In recent years, an increasing number of speech acts have been investigated from a diachronic perspective. How were they realized at a particular point in the history of a language, and how did they change in the course of time? Such investigations immediately expose a range of important theoretical issues about the ontological status of speech acts that need to be addressed. To what extent can a speech act from the fifteenth century, an apology, let us say, be compared to a speech act in the twenty-first century? Does it make sense at all to use the same (Present-day English) term to refer to both of them? And does the comparison concern the linguistic elements with which these speech acts are realized or does it concern the functional profile of the speech act itself? In what situations did speakers use this speech act then, and how do these situations differ from the situations in which it is used today?

So far, our knowledge about trajectories of individual speech acts is still disjointed and fragmentary and only a small number of speech acts have been investigated from a diachronic perspective at all (see Section 3 below). However, the corpus-linguistic tools that are necessary to trace speech acts systematically in large corpora are currently being developed with increasing success. This makes diachronic investigations more feasible and allows us to tackle the necessary theoretical issues.

It is the aim of this paper to contribute to the ongoing discussion by pulling together a range of recent research efforts investigating one particular speech act, viz. apologies at various points in the history of the English language, and by suggesting an integrated theoretical framework that combines these limited scope efforts into a more comprehensive understanding of the long diachrony of apologies. It transpires that apologies have undergone, and are still in the process of undergoing, a process of attenuation. What used to be a weighty and sincere act of admitting guilt and expressing remorse in a religious context has, via a series of steps, been weakened into a token acknowledgement of a minor mishap, such as, for instance, accidentally bumping into somebody on a crowded platform of a railway station. The older type of apologies still exists but the new types of fleeting and sometimes non-serious uses of apologies have been added, and they significantly increase the overall frequency of apologies.

In the following, I will first discuss some of the theoretical issues that are involved in the description of a long speech act diachrony, in particular the issue of the integrity of a speech act in the course of time and the tertium comparationis, i.e. the question of how much has to stay the same across the time periods in order to warrant a comparison. After that I will introduce the illustrative exemplar for this study, the apology, and suggest five stages of its diachronic development. In a final discussion section, I will add histories of some other culturally salient speech acts – in particular promises and greetings – which provide corroborating evidence for the suggested diachronic stages of development even if their histories have not yet been investigated in the same detail and over the same diachronic distance as apologies. At this stage, however, the claims about the range of application of these findings have to be modest. Our knowledge of the diachronic development of individual speech acts rests on a small number of culturally salient speech acts. It is still unclear how large the number of all possible speech acts might be for any given language (see Levinson 2015 for a discussion of the issue). But only some of them are salient enough for a speech community to develop terms, i.e. speech act verbs, to talk about them, and only a small number of these, so far, have attracted the attention of linguists and language historians.

An investigation of a speech act across time presumes that the speech act in question is – in some significant way – the “same” in its earlier and its later instantiations. To what extent can we still assume that we are talking about the same speech act in spite of the changes? What is the tertium comparationis, i.e. the element that stays the same for all instances that are compared with each other? What is the tertium comparationis of an apology in the fourteenth century that is compared with an apology today? Can we rely on the descriptive labels in a language, i.e. the metacommunicative expressions? It turns out that the noun “apology” is not attested in the English language before the sixteenth century (OED Second edition, apology, n). Does this mean that the speech act of an apology did not exist previous to the existence of the term? And how do we account for semantic changes in the meaning of the metacommunicative label? In order to tackle these questions, it is first necessary to disentangle the different aspects of a speech act; its functional profile, the metacommunicative expressions and the expressions and phrases that are employed to perform it.

