¹ For instance, pragmatics, functional grammar and sociolinguistics. For holistic perspectives on Latin that take into account the former two fields, students have recourse to Pinkster, H., Latin Syntax and Semantics (London, 1990)Google Scholar and now to Pinkster, H., Oxford Latin Syntax (Oxford, 2015)CrossRefGoogle Scholar. For the last-named field, sociolinguistics, readers may consult most recently Adams, J.N., Social Variation in the Latin Language (Cambridge, 2013)CrossRefGoogle Scholar, or, for an introduction, Clackson, J. ‘The social dialects of Latin’, in Clackson, J. (ed.), A Companion to the Latin Language (Malden, MA, 2011), 505–26CrossRefGoogle Scholar.

² For dates and further information on the Latin grammarians cited throughout this article on the –to imperative, see the prosopography with bibliographical references in Kaster, R., Guardians of Language: The Grammarian and Society in Late Antiquity (Berkeley, CA, 1988), 233–440 CrossRefGoogle Scholar. The first stop in attempting to understand the –to imperative is, of course, at the standard grammars, H.–Sz. 340–1, and K.–St. 1.196; see also Pinkster (n. 1 [1990]), 220. Numerous specialized studies also exist. E. Loch, Zum Gebrauch des Imperativus bei Plautus (Memel, 1871) offers a collection of material, as does Bennett, C., Syntax of Early Latin, vol. 1: The Verb (Hildesheim, 1966), 354–61Google Scholar. Riemann, O., ‘La question de l'impératif latin en –to ’, RPh 10 (1886), 161–87Google Scholar and Vairel-Carron, H., Exclamation, Ordre, et Défense (Paris, 1975)Google Scholar deal with issues that will be taken up below, as does R. Risselada's work, The Imperative and Other Directive Expressions in Latin: A Study in the Pragmatics of a Dead Language (Amsterdam, 1993)Google Scholar, which deals with the Latin directive, including the –to imperative, with the help of speech-act theory, politeness and discourse analysis.

³ This approach was identified as ‘21^st-century philology’ by Goldstein, D., ‘Review of A. Mercado, Italic Verse: A Study of the Poetic Remains of Old Latin, Faliscan, and Sabellic (Innsbruck, 2012)’, Gnomon 87 (2015), 695–703 CrossRefGoogle Scholar, at 695 with n. 1.

⁴ These were gathered by reading through the editions of Lindsay (1910, repr. with additions) for Plautus (with consultation when necessary of the other major reference editions: Ritschl et al. [1878–1902] and Leo [1895]) and the second edition of Kauer and Lindsay for Terence. Approaches to the Latin –to imperative similar to my own: Risselada (n. 2), who draws from Vairel-Carron's (n. 2) data, in addition to data from Plaut. Mostell. and a selection from Cicero's letters; Gibson, R., ‘Didactic poetry as “popular” form: a study of imperatival expressions in Latin didactic verse and prose’, in Atherton, C. (ed.), Form and Content in Didactic Poetry (Bari, 1997), 67–98 Google Scholar, who offers a fascinating essay taking into account Risselada's work, while drawing on a different corpus (Latin prose and verse didactic treatises), and Unceta-Gómez, L., La petición verbal en Latín: estudio léxico, semántico y pragmatico (Madrid, 2009), 43–4Google Scholar, who provides a brief account focussing on speech-act categories. Finally, Barrios-Lech, P., Linguistic Interaction in Roman Comedy (Cambridge, 2016), 59–62 CrossRefGoogle Scholar provides a summary of some of the issues raised here. I take this opportunity to explain a discrepancy. In my book, I identify 354 –to imperatives in Plautus; in this article, 369. For this article, I add to the 354 –to imperatives identified in Barrios-Lech (this note) fifteen (15) more tokens: eleven greetings (salueto) or leave-takings (ualeto); three instances of nolito (Cist. 108, Poen. 872 and 1320), and one instance at Persa 667. These I excluded from my book for various reasons, but I see fit to include them here for the sake of presenting the complete data. See further n. 38 below. Henceforth, I will indicate in a footnote wherever this article discusses more fully an issue summarized in Barrios-Lech (this note).

