You are on page 1of 38

j~a'n~hff

ELSEVIER Journal of Pragmatics 32 (2000) 289 326

From tf to iff:
Conditional perfection as pragmatic strengthening
Laurence R. Horn
Department of Linguistics, Yale University,
P.O. Box 208 236 Yale Sta., New Haven. CT 06520-8236, USA

Received 1 November 1997; revised version 22 March 1999

Abstract

Geis and Zwicky's squib on "conditional perfection" (1971) released a hornet's nest of
rebuttal and counter-rebuttal into the pragmatic atmosphere, with many scholars in the area
playing alternately the roles of stinger and stingee. Geis and Zwicky's goal was to explain the
notorious tendency among introductory logic students and ordinary speakers to 'perfect' an if-
then conditional of the form of If you mow the lawn, I'll give you five dollars into the corre-
sponding iffbiconditional of the form I f and only if you mow the lawn, I'll give you five dol-
lars. In an overlapping pair of studies, van der Auwera (1997a,b) has recently contributed a
comprehensive survey of the literature on conditional perfection before and since Gels and
Zwicky as a microcosm of the development of post-Gricean pragmatic theory. While agree-
ing with van der Auwera on the centrality of the if --~ lJf move for contemporary pragmatics,
I will critique his treatment and offer my own perspective on the data and their ramifications,
along with an expanded history that touches on the manifestations of conditional perfection
and related inferential fallacies addressed in philosophical treatises, empirical psycholinguis-
tic studies, and self-help primers. By extending the data base to include counterfactuals and
other non-predictive conditionals, I will also present problems for Dancygier and Sweetser's
(1997) alternative recent account of the CP phenomenon within the mental-spaces frame-
work. The overall focus will be on the implications for the theory of conversational implica-
ture, and in particular of R-based pragmatic strengthening, that can be drawn from the condi-
tions on - and motivation for - conditional perfection, © 2000 Elsevier Science B.V. All
rights reserved.

Keywords: Conditionals; Conditional perfection; Fallacy; Implicature; Invited inference;
Pragmatic strengthening

r~ Briefer (and hence less persuasive) versions of this paper were presented as the Second Annual
Emeritus Lecture in honor of Michael Geis and Arnold Zwicky at the Ohio State University in May,
1997 and at Northwestern University in October, 1997. Thanks to those attending the earlier presenta-
tions for their helpful comments, to Johan van der Auwera, Scott Schwenter, Kai von Fintel, Kent Bach,
Jacob Mey, and an anonymous JoP referee for some pertinent pointers, suggestions, and complaints, and
to Dimitri Gutas, my informant for Ancient Greek.

0378-2166/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 2 1 6 6 ( 9 9 ) 0 0 0 5 3 - 3

290 L.R. Horn / Journal of Pragmatics 32 (2000) 289-326

1. Introduction

Michael Geis and Arnold Zwicky's (1971) seminal squib on the inference schema
they dubbed CONDITIONALPERFECTION(henceforth CP) recently celebrated its twenty-
fifth birthday. The highlight of the silver anniversary ceremonies was the circulation
and subsequent appearance of two overlapping papers by Johan van der Auwera
(1997a,b). These twin treatises were designed not only to provide the correct treat-
ment of the CP inference but also - as the title of van der Auwera (1997a) suggests
- to characterize the history of scholarship on this question as a microcosm of the
development of pragmatic theory and description over the last quarter century.
Despite their usefulness and cogency, however, van der Auwera's are not the final
words on conditional perfection either from a theoretical or historiographic perspec-
tive, as I shall attempt to demonstrate in the present study.
The aim of Geis and Zwicky (1971) was to baptize and illuminate the notorious
tendency among introductory logic students and ordinary working speaker/hearers to
'perfect' an if-then conditional of the form of (la) into the corresponding if and only
tf biconditional of (lc) through the mediation of the corresponding only if condi-
tional of (lb). 1 CP, the tendency to move from (la') to (lb'), was taken to be an
instance of INVITED INFERENCE, a relation weaker than logical or semantic entail-
ment. 2

(1) a. If John leans out of that window any farther, he'll fall.
a t'
LDF
b. If John doesn't lean out of that window any farther, he won't fall.
b p" - L ~ - F
c. If and only if John leans out of that window any farther, he'll fall.
C t" L <--->F

In the same way, the utterance of (2a) was claimed to invite the inference of (2b) or
equivalently (2b')/(2b"), thus yielding the conveyed meaning of (2c): 3

As van der Auwera stresses (and as we shall illustrate below), Geis and Zwicky (1971) were unwit-
tingly echoing Ducrot (1969) in a number of respects. One such echo is in the consequences of this ten-
dency or linguistic habit: "C'est sans doute cette habitude linguistique qui rend si difficile aux appren-
tis math6maticiens de distinguer les conditions n~cessaires des conditions suffisantes" (Ducrot, 1969:
18).
2 There is some indeterminacy in the literature about whether it is the 'invited inference' of (lb) itself
that constitutes CP, or the biconditional (lc) conveyed by the union of this inference with the literal
meaning of (la). Since it is the status of the move from (la) to (lb), or from (2a) to (2b-2b") below, that
is the critical step here, this indeterminacy should not trouble us.
3 As McCawley has observed (1981: 49-54; cf. also McCawley, 1974, 1993: 81-88; and, indepen-
dently, Barker, 1993), q only ifp is better paraphrased by 'If not-p then not-q' than by 'If q then p' (or
by 'p if q'), at least when p and q have different temporal and causal implications. Thus, as he points out,
(i) is equivalent not to (ii) but rather to (iii):
(i) You're in danger only if the police start tapping your phone.
(ii) If you're in danger the police start tapping your phone.
(iii) If the police don't start tapping your phone you're not in danger.

L.R. Horn / Journal of Pragmatics 32 (2000) 289-326 291

(2) a. If you mow the lawn, I'll give you $5.
b. If you don't mow the lawn, I won't give you $5.
b p" Only if you mow the lawn will I give you $5.
bH ,
I will give you $5 only if you mow the lawn.
C. If and only if you mow the lawn, I'll give you $5.

While invoking the (then largely unpublished) work on non-logical deduction by
Paul Grice, Geis and Zwicky maintain that their invited inferences "are clearly dis-
tinct from the 'conversational implicatures' which are [Grice's] principal concern"
(1971: 565). As we shall see, their resistance on this point has proved more of a
speed-bump than a roadblock for others in search of the conditional-perfectionist
grail. In any event, they proceed from CP to the parallel phenomenon of INFERRED
CAUSATION, the fact that "sentences which express a temporal sequence of situations
invite the inference that the first situation is a cause or reason for the second" (Geis
and Zwicky, 1971: 564), as in (3a-b):

(3) a. After a large meal, we slept soundly.
[a'. As the result of a large meal, we slept soundly.]
b. Having finished the manuscript, she fell into a swoon.
[b'. Having finished the manuscript, she consequently fell into a swoon.]

In their discussion of inferred causation, Geis and Zwicky observe in passing
(1971: 564) that "this principle of inference corresponds to the familiar fallacy post
hoc ergo propter hoc (just as CP has its related fallacies)". It is with this correspon-
dence between invited inferences and traditional fallacies that we will launch our
study. The post hoc ergo propter hoc cases we shall return to in a bit, but we begin
with an expansion of that Geis-Zwicky parenthetical.

2. CP and inferential fallacy

The 'related fallacies' to which Geis and Zwicky refer are those of AFFIRMING(or
ASSERTING) THE CONSEQUENT and of DENYINGTHE ANTECEDENT.Each of these can be
seen as a perversion of the corresponding valid laws of inference for conditional
statements, MODUS PONENS and MODUS TOLLENSrespectively, as schematized below
(cf. Mackie, 1967):

(4) ~ODUS PONENS MODUS TOLLENS
[= AFFIRMINGTHE ANTECEDENT] [= DENYING THE CONSEQUENT]
If p then q. If p then q.
p. Not q.
Therefore, q. Therefore, not p.

(5) AFFIRMINGTHE CONSEQUENT DENYING THE ANTECEDENT
If p then q. If p then q.
q... Not p.
Therefore, p. Therefore, not q.

are attributable to the presence of what they call collateral information. Thus (8'a) may fail to trigger the inference of (8'b). 4 It is not necessary because the 'invited inference' remains even if the conditional form is removed.292 L. Horn / Journal of Pragmatics 32 (2000) 289-326 By the same token. When such information is present in the context. if not p then not q. In place of Mackie's (1967: 170) unambiguous but rather cumbersome labels for (Va) and (Vb). these inferences are sometimes distinguished as CONVERSION and OBVERSION respectively. The presence of a conditional is not sufficient either. the inferences are natural enough. declaring the presence of a conditional neither necessary nor sufficient to generate the CP inference. (8) I'll pay you $5 for mowing the lawn. as in (8"). Mow the lawn and I'll pay you $5. however fervently John might wish it to do so. The size of a monograph determined by a geometric progression will be left as an exercise for the reader. Therefore. 1973. Want to earn $5 ? Mow the lawn. the well-born inferential principle of contraposition (the infer- ence of the negation of the antecedent from the negation of the consequent) illus- trated in (6) gives birth to the two bastard children in (7). . thus the non-conditionals of (8) license the inference of (2b) just as strongly as the authorized licenser in (2a). if q then p. two of their then fel- low Buckeyes (BoEr and Lycan.R. CONVERTING A CONDITIONAL b. the latter corresponding to our invited inference in (1)-(2): (6) CONTRAPOSITION If p then q. 1972) sought to recall the invi- tations. the 'invited inferences'. b. If John doesn't quit. since invited inferences disap- pear if the context is altered. but when the context renders such inferences implau- sible. he won't be replaced. as in (3). If John quits. no causative relation will be assumed to hold between the prior and subsequent events: 4 It may be significant that BoEr and Lycan's broadside against Geis and Zwicky's 4. he will be replaced. If p then q. cf.1-page squib ran to twenty-two pages. including those associated with inferred causation as well as CP. For BoEr and Lycan. on the assumption that the progression has been arithmetic. Two years after the appearance of Geis and Z w i c k y ' s squib. also Lilje. I calculate that any full rejoinder to their critique would at this rate now require a 241-page treatise. (8') a. Therefore. (7) a. NEGATINGANTECEDENTAND CONSEQUENT If p then q. Therefore. if not q then not p.

I shall argue that their pessimistic conclusion is not warranted. to what can we attribute the invalid inference schemata noted by Geis and Zwicky and by innumerable hordes of fallacists before and since? 5 One hypothesis.CP and inferred causation . Now. That is. If you don't shut up I'll scream. Nor does the polarity of the antecedent and con- sequent seem to affect this tendency. Where the differences arise is in determining precisely which Gricean (sub)maxim applies and how it applies. 1 in van der Auwera 1997a: 268-269). the natural explanatory move is indeed (contra Geis and Zwicky) a turn toward Grice's Cooperative Princi- ple and its attendant maxims of conversation. This is confirmed by the results of a study by Fillenbaum (1986). Having finished the manuscript. this is in fact the step taken by the preponderance of those toiling in the rel- evant vineyards. and irrespective of his linguistic competence of any linguistic facts that he has internalized. 1973: 497~498) But while endorsing BoEr and Lycan's observation that the presence of a conditional is neither a necessary nor sufficient condition for the generation of CP. a'.b) demon- strates. ---) If you don't fix the car I won't give you $100. the inference-uninviters ultimately adopt a more non-committal stance. in each case we can completely account for the hearer's coming to have a certain belief or expectation just by pointing out beliefs and attitudes that he has simply as a result of his inductive observations of the way things typically work in his society or community. On this view. Horn / Journal of Pragmatics 32 (2000) 289-326 293 (8") a. the G-Z phenomena have nothing to do with language at all.R. any additional notion of invited inference will be otiose: "Most if not all of the G-Z phenomena can be explained in purely anthropological terms. If you fix the car I'll give you $100. In fact. ~ If you shut up I won't scream. L.do indeed constitute grist for the mills of linguistic pragmatics (BoEr and Lycan. Whatever the ultimate explanatory mechanism. b'. Since this collateral information can be used to predict where the inferences are nat- ural and where they are excluded. b. is to posit lexical ambiguity: 5 Two unusually thorough historical treatments of fallacies are those of Hamblin (1970) and Ebbesen (1981). . (9) a. assuming the context has been set up in the appropriate way. As van der Auwera (1997a. from Ducrot (1969) and Horn (1972) to van der Auwera himself (see Fig. we went to a movie. she decided to publish it with Random House. b.). allowing for an alternate approach in which the phenomena in question . On this view. 1973: 498ff. as unilluminating as it is uneconomical. in which 80% to 90% of his subjects readily accepted the 'falla- cious' inferences from the conditionals in (9a-b) to their primed counterparts. they maintain." (BoEr and Lycan. After a large meal. there can be no doubt that the ten- dency to draw Geis-and-Zwickian inferences is very real.

