Community-specific grammar? The case of Aelius Aristides

Community-specific grammar? The case of Aelius Aristides Dmitry Dundua University of Oxford [email protected] Language and Identity in Antiquity University of Lausanne July 19, 2023 1 Background 1.1 ‘Atticist Greek’ • Milieu: the ‘renaissance’ of Greek elite culture under Rome, i.e. the ‘Second Sophistic’ (Bowersock 1969, Swain 1996, Whitmarsh 2005 a.o.); • Time: late first to mid-third centuries CE; • Place: for the most part, Greek cities of the Roman East (but see Bowersock 1969, Bowie 1982 and Swain 1991 for a more detailed picture); • Protagonists: Hellenophone intellectuals, mostly Eastern notables (Veyne 1976, Quaß 1993) with important local and Roman connections; • ‘Atticist Greek:’ the language of the Greek belles lettres in the High Imperial era; • ‘Core:’ constituted by prose texts delivered orally and in public, as was already observed in contemporary theory (cf. the notion of λόγος πολιτικός in Alex. Num. De figuris 12; [Arist.] Ars rhet. I, II; Herm. De id. 2.10-12, Phryn. Praep. soph.) but often overlooked in modern research;1 • A linguistic practice strongly oriented towards a canon and used to reconstitute and renegotiate elite Greek identity (Swain 1996, 17-42, Schmitz 1997, 160-96, Whitmarsh 2005, 41-56); • Common assumption: Atticist Greek as a—more or less sucessful—attempt to replicate the canon in the grammatical and lexical domains; • Divergences from actual Attic prose are seen as deviations. Cf. Wilson (2010, 403): Aelian was an Atticist, and unfortunately he was not as proficient as other practitioners of that fashionable linguistic archaism. On every page there are unsuccessful efforts to imitate the idiomatic usages of classical authors. There are many expressions which, if transmitted in the text of those authors, would almost certainly need to be emended. 1 Cf., however, Schmid (1902, 215-6), Rutherford (1998, 43-48) on Hermogenes and Tribulato (2022) on Phrynichus. 1 • However, the mismatches are systematic, particularly in various optative constructions, which had been almost ousted from some registers of the language by the Imperial period2 —to be subsequently ‘revived’ in higher ones, the exact nature of this process being a matter of debate.3 1.2 Community-specific speech • A pragmatic notion dealing with felicity conditions of speech acts (Austin 1962); • Speech acts have input and output conditions, specifying what agents can perform them and what agents they target in uptake—that is, as part of their very pragmatic structure (Kukla & Lance 2009); • Different speech act types can defined with reference to different conditions: imperatives, obligations, baptisms, etc.; • Community-specific speech: a type of speech act targeting a community in either input or output. Cf. Herbert & Kukla (2016, 581): Speech with a community-specific output functions, as a matter of its pragmatic structure, to call for uptake from members of a community qua community members, and not just qua individuals with the shared property of belonging to a community… In recognizing that you are being called to give uptake to a community-specific speech act, part of what you are giving uptake to is your community membership and your recognition of the community membership of the speaker. This is insider speech, not (just or necessarily) in virtue of the content, but in virtue of its function as a second-personal social transaction. • Thus, certain speech acts serve to reconfigure and reinforce communal identity both of the speaker and of the listener; • Community-specific speech has various guises but it is often represented by markers such as technical jargon and specific terms or acronyms; • These are linguistic shibboleths (Silverstein 2017): particularly salient expressions which reveal something about the speaker’s social being, including her identity and group identity, as their social meaning (Eckert 2008). Questions: 1. If Atticizing Greek is considered to be a means to renegotiate and reinforce collective identity of the Greek elites, can it be a case in point? 