The functional profile corresponds largely to what Searle (1969: 30) called the illocutionary force of a speech act. Bertuccelli Papi (2000: 58) talks of the illocutionary potential. Searle captured the illocutionary forces of specific speech acts by a range of felicity conditions, i.e. those conditions that must obtain for us to agree that a certain speech act has been successfully performed. A promise, for instance, needs to talk about (i.e. predicate) something that the speaker wants to do at some point after the time of uttering the speech act and which he or she deems to be in the interest of the addressee. If this future action is not in the addressee’s best interest, we would not want to call this speech act a promise, but perhaps a warning or a mere assertion of a plan. Searle used philosophical tools to come up with such felicity conditions which create profiles of specific speech acts. Such profiles are not always very precise and they sometimes conflict with our everyday understanding of these speech acts. Questions, for instance, are a case in point. According to the felicity conditions proposed by Searle, the speaker asking the question does not know the answer (in the case of real questions) or wants to know whether the hearer knows the answer (in the case of exam questions). In everyday usage there seems to be a wide variety of additional questions with varying degrees of the speaker’s ability to predict the answer, not only in the case of teachers asking exam questions but also in journalistic interviews, police or courtroom interrogations and so on. And the felicity conditions for advice could also apply to recommendations or suggestions.

In more recent work on speech acts, researchers use different terms to talk about their functional profiles but generally they still define a range of conditions that must obtain to talk about a specific speech act in order to delimit the object of investigation (see Section 4 below for a discussion of the defining elements of apologies). But in some cases, researchers also adopt a more discursive approach, i.e. they do not provide a predetermined set of conditions but they investigate how the members of a speech community themselves talk about a particular speech act and how they negotiate it, that is they investigate the metacommunicative expressions that are used to talk about speech acts, such as, in the case of apologies, apology, apologize or apologetic. Such metacommunicative expressions, in particular in their form as verbs, have often been used to establish elaborate classifications of the speech acts themselves (e.g. Verschueren’s 1985 or Traugott’s 1991 early work in this respect; see also Taavitsainen & Jucker 2007).

The third aspect to consider at this point is the range of linguistic forms and expressions that can be used to carry out a particular speech act. Expressions that are regularly used come to be recognized as conventional means of carrying out a particular speech act. They become conventionally associated with it and, therefore, serve as what has come to be known as illocutionary force indicating devices, or IFIDs for short (Levinson 1983: 238). Speech acts differ considerably in the extent to which they make use of conventionalized expressions, presumably in relation to the routine nature of the speech act itself. In Present-day English, greetings and farewells, for instance, appear to be heavily conventionalized, apologies and compliments somewhat less and assertions not at all.

Figure 1 provides a schematic representation of how the three elements are related. A metacommunicative expression is used to talk about a speech act with a specific functional profile or illocutionary potential, and IFIDs are one possible way of performing such a speech act. In some cases, IFIDs also function as metacommunicative expression, as for instance the verb apologize, which can be used both to perform and to refer to this particular speech act (“I apologize” versus “No need to apologize”).

In the course of time, typically all three elements are subject to change. The functional profile may shift. Some felicity conditions may no longer be equally important, while new ones may be added, and – as a result – a particular speech act may increasingly be used in situations that differ from earlier contexts of use. These changes may be reflected in concomitant shifts in the meaning of the metacommunicative expressions or in the adoption of new metacommunicative expressions to refer to this now modified speech act. And parallel to these developments the linguistic resources available to the speakers to perform this speech act may change as well.

Such a perspective presumes a prototype approach, that is to say speech acts are considered to be fuzzy entities with definitions that include not only focal instantiations but also instantiations that are less focal and more peripheral to the definition (see Jucker & Taavitsainen 2000). In a long diachrony the assumption cannot be that a speech act in a previous century is exactly the same as it is today. It is more likely that the functional profile has gradually shifted. Some of the previously defining elements perhaps have become less important, have fallen away or have been replaced, and others have perhaps been added. Only some defining elements are affected at a time while others stay in place and make the speech act comparable across short diachronies even if the development over many centuries may lead to a speech act that is only vaguely related to its long-term ancestor. The change in the semantics of the metacommunicative expressions and the adoption of new expressions reflects at least to some extent these gradual changes of the functional profile. It is such a dynamic view of the nature of speech acts that allows a long diachronic perspective as will be illustrated with the case of apologies below.

Work on the history of specific speech acts so far has focused on a relatively small number of speech acts which have in common that they are culturally salient and have metacommunicative expressions that are used by members of the speech community to talk about them, and in most cases they show a relatively high level of routinization, i.e. they can be carried out by a small range of conventionalized expressions. One obvious reason for this focus is that culturally salient speech acts are perhaps more likely to attract the attention of researchers. Moreover, metacommunicative expressions and conventionalized IFIDs help to locate these speech acts in historical data.

