including communication, that is perceived by another sentient being, and inter- preted as reason for volitionally performing a particular action’’ (Talmy 2000: 438). Force Dynamics in the Psychophysical and Interpersonal Domains Force-dynamic concepts in the physical realm also transfer easily to the psycho- physical and interpersonal domains, as can be seen from the fact that the basic deontic uses of the English modals—core modals as well as honorary modals—can be defined in force-dynamic terms. 5 As such, can ‘‘in the context of not indicates that the subject has a tendency toward the action expressed by the following verb, that some factor opposes that tendency, and that the latter is stronger, blocking the event’’ (Talmy 2000: 441); likewise, ‘‘may not indicates an authority’s blockage to the expression of the subject’s tendency’’ (2000: 441); and ‘‘must and had better in the context of not suggest an active social pressure acting against the subject to maintain him in place’’ (2000: 441). The subject slots of these verbs are mostly filled by the Agonist, a sentient being involved in a psychosocial, rather than a purely physical, interaction: (20) John can/may/must/should/ought/would/need/dare/had better not leave the house. In (21) and (22), however, the same verbs allow nonsentient beings as subjects: (21) The cake can/may/must/ stay in the box. (22) The pear could/may/must/ be ripe by now. Still these examples do not contravene the idea that the verbs refer in their basic usage to psychosocial interaction. In (21), the Agonist does not fill the subject slot, but it is implicitly present, as an Agent controlling the actions of the Patient, which fills the subject slot (Talmy 2000: 442). In (22), the modals are used with an epi- stemic meaning; following Sweetser (1984, 1990), Talmy (2000: 443) analyzes these epistemic uses as resulting from a metaphorical transfer from the psychosocial domain to the domain of semantic inference, from ‘‘the interpersonal impinge- ments to the impingements of arguments on each other or on the reasoner, con- straining him towards certain conclusions’’ ( 2000: 443). Since verbs such as make, let, have, and help take a to-less infinitive just like the standard modal verbs (as in examples 2–5) and since these verbs can be analyzed in force-dynamic terms, Talmy (2000: 443–44) groups them in the ‘‘greater modal system,’’ which, in English, makes up one grammatical expression of the semantic model of force dynamics. However, unlike the core and ‘‘honorary’’ (see note 5) modals whose subject position is occupied by the Agonist, make, let, have, and help select the Antagonist as their subject. They share this characteristic with open-class verbs such as forbid and require, whose meaning can also be defined using force- dynamic concepts: (23) I forbid you to leave the house. (24) I require you to stay in the house. 300 walter de mulder Force Dynamics in Discourse Argumentation in discourse can be interpreted in terms of forces opposing and reinforcing particular positions or points of view, an idea that is already implicitly present in Lakoff and Johnson’s (1980) description of the metaphor argument is war. A force-dynamic interpretation of this domain also permits an analysis of the meaning of ‘‘logic-gators’’ (Talmy 2000: 452), words and expressions such as yes but, besides, nevertheless, moreover, granted, instead, after all, and on the contrary, which can be used to direct the flow of argumentation by expressing opposition or reinforcement with respect to points argued for or against (see also Oakley 2005). Force dynamics also operates in other discourse phenomena, such as ‘‘dis- course expectation,’’ which Talmy ( 2000: 453) defines as ‘‘the moment-by-moment expectations of participants in a discourse as to the direction and content of suc- ceeding terms.’’ ‘‘Vector reversal,’’ for instance, refers to a situation where a dis- course participant discovers that his or her assumptions about the direction of the discourse are exactly the opposite of those held by the other participant. The fol- lowing dialogue on a campus e-mail system may be cited as an illustration: (25) A titles message: ‘‘For Chinese students only.’’ B protests that it is exclusionary. A responds that the intent was: ‘Others need not bother to look’. Whereas B interpreted A’s title as exclusionary, A signals that it was his or her as- sumption that others would not want to read the message and that he or she wanted to spare them the trouble. 