How Style Met the City

1 What is style?

Style is an elusive and multifarious concept, despite being almost ubiquitous in the classical studies.^[1] As a notion, style is intuitively rooted in the perception of contrasts – that between ‘form’ and ‘content’, λέξις and διάνοια (e. g., Arist. Rh. 1404a18), τὸ λεκτικόν and τὸ πραγματικόν (D.H. Pomp. 4) – in short, a ‘what’ and a ‘how’ – and those between many discernible ‘hows’ that may be applied to the same ‘what’.^[2] In O’Sullivan’s (1992, 3) words, “the notion of stylistic distinction is […] implicit in the very idea of style”.^[3] Traces of these intuitions certainly help us track the emergence of an awareness of style to the dawn of Greek rhetoric,^[4] but such an understanding of style gives rise to several problems under close scrutiny. For one thing, form and content are hardly ever completely separable from each other.^[5] For another, one may argue that style is an intrinsic feature of any text regardless of the identification of a style by contrast with other styles, much in the same way as language is a feature of an utterance regardless of the identification of the language in which the utterance is produced.^[6]

A relatively recent discussion of style may serve as a good illustration of such difficulties. In his book on the notion of sublimity in antiquity, Porter takes issue with Russell’s remark that the sublime is “a special effect, not a special style”^[7] and rightly notes that if effect it is, it is one of style; the sublime – as he argues – does not ‘transcend’ style but is a style, so long as we do not understand style as “a textbook abstraction […] reducible to a checklist of criteria and rules”.^[8] Porter fundamentally identifies the sublime with the ‘grand’ style that figures in all ancient classificatory systems – from the ‘grand’ vs ‘simple’ dichotomy of early rhetoric^[9] to the tripertita varietas (Cic. Orat. 70) that dominates Hellenistic and Roman rhetoric and the quadripartite division of ‘Demetrius’ (e. g., Eloc. 36–37)^[10] – but seems unhappy with the conceptualization of ancient divisions of style as purely formal sets of rules concerning language use. In his view, the genus grande of ancient rhetoric “is not reducible to style because it includes the totality of what language delivers, from thought and the characteristics of ‘the soul’ […] to what ordinarily gets filed under ‘content’”.^[11]

I shall not venture too far into the discussion of Porter’s argument, which is not uncontroversial,^[12] but I will briefly note that ancient discussions of sublimity do not normally present it as a sustained ‘style’ to be identified throughout a text or a passage, but rather as a psychological reaction produced by a combination of subject matter and formal devices (cf. [Longin.] Subl. 8.1) at specific points in a text – “‘highlights’ that stand out from their contexts”, as de Jonge (2020, 157) puts it – which Russell’s notion of a ‘special effect’ seems to capture quite adequately. For example, sublimity (ὕψος) is listed by Dionysius of Halicarnassus (Th. 22–23) among the “ancillary” (ἐπίθετοι) virtues of speech (ἀρεταὶ τῆς λέξεως) – qualities that need not be present in all expressions (unlike the ἀναγκαῖαι ἀρεταί) but may serve to showcase the writer’s ability.^[13] In Dionysius’ critical method, virtues of speech form a system that functions as a framework for the description of the style of each writer (cf. D.H. Th. 21, Lys. 13, Pomp. 3).^[14] At the same time, Dionysius adopts a tripartite classification of styles – the simple, the grand, and the mixed style (which results from the combination of the qualities of the two polar styles, Dem. 1–3) – and presents the writers he surveys as representatives of each of them. Lysias is the champion of the simple style, Thucydides and Gorgias epitomize the grand style, and Isocrates, Demosthenes, and Plato offer good examples of the mixed one.^[15]

In principle, individual virtues are not connected to any style in particular, but it is true that some tend to be more at home in one style than the others. Necessary virtues should be present in all styles, but clarity (σαφήνεια), for example, is easier to preserve in the simple style. In Dionysius’ critical works, ὕψος normally goes together with virtues that point to the grand style (e. g., καλλιρημοσύνη “elegance”, σεμνολογία “solemnity”, and μεγαλοπρέπεια “grandeur”).^[16] However, nothing stops such virtues from figuring in the mixed style, which has no preferential virtues of its own and displays both the persuasive and pleasant qualities of the simple style and the penchant for emotion and impressiveness of the grand style (see e. g., Dem. 4 on the style of Isocrates). From this perspective, sublimity may not be confined, nor correspond, to the grand style.

To sum up, if we conceive of ‘style’ as ‘form’, it would be odd to say that sublimity is not a style, in that it is possible to pinpoint technical, formal devices that help to achieve (and recognize) sublimity in expression. Yet sublimity may not be called ‘a style’, if by style we mean a ‘type of style’ identifiable as an item in a classificatory framework. To complicate things, we may add that a strict understanding of ‘style’ as ‘form disconnected from content’ may fail to capture the true nature of the types of style singled out by ancient rhetoricians, as Porter would have it.

As Porter himself notes, much of the problem lies with terminology.^[17] What we call ‘styles’, the Romans called genera dicendi (e. g., Cic. Orat. 69–70) and the Greeks referred to as χαρακτῆρες^[18] λέξεως (e. g., D.H. Dem. 2) or τῆς ἑρμηνείας (e. g., Demetr. Eloc. 34), or φραστικοὶ χαρακτῆρες (Marcellin. Vit.Thuc. 39). There is no hypernym that refers both to ‘style’ as an (uncountable) phenomenon and to its (countable) types. It is telling, for example, that Aristotle often qualifies λέξις using adjectives that we may easily construct as characterizing types of style (ποιητική Rh. 1404a25, εἰρομένη/κατεστραμμένη Rh. 1409a24–27, γραφική/ἀγωνιστική and δημηγορική/δικανική Rh. 1413b3–4) but never uses this term in the plural in the Rhetoric or in the Poetics. In fact, λέξεις normally means “words” in Greek; the plural ἑρμηνείαι can in turn be interpreted as “stylistic qualities” or “powers of expression”, not as “styles”.^[19]

The original sin at the root of the overextension of the meaning of ‘style’ lies in the origin of this term. Latin stilus simply means “pen”. Its figurative use – by which words meaning “pen” are employed as a metonymy for “writing skill”, as is common in many European languages – appears as early as in Cicero (Or. 44).^[20] Crucially, stilus does not function as a synonym for genus dicendi but refers to (uncountable) ‘style’ as a feature of texts. At the same time, as theses expressions show, Greek rhetoricians and critics had quite a few options in their choice of a word to refer to (uncountable) ‘style’. These include terms such as λέξις, ἑρμηνεία, or φράσις, which may have broader or narrower meanings pointing to slightly different aspects of ‘style’. Λέξις refers specifically to wording and, because of the polysemy of the root of λέγω, its meaning may occasionally correspond to that of ἐκλογὴ τῶν ὀνομάτων^[21] (Porter renders it as “speech”).^[22] Ἑρμηνεία, in turn, is often rendered as “expression” and, when employed as a technical term, it would make it unambiguous that more than the choice of words is at play while still being interchangeable with λέξις in its broader sense. Etymologically, this term is connected to ἑρμηνεύς (“interpreter, translator”), which evokes the image of ἑρμηνεία as the trait d’union between thoughts and their verbal expression.^[23] “Expressive possibility”, a translation suggested, once again, by Porter, would be an appropriate rendering of this meaning.^[24] Φράσις, which appears, among others, in Dionysius of Halicarnassus and Quintilian,^[25] also seems to be interchangeable with λέξις, but in Diogenes Laertius’ account of the Stoic classification of ἀρεταὶ λόγου (7.59) these two terms seem to differ in their level of abstraction – φράσις would indicate a “way of speaking” whereas λέξις would refer specifically, once again, to the actual wording of a verbal expression.^[26]

