Professional Documents
Culture Documents
The Sigmatic Future in Archaic Latin
The Sigmatic Future in Archaic Latin
The Sigmatic Future in Archaic Latin
(1) Sı̄ nox fūrtum faxit, sı̄ im occı̄sit, iūre caesus estō. (Lex XII tab. ap. Macr.
Sat. 1. 4. 19.)
If he (x) shall have committed larceny at night, if he (y) shall have killed him,
he (x) shall have been killed lawfully.4
(2) Vtı̄ lēgāssit suae reı̄, ita iūs estō. (Lex XII tab. ap. Gaius inst. 2. 224.)
As he shall have made legacies of his possessions, so it shall be sanctioned.
There is little doubt that the indicative forms mark some kind of
future, but that is not sufficient. Five main questions arise:
1
This chapter is largely based on de Melo (2002b and c).
2
For legal style in general see Crawford (1996: i. 16–17).
3
For Cicero’s laws see Powell (2005) and Ch. 12.
4
For such inexplicit changes of subject see Daube (1956: 57). For this law see
Crawford (1996: ii. 612–13).
172 Synchronic Analysis
Table 6.1. Sigmatic futures in subordinate and main clauses in
Plautus and Terence: tokens
Plautus 53 70
Terence 1 9
5
For a general discussion of tense and aspect in Latin see Chs. 2 and 3 as well as
Pinkster (1983).
6
The sigmatic subjunctives are not restricted in this way, see Ch. 7.
The Sigmatic Future 173
Table 6.2. Sigmatic futures in subordinate and main clauses in
Plautus and Terence: types
Plautus 29 1
Terence 1 1
must be added that the verbs in subordinate clauses occur in all per-
sons and numbers, while the only form in main clauses is faxō. This
pattern does not change if we look at the rest of Archaic Latin. There
are thirty-four tokens.7 One of them is faxō in a main clause, the other
thirty-three are in conditional, temporal, and relative clauses. At first
sight, this seems to contradict one of the above findings based on
Plautus and Terence, namely that there are more tokens in main than
in subordinate clauses. However, this is not necessarily the case. Nine
of the tokens are citations from Archaic Latin literature. If we assume
that main clause faxō was relatively common here, it was probably
still understood by the grammarians and their contemporaries, and
so there was no need to gloss it. The one instance of faxō in a main
clause among these nine tokens is in Afran. com. 67. Non. p. 543. 23–4
cites the line not because of faxō, but because it contains the adjective
mı̄tis. The forms in subordinate clauses were rarer and more varied
and hence more interesting for the grammarians. The other twenty-
five tokens are in laws—hardly a sample of normal speech. They are
typically in the form sı̄/nı̄/quı̄ faxit, . . . estō, that is, the main clauses
regularly contain non-indicatival directives (future imperatives or
present subjunctives). In this connection compare Lex reg. ap. Fest.
p. 190: sı̄ hominem fulminibus occı̄sit, nē suprā genua tollitō, ‘if he
(Jupiter) has smashed a man with thunderbolts, one must not lift
him above one’s knees.’8 Obviously, texts like these do not allow main
clause faxō.
Because of this distributional pattern it is essential to look at the
semantics, register, and productivity of the forms in subordinate and
7
I leave out faxit in the treaty between Rome and Callatis because the syntactic
context is not entirely clear; the passage is in Warmington (1940: 292 + 294).
8
‘Lifting the corpse above one’s knees’ would mean ‘carrying the corpse away’, see
Latte (1932: 270–1).
174 Synchronic Analysis
main clauses separately (first two sections).9 After this I shall attempt
to explain this unusual distribution. Finally, I shall summarize my
findings and try to find a reason why the forms die out in the way
they do.
By Hercules, if you shall have gone a finger’s or a nail’s breadth from your
place, or if you shall have looked back until I’ve told you, I shall, by Hercules,
immediately put you on the cross for a lesson.
9
Bertocci (2001), who also discusses the sigmatic futures, fails to recognize this
pattern, which leads to some erroneous interpretations.
