ILLOCUTION AND COGNITION: THE CASE OF APOLOGIES

This paper analyzes the speech act of apologizing from the point of view of the conceptual tools provided by Cognitive Linguistics. Speech acts have not received a lot of attention in this approach to language. Notable exceptions are Panther & Thornburg (1998), Pérez (1997, 2001), Sweetser (2000), and Pérez & Ruiz de Mendoza (2002). While Sweetser’s study focuses on the performative value of metaphor, the studies carried out by Panther & Thornburg, on the one hand, and Pérez & Ruiz de Mendoza, on the other, share a commitment with reinterpreting traditional speech act theory in terms of cognitive modeling. Thus, Panther & Thornburg (1998) claim that illocutionary force is a form of inferential activity grounded in metonymy. Pérez & Ruiz de Mendoza (2002), in turn, have developed a more sophisticated theory according to which the illocutionary value of an utterance is a function of a number of cognitive models which interact in different but predictable ways. Pérez & Ruiz de Mendoza have thus analyzed the interpretation of directive speech acts as a combination of metaphorical and metonymic operations performed on propositional cognitive models of interpersonal nature, in connection to pre-conceptual sensorimotor and spatial representations, such as the FORCE image-schema. Here we extend Pérez & Ruiz de Mendoza’s account to the domain of expressive speech acts with an especial focus on apologies. We show that expressive speech acts are also sensitive to an analysis based on interpersonal cognitive models and image-schematic representations. In order to do so, we outline the essentials of an illocutionary scenario for apologies and explore the way it is exploited by means of other cognitive operations, especially metonymy. We additionally show that Pérez & Ruiz de Mendoza’s cost-benefit idealized cog249 Cuad. Invest. Filol., 29-30 (2003-2004), 249-268 nitive model, which these authors have postulated for directive and commissive speech acts, also lies at the base of our understanding of expressive categories.

nitive model, which these authors have postulated for directive and commissive speech acts, also lies at the base of our understanding of expressive categories.

Introduction
As is well known several studies have been carried out within the framework of Cognitive Linguistics that have provided evidence supporting the idea that language is a part of general cognition and that, as a result, linguistic categories are subject to prototype effects (Lakoff, 1987). The evidence so far has mainly focused on levels of linguistic description like syntax, morphology, phonology, and the lexicon, but little attention has been paid to speech acts. A notable exception is the contributions by Pérez (1997) and Pérez & Ruiz de Mendoza (2002), where directive speech acts are studied from the point of view of cognitive operations carried out on different cognitive models. More specifically, Pérez & Ruiz de Mendoza (2002) argue that underlying the interpretation of directive speech acts is a social cognitive model, of an interactional nature, called the cost-benefit model, which is exploited metaphorically and metonymically by the hearer in order to derive the illocutionary meaning of directive utterances. Interestingly enough, in this approach, the social directive dimension of speech acts is interpreted to run parallel to the physical dimension of Johnson's (1987) FORCE imageschema, which allows Pérez & Ruiz de Mendoza to postulate a mapping from the physical world of force to the non-physical interactional understanding of social action.
In this context, the aim of this paper is to prove that expressive speech acts can be explained from the perspective of the interaction between cognitive models like metaphor, metonymy and image-schemas. I shall do so by combining insights from Lakoff's theory of cognitive models within Cognitive Semantics (Lakoff, 1987(Lakoff, , 1993Lakoff & Johnson, 1999) and from post-Gricean pragmatics, especially relevant aspects of Leech's Politeness Principle (e.g. the well-known cost-benefit scale; cf. Leech, 1983). Then I will apply these theoretical notions to the analysis of the speech act of apologizing.

