Dionysius Thrax & Translating Πάθος

Click below to follow us via e-mail or RSS or Facebook.


Bernard Taylor’s paper in the 2010 SBL panel discussion on Greek voice and deponency surveyed the ancient grammarians to demonstrate that the concept of deponency has no grounding in what native Greek have to say about voice. In my view, it was convincingly argued, as I noted previously. In the question and answer time afterward, I asked Dr. Taylor if in his studies of the ancient grammarians whether or not he viewed the translation of πάθος as “passive” to be anachronistic or not. That is to say, have we been reading back into the ancient grammatical texts English syntactic categories.*

* On a side note that I will be picking up, those who argue that language typology is dangerous for the grammatical research on the basis that it might cause typology to be read back into the grammar of Greek must contend with what I would suggest is a more dangerous phenomenon: when one does not have a grounding in how languages function generally, the tendency to make Greek either more like English than it is (and thus safe and familiar) or less like English than it really is (and thus exciting and exotic). Four or five dissertations come to my mind that have done one or the other, but I will refrain from naming them out of respect for their authors who, despite that, are good scholars. Nobody is perfect, after all.

Rutgar Allan (2003) has already demonstrated that both the –μαι and –θη forms are basically semantic in nature and not syntactic. That is, they do not necessitate passive syntactic constructions. Consider for example: Josephus, Life 138 (the translation is Steven Mason’s and all the Greek below links back to Perseus):

μὲν ταῦτα ἔλεγεν, ἐγὼ δὲ τῷ θεῷ τὰ κατ’ ἐμαυτὸν ἐπιτρέψας εἰς τὸ πλῆθος ὡρμήθην προελθεῖν.
Although he was saying these things, I, having entrusted my affairs to God, set out to meet the mob in advance.

No passive here. The verb ὡρμήθην “set out” (or perhaps better, “rushed out”) can in no way be construed as a passive. Just a –θη form with its middle meaning.  But we’ll spend more time talking about Rutgar Allan’s work a couple posts from now. For now we can just observe that verbs denoting body motion consistently appear with the -θη form and do not have passive meaning and are not necessarily deponent (e.g. ἔρχομαι). ὁρμάω, the verb behind ὡρμήθην, also appears in Matt 8:32, Mark 5:13, Luke 8:33, Acts 7:57; 19:29.

Anyway, I think that if I had the time and opportunity to explain myself better than the 10 seconds that it took to ask the question that Dr. Taylor would be in basic agreement with me, but I would like to make clear where my question came from. What follows is an extended quotation from Masayoshi Shibatani’s (2004) article, “Voice.” For those of you non-linguists, who do not know who Shibatani is, see here. Stephen Carlson has provided the relevant excerpt from Dionysius in the comments.

Pardon any typos, I typed this whole quote out from the print edition sitting on my shelf. I think I’ve corrected them all, but I might have missed a few.

The middle (or medial) voice is considered to be the most heterogeneous voice category. But this heterogeneity can be only apparent because the variety of expressions encoded as middle are expressed by separated constructions in other languages. A typical middle form and its ‘ambiguity’ can be observed in the following Classical Greek middle expression from Homer (Andersen 1991:51)

(8) és hr’ asamínthous
to then bathing.tub (acc.pl)
go (aor)-part.aor-nom.pl
eüxéstas loú-sa-nto
well.polished (acc.pl) wash-aor-3.pl.mid

‘Then having climbed into the well polished bathing tubs they (a) washed themselves, (b) washed each other, (c) were (automatically) cleansed, (d) were washed.’

The middle inflection in Classical Greek expresses a range of concepts that are expressible, in English and some other languages, by four (and more) separate constructions; (a) the reflexive, (b) the reciprocal, (c) the spontaneous, and (d) the passive construction. Because of this, modern grammarians tends to view the middle voice as having a number of functions or senses each of which is uniquely expressed by a separate construction in some other languages. But such a view fails to capture the essence of the voice mechanisms of early Indo-European languages, from which the modern notions of voice as well as voice-related grammatical terms are inherited – some via Latin translations.

