ἐφίορκος, ἐπίορκος
(Phryn. Ecl. 279)

D. General commentary

The above entry in Phrynichus’ Eclogue (A.1) deals with ἐφίορκος, a variant of the adjective ἐπίορκος ‘forsworn’. Both the adjective and its denominal verb ἐπιορκέω ‘to swear falsely, forswear; to break one’s oath’ (which likewise has a variant ἐφιορκέω) are attested since Homer as antonymsAntonyms of εὔορκοςεὔορκος ‘keeping one’s oath’ and εὐορκέωεὐορκέω ‘to swear truly; to keep one’s oath’. In particular, the antonymic pair εὐορκέω/ἐπιορκέω recurs, in participial form, in the epigraphic formula εὐορκοῦντι μέν μοι εὖ εἴη, ἐπιορκοῦντι δὲ τὰ ἐναντία ‘if I swear truly may it be well with me; if I swear falsely, [may it be] the opposite’ (vel sim.; less often, in an explicit hypothetical clause of the kind εἰ μὲν εὐορκήσαιμι … εἰ δ’ ἐπιορκήσαιμι). This formula typically occupies the final part of oaths and is ubiquitous in literary and non-literary texts (for a history of the formula see Jouanna 2018, 81–5, with references to earlier studies). The respective meanings of εὐορκέω and ἐπιορκέω were explained by Cleanthes (fr. 581 SVF = Stob. 3.28.17).

Phrynichus proscribes ἐφίορκος in favour of ἐπίορκος. Indeed, ἐπι- is the only spelling found in literary sources: both the verb (C.1) and the adjective (C.2) are first attested in the Iliad, while, according to Jouanna (2018, 81), the first prose appearance occurs in the Hippocratic Oath (C.3), although the dating of this work is controversial. The adjective is rarer than the verb in classical times, but both are well-attested in Attic drama (see e.g. C.4, C.5) and prose (particularly in the orators but also in Plato and Xenophon: C.6, C.7, C.10) as well as in later literature. The aspirated variant, meanwhile, occurs frequently in inscriptionsInscriptions from the 5th century BCE onwards (see e.g. C.9, an amphictyonic decree in Delphic dialect; ἐ̣φ̣[ιορκο͂σι in the Athenian decree for Erythrae [IG 1,3.14.17, 453/2 BCE] is uncertain, see Threatte 1980, 505) and in papyriPapyri (see Gignac 1976, 137; Mayser, Gramm. vol. 1.1, 176; Jouanna [2018, 84] observes that in papyri ἐφιορκοῦντι outnumbers ἐπιορκοῦντι by 35 to 24); it is also found as a varia lectio in some manuscripts of the New TestamentNew Testament (Ev.Matt. 5.33, 1Ep.Tim. 1.10; see Blass, Debrunner 1976, §14 n. 2). This cumulative evidence, corroborated by attestations in Medieval Greek (E.), suggests that the stem ἐφιορκ- was ubiquitous in low-level koine. However, it is notably absent from Attic inscriptions (C.11): ἐπιορκοῦμαι ‘was the normal spelling of the word at Athens […] although the form in most dialects was ἐφιορκοῦμαι’ (Threatte 1980, 505), especially in North-West Greek (Sommer 1905, 52; see, e.g., C.12). Given this distribution, it is clear why Phrynichus condemned ἐφίορκος as non-Attic.

The origin of the forms in ἐφι-, which may be key to understanding their status and distribution, remains inadequately studied. In principle, the anticipation of aspiration on /p/ manifests only when the final /i/ of the prefix undergoes elisionElision before a second member with initial aspiration (cf. ἐπ(ι)- + ἵππος > ἔφιππος ‘on horseback’). For the elision to be possible, however, the aspiration itself must no longer be articulated as a full consonant (a phenomenon that Jatteau 2016, 193–206 explains with reference to the ‘non-sequential’ character of the Greek aspiration). However, there is no apparent reason for the anticipation of aspiration without the elision of /i/. Parallels are few and not wholly unproblematic: ἐφιάλτης ~ ἐπιάλτης ‘nightmare’ is of uncertain etymology and may have undergone tabooistic deformations (see DELG, EDG s.v.); the Pamphylian dialect offers two epigraphic examples of an apparently similar phenomenon (Brixhe 1976, 72–3).

