The optative is notoriously employed to encode modal meanings and functions in Ancient Greek, and there is general agreement that this mood is in decline in post-classical Greek and in the language of the New Testament (Horrocks 2010: 102; McKay 1993; Debrunner 1969: §§ 189–191; Turner 1963; Schwyzer 1950: 337 ff.). However, it is still unclear what the precise reasons for this decline and eventual demise are, and—more interestingly—why this mood survives at all to a certain extent. Despite the number of studies of post-classical (and, specifically, New Testament) syntax, there does not seem to be an up-to-date study of this topic that takes into account recent advances in research on typology and modality, along with the textual and pragmatic dimensions. This paper, which also offers a reanalysis of the entire relevant material from the New Testament (henceforth, NT), is a first step to fill this gap.

In what follows, we first offer an outline of the diachronic evolution of the optative from Homeric¹ to Late Greek (section 2). A review of the existing literature provides a useful background for an in-depth analysis of the NT corpus. Also, a somewhat comprehensive “history of the Greek optative” is still missing. Here, its functions are not only classified according to the traditional grammatical distinctions but also evaluated within a theoretical framework based on the recent literature on typology, modality and the realis-irrealis continuum, with a focus on the use of optatives as directives (section 3). In addition, we propose a new quantitative and qualitative analysis of all optative forms attested in the NT (section 4; cf. the Appendix for the dataset). We conclude (section 5) with some observations on the survival of the optative, discussing the role of Atticism, language contact, and the evidence from ancient grammarians. We argue that these factors contribute of an explanation of how the optative is preserved in fixed constructions which are associated to specific speech acts (pseudo-directives and pseudo-questions) and at the same time becomes somewhat detached from the inherited modal system.

In Homeric Greek (henceforth, HG), the optative is employed in a variety of functions, some of which continue into Classical Greek (henceforth, CG), while others are taken over by different forms. Here, we offer a synthetic account, mostly based on Willmott’s recent work.²

In Classical Greek, the optative still covers a wide range of functions, which we summarise here by taking examples, translations and terminology (in most cases) from the recent CGCG.

It is generally agreed that in κοινή Greek (henceforth, KG) the optative undergoes further decline, losing ground to the subjunctive, to the indicative, and to different periphrases.¹³ Before moving on to some considerations on the Biblical corpus, with reference to both the Septuaginta (henceforth, LXX) and the Greek of the New Testament (NTG), it is worth noting that the three main uses one finds in CG—cupitive and potential in main clauses, and oblique in subordinate clauses—decline at different rates. As Evans (1999: 489) puts it, “the historic sequence function is the first to be lost, while the volitive and potential uses last rather longer, the volitive proving most robust”.

Two reasons for the demise of the optative are generally adduced: the phonological identity with subjunctive forms in most of the paradigm (both /oi/ and /ε:/ eventually merge with /i/)²³ according to some, a purely syntactic change (the optative began to be replaced by other moods and constructions) according to others.²⁴ We do not think that one thesis necessarily excludes (or, vice-versa entails) the other, but on this occasion we focus, instead, on the reasons for the survival of the optative and on its most frequent and long-lived functions. Its use in the Biblical texts is sometimes motivated by stylistic features (some forms are thought to be tied with the overt attempt to imitate Attic prose) and the textual genre (the liturgical content and value of the Scriptures); some scholars also wonder whether sociolinguistic factors and linguistic contact played a role in the fate of the optative.²⁵ These questions are hinted at sparingly in the literature, but still remain mostly unanswered. As we see from the discussion and the analysis presented in the next sections, an updated theoretical framework and a fresh look at the data have the potential to shed some light on these issues.

Up to this point, we have adopted the terminology and reported the classification that grammars of Ancient Greek employ when they deal with the optative. In this section, we review those paradigms which have the potential to help us shed light on the evolution of the Greek optative. As becomes clear, its most persistent functions are better understood through a multi-level analysis which considers typology, reality status, modality, and pragmatics.

