The translation and transmission of ‘diatribal’ verbs in the textual traditions of the Zlatostruj collection

The ‘diatribe’ is a dialogical mode of exposition, originating in Hellenistic Greek, where the author dramatically performs different voices in a polemical-didactic discourse. The voice of a fictitious opponent is often disambiguated by means of parenthetical verba dicendi, especially φησί(ν). Although diatribal texts were widely translated into Slavic in the Middle Ages, the textual history of the Zlatostruj collection of Chrysostomic homilies especially suits an investigation not only of how Greek ‘diatribal’ verbs were translated, but also how the Slavic verbs were transmitted or developed in different textual traditions. Over time, Slavic redactional activity led to a homogenization of verb forms. The initial variety of the original translation was partly eliminated, and the verb forms and became more firmly established as prototypical diatribal formulae. Especially the (increased) use of the 2sg form has theoretical consequences for the text’s dialogical structure. Thus, an important dialogical component of the diatribe was reinforced in the Zlatostruj’s textual history on Slavic soil.


Introduction
This article investigates the phenomenon of the 'diatribe' in a selection of homilies from the Church Slavonic Zlatostruj collection. A basic characteristic of 'diatribal' discourse is that the author performs different voices in the context of a polemical clash with a fictitious opponent. A more precise definition of the diatribe will be given below, but one of its major features in (Hellenistic) Greek is that the fictitious opponent's voice can be explicitly marked by parenthetical verbs, most notably φησί(ν) 'says (he)'. In Greek, this 3SG.PRES verb form had petrified into a quotative particle. Previous research has established that its most prototypical Slavic equivalent in a variety of medieval sources is the parenthetical 3SG.AOR form рече 'said (he)' (e.g. Kakridis, 2019). However, a significant range of other verb forms is also in use. This article will show that речеши (2SG.PRES) is in fact as important a diatribal formula as рече, and that the former outpaced the latter over the course of the Zlatostruj's textual history. Its increased use will turn out to have theoretical implications for a pragmatic analysis of diatribal verbs in terms of participant roles and disambiguation of the referential framework. It will be argued that the increased use of 2SG.PRES forms led to a reinforcement of the dialogical character of diatribal discourse. This is a significant development of the diatribe on Slavic soil.
The present article investigates the way in which Greek parenthetical verbs (functioning within the discourse tradition of the diatribe) are rendered in Slavic in a substantial and representative sample of 22 Chrysostomic homilies in the Zlatostruj. The verbs will be analyzed not just in one source, but we shall see how they survived in the complex web of the Zlatostruj's transmission history. Thus, the extremely wide-ranging manuscript traditions of the Zlatostruj collection allow us to see not just one final result, but the development over time of some of the main diatribal formulae. This is an advantage over against previous case studies into the diatribe in the Slavic realm, where only the end result of the process in one particular monument, such as the Codex Suprasliensis, was investigated (e.g. Dekker, 2021a). Connected to this, some observations will be made about the textual history of the Zlatostruj. The distribution of parenthetical verbs will be discussed in relation to earlier researchers' conclusions on the relationship between different redactions of the Zlatostruj. In addition, the article provides a theoretical explanation of the function of the parenthetical verbs from the perspective of (historical) pragmatics.
The diatribe is a well-known phenomenon in circles of Classical scholars. In Slavic studies, on the contrary, it is largely off the radar. This is unjustified. The larger part of (Old) Church Slavonic literature was translated from Greek, and thus constitutes a potentially valuable source of information about the way in which Greek discourse strategies, such as the diatribe, were reflected in translation. In addition, a promising field of research is opened not only to gain an understanding of the way in which the diatribe was rendered in translated texts, but also of the extent to which the diatribe developed into an independent Slavic discourse tradition. To this end, the study of translated texts is a first prerequisite, towards which this article is a contribution. On the one hand, it investigates diatribal features in translated texts; on the other hand, it provides insights into the way these features were reproduced and maintained in manuscript traditions within the orthodox Slavic realm. To what extent did the translators of the Zlatostruj correctly interpret the structure of diatribal discourse? And to what extent did later scribes, who did not have access to the Greek sources, recognize the meaning of diatribal formulae and pass them on unscathed? We shall see that in most cases the interpretation of diatribal formulae was felicitous, and in some cases even enhanced by Slavic scribes in the later development of the translated homilies.
It might seem far from obvious that logical and coherent argumentation was widely employed in the Orthodox Slavic realm in the Middle Ages. It has been argued more than once that for the Slavs argumentation and polemical strategies were secondary to the liturgy (e.g. Thomson, 1999;Garzaniti, 2013, p. 204; for recent counterarguments see Kakridis, 2018;Ostrowski, 2018;Prestel, 2019). Our investigation into the use of diatribal verbs will show that both the Slavic translators of the Zlatostruj and later scribes were well able to interpret intellectually challenging texts and translate or transmit them in a coherent way. Thus, even the era allegedly characterized by Florovsky's (1962) "intellectual silence" and Thomson's (1999, p. 9) so-called "obscurantism" produced texts that were permeated with features of Hellenistic philosophy. That is worth emphasizing once more, over against the "anti-Orthodox bias of some investigators, including most recently Francis Thomson" (Prestel, 2019, p. 180), in order to provide a rehabilitation of logical thinking and show that the employment of rhetorical strategies was successfully transferred into and transmitted within the Orthodox Slavic realm.
The rhetorical strategies in question will be studied through a linguistic lens. In other words, the diatribe is to be demarcated by its main linguistic manifestations (see Sect. 2). Thus, qualitative methods of traditional philological close reading are used to select data to be incorporated into a quantitative investigation into the use and development of parenthetical verb forms. Linguistic means are identified on the micro-level that play a role in expressing a rhetorical function in the polemical discourse on the macro-level. In this way, rhetoric and pragmatics mutually inform each other by what Ilie (2018, p. 88) calls "cross-fertilization". At the same time, the emphasis remains on the relevant linguistic elements as identified in Sect. 2, which provide the building blocks for the rhetorical strategies of the diatribe. Ilie (2018, p. 92) succinctly gets to the heart of the distinction between pragmatics and rhetoric when stating that "pragmatics focuses on language as it is used by human beings, whereas rhetoric focuses on human beings as they use language". Keeping this distinction in mind, the main results of this study will turn out to be of a linguistic nature.

The diatribe
The diatribe, also known as 'Cynic-Stoic diatribe', is a dialogical mode of exposition that is best known for its use in Greek polemical-didactic texts from the Hellenistic period onwards. The diatribe originated in the works of the Hellenistic philosophers Bion (c. 335-245 BC) and Teles (fl. c. 235 BC), and reached its zenith in Roman times, with Epictetus (c. 50-135 AD) as its most prominent exponent. This discourse tradition was carried over into Christianity by the apostle Paul, most notably in his epistle to the Romans, as was first shown convincingly by Bultmann (1910), whose observations were subsequently refined by Stowers (1981), Song (2004) and, more recently, King (2018). The diatribal style of argumentation was consequently adopted by the Church fathers in the following centuries, also under the influence of the Second Sophistic. This leads us to John Chrysostom (c. 347-407), whose homilies provide an especially favourable sample of a lively style of diatribal discourse. Many of his works were translated into Slavic from the 9th/10th century onwards, so that the diatribal style entered the Slavic realm many centuries after its initial implementation in Hellenistic times. This line of development presents an intriguing case of how a well-known style of dialogical discourse that had developed fixed linguistic manifestations in Greek (see Table 1 below) was received and appropriated in a newly Christianized culture where the written word was an emerging phenomenon in its incipient stages. Moreover, the following manuscript traditions of the different redactions of the Zlatostruj show us how the features of the diatribe were maintained or developed in subsequent centuries.
