In Part 2 we made a few distinctions to help thinking about voice. Namely, noting that there is a difference between meaning and morphology in voice. Both are relevant to account for and they overlap in complicated ways, but they are not the same thing. Thinking about the meaning of voice, we are aided by talking in terms of the semantic roles of actor and patient/recipient rather than just using the same old familiar terminology of subject and object. A lot of the work done on voice recently addresses the semantics of the middle voice in new, interesting, and often times enlightening ways. Various studies have shown that the middle voice morphology is not used randomly, but tends to be used in certain types of verbs with certain types of meanings.

Beyond what the voices mean in the abstract and what their morphology is, there is another important component of the voice question: syntax. That syntax should be so important is implied by the history of Greek. An analysis of voice across the history of the language tells us that the most ancient form of Greek was probably an active-middle voice system, inherited from proto-Indo-European. The -θη- passive morphology in the aorist and future is a late-comer that gradually spread throughout different parts of the verbal system. This development continued to the point where, morphologically speaking, Modern Greek is an active-passive voice system, similar to English. At various points along the way, we will find cross-overs between morphology and syntax as the language changed. As the voice system changed, meanings which were at one time carried out by middle voice forms had to be farmed out to either active or passive voice verb forms.

In trying to wrap our minds around voice in the Greek verbal system, here are two important syntactic considerations:

  • the fact that the voices are not in equal syntactic relationship with each other
  • that the relationship between the middle and passive voice morphologies in the aorist appears to collapse over time into a variety of different features, prominently involving transitivity.

Both these developments deserve some comment.

Syntax, voice, and meaning: how do the voices relate in terms of sentences?

Syntax obviously is involved in voice. Remember that voice is a grammatical category that expresses the semantic functions attributed to the referents of a clause, indicating whether the subject is an actor or patient/recipient.[1] With this fresh in our minds, it is obvious that voice is tied into syntax at every step along the way.

The core of the English voice system, for instance, is an active—passive dichotomy. This is a syntactic relationship. Active voice and passive voice do not reflect a certain meaning, per se, but a syntactic relationship in which every transitive active sentence has a passive counterpart which means the same thing.[2] Consider the example from the last post:

  1. Actor[John] hit Recipient[the ball].
  2. Recipient[The ball] was hit Actor[by John].

These two sentences mean the same thing. They present the situation differently, but in terms of semantics, the two syntactic arrangements are equivalent. At its heart, the active—passive voice distinction in English is a matter of syntax, not meaning or morphology. Certain situations call for a passive, like when the Actor is not known, but generally the use of active is default and passive is an option. This is also more or less true in Greek.

As pointed out recently by Liana Tronci, and even more recently by Geoffrey Horrocks, in practice the voice system in Greek rarely shows an actual, free, three-way distinction between the voices. For example, summarizing some numbers from Tronci’s study of the NT, in the NT there are 1016 verbs which occur as an aorist finite verb. Of these, 805 of them (~80%) are attested only in one voice, with the active accounting from 580 of these. Of the 211 which occur in more than one voice form, the vast majority (167) are seen in active and passive. By contrast, only 9 verbs are attested in all three voices.[3] While these are summary statistics (for more detail, go out and read her article), they are telling in that in the NT robust voice variation is the exception rather than the rule. Most relevantly, the least likely voice to show voice variation with other voices is the middle.

In sum, there is rarely the option to use all three voices for a given verb without some sort of change happening to the meaning of the verb. Horrocks summarizes:

  • any active transitive verb may take medio-passive morphology in passive function and co-occur optionally with an agentive phrase (ὑπό + genitive etc.)
  • any active transitive verb with the appropriate lexical semantics may also take medio-passive morphology in a middle function, but cannot then co-occur with an agentive phrase.[4]

Here Horrocks points out that active transitive verbs have a syntactically productive relationship with the passive voice. They can all have a passive meaning equivalent. The active and the middle, by contrast, do not have a syntactically productive relationship across the verbal system. Only certain verbs with certain types of meanings have a middle meaning/syntactic version of the same verb. These verbs are common, which is what gives the impression that the middle voice is so common, but outside the core common verbs where the middle stands in relationship to the active, the middle is pretty much defunct already in Classical Greek.

Similarly, Tronci writes regarding voice in the aorist (the most complex part of the system, when it comes to voice) in the NT:

“The productivity of middle and passive aorists is not comparable. The aorist middle became a non-oppositive marker, whereas the aorist passive extended as an oppositive marker with respect to the active voice, by also replacing the intransitive middles, especially the unaccusative and the reflexive.”[5]

What Tronci is pointing out here is that the middle voice in the aorist—which has distinct “middle” and “passive” morphology—largely ceases to be used in a live system of voice distinction where a given verb can have all three different voices which present three different syntactical versions of the same verbal idea. I will have more to say in interacting with Tronci’s work as my thinking on voice continues.

So why do we still have middle verb morphology in the aorist and middle voice in certain verbs (“deponent”)? What is the middle voice actually doing?