3. Force Dynamics as a Cognitive System: Talmy and Jackendoff Compared Talmy’s work on force dynamics has been taken up, integrated, or developed fur- ther by various linguists of cognitive persuasion. In this section, I will look at Jackendoff’s (1990, 1996) integration of Talmy’s account of force dynamics into his system of conceptual structure and conceptual semantics formalization. Before turning to the comparison proper, I will first shortly sketch Jackendoff’s views on conceptual structure. Jackendoff (1990: 9) conceives of conceptual structure as composed of ‘‘a finite set of mental primitives and a finite set of principles of mental combination’’; the combinatorial structure of the theory should allow it to explain, among other things, the ‘‘creativity’’ of language (in a Chomskyan sense; see Jackendoff 1990: force dynamics 301 8–9). Each of the primitive units, called ‘‘conceptual constituents,’’ belongs to a small set of major conceptual categories such as Thing (or Object), Event, State, Action, Place, Path, Property, and Amount (Jackendoff 1990: 22, 43). The major syntactic constituents making up a sentence map onto these conceptual constit- uents: the sentence John ran toward the house, for instance, corresponds to an Event, John and the house to Things, and toward the house to a Path (Jackendoff 1990: 22). The combination of these constituents into larger structures produces predicate-argument structures, following elaboration rules such as (26): (26) [EVENT] [ Event GO ([THING], [PATH])] [ Event STAY ([THING], [PLACE])] (Jackendoff 1990: 43) This rule ‘‘says that a constituent of the category Event can be elaborated as either of the two Event-functions GO or STAY, each of which takes two arguments. The arguments of GO, which denotes motion along a path, are the Thing in motion and the Path it traverses. This structure is seen most transparently in a sentence like Bill went to New York. The arguments of STAY, which denotes stasis over a period of time, are the Thing standing still and its location, as seen in Bill stayed in the kitchen, for instance’’ (Jackendoff 1990: 44). In representations of this kind, the thematic roles introduced by Gruber (1965) can be redefined as ‘‘shorthand’’ for particular structural configurations. Theme, for instance, the thematic role which Gruber defined as the object in motion or being located, now corresponds to the first argument of the functions used in (26) and (27) (Jackendoff 1990: 46): 6 (27) [EVENT] [ State BE ([THING], [PLACE])] [ State ORIENT ([THING], [PATH])] [ State EXT ([THING], [PATH])] (Jackendoff 1990: 43) In these representations, the BE-function serves to specify the location of objects (The dog is in the park), the ORIENT-function to specify their orientation (The sign points toward New York), and the EXT-function to specify the spatial extension of linear objects along a path (The road goes from New York to San Francisco) (Jackendoff 1990: 44). If Theme is defined as the object in motion or being located, it cannot cor- respond to the semantic role filled by Fred in sentence (28), since Fred is not an object in motion or being located: (28) Sue hit Fred. Sue obviously fills the Agent role in the sentence, but Fred cannot constitute the Theme; rather, Fred is the Patient, the role designating the person affected. In (28), the Patient is at the same time the Goal, but this is not necessarily the case, as can be seen in (29): (29) Pete hit the ball into the field. 302 walter de mulder In view of the fact that being a Patient does not exclude taking up other roles, Jackendoff (1990: 126) concludes that conceptual roles fall into two tiers: a thematic tier, with Source-Theme-Goal relations, and an action tier, with Actor-Patient relations. Sentence (28), then, would be represented as (28’): (28') Sue hit Fred Theme Goal Actor Patient Actor and Patient are defined respectively as the first and the second argument of a new ‘‘affect’’ (AFF) relation. These elements allow Jackendoff to propose the following representation of sentence (30): (30) Harry prevented Sam from going away. CAUSE ([HARRY], [NOT GO ([SAM], [AWAY])]) AFF ([SAM], ) AFF ([HARRY], [SAM]) (Jackendoff 1990: 131) Talmy’s Agonist and Antagonist can now be identified with, respectively, the Pa- tient and the Actor on the action tier, the first and the second argument of the AFF- function; note