Idealized Cognitive Models
Two basic assumptions on Cognitive Linguistics are that linguistic categories show prototype effects which occur at every level of linguistic description and that prototype effects are not organizing structures of knowledge themselves, but the result of such an organization. Among the attempts made to look for the sources of prototype effects we find Lakoff's Idealized Cognitive Models (or ICMs), which he divides into propositional, metaphoric, metonymic, and image-schematic. Propositional models specify elements, their properties and their different forms of interaction. For example, our propositional model of 'water' includes our knowledge of its boiling point at 100°on the Celsius scale and of its freezing point at 0°on the same scale. Metaphoric models are mappings across domains. Thus, we may understand being in love (which is an abstract mental representation) as being inside a container (which is a concrete representation) whose internal conditions affect the protagonist. (e.g. in I am deep in love the emotional state of feeling a lot of love is seen in terms of the depth that the protagonist has reached inside the container). Metonymic models are internal mappings within the same domain. For example, when we talk about a customer who has eaten a cheese hamburger by saying The cheese burger has left without paying, we are using one characteristic of this customer (his eating a cheese hamburger) to refer to him. Image-schematic models are schematized images or spatial concepts (such as the notions of trajectory, path or container) and orientational concepts (up/ down, front/ back, left/right). For example, the sentence Petrol prices have gone up in the last month makes use of the up-down orientational schema to talk about quantity. This is based upon a correlation between height and quantity derived from our common everyday experience with levels going up as the amount of a substance increases (think of water being poured into a glass).
Within the theory of propositional models, Lakoff cites frame semantics as an important development (cf. Fillmore & Atkins, 1992). Fillmore's focus is mainly upon the description of encyclopedic knowledge through using frame elements and their syntactic realizations. Here it is worth mentioning Ruiz de Mendoza's approach to encyclopedic knowledge, which differs from frame semantics substantially. For every concept he proposes a general definer (understood as a set of necessary, though not sufficient, attributes of the concept) and their related prototypical associations. Ruiz de Mendoza (1996) exemplifies his theory with the concept of 'mother', whose general definer is 'woman who has (at least) one child' and whose prototypical associations are defined in terms of all our relevant knowledge about human mating, pregnancy, birth-giving, nurturing a baby, and so on. We will make use of Ruiz de Mendoza's notion of 'definer' in our characterization of apologies by postulating that the FORCE imageschema serves as a general definer for the source domain of a metaphorical mapping from physical to social action.
Image-schematic models have been originally dealt with by Johnson (1987). According to him, there are invisible forces around us which manifest structures that are very much a part of our having coherent, meaningful experiences that we can call into consciousness, understand, reason about, and communicate in language. These structures are image-schemas. Johnson's conceptualization of image-schemas and classification of FORCE image-schemas (i.e. COMPUL-SION, BLOCKAGE, COUNTERFORCE, DIVERSION, REMOVAL OF RES-TRAINT, ENABLEMENT, and ATTRACTION) lie at the basis of our study, too, especially the COMPULSION and REMOVAL OF RESTRAINT image-schemas, which we will deal with below.

Leech's Politeness Principle
Finally, Leech's contributions to our analysis are to be mentioned, too. We are referring to his Politeness Principle (PP) and his proposal of the existence in our culture of the cost-benefit scale as a central part of this principle (cf. Leech, 1983). The two other scales, indirectness and optionality, seem to play a more subsidiary role.
The cost-benefit scale is crucial to our understanding of the expressive speech act of apologizing. The Politeness Principle, which makes extensive use of this scale, is considered as part of the Interpersonal Rhetoric, i.e. as one of the principles which are observed in the planning and interpretation of messages. This principle is regulated by five Maxims: the Tact Maxim (Minimize cost to other/ Maximize benefit to other), the Generosity Maxim (Minimize benefit to self/ Maximize cost to self), the Approbation Maxim (Minimize dispraise of other/ Maximize praise of other), the Modesty Maxim (Minimize praise of self/ Maximize dispraise of self), the Agreement Maxim (Minimize disagreement between self and other/ Maximize agreement between self and other) and the Sympathy Maxim (Minimize antipathy between self and other/ Maximize sympathy between self and other).