10
Lex reg. ap. Fest. pp. 190, 260, Paul. Fest. p. 5; Lex XII tab. ap. Macr. Sat. 1. 4. 19
(two tokens), Fest. pp. 496, 508, Vlp. dig. 9. 4. 2. 1 (two tokens); Lex Sil. ap. Fest. p.
288, two tokens; Lex Aquilia ap. Vlp. dig. 9. 2. 27. 5; fr. legis ap. Gell. 20. 1. 12; Lex luci
Lucer. (CIL i2 . 401) 4; Lex luci Spolet. (CIL i2 . 366) 2. 1 and 2. 4; CIL i2 . 2872, ii. 4 and
7; Lex agr. (CIL i2 . 585) 25; Enn. ann. 335; Acc. trag. 454 and 293; dēuōtiō in Macr. Sat.
3. 9. 11.
11
Lex XII tab. ap. Gaius inst. 2. 224, Fest. p. 176, Plin. nat. 28. 18 (two tokens); Lex
agr. (CIL i2 . 585) 25 and 71; Cato agr. 14. 1; dēuōtiō in Macr. Sat. 3. 9. 11 (two tokens).
The Sigmatic Future 175
The clause with the future perfect excesseris is co-ordinated with the
clause containing the sigmatic future respexis. It is unlikely that re-
spexis has a temporal reference different from that of excesseris, since
it merely introduces a more specific condition, not a new one. Similar
co-ordination of sigmatic future and future perfect in conditional
clauses can also be found in Bacch. 847–9, Cas. 1001–3, and Rud. 775–
6. Compare also:
(4) (The speaker is in a catch-22 situation.)
Perı̄bō sı̄ nōn fēcerō; sı̄ faxō, uāpulābō. (Plaut. Fretum.)
I shall perish if I shall not have done it; if I shall have done it, I shall be beaten
up.
Here the two conditional and main clauses are not co-ordinated, but
parallel to each other and have the same time reference; the future
perfect or sigmatic future stands in the conditional clauses, and the
simple future in the main clauses. Other instances of parallelism in
conditional or relative clauses can be seen in Capt. 122 + 124, Cas.
1015–18, Mil. 156–7 + 163, and Pseud. 531–4.12
These instances of co-ordination and parallelism seem to indicate
that the sigmatic futures have the same temporal reference as the
future perfects in subordinate clauses. I have discussed the semantics
of the future perfect in Ch. 2. Future perfects are used to show that
the event in the subordinate clause is over before the one in the
main clause begins. Consequently the sigmatic futures are also likely
to indicate anteriority. Just like the future perfects, they can all be
interpreted as perfective,13 but unlike them, they are restricted to
telic verbs, that is, verbs which imply the existence of an endpoint of
an action.14 This restriction may not be relevant for the synchronic
12
There are also two instances of co-ordination with and parallelism to rhotacized
forms: sı̄ . . . adiŭuerō cūramue leuāssō (Enn. ann. 335) and sı̄ . . . uerberit, ast . . . plōrāssit
(Lex reg. ap. Fest. p. 260). These rhotacized forms will be discussed at the end of Ch. 7.
In Lex XII tab. ap. Fest. p. 496 there is parallelism between a sigmatic and a present
form: sı̄ membrum rupsit (or rūpit?), nı̄ cum eō pacit, tāliō estō.
13
Under perfectivity I understand that an event is described as a whole, not as
being in progress and incomplete.
14
In Cas. 1001 amāre seems to mean ‘make love’ rather than static, atelic ‘love’, ‘be
in love’: si umquam posthāc aut amāssō Casinam aut occepsō modo should be translated
‘if after this I shall ever have made love to Casina or if I shall merely have begun to do
so.’ This restriction does not seem to apply to the sigmatic subjunctives, which can
176 Synchronic Analysis
Well, if you shall like it, use it; if you shall not like the plan, find a more
suitable one.
In Ex. 5 the addressees must still like the plan when the main clause
action begins.
