Abstract
The immediate background for what I discuss is “factproposition-event (FPE) theory”. The three-way distinction between facts, propositions, and events (as entities and as concepts) has important support from the semantics of natural languages. First, there is the test for “factive” predicates to the effect that the truth of the complement of a factive predicate (the factive clause) remains implied or presupposed through negation of the predicate. Asserting either “John realized Mary left” or “John didn’t realize Mary left” presupposes or implicates the truth of “Mary left”. Predicates that are candidates for this test, but which fail it are deemed “propositionals” (in FPE theory) and their complements “propositional clauses”. “Believed” is a propositional, and “that Mary left” a propositional clause, in “John believed that Mary left”. “Eventive” predicates are those which seem, at first, to be additional factives. But eventive predicates resist associated subjects or objects which are full clauses.
This chapter is drawn from Peterson 1985a, a presentation to the Parasession on Causatives and Agentivity of the 21st Regional Meeting of the Chicago Linguistic Society, April 27, 1985.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
Thus, even Thomson’s “super-event” (a fusion of all events over all times)—cf. Thomson 1977, p. 85—is a candidate for being a cause and/or an effect (though it fails to be either). Mentioning Thomson reveals the central character of the wider context for this discussion—viz., philosophical action theory (which I commend to linguistic semanticists) such as in Davidson 1980, Goldman 1970, Thalberg 1977, Thomson 1977, Brand 1984, Brand & Walton 1976, and White 1968. Also, I am greatly influenced by Vendler 1967, 1972, 1975, Bennett 1973, Grimm 1977, and Kim 1973.
So, many investigators—from McCawley 1972 (p. 141) to Thomson 1977 (Ch. X)—who reduce expressions of agent causation to non-agentive cause-relations must be wrong.
Herein, I shall not try to refute those who think that so-called lexical causatives like “kill” only imply a related agent causation proposition (containing causing-to-die) (Thomson) or that such pairs only semantically overlap (Shibatani). I believe that examples like those in (5) and (6) are exactly synonymous (on certain readings of their components). In one way, of course, this must be true—viz., when “cause a” is stipulated to express a sense which makes the pairs synonymous. However, the structure and evidence for the theory of agent/non-agent causation meanings I offer here (offer within the wider FPE theory) as a whole tends to confirm the hypothesis (not just a stipulation) that the sense of “causeA” is not only a real one (unreducible to “causeN” and/or “causeE”)- (Incidentally, often herein a more convincing case for a causative sentence being synonymous with (2)-sort of paraphrase can be made by using “make” instead of “cause”; i.e., in many dialects “make” is the verb which expresses CAUSEa for certain examples—rather than “cause”.) One sort of objection to “cause” paraphrases of causatives is that one can cause something (say, Mary coming to lunch) without the relevant causative applying (e.g., without bringing her to lunch). For example, one can get someone else to bring Mary to lunch and, thereby, cause her to come but not bring her yourself (or cause someone else to feed the animals while not feeding them oneself, etc.) I myself think that when you cause someone else to cause something you might have (but didn’t) cause yourself, you indirectly cause it (i.e., often conditions are such that you do). If I get Mary to feed the animals, I indirectly feed them (i.e., “feed” can directly express “cause someone to feed”). This is not always acknowledged, but in the crucial cases—like killings—it certainly is. If you get Smith to kill Jones by threatening Smith’s children, you do indirectly kill Jones. You didn’t do it yourself, as we say (do it directly). (The direct/indirect distinction is clearly distinct from the basic/non-basic distinction—both applying to actions.) The more your use of someone else to get something done is like the use of inanimate tools, the more the indirect causing is like direct causing.
