οἶσθα, οἶσθας, οἶδας
(Moer. ο 24)

D. General commentary

This entry in Moeris’ lexicon (A.1) opposes two forms of the second person singular of the verb οἶδα ‘I know’, promoting the inherited οἶσθα against the innovative οἶδας. However, a third is implied by the specification that the approved form is οἶσθα ‘without σ’, implicitly rejecting the form οἶσθας. Both οἶδας and οἶσθας are analogical innovationsInnovative forms that spread over time at the expense of the inherited οἶσθα, but their reception in Atticist lexicography was somewhat different.

As is well known, the perfect (with present meaning) οἶδα is a remarkable archaism that is unique in Greek for its lack of reduplication; in classical Attic, it preserves other inherited traits as well, including the ablaut (ϝ)οιδ-/(ϝ)ιδ- and the ending -θα in the second person singular οἶσθα, which is the regular reflex of Proto-Indo-European *u̯oi̯d-th₂e (cf. Vedic véttha; see Willi 2018, 8–9; 474; on the phonetics see Norbruis 2023). Indeed, οἶσθα is the norm in Homer (18x, e.g. C.2) and in Attic drama (e.g. C.3; C.4 is a possible exception) and prose (it is, for instance, the only form attested in Plato, with 188 occurrences). The synchronically irregular paradigm of οἶδα, however, was subjected to considerable analogical pressure, leading to the creation of an inflection οἶδα, οἶδας on the model of perfects such as λέλυκα, λέλυκας, πέφηνα, πέφηνας, etc., which also extended to the plural (1st pers. plur. οἴδαμεν, with the same ablaut grade as the singular, in place of inherited ἴσμεν, etc.). The analogical forms are first attested in the Odyssey (C.1) and in Ionic poetry (h.Merc. 456 and 467, Thgn. 1.491 and 1.957, Hippon. fr. *177 West) and prose (Herodotus, e.g. 3.72, 4.157; the Hippocratic corpus). In particular, the single Homeric occurrence in C.1, which may have been motivated by assonance-based wordplay with ἀοιδός, ἀείδω, ἀοιδή, was discussed in ancient scholarship (see B.10; F.3; see F.2 for a different explanation). It is likely, therefore, that the analogical paradigm first arose in Ionic, whence it penetrated later Attic and the koine (see Solmsen 1906, 206–7). The first undisputed Attic attestations of οἶδας occur in Xenophon (C.6) and the corpus Aristotelicum (6x in Analytica priora and Sophistici elenchi), and the other forms of the analogical paradigm are also attested around the same time. However, a possible attestation in Euripides (C.4; see F.4) suggests that οἶδας likely coexisted with οἶσθα as early as the 5th century BCE (Stevens 1976, 60). οἶδας does not occur elsewhere in tragedy but is attested twice in New Comedy (Philem. fr. 45.3, Phoenicid. fr. 3.2; cf. Arnott 2002, 203–4) and subsequently becomes standard in the Hellenistic period. While the innovative paradigm is scarcely attested in Attic inscriptions (only a single example of οἴδαμεν from the late 1st century CE is recorded by Threatte 1996, 570), it spread widely throughout the koine, predominantly – though not exclusively – in the lower registersRegister: it is the norm in Ptolemaic papyriPapyri (see Mayser, Gramm. vol. 1.2, 81) and in the New Testament (see Blass, Debrunner 1976, 72) as well as in Roman and early Byzantine papyri (where the further innovation οἶδες, with the ending of the aorist and imperfect, is also frequent: see Gignac 1981, 353–4; 409–11; Zinzi 2013, 159–61), but it is also common in literary texts that aim at a more classicising language. In Plutarch, for example, οἶσθα (27x) occurs almost four times as frequently as οἶδας (7x), but the distribution appears random – or, at least, not guided by the avoidance of hiatus, since both forms are used before consonants and vowels alike: rather, it signifies the mingling of the forms that were current in the koine and those that belonged to classical Attic (see Sanz Morales 2019, 589–90). Indeed, the analogical forms were surely so common by the imperial period that even Atticising writers use them to some extent, although οἶσθα is the majority form in Dionysius of Halicarnassus, Dio of Prusa, and Lucian, and the only one in Aelius Aristides (see Schmid, Atticismus vol. 1, 85 and 232; vol. 3, 13 and 16; vol. 4, 38 and 599; Sanz Morales 2019, 589 n. 10).