that the notation [AFF ([SAM], )], which is part of the thematic tier, signals ‘‘Actor only’’ (Jackendoff 1990: 128). Jackendoff subsequently analyzes other force-dynamic structures by reformulating existing functions or adding new ones. This can be illustrated by his analysis of sentence (31): (31) Sam resisted Harry. CS u ([HARRY], [ Event ]) REACT – ([SAM], [HARRY]) (Jackendoff 1990: 137) The thematic tier signals that ‘‘Harry is exerting effort toward the realization of some implicit Event, with undetermined outcome’’ (Jackendoff 1990: 137); CS de- notes a function expressing the application of force, which can have three different values (cf. Talmy’s ‘‘resultant’’): CS u denotes undetermined outcome as expressed by try or pressure;CS þ denotes successful outcome and supplants the previous CAUSE; and CS – denotes unsuccessful outcome as expressed by fail or impede. The new function REACT on the action tier is a sort of mirror image of AFF, with the Agonist as the first argument and the Antagonist as the second. However, the REACT function entails more than a simple reversal of the linking between semantic roles and syn- tactic functions, since verbs of reaction assign a more active role to the Agonist than the one described by the passive Patient role (Jackendoff 1990: 138). Nevertheless, without pursuing the issue, Jackendoff suggests that AFF and REACT are to be seen as alternative realizations of a more abstract function. Jackendoff claims to have preserved Talmy’s force-dynamic generalizations, but to have adapted them to conceptual semantics through the introduction of the following system (see Jackendoff 1996: 120): force dynamics 303 a. Distinction between two opposed force entities: Antagonist (¼ Agent) and Agonist (¼ Patient) 7 b. Patient action desired by Antagonist c. Success of Antagonist However, as already pointed out above, Jackendoff’s conceptual semantics is a combinatorial system, where semantic representations are built up out of atomic ‘‘building blocks’’ (Taylor 1996). In this respect, his theory is different from Tal- my’s, where all force-dynamic elements are defined with respect to a basic scene, such that the presence of one part entails that of the others (Talmy 1996b: 267, cited by Lampert and Lampert 2000: 228; see also section 2) and where all elements are always present, although some may be backgrounded. This is relevant, for instance, with respect to the use of the prepositions from and to in examples (30) and (32): (32) Harry forced Sam to go away. CAUSE ([HARRY], GO ([SAM], [AWAY])) AFF ([SAM],) AFF ([HARRY], [SAM]) (Jackendoff 1990: 131) As pointed out by Deane (1996: 60), the use of the prepositions in these examples is motivated, as can be seen from a comparison of from and to in (30) and (32) with the prepositions in italics in examples (33)–(35): 8 (33) a. In his foolishness, Sam has turned to robbery. b. Sam has turned (away) from the crimes of his youth. (34) a. Sam is inclined toward going away. b. Sam is inclined against going away. (35) a. Harry talked Sam into going away. b. Harry talked Sam out of going away. In these examples, to, toward, and into denote that Sam’s tendency (toward action) is directed toward a particular target (expressed in the subsequent noun phrase); (away) from, against, and out of (and the subsequent noun phrases), on the con- trary, signal a tendency of Sam’s that he is no longer inclined to. Briefly, the pre- positions get their meaning by referring to the direction of the Agonist’s tendencies as represented within Talmy’s force-dynamic scene. Jackendoff’s representations, however, do not contain any such reference to these tendencies and thus seem unable to capture what is shared by both sets of prepositions. In fact, the only difference between (30) and (32) is the negative marker, which corresponds not to the difference between the individual items to and from, but to that between force to and prevent from. In other words, in Jackendoff’s approach, to is represented as a meaningless infinitive marker, without any link to the preposition to; in Talmy’s approach, on the contrary, as in other cognitive approaches, the use of to as an infinitive marker is related to its uses as a preposition. 