Aorist voice patterns in the diachrony of Greek The New Testament as a sample of Koine

This paper deals with the aorist voice system in NT Greek and focuses on middle-passivemarkers,namelymiddleinflection,e.g.inthemiddlesigmaticaorist,andaffixes - η -/- θη -, in the so-called passive aorist. The research is corpus-based and investigates the occurrences of ca. 1800 verbal items. According to the grammarians, in the NT both middle and passive aorists spread.The present study confirms this observation by providing a comprehensive account of the distribution of these forms, but also shows how they have functionally reorganised. Passive aorists spread at the expense of middle aorists in all kinds of intransitive constructions, namely passive, unaccusative, and reflexive, whereas middle aorists are either found in transitive middles, e.g. possessive, benefactive etc., or occur as deponent verbs in both transitive and intransitive clauses. The parameter transitive vs intransitive appears to be relevant for this functional reor-ganisation.


Introduction
This paper deals with the verbal category of voice in Koine Greek and focuses on the aorist.As Browning (1983) and Horrocks (2010) pointed out, the aorist underwent major changes in voice from Ancient to Modern Greek: "the endings of the aorist middle (-(σ)άμην [(s)áme:n], -όμην [-óme:n], etc.) were increasingly replaced by those of the aorist passive (-(θ)ην [-(th)e:n], etc.)" (Horrocks 2010: 103) and, some centuries later, the replacement led to "[t]he reduction tronci Journal of Greek Linguistics 18 (2018) 241-280 of the three voices of Classical Greek to two" (Browning 1983: 30).The avoidance of middle aorists increased over the centuries and overlapped with the collapse of the functional distinction between aorist and perfect.In the early Byzantine period, for instance, the perfect γέγονα 'I have become' replaced the middle aorist ἐγενόμην 'I became' , which had already been in competition with the passive ἐγενήθην in the Koine (cf. Horrocks 2010: 256).
The aim of the paper is twofold.First, a synchronic analysis of the Hellenistic Greek system of voice is provided, through an in-depth investigation of the New Testament (henceforth NT).Second, some insights are developed concerning diachronic changes in the voice system from Classical to Hellenistic Greek.Since the language of the NT is "a reasonably close reflection of the everyday Greek of the majority of the literate population in the early centuries AD" (Horrocks 2010: 147), the NT offers a good sample for investigating the changes that occurred in Greek in the first centuries CE.
The data analysis concerns both quantitative and qualitative aspects.To carry out the quantitative analysis, all the aorist occurrences were collected, with special emphasis on middle and passive aorists.The search for all the relevant forms was made by combining a manual search through the NA28 edition of the Greek New Testament (text available on the website http://www .nestle-aland.com/en/the-28-edition/) with electronic tools (e.g.TLG, Perseus 4.0 released on www.perseus.tufts.edu/hopper/).The research was refined by using lexica, e.g.Kittel & Friedrich (1932-1936) and Thayer (1995).Our goal was to check how many middle aorists were replaced by passive aorists, how many were still functionally oppositive and how many were deponent or quasideponent forms.This quantitative analysis provides the basis for the qualitative one, which focuses on the values still retained by middle inflection in the Koine and the role of deponency within the voice system, with attention to aspects of both form and function.
Even though many studies have been devoted to the topic of voice, particularly middle voice, in Ancient Greek, most of them are not corpus-based, adopt a predominantly synchronic point of view, and mostly concern Homeric and Classical Greek.1 The perspective of this study is different in both respects.The corpus-based research provides a picture of how the voice system is organised within a single, but entire text.Even though the results obtained concern only the text investigated, they presumably reflect the voice system of the language in the first centuries CE.Further research could carry out a Journal of Greek Linguistics 18 (2018) 241-280 corpus-based analysis of texts contemporary to the NT, with the aim of giving a more complete description of the voice system(s) in Koine and Hellenistic-Roman Greek.Some differences may also be found from one text to another, depending on the language register, the regional varieties, the speakers'/writers' knowledge of language and further sociolinguistic parameters.The issue of sociolinguistic variation might also be dealt with by comparing the texts which compose the NT, e.g.Gospels vs Acts vs Letters, which are different in terms of genre, style, author, and addressee, and, among Gospels, the more literary and Classical-shaped Gospel of Luke vs the more popular and spoken-like Gospel of Mark.However, this issue is too wide and complex to be dealt with here.The reader is referred to Porter (2000aPorter ( , 2000b) ) for some preliminary analysis.
As far as the synchrony vs diachrony dichotomy is concerned, this study provides both a synchronic analysis of the aorist voice system in the NT and a diachronic comparison with both Homeric and Classical Greek.
The structure of the paper is as follows.In Section 2 a brief survey of the literature is provided.Section 3 is devoted to the analysis of data, by focusing on middle-inflected aorists in 3.1, passive aorists in 3.2 and some problematic cases in 3.3.In Section 4 the diachronic aspects of the aorist voice system are addressed, by comparing data from the NT with Homeric Greek (4.1) and Classical Greek (4.2).Finally some conclusions are drawn in Section 5.
Before starting the next section, a terminological remark is necessary.By passive and middle aorists, I mean the passive-and the middle-inflected aorists, namely the inflectional classes marked by the affixes -η-/θη-and the middle inflection respectively, no matter what syntactic or semantic values these forms carry.Both labels are misleading when referring to the aorist, since not all passive aorists are true passives and middle aorists can overlap with the passive ones in some middle values, the term "middle" being applied in some recent studies, e.g.Kemmer (1993), also to some functional values regardless of the verbal forms which express them.