οἶσθας is the product of a different analogical innovation that took place in Attic and affected not the entire paradigm but the second person singular alone, re-characterising it with the usual ending -ς (see K–B vol. 2, 44; Schwyzer 1939, 662). This form occurs quite frequently in Middle and New Comedy (Alex. fr. 15.11 = C.7; Men. Epit. 480–1; Mis. 651; fr. 246.5; Pc. 152; Philem. fr. 45.4–5; Posidipp. fr. 29.2; Strato fr. 1.26; com. adesp. fr. 1017.65; see Arnott 1996, 94; Arnott 2002, 203–4), but according to Choeroboscus’ testimony (B.4), it was already used by Cratinus (C.5). The word’s distribution in comedy reveals that it was a useful metrical variant, as recognised already by ancient erudition (B.6, B.7, B.8). For a time, the two different innovations likely competed with one another and the inherited form, as suggested by Philemon (C.8), who has the same speaker use all three forms οἶσθα, οἶσθας, and οἶδας. In later Greek, οἶσθας is rarer than οἶδας. Notably, the distribution of οἶσθας continues to be skewed towards metrical texts, suggesting that it was still considered a metrical variant. οἶσθας occurs occasionally in literary texts from the Hellenistic and imperial periods: Philodemus Πρὸς τοὺς ἑταίρους 7 Angeli (P.Herc. 1005.col. xviii.14 = TM 62437); the Ninus romance, col. A.2.22 and col. A.3.25 Stephens–Winkler (notice that in the first passage οἶσθας occurs before a consonant); Arrian Epict. 1.12.26, and also in a metrical oracle quoted by Plutarch (De Pythiae oraculis 408a = 41 Parke–Wormell, where the reading οἶσθας ἄρειον is at variance with οἶδας ἄρειον transmitted by Hdt. 4.157.2 and with οἶδας ἄμεινον transmitted by adesp. AP 14.84.1). The fact that οἶσθας was, on the whole, less successful than οἶδας may also be explained by the fact that the former, unlike the latter, was not part of a wholly regular paradigm.

The lexicographers who rejected οἶδας – and, with it, the entire analogical paradigm in οἰδ-: Moeris (B.1), indeed, also condemns the 3rd pers. plur. οἴδασιν, but see F.1 – were more tolerant of οἶσθας. According to Eustathius (B.8), Aelius Dionysius admitted of both οἶσθα and οἶσθας as well as of ἦσθα and ἦσθας. The tradition reflected by Photius (B.6) and the Suda (B.7) and that likely also goes back to Aelius Dionysius specifies that οἶσθας could only be admissible for metricalMetre reasons (to lengthen by position the second syllable before a consonant-initial word) or to avoid a clashing of sounds (i.e., vowel hiatus; see F.1). The very existence of the debate as to whether οἶσθας should be written in the Homeric text also demonstrates that it was perceived, in principle, as an admissible variant even by scholars who were ready to condemn οἶδας (see B.9; F.2). It may be the case that οἶσθας was perceived as a compromise between the inherited οἶσθα and analogical οἶδας. This would have made it possible for Attic authors to use it with relatively greater freedom, as demonstrated by its attestations in comedy (possibly going back to Cratinus), and for imperial-period scholars to regard it as more acceptable – at least, under certain conditions. It is instructive to compare this relative tolerance with the Atticists’ rejection of ἦσθας, which arose from the same kind of analogy but could not boast any classical attestation (see entry ἦς, ἦσθα, ἦσθας). Later lexicographical sources simply state the equivalence between οἶσθα and οἶδας (B.2, B.5) or present οἶσθας and οἶδας as equally Attic (B.3); meanwhile, Thomas Magister, who is known to directly consult the Atticist lexicographers of the 2nd and 3rd centuries CE (see entry Thomas Magister, ’Ονομάτων Ἀττικῶν ἐκλογή), follows them in recommending οἶσθα (B.11).

