2003), modal adjectives and adverbs (e.g., Nuyts 1994, 2001, 2002), mental state predicates (e.g., Nuyts 1994, 2001, 2002; Pelyva ´ s 2001), evidential markers (e.g., Floyd 1999; Matlock 1989; Casad 1992; Lee 1993), lexical verbs acquiring episte- mic meanings (e.g., Verhagen 1995, 1996; Cornillie 2005a)—which venture into languages other than English do exist, but often lack a common core: they are like scattered pieces of a highly complex puzzle. The main focus in this chapter on modality within Cognitive Linguistics will therefore also lie on modals; at the same time, however, I will try to capture some of the basic insights that have arisen from research on other modal expression types in languages other than English and try to sketch possible future lines of research. Let us take up the initial question again: what is modality? A traditional exten- sional characterization is provided in Bybee, Perkins, and Pagliuca (1994: 176), in which a distinction is made between ‘‘grams with uses that are traditionally associated with modality—for instance, those indicating obligation, probability, and possibil- ity—and those traditionally associated with mood—imperative, optative, condi- tional, and subordinate verb forms.’’ 1 More schematically, modality can be taken to signal ‘‘the speaker’s attitude toward the proposition’’ (Givo ´ n 1994: 266), whereby ‘‘attitude’’ subsumes both epistemic (relating to issues of truth, belief, certainty, evidence, and the like) and valuative (dealing with desirability, preference, intent, ability, obligation, and manipulation) attitudes. Givo ´ n’s distinction between valua- tive and epistemic attitudes is reflected in the well-known polysemy of the modal verbs, which carry (at least) two kinds of related meanings: a basic root meaning and an epistemic meaning—both of which are taken, within Cognitive Linguistics, to involve some element of force (see section 2.2). The link between root and epistemic modality is usually regarded as metaphorical, whereby the real-world, sociophysical force associated with root modality is mapped onto the epistemic domain of rea- soning; this issue is, however, not uncontroversial, as section 2.3 will show. From a diachronic perspective, the evolution of the English modals has been described in terms of progressive subjectification (Langacker 1990, 1991a, 1999, 2003), whereby they are claimed to have acquired the status of (highly grammaticalized) ‘‘grounding predications,’’ which, together with tense and person inflections, relate the complement to the speech situation (the ground). This process of subjectificat- ion can also be witnessed in other modal expressions (as in the French, Spanish, and German modal verbs or in the lexical verbs threaten and promise and their German, Spanish, and Dutch equivalents), without them acquiring the status of grounding predications, however. In languages such as French and German, which still have a highly functional mood paradigm, it is the highly grammaticalized moods which are normally attributed the function of grounding a finite clause (see section 3). In their epistemic use, the English modals are mainly concerned with the degree of likelihood or the degree of personal commitment of the speaker toward the truth of the proposition. Evidential qualifications, 2 pertaining to the (type of ) informa- tion source through which the speaker has accessed the proposition, do not seem to play an important role in the (grammaticalized) system of English modality— although the inferential meaning carried by verbs like must and should has been 870 tanja mortelmans considered as evidential rather than genuinely epistemic by a number of scholars (van der Auwera and Plungian 1998; Nuyts 2001: 173). 3 More generally, there does not seem to be a consensus regarding the relationship between epistemic modality (in terms of speaker commitment) and evidentiality (information source). Are they to be seen as separate qualificational categories (van der Auwera and Plungian 1998; Nuyts 2001)? Or, do they both pertain to a broadly defined domain of epistemic modality which would also include the grammatical coding of speaker attitudes like surprise (Floyd 1999) or of the evaluation of a state of affairs in terms of its ac- cordance with background expectations (Lee 1993)? The fact that there appears to be a strong cross-linguistic connection between speaker commitment and infor- mation source to the extent that direct evidence generally evokes strong commit- ment, whereas indirect evidence (reported or inferred) pairs with weaker degrees of commitment (Givo ´ n 1982; see also Sanders and Spooren 1996), would support the latter position. In view of the relatively small amount of cognitive linguistic studies of evidential categories (see Floyd 1999 for a notable exception), 4 we will not elab- orate this issue any further. One of the main merits of a cognitive linguistic analysis of modality is its focus on semantics, which has resulted in a considerable number of fine-grained semantic (network) analyses of modal markers, both from a diachronic and a synchronic point of view. Moreover, Talmy’s force dynamics (see, e.g., Talmy 1985) has pro- vided a schematic conceptual background, against which a number of different, but related, models of modal meaning (see, e.g., Johnson 1987; Sweetser 1990; and Langacker 1990, 1991a) have been developed. 