Directives, for instance, have received a fair amount of attention, especially requests. See, for instance, the work by Busse (2002; 2008), Kohnen (2002; 2004a; b; 2008a; b), Culpeper & Archer (2008) and Moessner (2010). Advice giving also falls into the larger group of directives (e.g. Fitzmaurice 2002: Chapter 4; Milfull 2004). Commissives in the form of promises have also received some attention, e.g. by Arnovick (1994a; b; 1999), Alonso Almeida & Cabrera-Abreu (2002), Pakkala-Weckström (2002; 2008), Del Lungo Camiciotti (2008) and Valkonen (2008).

Most work, however, seems to have been focused on expressives, such as greetings and farewells (Arnovick 1996; 1999; Grzega 2005; 2008; Jucker 2011a; b; 2017), apologies (Jucker & Taavitsainen 2008; Kohnen 2017a; b; Williams 2018; Jucker 2018b), compliments (Taavitsainen & Jucker 2008), thanking (Jacobsson 2002), and in particular swearing and insulting (Hughes 1991; Arnovick 1995; Jucker 2000; Jucker & Taavitsainen 2000) and cursing (Danet & Bogoch 1992; Arnovick 1999).1

But there is also some historical work on less culturally salient speech acts which lack an everyday metacommunicative expression, for instance the opening moves in Middle English courtly literature (Honegger 2000), or work on speech acts that do not fit easily into a Searlian speech act classification system, for instance refusals, which by definition are not initiating but reactive speech acts (see Martínez-Insua 2010).

Most of this work tends to take the functional profile of the speech act as a starting point and investigates the ways in which it is performed in specific historical contexts and, in many cases, how the realizations change in the course of time. In some of this work, specific diachronic processes have been identified that are of larger theoretical interest, in particular the processes of pragmaticalization and subjectification. They will briefly be introduced in the following.

The term “pragmaticalization” describes the process in which lexical material undergoes a diachronic process of “becoming pragmatic” or “becoming more pragmatic” (see Claridge & Arnovick 2010), that is to say that a certain linguistic element loses at least some of its semantic transparency and increasingly needs a specific context of use for its interpretation. The farewell greeting goodbye may serve as an example (see Arnovick 1999; Claridge & Arnovick 2010). In Early Modern English, the phrase God be with you was regularly used in the double function of a blessing and a courteous closing of a conversation as in the following examples taken from the Early English Books Online (EEBO).

In both cases, the double illocution is clear from the context. God be with you is both a sincere blessing and a formula closing the conversation to which the addressee often responds with a similar blessing (Claridge & Arnovick 2010: 174). In the late sixteenth century, the phrase starts to appear regularly in contracted forms. The OED provides a long list of such contracted forms, e.g. godb’w’y, godb’w’y’, godb’w’you, goodb’wy, godb’w’ye, goodbwi’t’ye. In conjunction with this contraction, the closing function of the phrase becomes dominant. Claridge & Arnovick (2010: 175) date this shift to the end of the seventeenth and the beginning of the eighteenth century. They cite extract (3) as an example in which the blessing function was no longer apparent.

In a second step the form good was substituted for God. According to the OED, this probably happened in analogy to similar forms as in good day or goodnight. In this reduced form, and even more so in the further reduction into bye, any trace of the original blessing disappeared. And today, the form is fully pragmaticalized. This interpretation is reinforced by the re-appearance of the full form God be with you as a sincere blessing, a function that can no longer be fulfilled by goodbye as in extracts (4) and (5).

Arnovick (1999: 116) argues that the term “pragmaticalization” might be inadequate to describe the diachronic processes involved because in this case the process did not start with lexical material that adopted pragmatic meanings but it started with pragmatic, or rather illocutionary, material.

In other words, the term pragmaticalization describes processes affecting the linguistic material used to perform a speech act, while the term discursization describes the process that affects the speech act itself, or rather its illocutionary potential. I prefer the term attenuation for this process, as I will show below, because it makes a stronger claim about the way in which the process affects the illocutionary potential of a speech act and because it makes a claim that this kind of diachronic development is not unique to apologies but has wider applications.

