Marina Sbisà (University of Trieste)

 Paper read at the International Conference Pragma99 "Pragmatics and Negotiation", Tel Aviv and Jerusalem, 13-16 June 1999

 (Provisional version)

Apologies are an interesting case of a speech act under many respects.

 1) They are actions performed in issuing linguistic utterances.

2) They have felicity conditions.

3) They achieve effects on the relationship between the interlocutors.

4) Their effect is subordinated to their being recognized as apologies by the addressee.

5) The utterances in which they are performed, although they often contain "illocutionary force indicating devices", do not always have the form of a single sentence.

Facts (1) and (2) suggest to consider apology as an illocutionary act fitting the standard version of speech act theory. But fact (3) plays no role in that theory, where the illocutionary effect consists merely of the hearer's recognition of the speaker's communicative intention: the effect achieved by successful apologies cannot be reduced to this recognition, but consists of the complete or partial extinction of an interactional debt threatening various aspects of the interlocutors' "faces". It should be noted that this effect cannot be dismissed as merely "perlocutionary", because it is clearly conventional (or even ritual), and is distinct from the actual forgiveness on the part of the addressee (who might recognize the apology as a ritual move, but still remain angry or give a negative judgement of the speaker). Moreover, fact (4) fits, rather than standard speech act theory, the old idea put forward by Austin that "unless a certain effect is achieved, the illocutionary act will not have been happily, successfully performed" (1962: 116). Finally, fact (5) does not fit standard speech act theory at all.

Therefore, from the study of the speech act of apologizing there may come suggestions for going beyond the received version of speech act theory (while not dropping the notion of speech act altogether). This paper will consider apologies in this perspective, namely, as actions consisting of the bringing about of an effect on the interpersonal relation, performed in issuing linguistic utterances which may be syntactically composite.

I shall attempt to formulate a conceptualization of the speech act of apologizing (based on previously given definitions and on my research¹), and shall claim that in the interactional dynamics of apologies there is a room for negotiation, namely, that the successfulness of an apology depends both on the way in which it is formulated and on whether the addressee takes it to be an apology. The extent of the room left to the addressee's decision varies in correlation with certain features of both form and content of the utterance used by the speaker.

Since the linguistic data I have collected are Italian, reference will be made to the Italian speech act of scusarsi, which conflates apologies and excuses. This is perhaps an additional complication, but, as we shall see, the relationship between apologies and excuses is itself an interesting issue and might be illuminating with respect to the understanding of the interactional function of apologies.

2. The conceptualization of apologizing: some open problems

Sociology, sociolinguistics, pragmatics have greatly contributed to the analysis of the speech act of apologizing as well as to the description of the linguistic means by which it is accomplished. It is interesting to notice that at least some authors have observed the relationship between apologies and excuses, which is emphasized in Italian by the availability of only one verb for both acts (scusarsi). However, as I shall try to show, the conceptualization of apologizing and the description of the interactional processes by which apologizing is achieved are still unsatisfactory.

Let us consider first some definitions which have been given to apologies.

Goffman (1971) defined apologies as a kind of remedial work which involves the splitting of the speaker's self into two parts, the one guilty of having offended the addressee, the other aligning him or herself with the addressee and with the violated norm. Excuses, according to him, have a remedial function too, but belong to another kind of remedial work, namely, accounts, which consist in redefining a potentially offensive act, so as to make it acceptable.

Brown and Levinson (1978, 1987) consider apology as a face threatening act, which damages to some degree the speaker's positive face, since in doing it the speaker admits that he or she has done a transgression (1987: 68). At the same time, by apologizing, the speaker pays the debt created by his or her transgression, thus restoring the interactional balance (1987: 236). Apologizing can also be a negative politeness strategy: by apologizing for doing a face threatening act (different from the apology itself), the speaker indicates his or her reluctance to impinge on the hearer's negative face and thereby partially redresses that impingement (1987: 187). As to excuses, Brown and Levinson consider them as face threatening acts which offend the speaker's negative face (the reason they give is not clear: "S indicates that he thinks he had good reasons to do, or fail to do, an act which H has just criticized; his may constitute in turn a criticism of H, or at least cause a confrontation between H's view of things and S's view"; in fact this is more likely to happen with explanations or justification than with excuses, and threatens the hearer's positive face rather than the speaker's negative one) (1987: 67). However, in discussing interactional balance they seem to conflate the function of excuses with that of apologies (1987: 237-38).

According to Fraser (1981), "an apology may be perfomed just in case two basic conditions are met: first, the speaker acknowledges responsibility for having performed some act; and second, the speaker conveys regret for the offense which came about as a result of the commission of the act", so that an apology may be said to do both things. Such a definition of apologies leaves no room to their connection with excuses, since in excuses the speaker attenuates or even disclaims his or her responsibility for the offending act. Fraser argues that an utterance of "I apologize for Xing, but I deny that I did it" cannot be used as an apology, but does not notice that "I apologize for Xing, but I did not intend to do it/ did it under duress/ did it by accident" can be so used.