2. The Modal Verbs in Cognitive Linguistics 2.1. Polysemy versus Monosemy A recurrent theme in the study of modal verbs is their semantic ambiguity; that is, modals display a wide array of senses, among which two stand out: the dia- chronically more basic root meaning, 5 on the one hand, and the epistemic meaning, on the other. 6 A modal is regarded as epistemic, when its sole import is to indicate the likelihood of the designated process. In a root modal, there is additionally some concep- tion of potency directed toward the realization of that process, i.e. some notion of obligation, permission, desire, ability, etc. (Langacker 1991a: 272) It is generally agreed upon to treat the semantic ambiguity in modals as a case of polysemy; that is, the modals are taken to code a variety of interrelated meanings. modality in cognitive linguistics 871 However, there is less agreement when it comes to the exact number and the theoretical status of these various senses. Goossens (1992), for instance, provides a semantic analysis of the multiple meanings of English can in terms of a radial cat- egory organized around a number of prototypical cores (see figure 33.1). These cores correspond to a number of salient uses, established on the basis of frequency, centrality in the network of uses, and onomasiological contrasts with other modals (Goossens 1992: 377). These prototypical usages have acquired some degree of entrenchment, they are—to some extent—conventionalized. Sweetser (1990), on the other hand, allows for a greater indeterminacy by pro- posing only three different meanings or senses, which correspond with conven- tionalized metaphorical mappings of the modals to three domains of human ex- perience: (i) to the sociophysical world (in their root sense; see 1a); (ii) to the world of reasoning (in their epistemic sense; see 1b); or (iii) to the conversational world (so-called speech act modality; see 1c). (1) a. John may go. b. John may be there. c. He may be a university professor, but he sure is dumb. Whereas this threefold ambiguity belongs to the conventionalized semantics of the modals, any other specification (e.g., the identification of the imposer and imposee of the modality) is a matter of pragmatics and should therefore not be included in a modal’s semantics (Sweetser 1990: 65–68). The polysemy of modals is even more downplayed by Langacker (1990, 1991a). For the present-day English modals, he presents a schematic semantics that focuses more on the strength and existence of the directed potency than on its exact nature. The potency associated with must, for instance, is vaguely referred to as ‘‘a kind of necessity’’; as a grounding predication (see below), must is claimed to designate (or profile) the process deemed necessary in the complement (Langacker 1990: 27). Whereas Langacker still distinguishes between root and epistemic senses, though, Wierzbicka (1987: 38) advocates a purely monosemic approach, claiming that the different interpretations of can (and of every other modal, for that matter) ‘‘have more to do with the context, either explicit or implicit, than with the mean- ing of the modal as such.’’ Such an account, however, does not seem to be able to offer an explanation for the semantic restrictions of the modals. English can, for Figure 33.1. Present-day English can as a radial category (based on Goossens 1992: 389) 872 tanja mortelmans instance, does not allow an epistemic meaning, which cannot easily be accounted for based on Wierzbicka’s very general semantic description. 2.2. The Modals as a Force-Dynamic Category The notion of ‘‘force’’ has been highly influential in the way modality is concep- tualized in Cognitive Linguistics (Sweetser 1982, 1984, 1990; Talmy 1985, 1988, 2000; Johnson 1987; Langacker 1990, 1991a, 1999; Pelyva ´ s 1996, 2000; Achard 1998). Most prominent in this respect has been Talmy’s force-dynamic framework (see Talmy 1985, 1988, 2000), which was originally developed to provide a generalization over the traditional semantic category ‘‘causation’’ (Talmy 1976). According to Talmy, force dynamics involves four crucial parameters (see also De Mulder, this volume, chapter 12): a. A force opposition between an Agonist—the focal force entity—and an Antagonist, opposing the former; b. An intrinsic force tendency (either toward action or toward rest); c. The relative strengths of the opposed forces, whereby a stronger entity will be able to manifest its tendency at the expense of the weaker one; and d. The resultant of the force interaction—either action or rest, assessed only for the Agonist. The English modals, then, are viewed as constituting the grammatical category that corresponds to the semantic category of force dynamics; that is, the modals are the grammaticalized encodings of the various ways in which entities interact with respect to forces and barriers. In the case of cannot, 7 for instance, a typically sentient subject (the Agonist) is inclined toward the action expressed by the infinitive, but some opposing factor (force) blocks the realization of the event. With must, the Agonist is exposed to ‘‘an active social pressure’’ (Talmy 1988: 79) that tries to keep him or her in place. 