Throughout below, I ignore the difference between “be” and “become” in sentences expressing agent causation. Many investigators in this area always use sentences and clauses with “become” where I use “be”. Their reasoning I assume is that what is caused (either cause or causeN) is an event and that no states are events. Thus, when it appears x causes y to be something (asleep, happy, dead, etc.) that must be because x caused y to become so (asleep, happy, etc.)—where becoming so is a nonstative event. This reasoning is not entirely convincing. For one thing, the relation expressed by agentive uses of “cause” is transitive. Thus, if x causesa y to become happy and y is caused to be happy by becoming so, then x causesA y to be (not just become) happy. (Becoming happy, then, is a causedA-event which is a means to being happy in a director way than, e.g., shooting y fatally is a means to killing y.) There is at least one interesting reason, however, for distinguishing x causing y to be happy from x causing y to become happy. This concerns the complex event which is the event of (say) x causing y to be happy. Is y’s being happy a proper part of that event (x’s causing y to be happy)? If so, this causing might last as long as y remains happy (maybe a long time). A good candidate for what might ordinarily be referred to by “x’s causing y to be happy” might well be x causing y to become happy, since the latter causing is a briefer event (for one thing). (Thomson refers to the causings-of-becomes as causings of “onsets” of states; cf. her Chapter VIII. I myself think that the limiting of events to non-states is mainly based on the intuition, if it is that, that events are “changes”, or contain them.)
I also ignore herein the very interesting question of causals and causatives manimate subjects; e.g., “The taxi brought Mary (home)”, “The oven baked the cake”, “The oven’s heat(ing up) dried our clothes”, etc. I believe that semantically they are all derivative cases (a fact which is not obvious in English examples). Here are some hints about how I would try to demonstrate this, if time and space permitted. Many investigators (especially philosophers, but not only them) assume that in every case wherein a subject is said to causeA something Ej (event, state, etc.), there must be involved a causing event Ei which causesN Ej. They evidently think that it is only via presumption or implication of Ei causing Ej that use of “cause” (superficially “causeA”) is justified. On my view, such an approach is correct only for inanimate subjects of “causeA” predicates (and paraphrases of them with causative predicates). The main cases of “causeA” predicates (all those classified in (22) below) are those with animate subjects, in my view. And the main reason I adopt this claim is that only animate-subjected agent-causation sentences can describe or report basic actions (where basic actions are just those causingsA for which the reduction of “x causesa Ej” to “Ei causesN Ej” fails absolutely). Actions—non-basic as well as basic (but basically the basic ones)—are the paradigm cases for what agent-causation sentences describe. Of course, some events which are non-actions can be reported in a form that makes them appear to be actions (e.g., the oven baking our cake). But they aren’t in the end real actions (even if reported so by not-just-metaphorical expressions). The reason is that none of these could be basic actions (i.e., they aren’t even candidates). The oven bakes my cake in that the oven is used (turned on, thermostat adjusted, etc.) by me or someone to bake the cake. The oven as a semantic agent does not do anything because it is no agent at all. I conjecture that one reason why so many contemporary Anglo-American investigators do not see this strict difference between action and non-action events (reflected in the difference between agent-causation verbs with and without animate subjects) is their general preference for philosophical materialism (e.g., recent “identity (brain/mind) theory”). That is, they believe that there will be an explanation of how persons (and other animate objects) are really just inanimate devices or systems of them (like ovens, thermostats, or even computerized guidance and control systems) in the end (in which case so-called basic actions will be explained away in terms of constitutive physical causes). I suggest that if they succeed at this reductive (mental-to-physical) analysis it will only be at very great cost in terms of the strain it puts on conceptual structures and categories utilized by natural language semantics (as internalized in ordinary speakers).
What is the agent in Anne’s feeding of Emma an agent of? Emma’s eating, of course. Causing Emma to eat is being the agent of her eating (cf. the section “Agency” below). But what is the agent an agent of in throwing a ball? If we had all of the necessary and sufficient conditions for an action to be a throwing, then the agented-of event might be obvious—concerning holding and releasing what’s thrown, moving one’s arm, using the right speed, etc., etc. (Maybe too much to ask for a dictionary, or internalized lexical entry?)