Starting with typology, for Dobrushina et al. (2013) the term ‘optative’ refers to “an inflected verb form dedicated to the expression of the wish of the speaker”. Although most languages regularly express this meaning, relatively few have a morphological category which is exclusively dedicated to it. The following examples show different ways of expressing optative meanings: Engl. May he rest in peace! and Lat. Requiescat in pace. Considering that the optative domain is heterogeneous from both a formal and a semantic point of view, Dobrushina (2011) also proposes a distinction between two different types: the ‘performative optative’, which is used to bless or curse, and the ‘desiderative optative’, which is employed to express the speaker’s wishes or dreams.²⁶ In Dobrushina et al. (2013), these are called “opt1” and “opt2”. One may find the first type in those European languages which have a set of idiomatic blessing or cursing formulae based on non-productive formal patterns. For example, Germanic languages may use obsolete forms of subjunctive: cf. Engl. God save the queen! and Nor. Leve fedrelandet! ‘Long live the fatherland!’ (Steblin-Kamenskij apud Dobrushina 2011).

Desiderative optatives are even more rarely expressed by dedicated moods and, in most European languages, pure wishes (those over which the speaker has no power whatsoever) are normally encoded either by non-morphological means (such as modal verbs), or by modal forms which are also used for other functions (e.g. conditional clauses, volitional complement clauses, epistemic usages, etc.). This is the case of the subjunctive mood: cf. Engl. If only life were lived in reverse and It. Se solo avessi più tempo! ‘If only I had more time!’.

As Dobrushina (2011) shows, Caucasian languages display an extraordinary density of optatives, and also separate forms for performative and desiderative functions; the uniqueness of this area is even more evident in the WALS map (Dobrushina et al. 2013), in which only 48 languages out of 319 (that is, 15 % of the total) have an inflectional optative.

It is clear that the category identified by typologists coincides only in part with the functions of the Greek optative, which is in fact tied to other notions and domains such as the potential, which is a commonly accepted catch-all term (cf. Willmott 2007), and the irrealis, which needs to be qualified further. Before doing that, however, a brief terminological detour on the label ‘optative’ gives us the chance to underline a rather substantial issue.

Strictly speaking, as van der Auwera & Schalley (2004: 87–90) point out, in the classical languages the term is a misnomer because in Latin it denotes a functional category (which the authors call ‘use category’) and a specific meaning of the subjunctive mood, whereas in Ancient Greek it refers to a formal category (called ‘form category’).²⁷ However, it turns out that the expression of wishes is not the main function of such a form. As Duhoux (2000: 226) notes, the overwhelming majority of optatives found in Classical Greek corpora (99.6 % following the counts by Sanspeur and 93.3 % according to those of Chanet) have potential values. It is only by NTG times that the optative of wish becomes relatively dominant (63,8 % vs. 36.2 %, cf. also van der Auwera & Schalley 2004: 90) .²⁸ Now, if the encoding of wishes is a marginal function, not only does the term ‘optative’ remain inconsistent with the definitions proposed by typologists, but other problems also arise when one looks at the synchronically attested values in a diachronic dimension.

Any hypothesis on the original features of the optative and its position within the Indo-European modal system must remain speculative, but the ongoing debate seems to have found some firm points and promising developments. The comparison between the languages in which the optative is best attested, i.e. Greek, Vedic Sanskrit (Whitney 1889: 215, 219; Dahl 2010: 230; Gotō 2017: 362), and Avestan (Hoffmann & Forssman 2004: 178), allows scholars to reconstruct ‘potentiality’ and ‘wish’ as basic functions. Also, the difference between optative and subjunctive, which was traditionally described in terms of degree along a scale of possibility and probability (e.g. Whitney 1889: 216), is currently often conceptualized within the domain of irrealis (Greenberg 1986, Tichy 2002, Tichy 2006: 304–305).²⁹

In this framework, Willmott criticises the idea that the optative is a past-time variant of the subjunctive and expresses remote possibility (cf. 2.1.2 supra). In her opinion, the use of both cupitive and potential optatives with present- and future-time reference, along with the absence of the hic et nunc specification in secondary endings, point to the ‘timeless’ and ‘unreal’ nature of this mood (Willmott 2007: 115, 123 and 150–151).³⁰ In this perspective, she elaborates on Greenberg’s continuum, which places the uses of the Greek moods along a gradient in which the optative is more irrealis than the subjunctive and the subjunctive is, in turn, more irrealis than the indicative. Her discussion of the ways in which this cline intersects the epistemic and deontic uses of the moods, however, leaves two issues aside: the definition of the realis-irrealis dimension and its relationship with modality.