8 Talmy does not restrict his force-dynamic analysis to the tra- ditional modal verbs, but also accords ‘‘honorary’’ modal status to the less gram- maticalized verbs have to, be supposed to, be to, and get to. Moreover, a number of lexical verbs (make, let, have, help) are integrated into the so-called ‘‘greater modal system’’ (Talmy 1988: 81): syntactically, these verbs are on a par with the core modals insofar as they take a to-less infinitive complement; semantically, they are shown to have force-dynamic reference, but they differ from the more grammaticalized modals in coding the Antagonist as subject (I made him push the car to the garage vs. He must push the car to the garage). 9 In their root usage, the modals are taken to refer mostly to psychosocial (rather than physical) interaction involving a sentient Agonist as subject (Talmy 1988: 79). 10 There are two notable exceptions to this basic pattern. In the case of Agonist demotion, the Agonist is backgrounded in favor of a promoted (typically non- sentient) Patient (as in The cake must stay in the box). Second, nonsentient subjects also occur in the epistemic usage, which is regarded as ‘‘the application of modality modality in cognitive linguistics 873 to the domain of our reasoning processes about various propositions’’ (Talmy 1988: 80). The use of modals in epistemic contexts has been elaborated by Sweetser (1984, 1990): in the epistemic realm, the sociophysical forces and barriers are said to be metaphorically mapped onto premises in the world of reasoning. 11 In the real world must is taken as indicating a real- world force imposed by the speaker (and/or some other agent) which compels the subject of the sentence (or someone else) to do the action (or bring about its doing) expressed in the sentence. Here [i.e. in the epistemic world] must is taken as indicating an epi- stemic force applied by some body of premises (the only thing that can apply epistemic force), which compels the speaker (or people in general) to reach the conclusion embodied in the sentence. (Sweetser 1990: 64) Sweetser’s approach is not unproblematic, as she herself admits. For one thing, she is unable to explain why some metaphorical mappings are better than others. For instance, why does positive can hardly allow an epistemic reading, unlike can’t, could,orcan in interrogatives? The same holds for need, whose epistemic potential only arises in negative or interrogative environments. And why is an ep- istemic meaning ruled out altogether for shall? In his review of Sweetser (1990), Foolen (1992) argues that a purely monosemic account a ` la Wierzbicka, which views the different senses of the modals as pragmatic ambiguities, cannot account for these restrictions. It is difficult to see, though, how Sweetser’s approach could (as Sweetser herself admits). 12 Langacker (1990, 1991a, 1999) also takes a force-dynamic stance in his char- acterization of the English modals, both for their root and epistemic uses. Lan- gacker’s account of force dynamics in epistemic modals differs considerably from Sweetser’s, though. In his view, it is not so much the force of evidence which pushes the speaker toward a certain conclusion, but rather the highly abstract force residing in reality’s evolutionary momentum, that is, reality’s constant evolution based on (the speaker’s conception of) its structure. Note, however, that the notion of evolutionary momentum remains to some extent speaker-related as well, be- cause ‘‘the speaker is involved in any case as the primary conceptualizer and the person responsible for assessing the likelihood of reality evolving in a certain way’’ (Langacker 1991a: 274). An analysis in force-dynamic terms is also presented by Achard (1996b, 1998) for the French modals savoir, pouvoir, and devoir. 13 Achard divides the French modals into three usage groups (main-verb constructions, root senses, and epistemic sen- ses), 14 which correspond to the three subjectification stages presented by Langacker (1990, 1991a) to capture the diachronic evolution of the English modals (see below). The French modal pouvoir in its ‘ability’ sense, which represents the main-verb construction, is said to exhibit the following force-dynamic configuration: ‘‘Theforce stored in the locus of potency allows the latter to overcome the resistance (or force) coming from the activity profiled in the landmark, and therefore perform an oc- currence of that process’’ (Achard 1998: 143). Note, furthermore, that in the ‘ability’ sense the locus of potency—that is, ‘‘the origin of the force responsible for the 874 tanja mortelmans potential realization of the landmark process’’ (142)—is located in the subject. At the same time, the speaker’s role is limited to that of a mere conceptualizer: the speaker does not participate in the force-dynamic configuration. In its root sense of ‘pos- sibility’, pouvoir focuses on the presence (in the case of negated possibility) or ab- sence of a barrier between the subject and the infinitival process. Pouvoir’s root sense of ‘possibility’ differs from its ‘ability’ sense in two respects. First, the locus of po- tency is no longer equated solely with the subject, but becomes more diffuse, as it may also refer to the circumstances in general preventing or enabling the realization of the infinitival process (compare Il peut venir vous voir demain ‘It is possible for him to come and see you tomorrow’ with Marie peut soulever cent kilos ‘Marie is able to lift one hundred kilos’). 