In Ancient Greek, verbs often take a "middle voice", neither active nor passive. The forms usually look identical to the passive on the surface, but can take direct objects and cannot take an agent (unlike true passives). Some verbs only ever occur in the middle voice, with no active forms at all!

To a Latin-speaker, this should sound somewhat familiar: verbs which look passive on the surface, but can take direct objects, can't take agents, and have no active forms? That fits the deponent verbs (sequor, hortor, arbitror, etc) perfectly. In other words, the deponent verbs seem to act like the Greek middle voice, rather than the Latin active or passive.

But in Ancient Greek, pretty much any verb can be put into the middle voice, not just the deponents.

Was this true in any attested stage of Latin? That is, do we ever see non-deponent verbs with passive morphology, but able to take accusative direct objects and not able to take ablative agents?

  • Does gaudeo count? Or does it have to have active and passive forms in the same tense?
    – C Monsour
    Commented Aug 5, 2019 at 19:31
  • @CMonsour The same tense would be ideal, but that's still worth noting!
    – Draconis
    Commented Aug 5, 2019 at 19:55
  • Some Latin verbs (e.g. doceo) that take 2 accusative objects in the active can retain one of those accusative objects in the passive (so-called 'retained accusatives'). But doceor can take an agent too, because it's passive, not middle, in meaning. For a middle meaning in Latin, passive and reflexive forms can be used (e.g. movetur/se movet, 'It's moving'); but to my knowledge, additional accusative direct objects are never used in these instances.
    – cnread
    Commented Aug 5, 2019 at 20:00
  • @cnread The passive voice of verbs with two accusative objects was studied in an earlier question. Some ideas from there might indeed be useful in this discussion.
    – Joonas Ilmavirta
    Commented Aug 5, 2019 at 20:06
  • As for your question, “do we ever see non-deponent verbs with passive morphology, but able to take accusative direct objects and not able to take ablative agents?”, what about exs. like neque mei neque te tui intus puditumst factis quae facis (Pl. Bacch. 3, 1, 12)? As for your grammatical conditions, (i) there is an accusative direct object (te), (ii) no ablative agent is possible here, and (iii) the verb has passive morphology but is not "prototypically" deponent as such, though it is true it can be considered as semi-deponent in competition with puduit (cf. also C Monsour's point).
    – Mitomino
    Commented Aug 5, 2019 at 21:35

4 Answers 4


Thinking about your very interesting question ("That is, do we ever see non-deponent verbs with passive morphology, but able to take accusative direct objects and not able to take ablative agents?"), here is an example that could be argued (at least, claimed!) to satisfy your grammatical conditions:

flava caput nectentur oliva (Verg. Aen. 5, 309)

'They shall have their head(s) crowned with yellow olive' Source of Engl. transl.

(i) there is an accusative direct object. NB: caput is often regarded as an "accusative of relation" (aka "Greek accusative", which, by the way, nicely relates to your present question) but HERE, in this use, it can also be claimed to occupy the direct internal argument slot of the verbal predicate. Accordingly, it can be claimed to be more properly regarded as a "retained accusative of specification" in the sense that it can be analyzed as the corresponding direct object of an active sense (caput nectent; please see this link for the notion of "retained accusative of specification").

(ii) no ablative agent is possible here due to its "middle" meaning, and

(iii) the verb has passive morphology but it is not a deponent verb (NB: nor a prototypical one nor a "peripheric" one like the semideponent use of pudet/puduit in my comment above).

  • Verb "necto" (not the most appropriate for "to crown") appears to be passive future of a nominative: the head(s) will be crowned. What's the difference between that and e.g. agricola videtur; pueri audiuntur; donum datur; agricola nectetur? Are you sure that "caput" is an accusative? Recall "vinum est Bibendum" (the wine: it-ought-to-be-drunk); some, with an instinctive desire to treat "vinum" as a direct object (accusative) and not a nominative, which it is.
    – tony
    Commented Aug 6, 2019 at 9:02
  • 2
    @tony: As for the English translation of nectentur, I agree with you this is not a literal one (since it is not mine I previously added the source link. In any case, please note that the literal translation is not always the "most appropriate" one). As for caput, yes, I'm sure that it is accusative. Notice that it can only be accusative (as nominative, it would be capita, in agreement with 3rd plural of nectentur).
    – Mitomino
    Commented Aug 6, 2019 at 17:45
  • Mitomino: Yes, in the English translation, plural "head(s)" was optional; should have written "caput/ capita". Could yourself, please, explain how caput/ capita nectentur differs from a nominative noun + passve-verb; thus, becoming an accusative + passive-verb? The "flava oliva" = "with yellow oil" is easy; but, having separated "flava oliva", all that remains is "caput (capita) nectentur". This, in grammatical terms, is the same as the examples e.g. agricola nectetur, I gave earlier.
    – tony
    Commented Aug 7, 2019 at 15:55
  • @tony: Perhaps it can be useful to point out that the nominative plural subject of nectentur is mentioned in the previous verse: tres primi ('the first three'). Since caput is singular accusative but the verb nectentur (3rd.pl) is passive -and passive verbs don't generally have direct objects-, a natural conclusion seems to be that caput is an "accusative of relation". This said, my point is that one can claim that HERE caput, the alleged accusative of relation, is in fact occupying the direct object position of the verbal predicate: cf. "retained accusative of specification".
    – Mitomino
    Commented Aug 7, 2019 at 17:00
  • Thank you. Put an attempt at understanding in Answers.
    – tony
    Commented Aug 9, 2019 at 13:31