In spite of their variety, the Greek expressions that we more often than not translate as “style” fail to capture the same wide range of phenomena that the offspring of the Latin notion of stilus has grown to be able to.^[27] We can talk about ‘style of management’, ‘style of persuasion’, ‘historiographical style’, ‘oratorical style’, and so on^[28] – in short, we can call ‘style’ all that can be perceived as an abstractable formal feature, intuitively (but not necessarily practically) suitable for description, and potentially categorizable – but the Greek notions of λέξις, ἑρμηνεία, and φράσις only correspond to what Dover calls “linguistic style”^[29] – a restricted, specific ‘stylistic’ phenomenon. However, the fact that Greek lacks a truly coextensive term to the Latinate cross-linguistic family spawned by stilus should not make us think that the Greek sensitivity to style and stylistic variation was limited to the linguistic aspects of mere elocutio. We only need to look beyond technical discussions of style in the ancient rhetorical tradition.

If we think about paratragedy, for example, we can certainly agree with Dover that it consists in either direct quotation of tragedy or “composing for humorous purposes in a style easily recognized by the audience as tragic – by virtue of its vocabulary, syntax, and paucity of resolved long syllables” and that the comic effect results from “incongruity of style”^[30] – a mismatch between the ‘what’ and the ‘how’. However, the ‘how’ is not limited to elocutio but also includes inventio, so to speak. Take for instance the way in which Blepyrus laments the loss of his Assembly wages in Aristophanes’ Assemblywomen (391–395):

    οἴμοι δείλαιος.

Ἀντίλοχ’, ἀποίμωξόν με τοῦ τριωβόλου

τὸν ζῶντα μᾶλλον· τἀμὰ γὰρ διοίχεται.

Oh wretched me! Antilochus, don’t cry for the three obols, cry for me, still alive! For I have lost everything that was mine (transl. Farmer 2017, 67).

As Farmer comments, here the ‘tragic style’ is alluded to through several features of λέξις:

ἀποιμώζω and διοίχομαι are tragic vocabulary; there is a stylized poetic artificiality to the distorted word order, with με separated from τὸν ζῶντα by the phrase τοῦ τριωβόλου, the meaning of whose genitive case becomes clear only with the final μᾶλλον. Characters in tragedy are always receiving bad news, and when comic characters get bad news, they tend to react with tragic language; elevated language over the loss of a mere three obols is funny enough on its own.^[31]

However, ‘tragic style’ here is not limited to ‘tragic language’, but it also includes the character’s ‘tragic’ reaction to such a trivial situation. With the ‘what’ being equal – regret for a missed day’s wage – out of the many ‘hows’ that Aristophanes could have chosen for its verbal expression (ἑρμηνεία, if we like), he picked the ‘tragic’ one, which involves saying tragedic things (‘what’ Blepyrus says) in a tragedic manner (‘how’ he says it). ‘Tragedic things’ represent a second-order ‘what’ which is itself abstractable, describable, and categorizable as a ‘how’ (as opposed to, say, ‘comedic things’ or ‘forensic things’), and may thus be analysed as a fact of style, albeit at a different level from ‘linguistic style’.

A rather explicit formulation of this multi-layered sense for style can be traced in an illuminating passage from Isocrates’ Panegyricus (8–9):

ἐπειδὴ δ’ οἱ λόγοι τοιαύτην ἔχουσι τὴν φύσιν ὥσθ’ οἷόν τ’ εἶναι περὶ τῶν αὐτῶν πολλαχῶς ἐξηγήσασθαι καὶ τά τε μεγάλα ταπεινὰ ποιῆσαι καὶ τοῖς μικροῖς μέγεθος περιθεῖναι, καὶ τά τε παλαιὰ καινῶς διελθεῖν καὶ περὶ τῶν νεωστὶ γεγενημένων ἀρχαίως εἰπεῖν, οὐκέτι φευκτέον ταῦτ’ ἐστὶ περὶ ὧν ἕτεροι πρότερον εἰρήκασιν, ἀλλ’ ἄμεινον ἐκείνων εἰπεῖν πειρατέον.

αἱ μὲν γὰρ πράξεις αἱ προγεγενημέναι κοιναὶ πᾶσιν ἡμῖν κατελείφθησαν, τὸ δ’ ἐν καιρῷ ταύταις καταχρήσασθαι καὶ τὰ προσήκοντα περὶ ἑκάστης ἐνθυμηθῆναι καὶ τοῖς ὀνόμασιν εὖ διαθέσθαι τῶν εὖ φρονούντων ἴδιόν ἐστιν.

But since oratory is of such a nature that it is possible to discourse on the same subject matter in many different ways – to represent the great as lowly or invest the little with grandeur, to recount the things of old in a new manner or set forth events of recent date in an old fashion – it follows that one must not shun the subjects upon which others have spoken before, but must try to speak better than they. For the deeds of the past are, indeed, an inheritance common to us all; but the ability to make proper use of them at the appropriate time, to conceive the right sentiments about them in each instance, and to set them forth in finished phrase, is the peculiar gift of the wise (transl. Norlin 1928, 123–125).

Isocrates demonstrates an awareness of the separation between ‘form’ and ‘content’. ‘Form’ here, however, is a rather broad notion and is not limited to the verbal expression of thoughts. A good orator is one who is able to accomplish three tasks: (1) “to make proper use of the facts at the appropriate time” (τὸ ἐν καιρῷ ταῖς πράξεις καταχρήσασθαι), (2) “to conceive the right sentiments about them in each instance” (τὰ προσήκοντα περὶ ἑκάστης (τῆς πράξεως) ἐνθυμηθῆναι), and (3) “to set them forth in finished phrase” (τοῖς ὀνόμασιν εὖ διαθέσθαι). The first of these tasks falls neatly within the scope of inventio, but it is still presented as an abstractable feature subject to variation – it is something one can be better at than others – in other words, it is part of the ‘how’.^[32] The third of these tasks maps without difficulty onto λέξις or ἑρμηνεία, whereas the second can be pictured as the missing link between thought and expression – another component of the ‘how’ that determines, but lies behind, the actual wording.^[33]

The identification of these three tasks closely resembles the identification of ἔννοιαι, μέθοδοι, and λέξεις as components of stylistic categories (ἰδέαι λόγου) in Hermogenes’ On Types of Style. In Hermogenes’ system, each ἰδέα^[34] carries its own ‘thoughts’. For example, “purity” (καθαρότης) needs easy thoughts, “solemnity” (σεμνότης) comes with topics such as gods or illustrious deeds, while the subject matters of “plainness” (ἀφέλεια) are simple characters and trivial, ordinary things (e. g., animals, plants, and the like).^[35] At the same time, each ἰδέα has its own λέξεις (“word(ing)s”), which result from the combination of specific word-choice, figures, manners of segmenting discourse into κῶλα, manners of combining words with each other euphonically (συνθέσεις), rhythms, and cadences (i. e., clausulae).^[36] As to the μέθοδοι, Hermogenes presents them as the ways in which a thought can be handled before being formulated verbally (Id. 219.1–8):

ἔστω γὰρ βούλεσθαι ἡμᾶς ποιῆσαι γλυκύτητα. οὐκοῦν ἔννοιαι μὲν γλυκύτητος αἵ τε μυθικαὶ καὶ αἱ ταύταις ὅμοιαι καί τινες ἕτεραι, περὶ ὧν ὕστερον λέξομεν ἐν τῷ περὶ αὐτῆς τῆς γλυκύτητος λόγῳ. ἀλλ’ ἔννοιαι μὲν αὗται, μέθοδοι δὲ τὸ προηγουμένως καὶ σὺν ἀφηγήσει αὐτὰς διεξιέναι, ἀλλὰ μὴ ἐξ ὑποβολῆς ἤ πως ἑτέρως.

