No CrossRef data available.
Article contents
EARLY TRACES OF THE GREEK QUESTION MARK
Published online by Cambridge University Press: 28 July 2023
Abstract
According to the standard view on the issue, the habit of marking questions with a particular typographical sign in Greek and Latin script does not arise prior to the eighth or ninth century. This period is generally credited with the ‘invention’ of the question mark (excepting Syriac evidence, which points to the fifth and sixth centuries). The purpose of the present article is to correct this view. It argues that the first indication for the use of a typographical sign that marks questions can actually be detected no later than during the reign of Emperor Hadrian (a.d. 117–38), that is, more than half a millennium before the traditional date of the invention. The chief witness is Nicanor of Alexandria, who does not seem to have used question marks himself, but criticizes the misappropriation of another punctuation mark to that purpose. He thus indirectly testifies to the existence of question marks. Comparable traces can also be found later in the exegetical works of some Christian commentators.
- Type
- Research Article
- Information
- Copyright
- Copyright © The Author(s), 2023. Published by Cambridge University Press on behalf of The Classical Association
References
1 For concise summaries on Nicanor and his works (all fragmentary) in general, see Wendel, C., ‘Nikanor (27)’, RE 17 (1936), 274–7Google Scholar; Matthaios, S., ‘Nikanor aus Alexandreia (12)’, DNP 8 (2000), 903–4Google Scholar. The name and function of the eight punctuation marks, specifically, are explained and illustrated with examples in Nünlist, R., ‘Nicanor's system of punctuation’, GRBS 60 (2020), 124–38Google Scholar. That account is based on the two chief witnesses known at the time, a commentary on the grammar of Dionysius Thrax (pp. 26.4–28.8 Hilgard), which includes Homeric examples, and Nicanor's actual practice that can be derived from the fragments of his commentary on the Iliad. (The newly discovered summary of Nicanor's theory does not affect the present argument; for the new summary, see Sandri, M.G., ‘Nuovi frammenti sulle interpunzioni dall'opera di Nicanore di Alessandria', Glotta 98 [2022], 279–301CrossRefGoogle Scholar; Nünlist, R., ‘New evidence on Nicanor's theory of punctuation’, Philologus 167 [2023], 8–21.CrossRefGoogle Scholar) Nicanor's fragments have been edited, with extensive introduction, by L. Friedländer, Nicanoris Περὶ Ἰλιακῆς στιγμῆς reliquiae emendatiores (Berlin, 18572, repr. Amsterdam, 1967), but the text of the fragments is now best consulted in H. Erbse's edition of the scholia to the Iliad: Scholia Graeca in Homeri Iliadem (Scholia vetera), 7 vols. (Berlin, 1969–88). The attribution of Odyssean scholia to Nicanor is encumbered by uncertainty.
2 This is the standard view on the fate of Nicanor's system (e.g. Wendel [n. 1], 277; Matthaios [n. 1], 904). But there are exceptions. The commentaries on the speeches of Gregory Nazianzen from the pen of Basilius Minimus (tenth century), for instance, are indebted to Nicanor's system. See Schmidt, T.A., Basilii Minimi in Gregorii Nazianzeni orationem XXXVIII commentarii (Leuven, 2001), XXI–XXIIIGoogle Scholar, who, however, underestimates the degree to which Basilius’ practice actually differs from Nicanor's. Basilius probably meant to follow Nicanor, but did not fully understand his system. The terminology used in sch. Dion. Thrax pp. 54.27–55.5 Hilgard is reminiscent of Nicanor's (D. Blank, ‘Remarks on Nicanor, the Stoics and the ancient theory of punctuation’, Glotta 61 [1983], 48–67, at 57 n. 40).
3 On the function of the ἐνυπόκριτος ὑποστιγμή, see Nünlist (n. 1 [2020]), 128–9 (§3.1). It is one of the three softer punctuation marks. Nicanor's eight marks fall into two groups (Blank [n. 2], 51).
4 Translations of the Iliad follow Lattimore, R., The Iliad of Homer. Translated with an Introduction (Chicago, 1951)Google Scholar, occasionally with modifications in order to render the passage as the ancient critic understands it. Translations of the scholia are my own.
5 On the function of the διαστολή, see Nünlist (n. 1 [2020]), 130–3 (especially §4.1 for the specific function of separating the preceding main clause from the subsequent subordinate clause).
6 ὑποστιγμὴ δὲ διανοίας μηδέπω ἀπηρτισμένης ἀλλ’ ἔτι ἐνδεούσης σημεῖον (Dion. Thrax §4, p. 7.6–7 Uhlig, p. 44.4–5 Lallot). See Blank (n. 2), who also demonstrates that the differentiation between complete and incomplete phrases is common among ancient grammarians and originates in Stoic linguistics.