E. Byzantine and Modern Greek commentary

In the later history of Greek, οἶδα – which, even with regularised endings and ablaut, was still an anomaly in being an unreduplicated perfect with present meaning – underwent lexical replacement by other verbs. In particular, the present ἐξεύρω (from the aorist ἐξεῦρον of ἐξευρίσκω ‘to find out’), which later evolved to ξέρω, began to replace οἶδα in late Medieval Greek, as the latter became homophonous with εἶδα ‘I saw’ owing to the merger of /y/ and /i/ (see CGMEMG vol. 3, 1381). Nonetheless, οἶδα is still found throughout the Byzantine period as one of few relics of the old perfect. The forms employed for the 2nd person singular vary according to the register, with the classicising οἶσθα prevalent in higher-register texts, down to Neophytus Ducas in the 18th–19th century. The Byzantine chronicles, meanwhile, use only οἶδας (see Psaltis 1913, 241), whereas οἶσθας, despite the Atticists’ limited approval, apparently died out after the imperial period, with very few exceptions (Anonymus Professor Epist. 30.96, Theodorus Prodromus Epistula ad Constantinum Vardachlam 50–1). In Modern Greek, the common forms for ‘to know’ are γνωρίζω and ξέρω, while the archaism οἶδα appears only in set phrases such as τις οίδε ‘who knows’, Κύριος οίδε ‘God knows’, and the proverbial εν οίδα, ότι ουδέν οίδα ‘one thing I know, that I know nothing’, derived from Pl. Ap. 21d (see LKN s.v.).

F. Commentary on individual texts and occurrences

(1)    Phot. ο 150 (B.6)

This entry in Photius’ lexicon is also found in the Suda (B.7), where it is expanded with quotations from Sophocles and Synesius. Erbse (1950, 131) traced back the entries to Aelius Dionysius, comparing the similar prescription given by Eustathius (B.8), who quotes Aelius by name. The entry prescribes οἶσθα against οἶδας, permitting οἶσθας only on metrical–prosodic grounds. συγκρούω is a technical term for a ‘collision’ or ‘clashing’ of two or more successive sounds, whether vowels (i.e., hiatus) or consonants. Both phenomena were extensively discussed in ancient euphonic theory: such sound successions were generally considered disagreeable, although they could be exploited to achieve certain stylistic effects. Indications suggest that some Atticists, including Aelius Dionysius, drew on euphonic principles in their judgements and occasionally allowed such concerns to override other criteria (see entry ἄρκτος, ἀπαρκτίας). Both Photius and the Suda transmit readings that presuppose the noun σύμφωνα ‘consonants’. However, it is unclear how an additional consonant at the end of the word might help avoid a clashing of consonants as opposed to (potentially) creating one; therefore, Erbse followed Dindorf’s conjecture and restored φονήεντα ‘vowels’, which is not only makes sensible but is also supported by Eustathius’ statement that οἶσθας would be admissible before a vowel. In fact, οἶσθας is used by playwrights of Middle and New Comedy precisely in the two functions admitted by Aelius Dionysius: before a vowel (5x), or, less often, before a consonant to create a long syllable (3x), whereas it is unattested at verse end. This correspondence reveals that the lexicographer was an attentive observer of the forms’ distribution in classical texts. We may add that Aelius Dionysius could also be the source of Moeris’ entry (A.1) in that Moeris’ frequent dependence on Aelius is well established (see entry Thomas Magister, ’Ονομάτων Ἀττικῶν ἐκλογή and Erbse 1950, 58; Hansen 1998, 42–7), and the doctrine that the texts expound is essentially the same, the sole difference being that Moeris’ more succinct formulation omits the possible reasons for admitting οἶσθας. Thus, it is not necessarily the case that Aelius Dionysius is less vehemently opposed than Moeris to the innovative form(s) owing to ‘different standards’ of correctness (as assumed by La Roi 2022, 210–1), but it might also be the case that the highly compressed and schematic format of Moeris’ lexicon as it is transmitted omits several of the interpretative nuances present in his sources.