The main characteristic of the diatribe is that its dialogical character is performed dramatically. In the context of a polemical exchange, the author engages in a discussion with a fictitious opponent. Most importantly, the fictitious opponent's arguments are not just quoted, but acted out dramatically. This is the main distinctive feature that separates the diatribe from other polemical or dialogical strategies. Thus, the author (or rather, speaker or preacher in this case) performs both his own voice and the voice of his fictitious opponent. In this way, an alternation of voices occurs in the discourse, resulting in a lively exchange of polemical positions.
It should be emphasized that the diatribe is not a genre, but a mode of discourse organization. Yet, the genre of the homily is particularly suitable for diatribal style to be employed. Thus, even though sermons are often considered to be among the most prototypically 'distant' and monological kinds of discourse, 1 the use of the diatribe ensures that there is indeed a variety of voices being heard in the words of the preacher. Examples of typical diatribal formulae will be presented in Sect. 2.

Greek source
A Greek critical edition of the vast array of Chrysostom's homilies is unavailable and this will remain so in the foreseeable future. The most useful edition is part of the Patrologia Graeca (PG) series, compiled by Jacques-Paul Migne (1862) on the basis of various 17thcentury editions. Bady (2010, p. 162) calls this edition a "textus receptus" of the Church fathers; at the same time, he voices a negative opinion about its value and calls it a "texte imprimé largement insatisfaisant" (Ibid., p. 163). However, even a critical edition would not give us absolute certainty about the text the Slavic translator had in front of him. Dimitrova (2016b) tries to identify the Greek version the Slavic translator must have used for Chrysostom's homilies on Titus and the Acts of the Apostles, but she ultimately concludes that the question has to remain unresolved. This is not a serious problem for the present investigation, as the variation we shall encounter is primarily intra-Slavic; i.e. it arose as a result of subsequent editing of the translation without referring to the Greek original (see below). Thus, variation between the different Slavic versions can never be due to Greek manuscript variation. An investigation of the different Greek manuscript traditions is, therefore, not an urgent necessity at this point. References to the Greek examples quoted below are made to line numbers in the electronic edition 2 or, when the latter is not available, to the PG volume and text column (not page) numbers.

Textual history of the Zlatostruj
The textual history of the Zlatostruj is complicated. The original translation (autograph), which was produced in the Preslav literary centre under the reign of czar Simeon I (893-927), 1 Sermons are classified by Koch and Oesterreicher (1985) as prototypical exponents of the "language of distance". However, the dialogical elements of the diatribe that will be discussed below are to be seen as features of 'conceptual orality' ("konzeptionelle Mündlichkeit," Ibid.). Karvounis (2016, p. 57, fn. 123) applies these terms to different genres of Ancient Greek dialogues. Thus, the relationship between monologue and dialogue in Chrysostom's homilies is to be analyzed as functional variation within one and the same linguistic system. It is not related to Greek diglossia, which (if understood as a phenomenon permeating all levels of society, according to Karvounis's terminology) emerged only in the Middle Ages (Ibid., p. 55) and can, therefore, not be a relevant term for Patristic texts like these. 2 https://www.documentacatholicaomnia.eu/.
has not survived. This proto-collection of Chrysostomic homilies was not translated by a single translator; it must have been a collective enterprise (cf. Dimitrova, 2016a, p. 450). The proto-collection has not survived in its original form, either; even the exact number of homilies contained in it is unknown, though Miltenov (2013, p. 74) estimates that it must have contained at least 150 homilies. Instead, what has survived from the original Zlatostruj collection has come through to us in different 'redactions'. Following Thomson's (1982) terminology, the two main redactions are the Longer collection (henceforth: L) and the Shorter collection (henceforth: S). Miltenov (2013) also distinguishes a third (Hilandar) and a fourth (Voskresenski) redaction, but these will not be included in the present investigation.
As we shall focus on L and S, it is necessary to consider the relationship between these two redactions. L consists of 138 homilies (numbered L1-L137, with the double count of L78a 3 ), although several manuscripts contain only homilies L1-L45. It has been argued that these first 45 homilies form the most archaic core of L, and that the rest of the homilies up to L137 were added slightly later (although still in the mid-10th century at the latest), both from the Zlatostruj proto-collection and from other sources (Miltenov, 2013, p. 72;pp. 76-77). 4 Both parts of L (and also S) contain a selection of homilies from one and the same proto-collection (Dimitrova, 2019, p. 421). It is, therefore, justified to treat the Zlatostruj as different redactions of what was originally one whole, i.e. it is to be traced back to one original translation, part of which has come down to us in L and S.
The manuscript tradition of L is very stable; variant readings are rare. Thus, although the archetype of L is not extant, we can ascertain with a reasonable degree of certainty that later copies are a faithful reflection of the archetype of L. In addition, L is generally recognized to be the redaction that is closest to the original translation. Thus, apart from some exceptional cases, L will normally tell us more about the original translation than S. S constitutes a later reworking of a number of homilies from the proto-collection (Thomson, 1982). It contains 81 homilies, 62 of which are also available in L. The other 19 homilies in S were also taken from the proto-collection, but were not included in L.
The next point to be noted is that L and S are independent selections from the protocollection. In addition, S shows more deviations and omissions. Importantly, the scribe who edited the homilies in S did so without consulting the original Greek text (Miltenov, 2010, p. 387;Dimitrova, 2016a, p. 30). Consequently, changes to the Slavic text were made rather haphazardly, based on the scribe's understanding of the initial translation, but not necessarily reflecting a better interpretation of Chrysostom's intended meaning. In any case, the differences between L and S that will be discussed in the remainder of this article cannot be due to Greek manuscript variation, but must have arisen on Slavic soil. Accordingly, S is not a direct 'descendant' of L; nevertheless, L shows us a 'purer', more archaic state that is closer to the original translation. This is an important methodological point, as it enables us to compare diatribal features in the original translation, of which L is a faithful reflection, and subsequent alterations in S. The caveat remains that some differences between L and S might be due to manuscript variation and corruptions in L (of which I have found some instances), but the overall tendency is that the deviations stem from a reworking of the original translation in S. As we shall consider the distribution of diatribal verbs throughout our sample of homilies from L and S (see the Appendix for an overview), it will be noted that L shows a lot more variety, whereas S is more homogeneous. This is an additional argument in favour of L as a faithful reflection of the original translation, which was made by various translators. The variety in L is unlikely to be the work of a later scribe, as a later reworking by one scribe would have resulted in more homogeneity of parenthetical verbs, as is the case in S.

