How is the gerundive construction to be analyzed in the following example?

Cariotae cum ficis certandum habent. (Plin. Ep. 1,8)

'Dates have to fight with figs'.

Could you please provide me with some similar examples of "(in)transitive" HAVE constructions in Latin? Here I write "(in)transitive" because there is no explicit NP object. NB: these "(in)transitive" examples are different from the typical (and, indeed, expected!) ones, namely, the clearly transitive ones where there is an explicit object: e.g., see the following example from Joonas's answer below, which is also included in a previous related question (How many types of so-called “predicative Gerundives” can be differentiated in Latin? ):

Ibi agrum de nostro patre colendum habebat. (Ter. Phorm. 364-365)

‘He had there a piece of land from my father to farm’.

Importantly, as pointed out below by Joonas, these transitive examples of predicative gerundive are by far more frequent than the "(in)transitive" ones at issue here.

As is well-known, there is an important parallelism between transitive habere + accusative and intransitive esse + dative constructions, which holds in different domains of Latin grammar. E.g., cf. Caesar equitatum coactum habebat (Caes. Bell. Gall. 1,15) with Caesari equitatus coactus erat.

According to the abovementioned parallelism, the nominative subject cariotae can also be converted into the so-called "dativus auctoris" or "dative of agent": cariotis.

Cariotis cum ficis certandum est.

This second construction is clearly impersonal, i.e., with no personal subject (cf. the typical impersonal passive construction Pugnatum est acriter; see also the previous post for the impersonal construction Nunc est bibendum: Nunc est bibendum: gerund or gerundive? ) but what about the object of the allegedly transitive verb habent in the first example above from Pliny? Assuming that certandum is gerundive (and hence a verbal adjective), how should one analyze the (null) object of transitive habent in Pliny's example? Basically, we do have "null expletive subjects" in Latin (e.g., cf. Lat. Pluit and Engl. *(It) is raining or cf. Lat. Pugnatum est acriter and Engl. *(It) was fought fiercely Vid. http://www.glottopedia.org/index.php/Expletive_noun_phrase // https://en.wikipedia.org/wiki/Dummy_pronoun ) but do we have "null expletive objects" in Latin as well? How is the grammatical construction certandum habent above to be analyzed? I know what Pliny's example means but I do not know how to analyze it properly. My preliminary analysis would be that in this example there is a null expletive object pronoun involved (which would be the counterpart of the null expletive subject pronoun in the impersonal construction above with certandum est) and the verbal adjective certandum is then predicated of this null object pronoun in Pliny's example. But I'm not entirely convinced by this analysis because positing a null expletive object pronoun here can be said to be too stipulative: after all, in which other constructions should one posit such a null expletive object pronoun? In striking contrast to that, notice that the analysis of Caesar equitatum coactum habebat is indeed very transparent: i.e., the passive participle coactum is predicatively applied to the direct object of the transitive verb habebat, namely, equitatum. Exactly the same with Terence's example above.

NB: the Latin construction at issue developed into a grammaticalized verbal periphrasis of obligation, which is, for example, formed in Catalan with 'haver' de + infinitive (cf. also Engl. above: Dates have to fight with figs). Given this, one could simply claim that in Latin the very same happens (i.e., the construction in Latin is fully grammaticalized like in Romance and English) but I don't think so. My intuition is that the verb habere in the Latin construction above is not as (strongly) grammaticalized as in Romance and English. So a different analysis is involved.

  • 1
    I updated my answer with a reading of this construction. I would be curious to see what you think of it.
    – Joonas Ilmavirta
    Jul 20, 2019 at 17:46
  • I fully agree with you. Yes, possession is the BASIC semantic notion involved in habere + accusative constructions and esse + dative constructions. There are other CONTEXTUAL (i.e., non-basic or derived) notions involved in the latter constructions (e.g., "dative of agent", "dative of relation" or "dativus iudicantis", etc.) but the central or prototypical one is possession: e.g., as for "dative of agent", see your answer and my question. As for dativus iudicantis, one can also establish a parallelism between Quintia formosa est mihi and Habeo Quintiam formosam.
    – Mitomino
    Jul 20, 2019 at 20:51
  • Still, the structural, i.e., syntactic, problem I have with currendum habeo remains. In principle, an allegedly transitive verb like habere would involve an object. Assuming that currendum is a gerundive form (i.e., a verbal adjective), one would not expect currendum habeo to be well-formed for the same reason that one would not expect Habeo formosam to be well-formed without any direct object. For an admittedly controversial explanation, see my preliminary analysis above: there is a null expletive object that the gerundive is predicatively applied to.
    – Mitomino
    Jul 20, 2019 at 21:02
  • An alternative proposal is to consider currendum as a gerund (hence a substantive) in currendum habeo. That option would indeed satisfy the object requirement of a transitive verb like habere but would involve other problems. Ditto for currendum est mihi, where currendum could also be considered a gerund in...nominative case? No way! I would then stick to the gerundive analysis in both transitive (currendum habeo) and intransitive (currendum est mihi) constructions and would pay the price of positing a null expletive noun (marked with accusative and nominative, respectively).
    – Mitomino
    Jul 20, 2019 at 21:37

2 Answers 2


Here is one possible way to parse this kind of constructions. (The instance you quote is not unique as the examples below indicate.) Recall that there are two ways to indicate possession: domum habeo and mihi domus est are more or less equivalent. The usual way to express obligation is mihi currendum est. If one analyses this as a dative of possession, then the other way to express possession opens up: currendum habeo.