Ordinary dictionaries do not give such meanings; cf. the entries in your dictionary for “cousin” and “tree” and notice how the former do give an explanation of meaning of the word whereas the latter tells you something about what “trees” are. Of course, maybe your dictionary like mine tends to define all nouns by adopting the style of saying what the nouns refer to—e.g., “a tree is a plant…”, “a cousin is a relative…” However, the way in which “cousin” is defined in referential style more easily translates into an explanation of word meaning (as vs. reference or denotation) than does the definition of “tree”.
This does not, by the way, undermine the view that meanings in a narrow sense are universals, nor confuse meaning with reference; cf. Peterson 1983 (Appendix II) and Peterson 1986a.
Consequently, their dictionary entries may be less detailed semantically than might have been anticipated. Much of what one finds in an ordinary dictionary for “throw”, like what one finds for “tree”, ought to he thought of as not representing semantic components but rather representing factual knowledge about what the word denotes. This does not mean the idealized dictionary in a semantic component of a linguistic description of a natural language ought to contain factual information about what a word refers to. It probably shouldn’t. And that means idealized dictionary entries for natural kind and natural action words will probably contain much less than might have been thought—e.g., two such terms may have the same semantic-content components in their lexical entries, even though they refer to different natural kinds. My basic idea for natural kind meanings is derived from Kripke 1979 and Putnam 1975. One thing Putnam has, which we might expand on, is an image sort of component of a natural kind term’s meaning-and-reference—viz., his “stereotype”. The stereotype is not a new kind of “sense” component in the meaning of a term (since, for one thing, it is not a necessary component and, for another, it is an image sort of thing rather than a concept or sense). One might think of the visual stereotypes for “tree” and “elephant” as sorts of degraded or fuzzy images of what trees and elephants look like. Are there the same sorts of stereotypes for natural action verbs? R. Jackendoff (pers. com.) suggests that there are—viz., something like D. Marr’s 2-l/2-D-sketch/3-D-models of movements of shapes (cf. Marr and Vaina 1980). (Of course, only a subset of such movement images would be stereotypes for natural actions. Others would be for non-agentive natural motions, activities, processes, etc.) (On the other hand, for a quite different approach to these verbs, via event “fusions” ignoring the natural kinds analogy, cf. Thomson 1977, pp. 220–222.)
Alternatively, obtaining one and the same reading for two sentences one with “feeds” and the other with “causes…to eat” can result from some lexical rules that relate the concepts (if there are such) of FEEDS and CAUSES TO EAT, where FEEDS is the central semantic-content component of the lexical entry for “feeds” and CAUSES TO EAT is the semantic-content representation that results from combining lexical entries for “causes” and “eat” appropriately. The lexical rule that would apply here would be an appropriate formulation of the generalization that for all x and y, x feeds y if & only if x causes A y to eat, where this generalization is taken to express a meaning postulate—true by virtue of the predicates in it.
Some investigators (Talmy? Jackendoff?) might call what I call agency “authorship”, since agency according to them necessarily involves intentionality. (They could, then, use Davidson 1980 for philosophical support.) But this appears to be only a terminological variation, since they admit that there is something like what I call mere agency—viz., authorship. Care must be taken, however, to distinguish their authorship (=mere agency) from other very similar relations (such as patiency, see below). (Thomson, it seems to me, gives some of the right kind of arguments for separating intention from agency; cf. Thomson 1977, Chapter XIX.)