Suppose you want to create Sweetness. The thoughts that are characteristic of Sweetness are those that are related to mythology and similar topics, which we shall discuss later in the section on Sweetness. The μέθοδοι involve dealing with the topic as the principal theme and in a narrative fashion rather than treating it allusively or in some other indirect way (transl. Wooten 1987, 4, adapted).

The concept of μέθοδος, “the strangest stratum in Hermogenes”, as Rutherford calls it,^[37] is remarkably similar to Isocrates’ second task (τὰ προσήκοντα περὶ ἑκάστης τῆς πράξεως ἐνυθμηθῆναι), albeit at a higher level of abstraction and, seemingly, independence from the πράξεις themselves.^[38] Μέθοδοι have the ability to transform the quality of a thought according to rhetorical or stylistic needs.^[39] Harsh, offensive thoughts can be presented sweetly, mild thoughts may be presented forcefully, and so on. The realm of μέθοδοι also includes the ordering of ἔννοιαι – which fits more naturally within dispositio rather than elocutio but nevertheless is an element that Greek critics had been taking into account in their judgments of style. Several discussions of clarity (σαφήνεια) in ancient rhetoric, for instance, draw a distinction between clarity concerning words (ἐν τοῖς ὀνόμασιν) and clarity concerning the subject matter (ἐν τοῖς πράγμασιν),^[40] which can be achieved precisely through an effective ordering of thoughts in discourse.^[41]

Another interesting concept that reveals how the sensitivity of Greek critics for the ‘how’ was not restricted to ‘linguistic style’ is that of ἀπαγγελία – a word which does not appear at all in Hermogenes.^[42] Even though it is certainly not coextensive with stilus, this is arguably the Greek term that comes closest to conveying a broad understanding of ‘style’ as an all-encompassing ‘mode of expression’ including, but not limited to, the mise en parole of the subject matter.

Roberts includes ἀπαγγελία in his survey of Greek words for ‘style’ with the brief remark that this word is “properly used of narrative and especially of historical narrative” and “is another late word for ‘style’ in general”.^[43] However, ἀπαγγελία has a long history in Greek rhetoric and literary criticism.^[44] This word means “narration” in Aristotle (Po. 5 1449b11, 6 1449b27) and “report of past events” in the Rhetoric to Alexander (30.2 1438a8, 30.11 1438b10, 31.3 1438b23), where it may also refer to the corresponding section (the narratio) of forensic speeches (36.16 1442b29). Later on, in a fragment of Theophrastus (fr. 686 F = Demetr. Eloc. 114) whose wording may well be ascribed to ‘Demetrius’, this term shows up with the meaning of “manner in which a thought is expressed”.^[45]

In the critical essays of Dionysius of Halicarnassus, ἀπαγγελία may refer specifically to narrative (Th. 9) and to how its contents are expressed verbally (e. g., τὸ τάχος τῆς ἀπαγγελίας, Th. 24 and Comp. 20.16 Aujac-Lebel), but its use is not limited to discussions of this type of discourse. For example, Dionysius’ talks about the ἀπαγγελία of a passage from Demosthenes’ On the Symmories (13) as he describes how he “adorned and constructed” (κατεσκεύακέν τε καὶ ἐσχημάτικεν) a νόημα (Th. 54). Similarly, Dionysius uses ἀπαγγελία to refer to the manner in which νοήματα and the linguistic κῶλα that express them are structured (Dem. 9):

τί δὴ πάλιν ἐστὶν ἐν τούτοις τὸ συνταράττον τὴν κατὰ φύσιν ἀπαγγελίαν; πρῶτον μὲν τό, πρὶν ἀπαρτίσαι τὸ ἡγούμενον εἴτε νόημα χρὴ λέγειν εἴτε κῶλον, ἕτερον παρεμβαλεῖν καὶ μηδὲ τοῦ δευτέρου τέλος ἔχοντος τὸ τρίτον ἐπιζεῦξαι, εἶτα τὴν τοῦ δευτέρου νοήματος ἀκολουθίαν ἐπὶ τῷ τρίτῳ τέλος εἰληφότι θεῖναι, κἄπειτα ἐπὶ πᾶσιν, ὃ τοῦ πρώτου μέρος ἦν, διὰ μακροῦ καὶ οὐκέτι τῆς διανοίας αὐτὸ προσδεχομένης ἀποδοῦναι.

What is it that destroys the natural ἀπαγγελία in this as before? In the first place, before rounding off the first idea (or clause if it should be so called), a second idea is introduced; then a third is subjoined before the second is complete, and material belonging to the second is tacked on after the third has been completed; and at the very end the remains of the first subject, after a long interval, are added when the sense can no longer accommodate it (transl. Usher 1974, 273, adapted).

The ‘how’ that ἀπαγγελία refers to includes both the segmentation of thoughts into linguistic units (κῶλα), which is encompassed by Hermogenes’ concept of λέξις, as well as their arrangement – a task which Hermogenes assigns to his μέθοδοι. In other passages of Dionysius’ critical essays, ἀπαγγελία is used as a near-synonym for ἑρμηνεία and φράσις.^[46]

The use of ἀπαγγελία in Theon’s Progymnasmata matches closely that by Dionysius. Here, this word occurs in the presentation of exercises in narrative or descriptive discourse modes, where it indicates (a) an exercise consisting in repeating and/or reformulating the expression of specific contents (101.3–9 Patillon, cf. 85.29–86.5 Patillon; this meaning goes back to that of “report”), (b) oratorical narratio (διήγημα, 107.21–24 Patillon), or (c) generically, “presentation” (119.14). In the latter sense, ἀπαγγελία may indicate the verbal expression of thoughts as opposed to the ὑποκείμενα (the subject matter) as a synonym of φράσις and ἑρμηνεία.^[47] However, in the Progymnasmata ἀπαγγελία seems to remain distinct from λέξις, which is presented as its component.^[48]

It is hard to identify a consistent meaning of ἀπαγγελία in later Imperial rhetoric^[49] – this term will not become a ‘standard’ one, so to speak, in the technical vocabulary of the discipline; however, it is possible to discern a general meaning of “way of expression (of thoughts) ” in in both Dionysius and Theon – a ‘how’ that is not restricted to “diction” (λέξις) alone but would seem to encompass the structuring and organization of the contents at the level of discourse.^[50]

Last but not least, the adherence to τὸ πρέπον, a central concept in ancient stylistics, was also perceived as a formal feature subject to variation – a quality of the style of individual authors – and hence as an aspect of the ‘how’.^[51] The idea that a text should be ‘appropriate’ to its context is deeply rooted in Greek literary culture.^[52] In classical rhetoric, this idea is expressed by the notion of καιρός, which played a prominent part in sophistic debates (especially those concerning extemporization)^[53] but also took a more technical turn: a speaker’s language should fit both the subject matter and the speaker’s own character. In this sense, τὸ πρέπον appears alongside καιρός in Isocrates’ Against the Sophists (13) and in Aristotle’s Rhetoric (1408a10–15), where it is defined as follows:

τὸ δὲ πρέπον ἕξει ἡ λέξις, ἐὰν ᾖ παθητική τε καὶ ἠθικὴ καὶ τοῖς ὑποκειμένοις πράγμασιν ἀνάλογον. τὸ δ’ ἀνάλογόν ἐστιν ἐὰν μήτε περὶ εὐόγκων αὐτοκαβδάλως λέγηται μήτε περὶ εὐτελῶν σεμνῶς, μηδ’ ἐπὶ τῷ εὐτελεῖ ὀνόματι ἐπῇ κόσμος· εἰ δὲ μή, κωμῳδία φαίνεται, οἷον ποιεῖ Κλεοφῶν.