Therefore. 1967: 170) Of course the 'unfortunate ambiguity' dodge is one not particularly congenial to lin- guists . 6 An anonymous referee has called my attention to Wirth (1975). For example." (Mackie." (Fearnside and Holther. and the exclusive disjunc- tion the inclusive) guarantees that this burden will be hard to bear. A is C.294 L. Wirth claims (1975: 40) that the alternative pragmatic line "postulates a new type of inference in natural logic. therefore. As should have been clear from the literature when these words were written (cf. IrA is B. Barrett and Stenner. Hamblin. Horn." (Keynes. It is light. 1972. A is C. But as noted. 1884: 232-233) Not that Keynes actually discovered the fallacies: recognition of the invalidity of the inference schemata in (5a-b) can be traced back at least two millennia further to the Stoics. 1973) and is certainly clear by now (cf. A is C. the nineteenth century logician I like to think of as the grandfather of modern liberal economics. this 'well known' ambiguity is more of a chimera. A is not B: therefore. Not only does the burden of proof rest con- ceptually on anyone arguing for the semantic ambiguity of if and or. 1994. especially when an independent pragmatic account predicts the dual understandings of these operators.R. The lit- erature on the so-called ambiguity of inclusive vs. Jennings. then it is light. l f A is B. and references therein).the term is from Barrett and Stenner (1971) . whereas the logical ambiguity hypothesis does not involve any new type of phenomenon". it is day. exclusive and on the prag- or - matic recasting of that 'ambiguity' . 1959: 156) "An expression that actually asserts only a proposition of the form 'If p then q' may be wrongly taken as asserting 'q if and only if p'. 1971. sources (e. Likewise. Horn / Journal of Pragmatics 32 (2000) 289-326 "In common speech there is the ambiguity of 'if'. or it may mean 'possibly both. which may mean simply qf' or may mean 'if and only if'. 3) for a compendium of illustrations from dozens of logic texts of the persistent myth . for the reasons cited by Zwicky and Sadock (1975). 6 As to CP's cousins in (5a-b). as later catalogued by Sextus Empiricus. It is a fallacy if we regard the denial of the antecedent as justifying the denial of the consequent. for christening these fallacies: "It is a fallacy if we regard the affirmation of the consequent as justifying the affirmation of the antecedent. but the privative nature of the oppositions involved (the biconditional unilaterally entailing the conditional. .of a truth-functional exclusive disjunction in English and other languages. A is not C. vir- tually every pragmatic analysis of CP involves some combination of standard multi-use tools from the pragmatic workshop. which echoes the Fearnside and Holther line above: "It is well known that English or is ambiguous in that an English statement of the form P or Q may be true in the exclusive sense or in the inclusive sense" (Wirth. The same point applies to Wirth's related analysis of disjunction.g. but certainly one or the other'. and as we shall see in more detail below.is even more voluminous than that for condi- tional perfection. the inferences at issue are thus not fallacious at all but valid deductions from a biconditional interpretation. A is B.q. see Jennings (1994: ch. Both affirmation of the consequent "Such an argument as this is false: If it is day. a paper which does in fact 'explain' CP by taking English sentences of the form ifp then q to be semantically ambiguous between the propo- sitional logic expressions p D q and p --. 1970: 36) generally award pride of place to John Neville Keynes. 'or' may mean 'either one or the other but not both'. For example.or to philosophers weaned on the principle that Grice (1989 [1967]: 46) dubs Modified Occam's Razor: 'Do not multiply senses beyond necessity'. 1975: 41-42).

(Sextus Empiricus. A training manual for intellectual subversives. from Aristotle's rhetoric to t o d a y ' s pop-philosophi- cal guidebooks festooned with cute cartoons. affirming the consequent is a fallacy which comes naturally .. by use of a skilfully affirmed consequent: 'She's just a tramp.. 17ft.u p arguer can subtly morph into crafty sophist: 7 "This is an extremely good fallacy to use when you wish to impute base motives to someone. VIII. or The use of reason in everyday life (see references for full bibliographic information). so obvi- ously there is a rabid hedgehog about' . 1961:83 and Egli. the latter differ palpably from those sophisms and fallacies in dictione. In this respect.. the modem tradition has obscured the classical distinction between the inad- vertentfallacy of the hopelessly confused nal'f and the deliberate sophism of the self-conscious deceiver and rhetorical rou6. Adv. Adv. 1985: 9) W h e n Pirie shifts to the corresponding fallacy of (5b). and she did appear at the office party wearing a dress that was practically transparent!' " (Pirie. Math. 1961: 107) and denial of the antecedent If it is day. and false cause (post hoc ergo propter hoc) with the fallacies extra dictionem. Horn / Journal of Pragmatics 32 (2000) 289-326 295 for it allows us to infer a false conclusion from true premises. then it is light. 378. in Mates. instances of fallacious reasoning from commercials to political sound bites (perhaps not that long a stretch) and titles like The art of deception or Attacking faulty reasoning and subtitles like The counterfeit of argument. Math. Therefore. denial of the antecedent. F r o m Aristotle's treatment in the Sophistical refutations (De sophisticis elenchis) on. the fallacy of denying the antecedent is for those who do not really care if their brain is going forward or backwards. . Not for nothing is sophist the root of sophisticated. But whence such faux pas in the inferential shuffle? Some detect n o t h i n g more interesting than stupid blunders in reasoning: "To those who hopelessly confuse the order of horses and carts." (Pirie. ety- m o l o g y or other aspects of linguistic form and development. Girls like that always flaunt themselves before men.. the diabolical forensic schemer is once again forgotten and we are back to the poor confused schnook: "Like affirming the consequent. A practical guide to fallacy-free arguments. (Sextus Empiricus. 1985: 50) P i r i e ' s fool-vs." (Pirie. taxonomists have standardly reckoned affirmation of the consequent. L.. it is not light. 1983: 86) are on perspicuous display.-knave a m b i v a l e n c e on this point is amply replicated by other sources in the fallacist tradition. 'When cats are bitten by rabid hedgehogs they die. Here is a dead cat. It is not day. those i n d e p e n d e n t of language. You can always gain a hearing for your suggestion of less-than-honourable motives. in Mates. VIII. those dependent u p o n ambiguity. 1985: 7-8) But hopelessly confused and m i x e d . as in the classic casu- 7 Perhaps unfortunately.R. The arguer has mixed up the antecedents and consequents.

). not q. 8 But there is no i m m u n i t y against the d r a w i n g o f u n n a m e d fallacious conclusions. you are in Canada'. just as the traffic code is m o r e likely to address the failure to stop on a red light than the failure to p r o c e e d on green. even for those w h o should k n o w better. Horn / Journal of Pragmatics 32 (2000) 289-326 istry o f Ctesippus in P l a t o ' s Euthydemus (298): 'This dog is yours. If p then q. Not p. . Not p. q ( a n y w a y ! ) Or just as prescriptive grammarians keep reminding us that double negation 'really' affirms while let- ting us figure out for ourselves that a reinforced affirmation never amounts to a negation. You're not in Toronto. be dignified with a special name and treated similarly. 1987: 74) Indeed. But it does not follow that you're in Canada . Why not the Fallacy of Infer- ring the Conjunction of Two Propositions from their Material Equivalence. quite consistent in treating it [the Fallacy of Affirming the Conse- quent] separately . e s p e c i a l l y since W a l t o n is not instantiating the c o n v e n t i o n a l (5b) m o d e for the denial o f the antecedent but is rather inventing the brand n e w fallacy in ( 5 b ' ) : (5b) DENYING THE ANTECEDENT: (Sb') DENYING THE ANTECEDENT: TRADITIONAL MODE WALTONIAN MODE If p then q. the following argument is incorrect: 'If you're in Toronto then you're in Canada.. (5b). or (3)? (10) p ~ q Therefore.could. W h a t bars H a m b l i n ' s ' F a l l a c y o f Inferring the C o n j u n c t i o n o f T w o Propositions from their Material E q u i v a l e n c e ' from the ranks o f the fallacies extra dictionem is s i m p l y the u n l i k e l i h o o d that a n y o n e w o u l d inadvertently c o m m i t (or insidiously apply) this particular p a r a l o g i s m . 1970: 36-37) That is. say. Therefore. ergo this dog is y o u r father' (Hamblin. Perhaps unfortunately. w h a t m a k e s the fallacious inference s c h e m a in (10) any less deserving o f special status than those in (5a)." (Walton. of course.. for that matter. of other logical systems .R.296 L. this dog is a father. What is less clear is why it is singled out at all. This argument is invalid because the premisses could be true in the case where you're in Chicago. Here is D o u g l a s W a l t o n . as noted in Horn (1973: 213). 55ff. not so straightforward at all.. in theory. apply- ing deductive logic is not so straightforward as one might think or like. "We are only warned against those sins we are considered susceptible of committing". Therefore. denying the antecedent turns out to be invalid as a form of deductive argument. p and q. professional C a n a - dian fallacist and author or co-author o f o v e r two d o z e n important b o o k s and papers on the topic: "For similar reasons [to those determining the invalidity of inferences involving affirmation of the con- sequent]. 1970: 27. Therefore. For example. yet we do not hear of any others. Every invalid inference-schema of the propositional calculus or.. the Fallacy of Dis- tributing Quantifiers without regard to Negation Signs?" (Hamblin. or. But what motivates the classical and m o d e m traditions to isolate and treat just cer- tain s y n d r o m e s within a indefinitely large set o f paralogical p a t h o l o g i e s ? "'The modern books are.

the streets are wet. it has been raining. but not to universal affirmatives (all Fs can be Gs without all Gs being Fs) or conditional (hypothetical) statements. Aristotle introduces what will prove to be an important tool for the analysis of error (or deceit) within hypothetical reasoning: "The mistake seems to consist of treating the conditionals as expressing. no Gs are Fs). they then suppose also that if B is.. r. L. Sophistical refutations. Since the ground is wet after it rains. In the passage above." (Sextus Empiricus. we suppose that if the ground is wet. the natural starting point is as usual Aristotle. in Mates. In traditional logic. then. some Gs are Fs) and universal negatives (if no Fs are Gs.. Therefore.. B necessarily is. Adv. that the streets will not be wet. Pyrrh. "based on a false a s s u m p t i o n of u n b r e a k a b l e unity b e t w e e n elements which are. 167blff. A necessarily is . For an equally implausible mode of fallacious (or sophistical) reasoning that is less likely than W a l t o n ' s to have arisen from the failure to proofread o n e ' s copy. instead of merely predicating B of (all) A. suppose A is. In the most influential treatise on fallacies ever written. not only sufficient. then it is light. we have this curious n o n sequitur from the Stoics' c o m p e n d i u m : "Incoherent arguments are arguments which are invalid because there is no logical connection of the premises with one another or with the conclusion. s. capable of being separated" (Ebbesen 1981: 8). conversion applies validly to particular affirmatives (if some Fs are Gs. For whenever. W h e n c e the trap for the u n w i t t i n g victims of carelessness or sophistry: "The refutation which depends upon the consequent arises because people suppose that the relation of consequence is convertible." (Aristotle. the defining characteristic of the fallacy of consequence is the translation of a o n e . more generally. rain is not a necessary condition for the wetting of streets . (11) If p then q.w a y equivalence. but not a necessary condition. Suppose I can affirm that if and only if it rains.w a y implicational relation into a t w o . Horn /Journal of Pragmatics 32 (2000) 289-326 297 .) For Aristotle. E v e n the Socrates of the Platonic dialogues n e v e r tried a ploy quite that blatant. in fact. Rain is a sufficient condition for wet streets. Therefore. . he attributes the t e n d e n c y to affirm the c o n s e q u e n t as in (5a) to an inappropriate conversion..146. I can affirm that it has been raining. whereas that does not necessarily follow. B is equated to A. 1981: 27) Or. VIII. If it is day.not a mode of inference likely to tempt too m a n y d o w n the paralogical primrose path. But the conditional in the if-and-only-if interpretation is false: as a general principle. 2. But in fact wheat is being sold in the market. Math.R. but also necessary conditions. Or in the case when the streets are wet. 1961 : 82 and Ebbesen. when it has not been raining. R e t u r n i n g to the question of how we are to motivate CP or the corresponding fal- lacies. I can now affirm.the water-wagon may have washed them down . Dion is walking.

NEGATING ANTECEDENT AND [= A r i s t o t l e ' s c o n v e r s i o n ] CONSEQUENT [= CP] If p t h e n q.C. however. (7) a. many who behave in this way of whom the charge in question [of adultery] does not hold. since that is the way such people act. r e p e a t e d here) p e r t a i n s to w h e t h e r the l o g i c a l l y i n n o c u o u s m o v e o f c o n t r a p o s i t i o n has also applied. if q t h e n p. Sophistical refutations 167b10-13. b u t they usually do. If y o u ' r e look- i n g for a nice.298 L. It will be n o t i c e d that A r i s t o t l e ' s p r e s e n t a t i o n o f the fallacy o f c o n s e q u e n c e as an error in c o n v e r s i o n r e n d e r s this fallacy in a f o r m quite c l o s e l y a k i n to CP. the universe has not come to be. it does not also follow that what has a first beginning has come to be. Horn / Journal of Pragmatics 32 (2000) 289-326 This being the case. proofs based on signs also depend on consequences. based on the texts and consultation with informants. that w h a t is a d e d u c t i v e i n v a l i d i t y m a y well c o u n t as a r a t i o n a l step in a practical i n d u c t i o n . if n o t p t h e n n o t q.a n d .or a sociologist t r o l l i n g for data. c o u r t e s y o f M e l i s s o s (5th c e n t u r y B. Sophistical refutations. in the Rhetoric as well as the Sophistical refutations.R. I f p t h e n q. safe m a r r i e d m a n (not to be c o n f u s e d w i t h l o o k i n g for a h u s b a n d ) . someone who is hot must have a fever. therefore." (Fearnside and Holther. as Aristotle w o u l d n o d o u b t c a u t i o n . W e t streets m a y n o t always result f r o m rain.e.) a n d his p r o o f o f the e t e r n a l n a t u r e o f the u n i v e r s e . nothing follows from 'denying the antecedent' or 'affirming the consequent'." (Aristotle. and is therefore eternal. T h e o n l y d i f f e r e n c e b e t w e e n the A r i s t o t e l i a n a n d the G e i s . 167b13-20) 9 For reasons best left to the psychobiographers. suggests an a w a r e n e s s that these t w o i n f e r e n c e s are i n d e e d alternate v e r s i o n s o f the fallacy o f c o n s e q u e n c e : "Melissos's argument that the universe is eternal assumes that the universe has not come to be (for from what is not nothing could come to be) and that what has come to be has done so from a first beginning. a l t h o u g h .b . y o u r p i c k . any more than it follows that if someone who has a fever is hot. A r i s t o t l e ' s n e x t e x a m p l e . it has no first beginning. c h e c k i n g out the bars for a spiffy l o o k i n g f e l l o w w i t h a r o v i n g eye m i g h t n o t be a b a d place to start. it m a y a c c o m - p a n y most i n s t a n c e s o f Q in the actual world. There are." (Aristotle. If.Z w i c k i a n m o d e s (i.s t o r e ring . see also Darner. . The translation is mine.u p m a y b e a s i n g l e g u y with a d i m e . t h o u g h . P m a y thus serve as a ' s i g n ' o f Q:9 "In rhetorical arguments. 1959: 155. T h e r e f o r e . T h e r e f o r e . 1995: 128-129) O n e p l a u s i b l e m o t i v a t i o n for this c o n v e r s i o n is the fact that e v e n w h e n the P o f the i f P then Q c o n n e c t i o n is n o t a l o g i c a l n e c e s s i t y for the truth of Q. 7 a . Aristotle was so fond of the case of the unjustly accused adulterer he used it twice. CONVERTINGA CONDITIONAL b. I follow George Kennedy's (1991) Oxford Rhetoric in using 'cruises' as the gloss for the Greek xXav~t'~at here since it's appropriately neutral as to sex of cruisee. or in A r i s t o t l e ' s terms a p r o o f b a s e d o n signs. Rhetoric 1401b23-24) Note. But this does not follow: for even if what has come to be always has a first beginning. as for those who seek to show that a man is an adulterer because he wears fancy clothes and cruises at night.