2. Further, if it is seen as an attempt to replicate the language of the canon, how do the discrepancies between them fit into the picture? 2 3 Bianconi & Magni (2022) provide a useful up-to-date summary on the NT. See Higgins (1945), Anlauf (1960); Reardon (1971, 81-96) and Schmitz (1997, 67-73) for surveys. 2 2 Aristides • Aelius Aristides (118–180/185 CE) is often taken to be the arch-representative of Imperial Greek eloquence, partly due the degree of purism he adopted (Schmid 1889, 309-13, Boulanger 1923, 395412); • A rather diverse corpus, including long polemical essays, declamations, speeches addressed to city audiences on subjects relevant to local politics, epideictic oratory of various sorts, prose hymns and the Sacred Tales, a peculiar blend of autobiography, medical case history, and dream interpretation; • Most of it belongs to the ‘core’ category of Atticist prose, i.e. was meant to be delivered orally and in public; • Aristides’ achievement was acknowledged already in his age, cf. Rutherford (1998, 96-104) on Hermogenes and ‘Ps.-Aristides A’4 and Jones (2008) on Phrynichus; • However, a number of divergences from classical Attic is attested regularly, particularly in optative use;5 • Goal: to set Aristides’ usage against classical Attic in a systematic way. 3 Case study: main clause optatives 3.1 Baseline corpus • Comparandum: a representative sample of classical Attic authors that were part of the canon in the Imperial era; • Demosthenes 18; Isocrates 8; Lysias 1; Thucydides 2, Xenophon, Anabasis 3, Plato, Protagoras; • Total token count: 73120, with 563 optatives; • Target: main-clause optatives with the modal particle ἄν,6 traditionally ‘potential optatives,’ total number 266; • From a broadly constructional perspective, a family of constructions (Goldberg & Jackendoff 2004), i.e. similar formal patterns which partly differ and partly coincide in their import on the plane of sense; • Functionally, the potential family has two basic kinds of import (but often it is hard to distinguish); 4 Who mention Aristides precisely in a discussion of πολιτικὸς λόγος. Cf. the ironic comment of Reardon (1971, 83): “il y a de quoi surprendre dans le fait qu’Aristide, repute Atticiste, et employant un optatif censé représenter I’usage attique, l’emploie incorrectement — voire, commette des fautes qui ne passeraient pas, de nos jours, chez un élève de troisieme.” 6 But see Bers (1984, 128-42) on exceptions. 5 3 • Modality: possibility or necessity of different types, often epistemic, cf. (1), and • Nonfactuality: signaling that a state of affairs expressed by the predicate needs additional circumstances to obtain, cf. (2): (1) καίτοι πολλά γε ἔτη ἤδη εἰμὶ ἐν τῇ τέχνῃ· καὶ γὰρ καὶ τὰ ξύμπαντα πολλά μοί ἐστιν· οὐδενὸς ὅτου οὐ πάντων ἂν ὑμῶν καθ᾿ ἡλικίαν πατὴρ εἴην· (Pl. Prot. 317c) And yet many long years have I now been in the profession, for many in total number are those that I have lived: in age I might be the father of anyone among you.7 (2) καὶ οὕτως ἂν καλλίστη ἡμῖν ἡ συνουσία γίγνοιτο· ὑμεῖς τε γὰρ οἱ λέγοντες μάλιστ᾿ ἂν οὕτως ἐν ἡμῖν τοῖς ἀκούουσιν εὐδοκιμοῖτε καὶ οὐκ ἐπαινοῖσθε· εὐδοκιμεῖν μὲν γὰρ ἔστι παρὰ ταῖς ψυχαῖς τῶν ἀκουόντων ἄνευ ἀπάτης, ἐπαινεῖσθαι δὲ ἐν λόγῳ πολλάκις παρὰ δόξαν ψευδομένων· ἡμεῖς τ᾿ αὖ οἱ ἀκούοντες μάλιστ᾿ ἂν οὕτως εὐφραινοίμεθα, οὐχ ἡδοίμεθα· εὐφραίνεσθαι μὲν γὰρ ἔστι μανθάνοντά τι καὶ φρονήσεως μεταλαμβάνοντα αὐτῇ τῇ διανοίᾳ, ἥδεσθαι δὲ ἐσθίοντά τι ἢ ἄλλο ἡδὺ πάσχοντα αὐτῷ τῷ σώματι. (Pl. Prot. 337b-c) In this way [if Socrates and Protagoras agree to dispute on friendly terms] our meeting will have highest success, since you the speakers will thus earn the greatest measure of good repute, not praise, from us who hear you. For good repute is present in the hearers’ souls without deception, but praise is too often in the