According to Leech (1983: 79-151), the Tact Maxim applies to Searle's directive and commissive categories of illocutions, which refer, in their propositional content X, to some action to be performed, respectively, by the hearer (H) or the speaker (S). It is an 'other-centred' Maxim. In the utterance Have another sandwich there is a high degree of politeness as X is beneficial to H and implies no cost for him. The Generosity Maxim applies to impositives and commissives and is a 'self-centred' Maxim. In the sentence I can lend you my car there is an implied benefit to H and a cost to S so we can see the activation of the Generosity Maxim and, as a consequence, of the PP. The Approbation Maxim is used mainly in expressive and assertive speech acts such as What a marvellous meal you cooked! in which S is maximizing praise of other, and hence, politeness is achieved. The Modesty Maxim works with expressive and assertive speech acts, too. In the following examples we can see how it is felicitous to agree with another's commendation (e.g. (1)) except when it is a commendation of oneself (e.g. (2)).
(1) A: They were so kind to us. B: Yes, they were, weren't they?
(2) A: You were so kind to us. B: Yes, I was, wasn't I?
The Agreement and Sympathy Maxims are activated mainly in assertives. The Agreement Maxim is based on the fact that there is a tendency to exaggerate agreement with other people (e.g. (3)), and to mitigate disagreement by expressing regret, partial agreement (e.g. (4)), etc.
(3) A: A referendum will satisfy everybody.
(4) A: English is a difficult language to learn. B: True, but the grammar is quite easy.
The Sympathy Maxim explains why congratulations and condolences are polite speech acts, even though condolences express beliefs which are negative with regard to the hearer: I'm terribly sorry to hear that your cat died, in which the PP principle is at work as the S is maximizing sympathy between self and other.
The Pragmatic Scales that Leech proposes as having a bearing on the degree of tact appropriate to a given speech situation are: the cost-benefit scale, which regulates the cost or benefit of the proposed action A to the speaker (S) or to the hearer (H), the optionality scale on which illocutions are ordered according to the amount of choice which S allows to H and the indirectness scale on which, from S's point of view, illocutions are ordered with respect to the length of the path (in terms of means-ends analysis) connecting the illocutionary act to its illocutionary goal.
The PP is an interpersonal cultural model that could be defined as an abstract ICM. In the same way that 'mother' is defined as an ideational and non-abstract model and 'love' as an ideational and abstract one, in Pérez & Ruiz de Mendoza (2001) we find a reformulation of Leech's cost-benefit scale as an ICM: (a) If it is manifest to A that a particular state of affairs is not beneficial to B, and if A has the capacity to change that state of affairs, then A should do so.
(b) If it is manifest to A that a potential state of affairs is not beneficial to B, then A is not expected to bring it about.
(c) If it is manifest to A that a potential state of affairs is beneficial to B, then A is expected to bring it about.
Whereas Leech's cost-benefit scale just predicts the existence of an interactional rule, the ICM of cost-benefit formulated by Pérez & Ruiz de Mendoza facilitates our task of recognizing the cognitive operations that underlie such a rule.

Operational and non-operational cognitive models
In Ruiz de Mendoza (1996) we find a distinction between operational cognitive models and non-operational cognitive models. Operational cognitive models are defined as dynamic and involve a cognitive operation, either metaphoric or metonymic, as they are the result of a mapping between two conceptual domains (metaphor) or between a domain and one of its subdomains. Non-operational cognitive models are static and resultative, as is the case with Filmore's frames and Johnson's image-schemas. Operational cognitive models make use of non-operational cognitive models for their functioning, which does not mean that they are less central to our conceptual systems and as a consequence, for a theory of knowledge organization.
Metaphorical operations (correlation, comparison; cf. Grady, 1999) act on propositional ICMs and on image-schemas. In so doing they generate metaphorical models. For example, in the conceptual metaphor LOVE IS A JOURNEY we find a correlational operation in which lovers are seen as travelers, the love relationship as their vehicle, lovers' common goals as the destination at the end of the journey and the difficulties in the relationship as impediments to travel.