for people take what happens later as though it happened because of what preceded.k. it is not an animal. at least one prominent scholastic does explicitly treat (7a-b) as free vari- ants of the same fallacy. If it is not a man. If he is sleekly dressed. 7b) (Peter of Spain. Tractatus VII: 158-59 (1972: 171-172. x has a first beginning. for the war took place after it." (Aristotle. he is an adulterer. 7a) While (7b) may not appear as a straightforward syllogism in the classical fallacist literature. (cf. (cf. Rhetoric. Pope John XXI)." . Aristotle moves swiftly to the fallacy o f non-cause as cause. 7a) c. he is sleekly dressed. For the thirteenth century logician Peter of Spain (a. L. with his dark asides during the 1988 campaign about all those ' D e m o c r a t wars' . (cf. If he is an adulterer.) The m o d e m reader m a y discern a contemporary Demades in Bob Dole. 7b) d. non est animal' a destructione antecedentis. (cf. he has neither earned nor borrowed it. for example when something has taken place at the same time or after [something else].R. it is an animal. x has come to be. est animal. henceforth P H E P H : "Another [fallacy] is taking a noncause as a cause. he has stolen it. all begun with a Democrat in the White House. and especially people involved in politics. to the fallacy of consequence runs as follows (Tract. the inferences in (13) .the first and third obviously adapted from the m a s t e r . (cf. VII. If it is not made. If it is a man. and lexical. it has no source. it has a source. If he has stolen something. 1401b30ff. for example Demades [regarded] the policy of Demosthenes as the cause of all evils.all illustrate falsa consequentia. The inductive.a. 1990: 151-152)) After exposing the fallacy of consequence in the Rhetoric with an assist from the man taken in alleged adultery. If he has neither earned nor borrowed it. lines 18-22): "Hic est consequens: 'si est homo. 7b) If x has a first beginning. import of the treatment o f sufficient conditions as nec- essary ones is nicely brought out in a passage from De Morgan (1847) addressing PHEPH: to The original of Peter's assimilation of 'destruction of the antecedent'. If x has not come to be. (cf. 7a) b. If it is made. ergo si non est homo. and more specifically the post hoc ergo propter hoc. x has no first beginning. 159. ~° (13) a. as in (13d). Horn / Journal of Pragmatics 32 (2000) 289-326 299 Aristotle here portrays Melissos as deriving two interchangeable invalid conclusions from one (hypothetically) valid premise: (12) If x has come to be.t h e two World Wars and the Korean and Vietnamese wars.

or at least incipient. ! is black Crow No. n is black' Therefore. it m u s t be a duck. the tendency to give as examples of the Fallacy of the Consequent instances of what might be construed as valid. E n g l i s h c o n t r i b u t e s the shifts illustrated in (14): (14) if P then Q Q. there is a probability (the amount of which depends upon the number of instances observed) that the removal of one would be the removal of the other. or ' b e c a u s e ' as w e l l a l o n g w i t h the p u r e l y t e m p o r a l ' w h e n ' .and c e l e b r a t e d in l a r g e l y r e l i a b l e but l o g i c a l l y i n v a l i d rules o f t h u m b o f the ' W h e r e t h e r e ' s s m o k e . When the level of a billiard table is not good. S o m e A m e r i c a n r e a d e r s o f a c e r t a i n age m a y r e c a l l f r o m the R e d S c a r e era the o l d trade u n i o n a d a g e on h o w to spot a c o m m u n i s t .. which is often illustrated by the words post hoc. Thus we say that a tree throws a shadow. 1847: 268-269) D e M o r g a n ' s c a s e c a n be s u p p o r t e d by a d d u c i n g e v e n c l e a r e r l i n g u i s t i c c o r r e l a t e s o f P H E P H f r o m the p r o v i n c e o f d i a c h r o n y . T h e thin line b e t w e e n d e d u c t i v e f a l l a c y and i n d u c t i v e p r o o f s u r v e y e d by A r i s t o t l e and D e M o r g a n . without intent to assert real connexion.5 6 6 ) Q follows from P Q is a consequence o f P e v e n if P." (Hamblin.300 L. ' w h i l e ' .. there is frequently an assumption of necessary connection. still Q S i m i l a r l y . The idioms of language abound in it. Horn / Journal of Pragmatics 32 (2000) 289-326 "The non causa pro causa.. 1970: 47) W e h a v e o n l y to shift a v i a n species to r e i n f o r c e H a m b l i n ' s point. t h e r e ' s f i r e ' f a m i l y ." (De Morgan. [Whence] . This is the mistake of imagining necessary connexion when there is none . ' a l t h o u g h ' .. When things are seen together. all crows are black as examples of conditionally valid inductions and in another we are given comparable arguments as uncon- ditionally fallacious. to describe that it hinders the light. . A particular case of this fallacy. make their mere expressions of phenomena attribute them to apparent causes. the favoured pocket is said to draw the balls. a presumption of connexion: if A and B have never been seen apart. since P (cf. ergo propter hoc. of course. a p r i n c i p l e W a l t e r R e u t h e r w a s f o n d o f i n v o k i n g in the late 1 9 4 0 ' s : (15) I f it w a l k s like a d u c k and q u a c k s like a d u c k . 1971 : 5 6 5 . G e i s and Z w i c k y .R. is the conclusion that what follows in time follows as a consequence.is also s u r v e y e d by H a m - b l i n (1970): "Logicians are in the habit of presenting induction as an argument from particular to general in such a way as to guarantee that it commits the Fallacy of the Consequent. or 'after'. the L a t i n c o n n e c t i v e cum a c q u i r e d the l o g i c a l r e l a t i o n s o f ' s i n c e ' . There is. 2 is black Crow No. inductive probability might be said to begin. that is. It is when there is only one instance to proceed upon that the assumption falls under this fallacy. In one chapter of a textbook we are shown schemata such as Crow No. were there but two. inductions.

I take to be a very common form of error . Yet hardly anything is more common than for people. inductive signs o f d u c k e r y . The proper converse of an hypothetical proposition is this: If the consequent be false. 53). As when the conclusion is accepted. Sommers (1982). with the e m o t i v e i n v o l v e m e n t of those perform- ing the c o m p u t a t i o n (affective or ' s o c i a l l y c o n t r o v e r s i a l ' material p r o d u c i n g m o r e instances o f fallacious reasoning). see also Horn (1989: 473ff.. L. Hall of Fame outfielder for the St. and with the cognitive load o f the p r o b l e m .R. by no means holds good. Lefford (1946). as rec- ognized by p s y c h o l o g i s t s investigating the circumstances in which All A s are Bs is taken to license the inference o f its converse. e s p e c i a l l y in light o f the m o v e a p p a r e n t l y first suggested b y John W a l l i s (1687: ii. ~ The reference is to Joe 'Ducky' Medwick. see W i l k i n s (1928). often c o r r e s p o n d to our e x p e r i e n c e o f reality. or b a l l p l a y e r s II who acquired the rele- vant habits by o s m o s i s or genetic predisposition." (Mill. But this is not too surpris- ing. the two p r o p o s i t i o n a l forms are o b v i o u s l y semantically close.. therefore all B are A. and W a s o n and J o h n s o n . 1843: 803-804) As with C P itself. W a s o n (1964). to draw this inference. 10) on which ' i f ' is assimilated to ' e v e r y case in w h i c h ' . If the con- sequent be true.but then are all adulterers sleekly d r e s s e d ? ) . this is an inductively natural. in their private thoughts. swans. perhaps . ~2 Wallis's equation of si to omnis casus quo has more recently been exploited in one form or another by Kratzer (1978). Horn /Journal of Pragmatics 32 (2000) 289-326 301 Even if we assume that a particular m a n n e r o f w a l k i n g and talking is c o m m o n to all ducks (a questionable assumption. It m a y have been noticed that we have shifted here from the illicit c o n v e r s i o n o f conditionals o f (15) to the illicit c o n v e r s i o n o f universals. but is an error corresponding to the sim- ple conversion of an [sic] universal affirmative.L a i r d (1972). although l o g i c a l l y invalid. . for proof of the premises .t r a v e l l i n g geese.. the antecedent is true. All A are B.. and yon Fintel (1997). 31 March 1975. for CP) is naturally v i e w e d as one arising in the mental processing o f universals: "The simple conversion of an universal affirmative proposition. by extension. these can only be plausible. if d e d u c t i v e l y invalid. And so with another form of fallacy. 12 Even for those unwill- ing to follow W a l l i s along this reductionist road. Louis Cardinals et al. Indeed. 1 am indebted to Kai von Fintel and Victor Szinchez Valencia tk~rdiscussion of this point. but this. there m a y well be f e l l o w . for Mill (1843) the illegitimate c o n v e r s i o n responsible for the fallacy o f c o n s e q u e n c e (and. d o w n to the inapplicability o f simple c o n v e r s i o n to each. step.). and being g u i d e d by e x p e r i e n c e are usually r e g a r d e d as justifiable p r o c e d u r e s " ( C h a p m a n and C h a p m a n . 1959: 224). the antecedent is false. who "walked with a waddle and swung with a vengeance" (Newsweek obituary. not substantially different from the preceding: the erroneous conversion of an hypothetical proposition. which it so often is. All collies are dogs). The c o n v e r s i o n o f ' E v e r y duck Q s ' to ' E v e r y Qer is a d u c k ' is as logically unsound as it is e m p i r i c a l l y natural. All Bs are As: " S u c h interpretations. from 1932 to 1948. with the binarity o f the situation (whether the antecedent represents one o f only two possible states o f affairs). My sources do not record any remarkable aspects of Mr. E m p i r i c a l studies indicate that this justification for such pro- cedures .along with that for d e n y i n g the antecedent or affirming the consequent - varies with the degree o f abstraction o f the data (All A s are Bs vs. not telltale. Medwick's elocution or politics.

1.to paraphrase Grice's memorable remark . every case in which P is true is a case in which Q is true) is 'per- fected' to the point at which Qness is neither smaller nor larger than Pness.A . March 1973. as Mill would have it. or B . in the Wallis translation. Every A is a B. (16) (16') a.A--B c. Rolled coins & large cash transactions are welcome inside the bank Inbbg. If P then Q. P l e a s e have identification readg. b. c. University of Texas. Every B is an A. 13 In the same way.e. Austin. P ~--~Q Fig. and Implicatures. b. 1. The universal and hypothetical forms of error. a. If Q then P.R. Fig. as we have seen.. 2 13 Delivered at the 'Performadillo' Conference on Performatives. Presuppositions. W h e n you begin working on these patterns in which sufficiency is not enough. Thus consider the following sign posted outside the drive-through window at m y branch of the Bank of Boston: Ige w i s h t n s e r u e all o f o u r customerspromptl U. Horn / Journal of Pragmatics 32 (2000) 289-326 The parallel Mill draws between the universal and hypothetical 'form[s] of error' can be schematized as in Fig. space in (16) is virtually reduced until the B set is function- ally coextensive with A. Thus the symptoms of CP or illicit conversion m a y lurk within a conditional (or. to the Sophistical refutations) of classifying the fallacies in question as extra dictionem or language-independent. or when no overt logical connective is present at all. 'an hypothetical') proposition.until Bness is neither narrower nor wider than Aness. i f P then Q (i. reflecting B6er and L y c a n ' s (1973) obser- vation that nothing crucial about CP (or the related fallacies) hinges on the form of conditionals. or . . within the corresponding universal statement. you begin to see them under every bush. The non-B A.302 L. not to mention the tradition (dating back.