words of liars who hide what they really think. Again, we listeners would thus be most comforted, not pleased; for he is comforted who learns something and gets a share of good sense in his mind alone, whereas he is pleased who eats something or has some other pleasant sensation only in his body. • Distributionally, there are 104 presents and 162 aorists; • The two groups differ in the number of lexemes attested: 40 in the present vs. 84 in the aorist; • The picture is reversed with respect to the overall distribution of present and aorist forms in the corpus, where the total number of presents (7779) is almost twice as high as that of the aorists (4174); • In principle, this could suggest that the potential optative has a preference for the aorist on the whole, and the bias could be due to a semantic preference of optatives for the perfective aspect, but this is not borne out by the examples; • A closer look at the data is necessary. Top frequencies and overall frequencies for the two groups in decreasing order.8 7 Translation Lamb (1924). These and other calculations were made using R (R Core Team 2021), visualized using ggplot2 (Wickham 2016, https://ggplot2. tidyverse.org) and tikz (Tantau 2013, http://sourceforge.net/projects/pgf/). 8 4 15 20 15 10 10 5 5 0 Figure 1: Present (left) and aorist (right) potential optatives (baseline) Present (104) Aorist (162) εἰμί (23, 22.1%) λέγω (17, 10.5%) ἔχω, φημί (14, 13.4%) γίγνομαι (14, 8.6%) θαυμάζω (6, 5.8%) δύναμαι, ὁράω (7, 4.3%) Table 1: Top frequencies for potential optatives (baseline) • Little ‘middle ground’ in the present: roughly a third represented by three lexemes only, εἰμί, ἔχω and φημί. This is unsurprising in the case of εἰμί and ἔχω due to overall high frequency, φημί is perhaps somewhat less expected; • In the aorist, a ‘smoother’ decrease in frequency, with only two overall productive verbs (λέγω and γίγνομαι) standing out significantly from the rest; • We may assess whether the potential optative has a predilection for particular verbs and, if so, to what extent, with basic Collostructional Analysis, a number of corpus methods used to measure degrees of association between lexemes and constructions;9 • Collostructional strength is the degree to which a given lexeme is associated with a given construction; • That is, we need to know how often it occurs with respect to the expected frequency, i.e. the one it would have were it randomly distributed accross the corpus. 9 Stefanowitsch & Gries (2003) 5 Constr ¬Constr Totals Lex O11 O12 R1 ¬Lex O21 O22 R2 Totals C1 C2 N Table 2: A sample contingency table • We need only four figures from this contingency table to calculate it: R1 , the total tokens of the lexeme in question,10 C1 , the totals for the construction, O11 , the number of co-occurrences of the two, and N, the total corpus size; • For the baseline corpus, the results are the following:11 Verb Corp. freq. Constr. freq. Exp. freq. Coll. strength εἰμί 1192 23 0.9 25.71 ἔχω 337 14 0.3 20.01 φημί 107 14 0.1 27.20 θαυμάζω 19 6 0.0 14.33 Table 3: Collostructional strength of most frequent potential optative presents (baseline) Verb Corp. freq. Constr. freq. Exp. freq. Coll. strength λέγω 216 17 0.1 31.27 γίγνομαι 209 14 0.1 24.62 δύναμαι 12 7 0.0 19.79 ὁράω 48 7 0.0 14.82 Table 4: Collostructional strength of most frequent potential optative aorists (baseline) • As most productive verbs are different in present and aorist potential optatives, the two groups are treated separately; • In computing collostructional strength e.g. for φημί in present potential optatives we look for the total occurrences of φημί in the present in our corpus; • What collostructional strength tells us here is how φημί would be expected specifically in a present potential optative. 