Some proposed explanations assume that ἐφιορκ- arose not from a phonological process but from some kind of analogyAnalogy. In Ionic inscriptions, Dunkel (2014, 250–1 n. 46) assumes that it is a hypercorrection, ‘ein psilotischer Versuch, ἐπι ὁρκέω (sic) auszusprechen’. The idea that the ‘misplaced’ aspiration in ἐφιορκέω is an Ionic peculiarity goes back at least to the Etymologicum Gudianum (B.2, F.1); however, it is problematic, because this variant is in no way confined to Ionic inscriptions. Thumb (1889, 72) explained ἐφιορκέω as a compromise form between ἐπιορκέω and ἐφορκέω, but the latter is essentially unattested in the classical age (see E. for Byzantine forms), with the lone and late exception of ἐφορκέοντι (Chaniotis, Verträge 61B.203 [Souda, 110/09 or 109/8 BCE]), which is known only through an unreliable transcription (see Chaniotis 1996, 362–76); in BGU 16.2593.10 (= TM 23316) [Phebichis, 30 BCE–14 CE], the integration ἐφορ]κοῦντι is unwarranted, since in the missing left half of the papyrus the lacuna is ample enough (ca. 25 letters: see Brashear 1995, 70–1) to allow for a reconstruction ἐφιορ]κοῦντι, the usual spelling in papyri from the same time and place. Finally, it may be argued that ἐφίορκος arose by means of false segmentation based on the analogy of forms such as ἔφιππος, in which the /i/ belongs to the second member.

However, a phonological explanation may be supported. Sommer (1905, 47–9) proposed that the initial aspiration of the second compound member (ὅρκος, ὁρκέω) was anticipated to the prefixPrefixes, as in *προ-ὁρά > φρουρά (see the entry φροίμιον, φροῦδος). Given that the final vowel of ἐπι- (unlike that of προ-) is typically elided before another vowel, it is necessary to explain why the preverb was not elided in the first place to give the expected *ἐφορκέω. One may compare the much later formation ἐφορκίζωἐφορκίζω, a variant of the commoner ἐπορκίζω, attested in Christian texts since the late 1st century CE as a synonym of (ἐξ)ορκίζω ‘to exorcise’ (on which see entry ὁρκόω, ὁρκωτής, ὁρκίζω) and the epithet ἐφόρκιοςἐφόρκιος (Hsch. ε 7527Hsch. ε 7527 ἐφόρκιος· Ζεύς, ἐν †Κιτάνω τάματι), possibly a later formation and almost certainly unconnected with ἐπίορκος, since the prefix ἐπι- here must mean ‘upon’ rather than ‘against’ (cf. Ζεὺς ὅρκιος ‘Zeus guardian of oaths’). This question is connected with the origin of ἐπίορκοςἐπίορκος: if it came from the univerbation of a phrase such as ἐπὶ ὅρκῳ/-ον βαίνειν ‘to go against the oath’ (Schwyzer 1927, 255–8) or ἐπὶ ὅρκον ὀμόσσαι ‘to add an oath (to one’s declaration)’ (Leumann 1950, 79–92), it would stand to reason that the phonetic integrity of ἐπι- was preserved. De Lamberterie (apud Jouanna 2018, 94 n. 18) argues instead that ἐπιορκ- is a phonological archaismArchaisms, reflecting a time when /h/ was a fully-fledged consonant that could prevent elision (as in ἐπιάλμενος besides ἐφάλμενος or Ἀντίοχος besides ἀντέχω). The ‘interaspiration’ of a second compound member is indeed attested in inscriptions and several old manuscripts as well as in learned Latin loans, such as parhedrus; according to several ancient grammatical sources, including Aristarchus, Apollonius Dyscolus, and Herodian, its preservation depended on the perception of the word as a true compound (on the problem see Schironi 2004, 507–10; Cassio 2018; Jatteau 2016, 187–92 argues for the phonetic reality of interaspiration). Lejeune (1987, 318–9) assumed that anticipation of aspiration always preceded elision (*ἔπι+hέρπω > *ἐπhιέρπω > ἐφέρπω), making ἐφιορκέω the more archaic form, preserved in virtue of its formulaic nature. Alternatively, the /i/ of ἐπι- may have undergone loss of syllabicity before another vowel, and the aspiration – assuming that it had been phonetically preserved up to that point – ‘jumped’ to the voiceless labial stop, similar to that which occurred with the crasis of προ- in forms such a φρουρά (see Batisti 2022, 16–9). Loss of syllabicity, like crasisCrasis, is a strategy for resolving hiatus, typical of the spoken language (Devine, Stephens 1994, 266–9); forms resulting from crasis tended to be accepted into the written language only if they had a high frequency of use (Pardal Padín 2015). The diffusion of ἐφιορκ- in lower-level texts and its absence from literary contexts may thus reflect its origins as a spoken formColloquial language. Should this reconstruction be correct, the case of ἐφιορκέω would be somewhat peculiar, in that it likely reflected a colloquial realisation of a form that in itself was phonologically conservative. It appears that Attic reacted against the spread of a lower-level variant in a standardised official formula, a fact which explains the different treatment of ἐφιορκέω and ἐφίορκος in Atticist lexicography compared to forms like φροίμιον or φροῦδος.