A brief survey of the literature
It is commonly agreed among scholars that the system of three voices-active, middle, and passive-in general remained the same in the NT, and more generally in Hellenistic Greek, as in the Classical language (cf.Moulton 1906: 158-160;Blass, Debrunner & Funk 1961: 161).Besides the common idea of continuity with respect to Classical Greek, scholars also remark that there existed some differences between Classical and NT Greek, concerning in particular the relation- ship between middle and passive aorists.In his historical grammar, Jannaris (1897: §1478) stated (emphasis mine): [The middle and passive voices] should sooner or later finish by being completely identified […].The final outcome was then that the passive form soon gained ascendency over the middle and began to extrude it.One, probably also the primary, cause of this issue was that the subtle functions of the middle verb […] had been lost sight of in the popular language, which in the great majority of cases dispensed altogether with such fine shades of thought […], or, when indispensable, preferred to express them in a different, clearer mode, […] that is by means of a special complement […].This circumstance naturally led to the disuse and final disappearance of all indirect middle verbs […].It is now obvious that, as this class had been the commonest, the number of middle verbs was reduced to a very small group as compared with the hosts of their rival or passive verbs.
In accordance with this analysis, McKay (1994: 26) stressed that the deponent middles of Classical Greek also take on passive morphology and he quotes the example of the deponent middle ἀπεκρίνατο 'he replied' , which was replaced in the NT by "the normal Hellenistic aorist" ἀπεκρίθη 'he replied' (cf.§ 3.1).This example also occurs in Caragounis (2004: 152), who noted that the phenomenon "is old having begun to occur already in A [= Classical, LT] authors".He also refers to Hatzidakis (1892: 193-194), who remarked that "[v]on vielen Verbis kannten aber auch die Attiker nur die Formen des Passivaoristus in medialer Bedeutung" ("for many verbs, however, even Attic authors used only passive aorists in middle meanings"), e.g.ἐδυνήθην 'I was able' , ἠγέρθην 'I woke' , ἐκοιμήθην 'I fell asleep' , ἐκορέσθην 'I satisfied myself' , ᾠήθην 'I believed' , ἐπειράσθην 'I tried' , ἐσεβάσθην 'I was afraid' , ἐχάρην 'I rejoiced' , etc.This view is also shared by Aubrey's (2016: 573) study, which is devoted to the topic of voice in Hellenistic Greek: "[f]rom Homeric to Classical and into Hellenistic, -(th)ēslowly expanded in two directions: (1) enveloping the semantically and syntactically passive events, and (2) encroaching on and gradually displacing sigmatic middles as the marking of choice for certain middle event types".Some scholars also remarked that middle aorists were still abundant in NT Greek.This statement seems to conflict with the idea that passive aorists spread at the expense of the middle ones.Blass, Debrunner & Funk (1961: 161), for instance, pointed out that in NT Greek there is a "tendency to merge the middle and the passive into a single voice", formally represented by the passive aorist, and that "NT authors in general preserve well the distinction between middle  (1961: 165).Robertson (1934: 809) stated that "[i]n the flourishing period of the language [i.e.Classical Greek, LT] this [i.e.indirect middle, LT] was by far the most frequent use, but it finally faded before the active and the intensive (reflexive) pronoun or the passive.[…] But in general the indirect middle is abundant and free in the N.T.".
Summing up, two facts have been considered essential with respect to the aorist voice system of the NT: the spread of the "passive" aorists, which increasingly replaced the middle aorists, on the one hand, and the abundance of the middle aorists, on the other hand.Even though these two facts might appear to be inconsistent, this is not so.The fact that both classes are robustly attested in the NT seems to be in line with the state of affairs in Classical Greek.However, their functional distribution has changed with respect to Classical Greek.Our aim is to investigate the distribution of the two classes, in terms of both forms and functions, in order to determine the role of both inflectional opposition (middle vs active endings) and affixes contrast (-η-/-θη-vs -σα-) in the aorist voice system of NT Greek.

Analysis of data
The database for this research comprises 1838 verbal items (types) and a larger number of tokens.Some types are hapax legómena in the NT and are exclusively inflected in the present stem, e.g.γυμνιτεύω 'I am naked' , κουφίζω 'I am light, I relieve' , λυμαίνομαι 'I outrage' , μηκύνομαι 'I delay' .The types attested in aorist forms amount to 1016 by taking into account only finite moods (indicative, subjunctive, optative and imperative) and to 1222 by also counting non-finite forms (infinitives and participles).For simplicity's sake, our analysis only concerns finite forms.Before turning to the analysis of voice, some rough quantitative data is useful.This data strictly concerns the level of forms, so "active", "middle" and "passive" refer to the verb morphology.Within the 1016 verbal items inflected in the aorist, more than half ( 580 1 for a summary. In the following sections, I show the semantic and syntactic values of the three aorist voices in both their oppositive and non-oppositive uses.More emphasis is given to middle and passive voices, because in Hellenistic Greek they increased as types and entered into competition as strategies for marking middle meanings such as reflexive (e.g.I washed myself ), unaccusative (I grew), benefactive (I bought the book for myself ), possessive (I touched my eyes), etc.2 The middle voice tended to correlate with middle and deponent transitives, whereas the passive voice spread as a marker of intransitive structures, namely passive and unaccusative.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.

3.1
Middle aorists In my database, most of the middle aorists are deponent verbs, namely verbs with middle morphology and an active meaning in all forms of the paradigm.The functional equivalence between the middle morphology of deponent verbs and the active morphology of non-deponent verbs had been already remarked on by Mayser (1906: 112) in his grammar on Ptolemaic Greek papyri and was recently emphasised by Lavidas & Papangeli (2007: 105) who state: "the active and the middle/passive transitive interchange without any difference in meaning since the Koine period".Further evidence of this functional equivalence is given by the passive counterparts for the transitive deponents.In 9 of the 13 types which are inflected as both middle and passive, the morphological difference correlates with the functional opposition between transitive and Out of the total number of 148 types inflected as middle aorists, 105 occur in biargumental structures and 43 in intransitive structures.Among the biargumental verbs, 99 are fully transitive and 6 occur with the genitive (ἀντιλαμβάνομαι 'I help' , ἐπιλαμβάνομαι 'I help' , τάσσω 'I appoint' , φείδομαι 'I spare') or the dative (ὀνίναμαι 'I buy' , χράομαι 'I use').A summary of these data is provided in Table 2.