In contrast to what is the case with a metaphor like LOVES IS A JOURNEY, other metaphors exploit resemblance between the source and target domains. Thus, in John is a shark we are saying that a person is a member of the class of aggressive, voracious, predatory creatures. We are using, then, a comparison operation in which the behavior of a person (John) is compared to the behavior of an animal (a shark). According to Ruiz de Mendoza (2004) it may be argued that the 'is-a' construction, in its application to the expression John is a shark, focuses on culturally interpreted behavioral features of sharks that are set in correspondence with knowledge about people's behavior to yield the interpretation 'John is fierce, aggressive, dangerous, strong, powerful'. Metonymic operations (of expansion and reduction; cf. Ruiz de Mendoza & Santibáñez, 2003) act on propositional ICMs or on image-schemas, but they can work on metaphorical models by creating all necessary conditions for a fullyfledged mapping to take place.
In The sax has the flu we find a metonymic operation of expansion in which the instrument a musician plays (in this case, the sax) stands for the musician. In Repsol is on strike the whole company stands for the workers, i.e. we have a metonymic operation of reduction.
In the following section we apply this distinction to the study of the speech act of apologizing.

A cognitive study of the speech act of apologizing
According to Panther and Thornburg (1998) inferencing in speech act interpretation can be explained through a metonymic cognitive operation in which each of the components of the illocutionary scenario may stand for an act of requesting. Panther and Thornburg (1998:759)  An analysis carried out in terms of illocutionary scenarios and cognitive operations (like the ones associated with metonymies) is capable of revealing the different shades of meaning of both conventional and non-conventional linguistic expressions carrying an illocutionary load. This kind of analysis has the advantage over previous explanations that it goes beyond the mere observation that some expressions are explicitly performative while others are not, as is the case in the Searlean and all Post-Gricean approaches (e.g. Leech (1983) and Sperber & Wilson (1985), among others.

Consider the following examples:
(1) I apologize for being late.
(3) I owe you an apology for being late.
(5) It is with great/ extreme regret that I am late. (6) I am very/really/ awfully/ terribly/ so sorry I am late. (11) My dentist meeting was not supposed to take that long.
A relevant part of the semantic import of these utterances can be accounted for by considering the way different ICMs interact.
In example (1) there is an explicit activation of the speech act by means of the performative verb 'apologize'; in the rest of the examples (2) to (11) there is not such an explicit performative verb but different expressions that are understood as apologies.
The speech act of apologizing is soundly explained by taking into account points (a) and (c) of the ICM of Cost-Benefit as formulated by Perez & Ruiz de Mendoza (2001). If a person is aware of the fact that the addressee may feel hurt, he is supposed to change that negative state of affairs. Since it is physically impossible to change an unfortunate past situation, the only thing that the speaker can do is soften the adverse consequences by making the addressee aware of his discomfort and showing regret about the situation.
The speech act of apologizing in (1) is explicitly conveyed because of the use of the verb 'to apologize' which brings about the full activation of the illocutionary scenario for apologies formulated above.
However, in example (2) it is the hearer that is made to bear the brunt of speech act interpretation. By means of the activation of the AFTER part of the illocutionary scenario (i.e. 'H is expected to accept S's apologies') we understand example (2) as an apology. The speaker is not apologizing directly; he is just activating the last part of the illocutionary scenario somehow forcing the hearer to implicate that if he is expected to accept S's apologies it is because S has actually done something wrong to him. In this way the person apologizing is placing himself in a position of inferiority with respect to the addressee; and it is up to the addressee to determine whether there will be a situation of balance again.
