διδῴη and other optative forms of δίδωμι
(Phryn. Ecl. 324, Moer. δ 5)

D. General commentary

Phrynichus (A.1) and Moeris (A.2) discuss different forms of the present and aorist optative of the verb δίδωμι ‘to give’. Both lexicographers reject the innovative formsInnovative forms with a ‘long diphthong’ -ῳ- in place of -οι-, although each entry also touches on other morphosyntactical issues – namely, the confusion between optative and subjunctive, and the spread of -η- in plural forms of athematic optatives. The Atticists’ interest in the optative of δίδωμι is further confirmed by an entry in Pseudo-Zonaras’ lexicon, plausibly identified as a fragment of Orus (B.1), which agrees with the prescriptions of Phrynichus and Moeris, and is line with other traces of Orus’ lexicographical activity (see F.2).

As is often the case, the forms rejected by the Atticists were the product of analogical mechanisms (there is a growing body of recent literature on how the Atticists dealt with morphosyntactic analogy against the backdrop of the koine: see e.g. La Roi 2022, Monaco 2025). The inherited optative of athematic verbs with an ablauting stem in -ω/-ο (< *-eh₃-/*-h₃-), such as δίδωμι, featured the diphthong -οι-, arising from the contraction of the zero-grade stem with the zero-grade allomorph of the optative suffix *-i̯eh₁-/*-ih₁-. The present optative in διδοι- and the aorist optative in δοι- were the norm in classical Attic, with abundant attestations in both drama (e.g. C.4) and prose (e.g. C.5). By contrast, the forms in (δι)δῳ- lack secure attestations in classical literature (see below) and are absent from Attic inscriptions (Threatte 1996, 593); epigraphic occurrences from other locales are scarce and late (e.g. I.Napoli 1, 43.11 διδῷ [late 1st c. BCE – early 1st century CE]).

In post-classical times, analogical forms with a ‘long diphthong’ begin to occur, possibly under the influence of contract verbs in -άω (e.g. νικῷμεν < *-α-οι-μεν ‘may we win’ from νικάω ‘to win’, cf. F.2), which also influenced the verbs in -έω (a development censured by Phryn. Ecl. 321Phryn. Ecl. 321, Moer. π 74Moer. π 74), and/or by athematic verbs with a stable long vowel -ω- in the present or aorist indicative, which introduced the same vocalism into the optative (e.g. βιῴην ‘may I live’ after ἐβίων ‘I lived’; see Schwyzer 1939, 795; Chantraine 1961, 264). The innovative forms are first attested for the aorist stem in Hippocrates (C.7) and Aristotle (EE 1227a.19 δῴη, 1423b.10 ἀποδῴη), and later for the present stem in Philo (C.8); they subsequently gained ground in the koine (Crönert 1903, 214–5). Despite the lack of secure occurrences in the Ptolemaic papyri – where the optative of δίδωμι displays considerable variation (Mayser, Gramm. vol. 1,2, 88; Lobeck 1820, 346) – forms in -ῳ- are frequent in the Septuagint and in the New Testament (Blass, Debrunner 1976, 70), as well as in authors who employed a higher-register koine, such as Josephus and Plutarch. In keeping with this distribution, modern editors usually emend spellings with (δι)δῳ- occasionally found in the MSS of classical authors (e.g. Eur. Andr. 225 ἐνδῴη, v.l. for ἐνδοίη), attributing them to the influence of the koine (see Rutherford 1881, 456; Lautensach 1916, 99–100).

The lack of canonical attestations is not the only reason for Phrynichus’ proscription of the innovative forms (A.1). Indeed, he also warns against the use of δῴη on the grounds that it might be mistaken for the aorist subjunctive δώη (20x in Homer; see further F.1). Phrynichus reproaches the contemporary sophist Alexander of SeleuciaAlexander of Seleucia for his use of the innovative forms δῴη and διδῴη. Alexander is also criticised in Ecl. 124Phryn. Ecl. 124 for his use of the acc. sing. υἱέα ‘son’; for the possible reasons behind Phrynichus’ hostility towards him, see entry υἱεύς, υἱέως, υἱέα.