Let us return to the issue of deponency, which has been debated ever since the 19th century (cf.Kühner & Gerth 1892: §324;1898: § 377) and has received renewed attention in recent years (cf.Lavidas & Papangeli 2007, Ladewig 2010, Grestenberger 2014, Campbell 2015).In particular, Lavidas & Papangeli propose an account of both semantic and syntactic approaches to deponency and an insightful diachronic analysis from Ancient to Modern Greek.As they cogently argue, the proposed unified analysis of deponent verbs in terms of either semantic features or syntactic parameters does not match with Greek data.According to their investigation, "the data [i.e.Greek diachronic data, LT] lack any systematic pattern: an initially deponent verb may have an active variant with identical interpretation that is used in parallel or, even, instead of the deponent" (Lavidas & Papangeli 2007: 116).3 In the light of these considerations, one could interpret as "quasi-deponent" the verbs in the database which feature middle morphology in all aorist occurrences of the NT, while not being elsewhere deponent, namely ἀπονίπτω 'I wash' (hapax legómenon), ἀποτίθημι 'I put, lay down' , ἀσφαλίζω 'I make sure' , and κολάζω 'I punish' .Among the remaining media tantum aorists, ἅπτω 'I touch' and κομίζω 'I receive' are each attested only once with active morphology, ἅπτω in the present stem (Luke 15:8) and κομίζω in the aorist participle (Luke 7:37), and ἄρχω 'I begin' is attested only twice in the active voice (Mark 10:42 and Romans 15:12).For some of the remaining verbs, namely ἀπειλέω 'I warn' , συμβάλλω 'I help' , συμβουλεύω 'I counsel' , ὑποστέλλω 'I keep back' , there is no semantic difference between active and middle forms, whereas there is a difference for the last verbs of our list, namely καταρτίζω 'I prepare' vs καταρτίζομαι 'I am perfect' , κείρω 'I cut off (hair)' vs κείρομαι 'I am shaved' and κόπτω 'I cut' vs κόπτομαι 'I mourn' .For instance, the active forms (present stem) of κόπτω are transitive, e.g.(4a), whilst its middle forms (both aorists and imperfects) are transitive/intransitive, e.g.(4b): the meaning 'to mourn, to mourn for' of middle forms is presumably a semantic extension from the reflexive middle 'to beat or strike oneself' .The semantic disjunction between active and middle forms of the same verb is evidence of the loss of the syntactic, i.e. diathetic, relationship between them (see below for further details).This phenomenon had already begun in Late Classical Greek and the new deponent verbs attested in Hellenistic Greek are evidence of this (cf.Lavidas & Papangeli 2007: 105-107).
In the NT, there are also aorists which feature oppositive middle inflection.They are less frequent than the media tantum aorists and they also occur in both intransitive and transitive structures.In (5a)-(5c) I provide examples of some oppositive values of the middle, namely possessive, benefactive and unaccusative: ( The distribution of middle aorists is evidence of the functional decline of middle inflection in Hellenistic Greek.Even though it is widespread in many verb forms, the occurrences in which middle inflection expresses some middle  Linguistics 18 (2018) 241-280 values are very rare.It is preserved as a form, though it has lost its oppositive values.According to Caragounis (2004: 152), middle inflection in its entirety, the present stem included, is no longer productive as a voice marker in NT Greek.In interpreting the passage 15:2 of the Gospel of Matthew, in which the middle present νίπτονται [wash:prs.mid.3pl]'they wash' is combined with the object τὰς [art.acc]χεῖρας [hands:acc] αὐτῶν [3pl.gen]'their hands' , Caragounis remarks: "[t]he use of the middle with an expressed object indicates that the feeling for the fine distinction between the active and the middle has been lost" and quotes as further evidence the passages Acts 7:58, 26:21, Ephesians 5:16, and Titus 2:7.In my opinion, there are some essential differences between the present middle and the aorist middle in the NT.The present middle preserved all the functions of middle inflection, from the intransitive (the passive and the unaccusative) to the transitive ones (see Tronci forthcoming, who investigates the functional distribution of the middle forms belonging to all tenses and moods in the Gospel of Matthew for a total number of 979 occurrences).The aorist middle continued to exist as a form but lost its main functional values, e.g.unaccusative and reflexive, which were increasingly expressed by the passive aorist.The present middle, instead, still correlated with the intransitive middle values and tended to be replaced by the active in transitive middle values, namely benefactive and possessive.The analysis of changes in voice morphology by Lavidas (2009: 108) is in line with this view: "[f]rom the Hellenistic period, the marking of the benefactive ('personal interest') meaning by non-active [i.e.middle, LT] morphology begins to be abandoned".

3.2
Passive aorists The distribution of passive aorists within the database is as follows: 121 verbs are exclusively inflected as passive aorists, 167 are coupled with active aorists, 13 with middle aorists, and 9 with both of them.This picture allows us to have a clearer idea concerning the functions of passive aorists in the NT.The main function is to express the passive, e.g. ( 8a), and the unaccusative, e.g.(9a), by means of opposition to active morphology, e.g. ( 8b) and ( 9b The occurrence of the same verbal morphology in passives and unaccusatives is syntactically motivated.Both structures are characterised by the movement of an argument from underlying object status to subject status.The difference between the two clauses concerns the underlying subject of the clause; there is a subject present in the deep structure in the passive (whether expressed or absent in the surface structure), but one does not occur in the unaccusative (cf., among others, La Fauci & Tronci 2009).For instance, the anticausative events which occurred after the death of Jesus are shaped into a sequence of unaccusative passive aorists in the Gospel of .