304 walter de mulder Jackendoff can, of course, enrich his representation—whose sole aim in Se- mantic Structures was to express the correspondence between conceptual structure and syntactic argument structure. He could, for instance, bring the prepositions closer to their spatial meanings (Jackendoff 1996: 123). The question remains, how- ever, whether he could also explain the motivated changes of interpretation these prepositions go through in context. 4. Force-Dynamic Analyses of Modals Talmy’s research on force dynamics, and his views on modals in particular, has brought about various studies developing or taking issue with Talmy’s account. In sections 4 .1 and 4.2, I will consider Sweetser’s and Johnson’s analyses of modal meaning and compare their account with Talmy’s on four key issues. In section 4.3, I will pay attention to Langacker’s and Achard’s conception of modals as a more grammaticalized category. 4.1. Sweetser’s (1990) and Johnson’s (1987) Analyses of Modal Meanings An important result that has come out of Sweetser’s (1990: 28, 50) research is that there exists a general tendency among language users to speak of our internal world by employing language that normally refers to the external world. Reasoning pro- cesses, for instance, are modeled after real-world actions, as subject to compulsions, obligations, etc. (Sweetser 1990: 49–50). This metaphorical transfer explains the systematic polysemy of modal verbs in many unrelated languages, 9 whereby their ‘‘epistemic’’ meanings denoting necessity, probability, and possibility have been derived from their ‘‘root’’ meanings, denoting obligation, permission, or ability. 10 For Sweetser, modality is characterized as ‘‘basically referring to intentional, di- rected forces and barriers’’ (52), and its experientially basic level of operation is the sociophysical world; this is in contradistinction to Talmy, who holds that the physical level of force dynamics is the experientially basic one and that the socio- physical is already structured by a metaphorical projection of the basic folk model of physical force (Lampert and Lampert 2000: 221, 248, 278). Nevertheless, since Talmy also believes that the modals in their basic usage refer to psychosocial rather than to physical interaction (Talmy 2000: 441; see also Lampert and Lampert 2000: 248), Sweetser can adopt Talmy’s definitions for some of the modals. May, for force dynamics 305 instance, is said to express ‘‘a potential but absent barrier’’ (Sweetser 1990: 52). For other modals, Sweetser prefers to change Talmy’s definitions: instead of Talmy’s analysis of must ‘‘as a barrier restricting one’s domain of action to a single act’’ (Sweetser 1990: 52), she describes must as expressing a positive compulsion to do something. As pointed out by Pelyva ´ s(1996: 124), the contrast between these two definitions could be more apparent than real, as the notion of ‘compulsion’ is relevant when one adopts the doer’s point of view, and that of ‘restricting the subject’ when one adopts the speaker’s point of view. According to Sweetser (1990: 61), only the ‘‘image-schematic’’ properties of semantic structure are preserved in the metaphorical transfer that maps the struc- ture of the ‘‘root’’ (source) domain onto the ‘‘epistemic’’ (target) domain (see also Lampert and Lampert 2000: 252). Image schemas have been defined by Johnson (1987: 29) as recurrent patterns, shapes, or regularities which ‘‘emerge as mean- ingful structures for us chiefly at the level of our bodily movements through space, our manipulations of objects, and our perceptual interactions’’ (see also Lakoff 1987; Gibbs and Colston 1995; Oakley, this volume, chapter 9). These patterns have Gestalt-like qualities: they are seen as unified wholes whose parts only get meaning from the unity and coherence of the whole (Johnson 1987: 41, 44). Johnson (1987) proposes to view the notion of force in the root senses of the modal verbs as image-schematic force Gestalts. Thus, must is analyzed in terms of the image schema of compulsion, which develops from our experience of being moved by external forces. It can be summarized by the schematic representation in figure 12.2, where the dark arrow represents an actual force vector and the broken arrow a potential force vector (Johnson 1987: 45, 51): May, on the other hand, is defined by the absence or removal of restraint schema, which originates in our experiences of the removal of barriers and of the absence of possible restraints (e.g., door openings) and is represented by figure 12.3 (Johnson 1987: 46–47, 52). Following Sweetser, Johnson (1987: 53–55) argues that the epistemic senses of the