Three decades or so have elapsed since Austin""s initial formulation of speech act theory in 1962. His conception of the locutionary and the illocutionary acts has been scrutinized, challenged and modified (and drastically so with respect to the illocutionary act). The pertinent literature is now so well-known that it would be a clich¨¦ to cite it. The perlocutionary act, in contrast, has received the least attention. As far as I can gather, there are only four papers devoted to the topic: Cohen (1973), Campbell (1973), Gaines (1979) and Davis (1980). Peripheral discussions of it are found in e. g. Black (1969),Sadock (1974), Van Dijk (1977), Bach and Harnish(1979), and Leech (1983). These efforts have contributed, in one way or another, to our understanding of the perlocutionary act. Nevertheless, there are some important issues which call for further scrutiny and criticism. This paper first identifies these issues, then discusses the problems which are shown to have led to the impasse of the perlocutionary act. It is argued that the conception of the perlocutionary act is fundamentally misguided, and a fresh approach is called for.
2. An anatomy of perlocution
Let us first recall Austin""s initial statement of the act:
"Saying something will often, or even normally, produce certain consequential effects upon the feelings, thoughts, or actions of the audience, or of the speaker, or of other persons. . . "(1980: 101)
There are four defining features discernible in Austin""s conception of the act:
(1)S says something to H;
(2)H is affected in a certain way;
(3) That H is affected is treated as a consequential effect of S""s saying something;
(4)S is therefore attributed with the performance of the perlocutionary act.
In what follows we take a look at the various views of these four features elaborated by Austin and his proponents/opponents.
2. 1. S says something to H
It is held that perlocutionary effects can be brought off by the locutionary act, thus resulting in what Sadock (1974) calls "sense perlocution", or by the illocutionary act with or without the aid of the mediating locutionary act, thus giving rise to "force perlocution"(1974:153). Cohen (1973:496)even suggests the possibility that one locutionary act may give rise to several illocutionary acts, which in turn produce as many perlocutionary acts.
However, as far as the perlocutionary act is concerned, "say"need not be in its full normal sense. Austin tells us
"[i]t is characteristic of perlocutionary acts that the response achieved, or the sequel, can be achieved additionally or entirely by non-locutionary means: thus intimidation may be achieved by waving a stick or pointing a gun . . . " (1980: 119; emphasis mine)
Austin""s nonverbal means of achieving a perlocutionary effect is generally ignored. Cohen (1973:495) however holds that the perlocutionary act can be performed "simply by the sound I produce when I say whatever I say". Sadock exemplifies a similar view:
"Some perlocutionary effects can be brought off independently of, or even in spite of, the meaning of an utterance, as when one yells in the ear of a sleeping person. Don""t wake up. One can succeed in frightening someone by saying Boo!"(1974:153)
It is echoed by Bach and Harnish (1979: 82): "L1. By uttering ""Dont""t wake up""S awakened H".
Notice that, although "Don""t wake up" is a full-blooded utterance, it functions, with respect to the perlocutionary act of awakening H, merely as acoustic sounds uttered loud enough to rouse H from his/her slumber (see 3. 1 below for further analysis).
2. 2. H is affected in a certain way
Austin and the others allow the following three possibilities. (1) S""s saying something may produce multiple and different (though related) effects on one H. For instance, it may have some effect on H""s thought apart from that on his action. (2) The number of individuals to be affected can be more than one, and the effects may differ from one individual to another. And (3) the previous two cases may cooccur. (1), (2) and (3) can be graphically represented in Fig. 1.
S¡¯saying something may affect
Let us call the view that S""s saying something may produce multiple effects on multiple persons the Multiplicity Thesis. 
Austin is also the originator of another thesis, to be called the Infinity Thesis. He observes:
"For clearly any, or almost any, perlocutionary act is liable to be brought off, in sufficiently special circumstances, by the issuing, with or without calculation, of any utterance whatsoever, and in particular by a straightforward constative utterance (if there is such an animal). "(1980: 110)
Bach and Harnish restate Austin""s view thus:
". . . there is virtually no limit to the sorts of things that can result from speech acts - almost anything is possible. "(1979: 17; emphasis mine)
In other words there is no way to predict or tell, in view of the utterance, what the subsequent effects are; the utterance exerts no binding force on the effects produced.
Sadock voices another sense of limitlessness, or infinity in terms of quantity:
"it is also characteristic that the number of intended perlocutionary effects associated with an utterance is not limited. "(1974:9; emphasis mine)
In short, the issuing of an utterance may produce an infinite and indefinite number of perlocutionary effects.
2. 3 H being affected is treated as a consequential effect of S""s saying something
Although H being affected is some event occurring in H, it is regarded as something caused by S""s saying something. In Davis""s (1980: 39) terminology, S""s saying something is a "perlocutionary cause", while H being affected is a "perlocutionary effect". The view that S""s saying something causes H being affected is termed the Causation Thesis. If S""s saying something counting as locution or illocution or both fails to cause H being affected, it will at most constitute a perlocutionary attempt, far from being a perlocutionary act. If, on the other hand, some events take place in H prior to S saying anything, these events cannot be regarded as consequential effects caused by S""s saying something, and hence it will not lead to the claim that a perlocutionary act has been performed. Only when both the perlocutionary cause and the perlocutionary effect occur will S be said to have performed a perlocutionary act. The Causation Thesis is therefore fundamental to the conception of the perlocutionary act. Austin (1980) himself however nowhere elaborates it except in a footnote in Lecture IX (p. 113). The thesis has been generally acknowledged explicitly or implicitly, and elaborated in Davis (1980).No serious challenge to it has yet been made (to be examined in detail in section 3 below).
2. 4. S is attributed with the performance of a perlocutionary act
Since H being affected is something caused by S,S is attributed with the performance of a perlocutionary act. This gives rise to three questions. First, how can we decide which act is to be attributed? Second, should we take S""s intention into consideration? Third, what is the nature of such an attribution?
The first problem will be intractable, if we try to determine the perlocutionary acts from S""s point of view. For one thing, there is no "performative formula" for S to perform the act by uttering it. For another, by the Multiplicity Thesis, we cannot infer, from what S says, what effect is to be brought about -- anything is possible. Austin, being fully aware of the problem, seems to suggest a solution by the notions of "perlocutionary object" and "perlocutionary sequel". He observes (1980: 118): "the act of warning may achieve its perlocutionary object of alerting and also have the perlocutionary sequel of alarming". Cohen develops Austin""s idea into what he terms "direct associated perlocution". That is, for some illocutionary acts there are certain perlocutions typically associated with them, e. g. arguing with persuading or convincing, warning with alarming, threatening with intimidating, and so on. It takes little imagination to see that the solution gets us nowhere. For it works only in a few limited cases. 
The difficulty seems to be by-passed if we work backwards, i. e. by examining the effects produced in H. Gaines""s observation that it is the Austinian doctrine to define perlocutionary acts "strictly in terms of audience""s effects" (1979: 214). Overstating as it may sound, Gaines"" remark does capture the general practice among philosophers and linguists alike. This effect-to-act practice will be put to scrutiny later (see 3. 3 below). 
As for the second issue, viz. the relevance of S""s intention to the attribution, it is generally held that perlocutionary effects can be brought about intentionally or unintentionally, although Bach and Harnish (1979: 17), and Gaines (1979:213) argue that a limit has to be set to the class of intentional perlocutionary acts. S""s intention is therefore relevant to the distinction of intentional perlocutionary acts from unintentional ones. It is however irrelevant as to the decision of whether a given act is perlocutionary or not. This view will be referred to as the Intention Irrelevance Thesis.