Slavic sources
The Zlatostruj became extremely popular in the East Slavic lands, so that the vast majority of extant copies are of Russian provenance. For the present investigation, five different sources have been selected. For L, not even a diplomatic edition of one single manuscript exists. Therefore, I rely on three manuscript sources of L. My main source is codex 43 from the Moscow Theological Seminary (MDA 43), "copied in 1474 but clearly from a much earlier codex" (Thomson, 1982, p. 2). This can be characterized as an archaic manuscript. It is now kept in the Russian State Library and is easily available online, in high-quality colour images. 5 Although variant readings among the manuscripts of L are rare, I use two additional sources to exclude the possibility that unexpected verb forms are merely a distinctive feature of MDA 43. These two additional manuscripts are MAB F.19 № 238 (first quarter of the 16th century, kept in the Lithuanian Academy of Sciences, Vilnius), 6 which contains homilies L1-L45, and RGB ф.98 №240 (third quarter of the 16th century, kept in the Russian State Library, Moscow), 7 which contains homilies L58-L137. Thus, L is represented by three sources, two of which run parallel for each homily. 8 S is attested in a smaller number of manuscripts, but these are relatively well accessible in various editions. I use an electronic version of Georgieva (2003), 9 which is a diplomatic edition of the 12th century F.п.I.46 manuscript kept in the Russian National Library, St. Petersburg. This edition supersedes the older partial edition by Malinin (1910). The codex has been characterized as "extremely defective in many ways" (Thomson, 1982, p. 2), but its availability in an edition and its exceptionally old age make it an indispensable source for the present study. However, although it is the oldest available manuscript, it is by no means the most reliable source for the reconstruction of the original translation. A second, semi-critical edition of S is the one contained in the Macarian Menologion, i.e. Metropolitan Macarius's Velikie Мinei Čet'i for the 13th-15th of November, edited by the Arxeografičeskaja Kommissija (VMČ, 1899). 10 The edition is based on the Uspenskij copy, but provides deviant readings from two other manuscripts, viz. the Carksij and Sofijskij copies (all three copies are dated to the mid-16th century). Thus, the development of S in its textual transmission in the Russian lands is represented in the present study by four manuscripts.
In the examples provided below, the Greek text is translated by the present author with due attention to the earlier English translations edited by Schaff (1886-1890). For the Slavic text of the examples, only those manuscripts are cited that are relevant to the discussion. Thus, e.g. if two witnesses of L have no or only minimal variation, only one will be quotedusually MDA 43.

Major diatribal features
I shall now provide a short outline of the main diatribal features in Greek and Slavic. An overview of corresponding formulae and their functions is provided in Table 1. These formulae serve as diagnostic criteria that are often used as evidence that a text should be considered an exponent of the discourse tradition of the diatribe (e.g. Stowers, 1981;King, 2018). These typically diatribal formulae allow us to treat Chrysostom's homilies under consideration as exponents of the Cynic-Stoic diatribe, rather than just showing general features of orality or dialogue. Chrysostom's homilies are generally recognized to have been delivered extemporaneously and written down by means of tachygraphy (e.g. Goodall, 1979, p. 66). This provides ample room for an analysis in terms of 'oral' and 'literate' features. This is, however, not the objective of the present study. Instead, the formulae serve as diagnostic criteria of a more narrowly defined style. These linguistic features can be traced down from Hellenistic times onwards, through the New Testament and Greek church fathers, and are specific to the Cynic-Stoic diatribe, rather than performed dialogue as such. 11 It should be noted, though, that the Greek formulae enumerated in Table 1 have been elicited from the most prototypically diatribal sources, such us Epictetus, the apostle Paul, etc. In the Chrysostomic homilies under investigation, they do not all occur equally frequently. This does not mean that these homilies are therefore less diatribal; it just means that the phenomenon of the diatribe is broader and more diverse than would have appeared from the traditional definition. While its scope is wider than Epictetus and Paul, the main feature of the diatribe remains the use of 'speaker metalepsis' and, connected to that, the use of parenthetical verba dicendi to demarcate the different voices in the polemical discourse.
Speaker metalepsis is a term originating in Genette's (2004) narratological theory. Kakridis (2019, p. 248) first applied it to the phenomenon of the diatribe. This application means that the use of the term speaker metalepsis is taken beyond its original narratological meaning intended by Genette and transferred into the realm of rhetoric. In this slightly adapted meaning, speaker metalepsis involves a narrator (in our case, Chrysostom) who not only creates a narrative, but also enters into his own narrative by addressing one or more of his characters. As applied to the diatribe in Chrysostom's homilies, this means that the preacher addresses a fictitious opponent, like a Biblical character, a heretic, a fictitious Jew, or an abstract specimen of his hearers in church. The latter applies to the following example, taken from L13, where the hearers are entreated to give alms to the poor, even if they themselves do not have great riches. In his speaker metalepsis, the preacher challenges an abstract, and therefore fictitious, hearer of the sermon to give alms, even though it be just a little, by analogy with the apostle Peter; the fictitious opponent then voices an objection to Chrysostom's argumentation, which is signalled by φησί, to which the preacher responds again with the brisk and censuring vocative ἄνθρωπε 'man': ( что̀бо̀велико далъ петръ р�ци ми (…) нъ тъ петръ | рече, тоѓо дѣлѧ члч҃ е. не петрови тъ̀о῎ бѣщалъ точїю́(L13; MDA 43, 108v-109r) Both of the preacher's utterances in (1) are cases of speaker metalepsis: he no longer addresses his congregation as a whole, but singles out one prototypical specimen, who is identified by the general vocative ἄνθρωπε / члч҃ е 'o man', to engage in a polemical exchange. We see four of the prototypically diatribal formulae reflected in example (1) (cf. Table 1). We shall come back to this example in (3) below, as there is some interesting variation to be noted between the versions in L and S. Note also the condemnatory tone of the vocative: it not only serves to single out one fictitious interlocutor, but also to censure his (purported) views. The use of the vocative renders explicit that the utterance is addressed to a fictitious opponent, though this is a rather rare strategy in the Zlatostruj. Some of the more explicit vocatives show even more clearly that their function consists in introducing, discrediting and censoring a fictitious opponent in an aggressive, ad hominem way: (2) [01712] Εἰπὲ δή μοι, ἄθλιε καὶ ταλαίπωρε καὶ παντὸς ἀνδραπόδου δουλικώτερε, (In epistolam secundam ad Timotheum homilia VII) 'Tell me then, wretched and miserable one, more servile than any slave,' нъ̏повѣжьⷣ ми грѣшниче и҆ о῎ кан�нѣ, и҆ всѧќого̀сѣмене работнѣи́(L38; MDA 43, 232r) The mere fact that Chrysostom uses the singular would not in itself be sufficient evidence for positing an abstract, hypothetical addressee. However, it is obvious, too, that he does not address his hearers in church in this defamatory way, nor does he address the whole of humanity in a general way, as the "Christian diatribe" has sometimes been claimed to do. 13 These considerations lead us to interpret this phrase as speaker metalepsis, addressed to a fictitious opponent devised by Chrysostom to function as a 'whipping boy' against which he can voice his own convictions in a sharp, polemical tone. The vocative ταλαίπωρε 'miserable one' was already a well-established diatribal formula in the works of Epictetus, and was identified as such already by Bultmann (1910, p. 14). This particular form of address provides conclusive proof that it belongs to the arsenal of markers specific to the Cynic-Stoic diatribe and should not be analyzed along the lines of a mere marker of 2SG or a general feature of orality or dialogue. Thus, the analysis of second-person singular address and speaker metalepsis as a feature of the diatribe is to be traced to its Cynic-Stoic origins and should not be generalized beyond the Hellenistic and Byzantine realm of influence. Parallels with medieval preachers from e.g. the Latin West are at best partial. Although dialogical elements and dramatic performance may be present in e.g. the Italian vernacular sermons of the Franciscan preacher Bernardino da Siena (1380-1444; cf. Berardini, 2010), the notion of polemicizing with a fictitious opponent is absent. 14 The relatively simple, two-tiered division of labour as it is envisaged in speaker metalepsis (i.e. alternating between the preacher addressing his congregation and the same preacher addressing a fictitious opponent) can be enlarged on using Goffman's (1981) and Levinson's (1988) more detailed terminology for speech act participants. This will be done in Sect. 4. We shall now first concentrate on the use of parenthetical φησί and its Slavic equivalents. The comprehensive size of the Zlatostruj collection ensures that the data elicited in this case study can be analyzed quantitatively as well as qualitatively. The full data set can be consulted in the table in the Appendix.