Having thought about the matter for a while, this is indeed how I would interpret this: This construction arises by analogy to the usual obligation with dative and gerundive and the possibility to express possession with either dative or habere. Just like the usual obligation with dative can have or not have an object, the option with habere also has this freedom. That is, mihi tempus est > tempus habeo is like mihi currendum est > currendum habeo. The semantics of the gerundive work the same way whether it's esse or habere.

Although I imagine the dative of obligation is far more common, there are good reasons to occasionally prefer habere instead. The syntax is different, and so the pronouns can align more naturally with the surrounding sentence. After all, a form of habere indicates the person (to some extent at least), whereas with esse you would need to add a separate pronoun to have such indication. And if there are other datives around already, using habere can reduce ambiguity.

Below are more examples of similar uses of habere. Similar constructions are found in a number of places, but I had never seen them before. The use is often transitive, but there are also intransitive attestations. Here are some picks from a corpus search:

  1. Terentius, Phormio 361–

    nam iam adulescenti nil est quod suscenseam,
    sĭ ĭllum minu' norat; quippe homo iam grandior,
    pauper, quoĭ opera vita erat, ruri fere
    se continebat; ibi agrum de nostro patre
    colendum habebat.

    The presence of agrum makes this different from your Pliny example, but I still find this a worthwhile instance for comparison.

  2. Caesar, Bellum Civile 3.57, 3.84

    ut praeter auctoritatem vires quoque ad coercendum haberet

    quidnam Pompeius propositi aut voluntatis ad dimicandum haberet

    These are crucially different due to the preposition ad with the gerund. This kind of construction is quite common, but I would consider it a distraction and will not list more.

  3. Columella, De Re Rustica 1.4, 2.9, 2.14, 12.3

    in uniuersum tamen quasi testificandum atque saepius praedicandum habeo, quod…

    Illud deinceps praecipiendum habeo, ut demessis segetibus iam in area futuro semini consulamus.

    illud quoque praecipiendum habeo, stercus omne, quod…

    Illud uero etiam in perpetuum custodiendum habebit

    This piece of literature seems to be full of this construction, at least in comparison to most other books.

  4. Pomponius Mela, De Chorographia 2.29

    sed ea magis id memorandum habet, quod…

    By this point it is clear that this construction is pretty widespread. Not all authors make use of it, but many do. However, I found far fewer examples of a gerund(ive)1 with habere without an object. The remaining examples shall be those.

  5. Plinius, Naturalis Historia 10.154, 31.35

    Caesare ex Nerone gravida, cum parere virilem sexum admodum cuperet, hoc usa est puellari augurio, ovum in sinu fovendo atque, cum deponendum haberet, nutrici per sinum tradendo, ne intermitteretur tepor; nec falso augurata proditur.

    Item confitendum habent nec statim amnium utilissimas esse, sicuti nec torrentium ullius, lacusque plurimos salubres.

    It appears that Pliny is more fond of this construction than many others.

  6. Quintilianus, Declamationes Minores 368

    Cum patre nimirum mendicandum habet.

    This is very similar to your example.

  7. Seneca, Controversiae 2.7, 10.2

    Post tantos inpudicitiae quaestus si tacere possum, confitendum habeo hac me causa afuisse, ut in accessionem patrimoni peregrinando cum uxore certarem

    Pugnandum habebam non imperatori tantum sed patri.

    In the first example the ACI me afuisse is a kind of an object, but a different kind than encountered previously. The second one is a very clear case with no object. It also has plenty of other datives, making it reasonable to use this construction over the usual dative of obligation as I mentioned before the examples.

  8. Frontinus, Strategemata 1.4

    Idem, tenentibus angustias Thebanis, per quas transeundum habebat

1 Regarding gerundives and gerunds, you may want to take a look at When and how did the distinction between the gerund and the gerundive develop?, Is the nominative gerund attested?, and Nunc est bibendum: gerund or gerundive?.