Olshtain (1989) defines apology along the lines set by Brown and Levinson, as a speech act which is intended to provide support for the hearer insofar as he or she is affected by a violation; she considers admission of fault and responsibility as an essential component of the apology, and adds a component of self-humiliation. The latter component is consistent with the recognized face-threatening nature of apology with respect to the speaker's positive face, but it is not clear if it is related to the admission of fault or somewhat independent of it.

Holmes (1990) considers apology as a speech act directed to the addressee's face needs and intended to remedy an offense for which the speaker takes responsibility, and thus to restore equilibrium between speaker and addressee. She thinks that apology as an illocutionary act has some minimal felicity conditions, namely

(1) an act has occurred

(2) the speaker believes the act has offended the addressee

(3) the speaker takes some responsibility for the act.

The third condition admits of attenuated responsibility, so that Holmes's definition can account for the connection between apologies and excuses.

Criticism of this conception of apologies stems from more general doubts about the role of felicity conditions with respect to the performance of speech acts. Holmes claims that when the circumstances indicated by the felicity conditions obtain, what the speaker will say is likely to be interpreted as an apology. But the hearer does not in general know that the speaker believes that an act of his or hers has offended the addressee, nor that the speaker takes some responsibility for the act, independently of the utterance which is taken to count as an apology. So we cannot see how felicity conditions could play any role in determining whether the utterance has to be taken as an apology or not.

In connection with these definitions, the following linguistic means for performing apologies have been identified.

According to Brown and Levinson (1978, 1987), the apology's function of redressing the hearer's negative face can be achieved by: admission of responsibility for the virtual offence, indication of reluctance, giving overwhelming reasons, and begging forgiveness. They focus on the apology's redressing function with respect to a face threatening act which is still to come; indication of reluctance is clearly a linguistic strategy limited to this kind of situation. Expression of regret is classed by them as a way for begging forgiveness. The giving of overwhelming reasons may amount to offering an excuse.

Fraser (1981) distinguishes four direct strategies for apologizing (announcing the apology, stating one's obligation to apologize, offering to apologize, requesting acceptance) and five indirect strategies (expressing regret, requesting forgiveness, acknowledging responsibility, promising forbearance, offering redress). It is puzzling to notice that the two basic things which, according to him, are done in apologizing are listed among the indirect strategies, not among the direct ones; this perhaps depends on the fact that only linguistic forms containing the verb "apologize" or the name "apology" are considered as direct. Consistently with the definition given to apology, offering accounts or excuses is not considered as a strategy for apologizing, not even an indirect one.

Olshtain and Cohen (1983) and Olshtain (1989) distinguish five strategies for apologizing: two general strategies, applicable in any situation, that is using an illocutionary force indicating device (the formulaic, routinized forms of apology) and expressing the speaker's responsibility; and three situation specific strategies, that is explanation, offer of repair and promise of forebearance. Expressions of responsibility may consist in excuses, since they contain substrategies which invoke reduced competence, admission of carelessness, and the like.

Holmes (1990) distinguishes four main strategies: explicit expression of apology (comprising: offering an apology, expressing regret and requesting forgiveness), explanations (together with accounts, excuses and justifications), acknowledgement of responsibility, and promise of forebearance. There is an overlap between the second and the third strategy, since they both can be realized by offering excuses (since one of the substrategies for acknowledging responsibility is expressing lack of intent). It is interesting to notice that Holmes, in her wide sample of New Zealand English apologies, has never found realizations of the third strategy alone: this casts some doubt on the idea that expression of responsibility can by itself amount to an apology. In contrast, according to her, expression of regret and request of forgiveness (which were considered by Fraser as indirect strategies and were not specifically taken into account by Olshtain) can be considered as explicit strategies.

Kerbrat Orecchioni (1994), studying apologies in French, distinguished explicit realisations of apologies (consisting of requests of forgiveness) and implicit realisations, which may consist of descriptions of appropriate internal states, accounts, or explicitations of the committed offense. Excuses, which French conflates with apologies just as Italian does (1994: 162), are collocated within the group of implicit apologies, but without distinguishing them sharply from other kinds of accounts. It depends perhaps on the specific linguistic conventions of French that descriptions of appropriate internal states count as implicit apologies too (while Holmes counts expression of regret as an explicit strategy). But the precise status to be granted to the expression of regret is something on which the authors who have studied English data (such as Fraser, Olshtain, Holmes) do not wholly agree either.

Given the disagreements among the scholars who have studied apologies, as well as some recurrent unclarities in their analyses, we can conclude this brief survey by pointing at some open problems.

- What are the strategies for apologizing explicitly? is it legitimate to expect that English illocutionary force indicating devices translate into other languages literally?

- Is assumption or expression of responsibility really a central move for apologizing? How is it related to the remedial moves which Goffman called accounts, and in particular to excuses? Have all kinds of accounts the same relation to apologies, or a sharp distinction should be drawn (as suggested by Austin 1960) between excuses and justifications?

- What makes it so that apologies, however formulated, can restore the interactional equilibrium?