The subjects in question essentially arrived at true con- ditionals of the types displayed in (19). The better (worse) x is. allowing us to convert the bank's regulation to what we shall refer to as CP-normal form: (17) a. and so on. The subjects were told that they could propose as m a n y triples as they liked in search of the correct rule. x is (must have been) good. Horn / Journal of Pragmatics 32 (2000) 289-326 303 Inside must be understood here as 'inside and only inside'. which they then incorrectly promoted to biconditionals. unfalsifiable systems such as "existentialism and the majority of religions". with the first line: (18) I ' m gonna live forever if the good die young. In the original task. b.but this time with a scalar twist. a majority of the subjects advanced incorrect rules representing specific subcases of the correct one. c.000. we have the conversion of (18'a) into (18'b). obsessional behaviour" characteristic of closed. The implicit argumentation here is somewhat more sophisticated. On a related note.000 deposit.000 deposit. the target rule was 'numbers in ascending sequence'. 1993). d. or even in the "fixated. but I will eschew the formal proof here. Beginning with the assumption presented in the postposed protasis of (18). x dies young. sub- jects were presented with the triple 2-4-6 and told that this sequence exemplified a rule or pattern that they were to determine. Fortunately. must differ from each other by two. x is welcome to make a $1.000. b. I suspect (18) remains technically unsound as a conclusion even on a generous read- ing of this schema. For one more venue in which sufficient conditions are strengthened to neces- sary-and-sufficient ones we turn to a family of conceptual tasks empirically inves- tigated by Wason (1960) and his followers and critics. my cash transaction was small enough to be administered externally. If x is good. If x dies young. As it happens. Iff x is in the bank lobby. 14 ~4 Wason later reflects (1968: 173-174) that his study demonstrates "how dogmatic thinking and the refusal to entertain the possibility of alternatives can easily result in error". must form an arith- metic progression. resulting as usual in the equivalence in (18'c) . x dies young. Atlantic #82483. Subjects were to identify the rule only when they were sure they knew it. x is welcome to make a $1.R. L. If x is in the bank lobby. . Iff x is good. typically assuming (in the absence of negative feedback) that the integers must all be even. as approximated in (18'd): (18') a. Neal Whitman has reminded me of the country music hit ' I f the good die young' by Tracy Lawrence (on Alibis. and that the experimenter would let them know if the proposed triple did or did not con- form to the rule. the younger (older) x dies.

q . A-B-C is accepted. song-writers. 1983: 247) Van der Auwera reconstructs this analysis by way of the now standard conversa- tional-implicature-inducing quantitative scales (Horn. the analysis of CP and its family of fallacies is a natural target for pragmaticists." (Cornulier. 1997a: 262-263). Using modem terminology. b. But which items are to be pulled from the pragmatic toolkit and how they are to be deployed? 3. 1972.B = B .304 L. Conditionals 'R' us: CP and pragmatic strengthening "It is undoubtedly true that Geis and Zwicky are responsible for quickening linguistic interest in condi- tional perfection. A-B-C is accepted. given the diffi- culty of establishing causal correlations. If [C>B>A and B is the mean of A and C]. we have surveyed the explanations proffered for the tendency to con- vert or perfect conditional statements and their functional kin." (van der Auwera. Hirschberg. C. and bankers alike. He moreover does not merely envisage a Gricean analy- sis but surpasses Geis and Zwicky in actually proposing one. d.A] (i. Fauconnier.implying or inferring that a formal impli- cation is to function as an informal equivalence . A-B-C is accepted.e. Since Aristotle. its application is subject to the vagaries of context and in particu- lar to the beliefs and goals of speaker and hearer. This merit goes to the French linguist Ducrot (1969)... q if p. q and if s. whose job description includes the explanation and delimitation of defeasible inference. if A-B-C form an arithmetic progres- sion). 1979. very roughly. While the general practice of strengthening a sufficient condition into a necessary one . this tendency has often been ascribed to the fact that the deductive invalidity of arguments resulting from this tendency is offset by their inductive naturalness. 1997b: 170) For van der Auwera (1997b: 173-174. which though incorrect is definitely on the right track . Horn / J o u r n a l of Pragmatics 32 (2000) 289-326 (19) a.R. B.is reflected in realms from rhetoric to empirical psychology to lexical diachrony and exploited by sophists. If [C>B>A and A. q q'if p. and that the utterance situation suggests that if other sufficient conditions (allowing to sit there) did exist. q and if r.. In this respect. 1985): (20) if p. the 'correct' scalar analysis derives from a suggestion presented informally in Horn (1972: 107-108) and semi-formally in Cornulier (1983: 247-248): "We can suppose. they would have been men- tioned. that in [One is allowed to sit in this seat if one is disabled or if one is older than 70] the word if keeps its merely sufficient condition meaning. one would now say that Ducrot derives [2b] or [2c] as a scalar conversational implicature. But they were not the first to introduce the problem into modem linguistics . A-B-C is accepted. so that the only mentioned property (to be disabled or older than 70) is the only property which gives the right to sit there (presumption of exhaustivity).. q and if r. If [B = A + 2 and C = B + 2]. C are even]. 1975. If [C>B>A and C . In this condensed retelling of the family saga of interrelated non-language-depen- dent fallacies. Gaz- dar.

claim" (Strawson. why assume the scale in (20) rather than the more straightforward one in (22)? (22) if and only i f p ... 1984) is dismissed by van der Auwera as unconvincing. W .. (ifp then q) here communicates not the negation o f ' .. Atlas and Levinson. as when an assertion about some Fs being Gs is taken to implicate that (for all the speaker knows) not all of them are..although these conditions never seem to figure directly in the reasoning that takes us from i f p then q to if not-p then not-q. If .. 1981.. a 'pragmatic' consideration.... q (iffp..P. 1984) determined by the scales in (21) and (21').g. unilaterally entails . or more generally how the assertion of c~ . L.or Q-based implicature as in Horn. S ... a general rule of linguistic conduct . The response offered by Levinson (1987: 69-71) and others (e. It will be noticed that the scale invoked by van der Auwera in (20) explicitly refers to propositional variables for hypothetical states of affairs to be excluded - Cornulier's 'other sufficient conditions' .. As we shall see. H..R. 1952: 179. the key step is that of showing "how one can stop if from Q-implicating the negation of if and only if'. S . that one does not make the (logically) lesser.k. But this seems to yield just the opposite prediction from the one we want. Strawson's parenthetical rider will play a crucial role in our retelling of the CP story.' itself (iffp then q)? As van der Auwera (1997b: 175) notes. emphasis added). . but this reac- ..' as in (21) and (21') but rather '. while the assertion of a lower or weaker proposition implicates that the speaker was not in a position to have substituted a higher or stronger proposition salva veritate (implicature proceeds upward from W to the negation of S)... how is it that . one might ask.. For that matter.. W . Grice') attributes the inferences in question to ". quantity.. Horn. While this principle is not novel with Grice (see Horn. q To be sure. when one could truthfully (and with equal or greater linguistic economy) make the greater. (21) All of them left (21') None of them left $Some of them left "['Not all of them left The motivating principle at work here is most familiar in the form of Grice's first maxim of quantity (1989 [1967]: 26-27): "Make your contribution as informative as is required (for the current purposes of the exchange)"... the biconditional in the upper line does unilaterally entail the one-way conditional in the lower line. Horn / Journal of Pragmatics 32 (2000) 289-326 305 In this ordered structure...a. q) "]'if p. Van der Auwera regards the pattern evinced in (20) as parallel to the classic instances of Gricean scalar implicature (a.). a proposition on a given line unilaterally entails any propo- sition on a lower line (entailment proceeds downward from S(tronger) to W(eaker) schemata).. can be blocked from implicating ~(a and only c~ . one particularly significant precursor of his maxim is due to a fellow Oxonian who (with acknowl- edgments to his colleague 'Mr.. . S . 1990 for discussion and references).

..) only if S is at least as lexicalized as W within the relevant domain. 1987: 71. . not those in which it is more lexicalized or briefer. S . 3 and Horn (1996) for arguments that if and only if is a semantically compositional coordination of if and only if. W .e.. while also largely technical. emphasis added) b. though not entirely successful. Kelley in which iff appears. 1993) that the factors determining when a potential Q-based scalar implicature will be overridden can be quite difficult to pin down.. there is a significant overlap between the considerations of brevity and lexicalization invoked by Levinson. Genuine Q(uantity)-implicatures [involve] tight Horn scales and similar contrast sets of equally brief.. Nor is it. But in fact the correct generalization is not the justly maligned (23a-b): (23) a. fails to hold (for all the speaker knows). van der Auwera rejects this argu- ment on the grounds that other instances of successful Q-implicatures do clearly dif- fer in degree of lexicalization (see also Matsumoto. S . (Levinson. 1981: 44. 1989. (where .. (Atlas and Levinson.. emphasis added) but something more in the nature of (24): (24) The utterance of . 1993). W . will (correctly) be barred from implicating that the stronger expression. 1995: 42ff. since . equally lexicalized linguistic expressions 'about' the same semantic relations.. logicians and linguists. but the generalization in (24). To constitute a genuine scale for the production of scalar implicatures.despite the existence of the strength or informativeness scale in (22) . each item must be lexicalized to the same degree.more generally - more marked as a contribution than W.). has a wider distribution than the abbreviation but is also clearly less lexicalized than if.iff is not as lexicalized as if.. i. This is shown by both reflection and a consultation of other pragmatic work in this spirit: see e. Thus if.. (containing an unembedded occurrence of a weak scalar value) will Q(uantity)-implicate that the speaker was not in a position to affirm . Van der Auwera takes Levinson and the other authors to claim "that the relevant linguistic items must be lexicalized to the same degree" (1997b: 175). 15 Now. 1990. . Matsumoto shows (see also Horn. or .. 16 The conditions under which Q-based implicature is constrained are those in which S is less lexicalized. unilaterally entails .306 L. 16 AS Matsumoto (1995: 44-45) observes.e. is informationally stronger than. The first citation in the OED is from a 1955 topology text by J. Horn (1984. i. as it does in most early instances.. to conform to the Maxim of Man- ner" (Levinson.. if and only if . necessarily idiomatic: see the McCawley and Barker references in fn.L...... . longer. The fuller if and only if. 1983: 135). Horn /Journal of Pragmatics 32 (2000) 289-326 tion is based on a misconception of the substance of that response.g. in the definition of technical terms. ~5 lff is obviously a self-conscious innovation within the vocabulary of mathematicians.R. inter alia. will serve the current pur- poses. as well as Levinson's actual argumentation for a 'Brevity Condition' to ensure that the speaker didn't choose W over S "simply in order to be brief.. 1989. as often assumed..

Thus. possible>. 1990: 464. you will tend to infer that I didn't know for a fact that the Democrats are corrupt.the value(s) on its left could not have been substituted salva veritate. OK> constitutes a Q-relevant scale such that the affirmation of any weak or intermediate value will implicate (ceteris paribus) that . <thumb. much less that both parties are). When degree of lexicalization is not a factor. 1955 and Fogelin. good. 1997b). 3. <and. But while the assertion of a disjunction does implicate that (for all the speaker knows) the singular expression and conjunction fail to hold (from my assertion that The Democrats are corrupt or the Republicans are corrupt.. sometimes>. <certain. entailment proceeds downward along the usual gravitational lines. and each of the two unilaterally entails the I-form disjunction. each of the ordered n-tuples of items in (25) (25) <always. that "the requirement of an equal lexicalization seems unduly severe" (van der Auwera. 1997b: 176) and that it is rejected in prac- tice by Levinson himself.for all the speaker knows . Newt. <excellent. a singular statement (Bill is corrupt) does not implicate anything about the character of. after Czezowski. . A E [strong values] ~//~p andq neitherp [ ' . warm>. <hot. 1989: 497ff. 1967) The A-form conjunction unilaterally entails its (umlauted) simplex counterpart. the fact that not all is less lexicalized than none strengthens the tendency to draw the upper-bounding (not all +> -none (= some)) scalar implicature here compared to its positive (some +> -all) counterpart where the items that differ in strength are equiv- alent in degree of lexicalization (see Horn. Horn / Journal of Pragmatics 32 (2000) 289-326 307 Thus it is true. 3. A~ ~ ~ p not both~¢] n o / ~ p 0 ues] [intermediateval or q p and q l / [wesk vsl uesl I O Fig. Significantly. My affir- mation of A fails to implicate that (for all I know) not-A. often. scalar implicature goes through. <cold. cool>. say. In the extended standard square of opposition for the binary connectives in Fig. as in the invocation for a Q-implicature from not all to the negation of 'none' in (21') above (24 in van der Auwera. . or>. but irrelevant. L. The extended standard square of opposition (Horn. likely. But when S is less economical than W in the appropriate sense. finger>. for a detailed discussion of the effect of this asymmetry in lexicalization on quantifier scope facts). no Q-implicature is triggered even when unilateral entailment provides the necessary strength differ- ential. usually.R.

you'll pass'.) . c. a fortiori. K i m won. get the answers ahead of time. etc. the use of finger m a y c o n v e y ' n o n . John ate the BEANS. to affirm the truth o f 2 + 2 = 4 is not to i m p l i c a t e that this truth is contingent. (26) a. whence the exhaustiveness effect in clefts. While (i) does not Q-implicate that John ate nothing besides the beans. M a n n e r does matter. its focal counterparts in (ii) do: (i) John ate the beans.you don't cheat and you study hard. but so too is any scale of the form in (28) or. M a t s u m o t o . "]'Possibly. I hurt m y finger -/-> N[I hurt m y p i n k y (finger)] c. c. 1995). As suggested by the parallel b e t w e e n (26) and (26'). b. not only are (20) and (22) barred as direct sources for scalar Q . is the structural a n a l o g u e o f the thumb. (ii) It was the BEANS that John ate.t h u m b ' given the existence o f thumb as a functional alter- native tofinger (cf. What John ate was the BEANS. K i m won. K i m won. I hurt m y toe -/-> . Necessarily. nor does the use o f toe c o n v e y ' t o e other than the big t o e ' . W h a t is crucial is the status o f thumb (as o p p o s e d to pinky and big toe) as a viable lexical- ized alternative to finger. If I tell you that 'if you study hard.308 L. K i m won and Chris won.d r i v e n a s y m m e t r y obtains for m o d a l statements. an assertion that q0 is p o s s i b l y true does i n d e e d Q . copy your neighbor's answers. Rather. (26') a. ~8 As noted in Horn (1981). G i v e n the p r o p e r i m p l e m e n t a t i o n o f the principle in (23'). although the big toe. and bare focus constructions.i m p l i c a t e that for all the s p e a k e r k n o w s q~ is not true (much less n e c e s s a r i l y true). K i m won. pseudo-clefts. I don't implicate that you'll pass whateveryou do (e. the assertion of p will implicate the negation of p and q within focus con- structions. ~7 Once again.R. but an assertion o f q0 does not itself i m p l i c a t e that (for all the speaker k n o w s ) it is not n e c e s s a r y that q0. This point extends to non-quantitative ' s c a l e s ' o f items differing in informative strength: (27) a.[ I hurt m y big toe] A l t h o u g h thumbs are fingers (or w e ' d all have eight fingers instead o f ten). But the use of finger does not c o n v e y ' n o n - p i n k y (finger)'.(John ate) the beans. ] ' K i m won or Chris won. in the following sense. Now it might be claimed that our CP inferences involve similar focal contexts. pace van der A u w e r a . the asymmetry of (27a-c) would instantly vanish. In the same way. what I implicate is that within the focus set of alterna- tives .studying is . you don't cheat and you don't study hard .g. an assertion that I saw the movie will implicate that I didn't read the book precisely when it comes as a response to the question 'Did you read the book?'. [What did John eat?] . I hurt m y finger -Q-> N[I hurt m y thumb] b.this is in fact a problem for both the Cornulier line represented in (20) and the unconditional line cited below. qua p l u m p inside digit.b a s e d implicata. b. o f that in (28'): ~8 17 We would predict that if the colloquial language replaced its thumb with the polymorphous poller (the Latin and scientific English term for both 'thumb' and 'big toe'). Horn / Journal of Pragmatics 32 (2000) 289-326 The same M a n n e r .