The third issue concerning the nature of perlocutionary attribution is so far ignored and will be examined in 3. 4 below.
Our critique of the current theory of perlocution begins with the examination of the Causation Thesis which it will soon be clear, is the source of many problems. Then we turn to the theses of Multiplicity, of Infinity and of Intention Irrelevance, and the Effect = Act Fallacy. This is followed by the analysis of "By sayig x I did y". Finally we conclude this paper by pointing out some implications our critique will have for pragmatics.3. 1. The Causation Thesis: Three types of causation
Austin himself seems to be uneasy about using causative terms, which he tries to avoid as far as he can manage. For example, "Saying something,. . . produces . . . consequential effects" (1980:101);". . .what we bring about or achieve by saying something" (1980:109); ". . . any . . . perlocutionary act is liable to be brought off. . . "(1980: 110; emphasis mine). The verbs "to produce", "to bring about", "to achieve", and "to bring off "contribute to the impression that they designate the acts S performs. But how does S produce, or bring about, or achieve, or bring off the consequential effects? Well, s/he causes them. Austin, in a footnote, explains what he means by "causes things":
". . . the sense in which saying something produces effects on other persons, or causes things, is a fundamentally different sense of cause from that used in physical causation by pressure, etc. It has to operate through the conventions of language and is a matter of influence exerted by one person on another. "(1980: 113; emphasis mine)
That is, perlocutionary causation is a matter of interpersonal influence exerted through linguistic communication. Let us call this sense of causation verbal-influential causation. It is however not the only sense of causation found in the literature. There are others detected either in Austin""s own work or in his followers"".
Physical (or mechanical) Causation. This can be illustrated by Sadock""s, and Bach and Harnish""s example of S awakening H by "Don""t wake up" (see 2. 1 above). The situation for this perlocutionary act is not difficult to imagine: H was asleep; S intended to wake H up by a verbal means; and H was awakened as a result of S""s uttering "Don""t wake up". Two things immediately strike us as being unusual. First, S and H were not engaged in communication, since H was asleep, and could not process the utterance. Second, if S sincerely intended to wake H up, the propositional content of the utterance is at odds with his intention.
This example was obviously cooked up to illustrate that the perlocutionary act can be performed by merely uttering acoustic sounds, although "Don""t wake up" is a full-blooded utterance. H""s being awakened is beyond H""s control if the noise is loud enough to wake him up. There is a simple chain of physical causation: S caused his speech organs to vibrate, which caused the air to vibrate, which caused H""s eardrum to vibrate, which caused H to wake up. By transitivity of causation, S caused H to wake up. Hence S performed a perlocutionary act of awakening H. To allow the perlocutionary act to be brought off by acoustic means in fact introduces to the theory of the perlocutionary act the sense of physical causation in spite of Austin""s own warning against such a move. 
Verbal Causation. This can be exemplified by Davis"" S frightening H by saying "There""s spider on your lap" (1980; see further analysis later on). Verbal causation, differing from physical causation, operates through linguistic communication. S causes his speech organs to utter, according to (say, English) grammatical rules, some sounds which become an utterance, e. g. "There""s a spider on your lap", which is understood by H, whose understanding of what is said causes him/her to be frightened. By the transitivity of causation, S is attributed with performing a perlocutionary act of frightening H.
Let us go back to the verbal-influential causation mentioned above. Austin nowhere elaborates what he means by "influence". I construe his notion of causation as "a matter of influence" in terms of what Hume calls "moral cause": "By moral causes, I mean all circumstances , which are fitted to work on mind as motives or reasons"(Hume, 1985:198; emphasis mine). "Influence" can thus be defined as force derived from all circumstances of the interaction which are fitted to work on H""s mind as motives or reasons for adopting a certain behaviour. Verbal-influential causation differs from verbal causation in that the effect caused is not assigned alone to the propositional and/or illocutionary act performed, but also to other factors such as interpersonal relations, power relations (political or administrative or academic), motives and other contingencies. Suppose that S and H are fellow travellers who have just met on the train. S wants to convince H that an earthquake is imminent. Thus S tells H, with all seriousness: "An earthquake is imminent". It is very unlikely that H will jump at accepting the proposition, or even take the remark seriously. But once he is told that S is a well-known seismologist, H may change his/her attitude drastically and accept it as being true. H""s acceptance is not due to S""s utterance (counting as propositional or illocutionary act) alone, but also due to S""s influential power derived from his expertise.
Arguments can be made in favour of each sense of causation against the other two. Those perlocutionary acts such as the "Don""t wake up "example favour physical causation against the verbal and verbal-influential ones. For to explain the act of awakening H in terms of linguistic communication or interpersonal influence sounds bizarre if not ridiculous. Similarly, it seems unnecessary or bizarre to evoke interpersonal influence to account for the frightening-with-a-spider example (but see later).
Evidence abounds in favour of verbal-influential causation. The previous earthquake example is one instance. It will become much clearer if we consider the following. Suppose S says to H, seriously of course: "Would you please lend me a hundred pounds?" The perlocutionary effect intended here is H""s handing over the money requested. Should perlocutionary effects be physically caused in the way the loud cry of "Don""t wake up" causes H to rise, then perlocutionary acts would be the simplest way in the world to get rich. Similarly, if H""s understanding of what is said or meant by the utterance were sufficient itself to cause H to draw out the money, desires and wants would be satisfied in a full and effortless manner. This is not denying that the utterance cannot be followed by H""s handing over a cheque. The point is that the effect is not caused physically or verbally: it has to be explained in such terms as interpersonal relations (e. g. S and H are close friends), power relations (S is H""s employer) , motives (e. g. H has long wanted to please S), or other contingencies (e. g S was just pickpocketed). These parameters jointly or separately exert influential force on H""s compliance with S""s request.
3. 2. Evils of the Causation Thesis
In what follows, however, I shall argue against the very notion of causation, physical or verbal or verbal-influential. Some ground preparation is in order here. Let us first make clear what is meant by perlocutionary effects. Perlocutionary effects, except intrapersonal consequential effects (which will lead to many philosophical and psychological pitfalls and hence cannot be adequately dealt with within the reasonable space of this paper) are in fact H""s responses to S""s verbal behaviour. H""s responses can be characterized by the following six major categories (cf. Gaines, 1979: 209).
4. Negative response -- H stopped the present behaviour or refrained from doing certain behaviour.
5. Verbal response -- performing a speech act: e. g. H answered S""s question, etc.
6. Physical response ¡ª¡ª performing a physical act:e. g. H pulled a gun and shot the woman at S""s ordering "Shoot her!"
(Notice that a higher-order response may presuppose a lower-order one, e. g. a physical response may be preceded by a cognitive response. All the categories except the first presuppose the process of linguistic communication. )
Now, to say that S""s saying something (abbreviated as U hereafter) produces some consequential effect is to say that U causes, physically or verbally, or verbal-influentially, one of the six responses (abbreviated as R). That U causes R allows for the ensuing interpretations.
Given the normal standing conditions, there are:
A. Universal correlation:
a"" for all Hs, regardless of who S is, it holds true that if U, then always R; if not U, then never R.
a" for all Hs, regardless of who S is, it holds true that if U, then always R.