⁵ Plaut. Mostell. 579: redito huc circiter meridie.

⁶ Plaut. Asin. 740: Leonida, curre opsecro, patrem huc orato ut ueniat. On this construction, see K.–St. II.1§50.3.c.

⁷ Plaut. Stich. 287 (a seruus currens, in a self-address): si rex opstabit obuiam, regem ipsum priu’ peruortito. The future verb may be a future indicative (as in this example), a future perfect or a future sigmatic verb. See K.–St. II.1§50.3.b.

⁸ Plaut. Mostell. 255: ubi tu commoda es, capillum commodum esse credito; here ubi has the sense ‘whenever’. See e.g. Vairel-Carron (n. 2), 236 n. 2.

⁹ Plaut. Asin. 240: modo tecum una argentum adferto, facile patiar cetera. See for this construction Loch (n. 2), 8–9. The future verb may be a future indicative (as in the previous example), a future perfect or a future sigmatic form.

¹⁰ The ablative *–tod of the demonstrative *–to paired with an order for immediate execution produced *agetod, which would mean ‘from then on, lead/do/drive’; or ‘en s’éloignant de maintenant’, as Vairel-Carron (n. 2) translates; ‘von da an’: H.–Sz. II.§188. For further details, see L.–H.–S. I.§423, Vairel-Carron (n. 2), 257–9 and Meiser, G., Historische Laut- und Formenlehre der lateinischen Sprache (Darmstadt, 1998), 41 and 220–1Google Scholar. I will not discuss here the well-known ambiguity of person present in the –to imperative form, between the second and third persons. Fuller discussion on this issue at Vairel-Carron (n. 2), 224–8, arguing at 227 that ‘pour les Latins de l’époque de Plaute’ the form was not ‘indifférenciée du point du vue de la personne’, but Rosén, H., Latine Loqui: Trends and Directions in the Crystallization of Classical Latin (Munich, 1999), 115 Google Scholar disagrees: ‘[o]n the contrary: [t]he meaning of the –to imperative … can be brought to one common denominator, that of the … apersonal imperative’. Risselada (n. 2), 131 argues for ‘indefinite reference’: ‘this kind of directive … is addressed to all potential hearers or readers, but is relevant … only for a particular subset of addressees that meet with particular conditions'. Augustine, bishop of Hippo, notes that there is an explanation why facito is used both for the second and for the third persons but then, frustratingly, declines to offer it: sed nunc [sc. ratio] praetermittenda est, ne ad aliam rem transeamus (Keil 5.510.11–12).

¹¹ All translations are the author's unless noted otherwise.

¹² Blase, H., Landgraf, G., Golling, J. (edd.), Historische Grammatik der lateinischen Sprache, vol. 3.1: Syntax des einfachen Satzes (Leipzig, 1903), 236Google Scholar; Löfstedt, L., Les expressions du commandement et de la défense en latin et leur survie dans les langues romanes (Helsinki, 1966), 22–9Google Scholar and 38–40 discusses the post-classical diminution in use of the –to imperative within the spoken language.

¹³ Blase, Landgraf, Golling (n. 12), 235, who note that the prevalent use of the –to imperative in laws ‘bleibt durch die ganze Latinität’.

¹⁴ Adams, J.N., Pelagonius and Latin Veterinary Terminology in the Roman Empire (Leiden, 1995), 460 CrossRefGoogle Scholar on Pelagonius; Gibson (n. 4), 82 on Pelagonius and Palladius (a writer of an agricultural treatise); Rosén (n. 10), 12 on Cicero's legal language in De legibus; and Hine, H., ‘“Discite … agricolae”: modes of instruction in Latin prose agricultural writing from Cato to Pliny the Elder’, CQ 61 (2011), 624–54CrossRefGoogle Scholar, at 649–50 on Pliny the Elder's use of this form in HN, as a ‘tribute to Cato’.