The term modality refers to a conceptual domain that can be encoded through a wide range of linguistic means and grammatical categories. Scholars define and subdivide this realm according to different criteria: Bybee et al. (1994: 176–181) distinguish four broad sub-domains (agent-oriented, speaker-oriented, epistemic and subordinating modality).³¹ In their opinion, the Greek optative represents “the wish or hope of the speaker expressed in a main clause” and is part of the speaker-oriented modalities.³² These are typically found in directive utterances³³ and are encoded by imperative, prohibitive, optative, hortative, admonitive or permissive forms and constructions. Conversely, notions like volition, desire and evidentiality are excluded from the realm of modality in van der Auwera & Plungian (1998), who restrict the use of the term modal to those categories whose functions can be described by the concepts of possibility and necessity. More recently, Portner (2009: 1) suggests that “modality is the linguistic phenomenon whereby grammar allows one to say things about, or on the basis of, situations which need not be real”. Leaving aside the (not-so-immediate) definition of the term real, the very reference to the notion of reality reminds us of the realis-irrealis dimension, which largely overlaps with modality, but also encompasses notions that are commonly viewed as non-modal (e.g. the future tense).³⁴

Greek and Indo-Iranian employ a variety of formal means to encode various types of directives: injunctives and subjunctives for prohibitions (with mā in Sanskrit and μή in Greek, respectively), optatives for wishes, subjunctives for exhortations, and of course imperatives (but also infinitives and futures) for orders. Such variation is motivated by the inherently hybrid and multifaceted nature of directive situations, i.e. situations which, in principle, can be characterised as being logically unactualized but, in fact, display different levels of realisness according to various factors.

Mauri & Sansò (2012b) suggest that positive directive situations presuppose three main functional components: a) the speaker’s wish that a SoA becomes true, b) the appeal to the addressee(s) to help make this SoA true, and c) the expectation that the desired SoA is brought about in the near future. In addition, the performer(s) of the action(s) required may coincide: (i) with the addressee, (ii) with the speaker, (iii) with a third party, or (iv) with any possible combination of (i)-(iii). The most typical directive is thus a manipulative speech act in which the speaker has authority over the addressee and, therefore, high expectations concerning the fulfilment of their wish. Moreover, the situation “is deeply rooted in the deictic here-and-now in which the speech act is uttered” (Mauri & Sansò 2012b: 151).

In this section we have seen that no theoretical paradigm precisely captures all the functions of the Greek optative, and that—on the contrary—the terminology can sometimes be misleading. At the same time, we have suggested that the interplay between modality, realisness, and pragmatics may open new perspectives on the evolution of the optative and its emerging role as a modulator of illocutionary force in pseudo-directives. To complete the diachronic picture, in the next section we present a quantitative and qualitative analysis of the dataset from the NTG corpus.

In the NT, there are 73 tokens (and 31 types) of optatives:⁴⁴ 41 are found in main clauses and 32 in dependent clauses. Only two (Acts 8:31 and 17:18) out of the 41 main clause optatives are accompanied by the modal particle ἄν; the fixed expression μὴ γένοιτο accounts for 15 more main clause optatives. As for dependent clauses, 13 optatives are in indirect questions,⁴⁵ three are in indirect statements/object clauses,⁴⁶ eight are in conditional clauses,⁴⁷ two are in temporal clauses, one is in a purpose clause, and one in a modal clause.⁴⁸ A relatively high concentration is found in Luke and Acts. It is generally said that, among the other Gospels, only Mark has one secure example (5:43), but our count reveals that there are five more instances, sometimes found in variae lectiones, and often neglected because they are thought to be due to the corrections of Atticistic copyists or redactors.⁴⁹