It is worth noting that later sources did not necessarily agree with Phrynichus’ rule. Choeroboscus (B.2) explains δῴην as a lengthened variant of δοίην, and the Epimerismi Homerici (B.3, B.4) – probably authored by Choeroboscus himself – regard δῴη(σι)ν as an admissible optative form, to be distinguished from the subjunctive δώῃ(σι)ν. In fact, B.4 alludes to a scholarly debate concerning the correct spelling of the latter form, which occurs twice in Homer (Il. 1.324, 12.275; cf. also h.Merc. 174), in addition to the slightly more common contracted variant δῷσι (Hom. 4x). From the perspective of modern scholarship, these forms are best understood as aorist subjunctives expanded with the athematic ending -σι (Chantraine 1958, 462).

In the Palaeologan age, while George Lacapenus (B.6) promotes the Attic forms in -οι- in line with the imperial-period Atticists, Thomas Magister (B.5) directly quotes A.1 and criticises Phrynichus’ stance. Thomas defends the forms with a long diphthong on the basis of their parallel with verbs that display ω-vocalism in the aorist indicative (ἐβίων, ἑάλων), although he explains the change as a Doric feature. Nevertheless, he acknowledges that (δι)δῴην coexisted to some extent with (δι)δοίην, adducing an occurrence in Herodotus – where, however, the textual tradition is divided (see the apparatus to C.3) – as a precedent for the usage of later authors, including his favourite Libanius (however, see F.3).

Moeris’ entry (A.2) expounds a doctrine similar to that of Phrynichus and Orus. Quite surprisingly, however, Moeris adduces 1st- and 2nd-person plural forms with an -η- analogically extended from the singular, not only in the rejected koine forms δῴημεν δῴητε, but also in the Attic ones δοίημεν δοίητε. This is at odds with his own prescriptions against the analogical -η- (cf. Moer. β 5Moer. β 5, γ 13Moer. γ 13, and see AGP vol. 2, Verbal morphology, forthcoming), and is only partly supported by classical precedent. Although δοίημεν and δοίητε (including prefixed forms) do occur in 4th-century BCE Attic prose (X. 4x; D. 21.170; Is. 7.41), and δοῖτε is admittedly unattested in classical Attic, δοῖμεν was favoured by Plato (6x). One may therefore either assume inconsistency on Moeris’ part or posit an alteration of the expected δοῖμεν δοῖτε to δοίημεν δοίητε in the course of textual transmission.

The Atticising authors employed the classical forms in -οι-, with very few exceptions. Lucian, for instance, employs (-)δοίη(ς) 8 times in his genuine works, whereas δῴη occurs only once, in the probably spurious Cynic (17), and the ambiguously spelled μεταδωης in Par. 1 appears as a varia lectio for μεταδοίης (see Schmid, Atticismus vol. 1, 231; vol. 4, 598; Deferrari 1916, 24–5). By contrast, authors less influenced by Atticism appear to have had fewer reservations about the innovative forms: the Church Fathers, for instance, regularly use δῴη (see E.). Interestingly, however, the innovative forms are not attested in Roman and Byzantine papyri. According to Gignac (1981, 390), their failure to displace the classical forms in -οίην may be due to the popularity of the thematic forms of the present and aorist subjunctive of δίδωμι: ‘since those thematic forms resembled the corresponding forms of contract -όω verbs, the optative -οίην, likewise identical with the present optative of contract -όω verbs, would have appeared regular’. Therefore, the prevalence of forms in -οι- among Atticising authors need not be interpreted as a markedly puristic choice, since it would have coincided with contemporary spoken usage (though not with the literary koine). Nonetheless, a puristic motivation remains likely in view of the debate among Atticist lexicographers.