The term “subjectification” applies to the motivation underlying diachronic processes and describes the way in which meanings turn from objective descriptions to expressions of the speaker’s internal perspective or attitude. It has been proposed as a description of the motivations for grammaticalization (Traugott 1989; 1995; López-Couso 2010). It generally proposes that meanings shift from an objective description of the situation to meanings based in the textual and metalinguistic situation and finally to the speaker’s subjective believes and attitudes towards the proposition expressed. The classic example for this is the Old English phrase þa hwile þe ‘at the time that’, which diachronically changed into a temporal connective while meaning ‘at the same time as’ and finally to a concessive marker while meaning ‘although’.

But subjectification has also been argued to be the motivation underlying processes of pragmaticalization, for instance in the case of curses. According to Arnovick (1999: 77) the extant material suggests that cursing in Old English was invested with spiritual power. Curses were a serious matter. They “were not performed lightly or indiscriminately”. Danet & Bogoch (1992) study a set of such curses in Old English legal documents. The curses were intended to protect the document from interference and clearly were intended to have performative power, as for instance in example (6).

These curses are formulated as conditionals, consisting of a “whoever” clause specifying what the writer wants to prevent and a second clause that spells out the punishment that will fall on anybody who violates this condition and tampers with the document or interferes with its execution. If the condition in the “whoever” clause is fulfilled, the curse is activated. It has what Arnovick calls deontic power. For the period of Middle English, Arnovick (1999: 85–87) identifies both religious curses that maintain some of this deontic power including a desire for God’s wrath on some wrongdoer and more expressive curses that indicate the speaker’s exasperation. The religious impact of curses is increasingly diluted. In Present-day English, cursing increasingly turns into cussing, i.e. into a more personal and more subjective expression of anger or annoyance. As Arnovick (1999: 90) puts it, “when someone in fin de siecle Vancouver, British Columbia yells, ‘Damn you,’ he or she does not usually wish actual damnation on the hearer.”

In this paper, I focus on the process that affects the functional profile of a speech act rather than the metacommunicative expressions or the IFIDs. The relatively detailed history of apologies (understood in a dynamic and prototypical sense) shows a process of attenuation, that is to say their force becomes increasingly weaker, and at the same time the linguistic resources used to perform the speech act undergo a process of reduction and conventionalization. In its early form the speech act is semantically explicit and spells out its illocution while later it is reduced to a conventionalized expression that requires increased pragmatic processing for its interpretation. Figure 2 gives a schematic representation of the stages that I will illustrate for the case of apologies in Section 4. In Section 5, I will provide supporting evidence from a range of other speech acts which also move along at least part of this scale even if not all of it.

The scale in Figure 2 accounts both for the diachronic weakening of the illocutionary force and for the increasingly conventionalized expressions used to perform the speech act. It is, of course, not meant to suggest that all speech acts originate in a speech act to God. This is clearly implausible. But in the following I want to show that apologies, as one culturally salient speech act can be argued to have gone through all five stages of this development. Other speech acts, e.g. salutations, can be argued to have early examples that correspond to a stage 2, and so on. But the claim is that several culturally salient speech acts can be shown to move from left to right somewhere along this scale, and they do not move in the opposite direction. Figure 2 also does not claim that once a certain stage in the development is reached, earlier stages can no longer be accessed. The different stages can clearly co-occur synchronically. But the claim is that the stages to the right of the scale are later additions than the ones on the left.

In the case of apologies, a long diachrony is emerging from recent research on its history. Kohnen (2017a; b) investigated apologies in Old English, Williams (2018) extended the perspective from Old to Middle English, Jucker & Taavitsainen (2008) covered the Early Modern period (from 1500 to 1660) and Jucker (2018b) looked at the last two hundred years. For Present-day English we have a rich literature on apologies using a variety of frameworks and methodologies, for instance Blum-Kulka et al.’s (1989) cross-cultural work on apologies based on discourse completion tasks, Trosborg’s (1995) work based on role plays and role enactments, and Deutschmann’s (2003) work based on the spoken part of the British National Corpus. More recent work on apologies by Lutzky & Kehoe (2017a; b), Aijmer (2019) and Haugh & Chang (2019) might already be pointing towards the future of apologizing in English. Together these studies provide a fairly detailed long diachrony of one specific and culturally salient speech act, in spite of the data limitations for the early centuries and some missing periods (in particular the eighteenth century).