In some cases, the choice between passive and unaccusative is a matter of interpretation, e.g. in (10a) compared to the transitive in (10b).This fact is hardly surprising, since the two voices, not being formally distinguished in Greek, are not expected to diverge much in meaning either (cf.Jankuhn 1969: 18-25 Many passive aorists occur in the NT as passiva tantum (121 types).A little less than half (50 types) are either deponent verbs, e.g.αὐλίζομαι 'I spend the night' , βούλομαι 'I desire' , δέομαι 'I pray' , δύναμαι 'I am able' , ἡττάομαι 'I am overcome' , ἱλάσκομαι 'I make reconciliation' , or quasi-deponent verbs, namely verbs which were coupled with a corresponding transitive causative in Classical Greek, but are exclusively attested in non-active voices in the NT, e.g.αἰσχύνομαι 'I am ashamed' , θαμβέομαι 'I am astonished' , κοιμάομαι 'I sleep' , ὀργίζομαι 'I am angry' , σκοτίζομαι 'I am darkened' , φοβέομαι 'I fear' .In the former group, the following are hapax legómena: ἀναγνώριζομαι 'I recognise' , ἀχρειοῦμαι 'I am corrupted' , ἐκκαίομαι 'I am kindled' , θυμοῦμαι 'I am enraged' , ματιοῦμαι 'I become futile' , πιστοῦμαι 'I am assured' , προσκληροῦμαι 'I am attached' , προσκλίνομαι 'I am joined' , προσορμίζομαι 'I draw to the shore' , ῥυπαίνομαι 'I am filthy' , σεβάζομαι 'I worship' , συγκαταψηφίζομαι 'I am numbered' , συνδοξάζομαι 'I am glorified' , συνυποκρίνομαι 'I act hypocritically' .In the latter group, there are both verbs which are inflected as passive aorists in Classical and Hellenistic Greek, e.g.αἰσχύνομαι (11a)-(11b), and verbs which occur as passive aorists in Hellenistic Greek but are active-inflected in Classical Greek, e.g.θαμβέομαι (12a)-( 12b The verb θαμβέομαι is a good example to illustrate the voice changes in Ancient Greek.The verb θαμβέω is attested in both present and aorist stems with active voice from Homer until Classical Greek, as shown by the aorist ἐθάμβησεν in (12b) and the imperfect ἐθάμβεον attested in Homeric poems, Pindar and Herodotus.The verb is commonly considered to be intransitive, the occurrences with a noun in the accusative commonly being interpreted as constructions with an accusative of relation, e.g.Summing up, the verb θαμβέω was activum tantum in Homeric and Classical Greek4 and became deponent (medium or passivum tantum) in Post-Classical Greek.This phenomenon is part of a larger phenomenon which concerned many verbs throughout the history of the Greek language.According to Lavidas & Papangeli (2007: 108), "[n]ot all deponents display stable behaviour through the different stages of Greek.In many cases, the deponent verbs have changed to actives, and, in other cases, they have attracted other verbs from the class of actives".The syntactic and semantic behaviour of the verbs which changed from active to middle/passive and vice versa remains the same, because the change concerns only verb morphology; cf.Lavidas & Papangeli (2007: 121).Deponency does not involve diathesis, since the morphology of deponent verbs, be they activa or media/passiva tantum, is not oppositive.Let us now turn to the 71 types of passiva tantum aorists that belong to non-deponent verbs.Most of them are attested in the NT with both active and middle morphology in stems other than the aorist.As for the aorist, only the passive is attested.See for instance the present transitive of ἀνακρίνω 'I judge' vs the aorist passive in (13a), the future transitive of πληθύνω 'I multiply' in (13b) vs the aorist passive in (13c), the perfect transitive of πιπράσκω 'I sell' in (13d) vs the aorist passive in (13e): ( 13 poor:dat 'Why was this ointment not sold for three hundred denarii and given to the poor?' ( John 12:5) Within the database, only the verb χαίρω is inflected as a passive in the aorist stem and as an active in the present stem without any difference in meaning.In both cases it means 'to rejoice, to be glad' (on this verb, cf.Among the 71 types of passiva tantum aorists, there are also some hapax legómena, namely διαλλάσσω (διηλλάγην 'I was reconciled'), διαλύω (διελύθην 'I was dispersed'), διανέμω (διενεμήθην 'I spread'), ἐξουδενέω (ἐξουδενήθην 'I was set at nought'), ἐπικαλύπτω (ἐπεκαλύφθην 'I was covered'), καταστρώννυμι (κατεστρώθην 'I was spread'), κληρόω (ἐκληρώθην 'I obtained an inheritance'), μετατρέπω (μετετράπην 'I was turned'), μορφόω (ἐμορφώθην 'I was formed'), πλήσσω (ἐπλήγην 'I was struck'), πρίζω (ἐπρίσθην 'I was sawn in two'), τυμπανίζω (ἐτυμπανίσθην 'I was tortured'), ὑπολείπω (ὑπελείφθην 'I was left'), and some verbs which are attested in all stems with non-active morphology, namely passive morphology in the aorist and future, and middle morphology in the present, imperfect, future and perfect.A couple of examples of each type follow in (15a)-( 15b A few passive aorists expectedly feature the affix η-.Following the alphabetic order of the dictionaries, they are ἀπεστράφην 'I turned away' , διηλλάγην 'I was reconciled' , διεσπάρην 'I was scattered' , διεφθάρην 'I was destroyed' , ἐνετράπην 'I was put to shame' , ἐξεπλάγην 'I was astonished' , ἐξετράπην 'I turn aside' , ἐπεφάνην 'I appeared' , κατεκάην 'I was burned up' , κατηλλάγην 'I was reconciled' , κατενύγην 'I was pierced' , μετετράπην 'I was turned' , ἐπλήγην 'I was struck' , συνετάφην 'I was buried' , συνεχάρην 'I rejoiced with' , ἐχάρην 'I rejoiced' .They are forms inherited from Homeric and Classical Greek which continued to exist in Post-Classical Greek, even though the affix η-was not productive.The semantic and syntactic values of the passive aorists with η-are comparable to those of the aorists with -θη-; see the passive (17a) and the unaccusative (17b): Before summing up, I discuss the case of the 5 verbs, namely ἀποκρίνομαι 'I reply' , γίνομαι 'become' , διαλέγομαι 'I reason' , ἐμβριμάομαι 'I groan' , χαρίζομαι 'I gratify' , which occur as both middle and