In other words, the co-ordination of sigmatic forms and future
perfects is significant. We may see whether an anterior meaning
is suggested or contradicted by other evidence. Unfortunately, the
temporal adverbials are not very helpful. Posthāc (Cas. 1001) merely
situates an action in the future, and praeterhāc (Men. 112, Rud. 1117,
Stich. 345) need not even be taken as temporal because it can have
a discourse function as well. Ante (Asin. 818) and prius ( . . . quam,15
Epid. 122–3) suggest anteriority, even if they themselves are not suf-
ficient to prove it. Yet none of the occurrences of the sigmatic in-
dicatives points in other directions, and we must conclude that these
forms mark future anteriority.
If this is so, how do the sigmatic futures differ in usage from the
future perfects? Conceivably, the distinction is not one of meaning,
but of register. In order to see if this is possible, we ought to check
how the forms are distributed over cantica and senarii in Plautus, on
be static, as in ne is hercle ab istā nōn pedem discēdāt sı̄ licēssı̄t (Asin. 603) ‘really, he
wouldn’t go one foot away from her if it were allowed.’
15
Hodiē in the same passage may be temporal or may express anger, see Don. Ter.
Ad. 660.
The Sigmatic Future 177
whom we shall concentrate for the moment, even though this in itself
is insufficient.16 Plautine cantica contain more elements of elevated
language than the senarii. Put differently, the higher style in cantica
is marked by the addition of special figures of speech or moribund
forms. The use of archaic forms (like the sigmatic futures) is, however,
not obligatory, as is for instance proved by the occurrence of sigmatic
future forms which are co-ordinated with ‘regular’ future perfects.
The ratio of cantica to senarii is 3 : 1, and stylistically unmarked forms
should have a corresponding distribution. The two tokens in relative
clauses are in cantica, and of the three tokens in temporal clauses the
one that is not in a canticum is in a parody of a law (Persa 70), where
higher register is appropriate.
At first sight, the situation seems to be different in conditional
clauses: of the forty-eight tokens in Plautus, thirty-four occur in
cantica and fourteen in senarii, which is the distribution we should
expect for stylistically unmarked forms.17 However, when we look
at these fourteen tokens more closely, we can see that the criterion
of distribution alone does not establish much because most of the
forms are found in contexts where Plautus normally uses elevated
language:
1. invocations of gods (Bacch. 848, Rud. 1348); compare:
(6) (A soldier is pretending to be angry.)
Nam neque Bellōna mi umquam neque Mārs crēduāt,
ni illum exanimālem faxō, sı̄ conuēnerō,
nı̄ue exhērēdem fēcerō uı̄tae suae. (Bacch. 847–9.)
For never shall Bellona or Mars trust me any more, if I shall not have
done him to death, if I’ve found him, or if I shall not have disinherited
him of his life.
2. legal language (Asin. 770, 794, Rud. 81118 ) and threats, which
are often phrased like legal language (Capt. 124, 695);
3. asides by someone overhearing a conversation (Happ 1967: 182)
and moralizing soliloquies (Asin. 818, Stich. 192).
16
See Ch. 1 for a general outline of metre and register.
17
Only five of the tokens in cantica are in sung passages, the others are in recited
long verses.
18
See Marx (1928: 160 ad loc.).
178 Synchronic Analysis
Five tokens remain. All of them are in passages where high register
is not unusual. The form in Persa 393 for example is prefaced by the
bombastic statement pol deum uirtūte dı̄cam et maiōrum meum, ‘by
Pollux, I shall speak by the excellence of the gods and of my ancestors.’
Respexis in Aul. 58 is used by Euclio, who has a predilection for archaic
forms, as we shall also see in Ch. 7. Agorastocles in Poen. 428 says
si istuc lepide effexis just before he attempts a grand speech with an
adynaton. In Pseud. 533, sed sı̄ nōn faxis follows a hyperbolical com-
parison with King Agathocles. And Epid. 441 is spoken by Periphanes,
a boastful character who used to be a mı̄les glōriōsus.
