ἀπέκτονα, ἀπέκταγκα
(Moer. α 70, Philemo [Laur.] 355, Thom.Mag. 8.9)

D. General commentary

An entry in Moeris’ Atticist lexicon (A.1) discusses the perfect forms of the verb ἀποκτείνω ‘to kill’, rejecting the innovative formInnovative forms ἀπέκταγκα in favour of the older ἀπέκτονα. Entries in the lexica of Philemon (A.2) and Thomas Magister (A.3) likely address the same issue, although their textual status is problematic (see F.1).

The ablauting root *tken-/*tkon-/*tkn̥-, originally formed an o-grade active perfect ἀπέκτονα (never found uncompounded) next to the e-grade *-i̯ō-present (ἀπο)κτείνω (the compound is the norm in Attic prose and comedy), the athematic nasal present (ἀπο)κτ(ε)ίννυμι (with a vocalism influenced by the thematic present: see F.3), the ‘asigmatic’ aorist (ἀπ)έκτεινα, and the thematic root aorist (ἀπ)έκτανον. While ἀπέκτονα is the norm in classical Attic prose, with abundant attestations in Plato (e.g. C.1), Lysias (e.g. C.2), Xenophon, and Demosthenes, slightly later a new perfect ἀπέκταγκα was created, with the ablaut grade of the aoristAorist ἀπέκτανον and the productive kappatic suffix. The innovative perfect is first attested in Aristotle (C.3) and apparently in a fragment of Menander (C.4) transmitted by an Atticist entry in the Synagoge tradition (B.4), possibly going back to Orus (see F.3 for the textual difficulties). Forms of ἀπέκταγκα are also occasionally found in the koine (Plb. 3.786.11 = C.5, 3x in LXX, D.S. 4.55.4 = C.6, who also employs the pluperfect ἀπεκτάγκει at 14.47.2). Polybius (11.18.10), however, also attests to ἀπεκτακότες, from a nasal-less stem ἀπεκτακ- that is found as a varia lectio in the other texts too (see C.3, C.5, C.6). It is possible that both innovative variants ἀπέκτακα and ἀπέκταγκα coexisted; indeed, Thomas Magister’s entry, as restored by Buttmann (see F.1), would imply a descending order of preference from ἀπέκτονα to ἀπέκτακα to ἀπέκταγκα. From a morphological perspective, ἀπέκταγκα apparently copied the root allomorph κταν- from ἀπέκτανον, while ἀπέκτακα shows the original pre-consonantal reflex of zero-grade *tkn̥- as seen in Homeric athematic aorist forms (sometimes taken up by Attic tragedy) such as ἔκτα, ἔκταμεν, κτάμενος, κτάσθαι, and ἔκτατο (on these forms and their relationship with the thematic aorist, see Willi 2018, 329‒30).