7 Therefore Nicanor's differentiation between the punctuation of ‘regular’ and ‘inverted periods’ is less arbitrary than might seem at first sight. A preceding main clause might well be complete; the subsequent subordinate clause is not strictly necessary. Conversely, a preceding subordinate clause needs to be followed by a main clause in order to result in a complete period. Nicanor's notion of completeness (αὐτοτέλεια) is further explored in R. Nünlist, ‘Nicanor of Alexandria on grammatical completeness’, forthcoming in Glotta.
8 The name ἐνυπόκριτος ὑποστιγμή is transmitted by sch. Dion. Thrax p. 27.12–13 Hilgard (cf. n. 1 above). Instead of the adjective ἐνυπόκριτος, the relevant Homeric scholia use the alternative expression ἐν ὑποκρίσει (sch. A Il. 1.512b, 2.123–7, 2.139b, 2.148b, etc., all attributed to Nicanor).
9 On the function of the ἀνυπόκριτος ὑποστιγμή, see Nünlist (n. 1 [2020]), 129–30 (§3.2). That other ὑποστιγμή and its function are likely to be Nicanor's invention (130). On Nicanor's treatment of insertions (διὰ μέσου), see Nünlist, R., ‘Two cornerstones of Nicanor's syntactic explanations’, RFIC 147 (2019), 395–416Google Scholar.
10 In practice, the relevant Homeric scholia repeatedly fail to specify which of the two ὑποστιγμαί is actually meant (the same observation applies to the five different στιγμαί: Nünlist [n. 1 (2020)], 127 [§2.6]). This state of affairs may be the result of the multi-stage abbreviation process that the scholia underwent. Truly ambiguous cases are nevertheless rare, though the two fundamental errors in the description of sch. Dion. Thrax pp. 26.4–28.8 Hilgard (cf. n. 1 above) both concern the function of Nicanor's ὑποστιγμαί (Nünlist [n. 1 (2020)], 129 n. 18).
11 Friedländer (n. 1), 59, tacitly accepted by Blank (n. 2), 50.
12 This intonation applies, in particular, to yes-no-questions, whereas wh-questions have an intonation peak on the interrogative word. As far as the intonation of questions specifically in ancient Greek is concerned, there is, admittedly, little hard evidence to rely on. The fact that interrogative words such as τίς/τί always retain their acute accent is generally taken as an indication of the intonation peak in wh-questions: see P. Chantraine, Grammaire homérique, vol. 2 (Paris, 1963), 10; Devine, A.M. and Stephens, L.D., The Prosody of Greek Speech (New York, 1994), 454–5CrossRefGoogle Scholar. The assumption that yes-no-questions were indicated by raising the pitch of the voice towards the end of the sentence (presupposed without proof by various grammars: e.g. R. Kühner and B. Gerth, Ausführliche Grammatik der griechischen Sprache. Zweiter Teil: Satzlehre, 2 vols. [Hannover, 1898], 2.523, Chantraine [this note]) can be supported by a typological argument: 95% of the world's languages use this marker in order to differentiate between yes-no-questions and statements; see P. Siemund, ‘Interrogative constructions’, in M. Haspelmath et al. (edd.), Language Typology und Language Universals. An International Handbook, vol. 2 (Berlin, 2001), 1012–13 (I owe this reference to Alessandro Vatri). As a ‘null-subject language’, ancient Greek cannot mark yes-no-questions by means of an inversion (e.g. ‘are you ready?’). Homeric Greek, in particular, has no equivalent to the Attic interrogative word ἆρα (Kühner and Gerth [this note], 2.527). The particle ἦ (‘surely, truly, really’), in turn, can be used to introduce both statements and yes-no-questions. As a result, the illocutionary force of several Homeric sentences remains disputed to this day (see e.g. the examples mentioned in nn. 16, 26).
13 The five notes are mentioned by Friedländer (n. 1), 70–1 as examples for wrongly placed ἐνυπόκριτοι ὑποστιγμαί, without, however, touching upon the history of the question mark. Schmidt, K.E.A., Beiträge zur Geschichte der Grammatik des Griechischen und Lateinischen (Halle, 1859), 536Google Scholar with n. 34 underestimates their relevance because his focus is on finding the unique question mark.
14 Nünlist (n. 1 [2020]), 125–6 (§2.1).
15 The note does not specify the ὑποστιγμή, but there can be no doubt that the ἐνυπόκριτος is meant; cf. n. 10 above.
16 The relevant notes, at least in their extant form, do not make it clear whether Nicanor accepts the analysis of Il. 3.428a as a question. In modern scholarship this is done by, among others, K.F. Ameis, C. Hentze and P. Cauer, Homers Ilias, vol. 1 (Leipzig, 19137, repr. Amsterdam 1965), ad loc. or the Basel commentary, J. Latacz (ed.), Homers Ilias: Gesamtkommentar. 3. Gesang (Berlin, 2009), ad loc.