B. Statistical correlation:
b""(i)for some Hs, regardless of who S is, it holds true that if U, then always R; if not U, then never R.
(ii)for some Ss, regardless of who H is, it holds true that if U, then always R; if not U, then never R.
b" (i)for some Hs, regardless of who S is, it holds true that if U, then always R.
(ii)for some Ss, regardless of who H is, it holds true that if U, then always R.
C. Unique correlation:
c"" for a particular H and with a particular S under a particular circumstance, it holds true that if U, then always R; if not U, then never R.
c" for a particular H and with a particular S under a particular circumstance, it holds true that if U, then R. 
(Notice that A entails B, and that B entails C. That is, if U-R relations satisfy A, it will logically satisfy B and C; if they satisfy B, it will logically satisfy C. But C negates B, and B negates A. That is, if U-R relations satisfy C only, then they will be logically disqualified for B and A; if they satisfy B only , they will be disqualified for A. These logical relations are taken for granted in the arguments to follow. )
As is known, given two events, X and Y, if a universal correlation holds between them, the causal statement "X causes R" offers the most exhaustive and fullest explanation of them; if a statistical correlation holds, the causal statement offers only partial explanation of some co-occurrence of the total events, leaving the remaining unaccountable; if a unique correlation holds, the causal statement is worse than useless, for it presents a misleading and oversimplified picture of what actually takes place. For each instance of the co-occurrence of X and Y is unique and distinct from each other, thus calling for explanation in its own terms.
Now let us see how the three kinds of perlocutionary causation fare against A-, B- and C- correlations. A little reflection will show that a""-,b""-and c""- versions, which are equivalent to the statement that ""R if and only if U"", are too strong for perlocutionary causation of any kind. R of the six categories (a"", a", b"", b"; c""and c")can occur, in theory, independently of U- H may produce R of his/her own accord, or due to causes other than U. This entails that U cannot be necessary for R. (This is another way of saying that there is no one-to-one steadfast and biconditional correlation between an utterance and a certain perlocutionary effect. )
a"-, b"-and C"-versions which have the form "if U, then R", imply that U is sufficient for R. In fact this is the assumption encouraged by the current theory of the perlocutionary act. Let us first deal with this assumption before we look at the three versions in detail. It is derived from an unstated argument as follows. H will not produce R without S""s saying something (which is like the preparatory condition of the perlocutionary act). S says something, and H produces R. Therefore S""s saying something is sufficient for R. Valid as it may appear, its conclusion is however false. The money borrowing and the earthquake examples above clearly show that U is not sufficient for R. The awakening-H instance, and the spider-frightening act, however, seem to support the conclusion that U is sufficient for R. It may immediately appeal to us that a loud enough cry is sufficient (physically) to wake up a sleeping H. But is it sufficient by itself? In fact it is not, in spite of its immediate appeal. It is what Mackie (1965)terms "an insufficient but necessary part of a condition which is itself unnecessary but sufficient" (INUS for short) for the outcome. In plain words, H can wake up with or without S""s uttering anything. The latter is therefore not necessary for the former. That S""s uttering the noise does wake H up offers no proof that it is sufficient itself for the consequence. For it will not cause H to rise without everything else being normal(e. g. H""s ears work well; H""s nervous system functions as usual; etc. Or in negative terms. H is not dead or anaesthesized, and so on). The given normal conditions plus S""s uttering the sounds loud enough make up the unnecessary but sufficient condition which cause H""s waking up. 
Davis"" analysis of "There""s a spider on your lap" also encourages the assumption that U is sufficient (verbally) for R. Davis"" account deserves a detailed examination, for it is the latest and most elaborate statement on verbal causation found so far. Davis"" notion of verbal causation is derived from Austin""s verbal-influential causation. He finds that Austin""s use of the term "influence" is "not quite the right word". Suppose that S tells H that it is raining, and because of S""s telling, H comes to believe that it is raining. This is "a perfectly good example of a perlocutionary act". In performing this act, so Davis argues, S has not exerted any influence on H(see Davis,1980: 43).
Davis"" scheme of perlocutionary causation runs as follows. It operates through the linguistic competence of both S and H. But this is not enough. It has to work through perlocutionary uptake, and Davis insists that it or something like it be necessary for the performance of the perlocutionary act. S""s saying something (by exercising his linguistic competence) counts as the illocutionary act which causes H to understand (by exercising his linguistic competence) what is said, H""s understanding of which causes H to produce a perlocutionary effect. This is not the whole story yet.
"what H understands S to mean might not be the immediate cause of the effect in question, but only the proximate cause. Because of what H understands, he might form some belief, thought, want, desire, etc. which causes the perlocutionary effect. "(1980:52)
The causal chain is diagrammed in Fig. 2.
(Notice that cause 1 is interpersonal causation, whereas causes 2a, 2b, and 3 are intrapersonal causation. Davis talks about them as if they belonged to one and the same category. Causes 1, 2a, and 3 on the one hand, and causes 1, and 2b on the other, comprise two subtypes of what we have termed verbal causation. )
The spider example is anatomized thus:
(i)I tell you that there""s spider on your lap by uttering "There""s a spider on your lap".
(ii)I mean by "There""s a spider on your lap" what it means in English.
(iii)You understand that I have told you something by uttering, "There""s a spider on your lap".
(iv)You understand what I mean by "There""s a spider on your lap".
(v)What you understand me to mean causes you to become frightened.
(vi)I intend that you become frightened because you understand what I mean. (1980: 53)
This is supposed to be the standard performance of the intentional perlocutionary act of frightening H. In this instance, U causes R through causes 1 and 2b. Davis""(i) and (ii) constitute cause 1, of which (iii) and (iv) are the effect, which in turn serves as cause 2b immediately causing the perlocutionary effect [i. e. (v) above].Cause 1 operates so long as what Searle(1969) calls the normal input and output conditions of communication obtain. It is cause 2b (i. e H""s understanding of the message) that is decisive. Is it sufficient for causing H""s being frightened? There are at least the following five possibilities overlooked by Davis.
(1)H was frightened by the mere mentioning of the word "spider". H did not even check whether there really were any spiders on his/her lap.
(2)H was frightened by his/her understanding that there was a spider on his/her lap. Being frightened took place before H actually saw the creature on his/her lap. H reacted on the belief that S was telling the truth.
(3)H was not initially frightened because s/he did not believe that S was telling the truth until s/he saw the creature on his/her lap.
(4)H was not frightened, for s/he thought that S was kidding, and indeed it was the case.
(5)H was not frightened but delighted at the news, for s/he was a spider collector.
The previous analysis vindicates the following observations. (1)For the utterance "There""s a spider on your lap" to cause the effect of H being frightened at all, H, no matter who, must be afraid of spiders, real or sham, big or small. Without this necessary condition, the utterance or even spiders in the flesh will never make H frightened.
(2) In the first case, the utterance is made sufficient by the fact that H suffers from a pathological fear of spiders. In the second case, it is made sufficient by the intervening belief that S was telling the truth. In the third case, the utterance, i. e. as a symbolic substitute for a real spider, is insufficient: H refuses to be frightened unless s/he is confronted with a live creature. In the fourth the utterance fails again, but for a different reason. This instance represents a situation in which S intended to frighten H by telling a lie that there was a spider on his/her lap so that he might play a nasty trick on H. But S""s intention of frightening H and of playing a trick on H was seen through. In the last instance, S""s intention of frightening H, the spider and the speech act, are all futile, for H was not afraid of spiders after all. The two successful cases and three failures all point to the fact that U is by itself insufficient for R. Verbal causation, as physical causation, is an INUS condition and plays no more than a triggering role.