In the following parts of this paper we shall attempt to answer these questions with reference to the data we have collected as regards the Italian speech act of scusarsi. This is not to be straightforwardly identified with the speech act of apologizing, since it definitely includes the offering of excuses: in Italian, both moves aiming at obtaining forgiveness (by requesting it, or by displaying sorrow or repentance, and so forth) and moves aiming at redefining the offending behaviour in order to make it less offending or perhaps not offending at all, are called by one name (scuse). Our data regard both linguistic production in situations requiring apologies, and the understanding of apologies and other functionally related speech acts.

We have collected data about scuse by means of a discourse completion test inspired by the one used in the CCSARP research on apologies (Blum Kulka et al. eds., 1989). From the 7 situations used to elicit apologies in the CCSARP questionnaire, 4 have been selected and reformulated in Italian. Our questionnaire comprised a short description of each situation and a first turn by the offended participant and was submitted to 100 italian native speakers (students). This is the English translation of the questionnaire.

1. Professor A promised student B that that day he would give back the essay, but he hasn't finished correcting it yet.

B: Excuse me, professor, have you got my essay?

2. Student A had borrowed a book from Professor B, promising to bring it back that day, but he hasn't brought the book.

B: Have you brought back the book I lent you?

3. A hits B's car while parking.

B: You hit my car!

4. In an expensive restaurant, waiter A brings customer B some fruit instead of the speciality dessert that B had ordered.

B: I ordered the speciality dessert!

In all these situations, the response which is expected from A is an apology (and therefore, a fortiori, an act of scusarsi). A has done something wrong with respect to B: he has broken a previous agreement (1,2) or has run a concrete risk to damage B (3), or has disregarded B's command (4). B's turn makes this situation explicit. Thus A turns out to have threatened B's negative face (B's interactional rights) or, as we prefer to put it, has incurred in an interactional debt with respect to B, which has to be paid in order to restore the interactional balance. This can be done by apologizing (or, to say it in Italian, by issuing a scusa).

Responses to the discourse completion test were classified according to two different sets of criteria. At first, a comparison between the Italian data and the data collected by the CCSARP researchers (discussed by Olshtain 1989) was attempted, in order to detect possible specific features of the Italian procedure of scusarsi. In this case, the same criteria used by Olshtain were adopted. In a second phase, we realized that not all of the responses to the discourse completion test were full-fledged apologies and not even scuse, and that at least some of them were not apologies nor scuse at all. Therefore, we classified our data again trying to separate the utterances we found it sensible to call scuse from other speech acts whether having remedial function or not. Here are some considerations which stem from these two attempts to classification.

The two general strategies for apologizing proposed by Olshtain, use of IFIDS and expression of responsibility, are present in the Italian data with a frequency comparable to the CCSARP data for Hebrew, Australian English and Canadian French. Among the situation specific strategies, explanation, offer of repair and promise of forebearance, the third has never been used in Italian, as well as in the other three languages, but the first and the second (explanation and offer of repair) have been used by Italian speakers more often than by Hebrew speakers in the same four situations:

as well as by both Hebrew, Canadian French and Australian English speakers in all of the seven situations of the CCSARP questionnaire (Olshtain 1989):

This finding suggests that there might be greater cross-linguistic or cross-cultural differences in the choice of strategies for apologizing than is claimed by Olshtain (1989). Moreover, it should be noted that in this first classification of our data we have counted such forms as "Mi dispiace" or "Spiacente" as IFIDs, just because they are the literal translation of the English IFIDs "I'm sorry" or "Sorry". This choice is not justified; there are reasons to doubt that "Mi dispiace" functions in Italian as an IFID for apologies irrespective of the linguistic context in which it is embedded. If "Mi dispiace" were not counted as an IFID, the frequency of the IFID strategy in Italian would turn out to be by far lower than in the languages investigated by Olshtain (only 18% in situation 1 where the speaker is neatly superior in status; 35% in situation 2, 36% in situation 3, and 51% in situation 4). The fact that the student who has forgotten the professor's book uses less IFIDs (both in the broad and in the restricted sense) than the waiter who has brought the wrong dish cannot be explained by reference to power relations, but probably depends on the fact that people conceive of the waiter as basically reluctant to use the second general strategy for apologizing, that is, expression of responsibility (as already suggested by Olshtain, he is supposed to be reluctant to admit of incompetence in the performance of his role), so that the only main strategy which is open to him is the use of an IFID..

We have found it difficult to draw the distinction between expression of responsibility and use of explanations. In fact, expressions of responsibility often present the speaker's responsibility as partial or attenuated (e.g. as a responsibility for unintentionally forgetting the book at home instead of not returning it intentionally) and act therefore as excuses, just as those explanations which cite extenuating circumstances (being too busy, being absent-minded, having problems in the family, and so forth).

Our second classification of the data originated from the consideration that not all of the responses we had collected were clear cut apologies or scuse, and some had no remedial character at all. We have therefore attempted to distinguish scuse, remedies, justifications and non-remedial speech acts. Our attempt might recall the idea, put forward by Kerbrat Orecchioni (1994), of a continuum ranging from speech acts which are incontestably apologies to speech acts which are incontestably non-apologies through various forms of implicit apology. But we do not want to take it for granted that remedies and justifications are scuse and differ from scuse only in their degree of explicitness. We hypothesize that they are distinct, although functionally related, kinds of utterances.