E. Byzantine and Modern Greek commentary

Although the innovative forms in (δι)δῳ- appear to have retreated from the spoken language after the Hellenistic period (see D.), they continued to be used throughout the Byzantine period on the strength of their biblical occurrences, especially – though not exclusively – in theological works, but also in lower-register writings. As already in the New Testament, these forms often occur in fossilised blessing or cursing formulas of the kind (μὴ) δῴη ὁ Θεός ‘may God (not) grant’ (see Bianconi, Magni 2023), a distribution traditionally connected to an alleged general decline of the optative in Post-classical Greek, particularly in dependent clauses (see e.g. Horrocks 2010, 102–3; 117; 130). It is important to note, however, that this perspective has been challenged by more recent corpus-based studies (La Roi 2025), which suggest that the optative in fact remained vital during this period and even developed innovative uses. At any rate, as far as the formal variation in the optative of δίδωμι is concerned, the more cultivated Byzantine authors continued to prefer the classical forms in (δι)δοι-. A case in point is Michael Psellus, who employs the latter 62x, across a wide range of paradigmatic forms and syntactic constructions, and uses the innovative form only once, in a letter of dubious authorship (Epistulae 54.245.33: εἰ δὲ δῴη θεὸς καθαρᾶς με ὑγείας ἐπαπολαῦσαι, ‘If God would grant me to profit from good health’).

In Medieval Greek, the verb δίδωμι itself was thematised to δίδω, and then secondarily remodelled as δίνω after the presents in -νω, with regional variants διδῶ, δούνω, and δώνω, while the perfective stem was either δωσ- or δωκ- (CGMEMG vol. 3, 1376–8; Kriaras, LME s.v. δίδω). Standard Modern Greek has the pres. δίνω, aor. έδωσα.

F. Commentary on individual texts and occurrences

(1)    Phryn. Ecl. 324 (A.1)

Phrynichus supports his rejection of the Hellenistic aorist optative δῴη with an argument based on Homeric usage: δώῃ (with an iota after the second vowel, not the first) does indeed occur, but only as a Homeric subjunctive (C.1); Homer himself, when an optative is required (C.2), employs forms in -οι-, which are also the correct Attic ones. As Radermacher (1915, 27) correctly pointed out – against Moulton (1906, 193–4) – the Eclogue’s entry does not therefore lend support to an alleged koine subjunctive δώῃ. At any rate, since Phrynichus’ reasoning presupposes that he – like modern scholars – interpreted the Homeric form in C.1 as a subjunctive, δώῃ should be printed in this entry (as Lobeck 1820, 345 and Rutherford 1881, 429 did). On the other hand, Choeroboscus (B.2) quotes the same Homeric line as an example of the optative δῴη, suggesting that a certain degree of confusion surrounded the interpretation of this form. In addition to the homophony of δῴη and δώῃ in post-classical pronunciation, a source of confusion may lie in the fact that the construction εἰ … κε(ν) with the optative – foreign to Attic syntax – is indeed common in Homer, alongside the more usual construction with the subjunctive (Chantraine 1963, 277–80).

(2)    Orus fr. A 31 (= [Zonar.] 566.12) (B.1)

There is evidence that Orus took an interest in the spelling of several other optative forms in his Orthography, as can be inferred from an excerpt preserved in cod. S. Salvatore 118 (edited by Rabe 1892; 1895). This passage discusses the thematic optative of the contract verbs νικάω ‘to win’ (Lex.Mess. fol. 281r.16–20, cf. Orus fr. B 31), ὁράω ‘to see’ (Lex.Mess. fol. 282r.17–8), and χράομαι ‘to use’ (Lex.Mess. fol. 282v.21–2). See, most recently, Fiori (2025, 176–81), who, following Reitzenstein (1897, 295), argues that Et.Gen. AB s.v. τρυγῷ καὶ τρυπῷ (= [Zonar.] 1752.8–12 ~ EM 771.4–9) also ultimately goes back to Orus’ Orthography.

(3)    Thom.Mag. 91.11–92.15 (B.5)

The entry’s wording is ambiguous with regard to Libanius’ usage. His mention immediately after Alexander of Seleucia suggests, on the most straightforward reading, that even Libanius – an author normally endorsed by Thomas Magister as a model – did not refrain from using the post-classical form (δι)δῴην (see e.g. Lobeck 1820, 345–7). Foerster (1922, 655–6), by contrast, understood Thomas’ words as implying that Libanius used the classical forms διδοίη and δοίη, which he printed as Libanius’ fr. 61. In fact, the forms in -οι- are the norm in Libanius’ works, with around 60 occurrences, whereas those in -ῳ- are entirely unattested in his corpus.