¹⁵ I thus exclude 19 forms with third-person reference from my corpus in the above table; cf. Vairel-Carron (n. 2), 230–2; to her list of 18 forms I add Plaut. Merc. 1021, neu quisquam posthac prohibeto.

¹⁶ I use the exhaustive word-counts conducted by M. Gilleland, ‘The linguistic differentiation of character type and sex in the comedies of Plautus and Terence’ (Diss., University of Virginia, 1979), 80–3.

¹⁷ The statistic relevant to Cato's treatise on agriculture is calculated by Hine (n. 14), 628.

¹⁸ We return to this point below.

¹⁹ Adams, J.N., ‘Female speech in Latin comedy’, Antichthon 18 (1984), 43–77 CrossRefGoogle Scholar, at 50.

²⁰ For a description of the z-test and a sample calculation, applied to a simple problem in linguistics, see e.g. Butler, C., Statistics in Linguistics (New York, 1985), 92–5Google Scholar.

²¹ Cf. Barrios-Lech (n. 4), 59–60, K.–St. II.1, 198 (§50.c. Anmerk. 1) and Vairel-Carron (n. 2), 328, who argued for the diminution of the form along a different tack: according to her, 77.25% of the –to imperatives in Plautus do not appear with a subordinate clause, compared with 67.70% in Terence. 22.75% of –to imperatives in her data for Plautus appear with the subordinate clause containing a future indicative; her corresponding figure for Terence is 32.30%. She suggests that the cause for these divergences is that the –to imperative solitarium was receding in the spoken language from Plautus' time to Terence's, but I disagree with the underlying assumption that future imperatives appearing with a subordinate clause are ‘frozen’ expressions, ‘qui sont plus susceptibles de se figer et de se conserver uniquement par la tradition’.

²² Goldberg, S., Terence Hecyra (Cambridge, 2013), 100 Google Scholar. Rather, the –to imperative, since it conveys instructions, perfectly suits the tricky and ‘wise adviser’ slave character-type: see section 5 below.

²³ Gratwick, A., Plautus Menaechmi (Cambridge, 1993), 140 Google Scholar and de Melo, W., ‘Review of E. Karakasis, Terence and the Language of Roman Comedy (Cambridge, 2005)’, Mnemosyne 59 (2006), 268–74Google Scholar, at 270.

²⁴ Haffter, H., Untersuchungen zur altlateinischen Dichtersprache (Berlin, 1934)Google Scholar, followed, in the main, by Happ, H., ‘Die lateinische Umgangssprache und die Kunstsprache des Plautus’, Glotta 45 (1967), 60–104 Google Scholar.

²⁵ For the markings C (canticum) and DV (deuerbium) in the manuscripts, see e.g. Moore, T., Music in Roman Comedy (Cambridge, 2012), 13–14 Google Scholar.

²⁶ Haffter (n. 24), 44–5, Happ (n. 24), 88–92.

²⁷ Moore (n. 25), 103.

²⁸ See Moore (n. 25), 15–16 for discussion and see further the helpful tables at 382–3.

²⁹ Not the total, 369: the metrical setting of Cas. 870 cannot be determined owing to the fragmentary nature of the text there, so that example is excluded.

³⁰ The chi-square test allows us to determine whether a distribution, like that for the –to imperative, accords with the expected distribution, and thus supports a null hypothesis that the token is randomly distributed throughout the plays. Such random distribution would be the expected one for a stylistically unmarked element. The chi-square test in this case forces us to abandon the null hypothesis. For more on this test, see, for instance, Butler (n. 20), 112–26.

³¹ de Melo, W., The Early Latin Verb System: Archaic Forms in Plautus, Terence, and Beyond (Oxford, 2007), 10–11 CrossRefGoogle Scholar and passim.