Very few main clause optatives (Mark 8:37, Acts 8:31, 17:18) are of a potential nature, the rest being used to express wishes, blessings, or curses (this group includes the 15 μὴ γένοιτο). A famous example features in the cursing of the fig tree, an episode which is found in Mark (11:14 καὶ ἀποκριθεὶς εἶπεν αὐτῇ· μηκέτι εἰς τὸν αἰῶνα ἐκ σοῦ μηδεὶς καρπὸν φάγοι. καὶ ἤκουον οἱ μαθηταὶ αὐτοῦ ‘Then he said to the tree, “May no one ever eat fruit from you again.” And his disciples heard him say it.’). The presence of φάγοι in a speech delivered by Jesus has led to the hypothesis of an intentional high-register feature which is supposed to underline his status (Lee 1985); other ‘high-register’ features of Jesus’ language are adduced, but as far as the optative is concerned this claim seems to rest exclusively on the assumption that at the time its potential use was an Atticistic feature. From our perspective, this should be seen as a remarkable instance of performative optative (cf. 3.2 supra), and the power of Jesus’ words is confirmed by the perlocutive sequel added in the alternative version of Matthew, where, however, the curse is in the subjunctive (21:19 καὶ λέγει αὐτῇ· μηκέτι ἐκ σοῦ καρπὸς γένηται εἰς τὸν αἰῶνα· καὶ ἐξηράνθη παραχρῆμα ἡ συκῆ ‘Then he said to it, “May you never bear fruit again!” Immediately the tree withered’).⁵⁰ A parallel passage such as this is particularly useful for an analysis of variation in the use of moods. There are at least three differences between the versions attested in Mark and Matthew, and these could account for the different moods. In Matthew 21:19 all actants involved in the curse are present, the actualisation of the SoA does not depend on a third agent, and the consequences of the curse take place immediately. Conversely, in Mark 11:14, where the optative is used, the same curse involves indefinite third-person actants (μηδεὶς καρπὸν φάγοι), either present or absent, and the consequence of the curse is not spelled out. In other words, the situation denoted by the subjunctive is more ‘realis’ and better framed within a hic et nunc, which is otherwise suspended when the optative is employed. One should however note the varia lectio γένοιτο in ‮א‬‎, Θ and Origen, which seems to point to a certain degree of functional overlap, at least in later Greek, or influence from the frequent μὴ γένοιτο. Two more passages (Acts 8:20, Jude 1:9) feature an optative in a curse or imprecation, which in NT is usually expressed through the imperative (e.g. the fixed expression ἀνάθημα ἔστω ‘let him be accursed’ in Gal. 1:8); a confirmation of this comes from Acts 1:20, which contains the imperative λαβέτω to quote Ps. 108:8, which originally has the optative λάβοι (Boyer 1988: 132). The so-called ‘less vivid’ future condition, which features optatives in both the apodosis and the protasis (cf. 2.1.4 and 2.2.4 supra), is another Classical use that is lost in the NT.

Optatives have disappeared from most subordinate clauses, often being replaced by the subjunctive (as in, e.g., purpose clauses). However, the ‘iterative optative’ (cf. 2.2.4 supra) is still found in εἰ τύχοι (1 Cor. 14:10, 15:37), which seems to be a fixed expression. Elsewhere, this construction is replaced by ἄν + indicative (e.g. Mark 6:56, Acts 2:45, 4:35, 1 Cor. 12:2) or, more rarely, by ὅταν + imperfect or aorist (only 1× in Mark 3:11). An interesting peculiarity of Luke and Acts is the relative frequency of optatives in indirect questions (6× in both). This, along with the fact that 20 out of 29 potential optatives are in questions led Boyer (1988: 133–134) to suggest that “the added “potentiality” which inherently is involved in a question may have made it more likely that the optative should survive there”. We can modify this assumption in light of what we have argued above about the use of the optative in pseudo-directives (cf. 3.6.1 supra). Since all potential optatives in main clauses are in rhetorical questions introduced by an interrogative pronoun or adverb (τί in Mark 8:37 and Acts 17:18, and πῶς in Acts 8:31),⁵¹ one can assume that the use in pseudo-questions favoured the survival of the optative in indirect questions as well. Also, it is worth noting that these are introduced by the same formal means as direct ones, so it is not surprising that optatives are relatively abundant in interrogative contexts.

The preceding discussion has confirmed the retention of optatives in fixed expressions but has also highlighted their frequency within specific types of speech acts. In the next paragraphs, we discuss three factors that may help explain the usages observed above and the functional evolution of this mood. We consider the role of Atticism, language contact, and the Stoic grammatical theory.

As mentioned in 2.3 and 4, the use of the optative in the LXX and NT has often been tied with the overt attempt to imitate Attic prose by later authors. The pressure to ‘Atticize’ came mainly from rhetorical schools, lexicographers, and grammarians, who included the optative in its full range of classical functions among the basic hallmarks of correct Attic usage. According to Boyer (1988: 140): “[t]he optative had practically disappeared from the common language, and only later received a temporary revival by Atticizing purists who were attempting to restore the literary language of Greece’s golden age”.⁵² On the other hand, these tendencies go hand in hand with the typology of these texts, whose peculiar genre is highlighted in Turner’s observations (1963: 131–132): “one must not reject too lightly the possibility that the optatives in the NT owed their preservation in some measure to their incidence in the pompous and stereotyped jargon of devotion. These optative phrases are decidedly formal. […] The retention of the optative at a time when everywhere they were diminishing need not surprise us in view of their value for the liturgy”.