Unfortunately, there is some terminological inconsistency in the Greek grammatical tradition handed down from the classical grammarians such as Dionysius Thrax, whose Techne grammatike (ca. 100 B.C.) is considered a standard work on Classical Greek (cf. Art.6). Against the two distinct sets of inflectional endings for persona and number, which are identified as ‘active’ and ‘middle’, Dionysius distinguishes three diatheses ‘state, dispositions’ or voices: energeia ‘activity’ (from which the modern term active’ evolved from Latin activum), pathos ‘affection’ ( > Latin passivum > ‘passive’), and mesotes ( > ‘medium’, ‘middle’). Among these, the first two voice categories are primary ones that correlate directly with the inflectional categories of ‘active’ and ‘middle’. Dionysius exemplifies the diathesis energeia with tupto (hit. 1.sg.act) ‘I hit’ and pathos with tuptomai ((hit 1.sg.mid) ‘I undergo hitting’. That is, as a first approximation, the major voice categories of Classical Greek of energeia and pathos can be said to be correlated with the semantic distinction of whether the subject of the verb affects others (energeia) or it itself is affected (pathos). The mesotes category, as its name implies, combines the features of the two major voice categories; i.e. the active inflection with the pathos meaning or the middle inflection with the energeia meaning. Thus, the form identified as the middle inflection does not correlate directly with Dionysius’ mesotes voice. Rather the middle inflection represents the pathos voice, and this is what is normally recognized as the middle voice by modern grammarians. (One needs to exercise discretion in reading Kemp’s (1987) translation of Techne, in which ‘pathos’ is straightforwardly translated as ‘passive.) Notice that the four ‘modern’ interpretations of the middle form in (8) all express pathos meaning, i.e. the subject represents an affected entity or ‘the locus of the principle effects of the verbally denoted action’ (Klaiman 1991: 106).

I am interested in my reader’s thoughts on Shibatani’s view, particularly the section of mesotes. I’ve already shared this with a few people and all of them have expressed agreement with Shibatani’s analysis of the meaning of πάθος.

He makes no reference to “deponency” in the entire article, but his words here might be interpreted as an approval a valid category. Personally I don’t think he would make such a claim, particularly because he makes it clear that in his view the category mesotes has no formal realization in the language. And further, the entire issue of tantum verbs (whether activa or media) is not broached in any way.

The interpretation of the meaning of mesotes is so extremely complicated in Dionysius Thrax, that I highly doubt that there’s any benefit to spending much time on trying to draw a conclusion. There’s so little information in the ancient grammatical texts that I would be incredibly nervous to place too much weight on any interpretation of mesotes. With that said, the words about energeia and pathos are sufficiently clear that can minimally recognize that 1) Dionysius is using the terms to refer to  the -ω forms and the -μαι forms and that they are basically semantic in nature: activity and affected. There is absolutely no reference made to the passive syntactic construction.

I also want to emphasize that the verbal inflectional form labeled as πάθος in Thrax is a middle form (-μαι). Nowhere does Thrax make any reference to the supposedly “real” passives: –θη. Fundamentally, πάθος needs to be read on its own terms rather than through the lens of 2000 years of western grammatical terminology that have developed mostly apart from it: “that which is affected;” “state or condition;” “what one has experienced.” Considering that the middle is what is in debate these days and that those who reject deponency also just so happened to have attached themselves to the phrase “subject affectedness,” the fact that pathos can be so readily translated as affected should grab our attention.


Subject affectedness , or in Klaiman’s (1991) terms, “the locus of the principle effects of the verbally denoted action” is a better way forward for both understanding middle voice in general and understanding Dionysius Thrax’s term pathos.

Works cited

Shibatani, Masayoshi. 2004. “Voice.”Pages 1145-64. In Morphology / Morphologie: Ein Internationales Handbuch Zur Flexion Und Wortbildung/an International Handbook on Inflection and Word-Formation. Edited by Geert E. Booij et al. Berlin: Walter de Gruyter.
Klaiman, M. H. 1991. Grammatical Voice. Cambridge Studies in Linguistics. Cambridge: Cambridge University Press.