10 All data for nonmodal categories, e.g. for the total number of present indicatives in a given subcorpus, was obtained using Perseus under PhiloLogic (https://perseus.uchicago.edu), the Diorisis corpus with Diorisis Search 3.3 (Vatri & McGillivray 2018), and the Thesaurus Linguae Graecae (https://stephanus.tlg.uci.edu), sometimes in combination. In every case the data was handchecked to discard false positives. I am indebted to Francesco Mambrini here. 11 The figures were calculated in R using the Collostructions package (Flach 2021, https://sfla.ch/collostructions/). Several statistics are available for collostructional strength measurement; here it is the negative decadic log-transformed p-value of the Fisher-Yates exact test, which was chosen because of its usefulness for small- and mid-scale corpora. See Stefanowitsch & Gries (2003, 218-220) and Levshina (2015, 213-214). 6 • Thus, φημί is indeed attracted to the optative; εἰμί and ἔχω in the present, and λέγω and γίγνομαι in the aorist are also strongly associated with potentials, their overall high frequency notwithstanding. 3.2 Aristides • Corpus: a (hopefully) representative selection of speeches; • Or. 2, To Plato in Defence of Oratory, the Sicilian declamations (orr. 5 and 6), or. 23 (To the Cities on Concord), or. 26 (Encomium of Rome), prose hymns To Zeus and To the Aegean Sea (orr. 43 and 44), Sacred Tales 1-2 (orr. 47 and 48); • The total token count: 67130, with 530 optatives; • Total potential optatives: 232, with 113 presents and 119 aorists; • Overall distribution and top frequencies in Aristides: 25 20 20 15 15 10 10 5 5 0 0 Figure 2: Present (left) and aorist (right) potential optatives (Aristides) Present (113) Aorist (119) φημί (24, 21.2%) λέγω (22, 18.5%) εἰμί (23, 20.3%) φημί (11, 9.2%) ἔχω (16, 14.1%) γίγνομαι (8, 6.7%) Table 5: Top frequencies for potential vs. nonpotential optatives (Aristides) • More than a half of the presents is furnished by three verbs: φημί, εἰμί and ἔχω; • In the aorist, the three most productive roots λέγω, φημί and γίγνομαι make for a third of the total; • On the face of it, the tendencies observed in the baseline corpus are here radicalized: again, no ‘middle ground’ in the present and a more even distribution in the aorist; 7 • A closer look reveals differences: φημί is the most frequent lexeme in the present group, while in the aorist it is second to λέγω, which, in turn, is well represented in the aorist in the baseline corpus but still not as frequent as in Aristides; • A collostructional test seems to confirm this: Verb Corp. freq. Constr. freq. Exp. freq. Coll. strength φημί 162 24 0.2 46.43 εἰμί 1066 23 1.0 24.77 ἔχω 320 16 0.3 22.84 Table 6: Collostructional strength of most frequent potential optative presents (Aristides) Verb Corp. freq. Constr. freq. Exp. freq. Coll. strength λέγω 256 22 0.2 42.25 φημί 33 11 0.0 27.71 γίγνομαι 133 8 0.1 13.76 Table 7: Collostructional strength of most frequent potential optative aorists (Aristides) • Is there something ‘special’ about φημί and λέγω? • It appears that a large number of examples involving them share both additional lexical material and a specific meaningful import; • In other words, they are constructions with a rather high degree of idiomaticity (Wulff 2008); • There are 35 instances of potential optatives with φημί in our subcorpus: 24 presents and 11 aorists; • Of these, 5 involve a sequence of φαίης ἄν and 13 of φαίη/φήσαι/φήσειεν τις ἄν. 3.2.1 φαίης ἄν • This construction can target a specific addresee but it is also used as a more general hedging device; • Cf. (3), where Aristides addresses Plato, and (4), respectively: (3) οὐ μέντἂν τοσοῦτον ἐλείποντο τοῦ τὴν Ἐπίδαυρον ἔχοντος θεοῦ, φαίης ἄν, καὶ μάλα ἐμοὶ γοῦν κατὰ νοῦν (2.153) In that case, you would say that they were not so far behind the patron divinity of Epidaurus, and I would be firmly of the same opinion.12 12 Translation Trapp (2017). 