passive aorists in the NT without any difference in syntactic distribution and meaning.The first verb, ἀποκρίνομαι 'I reply' , is attested 109× as a passive aorist and 7× as a middle aorist; see examples above in (3a)-(3b).The quantitative relationship between the two forms in the NT is inverted with respect to Classical Greek.In Classical Greek, the passive aorist is rare: it occurs 2× in Thucydides and 1× in Herodotus, whilst the middle aorist occurs 18× in Thucydides, 83× in Xenophon, 29× in Plato, 20× in Demosthenes.Furthermore, by comparing passive and middle aorists of ἀποκρίνομαι in Thucydides, two different meanings emerge, namely, 'to reply' for the middle and 'to be separated' for the passive: ( Let us now turn to the second verb of the list, namely γίνομαι 'I become' .The middle aorist ἐγενόμην and the passive aorist ἐγενήθην occur in the NT 301× and 40× respectively.The passive aorist is attested in the Gospels of Matthew, Mark and Luke, in the book of Acts, in the Pauline Epistles, and in the first Letter of Peter.In some cases, the two forms show the same textual distribution, e.g. ( 19a)-( 19b The syntactic and semantic distribution of the passive aorist is more restricted than that of the middle aorist.Even though some uses of the two forms overlap, they do not appear to be free variants in the NT. The verbs διαλέγομαι 'I reason' and ἐμβριμάομαι 'I groan' can be quickly dealt with because their middle and passive aorists are rarely attested in the NT (διελέξατο 2×, διελέχθησαν 1×, ἐνεβριμήσατο 1×, ἐνεβριμήθη 1×) and the occurrences of διελέξατο/διελέχθησαν, on the one hand, and ἐνεβριμήσατο/ἐνεβριμήθη, on the other hand, show the same syntactic distribution and meaning.
The last verb of the list is interesting because shows the emergence of a new passive voice in the aorist, expressed by the passive ἐχαρίσθην 'I was granted' , besides the deponent transitive ἐχαρισάμην 'I bestowed' .The middle aorist is attested 7× in the NT, whereas the passive is attested only once.The following pair shows the syntactic and semantic difference between the two forms, the former being transitive and the latter passive: ( 20 For it has been granted to you that for the sake of Christ you should not only believe in him but also suffer for his sake' (Philippians 1:29) Summing up, passive aorists are very productive in the NT.They spread at the expense of middle aorists in unaccusatives and intransitive deponents.Middle aorists continued to exist in transitive uses of deponent and quasi-deponent verbs, but lost most intransitive uses.As I show in Section 4, intransitive occurrences of middle aorists were attested in Homeric poems and had already decreased in number in Classical Greek.In Homeric poems we also find pairs of middle and passive aorists derived from the same verb and both occurring in unaccusative structures, e.g.ἐχολώθην/ἐχολωσάμην 'I became angry' (cf.Tronci 2011 and examples 27a-27b below).
The seemingly chaotic outcome of the competition between passive and middle aorists is evidence of some instability.However, there is some evidence that the system was being reorganised in binary terms: on the one hand, the diathesis expressed by both active and middle aorists, which increasingly correlated with transitive structures; on the other hand, the diathesis expressed by passive aorists, which embraced the core intransitive values of middle, namely the passive and the unaccusative, and extended into intransitive deponents.The reassessment shows the emergence of the syntactic feature transitive vs intransitive within the middle diathesis (cf.Section 5).

3.3
Some problematic cases Three occurrences in my corpus do not fit into the illustrated patterns.They concern the verbs ἀγαλλιάω 'I rejoice exceedingly' , ἀπολούω 'I wash' and βαπτίζω 'I baptise' .The first verb could be interpreted as a quasi-deponent, if it were not for the occurrence of one unaccusative active aorist: see discussion below.As for the latter two verbs, their middle aorists ἀπελουσάμην, ἐβαπτισάμην are commonly interpreted as passives, though they cannot be passive, because sigmatic middles are never passive in Ancient Greek, cf. Wackernagel (1926:137-138), Prévot (1935: 17-18), Schwyzer (1990: 757), Ambrosini (1996: 21).In the following we argue for a different reading through an investigation of all the occurrences of these verbs.The first case concerns the verb ἀγαλλιάω 'I rejoice exceedingly' , which is more frequent as deponent ἀγαλλιάομαι, according to LSJ s.v.In the NT, the verb is attested 4× with middle inflection, e.g. ( 21a), and 1× with active inflection, e.g. ( 21b), without any semantic and syntactic difference between them: ( 21 The inflectional opposition does not correlate with any difference concerning voice, unless we read (21b) as a transitive-causative clause 'my soul let my spirit rejoice in God' .This would be a contrived interpretation, however, and for this reason we prefer to adopt the common reading and try to explain otherwise the occurrence of active inflection (cf.also commentaries on Luke's Gospel, e.g.Carroll 2012: 47-48).The verb ἀγαλλιάω is attested 4× as a sigmatic middle (ἠγαλλιάσατο in Luke 10:21, John 8:56, Acts 2:26, 16:34), 1× as a sigmatic active (ἠγαλλίασεν in Luke 1:47) and 1× as a passive (infinitive ἀγαλλιαθῆναι in John 5:35).In the present stem, the verb occurs 4× as a middle (ἀγαλλιᾶσθε in Matthew 5:12, 1Peter 1:6, 1:8; ἀγαλλιώμενοι in 1Peter 4:16) and 1× as an active (ἀγαλλιῶμεν in Revelation 19:7), without any difference in meaning between active and middle.The verb is also attested in the Old Testament, consistently inflected in the middle voice.According to some commentaries, e.g.Plummer (1960: 31-32), the entire passage of the Song of Mary is modelled on the Song of Hannah in 1Sam.2:1-10, in which the verb ἀγαλλιάω does not occur, however.