As for verbal-influential causation, the way we have construed it is itself a recognition of the fact that U is not sufficient itself for R. Austin""s observation that perlocutionary causation as "a matter of influence exerted by one person on another" implicitly advises us to go beyond the immediate utterance act to seek for the true cause (which, I shall argue later, is also misleadingly futile).
What is to be gained after this laborious argument that U is neither necessary nor sufficient for R? The point lies in the next move: if U is neither necessary nor sufficient for R, then the statement that U causes R is misleadingly false and ought to be abandoned. The objection to this conclusion may go as follows. It is admittedly true that U is neither necessary nor sufficient for R. But U is an INUS condition, as has been shown. The statement that U (as an INUS condition) causes R does make sense. Though this counter-argument seems to have some plausibility, it is not difficult to dispense with. If we are able to show that U (as an INUS condition), except in some peripheral or even arguably negligible cases, produces a unique correlation only, the statement U (as an INUS condition or otherwise)causes R is worse than useless (the reason for this has been already given, see the first half of this section above and also later). For ease of reference, the part of the three types of correlation aforementioned which is pertinent to the argument here is reproduced below.
Given the normal standing conditions:
A. Universal correlation:
a" for all Hs, regardless of who S is, it holds true that if U, then always R.
B. Statistical correlation:
b" (i)for some Hs, regardless of who S is, it holds true that if U,then always R.
(ii)for some Ss, regardless of who H is, it holds true that if U, then always R.
C. Unique correlation:
c" for a particular H and with a particular S under a particular circumstance, it holds true that if U, then R.
Can we find U which satisfies a"? The loud cry "Don""t wake up" is perhaps a candidate. It appears intuitively true that, given the normal standing conditions, H, no matter who, is bound to wake up to the noise, no matter by whom it is uttered, if it is loud enough for the result. But it is hardly deniable that perlocutionary acts, such as awakening H by a loud cry, are peripheral, or even arguably nonperlocutionary (see e. g. Gaines, 1979: 211). What about b"? We may well think along this line. Some people are afraid of spiders (or snakes, or anything). Then regardless of who issues the relevant utterance, these people will always be frightened. This is however wishful thinking. Suppose that a schizophrenic sees a big spider on the lap of the caring nurse. The patient cries to the latter: "There""s a spider on your lap". Although the nurse is habitually scared of spiders, real or symbolic, she may pay no heed to the warning, showing not the slightest sign of being frightened.
Perhaps we may think of situations where both U and R are institutionalized, e. g. a wedding ceremony, a graduation congregation, a religious service, a court ruling, a military transaction, and so on. Even in these situations, however, failures are not unknown. For example, the bridegroom had his schizophrenic wife hidden in the attic; the graduate refused to accept the degree; the private rebelled against the lieutenant""s order. Moreover, U does not cause R, for UR correlations are either agreed upon beforehand, or secured and protected by institutional force, and from the point of view of linguists, the perlocutionary acts thus performed are of little interest to the theory of the perlocutionary act (although they may be important to sociopsychologists and anthropologists).
Now we may safely assume, with the misgivings just dealt with, that U causing R results in a unique correlation. Even in this case it will still be naive to assume that for a particular H with a particular S under a particular circumstance, it holds true that if U then R. Suppose that, on occasion A, John successfully frightened Jane with the utterance "There""s a spider on your lap", with no actual spiders backing it up. On occasion B, however, there was indeed a live creature on Jane""s lap, of whose presence she was totally unaware. Under this particular circumstance, we cannot rest assured that Jane will be frightened by the same utterance. This time Jane may take no more heed of John""s utterance than to dismiss it with disdain.
The point we have tried to drive home is that the U-R correlation is unique and that U is neither necessary nor sufficient for R. If U is ideed followed by R, U is an INUS condition, and plays no more than a triggering role. This applies alike to the three types of causation discussed above.
We rule out purely physical/nonverbal causation (with U functioning as acoustic sounds only) as perlocutionary causation (in fact it ought to be truly called "perlocutionary effects trigger"). For by doing so we are able to confine perlocutionary phenomena within the boundary of linguistic communication. I dismiss the verbal and verbal-influential causations as types of perlocutionary causation on the following grounds. 
One. The statement that U causes R is a gross over-simplification of what actually takes place, and is notoriously misleading.
Two. The theory is incapable of explaining cases where U is not followed by R. That is, it cannot explain why U fails to cause R.
Three. It is a myth to the theory that one U should cause more than one R (recall the Multiplicity Thesis). 
Four. It denies H his/her legitimate claim to agency. It has been pointed out (at the beginning of this section) that H""s response (i. e. R) falls into six categories. Let us first take a look at H""s physical and verbal responses. It seems to be undeniable that these two types of response (e. g. handing over a hundred-pound cheque, answering S""s question) are full-blooded acts performed by H. Here we have an interactive situation in which S performs a speech act to H, and H produces a non-verbal or verbal act in response to S""s speech act. Treating H""s response-act as a consequence of S""s speech act denies H""s status as agent of the response-act.  It is only through denying H such status that S can be truly said to have caused the response and that S can thus be attributed with the performance of a perlocutionary act. Otherwise the perlocutionary act is not something done by S alone, but a transaction, a joint endeavour involving S""s speech act and H""s response-act.
In fact Austin would not deny that H""s physical or verbal responseact is an act. Nor would he allow another act to be a consequential effect of a speech act, for "consequence-language" (Austin""s term)does not allow another act to be part of the initial stretch of an action(see Austin, 1980:117).  It would be inconceivable that one agent""s act should be allowed to have another agent""s as its component, while claiming that there is only one act performed by the former. Austin""s notion of "consequential effects on. . . actions "therefore must be construed in psychological terms. Take Austin""s own example of S""s persuading H to shoot the woman (1980:101). The "consequential effect" does not include H""s physical response-act of pulling a gun and shooting her, but refers to H""s attitude change, a disposition to the performance of the physical response-act.
Along this line of thinking, perlocutionary effects must be psychological or mental events, e. g. H""s accepting as being true that an earthquake is imminent; H""s being frightened, etc. That is ,H""s negative, cognitive, emotive and motor responses can be candidates of perlocutionary effects.
Yet, are there any mental acts? Austin was not prepared to accept mental acts. "Is to think something . . . to do an action?" Austin asks in brackets (see Austin, 1979: 179). Austin admits that things like thinking, trying, belong to "the most patent exceptions or difficulties" (ibid. ). According to Austin, an act or action involves at least some minimum physical doing (see Austin, 1980: 107). What is essential to Austin""s notion of perlocutionary effects, therefore, is the assumption that mental events taking place inside the addressee""s skull are not acts. So H has no claim of agency concerning the effects produced in him. The effects are indeed caused by the illocutionary act, etc, which is performed by S. By transitivity of causation, S can claim the agency of the perlocutionary effects. Hence S is credited with performing a perlocutionary act.
Taylor (1963, 1970), however, argues that the distinction between the things done by the agent and those done to the agent is also applicable to mental events. Thus: (1) and (2) are different from (3) and (4) (adopted from Taylor, 1963: 321).