The functional profile (or illocutionary potential) of an apology has often been described in the relevant literature. The Oxford English Dictionary defines the noun apology as follows:

The relevant elements are an offender who takes responsibility for some action, someone who was offended by this action or who was perceived by the offender to have possibly been offended, the offence or perceived offence itself and the speaker’s acknowledgment of the offence coupled with an expression of regret. See Figure 3 (Deutschmann 2003: 44–47).

In a narrow interpretation of what an apology is, or for a prototypical apology, all four conditions must be met. If the speaker does not take responsibility for the offence, if he or she does not display any regret, or if the speech act is made not to the offended party but to somebody else, the speech act is no longer an apology, or – as I would prefer to see it – it may still be an apology albeit not a prototypical one. Apologies clearly shade off into neighboring speech acts within a multidimensional pragmatic space (cf. Jucker & Taavitsainen 2000 on the notion of pragmatic space). If the speaker does not take any responsibility for the event, the speech act may turn into a general expression of regret or a spill cry2 as in the examples (7) and (8).

In both cases, the context leaves it open whether the speaker takes responsibility for what has happened. It is, of course, possible that the addressees had a clearer idea than the researcher inspecting the transcript whether the speaker meant to take responsibility or not, but in both cases, it does not seem to matter very much, and in fact, as I will argue, this is how speech acts change their illocutionary potential. Some of their felicity conditions become fuzzy; they may no longer be part of a narrow definition of the original speech act. Up to a point this may either go unnoticed or it may draw the attention of language purists. It is not unusual today to hear somewhat disapproving comments about people who apologize for things they have not done.

In the following, I will show how apologies started out as a very serious speech act addressed to God, and then gradually, by slowly changing some of the felicity conditions, turned into highly pragmaticalized and conventionalized speech acts that in some cases seem to be no more than spill cries.

The above case study has illustrated how the proposed new theoretical framework of speech act attenuation can help to trace a long diachrony of apologies. In the course of time, the illocutionary potential has changed very considerably, and in fact the early examples from Old English in the form of acts of repentance for sins committed before God show only a very faint resemblance to the minimalized expressions of regret that often enough appear to be less than heart-felt and sincere but merely token acknowledgments of minor infringements or perhaps just spill cries. But they are linked through a succession of stages with relatively minor differences from one stage to the next. The analysis presumes a dynamic definition of the speech act of apology which is based on a prototypical set of conditions but does not assume that all these conditions need to be fulfilled at all times. On the contrary, it is the shift in individual conditions which leads to a gradual attenuation of the speech act and ultimately to a significantly modified profile.

Figure 3 listed the four aspects of a prototypical apology: the offender, the offended, the offence and the remedy. We can now see that the speaker always identifies him or herself as the offender, where “offender” has to be understood in a very broad sense. The other three aspects undergo considerable changes. In the first period, described in Section 4.1 above, the offender is a Christian sinner who takes responsibility for serious wrongdoings before God and expresses emotions of deep remorse. In the second period (Section 4.2), the offended is no longer God but a fellow human being. In the fully secularized third period (Section 4.3), the offender, who has offended against a fellow human being, implores the offended to pardon and forgive. In the fourth period (section 4.4), there is a shift in focus from asking for forgiveness to expressing feelings of regret. And in the last period (Section 4.5), especially if we think of the examples given in (29) to (36), the “offender” is now someone who playfully draws attention to a minor infraction, such as an unorthodox spelling, which is not really presented as an offence, and it is left unclear who, if anybody, might have been offended by this.

To date there does not appear to be any other speech act whose diachronic development has received enough scholarly attention to test the theory of speech act attenuation in a similarly comprehensive way. However, from the different accounts of promises in the history of the English language, we can deduce a related story. In Present-day English, promises seem to have a bad press, not only those made by politicians during an election campaign. They are often dismissed as insincere and devoid of any great commitment of the speaker to actually carry out what he or she promised. Extract (37) may be typical in the sentiment that it expresses about promises.