One question I have not answered yet concerns other-than-agent subjects of agentive causation verbs (especially all the noncausals). My answer is implied in my treatment of causals(-) —viz., agentive causation verbs without animate subjects are derivative (somehow), perhaps not genuinely metaphorical but still dependent. There are two sorts: event-phrase subjects and (non-event-phrase) inanimate subjects. Event-phrase subjects only appear to occur with causals(-), for when an event-phrase occurs as a subject of “cause”, the verb can’t (by definition) be “causeA”(but is “causeN”). Yet, event-phrases sometimes occur as subjects of other agentive causative verbs—e.g., “Sirhan’s shooting of RFK killed him”, “Jimmy’s hammering the bowl flattened it”, “Nancy pushing Ronald moved him off the stage”, etc. On my approach, the use of some causatives as verbs that are eventive subject-wise is based upon their more fundamental use as agentive subject-wise (with the non-basic use somehow dependent on tacit replacement of CAUSEa in the interpretation of them with CAUSEn, e.g., CAUSEa (.. MOVE) replaced by CAUSEn (…MOVE) in the use of “move” when it is eventive subject-wise). (This order of dependency (non-agentive depending on agentive) reverses Thomson’s (1977). She takes the fact that Sirhan’s shooting of RFK killed him to explain, in part, the fact that Sirhan killed RFK by shooting him—a reversal of the correct direction, in my opinion.) Similarly, non-event-phrase inanimate subjects—e.g., “The bullet killed RFK”, “The paint black(en)ed the floor” (but why can’t you say “Nancy aquaed the walls”?)—are derivative. Here philosophical predilections become especially prominent. One reason for taking these cases to be derivative is that they all imply means events, since there are no basic actions (or basic “action-correlates”) that are events with inanimate “agents”. Briefly, the bullet kills because an event containing it causesN death. Similarly, the hammer flattens the metal because some event containing the hammer causesN the metal to be flat. But, many will ask (not only philosophers), isn’t this true for animate subjects? Isn’t there always an implied causing event (so that eventive “cause” is contained in the explanation of every use of an agentive causation sentence)? (A “yes” answer is implied by McCawley 1972, p. 141, for example.) My answer is “no”—for reports of basic acts don’t (can’t) have such causing events. Thinking that there are always implied causingN events (and that basic acts are suspicious) is the result, I believe, of widespread philosophical materialism of the type discussed above in note 6.
“Ronald frightened Nancy by predicting defeat” would be semantically represented, on my approach, “Ronald CAUSEDa (Nancy is frightened) BY (Ronald predicting defeat)” (roughly speaking). But there is another way of looking at it; i.e., regarding it as appears in the paraphrase “Nancy was frightened by Ronald’s prediction of defeat”. For then it can be proposed that the semantic representation (AFFECT-wise) is: “(Ronald’s prediction of defeat) AFFECTED Nancy THROUGH (Nancy becomes frightened)” —where (perhaps), by analogy to means-ends actions, we have means-ends passions (sufferings), since an event(=Ronald’s prediction) AFFECTS Nancy THROUGH (the “means” event of) her-becoming-frightened. And we might even propose that Nancy’s being frightened (the means event) is a candidate for being a “basic passion” involved in “generating” (in Goldman’s 1970 sense, hereafter “G-generate”) a nonbasic passion—viz., Ronald’s frightening her with his prediction. (The idea is that the embedded means event which is a basic passion G-generates a non-basic passion just like the basic action of say, my kicking the dog might G-generate the non-basic action of my killing the dog by kicking it.) Similarly, perhaps, “I saw Mary depart” might be regarded (conceptually) as “Mary’s departure AFFECTS me THROUGH (visual experience)”, where “(visual experience)” represents (here, very schematically) some basic passion (a basic experiencing, then, for which there is no more basic means event) which is a means to — and G-generates—the nonbasic passion of my seeing Mary’s actual (over-and-beyond-mere appearances-of-it-to-me) departure. Well, maybe.
One question for Mourelatos, then, is are there instances of all the four types of “situations” (=my “events”)— states, processes, developments, and punctual-events—wherein patients are in the events (not “of” the event, mirroring the “in/of” distinction for agents vis à vis events used above). The choice seems to boil down to +[Patient—]-process and +[Patient—]-development for my seeing Mary depart.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Peterson, P.L. (1997). Causation, Agency, and Natural Actions. In: Fact Proposition Event. Studies in Linguistics and Philosophy, vol 66. Springer, Dordrecht. https://doi.org/10.1007/978-94-015-8959-8_12
Download citation
DOI: https://doi.org/10.1007/978-94-015-8959-8_12
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4856-1
Online ISBN: 978-94-015-8959-8
eBook Packages: Springer Book Archive