A recent study of analogical levelling in Ancient Greek verbal inflection by Sims-Williams (2016, 333–4) demonstrated that the perfect active is particularly susceptible to replacement and more likely to be influenced by the mediopassive than the other way around. This is due to the fact that the allomorph on which the perfect active stem is built often stands isolated in the paradigm, while the perfect mediopassive stem never uses a unique allomorph but shares it either with the perfect active or with other passive forms (aorist, future). The perfect active of (ἀπο)κτείνω fits the profile for this kind of analogical levelling, since the allomorph (ἀπ)έκτον- is not shared with any other form in the paradigm, while the perfect mediopassive (ἀπ)έκταμαι is backed up by the aorist passive (ἀπ)εκτά(ν)θην and the future passive (ἀπο)κτανθήσομαι. However, the perfect mediopassive of ἀποκτείνω has very few and late attestations (LXX 1Ma. 5.51 ἀπεκταμμένων, Plb. 7.7.4 ἀπεκτάνθαι), making it an unlikely source of analogical pressure on the active. Moreover, the passive forms of (ἀπο)κτείνω are rare, with (ἀπο)θνῄσκω(ἀπο)θνῄσκω ‘to die’ normally used in their place. Thus, ἀπέκτανον remains a more plausible intraparadigmatic model. The analogy, however, may have been based on external models: Chantraine (1927, 140) suggested that ἀπέκταγκα was created on the model of πέφαγκα, itself a relatively recent creation (attested since Deinarchus, who uses ἀποπέφαγκεν 9 times) next to mediopassive πέφαμαι (Hom.+). In this vein, we may also cite verbs whose stems terminate in a resonant that had a zero-grade kappatic perfect next to an e-grade present in *-i̯ō and a contract future: Choeroboscus (B.3), indeed, puts κτείνω ‒ κτενῶ ‒ ἔκταγκα precisely in the company of verbs like κείρωκείρω ‒ κερῶ ‒ κέκαρκα (elsewhere Choeroboscus [in Theodos. GG 4,2.105.20] assigns to κτείνω the perfect ἔκτονα to illustrate a different derivational rule; on which, see Benedetti 2019). It should be noted that ἀπέκτα(γ)κα is not the only innovative form replacing ἀπέκτονα in the koine: an o-grade kappatic perfect ἀπεκτόνηκα is attested in Aristotle (SE 182b.19) and some later prose authors (Parth. 24.2.6, Ios. AI 4.220, Plu. Tim. 16.10, Serap. Man. 32); the corresponding mediopassive is attested only in LXX 2Ma. 4.36 (inf. ἀπεκτονῆσθαι). One may compare Phryn. PS 63.4‒7Phryn. PS 63.4‒7 and Moer. δ 31Moer. δ 31, contrasting Attic διέφθορενδιαφθείρω ‘s/he has destroyed’ with the innovative διέφθαρκεν.

The erudite debate about ἀπέκτα(γ)κα in the imperial age is witnessed not only by lexicographers but also by Sextus Empiricus (B.2), who opposes the analogist perspective, demonstrating that some verbal forms that would be paradigmatically regular, such as ἔκταγκε, are not, in fact, attested (see F.2). Interestingly, Sextus’ argument implies that ἔκταγκε is a regular form that analogist grammarians would have approved; predictably, the Atticists rejected it precisely as the product of a regularising tendency (‘analogy’ in the modern linguistic sense), valuing instead the synchronically more marked ἀπέκτονα thanks to its attestations in older Attic literature. Contemporary high-register authors apparently shared this judgement: among Atticising writers of the 2nd century CE, the perfect of ἀποκτείνω is found mainly in Philostratus (VA 10x, Her. 13.4) and Lucian (VH 2.7, IConf. 16, DMort. 24.1, Alex. 44.7), invariably with the stem ἀπεκτον- (see also Schmid, Atticismus vol. 2, 28 on ἀπεκτονώς in Aelius Aristides).

E. Byzantine and Modern Greek commentary

In Post-classical Greek, the perfect disappeared as a living morphological category, merging with the aorist first functionally and then formally. Although perfect forms are amply attested in the Byzantine period, they are used interchangeably with the aorist and are largely restricted to higher-register texts; the only finite perfect form that has survived in the spoken language is εὕρηκα, which was reinterpreted as an aorist (> Modern Greek βρήκα), while several periphrastic constructions emerged to express the functions of the old perfect (on all the above developments see CGMEMG vol. 3, 1342–3; Horrocks 2010, 131–2; Hinterberger 2014; Horrocks 2020). The verb (ἀπο)κτείνω itself, which already in antiquity suffered the competition of φονεύωφονεύω, occurs in the medieval and early modern periods only in high-register contexts, and should be considered an ‘internal borrowing’ from Classical Greek (see DELG s.v. κτείνω). During this period, the stem ἀπεκτον- is used by classicising authors such as Michael Psellus (4x) or Nicetas Choniates (6x), while lower-register texts tend to favour ἀπεκταγκ-. For the notion ‘to kill’ (in both its literal and figurative senses), Modern Greek uses the verb σκοτώνω, derived from Ancient σκοτόωσκοτόω ‘to darken, blind’ and employed as an equivalent of ἀποκτείνω already in medieval texts (see Chantraine 1949, 147‒8, who observes that sometimes the manuscript tradition oscillates between the two, as in Nicetas Choniates Historia 185.13 van Dieten τοῦτον ἀπεκτονότες, v.l. τοῦτον σκοτώσαντες). The learned borrowing φονεύω remains in use as a more formal synonym for ‘to murder, kill (a person)’.