However, the intentional use of such high-register features seems to contradict the assumption that “the Scriptures were written and rendered in the language of the people” (Turner 1963: 131).⁵³ But the inconsistency becomes only apparent if, looking at the history of the optative, one distinguishes between a revival, mainly due to cultivated redactors, and a survival, partly due to natural speakers. Since these two phenomena involve different motivations, their linguistic effects are also different. On the one hand, stylistic needs tend to preserve structures and meanings that were recessive at the time; on the other, popular speech continues to employ expressions and functions that reflect the last relic of the optative. Yet, we must not forget that non-standard forms of Greek were also spoken by large communities of non-native speakers.⁵⁴

In this respect, a topic that needs to be briefly re-addressed is the role that language contact might have played in the fate of the optative in post-Classical Greek. Of course, a complete reassessment of the issue goes well beyond the limits of this paper but, in the literature, there are hints at the fact that non-native speakers must have found it difficult to acquire the modal system of CG and that this is a plausible reason for the decline of the optative.⁵⁵ This is probably true, but in our view the role of language contact in contributing to the survival of the optative is far more interesting. Users of Greek as a foreign language probably learned isolated forms in fossilized locutions such as μὴ γένοιτο, θεὸς φυλάξοι, εἰ τύχοι, which were part of everyday language (cf. 2.3.3 supra). But the optative was also employed in more extended formulas and multiword ‘chunks’⁵⁶ like ready-made greetings and wishes, or set legal phrases, as one can also see from the evidence in first-millennium CE papyri, e.g. χαίροις, κυρία μου (P. Oxy. 112), ἢ ἔνοχος εἴην τῷ ὅρκῳ (P. Oxy. 82, 7–8), ὃ μὴ εἴη (P. Oxy. 1473, 12–13).⁵⁷

In this scenario, we can therefore imagine two coexisting dynamics: on the one hand, the bleaching of the syntactic rules that govern the usage of the optative (e.g. the ‘oblique’ type), and on the other, the strengthening of the pragmatic functions that maintain this mood alive in language use.⁵⁸ As for the connection with particular kinds of speech acts, further motivations can perhaps be found by going back to the origin of the term ‘optative’ and its conception within the ancient grammatical theory—which has been hitherto partially neglected.

If in Classical Greek the expression of wishes was not the prevailing function of the optative (as we noted in 3.4 supra), how does one explain its traditional denomination? As is well known, the Latin term optativus is a calque on the Greek adjective εὐκτικός ‘expressing a wish, a prayer’, and O’Sullivan (2011: 83) wonders why the Greek grammarians privileged this label, at the expense of the potential meanings and functions. According to Jannaris (1897: 563), the name could have been chosen late, probably when the wishing or praying function was virtually the only function left. An alternative explanation can be found in the division of the λόγος into types, which goes back to Protagoras and is later elaborated and expanded by the Stoics. In a passage by Diogenes Laërtius (Diog. Laert. 9.53–54) we read that διεῖλέ τε τὸν λόγον πρῶτος εἰς τέτταρα· εὐχωλή, ἐρώτησιν, ἀπόκρισιν, ἐντολήν […], οὓς καὶ πυθμένας εἶπε λόγων ‘He was the first to divide speech into four kinds—prayer, question, answer, command […], which he even named ‘foundations of speech’.’ [transl. Huitink & Willi 2021]

Evidently, the four ‘bases’ of the λόγος correspond to different kinds of speech acts (Schenkeveld 1984: 330, Lallot 1989: 162, Shalev 2008: 249), which can be roughly mapped onto the traditional moods: “prayer and command immediately suggest optative and imperative, and the indicative is the invariable mood of an answer” (O’Sullivan 2011: 84–85).⁵⁹ Prayers are also included in the Stoic taxonomy of the so-called λεκτὰ αὐτοτελῆ ‘complete sayables’,⁶⁰ which for Schenkeveld (1984: 351) “were the starting-point for a Stoic distinction of moods, which comprised grammatical moods as well as a specific group of adverbs and some particles”.⁶¹

If we accept this view, it is therefore plausible that the concept of optative could be derived from its recurrent association with prayers and curses (as can be seen from the Stoic label ἀρατικόν) or wishes and solemn requests (as is evident from the usual term εὐκτικόν).⁶² The Greek perspective on the moods, which diverges from that of modern scholarship, in its focus on actual uses rather than on abstract meanings, could thus motivate not only the naming of the optative as the mood for prayers and wishes (εὐκτικὴ ἔγκλισις), but also its tenacious survival in specific kinds of speech acts.