³² Plautus: the –to imperative in low-register contexts at Amph. 770, Aul. 55–6, Capt. 112–14, routine commands to a slave; in Bacch. 327–8 it co-occurs with the colloquial heus; the –to imperative in high-register contexts at Mil. 161–5, in language which clearly mimics that of a formal edict, introduced at Mil. 159 with the phrase nunc adeo edico omnibus, though the present imperative is equally at home in such mock edicts (see Capt. 803–4, prius edico … | continete uos domi, prohibete a uobis uim meam), mock edicts which the eavesdropping old man calls basilicas (Capt. 811); in Persa, the slave has wrapped his pallium around himself gloriose (307), and spits magnufice (308). He then utters a string of future imperatives: adito | uidebitur. factum uolo. uenito. promoneto (310–11), although here it is probably the asyndetic pile-up of these imperatives that conveys the tone, not the future imperatives as such; at Rud. 103, the young man's salueto may strike a loftier tone in his address to an ignotus: pater salueto; at Rud. 1342, the slave and pimp stand before the temple of Venus, as the slave leads the pimp in an oath, tum ego huic Gripo, inquito et me tangito. On inquito in formal/ceremonial contexts, see below.

³³ Bagordo, A., ‘Langversstil und Senarstil bei Terenz’, in Kruschwitz, P., Ehlers, W.-W. and Felgentreu, F. (edd.), Terentius Poeta (Zetemata 127) (Munich, 2007), 127–41Google Scholar. It should be mentioned, however, that this study is carried out only for rhetorical figures and on a single play. Confirmation awaits a more thoroughgoing analysis.

³⁴ Terence: the –to imperative in high-register contexts at Ad. 970, manumission formula; Phorm. 984, the legal formula lege agito; in low-register contexts, at Ad. 281–2, absoluitote co-occurs with the colloquial heus; An. 865, order to a lorarius to bind a slave; Eun. 503, in a string of commands to a maidservant; Eun. 593–4, facito and lauato co-occur with heus tu.

³⁵ See for instance the nice analysis of Unceta-Gómez (n. 4), 43–4 and, earlier, K.-St. II.1, §50(c)5, page 199.

³⁶ See the note at Loch (n. 2), 6 for the earlier view, since refuted, that the future imperative was stronger than the present.

³⁷ Loch (n. 2), 11–12. See, in addition to the work cited at n. 35, Bennett (n. 2), 359–61, especially at 359: ‘the future imperative covers substantially the same field of meaning as the present.’

³⁸ Present imperatives from three plays have been analyzed: Plautus' Captivi and Truculentus and Terence's Phormio. 421, and not the grand total (434), –to imperatives in extant Roman comedy are considered, since I exclude greetings and farewells: these are, strictly speaking, not directives; rather they are ritualized speech-acts: see Risselada (n. 2), 117–18. I also exclude Plaut. Amph. fr. XII Lindsay, ibi scrobes ecfodito plus sexagenos in dies, because the context is absent, and Persa 667, where the reading is uncertain.

³⁹ For the z-test, see n. 20 above. I came to the same conclusion, but with an expanded data-set, at Barrios-Lech (n. 4), 42. However, I believe the new method here introduced to be more rigorous.

⁴⁰ On its permissive force, see e.g. Blase, Landgraf, Golling (n. 12), 248 and, more recently, Risselada (n. 2), 137.

⁴¹ One type of permission that recurs is the phrase audacter dicito (‘you can say that as boldly as you please’ [the implication is that you can say that whenever you like]): e.g. Plaut. Epid. 16, Merc. 726, Mostell. 916, Pseud. 828.

⁴² Thanks to the anonymous reader of the article for pointing this out to me. Again, the method used here is new, and more rigorous.

⁴³ This corroborates the position taken at Barrios-Lech (n. 4), 60.

⁴⁴ See e.g. K.–St. II.1, §3.a.

⁴⁵ Cf. Blase, Landgraf, Golling (n. 12), 237: ‘Auch ohne Bestimmung durch einen subordinierten Futuralsatz oder einen vorausgehenden Imper. I findet sich oft der Imper. II, um einen Befehl auszudrücken, der erst nach einer bestimmten Zwischenzeit in Zukunft ausgeführt werden soll, oder allgemein für alle Fälle, wo die Gelegenheit der Ausführung sich ergibt.’ (emphasis mine).