15 Second, the speaker’s role gains more prominence, as it is the speaker who points to the absence or presence of a barrier, that is, to the circumstances surrounding the modal situation. The speaker can therefore be taken to partake, albeit minimally, in the force-dynamic configuration, as he or she has mental access to the locus of potency. If the speaker removes the obstacle by himself or herself, the speaker’s involvement becomes stronger. This is particularly clear in the social domain, in which case the removal of a social barrier equates with granting permission (Tu peux aller au cinema ‘You can go to the movies’). The ‘obligation’ sense of devoir can be described along the same lines: the speaker’s role is ‘‘strong’’ when the speaker associates himself or herself with the locus of potency (the source of obligation); a weaker role is attributed to the speaker when he or she only conveys the source of obligation. It should be noted that Achard’s distinction between a weaker and a stronger speaker role in the root domain considerably deviates from Langacker’s descrip- tion, who—in view of the fact that the evolution of modal verbs can be described in terms of the locus of potency becoming progressively less salient and well defined— explicitly states that the identification of the source of potency (the locus of obli- gation or permission) is not a crucial matter: ‘‘An analysis of the modals ought to focus more on the existence and strength of the directed potency than on pinning it down to a specific source and type’’ (Langacker 1991a: 272). In fact, Achard’s ideas on weaker and stronger speaker roles seem to be more in line with Traugott’s (1989) more pragmatic interpretation of the subjectification process (see section 2.5). 2.3 From Root to Epistemic Modality: Metaphor, Metonymy, Minimal Shifts/Partial Sanctioning? Whether the development of epistemic modality out of root modality is indeed an instance of metaphorization (Johnson 1987: 48–61; Sweetser 1990; Pelyva ´ s 1996, 2000) is a matter of considerable debate (for an overview, see, e.g., Heine 1992: 37– 46; Nuyts 2001: 182–83). For one thing, it remains rather unclear how the source domain for the metaphorical mapping has to be established, that is, which aspects of the image-schematic structure of the root modal are to be mapped onto the epistemic world. For may, for instance, Sweetser, following Talmy, takes the modality in cognitive linguistics 875 ‘permission’ meaning of the verb as a starting point, with may denoting a potential, but absent barrier: ‘‘May denotes lack of restriction on the part of someone else’’ (Sweetser 1990: 53). Pelyva ´ s(1996, 2000), however, correctly remarks that the ep- istemic meaning has not developed out of the ‘permission’ meaning (as is shown, for instance, in Bybee, Perkins, and Pagliuca 1994: 199), but rather out of a prior ‘possibility’ sense of the verb. 16 A number of authors, mainly working within grammaticalization theory (Traugott and Ko ¨ nig 1991; Hopper and Traugott 1993; Diewald 1999), reject the metaphorical analysis, since the conceptual leap evoked by a metaphorical mapping from one discrete conceptual domain onto another is not in line with the gradual character of the actual development of the epistemic meaning. They regard the extension from root to epistemic meanings as a metonymic process based on contiguity. Must in the epistemic sense of ‘I conclude that’ derived from the obligative sense of ‘ought to’ by strengthening of conversational inferences and subjectification. If I say She must be married in the obligation sense, I invite the inference that she will indeed get married. This inference is of course epistemic, pertaining to a state of affairs that is anticipated to be true at some later time. (Traugott and Ko ¨ nig 1991: 209) Bybee, Perkins, and Pagliuca (1994: 196) in general advocate the metonymy position as well, whereby ‘‘inferences that can be made from the meaning of a particular modal become part of the meaning of that modal.’’ They do not accept this view for English must, however, arguing that ‘‘the epistemic use of must arises in contexts with aspectual interpretations distinct from the obligation uses’’ (201 ); the ap- propriate conversational implicatures, therefore, do not arise. Goossens (1999, 2000) takes yet another stance, as he rejects both metaphor and metonymy as the basic patterns of meaning extension in the development of epi- stemic out of root modality. On the basis of corpus material, Goossens describes the development from root to epistemic must as a concatenation of minimal and very gradual shifts of uses, which are only partially sanctioned with respect to the con- ventionalized uses of the verb. Genuinely metonymic uses in which a deontic and an inferential (epistemic) reading are simultaneously possible do occur (albeit less frequently than expected) and have supported the development of the subjective epistemic meaning. These semantic developments are to be seen against the back- ground of a more global shift: it is the process of ‘‘subjectification in the participant- external, more specifically the deontic, area [which] paved the way for the devel- opment of the (subjectified) epistemic sense’’ (Goossens 2000 : 167). 