Λέξις will have τὸ πρέπον if it expresses emotion and character and is proportional to the subject matter. Proportion exists if there is neither discussion of weighty matters in a casual way nor shoddy things solemnly and if ornament is not attached to a shoddy word. Otherwise, the result seems comedy, like the [tragic] poetry Cleophon composes (transl. Kennedy 2007, 235).

Aristotle is even more explicit in another passage of the Rhetoric (1404b1–5, 12–19) where τὸ πρέπον is presented as a component of the λέξεως ἀρετή and as principle that governs stylistic variation:^[54]

ἔστω οὖν ἐκεῖνα τεθεωρημένα καὶ ὡρίσθω λέξεως ἀρετὴ σαφῆ εἶναι (σημεῖον γάρ τι ὁ λόγος ὤν, ἐὰν μὴ δηλοῖ οὐ ποιήσει τὸ ἑαυτοῦ ἔργον), καὶ μήτε ταπεινὴν μήτε ὑπὲρ τὸ ἀξίωμα, ἀλλὰ πρέπουσαν· ἡ γὰρ ποιητικὴ ἴσως οὐ ταπεινή, ἀλλ’ οὐ πρέπουσα λόγῳ. […]

ἐπὶ μὲν οὖν τῶν μέτρων πολλά τε ποιεῖται οὕτω καὶ ἁρμόττει ἐκεῖ (πλέον γὰρ ἐξέστηκεν περὶ ἃ καὶ περὶ οὓς ὁ λόγος), ἐν δὲ τοῖς ψιλοῖς λόγοις πολλῷ ἐλάττω· ἡ γὰρ ὑπόθεσις ἐλάττων, ἐπεὶ καὶ ἐνταῦθα, εἰ δοῦλος καλλιεποῖτο ἢ λίαν νέος, ἀπρεπέστερον, ἢ περὶ λίαν μικρῶν· ἀλλ’ ἔστι καὶ ἐν τούτοις ἐπισυστελλόμενον καὶ αὐξανόμενον τὸ πρέπον· διὸ δεῖ λανθάνειν ποιοῦντας, καὶ μὴ δοκεῖν λέγειν πεπλασμένως ἀλλὰ πεφυκότως.

Let the matters just discussed be regarded as understood and let the virtue of style be defined as “to be clear” (speech is a kind of sign, so if it does not make clear it will not perform its function) – and neither flat nor above the dignity of the subject, but appropriate. The poetic style is hardly flat, but it is not appropriate for speech. […]

Many [kinds of words] accomplish this (sc. τὸ θαυμαστόν) in verse and are fitting there; for what is said [in poetry] about subjects and characters is more out of the ordinary, but in prose much less so; for the subject matter is less remarkable, since even in poetry it would be rather inappropriate if a slave used fine language or if a man were too young for his words, or if the subject were too trivial, but in these cases, too, propriety is a matter of contraction or expansion (transl. Kennedy 2007, 221–222, adapted).

These remarks echo an exchange between Euripides and Aeschylus in Aristophanes’ Frogs (1056–1061):^[55]

(transl. Sommerstein 1999, 121)

As the first passage of Aristotle quoted here indicates, failure to adhere to τὸ πρέπον characterizes the tragedies of Cleophon. In the Frogs, Aeschylus and Euripides agree that there is a difference between what we would call sociolects. Euripides accuses Aeschylus of using a not using the sociolect of the audience and Aeschylus responds that he used the sociolect of the characters he put on stage. The playwrights are made to represent clashing views of what τὸ πρέπον commands, and Aristotle would probably subscribe to the practice ascribed to Aeschylus. Aeschylus, in turn, does not attack Euripides explicitly on the grounds of the language he puts in his characters’ mouths, but condemns his choice of lowly characters in the first place (Ra. 840–846) – which Euripides defends (947–949) – as well as his way of portraying high-ranking characters (1063–1064).

The task of castigating Euripides’ failure at τὸ πρέπον is left with Theon’s Progymnasmata (60.27–31 Patillon):

διὰ τοῦτο πρῶτον μὲν Ὅμηρον ἐπαινοῦμεν, ὅτι οἰκείους λόγους περιτέθεικεν ἑκάστῳ τῶν εἰσαγομένων προσώπων, τὸν δὲ Εὐριπίδην καταμεμφόμεθα, ὅτι παρὰ καιρὸν αὐτῷ Ἑκάβη φιλοσοφεῖ.

Thus, we praise Homer first because of his ability to attribute the right words to each of the characters he introduces, but we find fault with Euripides because his Hecuba philosophizes inopportunely (transl. Kennedy 2003, 4).

The epitomator of Dionysius of Halicarnassus’ On Imitation (3.5 Aujac) criticizes Xenophon on similar grounds:^[56]

ἀλλ’ οὐδὲ τοῦ πρέποντος τοῖς προσώποις πολλάκις ἐστοχάσατο, περιτιθεὶς ἀνδράσιν ἰδιώταις καὶ βαρβάροις ἔσθ’ ὅτε λόγους φιλοσόφους, λέξει χρώμενος διαλόγοις πρεπούσῃ μᾶλλον ἢ στρατιωτικοῖς κατορθώμασι.

But he is often careless of what is appropriate (τοῦ πρέποντος) for his characters, as he now and then puts philosophical speech in the mouth of common people or barbarians, employing a λέξις that is more suitable for dialogues than for military feats.

Once again, respect of τὸ πρέπον is an aspect of the ‘how’ that contributes to the characterization of the style of an author and, as we read in Theon’s Progymnasnata, may be singled out as a distinctive element and as the basis of comparison between authors. What is important in these passages is that τὸ πρέπον does not concern the wording alone, as Aristotle and Aristophanes hint at, but it involves the very choice of the subject matter: Hecuba, ἰδιῶται, and βάρβαροι are neither expected to philosophize nor, as a consequence, to speak in a philosophical style.