At any rate, there is no indication that ancient scholars understood ἐφιορκέω in terms of the same rule that they recognised in those forms; rather, ἐφιορκέω was regarded as a non-Attic form (possibly Ionic, see F.1), similarly to other cases of oscillation between aspirated and unaspirated voiceless stop, like βρύκωβρύκω ~ βρύχω ‘to bite’ (Moer. β 24Moer. β 24, Philemo (Laur.) 356Philemo (Laur.) 356, prescribing the unaspirated form as Attic) or γρῖπος ~ γρῖφοςγρῖφος ‘fishing net; riddle’ (Philemo (Laur.) 357Philemo (Laur.) 357, prescribing the aspirated form as Attic). Indeed, Phrynichus himself offers relatively frequent remarks on pairs of variants distinguished by the presence of aspirated or unaspirated consonants. In compounds, the change of a voiceless stop to the corresponding aspirate before a vowel-initial second member was a useful indication that the latter had initial aspiration. The preservation of word-initial /h/ was regarded as a defining trait of classical Attic, but precisely for this reason it was likely to be unduly extended; Atticists debated cases when the aspiration (or lack thereof) in the genuine Attic form ran contrary to expectations (see Vessella 2018, 101–3). A case in point is the entry ἄπεφθονἄπεφθος (‘boiled down’) of the Praeparatio sophistica (25.16–25)Phryn. PS 25.16–25, under which Phrynichus treats several other exceptions to the rule, changing π and τ to φ and θ, respectively, in compounds (Vessella 2018, 163–5). In cases wherein the variation between aspirated and unaspirated consonants occurred in the word’s stem, the preferred variant was that more frequently employed by canonical authors, such as πανδοκεῖονπανδοκεῖον ‘inn’ prescribed against πανδοχεῖον (Phryn. Ecl. 275)Phryn. Ecl. 275, or conversely μοχλόςμοχλός ‘crowbar’ against μοκλός (Ecl. 277)Phryn. Ecl. 277. As such, ἐφιορκέω represents something of an intermediate case as a compound whose second member had initial aspiration, but, at the same time, it is irregular, given that with no elision of /i/, the aspiration rule should not have applied.