(1)I recall the telephone number (using, e. g. , mnemonic devices)
(2) Something makes me recall the number.
(3) The number occurs to me.
(4) Something makes the number occur to me.
Taylor shows that (1) and (2) are instances of mental acts, whereas (3) and (4) are merely happenings. Geach, in his book-length treatment of mental acts defines mental acts thus: ". . . if somebody puts into words, not parrotwise but with consideration, there occurs a mental act . . . "(Geach, 1956:9).
Let us come back to H""s negative and cognitive responses. Negative response involves changing one""s mind or attitude and refraining from doing something which one previously intended to do. This can be regarded as a special case of cognitive response. Before H produces a cognitive response, H first of all processes what S says to him. This includes H""s hearing and interpreting the message. Austin treats S""s saying something in terms of a series of acts, whereas H""s hearing the utterance, understanding the meaning of the utterance, inferring the illocutionary force, etc. are not given a similar analysis. How could Austin justify his unequal treatment? Is "hearing the utterance", etc. , merely a happening to H or something done by H? "Hearing the utterance" differing from "listening to something", carries the overtone that the person who hears something cannot help hearing it, so long as s/he has ears and the ability to hear. But in the case of interpersonal communication, is "hearing something" a passive thing happening to the addressee, and therefore not an act? Granted that it is not, what about the understanding of the locutionary meaning, or the recognition of illocutionary force, or the working out of conversational implicature? We may conclude, with some reservation perhaps, that the addressee""s understanding of sense, his/her recognition of force and his/her inferring of implicature are not merely happenings, caused to take place in him/her, but that they are things that s/he does, though they may be invisible to the naked eye, which distinguishes them from the acts involving physical doings. If we follow Taylor and Geach, they are mental acts. Since H""s cognitive response presupposes H""s performance of these mental acts, H""s negative and cognitive responses are all the more responseacts. They cannot therefore be perlocutionary effects.
Finally, one thing which is common among H""s emotive and motor responses is the varying degrees of incontrollability and lack of intentionality on H""part, i. e. H, to a certain degree, cannot help producing the responses, given the appropriate stimulus (e. g. a spider, a loud nose, etc. ). But the emotive response proper presupposes message processing. Hence, H""s emotive response involves H""s performance of mental acts. The motor/reflexive and motor/emotive responses, in contrast, occur without intervention of linguistic communication. Perhaps only such responses can be regarded as mere happenings to H, and will disqualify H from claiming agency. It is probably this kind of response that can be legitimately treated as consequential effects of S""s saying something; and only this kind of consequential effect can entitle S to claim agency for consequential effects, and to be attributed with the performance of perlocutionary acts.
Even if one is not prepared to acknowledge "mental acts", s/he has to admit that no addressee is a mere robot capable of being caused by S to produce automatic perlocutionary consequences. We take the following as facts:
(1)H is not a robot. S/he is just as much an agent as S.
(2)H""s role in interaction is just as active as S""s.
(3)H can always, in theory, choose the way s/he will respond to what is said.
Restoration of the addressee to the status of an active agent, and recognition of mental acts lead to the conclusion that the so-called perlocutionary effects are not in fact caused by S, but actively produced by H, who has the claim to the agency of the effects. Thus, the perlocutionary act cannot be said to be performed by S alone. It is a joint endeavour between S and H. It involves S""s performance of speech acts and H""s performance of response-acts. The relation between S""s speech acts and H""s response-acts is anything but causal.
3. 3. The Multiplicity Thesis, the Infinity Thesis, the Intention Irrelevance Thesis and the Effect = Act Fallacy
The Multiplicity Thesis is in essence a claim that one U may cause more than one R, hence giving birth to more than one perlocutionary act. The Infinity Thesis, on the other hand, asserts that the sorts of R, and the number of R to be caused by U know no bounds, and by one U one can possibly perform any kind of perlocutionary acts. These two theses together with the Intention Irrelevance Thesis will inevitably lead to what we propose to call the Effect = Act Fallacy. Let us first exemplify this fallacy. Suppose that A, seeing B""s house on fire, shouted to B: "Your house is on fire". As a result of A""s warning, B became alerted and alarmed. We say that by saying "Your house is on fire "A has performed a perlocutionary act of alerting B and a perlocutionary act of alarming B. But it may have turned out that B was seized with a heart attack and died as a result of A""s warning. We are not prevented, for perlocutionary effects can be unintended, from saying that by saying "Your house is on fire" A performed a perlocutionary act of getting B to succumb to a heart attack, in spite of the fact that A swore with his hand on the Bible that he had no such intention at all and that he had not even the slightest idea that B had suffered from a heart trouble in the first place. It may also be the case that at the warning. B appeared to be alerted and panicking, but inwardly he was delighted, for B him/ herself had deliberately set the house on fire in order to claim a large sum of insurance money. A will be in this situation attributed with the performance of the perlocutionary act of alerting B and the perlocutionary act of delighting B. There is no way for S to escape this iron grip of effects.
If we feel something absurd in this analysis, it is certainly justified. What S actually said is almost irrelevant to the perlocutionary acts attributed to him/her. It is the effects that say the last word. It is implicitly assumed that, where there is an effect, there must be an act. The effect is then used to define and determine what the act is to be. Once we recall that the effect is H""s response-act, we shall soon detect the implausibility, namely, although S is attributed with the performance of perlocutionary acts, what S him/herself says or does is not a defining or determining factor of the sorts of perlocutionary acts we have attributed to him/her. Rather they are defined and determined by the way H responds.
Now one may point out that in actual fact, S""s saying something may indeed be followed by various effects. We have no intention to deny this. Our quarrel with Austin and others is that such effects should not be expounded in terms of causation or the acts performed by S. They ought to be explained in transactional terms with the recognition of hearers as agents of the effects. The transactional approach (I have termed it the rhetorical approach elsewhere, see Gu,1987) will not only de-mystify the one-U-causing-many-R phenomenon which the Causation Thesis finds mysterious, but will also keep clear of the absurdity the Effect = Act Fallacy leads to. S""s saying something plays a triggering role in the whole transaction. H""s response is H""s contribution to the transaction. How H is to respond is mainly up to H. The trigger may remain the same, but the response varies with the individuals involved.
3. 4. The meaning of "By saying x I did y"
If our previous argument holds its ground, that is, it is treacherous to say that U causes R, and the perlocutionary act is not a single act, but a transaction, will Austin""s formula "By saying x I did y" still make sense? In other words, is it still plausible to talk about perlocutionary acts? Our answer is that there are only a few marginal cases where it can be truly said that S has performed a perlocutionary act of y, i. e. those cases where H""s response is motor reflexive and incontrollable (physically or mentally or by institutionalized force).Apart from these, the formula is utterly misleading if not nonsensical. But one may argue that in ordinary uses of language, people do sensibly talk about such things as "Jack frightened me", "John convinced me of the drug""s harmful effects", and many others. As for this appeal to the ordinary usage of perlocutionary verbs, we must bear in mind the ensuing three points.