may be motivated by cultural factors that conspire to avoid the direct expression of the stronger and more relevant conditional. in (29'). The circumlocution involved here. this is the very a p p r o a c h that van der A u w e r a (1997a: 2 6 8 . (29) q (29') w h a t e v e r the case. 1989: 192ff.b o u n d i n g scalar implicature. on Malagasy cultural diffidence. the latter is the UNCONDITIONAL o f Z a e f f e r e r (1991). Manner. q "[" if p. as in other cases of implicature generation. evidently employed precisely to avoid the ethical or legal commitment that would have accompanied the corresponding positive conditionals: (i) A: How does one open this door? B: If one doesn't open it from the inside. In any case. ~9 There may he additional societal reasons for presenting a causal relation as a sufficient condition. Thus consider the negative conditionals in (i) and (ii). also Karttunen. it is certainly a natural one: " B y stipulating P as a sufficient c o n d i t i o n for Q.). cf. L. for commentary) (ii) If it doesn't say LISTERINE it isn't clinically proven to prevent gingivitis. (Thanks to Johan van der Auwera and Scott Schwenter for helping to clarify my understanding of these matters. it w o u l d s e e m to be that in (29) or. although it would still prevent us from captur- ing the generalizations made possible by the R-based account defended here. alternatively.2 6 9 ) represents in the left-most c o l u m n o f his figure. (from label on Listerine Antiseptic) 20 The Q and R Principles alluded to here are the two driving forces in a dualistic functional model I have proposed (Horn.) A revision of the Cornulier approach that takes off from this point is worth pursuing. as noted above. Horn / Journal of Pragmatics 32 (2000) 289-326 309 (28) p andq (28') p and only p q'p "]'p In the terms o f M a t s u m o t o (1995: 46ff. while he takes the strengthened conditionals in (20) to constitute POSITIVELY ENHANCED stronger statements whose negations can in principle be i m p l i c a t e d by the c o r r e s p o n d i n g simple conditional. that such p r o p o s i t i o n s are i n v o k e d here as dei ex machina to assure a successful scalar account. 1971: -68: " W h a t w o u l d be the point o f stating a condition if it was not a necessary c. 1987) as a reduction and generaliza- .i m p l i c a t a in the case o f conditionals. also Levinson. also Prince. NEGATIVELY RESTRICTED stronger state- ments like those o f (22) and (28') n e v e r figure in u p p e r . w e implicitly suggest that P is a necessary condition as well . 1984. and/or the R Principle) 2° m u s t be taken to be (ceteris paribus) a necessary and sufficient condition for passing. the point remains.. ndi- tion?")19 O b s e r v e that it is indeed stronger (in both intuitive and e n t a i l m e n t . 1982. c o r r e s p o n d i n g to what he considers the ' i n c o r - r e c t ' scalar model. the door won't open.b o u n d i n g Q ..R. it contains extra infor- m a t i o n w h i c h (by R e l e v a n c e . q Indeed. q) "['if p. Not only is the c o n d i t i o n a l assertion o f i f p then q w e a k e r than the b a l d assertion o f q. w h y m e n t i o n i t ? " (Horn. 1993. q (whether p or not-p. cf.d e f i n e d senses o f the term) to assert q than it is to assert i f p then q. leaving its status as a necessary condition to be recovered pragmatically. (Keenan 1976: 71.else. cf. If any scale is appropriate for generating negative or u p p e r . 1973: 212.

2~ The characterization of CP in Horn (1973) appears within a discussion of various fallacies repre- senting what are clearly non-Q-based inferences in ordinary reasoning. I1 serait con- sid6r6 en effet comme assez anormal. The functional tension between these two antinomic forces motivates and governs not just implicature but a wide range of linguistic phenomena from politeness strategies to the interpretation of pronouns and gaps.and R-based accounts of the following decade. Si P i e r r e ne v i e n t pas. 1981. The R Principle is an upper-bounding correlate of the Law of Least Effort dictating minimization of form ('Say no more than you must. que le d6part de Jacques est subordonn6 b. b. calling into question van der Auwera's implication that I was later seduced away from my earlier Q-based insight by the wiles of (Atlas and) Levinson. The lexical strengthening of expressions like since has been cited elsewhere (Atlas and Levin- son. including the PHEPH-style infer- ences responsible for the logical sense of originally temporal expressions like English since and still (see (14)) or the interpretation of The flag is red as 'The flag is all red' (Horn. Q u a n t i t y and R e l a t i o n / M a n n e r c o n s p i r e to e s t a b l i s h the t e n d e n c y to p e r f e c t c o n d i t i o n a l s . K6nig and Traugott. the second Quantity maxim. corresponds to the Fallacy of Com- position. A n d w h a t c o u l d m a k e s u c h a c o n d i t i o n m o r e r e l e v a n t than its n e c e s s i t y ? O n this v i e w . and the last two submaxims of Manner (in particular. A s w i t h n e g a t i v e s . . Horn / Journal of Pragmatics 32 (2000) 289-326 r e l e v a n t to the s p e a k e r ' s c o n t r i b u t i o n . ou comme mensonger. I have not had the opportunity to consult this work. qu'elle enest la condition suffisante. it collects the first Quantity maxim and the first two submaxims of Manner. 1989: 201).t o g e t h e r w i t h the R - b a s e d p r i n c i p l e r e s p o n s i b l e f o r the s p e a k e r ' s e l i m i n a t i n g f r o m h e r m e s s a g e w h a t e v e r w o u l d i n c r e a s e p r o c e s s i n g e f f o r t w i t h o u t i n c r e a s i n g r e l e v a n t c o n t e n t ( H o r n . 22 Scott Schwenter (p. the m a r k e d status o f n e g a t i o n is d e r i v e d f r o m the the Q .3) that the pre- s u p p o s i t i o n a l i t y o f n e g a t i v e s t a t e m e n t s results f r o m the m u t u a l r e i n f o r c e m e n t o f Q and R. d'6noncer [30a] si l'on ne pense pas [30b]. R-based implicature (or the I-implicature of Atlas and Levinson. Horn.w h e r e p o s i t i v e s t a t e m e n t s are c h a r a c t e r i s t i c a l l y m o r e i n f o r m a t i v e t h a n n e g a t i v e o n e s . and is systematically exploited (as in scalar cases) to generate upper-bounding implicata." (Ducrot. euphemism) and negative strengthening. 2j N o r w o u l d this be u n p r e c e d e n t e d : it is a r g u e d in H o r n ( 1 9 8 9 : §3. J a c q u e s n e partira pas.R. 1969: 17) tion of the Gricean schema. Q-based implicature is typically negative in that its calculation refers crucially to what could have been said but wasn't: H infers from S's failure to use a more informative and/or briefer form that S was not in a posi- tion to do so. 1984.b a s e d r e q u i r e m e n t that s p e a k e r s be as i n f o r m a t i v e as p o s s i b l e . from lexical and semantic change to the analysis of conversational interaction. Si P i e r r e vient.) has alerted me to the fact that in more recent work. J a c q u e s partira. Ducrot (1990: 133) has rejected his earlier implicature-based approach to CP in favor of one based on argumentative topoi. la personne qui entend [30a] conclut non seulement que la venue de Pierre entraine le d6part de Jacques. 1981. modulo Quality and R'). Levinson. mais aussi qu'elle en est la condition n6cessaire. This last move. the predication of the part taken as predication of the whole. In p a r t i c u l a r . modulo Q'). 1973: 212-213). la venue de Pierre.or R-based implicature. and is exploited to induce strengthening or lower-bounding implicata. so w i t h c o n d i t i o n a l s .31 o L. T h i s line o f a r g u m e n t is p r e f i g u r e d by O s w a l d D u c r o t : 22 (30) a. 1987) typically involves social rather than purely linguistic motivation and is exemplified by indirect speech acts (in particular. "Dans la plupart des contextes imaginables. Thus the analysis of CP in Horn (1973) foreshadows the I.c. it collects the Relation maxim. 1988) as instances of conventionalized and 'on-line' I(nfor- mativeness). the submaxim of Brevity). The Q Principle is a lower-bounding hearer-based guarantee of the suffi- ciency of informative content ('Say as much as you can.

emphasis added) Note the o p p o s i t i o n o f sentence m e a n i n g and s p e a k e r meaning.. was not in a posi- tion to have affirmed that it holds for tous . not because the words mean it. as in (30'). 5) for an R-based account of such pragmatic strengthening processes. supposant que le locuteur a respect6 cette r~gle.. If I say to any one. . e. D u c r o t recognizes that the relation in (30) is anal- o g o u s to that i n v o l v i n g litotes or understatement. Jacques partira? T h e kinship b e t w e e n D u c r o t ' s s o u s . Horn / Journal of Pragmatics 32 (2000) 289-326 311 A n t i c i p a t i n g G e i s and Z w i c k y ' s discussion. les renseignements les plus forts qu'il poss~de. L.b o u n d i n g infer- ence from W to ' n o t S ' (as in that f r o m s o m e to ' n o t a l l ' ) but the strengthening infer- ence from W to ' S ' . 23 as e x e m p l i f i e d by the t e n d e n c y to 23 See Horn (1989: ch.e n t e n d u . the inference c o m - puted f r o m w h a t w a s n ' t said as well as w h a t was.e n t e n d u s . 1972: 134) D u c r o t ' s s o u s . h o w e v e r . Here is John Stuart M i l l writing in o p p o s i t i o n to Sir W i l l i a m H a m i l t o n o f E d i n b u r g h ' s definition o f s o m e as ' s o m e only. D u c r o t elaborates a loi d ' e x h a u s t i v i t ~ that c o r r e s p o n d s to a quantity m a x i m in its f o r m u l a - tion and its e x p l o i t a t i o n as a source o f u p p e r . 1867: 501.e n t e n d u and G r i c e ' s implicature is evident..all traits stressed a century later by both D u c r o t and Grice.e n t e n d u has a c o n s i d e r a b l y richer lineage within the t a x o n o m y o f non- logical inference than does its A n g l o p h o n e rival .. but had not yet a p p e a r e d in print). especially those occurring in negative contexts. although the descriptions were arrived at independently.S. si Pierre vient.. c'est qu'il n'y a que cela).. "I saw some of your children today". In other w o r k from the same p e r i o d (the years w h e n G r i c e ' s W i l l i a m James lectures were circulating in m a n u s c r i p t form in Britain and the U.. but because. si la rrserve du locuteur ne peut pas ~tre attribure une absence d'information. h interprrter route affirmation restreinte comme l'affirmation d'une restric- tion (s'il ne dit que cela. the infer- ence that a s p e a k e r affirming that a p r e d i c a t e holds for c e r t a i n s . Le destinataire. which does not occur with semantic inferences.. he observes. the t e n d e n c y is attenuated or s u p p r e s s e d when the c o n d i t i o n a l appears within the scope o f another illocutionary operator. s o m e but not a l l ' : "No shadow of justification is shown . D u c r o t sees the inference from (30a) to (30b) as an instance o f the p r a g m a t i c relation o f s o u s . and the role o f e p i s t e m i c uncer- tainty in d e t e r m i n i n g the force o f the inference . : "[La loi d'exhaustivit~] exige que le locuteur donne. et qui sont susceptible d'intrresser le destinataire . i n v o l v e s not the u p p e r .b o u n d i n g s o u s . alors qu'il sait ce qui s'est passr.R. (30') Est-ce que. if I had seen them all. In particular." (Ducrot. sur le th~me dont il parle. it is most likely that I should have said so: even though this cannot be presumed unless it is presupposed that I must have known whether the children I saw were all or not. for adopting into logic a mere sous-entendu of common con- versation in its most unprecise form.g. T h e c o n d i t i o n a l s o u s . he might be justified in inferring that I did not see them all.even in the A n g l o p h o n e literature itself.e n t e n d u . F o r one thing." (Mill. aura tendance.