Data selection and discussion
I have investigated 40 homilies that occur in both L and S, to search for the general features of the diatribe set out in Table 1. 22 of these show a full array of diatribal features, at least in parts of the homily. These 22 homilies have, therefore, been selected as a basis for the 13 Thus, Stewart-Sykes (1998) claims that whereas a "Classical" diatribe addresses an individual hearer who "is very present as an entity" (Ibid., p. 69), a "Christian" diatribe addresses a "group" of people who are "completely absent" (Ibid.). Stewart-Sykes bases his ideas on an outdated study by Thyen (1955), who exaggerates the continuity between Jewish and Christian preaching. The latter even goes so far as to claim a "gnomic" meaning of the second-person singular, akin to the "thou" of the Decalogue, addressing the whole of humanity (Thyen, 1955, p. 96). What is worse, Stewart-Sykes does not take into account any of the subsequent research literature on the diatribe. In any case, the alleged dichotomy "Classical" vs. "Christian" is not evidenced in Chrysostom's homilies. 14 Berardini (2010) calls dramatically performed elements "performance indicators", although in the sermons included in her investigation the notion of polemic is markedly absent. The preacher's theatrical performances are claimed to "liven up" his monologue, whereas in the diatribe the dramatically performed dialogues are meant to attain a rhetorical end in a polemical context (i.e. they do not serve a merely didactic purpose). In Bernardino da Siena's sermons, there does not seem to be a fictitious opponent (a heretic, a Jew, etc.). Thus, parallels between the Cynic-Stoic diatribe and Western dialogical strategies in Latin or vernacular sermons are disputable and should not be overstretched. Bernardino da Siena's "dilettissimi citadini" (plural and respectful, cf. Berardini, 2010, p. 84) is very different from "ὦ ταλαίπωρε" (singular and censuring, cf. King, 2018). present investigation. 15 Most of them are taken from Chrysostom's 'commentary series' on some of the New Testament books, most notably Acts, Romans, Corinthians and Timothy. These homilies can be considered a core component of the Zlatostruj proto-collection, and Dimitrova (2015, p. 20) characterizes them as a relatively homogeneous group in terms of translation principles.
This type of Chrysostomic homilies originally consists of two parts, viz. an exegetical part, which provides a verse-by-verse commentary on the Bible text, and an instructive part known as ethicon, which is usually concerned with exhortations pertaining to practical matters of everyday life, rather than with the resolution of doctrinal controversies. Most homilies in the Zlatostruj only contain the ethicon. Thus, it is not surprising that most instances of speaker metalepsis are addressed not to a fictitious heretic (whose views are typically dealt with in the exegetical part), but to an abstract specimen of a random opponent from among the audience in church.
Most diatribal formulae are rather stable throughout the various witnesses of L and S, e.g. видиши ли, человѣче, рьци ми. The use of parenthetical verba dicendi, however, shows a lot more variation. One reason why this is so may be the broad array of usages of φησί, and a felt need to differentiate in Slavic between its quotative and diatribal functions. Whereas φησί in its quotative function (often used to identify Biblical quotes) is almost exclusively translated as рече, the diatribal use of φησί results in a wider array of verbs; the most important Slavic correspondences are enumerated in Table 2; a full overview is given in the Appendix. This variety as opposed to the uniformity of the quotative marker рече clearly shows that the translators were aware of the difference in function between quotative and diatribal φησί and took this difference into account in the translation choices they made. 16 Before we embark on a review of the different verb forms, a general statement is in place about a conspicuous difference between L and S. In Greek, our sample of homilies has only three different parenthetical verbs of the diatribal type, of which the vast majority is φησί(ν) (65 instances), over against only one instance of λέγεις (L2) and two of ἐρεῖς (L18 and L40). In the Slavic translation as transmitted in L, these Greek verbs are represented by 13 different Slavic verb forms. This remarkable variety shows that a fixed equivalent of diatribal φησί had not yet developed in Slavic and should partly be explained by the variety of translators working on the Zlatostruj collection. However, the 13 verbs forms are reduced to seven in S. This reduction is to be explained as a conscious editing practice by the scribe(s). We shall encounter examples below, and explain these scribal choices along pragmatic lines.
What interests us now is not just how Greek diatribal φησί was initially translated into Slavic, but also how it was preserved or modified in the manuscript traditions afterwards. Therefore, Table 2 shows the correspondence of verbs forms between L and S. I shall now discuss some of the main observations to be drawn from this table. Of the 75 diatribal verbs 15 This does not mean that the 18 excluded homilies do not show any dialogical features at all. The phenomenon of the diatribe is more restricted than the broad range of dialogical features that belong to 'orality' in a more general sense. 16 The same distribution can be seen in the Izbornik 1073. Quotative φησί is regularly translated as рече, whereas diatribal φησί shows a range of at least seven different verb forms (Kakridis, 2019, p. 145;2020, p. 260). A similar state of affairs exists in John the Exarch's Hexaemeron (cf. Dekker, 2021b). It might be argued that quotative φησί expresses an impersonal meaning, such as "heißt es, lautet es", as is posited by Bauer and Aland (1988, p. 1707-1708. This is plausible for its quotative use. However, such an impersonal meaning cannot be posited ex nihilo for diatribal φησί. Also, Thyen's (1955, p. 96) unsubstantiated claim about an impersonal meaning for 2SG address in general in diatribal homilies remains unconvincing. Whereas the impersonal, quotative meaning of φησί can be circumscribed as 'it is said', the more subjective stance and polemical burden of a diatribal verb might be rendered more felicitously as 'it is claimed'. This subject will turn out to be relevant to our discussion of the 'distancing' function of diatribal verbs in Sect. 4.  extracted from the 22 homilies under consideration, the numerically strongest forms throughout all manuscripts are рече (3SG.AOR) and речеши (2SG.PRES). рече has so far been considered the 'standard', 'canonical' translation of φησί. This is not denied by the present data, but the strong supplementary use of речеши should not be underestimated. Its use must be explained from a pragmatic point of view. In the following discussion, the main observations drawn from Table 2 will be illustrated and explained by means of examples. This concerns the five major correlations presented in Table 3.