Homer’s mastery of τὸ πρέπον must have had an almost axiomatic status among Hellenistic critics, if we are to trust what scholia tell us about the activity of Alexandrian scholars. Schol. Il. 3.423a, for instance, reports that Zenodotus replaced Il. 3.423–426^[57] with the line ἀυτὴ δ’ ἀντίον ἷζεν Ἀλεξάνδροιο ἄνακτος (“and she sat down facing lord Alexander”) because he found it ἀπρεπές that Aphrodite should set down a chair for Helen. Similarly, according to schol. Il. 4.88a, Zenodotus is blamed for turning εἴ που ἐφεύροι (“hoping to find him somewhere”, Il. 4.88) into εὗρε δὲ τόνδε (“she found him”) and omitting line 89 (εὗρε Λυκάονος υἱὸν ἀμύμονά τε κρατερόν τε “she found the son of Lycaon, blameless and mighty”) because of his belief that looking for someone is ἀνθρώπινον and inappropriate for Athena. As Pfeiffer notes, these scholia ultimately seek for reasons for Zenodotus’ interventions and their explanation may well be guesswork; it is nevertheless telling that they invoke alleged convictions about τὸ πρέπον as a ‘stylistic’ trait as a plausible criterion.^[58]

If we remain in the domain of ancient philology, a sense for what the broadly-defined ‘style’ of an author was – one that looks at certain aspects of the ‘what’ as aspects of the ‘how’ – informed the textual criticism of Aristarchus.^[59] This approach will also be central to efforts towards the solution of questions of ascription and authenticity, starting, if Ronconi is right, with the school of Pergamum, whose anomalist views in grammar would nurture the idea that each author has an individual χαρακτήρ.^[60] A ‘stylistic’ criterion in matters of ascription surfaces in a fragment of the polymath Demetrius of Magnesia, who authored a book On Homonyms and was reproached by Dionysius of Halicarnassus along with Callimachus (evidently qua author of the Pinakes) and, precisely, the grammarians from Pergamum (D.H. Din. 1 = fr. 1 Mejer).^[61] Dionysius’ criticism is constructive, in that he endeavours to develop and systematize this critical approach. The ‘stylistic’ ἰδιότης of each writer is what he set out to expound in his essays on the ancient orators; in that on Dinarchus – a late one in his production – Dionysius adopts it as a guideline to distinguish the speeches ascribed to this orator from those of his colleagues, alongside an external criterion such as chronology (Din. 4). One must look at both the λεκτικόν and the πραγματικὸν τόπον – diction and treatment of the subject matter (Din. 5) – a kind of ‘what’ that is more abstractable and formalizable than the ‘(true) what’ that is the subject matter itself.

Broadly-defined ‘style’ – or shall we say usus scribendi^[62] – has also played a major role in modern textual criticism as an internal criterion for the emendation of texts. However, when we read about style in contemporary classical scholarship, what we normally find are discussions of a restricted notion of ‘linguistic style’.^[63] Chapters on (linguistic) style are staple in the introductions to literary commentaries to classical prose texts; curiously, they tend to be replaced by chapters on ‘language (and metre)’ as their functional equivalents in commentaries on poetry, as if the very notion of ‘style’ belonged in the tradition of rhetoric rather than poetics or literary criticism at large. On this point, we may note that while stylistic analysis may focus – at least in principle – on any set of formal/formalizable features of a text, properties such as language/dialect, metre, or the organization of structural units (such as those we identify in plays or oratorical speeches) are de facto normally left out of discussions of style.^[64] This raises the question what it really is that modern scholars perceive as ‘style’.

The difficulty of answering this question is also reflected by the elusive status of ‘stylistics’ as a discipline beyond the classical studies. Just like its object of interest, stylistics is difficult to define and delimit. Some specialists restrict the scope of the discipline to literary texts, whereof stylistics represents the ‘formal’ study – as Stockwell and Whiteley (rather hybristically!) put it, the “proper study”.^[65] According to Simpson, for example, “stylistics is a method of textual interpretation in which primacy of place is assigned to language” and whose “preferred object of study […] is literature”, with the caveats that “creativity and innovation in language use should not be seen as the exclusive preserve of literary writing” and that “the techniques of stylistic analysis are as much about deriving insights about linguistic structure and function as they are about understanding literary texts”.^[66] Burke, in turn, glosses stylistics as “literary linguistics” and traces its roots in ancient rhetoric, equating λέξις with ‘style’,^[67] while Fix, Gardt, and Knape insist on how stylistics and rhetoric share a “pragmatic-communicative as well as a structural dimension” and on how difficult it is to disentangle them from each other if looked at from this perspective.^[68] On a narrower understanding, stylistics may be conceived of as the study of ‘literariness’ in the tradition of the Formalist schools of the twentieth century.^[69]

Jeffries and McIntyre, conversely, present stylistics as “a sub-discipline of linguistics that is concerned with the systematic analysis of style in language and how this can vary according to such factors as, for example, genre, context, historical period and author”. In their view, stylistics can concern itself with both literary and non-literary texts, and the analysis of style consists in “looking systematically at the formal features of a text and determining their functional significance for the interpretation of the text in question”.^[70] On such an understanding, the notions of style and stylistic variation come very close to those of register and register variation, if by register we mean a linguistic variety defined by its communicative circumstances and purposes^[71] and whose study falls especially in the domain of ethnography and sociolinguistics. Seen in this light, the notion of style would differ from that of register would only be distinguished from that of register by resorting, once again, to the notion of ‘literariness’. According to Biber and Conrad, for example, style and register are essentially the same thing – they can be analysed using the same techniques – but can be distinguished because they express different motivations for linguistic variation:^[72]

The register perspective combines an analysis of linguistic characteristics that are common in a text variety with analysis of the situation of use of the variety. The underlying assumption of the register perspective is that core linguistic features like pronouns and verbs are functional, and, as a result, particular features are commonly used in association with the communicative purposes and situational context of texts. […] The style perspective is similar to the register perspective in its linguistic focus, analyzing the use of core linguistic features that are distributed throughout text samples from a variety. The key difference from the register perspective is that the use of these features is not functionally motivated by the situational context; rather, style features reflect aesthetic preferences, associated with particular authors or historical periods.

Still, even the aesthetic factors motivating variation in literary language can in principle be explained as the result of the conventionalization of linguistic forms in connection with socio-cultural aspects of communication. On these grounds, the substitution of ‘register’ for ‘style’ – at least in the classical studies – has been advocated by Willi:^[73]

One may speak of the ‘style’ of an author or even of an epoch (e. g., the ‘style’ of Thucydides/of Hellenistic literature), but given the wide range of linguistic usages adopted by Thucydides in different parts of his work or by different Hellenistic authors, any overall description of these would end up being banal; and one might even argue that the peculiar ‘style’ of a poet like Aristophanes arises precisely from the mixture of ‘registers’ belonging to different communicative situations. Hence, the danger of imprecision is at least reduced when we use the term ‘registers’, and as long as we define at what level of generality we are conducting our investigations, ‘registers’ is actually quite a useful concept. After all, the same is true for its counterpart ‘genre’: the fact that we may refer to, say, love-letters as a ‘genre’ at a low level of generality, whereas at a higher level personal letters on all kinds of topics might constitute a ‘genre’, does not reduce the usefulness of the concept of ‘genre’. Rather than set registers and genres against each other, by associating the former with regularly recurring communication situations and the latter with regularly recurring message types (Ferguson 1994, 20–21), we should therefore understand register as the form (or signifiant) plane of an utterance or text, which corresponds to genre as the content (or signifié) plane: genres are “text categorizations made on the basis of external criteria relating to author/speaker purpose” or “text categories readily distinguished by mature speakers of a language” (Biber 1988, 68; 1995, 9), whereas registers are constituted by the linguistic features identifying these text categories. For instance, all those features (of intonation, syntax, lexicon, etc.) which were typically used in a funeral speech constitute the ‘register’ of the ‘genre’ ἐπιτάφιος λόγος.