There cannot be any certainty that all the passages listed above
belong to a higher style, but the assumption must be that most of
them do. If so, we can reasonably conclude that the sigmatic indicative
forms in subordinate clauses belong to a higher register, and the
same will be true of the one token in Terence. The other sigmatic
forms in subordinate clauses in Archaic Latin do not contradict this
finding. Of the thirty-three tokens, twenty-five are in laws, where
formal language is normal.19 Higher register is also appropriate in
the remaining instances: three of them occur in tragedies by Pacuvius
and Accius, one is in Ennius’ Annals, and three are in a dēuōtiō cited
by Macrobius:
(7) Sı̄ haec ita faxitis ut egō sciam sentiam intellegamque, tunc quisquis
uōtum hoc faxit ubiubi faxit rēctē factum estō ouibus ātrı̄s tribus. Tellūs
māter tēque Iuppiter optestōr. (Macr. Sat. 3. 9. 11.)
If ye shall so have done this that I shall know, feel, and understand, then this
offering, whoever shall have made it, wherever he shall have made it, shall
have been made correctly with three black sheep. I invoke thee, Mother Earth,
and thee, Jupiter.
Note: I include sı̄ue and other alternatives under sı̄, and also nı̄ue and similar expressions
under nisı̄. In the table above I have excluded the 156 regular future perfects in main clauses,
the four regular future perfects dependent on faxō, and one token that is in an indirect
question.
Terence? In relative and temporal clauses they are indeed so rare that
they can hardly be called productive in Plautus, and in Terence they
do not occur at all in these contexts. However, in conditional clauses
there are forty-eight tokens in Plautus (as opposed to the one form
in Terence). This is not a negligible number; can we therefore call the
sigmatic indicatives productive in Plautine conditional clauses? Since
they are not found in set phrases, and since the forty-eight tokens
belong to twenty-seven different verbs rather than just three or four,
the sigmatic future is unlikely to be fossilized. We may qualify this
assertion by comparing the figures for the sigmatic indicatives with
those for the future perfects, which, as we have seen, have the same
meaning, but a different register (Table 6.3).
There are 53 sigmatic futures and 354 regular future perfects in
Plautus’ subordinate clauses, which make up 13.02% and 86.98%
of the total, respectively. This ratio does not recur among all clause
types: relative and temporal clauses with sigmatic forms are dispro-
portionately rare, while conditional clauses with sı̄ and nisı̄ have more
sigmatic forms than the average 13.02%.20 Thus the sigmatic fu-
tures have a certain productivity in conditional clauses with a higher
register in Plautus. In Terence, on the other hand, the formation is
fossilized. The regular future perfects in Terence are more or less
distributed as in Plautus, except that they are disproportionately rare
in nisı̄-clauses; there are 41 tokens in main clauses, 51 in conditional
20
The likelihood that this finding is statistically significant is higher than 99.95%,
as a t-test shows.
180 Synchronic Analysis
clauses with sı̄ and the like, only 2 in conditional clauses with nisı̄,
13 in relative clauses, and 32 in temporal clauses. But while we have
98 regular future perfects in subordinate clauses, there is only one
sigmatic token in Terence, faxis, which again occurs in a sı̄-clause.
The only form that occurs in main clauses is faxō: there are forty-
one instances in Plautine cantica, twenty-nine in his senarii, nine
in Terence, and one in Afran. com. 67. They are found in different
constructions, for example in the type faxō hı̄c aderit, which could
provisionally be translated as ‘I shall bring it about that he shall be
here’. Consequently this is a causative construction.21
Exx. 8 and 9 are typical:
(8) (A slave wants his master to be patient; he will reveal everything a bit
later.)
Cētera haec posterius faxō scı̄bis ubi erit ōtium. (Epid. 656.)
I shall take care that you will know the other things later when we are free.
(9) Immo ollı̄ mı̄tem faxō faciant fūstibus. (Afran. com. 67.)
I shall see to it that they make him soft with cudgels.