The less frequent forms in Tables 2 and 3 (вѣща, дѣеши) will turn out to be pragmatically most interesting. They tell us most about the scribe's adequate or inadequate understanding of diatribal strategies of textual organization. The most 'canonical' correspondence, viz. φησί = рече, occurs 14 times where all consulted manuscripts agree (plus one as a translation of λέγεις). This is an unsurprising distribution that has been amply discussed on the basis of other texts (Dekker, 2021a(Dekker, , 2021b and thus hardly needs to be illustrated by examples at this point. Three additional instances of рече in L do not occur in S (n/a), because the passages in question have been deleted in their entirety; these are not significant, either. The deletion of passages in S is nothing unusual; it is part of the general 'editorial' principles of S. Two instances of рече in L have речеши in S. This is a significant observation. One more instance of рече is absent from F.п.I.46 but has речеши in VMČ.
The textological considerations discussed in Sect. 1.3 lead to the conclusion that речеши must be an innovation in S in those cases where L has (and, presumably, the original translation had) рече. Of course, the diatribal use of речеши as such is not an innovation in S, as it also occurs in L, although on a somewhat more limited scale. It was already a wellestablished formula in the proto-collection, as reflected in those instances of речеши where L and S correspond, but its use was reinforced in S: nine out of 23 uses of речеши in S are not to be traced back to the original translation. This means that a significant part of its occurrences in S are to be characterized as later copyists' innovations. (Recall that they had no recourse to the Greek text.) The use of речеши as an equivalent of φησί is not unique to the Zlatostruj. Its use has been identified in John the Exarch's Hexaemeron, too (Dekker, 2021b), although to a noticeably lesser extent: of the 15 instances of diatribal φησί, four are translated as речеши, over against nine instances of рече. However, it could not be determined in that study whether the Slavic manuscript tradition showed any development in this respect. This lacuna can now be filled with reference to the Zlatostruj, which shows that речеши was partly a feature of the original translation, and partly the result of a slightly later reworking by Slavic scribes in S. Accordingly, речеши had become a well-established diatribal formula already early on in the Preslav literary centre.
In the original translation, as reflected in L, рече and речеши were co-equal renditions, although рече is slightly more frequent (23 over against 20 tokens). Thus, речеши was a rather widespread and fully acceptable diatribal formula already in L, as it occurs in 10 out of the investigated 22 homilies. At a later stage, reflected in S, речеши even became the preferred form (23 tokens), although рече still remained an acceptable variant (19 tokens). An interesting observation in this respect is that diatribal рече occurs only twice in the investigated homilies from the second part of L, i.e. from L46 upwards, which were incorporated into L from the proto-collection at a slightly later date, after the initial selection of L1-L45 had been made. This concerns one instance each in L72 (which is changed into речеши in VMČ, S44) and L99. This observation underscores that the dominant use of рече was indeed concentrated in the earliest stages of the Zlatostruj's textual history.
Importantly, out of the 23 instances where L has рече, only three were changed into речеши in S (twice in L13, once in L72). This implies that рече was quite an acceptable diatribal formula for the scribe(s) of proto-S, whereas other verbs, such as вѣща and дѣеши, were more likely to be changed into речеши (one and two out of four instances, respectively), due to their relative infrequency. This development is part of a general principle in S, viz. that atypical verb forms are likelier to be exchanged for the more usual diatribal formulae рече and речеши.
The cases where рече was changed into речеши have an important implication for our assessment of the successful adoption of diatribal discourse in Slavic. Although the three cases are not statistically significant, they do show the scribe's adequate understanding of their diatribal function. The scribe correctly identified рече as a marker of the fictitious opponent's voice, i.e. not just of an 'ordinary' quote, and decided to reinforce this interpretation by using the more explicitly diatribal verb речеши, which cannot be used to identify quotes and is, therefore, a more unambiguous marker of the diatribe. 17 The following example is part of the same exchange with the fictitious opponent as in (1), but now in its 'revised' version as represented in S: (3) нъ речеши ми. тъ бо петръ бѣ. да чьто ͞ члвче. не петрови того обѣщалъ тъчию. (S10, F.п.I.46, 18в) Но речеши ми: тъ бо Петръ бѣ. Да что, человѣче? Не Петрови бо того обѣщал точью. (S10; VMČ 1226) If речеши appears in S as a substitute for another verb in L, the eliminated verb can sometimes be рече, as seen in (1) and (3) above, but more often, it concerns verbs that are less frequent as diatribal markers, such as вѣща: (4) Ἀλλ' ἐκεῖνος Ἀβραὰμ, φησὶν, ἦν. Καὶ μὴν σὺ ἐπὶ μείζονα ἐκλήθης σκάμματα. We can see that the final phrase has been omitted in S. In addition, the referential framework of the middle phrase is different in L (а῎ тъ̏н� ⷭ ѣ) and S (а азъ нѣсмь). The version in L seems to be closer to the Greek, which contains a 2SG pronoun σὺ, but the phrase as such is constructed differently even in L, as compared to the Greek edition. I suppose the difference is either due to Greek manuscript variation (μὴν vs. μή) or to the translator misreading μὴν as μή. His already corrupted rendition is then further mutilated in S. In L, the prase а῎ тъ̏н� ⷭ ѣ forms the beginning of the preacher's rebuttal, but in S, а азъ нѣсмь is a continuation of the opponent's objection. The use of вѣща as a translation of diatribal φησί in our sample of homilies is limited to L10. Dimitrova (2016a, p. 148) characterizes the lexicon of L10 as somewhat atypical. Although she does not mention вѣща, it would certainly fit into her list of atypical correspondences in L10. It is, therefore, not completely surprising to see that it was changed into речеши in S53. More significant than the change of verb, however, is the change of tense (aorist → present) and person (3SG → 2SG). These cannot be explained away as the replacement of a relatively rare Preslavism by a more common lexeme. I shall come back to this issue in greater detail in Sect. 4 below.