Such a perspective on style and genre squares quite well with ancient remarks on τὸ πρέπον and the individual style of characters in literature. The identity of the speaker (defined as the ‘source’ of a verbal message) is one of the elements in a linguistic exchange that can be recognized as a component of any socio-cultural situation in which language is used (a ‘speech situation’) and whose variation may determine linguistic variation. This is, for example, what Aristophanes’ distinction between φράζειν ἀνθρωπείως and the language of the demigods points to. Likewise, the distinction between a λέξις διαλόγοις πρέπουσα and a λέξις στρατιωτικοῖς κατορθώμασι πρέπουσα fits well an understanding of genres as ‘recurring message types’ identified by sets of formal characteristics that texts are expected to share on the grounds of an established tradition in a community of language users (a ‘speech community’). Defined in this way, ‘genre’ may also be construed as a component of a speech situation and thus as a sociolinguistic variable;^[74] as a consequence, a linguistic variety like the λέξις διαλόγοις πρέπουσα should rather be regarded as a ‘register’ than as a ‘style’.

Classicists, however, seem quite reluctant to abandon the notion of ‘style’ altogether,^[75] and perhaps Biber and Conrad do have a point in underlining the ‘aesthetic’, ‘literary’ connotation carried by this notion as opposed to its ‘technical’, ‘socio-anthropological’ counterpart. As Jeffries and McIntyre point out, “a principle that all stylisticians subscribe to is that meaning in language comes about through the choices that a writer makes (either consciously or unconsciously)”.^[76] The study of style aims to contribute to the study of meaning – it is geared towards the interpretation of texts, and not towards the formal description of all possible elements of variation – and would thus differ from the study of register in its goals more than anything else. Such an understanding of style and stylistics may perhaps explain why features such as metre, language/dialect variety, or structure are not normally described as features of ‘style’, as noted above. These features, I would surmise, tend to be perceived as independent from the ‘meaning’ of a text; from a semiotic point of view, they are signs that refer primarily to text-types. If this holds true, the reason why the structure of classical Attic tragedy is not commonly considered an element of tragic style would be that the referent it is perceived to evoke is nothing else than the genre ‘classical Attic tragedy’ itself.

2 How style met the city

As we have just seen, certain ancient remarks on linguistic variation square well with a sociolinguistic interpretation of the notion of ‘style’ as ‘register’, and it is worthwhile to explore to what extent a sociolinguistic sensitivity may be seen to underlie ancient conceptualizations of style as a phenomenon. For a start, let us think about from the ancient classificatory systems.

That of the tripertita varietas – the tria genera of the Rhetorica ad Herennium (4.8.11) – is quite a persistent presence in Western rhetorical tradition. Apart from Cicero and Dionysius of Halicarnassus, the threefold distinction between simple, grand, and middle style is repeated by Quintilian and we find it, later on, in Iulius Victor (p. 438 Halm), Fortunatianus (3.9, as part of a more complex system), in Isidore (Orig. 2.17.1), and throughout medieval and early modern Europe.^[77] If we look at the Greek side, however, this system pretty much gets out of the radar from the Imperial period onwards. It is fundamentally ignored by the Progymnasmata – which may be fair enough, if we consider that these were preliminary exercises^[78] – it is not mentioned in the treatise On the Sublime, and it does not really seem to have survived the ‘rhetorical revolution’^[79] of the second century, when the system was fundamentally supplanted by Hermogenes’ ἰδέαι which would then dominate Medieval Greek rhetoric.^[80] In fact, the tria genera look oddly marginal in Quintilian’s Institutio (12.10.58–80) as well, where their presentation is relegated to the end of the last book and does not reveal much enthusiasm (66–72):

Sed neque his tribus quasi formis inclusa eloquentia est. […] Prope innumerabiles species reperiuntur, quae utique aliquo momento inter se differant. […] Plures igitur eloquentiae facies, sed stultissimum quaerere ad quam se recturus sit orator, cum omnis species, quae modo recta est, habeat usum, atque id ipsum non sit oratoris quod vulgo genus dicendi vocant: utetur enim, ut res exiget, omnibus, nec pro causa modo sed pro partibus causae. […] Non unus color prohoemii narrationis argumentorum egressionis perorationis servabitur. Dicet idem graviter severe acriter vehementer concitate copiose amare, idem comiter remisse subtiliter blande leniter [dulciter] breviter urbane, non ubique similis sed ubique par sibi.

But eloquence is not limited to these three patterns, as we may call them. […] An almost infinite number of species can be found, all differing from one another in some respect. […] Eloquence thus takes many forms; but it is very foolish to ask which of them the orator should take as his standard. Every variety which is correct has its use, and what is commonly called a ‘style’ (genus dicendi) is not something that belongs to the orator. He will use all ‘styles’, as circumstances demand, and as required not only by the Cause as a whole but by its various parts. […] He will not keep to one tone in Prooemium, Narrative, Arguments, Digression, and Peroration. He will speak gravely, severely, pungently, vehemently, energetically, copiously, bitterly, or again affably, quietly, simply, flatteringly, gently, [sweetly], briefly, wittily. He will not always be the same, but he will nowhere fall below his own standard (transl. Russell 2002, 317–321).

Quintilian seems to anticipate Hermogenes in his approach to stylistic composition as the ability to vary one’s expression fittingly throughout a text rather than sticking to a rigid but coarse-grained inventory of style types. The suspicion arises that the tripartite classification of style might have enjoyed more popularity with literary criticism than with practical rhetorical education. After all, the classification itself seems to be traceable to philosophical, rather than rhetorical, schools.

As far as we can tell from pre-Aristotelian sources, the ancient Greek stylistic sensitivity is based on a fundamental distinction between a ‘grand’ and a ‘simple’ style. This dichotomy can be traced back at least to the beginning of rhetorical teaching, with the most conspicuous evidence being the contest of Aeschylus and Euripides in Aristophanes’ Frogs.^[81] Such a distinction appears to be based on a combination of aesthetic perceptions (e. g., heaviness/lightness, Ra. 941), a sense for what we would call register variation (e. g., στωμύλματα ἀπὸ βιβλίων “bookish persiflage”, 942, λαλιὰ καὶ στωμυλία, 1069, etc.), and a practical preoccupation with the effectiveness of communication (ῥήμαθ’ ἱππόκρημνα / ἃ ξυμβαλεῖν οὐ ῥᾴδι’ ἦν, 928–929, cf. 1434).

The ‘discovery’ of a third, middle style is often ascribed to Theophrastus^[82] and certainly fits well with the Aristotelian ideal of μεσότης. A possible precursor may be traced into a fragment of the fifth-century philosopher Diogenes of Apollonia (a contemporary of Socrates), who maintained that those who begin a discourse should adopt an ἑρμηνεία ἁπλῆ καὶ σεμνή (fr. 64 B1 DK)^[83] – a piece of advice that resonates, for example, with later views of Dionysius of Halicarnassus on the primacy of the mixed style of Demosthenes (Dem. 34). At any rate, the fact that a rhetorician like ‘Demetrius’, imbued with Peripatetic ideas as he was, adopts a quadripartite classification that excludes any intermediate styles of sorts (cf. Demetr. Eloc. 36–37) and does not feel compelled to defend his model against the tripartite one, raises the suspicion that the idea of a middle style must have been developed at a later time.^[84] As Innes suggests, Theophrastus may have “recognised only one good style (or diction), a mean between excessive plainness and elaboration. […] If so, Theophrastus developed what was in essence already in Aristotle, who advised appropriate diction, neither low nor too elaborate”.^[85] As a matter of fact, Theophrastus seems to have regarded τὸ πρέπον – Aristotle’s governing principle of expression – as one of the ἀρεταί, as mentioned above. If this is true, it would look like the initial ‘intuitive’ dichotomy between a ‘grand’ and a ‘simple’ style was replaced by the Peripatetic ‘middle’ as the only acceptable style and that the two extremes (which were probably acknowledged, but not sanctioned, by the early Aristotelians) reentered the classificatory system at a later stage within the tradition of the Peripatos. This idea would be supported, according to Chiron, by the fact that Cicero (Orat. 69–70) associates the three genera to three officia that may be mapped onto the three Aristotelian πίστεις: the officium of genus subtile is probare (λόγος), that of the genus modicum is delectare (ἦθος), and that of the genus vehemens is flectere (πάθος).^[86] In turn, the four styles of ‘Demetrius’, if Chiron is right, may also have a theoretical, rather than practical origin, and could be traced to the rhetorical ideas of Middle Stoicism.^[87]