The way I see it, there is no middle voice in Latin, but there can certainly be traces of it. The situation is similar to the dual number. That said, looking for such traces is interesting. Here is one candidate worth considering.

Dictionaries tend to list contemplare and contemplari as separate entries. However, if we analyze them as the same word, something resembling a middle voice arises. The passive forms are not passive in nature, and they can take accusative objects as mentioned in the linked L&S entry.

The slight difference in meaning sounds suitable for a middle reading. Namely, contemplo is roughly "I observe" whereas contemplor is more "I consider" which could be seen as "I observe for myself". For practical purposes the verbs are essentially synonymous, but one can contemplate whether the deponent verb arose from something middle. L&S mentions that the active forms are ante- and post-classical. Perhaps the passive forms took over and there was no need to use the active forms in parallel if the distinction was minor or non-existent.

Plautus appears to be the only writer to use both active and passive forms to any significant extent, so it is worthwhile to take a look at his 15 uses of contempl-. Upon a quick inspection, I get the feeling that the active forms are more "look" and the passive ones "think". This gives some support to the idea of a middle. (But I won't stretch the analysis too far with just these data points or claim that it is absolutely so.)

So, perhaps the deponent verb contemplari could be seen as the middle voice of contemplare. This is not the only verb which has normal and deponent versions. Perhaps analyzing more such pairs could reveal more solid evidence for middle, if these kinds of cases count as middle for you. I happened to be reminded of this particular pair in a recent question.

  • llmavirta: Does yourself agree with Mitomino's analysis of "flava caput nectentur oliva", above, or not?
    – tony
    Commented Aug 18, 2019 at 10:06
  • @tony I agree that there is indeed a trace of something like a middle there in a plausible reading. I don't know if there's anything in Latin that I'd call "truly middle", but similar phenomena do appear. The point of my answer was to add to the variety of middle-like constructions and let the folks decide what's most appropriate or interesting.
    – Joonas Ilmavirta
    Commented Aug 18, 2019 at 10:31
  • @tony: I think that my last example (see below: nuda genu nodoque sinus collecta fluentis) is in a sense even better as an answer to Draconis's question (hence my third try!). It's "better" in the sense that the accusative-of-relation interpretation of sinus fluentis is a bit contrived here (cf. the body part caput in the previous example (caput nectentur), where such an interpretation is in fact the "typical" one). If so, it seems that sinus fluentis is the true direct object of collecta, which behaves like a Greek middle participle.
    – Mitomino
    Commented Aug 19, 2019 at 1:40
  • @JoonasIlmavirta. After reading your very interesting answer on different meanings of contemplo and contemplor, I was wondering if for native speakers of Latin there was an (indeed, very subtle, if any!) semantic difference between, for example, the impersonal variants puduit and puditum est. Cf. my comment & example above (my first try!): neque mei neque te tui intus puditumst factis quae facis (Pl. Bacch. 3, 1, 12). After all, many functionalist linguists consider it, at least, as a default hypothesis that a difference in form involves a difference in meaning.
    – Mitomino
    Commented Aug 19, 2019 at 2:07
  • @Mitomino: Every time I read this, something new registers! In "...with bare knees having been gathered in a knot of flowing robes..."; isn't "sinus fluentis" a genitive phrase": of-the-flowing-robes? But the thing (flowing robes) referred to by the action (having-been-collected [collecta]) gives the "accusative-of-relation"--is that correct?
    – tony
    Commented Aug 20, 2019 at 9:39

Third time lucky! Here is then my third try in answering Draconis's intriguing question ("do we ever see non-deponent verbs with passive morphology, but able to take accusative direct objects and not able to take ablative agents?"). Consider the following example:

Namque umeris de more habilem suspenderat arcum
venatrix dederatque comam diffundere ventis,
nuda genu nodoque sinus collecta fluentis.

(Verg. Aen I, 318–320)

Notice how the "Draconian" conditions above are fulfilled:

(i) collecta has passive morphology but is not a deponent verb.

(ii) there is an accusative direct object (sinus fluentis), and

(iii) no ablative agent is possible here due to its "middle" sense.