The following exchange is part of a discourse on the mourning of the dead. The fictitious interlocutor justifies his mourning by referring to the unknown destiny of the deceased one, which is rebutted by Chrysostom by means of a reproving question (5). Chrysostom then insists that the opponent should know the destiny of the dead, according to whether he lived well or not. Once again, the rather exceptional use of вѣща 'said-3SG.AOR' as a diatribal verb is amended in S, where another clarifying word is added (умрый), to make the subject of the phrase ex-plicit. In addition, the position of the parenthetical verb was shifted backward in the sentence. The opponent's нъ, used to introduce his objection (which is a typical diatribal formula, cf. Table 1), is also eliminated in S. This shows that the scribe(s) who reworked S did not always felicitously interpret the diatribal features in the absence of the Greek original. In (5), this is only a minor issue, but there are examples, such as (10), where the elimination or mutilation of diatribal features can even become a drawback to the intelligibility of the discourse structure in S. As follows from Table 2, рече can sometimes be turned into речеши (thrice), but the reverse direction is not attested in our data: where L has речеши, this is never amended to рече in S, which underscores once more that речеши was a fully accepted variety of a diatribal formula. There would seem to be one exception to this rule in VMČ, but the use of a 2PL.PRES form in F.п.I.46 18 and some other peculiarities in VMČ 19 would indicate that S is corrupted. These irregularities cause this particular example to be classified under 'other' in Another verb form requiring some attention is дѣеши, as illustrated by (7). In L45, a large part of the ethicon is devoted to an admonition to obey the priest. The fictitious opponent voices an objection, namely that the priest often does not live up to his own teaching. Chrysostom rebuts that it is not the priest who speaks, but Christ speaks through him: The opponent's objection is identified by a parenthetical φησὶν, which is turned into a 2SG дѣеши in Slavic (L). This is a more explicit signal of a diatribal strategy than the more 'neutral' φησὶν. It is, however, a rather rare diatribal verb (in our sample of 22 homilies, it occurrs only four times in L and twice in S); Dimitrova (2016a, pp. 307-308) considers the use of this verb to be a typical Preslavism. This 'rarity' is mended towards the more usual речеши in S. It should be noted, though, that this is not the only thing that is changed. Cases are rare where only the verb is replaced in S and all else remains as it was. Thus, in this example, it may be argued that the verb is not parenthetical anymore in S. 20 Interestingly, in three of our four cases, дѣеши is used where the Greek edition has no verb. The fourth case (7) is a translation of φησὶν. All four cases are concentrated in a cluster of homilies towards the end of the 'core' part of L, viz. L37, L40 and L45 (2x), on the Acts of the Apostles and Paul's Epistles to Timothy. This may be an indication that this cluster of homilies was translated by the same person. Dimitrova (2016a) characterizes the translation of most of these homilies as "свободен", "не е буквален" etc. She notes the use of the verb дѣти as a peculiar feature when discussing homilies L37 and L45 (Ibid., p. 274; 308). 21 Thus, we can consider the form дѣеши to be a peculiarity of one particular translator, whose idiosyncrasy was eliminated in two of the four cases in S, and exchanged for the more common form речеши. The scribe's intervention in S can sometimes result in a wrong placement of a verb where there is none in L. In the following example, the Greek edition does not have a diatribal verb, nor does L, but рече is inserted in VMČ. Apparently, the scribe misinterpreted the phrase, as if it rendered the fictitious opponent's voice, in which case a diatribal verb would have helped the reader's understanding. In this case, however, the phrase is the preacher's voice responding to the opponent's earlier objection that a priest who misbehaves cannot be ordained of God: (8) [00415] Εἰ γὰρ δι' ὄνου, καὶ διὰ Βαλαὰμ, διὰ μιαροῦ ἀνθρώπου, τοῦ λαοῦ ἕνεκεν ἐλάλησε, πολλῷ μᾶλλον διὰ τοῦ ἱερέως.

(In epistolam secundam ad Timotheum homilia II)
For if he spoke, for the sake of the people, through an ass, and through Balaam, an abominable man, much rather (will He speak) through the priest.
Есть лѝбо о῎ слѧтеⷨ . и҆ валомоⷨ скврьномъ члк҃ мь, людѣи́дѣлѧ бесѣдоваⷯ . то множае паче чиститель (L45; MDA 43, 267r) Елико бо ослятемъ и Валаамомомъ сквер̾ ныимъ человѣкомъ бесѣдовах, рече, то множае паче чистителемъ. (S23; VMČ 1273) 20 Note also that the negation in all Slavic versions (не вѣроуе҆ ши) is due to the translator's flawed understanding of the phrase. The sense in Greek is that if you just trust or obey the priest as such, i.e. as an ordinary man, you do not get a reward in the afterlife. You are supposed to believe him because he is divinely accredited, i.e. Christ is speaking through him. The Slavic translator does not seem to identify this meaning and translates that if you disbelieve the priest, you do not get a reward. 21 Dimitrova (2016a, p. 303) also notes the use of the verb form дѣѥ, as translating φησίν, in L44. This concerns two cases of a non-parenthetical use of the verb. The same form occurs both in L (MDA 43, 265r) and in S (VMČ 1296). In F.п.I.46, the passage is corrupt; the first instance of дѣѥ is left out, the second has дѣ (56в).
Cases like this are relatively easy to explain, as they show that the scribe was concerned to make the text more explicit where the distribution of the different voices might be difficult to disentangle. However, in this case the scribe himself became confused, too. Another clear example is found in L18, where neither L nor the Greek edition have a diatribal verb. Accordingly, the likeliest scenario is that the original Slavic translation did not have a verb, either. The verb (речеши, according to F.п.I.46) must have been inserted in S by the scribe who edited the homilies he selected for proto-S. VMČ has рещи, which is most likely to be a scribal error. This example shows once more that речеши was an active and productive diatribal formula, as it could spontaneously be inserted without any incentive in the scribe's Vorlage. However, in spite of an added diatribal verb, it remains unclear who says what, even when taking the wider context into account. After all, речеши can be meant to refer back to the first phrase or forward to the second, and punctuation in F.п.I.46 is unreliable and inconsistent. Thus, even though the scribe's objective in adding a parenthetical verb must have been to clarify the passage, his scribal intervention does not necessarily result in greater clarity. The opposite direction is harder to explain, i.e. where L has a diatribal verb which is absent from S. The first assumption that might come to mind, based on the textological considerations discussed above, is that the editor of proto-S just chose to leave the verb out, for reasons best known to himself. This is, however, not very probable. Why would a scribe on purpose leave out a disambiguating element, when this deletion makes the text less understandable for its intended readership? The following sentence exemplifies that the leaving out of the verb often goes hand in hand with the elimination or attenuation of other diatribal elements: (10) [01832] Ἀλλ' οὐκ ἤνεγκάν τινες, φησὶ, τὴν χηρείαν, ἀλλὰ συμφοραῖς περιέπεσον· (In epistolam secundam ad Timotheum homilia VII) 'But some, he says, did not endure widowhood, and have fallen into many misfortunes.' нъ̏не стрьпѣша рече мноśи в�довьства. вь напасти в�падоша (L38; MDA 43, 235r) не сътьрпѣша бо мноꙁи въдовьства. нъ въ напасти въпадоша. (S61; F.п.I.46, 171а) не стерпѣша бо мнозѣ в̾ дов̾ ства но в напасти в пасти впадоша (S61; VMČ 1469) The diatribal formula ἀλλὰ / нъ has been eliminated from S. Instead, the conjunction бо is added, which establishes a continuity in argumentation, as opposed to the contrastive conjunction нъ, which serves to introduce an objection, i.e. an opposing line of thought. Thus, the logical coherence of the text is ruined in S. 22 Instances like this are typical of S, where not just the verb is changed, but also other changes tend to be made to the structure of the dialogue. There are 25 cases where L and S have different verbs. Apart from little differences in spelling or word order, only 8 instances show no significant deviations other than the verb itself. Thus, in 17 out of 25 cases, there are deviations in the referential structure of the sentence, or stretches of text are left out (usually in S). The most important observation so far has been that the less typical forms in L, such as дѣеши and вѣща, tend to be suppressed in S and replaced by other, more common forms, most notably рече and речеши. The variation in L shows that different translators must have been at work: the Zlatostruj as a whole is certainly not to be traced back to one translator. This is in accordance with statements in the research literature (e.g. Dimitrova, 2016a, p. 450).
Some clusters can be identified, based on the sequence of homilies on each New Testament book. Thus, for instance, the six investigated homilies from the series on the Acts of the Apostles never contain diatribal рече, but seven instances of речеши and seven other diatribal verb forms. In the three homilies from the sequence on Romans, on the other hand, речеши does not occur, but diatribal рече occurs 12 times. These are just some preliminary observations; more research into the composition of the original Zlatostruj is needed. The whole work needs to be taken into account; even a substantial sample like our 22 homilies is not quantitatively significant enough to assign homilies or clusters of homilies to individual translators.