modals are derived by interpreting the notions of force and barrier meta- phorically, the forces being the premises of an argument that ‘‘force us along a path toward some conclusion’’ (54). As is suggested by this formulation, argument and reasoning are also, at least partly, metaphorically structured in terms of a source- path-goal scheme: we follow a path to reach a destination, in this case, a con- clusion, and various propositions can act as blockage, such that we do not reach that conclusion (54). Figure 12.2. The compulsion schema 306 walter de mulder 4.2. A Selective Comparison Space limitations do not allow an exhaustive comparison of the different proposals presented above (but see Mortelmans, this volume, chapter 33); we will limit ourselves to the following four key issues: (i) the basic level of force dynamics; (ii) the exact nature of the source and target domains of the metaphorical transfer; (iii) the metaphorical nature of the relation between the deontic and the episte- mic meanings; and (iv) the image-schematic nature of force dynamics. The Basic Nature of Force Dynamics Sweetser holds that modality, conceived of in terms of intentional, directed forces and barriers, basically operates at the level of the sociophysical domain. This view is also subscribed to by Pelyva ´ s(1996: 125, 144–46), who, like Sweetser (1990: 152, note 5), believes that we conceive the physical world in terms of basic aspects of human experience, such as actions and intentions. Despite some minor differences (see Pelyva ´ s 1996: 135), both authors thus disagree with Talmy, who recognizes that the modals refer in their basic deontic usage to ‘‘psychosocial’’ interactions, and not to physical ones (Talmy 2000: 441), but sees force dynamics nevertheless as ultimately related to our kinesthetic system (2000: 467). In other words, Talmy presents the fundamental elements of force dynamics with respect to a purely physical scene (2000: 413), adding that he regards nonagentive forms of force dynamics more basic than forms containing an agent (2000: 421). 11 The Exact Nature of the Source and Target Domains of the Metaphorical Transfer According to Sweetser (1990: 50), the English modals developed their root mean- ings from nonmodal meanings (e.g., OE magan ‘be strong, be able’) before they acquired their epistemic meanings. Pelyva ´ s(1996: 133–34), however, concludes from Figure 12.3. The removal of restraint schema force dynamics 307 diachronic evidence (cf. Traugott 1989: 36, who relies on data by Bybee and Pagliuca 1985 and Bybee 1988) that both the root and the epistemic meaning of may are derived from a (now extinct) ability meaning and that the epistemic meaning is attested before the deontic one. This course of development could explain why the doer’s intentions need to be taken into account to describe the deontic, but not the epistemic meaning (Pelyva ´ s 1996: 134). It implies, moreover, that Sweetser’s root domain conflates two meanings that would better be distinguished: sociophysical meanings implying intentionality and meanings such as ‘ability’, in which inten- tionality only plays a peripheral role (Pelyva ´ s 1996: 125–26). 12 The Metaphorical Nature of the Relation between the Deontic and the Epistemic Meanings As pointed out by Pelyva ´ s(1996: 154), the development of the epistemic meaning need not be conceived of as a metaphorical transfer. Traugott (1989) analyzes it as the conventionalization of conversational implicatures through pragmatic stren- gthening: for instance, ‘‘if one says You must go in the meaning ‘You ought to go’, one can implicate that one believes/concludes that it is true that you have to go’’ (51). The metonymic nature of the relation between deontic and epistemic meaning of must may also be suggested by the following observation of Lampert and Lampert (2000: 252): whereas the deontic meaning focuses on the (entire) path which the subject must follow as it is impinged upon by the Antagonist (note the infinitive come home in example 36), the epistemic meaning in (37) focuses rather on the path’s terminal point, namely the conclusion: (36) You must come home by ten (Mom said so). (Sweetser 1990: 61) (37) You must have been home last night. (Sweetser 1990: 61) There is, then, a difference with respect to the windowing of attention in the two meanings, which each focus on adjacent elements within the same frame (Talmy 1996b). The evolution from deontic to epistemic can also be analyzed in terms of image-schematic transformations (see Gibbs and Colston 1995: 361, cited in Lamp- ert and Lampert 2000: 252–53). 