The verb with which faxō is construed can be (i) a simple future as in
Ex. 8 (forty-eight times in Plautus, five times in Terence), (ii) a present
subjunctive as in Ex. 9 (fourteen times in Plautus, twice in Terence,
once in Afranius), (iii) a future perfect (five times in Plautus), and
(iv) a perfect subjunctive (once in Plautus). In addition, there is one
instance of faxō ut (Asin. 897), and there is also faxō with a double
accusative (once in Plautus and twice in Terence).
If we compare these constructions with those of Classical Latin,
one obvious contrast is in the use of ut: faciō ut + subjunctive is
the normal causative formation of Classical Latin, while ut is almost
always missing here. In Archaic Latin, however, we have many se-
quences of verbal forms followed by subjunctives without ut. Partic-
ularly frequent is the type fac + subjunctive, where the imperative fac
21
A general typology of causative constructions can be found in Dixon (2000).
The Sigmatic Future 181
alternates with fac ut; less frequent, but attested is the future faciam +
subjunctive,22 which is found alongside the ordinary faciam ut + sub-
junctive. Thus in the constructions of faxō we must assume that the
forms without ut are the original ones, while the one example with ut
is an innovation based on faciam ut. What is a striking phenomenon
is not the absence of ut, but the presence of a future or future perfect
in the slot otherwise occupied by the subjunctive. This could perhaps
be explained if faxō began as a parenthetic causative construction.23
Are there any semantic differences between the constructions? As
far as I can see, faxō + simple future and faxō + present subjunctive
are not distinct from each other. They occur in the same contexts, and
there are no restrictions as to what verbs can be governed by faxō. We
find both faxō scı̄bis and faxō sciās (‘I shall make you know’). In other
words, faxō can be combined with stative verbs. However, we also
encounter both faxō dı̄cēs and faxō dı̄cāt (‘I shall make you/him say’).
Thus faxō can be combined with telic events as well.24
Just as faxō sciās means ‘I shall bring it about that you know’, the
only instance of faxō + perfect subjunctive is of the type ‘I shall bring
it about that you have done’:
(10) (A hanger-on is angry because he did not get the lunch with Menaech-
mus which he had hoped for.)
Faxo haud inultus prandium comēderı̄s. (Men. 521.)
I shall take care that you have not eaten the lunch unpunished.
22
The two instances of a 1st pers. sg. future in Plautus and Terence are in Amph. 63
(faciam sı̄t) and in Amph. 876 (faciam . . . fı̄āt). The form faciam with a subordinator
occurs 38 times in Plautus.
23
The status of the clauses in the future tense that are combined with faxō is
unclear. Are they main or subordinate clauses? For other types that could be main
or subordinate clauses see Pfister (1995).
24
For the concept of telicity see Ch. 5.
25
Future perfects stand for events anterior to another future event, but this ante-
riority does not extend as far as a time before the moment of speech.
182 Synchronic Analysis
Pseudolus uses the future perfect not to indicate that he will be doing
something, but rather that he will have done it and that the results
will then be obvious to everyone.
In the other four instances, the future perfect is employed to assess
the consequences of an event and does not introduce a new action.29
In the English translation, we can always add ‘thereby’ to make it clear
that no new event is introduced:
(12) (Megaronides thinks that his wife is worse than Callicles’.)
Vı̄n commūtēmus, tuam egō dūcam et tū meam?
Faxo hau tantillum dederis uerbōrum mihı̄. (Trin. 59–60.)
Do you want us to swap, that I take your wife and you mine? I shall (thereby)
bring it about that you shall not have fooled me at all.
The future perfect dederis does not introduce any new action. The
clause is an assessment of the consequences of exchanging wives.
26
Similarly, the evaluation nēquı̄quam cannot be left out in Bacch. 701 (ēmungam
hercle hominem probe hodiē, ne id nēquı̄quam dı̄xerı̄t, ‘I’ll trick the chap properly today
so that he hasn’t said it for nothing’). Cf. also nihilo hercle e ā causā magis faciētis ut egō
hodie apstulerim pallam et spintēr, ‘you won’t bring it about any more for that reason
that I took away the mantle and bracelet today’, in Men. 1060–1, where nihilō . . . magis
is obligatory.