E. Byzantine and Modern Greek commentary

The forms ἐφίορκος and ἐφιορκέω occasionally appear in late ancient and medieval Greek texts, from Didymus Caecus (in Eccl. 234.15, 235.10, in Ps. 69.14) in the 4th century CE to Byzantine prose and poetry, especially in works that reflect the spoken language (the Alexander Romance [C.13], the Chronicon Moreae, the Bellum Troianum, Michael Ducas’ Historia Turcobyzantina), up to the 17th century (Neophytus Rhodinus). Higher-register authors, such as Nicetas Choniates, only use ἐπιορκ-. In texts of a more vernacular register, these words show phonetic variation, reflecting sound changes typical of the spoken language (see Kriaras, LME s.vv.), such as aphaeresis in ’φιορκία, ’φιορκώ, changes in the timbre of the initial vowel in αφιορκία, αφίορκος, αφιορκώ (especially frequent in cod. H of the Chronicon Moreae; for initial /e/ > /a/ see CGMEMG vol. 1, 95–8), and – most interestingly – synizesis of /i/ (implied by the accent shift in αφιόρκος), followed by hardening of the glide /j/ to a palatal fricative /ç/ or a palatal stop /c/ (CGMEMG vol. 1, 105), implied by the spelling <κι> in ’φκιόρκος and ’φκιορκώ in the Cretan Apocalypse of the Virgin (see now Lampakis 2017). Of course, given that synizesis of /i/ in hiatus is a widespread feature of Medieval and Modern Greek (CGMEMG vol. 1, 61–3), the latter forms do not by themselves prove that the metathesis of aspiration in ancient ἐφιορκέω was triggered by synizesis; however, they do verify that variants with -φ- continued to be part of the living language. Also attested in the same period are the forms ἐφορκέω, ἐφορκία, ἔφορκος (see LBG s.vv.), which show the expected treatment with elision and aspiration. In the absence of any classical antecedents for these forms (see D.), they may be regarded as analogical renewals on the model of other compounds in ἐπ(ι)- and possibly under the influence of ἐφορκίζω ‘to exorcise’, unless they represent attempts to spell ἐφιορκ- with loss or non-notation of the glide /j/. Despite the diffusion of innovative forms in Byzantine sources, the Atticist-approved stem ἐπιορκ- ultimately won out, likely aided by its nature as a technical term of legal language: only επιορκία, επίορκος, and επιορκώ survive – as learned terms – in Modern Greek (see LKN s.vv.).

F. Commentary on individual texts and occurrences

(1)    Et.Gud. 522.36–42 (B.2)

This entry in the Etymologicum Gudianum seeks to explain the aorist participle ταφώνταφών ‘astonished’ from an (unattested) present *θήπω (in addition to the aorist, only the perfect τέθηπα is attested for this verb: see Barton 1993 on the distribution of these and related forms in Homer), positing an intermediate form *θαφών and an ‘Ionic’ change from θ to τ. The alleged IonicIonic tendency to exchange aspirated and plain voiceless stops is then exemplified by an instance of the opposite treatment: ἐπιορκῆσαι > ἐφιορκῆσαι. This doctrine is oddly combined with a different explanation of ταφών as derived metaphorically from τάφοςτάφος ‘tomb’, which was widespread in antiquity (Neitzel 1977, 289–90). The derivation of ταφών from a present *θήπω, which finds support in the cognate forms θαμβέωθαμβέω ‘to be astounded’ and θάμβοςθάμβος ‘amazement’, goes back to the Augustan-age grammarian Apion (fr. 136 Neitzel = Apoll. Lex. 149.32–5) and is reiterated in several erudite sources (schol. (Ep.Hom. | Did.?) Hom. Il. 9.153b (A), Et.Parv. θ 9, EM 748.53, [Zonar.] 1715.12, Eust. in Il. 1.741.4–15, 2.696.2–697.2, in Od. 1.62.18, etc.), often combined, as here, with the etymology from τάφος. A fuller exposition is found in the parallel entry of the Etymologicum Magnum, which lays out the intermediate steps more clearly: ἐκ τοῦ θήπω, ἔθαπον, καὶ ἔταφον Ἰωνικῶς, καὶ ταφών ‘from (present) θήπω [is derived the aorist] ἔθαπον, and [it becomes] ἔταφον in Ionic, and [the participle is] ταφών’. For the idea that Ionic was characterised by metathesis of aspiration in both directions with respect to Attic, see F.2. The example of ἐφιορκῆσαι, however, is adduced only in the Gudianum.