17 The στιγμή in question is the πρώτη ἄνω (Friedländer [n. 1], 56), on which see Nünlist (n. 1 [2020]), 126–7 (§2.3).
18 For this translation of οὐ πάντως here, see Friedländer (n. 1), 71 with n. 9. From a terminological point of view, the note seems to indicate that Nicanor does not observe the distinction made in ancient grammar and rhetoric between πεῦσις (wh-question) and ἐρώτησις (yes-no-question), on which see Rijksbaron, A., ‘A question of questions: peusis, erôtêsis and [Longinus] Περὶ ὕψους 18.1’, Mnemosyne 56 (2003), 733–7CrossRefGoogle Scholar. His notes discussed in this article, at any rate, deal with yes-no-questions. See also n. 27 below.
19 For the notion that questions are spoken ἐν ὑποκρίσει, see also Ps.-Herodian, Fig. 37 Hajdú (illustrating the phenomenon with a literary yes-no-question of uncertain paternity, see apparatus criticus); cf. sch. vet. Ar. Plut. 651b.
20 Nicanor's notes on ἐνυπόκριτοι ὑποστιγμαί regularly identify the conjunction on which the subordinate clause depends, e.g. (all attributed to him) sch. A Il. 1.512b (ὡς), 2.80a (εἰ), 2.198a (relative ὅς), 3.221–2 (ὅτε), 12.13a (ἐπεί), 18.88–90 (ἵνα).
21 The illocutionary force of lines 83b–84 is not discussed in the extant notes. Nicanor may have read the entire speech as an alternative question (see n. 22 below).
22 This extension is hardly surprising because alternative questions such as Il. 4.15–16 (taken as independent questions) can easily be analysed as a subtype of yes-no-questions (‘Shall we do A? Or B?’); see Rijksbaron, A., The Syntax and Semantics of the Verb in Classical Greek (Chicago, 2006 3), 56Google Scholar. The preceding example Il. 4.82–4 might in fact be read as an alternative question.
23 In sch. A Il. 9.673–4 Nic., sch. Od. 4.487a Nic. and sch. Od. 4.712a Nic. both options, dependent and independent, are considered feasible with no indication of a preference. In sch. A Il. 21.226 Nic. the two alternatives are an independent statement (ἀποφατικός) or a dependent question.
24 The wording of the note contains a small oddity: ‘She [sc. Helen] asks ironically. Therefore, one must not put a hupostigmê on parestês’ (ἠθικῶς πυνθάνεται· διὸ οὐ δεῖ ὑποστίζειν εἰς τὸ “παρέστης”). One might have expected Nicanor to write something like ‘nevertheless’ instead of ‘therefore’ (διό).
25 According to Erbse (n. 1), ad loc., the note refers to the punctuation mark at the end of line 298, but this cannot be right.
26 In modern scholarship, line 299 is read as a question by, among others, Janko, R., The Iliad: A Commentary. Volume IV: Books 13–16 (Cambridge, 1992)Google Scholar, ad loc. and the Basel commentary, J. Latacz (ed.), Homers Ilias: Gesamtkommentar. 14. Gesang (Berlin, 2015), ad loc.
27 For parallels where Nicanor's note spells this out, see (all attributed to him) sch. A Il. 1.290–1, sch. AbT Il. 3.46–52, sch. A Il. 9.673–4, sch. A Il. 10.61a, sch. A Il. 10.545–6a (with a remarkable transition from indirect to direct question), sch. A Il. 13.727–8, sch. A Il. 24.381. Unsurprisingly, all these notes deal with yes-no-questions, which are more difficult to detect for the reader, whereas wh-questions are marked by interrogative words; cf. n. 12 above. The question mark found in early Syriac manuscripts indicates yes-no-questions (J.F. Coakley, ‘An early Syriac question mark’, Aramaic Studies 10 [2012], 193–213, at 201); cf. n. 38 below. The Latin question marks found in Carolingian manuscripts, on the other hand, signal yes-no-questions and wh-questions alike: see B. Bischoff, Paläographie des römischen Altertums und des abendländischen Mittelalters (Berlin, 20094 [19791]), 225.
28 Rinas, K., Theorie der Punkte und Striche. Die Geschichte der deutschen Interpunktionslehre (Heidelberg, 2017), 196–7Google Scholar. This punctuation applies to preceding conditional clauses that lack a conjunction. Instead of such a conjunction the word order is inverted, which has the clause resemble a yes-no-question (cf. n. 12 above). An example is (Logau [1605–55], quoted from Rinas, with contemporary spelling): ‘Kan die Deutsche Sprache schnauben, schnarchen, poltern, donnern, krachen? Kann sie doch auch spielen, schertzen, liebeln, güteln, kürmeln, lachen.’ (‘[If] the German language can snort, snore, bluster, thunder, crash [question mark] it can also play, jest, flirt, cuddle, babble, laugh.’).