⁴⁶ This corroborates the conclusion reached at Barrios-Lech (n. 4), 60–1. Again, the new method used here is more rigorous.

⁴⁷ Blase, Landgraf, Golling (n. 12), 246; Bennett (n. 2), 357 and K.–St. II.1, 199 (§50.c.5).

⁴⁸ According to the z-test. Ultimately, though, the result should not be treated as definitive, for, in some cases, to assess whether an imperative is softened or not is a subjective matter.

⁴⁹ There is a similar pattern in Terence: 7.8% of all present imperatives are softened, whereas a slightly smaller proportion (6.2%) of –to imperatives are. Yet, the z-test shows that there is no significant variation between these two proportions. Is it possible that, by the time Terence was writing, the –to imperative was not necessarily felt to be milder than the present imperative?

⁵⁰ Neue, C., Formenlehre der lateinischen Sprache (Berlin, 1875), 403–5Google Scholar with Thurot, C., ‘De l'impératif futur latin’, RPh 4 (1880), 113–17Google Scholar, quote from 117.

⁵¹ Riemann (n. 2), 185.

⁵² Vairel-Carron (n. 2), 243 n. 1: ‘passages de ce genre (that is, imperatiuus futuri pro imperatiuo praesentis) sont extrêmement rares’, cf. H.–Sz. II.§188, page 341: ‘Die altlateinische Belege für Imper. Fut. statt Imper. Praes. sind noch spärlich und nicht eindeutig’; they cite the interesting example of Enn. scaen. 420: extemplo … me necato et filiam.

⁵³ Vairel-Carron (n. 2), 327: ‘there are no strict breaks or boundaries between the immediate future [the relevant time-sphere for the present imperative] and the less-immediate future; between the distant future and an indefinite future. One passes from one to the other imperceptibly and gradually.’

⁵⁴ Riemann (n. 2), 182.

⁵⁵ On scito, memento and esto as suppletives, see H.–Sz. II.§188 and Rosén (n. 10), 115. For the two other tokens excluded, see n. 38 above.

⁵⁶ Vairel-Carron (n. 2), 240 says that the use of the imperative with –to indicates that ‘the order is valid for a period further than the immediate present, with the context showing, in a sufficiently clear way, that it is also valid for the immediate future.’ Thus, according to her, salueto = ‘fare well’ for the distant future but also (by implication) for the present time. I prefer as simpler the explanation that verbs with atelic aspect, when conveyed as commands, tend to attract the –to imperative, though it may well be that in some cases the speaker wishes to conversationally implicate that the –to imperative also applies to the present (on conversational implicature, see Grice, H.P., ‘Logic and conversation’, in Cole, P. and Morgan, J. (edd.), Syntax and Semantics 3: Speech Acts [New York, 1975], 41–58)Google Scholar.

⁵⁷ Pieces of wisdom conveyed as ‘timeless’ advice: Plaut. Cas. 209, Cist. 62, 64, Mostell. 255, Pseud. 312, Trin. 323, 485; Ter. Haut. 221, 590, Ad. 417. Fatherly advice: Trin. 295–6.

⁵⁸ Similar: Plaut. Aul. 55–6, Men. 334, Poen. 890, Rud. 836, Truc. 197, 953. Cf. Riemann (n. 2), 171.

⁵⁹ For an introduction to this concept, see Filip, H., ‘Lexical aspect’, in Binnick, R.I. (ed.), The Oxford Handbook of Tense and Aspect (Oxford, 2012), 721–51Google Scholar, especially 721–4 with the useful tests for telicity there.

⁶⁰ Loch (n. 2), 10–11, however, and Riemann (n. 2), 179 see instances of ualeto as ‘present for future’, but this cannot be the case given the meanings of the verbs, as argued above. Of course, the present imperatives salue and uale are much more frequent, but this does not imply that, when salueto and ualeto are used, they are replacements for salue and uale. Rather, the two forms overlap in temporal scope (= now and in the future). Bene ambulato refers to a journey the addressee will make in the not-too-distant future, after he parts ways with the interlocutor.