8 (4) φαίης ἂν ἄστη πολλὰ συνελθόντα εἰς ταυτὸν οἰκεῖν καὶ εἶναί τινας ὥσπερ ἀστυγείτονας χρόνῳ συνοίκους γεγενημένους, αὐξηθέντων αὐτοῖς τῶν ἐπ᾽ ἀλλήλους ἀεί. (23.13) You would say that many towns have merged and settled together, and that as if they had been neighbors, some people in the course of time had come to live together as their mutual borders had considerably expanded.13 • The problem of the construction’s sources is peculiar. In the classical corpus, we find 2 tokens in Xenophon and 23 in Plato; • Most of the latter (18) occur in questions and serve an important discursive function of highlighting an answer or biasing the interlocutor towards it: (5) —ἀνδρεῖον γὰρ τοῦτο γε ἴσως αὖ φαίης ἂν εἶναι. —ἔγωγε. (Pl. Alc. I 127a) —Because, I expect you will say again, that is a man’s affair. —I would.14 • The construction resurfaces only in Imperial prose and quickly becomes quite widespread; • There are 3 instances in Plutarch, 8 in Lucian, Galen and Aelian. Most are used in declaratives, but unlike in Plato, they do not highlight an answer but seem to have a more general hedging function; • While the construction’s form is identical in both classical and Imperial authors, there is a discrepancy in its meaning: it is hardly ever used as a hedging device in the classical corpus.15 • Thus, what we find in Aristides and in Imperial-era prose more generally cannot be attributed to mere copying, as its meaning range is broader. 3.2.2 φημίprs/aor.3sg τις ἄν • Another frequent construction in Aristides is φαίη/φήσαι/φήσειεν τις ἄν: it is represented by 13 tokens, with 11 presents and 2 aorists; • Like φαίης ἄν, it is used as a speech act modifier of sorts: (6) μέχρι μέντοι τῶν νῦν τούτων ὑπ’ αὐτῶν τῶν λόγων ἐξελέγχεσθαι Πλάτωνα φαίη τις ἄν· βούλομαι δ’ αὐτὸν καὶ ὑπὸ τῶν αὑτοῦ λόγων ἐξελεγχόμενον φανῆναι. (2.50) Up to this point it might be said that Plato is shown to be in the wrong by the arguments themselves; but I want it also to be clear that he is proved wrong by his own words as well. 13 Translation Behr (1981). Translation Lamb (1927). 15 On a side note, it seems to have a more similar function in Homer, with the form of φαίης κε and a fixed metrical position in the beginning of the line (or, if negated, at its end). 14 9 (7) οἱ δὲ ἄρχοντες οἱ πεμπόμενοι ἐπὶ τὰς πόλεις τε καὶ τὰ ἔθνη τῶν μὲν ὑφ᾽ ἑαυτοῖς ἕκαστοι ἄρχοντές εἰσι, τὰ δὲ πρὸς αὑτούς τε καὶ πρὸς ἀλλήλους ὁμοίως ἅπαντες ἀρχόμενοι. καὶ δὴ καὶ τοῦτο φαίη τις ἂν αὐτοὺς τῶν ἀρχομένων διαφέρειν, ὅτι πρῶτοι δεικνύουσιν ὅπως ἄρχεσθαι προσῆκε (26.31) The rulers who are sent to the cities and to the peoples are each the rulers of those under them, but in regard to their personal position and their relations to each other are equally subjects. And, indeed, one would say that in this respect they differ from their subjects, in that they first teach the duties of a subject.16 • This construction, too, has an Attic genealogy (though it is not found in the baseline corpus): 2 tokens in Isocrates, 5 in Xenophon, 16 in the Platonic corpus, and 10 in Demosthenes; • The following is a representative example: (8) νῦν γὰρ φαίη ἄν τίς σοι λόγῳ μὲν οὐκ ἔχειν καθ’ ἕκαστον τὸ ἐρωτώμενον ἐναντιοῦσθαι (Pl. Resp. 487c) In the present situation someone might say to you that he can’t argue against each individual question theoretically.17 • Compared to φαίης ἄν, there is more continuity throughout the history of Greek: a handful of examples occur in Hellenistic prose as well. • However, a noticeable increase in frequency is attested only from the early Imperial period onwards, beginning with Dionysius of Halicarnassus and Philo. 3.2.3 εἴποι/λέγοι τις ἄν • Another construction which seems to have an idiomatic character, albeit to a somewhat lesser degree than the ones just discussed, is εἴποι or λέγοι τις ἄν;18 • It accounts for 11 of the 25 instances of λέγω in the potential group, present and aorist combined; • On the functional side, it seems to have the range of functions typical of the potential optative generally: (9) οἶμαι τοίνυν ἅπαντας ἂν οἷς δυνατὸν συμφῆσαι τὸ μὴ κατὰ ταὐτὰ φῦναι τοὺς ἀνθρώπους, ἀλλὰ δυοῖν μερίδοιν τὴν μὲν οἵαν βιάζεσθαι καὶ πλεονεκτεῖν εἶναι, ἣν τῶν κρειττόνων τις ἂν εἴποι, λέγω δ’ οὐκ ἀρετῇ βελτιόνων, ἀλλ’ ἰσχυροτέρων, τὴν δ’ οἵαν ἐλαττοῦσθαι καὶ παρὰ γνώμην συγχωρεῖν ἀπορίᾳ τοῦ κωλύειν, ἣν τῶν ἀσθενεστέρων οὖσαν γιγνώσκομεν. (2.206) I think that all who are in a position to do so would agree that human beings are not all naturally the same, but divide into two classes. One of these two classes characteristically uses force and takes 16 Translation Behr (1981). Translation Emlyn-Jones & Preddy (2013). 