~ quoi bon ajouter cette d6termination .. pour s'expliquer la pr6cision apport6e par le mot 'policiers'. Horn / Journal of Pragmatics 32 (2000) 289-326 interpret the negative assessment in (31a) as an expression of the stronger negative j u d g m e n t in (3 lb). il est cens6 ne pas parler en vain. ou sim- plement s'il devait partir m~me au cas o~ Pierre ne viendrait pas. x ~ novels) (If x is a detective novel. (31) a. 1988). d'autre part. une rbgle de convenance s'oppose." (Ducrot. 1969: 22) In both (30) and (31). Si X et X' sont deux 6nonc6s situ6s sur une m~me 6chelle de signification. 1969: 22) W h e n c e . the prescriptivist tends to find such shifts illogical. et si. to posit any restriction will tend to suggest that the general predication holds not just when.312 L. must be assumed to be relevant. dans le choix de son 6nonc6. je peux conclure avec une certaine vraisemblance que. but in (30) there is in addi- tion the role played by Manner. 1969: 22-23) As we have also seen. but only when. les autres romans. l'auditeur qui entend X a tendance h l'interpreter comme X'." (Ducrot. ou peut sembler s'opposer ~ l'emploi d'X.. the strengthening of sufficient to necessary conditions applies in a wide range of cases that carry conditional force but often lack condi- tional form. "L'auditeur ne cherche une litote que lorsque l'utilisation d'un 6nonc6 plus fort aurait quelque chose de d6plac6. si Jacques devait partir de toute fa~on.. Si on dit d'une per- sonne qu'elle aime les romans policiers. si elle aimait tousles romans. K 6 n i g and Traugott. et si le second ne diff~re du premier que parce qu'il occupe un degr6 sup6rieur de cette 6chelle. but this verdict does not stand in their way: . Agatha likes x) b. pour lui. x is a detective novel) c. une esp~ce de loi d'6conomie. ou moins. then. l'auditeur est tent6 de conclure. Jacques ne d6teste pas le vin." (Ducrot. Si mon interlocuteur a tenu a subordonner l'6noncia- tion du d6part h l'6nonciation de la venue. the fact that the speaker seems to have gone out of her way to provide additional material which. Mais. as we have seen. D u c r o t ' s fan of detective novels can be adjusted to fit our Procrustean bed: (32) a. there is the ' W ' +> S move.R. x ~ novels) (Iff x is a detective novel. Thus. x ~ novels) (If Agatha likes x. on ne pose le d6part de Jacques qu'apr6s avoir demand6 ~ l'interlocuteur de faire l'hypoth~se pr6alable de la venue de Pierre. in which an expression gradually comes to refer to a salient subset or m e m b e r of the set of enti- ties to which it previously referred. (Vx. Agatha likes x) A particularly rich vein of strengthened sufficiency is provided by the lexical process of R-based narrowing (Horn. 1984. d'autre part. Jacques aime beaucoup le vin. the CP effect: "Lorsque l'on utilise [30a]. le fait du d6part est subordonn6 au fait de la venue. Car.. (~/x. de r6pr6hensible. b. this restriction is satisfied: "Le locuteur observe. qu'elle aime peu. As with the related p h e n o m e n a of CP and PHEPH. given the operation of linguistic econ- o m y (our R Principle). d'inconvenant. Unsurprisingly. si. (Vx. a quoi bon subordonner l'annonce de son d6part ~tl'6vocation de l'arriv6e de Pierre? .

and animal (excluding humans. If x is a person. where A +-. fenum 'produce' --~ ' h a y ' and Greek dlogo(n) [lit.R. d'inconvenant. x is from the earth [humus]. and/or fish). because man is born of the earth. b. and grays)." (Br6al. excluding black. in which the general sense survives in privative opposition with a specific and stronger sense or use. intransi- 24 Notice that in several of these cases there is a specialized term already in existence. Ger.is most likely when "l'utilisation d ' u n 6nonc6 plus fort aurait quelque chose de d6plac6. Illustrations include the par exemplar narrowing of Lat. is in the pattern of R-based. 1897:114) Putting Br6al's derivation into CP-normal form yields (33). Tier) and hound (cf. inter alia disease. it is not surprising that the night-soil of euphemism should prove especially fertile for the development of R-narrowed meanings: cf. Frau/femme/mujer ( ' w i f e ' < ' w o m a n ' ) . however. In Q-based. 'speechless one'] ~ ' h o r s e ' . In both euphemistic and general domains. the R-based narrowing m a y result in development o f AUTOHYPONYMS. the originally argot-based restriction of deer (cf. Ger. positive narrowing to a salient or stereotypical exemplar. and toilet.. and the loaded senses acquired by poison (cf. so a Q-based narrowing of e. Recalling D u c r o t ' s observation (1969: 22) cited above that speaker's understatement . temperature or number would have yielded such complementary restricted senses as 'a . L. Examples include the development of specific uses or senses o f dog (excluding bitches). 24 A m o n g the euphemisms. as if all animals had not the same origin?' Yet it is most certain that homines did signify 'the inhabitants of the earth'. We know the observations of Quintilian on the subject of homo: 'Are we to believe that homo comes from humus. rectangle (excluding squares). liquid). since earth- ling instantiates precisely the same R-based narrowing effect. Dii or Superi. Hund) to animals constituting a hunter's salient goal and assistant. undertaker. Thus we have color (for ' h u e ' . number (for 'inte- ger'). the target is the complement o f a previously named subset: the existence of a specific h y p o n y m H o f a general term licenses the use of the general term for the comple- ment of the extension o f H.B signals that a term once denoting ' B ' has been extended to refer to ' A ' (either instead of or in addition to ' B ' ) : (33) a.g. cow (excluding bulls). pass away. we have drink (for 'imbibe alcohol'). x is a person. de r6pr6hensible". spe- cial education. Horn / Journal of Pragmatics 32 (2000) 289-326 313 "Restriction of meaning has at all times been a cause of astonishment to etymologists. The process illustrated in (33) represents one o f two distinct sources of narrowing within diachronic lexicography. whites. Our interest. x is from the earth. It was a way of opposing them to the inhabitants of the sky. temperature (for 'fever'). birds. linguistically motivated narrowing. c. [homo] ( / N o r should this development seem terribly exotic to English speakers.or bearer's strengthening . finger (excluding thumbs). Iff x is a person. If x is from the earth. man/homme (for 'male adult h u m a n ' ) . accident. potion) and liquor (cf.

Though we must have made love a thousand times. as soon as one can speak of a dog going to the bathroom on the kitchen floor. as in 35) that when two people are sexually engaged. If x makes love with y. b.with the appropriate pause and/or circumflex eyebrows . Historically. Horn / Journal of Pragmatics 32 (2000) 289-326 tive smell (for 'stink'). i. Iff (and whenn) x excretes. Iff x makes love with y. (35') a. x [literally] sleeps with [= alongside] y. we have the narrowing that results when what is "d6plac6. from a sufficient to a necessary condition for bathroom visits. x sleeps with y.just as a straightforward Q-based strengthening of ifp then q would have resulted in the reverse CP effect noted earlier.314 L. we have the following pattern. we avoid invoking the loaded meaning in a temperature other than that in the fever range' and 'a non-integer' respectively. For we were just young. Didn't have no place to go. c. Hannibal #4438. 1" / In euphemism. 1967) More prosaically. i. resulting (as with the homines of 33c) in the identification of the two activities: (34) a. from Layers of the onion. Once again. a speaker achieves protective coloration by virtually referring to a taboo object without technically referring to it. If (and when) x excretes. in the negation of iff p.' Consider the evolution of go to the bathroom. r6pr6hensible" involves sex and reproduction.' or alternatively . etc.ideally after- ward rather than during. x goes to the bathroom. c. . q. If x sleeps with y [ceteris paribus]. building from the generalization (occasionally honored in the breech. and friend. we have the devel- opment illustrated in (34) in which the execution of the relevant function moves. x goes to the bathroom. as it were.R-implicate 'more than friend. Folk-rocker Ian Robertson plays off both literal and euphemistically standard senses of one particular autohyponym in this verse: (35) I never slept with you. which may either Q-implicate 'no more than friend. spouse. If (and when) x goes to the bathroom. (Incredible String Band. 1" / Elsewhere in the land of unmentionabilia. as Morgan (1978: 263) points out. 'First girl I loved'. they typically end up sleeping together ('condorming') too .e.R. inconvenant. b. not lover.e. x makes love with y. x excretes. whose conventional meaning has undergone a final shift. An analogous source of R-based restriction that shares the social or cultural motivation of euphemism is the attenua- tion of negative force.

1 8 8 8 : 2 8 1 ) ." (Mill. If x does not b e l i e v e that p. 25 See Horn (1989: ch. L. un-American. Horn / Journal of Pragmatics 32 (2000) 289-326 315 straightforward and u n a m b i g u o u s w a y b y p a c k a g i n g contrary negatives in contradic- tory clothing. 1888: 319) Thus we have the R .r a i s i n g ' understanding.. f o r m a l l y c o n t r a d i c t o r y nega- tion to a specific. c. this s a m e practice is responsi- ble for the ' n e g .s t r e n g t h e n e d (contrary) interpretation rather than s i m p l y taking m e at m y (contradictory) word. and of attending to the logical canon. Compare our common phrase 'I don't think that' ." (Bosanquet. One is the t e n d e n c y for n e g a t i v e l y .. b. immoral) or verbs (dislike. Iff x d i s b e l i e v e s that p. I say I d o n ' t like a p a p e r or that I ' m not e s p e c i a l l y thrilled with it p r e c i s e l y to a v o i d saying that I dislike it.R. is the practical form of a logical error with respect to the Opposition of Propositions. of mistaking reverse of wrong for right. e s p e c i a l l y with posi- tive predicates. 1888: 293) "[Consider] the habitual use of phrases such as I do not believe it."[t]he essence o f formal negation is to invest the contrary with the character o f the contra- d i c t o r y " (Bosanquet. It is committed for want of the habit of distinguish- ing the contrary of a proposition from the contradictory of it. x d i s b e l i e v e s that p [= b e l i e v e s that not-p]. 1843: 804) Once again. or likelihood. as if it had lower- clause scope. 1" / In each case. disbelieve) to acquire strengthened contrary readings. 5) for elaboration. that contrary propositions.. x does not b e l i e v e that p.p r e f i x e d adjec- tives (unhappy." (Bosanquet. 25 This yields another o f the M i l l i a n fallacies: "The very frequent error . x does not b e l i e v e that p.not sur- prisingly. yet it is treated as if it were n e c e s s a r y . may both be false. the contrary m e a n i n g is sufficient but not l o g i c a l l y n e c e s s a r y to estab- lish the truth o f the contradictory. contrary interpretation in the case o f litotes.and there are specific contexts in which this tac- tic is e s p e c i a l l y likely to surface. In an e m b e d d i n g environment. at the s a m e time. desire. since it represents the i n d u c t i v e l y salient case that m a k e s the c o n t r a d i c t o r y true and since there m a y be social constraints against direct e x p r e s s i o n o f the stronger contrary. C o n v e r t i n g again to C P .which is really equivalent to "I think that _ _ not"..n o r m a l form: (36) a. unfriendly. the t e n d e n c y to interpret a higher-clause negation o v e r certain predicates o f opinion. I count on your willingness to fill in the intended R . If x disbelieves that p [= believes that not-p]. though they cannot both be true. T w o other contexts are cited b y B o s a n q u e t as illustrations o f his i n v e s t m e n t principle: "From 'he is not good' we may be able to infer something more than that 'it is not true that he is good'. this ' v e r y frequent e r r o r ' is a m p l y reflected in ordinary usage .b a s e d strengthening o f a general. which refer grammatically to a fact of my intellectual state but actually serve as negations of something ascribed to reality . w h o s e a n t o n y m s are most likely to trigger a v o i d a n c e m e c h a n i s m s . .

while contemporary logicians from Quine to Hintikka typically bemoan the 'quirk of English'.b a s e d strengthening o f a sufficient to a necessary and sufficient condition. since the denial o f the antecedent (or affirmation o f the consequent) arouses the ire o f fallacists in precisely the same w a y and. The fourteenth century logician R o b e r t B a c o n begins by distin- guishing a m o n g three varieties o f interaction b e t w e e n negation and its focus: the ordinary negative n a m e (nomen negatum) ' i s n ' t j u s t ' (non est iustus). as s c h e m a t i z e d in Fig. ergo est non ius- tus. Three varieties of interaction between negation and its focus. Technically. 26 This should strike a familiar chord. refers to the logically unwarranted placement (logisch ungerechtfertigte Stellung) of negation. Syncategoreumata: Braakhuis. 144-145. ergo non est iustus.' E converso autem non tenet.r a i s e d ' understanding o f I don't believe that p as c o n v e y i n g ' I believe that n o t . and the privative n a m e with incorporated negation (nomen privatum) 'is unjust' (est iniustus). 1989: 252.' Similiter: ab infinito ad negatum. ut: 'est iniustus. for precisely the same reason. Each represents essentially the same p h e n o m e n o n : R . ustus itJ~( Fig. see Horn (1989: §5. O u r m o v e to assimilate c o n d i t i o n a l fallacies and negative strengthening has an ancient lineage. he notes. Horn / Journal of Pragmatics 32 (2000) 289-326 L i k e the derivation o f homo from humus. 1979. emphasis added) 26 The first systematic treatment of the neg-raising phenomenon. but o r d i n a r y usage is not always consistent with this: "Ex hiis patet quod bene sequitur argumentum a privato ad infinitum. or 'unfortunate ambiguity' responsible for the offending readings. the ' n e g . the infinite n a m e (nomen infinitum) 'is not-just' (est non iustus).2) for references and discussion. we have argued. .p ' tends to arouse the ire o f the p h i l o s o p h e r and g r a m m a r i a n .316 L. sed est paralogismus consequentis. 4. 'peculiarity'. Spruyt. the third unilaterally entails the second and the second the first (see Figure 4).R. that of Tobler (1882). ut: 'est non iustus. I. 4. vol." (Bacon.

1" / 4.a n d . K i m is unhappy. I'm Nick'. If K i m isn't happy. If K i m is unhappy. a fi)rtiori. 1972.s u f f i c i e n t condition: (37) a. T h e r e ' s s o m e b e e r in the fridge (whether y o u ' r e thirsty or not). W e have also noted (following Lilje. S o m e conditionals are i m m u n e by their very seman- t i c o . y o u ' r e not thirsty. As A u s t i n (1956: 210) notes (cf. 1973. since p does not offer a sufficient condition for the truth o f q here (being entirely irrelevant to the truth conditions o f q). and Zuber. b. who are you then?' one of the 'ladies' asked. 1974) that not all conditionals license them. Zaefferer. 1991). . One case in point is the non- causal or ' A u s t i n i a n ' c o n d i t i o n a l whose locus classicus has b e c o m e (38a). BoEr and Lycan. 'And if we don't need anything. but is the fallacy of consequence. However. and hence o f CP. from the infinite to the negative. so that if (38a) is true. c." Thus the m o v e from the c o n t r a d i c t o r y K i m isn't h a p p y to the contrary K i m is u n h a p p y is an instance o f the fallacy o f consequence. 27 (38) a. Horn / Journal of Pragmatics 32 (2000) 289-326 317 "From these it is apparent that the argument follows validly from the privative to the infinite. not only conditionals license C P effects. 28 In fact the j o c u l a r effect o f such a response to non-causal conditionals 2v Austin's actual example was There are biscuits" on the sideboard if you want some. 'Ladies'. thus 's/he is not-just. K i m i s n ' t happy. which require a sufficient c o n d i t i o n to p r o m o t e to necessity. 1974: 105). 28 Compare this putatively humorous anecdote from the New York Times 'Metropolitan Diary' (13 July 1998): Rebecca Mai and Tiffany Otto were in a SoHo boutique when they were approached by an eager salesman. as we see in (38c). If y o u ' r e thirsty.p r a g m a t i c nature from the fallacy o f perfection. there can be no CP effects. Iff K i m is unhappy. The formal antecedent in an A u s t i n i a n i f p then q conditional offers grounds for the relevance o f the state- m e n t that q. ? I f t h e r e ' s no b e e r in the fridge. as in Zuber. thus 's/he is unjust. 'If there's anything you need. the d e d u c - tively invalid but i n d u c t i v e l y plausible strengthening o f a sufficient condition to a n e c e s s a r y . A response to (38a) o f the form ' S o if I ' m not thirsty. therefore s/he isn't just'. side- boards. these (un)con- ditionals entail their consequents. Unperfectable conditionals (and others) A s we have seen in s o m e detail. L. Similarly. the converse does not hold. also Zuber. (38b) will be true as well. t h e r e ' s some beer in the fridge. b. but the beer/fridge case has tended to supplant it (even in French. he said. but not a sin- cere one. t h e r e ' s no b e e r ? ' is a typical c o m e b a c k . therefore s/he is not-just'. 1974. K i m i s n ' t happy. Nor do they allow ordinary contraposition.R. perhaps because so many neo-Austinians are epistemically insecure with respect to British biscuits and. c.