Metonymy, too, pays an important role in the conceptualization of example (3) since by activating the CORE of the illocutionary scenario for apologies (i.e. 'S makes H aware of his purported uneasiness') the whole speech act is called upon. If we take into account part (b) of the ICM of cost-benefit ( i.e. 'If it is manifest to A that a potential state of affairs is not beneficial to B, then A is not expected to bring it about') we can note that the activation of the CORE of the illocutionary scenario directly calls upon this part of the illocutionary model. Thus, the interaction becomes a debt-solving situation in which the speaker is a debtor of the addressee, which places the former in a situation of submission until the latter accepts his apologies. Underlying the semantic value of this expression, we also find the COMPULSION image-schema, according to which the apology is seen metaphorically as an object that goes from speaker to addressee. We will deal with the involvement of the FORCE image-schema in the interpretation of apologies in some more detail below.
Examples (4) and (5) focus on the second aspect of the BEFORE part of the illocutionary scenario for apologies (i.e. 'S regrets X'). By mentioning a part of the illocutionary scenario the whole of it is accessed. If we pay attention to the modifiers accompanying both examples we realize that they enhance the truthfulness and sincerity of the apology rather than work on the whole illocutionary model.
Example (6), again, mentions just a part of the illocutionary scenario for apologies to refer to the whole of it. It activates one of the parts of the BEFORE component of the scenario for apologies, i.e. 'S feels guilty for X'. Examples (7) to (8) focalize the speech act of apologizing from the perspective of the addressee, i.e. they emphasize forgiveness by means of two explicit verbs; "pardon" and "forgive", which invoke the AFTER part of the illocutionary scenario for apologies. We observe that two verbs belonging to the expressive category of 'pardoning' and 'condoling' (cf. Leech, 1983) are understood, by means of a part-whole metonymy, as an apology. This activation of the AFTER part of the illocutionary scenario is what explains that verbs belonging to the expressive category may stand for verbs from a different category since the former only have the function of activating a part of the illocutionary scenario. We grasp the whole illocutionary scenario by means of inference. This observation is related to the more general observation that it is possible to make use of a speech act category to convey by means of inference even an opposing category, as is the case with the use of the verb "promise" to make threats (e.g. I promise I won't allow you to go to the party).
The apology in example (9) is implicitly activated, too. Metonymy also plays a role here. In this utterance, however, the relationship is established between the fear of the speaker to be punished by the addressee and the appropriateness of the apology in such a context to avoid negative consequences. There is a close link between the act of apologizing and the desire on the part of the speaker to escape punishment or future negative consequences of his past action.
Examples (10) and (11) are cases of implicit activation, too, and they are even more indirect because of the two implications needed for their comprehension. In example (10) we find a metonymy in which the speaker does not apologize directly but rather expresses the speaker's recognition that he did wrong. It is this recognition that stands for the whole speech act category. (10) conventionally implicates that S did not call H; once this implication has been derived it activates part (a) of the BEFORE component of the apology scenario. Example (11) is a clear instance of metonymy activated by a semantic connection between the uncontrolled reason for a person being late and the fact of being late. A case of metonymy, too, in which the fact of the uncontrolled reason for being late (and trying not to be responsible for this) stands for the whole illocutionary scenario for apologies. Two implications are needed, too, in order to understand it. S did not do what he should have done, which activates part (a) of the illocutionary scenario.

A general definer: FORCE image-schema.
In the last section above we have dealt with the metonymic basis of the comprehension of speech acts; however, it is also possible to understand speech acts by looking into other cognitive models that help us to understand social interaction in terms of physical interaction. Pérez and Ruiz de Mendoza (2001) have done this kind of analysis in connection with directive speech acts. In this category of speech acts the metaphorical relationship between physical force and social force happens to be more evident since there is an impositive dimension. In an expressive speech act this does not seem so evident. However, we are going to show that the connection exists.
Apologies are conceptually tied to FORCE image-schemas to the extent that they are capable of activating the cost-benefit cognitive model. An apology can be understood by the speaker as a way of avoiding or weakening the consequences of a non-beneficial action for the hearer. In this connection it will be noted that apologies make a more restrictive use of this schema than direct speech acts. This is due to the fact that apologies only exploit the cost-benefit ICM, as indicated above, in a very partial manner. Thus, while we may expect such notions as the relationship between FORCE and COUNTERFORCE to be operational in a speaker-addressee directive relationship (if the speaker puts pressure on the addressee to perform an action the addressee may challenge or not the speaker), such notions are immaterial in the case of attitudinal expressions made by the speaker with respect to a directive relationship. In what follows, we shall apply Johnson's description of the internal gestalt structure of FORCE image-schemas to apologies.