As seen in section 4, NTG optatives are used to issue requests that expect no fulfilment and ask questions that expect no answer. Because of its intrinsic emphasis, the εὐκτικὴ ἔγκλισις is indeed perfectly suitable to prayers and blessings, but also to rhetorical questions, i.e. pseudo-questions that do not require information and may correspond by implicature to indirect assertions (both affirmative, as in Acts 17:18 τί ἂν θέλοι ὁ σπερμολόγος οὗτος λέγειν; ‘What is this babbler trying to say?’ = ‘he is just talking nonsense’, and negative, as in Mark 8:37 τί γὰρ δοῖ ἄνθρωπος ἀντάλλαγμα τῆς ψυχῆς αὐτοῦ; ‘What can anyone give in exchange for their soul?’ = ‘nothing can be given in exchange’) or indirect directives (as in Acts 8:31 πῶς γὰρ ἂν δυναίμην ἐὰν μή τις ὁδηγήσει με; ‘How can I [understand it], unless someone explains it to me?’ = ‘someone should explain’). In the last example, we may observe a sort of ‘deferred’ perlocutionary effect (Philip sits next the Ethiopian official and answers his questions on the Scriptures). However, in most utterances with an optative, expected results of the speech act tend to remain in the background (cf. the cursing of the fig tree in Mark 11:14), because both rhetorical questions and positive/negative wishes are to be taken as ‘less real’ and ‘more indirect’ than actual questions, commands, and prohibitions.

As mentioned above (cf. 3.6.2), phenomena pertaining to modality can serve as indicators and modulators of illocutionary force. The weakening direction of modulation is mitigation, a concept first applied to pragmatics by Fraser (1980), who inspired research on the strategies for attenuating or boosting illocutionary force (cf. e.g. Holmes 1984).⁶³ In this perspective, the optative turns out to be a mitigating device which allows speakers to obtain ‘indirectedness’ on different levels.⁶⁴ As we have seen, pseudo-directives are characterized by a de-focalization of the deictic origin of the utterance, since optative situations usually imply a spatio-temporal displacement, which suspends the here-and-now of ongoing communication (cf. 3.6.1). The detachment of the utterance from its deictic origin through the strategic manipulation of one or more components of the I-here-now triad is a form of mitigation through de-actualization (Caffi 2007: 66). In pseudo-directive contexts, the relation of the optative with the non-actual dimension of irrealis unfolds in terms of non-imminence.

Furthermore, even the speaker can be de-focused when the optative utterance features some ‘impersonalization mechanism’ (Brown & Levinson 1987: 273) to obtain generalising or gnomic effects⁶⁵ (as in the rhetorical question seen above, Mark 8:37 τί γὰρ δοῖ ἄνθρωπος …), or to attribute responsibility to an external agent or factor (as in Acts 8:20 τὸ ἀργύριόν σου σὺν σοὶ εἴη εἰς ἀπώλειαν ‘May your money perish with you’, or Jude 1:2 ἔλεος ὑμῖν καὶ εἰρήνη καὶ ἀγάπη πληθυνθείη. ‘Mercy, peace and love be yours in abundance’). The use of the optative for ‘strategic mitigation’ is also evident in 1 Peter 3:17, where the parenthetical εἰ θέλοι τὸ θέλημα τοῦ θεοῦ ‘if it is God’s will’ attenuates an indirect directive utterance beginning with κρεῖττον γάρ ‘for it is better […]’. But formulaic expressions, like the enunciative mitigating device εἰ τύχοι, more often serve for ‘ritual mitigation’, “i.e. mitigation typically realized by routine formulae or ‘indirect speech acts’ that have become catachreses, such as clichés expressing politeness” (Caffi 2007: 86). These, as we have seen, are the most frequent contexts and long-lasting functions of the optative, the evolution of which now seems clearer, in the light of the interaction between irrealis, illocution, and mitigation. The fact that the optative tends to survive in these usages seems consistent with the progressive loss of modal functions which leads to its “removal from the verbal system” (Evans 1999: 490). As the possibility-related uses begin to disappear, the optative gradually slips out of modality, towards the end of the irrealis cline and into the domain of illocution.