Let us return to the active inflection of the verb in (21b), which is awkward with respect to the consistent picture provided by the other occurrences of the verb and needs, therefore, some further explanation.The passage has received much attention among scholars because the two verbs that occur in it present some peculiarities with respect to both voice and tense.Firstly, μεγαλύνει is a present and ἠγαλλίασεν is an aorist.The combination is unusual because the two verbs are coordinated and both refer to the present time.To explain this oddity, Buth (1984) argues that the passage was literally translated by Luke from an original Hebrew poem.In Hebrew, tense shifting is a matter of aesthetics and poetry and does not correlate with any shift in time reference.The Greek translator, presumably the evangelist Luke, "accurately transmitted the Hebrew poetic form and preserved the poetic tension of the Hebrew in the Greek" (1984: 74).By following this suggestion, it can also be argued that the active voice of ἠγαλλίασεν depends on the wish of the translator "to parallel μεγαλύνει", as proposed by Méndez (2016: 558, fn. 3).In sum, the peculiarity of ἠγαλλίασεν as far as tense and voice are concerned could be explained in terms of an ad hoc creation to translate an original Hebrew poem (cf.Plummer 1960: 30 who argues that Jews knew many of the Old Testament lyrics by heart and that Mary had recourse to these in the moment of exultation).A comparable hypothesis can be proposed for the active present ἀγαλλιῶμεν (Revelation 19:7).The verb is preceded by χαίρωμεν and followed by δώσωμεν, both of them active.The morpho-phonetic parallelism, on the one hand, and the semantic closeness to χαίρωμεν, on the other hand, could have allowed the verb ἀγαλλιῶμεν to take an active inflection.Consider, however, that the middle form ἀγαλλιώμεθα is provided by some manuscripts, even though it is presumably to be interpreted as lectio facilior.
The second problematic case concerns the middle aorist of ἀπολούω 'I wash' in (22a), which is commonly interpreted as a passive (cf.commentaries by Phillips 2002: 129 andHays 2011: 97 who translate it with 'you have been washed' and 'they were washed' respectively) along with the two subsequent passive aorists ἡγιάσθητε 'you were sanctified ' and ἐδικαιώθητε 'you were justified': (22)  With respect to the transitive clause in (22b), the clause in (22a) could be either passive 'you were washed' or reflexive 'you washed yourselves' .In both cases, the act of washing is metaphorical and implies 'to wash of sin': according to Hays (2011: 97) "Paul alludes here to [the] baptism".The interpretation as a passive has presumably been encouraged by the occurrence of the passive aorists ἡγιάσθητε and ἐδικαιώθητε in the sequence.We consider that the difference in voice between the middle ἀπελούσασθε and the passives ἡγιάσθητε and ἐδικαιώθητε is meaningful.The middle aorist could be interpreted as a reflexive, thus implying an involvement of the subject in the event.The act of washing themselves (ἀπελούσασθε) could therefore imply sanctification (ἡγιάσθητε) and justification (ἐδικαιώθητε): the three verbs "are three descriptions of the one fundamental transformation that has occurred for those who now belong to Christ" (Hays 2011: 98).This reading allows us to account for the sigmatic middle ἀπελούσασθε in accordance with the syntactic and semantic values of middle aorists which are attested in the NT as well as in Homeric and Classical Greek.Let us now turn to the third problematic case, which concerns the following occurrence of the verb βαπτίζω 'I wash, I baptise': The middle aorist ἐβαπτίσαντο is commonly interpreted as a passive (see also the modern translations on the website http://biblehub.com/),but this reading is not convincing and the syntactic structure of the clause leads to another interpretation.The verb is combined with two locative complements which designate the cloud and the sea.These complements also occur in the pre-  :aor.act.3pl](1 Corinthians 10:1) 'our fathers were all under the cloud, and all passed through the sea' .The reference to the Hebrews' flight from Egypt could also be a figure of speech for baptism.Modern commentaries on this passage point to the fact that "Paul reads the exodus story and the Christ as a double exposure" (Montague 2011: 166).Because of this double reference, the verb ἐβαπτίσαντο could have two complementary readings.It can be interpreted as a true reflexive 'they dipped themselves' and refer to the real fact that the Hebrews dipped into the cloud and the sea, which are, respectively, the symbol of God in the Old Testament and the prefiguration of baptism.If the figured interpretation and the reference to baptism prevail, the verb can be read as a causative middle 'they got themselves baptised' .The passive reading 'they were baptised' is also possible, if we adopt the lectio facilior ἐβαπτίσθησαν, accepted in the NA28 edition, instead of the lectio difficilior ἐβαπτίσαντο, transmitted by papyrus P46 of the Chester Beatty collection and accepted in both TLG and the Perseus Project.5 Both reflexive and middle causative analyses of ἐβαπτίσαντο are consistent with the other occurrences of the verb.In the aorist stem, the verb is attested 15× as a passive aorist with passive ('I was baptised') and reflexive ('I washed myself') meanings, e.g. ( 23b)-(23c) respectively, and 8× as an active aorist with an active and transitive meaning ('I baptised'), e.g. ( 23d The verb is also twice attested with middle morphology, here in (23e)-(23f).In the first occurrence, the verb means 'to dip, to wash' and is a reflexive.The comparison with (23c), in which the reflexive is expressed by a passive