In order to show the reader how apologies can be conceptualized by means of a FORCE image-schema let us see how Johnson's description of the internal gestalt structure of FORCE image-schemas applies to this speech act category. Johnson (1987: 42-48) distinguishes these five characteristics as constitutive of force gestalts: -Forces are always experienced through interaction.
-Forces are provided with a vector quality or directionality.
-Forces tend to describe a single path of motion.
-Forces have origins or sources that can be moved by agents to targets or destinations.
-Forces have degrees of intensity.
-Forces constitute one of the possible ways in which we understand causal sequences All of these characteristics constitute what, using Ruiz de Mendoza's (1996) terminology, may be called the general definer for the FORCE image-schema, i.e. a set of necessary but not sufficient conditions for the notion. Other possible conditions, which would fall out of the general definer, would have to do with the way humans perceive forces and with their psychological impact in terms of control. Now, each of the necessary characteristics of the FORCE image-schema listed above can be applied to the logical structure of apologies in the following way: -Apologies are always experienced through interaction. Prototypically we cannot apologize if we have not done something that has influenced another person in a negative way and we need to apologize to the addressee via interaction. In this sentence 'I'm sorry about that', I apologized to the household as we ate our dinner in silence (BNC) it is via interaction between the speaker and the hearer that the speech act of apologizing takes place.
-Apologies are directed from a speaker to a hearer, e.g. I wish to express my apologies to those I let down. (BNC). In this sentence the apology is directed from the speaker (as expressed by the first person singular personal pronoun) to the hearer (the people that the speaker has let down).
-The apology will tend to figuratively describe a single trajectory between speaker and addressee.
-Apologies, in the same way as forces, are figuratively seen as objects that can be moved by an agent (the speaker) to a target (the addressee) as we can observe when we are reporting on somebody's apology. In the sentence Tse Ling himself came into the room, gave her his apologies for keeping her waiting, and bowed low (http://www.sluttpuppy.us/story/?cat=20&jump=T) we observe that the apology is moved from an agent to a target, in this concrete case this fact is emphasized by the verb 'to give', which implies movement.
-Apologies, as forces, present different degrees of intensity. We can be confronted with a simple apology such as Forgive me for being late or with a more intense one as the following: Please, accept my most sincere apologies for being late, please do.
-Apologies can be the explanation of a subsequent action, e.g. a father forgiving his son after the latter has apologized for borrowing the family car without parental permission.
We mentioned in the first section of this paper that Johnson (1987:45  The COMPULSION image-schema is a FORCE image-schema constituted by an external force physically or metaphorically pushing, or tending to push, an object. A clear example is the experience of being pushed by people in a crowded place. F1 ROCÍO MANCEBO FRANCISCO 262 Cuad. Invest. Filol., 29-30 (2003-2004. COMPULSION IMAGE-SCHEMA ( TAKEN FROM JOHNSON, 1987:45) The REMOVAL OF RESTRAINT image-schema is a FORCE image-schema that involves either the physical or metaphorical removal of a barrier from the action of a force, or the absence of a barrier that was potentially present. An example of the application of this schema is a person trying to enter a place blocked by a door that he has to open in order to go in.