27
For this value see Ch. 2.
28
Hoc < ∗ hod-ke normally scans hocc, but here we have iambic shortening. Iam
ego hoc counts as two light syllables.
29
For this value of the future perfect see my discussion in Ch. 2, Risselada (2000),
and also Gaffiot (1933).
The Sigmatic Future 183
causative verb with himself as the subject. For similar reasons, faxō
is probably causative in Asin. 132, Poen. 1228, Truc. 761, and Ad.
847; the speakers are emphasizing that they will make something
happen.
If faxō as a marker of certainty is not verbal any more because
it has been reanalysed in some contexts, we can look at syntactic
criteria as to where faxō has to be verbal and thus causative: in faxō
ut, faxō is verbal because otherwise there could be no subordinator
ut. And when there is egō faxō instead of mere faxō, this must be
verbal; otherwise it would make no sense to express a subject egō.
Egō is used eighteen times. Similarly, when faxō is combined with a
double accusative it has to be verbal, or the clause would have no
verb at all. But if faxō as a marker of certainty is still verbal, only
faxō + double accusative has to be causative from a syntactic point of
view.
What is the temporal meaning of faxō? Again, temporal adverbials
do not tell us much because most of them are very general in mean-
ing, and because in a phrase like iam faxo hı̄c aderit (Phorm. 308) it is
not clear whether iam modifies faxō or aderit.35
I used ‘I shall bring it about that’ as a provisional translation for
faxō. In subordinate clauses, the type faxō can always be interpreted as
anterior; in main clauses, no such interpretation is possible, compare
Ex. 9 above. In this example, the speaker simply refers to the future; it
is impossible to detect any anteriority. The example can be translated
perfectively, but not imperfectively: ‘I’ll be in the process of making
them make’ does not make any sense. However, this perfective inter-
pretation can simply be the result of the kinds of speech acts we are
dealing with: faxō is used in promises and threats, and one normally
promises or threatens to do things completely, from beginning to
end.
We learn more from the tenses with which faxō is co-ordinated.36
In Ex. 16 it is the simple future:
35
The most frequent adverbial is iam (25 instances), followed by hodiē (three
instances). There are only two examples where the adverbial must belong to faxō:
Rud. 800 and Haut. 341. In both cases faxō takes a double acc. rather than a finite
verb.
36
Only those cases in which faxō has to be verbal are considered.
186 Synchronic Analysis
Here faxō is co-ordinated with two simple futures, ı̄bō and perdam.
Faxō in Ad. 847 is co-ordinated with a future perfect uı̄derō (l. 845),
but as I pointed out in Ch. 2, uı̄derō is a fossilized idiom that does
not indicate anteriority;37 besides, faxō is parallel to the following
faciam (l. 848). In Ex. 17 it is not clear whether there is co-ordination
between faxō and faciet:
(17) (Milphio has to pacify Adelphasium.)
Ego faxō, sı̄ nōn ı̄rāta’s, ninnium, prō tē dabit
ac tē faciēt ut siēs cı̄uis Attica atque lı̄bera. (Poen. 371–2.)
I shall take care that, if you are not angry, he will pay for you, little horse,
and he will take care/and that he will take care that you are a free Athenian
citizen.38
Plautus 40 26
Terence 7 1
Note: Under faciam ut I also include the types nōn faciam quı̄n (Mil. 283–4) and
ūnum hoc faciam ut (Aul. 365); these all have object clauses. I exclude the adverbial
ut-clauses in Capt. 385–7 and Persa 662. Under faciam + double acc. I include the
type te ego faciam . . . proinde ac meritus es, but I exclude the type ex mē . . . ūnam
faciam litteram longam (Aul. 77–8; longam is textually problematic).
188 Synchronic Analysis
the type fac or facitō + subjunctive (fac habeant etc.) is very frequent
in Plautus; its function would partly overlap with that of a causative
second person faxis + subjunctive, so that the latter construction
would be superfluous.
CONCLUSIONS