Between the ‘aesthetic’ distinction between grandeur and simplicity and the ostensibly speculative systems of the Hellenistic period, however, there are signs of a truly ‘practical’ approach to the distinction of styles – one that presupposes familiarity with, and consideration for, the city and the actual communicative contexts for which texts were composed.

The city in question is the city of Athens, where two slightly younger^[88] contemporaries of Aristophanes voiced their opposing opinions on the role of writing in matters of verbal communication.^[89] Alcidamas, to begin with, draws a distinction between written and improvised speeches in his pamphlet On Sophists. Written composition is betrayed by excessive precision (ἀκρίβεια, Soph. 13)^[90] and, while this may be fine in epideictic speeches (31), it is more often than not a drawback in contexts where one needs to speak off the cuff (Soph. 25), since written speeches are not adaptable enough to be truly helpful in decision-making contexts. This idea that is captured well by the modern notion of ‘desituatedness’: written language is produced separately from the circumstances of its reception, and as a consequence it may not adapt to such circumstances.^[91]

Isocrates shifts the focus from the role of writing in composition to the contexts of reception of written texts. He would agree with Alcidamas on the fact that speeches meant to be read (λόγοι ἀναγιγνωσκόμενοι, Isoc. 5.25) are unsuitable for decision-making contexts because of their desituatedness (Isoc. 5.26). On top of this, λόγοι ἀναγιγνωσκόμενοι incur the further problem that their oral rendering may be completely out of the author’s control. A bad reader may not be able to interpret and perform a written text correctly, because of incompetence or lack of talent (cf. Isoc. 15.189) or because of his inability to convey the character and sentiments of the writer (Isoc. 5.26). Those who compose λόγοι λεγόμενοι know that the text will be delivered on a specific occasion, either by themselves or by another speaker who could, at least in principle, consult them as they rehearse the text.^[92] Conversely, in the case of λόγοι ἀναγιγνωσκόμενοι, the occasion on which a speech will be received is potentially very distant and its contexts are mostly unpredictable. A savvy (or self-conscious) writer must take this into account and, as a matter of fact, Isocrates apologizes for not having indulged in the περὶ τὴν λέξιν εὐρυθμίαι καὶ ποικιλίαι (5.27) in the speech he wrote to be sent to Philip. Improvisation seems out of the question here; λόγοι λεγόμενοι may well be composed in writing, but their envisaged mode of performance is different from ἀνάγνωσις in that they are performed directly by their author (or his client) and the written page must remain ‘invisible’ to the audience.^[93]

Different practicalities and communicative concerns were thus perceived both to be reflected by formal features of texts – improvised texts are distinguishable from prepared texts – and to demand different stylistic properties – writing for ἀνάγνωσις requires different strategies from writing for ὑπόκρισις. The latter notion is fully formalized in Aristotle’s Rhetoric, which draws a distinction between λέξις γραφική and λέξις ἀγωνιστική (1413b3–1414a18). The most ‘written’ style, we are told, is that of epideictic literature, which is composed for ἀνάγνωσις – which, in Aristotle’s time, would still indicate oral reading.^[94] The ‘debating’ style, in turn, is connected to decision-making (forensic and political) contexts and requires fully-fledged oratorical ὑπόκρισις in order to be effective.^[95] Much in the same way as Isocrates refrains from εὐρυθμίαι καὶ ποικιλίαι, a stylistic feature such as asyndeton should be avoided when competent ὑπόκρισις is not guaranteed or not even envisaged as an option.^[96]

The idea that style should be sensitive to the envisaged contexts of communication falls into line with the almost sociolinguistic sensitivity to linguistic variation that we have observed in Aristophanes. Aristotle’s identification of speaker (ὁ λέγων), message (περὶ οὗ λέγει), and addressee (πρὸς ὅν) as components of λόγος as he defines the three rhetorical εἴδη (forensic, political, and epideictic, Rh. 1358a36–b1) goes in the same direction. The parameters that define each εἶδος easily fit into lists of components of speech situations drawn by modern sociolinguists.^[97] Apart from audience and place, what Pepe^[98] calls “communicative function” (προτροπή/ἀποτροπή for political oratory, κατηγορία/ἀπολογία for forensic oratory, ἔπαινος/ψόγος for epideictic oratory) corresponds quite neatly to what Hymes calls “purpose-goals”, that is to say, what the speaker wants to achieve by performing of a communicative act.^[99] What Pepe calls “end” (τὸ συμφέρον/τὸ βλαβερόν for political oratory, τὸ δίκαιον/τὸ ἄδικον for forensic oratory, τὸ καλόν/τὸ αἰσχρόν for epideictic oratory) maps onto Hymes’ “purposes-outcomes”, defined as the “conventionally recognized and expected outcomes […] of speech events”,^[100] that is, what members of a community expect to achieve by participating in a speech event in the first place.^[101]

Aristotle’s distinction was not taken on by later attempts at the systematic categorization of styles and perhaps should not be regarded as a conscious effort towards one either. If we are to trust the Rhetoric to Alexander as the best representative we have for what early technai looked like, the identification of χαρακτῆρες τῆς λέξεως was probably not much of a thing in early rhetoric anyway. In this treatise, for example, we read about how to make λέξις grand (μεγαλοπρεπής, Rh.Al. 35.16 1441b9–10) and to produce a λόγος ἁπλοῦς καὶ μὴ ποικίλος (31.2 1438b21), but these amount to qualifications of language or discourse. In the end, Aristotle himself does not talk about χαρακτῆρες, but of ‘qualified’ λέξεις, as we have seen. As it were, the notions of λέξις γραφική and λέξις ἀγωνιστική would refer to how λέξις is/should be like when γραφική or ἀγωνιστική, and not to two distinct types of λέξις, and the fact that ‘writtenness’ did not play a role in later classifications of styles seems to point in this direction.