2.4. Subjectification and ‘‘Grounding Predications’’ Subjectification, defined as the ‘‘shift from a relatively objective construal of some entity to a more subjective one’’ (Langacker 1999: 297), not only plays a crucial role in our understanding of the diachronic evolution of the English modals, but also 876 tanja mortelmans helps to clarify their grammatical status vis-a ` -vis the (less grammaticalized) modals in other languages like German, Spanish, or French. The distinction between ob- jective and subjective construal (see also Verhagen, this volume, chapter 3) is based on the perspective taken by the conceptualizer on a particular scene; it is thus a matter of viewing. In Langacker’s definition, an entity is objectively construed in- sofar as it is the explicit focus of attention, whereas it is subjectively construed to the extent that it remains offstage and unmentioned. Langacker draws explicit atten- tion to the fact that his technical use of the terms ‘‘subjective’’ and ‘‘subjectification’’ differs from Traugott’s (1989, 1995) definition. For Traugott, subjectification refers to the diachronic process whereby ‘‘meanings become increasingly based in the speaker’s subjective belief state/attitude toward the proposition’’ (1995: 31). Whereas Langacker’s subjectification is a matter of perspective and vantage point, sub- jectification a ` la Traugott pertains to the ‘‘domain in which a relation or property is manifested’’ (Langacker 1999: 393). In the case of the English modals, Langacker argues that subjectification can be witnessed to the extent that the locus of potency, which in the original main verb stage can be identified with the (onstage) subject, 17 has come to be gradually construed in a more implicit and diffuse manner. With root modality (exemplified in 2), the source of potency (Talmy’s Antagonist) ‘‘may be the speaker but need not be. It is not necessarily any specific individual, but may instead be some neb- ulous, generalized authority’’ (Langacker 1999: 308). (2) a. You must go home right away—your wife insists. b. Passengers should arrive at the airport two hours before their flight. The same holds for the target of the potency (Talmy’s Agonist), which may be a specific individual, but again, which need not be. In (3), for instance, the force is ‘‘simply directed toward realization of the target event, to be apprehended by anyone who might be in a position to respond to it’’ (Langacker 1999: 308). 18 (3) There may not be any alcohol served at the party. Epistemic modals, then, show a maximally diffuse source and target of potency, ‘‘inhering in the evolutionary momentum of reality itself as assessed by the speaker/ conceptualizer’’ (Langacker 1999: 309). In the case of the English modals, the process of subjectification is accompanied by formal grammaticalization to the extent that the English modals are claimed to function as ‘‘grounding predications.’’ Together with tense and person inflections, the English modals are said to ground a finite clause, that is, to locate the process designated by the content verb in a particular epistemic region vis-a ` -vis the ground (the speech event, its participants, and its immediate circumstances). The subjectification of the English modals has proceeded in two phases: a first phase involves the realignment of the potency relation from the subject (as an ob- jective participant) to a more subjectively construed participant (the default case being the ground itself). 19 This reoriented relationship, typically anchored in the ground, remains in profile, however; that is, it is construed with a considerable modality in cognitive linguistics 877 degree of objectivity (this situation, as Langacker claims, can still be observed for the German modals, which, being less grammaticalized than the English modals, are still grounded by tense and mood; for a more qualified assessment of the ground- ing status of the German modals, however, see Mortelmans 2000, 2001). For modals to function as genuine grounding predications, the potency relationship itself must be construed subjectively. The modal thus does not profile the potency relationship (anymore), but the complement process at which the potency is directed. 20 The typical formal characteristics of the English modals are interpreted as reflecting their grounding function. As grounding predications, the English modals are aux- iliaries that necessarily precede the clausal head and cannot occur within it. They lack participialand infinitival forms (unlike the German and French modals),whose atemporal character would be inconsistent with the function of grounding a finite clause. The modals lack a third-person singular marker -s, since the latter is viewed as a grounding predication in its own right, situating the state of affairs within im- mediate reality. 