⁶¹ For this meaning, see OLD s.v. gratia 3d.

⁶² Apart from the example quoted, see Curc. 699. Habeto is similar when it means ‘consider’; e.g. Poen. 542, per iocum itidem dictum habeto quae nos tibi respondimus ‘consider the answer we gave you as a joke’.

⁶³ See Risselada (n. 2), 44–5.

⁶⁴ The force of cogitato in these cases is made clear in the following: Mil. 1364: cogitato identidem tibi quam fidelis fuerim. Cogitato similarly precedes an extended comparison at Mil. 915–17; or wise advice that deserves further reflection: e.g. Plaut. Poen. 237, Ter. Eun. 759. Eight examples of cogitato in Roman comedy. Similar is the expression facito (ut) cogites: Plaut. Stich. 519 (hoc tu facito ut cogites), Trin. 485 (semper tu hoc facito … cogites), Ter. Ad. 500 (hoc tu facito cum animo cogites), Ad. 808 (quaeso hoc facito tecum cogites); seven examples in Roman comedy. Interestingly, both examples in Terence (there are only two) are put in the mouth of old characters: Ad. 500, 808. The examples of credito at Haut. 577 and Phorm. 874 imply belief not only right now but also continued into the future. It should be noted that the parallel idiom in Greek has a present imperative: Men. Sam. 697 and 703 (Arnott).

⁶⁵ Though not necessarily metadirectives, the expressions facito ut memineris and putato are similar, as they imply action that should be undertaken immediately and be continued: facito ut memineris: e.g. Mil. 354, Stich. 47; cf. Poen. 1278, patrue, facito in memoria habeas; Poen. 1418. Eight examples in Roman comedy. Also similar are examples at Cas. 94–6, the metadirective conicito with dehinc ‘from this time forward’ (dehinc conicito ceterum | possisne necne clam me sutelis tuis | praeripere Casinam uxorem, proinde ut postulas), and the metadirective at Phorm. 166–7 (tu conicito cetera | quid ego ex hac inopia nunc capiam et quid tu ex ista copia), where the addressee is to ‘put things together’ after he listens to the prompt, and then for some time afterward. Of putato there are no examples in Plautus; two in Terence, at Phorm. 424 and Ad. 817. On this verb in Terence = existimo, iudico, see Karakasis, E., Terence and the Language of Roman Comedy (Cambridge, 2005), 178 CrossRefGoogle Scholar. Obviously, putato, when it means existimato, refers to an action that should be undertaken and then continued into the future. As far as putato is concerned, Terence approaches the colloquial register of CL. It is avoided in Plautus; in Cicero putato occurs in the letters (seven times) and in speeches (seven times), but is absent from philosophical and rhetorical works.

⁶⁶ The three instances at Plaut. Aul. 458, Plaut. Mil. 453, Ter. Phorm. 984. In all, the speaker utters lege agito when refusing to deal with the addressee's claim; in a similar situation we would say, in English, something like ‘fine, go and sue me for all I care’. In all the passages, the speaker directs or has directed physical violence of some kind to the addressee (Euclio, speaker of Plaut. Aul. 458, has just beaten the cook; the speaker restrains Philocomasium in Mil.; the speaker similarly restrains the parasite in Phorm.).

⁶⁷ H.–Sz. II.§188. Terence never uses inquito, which appears only three times in Plautus.

⁶⁸ Thus, Plaut. Bacch. 883 occurs in a ‘stipulatio-scene’, as does Plaut. Pseud. 535–8, passage 11 above.

⁶⁹ Ter. Haut. 329, a passage like the one from Ter. Phorm. 917–19, passage 12 above.

⁷⁰ Transl. de Melo, W., Plautus, vol. 5 (Cambridge, MA and London, 2013), 163Google Scholar.

⁷¹ Riemann (n. 2), 185.