In order to decide which utterances had to be classed as scuse, we have chosen the following complex criterion: they had either to contain an IFID ("Mi scuso", "Mi scusi", "Chiedo scusa"), to be an excuse (scusa) (namely, an extenuation of responsibility, presupposing that the act at issue is to be blamed), or to encode at least two strategies among expression of regret, offer of repair, request of forgiveness. We have counted as IFIDs only expressions containing the verbs scusare or scusarsi or the noun scusa; the common formula Mi dispiace (I'm sorry) has been counted as encoding expression of regret, but not as constituting a real IFID, since we felt that the mere utterance of Mi dispiace is not unambiguously a scusa, while the mere utterance of, say, Scusami or Chiedo scusa is unambiguously marked as such. We have granted a privileged role to the extenuation of responsibility because of the connection between apologies and excuses involved in the meaning of scusarsi. In practice, the criterion admitted as scuse only utterances involving, at least implicitly, the admission that the offending act was not the right thing to do. There were also utterances which did not meet our criterion for scuse, but had nevertheless a more or less neatly remedial character, for example because they encoded one of the strategies for scusarsi, or because they displayed interest for or solidarity with the offended person. We have dubbed these utterances "remedies", borrowing the word from Goffman (1971), but giving it a more restricted sense, since for Goffman "remedies" include justificatory accounts too. We have classed separately as justifications those utterances which aimed at showing that after all, the supposedly offending act was the right thing to do. Justifications are remedial acts in Goffman's broad sense, since if they are accepted, the offensive act ceases to be such. Finally, we have dubbed "non-remedies" those utterances which did not contain remedial moves, at least with reference to the current interactional debt. This second classification gave the following results.

These data confirm that a significant percentage of the responses to the discourse completion test are not full-fledged scuse or even lack a neatly remedial character, cast some doubts on the possibility of empirically extracting the strategies for scusarsi (and a fortiori for apologizing) from the responses to a discourse completion test, and display an interesting correlation with the relative power of speaker and addressee. Where the speaker is superior to the addressee or the two are on a par, non-remedial responses are given more often than when the speaker is inferior to the addressee. A waiter, for example, is clearly not in a position to afford not remedying an offense, and thus the lowest rate of non-remedial responses is (probably not by chance) the one relative to situation 4. Conversely, when the speaker is inferior to the addressee there is a higher rate of scuse, namely, there is a tendency to pay off the interactional debt in the most recognizable and therefore most efficient way.

However, there is something which this kind of data cannot tell: namely, to what extent the linguistic forms and contents we had singled out as characteristic of scuse, remedies, and justifications determine the hearer's perception of the speaker as performing, or not performing, the act of scusarsi? In order to answer this question, we have elaborated another questionnaire. We have selected 9 responses to the discourse completion test for each situation (3 scuse, 2 remedies, 2 justifications and 2 non-remedies) and have asked a sample of 50 Italian native speakers who had not answered the first questionnaire whether the speaker of each utterance, relative to the situation proposed by the discourse completion test, had apologized (si è scusato) or not. In order to collect some additional information on the relationship between scuse and politeness, we have also asked whether the speaker of each utterance had shown him or herself polite, merely correct, or impolite.

The results of the second questionnaire proved very interesting. No clear cut boundary was found between those utterances which can and those which cannot count as scuse. The rate of responses which were recognized as apologies was higher for the group of scuse and progressively lower for the other three groups.

The utterances we had classed as scuse were recognized as such in the 89,5% of the cases; the rates of recognition as scuse decreased in the cases of remedies (49,2%), justifications (19,5%) and non-remedies (3,7%). This shows that the conception of a scusa is not clear cut; utterances meeting our theoretically posited criteria failed to be recognized as scuse in a significant percentage (10,5%), while a lot of utterances not meeting the criteria were recognized as scuse at least by some subjects.² That is, not only the addressee may not grant the status of a scusa to an utterance which is presented as a suitable remedy by the speaker (as already noted by Kerbrat Orecchioni 1994: 172) but also the addressee may grant the status of a scusa to an utterance which the speaker has not clearly formulated as such. Even with certain utterances with hardly any remedial character (at least with respect to the interactional debt under consideration) it seems to be open to the receiver to count them as scuse and therefore, we surmise, the speaker has some chance to pay off his or her interactional debt by uttering them. A speaker who is really willing to extinguish his or her debt is likely to choose an utterance which is most likely to be recognized as a scusa, but some collaboration on the part of the audience is required anyway.

