Three verbs are well known to have an irregular short imperative: fac, dic, duc. Do the imperatives remain short in the presence of a prefix? For example, which ones are correct out of effic/effac/effice, interdic/interdice, adduc/adduce? My impression is that they are almost always in their usual long form if there is a prefix, but there are some exceptions. For some reason I feel that I should undo vowel gradation if I use a short imperative; effac sounds better than effic.

Is there a rule or guideline for the imperatives of facere, dicere and ducere with prefixes?


To supplement Tom Cotton's answer—

There's one other verb which similarly shortens its imperative: ferō, ferre, tulī, lātus, imperative fer, ferte. Compounds always use the shortened/apocopated form (adfer/adferte, offer/offerte etc), with *adfere / *adferete, *offere / *offerete etc entirely absent from the PHI corpus.

So these two seem relatively certain: compounds of facere have regular imperatives (as Tom explains), while compounds of ferre have shortened ones. Dīce and dūce are less clear-cut.

For compounds of dūcere, the regular/long imperatives are almost exclusively found in Plautus, where they actually appear more frequently than the shortened forms: the PHI corpus (when restricted to Plautus) lists 7 occurrences of addūce and 7 of abdūce, but only one of addūc and three of abdūc.

In later authors, compounds using dūce are much less common: ten instances of prodūc compared to zero of prodūce. (Both addūc and abdūc become rare, though still more common than addūce and abdūce.)

For dīcere, imperatives in compounds are extremely rare in general. Aeneid 11.463 contains one notable instance of ēdīce, which is marked as unusual in Honoratus's Commentary:

'edice' autem plenus est imperativus; nam ab omnium coniugationum infinito detracta 're' ultima syllaba fit imperativus, ut 'amare ama', 'docere doce', 'legere lege', 'audire audi'. cum autem 'fac' vel 'dic' dicimus, apocopen verba patiuntur. ex 'face' 'dice' conposuit 'edice'. expressa autem festinantis oratio, quia Turnum non praedixit locuturum.


ēdīce however is a full imperative; for the imperative of every conjugation is formed by removing the final syllable -re from the infinitive, thus amāre - amā, legere - lege, audīre - audī. When, however, we say fac or dīc, the verbs undergo apocope. Ēdīce is composed from face or dīce. However, this is the rushed speech of someone in a hurry, because he did not mention Turnus being ready to speak.

Plautus even uses dīce and dūce [as imperative of dūcere, not ablative singular of dŭx] on occasion, which I can't find anywhere else, though he still prefers dīc and dūc: dīc appears 121 times, dīce 12. So the regularized -e imperatives may have been colloquial or dialectal forms which were dying out in Plautus's time, and disappeared almost entirely by the Classical era.

  • Thank you for reminding me of fero, which I'd quite overlooked. I wonder how many other, similar cases there are!
    – Tom Cotton
    Nov 1 '16 at 15:51
  • 1
    Could you please expand your answer? What was the syntax of your queries? What numbers did you get? And, most importantly, how did you analyze the data? Linguistics starts with manual, careful analysis of data obtained from corpus research.
    – Alex B.
    Nov 1 '16 at 16:55
  • 1
    Expanded. Hopefully this is more helpful.
    – Draconis
    Nov 1 '16 at 18:34
  • 1
    Thanks, this is great! One small comment: The noun dux has a short U in all forms, so dŭce (ablative) and dūce (imperative) are distinguishable if quantity is known.
    – Joonas Ilmavirta
    Nov 2 '16 at 8:53
  • @JoonasIlmavirta Ah, how I wish now for a corpus with macrons! That would be a much easier way to distinguish the two; my strategy here was eyeballing the context to see if it was verb-like or noun-like. Thanks!
    – Draconis
    Nov 2 '16 at 16:10

Generally, verbs of mixed 3rd/4th conjugation and their compounds (inf. -ere, 1st. pres. sing. in -io) all follow the same pattern, which includes compounds of facio, but not facio itself — which means that effice, etc. are the correct forms. Fac is a peculiarity.