F. Commentary on individual texts and occurrences

(1)    Philemo (Laur.) 355 (A.2)

This entry is transmitted in both recensions of Philemon’s lexicon, as well as in Thomas Magister (A.3), albeit in a textually problematic form that the editors attempted to emend in various ways. While both codd. of Philemon transmit ἀπέκτονας as the approved form, for the rejected one cod. L has ἀπέκτανας, corrected from ἀπέκτεινας, while cod. V has ἀπέκτονες. Neither version is likely to preserve the original text, since ἀπέκτανας and ἀπέκτονες are not otherwise attested. Thomas Magister, meanwhile, has ἀπέκτονα as the first and most approved form, ἀπέκτεινα as the second best, and ἀπέκτανον as the last and least favourable. Buttmann (1825, 174) argued that Thomas’ text makes no sense as transmitted, since it would be awkward to contrast a perfect with two aorist forms; accordingly, based on the comparison with Moeris’ entry, he proposed to correct ἀπέκτεινα and ἀπέκτανον to ἀπέκτακα and ἀπέκταγκα, respectively. In his edition of Thomas Magister, Ritschl (1832, 8) reported Buttmann’s conjecture in the apparatus. The editores principes of Philemon, however, corrected the respective texts of L and V based on the paradosis of Thomas: so, Osann (1821, 292) changed L’s ἀπέκτανας to ἀπέκτεινας, while Reitzenstein (1897, 392) changed V’s ἀπέκτονες to ἀπέκτανες, thus restoring in both cases one of the attested aorists of ἀποκτείνω. Cohn (1898, 355), who edited L and V synoptically, argued that ἀπέκτακας should be restored in both versions. Cohn’s emendations would make Philemon’s prescription one about competing forms of the perfect, in line with Moeris (A.1) and with the interests of Philemon himself, who often contrasts conservative and innovative verbal morphology (see Brown 2008, 217–9) but never comments on the use of different tenses of the same verb.

Meanwhile, a prescription recommending the use of the perfect over the aorist is not implausible and may find some support in Pollux’ Onomasticon (B.1), which lists both ἀπέκτονεν and ἀπέκτεινεν among verbal forms (all in the aorist, except for ἀπέκτονεν itself) that may be used to express that someone killed someone, suggesting that aorist and perfect might be perceived as equivalent in this context, at least by less strict Atticists. Indeed, the Greek perfect may be used to characterise the subject as having committed something (e.g., a crime) and thus presently bearing responsibility for it. For semantic reasons, ἀποκτείνω was a verb whose perfect was frequently used in this manner to emphasise the subject’s responsibility for the killing and could be contrasted with the aorist indicative, which merely reports the fact: ‘ἀπέκτεινε ‘he has killed (someone)’, ἀπέκτονε ‘he is (someone’s) murderer’’ (Rijksbaron 2002, 36; see also the influential discussion by Chantraine 1927, 168–71, with examples from Xenophon and Plato). To 2nd-century-CE speakers, living in an age when the functional distinction between aorist and perfect was being lost in spoken Greek (see E.), such nuances of classical usage may have been difficult to acquire; hence, some Atticists may have recommended the perfect as a more appropriate stylistic choice.

(2)    S.E. M. 1.238 (B.2)