First, perlocutionary verbs are convenient labels used to bracket some consequential effects, as Austin himself points out:
"a single term descriptive of what he did may be made to cover either a smaller or a larger stretch of events, those excluded by the narrower description being then called ""consequences"" or ""results"" or ""effects"" or the like of his act. "(1979: 201)
Given a whole stretch of events, an illocutionary verb is a single term which gives a narrower description covering a smaller stretch of events, excluding the remaining which will be called perlocutionary consequences. Thus Austin writes:
"the vocabulary of names for acts (B) [i. e. illocutionary acts] seems expressly designed to mark a break at a certain regular point between the act (our saying something) and its consequences (which are usually not the saying of anything), or at any rate a great many of them"(1980:112)
Austin""s idea boils down to the following. When we perform an illocutionary act, the act performed has its consequences. Thanks to the vocabulary of names for illocutionary acts, we can detach the acts (linguistic) from their consequences (usually non-linguistic). In other words, our language about illocutionary acts facilitates us to talk about them without mentioning their consequences. What perlocutionary verbs or expressions do is embrace a greater stretch of events which encompass both the action we do (i. e. the illocutionary act) and its consequences. This perlocutionary description of a greater stretch of events by way of perlocutionary verbs or expressions is then metamorphasized by Austin into perlocutionary acts, as shown in Fig. 3.
In this linguistic manoeuvre of transforming perlocutionary descriptions into perlocutionary acts, it is erroneous to assume that linguistic descriptions mirror and hence define acts. This is the second point to be borne in mind. Relying on the ordinary usage of perlocutionary verbs in classifying perlocutionary acts, as Gaines does, will fall victim to the Perlocutionary Verb Fallacy.
Finally, the perlocutionary formula "S performed a perlocutionary act of y by saying x" or "By saying x I did y", ought to be interpreted as a claim to responsibility or liability for some consequences on part of S. "Jack frightened me" claims, justifiably or otherwise, that Jack is responsible for my being frightened. It may serve some psychological or legal purpose, but it does not truly depict the nature of the interpersonal transaction. The formula also captures the afterthought, viz. H would not have produced R without S""s uttering U. It can never be overemphasized that U plays no more than a triggering role in the whole of perlocutionary affairs. Austin""s "By saying x I did y" should be recast as "S makes H do sth by saying x" or "S gets H to do sth by saying x". This formula has the obvious advantage over Austin""s in that it makes explicit that there are two acts involved, one by S and the other by H. 
4. Implications for pragmatics
The dismantling of the perlocutionary act certainly has some significant consequences on the current theory of pragmatics. In this final section we would like to outline two areas of pragmatics which are more prominently affected by our foregoing critique, namely the distinction between illocution and perlocution, and Grice""s theory of conversation.
4. 1. Illocution vs. perlocution dichotomy
As far as I know, Campbell (1973) is the only author who argues that the distinction between the two collapses, because all illocutionary acts produce perlocutionary effects. He concludes by observing:
"not only ""that the concept of illocutionary force, as Austin himself defines it, is empty"", but that the concepts of perlocutions and the production of perlocutionary effects are similarly empty. "(1973:296)
Sadock finds that illocutionary acts have all the characteristics of perlocutionary acts, but not vice versa. He then moves to suggest that "an illocutionary act is a special kind of perlocutionary act" with the distinctive features of being conventional and directly related to meaning (1974:153).
Gaines emphasizes the differences between the two which, Gaines holds, operate in two different systems. The system within which illocution operates is "strictly conventional"; the structure of this system allows for an analytic "counts as" relation. "The system wherein perlocution operates is quite different" (1979:209). Gaines does not continue to elaborate a perlocutionary system, but he points out that perlocutionary acts are "consummated". A perlocutionary act consists of two components: (1) an initiatory illocutionary utterance (I), and (2) a consequential effect (P) of I which consummates PA (i. e. perlocutionary act). The relationship between illocution and perlocution is "one of part to whole" (ibid. :214).
Van Dijk distinguishes two kinds of illocutionary acts: the I-successful illocutionary act and the P-successful illocutionary act. The former is performed when the speaker""s illocutionary intention is recognized by the hearer through his comprehension of the utterance. The latter is performed if the purpose of the speaker is reached (1977:198). A P-successful illocutionary act is "also called a perlocutionary act". "Hence a perlocutionary act is an act of which the conditions of success are given in terms of purposes of the speaker with respect to some change brought about in the hearer AS A CONSEQUENCE OF the illocutionary act" (1977: 198). Consequences of illocutionary acts are also perlocutionary effects. Van Dijk soon observes: "Perlocutionary effects are . . . beyond the domain of a linguistic theory of pragmatics". They are so because whether the hearer behaves in the way proposed by the speaker is "beyond the control of the speaker and beyond the conventional norms of communicative interactions" (1977:198).
Van Dijk, unlike Sadock and Gaines, clearly draws a line between illocution and perlocution. This view is shared by Bach and Harnish (1979) and Leech (1983). Bach and Harnish argue that the illocutionary act is an act of linguistic communication, "an act of expressing an attitude by means of saying something"(1979 :vx). The distinctive feature of the illocutionary act is its reflexive intention, whose fulfilment consists of its recognition by the hearer (1979: 15).The act is successful, that is communication has been achieved, if the hearer identifies by means of recognition of the reflexive intention, the attitude expressed in the way the speaker intends him to identify it (1979:xv). Unlike illocutionary intention, which is reflexive, perlocutionary intention can be overt (i. e. recognized or intended to be recognized), or covert (i. e. not recognized or intended not to be recognized), or reflexive. When it is relflexive, it is not communicative in the way that illocutionary intention is: the fulfilment of perlocutionary intention does not rest on the recognition but on the production of some further effects (1979: 80). Perlocutionary effects, including any effects on feelings, thoughts, and actions of the addressee, are thus quite different from illocutionary ones. Bach and Harnish therefore reach the conclusion that the illocutionary act is in the domain of linguistic communication, for "[i]t is sufficient that H recognize S""s R-intention, S""s expressed attitudes. This is what communication is about; anything more is more than just communication" (1979:16). Thus, the perlocutionary act, which goes far beyond H""s recognition of S""s expressed attitudes, lies beyond linguistic communication.
Unhappy with the tendency of talking about everything we do with words in terms of action (Leech, 1981:414), and with the flirting by Austin and Searle with "the Performative Fallacy" and "the Illocutionary-Verb Fallacy "(Leech, 1983:175), Leech defines the force of an utterance U by a set of conversational implicatures, a subset of which "defines the illocutionary force of U. This subset. . .defines a means-ends analysis of which the uttering of U is the central action, and thereby defines the presumed illocutionary goal of s in uttering U"(1983:153f. ). How is Leech""s theory of illocution to be related to perlocution? He finds no difficulty in doing so. "To include Austin""s perlocutionary act in a means - ends diagram, . . . one can simply add one further layer to the hierarchy "which consists of illocutionary, locutionary and phonetic levels (1983:200f. ). Thus, the perlocutionary act stands for the sequence of events enacted to reach the perlocutionary attainment. Leech, perhaps getting impatient with illocution vs. perlocution, observes: "it is unnecessary to be too deeply concerned with these distinctions: perlocutionary effects do not form part of the study of pragmatics" (1983:203).
Our foregoing critique of current perlocutionary theory endorses the stance taken by Van Dijk, Bach, Harnish and Leech on a line drawn between illocution and perlocution. The successful performance of the illocutionary act involves the utterance (i. e. the act of uttering and its product) and illocutionary intention on S""s part, and the understanding of illocutionary force or recognition of illocutionary intention on H""s part. The interactive relation between S and H is linguistic and communicative. Thus the illocutionary phenomenon falls within the domain of linguistic pragmatics or linguistic communication. We go beyond pragmatics or linguistic communication if the comprehension of illocutionary force is followed with perlocutionary effects.