18 Aristides’ fondness of it has been indicated before: Vix (2010, 498). 17 10 the larger share; one might call it the superior class, meaning by that not better in moral character but physically stronger. (10) ὥστε καὶ σύνδεσμον τὴν ῥητορικὴν τοῦ παντὸς ὀρθῶς ἂν καὶ <κατὰ> τοῦτο εἴποι τις. (2.424) So in this respect too, it would be right to call oratory the bond that holds the universe together. • Its high frequency is still remarkable; • It is also amply attested across the whole of pre-Imperial Greek prose, including Hellenistic, and cannot be attributed to a specific source. οὐδ᾽/μηδ᾽ ἄν εἷς + optative • Another peculiar construction with specified lexical slots has the form οὐδ᾽/μηδ᾽ ἄν εἷςsubj. ; • Here, the specificity is in the form of the negator, not in a preference for specific verbs; • There are 7 examples in our corpus, all of them in orations 2 and 23: (11) φυγὰς δὲ καὶ σφαγὰς κοινὰς τῶν Ἑλλήνων καὶ πανωλεθρίᾳ πράγματα ἀπολλύμενα οὐδ᾽ ἂν εἷς λέγων ἐφίκοιτο. (23.49) No one would sufficiently describe the exiles, the mutual slaughter of the Greeks and the fortunes ruined in the general destruction. (12) τῷ μὲν γὰρ αὔξειν ἐθέλειν τὰς αὑτῶν πατρίδας καὶ τὰ ὑπάρχοντα κοσμεῖν οὐδ᾽ ἂν εἷς νεμεσήσειεν, ὃ δὲ τούτῳ πρόσεστιν ἀηδίας ἀφελεῖν ἔγωγε συμβουλεύω (23.68) No one would resent your wish to exalt your countries and to honor what you have but I advise you to dispense with the unpleasantness connected with this. • In the whole of the Aristidean corpus, there are 25 tokens; • In Attic prose we have two instances in Isocrates, Lysias and Xenophon, 9 in Plato, and 26 in Demosthenes; • It is found in our baseline corpus also (3 times in Demosthenes 18 and once in the Protagoras): (13) ποῦ δὲ παιδείας σοὶ θέμις μνησθῆναι; ἧς τῶν μὲν ὡς ἀληθῶς τετυχηκότων οὐδ᾽ ἂν εἷς εἴποι περὶ αὑτοῦ τοιοῦτον οὐδέν, ἀλλὰ κἂν ἑτέρου λέγοντος ἐρυθριάσειε (Dem. 18.128) And what gives you the right to talk about education? No truly educated person would ever say such things about himself, but he’d blush just to hear someone else say them.19 • Given its prominence in Demosthenes, it may have been particularly strongly indexed to his style; 19 Translation Yunis (2005). 11 • In the Imperial period, it is especially favored by Aristides, also sporadically present in Plutarch, Dio, Marcus Aurelius, with one instance in each, and in Lucian,20 who has two; • With this in view, if we compare Aristides with the baseline data in tables 3 and 4, we can see that many of the lexemes overlap but at the same time display a significantly higher degree of association; • This is especially true of φημί in present potential optatives: 46.43 in Aristides vs. 27.20 in the baseline subcorpus; • Again, λέγω presents a similar picture in the aorist: 42.25 in Aristides vs. 31.27 in the classical subcorpus; • Finally, the very presence of φημί in the potential aorist group and its rather high associational strength in Aristides are also remarkable: there is simply no parallel to this in the baseline data. 4 Conclusion • Three of the four constructions discussed likely have identifiable classical sources; • This strong presence of idiomatic potential constructions is an instance of what may be called constructional mimesis, which is a hallmark of Aristides’ prose on the whole; • As shown by Pernot (1981, 90-2) on the basis of the Sicilian declamations, Aristides reuses material from classical prose (particularly, but not exclusively, from Demosthenes), from more ‘grammatical’ constructions to much more specialized turns of phrase; • They thus become salient shibboleths indexed to a particular author and thereby to that ‘core’ stratum of Imperial-era prose which may be rightly called Atticizing; • They mark his speech (recall that most of it was meant to be delivered orally and publicly) as targeted at those who can recognize and appreciate the way in which he inherits the canon and in particular Demosthenes, the highest canonical authority; • This of course constructs and reinforces Aristides’ own authority, but also that of his listeners, which is the prime effect of community-specific speech; • The audience of ‘core’ Atticizing prose consisted of two large groups, ‘general listeners’ and ‘insiders,’ the latter being the primary target audience and potential rivals of the practising sophist (Korenjak 2000, 52-6, Zweimüller 2008, 89-107, Matthaios 2013); • One may surmise that they, as a community, were the ‘target audience’ of shibboleths, whose effect was something like the following (Pernot 1981, 92): 20 If we accept the authenticity of his encomium of Demosthenes; see Marquis (2010) for an argument in favor of this view. 12 Pris isolément, chacun de ces rapprochements pourrait être imputé au hasard ; mais leur nombre interdit cette interprétation. En raison de la banalité des expressions empruntées, on ne peut penser non plus à des citations intentionnelles. Aristide est imprégné de Démosthène au point de calquer involontairement ses expressions… Si l’auditoire n’identifiait pas chaque emprunt au passage, leur accumulation lui devait laisser l’impression d’un style démosthénien. • Importantly, the functional import of the shibboleths could be different, as we have seen that in some cases their functional range is extended compared to classical prose; • What matters more is their social, not descriptive, meaning; • Grammatical divergences between classical and Atticizing Greek should not (or at least not a priori) be written off as deviations, since more complex mechanisms may be at work; • Imperial Greek mimesis, whether literary or linguistic, was a fundamentally creative process. References Anlauf, G. (1960), Standard Late Greek oder Attizismus? Eine Studie zum Optativgebrauch im nachklassischen Griechisch, PhD thesis, Universität zu Köln. Austin, J. L. (1962), How to Do Things with Words, Harvard University Press, Cambridge, MA. Behr, C. A. (1981), Aelius Aristides. The Complete Works. Vol II: Orations XVII-LIII, Brill, Leiden. Bers, V. (1984), Greek Poetic Syntax in the Classical Age, Yale University Press, New Haven. Bianconi, M. & Magni, E. (2022), ‘The survival of the optative in New Testament Greek’, Journal of Greek Linguistics 23(1), 36–78. Boulanger, A. (1923), Aelius Aristide et la sophistique dans la province d’Asie au IIe siècle de notre ère, de Boccard, Paris. Bowersock, G. W. (1969), Greek Sophists in the Roman Empire, Clarendon, Oxford. Bowie, E. (1982), ‘The Importance of Sophists’, Yale Classical Studies 27, 29–59. Eckert, P. (2008), ‘Variation and the indexical field’, Journal of Sociolinguistics 12(4), 453–476. Emlyn-Jones, C. & Preddy, W. (2013), Plato: Republic, Volume II, Books 6-10, Loeb Classical Library 276, Harvard University Press, Cambridge, MA. Flach, S. (2021), Collostructions: An R implementation for the family of collostructional methods. García Valdés, M., Llera Fueyo, L. A. & Rodríguez-Noriega Guillén, L. (2009), Claudius Aelianus, De Natura Animalium, Teubner, Leipzig. 