In Old English, however, the situation was very different. “Statements of intended future action have the force of vows, utterances not lightly made. Promises thus entail obligations which must be honored even if death is the route to their successful fulfilment.” (Arnovick 1999: 58). Pakkala-Weckström (2002; 2008) focuses on Middle English promises in the context of Chaucer’s work. She identifies binding promises that result in an irreversible obligation by the speaker even in cases in which the promise was made in jest. As a particularly telling example, she gives Dorigen’s promise in Chaucer’s Franklin’s Tale. While her beloved husband, Arveragus, is absent, Dorigen tries to get rid of an unwanted suitor, Aurelius, by promising him that she will love him if he can remove all the rocks on the coast of Brittany. Against her expectations, Aurelius, who secures the help of a magician, manages to let the rocks disappear. On his return her husband agrees that the distraught Dorigen must keep her promise because “trouthe is the hyeste thyng that man may kepe” (Franklin’s Tale V (F), 1479). It is only Aurelius himself who can release Dorigen from her promise once he realizes the true love between Dorigen and Arveragus.

This contrasts very considerably with today’s promises. Today it often seems necessary to reinforce the speaker’s intention to keep the promise by combining several formulations (see Arnovick 1999: 61–61). Extract (38) gives a relevant example from a recent American Soap Opera.

The words I will never do it again do not seem to be sufficient to convince the addressees of the sincerity of the speaker’s commitment. Thus, promises appear to undergo a very similar process of attenuation from a speech act with considerable weight to one that often is perceived as having very little weight and which needs to be reinforced appropriately if it is to be taken seriously.

Jucker (2017) has looked at greetings and farewells. They are largely defined by their occurrence at the boundaries of conversations, either at the opening or at the closing, and they are often claimed to have very little propositional content. In the Corpus of Historical American English (COHA), the most frequent expressions used as greetings at the beginning of the nineteenth century were good morning and how are you? The most frequent leave-taking expression was farewell. Today it is the expressions hi and hello as greetings and goodbye and bye bye as farewells. He uses how are you? and how do you do? to investigate the claim that these are just conventionalized and empty formulae without any propositional content and finds that both at the beginning and at the end of the period covered by COHA these forms could be used both as serious questions about the well-being of the addressee or as formulaic openings of a conversation that do not require any explicit answer that goes beyond responding to the greeting. However, in recent decades the form how do you do? has come to be restricted to relatively formal introductions between strangers and for both forms it is often left to the addressee to interpret it as formulaic or as a serious request for information. These forms, therefore, add further evidence for the unidirectionality of change from a situation in which the formulation of a speech act was regularly interpreted at face value, i.e. in this case as a real question, to a situation in which contextual information is required to pragmatically interpret the utterance which may still have its full semantic weight or may have lost it altogether in favor of a maximally attenuated speech act.

Our understanding of the long term diachronies of promises and greetings is not as detailed as it is for apologies, for which research published in the last few years has provided access to a more or less step by step development from Old English up to Present-day English. But the limited knowledge that we have appears to be supporting the analysis of a continuous diachronic attenuation. The speech acts lose their earlier weight, at least in some of their applications. What used to be strong commitments turned into increasingly weaker and less imposing expressions of the speaker’s attitude. In Section 3 above, I mentioned the cases of the farewell greeting and curses. In both cases, there is a clear religious, or at least spiritual, origin which again parallels the development of apologies. Goodbye developed out of the farewell blessing God be with you, and today’s cussing developed out of the Anglo-Saxon curses that in a very real sense were meant to protect legal documents by afflicting anybody who tampered with them. And even if the parallelisms are not identical developments in any strict sense, they are certainly sufficiently similar to create a strong curiosity to learn more about the long diachronies of other speech acts and to further explore the similarities and differences between them.

Birmingham Blog Corpus (BBC): https://wse1.webcorp.org.uk/

British National Corpus: https://corpus.byu.edu/bnc/

Corpus of American Soap Operas: https://corpus.byu.edu/soap/

Corpus of Contemporary American English: https://corpus.byu.edu/coca/

Corpus of Historical American English: https://corpus.byu.edu/coha/

Early English Books Online: https://corpus.byu.edu/eebo/

Global Web-Based English (GloWbE): https://corpus.byu.edu/glowbe/

Google Books Ngram Viewer: https://books.google.com/ngrams/