As is argued in 3. 3 above, perlocution is not a single act performed by S. Nor is its effect being caused by an utterance. It involves a transaction consisting of S""s speech act(s) and H""s responseact (s). How the transaction is carried out is an issue which belongs to the domain of rhetoric. (Elsewhere I use the term ""conversational rhetoric"" so as to disassociate it from the unfavourable connotations of traditional rhetoric, see Gu, 1987, 1989,)
4. 2. Pragmatics, rhetoric and Grice""s theory of conversation
To show the interaction between linguistic pragmatics (pursued by Van Dijk, Bach and Harnish, as well as Leech, and in Sperber and Wilson, 1986), and conversational rhetoric (I have been working on since 1985), the best thing to start with is to take a look at a rhetorical transaction of minimal size (""rhetorical transaction"" is a key concept of conversational rhetoric).Compare:
A is an instance of linguistic communication. The physical response of nodding the head indicates that the communication is successful. B is an instance of rhetorical transaction, which also involves linguistic communication as shown in B"". C is an instance of rhetorical transaction as well as a "talk exchange" (Grice""s term, see Grice, 1975).
There are of course many other complications concerning "rhetorical transaction" and "talk exchange", which cannot be dealt with here. The point I intend to show is that linguistic communication, hence linguistic pragmatics (as defined by the aforementioned linguists) is basic with regard to rhetorical transaction and talk exchanges, but it is inadequate to account for talk exchanges. B and C are infiltrated with goals beyond linguistic communication, which are clearly shown in B"" and C"". Linguistic pragmatics therefore needs conversational rhetoric in order to accommodate non-communicative goals in talk exchanges.
Grice""s CP and the maxims are formulated with talk exchanges in mind. In the last ten years or so, linguists, and Sperber and Wilson in particular, have been moving in the direction of linguistic communication by filtering out noncommunicative goals in talk exchanges. In his analysis of meaning, Grice observes that
"""[S]meant something by x""is (roughly) equivalent to""[S]intended the utterance of x to produce some effect in an audience by means of the recognition of this intention"". "(1957:385)
Strawson (1964) reformulates the Gricean intention into three subintentions. To mean something by x, S must intend
(a)S""s utterance of x to produce a certain response r in a certain audience A;
(b)A to recognize S""s intention (a);
(c)A""s recognition of S""s intention (a) to function as at least part of A""s reason for A""s response r.
According to Schiffer, """a certain response r""[in (a)]. . . is equivalent to ""a certain propositional or affective attitude or action r"""(1972:10). Schiffer construes "propositional or affective attitude"as referring, respectively, to e. g. believing that p, or feeling distressed or humiliated and "action r" including physical behaviour, such as jumping to a drowning man""s rescue at his shout "Help!"
Now the connection between Austin""s notion of the perlocutionary act and Grice""s S meaning something by x becomes only too obvious. In Austinean language, one would say: S performed a perlocutionary act of y by saying x through producing a certain perlocutionary effect in an audience. In Gricean language, on the other hand, one would say: S meant something by intending the utterance of x to produce some effect in an audience by means of the recognition of the intention. There is certainly much more to be said concerning this connection. The direction oriented by Sperber and Wilson however is reductive. They argue that, in Strawson""s interpretation of the Gricean intention, (a) is not the communicative intention. The true communicative intention is (b). Communication can succeed without either (a) or (c) being fulfilled. They have dropped out (c) and reformulated (a) into what they call "informative intention", namely, "to inform the audience of something" (Sperber and Wilson, 1986:28 29). Their view of successful communication includes S""s ostensive behaviour which makes S""s informative intention manifest and H""s inferential process through which H interprets S""s ostensive behaviour and correctly recognizes S""s information.
I have no quarrel with Sperber and Wilson""s reductionist programme. My reservation however is that talk exchanges (either in Grice""s sense or in ordinary usage) are infiltrated with motives or purposes or goals which are quite significant although they go far beyond linguistic communication. The ultimate object of conversational rhetoric is to accommodate these motives or purposes or goals and provides a bridge linking linguistic communication and social interaction at large. Thus linguistic pragmatics, conversational rhetoric, and sociopsychology (as a study of social behaviour in general) are collaborators in coping with the extreme complexity of talk exchanges. (Interested readers may go to Gu, 1993, for details. )
Austin, J. A. , 1980[ 1962]. How to do things with words. Oxford: Oxford University Press.
Austin, J. A. , 1979, 3rd ed. Philosophical papers. Oxford: Oxford University Press.
Bach, K. and R. M. Harnish, 1979. Linguistic communication and speech acts. Cambridge, MA: MIT Press.
Binkely, R. , R. Bronaugh and A. Marras, eds. , 1971. Agent, action and reason. Toronto: Toronto University Press.
Black, M. , 1969. Austin on performatives. In: K. T. Fann, ed. , 1969, 401 411.
Brand, M. ed. , 1970. The nature of human action. Glenview, IL: Scott, Foresman and Company.
Campbell, P. N. , 1973. A rhetorical view of locutionary, illocutionary and perlocutionary acts. Quarterly journal of speech 59:284-296.
Cerf, W. ,1969. Critical review of How to do things with words. In:K. T. Fann, ed. , 1969, 351-379.
Cohen, T. , 1973. Illocutions and perlocutions. Foundations of Language 11:492503.
Cole, P. , and J. L. Morgan, eds. , 1975. Syntax and semantics, Vol. 3: Speech acts. New York: Academic Press.
Cole, P. ,ed. 1978. Syntax and semantics, Vol. 9: Pragmatics. New York: Academic Press.
Cornman, James, 1971. Comments. In: R. Binkely et al. , eds. , 1971, 26-37.
Davidson, Donald, 1971. Agency. In: R. Binkely et al. , eds. , 1971,1-25.
Davidson, Donald, 1980. Essays on actions and events. Oxford: Clarendon Press.
Davis, S. ,1980. Perlocution. In: John Searle et al. , eds. ,1980,37-55.
Dijk, Teun A. van, 1977. Text and context. London: Longman.
Dijk, TeunA. van, 1981. Towards an empirical pragmatics: Some social paychological conditions of speech acts. Philosophica 27(1): 127 -138,
Fann, K. T. , ed. , 1969. Symposium on J. L, Austin. London: Routledge.
Flew, Autony and Godfrey Vesey, 1987. Agency and necessity. Oxford: Blackwell.
Franck, Dorothea, 1981. Seven sins of pragmatics: Theses about speech act theory, conversational analysis, linguistics and rhetoric. In: H. Parret et al. ,eds. ,1981, 225-236.
Gaines, R. N. , 1979. Doing by saying: Toward a theory of perlocution. Quarterly journal of speech 65: 207-217.
Geach, Peter, 1956. Mental acts. London: Routledge.
Grice, H. P. 1957. Meaning, Philosophical Review 66:377-388. [Reprinted in Steinberg and Jakobovits, eds. , 1971, 53-59, and in Strawson, ed. , 1971,39 -48. ]
Grice, H. P. ,1969. Utterer""s meaning and intentions. Philosophical review 78: 147-177.
Grice. H. P. 1975. Logic and conversation. In: P. Cole and J. L. Morgan, eds. ,1975, 41-58.