13 Goldberg, A. E. & Jackendoff, R. (2004), ‘The English Resultative as a Family of Constructions’, Language 80(3), 532–568. Herbert, C. & Kukla, R. (2016), ‘Ingrouping, Outgrouping, and the Pragmatics of Peripheral Speech’, Journal of the American Philosophical Association 2(4), 576–596. Higgins, M. J. (1945), ‘The Renaissance of the First Century and the Origin of Standard Late Greek’, Traditio 3, 49–100. Jones, C. P. (2008), Aristides’ First Admirer, in W. V. Harris & B. Holmes, eds, ‘Aelius Aristides between Greece, Rome, and the Gods’, Brill, Leiden, pp. 251–262. Korenjak, M. (2000), Publikum und Redner. Ihre Interaktion in der sophistischen Rhetorik der Kaiserzeit, C. H. Beck, München. Kukla, R. & Lance, M. (2009), ‘Yo!’ and ‘Lo!’: The Pragmatic Topography of the Space of Reasons, Harvard University Press, Cambridge, Mass. Lamb, W. M. R. (1924), Plato, Volume III, Loeb Classical Library 165, Harvard University Press, Cambridge, MA. Lamb, W. R. M. (1927), Plato: Charmides. Alcibiades I and II. Hipparchus. The Lovers. Theages. Minos. Epinomis, Loeb Classical Library 201, Harvard University Press, Cambridge, MA. Levshina, N. (2015), How to do Linguistics with R: Data exploration and statistical analysis, John Benjamins, Amsterdam. Marquis, É. (2010), ‘Des voix pour montrer la voie dans l’Éloge de Démosthène de Lucien’, Bulletin de l’Association Guillaume Budé 2, 73–87. Matthaios, S. (2013), Pollux’ Onomastikon im Kontext der attizistischen Lexikographie, in C. Mauduit, ed., ‘L’Onomasticon de Pollux : aspects culturels, rhétoriques et lexicographiques’, de Boccard, Paris, pp. 67– 140. Pernot, L. (1981), Les Discours siciliens d’Aelius Aristide, Arno Press, New York. Quaß, F. (1993), Die Honoratiorenschicht in den Städten des griechischen Ostens, Franz Steiner, Stuttgart. R Core Team (2021), R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna. Reardon, B. P. (1971), Courants littéraires grecs des IIe et IIIe siècles après J.-C., Les Belles Lettres, Paris. Rutherford, I. (1998), Canons of Style in the Antonine Age, Oxford Univeristy Press, Oxford. Schmid, W. (1889), Der Atticismus in seinem Hauptvertretern von Dionysius von Halikarnass bis auf den zweiten Philostratus, Vol. II, Kollhammer, Stuttgart. 14 Schmid, W. (1902), Bericht über die Literatur aus den Jahren 1894-1900 zur zweiten Sophistik, in L. Gurlitt & W. Kroll, eds, ‘Jahresbericht über die Fortschritte der klassischen Altertumswissenschaft’, Vol. 108, Reisland, Leipzig, pp. 212–280. Schmitz, T. A. (1997), Bildung und Macht. Zur sozialen und politischen Funktion der zweiten Sophistik in der griechischen Welt der Kaiserzeit, C. H. Beck, München. Silverstein, M. (2017), ‘Standards, Styles, and Signs of the Social Self’, Journal of the Anthropological Society of Oxford Online 9(1), 134–64. Stefanowitsch, A. & Gries, S. T. (2003), ‘Collostructions: Investigating the interaction of words and constructions’, International Journal of Corpus Linguistics 8(2), 209–243. Swain, S. (1991), ‘The Reliability of Philostratus’s “Lives of the Sophists”’, Classical Antiquity 10(1), 148–163. Swain, S. (1996), Hellenism and Empire: Language, Classicism, and Power in the Greek World AD 50–250, Oxford Univeristy Press, Oxford. Tantau, T. (2013), The TikZ and PGF Packages. Trapp, M. (2017), Aelius Aristides. Orations, Volume I, Harvard University Press, Cambridge, MA. Tribulato, O. (2022), Stylistic terminology in the Praeparatio sophistica. Talk delivered at the 2nd PURA (Purism in Antiquity) Workshop, Venice, 26 September 2022. Vatri, A. & McGillivray, B. (2018), ‘The Diorisis Ancient Greek Corpus’, Research Data Journal for the Humanities and Social Sciences 3(1), 55–65. Veyne, P. (1976), Le Pain et le cirque : sociologie historique d’un pluralisme politique, Seuil, Paris. Vix, J.-L. (2010), L’enseignement de la rhétorique au IIe siècle ap. J.-C. à travers les discours 30-34 d’Ælius Aristide, Brepols, Turnhout. Whitmarsh, T. (2005), The Second Sophistic, Cambridge University Press, Cambridge. Wickham, H. (2016), ggplot2: Elegant Graphics for Data Analysis. Wilson, N. G. (2010), ‘Review of García Valdés et al. 2009’, Exemplaria Classica 14, 403–405. Wulff, S. (2008), Rethinking Idiomaticity, Continuum. Yunis, H. (2005), Demosthenes, Orations 18 and 19, University of Texas Press, Austin. Zweimüller, S. (2008), Lukian, »Rhetorum praeceptor«. Einleitung, Text und Kommentar, Vandenhoeck & Ruprecht, Göttingen. 15