Johnson's representation of the REMOVAL OF RESTRAINT image-schema is horizontal, but forces may have a vertical orientation, which applies in the case of the schema in question as shown in figure 5 above. We propose, then, a combination of the VERTICALITY and the REMOVAL OF RESTRAINT imageschemas. Drawing on work on orientational metaphors by Lakoff and Johnson (1980: 15) we argue that the expressive speech act of apologizing exhibits initially (i.e. before the acceptance of the speaker's apologies takes place) a clear tendency to be understood in terms of the metaphoric mapping HAVING CON-TROL or FORCE IS UP and BEING SUBJECT TO CONTROL or FORCE IS DOWN 1 . In the case of apologies the speaker places himself in a voluntary situation of inferiority (the DOWN position) with respect to the hearer (the UP position), who feels wronged or otherwise harmed by the speaker and, thus, with a metaphorical possession of force and control over the speaker. The speaker's pur-ILLOCUTION AND COGNITION: THE CASE OF APOLOGIES 263 Cuad. Invest. Filol., 29-30 (2003-2004, 249-268 1. In Peña (2003:123-195) we find a study in which both the FORCE image-schema and the VERTICALITY image-schema are considered as subsidiary to the PATH image-schema. As some kind of force causes motion and the concept of motion cannot be understood without the notion of path it may be postulated that the PATH and FORCE image-schemas are interrelated in structurally significant ways. Disagreeing with Johnson (1987: 122-123), Peña (2003 argues that it is possible that the basic logic of the VERTICALITY construct draws its main conceptual structure from the PATH schema.  (TAKEN FROM JOHNSON, 1987: 46) F 1 ported wrongdoing is the figurative obstacle in his relationship with the hearer so if removal is successful (i.e. if the hearer accepts the speaker's apologies) the initial situation of imbalance between speaker and hearer will be solved and normal force balance restored.
We argue, then, that the expressive speech act of apologizing can be understood by means of the COMPULSION and the REMOVAL OF RESTRAINT image-schemas with a vertical orientation, thus applying the VERTICALITY image-schema, as is shown in the examples we present below.
(1) Pardon me if I write vehemently (BNC) In these instances of language the speaker's cause of discomfort and uneasiness towards the hearer (in italics) is figuratively moved from the speaker himself to the hearer; then, the figurative barrier that separates both speaker and hearer, i.e. the apology acting as the restraint that impedes a balance of forces, is removed through the acceptance of the speaker's apologies by the hearer.
The application of the VERTICALITY image-schema in the sense of the speaker feeling inferior to the hearer can be grasped through the expressions in italics, too. These expressions indicate that the speaker himself feels responsible for the negative situation he is apologising for and, thus, in a position of submission and inferiority with respect to the hearer; a position which is graphically represented by the DOWN situation in the VERTICALITY image-schema.

Conclusion
This paper has been able to show that expressive speech acts can be explained by using as methodological tools some of the conceptual categories popularized in Cognitive Linguistics, more specifically the interaction between two of Lakoff's idealized cognitive models (i.e. metaphor and metonymy), two of Johnson's FORCE image-schemas (i.e. COMPULSION and REMOVAL OF RESTRAINT), and Ruiz de Mendoza & Pérez's (2002) formulation of the costbenefit ICM, which is based on previous proposals by Leech (1983) within the domain of pragmatics.
What is innovative in this paper is twofold: on the one hand we have noted that expressive speech acts can be explained from a cognitive perspective, and not just from the point of view of semantics, pragmatics or discourse analysis, as had been previously done; on the other hand, we have also shown that apologies, even though generally classified as expressive speech acts, respond to the same conceptual constructs as directive and commissive categories. This should come as no surprise since apologies differ from other expressive speech acts (e.g. condoling, sympathizing) in that they are directly intended to avoid the dire consequen-ces of speaker's purported misbehavior. This does not mean that other expressive acts may not make occasional use of the force dynamics logic, as when the expression of condolences is intended to move the addressee to some form of action. However, these exploitations of the expressive category would seem to be more subtle and indirect, the expressive category itself being but an intermediate step in a more complex inferential chain (in some contexts condoling would open the way for the hearer to feel thankful and then do something for the speaker). In this context, what our observations ultimately mean is that apologies are closer to the directive-commissive continuum than other expressive speech acts, a fact which is more forcefully brought to the fore by looking into the image-schematic grounding of the speech act than by simply inquiring into its propositional nature as an illocutionary scenario.