Grice, H. P. 1978. Further notes on logic and conversation. In: P. Cole, ed. ,1978, 113-127.
Gu, Y. G. , 1987. Towards a model of conversational rhetoric: An investigation of the perlocutionary phenomenon in conversation. Unpublished Ph. D. Thesis, University of Lancaster.
Gu, Y. G. ,1989. An outline of conversational rhetoric. Journal of Foreign Languages Teaching and Research 2:8-15.
Gu. Y. G. ,1993. Pragmatics and rhetoric: A collaborative approach to conversation. In: H. Parret, ed. , Pretending to communicate, 112-127. Berlin:Walter de Gruyter.
Hornsby, Jennifer, 1980. Actions. London: Routledge.
Huge, Jerald and Barbara Foley Meeker, 1988. Social causality. Boston, MA: Unwin Hyman.
Hume, D. , 1985. Essays moral, political and literary. Ed. by E. F. Miller. Indianapolis. IN: Liberty Classics. [1741-77].
Leech, Geoffrey N. , 1981. Pragmatics and conversational rhetoric. In: H. Parrett, M. Sbis¨¤ and J. Verschueren, eds. , 1981, 413-442.
Leech, Geoffrey N. , 1983. Principles of pragmatics. London: Longman.
Levinson, Stephen C. , 1983. Pragmatics. Cambridge: Cambridge University Press.
Mackie, J. L. , 1965. Causes and conditions. American philosophical quarterly 24: 245-64.
New, C. G. , 1969. A plea for linguistics. In: K. T. Fann, ed. , 1969, 148-165.
Parret, H. , M. Sbis¨¤ and J. Verschueren, eds. , 1981. Possibilities and limitations of pragmatics. Amsterdam: Benjamins.
Sadock, Jerrold M. , 1974. Toward a linguistic theory of speech acts. New York: Academic Press.
Schiffer, S. R. , 1972. Meaning. Oxford: Clarendon Press.
Searle, John R. , 1968. Austin on locutionary and illocutionary acts. Philosophical review 77:405 - 424.
Searle, J. R. ,1979. Expression and meaning: Studies in the theory of speech acts. Cambridge: Cambridge University Press.
Searle, John R. ,1969. Speech acts. Cambridge: Cambridge University Press.
Searle, John R. , ed. 1980. Speech act theory and pragmatics. Dordrecht: Reidel.
Sosa, E. ed. , 1975. Causation and conditionals. London: Oxford University Press.
Sperber,D. and D. Wilson, 1986. Relevance. Oxford: Blackwell.
Strawson, P. F. ,1964. Intention and convention in speech acts. Philosophical review 73:439-460.
Taylor, R. ,1963. The stream of thoughts vs. mental acts. Philosophical quarterly113:311-321.
Taylor,R. ,1980[ 1966]. Action and purpose. Humanities Press.
Taylor, R. , 1970 [ 1969]. Thought and purpose. In:M. Brand, ed. , 1970,267283.
Gu Yueguo(¹ËÔ»¹ú), Ph. D. ,resarch professor of linguistics, head of Contemporary Linguistics Department, the British Academy K. C. Wong Fellow, is interested in pragmatics, discourse analysis, the philosophy of language and rhetoric. His published works include 17 research papers in international journals, 13 artcles in national journals, and 15 English textbooks. He is one of the editors of the Journal of Contemporary Linguistics, and special issue editors of Text and Journal of Pragmatics.
. Related to this thesis is the view that S""s saying many things may jointly contribute to one effect. This many to one causal relation poses no problem nor offers any support to the Causation Thesis (to be dealt with later on). It is therefore ignored in this paper.
. In the majority of cases, acts like insulting sb, making sb sad, scaring sb, do not have the so-called called perlocutionary object or sequel, nor direct associated perlocutions.
. Cohen, Sadock, and Gaines embark on classifying perlocutionary acts. Cohen and Sadock classify the acts by way of perlocutionary causes, while Gaines does so by way of perlocutionary effects. Note that attributing perlocutionary acts is different from, though related to classifying perlocutionary acts.
. Some may point out that for Austin this may not be a case of perlocution. However, Austin certainly allows for the perlocutionary act to be performed by non-locutionary means such as waving a stick or pointing a gun (Austin,1980:119). From this we may well infer that it was likely that Austin would allow the "Don""t wake up"case to be perlocutionary.
. The three-type correlations of U causing R can be exemplified in plain language as follows. Given the normal standing conditions, and the utterance ""Shoot her! ""(Austin, 1980:101) uttered in its full normal sense and in all earnest:
A. it satisfies the universal correlation, (a"") if no matter who utters it, the addressee, no matter who, upon hearing the utterance, always shoots the woman, and the addressee, no matter who, never shoots the woman without hearing the utterance no matter by whom it is uttered.
(a") if no matter who utters it, the addressee, no matter who, upon hearing the utterance, always shoots the woman.
B. it satisfies the statistical correlation, (b""i) if no matter who utters it, there is/ are always some addressee (s) who, upon hearing the utterance, always shoot (s) the woman, and there is/are some addressee (s) who never shoot (s) the woman without hearing the utterance no matter by whom it is uttered.
(b""ii)if it is uttered by some certain speaker(s), the addressee, no matter who, upon hearing the utterance, always shoots the woman; and the addressee, no matter who, never shoots the woman without hearing the utterance uttered by some certain speaker(s).
(b"i)if no matter who utters it, there is/are some addressee(s)who, upon hearing the utterance, always shoots the woman.
(b""ii)if it is uttered by some certain speaker(s), there is/are some addressee(s)who, upon hearing the utterance, always shoots the woman.
C. it satisfies the unique correlation, (c"") if it is uttered by a particular speaker to a particular addressee under a particular circumstance, the addressee, upon hearing the utterance, always shoots the woman; and a particular addressee never shoots the woman without hearing the utterance uttered by a particular speaker under a particular circumstance.
(c")if it is uttered by a particular speaker to a particular addressee under a particular circumstance, the addressee, upon hearing the utterance, always shoots the woman.
. This is a complex condition. It is unnecessary because S""s saying something is unnecessary. In this whole affair, S""s uttering the noise plays no more than a triggering role.
. One may observe that verbal and verbal-influential causation (if we may use these two terms) involve ultimately physical causation. My argument is that the former cannot be reduced to the latter, for meaningful linguistic communication between human beings transcends physical causation.
. It poses no problem to any theory of causation that two Xs(X = event) or more jointly cause one Y. But no causal theory, as far as I know, can explain adequately how one X causes many different Ys.
. We uphold the traditional distinction between action and behaviour, between things people do and things which merely happen to them. Although it may be controversial to take intention as the hallmark of agency, as Davidson (1971)does, who is challenged by Cornman (1971), we regard H""s awareness or consciousness of and his exertion of some control over what he is doing as necessary for the claim of agency.
. Austin (1980:117) regards one agent""s act to another""s act as a sequel act or a response-act. He explains the relation between the two acts not in terms of causation, but in terms of convention. See Gu (1987:ch. 2) for detailed criticism.
. Austin(1980:102) also uses this formula to report perlocutionary acts, but he obviously favours single perlocutionary verbs.
. My conception of conversational rhetoric is different from, though related to, that formulated by Leech (1981, 1983). The key idea of conversational rhetoric of mine is "rhetorical transaction" whereas Leech is mainly concerned with conversational principles which he prefers to call rhetorical principles.