These findings, in the light of a dialogic or interactional view of the performance of speech acts, suggest that an utterance's counting as a scusa is not only a matter of its linguistic form and content, but also of the hearer's willingness to take it as the payment for an interactional debt. There is a fairly wide range of cases in which the fact that an utterance counts as a scusa seems to depend on an implicit negotiation between speaker and hearer. The speaker offers what he or she thinks he or she can afford (compatibly with his or her own face needs) and the hearer has to decide, as it were, if what is offered is the right kind of thing and is enough from the quantitative point of view). It is sensible to hypothesize that the closer an utterance is to the ideal model of a scusa, the easier it is that the addressee takes it as such. In contrast, only favourably disposed speakers in favourable circumstances will acknowledge as scuse utterances not presented as such clearly enough, or containing justifications of the speaker's behaviour instead of excuses, or even not remedial at all.

Our data do not give us any direct means for studying negotiation processes directly. Nevertheless, we can attempt to better understand the interactional dynamics of scuse by investigating either the circumstances in which a mere remedy, or a justification, or a non-remedy may count as scuse (as well as the circumstances in which a scusa may not be taken as such); or by investigating those linguistic features of the utterances in the questionnaire, which might facilitate or hinder their acknowledgement as scuse.

Some differences in the rates of recognition as scuse obtained by the utterances in our questionnaire seem to be correlated with the situation in which each utterance is imagined to be produced. For example, it is probably because of the expectations we have about the role of a teacher - he or she ought to care for his or her students - that the response "Mi dispiace ma la ho dimenticata a casa!" (I'm sorry but I forgot it at home!) in situation 1 gets one of the lowest ratings among scuse (although technically such): those who forget something do not care about it, and although forgetting is unintentional, it is nevertheless a violation of the teacher's obligations towards the student. Thus only 72% of the subjects recognize this utterance as a scusa or, to put it differently, up to 28% of the subjects would not say that the teacher has issued a scusa. Or it might be because of the respect we have for the waiter's professional role (and, as noted by Olsthain 1989, for his interests as a dependent worker) that his face-saving justifications in situation 4 are taken as scuse (that is, as sufficient for paying the interactional debt) at a much higher rate than in the other three situations (31% instead of 22%,23% and 2% respectively). In contrast, if the driver who has hit someone else's car while parking attempts to justify his or her behaviour (which presupposes that what he or she has done is allright), this is in general not perceived as a scusa. We have not tackled these problems in detail, however. We believe that what is most necessary in order to evaluate the import of our findings correctly, is a closer consideration of those linguistic features of the utterances in the questionnaire, which might facilitate or hinder their acknowledgement as scuse.

The utterances we had classed as scuse are not homogeneous, and the group of "remedies" is heterogeneous too. The rates of recognition of each kind of utterances as scuse, reported above, are just an average of the different individual rates obtained by the utterances classed in each group. This suggests that further distinctions should be introduced and that the rates of recognition of utterances displaying similar linguistic features should be compared and contrasted with those of other utterances classed in the same group. In our closer examination of the linguistic features of the utterances in the questionnaire, we have addressed two specific groups of questions..

1) One linguistic factor which certainly influences the recognition of an utterance as a scusa is the presence of an IFID. But which expressions really play the role of IFIDs in our data? Is our hypothesis that "Mi dispiace" in Italian is not a full-fledged IFID confirmed by our findings or not?

2) Another factor which might influence the recognition of an utterance as a scusa or interact with it is the overall polite or impolite character of the utterance under consideration. Are scuse inherently polite? Is an impolite utterance less likely to be recognized as a scusa? Is a polite utterance which is not a standard scusa more likely to be recognized as such than a less polite one?

We had considered as IFIDs only formulas such as Mi scusi ("Excuse me"), Mi scuso or Chiedo scusa ("I apologize"). That these are conventional IFIDs in Italian is confirmed by the data, since utterances containing them have been recognized as scuse in the 98,4% of the cases. It is reasonable to think that the use of an explicit speech act marker reduces the room for negotiation or even creates some obligation for the audience to recognize the speech act. Our choice of not considering as an IFID Mi dispiace ("I'm sorry") has turned out to be more problematic. The results obtained by the group of "remedies" display a clear split between those utterances which contain Mi dispiace and those which encode various remedial strategies, expression of regret included, without containing this formula. Some of the remedies containing Mi dispiace have obtained rates of recognition as scuse comparable and even higher than those obtained by the scuse lacking an IFID. It seems clear that not only the strategy of expressing regret has a remedial value (sustaining the alignment of the "good" part of the speaker with the victim of the offending act), but also the choice of Mi dispiace (instead of more informal expressions, such as the adverb purtroppo ("unfortunately")) plays some role in facilitating the recognition of the utterance containing it as a scusa. The remedy which has obtained the highest rate (from situation 2: "Mi dispiace ma non ho terminato la mia ricerca. Non le spiace se glielo restituisco domani?", I'm sorry, but I haven't finished my research yet. Do you mind if I bring it back to you tomorrow?) has been recognized as a scusa in the 86% of the cases. Is this an indication that Mi dispiace should be counted as an IFID and that at least some of the utterances we had classed as "remedies" should have been classed as scuse?