21 The semantic import of the English modals is characterized with respect to a number of Idealized Cognitive Models, the most essential of which is the so-called ‘‘basic epistemic model’’ (which is also inherent in Achard’s ‘‘basic reality’’; see below). This model is made up of ‘‘known reality’’ (comprising those situations that are accepted by a conceptualizer as being real), ‘‘immediate reality’’ (reality at its latest stage of evolution functioning as the vantage point from which the concep- tualizer views things), and ‘‘irreality’’ (everything other than known reality). Note that Langacker adheres to a dynamic view on reality, which is con- ceptualized as ‘‘an ever-evolving entity whose evolution through time continu- ously augments the complexity of the structure defined by its previous history’’ (1994: 139). Roughly, then, the absence of a modal is said to indicate that ‘‘the speaker accepts the designated process as part of known reality’’ (1991a: 245)—the unmarked option. A modal, by contrast, locates the process somewhere within irreality. In order to arrive at a finer characterization of the semantic contribution of modals, especially in their epistemic usage, Langacker introduces the ‘‘dynamic evolutionary model’’ (reflected in Achard’s 1998, 2002 conception of ‘‘elaborated reality’’; see below), which integrates the notion ‘‘structured world’’ with force dynamics. The notion ‘‘structured world’’ recognizes the fact that we conceive of the world as being structured in a particular way, that we do not feel surrounded by mere chaos. It tries to capture the difference between (i) incidental events (which simply occur, but cannot be predicted or anticipated) and (ii) those events that are ‘‘direct manifestations of the world’s structure’’ and as such exhibit a degree of reg- ularity and predictability (Langacker 1991a: 264). Future events of the latter kind can be projected, confidently anticipated, as ‘‘present circumstances include those under which the world is biased toward the occurrence of particular sequences of events’’ (277). It is this bias toward certain developmental paths, ‘‘this tendency for reality, having evolved to its present state, to continue its evolution along certain 878 tanja mortelmans paths in preference to others’’ (277) that can be regarded in force-dynamic terms, namely as the force inherent in reality’s evolutionary momentum. The locus of potency can therefore be said to be maximally vague und undifferentiated. 22 The cases in which this force is conceived as strong enough to push reality toward certain future events (which can be confidently anticipated) are referred to as ‘‘projected reality’’; the term ‘‘potential reality’’ is used for those future paths which are not excluded from being followed. The epistemic modals may and will, 23 then, are said to place the designated processes in the realm of potential and projected reality, respectively. A similar analysis is put forward by Achard (1996b, 1998) for the epistemic uses of the French modals devoir and pouvoir and by myself (Mortelmans 2001) for the German modals m € uussen and k € oonnen. In particular, devoir and m € uussen are said to situate the infinitival process within projected reality, while pouvoir and k € oonnen situate it within potential reality. 24 Achard describes the evolution of the French modals from main-verb constructions to epistemic senses in terms of a subjective realignment of their modal force; that is, ‘‘the modal force anchored by the subject becomes progressively more and more aligned onto the conceptualization relation anchored by the speaker’’ (1998: 163). This process of subjectification is mirrored in a number of formal properties: to the extent that the subject’s intentionality is less at stake (and its role as a locus of potency is therefore minimal), there are fewer grammatical constraints imposed by the modal on the complement process; that is, the infinitive can be marked for voice (passive) or perfect aspect. In other words, to the extent that the speaker becomes more clearly associated with the locus of potency, possible constraints on the infinitival complement are loosened, since ‘‘the speaker gains independent access to the complement process’’ (Achard 1998: 169). Markers of aspect and voice—both instances of conceptual manipulation of the complement process—are interpreted as signals that the speaker (instead of the subject) has gained conceptual control over the infinitival process. 25 In spite of the fact that the epistemic usage is strongly subjectified, Achard does not take the pos- sible grounding status of the epistemic French modals into consideration. Follow- ing Achard, clausal grounding in French is effected by the tense and mood oper- ators only (for the latter, see section 3). Figure 33.2 . Langacker’s basic epistemic model modality in cognitive linguistics 879 . belongs to the conventionalized semantics of the modals, any other specification (e.g., the identification of the imposer and imposee of the modality) is a matter of pragmatics and should therefore. at the expense of the weaker one; and d. The resultant of the force interaction—either action or rest, assessed only for the Agonist. The English modals, then, are viewed as constituting the. that the identification of the source of potency (the locus of obli- gation or permission) is not a crucial matter: ‘‘An analysis of the modals ought to focus more on the existence and strength of