appear in the sentences in 12 because focus on the auxiliary does not allow focus on the rest of the sentence: (15) A: John never takes [Mary] F to the movies. B: No! John always [does] F take Mary to the movies. B here cannot have the the interpretation that whenever John takes someone to the movies, it is Mary. This is because there is no second-occurrence focus on Mary. This is likewise the case with sentences containing focus sensitive particles (16) a. Only [mammals] F bear live young. b. Even [mammals] F lay eggs. 5
Focus is not associated with the generic quantifier, but with the focus-sensitive particle, and so no alternatives are introduced. In the case of additives, there are two ways to view things, according to Cohen. (17) a. Mammals also lay eggs. b. Mammals lay eggs too. If we view them as focus sensitive, then the same line of reasoning used for other focussensitive operators applies. The second way to look at additives is as particles that are focused which have an associated element that is a contrastive topic. Both of these views leave no focused material in the scope of the generic quantifier, and so no alternatives are introduced. In the case of unrestricted habituals, Cohen proposes that there is no narrow focus there at all. All sentences have wide focus (i.e. introducing something new), but there is no reason to expect all sentences to also have narrow focus. If there is no narrow focus associating with the generic quantifier, no alternative set is introduced. V An explanation of these facts So far: a) there are a variety of generics that get quasi-existential readings b) a common characteristic of these generics is that no alternatives are introduced. What is the causal relationship between a) and b)? First, preliminary remarks of the semantics of generics Cohen s LF: (18) gen x [mammal(x), bear-live-young(x)] In general, if x is a mammal, x bears live young. The disjunction of the set of alternatives introduced by focus restricts the domain of the generic quantifier. So 18 above is evaluated with respect to the following: (19) A = { bear-live-young, lay-eggs, undergo-mitosis } If x is a mammal and x satisfies A(x) (the disjunct of alternatives), then, in general x bears live young. 6
What happens if there are no individuals that satisfy this? Since quantifiers presuppose the nonemptiness of their domain, and the generic quantifier should behave the same way, this should yield sentences that are odd. However, these sentences seem to be false, rather than odd: (20) Male hedgehogs bear live young. Cohen s previous work suggests that in these situations, when a presupposition is imminent, a different set of alternatives is accommodated. This set consists of a tautology: (21) A = { bear-live-young, bear-live-young } Every male hedgehog automatically satisfies the disjunction of these alternatives. Because of this accomodation, 20 expresses quantification over all male hedgehogs, not only over the (nonexistent) procreating ones, thereby avoiding presupposition failure. If there are no alternatives to a proposition φ introduced by narrow focus, then the alternative set will be the singleton set {φ}. So: (22) a. In general, if ψ(x) A(x), then φ(x) b. A = φ c. In general, if ψ(x) φ(x), then φ(x) This is trivially true. But what if there is no x that satisfies this? Then the domain of the generic would be empty, leading to presupposition failure, and accommodation occurs, yielding the alternative set: (23) A = {φ,, φ} Since this is a tautology, the generic is true iff: (24) In general, if ψ(x) then φ(x) However, since this accommodation is triggered only when no ψs are φs, then 24 must be false. To summarize: when the alternative set is a singleton, the generic is true iff there is some individuals that satisfy ψ(x) and φ(x) when they are assigned to x. So in a sentence like the following, it is true just in case mammals are the only class that has some members that bear live young: (25) Only [mammals] F bear live young The restriction of the domain of the generic quantifier by a singleton set of alternatives gives the generics discussed here their existential reading. They are not truly existential, however, as the generic quantifier is still contributing a lawlike quality. But exceptional cases will not invalidate the generic the only requirement is that there exist some case where the generic is true. 7