Unlike those of facio, the compounds of dico and duco mainly follow the pattern of the simple verbs, including the abbreviated imperatives. There are one or two exceptions, such as the occurrence of adduce for adduc and dice for dic, though my impression is that they are most often pre-classical. The compounds usually fall into the 3rd conjugation, but there may be exceptions: indico, for example, is a completely regular 1st conjugation verb with the imperatives indica, indicato, and so on.

  • You say this as if it was instinctive, which of course it is after a time. But did you have to look it up at all?
    – Hugh
    Nov 1 '16 at 15:23
  • And speaking meta, this is a really useful set of misc information to have all in one place. Ideal SE.
    – Hugh
    Nov 1 '16 at 15:27
  • 1
    @Hugh I'm not sure that 'instinctive' is the right word, but I simply wrote it as I remember after years of translating from and into Latin.
    – Tom Cotton
    Nov 1 '16 at 15:48

As Weiss 2009/2011 mentions, word-final –e is “normally retained” (p. 147; emphasis mine – Alex B.). That being said, there are some few cases when word-final –e was lost (or apocopated).

In your question, you ask us specifically about imperatives from compound verbs of dico, duco, and facio. In my answer below I will address this issue only. If you are interested in imperatives of simplex dico, duco, facio, and fere; how to interpret data from Archaic Latin (Plautus and other authors) – it’s not that simple as Draconis has been trying to convince us here – I am willing to write a separate answer, just post another question.

Now, revenons à nos moutons.

First, facio. Based on whether the root vowel is changed or not, the derivatives of facio can be grouped into two classes:

Type A. calefacio [root vowel preserved], compound verbs (two roots).

Type B. conficio [root vowel weakened], prefixed verbs.

However, the 2nd singular imperative of both types is formed the same way: calface! confice! areface!

Secondly, dico. Weiss argues that compound 2nd singular imperatives from dico are not attested in Classical Latin.

Finally, duco and fere. Apocopated forms are standard in Classical Latin:

confer! infer!

Apocopated means that originally they had –e at the end (in fact, such forms are attested) but eventually it was lost. Note that there are no root vowel changes in these compounds. Another strong evidence in support of this analysis comes from accentology. Such apocopated singular imperatives are oxytones, i.e. the last syllable is stressed:


The most interesting question that has not been addressed in the other answers – even though you clearly ask us – is how we know when the root vowel changes or not.

Weiss discusses it on pp. 121-122 (footnote 25, citing Garrett 2005 and Pultrova 2006).

Back root vowel [no change in compounds]: duco – adduco, educo etc. and, respectively, adduc, educ etc.

More diverse morphology [vowel weakening, a changes to i]: facio – conficio and its 2 sg. imperative confice.

To recap,

Imperatives from compounds of duco and fero: short (apocopated) forms, e.g. educ! effer!

Imperatives from compounds of dico: not attested in Classical Latin

Imperatives from compounds of facio: unapocopated forms, e.g. calface! confice!

  • 1
    Re archaic: PHI lists 121 instances of dīc in Plautus's work, compared with 12 of dīce. So the apocopated form was already preferred for the uncompounded verbs.
    – Draconis
    Nov 1 '16 at 16:28
  • @Draconis let's not get sidetracked here. The OP is about imperatives of compound verbs. But thanks, I'll look into this later.
    – Alex B.
    Nov 1 '16 at 16:35
  • 3
    Indeed. I just find it interesting (and relevant) that Plautus preferred -e for compounds but no -e for the base words, which might indicate how the apocopated imperatives evolved or show some sort of hypercorrection; unfortunately, I don't know enough about the historical context to do more than speculate on that front.
    – Draconis
    Nov 1 '16 at 16:42
  • @Draconis Sorry, cannot comment on that. Need time and more data. So far, I'll stick to "seem". Willing to make changes later.
    – Alex B.
    Nov 1 '16 at 16:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.