aorist, clearly shows that the two voices are in competition.In the second occurrence (=22b), the verb means 'to baptise' and can be interpreted as a causative middle (lit.'to get yourself baptised'), consistently with (23a): ( 23 The problem concerning the interpretation of βαπτίζω and its voices was posited by Moule (1959: 26) in his grammar of the NT.He lists the verb among "the specimens of problematic passages" and remarks: "[i]s there any significance in the comparison between Acts 9:18 ἀναστὰς ἐβαπτίσθη and 22:16 ἀναστὰς βάπτισαι καὶ ἀπόλουσαι τὰς ἁμαρτίας σου?Did a Christian in fact generally 'baptize himself' , 'get himself baptized' , or 'submit to baptism'?".It is not surprising that the interpretation of the verb βαπτίζω, which designates one of the core notions and practices of Christianity, is so problematic.In this verb, indeed, the linguistic issue of diathesis, in its etymological meaning of 'disposition' (of the arguments), meets the issue of baptism, its sacral value and its theological meaning.Which semantic roles do the subject and the object display?Are they 'one who baptises' vs 'one who is baptised' or, rather, 'one who lets someone get baptised' vs 'one who gets baptised'?The answers to these questions are not trivial for Christian religion and theology.In grammars and studies concerning the diachronic changes in the verbal system of Ancient Greek, the most hotly debated topics are the changes in future expressions from analytic to periphrastic strategies, the restructuring of verbal aspects and the disappearance of the optative (cf., among others, Horrocks 2010: 82, 102, 130-131).As far as the aorist is concerned, the merging with the perfect yielded major changes in the tense and aspect systems; see chapters 1-4, 14 and 17 of the book recently edited by Runge & Fresch (2016) and references therein.Not many studies have been devoted to the diachrony of the aorist voice system.Scholars limit themselves to remarking that passive aorists spread and middle aorists were reduced in number and were no longer productive (see the studies cited in Section 2 above).Since the aorist forms are the same as in Classical Greek, the functions are also expected to have been preserved, as Aubrey (2016: 620) argues for Hellenistic Greek: "-(θ)η-is used predominantly within its original five types, marking those events that shift the focus of attention to the more patient-like participant.The sigmatic middle aorist morphology predominantly marks the other middle types, which involve a more agent-like primary figure".This assumption proves to be mistaken, as I show in what follows; here, I compare the aorist voice system of the NT with that of Homeric and Classical Greek.

4.1
Homeric Greek In Homeric poems two different systems of voice are attested for the aorist (cf.Benedetti 2017 for more details).In the first system, inflection (active vs middle) marks the opposition between active and middle/passive voice, as in the voice system of the present.This system is presumably inherited and is not productive.It is only found with root and thematic aorists, see ( 24) and ( 25) respectively, and occurs with a few verbal items.What is peculiar is that middle inflection also marks the passive, e.g. ( 24b) and (25b), compared with the active transitive in (24a) and ( 25a The second system involves the most productive aorist classes, namely the sigmatic and the passive aorists.Two patterns of voice markers characterise this system.On the one hand, the affixes -σα-vs -η-/-θη-codify the opposition between active and middle-intransitive, namely passive and unaccusative.On the other hand, the inflection (active vs middle) codifies the opposition between active and middle-transitive, namely reflexive, benefactive, reciprocal, etc.The three terms are exemplified in (26a)-( 26c (Il. 14:214) In the pair (26a)-(26b), the opposition between active transitive and middleintransitive is marked by affixes, the inflection being active in both cases.In the pair (26a)-(26c), instead, the opposition between active transitive and middle- transitive is marked by inflection.The tripartite voice system is innovative with respect to the bipartite one, since it makes one further distinction within middle diathesis relevant, namely the syntactic difference between intransitive and transitive.The intransitive term is constituted by passives and unaccusatives.They are mostly expressed by passive aorists, but some sigmatic middles are also found in unaccusatives, e.g. ( 27a), besides passive aorists (27b  ' (Il. 13:206) The coexistence of the two forms in the same structure is evidence that the change is in progress.The verbal markers for the unaccusative are middle inflection in (27a) and the affix -θη-in (27b).They relate to the bipartite and the tripartite voice systems, respectively.As we argue in Section 4.2, in Classical Greek there was a tendency to reduce redundancy, by selecting one of the two forms according to their transitive or intransitive behaviour.

Concluding remarks
In this paper it has been pointed out that the aorist voice system changed in Hellenistic Greek with respect to Homeric and Classical Greek.The major changes concerned the middle and passive voices.Starting from Homeric Greek and even more in Classical Greek, sigmatic aorists and passive aorists became the most productive aorist classes.Sigmatic aorists display a bipartite voice system, based on inflection opposition.Passive aorists are opposed to sigmatic aorists through the affixes -θη-/-η-vs -σα-.The coexistence of these two classes created a tripartite voice system, which included (sigmatic) activeinflected forms, (sigmatic) middle-inflected forms and (passive) affixed forms.