All in all, stylistics’ flirtation with the city – if any – was not there to last. That between style and the city, though, seems to have been a true relationship. In a recent exercise,^[102] I have attempted to interpret the ancient perceived dichotomy between prepared and extempore speech as a contrast between perceived planning and spontaneity, and that between ‘theatrical’ ἀγωνιστικὴ λέξις and ‘bookish’ γραφικὴ λέξις as one between (inter)personal involvement and detachment. Modern sociolinguistic and corpus studies have found that these contrasts correspond to dimensions of linguistic variation along which registers may be compared with each other, with oral registers leaning towards the ‘involved’ and ‘spontaneous’ ends of the respective dimensions and written registers leaning towards the ‘detached’ and ‘planned’ ends.^[103] In order to compare texts and assign them a score along these dimensions, I have identified sets of linguistic features whose pragmatic function can be interpreted as contributing to the effects that characterize each dimension (for example, I have taken second-person references as indicators of involvement) and have run a statistical analysis on the corpora of Demosthenes, Isocrates, Lysias, and Aeschines. The results were shockingly encouraging. Based on the average values for each dimension, the corpora of Demosthenes and Isocrates appear at the opposite ends of the spectrum, with Demosthenes displaying significantly more involvement and spontaneity than Isocrates – precisely what one would expect – with Lysias and Aeschines clustering somewhere in the middle (Figure 1).

Figure 1:

Average detachment/involvement and planning/spontaneity scores for oratorical corpora (95 % confidence intervals).^[104]

These results would indicate that the practical, ‘sociolinguistic’ sensitivities of early rhetoricians did correspond to actual linguistic variation in oratorical texts – their style did consider the communicative needs and expectations of the city.

3 This special issue

Ancient rhetoric found its breeding ground in Attic public speaking and matters of style have remained a central concern for critics and practitioners of oratory ever since classical Antiquity. Attic oratory was not – and, of course, still is not – the only genre that stylistic analysis has concerned itself with, but hardly any literary study of these materials fails to take into consideration questions of style. Typical modern treatments develop and update traditional analytical categories; as Carey notes, for example, “all modern judgments on Lysias’ style take as their starting-point the perceptive essay of Dionysios of Halikarnassos in his collection On the ancient orators”,^[105] and he himself discusses such features as vocabulary, metaphors, ‘periodic’ structures, figures, and divisions.^[106] Similarly, Ronnet’s 1971 monograph on the style of Demosthenes deals with figures of speech, the periodic style, figures of thought, and metaphors, as well as vocabulary and syntax. Old questions can stimulate fresh and methodologically innovative answers. Gagarin, for example, revises the question of the writtenness/orality of ancient oratory with a view on communication,^[107] while MacDowell’s review of Demosthenes’ style focuses on the communicative effect of rhetorical devices^[108] and Dover and Usher respectively devise stylometric solutions for problems of attribution or stylistic variation.^[109]

Attic oratory has also served as valuable repertoire of forms and usages for the study of the Greek language at large – one may think, to give but one example, of the numerous discussions of the use of the aorist and the present imperative in Demosthenes’ addresses to the clerk.^[110] Still, much work needs to be done to connect the rhetorical/literary study of style with a linguistic approach seeking to unearth the mechanisms that lie behind the purported effect of stylistic composition in Attic oratory – persuasion.

The present special issue brings together seven classicists and linguists in an attempt to contribute to this effort by examining the language of Attic oratory both in its forms and in its functions. Individual chapters focus both on what rhetorical and psychological effects the manipulation of language intended to achieve (the functions) for the purpose of winning over the audience, and on the linguistic devices deployed to achieve such effects (the form), examining specific aspects of what one may call the “grammar of persuasion”.

Studies in this volume adopt either a bottom-up approach, aiming at identifying the function of linguistic features, or a top-down approach, starting from specific rhetorical effects and aiming to identify the linguistic features, or the cognitive mechanisms underlying the linguistic features, that produce them.

The latter approach informs three contributions. Evert van Emde Boas discusses the complex question of ēthopoiia in Lysias – an often misconstrued concept – as a proof of concept for the modern notion of ‘mind style’, that is to say, the idea that “the idiosyncratic world view of a character or narrator [is] exhibited through the linguistic structure of a text”.^[111] Van Emde Boas seeks to identify linguistic patterns that characterize Euphiletus in the diēgēsis of Lysias 1 by adopting a quantitative approach and using the diēgēseis of several other speeches with lengthy first-person narratives (3, 12, 19, 32) as comparative material, as well as providing an interpretation of the data from a cognitive and narratological perspective.

Jakub Filonik starts from a long-recognized feature of ‘style’ – metaphor – and explores its communicative effectiveness in Attic oratory. In particular, he expounds the working of the conceptual metaphor the polis is a possession and related ones (e. g., citizens are stake-holders) highlighting their emotional effectiveness and demonstrating its relevance for contemporary political discourse.

The diēgēsis of Lysias 1 is also the object of interest of Rutger Allan’s contribution. Allan examines the enargeia that Dionysius of Halicarnassus ascribes to Lysias as a way to induce the experience of ‘immersion’ in the audience. This concept – which Allan himself has profitably explored^[112] – has emerged in cognitive studies and has been especially helpful in the study of virtual reality. Allan brings together insights from narratology and fine-grained linguistic analysis to expose Lysias’ immersive devices as well as their contrary, distancing devices, and discuss their interplay and role in the connection with Lysias’ persuasive purpose.

Bottom-up approaches are also represented by three contributions. Andreas Serafim examines the use of, and the differences between, imperatives and subjunctives in Attic oratory. In particular, he explores the different usage patterns of imperatives across the three Aristotelian εἴδη, the pragmatic differences between commands and prohibitions in the subjunctive, and the effects of the two moods on judges and onlookers. Serafim concludes that the subjunctive is not a mitigator – commands and prohibitions are as forceful in this mood as they are in the imperative – but its use is sensitive to the personal involvement of the speaker in the situations he talks about. This conclusion may also serve as a starting point for the solution of a difficult problem in Greek grammar – namely, the preferential selection of the subjunctive for second-person prohibitions in the aorist.

The focus of Liana Tronci’s contribution is an important yet underexplored linguistic phenomenon – impersonal constructions. This linguistic device is often functionally interpreted as a detachment marker^[113] but its pragmatics and textual distribution in Classical Greek have not been specifically studied. Tronci sets out to fill this gap by looking at impersonal verbs and ‘generic’ uses of personal constructions (e. g., unspecified first-person plural subjects and indefinite-pronoun subjects) in Aeschines and Demosthenes, and concludes that Aeschines’ rhetorical use of impersonal constructions aims at concealing the referent, while Demosthenes uses such constructions to put forward universal statements that are valid for all people. As a side note for linguists, Tronci takes the thought-provoking step of extending the range of impersonal constructions to instances of the s.c. schema Atticum (neuter plural subject with singular verb) that involve subjects that rank low on the animacy hierarchy, and presents the idea that, in these constructions, verbs are ‘personless’.

A bottom-up approach to the study of Lysias is offered by Edith Hall, whose contribution examines the orator’s use of rhetorical questions. Hall begins with a review of the cognitive effects of this linguistic/rhetorical device and describes how it may be deployed to convey indignation and create the impression of spontaneity and emotion, while being at home in arguments from probability (eikos). Importantly, Hall discusses the function of rhetorical questions in characterization and shows how their use should fit the character of the speaker. Untimely and badly delivered rhetorical questions may sound ‘rhetorical’ and fake and may undermine the efforts of certain speakers to sound sincere, self-controlled, and down-to-earth. The results of this study are put to work in a case study of Lysias’ Against Eratosthenes.

The volume is rounded off by Victor Bers’ article on, we may say, ‘meta-stylistics’ in Attic oratory. In particular, Bers examines the linguistic awareness – and self-consciousness – of Athenian public speakers in the classical period, focusing both on the pretence of incompetence and on the ostentation of rhetorical ability, and highlighting cases in which a (for once, non-Aristotelian!) golden mean is pursued.