This passage is taken from the first book of Sextus Empiricus’ Against the Mathematicians, comprising a broad attack on the technical study of grammar, and in particular from the section (176‒247) discussing two kinds of ἙλληνισμόςἙλληνισμός or ‘linguistic correctness’, one based on analogy, which Sextus rejects, and one based on συνήθειασυνήθεια or ‘common usage’, which he embraces (on Sextus’ attack on technical grammar see Sluiter 2000; on the evolution of the concept of Ἑλληνισμός, see Pontani 2011; Pagani 2014). Here, the author responds to the grammarians’ argument that common usage is inconsistent by pointing out, with examples taken from both nominal and verbal morphology, that the analogical rules extracted from that usage are likewise inconsistent. Sextus shows that, on the one hand, nouns and verbs that are similar in one case or tense, respectively, are dissimilar in others and that, on the other hand, some verbal forms that are predicted by the rules of analogy are not attested, with others used in their place. Several of the examples adduced in this passage are textually problematic. Blank (1998, 253‒5) accepts the corrections made by Theiler (1956, 286); in particular, the codd. transmit the present mediopassive κτείνεται, which must be incorrect, as it makes no sense in opposition to the perfect active ἔκταγκε. Theiler’s correction of the former to ἔκτονε is likely right, particularly considering that the other example given (attested ἀλήλιπται for expected ἤλειπται) likewise concerns the formation of the perfect stem; ἔκτονε might have become κτείνεται under the influence of the immediately following λέγεται. Interestingly enough, the other example also echoes entries in Atticist lexica (cf. Phryn. Ecl. 23Phryn. Ecl. 23, Phryn. PS 32.12‒4Phryn. PS 32.12‒4; see entry ἤλειπται, κατώρυκται, ὤμοκε); on other coincidences between forms discussed by Sextus and the Atticists, see Blank (1998, 182 and passim). Dickey (2019) , however, argues that Sextus’ main polemical target is linguistic expertise based on analogy rather than Atticism, implying that Sextus (whose date may be only approximately placed sometime between 100 CE and the early 3rd century, see Blank 1998, XV) was writing at a time before Atticism became fully established (on the contrast between analogist and Atticist views regarding correct Greek, see Probert 2011).

(3)    Σ^b α 1872 (= Su. α 3372, Phot. α 2534, ex Σʹ) (B.4)

This entry is transmitted in the expanded Synagoge, Photius, and the Suda; Alpers identified it as a fragment of Orus (see Alpers 1981, 56‒69 on the attribution to Orus of Atticist glosses in the Synagoge). The first part of the entry discusses the correct form of the nasal-suffix present built from the root *tken-, arguing for athematic ἀποκτίννυμι (supported by a quotation from Cratinus’ Boukoloi, on which see Bianchi 2016, 128‒33) against thematic ἀποκτιννύω (this present is treated elsewhere in Atticist lexica: Phryn. PS 51.12Phryn. PS 51.12 argues for the spelling with a single ν, while Moer. α 56Moer. α 56 prefers ἀποκτειννύναι to ἀποκτείνειν). The second part of the entry turns to the perfect, expressing a preference for 3rd pers. plur. ἀπεκτόνασιν over ἀπεκτάγκασιν: the latter form is exemplified with a quotation from Menander’s Misoumenos (C.4). However, the textual status of this quotation is highly problematic: both the text transmitted by the Synagoge (Μισουμένῳ πάτερ μὲν Θράσωνι…) and that transmitted by the Suda (μισοῦσι μὲν, ὦ πάτερ, Θράσωνα…) must have undergone some corruption, but consensus has yet to emerge as regards how the text should be reconstituted (see Arnott 1968; Furley 2021, 194; Kassel, Schröder PCG vol. 6,1, 317). The only words that all editors agree belong to Menander’s fragment are πάτερ and ἀπεκτάγκασι δ’ οὔ, with the important exception of Alpers (1981, 208), who follows Cobet (1858, 55‒6) in restoring ἀπεκτόνασι. While Cobet’s argument that the form ἀπέκταγκα belongs to Byzantine Greek is weak in view of its attestations in koine prose (C.3, C.5, C.6), Alpers may be correct in observing that the quotation should exemplify the approved rather than the rejected form; moreover, Orus (unlike Phrynichus) usually cites Menander as a model of good Attic. However, the editors of Menander have not generally accepted this correction. If Menander did use ἀπεκτάγκασι, this would be one of several innovative perfects for which he provides the first – or one of the first – attestations together with some other 4th-century-BCE author – in this case, Aristotle (C.3) (see Goldberg 1996, 27‒57; Cartlidge 2014, 229‒35).