It seems that there are two ways of analyzing the nominal phrase sinus fluentis: (A) it is the true direct object of collecta, which behaves like a Greek middle participle (see this philological comment on verse 320 // see also Allen and Greenough: 247) and (B) it is an "accusative of relation" (see second alternative on verse 320), perhaps in nice symmetry with genu in the previous Adjectival Phrase nuda genu. See also the very interesting comment on verse 320 made by the eminent classical scholar Robert Seymour Conway.

In case you think (as I do, not without some hesitation) that HERE the two abovementioned possibilities (A and B) can be conflated into only one, I can conflate the present answer into my previous one (for the time being, I'll leave them as two different (but related) possibilites of answering the present question). So far what seems clear (at least to me!) is that Draconis's very interesting question can only be answered with examples that, unsurprisingly, turn out to be clearly influenced by Greek.

  • This Q has certainly generated some labour! One (hopefully) final point, please. A middle-voice verb can take a direct-object/ accusative; but not an agent. Why not? If we can slay one sacred cow, why not another? Why can't Joonas's "tu rem doceris a me" qualify.
    – tony
    Commented Aug 11, 2019 at 11:32
  • Continuing: our favourite phrase could be: "flave caput nectentur oliva, a rege", could it not?
    – tony
    Commented Aug 11, 2019 at 11:46
  • The passivization of docere is a case different from the "middle" example we're discussing here. Notice, for example, that you can add an agent ablative phrase in Tu rem doceris a rege (cf. Rex docet te rem) but you can't do it in Tres primi flava oliva caput nectentur (*a rege). Remember that in the relevant middle example the (elliptical) nominative subject is tres primi ('the first three'), which is the agent and patient of the action (hence its middle sense). The incompatibility with a rege is then very clear.
    – Mitomino
    Commented Aug 11, 2019 at 15:27

As comments-section grows, this is not so much an answer as an interpretation of Mitomino's "flava caput nectentur oliva": the assertion that "caput" is an accusative; not a nominative.

Expressions like "accusative-of-relation" & "retained-accusative-of-specification" can be baffling to those who have not studied linguistics. Extra study is required: "accusative-of-relation" denotes the thing or person (caput) referred to by the action (nectentur) rather than the object proper.

Greek: a retained accusative is based on the supposition that such constructions are converted from active to passive verb-forms and that the accusative is retained from the active construction. At this point it may be appropriate to study Joonas's earlier Q. Here, Brianpck adapted "te rem doceo" = "I teach you something" to give "rem a me doceris" = "you (subject) are taught (passive) something (retained accusative) by me (agent). In the passive construction, the accusative-of-the-person becomes the subject; the accusative-of-the-thing is retained.

In the OP's original Q., a "middle-voice" verb is not to take an agent. If the person doing the crowning had been identified, that soul would have been the agent (ablative): "flava caput nectentur oliva, a puella"; presumably, the use of an (ablative) agent restricts the verb to passive-use only; therefore, disqualifying its elevation to the "middle-voice" genre?

This links into "caput" being a "direct internal argument" of "the verbal predicate". The internal argument of a verb has to be realised inside the maximal projection of that verb. The one closest to the verb (here, "caput") is the "direct internal argument"; the other/s--"indirect internal argument" (here, "flava oliva").

The argument-structure of English transitive verb, "open", has an EXternal argument (the Agent) and two internal arguments (theme & instrument) e.g. "Tom opened the door with his key", where "door" is the "direct internal argument" (closest to the verb, "open") & "with his key", the INdirect internal argument.

Hebrew: Psalm 3:8 "For you have stricken all of my enemies with respect to the cheek"; here, "cheek" is an indefinite primary-noun (no specific or named owner: "his", "John's") and functions as an "accusative-of-specification"--it specifies/ clarifies the verb--the striking is limited to the cheeks, of the enemies. Alternative translations identified "adverbial-accusatives" (just to add to the "accusative" confusion) "on the cheek" & "upon the cheekbone"; how the striking was done.

Of course, this is not a retained "accusative-of-specification"; but, it helps to explain arcane terminology.

Applying to the Latin: the crowning will be limited to the heads, suppose it would be, by definition! Hence, the thing or person (the heads) referred to by the action (the crowning).

The conclusion: "caput" in "flava caput nectentur oliva" is an accusative.

(Have not studied Greek, or Hebrew, but they seem to involve similar concepts, on this topic.)

  • Yes, my claim is that in this particular example caput is occupying the direct internal argument position since it can be understood as a "retained accusative of specification" (see the link in my answer). In fact, it is caput what turns out to be nexum flava oliva. Cf. also necte comam myrto (Ov. Am. 1, 2, 23).
    – Mitomino
    Commented Aug 10, 2019 at 17:02

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.