Middle voice = lexical specialization

Two things happen with the middle voice instead: lexical specialization and syntactic specialization. Syntactic specialization will be a topic for the future. Here we will take up the reality of lexical specialization. The middle takes on a special lexical meaning which differs from the meaning of the verb in the active—passive pair. Rather than calling this voice alteration, we really should consider this a case of separate words. Consider, for example, this list of words in Daniel Wallace’s grammar, pointing to this phenomenon of lexical specialization as opposed to live-voice variation:[6]

αἱρέω – I take
ἀναμιμνήσκω – I remind
ἀποδίδωμι – I give away
ἀπόλλυμι – I destroy
δανείζω – I lend
ἐνεργέω – I work
ἐπικαλέω – I call upon, name
ἔχω – I have, hold
κληρόω – I appoint, choose
κομίζω – I bring
κρίνω – I judge
παύω – I stop (transitive) π
είθω – I persuade, convince
φυλάσσω – I guard
αἱρέομαι – I chose, prefer
ἀναμιμνήσκομαι – I remember
ἀποδίδομαι – I sell
ἀπόλλυμαι – I perish
δανείζομαι – I borrow
ἐνεργέομαι – I work (only impersonal)
ἐπικαλέομαι – i appeal
ἔχομαι – I cling to
κληρόομαι – I obtain, possess, receive
κομίζομαι  – I get, receive
κρίνομαι – I bring a lawsuit
παύομαι – I cease (intransitive)
πείθομαι – I obey, trust
φυλάσσομαι – I am on my guard
Active–Middle voice verbs showing lexical specialization

At its most pronounced, we see this pattern of lexical specialization reach extremes like ἅπτω “I light, kindle” and ἅπτομαι “I touch,” or ἄρχω “I begin” and ἄρχομαι “I rule.” These “voice” forms have become completely detached from one another such that they clearly should not be thought of as voice-variations of the same verb; they are distinct verbs. The various verbs in the above list, to varying degrees, also show this lexical specialization. They are a unique lexical item which probably should have its own entry in the lexicon rather than be tucked under an active headword, implying that this meaning is somehow tied up to the regular function of voice morphology.[7] Horrocks rather strongly summarizes this aspect of the Greek voice “system” this way:

when learners look up a given transitive verb [middle voice form, NJE] in a lexicon, they typically find that its middle in fact has a special sense, one that can only be connected with the supposedly “regular” indirect-reflexive sense via some tortuous special pleading of the type that tries to persuade us that “choose” is a semi-paraphrase of “take for oneself” etc. Pretending that these are somehow the straightforward middles of the corresponding actives in anything other than form is a disservice to students. They are clearly lexicalized verbs in their own right, with unpredictable meanings, and as such they deserve entries of their own in the lexicon.[8]

While saying things in a rather strong way, Horrocks has a good point here. One might paraphrase the basic contention in this way.

  • It is unlikely that Greeks saw any profound meaningful connection in the voice forms which created a unified meaning between ἄρχω “I begin” and ἄρχομαι “I rule.”
    • Learning Greek, you would have simply learned these as two completely distinct words that only happen to look and sound like a “normal” voice variation between active and middle.
    • Likewise with ἀποδίδωμι/ἀποδίδομαι. Again, these simply mean different things and should not be conceived of as an example of how the middle voice relates to the active voice in terms of mere voice variation.
  • The semantics of the core verb change between these two forms which were historically related, but which have gone their separate paths. There is no reason to believe Greek speakers thought of these two forms as voice variation because they do not work like voice variation. When you “change” to the middle voice, you are actually change the semantics, not just the ending.
  • While it is possible that historically there was a clear and predictable semantic extension of a “normal” middle meaning of subject-affectedness, by the time we reach Koine (and often already in Classical), such a voice distinction has given way to lexicalized verbs which have historically related forms but which users would no longer have perceived as related anymore because you can’t use an active version of ἀποδίδομαι in a sense meaning “to sell.” It doesn’t vary in voice that way.


What both these scholars are getting at is that in practice, the usage of the middle voice morphology is asymmetrical with the passive voice morphology. Actives and passives get along in a complementary fashion, while middles are the awkward third wheel. While the active—passive voice distinction in Greek is essentially syntactic, the middle voice has certain semantic limitations in its usage. Stretching back at least to Classical Greek, not every verb can have middle function expressed via middle morphology. Further, a great many of the middles which remain in usage in Greek do not actually participate in voice distinctions, even if there is an active voice form still in common use. Through whatever historical processes, these middles have become distinct lexical items which speakers would have to learn as such, since they have unpredictable relationships with their active “counterparts.” A description of the middle voice may uncover interesting historical relationships which connect the various forms and their semantic developments together, but that does not mean that Greek speakers had even the slightest idea that ἀποδίδωμι and ἀποδίδομαι were voice variations of the same verb from some point back in the murky past.

[1] This definition from SIL Glossary of Linguistics.

[2] There are, of course, exceptional verbs whose meaning puts them outside the normal active—passive dichotomy.

[3] Liana Tronci, “Aorist Voice Patterns in the Diachrony of Greek: The New Testament as a Sample of Koine,” Journal of Greek Linguistics 18 (2018): 245–46.

[4] Geoffrey Horrocks, “What’s in the Middle? Two Voices or Three in Ancient Greek?,” Keria: Studia Latina et Graeca 22, no. 2 (2020): 12.

[5] Tronci, “Aorist Voice Patterns in the Diachrony of Greek,” 247.

[6] Daniel B. Wallace, Greek Grammar Beyond the Basics: An Exegetical Syntax of the New Testament (Grand Rapids: Zondervan, 1997), 416.

[7] As per Horrocks, “transitive middles that were not assigned such “developed” meanings tended simply to drop out of use over time,” Horrocks, “What’s in the Middle? Two Voices or Three in Ancient Greek?,” 14.

[8] Horrocks, “What’s in the Middle? Two Voices or Three in Ancient Greek?,” 14.