29 In his apparatus criticus on sch. A Il. 1.290–1 Nic., Erbse (n. 1) suggests understanding ἐπὶ δὲ τὸ “μυθήσασθαι” (Il. 1.291) στικτέον <πευστικῶς> (‘a [question] mark must be put after muthêsasthai’). Nicanor does understand the relevant line as a yes-no-question (cf. n. 27 above), but there is no evidence that he ever used a question mark. On the contrary, questions are rounded off with a στιγμή (usually a τελεία), whether they are wh-questions as in Il. 4.351, 10.61, or 18.80–2 (sch. A Il. 4.351–3 Nic., sch. A Il. 10.61a Nic., sch. A Il. 18.82a Nic.) or yes-no-questions as in Il. 4.93–4 or 10.204–6/13 (sch. T Il. 4.93c Nic., sch. AbT Il. 10.204–13a Nic.). Erbse's suggestion is unlikely to be correct. The claim that Nicanor puts a τελεία στιγμή ‘before and after interrogatives’ (Blank [n. 2], 56) must be a slip of the pen.
30 The list makes no pretence to completeness.
31 On the other hand, the two notes quoted in the main text are, to my knowledge, the only attestations of ὑποστιγμή (and cognates) in the sense of ‘question mark’ in Eusebius’ extant works.
32 As with the anonymous targets of Nicanor's criticism, the same punctuation mark can have several functions for Athanasius. The use of the ὑποστιγμή as a question mark recurs e.g. in PG 27.132.138–9 (on Psalm 21.3, essentially replicating Eusebius’ interpretation, quoted above), 27.97.32 (on 52.5), 27.124.29–33 (on 18.4).
33 See PG 69.785.23 (on Psalm 9.34), PG 69.809.15–19 (on 15.4), etc. One of Cyril's examples (Paul's letter to the Corinthians 1.6.3, with the commentary printed in P.E. Pusey, Sancti patris nostri Cyrilli archiepiscopi Alexandrini in D. Joannis evangelium, vol. 3 [Oxford, 1872], 262) also occurs among the instances that Coakley (n. 27) was able to identify in Syriac manuscripts of the fifth and sixth centuries (cf. n. 38 below).
34 Theodoretus, too, uses the ὑποστιγμή for multiple purposes. For another question mark, see PG 82.161.34–5 (on Rom. 9.30, on a very similar wh-question with subsequent answer).
35 E.g. sch. Eur. Hec. 1015, sch. Eur. Hipp. 1076a (ed. Cavarzeran), sch. Thuc. 1.35.3d (ed. Kleinlogel–Alpers); see also sch. rec. Ar. Plut. 139.
36 Besides, Nicanor's understanding of the passage is made clear in sch. A Il. 10.424–5a Nic., including the telling establishment of the natural word order (and the modification of the accent on ἦ/ἤ): πῶς γὰρ νῦν εὕδουσι, Τρώεσσι μεμιγμένοι ἢ ἀπάνευθε; (‘How then are they sleeping, mixed with the Trojans or apart?’).
37 The argument that ὑποστιγμή originally meant ‘weaker point’ is Schmidt's (n. 13), 515, accepted by Blank (n. 2), 52 n. 24.
38 For Latin script, see e.g. Bischoff (n. 27), 152, 225; Parkes, M., Effect and Pause. An Introduction to the History of Punctuation in the West (Cambridge, 1992), 35–6Google Scholar; Fees, I., ‘Interpunktion’, in M. Kluge (ed.), Handschriften des Mittelalters (Ostfildern, 2015 2), 48–55Google Scholar, at 53; all three assume that the mark is meant to have a bearing on the intonation as well. For Greek, see e.g. Randolph, C.B., ‘The sign of interrogation in Greek minuscule manuscripts’, CPh 5 (1910), 309–19Google Scholar. According to these and many other witnesses, the eighth/ninth century is to be credited with the ‘invention’ of the question mark tout court. In this comprehensive form, this opinion can no longer be upheld owing to Coakley (n. 27), who demonstrates that in Syriac manuscripts of the fifth and sixth centuries yes-no-questions are marked with a particular sign (a colon placed above a word within the relevant sentence, later called zawgā ‘elāyā, ‘upper pair’). But Coakley, too, follows the traditional date for the introduction of question marks by Greek and Latin scribes (Coakley [n. 27], 211) and does not address the issue whether they might have been influenced by the Syriac practice. Such an influence seems unlikely in the light of the present article.
39 This palaeographical argument lies beyond the scope of the present article. It is bedevilled by at least two factors: the scarcity of relevant material that predates the eighth/ninth century and the use of the same punctuation mark for multiple purposes.