North Vietnam will not agree to negotiate. C. K6nig. 1986: 236). 1994) that an assertion of even if p. The (even) if conditional in (41a) is not converted to (41b) or 'perfected' to the iff biconditional in (41c) because it is only sufficient conditions that implicate necessary-and-sufficient ones. 1991. the U. b. (42) a. indeed. whether subjunctive or indicative. . North Vietnam will not agree to negotiate. since it has been frequently noted (cf. even if conditionals. (40) a. he's bald.S. of CP within causal conditionals and other affirmations of sufficient condition. b. Antecedents presupposed by their consequents. If there's a King of France. North Vietnam will (still) not agree to negotiate. Horn / Journal of Pragmatics 32 (2000) 289-326 reinforces the psychological plausibility. b. If Jack has a wife.S. That is. (39) a. North Vietnam will agree to negotiate. Only if you gave me $1000 would I not mow the lawn. 1990. it is undeniable that (as in the case of the Austinian conditionals just mentioned) neither contraposition nor modus tollens holds. inter alia Pollock. (41) a. Barker. Note that this holds whether or not the even explicitly occurs in the antecedent or a still in the consequent.g. (Even) if the U.S.318 L. she's miserable. and antecedents entail- ing their consequents. C. I (still) wouldn't mow the lawn. 1982. This is hardly surprising. 1991. 1993. I (still) wouldn't mow the lawn. rather. Bennett. it is seen as insufficient grounds for the attenuation of an already established intransigence. Lycan. {will not halt/will not have halted} the bombing. if not logical validity. If the U. Halting of the American bombing campaign is not advanced in (39a) as sufficient grounds for North Vietnamese intransigence. as in (42). as McLaughlin (1990: 33-34) observes. the point of a con- ditional like (39a) (from Stalnaker.'. halts the bombing. fail to trigger CP effects (see e. If you gave me $1000. McLaughlin. If and only if you gave me $1000. 1976. b. The U. will halt the bombing. halts the bombing. 1968) is to stipulate that (39c) would be true regardless of whether (39b) is true. as in (43). q does not establish p as a sufficient condition for the truth of q.S. While there has been some dispute about whether every even- ~f conditional in fact entails its consequent (see above references for discussion). are also incompatible with CP. If North Vietnam agrees to negotiate. neither (40a) nor (40b) follows from an assertion of (39a) understood as an even if conditional. By the same token.S. The U.R. . will not halt (will not have halted) the bombing.

A stronger claim has recently been advanced by Dancygier and Sweetser (1997). threat. it has rained and If he's hot. Their approach to conditionals. Don't do the crime if you can't do the time. he has a fever. and from the conditionalized directives of (48) (48) Break glass in case of emergency. If you know what's good for you. You scratch my back. If she's a phonologist. and lex talionis in (47). while often subtle and insightful. and your parents. If you can't stand the heat. pluck it out. none dare call it treason. Ever since Aristotle pinpointed the temptation to infer If the streets are wet. Harry Truman. Stop here on red. Rather than listing those varieties of conditional statements that exclude CP effects. however. the promotion of sufficient conditions to necessary ones often seems causally dependent on the goals of speaker and hearer in the discourse context of the conditional. a tooth for a tooth. (47) An eye for an eye. (46) You help me. non-predictive instances of CP are legion. respectively attrib- utable to Jesus. Horn / Journal of Pragmatics 32 (2000) 289-326 319 (43) a. it's a mammal. Feeling sleepy? Pull off the road and take a nap. L. If it prosper. quid pro quo in (46). In addition. why not simply list those that allow them? As Mike Geis (p. it has also been clear that the conversion or perfection of conditionals cannot be restricted to warnings. a step that would make it impossible to generalize the pattern to those CP-inducing sufficient conditions that don't involve conditional form. and warning character of conditionals like those addressed to the window-leaner of (1) or the prospective lawn-mower of (2) are especially suggestive. But this cannot be the case. to the contexts of cops and robbers in (45). threats. entails that per- fectibility is built into the conventional meaning of (certain) conditionals. for one.c. Counterexamples to the conventionaliza- tion and/or prediction theses range from the injunctions of (44). If that's a cat. b.) remarks. beginning with the Aristotelian rain and fever cases. I scratch yours. the promise. Garbage in. maintains that CP is most natural with prohibitives. I help you. (44) If thy right eye offend thee. Comrie (1986).R. . get out of the kitchen. (45) One false move and I'll shoot. Dancygier and Sweetser explicitly argue that all cases of if --->iff strengthening involve predictive condition- als. garbage out. she's a linguist. or promises. in par- ticular. you'll stop that. Sir John Harrington.

320 L. Since paratactic protases are w i d e l y attested in the conditionals o f the w o r l d ' s languages (cf. (Comrie.no service. 1979: 6) (iii) If the butler had done it. I ' l l give you $5. If you h a d n ' t m o w e d the lawn.'No credit. In each case. 2. nothing gained. 1951: 37) (ii) If Mary were allergic to penicillin. If you d o n ' t m o w the lawn. CP has standardly been seen as a p p l y i n g to indicative conditionals only. he would have shown just exactly those symptoms which he does in fact show. If you m o w the lawn. if the 'understood' tf is an even ( f . Haiman. no washee. No sufficiency. not to subjunctives or counterfactuals. it will not straightforwardly invite the inference o f (50b). b.the conditional will be unperfectable. yet it is not obvious how to extend either C o m r i e ' s notion o f prohibition or D a n c y g i e r and S w e e t s e r ' s s c h e m a of prediction + c o n v e n t i o n a l i z a t i o n to the full range o f the e x a m p l e s o f ( 4 4 ) . I w o n ' t give you $5. she would have exactly the symptoms she is showing. no problem' . 3° But if we focus on the the actual connections b e t w e e n protasis and apodosis in the two cases. (Kart- tunen and Peters. N o t h i n g ventured. Among the examples that have cited to prove that subjunctive conditionals are not necessarily counterfactuals are those in (i)-(iii): (i) If Jones had taken arsenic. 1986: 90) See Comrie (1986) and von Fintel (1998) for more on the semantics of subjunctive vs. G i v e n that (50a) is generally felicitous only against the b a c k g r o u n d o f the p r e s u p p o - sition that you did not in fact m o w the lawn. however it is to be analyzed. If you had m o w e d the lawn.( 4 9 ) in any non-circu- lar way. the sufficient condition (51a) is strengthened p r a g m a t i c a l l y to the necessary-and-sufficient condition (51b). the sufficient condition identified in the virtual protasis is regularly p r o m o t e d to a necessary and sufficient one.R. (50) a. 1986). . no gain. which induces the opposite presupposition. no CP! 3o Notice that this 'presupposition'. but this ques- tion will have to be left to future research. Granted. it w o u l d be interesting to determine the gen- erality o f the cross-linguistic prevalence o f C P in such environments. we would have found just the clues that we did in fact find. Horn /Journal ~f Pragmatics 32 (2000) 289-326 to the l a u n d e r e r ' s e n t h y m e m e s of (49). there is an apparent a s y m m e - try b e t w e e n indicative and subjunctive conditionals with respect to CP: (2) a. no shoes . there is a parallel: in both cases. No shirt. Finally. we turn briefly to one class o f conditionals generally figured a m o n g the non-CP-inducers. cannot be part of the semantics of the subjunctive per se. b. 29 (49) No tickee. I ' d have given you $5. for the reasons we have noted. (Anderson. No pain. I w o u l d n ' t have given you $5.~ Of course. indicative condi- tionals.

Horn /Journal of Pragmatics 32 (2000) 289-326 321 (51) a. establishing an independent pragmatic motivation for the tendency in question allows a simpler and more general account of the con- struction's conventional meaning. PHEPH. What separates the peculiar (52b) from the reasonable (52a) is that even though it's just as true . Thus my watering the plant is not just a sufficient but also a necessary condition for the plant's survival.g. and indeed as assumed by the fallacists as well. in uttering (52a). I thereby implicate. litotes. L. sprayer in hand. Only those counterfactuals that are convertible in this manner are contextually plausible as well as true. as argued for by Gricean theory. I acknowledge that in the nearest world (the one as much like the actual one but) in which I remembered to water the plant.it's a sufficient condition that cannot naturally be understood as a nec- essary one: in the nearest world in which the plant survived. b. in the usual CP manner.as well as the related phenomena of euphemism and neg-raising . the tendency to read sufficient conditions (e. it survived. 5. [you mow the lawn] ~ [I give you $5] ('is-a-necessary-and-sufficient-condition-for') Where (2) and (50) differ is in whether the set of worlds against which we evaluate these connections includes the actual world or not.sufficient-condition-for') lI b. q) involves not the semantics of what conditionals say but the prag- matics of what speakers can be rationally assumed to implicate (or to sous-entendre). it would have survived. (it's because) I watered it. In each case.R.an adventitious visit by the teen idol. Last words? The problem raised by CP is one we encounter whenever we cruise the transi- tional neighborhood of the semantics/pragmatic border: how do we draw the line between what is said and what is implicated'? As stressed by Ducrot. young Leo wasn't the one who watered it. (52) a. and other instances of on-line inferencing . it would have survived. I was. I gave you $5. the converse: in the nearest world in which the plant survived. In this respect. unfortunately (given the usual presup- positions) a condition not met. If I had watered that plant. what I implicate is that in that world I gave you the money (not just if but also) only if you mowed the lawn. the protasis of ifp then q) as necessary-and-sufficient conditions (iffp.is complementary to the Q-based upper-bounding of scalar predication. By the same token. the R-based strengthening that motivates CP. [you mow the lawn] --> [I give you $5] ('is-a. . If Leonardo diCaprio had watered that plant. would have rescued my coleus as surely as my watering it would have done . in (50a) what I say is that in the closest world to the actual one in which you mowed the lawn.

Be that as it may. but rather must be taken and used as granted. Ref. Aristotle. 1928.R. and analogous fallacies? As it happens. edited and translated under the supervi- sion of W. is there any m o n e y to be made in per- petrating CP. 1970: 50ft. PHEPH. contentiousness ('for when agitated everybody is less able to take care of himself'). professional fallacists from yesterday's rhetoricians to t o d a y ' s pop philosophers are the moral equivalent of the purveyors of electronic detection equipment. than in being wise without seeming to be (for the art of the sophist is the semblance of wisdom without the reality. but I cannot be certain that I have m a n a g e d to reduce my readers to u n g r a m m a t i c a l i t y or babbling.322 L. they look ahead less'). References Anderson.. Horn / Journal of Pragmatics 32 (2000) 289-326 In tracing the fallacist tradition back to Aristotle's Sophistical refutations. 1547): "Aristotle . Alan Ross. but rather affirms it as a demonstrated conclusion: 'So-and-so is not true' . one does not put the final proposition in the form of a ques- tion. In this respect. Sophistical refutations. Sophistical refutations" 165a20-24) To this end. and the reduction of o n e ' s adversary to u n g r a m m a t i c a l i t y or babbling. Analysis 11 : 35-38. 174a17ff.or better yet by presupposing ." (Aristotle.. this question has been addressed: "'Now for some people there is more profit in seeming to be wise. 174b8-12. 3~ Any doubt that o1' Stagira Slim was a master of the art of the bluff should be dispelled by this remark cited by William Baldwin (A treatise ofmorall philosophie. speed ('for when people are left behind. Oxford: Oxford University Press.D. 39-40) While it is more of a challenge to utilize these useful suggestions in written argu- ment. and the sophist is one who makes money from an apparent but unreal wisdom). I am especially taken with Aristotle's tactic of the well-timed bluff: 31 "What often produces the strongest impression of refuting one's adversary is the most highly sophistical insinuation of all: having proved nothing. 1951.. Conclusions should not be put in a form in which one's respondent can deny them... I can only close by asserting with- out the slightest justification . also Hamblin.).. But more to the point. cf.that the last words on conditional perfection have now been spoken. we have seen that it's never quite clear whether the advice so generously offered is intended more as a shield against sophistry or as an instruction m a n u a l for would-be sophists. A note of subjunctive and counterfactual conditionals. the recipe of success for the sophist in an academic debate includes an elaborate and diverse list of forensic ingredients (Soph. Ross. I have done m y best to go on for far longer than I should have and to throw m y points together in such a way as to make it as difficult as possible for a n y o n e to follow them. 165b12ff. A m o n g these are length ('for it is difficult to keep sev- eral things in view at once')." (first citation under 'unbelieve' in OED) . But as a fellow poker-player. De sophisticis elenchis: The works of Aristotle." (Aristotle. being asked what vauntage a man might get by lying. he answered: 'to be unbelieved when he telleth truth'.