(2)    Eust. in Il. 1.741.4–15 (B.3)

Commenting on Hom Il. 4.243 τίφθ᾿ οὕτως ἔστητε τεθηπότες ἠύτε νεβροί ‘why do you stand around this way, dazed like fawns’, Eustathius explains the perfect participle τεθηπότες as deriving from the present θάπτωθάπτω ‘to bury’ (rather than from the *θήπω postulated in other erudite sources, see F.1): the immediately following remark ὅπερ ἄγειν οἶδε τοὺς συμπαθεῖς εἰς ἔκπληξιν alludes to the metaphorical derivation from τάφοςτάφος ‘tomb’. Intermediate forms θάπος and θαπών are then posited to explain τὸ τάφος (‘astonishment’, homophonous with the noun meaning ‘tomb’ and attested in the Odyssey) and ταφών. As examples of the Ionic tendency to exchange aspirated and unaspirated consonants in two consecutive syllables, Eustathius first gives three forms attested in the literary Ionic of Herodotus (ἐνθεῦτεν, κιθών, βάθρακος); further examples are taken from the uneducated speech of Eustathius’ own time, such as the form ἀχάντια ‘cotton thistles’, here attributed to ‘eastern men’ (in 2.696.10, the same form is attributed to contemporary ‘rustic speakers’: ἔστι δὲ μέχρι νῦν ἀκοῦσαι πολλοὺς τῶν ἀγροίκων οὕτω τὰ ἀκάνθια ἀχάντια λέγοντας), the compound μονόκυθρον ‘one-pot (meal)’ (attested in medieval authors, see LBG s.v.) implying the Ionic κύθρα ‘pot’ rather than the Attic χύτρα, or the variant πάσχα for φάσκα (the latter was the form used for the Hebrew Passover since Iosephus, while the former spelling tended to be reserved for the Christian Easter; see Kriaras, LME s.v. Πάσχα). Another alleged example is the word κύσθος ‘female genitals’, which Eustathius connects etymologically with χέω ‘to pour’ and χύσις ‘flow’. With the exception of πάσχα/φάσκα, all of the above examples are repeated in Eust. in Il. 2.696.2–697.2, where the derivation of ταφών is given in the commentary on Il. 9.193 ταφὼν δ’ ἀνόρουσεν Ἀχιλλεύς ‘and Achilles leapt up in amazement’.

(3)    Lys. 10.17 (C.8)

In his oration Against Theomnestus 1, Lysias quotes repeatedly from Solon’s laws and comments on their archaic lexical usages to prove that a law’s intention remains in full force even if the words’ meanings have changed over time. In this particular passage, Lysias claims that the participle ἐπιορκήσαντα in the law’s text is the equivalent of ὀμόσας ‘having sworn’, a meaning unparalleled in other Greek sources, where ἐπιορκέω invariably means ‘to forswear’. Lysias’ remark is repeated in ancient and medieval scholarship (Harp. ε 106, after which ὀμόσαντα is restored in the text; Su. ε 2499; Phot. ε 1653) and is mentioned by all modern dictionaries (see e.g. LSJ s.v.). Nonetheless, Jouanna (2018)    has recently shown that Lysias’ interpretation was mistaken, and that the phrase referred to the ‘the moment whereby the one who swears tells what should happen to himself if he swears falsely (ἐπιορκήσαντα), and gives the name of the God to call as witness at that time (Apollo)’. If this alleged exception is removed, the meaning ‘to forswear’, stably attested since Homer, may be accepted as the original.