We have so far not discussed those instances where no parenthetical verb is present at all. The differentiation of voices remains implicit there. There is no way of investigating this phenomenon quantitatively, as there are no tangible linguistic parameters involved according to which the instances might be measured. The voices just follow each other in their logical sequence, and the reader/hearer has to isolate them by referring to the contents and context. In the oral delivery of his homilies, the preacher would have used intonation to demarcate the voices. In the written sources that have remained, only a logical analysis of the polemical positions can lead to a hypothesis about the different voices. This task is not undertaken in the present study. We concentrate on the explicit disambiguation strategies provided by parenthetical verbs.

The pragmatic functions of diatribal parenthetical verbs
Building on the quantitative data and the ensuing observations made in Sect. 3, I shall now proceed to make the following two arguments: (a) 2SG forms (e.g. речеши) as used in diatribal discourse are not pragmatically equivalent to 3SG forms (e.g. вѣща, рече), (b) the development in the use of Slavic diatribal verbs, as discussed in Sect. 3, led to an enhancement of some of their diatribal functions.
In doing so, I want to enlarge on the two main functions of the hitherto discussed parenthetical verbs, viz. (i) disambiguating and (ii) distancing, from a pragmatic point of view. Again, marry at all than to marry, much more in this case. "But some, you say, could not endure widowhood, and have fallen into many misfortunes". Yes; because they know not what widowhood is" (transl. Schaff vol. 13, p. 503). The underlined phrase corresponds to (10) and shows clearly that the sense of the sentence is an objection that stands in contrast to the preacher's line of argumentation. Neither the elimination of нъ and рече, nor the insertion of бо is, therefore, conducive to the text's felicitous interpretation.
our focus is not primarily rhetorical, because we start from the linguistic elements on the micro-level and determine their function in discourse. The first function of parenthetical verbs in their diatribal usage (i.e. distancing) allows the preacher to position himself polemically over against the opponent's point of view. He uses parenthetical verba dicendi to signal his disapproval for the position attributed to the opponent and distance himself from it. The second function (i.e. disambiguation) is to help the hearers identify the dramatically performed voice as belonging to the fictitious opponent, and not to the preacher himself. Thus, one function places the preacher over against the attacked position; the other positions him in relation to the hearers in church. These two functions will turn out to be especially relevant when discussing the use of 2SG forms. Whereas essentially all diatribal verbs share the two functions (although to different degrees), the 2SG forms are better suited for both functions.
As has been demonstrated by the data discussed in Sect. 3, the use of 2SG parenthetical verb forms in the Zlatostruj is to be traced back to two different backgrounds, viz. the original translators and later Slavic scribes. It is, therefore, expedient to give a functional explanation for the more widespread use of 2SG forms in the Slavic translation generally, i.e. as compared to the Greek original, and the (albeit moderately) increased preference for 2SG forms in the Slavic manuscript tradition as represented by S. It was a Slavic innovation initially implemented by the translators, but reinforced and stabilized by later scribes. The reason for this was not grammatical, for there was an already well-established alternative (рече), which corresponded more closely to φησί in terms of grammatical equivalence.
The increasing preference for 2SG forms, most notably речеши, has some significant implications for the theoretical analysis of the fictitious opponent's voice in the field of pragmatics. 23 The change seems trivial, but when a 2SG form is substituted for a 3SG form, the parenthetical nature of the verb remains the same on the syntactic level, though not on the pragmatic level. This statement requires some explanation.
A first useful terminological distinction can be made based on Clark's (1996) theory of "layering". He distinguishes two layers "in all nonserious actions" (1996, p. 353). When the preacher addresses a fictitious opponent, he is acting on the fictitious level of Layer 2. The discussion on this fictitious level does not occur in real life (the opponent does not necessarily exist at all, or is long dead, absent or an abstract specimen of the hearers in church), and is therefore "nonserious" in Clark's terminology. The fictitious layer is embedded into the real-life setting of Layer 1: everything on this level is "serious" (1996, p. 354), i.e. what the preacher is really doing, viz. delivering a sermon to his hearers in church. His hearers in church are expected to know when the preacher is addressing them as a congregation, and when he is enacting a polemical discourse on the fictitious level. In Clark's terms, the hearers are capable of interpreting "layered meaning" (1996, p. 359 Thus, the fictitious level (Layer 2) is 'interrupted' by φησί. The author is enacting the fictitious opponent's voice, but interrupts his enactment in order to signal to his audience in church (Layer 1) that he is enacting the opponent's voice. Thus, φησί is not part of the opponent's voice and therefore does not function on the fictitious level. The same applies to Slavic 3SG forms such as рече. When a 2SG form is used, as in the Slavic translation of (11), both in L45 and S23, the theoretical constellation is somewhat more complicated. The verb is still not a part of the opponent's voice, and therefore parenthetical, but the verb's 2SG addressee is the fictitious interlocutor, so that the fictitious level as such is not interrupted. At the same time, the pragmatic function of the verb is geared towards the actual audience in church: they are to understand whose voice is enacted, and the verb is inserted for their benefit. Thus, the parenthetical verb is moved from the real-life level to the fictitious level, although its main function remains to play a role on the real-life level. 24 Interestingly, the opponent's answer to the preacher's question in (11) is rephrased as a question in Slavic by the particle ли. 25 Can it be that the translator interpreted this as a rhetorical question by the preacher? Or can it still be the opponent's reply, even though it is a question? The next example decisively shows that the opponent's answer (as signaled unequivocally by речеши) can be rephrased as a question in Slavic: I want to emphasize again at this point the importance of the distinction between quotative and diatribal parenthetical verbs. This distinction has implications first of all for the verbs' disambiguating function. Our whole discussion about layering, and the fictitious level, is not applicable to ordinary (Biblical) quotes, because they are not enacted dramatically, as 24 Goffman's notion of "frame" is applicable to speech acts that Austin dismisses as "non-serious" (McCawley, 1999, p. 618). In this sense, Chrysostom's performed exchanges with a fictitious opponent are a clear example of "framed" discourse: it occurs within another frame. This terminology is compatible with Clark's (1996, p. 354), who visualizes his layers by drawing a pyramid of rectangles, i.e. Layer 2 is on top of and enclosed by Layer 1. In these terms, речеши is 'framed' (or 'layered'), рече is not. The former involves an extra layer in the communication, because it is addressed to the fictitious opponent on Layer 2. Paradoxically, this more complex configuration more easily facilitates a correct interpretation for the audience, thanks to the enhanced disambiguating function of речеши. 25 Note a similar use of ли in (9). diatribal dialogue is. Thus, the diatribal verbs in examples (11) and (12), as well as the other examples discussed in Sect. 3, differ from the well-known, 'ordinary' quotative parenthetical рече (cf. Kamčatnov, 2004) in that they have a more narrowly defined sphere of use. This difference comes to the fore especially when 2SG forms are used. An ordinary quotative use of parenthetical 2SG forms such as речеши has not been attested in the literature so far. This implies that 2SG verbs are more specific to the diatribe and, therefore, better suited to their discursive task: they signal not just the presence of another voice in the discourse, but specify more unequivocally that it represents the fictitious opponent's voice within the diatribe. In other words: the fictitious opponent's objection, identified by a 2SG form, such as речеши, cannot be misinterpreted as a mere quote. This points to a more felicitous performance of the first function of diatribal verbs, viz. disambiguation. Their use is less ambiguous than that of, e.g., рече, which is a rather general marker of any non-authorial voice and has a broad bandwidth of applications: it can mark a Biblical quote (which the preacher adduces to support his own argumentation) as well a fictitious opponent's heretical objection (which the preacher then has to rebuff). That leads us on to the second function, viz. that of distancing. It is here that the difference between quotative and diatribal verba dicendi appears even more clearly. They serve not only to disambiguate voices for the benefit of the audience, but also to convey a subjective stance towards the content of the stretch of represented speech.