13 The Image-Schematic Nature of Force Dynamics Although both Johnson and Talmy underline the schematic nature of the force- dynamic structures, their conceptions can be traced back to different origins and, consequently, are different in nature: Johnson views force dynamics in terms of image schemes that emerge out of our concrete embodied experiences, at a nonlinguistic—or prelinguistic—level; Talmy, in contrast, conceives of force dy- namics as an abstract schema, common to different domains, most notably the linguistic and the kinesthetic ones (Lampert and Lampert 2000: 219–21). 14 308 walter de mulder 4.3. A More Grammatical Conception of the Force-Dynamic Nature of Modals Langacker (1991: 269–81) defines modals as ‘‘grounding predicates,’’ since they specify a relationship between the process profiled by the complement clause and some element of the ground, that is, the speech event, its participants, and its im- mediate circumstances. As is characteristic of grounding predicates, the grounding relationship expressed by the modals remains offstage and unprofiled, and the ground is construed with a high degree of subjectivity (Langacker 2002: 7, 13, 17). Consequently, Langacker (1991) describes the development of the modal verbs, from main verbs to more grammaticalized markers, in terms of subjectification. Initially, these main verbs expressed a physical capacity to do something, with their subject denoting a ‘‘locus of some kind of potency directed at the landmark process, i.e., a physical or mental force that, when unleashed, tends to bring about the occurrence of that process’’ (270). When the modals are used with their root meaning, the locus of potency (corresponding here with Talmy’s Antagonist) can no longer be iden- tified with the subject; it must be associated with the speaker or some other element associated with the ground (even some ‘‘nebulous, generalized authority’’ [Lan- gacker 1999: 308]), as can be seen in (38): (38) a. You may leave the table now! b. This noise must cease immediately! c. He absolutely will not agree with it. Epistemic modals constitute the end point of the development. Unlike with de- ontic modals, the locus of potency is no longer identified with a specific individual or another element associated with the ground; the ‘‘impetus toward realization of the designated process is not provided by any specific force, but rather by the gen- eralized force consisting in the fact that the world has a certain structure and reality is unfolding in a particular way’’ (Langacker 1991: 273). In Langacker’s view, this generalized force can be defined using elements of a cognitive model of the way we think about the world, the ‘‘dynamic evolutionary model.’’ In this model, the world is seen as a structured whole in which situations unfold, following an evolution that is due to an unknown force. Because of the structured nature of the world, its evolution can to a certain extent be foreseen (for more details, see Mortelmans, this volume, chapter 33); consequently, there are (at least) two types of future paths the evolution can follow: paths which can be projected from the present with reasonable confidence, called ‘‘projected reality,’’ and paths which reality is only not precluded from following, called ‘‘potential reality’’ (Langacker 1991: 277–78). These ideas allow Langacker to define the meanings of the English epistemic modals: may situates the designated process in potential reality, whereas will situates it in projected reality. Langacker’s conception differs from Sweetser’s in that he presents the modals, more explicitly than Sweetser, as a grammaticalized category. Moreover, in his force dynamics 309 . fills the Agent role in the sentence, but Fred cannot constitute the Theme; rather, Fred is the Patient, the role designating the person affected. In (28), the Patient is at the same time the Goal,. to the following four key issues: (i) the basic level of force dynamics; (ii) the exact nature of the source and target domains of the metaphorical transfer; (iii) the metaphorical nature of the. 1987: 45, 51): May, on the other hand, is defined by the absence or removal of restraint schema, which originates in our experiences of the removal of barriers and of the absence of possible restraints