(2)    Schol. (Ariston.) Hom. Il. 1.85e (B.9)

In the A scholium to Il. 1.85, Aristonicus warned against the variant οἶσθας, observing that Homer invariably uses the form without -ς. Van der Valk (1985, 376) assumed that this remark was aimed at Zenodotus, who, according to Eustathius incorrectly wrote οἶσθας in Homer, but Le Feuvre (2022, 14; 17) points out that nothing in the scholium obliges us to associate the condemned reading with Zenodotus and suggests that ‘the οἶσθας Eustathius attributes to Zenodotus is not a Zenodotean reading but ended up under his name’, as deviant readings tended to be ascribed to him by default. Meanwhile, in a scholium to the Odyssey (B.10), Aristonicus reports that Zenodotus read the pluperfect ᾔδεις instead of οἶδας in Od. 1.337 (C.1), while Aristarchus defended the vulgate text (not Zenodotus’ variant, although the scholium’s expression is ambiguous: see Pontani 2007, 176). The (alleged) contradiction that van der Valk (1985, 378 n. 7) perceived between Zenodotus’ rejection of οἶδας and his acceptance of οἶσθας disappears if the latter position is not ascribed to him, though the two opinions would not necessarily be contradictory, given that οἶσθας appears to have been considered more acceptable than οἶδας (see D.). Schwartz’ hypothesis that Zenodotus read θελκτήρι’ ἀείδεις is considered favourably by West (1988, 118); see further F.3. On this passage see now Cassio (2024, 71–6), who argues that οἶσθας was adopted here for the sake of episynaloephe – that is, to ensure a smooth transition to the initial vowel of the following line; οἶδας in C.1 would have had a similar motivation, given that the avoidance of hiatus became increasingly important in the Homeric poems’ later compositional stages.

(3)    Hom. Od. 1.336–40 (C.1)

The only Homeric attestation of οἶδας comes from this passage of the first book of the Odyssey, in which Penelope addresses Phemius, who has been singing of the Achaean heroes’ returns from Troy, and asks him to change the subject matter of his song. These lines are marked by the polyptoton and figura etymologica of ἀοιδός (2x) / ἀείδω / ἀοιδή; notably, all forms except for ἄειδε at v. 339 occur at the end of the line. West (1988, 118) suggested that the form οἶδας was selected so that v. 337 would echo the endings of vv. 336, 338, and 340. Building on West’s suggestion, Hardie (2000)    argued that this passage presupposes a folk-etymology of ἀοιδός as ‘one who knows’, from intensive ἀ- + οἶδα (this etymology is indeed attested in ancient erudite sources, sometimes with an analysis of the ἀ- as privative; see also Kaesser 2004 for a possible reprise in Callimachus). Hardie (2000, 174–5) further suggests, more speculatively, that Zenodotus’ reason for rejecting the reading οἶδας might have been his awareness that the etymological analysis linking ἀοιδός with οἶδα was later than Homer’s time (on the possibility that οἶδας was selected to avoid hiatus with the line that followed, see F.2).

(4)    Eur. Alc. 780 (C.4)

This Euripidean passage may preserve the earliest attestation of οἶδας in Attic literature and the sole example from tragedy. The reading οἶδας is transmitted by almost all the codd. (except L), and is retained by Dale (1954, 34), while Diggle (1984, 68) cautiously obelised it. Parker (2007, 33) accepts Blaydes’ τὰ θνητὰ πράγματ’ ἥντιν’ οἶσθ’ ἔχει φύσιν;, which restores the usual Attic form but at the cost of producing an usual word order (see Parker 2007, 207–8). Dindorf’s emendation οἶσθας is hardly more convincing, since it would introduce a form likewise unattested in tragedy. This verse is also quoted by [Plu.] Consolatio ad Apollonium 107b, where the MSS transmit several variant readings (οἶδας, οἶσθας, οἶσθά γ’, οἶσθ’); Parker (2007, 207) observes that ‘the variety of forms offered by the MSS of pseudo-Plutarch suggests that the textual problem manifested itself early’. The possible reasons for Euripides’ use of οἶδας have also been discussed. Rutherford (1881, 227), who accepted the MS reading, deemed it ‘probable that οἶδας should be acknowledged as old Attic’; Stevens (1976, 60) considered it, more plausibly, to be a colloquial form, reflecting the influence that Ionic exerted on the spoken Attic of the time. Note, however, that Collard has recently expressed a more noncommittal view in this regard (2018, 127).