Ebbesen. 79-96. eds.. Conditional perfection.O. 1991. Johan. PMLA 67: l 117-1144. De 13de eeuwse tractaten over syncategorematische termen (2 vol. L. Lon- don: Taylor and Walton. use. 1986. 1972. 1955. Br6al. 1959. Steven E. W. the calculus of inference. Bolinger. eds. and interpretation of language. 1970. second edition. vol.. yon Fintel. In: A.. Aristotle. Chapman. In: P. conditional assertion. 1975. Stephen J. Oxford: Oxford University Press. 204-232. informativeness.. Pragmatics in the last quarter century: The case of conditional perfection. I: The Greek tradition. Cali: Universidad del Valle. 1973. Reprinted in: O. Jonathan. Tadeusz. Cole. Czezowski. necessary and probable. Comrie. Gilles. Pr6suppos6s et sous-entendus. Barker. 1956. Logique et Analyse 28: 61-74. On certain peculiarities of singular propositions. Kennedy. Only if. 1991. Mind 64: 392-395. van der Auwera. 1985. still. New York: Holt. Linguistic Inquiry 6: 535-375. Ducrot. T. Meaning.. Fillenbaum. Barbara and Eve Sweetser. In: Traugott et al. . and Jean P. BoEr. Cornulier. In: Traugott et al. Then in conditional constructions. Bernard. On rhetoric. 1888. 1997. J. Radical pragmatics. It-clefts. On conditionals again. Stephen J. Braakhuis. Athanasiadou and R. The myth of the exclusive 'or'. 1. Dwight. Linear modification.D. 1-61. and "only if'.. Darner. Ph. Even. Belmont. New York: Oxford University Press. MIT Working Papers in Linguis- tics 25: 29-44. Warnock. H. Leiden: Brill. Ducrot. Edward and William B. The Stoic theory of arguments. 13-31. Polifonfa y argumentaci6n. Translated by Mrs. von Fintel. Kai. 1986. Chapman. 1983. Urs. New York: Brown. Conditionals in a dynamic context. 'Even if'. Journal of Pragmatics 27: 261-274. 'If' and the presumption of exhaustivity. Translated by George A. Fallacy: The counterfeit of argument. Atlas. and Stephen C. Atmosphere effect re-examined. 1897. Paris: Les l~ditions de Minuit. Sten. Barrett. 1911. 1981.R.. 1995. Kai.L.. Linguistics and Philosophy 5: 403-418. De Morgan. Attacking faulty reasoning: A practical guide to fallacy-free arguments. 1993. 169-190. Oxford: Clarendon. Austin. Reprinted in: J. The consequent-entailment problem for even if. Englewood Cliffs NJ: Prentice-Hall. 1983. 77-99. CA: Wadsworth. Ducrot.. Samuel. The presupposition of subjunctive conditionals. 1979. Paris: Hermann. Nicholas.. 1998. Schwarze and A. Philosophical papers. 1971. 1994. 1995. 1981. Dire et ne pas dire. Levinson. 1984. Johan. Loren J. Fearnside. eds. eds. The use of conditionals in inducements and deterrents. Oswald.. Bennett. 1959. Mind 80:116-121. 1997. Bosanquet. Jay D. Berlin: de Gruyter. Invited inference and other unwelcome guests. Stephen J.G. von Stechow. Henry Cust. Augustus. Beno]t de. Fauconnier. 1990.. Logic. 1847. second edition. Bernard. Le dire et le dit. Amsterdam: Benjamins. van der Auwera. ed. Holther. Dancygier. dis- sertation. Barker. Ducrot. Urmson and G. Dirven. and logical form. Analysis 53: 254-261. Linguistics and Philosophy 14: 1-38.A. Linguistics and Philosophy 17 : 249-260. Michel. Egli. van der Auwera. 1971. 1969. and counterfactuals. 179-195. Ifs and cans. Lycan. New York: Academic Press. Robert and Alfred Stenner. eds. 1982. Commentators and commentaries on Aristotle's Sophistici elenchi: A study of post- Aristotelian ancient and medieval writings on fallacies: vol. 1997b.J. 1997a. Journal of Pragmatics 7: 247-249. Paris: Librarie Philosophique J. traduction et commentaire par Louis-Andr6 Dorion. 1952. MIT. Introduction. Les r6futations sophistiques. Edward. In: C. Third edition. Journal of Experimen- tal Psychology 58: 220-226.. Essai de sdmantique. University of Leiden. and William G. Barker. Johan. Cognitive Linguistics 8: 109-136. Horn /Journal of Pragrnatics 32 (2000) 289-326 323 Aristotle. Oswald. Papers in Linguistics 6: 483-505. The art of deception. Conditionals: A typology. Oswald. Vrin. Conditional excluded middle. Formal logic: Or. Ms.). Capaldi. Pragmatic scales and logical structure.

1986. James D. Howard. Benson. The conversational condition on Horn scales. 1989. Studies in the way of words.. 215-228. Levinson. 1971. eds. Angelika.Modalw6rter . John Neville. Journal of Semantics 13: 140. H. 1991. Schiffrin.. eds.L. NELS 14: 125-142. The influence of emotional subject matter on logical reading. Journal of General Psychology 34: 127-151. Laurence R. CA: University of California Press. McCawley. Pragmatics. 1884. Linguistics and Philosophy 18: 21-60. Mates. Stephen C. Constraints on the form and meaning of the protasis. London: Macmillan. 1981. Uninvited inference. Robert.. eds. Studies and exercises in formal logic. 1988. 1972. Horn.. John. Economy and redundancy in a dualistic model of natural language. Fallacies. Laurence R. Johnson-Laird.-K. SKY 1993: 1993 yearbook of the Linguistic Association of Finland: 33-72. 1989 [1967]... J.. MA: Harvard University Press.. BLS 16: 454-471. In: Traugott et al.A. Gazdar. 1967. Horn. Evidence and meaning. In: D. New York: Humanities Press.N. 1973. Haiman. 169-179. Tfibingen: Niemeyer. Zwicky. dissertation. R. Logic and contemporary rhetoric: The use of reason in everyday life. 1995. Horn / Journal of Pragmatics 32 (2000) 289-326 Fogelin. Toward a new taxonomy for pragmatic inference: Q-based and R-based impli- cature. Julia. CA: Wadsworth.. Greek Grice: a brief survey of proto-conversational rules in the history of logic. Berkeley. presupposition and logical form. Papi and J. Horn. and Arnold M. Keynes. Seventh edition. Syntax and semantics 11 : Presupposition. Ekkehard and Elizabeth Traugott. New York: Macmillan.. 1995. Gerald. DC: Georgetown University Press. and use in context (GURT '84).. Hamblin. Hamburgers and truth: Why Gricean inference is Gricean. concessive conditionals and concessives. Laurence R. A theory of scalar implicature. 1961.Konditionals~tze. form. 1996. K6nig.D. Horn. On the semantic properties of logical operators in English. Kahane. Meaning. The genealogy of disjunction. Dinneen. Pragmatics: implicature. 1990. In: W. K6nig- stein: Scriptor. In: Traugott et al. Mackie.. Schulze. Cambridge: Cambridge University Press. 205-214. Even and even if. 1971. Keenan. CLS 9. 1993. H~illen and R. Fallacies. Jennings. Horn.. 1972. IL: University of Chicago Press. Language in Society 5: 67-80. Ekkehard. 1978. If and only if.. P. Amsterdam: Benjamins. U. Arthur. Oh and D. Cambridge. Pragmatic strengthening and semantic change: The con- ventionalizing of conversational implicature.. Linguistic Inquiry 2: 566-569. Karttunen. Understanding the lexicon... Horn. Lycan.. A natural history of negation. UCLA. Laurence R. 1967. Conventional implicature. Verschueren. Levinson. In: Traugott et al. The universality of conversational postulates. Conditionals. 55-76. Laurence R. Belmont.L. Entry in. Conditionals and mental models. Linguistic lnquiry 2: 561-566. ed. Matsumoto. Ph. Kratzer. 1979. Lilje.. Grice. Counterfactual conditionals. 3. Horn. On invited inferences. eds.. 1985. Stoic logic. Karttunen. Chicago. . In: C. 1979. Hirschberg. William. 1986. 1983. eds. 1970. 1986.. Michael L. Semantik der Rede: Kontexttheorie . In: M. Yo. Lefford. Laurence R. 1994. Washington.. G. 1984. 1974. Elinor Ochs. 1946. New York: Oxford University Press.P. Horn.W... eds. Linguistics and Philosophy 14:115-150. Encyclopedia of philosophy. Lauri.. C. New York: Academic Press. 110-123. New York: Academic Press. Minimization and conversational inference. Exclusive company: Only and the dynamics of vertical inference. Stephen C.. 11-42. Linguistic Inquiry 3: 540-542.. 61-129. of Pennsylvania dissertation.324 L.E. 229-246. 1987. London: Methuen. vol.R. K6nig. Lauri and Stanley Peters. Linguistic Inquiry 5: 632~635. Exhaustiveness and the semantics of clefts. The pragmatic perspective. Laurence R. 1-56. Laurence R. 1976. Geis.

. Joke. On con- ditionals. NY: State University of New York Press. New York: Academic Press. James D. Zwicky. Sommers.F. Oxford: Clarendon. Wallis. The effect of self-contradiction on fallacious reasoning. Wason. 201-205.. A pragmatic theory of fallacy. 1974. Conditionals and unconditionals: Cross-linguistic and logical aspects. 1976. P. Chicago. On the logic of ordinary conditionals. Jessica. Ferguson.. 1882.. eds. Amsterdam: Benjamins. Thinking and reasoning. P. Oxford: Amos Curteyne. Wason. Robert. Subjunctive reasoning. 1975. Mill. Wason. John Stuart. P. Pollock.M. Pirie. 1981. Quarterly Journal of Exper- imental Psychology 16: 3 ~ 3 4 . Traugott. 1972. London: Longman. Zaefferer. 1990. James D. ed. P. London: Routledge. In: D. 1928. Cambridge: Cambridge University Press.. Relevance: Communication and cognition. Syntax and semantics 4. Wilkins. 103-110.C. Implications and meaning. P.. Zaefferer. 1985. 1974. Tuscaloosa. Jerry. de Rijk. Mod61es logiques et niveaux d'analyse linguistique. Quarterly Journal of Experimental Psychology 12: 129-!40. Institutio logicae. Leipzig: Hirzel. 1987. On fallacies. The logic of natural language. Pragmatics (Syntax and semantics 9): ed.C. 1968. Paris: Librairie Klincksieck. 1986.C.C. Mill.. A theory of conditionals. In: J. Martin. Francis P. P. 1964.. On the failure to eliminate hypotheses in a conceptual task. Cambridge. Fred. London: Batsford. Harmondsworth: Penguin. 1687. Glossa 9: 36-52. 1-37. 1972. P.. eds. Arnold and Jerrold Sadock. Psychology of reasoning. Robert C. Alice ter Meulen. 1867. Ellen. The effect of changed material on ability to do formal syllogistic reasoning. Wason and P. John Stuart. Peter of Spain on composition and negation.. Archives of Psychology 102. 261-280. Dietmar. 1921.N. Toronto: University of Toronto Press.~matics 32 (2000) 289-326 325 McCawley. 210-236. S. L. Peter of Spain: Language in dispute.': A second look. I1 ne faut pas que tu meures 'du darfst nicht sterben'. Judy Snitzer Reilly and Charles A. Strawson. Walton. Assen: Van Gorcum. New edition. Stalnaker. McCawley. third edition. Everything that linguists have always wanted to know about logic but were ashamed to ask. 1975. 1989.. 165-174. Johnson-Laird.. Walton.N. McLaughlin. Peter of Spain. 1960. Elizabeth. The book of the fallacy: A training manual for intellectual subversives. New York: Academic Press. 1993. Theoria 35: 38-54. Douglas. O'Hair. Reprinted in: Vermischte Beitrtige zur franz6sischen Grammatik 1. 1986. American Philosophical Quarterly Monograph Series 2:98-112. Sperber. ed. Spruyt. John. English translation of above. eds. Madsen. Peter of Spain. Prince.. Din- neen. Douglas. University of Pennsylvania. 1991. 1982. AL: University of Alabama Press. John. Dordrecht: Reidel.. Ambiguity tests and how to fail them. Albany. Morgan. Horn / Journal of Pra. Johnson-Laird. Informal fallacies.. Zuber. David and R. 1952. 1982. 1969. trans. Berlin: Foris. Ryszard. Tractatus called afterwards Summule Logicales. and P. Semantic universals and universal semantics. Chicago: University of Chicago Press. Critical edition by L. 1843. Tobler. On the explanation of invited inferences. MA: Harvard University Press. Wirth. London: Methuen. An examination of Sir William Hamilton's philosophy. Cole. . Introduction to logical theory. Amsterdam: Benjamins. Grice and universality: A rejoinder..C. 1995. 1990. Kimball. 1968. Third edition..G. 1978. In: Book V: A system of logic ratiocinative and inductive. M. Dan and Deirdre Wilson. IL: University of Chicago Press. Alfred. Nijmegen: Ingenium.. Wason.C. Conditionnelle: Sdmantique ou pragmatique? In: J. Everything that linguists have always wanted to know about logic but were ashamed to ask. 'On the failure to eliminate hypotheses .R. Ms. Two types of convention in indirect speech acts.

1989) and of numerous articles and monographs on negation. 1945) received his Ph. He has taught at the Universities of California-Berkeley. Southern California. Horn /Journal of Pragmatics 32 (2000) 289-326 Laurence R. C. Since 1981 he has been on the faculty of the Department of Linguistics at Yale University. or both.D. IL: University of Chicago Press. Santa Cruz LSA Institutes. and at the Stanford and U.R. Horn (born in New York City. many of which appear in the bibliography of his paper in this issue. . Wisconsin.326 L. and Aix-Marseille. He is the author of A natural history of negation (Chicago. where he is now Professor and Director of Undergraduate Studies. from UCLA in 1972. pragmatic inference.