This system spread at the expense of the bipartite system, which only displayed inflection opposition, concerned thematic and root aorists and was attested in Aorist voice system in NT Greek

Middle Passive
Transitive Intransitive

Aorist forms
Activeinflected aorists Middleinflected aorists "Passive" aorists Homeric Greek.In the tripartite system, the functional distribution of voice markers is as follows: active aorists occur in active clauses, passive aorists in passive and unaccusative clauses, and middle aorists in middle-transitive clauses, e.g.benefactives and possessives, in deponent verbs, and in some unaccusatives, e.g.ἐχολωσάμην 'I was angered' from χολόω 'I anger' .In Classical Greek, a tendency towards a clearer distribution of voice markers developed, mostly in the middle.Passive aorists spread into unaccusatives and intransitive deponent verbs.Middle aorists increasingly occurred in middle-transitive structures.However, in some instances, e.g.(32b), middle inflection appears to be unable to express the middle meaning of the clause and other middle markers, e.g.reflexive pronouns, had to be used.
In the NT, the aorist voice system displays the same forms as Classical Greek, with some differences concerning their functional distribution.Passive aorists increasingly replaced middle aorists of deponent and quasi-deponent verbs, e.g.ἀπεκρίθην 'I replied' instead of ἀπεκρινάμην, and were also productive with the new items, e.g.βαπτίζω 'I baptise' .In brief, passive aorists spread in all intransitive uses of the middle.As far as middle aorists are concerned, they continued to exist and maintained middle transitive meanings, e.g.benefactives, possessives and so on.Most deponent transitives continued to be middleinflected in the aorist; in these verbs the opposition between middle and passive morphology was used to express the functional opposition between active and passive, e.g. ( 20a)-(20b).All these phenomena clearly show that the transitive vs intransitive feature became even more relevant in the middle voice and led to a reorganization of the functional distribution of aorist voice markers, as Table 3 roughly represents.
The expansion of passive aorists at the expense of middle aorists is not only a matter of forms that replace other forms; it also concerns the level of functions.In other words, the productivity of passive aorists in Hellenistic Greek is the effect, at the level of form, of the syntactic feature associated with them, namely intransitivity.As an oppositive term, transitivity is expressed by both active and middle aorists, middle inflection being completely defunctionalised as a voice marker, with some rare exceptions; see examples ( 5) and (6a) above.
So, in agreement with Blass, Debrunner & Funk (1961: 180), we contend that "[t]he system of three voices of the verb-active (transitive), passive (intransitive), and middle (i.e.transitive with reference to the subject)-remains on the whole the same in the NT as in the classical language".Our research allows us to add that the functional distribution of sigmatic middle and passive aorists changed in NT Greek, even though the forms seem to be the same as in Homeric and Classical Greek.In our opinion, the reorganisation of the middle markers on the basis of the transitive vs intransitive feature is the first step in the creation of the Modern Greek bipartite system.However, further studies are needed to investigate whether and how this feature spread in later Greek texts.

table 1
Distribution of voice markers in finite aorists

Finite aorist types: 1016 Non-oppositive types Oppositive types
ἐρημοῦμαι 'I set free' for the passive voice.As can be seen, a large number of verbs which belong to the two latter classes are media tantum.Besides these forms, which do not feature any voice opposition, in the NT are also attested aorists which feature some voice opposition.The distribution is as follows.Only 9 types feature a tripartite opposition of active vs middle vs passive, e.g.βαπτίζω 'I baptise' , εὐαγγελίζω 'I bring good news'; the remaining 202 types show bipartite oppositions: active vs passive (167 types), active vs middle (22 types), and middle vs passive (13 types).See Table They amount to 89 types, e.g.ἀσπάζομαι 'I greet' , δέχομαι 'I receive, welcome' , διηγέομαι 'I describe' , δύναμαι 'I am able' , ἐργάζομαι 'I work' , ἰάομαι 'I heal' , κτάομαι 'I acquire' , ὀρχέομαι 'I dance' , χαρίζομαι 'I gratify' , and are the main group among the aorists exclusively attested in the middle voice in the NT (104 types).
him of the robe and put his own clothes back on him' (Matthew 27:31) Much more frequent are the cases in which only the middle is attested, even though the verb is not deponent and the active aorist is elsewhere attested, for instance in Classical Greek, e.g.(7a)-(7b): Downloaded from Brill.com 10/26/2023 02:27:23PM via Open Access.This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC-BY-NC 4.0).https://creativecommons.org/licenses/by-nc/4.0/Journal of Greek Linguistics 18 (2018) 241-280 ).
via Open Access.This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC-BY-NC 4.0).https://creativecommons.org/licenses/by-nc/4.0/tronci Journal of Greek Linguistics 18 (2018) 241-280 Pl.
Journal of GreekLinguistics 18 (2018) 241-280The passive aorist is never found in these kinds of clauses in the NT.It occurs with a lexical subject and adjectival or nominal predicative of the subject, e.g.
Downloaded from Brill.com 10/26/2023 02:27:23PM via Open Access.This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC-BY-NC 4.0).https://creativecommons.org/licenses/by-nc/4.0/ Journal of GreekLinguistics 18 (2018) 241-280The second occurrence of the verb in the NT is again with middle inflection, but the clause is transitive: Downloaded from Brill.com 10/26/2023 02:27:23PM via Open Access.This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC-BY-NC 4.0).https://creativecommons.org/licenses/by-nc/4.0/ Downloaded from Brill.com 10/26/2023 02:27:23PM via Open Access.This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC-BY-NC 4.0).https://creativecommons.org/licenses/by-nc/4.0/ . Further examples are provided by κορέννυμι 'I satiate' , μιμνῄσκω 'I remember' , ὁρμάω 'I set in motion, start' , πειράω 'I attempt, try' and the deponent verbs ἄγαμαι 'I wonder, admire' and ἔραμαι 'I love' .All the data for Classical Greek come from Xenophon's works.