A more careful look at the data shows that Mi dispiace is not a marker of scusa independently of the linguistic context in which it occurs, or at least, it is not so in the same degree as those expressions we have considered as IFIDs proper. Among the utterances classed as scuse, in our questionnaire there were 5 containing Mi dispiace. They were classed as scuse because of their overall content. It is interesting to compare the rates of recognition as scuse of these utterances with those of the 5 remedies containing Mi dispiace. Scuse with Mi dispiace were recognized as such in the 87% of the cases, while remedies with Mi dispiace were recognized as scuse in the 68,4% of the cases. It seems therefore that the effectiveness of Mi dispiace in facilitating the recognition of an utterance as a scusa varies depending on the linguistic contexts in which it is embedded. It should also be noted that scuse with Mi dispiace are far from being recognized as such at a rate comparable to scuse with Mi scuso or Mi scusi (98,4%). Although the explicit expression of regret is a strategy which contributes to characterize an utterance as a scusa or, anyway, facilitates its being acknowledged as such, it still calls for the audience's cooperation, which the audience will grant or not depending on their evaluation of the speaker's overall behaviour.

Reasons for the low rate of recognition obtained by some scuse or remedies and for the relatively high rates obtained by some remedies or justifications may be found in those features of the utterances in the questionnaire, which make them polite or impolite irrespective of how close they are to a standard scusa. In fact, although scusarsi is a speech act endowed with a precise negative politeness function, it does not seem that all scuse are polite just because they are scuse, and conversely, in the situations envisaged by our questionnaire, one can be polite even without scusarsi. The percentage of utterances found impolite although recognized as scuse by the same subject is small but not irrelevant (8,4%) and the percentage of utterances judged polite although not scuse is remarkable (20,6%). In the evaluation of the utterances as polite or impolite, a role is played both by the use of multiple remedial strategies (as already noted by Holmes 1990), and by the display of positive politeness (concern for the addressee's well-being, needs, interests, feelings, and so forth).

In order to see whether impolite features hinder the recognition of a remedial utterance as a scusa, we have taken into consideration those among our scuse and remedies which have been found impolite by a relatively high number of subjects. The most striking example is the scusa relative to situation 1 we have already quoted above: Mi dispiace ma la ho dimenticata a casa! (I'm sorry but I forgot it at home!), which is recognized as a scusa with the relatively low rate of 72%, while is deemed impolite by the 54% of the subjects. A scusa relative to situation 2 Mi dispiace, l'ho dimenticato a casa, glielo riporto domani. ("I'm sorry, I forgot it at home, I'll bring it back to you tomorrow") is deemed impolite by the 18% of the subjects and reaches only a rate of 84% as to its recognition as a scusa (the average rate for the sub-group to which it belongs - scuse without IFID but with Mi dispiace - is 87,6%). In both cases the excuse offered invokes forgetfulness, which may be interpreted as involving lack of interest for the addressee and therefore as a violation of positive politeness. In the latter case, however, this violation is in part compensated by the addition of one more strategy for apologizing, offer of repair (the promise to bring the book back on the following day).

As for remedies, a striking case is (in situation 1) No, mi dispiace ma devi tornare domani.("No, I'm sorry but you'll have to come back tomorrow"): only 46% of recognition as a scusa (the average rate for remedies containing Mi dispiace is 68,4%) and 46% of assignments of impoliteness. It is clear here that the impolite character of the imposition to come back tomorrow, heavily expressed by the indicative devi (must) (a violation of negative politeness) hinders the recognition of the utterance as a scusa. Another interesting case is (in situation 2) the utterance Hem...ecco...veramente non l'ho portato, ma se le serve, farò in modo di farglielo avere al più presto...domani...dopodomani (Ehm...well...I haven't brought it, actually, but if you need it I'll make sure you get it as soon as possible...tomorrow...the day after), considered as a scusa only by the 10% of the subjects (although it encodes offer of repair and comes near to expressing regret) and deemed impolite by the 34%. Here the assignment of impoliteness may depend on the fact that the utterance expresses embarassment rather than regret (and thus does not observe positive politeness), or on the fact that the offer of repair is conditional on the addressee's declared need of the book (and thus an imposition is made on the addressee to declare he or she needs the book in order to get it back soon, which violates negative politeness).

As for justifications, an impolite one such as (in situation 3) Guardi che è lei che si è spostato mentre stavo posteggiando! (Listen, you were the one who moved while I was parking!) (assessed as impolite by the 84% of the subjects) has hardly any hope to be accepted as a scusa (2%).