In previous research on the diatribe in the Slavic realm, I have so far employed Genette's (2004) narratological theory of speaker metalepsis and Clark's (1996) theory of conversational pragmatics as tools for analyzing the fictitious layer of diatribal discourse. These approaches, though useful in themselves, can benefit from being supplemented by Goffman's (1981) theory of participant roles and its more linguistically-oriented sophistication by Levinson (1988). Now, if we analyse examples (11) and (12) in terms of Goffman's "participation roles", but slightly modified by Levinson (1988, p. 169), who divides them into "production roles" and "reception roles", we arrive at the following considerations.
When Chrysostom enacts the fictitious opponent's voice, the latter, even though he does not exist in real life, is the "principal" (the one who adheres to the ideas expressed). At the same time, Chrysostom always remains not only the "animator" (the one who performs the voice) but also the "author" (the one who puts the ideas into words). This implies that he is completely in charge. The opponent's objection is seen through the preacher's prism. 26 The fact that Chrysostom is the "author" makes him susceptible to distort (for rhetorical reasons) the viewpoint of the fictitious opponent whose voice he is enacting. This distortion is most obvious when the objection comes from a fictitious heretic, whose views are censured more briskly than those of an abstract specimen of the hearers in church, who is rather treated as an ignorant but benevolent conversation partner. As I pointed out already, the latter is more prevalent in most of the ethicon parts of the homilies investigated for this study. Still, evidence can be found that the fictitious opponent's objections are sometimes skewed to such an extent that it is unrealistic to assume such on objection to be based on a real-life context. The preacher uses (or abuses) this distortion for his own rhetorical ends, viz. imposing orthodox doctrine and practice on his hearers by exaggerating alternative points of view. A fictitious opponent is obviously unable to step in and rectify this unfair treatment. 'But you will say: What is the use to me of toil and hard labour, when it is possible to be acquitted of all (my sins) without fatigue?' но̏речеши на коую῎ ми потрѣбоу стртⷭ ь и҆ томленїе. а҆ могᲈщоу без� троуда разрⷣ ѣшити (L40; MDA 43, 253v) This example reflects a theological position that Chrysostom would hardly have attributed directly to any of his hearers in church. If any of them were slothful enough to hold this position, they would nevertheless not have phrased it in such an outright way. Thus, on the one hand, the intermediary function of the fictitious opponent allows the preacher to use exaggeration (hyperbole) when presenting the opposing point of view, so that he can contend that there is no reasonable alternative to his own orthodox views; on the other hand, it allows him to attack his congregation in an indirect way (attenuation), because the one who gets the punches is but fictitious. These two functions of the fictitious opponent coincide nicely with the two aforementioned functions of parenthetical verbs within the diatribe. The linguistic expression of distancing resides primarily in the parenthetical verb. 27 Again, this is a function of diatribal verbs as opposed to quotative ones. The latter have the opposite function, viz. to support the author's argument as he adduces a favourable quote. In the case of both φησί and рече, which can also be used as an impersonal quotative marker, this subjective stance has to be deduced from the context. In the case of речеши, which is less neutral and has a more overtly polemical tone, the distancing function is verbalized more explicitly, thanks to their more exclusively diatribal function.
The pragmatic functionality of речеши is thus straightforward: it serves as an explicit signal of the performance of a fictitious opponent's objection, rather than the reproduction of a quote. This is a feature of the diatribe that is consistently developed in Slavic and taken further than its equivalent in the Greek original. (Parenthetical λέγεις and ἐρεῖς do exist, but they occur only very sparingly; respectively once and twice in our sample of homilies.) Chafe (1982, p. 46) carefully posits that the use of second-person forms is "a symptom of involvement". This is an aspect of the diatribe that shows a significant development on Slavic soil, with the increase of 2SG forms. Involvement means that the speaker is more directly involved in immediate interaction with the audience or addressee(s), so that the discourse is of a more dialogical nature.
To return to the two statements made at the beginning of this section: we have argued that there is indeed no pragmatic equivalence between 2SG and 3SG forms. This does not mean that there is a difference in terms of e.g. speech acts, but both relate differently to the fictitious and real-life levels that are so central to a correct interpretation of the diatribe. In addition, the disambiguating and distancing functions come to light to different degrees in each form. Thus, both functions being represented more felicitously in речеши, this may be part of the explanation for why it gained popularity as a diatribal marker (a) in Slavic over against the Greek original and (b) in S over against L at the detriment of some (rare or frequent) 3SG forms, as was established in Sect. 3. Both рече and речеши form the core components of a more homogeneous set of diatribal parenthetical verbs. A more uniform set of verbs means that they are more distinctive and thus better recognizable as a specifically Slavic manifestation of the Cynic-Stoic diatribe.

Concluding remarks
We have seen that the initial variety of parenthetical verb forms in L is partially suppressed in S. This homogenization results in an increased conventionalization of the most common formulae, viz. рече and речеши. The diatribal use of 2SG forms such as ἐρεῖς was also known in Greek, but is, at least in Chrysostom's homilies, by far not so widespread as its Slavic counterpart речеши. The latter can, therefore, be characterized not so much as an absolute innovation in Slavic, as rather the reinforcement of a phenomenon of which the germ was already available incipiently in Greek. The Slavic translators of the Zlatostruj proto-collection picked this element up and established it as a solid feature of the diatribe in Slavic. Later scribes of S reinforced its use. It is, therefore, justified to speak of the development of a distinctive feature of the Slavic diatribe in its own right. The increased use of 2SG forms, which had been observed in its incipient stages in other (Old) Church Slavonic texts, has now been solidified in the Zlatostruj collection, most notably as a prominent innovation in S. At the same time, the already well-established formula рече remained prominent as a diatribal marker as well; it is pragmatically more akin to Greek φησί, both in its quotative and its diatribal use.
We discussed two functions of parenthetical verbs in their diatribal usage, viz. distancing and disambiguation. Although these functions come to the fore in all diatribal verbs, the 2SG речеши represents these two functions better than most alternative verb forms. Thus, its increased use in Slavic shows that the diatribal element of parenthetical verbs has been successfully received, perpetuated and reinforced in Slavia orthodoxa.
Topics for further research include other diatribal formulae in the Zlatostruj. The parenthetical verbs are the main diatribal feature, but there are others, as shown in Table 1. These should also be systematically investigated. In addition, the argumentative and rhetorical structure of fictitious dialogical exchanges without parenthetical verbs should be charted and analyzed. Most importantly, the topic of the diatribe should be pursued in its further development in the Orthodox Slavic realm. To this end, several original works (i.e. not translated from or based on Greek works) should be investigated, most notably the works of Maksim Grek and Iosif Volockij (Joseph of Volokolamsk). The present author is currently working on the latter's original Slavic compositions.

Appendix: Table of diatribal verbs in the Zlatostruj
Note: '/' refers to the absence of a diatribal verb. 'n/a' refers to cases where the entire corresponding section of text is absent from the manuscript in question. The two leftmost columns refer to the number of the homilies in L and S. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.