  • Many thanks, Joonas! Your examples are very useful, indeed! As for your VERY important remark "I found far fewer examples of a gerund(ive) with habere without an object", this is what is expected since the gerundive, as a verbal adjective, must have an object to be predicated of. The interesting point is that these examples without object already existed in Classical Latin, even in Quintilian! By the way, I see you consider the possibility that in these rare cases without object the -nd- form could be a gerund (cf. your parentheses in "gerund(ive)"). Very interesting, indeed!
    – Mitomino
    Jul 19, 2019 at 19:37
  • 1
    Tony, as for the "gerundive plus habere", the transitive construction (e.g., ibi agrum de nostro patre colendum habebat) is not surprising. For the syntactic reasons mentioned above, what is more intriguing is the intransitive one (e.g., pugnandum habebam). It could be the case that the latter was more typical of colloquial Latin, in contrast to mihi pugnandum erat. In any case, notice that habere + accusative and esse + dative are related constructions in many languages. In fact, some linguists derive HAVE from "BE + Preposition". E.g. doi.org/10.1016/j.lingua.2011.10.009
    – Mitomino
    Jul 23, 2019 at 23:25
  • @Mitomino: Would all of the expressions work without "habere"? "There is a field, from our father, it-ought-to-be-cultivated."; "Dates, with figs, it-ought-to-be-fought.": these do not require a form of "habere". It is necessary in "...ut praeter auctoritatem vires quoque ad coercendum haberet." (with the result that, beyond his authority, he would also have the power for the purpose of restraining.) because it is an "ut" clause requiring a subjunctive. The use of "habere + acc." is understood if "habere" is necessary; predicative dative e.g. auxilio esse = to be helpful. In "pugnandum habebam
    – tony
    Jul 25, 2019 at 9:10
  • @Joonas llmavirta: Apart from the "ut" clause, would the expressions in your answer work without a form of "habere"?
    – tony
    Jul 25, 2019 at 9:16
  • @tony I think a form of habere or esse is needed, but esse can sometimes be left implicit. I wouldn't leave it out here, though, so I'd say you can't drop the habere. The way I see it, this is simply a construction by analogy: mihi tempus est > tempus habeo is like mihi currendum est > currendum habeo. The semantics of the gerundive work the same way whether it's esse or habere. At least I see nothing more in it, but separate follow-up questions are always welcome.
    – Joonas Ilmavirta
    Jul 25, 2019 at 15:04

A summary, before the comments are deleted: the gerundive plus habere (referred to, here, as "habere plus accusative" was a little-used, colloquial equivalent of the better-known, more widely used "gerundive-of-obligation" with a form of esse and the person, upon whom the obligation falls, is put into the dative; this construction, referred to here as "esse plus dative". These two constructions are related: consider: mihi tempus est = for me there is time; equivalent to: tempus habeo = I have time.

Analogously: mihi currendum est = it-ought-to-be-run by me; equivalent to: currendum habeo = it-ought-to-be-run I-have/ must (I-am-obliged)--this lacks credibility, for me, but appears to be the linguistic reality that is.

Contrasting the intransitive: "pugnandum habebam non imperatori tantum sed patri" (it-ought-to-be-fought not only for the Emperor but for the father) with "pugnandum erat mihi non imperatori tantum sed patri" the point made that the latter involves too many datives.

This point is valid because, with awkward constructs e.g. gerundive genitives, established rules are relaxed; therefore, the gerundive is allowed to govern a direct object, in order to avoid clumsy-sounding jingles ending -orum -orum or -aram -aram e.g. "ars oppidorum oppugnandorum" becomes "ars oppugnandi oppida".

This summary distilled from the work of Mitomino, Joonas and (my) questions in attempting to get to the bottom of things.

  • 1
    I have actually updated my answer with quite a few of the insights arising from the comments along the way.
    – Joonas Ilmavirta
    Jul 27, 2019 at 10:52
  • @Joonas llmavirta: With Mitomino's permission we may put this one to bed? Did you read "Nunc Est Bibendum"? Sweated blood on that one; but, then we're supposed to!
    – tony
    Jul 27, 2019 at 13:28
  • I did indeed read that, and I'd be happy to close this case. The call is Mitomino's; marking an answer as accepted is what ends it definitively. (Or the question can be edited to tell what's missing, but often asking a more specific follow-up question works betters.) I never intended the discussion to meander as long as it did; a separate follow-up question is often (but not always) a better idea after the fifth comment).
    – Joonas Ilmavirta
    Jul 27, 2019 at 13:47
  • Thanks, tony, for your useful summary. Thanks, Joonas, for your helpful revision of your answer. I've just marked it as accepted since it addresses the important connection between habere + accusative constructions and esse + dative constructions (the important link being possession) by providing a lot of useful examples. Although I've decided to mark Joonas's answer as accepted, notice that the very specific question formulated in the title remains to be solved: namely, are there null expletive objects in Latin?
    – Mitomino
    Jul 27, 2019 at 18:45
  • @Mitomino I'm glad to be able help and it was interesting to learn about this construction myself. Having taken this step forward, you can always ask a narrower new question on whether such objects exist.
    – Joonas Ilmavirta
    Jul 27, 2019 at 18:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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