In order to see whether polite features can facilitate the recognition of an utterance as a scusa, we have examined those among our scuse, remedies and justifications which have been found polite by a relatively high number of subjects. The scusa Mi dispiace, non l’avevo vista, ma non si preoccupi, la mia assicurazione risarcirà l’eventuale danno (I’m sorry, I didn’t see you, but don’t worry, my insurance will pay any damage that may have been done), deemed polite by the 64% of the subjects, is recognized as a scusa at the relatively high rate of 96%, notwithstanding the fact that it does not contain an IFID. The remedy (in situation 2) Mi dispiace ma non ho terminato la mia ricerca. Non le spiace se glielo restituisco domani? (I'm sorry, but I haven't finished my research yet. Do you mind if I bring it back to you tomorrow?), deemed polite by the 56% of the subjects, is recognized as a scusa at the relatively high rate of 86%. Both the use of Mi dispiace and the concern for the feelings of the addressee expressed by the request for a delay (Non le spiace in Italian is not so idiomatic as do you mind in English and retains the value of an expression of personal interest) characterize the utterance as encoding positive politeness and it is quite reasonable to suppose that this facilitates the acceptance of the remedy as a payment for the interactional debt. Similar considerations can be made with reference to a remedy in situation 3 Mi dispiace, ma non si preoccupi. Se ha dei danni le dò i dati della mia assicurazione. (I'm sorry, but don't worry about it. If there's any damage I'll give you my insurance policy number), deemed polite by the 50% of the subjects and recognized as a scusa by the 84%. A slightly different case is a remedy (not containing Mi dispiace) in situation 4, which has got a rather good rate of recognition as a scusa (42%; average rate for the sub-group of remedies not containing Mi dispiace: 20,7%) although it only encodes offer of repair: Non dica altro: il dolce lo offre la casa (Say no more: the dessert is on the house). This utterance, clearly expressing positive politeness, has been deemed polite by the 56% of the subjects, and this evaluation seems not to be foreign to its rate of recognition as a scusa.

As to justifications, two of them have been deemed polite by the 66% of the subjects, and both have obtained a rate of 36% of recognition as scuse, the highest for justifications in our data. They are: No, ma è un buon segno; l'ho trovata molto ricercata e approfondita, mi ci vuole più tempo per assegnarle un ottimo voto (No, but that's a good sign; I found it very well researched and gone into, I need more time to give you a really good mark) in situation 1 and Certamente ora gliela porto, ma volevo farle assaggiare questa frutta fresca (Of course, I'm going to bring it to you now, but I wanted you to taste this fresh fruit) in situation 4. Both utterances avoid admitting any fault or mistake on the part of the speaker (while scuse at least implicitly do so), but display a lot of concern for the addressee and therefore positive politeness.

We can therefore conclude that the display of positive politeness facilitates the acceptance of an utterance not having the characters of a scusa or even of a remedy in our sense, as a payment for an interactional debt. In contrast, various kinds of violations of positive or negative politeness may make more difficult for utterances having the character of remedies or even of scuse, to be recognized as such. My hypothesis it that the polite or impolite features of an utterance have an effect not directly on the extent to which the utterance itself meets the requirements for being a scusa, but on the hearer's positive or negative disposition towards the speaker.

We shall now try to give some answers to the questions we have raised in §2 above.

(1) What are the strategies for apologizing explicitly? is it legitimate to expect that English illocutionary force indicating devices translate into other languages literally?

In Italian, the most explicit way for scusarsi (the only one which seems to force the hearer's recognition) exploits the use of the words scusare, scusarsi, scusa in various syntactic connections (the most typical of which are the performatives Chiedo scusa and Mi scuso and the imperatives Scusa(mi) or (Mi) scusi). Utterances with IFIDs containing these words have been recognized as scuse in a percentage of 98,4%. It is interesting to notice that most uses of these words encode at the same time request for forgiveness: the performative Mi scuso is by far less common than the request for forgiveness Scusami, and the performative Chiedo scusa in turn literally means "I beg to be excused". Expression of regret, although a common and effective strategy for scusarsi, does not generate fully conventionalized IFIDs in Italian. It might be true that "Sorry" is a conventional IFID for apologizing in English (the data collected by Owen 1983 and by Holmes 1990 are suggestive of this, since they show that a high rate of utterances understood as apologies contain "Sorry" in some form: 79% in Holmes's sample and 86,7% in Owen's). But if "Sorry" is an IFID in English, our data suggest that the literal translation of what in one language may count as an IFID does not necessarily itself act as a conventional IFID. The strategies for apologizing may be the same across different languages and cultures, but the extent to which formulas expressing them are conventionalized into IFIDs, and therefore are able to force the recognition of the apology as such, may vary.

2) Is assumption or expression of responsibility really a central move for apologizing? How is it related to the remedial moves which Goffman called accounts, and in particular to excuses? Have all kinds of accounts the same relation to apologies, or a sharp distinction should be drawn (as suggested by Austin 1960) between excuses and justifications?

In our second questionnaire Non la ho ancora guardata. Arrivederci. (I haven't looked at it yet. Goodbye) was not taken as a scusa by anyone of the subjects. Undeniably, in this utterance the professor assumes responsibility for not having corrected the student's paper. But this assumption of responsibility, not being accompanied by any hint at self-disapproval, does not amount to an excuse (which would presuppose disapproval of the committed act), nor therefore to a component of an act of scusarsi. Examples of this kind have lead us to think that mere expression of responsibility is not a central move for apologizing, and perhaps no move for apologizing at all. It may become a move for apologizing only when something else in the utterance suggests that the speaker disapproves of his or her own act, so that admitting to have done it implicitly amounts to humiliating oneself, thus acquiring a remedial character.