⁷² Karakasis (n. 65), 40 and 89 shows that slaves' speech, at least in Terence, is marked by the presence of colloquial features. Both Plautus and Terence have slaves speak markedly more Greek words; on this, see Maltby, R., ‘The distribution of Greek loan-words in Plautus’, in Brock, R. and Woodman, A.J. (edd.), Papers of the Leeds International Latin Seminar, vol. 8 (Leeds, 1995), 31–69 Google Scholar, at 34–5.

⁷³ See also Karakasis (n. 65), 5.

⁷⁴ For this use of nunc, see OLD s.v. nunc 10a: ‘Since or if this is the case, in light of these circumstances, then.’

⁷⁵ Cf. Barsby's translation in Barsby, J., Terence, vol. 2 (Cambridge, MA and London, 2001), 37 Google Scholar: ‘You attack him first; I'll be lying in wait here in reserve in case you need some support.’

⁷⁶ See Wessner, P. (ed.), Aeli Donati quod fertur Commentum Terenti: accedunt Eugraphi commentum et Scholia Bembina, 3 vols. (Leipzig, 1902–1908)Google Scholar, 1.9 for Caesar's epigram on Menander and Fantham, E., ‘Caesar as an intellectual’, in Griffin, M. (ed.), A Companion to Julius Caesar (Oxford, 2009), 141–56CrossRefGoogle Scholar, at 148–51, for Caesar's style and views on language.

⁷⁷ The confidence interval is 99.92%; this indicates that there is good reason to reject the null hypothesis that the futurum pro praesente form is randomly distributed over free and slave speech in the comedies. See further Barrios-Lech (n. 4), 17 on the confidence intervals.

⁷⁸ 361 (and not 369) tokens appear for Plautus, because in eight cases, the addressee is indeterminate, being either the audience or mixed addressees, or because the context cannot be determined owing to the state of the text.

⁷⁹ As the z-test confirms, with the calculation conducted on each of the three pairs of proportions. Thus, in Plautus, 39.3% of all –to imperatives are directed ‘downward’, and there is no significant difference between that proportion and the corresponding one in Terence, 38.5%.

⁸⁰ Adams (n. 19), 76.

⁸¹ We can take Gilleland's (n. 16) word-counts as the basis for the proportion of speech allotted to women and to men: 13.9% of the total speech in Plautus is allotted to women; 86.1% to men. For Terence, the figures are 13.3% (women) and 86.7% (men). In each author, the future imperative is distributed in roughly these proportions.

⁸² A summary of this discussion on the –to imperative and linguistic characterization can be found at Barrios-Lech (n. 4), 61–2.

⁸³ Measured as x –to imperatives/100 lines, where a line = to a ten-word unit. The last measure has been chosen to avoid giving unwieldy statistics.

⁸⁴ Ter. Eun. 595, 596.

⁸⁵ The relevant statistics are, for the ‘vitriolic wife’ type, 4 tokens of the –to imperative, representing 2 instances per 100 lines, and for the ‘good wife’ type, speaking 6 instances, at a corresponding average frequency of 2.3 per 100 lines.

⁸⁶ Apuleius (Flor. 16), in describing the comedies of Philemon, says: rarae apud illum [sc. Philemonem] corruptelae, tuti errores, concessi amores. nec eo minus et leno periurus et amator feruidus et seruulus callidus et amica illudens et uxor inhibens et mater indulgens et patruus obiurgator et sodalis opitulator et miles proeliator, sed et parasiti edaces et parentes tenaces et meretrices procaces.

⁸⁷ Measured as x –to imperatives/100 lines.

⁸⁸ Adams (n. 19), 56.

⁸⁹ It should be noticed, however, that two of the instances of –to imperative put in the mouth of slaves are patently mock politeness. See Asin. 375 quaeso, aequo animo patitor [sc. ‘when I punch you in the face’], and Mil. 865 ‘please, do at any rate accept my share in my absence, if the misfortune is divided between us’, quaeso tamen tu meam partem, infortunium | si diuidetur, me absente accipito tamen.