As to Goffman's "accounts", we believe that our data support a sharp distinction between excuses, closely connected to apologies, and justifications. Utterances containing justifications not accompanied by further remedial moves got an average rate of recognition as scuse of only 19,5% (with a minimum of 2% in situation 3 and a maximum of 31% in situation 4). It should be noted, however, that justifying one's behaviour is not incompatible with attempting to remedy an offense (while it is incompatible with that element of self-disapproval which is central to excuses); as Holmes (1990) has remarked, a justification may accompany an apology or, as we would put it, be included in a remedial turn:. We suppose that the inclusion of a justification in a remedy has the function of mitigating the threat to the speaker's positive face (it is aimed at seeking approval). But the justification is not understood as a compensation for the offended person unless it is accompanied by other moves oriented in this direction (such as, at least, expression of concern for the addressee), or unless the hearer is willing to lower the cost of the extinction of the interactional debt.

3) What makes it so that apologies can restore the interactional equilibrium?

Both scuse and other remedial moves can restore the interactional equilibrium by being taken by the addressee as the payment for an interactional debt. We think that it is possible to apply to the act of scusarsi the definition given by Goffman to apologies in terms of the splitting of the self into one part who is guilty of an offense and another part who aligns him or herself with the victim of the offense. We believe that, in apologies as well as in our scuse, it is just this split which counts as a symbolic compensation for the addressee, namely, as the payment of the speaker's interactional debt. In particular, the threat to the speaker's positive face (relative to the part of him or her who has committed the offending act) balances the threat to the hearer's negative face that characterizes the offending act. However, it is only one of the two parts into which the speaker's self divides who is threatened in his or her positive face; the other part retains his or her claim to positive face and aligns him or herself with, or gives support to, the hearer's positive face. The various strategies for apologizing or for scusarsi encode partial aspects of this complex operation and it is not by chance that often more than one strategy is used. The role of excuses in this picture includes: conveying self-disapproval (by presupposition); making it easier for the offended person to accept the restoring of the equilibrium by extenuating the speaker's fault and therefore making his or her debt lighter; further dissociating the speaker's intentional behaviour from the offending act, thus facilitating solidarity between the "good" part of the speaker's self and the addressee. This explains the close connection between apologies and excuses, which comes to a conflation in the Italian speech act lexicon. In contrast, the fact that the speaker retains at least in part his or her claims to positive face explains the partial compatibility of scusarsi with justifications.

Notes

1. I would like to acknowledge the contribution of my student Iva Macaluso to the collection of data and to the first phase of their discussion.

2. It should be noted that one utterance turned out to be misclassified. It is the utterance relative to situation 4: Purtroppo la torta della casa è terminata e ho pensato che la frutta fresca di stagione potesse degnamente sostituirla! ("Unfortunately, the cake is finished and I thought that fresh fruit of the season might be a suitable substitute!"), which we had classed as a scusa because it encodes two strategies for apologizing (expression of regret and offer of repair), but has obtained a comparatively very low rate of recognition (62%).This result, which cannot be fully explained by other factors such as the remarkable but not really high rate at which the utterance has been considered impolite (18%), can be explained by an ambiguity of the utterance itself: in fact, ho pensato che la frutta fresca di stagione potesse degnamente sostituirla may be taken as a justification rather than an offer of repair and under this interpretation the utterance should have been classed as a mere remedy. The misclassification of this utterance as a scusa unduly lowers the total rate of recognition of scuse as such. Its inclusion in the group of remedies would not have remarkable repercussions on their total rate of recognition as scuse.

Austin, John L. (1960). Philosophical Papers. Oxford: Oxford University Press. III ed. 1979.

Austin, John L. (1962). How to Do Things with Words. Oxford: Oxford University Press. II ed. 1975.

Blum-Kulka, Shoshana, Juliane House and Gabriele Kasper (eds.) (1989). Cross-cultural pragmatics: requests and apologies. Norwood (N.J.): Ablex.

Brown, Penelope and Stepher C. Levinson (1978).

Brown, Penelope and Stephen C. Levinson (1987). Politeness. Some universals in language usage. Cambridge: Cambridge University Press.

Fraser, Bruce (1981). "On apologizing". In: Florian Coulmas (ed.), Conversational routine. The Hague: Mouton, pp. 259-71.

Goffman, Erving (1971). Relations in public. New York: Basic Books.

Holmes, Janet (1990). "Apologies in New Zealand English". Language in Society 19, pp. 155-199.

Kerbrat Orecchioni, Catherine (1994). Les interactions verbales. Tome III. Paris: Colin, 1994

Olshtain, Elite (1989). "Apologies across languages".In: Blum-Kulka et al. (eds.) (1989), pp.155-174.

Olshtain, Elite and A.D. Cohen (1983). "Apology: a speech act set". In: Nessa Wolfson and E. Judd (eds.), Sociolinguistics and language acquisition. Rowley, MA: Newbury House, pp. 18-35.

Owen, Marion (1983) Apologies and remedial interchanges. Berlin: Mouton.